Você está na página 1de 242

Extending performance-based design methods by applying

structural engineering design patterns

by

John-Michael Wong

B.S. (University of California, Berkeley) 2003


M.S. (University of California, Berkeley) 2004

A dissertation submitted in partial satisfaction of the

requirements for the degree of

Doctor of Philosophy

in

Engineering—Civil and Environmental Engineering

in the

Graduate Division

of the

University of California, Berkeley

Committee in charge:

Professor Bozidar Stojadinovic, Chair


Professor Stephen A. Mahin
Professor James W. Demmel

Fall 2008
UMI Number: 3353354

INFORMATION TO USERS

The quality of this reproduction is dependent upon the quality of the copy
submitted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.

UMI
UMI Microform 3353354
Copyright 2009 by ProQuest LLC.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.

ProQuest LLC
789 E. Eisenhower Parkway
PO Box 1346
Ann Arbor, Ml 48106-1346
Extending performance-based design methods by applying
structural engineering design patterns
©2008
by
John-Michael Wong
1

Abstract

Extending performance-based design methods by applying

structural engineering design patterns

by

John-Michael Wong

Doctor of Philosophy in Engineering—Civil and Environmental Engineering

University of California, Berkeley

Professor Bozidar Stojadinovic, Chair

Performance-based design methods are powerful tools for designing structures that are

safe, reliable, and predictable in an environment with rare but potentially catastrophic haz-

ards, such as earthquakes. The design methods used today are made for the just-built state

of the structure, and place relatively little emphasis on other stages in the structure's life cy-

cle. Current design methods need extensions that specifically consider performance across

the entire life cycle of the structure, for example for the construction stage or the stage

of post-earthquake repairs. Such considerations are challenging during the early stages of

design because the precision necessary for analyzing structural performance during, for

example, construction and post-earthquake repair stages is typically unavailable until too

late after design decisions are made. Therefore, the task of extending performance-based

design requires an analysis of design information, analysis methods, and decision making.
2

Those needs are fulfilled in this thesis by investigating structural engineering analysis and

decision making methods using the design pattern method, and by defining the qualities

and types of design information.

Patterns describe procedures and practices for solving recurring problems in a general

way, which allows comparison and understanding of the tradeoffs involved in their use.

The definition of structural engineering design information categories reveals what infor-

mation is needed to perform certain types of analysis for construction and repair, and how

to communicate it at the appropriate level of detail. This description explains the different

types of information needed for conceptual design and for construction. The discussion

of analysis and decision methods using patterns identifies unnecessarily premature deci-

sions in existing structural design methods, and locations where mechanics models alone

are insufficient for understanding structural performance. The pattern definitions also allow

concise expression of design intent within structural engineering problem solving methods.

The modifications to performance-based design methods, developed through examples

using the pattern framework and presented in this thesis, incorporate performance consid-

erations for all phases of the structure life cycle. These recommendations include: con-

sidering additional analysis patterns during design, using decision patterns that account

for performance in all life cycle phases, and implementing set-based design methods for

evaluating multiple design options.


1

For from Him and through Him and to Him are all things.

To Him be the glory forever.

Amen.

— Apostle Paul (Romans 11:36 NASB)


ii

Contents

List of Figures v

List of Tables vi

List of Symbols and Acronyms vii

1 Introduction 1
1.1 What is this thesis about? 1
1.2 Why might this thesis be worth reading? 3
1.3 How can this thesis be read most effectively? 5
1.3.1 Other work 6
1.3.2 Chapter descriptions 9
1.3.3 Patterns 11

2 Performance-centric design 16
2.1 Concept phase 18
2.2 Design phase 19
2.3 Construction phase 22
2.4 Operation phase 23
2.5 Repair phase 24
2.6 Knowledge reuse 24
2.7 Summary 26

3 Information in structural design 27


3.1 Content of information 30
3.1.1 Material 30
3.1.2 Functional 30
3.1.3 Spatial 30
3.1.4 Process 31
3.1.5 Abstract 31
iii

3.2 Quality of information 32


3.2.1 Detail 32
3.2.2 Ambiguity 32
3.2.3 Maturity 33
3.2.4 Precision 36
3.2.5 Accuracy 37
3.2.6 Grouping: agglomeration and aggregation 37
3.3 Importance of content and quality 39
3.3.1 Mechanics models 39
3.3.2 Communicating quality of information 42
3.4 Summary 44

4 Design communication patterns 45


4.1 Communication fundamentals and hermeneutics 46
4.2 Patterns for communication 53
4.2.1 Narrative pattern 53
4.2.2 Sketch pattern 59
4.2.3 Annotation pattern 64
4.2.4 Spreadsheet pattern 68
4.2.5 Mechanics Model pattern 73
4.2.6 Rebar Placing Drawings pattern 76
4.2.7 Database pattern 83
4.2.8 Schedule pattern 87
4.3 Application to highway bridge loss estimation 91
4.4 Application to building information modeling 102
4.5 Summary 110

5 Analysis and decision patterns 111


5.1 Patterns for analysis 113
5.1.1 Abstraction pattern 113
5.1.2 Aggregation pattern 118
5.1.3 Agglomeration pattern 122
5.1.4 Updating pattern 125
5.1.5 Engineering Performance pattern 129
5.1.6 Social Performance pattern 132
5.2 Patterns for decision making 136
5.2.1 Pick Any pattern 136
5.2.2 Minimum Weight pattern 140
5.2.3 Minimum Cost pattern 143
5.2.4 Optimization pattern 146
5.2.5 Prescription pattern 149
5.2.6 Set Exploration pattern . 153
iv

5.3 Metadata for design process and design product 157


5.4 Examples of design and decision patterns 159
5.4.1 Scope of examples 161
5.4.2 Reinforced concrete beam design 161
5.4.3 Preliminary frame design 168
5.4.4 FEMA 350 steel moment connections 174
5.5 Summary 180

6 Extending performance-based design 181


6.1 Extension 1: Additional analysis for decision support 182
6.2 Extension 2: Change decision patterns 188
6.3 Extension 3: Use set-based decision patterns 192
6.4 Summary 197

7 Conclusion 198
7.1 What I have done here 198
7.2 Future work 200
7.2.1 Expanded pattern catalog 200
7.2.2 Communication patterns with BIM 201
7.2.3 Teaching and training engineers 202
7.2.4 Design with high ambiguity and unclear mechanics 204
7.2.5 Reliability assessment during design 205
7.2.6 Development of computer tools for supporting set-based design . . 206

A Pattern summary 223


A.l Communication patterns 223
A.2 Analysis patterns 224
A.3 Decision patterns 225
V

List of Figures
Fig. 2.1 Project phases 17
Fig. 3.1 Design variables for structure configuration 28
Fig. 4.1 Narratives describe structural design at a functional level 55
Fig. 4.2 Sketch example showing preliminary structural system 62
Fig. 4.3 Sketch example showing preliminary floor system 63
Fig. 4.4 Annotations enable proper interpretation 66
Fig. 4.5 Unit repair cost spreadsheet 71
Fig. 4.6 Mechanics model example 75
Fig. 4.7 Placing drawings example 79
Fig. 4.8 Relational data model for bridge performance database 86
Fig. 4.9 Schedule for bridge repair scenario task 89
Fig. 5.1 Updating pattern 127
Fig. 5.2 Social performance 134
Fig. 5.3 Pick Any pattern 139
Fig. 5.4 Minimum weight design aid 142
Fig. 5.5 Set exploration example for beam-column rebar 155
Fig. 5.6 Inelastic behavior of frames with hinges in beam span 176
Fig. 5.7 Location of plastic hinge formation 178
Fig. 5.8 Reduced beam section connection detail 179
Fig. 6.1 Additional analysis 183
Fig. 6.2 Considering construction phase performance 185
Fig. 6.3 Labor productivity rates for different bar sizes 186
Fig. 6.4 Changing decision patterns 191
Fig. 6.5 Using set-based decision patterns 193
Fig. 6.6 Required steel area based on mechanics model 195
Fig. 6.7 Bar layout options for beam and column 196
Fig. 7.1 Set-based design prototype tool 207
VI

List of Tables

Table 1.1 Pattern template 14


Table 3.1 Information content types answer different questions 29
Table 3.2 Information quality types answer different questions 29
Table 4.1 Qualities of communication patterns 49
Table 4.2 Loss estimation procedure for reinforced concrete bridges 93
Table 5.1 Design procedure for known b and h 163
Table 5.2 Target drift and yield mechanism method 168
Table 5.3 Connection design procedure based on FEMA 350 basic design ap-
proach 175
Table 5.4 Steel moment-resisting connection design procedure 178
Table 6.1 Original procedure for single column bent design 187
Table 6.2 Improved procedure for single column bent design 189
Vll

List of Symbols and Acronyms

The following symbols and acronyms are used in this dissertation:

a = depth of equivalent rectangular compression stress block


ab = depth of rectangular stress block at balanced strain conditions
Gtcl = depth of stress block at the tension-controlled limit
A = cross-sectional area
Aeff = effective area
As = cross-sectional area of steel
ACI = American Concrete Institute
ASTM = ASTM International, formerly American Society for Testing and Ma-
terials
ATC = Applied Technology Council
b = beam width, or mode shape factor
K = width of web
BDA = Caltrans Bridge Design Aids
BDD = Caltrans Bridge Design Details
BIM = building information modeling
Ce = wind exposure factor for gust and height, or seismic coefficient
CAD = computer-aided design
CBC = California Building Code
CRSI = Concrete Reinforcing Steel Institute
d = depth from top of beam to steel centroid
Dc = column cross section depth
Ds = bent cap depth
DL = dead load
DM = damage model
DS = damage state
DV = decision variable
E = elastic modulus
Ec = elastic modulus of concrete
Vlll

Es = elastic modulus of steel


EDP = engineering demand parameter
FEMA = Federal Emergency Management Agency
f'c = unconfined concrete compressive strength
Fi = design forces at the /th story
FiU = updated design forces at the nth story
Fy, fy = steel yield strength
h = beam height
H = building height
/ = moment of inertia for bending
IM = intensity measure
jd = distance between compression and tension resultants inside a cross sec-
tion
LL = live load
LLRCAT = local linearization repair cost and time methodology
LRFD = load and resistance factor design
M = moment
Mc = moment at centerline
Mf = moment at column face
Mn = modal mass, or nominal moment strength
Mp - plastic moment
Mpb = plastic moment of beam
Mpt,r = reference beam plastic moment
Mpc = plastic moment of column
Mpr = probable plastic moment
Mu = ultimate moment
My = yield moment
Myf = yield moment at column face
n = modulus ratio
PBEE = performance-based earthquake engineering
PEER = Pacific Earthquake Engineering Research Center
PG = performance group
PPC = percent planned complete
RBS = reduced beam section
RCR = repair cost ratio
Sf, = plastic hinge location
s = spatial distribution of external forces
SDC = Caltrans Seismic Design Criteria
T,T\ = fundamental period
Vf, = base shear
Vp = shear at plastic hinge
W = building weight
wc = unit weight of concrete
Zx = plastic modulus, plastic modulus about the strong
Pi = rectangular stress block coefficient
Pi = shear proportioning factor
r„ = modal coefficient
y = shape proportioning modification factor
&CU = compressive strain at crushing of concrete
e = curvature
e, = target drift
%> = yield drift
h = shear proportioning factor
V = displacement ductility
& = overstrength factor at /th story
P = reinforcement ratio
Pb = balanced reinforcement ratio
• = resistance factor
§n = mode shape of nth mode
(0 = reinforcing index
X

Acknowledgments

I am thankful to the Lord Jesus Christ for giving me new life, the privilege of learning

how to learn, and opportunities to serve others while studying at the University of Califor-

nia, Berkeley.

I am particularly thankful for my advisor, Professor Bozidar Stojadinovic. He provided

me the freedom, resources, and guidance to pursue a broad spectrum of research topics

within structural engineering and beyond. His flexibility and consistent encouragement

helped me navigate the course of my studies since my junior year of college. Boza is

among the best of the best and I am privileged to have worked with him.

The research on reinforced concrete bridges would have been impossible without the

teamwork of Professor Kevin R. Mackie. He helped me understand the ongoing work

within the Pacific Engineering Earthquake Research (PEER) Center, and his insightful ad-

vice and active work ethic made our research progress at record speed. I am thankful for his

assistance in helping develop my technical communication skills and fostering an attitude

of academic diligence.

The completion of my doctoral studies would not have been possible without the cama-

raderie of my fellow students. My long time officemates, Ady Aviram and Jeff Hunt, have

been a consistent source of academic sharpening and motivation. Kristen Parrish helped

me discover the world of design methodology and project management, and has been a

consistent source of cheerfulness. Matthew Dryden, Hong Kim, and Seongeon Jeong have

provided me with enduring fellowship throughout my doctoral studies and I very much
xi

look forward to co-laboring with them in the time to come.

I am thankful for the guidance and encouragement of Professors James W. Demmel,

Steven D. Glaser, Stephen A. Mahin, and Khalid M. Mosalam. Their service on my quali-

fying examination and dissertation committees completed the shape of this dissertation.

A special note of acknowledgment is due to Professors Iris D. Tommelein and H. Glenn

Ballard, for their gracious support and instruction in many academic areas beyond structural

engineering. Much of this dissertation is a product of the many synergies developed through

my interactions with them and the research initiatives within the P2SL laboratory.

My family has been a consistent source of strength, and has contributed to the comple-

tion of this dissertation in ways that 1 am unable to fully articulate. I am thankful for my

mom's sacrificial love and labor towards me. My brothers and sisters in the church have

spurred me on to love and good deeds and helped me maintain a proper balance among all

of my numerous roles and responsibilities. I am also particularly thankful for my fiancee

Clara and her encouragement, prayers, and flexible support during the final stages of my

doctoral studies.
Chapter 1

Introduction

This chapter answers the following questions: What is this thesis about? Why might it

be worth reading? How can it be read most easily and effectively?

1.1 What is this thesis about?

Current performance-based design methods focus primarily on structural behavior dur-

ing normal service and in a single earthquake event, but with relatively little emphasis on

other stages in the structure life cycle. This thesis develops new methods for specifically in-

cluding constructibility and reparability as performance measures for consideration during

design.

In order to accomplish this, I need to investigate basic design questions such as: When

are mechanics models sufficient, or insufficient, for making design decisions? If the me-

chanics models are insufficient, then what additional information is needed for making
2

decisions? What design prescriptions exist that must be followed, and why? What non-

engineering factors drive design decisions, and where should they come into the design

process? If structural engineering-based models lack content needed to make design deci-

sions, could a new model or methodology connect decisions back to mechanics fundamen-

tals?

These types of questions require fresh thinking about the nature of the design process

and the intentions behind current structural engineering design procedures. Basic terms

are defined which describe the categories and qualities of design information. The "pat-

tern" paradigm is employed to describe methods of communicating preliminary design

information, and the basic types of analysis and decision making that occur inside design

procedures. This provides a common platform for critiquing current design methods and

for understanding where the recommended design extensions fit into a design process.

This thesis describes the differences between quantities that can and cannot be deter-

mined using structural engineering methods. Conventional mechanics models for structural

analysis are unable to describe a fully complete design problem because those models are

simply unable to consider non-mechanical issues. Parameters such as cost, construction

time, and material availability are essential for the design process, yet are difficult or im-

possible to characterize within a mechanics-based model without using prescriptive design

constraints . The PEER PBEE methodology (Cornell and Krawinkler, 2000; Moehle et al.,

2005) is one example of attempting to bridge the gap between mechanics-based analysis

results ("engineering demand parameters") and non-mechanical factors such as cost and
3

time ("decision variables").

This description is important for achieving true performance-based design because

"performance" need not refer solely to structural response criteria (e.g. inter-story drift,

maximum moment, peak strains), but should also involve other factors such as cost, con-

struction time, reduced uncertainty, extent of earthquake damage, and anticipated downtime

and repair. In this expanded sense of "performance", correct performance-centric thinking

must be applied for all project phases, not just for the as-built structure.

This thesis applies the set-based design method for incorporating these additional mea-

sures of performance into structural engineering design. Set-based design evaluates many

feasible design solutions for as long as possible in order to gain the advantage of acquiring

more information. Design decisions are postponed until the last responsible moment so

that viable alternatives may be considered for their influence on the entire project.

1.2 Why might this thesis be worth reading?

This thesis is worth reading because it explains a method for including non-mechanics

based criteria into the design process, such as construction performance. Construction per-

formance is an important design consideration, but is often based on factors outside the me-

chanics model. This thesis also considers structural performance beyond the initial as-built

state by including possible post-earthquake repairs and constructibility issues. Earthquake

repairs are an appropriate consideration in high seismic hazard areas like California, and
4

especially for critical infrastructure projects such as highway bridges.

The main intellectual contributions of this thesis include:

1. Identification of unnecessarily premature decisions in existing structural engi-

neering design methods. This is accomplished by showing where and how decisions

are made using pattern descriptions. Describing decision-making methods that are

tacitly embedded in textbook design procedures reveals shortcomings that can be

easily improved by substituting a better decision making pattern in its place.

2. Identification of decision points in existing structural engineering design meth-

ods that assume the information from a mechanics model is sufficient, when in

fact the controlling variables may not be based on a mechanics model. Other

information not based on mechanics models such as cost, constructibility, repara-

bility, time, and material availability are also important for making sound structural

engineering design decisions. This is examined by analyzing patterns for structural

engineering analysis and decision making.

3. Definition of different categories of structural engineering design information.

These categories are needed for expressing design intent and selecting the appropriate

methods of communicating design information.

4. Explanation of patterns for design communication, analysis, and decision mak-

ing. Different patterns of design communication are explained with their suitability

for communicating the different types of structural engineering design information.

The patterns of analysis and decision making separate evaluation from design selec-
tion and reveal areas where more analysis can help inform better decisions.

5. Definition of metadata for analysis and decision making patterns, which are

useful for expressing design intent in structural engineering problem solving

algorithms. Knowing only the final design information, but without the underlying

information used to arrive at those decisions, can cause rework when the final design

turns out to be infeasible later in the design or construction process. Documenting

the reasons why decisions were made and why alternatives were discarded would

reduce this rework and promote learning.

6. Explanation that set-based design naturally fits with a proper understanding

of structural design information and the results from mechanics models used in

design. Mechanics-based design criteria are typically unable to indicate only one op-

timal design, but often leave several choices that can fulfill the requirements. For ex-

ample, designing the required area of reinforcing steel for reinforced concrete beam

flexure leaves several possible options for bar size and configuration. This leaves

room for a set of options to be evaluated.

1.3 How can this thesis be read most effectively?

This thesis can be read most effectively by understanding its context in light of my

other works, by understanding the organization of the chapters, and understanding the use

of patterns to describe methods of design.


6

1.3.1 Other work

This thesis is best understood in the context of my other works, which generally fall into

three categories: set-based design, highway bridge loss modeling for earthquake hazards,

and structural state data and metadata. The work on set-based design complements the

understanding of the Explore Set pattern and the advancements proposed in Chapter 6.

Chapter 4 uses an example of damage scenarios and performance groups from my work

on reinforced concrete bridge loss modeling. This thesis discusses the creation of damage

scenarios from a more general perspective than the detailed examples in Mackie et al.

(2008a), and suggests how bridge loss information might fit into a bridge design process.

Set-based design

Set-based design is a methodology for maintaining feasible design solutions for longer

in the design process than is otherwise affordable using point-based design, for the purpose

of obtaining input from several project participants simultaneously and early on. Set-based

design supports communication between these parties so that they can develop a more

globally satisfactory design solution while reducing rework.

An introduction to set-based design for structural engineering and reinforced concrete

design is presented in Parrish et al. (2007). That paper describes the general concept of set-

based design from manufacturing and applies the concept to structural engineering. Set-

based design is also described in its connection to lean construction. Parrish et al. (2008a)

explains a particular examples of set-based design for a new hospital in San Francisco,
7

California. That case study example describes the selection of the lateral force resisting

structural system, beam orientation, wall penetrations, and structure skin. Those examples

show an example of the set-based design method and structural engineering decisions that

were made based on construction performance. Parrish et al. (2008b) develops the concept

of "value propositions" which provide information about how different project stakeholders

rank different design options. An example is given for performance measures related to

rebar fabrication and placing time with different sizes and bending types of rebar.

The work on set-based design explains concepts behind the Explore Set pattern (5.2.6)

in more complete detail, and elaborates on the rationale for postponing commitment from

the lean construction perspective. The case studies present real examples of how set-based

design methods can be used in structural engineering practice at the level of preliminary de-

sign and at the detail needed for construction. The work on value propositions and commu-

nication gives several examples of analysis patterns for Social Performance (5.1.6). These

provide a glimpse into the performance measures that are important for the construction

phase and repair phase of a project.

Highway bridge loss modeling

Mackie et al. (2008a) contains the entire description of a method for performance-

based evaluation of reinforced concrete highway bridges including post-earthquake repair

costs and time. This introduces the local linearization repair cost and time methodology

(LLRCAT) for computing probabilistic repair cost and repair times to bridge components
8

for varying degrees of damage. This method retains the simplicity of automated, closed-

form solutions and creates a well-behaved model that links damage to repair quantity. The

LLRCAT method is summarized in Mackie et al. (2008b,c), and the data needed to support

the methodology are summarized in Wong et al. (2008).

The work on highway bridge loss modeling demonstrates and addresses the challenge of

design information availability in extending performance-based design into the repair and

operation phases. The level of precision needed to detail repairs is not yet available early in

the project during preliminary design. But those early stages are exactly where information

about construction and repair are the most useful, because design changes are the most

feasible and effective earlier rather than later. This challenge is overcome by creating

damage scenarios that provide just enough information for estimating repair quantity, cost,

and time, but with only limited information about the structure. This procedure is explained

in Chapter 4 using communication patterns.

Structural state data and metadata

Stojadinovic et al. (2004) describes a framework for integrating structural state data

from sensors. This framework defines metadata standards for sensor information, included

web-based visualization tools, and was demonstrated using wireless accelerometer data

from a shaking table test of a reinforced concrete bridge column. Wong and Stojadinovic

(2005) summarizes the specific work on metadata descriptions and the web interface, and

Wong et al. (2005) summarizes the work on wireless sensors, data acquisition, and data in-
9

gestion. Wong et al. (2006) proposes a method for using sensor data to improve estimates

on bridge performance for post-earthquake repair cost and repair time evaluation. Wong

et al. (2007) describes state data and metadata for the design process and compares the

structure of point-based and set-based design methods. Discrete event simulations demon-

strate the types of communication needed for set-based design.

The idea of using structural state data to update our understanding of a structure's per-

formance is a parallel concept to the Updating pattern (5.1.4), but applied to the design

process instead. In a sense, design information updates our conception of the structure

design while still on the drawing board. The concept of using metadata to classify infor-

mation about structure state is applied to the design process where the patterns can form

a metalanguage description that can help describe the design intent behind the structural

design information.

1.3.2 Chapter descriptions

A brief overview of each chapter is provided here to help the reader navigate the differ-

ent sections of the thesis. Readers who want a top-level overview and some examples for

design might want to focus on Chapters 2 and 6. Understanding the vocabulary developed

in Chapter 3 will help the reader grasp the discussion of patterns and examples in Chap-

ters 4 and 5, and the proposed extensions to performance-based design in Chapter 6. Each

chapter contains a final section giving a summary of the chapter's main accomplishments.

Chapter 2 provides an overview on how to incorporate the notion of performance into


10

planning for each phase of a structure's life from initial concept through its operational life.

This sets the context for applying the concepts discussed in the subsequent chapters.

Chapter 3 describes different types of structural design information and defines terms to

discuss them. The information content of a design changes with time as designers refined

the design by adding more information at greater levels of accuracy. Designs also mature

over time, replacing preliminary information that aids decisions with final information after

decisions are made. This chapter is presented first in order to provide the reader with

fundamental concepts about design that will provoke thinking about the design process

itself, instead of thinking about just one particular design or design method.

Chapter 4 describes ways to communicate the different kinds of design information de-

fined in Chapter 3. The performance of the design phase can be thought of as the clarity of

the finished design. Poor designs are either lacking in information, or lacking in clear pre-

sentation. This chapter explains principles for clear design communication and discusses

the nature of communication. This chapter is by no means exhaustive, and provides just

enough breadth in order to explain the points in this thesis. Future research would expand

this chapter substantially.

Chapter 5 describes patterns of decision making in structural engineering and the infor-

mation used to make those decisions. Again, this chapter is by no means exhaustive, and

intends to provide only the theory behind my contributions to performance-based design, in

a generalized manner. Future work on engineering design methodology could easily extend

the contents of this chapter.


11

The end of Chapter 5 walks through a sampling of common structural engineering de-

sign procedures and shows how the general patterns apply. The examples are intended to

provoke thinking about the design process itself and the thinking and judgment implied

by the calculation steps. The design methods may be presented in some textbooks as very

mechanistic, but they are in fact full of decision points where the engineer must practice

"best judgment" (Kardon, 2003). Seeing how the patterns apply in the common design pro-

cedures taught in books and articles will motivate the design process innovations explained

in Chapter 6.

Chapter 6 explains my extensions to performance-based structural engineering design.

The first example walks through the computation of post-earthquake repair time and repair

cost for reinforced concrete bridge design. The second example walks through an example

of incorporating construction performance into the design of a reinforced concrete struc-

tural element. The procedures employed makes use of the theory and concepts explained

in the preceding chapters.

Chapter 7 provides some closing thoughts about how the ideas in this thesis might

be applied to workaday structural design. I also indicate some directions for follow-up

research related to structural engineering design.

1.3.3 Patterns

Understanding the use of patterns to document and describe methods of design com-

munication, analysis, and decision making will help the reader follow the arguments in this
12

thesis. Patterns are the essence of common solutions to common problems. In describing

the design process in this way, I hope that the reader will be able to pick up on my obser-

vations about design and understand where my proposed extensions to performance-based

design fit in.

Patterns describe procedures and practices for solving recurring problems in a general

way, which allows comparison and understanding of the tradeoffs involved in their use.

This "pattern language" or "design pattern" method was coined in the field of architecture

by Alexander et al. (1977), and has been applied elsewhere in the field of computer science

in describing object oriented programming techniques (Gamma et al., 1995). Chapters 4

and 5 use the pattern method to describe the processes inside structural engineering design

at a certain level of specificity needed to focus on the thesis examples, yet also with a level

of generality so that the same concepts can be applied in other domains.

Anwar et al. (2005) discusses the use of patterns in developing software for structural

engineering design. They develop the idea of reusable components for making structural

design decisions, but limit the discussion to implementation of of computer software. Sim-

ilarly, Yu and Kumar (2001) describe the use of design patterns, but only for the specific

domain of finite element analysis. Also, Heng and Mackie (2008) describe five basic pat-

terns for finite element analysis with an emphasis on object-oriented programming design

for analysis software. In this thesis, the method of patterns for description is applied to

the general design process itself, without a specific view on how to implement in software.

There is often a tendency to reduce problems down to computable components and only
13

work on those (Wing, 2006). But, it must be remembered that computers cannot think

outside the box, because they are a box. Computers cannot come up with something inno-

vative because they are constrained by their own limitations. Rather, the human cognitive

process of learning and creativity is what must be pursued. For this reason, the patterns

in this thesis include the conceptual processes of communication and decision making and

not the computational processes of software and analysis.

The template used in this thesis to describe the patterns is based on Gamma et al. (1995,

p. 6-7) with modifications to fit the context of structural engineering design (Table 1.1).

Only the headings necessary for describing elements of structural engineering design in-

formation are used, and the descriptions for each heading are modified accordingly. Using

this common template makes comparisons between the different patterns more obvious.

The template also provides a method for recording the various trade-offs involved in using

each pattern in the design process.


14

Table 1.1: Pattern template.

Template section Description

Pattern name The pattern's name communicates the pattern's essence suc-

cinctly. The name is important because it becomes part of design

vocabulary.

Intent The pattern's intent describes what the pattern does, what its ratio-

nale and intent are, and the design issues or problems it addresses.

Also known as These are other common names for the pattern, if any.

Assumed Description of the assumed background training of the reader

background needed to correctly understand and apply this pattern. The de-

scription uses the categories of civil engineering knowledge de-

fined in American Society of Civil Engineers (2008) at the bache-

lor's level, with additional references as needed.

Motivation A scenario that illustrates the design issue or problem and how the

pattern is used to solve the problem. The scenario gives a concrete

example to aid understanding of the pattern's abstract description.

Applicability The pattern's applicability describes the situations where the pat-

tern can be applied, poor alternatives that this pattern addresses,

and guidance on how to recognize when the pattern is applicable.


15

Table 1.1: (continued)

Template section Description

Structure A diagram, illustration, or short example showing the a general

use of this pattern.

Consequences and The pattern's consequences describe how the pattern fulfills its ob-

implementation jectives, explains the tradeoffs and results of using the pattern, and

indicates what the pattern allows one to accomplish. Describes

any trade-offs, pitfalls, tips, or other information that designers

should be aware of when implementing this pattern. Also dis-

cusses any tool-specific information important for pattern use.

Known uses Examples of this pattern found in real structural engineering ex-

amples.
16

Chapter 2

Performance-centric design

The contributions of this thesis are best applied in the context of achieving a per-

formance-based structural design methodology that fully involves "performance-centric"

thinking throughout all the phases of a project. The performance of a structure during an

earthquake, for example, depends on its structural design. But the structural design does

not emerge from a vacuum, but rather as a product of the concept phase of a project. The

structure itself is a product of the construction and operation phases. Therefore, perfor-

mance needs to be considered throughout the entire project delivery cycle (Figure 2.1).

This consideration of structural performance in all project phases is "performance-centric"

design. This chapter walks through the various phases of a project and points at places

where current research on design methodology can contribute.


17

Performance targets
Concept
and project scope

Design Design information

Knowledge reuse

Construction Building process

Operation As-built state

Repair Post-disaster state

Figure 2.1: Project phases.


18

2.1 Concept phase

The concept phase is where the general function, purpose, and use of a structure are de-

termined. This is where structure performance must be first considered because this phase

establishes the general direction for a project. The general building topology, materials,

and general concepts for the basic structural system are determined here. Non-engineering

factors such as financing, architectural features, and permitting also happen here and if not

properly considered, could derail a project from efficiently fulfilling its performance goals.

To address these issues, research has been conducted with industry participants through the

UC Berkeley Project Production Systems Laboratory (P2SL). The documentation of suc-

cessful case studies of structures (mostly buildings) has provided insight into the factors

that make projects successful.

The concept phase is the appropriate time for choosing the target performance of a

structure beyond service-level loads. This includes performance during earthquakes or

other hazard events. Determining the target performance requires adequate pre-event plan-

ning in order to incorporate the constraints on the designs and construction processes, and

indicate the appropriate needs for the operation phase. The structure owner should define

the acceptable risk of loss and risk of hazard exposure in service and extreme hazard con-

ditions. For example, in nuclear structures, the acceptable probability of a core accident in

service conditions is set descriptively by federal law (10 C.F.R. § 100). Risk management

consulting would be the most appropriate during this phase.


19

2.2 Design phase

The engineering design phase includes completing design information at the level of

detail needed for construction, and obtaining the permits necessary to begin construction

work. Three challenges to implementing fully performance-based thinking within the de-

sign phase exist today:

1. Producing analysis results that are meaningful to owners and other project partici-

pants is a difficult task that requires more supporting data and analytical procedures

than required for presenting the design to engineers, experienced builders, or a spe-

cialized audience who has expertise.

2. A disconnect often exists between what is designed in the office and what is optimal

for construction in the field.

3. An entrenched, point-based design process makes it difficult to explore design op-

tions for system-optimal structural solutions.

The first issue is addressed through the outcomes of the performance-based earthquake

engineering (PBEE) methodology developed through the Pacific Earthquake Engineering

Research (PEER) Center. This method of performance-based design enables consideration

of non-structural-engineering decision variables (DVs) by (1) extending quantification to

such decision variables, and by (2) providing a probabilistic acceptance criteria to deter-

mine whether performance is acceptable with an adequate reliability. The DVs could be

expressed in terms of dollar cost, material quantities, and probability. This analysis is per-
20

formed using analytical structure models subjected to earthquake ground motions, which

produce a relationship between hazard and demand. Then, that demand is correlated to

damage through empirical data. And finally, the damage is linked to repair methods, quan-

tities, and costs based on historical and projected construction data. An example of this

method was developed for determining the repair cost and repair quantities of a California

reinforced concrete highway overpass bridge after an earthquake.

The second issue touches upon limitations in the process capabilities of contractors

and the interpretation of structural design communication, such as drawings, which are the

principal means of expression in practice today. An example of limited process capabilities

can be seen when designing sections with large amounts of congestion. Even if the design

seems to work on paper, installation in the field might be unfeasible due to tolerance is-

sues and ambiguous details expressed in a 2D paper space. Or, the as-designed structure

might cost much more than expected due to hidden complexity. These issues are addressed

through design methodology research using the example of identifying factors contributing

to the notion of "congestion" in reinforced concrete structural elements.

Set-based design is proposed as a solution for the third issue, specifically for reinforced

concrete structures. Set-based design can be thought of as a funneling or narrowing pro-

cess. An example of reinforced concrete building structures are provided in this thesis to

illustrate this general design method. The point-based design method goes from one trial

solution to another attempting to arrive at a final, feasible solution. The set-based design

method begins with the whole space of possible design solutions, and then intelligently
21

narrows the design set by making decisions at the "last responsible moment" (Ballard,

2000). Set-based design reduces rework and provides an opportunity to document why cer-

tain decisions are made. This documentation step helps to express original design intent,

and could be useful when designs need to change due to unexpected circumstances in the

field. When design intent is described in addition to a design solution, richer collaboration

can occur between engineers, general contractors, and rebar fabricators. Expressing design

intent demonstrates the implementation of decisions made in the concept phase.

All of the communication, design, and decision patterns have strong bearing during the

design phase. The communication patterns are especially important to implement prop-

erly because of the different levels of conceptual and concrete information communicated

during the preliminary and final stages of design. The analysis patterns show the different

types of analysis that need to take place during design. The patterns for decision mak-

ing reveal different methods at choosing among various design options. The use of the

best decision patterns within design procedures can provide a solution with much greater

performance. Many design procedures focus solely on Engineering Performance (5.1.5)

without considering Social Performance (5.1.6).

Furthermore, the design itself can be measured for performance. The performance of

the design phase can be thought of as the clarity of the finished design. Poor designs are

either lacking in information, or lacking in clear presentation. The discussion of hermeneu-

tics and patterns for communicating design information at the preliminary and detailed

levels explain methods for presenting designs in a manner appropriate the information.
22

2.3 Construction phase

Structural performance is not only an issue for the completed, final product, but rather

should also be a consideration throughout the construction phase of the project. During

construction, some demands are much higher than during normal service. For example,

mobile cranes and concrete trucks can impose significant demands on supporting columns

and foundations. Moreover, high quality construction is necessary to ensure that critical

details are properly built in the final structure. Therefore, it is of vital importance that a

full discussion of structural performance considers not only service and extreme events, but

also the construction phase of a project.

Describing design intent fosters a dialog between designers and builders which can

promote selection of more efficient details. For example, certain size bars specified in a

design might be more costly or difficult to install than other options. Cases studies and

interviews with structural engineers, general contractors, and rebar fabricators have helped

document these factors for structural design (Parrish et al., 2008a,b). Work flow simulation

has been performed using a Java web services implementation to experiment with the type

of information necessary to collaborate in a set-based manner (Wong et al., 2007).

Communication patterns requiring high maturity are the most useful here. For exam-

ple, rebar placing information is essential for fabricating, delivering, and placing rebar for

reinforced concrete structural elements. A well-communicated design will aid construction

inspections by reducing the ambiguity in interpreting design intent among inspectors and

builders.
23

2.4 Operation phase

The operation phase of a structure allows verification of the structure's actual perfor-

mance against the original design estimates. Previous work on sensors and state data frame-

works provides the mechanism for capturing and querying performance data over time.

Using those frameworks, monitoring data can be made readily available for further use in

performance evaluation for the as-built state and post-hazard state, and estimation of future

performance related to long-term deterioration and degradation and estimates changes in

hazard exposure or service usage patterns. The major barriers to using the data involve

determining how to use the data for performance evaluation and identifying appropriate

questions to answer with the data.

One application of sensor data is to update a structure's demand model. A demand

model connects an engineering demand parameter (EDP) during an earthquake with a

given earthquake intensity measure (IM). Before an earthquake, the ground motion and

the structure's response are uncertain. After an earthquake, the IM can be determined from

seismograph records, and the EDP can be measured with structural monitoring sensors.

The observed response and expected response can be combined using the Bayesian update

rule to generate an updated demand model.

For a large earthquake, this method could be applied to post-earthquake rapid infras-

tructure assessment-like a "triage" for buildings and bridges. For moderate or small earth-

quakes, the data could be used to update performance estimates for future earthquakes.

The Updating pattern is useful here when applied to structural state data. The structural
24

state data from sensor monitoring information, or inspections can be used to update earlier

estimates on the structure's properties.

2.5 Repair phase

The repair phase of a project refers to the recovery process after a major catastrophe,

such as a large earthquake. The repair phase is typically not considered during initial

design for typical buildings and civil infrastructure, but current research is beginning to

address it. The LLRCAT methodology addresses the repair phase for reinforced concrete

bridges (Mackie et al., 2008c,a). Others have begun including the repair phase as part of

sustainability analysis during the design phase (Kang and Kren, 2006; Kneer and Maclise,

2008).

Special structures may have specific design procedures for repair and recovery after ex-

periencing a hazard. Hospitals might be designed to need little repair in small earthquakes,

but allowed to require some repair in large earthquakes. Nuclear structures are designed

for safe shutdown and immediate re-start after a design-basis hazard, and to accommodate

emergency shutdown in an extreme hazard beyond the design basis.

2.6 Knowledge reuse

Lessons are inevitably learned from every phase of every project. The operation pha-

ses provides data from as-built behavior; the construction phase can provide feedback to
25

design; the design phase can capture the reasons for choosing one option over others; and

the concept phase provides the rationale for selecting performance objectives. Once per-

formance is observed for a particular structure the data can be reused to improve future

designs. Future designs are often influenced through design code revisions which typically

require 5-10 years. Given this long feedback cycle, retaining design information is par-

ticular challenging. This learning process can be facilitated by using set-based design and

specifically documenting design intent.

In the case of retrofit, retaining knowledge of the original design intent and construction

documents would help engineers assess the structure's capabilities and options for retrofit.

The retrofit design process starts many years after the original structure is built and can

reuse much of the information from the original design.

Cost functions based on rebar reinforcement ratios and "congestion" in a joint is one

example of knowledge reuse. Data from the construction phase and performance data af-

ter an earthquake can be synthesized to enable efficient set exploration early in the design

phase. Research conducted through industry workshops has revealed the connections be-

tween communication during the design phase and the ability to reuse knowledge in the

future.

Knowledge reuse is encapsulated in this thesis through the use of patterns to describe

the concepts and procedures. The use of patterns allows one to reuse the methods and

understand why one certain procedure is more or less useful in different situations.
26

2.7 Summary

What I have done in this chapter:

1. Denned phases in the life of a structure by describing dominant activities in each

distinct phase.

2. Explained the importance of understanding performance in all phases.

3. Highlighted applications and research related to each phase.


27

Chapter 3

Information in structural design

This chapter defines terms for categories of information that express structural engineer-

ing design. These categories are used in the following chapters to explain the differences

between patterns of communication (Chapter 4) and the different emphases of the various

design analysis and decision patterns (Chapter 5). Using these terms helps clarify design

intent at the point of decision, and simplify discussion of how design variables converge to

their final values.

The discrete pieces of information are called variables because they develop with time

as the design process progresses. ATC-63 lists six basic design variables: occupancy and

use, elevation and plan configuration, building height, structural component type, seismic

design category, and gravity load. These design variables have related physical proper-

ties that express precise information (Figure 3.1). Additionally, Schon (1983, p. 98) de-

scribes "design domains", which encompass a larger range of design variables for building
28

Rt'laled Physic.il Properties


• Typical framing layout
• Distribution of seismic-force-resisting system
Occupancy and Use components
• Gravity load intensity
• Component overstrength
• Distribution of seismic-force-resisting
components
• Typical framing layout
• Permitted vertical (strength and stiffness)
irregularities
• Beam spans, number of framing bays, system
Elevation and Plan regularity
Configuration • Wall length, aspect ratio, plan geometry, wall
coupling
• Braced bay size, number of braced bays,
bracing configuration
• Ratio of seismic mass to seismic-force-resisting
components
• Ratio of tributary gravity load to seismic load
• Story heights
Building Height
• Number of stories
• Moment frame connection types
Structural Component • Bracing component types
Type • Shear wall sheathing and fastener types
• Isolator properties and types
• Design ground motion intensity
Seismic Design
• Special design/detailing requirements
Category
• Application limits
• Gravity load intensity
• Typical framing layout
Gravity Load
• Ratio of tributary gravity load to seismic load
• Component overstrength

Figure 3.1: Design variables for structure configuration have related physical properties
(ATC-63, p. 4-4).

projects including: program and use, siting, building elements, organization of space, form,

structure and technology, scale, cost, building character, precedent, representation, and ex-

planation. These domains involve architectural style and artistic expression in addition to

structural configuration. Design information can be categorized by content using descrip-

tors shown in Table 3.1 and by quality using descriptors shown in Table 3.2.
Table 3.1: Information content types answer different questions.

Content Typical Questions Answered

abstract how? parameters that are handy for design

functional why? what are the reasons?

material what is it made of?

process who will make it? when will it be made?

spatial where will it be?

Table 3.2: Information quality types answer different questions.

Quality Typical Questions Answered

accuracy does it hit the target?

ambiguity how well are the information relationships known?

detail what amount of precision is available?

grouping what groups of pieces are referred to?

maturity what will develop or change in the future?

precision how close to the bull's eye?


30

3.1 Content of information

3.1.1 Material

Material variables describe an object's composition and intrinsic characteristics. This

includes whether the object is made of concrete, steel, wood, or some other material. It

includes the different grades of the material and any measurements on the composition and

engineering properties.

3.1.2 Functional

Functional variables describe the object's purpose and role in fulfilling the performance

goals of the structure. This information would describe that a shear wall serves to resist

lateral forces in a building, a beam carries the gravity load for a floor system, and that a

column carries gravity loads for a portion of the building. The functions of the individual

elements sharpen as the design process continues. Items exist in the design because of their

function in fulfilling the structure's purpose.

3.1.3 Spatial

Spatial variables describe an object's location, size, or shape. This includes information

such as the location and configuration of reinforcing bars inside a concrete structural ele-

ment, the position of columns and beams inside a building, and the shape of a steel beam

or floor slab. The spatial information tells where the material goes.
31

3.1.4 Process

Process variables describe an object's method of construction or installation. This

would include phases and the sequence of construction along with any dependencies with

other pieces of the structure. This involves schedule information, procurement of the mate-

rials, positioning of the pieces, and guarantees that the built structure matches the designed

structure.

3.1.5 Abstract

Abstract variables deal with concepts that are handy for thinking and analysis. Abstract

variables are involved in the design process but are not necessarily needed for specifying

material and spatial properties nor for expressing function. For example, the reinforcement

ratio p is a design variable that is based on spatial and material information but is not

itself specifying the material composition or position, size, and shape of individual bars.

Other examples include the total cross-sectional area of reinforcing steel in a reinforced

concrete element, Aeff, the reinforcing index, co, and the ratio between repair cost and

initial construction cost, RCR. Abstract variables are typically derived from mechanics

models used for structural analysis.

Abstract variables can also be related to process information when dealing with mea-

surements of construction performance such as PPC, the percent of planned items that were

completed.

"Performance goals" are abstract variables, because they are not intrinsically part of
32

a real object. Structural "performance" refers to a measurement of a structure's behavior

usually compared to an estimate based on a mechanics model. Performance could also

refer to the state of a structure during or after a specific catastrophic event (e.g. earthquake,

fire) or over a specified duration of time (e.g. 50 year performance). Performance then is

just another variable with different kinds of functional qualities. Structural performance

metrics apply at a specific level of detail, scale, abstraction, and aggregation.

3.2 Quality of information

3.2.1 Detail

Detail refers to the level of geometric specificity at a given scale. Low detail means that

very few of the spatial details are available. A 3D view or 2D drawing of something with

low detail would perhaps show the general shape of an object, but would leave out other

elements. Description with high detail can be viewed at a large scale and have features that

are visually prominent. A high level of detail is needed to make the designed object look

like what real object does. The highest level of detail is only achievable with the actual,

real object itself. Other representations of the object have less detail.

3.2.2 Ambiguity

Ambiguity is defined by Schrader et al. (1993) in two levels. Level 1 ambiguity is "Char-

acteristic of a situation in which the problem solver considers the set of potentially relevant
33

variables as given. The relationships between the variables and the problem solving algo-

rithm are perceived as in need of determination." Level 2 ambiguity is "Characteristic of a

situation in which the set of relevant variables as well as their functional relationship and

the problem-solving algorithm are seen as in need of determination." The concept of model

uncertainty in structural engineering (Ditlevsen, 1982) is related to level 2 ambiguity since

both relate to missing or neglected relevant variables.

These two levels of ambiguity function differently in a design situation. When some-

one is "figuring out a problem" it usually involves starting with a level 2 ambiguity and

working towards a level 1 ambiguity. Then, once that is determined, the solution method

can be chosen and the ambiguity can be resolved at the proper level of precision, detail,

and stability.

Most design methods work on the basis of level 1 ambiguities. They serve as a prob-

lem solving algorithm that provides reasonable values for each of the important variables.

However, one problem is that many design methods assume a point-based solution by only

allowing expression of variable values to a high degree of precision. Identifying the preci-

sion tolerances on these variables would leave set exploration more open.

3.2.3 Maturity

Maturity is the characteristic of design information that it has developed into a stable

form, close to its final value, and has enough precision needed for construction. Mature

variables have been informed by all the prerequisite information. In the literature, maturity
34

with respect to the amount of information content and variable values is referred to as

certainty or uncertainty, and maturity with respect to changes with time as stability. The

concept of maturity in this thesis refers to both aspects.

Schrader et al. (1993) defines uncertainty as "Characteristic of a situation in which the

problem solver considers the structure of the problem (including the set of relevant vari-

ables) as given, but is dissatisfied with his or her knowledge of the value of these variables."

Terwiesch et al. (2002) defines stability as "the likelihood of changing a piece of informa-

tion later in the process." This also depends on the precision of the piece of information.

Low precision information is usually more stable, and high precision information is usually

less stable. When decisions are locked in as the design passes certain milestones, then the

locked down information is hopefully completely stable. A finished project is completely

stable, aside from information related to building operation.

Choosing a value that is highly precise at a stage of design that is too early might lead

to high uncertainty. Uncertainty involves not knowing the exact value of the variables at the

earlier stages of the design process. In some sense, progress during design involves labor

in order to reduce the uncertainty of the variables. Variables in the design are not always a

single numeric value. For example, the shape of an object might be a variable which needs

to be described in a more sophisticated way than a single linear dimension.

The term maturity is used instead "uncertainty" (Schrader et al., 1993) because it is

different in nature than that of statistical or model uncertainty (e.g. Ditlevsen and Madsen

(1996)). Using similar words would cause great confusion in a discussion of probabilistic,
35

performance-based structural engineering. Schrader et al. (1993) take the view that uncer-

tainty is related to the design, or selection, of values that satisfy a certain set of criteria.

This is not the same as uncertainty regarding measurement, or uncertainty about a future

event, because in these types of uncertainty there is no design. It characterizes the gap

between limited knowledge and lack of detail. The concepts could be thought of a similar,

if one can imagine "measuring" a phantom object that does not yet exist spatially and ma-

terially. This is quite like what is done when describing a structure in a narrative. However,

the concepts are dissimilar enough to warrant different vocabulary for each category. New

terms are needed to distinguish these variables properly.

For similar reasons, maturity is preferred over "stability" (Terwiesch et al., 2002) be-

cause "stability" has different accepted meanings in the domain of structural engineering.

An stable structure is one that is resistant collapse, and an unstable structure is prone to

collapse. Neither sense of the word "stability" communicates the refinement of design

information and the development with time.

Furthermore, maturity is used only to describe design information, and not for the ac-

tual as-built structure and its operational performance. A mature design variable, such as

the specified length of a steel column, may actually represent a probability distribution

based on AISC tolerance specifications. The design variable is fully mature, having all the

information required for construction and expressing design intent, but the as-built con-

dition may still be probabilistically uncertain to the designer. Aleatory uncertainty of the

structure in the sense of PBEE design methods considers the possible deviation between
36

the as-built structure and the mature design information used for construction.

3.2.4 Precision

Precision is defined by Terwiesch et al. (2002) as "the accuracy of the information

exchanged." As the design progresses the variables in the project will become more precise.

A finished project is completely precise because the real object contains all the spatial

information of the actual project. Numerical precision can be low with a range such as

10"-20" compared to a high precision range like 10.2"-10.4". Conceptual precision can

refer to generalities expressing purpose. For example the designer can use a lower precision

term like "lateral force resisting systems" and a higher precision description like "shear

wall". The combination of both numerical and conceptual precision is needed for project

completion.

Adequate control over the implied level of precision is important throughout the design.

If software tools are used to communicate design, there is often the mistake of allowing the

tool to imply a higher level of precision than intended. For example, it would make an

engineer nervous to see a rebar placer determining dimensions by using a scale and reading

off a number from a set of plans instead of using only the called out dimensions. This

is because the drawing itself is not intended to carry the degree of precision. It is the

dimensions which are supposed to express the precision. When a picture is drawn, it is

important to always carry along with it the text annotations that adequately describe the

intended level of precision.


37

3.2.5 Accuracy

Accuracy tells whether the variable's value is correct or not. It refers to how close a

variable's value is to its final state in the structure with reference to the implied precision.

In general, lower precision values are more likely to have high accuracy because there is a

wider range of possible values. Accuracy is important because it gives a measurement of

how likely values will change with time during the design process. This aspect of uncer-

tainty in the final value is not the same as epistemic or aleatory uncertainty in the sense of

probabilistic hazard and structural performance, because it relates to the process of matur-

ing design information and not the as-built condition of the real structure.

3.2.6 Grouping: agglomeration and aggregation

Agglomeration and aggregation are two different methods of grouping smaller physical

items into larger logical items. "Agglomeration" involves grouping dissimilar systems and

"aggregation" involves grouping similar systems (Yuan, 2008). For example, a bunch of

bananas is an aggregation of individual bananas, while a pizza is an agglomeration of crust,

cheese, and tomato sauce. Both the bunch and the pizza can be referred to as individual

items, but their underlying composition is different.

The rebar inside a reinforced concrete beam could be thought of as individual bars, or

as a single group of bars. This would be an aggregation of the individual bars into a group.

Similar objects are bundled together and considered as a single concept. The beam could

be thought of as a single object as well. The beam is an agglomeration of the concrete


38

portion and the rebar portion. Dissimilar objects are bundled together into a single object.

The issue of agglomeration and aggregation is distinct from that of precision. Precision

deals with how close the specification of the item is to its actual final outcome. Precision

operates at the appropriate level of grouping. For example, looking at a concrete beam

item to check clashes with mechanical equipment, specifying its overall dimensions to be

24" x 24" might be perfectly precise. But, if thinking about the individual rebar and con-

crete pieces as items, knowing the overall beam dimensions leaves a tremendous amount

of imprecision in the definition.

If the item being operated on is thought of at the group level, then it's constituent items

might be considered as a set of options because their individual specification is not given,

and might not yet be needed in the design process. For example, when performing structural

analysis on steel shear tab connection, the exact details of bolt spacing are not needed as

long as the agglomerate "connection" (consisting of nuts, bolts, and angle plates) can be

idealized as a non-moment-resisting pin. The exact bolts and plates to be used are left open

as a set of options until the group items need definition.

Another example is in the nailing configuration for a plywood shear panel. As long as

the nails can be treated as an aggregate group specified by a nail size and nominal spacing,

the exact position of each nail is not needed. Full spatial precision for design is obtained

for the "nail group" by giving the nominal spacing. But, full spatial precision is not given

for the individual nails, and is not required for construction. The exact position of each nail

can be left up to the installer. Good inspection evaluates individual nails for workmanship,
39

spacing, and size, at the same level of grouping as specified in the design.

3.3 Importance of content and quality

The importance of understanding the content and quality aspects of design information

are illustrated by looking at what is needed to define a mechanics model in order to inform

design decisions. Different qualities of information are needed to answer questions about

different parts of the structure. For example, bolt spacing might not be necessary to answer

initial questions about acceptable story heights, but will be necessary for the designer to

choose connection details, which may, in the end, impact the final story height through bolt

spacing and the number of bolts. The content and quality of information available during

the progression of a structure's design also influences the ways in which design information

can be communicated.

3.3.1 Mechanics models

Mechanics models are needed to determine whether a designed structure will perform

as expected before building it in reality. The mechanics models allows engineers to un-

derstand the behavior of something before it is built as a real object. The model allows

tinkering to determine the necessary design variables and relationships between them in or-

der to resolve level 2 ambiguity. The model also serves as a mechanical abstraction of the

designed structure prototype. This gives the model the power to test design concepts on pa-
40

per (or computer) before the costly expense (or impossibility) of making the real prototype.

The model can quantify decision variables such that an acceptance criteria for performance

objectives can be applied. Most structures are too large and too complex to build samples

of the real-thing because off the one-off nature of construction projects in general. Be-

fore mechanics models, complex structures could only be made by master builders who

intuitively understood mechanics based on experience in constructing previous prototypes.

The mechanics models employed require a certain amount of information in order to

function. The different information content types are important for the mechanics models

because they must all be present in order to make a practical model. They generally require

the material information enough to specify the stress-strain behavior of essential compo-

nents. The functional information must be known in order to determine what must be con-

sidered in the model and what may be ignored for the sake of simplicity. Enough spatial

information must be known in order to determine the location of the structure pieces, and

how the pieces interact with one another in compatible deformation. Process information

must be known in order to determine when different pieces are present in the structure dur-

ing construction, and whether the structure can maintain equilibrium even if certain pieces

are not yet built. The abstract variables are the ones that are needed to run the mechanics

model and are commonly the results of the model that then need to be applied to inform

engineering design judgment in order to proceed in the structural design.

The quality of information is also important for mechanics models. The right level of

detail is necessary because if there is too little detail, then it will not be possible to define
41

a meaningful mechanics model of the structure. On the other hand, too much detail might

be unnecessary because some detail might not be included in the mechanics model at all.

This occurs when the abstract variables to be determined with the mechanics model are

not dependent on information at a too higher level of detail. The ambiguity of the design

information influences the ability to construct a mechanics model in the first place. In a

situation with level 2 ambiguity, the engineer might be unable to construct a relevant me-

chanics model because the problem solving method is unknown and therefore would not be

able to determine what information must be gleaned from the model. For these cases, simil-

itude relationships may help by eliminating unnecessary variables, agglomerating groups

of similar variables (independent variables representing dimensions), and pointing to func-

tional relations among independent variables through Il-terms (Sonin, 2004; Harris and

Sabnis, 1999).

The level of maturity of design information might also be affected by the mechanics

model depending on its complexity and the amount of information needed to compute and

construct it. Mechanics models usually operate at a specific level of precision, and when

that changes then the outputs will also be affected and will most likely influence changes in

the downstream decisions that depend on the model results. Immature designs are also less

likely to permit more detailed kinds of mechanics models. Immature designs might be well-

suited for very simplified types of models that capture the general global behavior of the

structure, but might miss out on some of the local effects that require more mature designs

and higher precision. Typically capturing local effects requires more complex models and
42

therefore require more of a time investment into creating and computing them. So, it might

be undesirable to commit to working on a mechanics model unless the design is more

mature and ready for detailed mechanics-related analysis.

The issues of grouping, substructuring, and similitude are relevant to mechanics models

because they determine what model elements are necessary. For example, in analyzing the

global behavior of a concrete moment frame building, it might be convenient to agglomer-

ate the components that make up the columns (concrete, longitudinal rebar, and transverse

rebar) into an element that represents the column as a single prismatic frame element with

a stiffness EI and plastic moment capacity My. For a woodframe shear wall, it might be

good to aggregate the individual nails in the plywood panel so that it behaves as if the

panel were just rigidly connected. In connection design, it might be useful to agglomerate

all the different connection pieces and model them as a single semi-rigid connection object

idealized using springs.

3.3.2 Communicating quality of information

Clearly articulating the level of detail and accuracy necessary to define a design option

set at a given point in time during design requires open communication and understanding

of the values each party can bring and constraints that affect them. Lack of clarity on these

is an obstacle to set-based design. The right questions need to be asked at the right stage of

project design. Each project participant must understand not only what is asked, but also

the level of detail and accuracy that is required. Too much too early forces unrealistic and
43

undesirable commitment, while too little encourages rework.

The issue of level of detail appears in a variety of different contexts. In sensors and

health monitoring applications the information obtained from sensors can be found on the

local, element, and global levels. Estimating a global level parameter using only local level

data causes problems with data applicability. Estimating a local level parameter using only

global level information introduces too much uncertainty in processing the data. The right

information needs to be present in at the right level of accuracy in order to produce the

desired result. Different types of information at different scales have their own costs and

uncertainties.

In the design process, the issue manifests differently throughout the different stages.

In a sense, the stages of design themselves can be differentiated by the amount of detail

available for precise expression. At conceptual design, very few high accuracy details can

be expressed. By the construction phase, precise details are already clearly expressed in

the design.

There is a tension between freeform and parametric descriptions of structural systems,

elements, and details. A parametric description enables quick changes and easy use of

computer modeling tools. But, it also limits the ability to design something that is outside

the range of choices. For example, choosing to use an innovative structural system doesn't

easily fit in any conventional parametric model. A freeform description is necessary for

something new. Designing an innovation requires both creation of new objects and details,

and creation of new ontological categories to describe its function.


44

3.4 Summary

What I have done in this chapter:

1. Defined different aspects of the content and quality of structural design information.

2. Explained how different types of content and quality influence mechanics models.

3. Described challenges to communicating information quality.


45

Chapter 4

Design communication patterns

The clarity of communicating design intent is a performance goal for the concept and

design phases. The goal of communication is to transmit the intended level of information

without implying additional information and without neglecting any information. Good

communication instructs the receiver in the fullest detail with the greatest possible clarity

to achieve the purpose. When too little information is given, then recipient is left con-

fused. When information is too ambiguous then erroneous assumptions are often made.

When communication is noisy with distractions then the information becomes obscured.

Confusing, cluttered, and noisy information presentation blocks successful interpretation.

This chapter describes the different patterns of technical communication for represent-

ing and transmitting design content. Using the most appropriate pattern of communica-

tion with at the correct level of information with the proper presentation design simplifies

understanding the details behind and intent of an engineering design. Each pattern has
46

different advantages in being able to articulate aspects of design variables themselves and

their interconnectedness to other pieces of information. The chapter begins with an intro-

duction to the concept of genre as a communication pattern, and an overview of classical

hermeneutics. An understanding of hermeneutics allows analysis of the different patterns

of communication and the ultimate intent of conveying the author's meaning. After that,

the chapter discusses a list of various patterns describing their individual characteristics.

An example of communicating preliminary design information is given for the collec-

tion of estimates on bridge repair quantity, cost, and time (Mackie et al., 2008a). Specific

patterns of communication were necessary in order to communicate preliminary design in-

formation with bridge engineers at the California Department of Transportation (Caltrans).

Providing a mechanics model of the bridge was not helpful and neither were the spread-

sheets containing the repair quantities. Rather, a different combination of communication

methods were best for conversation and building common understanding. The informa-

tion needed was instead conveyed in damage scenarios containing (1) repair information

with spatial information, (2) dimensions and quantities, and (3) descriptions and names of

individual pieces.

4.1 Communication fundamentals and hermeneutics

Proper communication methods are needed to convey the information inside any build-

ing model. Different types of information need to be presented in different ways. The 3D
47

view of a building is not as good as a table for showing a list of materials to purchase, and

sometimes a 2D section cut can illustrate details better than a photo-real rendering with too

much unnecessary information. Each communication genre is characterized by a particular

form and purpose. Through the proper combination of appropriate methods, the engineer

will be able to communicate and document the design in the appropriate manner.

These different genres of communication are still necessary regardless of the manner

in which they are generated. Before the use of computers, the generation of drawings

involved laborious manual drafting. With the use of building information modeling (BIM),

the communication might be generated by extracting information from the building model.

Before investigating technical communication specifically, it is helpful to review exist-

ing, long-standing methods of communication in general. The various genres of commu-

nication have different strengths and are well-suited to different levels of abstraction. For

example, Fee and Stuart (2003, p. 22) list many different "genres", or patterns, of verbal

communication: narrative history, genealogies, chronicles, laws of all kinds, poetry of all

kinds, proverbs, prophetic oracles, riddles, drama, biographical sketches, parables, letters,

sermons, and apocalypses. Although all the genres of communication share common ele-

ments, the process of understanding each type can be drastically different because they use

a different form. Each genre also has different purposes. For example, a proverb or "rule

of thumb" has the purpose of communicating a general statement which may not be true

in all circumstances. A narrative history aims to explain the sequence of events and causes

leading up to a present condition. This genre might be most useful for communicating the
48

design process for a new hospital. The form and purpose of poetry might not be well-suited

for expressing construction details, but might be a suitable way of describing a building's

overall aesthetic impact on the community.

In a similar sense, there are different patterns of communicating design intent that are

appropriate for different phases in a project. The proper choice of a communication method

will assist the communication of the design and the design intent. Understanding these dis-

tinctions allows greater control over communication and allows the engineer to be more

clear, thus fulfilling the goal of helping the reader comprehend what originates in the mind

of the author. Different communication patterns are needed to adequately explain the con-

cepts, spatial relationships, and information needed at different stages of design where

different decisions have and have not been made yet. Each genre is needed to express the

appropriate level of precision at the right stage in design. Several patterns of design com-

munication are listed in Table 4.1 along with the information emphasis they are best suited

for.

A proper understanding of hermeneutics is important for communication, because of

the different concepts in design that must be clearly communicated. Hermeneutics is de-

fined by Terry (1974, p. 17), who explains that:

Hermeneutics is the science of interpretation. The word is usually applied


to the explanation of written documents, and may therefore be more specifi-
cally defined as the science of interpreting an author's language. This science
assumes that there are divers modes of thought and ambiguities of expression
among men, and, accordingly, it aims to remove the supposable differences
between a writer and his readers, so that the meaning of the one may be truly
and accurately apprehended by the others.
49

Table 4.1: Communication patterns have different spatial, concrete, and abstract qualities.

Pattern Information emphasis

Narrative abstract, functional, material, process

2D Sketch spatial

Annotation associating of spatial and other types

Structural Details spatial, material, functional

Spreadsheet non-spatial

Mechanics Model abstract

Rebar Placing spatial, process, material for rebar

Schedule process, with emphasis on time

A slightly broader definition is offered by Ramberg and Gjesdal (2005) who comment that

"hermeneutics covers both the first order art and the second order theory of understanding

and interpretation of linguistic and non-linguistic expressions." Non-linguistic expressions

such as drawings and figures are clearly also within the domain of hermeneutical applica-

tion.

Hermeneutics becomes critical to the communication of abstractions in structural de-

sign because misinterpretation of design intent causes serious problems. It is in everyone's

best interest when all project participants aim to understand the original design intent. Her-

meneutical principles exist in order to overcome the barriers to communication that exist

in the transfer of ideas. Understanding the principles identifies the areas where ambiguities
50

are common and highlights the issues that need to be overcome in the design process. An

understanding of the basic nature of communication helps the analysis of abstract commu-

nication in structural design. Thomas (2002, p. 27) lists the "time-honored definitions" for

hermeneutics, exegesis, meaning, and interpretation:

1. Hermeneutics is a set of principles.

2. Exegesis is an implementation of valid interpretive principles.

3. Meaning is the truth intention of the author.

4. Interpretation is an understanding of the truth intention of the author.

Challenges to the success of each process provide roadblocks to communication. For exam-

ple, reading a drawing requires proper exegesis. From this perspective, one can understand

the general fear of novel computer tools providing an unintended implementation of ex-

egetical principles, thus giving a wrong interpretation of the meaning. For example, many

engineers are somewhat wary of using electronic models as contract documents because of

the uncertainty in the ways information is expressed.

The goal in design communication is to accurately communicate the design intent all

the way through the project in a consistent manner. The handoff of information is where

comprehension of design intent can break down. Design intent is not subject to the reader's

interpretation. It is the writer's responsibility to communicate clearly, and it is the reader's

responsibility to apply correct principles in order to grasp the design intent properly. The

meaning of the design documents has its origin in the mind of the designer, and the goal

of handing off is to transmit that successfully to the next participant. Noisy communica-
51

tion can block that transmission. Doumont (2002a) notes three sources of transmission

"noise" in written communication: text (inconsistent paragraphs, complicated sentences,

faulty spelling), page layout (unclear structure, inconsistent typography, unusual fonts),

and illustrations (too many details on schematics, too many rules in a table, too many tick

marks on a graph). A good communicator would reduce the noise in all these areas.

A breakdown in communication results in problems. For example, interpretation might

be strained because of unfamiliar technical terms. Structural engineers are generally com-

fortable with discussing abstract quantities such as the reinforcement ratio p, which is com-

monly used throughout the design process. It is an abstract value based on a typical section

cut through a reinforced concrete element. Its value is not something that is calculated or

measured in reality. The code contains minimum and maximum values for p. Many design

equations use p and that number is used to help understand what is important to specify in

the design process. However, even though this quantity is important and well-understood

by designers, the quantity is largely unknown to rebar fabricators. This fact might give

the engineer pause to consider how to convey his meaning in a way that is more easily

interpreted.

Furthermore, Turk (2001) discusses the difficulties of conceptual product modeling

from a phenomenological perspective and his understanding of the process of the herme-

neutical task. Product models are not themselves part of objective reality, but are created

from how the modeler chooses to describe objective reality. Therefore, Turk suggests that

there can never be a single unified product model that can capture all design information,
52

but rather that "several correct or different models may and should exist" and that new soft-

ware "should not be built on a unified, centralized model, but on a combination of models".

Winograd and Flores (1986, p. 163) also note that "The most important design is ontolog-

icaF, which is concerned with creating new types of things, not just creating instances of

similar objects. All the more, this points to the importance of carefully considering various

forms of communication to convey design intent. This indicates that any information tools

and BIM software for design must provide designers with the ability to control communi-

cation at the appropriate level of conceptual abstraction, and that no automatic system will

exist that take the place of human communication.

This issue is also seen in other domains, such as cartography. A well-made map will

provide information beyond mere spatial relationship and a mere photograph. In the same

way, the communication of design intent must be designed in a manner that allows for the

right manner of thinking, both symbolically and physically, by properly understanding the

different types and qualities of information. Imhof and Steward (1982, p. 79-80) comment

that:
The map is not only a picture; it is primarily a means of providing infor-
mation. It has other responsibilities than has a painting. It must show not only
those features which appear in good light, but must also allow all similar things
to appear with the same emphasis wherever they are present. Above all, it must
include the depth of the valleys and the heights of the mountains, not only in
the visual sense but in the geometrical sense also. To do this one must trans-
form some of these direct image effects and extend them through symbolically
indirect elements.

To this end, the following patterns of communication explain which ones are best suited

for different types of information.


53

Four general-purpose patterns are described first, then four more patterns that are more

application-specific. The patterns for Narrative, Sketch, Annotation, and Spreadsheet are

best for communicating freeform information. The Mechanics Model pattern is suited for

specific information related to structural behavior. The Rebar Placing Drawings pattern are

intended for the information needed by rebar placers and fabricators during construction.

The Database pattern is useful for standardized information. The Schedule pattern deals

specifically with time.

4.2 Patterns for communication

4.2.1 Narrative pattern

Intent

Describe preliminary conceptual information using word descriptions instead of using

sketches and drawings. Using words alone focuses on communicating functional, material,

and abstract information without prematurely communicating spatial information.

Also known as

Preliminary description, conceptual description.

Assumed background

Foundational + Design + Problem recognition and solving + Communication.


54

Motivation

During conceptual design, different options need to be explored and described without

too much detail. In the beginning of a project, the general building topology, material, con-

struction sequence, and structural system need to be defined in a way that allows checking

for general feasibility without making premature commitments to details. This stage of im-

mature design requires the ability to communicate functional information along with low

precision spatial and material information, and some abstract information that can be used

to perform basic structural analysis. Here, narratives are useful because they emphasize

exactly the types of information present in the conceptual design phase.

Applicability

Narratives are used in situations where some information needs to be presented in a

more conceptual manner, often at a lower precision than necessary for construction. A

word description is good at communicating design function without implying any prema-

ture spatial information. Material information can be communicated without making com-

mitments to spatial decisions like shape and size. In the earlier stages of design, narratives

are preferred over detailed models with much 3D spatial information because of the ability

to describe preliminary designs at the appropriate level of detail.

A narrative should not be used when lots of spatial information needs to be commu-

nicated. Spatial data is difficult to extract from a word description, while a picture allows

the location and size of images to convey spatial relationships in a rapid fashion. For this
55

Line 2 and 3 at the Main Level (Plan 2/S3 on the structural drawings, the west and east walls of the stairwell
and hallway #1): The current scope of the remodel includes replacing the lateral load-resisting capacities of the
existing walls by adding plywood to remaining walls to replace plaster lost at the new openings. To upgrade the
remaining walls to the current code levels would require adding plywood at the alcove at the stairwell (by
eliminating the alcove, or by adding plywood from the living room side), and increasing the number of nails and
adding floor ties in the remaining walls.

Line 2 and!? at the Lower Level (Plan 1/S3 on the structural drawings, the west and east walls of the stairwell):
Upgrading this line would involve increasing the number of nails and adding other hardware at those walls
already set to receive plywood.

Figure 4.1: Narratives describe structural design at a functional level.

reason, narratives are often accompanied by a Sketch (4.2.2) because the sketch allows the

conveyance of spatial data without the precision needed for a model with 3D geometry.

Structure

Narratives often take the form of a short written report, or a letter. Narratives are often

accompanied by sketches that illustrate the spatial information for the items described in

the narrative. Figure 4.1 provides an example of a narrative describing the general scope of

work for a residential seismic strengthening project. The narrative explains basic functional

and material information without implying any more precision than is necessary at this

stage in the project.

Consequences and implementation

The narrative is able to fulfill its purpose of communicating information at a very spe-

cific level of precision because of the specificity and generality of words. Words can be

used to express a very specific detail, or can be used to express abstract categories and
56

types. An author can choose to use words to express the intent exactly. The communica-

tion can be done at the exact level of intended detail. The ability of verbal communication

to accomplish this strict control over level of implied precision is unparalleled in com-

parison with graphical communication alone. Doumont (2002b) comments that although

graphics are "indeed superior for conveying intuitive or global information. By contrast,

they do a poor job of expressing abstract concepts and lack the accuracy that words are

endowed with. Words, in a sense, are worth a thousand pictures."

Narratives are not suitable for communicating large amounts of complex spatial infor-

mation alone. It takes many more words to express the same spatial information that can

be communicated with a graphic.

Narratives are able to explain the scope of a project with clarity that is beyond the capa-

bility of a sketches alone. Narratives typically may include a background or that discusses

project history, and the main purposes of a building project. Understanding the context

of the building helps to communicate the underlying intent of the design decisions and

options.

Because a narrative is generally brief, then the fact that something gets mentioned im-

plies a significant weight. Each word in every sentence is intentional and chosen specifi-

cally. This is not the same as a 3D model inside a computer system where everything shown

on the screen is not necessarily intentional. For example, when a designer adds a steel beam

object to the model, there is usually no conscious decision to govern its appearance in every

conceivable view.
57

Known uses

Narratives are used when presenting a client with options for scope of work and general

design approaches. This allows the structural engineer to describe the nature and extent of

proposed work without committing time and effort to developing a full solution. The use

of a narrative helps build vocabulary between the engineer and owner, and can facilitate

communication at the functional level without needing to delve into spatial and material

design information. The narrative can also be useful in describing situations where the

construction process information controls the scope of work, such as in an existing building

remodeling project.

Narratives help describe general parameters constraining structural design. One case

study example from a medical office building contained the following sections:

1. Project description

2. Applicable standards and references

3. Material specifications

4. Geotechnical recommendations

5. Medical office building structural system description

6. Parking structure structural system description

7. Future bridge

8. Preliminary design sketches

The project description section includes the project location described by giving the
58

city, surrounding streets, and adjacent buildings. The building use is described as a medical

office building (MOB) and parking garage. The number of above grade and below grade

stories is given for the MOB and parking garage. Their orientation with respect to the side

of the project site is given along with the streets that the parking garage faces. Unusual

site conditions are described; there is a a natural creek that runs through the middle of the

site which requires special consideration. A statement of what is not included in the design

clarifies the scope of the structural engineer's role on the project.

The applicable standards section lists the standards used and the parameters assumed

for the schematic design. California Building Code (CBC) 2001 is used with parameters

for seismic design (zone, importance factor, source type, soil profile, near source effect)

and wind design (exposure class, basic wind speed, combined exposure factor for gust and

height Ce, importance factor).

The material specifications reference the applicable documents and specify certain pa-

rameters for use in the schematic design. The geotechnical recommendations provided a

summary of only the essential information required for completing structural design. The

structural system descriptions contain approximate dimensions for floor plan, roof pent-

house, and basement wall thicknesses; general description of the construction work re-

quired to build the parking structure; approximate sizes of seismic joints between the main

building and parking structure; depths and approximate spacing of steel members for floor

system; and two different options for seismic force resisting systems. The future bridge

describes plans for a possible future addition. The design sketches display the spatial infor-
59

mation related to the narrative description. Figure 4.3 shows a floor system from this case

study.

Narratives can also describe acceptable hazard, risk, and performance objectives in a

manner that describes outcomes at the right level of precision. For example, the regulations

for nuclear reactor sites (10 C.F.R. § 100) explain hazard and risk criteria for reactor site

selection, and also the desired performance objectives related to radiation exposure and

population risk. The regulations are written for general application to a wide range of

possible nuclear reactor types.

4.2.2 Sketch pattern

Intent

Describe preliminary conceptual information using sketches in order to show spatial in-

formation and relationships between objects without implying an artificially higher degree

of precision.

Assumed background

Foundational + Design + Problem recognition and solving + Communication.

Motivation

Sketches are useful for communicating spatial and non-spatial information together,

and are especially useful during stages of design when only preliminary spatial informa-
60

tion is available. The style of a sketch and the presence (or lack) of Annotations (4.2.3)

controls the amount of precision that is implied in the sketch. Sketches do not contain the

level of detail necessary to purchase and construct an item, but do provide enough con-

ceptual information in order to understand function, purpose, and some degree of process

information.

Applicability

Sketches address the need to communicate preliminary design information in a way

that expresses essential spatial information, yet without implying more precision than is

actually needed during the current stage of design. Sketches are applicable even before

sufficient design information is available to use modeling software. The use of computer

models too prematurely during design might cause "fixation" (Jansson and Smith, 1991)

due to the structured requirements and overly precise geometry needed to use modeling

tools.

Structure

Sketches are typically used with Annotation (4.2.3), and often linked with a Narra-

tive (4.2.1). Sketches are freeform graphics, and as such can be used to communicate

almost anything in a way that is best suited for the designer. Sketches can be custom tai-

lored to suit any situation where conceptual information needs to be communicated, unlike

views generated from databases in BIM software which tend to be more rigid and difficult
61

to customize within the software itself. Figure 4.2 shows an example of a sketch pointing

out the location of a lateral force resisting system, without providing too much premature

detail. Figure 4.3 shows a sketch of a floor system inside a medical office building. The

sketch gives the spatial arrangement and depth of steel floor framing members and the deck,

but without implying a commitment to specific steel shapes.

Consequences and implementation

Lack of control over the level of detail poses a serious problem to the implementation of

BIM into the entire engineering delivery process. The views of a model are generated au-

tomatically by the computer software. Because of this, the engineer has less direct control

over the spatial relationships that are implied by the rendered views. Unlike a static CAD

drawing sheet, the dynamic view in a BIM system often is unable to contain consistent

annotation. In a static drawing sheet, the engineer can refine the graphic to express only

the significant spatial relationships in order to amplify their intended meaning. Further-

more, the static drawing can be annotated in the most appropriate way to express design

intent with accuracy at the appropriate level of detail. The engineer need to be concerned

about the drawing's view being changed thus corrupting the combination of text and precise

control over the exact spatial relationships shown.

In sketches, every stroke and annotation is intentional to the extent that they express a

certain thing in the designer's mind. Because of this, the conceptual specificity of sketches

can be more than a computer model view. A computer model view is a rendering of precise
62

© 0 © 0 © © ©
* -t
©
i 1 i

- ®

—®

S>

Figure 4.2: Sketch shows general spatial information along with functional information for
preliminary structural system at an appropriate level of precision.
63

2£-4"

l—H.

3^" ^M/-C.
N
3" /ffz. p/^*-

Figure 4.3: Sketch shows the general depth and dimensions of floor system elements, but
without implying a commitment to specific steel shapes.
64

information already entered into some database system, while the conceptual sketch often

conveys information with a degree of imprecision in order to avoid premature specificity.

Known uses

Sketches are often used in communicating preliminary design information and can be

readily used even when the project is immature and highly precise design information is

not yet available.

4.2.3 Annotation pattern

Intent

Display information in a visual format consisting of a main graphic element with text

that explains the important components of the graphic.

Also known as

Call outs, notes, markup, labels, tags.

Assumed background

Foundational + Design + Problem recognition and solving + Communication.


65

Motivation

Graphic elements, on their own, lack the specificity needed to communicate what the

main point of the graphic is. Annotations link spatial and non-spatial design information,

and show which portions of the graphic are meant to convey information and which ones

are not. Annotations show material design information when items are called out, and can

explain functional information that would otherwise not be presentable in a graphic form.

Applicability

Annotations can be added to any graphic element in order to highlight and explain its

significant pieces of information. This addresses the increasing tendency to allow software

programs to generate drawings and models without seriously thinking about how to teach

the reader what information is significant. Annotations provide information at the point

of need and make communication more clear. When more information is provided, the

graphical communication becomes more clear and easily understood and used.

Annotations are useful when grouping needs to be communicated, or functional infor-

mation that describes the intent of an item that goes beyond its spatial arrangement. The

text annotations can describe categories and define terms, and express other information

that is not easily represented in graphical form.


66

Figure 4.4: Annotations enable proper interpretation.

Structure

An example of the importance of annotation can be seen in three drawings in Figure 4.4.

Each drawing has the exact same graphical, but with different verbal annotations. The first

drawing has no annotation, the second shows geometry plus a description of the bar size

and bar count, the third gives the individual position of the bars. No single drawing is

"correct", but each conveys a very different information content. The center drawing is

perhaps the most clear in communicating the original design intent for the rebar placement

and selection. The additional information given in the right drawing is not necessary to be

specified, and might even be incorrect because it implies a level of precision higher than

necessary.

This example illustrates that good communication is sometimes necessarily redundant

(Doumont, 2002a), and that there are multiple different ways to model the information
67

content of a single design expression. The position of the rebar could be modeled by

giving the individual position of each bar relative to the concrete member, as depicted by

the third drawing. Or, the position of the rebar could be modeled at a greater level of

aggregation as in the second drawing which represents more of a conceptual layout instead

of a physical layout. For visualization, it may be necessary to compute an exact position

of each individual bar, but in reality the specification of the rebar might only be described

on the level of detail of the annotation in the second drawing. This leads to problems of

over-precision that occur when too much information is inferred that was not intended.

Consider how an engineer would feel if he saw a subcontractor measuring a drawing

with a scale to determine the spacing between essential structural components? It would

be slightly disturbing to say the least, because the drawings were never intended to be mea-

sured. Rather, the engineer used the drawings to convey spatial relationships and intended

for the specifically-expressed dimensions to control the important, defining characteristics.

If the exact position of each bar were really necessary and critical to the design, then draw-

ing (c) would be used. If not, then drawing (b) would have been used.

This notion of level of detail and level of abstraction and aggregation becomes more

important in preliminary design. In the earlier stages, it is more important to remain open

and slightly "fuzzy" on the details. In the later stages of design, exact locations and de-

tail become more important because different kinds of performance questions are being

considered.
68

Consequences and implementation

Annotations accomplish the communication of non-spatial information that is linked

to spatial information presented in a graphic. Annotations need to be carefully thought

out so that the meaning can be communicated clearly. Text can be added to graphics in a

number of different ways, by using an arrow or some other pointer, or simply by positioning

text in the appropriate location. As free form tools, annotations are inherently able to

communicate the appropriate level of abstraction and concreteness.

Implementing good annotations may depend on the drawing tools that are used by the

designer. Adding annotations to at hand sketch is obviously simple, but adding annotations

to a view from BIM software might prove complicated.

Known uses

Annotations are used everywhere in order to add information on top of spatial infor-

mation in a graphic. It is used to describe non-spatial information that must be linked to a

spatial location on a graphic. Poor drawings that cause frustration and miscommunication

often have inadequate annotation.

4.2.4 Spreadsheet pattern

Intent

Present information in a tabular format in order to provide easy access to data that is

linked or sorted in a particular way using text and numbers.


69

Also known as

Tables, Excel sheets.

Assumed background

Foundational + Design + Problem recognition and solving + Communication, and gen-

eral knowledge of spreadsheet concepts and tools, such as Microsoft Excel.

Motivation

Spreadsheets are useful because they are general purpose tools for communicating all

different kinds of information and can be accessed in a computer program. The spreadsheet

is able to perform logic operations and basic computation.

Applicability

Use spreadsheets when

1. One needs a general purpose tool to convey text or numeric information.

2. Data entered by the designer needs to be transformed through computation. The

spreadsheet can automatically perform some computation through cell references

and formulas.

3. Reference data needs to be looked up by a person and by a computer program.

4. Linking the information to spatial location is not significant.


70

5. Different types of information need to be linked and presented together, as in a table

showing two variables across the rows and columns.

6. The information can be located by using multiple coordinates. For example, the row

and column of the data in a table identifies a piece of information.

7. Multiple tables need to be displayed together or in series.

Structure

A spreadsheet displays data in a tabular format. This permits a free-form logical ar-

rangement that can be specifically suited for a work task. The advantage of being able to

show only specific pieces of information helps make communication clear. However, this is

also somewhat of a disadvantage since spreadsheet tables often lack the spatial linking that

is often necessary to understand the data in the context of the whole structure. Figure 4.5

shows an example of a spreadsheet containing the unit repair costs for bridge repair items.

Consequences and implementation

An advantage of using spreadsheets is the freeform ability to present and store infor-

mation using a view that is the most needful for the specific application. The ability to

implement some logic within the spreadsheet is also a particular strength.

The barriers to implementation are extremely low because almost everyone knows how

to use and edit a spreadsheet and has access to spreadsheet software, such as Microsoft

Excel. The ability for most people to edit the information and customize it without having
71

A B I C I D I E
1 Unit costs associated with each repair quantity
2

3 Item Name Unit tiean std dev Notes


4 Structure excavation CY 165 $ 33 $250for<50CY
CY ""f
5 Structure backfill
SF
$ 220
38
$
$
44 $335 for < 14 CY, $ 1 1 5 > 2 5 0 C Y
6
7
Temporary support (superstructure)
Temporary support (abutment) SF
$ 38 $
8
8 placeholder
8 Structural concrete (bridge) CY
$ 2,225 $ 445 $2000 for > 10 CY
9 Structural concrete (footing) CY
$ 520 $ 104
10 Structural concrete (approach slab) CY
$ 1,625 $ 325
CY
$
11 Aggregate base (approach slab) $ 325 $ 65
LB
12 Bar reinforcing steel (bridge) $ 1.35 $ 0.27 $6.50 < 400 LB
LB
13 Bar reinforcing steel (footing, retaining wall)
LF
$ 1.20
215
$
$
0.24 placeholder
14 Epoxy inject cracks
SF
$ 43
15 Repair minor spalls $ 300 $ 60 s1 has100/sf
LB
16 Column steel casing
17 Joint seal assembly LF
$ 10.00
275
$
$
2
55 MR101-160mm
18 Elastomeric bearings EA
$ 1,500 $ 300
19 Drill and bond dowel LF
$ 55 $ 11
20 Furnish steel pipe pile LF
$ 55 $ 11 Class 625 (140)
21 Drive steel pipe pile EA
$ 2,050 $ 410 610mm diameter. Class 625.
22 Drive abutment pipe pile EA
$ 9,000 $ 1,800 610mm diameter. Class 70.
23 Asphalt concrete TON
$ 265 $ 53
24 Mud jacking CY
$ 380 $ 76
25 Bridge removal (column) CY
$ 3,405 $ 681
26 Bridge removal (portion) CY
$ 2,355 $ 471
27 Approach slab removal CY
$ 1,000 $ 200
28 Clean deck for methacrylate SF
$ 0.40 $ 0.08
29 Furnish methacrylate GAL
$ 85 $ 17
30 Treat bridge deck SF
$ 0.55 $ 0.11
31 Barrier rail LF
$ 2 $ 0.40 placeholder
32 Re-center column EA
$ 20 placeholder
$ 100 $

Figure 4.5: Unit repair cost spreadsheet (Mackie et al., 2008a).

specialized skill is a tremendous benefit for interaction.

The liability of using spreadsheets is the lack of spatial information relating the in-

formation to a position within the structure being designed. This poses a challenge to

understanding data in the proper context. For example, rebar bar list might be useful as a

spreadsheet for estimating purposes, but would be a poor way to show design information

for installation because of the lack of spatial connectivity. Using spreadsheets for recording

the repair quantities in bridge earthquake damage scenarios was not as good for communi-

cating with a cost estimator as was placing the same information on drawings Mackie et al.

(2008a).
72

Known uses

The use of spreadsheets is ubiquitous in engineering calculations. They are useful in all

circumstances because of their free-format ability to communicate the right type of design

information with the proper accuracy.

Spreadsheets are used commonly used for documenting design procedures involving

many computations. Christy (2006) provides examples of how to use spreadsheets for per-

forming and document structural engineering calculations. The examples cover basic math-

ematical operations such as integration and matrix math, and detailed design examples for

loading, connections, foundations, storage tanks, slabs, beams, and columns. Spreadsheets

can provide a "clear audit trail and uncomplicated code" and are useful in documenting cal-

culations inside a single document. The ubiquity of spreadsheets allows them to function

as a universal platform for communicating information.

Spreadsheets are used to implement the database described in Mackie et al. (2008a)

for reinforced concrete bridge repair cost and time. A series of spreadsheets contain basic

information about the material and geometry of the bridge, its analyzed performance sub-

jected to multiple earthquake ground motions, and data on repair cost and time estimates.

The spreadsheets are used to implement a database structure needed to store and access all

the information needed to support the methodology. In this example, the spreadsheets are

linked to each other using cell references so that changes in one spreadsheet update in the

other spreadsheets. It serves as both a data storage mechanism, and an easily programmable

data entry method.


73

4.2.5 Mechanics Model pattern

Intent

Use an abstract model of a whole structure, or structure components, based on mechan-

ics theory to show the structure's response to loads and deformations.

Also known as

Analysis model.

Assumed background

Foundational + Design + Mechanics + Materials science + Problem recognition and

solving + Communication.

Motivation

During preliminary design, it is often necessary to explore the feasibility and commu-

nicate the essential nature of a structural system. Describing how a structure works using a

mechanics model communicates the internal characteristics of a building and reveals how

the structure will hold itself up and resists its loads. The mechanics model is able to show

a functional view of how system components interact, because the mechanics model is ab-

stracted. This makes a mechanics model suitable for communicating preliminary functional

information along with basic spatial and material information at the same time.
74

Applicability

A mechanics model is useful for showing the essential features of a design that charac-

terize its structural behavior and demonstrate how it will perform its function. This does

not necessarily refer to computational mechanics model which is computed using software,

or a method suitable for hand calculation. Rather, the idea in this pattern refers to the way

that this information is communicated in the design process. For example, even a simple

force flow diagram, or deformation mechanism could be an appropriate mechanics model

for the concept phase of design.

Using a mechanics model to communicate design intent avoids the unintended effect of

premature precision when the design is still immature. It avoids the problem of implying

additional details that have not yet been designed. The mechanics model also makes clear

which parts are relevant to structural performance, and which parts are not. This distinc-

tion can reveal areas where other design changes can be made without affecting structural

performance.

Structure

Mechanics models are generally shown using a schematic diagram combined with nu-

merical data supporting structural performance and spatial information. Mechanics models

tend to show less detail than other forms of communication because they are highly abstract

in nature. An example of a simple 2D mechanics model with a nonlinear pushover curve,

moment demands, and displaced shape is shown in Figure 4.6.


75

Nonlinear Static Pushover for Frame 1

1200 h

4 5 6
Horizontal Roof Displacement

Figure 4.6: Mechanics models show essential design features and structural performance.
76

Consequences and implementation

The implementation of a mechanics model requires a simplification of the actual struc-

ture. Only the components that influence the outcome of the model are considered, and

other information that is not relevant to the abstract variables that come from the model are

generally ignored or not used. Much wisdom and practical experience is needed to judge

what aspects to include or exclude, and what level of accuracy and precision is sufficient.

Known uses

Analysis models are appropriate for showing abstract variables. The analysis model is

created for the purpose of manipulating objects related to structural engineering, mechan-

ics, and materials theories. They are well-suited for communicating the non-real variables

that are needed to make engineering decisions that are based on mechanics and calculation.

Of course, not all decisions are based on analysis. In fact, the limitations of the analysis

model for decision making highlights places where mechanics theories are insufficient for

completely describing the factors necessary for making the design decision.

4.2.6 Rebar Placing Drawings pattern

Intent

Provide drawings with information necessary for a rebar placer to tie and place rein-

forcing bars within concrete formwork.


77

Also known as

Sometimes inappropriately known as "shop" drawings.

Assumed background

Foundational + Design + Problem recognition and solving + Communication + Team-

work, and either familiarity with the CRSI design and detailing manuals (Concrete Rein-

forcing Steel Institute, 2000,2001, 2002, 2008) or experience comparable to an apprentice-

level ironworker.

Motivation

When an ironworker is out in the field, he has little time to analyze a set plans or compile

information that is found in difference places. All of the relevant information needs to be

readily available down to the level of individual bar locations. The information must also

include all the process information and enough spatial information to identify and place

each bar. This includes details such as bar chair location and intermediate supports for

stability.

Good placing information will locate all the relevant information the point of need. This

involves using graphics and appropriate annotation to explain what each graphic element

indicates.
78

Applicability

Placing drawings are used to communicate a fully mature design that has sufficient

precision for construction. These drawings emphasize spatial, material, and process infor-

mation for rebar, and abstract information is largely absent. These drawings show a tight

linkage between the spatial and material information by showing the location of each in-

dividual bar and indicating the bar size, type, and configuration of each one. This close

coupling of information makes the placing drawings particularly useful to field ironwork-

ers.

Placing drawings are an example of a communication pattern that is inappropriate for

preliminary design. The required amount of precision is too high for an immature design.

Structure

Placing drawings are typically created using general purpose CAD software augmented

with special tools for drafting rebar. They typically make heavy use of cross section and

elevation views and use annotations to describe the individual bars. Intelligent grouping

of the bars makes the drawings easier to read. The level of detail is much higher than

that of other drawings because placing drawings need enough precision for construction

(Figure 4.7). They must enable measurement and provide fixed reference points that can

be identified in the field, which allow placers to determine the location of individual rebar

pieces.
79

- i - as i r-i" TOP
4- MS X S'-O* LEFT SHE
2 - 16118 giCKT S!K
Mowspaw

^ u JJfi-
?- 2SIB 1

"WEFT—

StSECTM

.4-

SSMf

(b)
Figure 4.7: Placing drawings (a) show each individual bar, but other drawings (b) of the
same object might only show aggregations of bars (Concrete Reinforcing Steel Institute,
2000, p. 17-14,17-15).
80

Consequences and implementation

Placing information is typically communicated using a set of 2D drawings printed on

paper. The use of paper is good because of portability and the ability to see a large amount

of spatial information at once, providing context for the information presented.

General purpose drawing software is perhaps best suited for creating placing drawings

today. However, recent versions of Tekla Structures now provide a high enough level of

detail for modeling individual rebars. Currently, the placing drawings are created by de-

tailers, trained technicians who specialize in making placing drawings based on structural

engineering design information. Their expertise includes the addition of information that

makes it easier for an ironworker to perform his job well.

A barrier to implementation is the fact that the level of precision needed for generating

placing information is much higher than that typically needed for structural design tasks.

Placing drawings are "an entirely different matter" from other drawings, because they re-

quire the shop drawings of other trades, exact dimensions for concrete formwork, and the

precise construction sequence (Birley, 2008). The creation of placing drawings also reveals

areas where the design might not have the level of maturity necessary for construction. This

is because the level of precision needed to develop a mechanics model, estimate costs, and

estimate quantities is much less than that required to make placing drawings. Information

such as the location of bar chairs might not be needed to design the structural components,

but are required to build them and must be shown on placing drawings.

This requirement for more information poses a challenge for applying performance-
81

based design concepts that extend to placing information. During the earlier stages of

design, there is not enough information to generate placing information. Yet, it would be

helpful to know how preliminary design decisions might influence the overall ability to

place rebar.

Known uses

Placing information is ubiquitous in engineered projects made of reinforced concrete.

In current practice, placing drawings are typically created by rebar detailers. Placing draw-

ings are defined by several different industry bodies.

Concrete Reinforcing Steel Institute (2002) defines placing drawings and draws a clear

distinction between placing drawings and the inappropriate term "shop" drawings:

Placing drawings are working drawings akin to erection or assembly type


drawings, instructing the field Ironworker (Placer) where to place and tie the
reinforcing bars within the formwork. Placing drawings may also indicate
the bar support layout with a placing sequence, thus facilitating the efficient
installation of the reinforcing bars.
Placing drawings are prepared by detailers, trained technicians who are not
necessarily graduate engineers, but are extremely proficient in interpreting the
structural information shown by the contract documents. At no time does a
Detailer make an engineering decision. In fact, in today's litigious society, a
Detailer employed by a Reinforcing Bar Fabricator would be foolish to accept
this responsibility.
Detailers also prepare lists of the reinforcing bars that are shown on the
placing drawings. A bar list is a listing of reinforcing bars making up a bill
of materials. The bar list contains the quantities, sizes, lengths, and bending
dimensions of the reinforcing bars. These lists serve several purposes. The
Fabricator uses a bar list for shearing and bending, tagging, shipping, and in-
voicing. The Ironworker Foreman and the placing crew use a bar list for check-
ing delivery quantities, sorting bundles of bars in the job-site lay-down area,
and hoisting of proper bundles to the placing area on the formwork.
82

Placing drawings are not used in the fabricating shop, per se. Thus, the
generic term "shop" when applied to the detail drawings for reinforcing bars
in site-cast concrete construction is extremely inappropriate.

ACI 315, Sec. 3.1-3.2 mentions specific items that should be in placing drawings:

Placing drawings are working drawings that show the number, size, length,
and location of the reinforcement necessary for the placement and fabrica-
tion of the material. Placing drawings may comprise plans, details, elevations,
schedules, material lists, and bending details. They may be prepared manually
or by computer.
Placing drawings are intended to convey the Engineer's intent as covered in
the Contract Documents... The placing drawings must include all information
necessary for complete fabrication and placing of all reinforcing steel and bar
supports.

American Society of Civil Engineers (2000, Sec. 17.3.7) defines placing drawings as

essential for safety and performance and discusses the design professional's responsibility

to approve placing drawings:

Concrete placing drawings illustrate the reinforcing steel components that


will be part of a completed structure. These components are crucial to the
safety and performance of a completed facility.
Reinforcing steel components are furnished and placed according to the de-
sign professional's specifications in the construction contract documents. Ex-
amples of components for which placing drawings are prepared include cast-
in-place concrete and post-tensioned pre-stressed concrete structural elements.
The design professional has authority and responsibility for overall design
of the completed structure and for the review and approval of the placing draw-
ings for conformance with the project design concept and the information in
the construction contract documents.
The constructor and subcontractors have responsibility for preparing the
placing drawings, providing the materials specified, and completing the fabri-
cation and construction processes. This work is carried out in accordance with
the construction contract documents, approved placing drawings, and accepted
industry standards.
In most cases, placing drawings for reinforcing steel in cast-in-place con-
crete do not need design services and it is not necessary or appropriate for
83

the contract documents to call for certification by a registered professional en-


gineer. For post-tensioned pre-stressed cast-in-place concrete structures, the
design professional may delegate certain design activities to a specialty engi-
neer employed or retained by others and provide specifications for the load-
ing condition and other design parameters in the contract documents. In such
cases, the design professional retains responsibility for the overall safety and
performance of the completed structure. The specialty engineer is responsible
only for the design work delegated and certifies with signature and seal that
the related calculations and drawings meet the specifications provided by the
contract documents.

4.2.7 Database pattern

Intent

Store structure information in a database so that it can be reused by a computer program

for further analysis.

Assumed background

Foundational + Design + Problem recognition and solving + Communication, compre-

hension of relational databases (e.g., Codd, 1990), and knowledge of the capabilities of

BIM software for structural engineering (Eastman et al., 2008).

Motivation

When information is stored in a database, it can be accessed programmatically by com-

puter software. This is useful in BIM applications where communication methods such as

drawings, lists, and 3D renderings are constructed as views of database content. Databases
84

are also useful when tools for structural analysis are used to interact with other software

for design and coordination. Using a shared database avoids data duplication by requir-

ing information to be entered in only one location and used in several other applications.

This also reduces errors because a single change can be made once for the entire database

instead of multiple locations.

Applicability

Databases are applicable when the information needs to be retrieved by another com-

puter program, and when where the input data are already in a known form. This does not

work as well when information is of a free form nature, or is in a preliminary stage where

it cannot be defined with the precision necessary to fit into the database schema. In those

situations, it is better to use other communication patterns such as the Sketch (4.2.2) and

Narrative (4.2.1).

Structure

A database requires a data model that expresses the form of the information content to

be stored in the database. A relational type model is commonly used, representing data in

the form of tables and views (Codd, 1990). The relational data model for reinforced con-

crete bridge structural performance, earthquake hazard, repair methods, costs, and damage

states illustrates the Database pattern (Figure 4.8). This example shows the relationships

between the different components of the LLRCAT loss estimation methodology presented
85

in Mackie et al. (2008a).

Consequences and implementation

Using a database requires the definition of a data model to store all of the information.

Creating this data model could be very difficult when handling different types of infor-

mation content, especially when the different data types are combined with one another

at different levels of grouping. The relationships between different pieces of data are not

always well-defined, particular with functional design information, and for designs that are

at a lower level of maturity.

Defining the views needed to communicate database content is an additional step of

work that would not be needed if the communication were created directly. This would,

of course, depend on the kind of information that is already stored within the database.

These views must also implement a method for entering information into the database,

which might require substantial programming overhead in order to store information into

the appropriate fields.

Data validation is another barrier to implementation. If the data is assumed to be in a

particular format or type to be used by a computer program, proper care must be taken to

ensure that all of the data is valid before entering the database.

Databases might be inappropriate in cases where the rigidity of the data model prevents

certain objects from being properly communicated. In these cases, a freeform communica-

tion method such as a Narrative or a Sketch would be more appropriate. Databases might
86

Bridge Information

PK info id Downtimes

name PK time id
value
unit w downtime
metricValue -has- downtimeSD
metricUnit
w

Damage States
Performance Groups Repair Methods
—has-many-W PK PS..Jd PK repair id
1..*
PK PG id
"has many
FK1 repairjd
location FK1 qty_id
name
FK1 DS_id lambda repairDescription
FK2 EDP_id beta
description
V
state
has FK2 time id Repair Amounts
PK amount id
EDPs Performance Results Intensity Measures
1 amount
PK EDP id 1 PK result id
fe,PK IM id repair id
Q_id
unit FK2 INUd unit
description nany-^ EDPvalue description
FK1 result_id FK1 EDP_id
IM_value

Production Rates Repair Quantities Unit Costs

PK rate id PK QJd PK I cost id


-ha:
unit itemName unit
unitRate FK1 rate_id unite ost
unitRateSD -has- FK2 cost id unitCostSD
notes notes
PERTmin
PERTmax
PERTmode

Figure 4.8: Relational data model for bridge performance database.


87

also require a certain amount of precision that make it unsuitable for communicating pre-

liminary information when the designs are not mature enough to support such precision.

A single model is always unable to capture all relevant information, and the best use of a

database would be to combine it with other communication patterns to express the whole

design.

Known uses

Software for BIM commonly are examples of the Database pattern to communicate

general design information. Currently, the two most prevalent commercial programs for

structural engineering applications are are Autodesk Revit, and Tekla Structures. These

computer programs generate views of the information stored internally as a database, and

are intended to allow external programs to interface with the design information stored

internally.

4.2.8 Schedule pattern

Intent

Display process information with an emphasis on showing actions linked to time so that

concurrency and dependency can be easily seen.

Also known as

Timeline.
88

Assumed background

Foundational + Problem recognition and solving + Project management + Communi-

cation, and knowledge of the Lean Project Delivery System (Ballard, 2001).

Motivation

The schedule pattern is useful when the duration of a construction project needs to be

estimated. This allows communication of time dependencies and the sequence of activities.

Schedules constructed at different levels of detail allow the communication of different

information. During preliminary design phases, the schedule can show general feasibility

and milestones for design and construction.

Applicability

A schedule is one of the best methods for showing process information. It addresses the

problem of communicating how something will be built. The material and spatial informa-

tion needed to describe the final state of the structure can be broken down into construction

steps and shown in a schedule. Schedules communicate changes in material and spatial in-

formation with time during construction. This pattern addresses the poor alternative of only

describing the final state of the product without due consideration of the process behind it.
89

Mar 2008
ID Task Name Start Finish Duration
5 | 6 | 7 | 8\ 9 \10\11\12\13\14\15\16\17\18
1 Excavate Bent 5 3/5/2008 3/5/2008 1d
2 Set Column Casing and Weld 3/6/2008 3/7/2008 2d ^am
3 Grout and Paint 3/10/2008 3/17/2008 6d
4 Backfill Bent 5 3/18/2008 3/18/2008 1d •

Figure 4.9: Schedule for bridge repair scenario task (based on Mackie et al., 2008a, p. 177).

Structure

Schedules can take many different forms depending on depending on the method of

scheduling and the intent of the schedule. Schedules can be represented at different levels

of detail. A schedule at a lower level of precision might only show general tasks for the

overall project, while a schedule at a high level of precision might show individual tasks

for individual workers.

An example schedule based on an estimate for a bridge repair task is shown in Fig-

ure 4.9 (based on Mackie et al., 2008a, p. 177). This example provides information at the

level needed by Caltrans to perform estimates for budgeting.

Consequences and implementation

Implementing a schedule can be difficult because the process data needs to be expressed

as activities with time, durations, and sometimes also resources. The dependencies of

activities must be known and whether the activities will happen concurrently, or one at a

time. There might not be enough information to estimate a schedule unless the process

information for the design is mature enough.

Schedules usually have difficulty presenting information that is spatial in nature because
90

the focus is not on where, but on when. This deficiency can be overcome using particular

methods of showing schedules for repetitive work, such as the Line of Balance method

(United States Navy, 1962; Al Sarraj, 1990), or concentrating on site planning

Schedules need to be created at a particular level of grouping in order to maintain

clarity. The different units of work are broken down into appropriate levels depending

on how detailed the work plan needs to be in order to satisfy the user of the schedule.

For example, when doing initial preliminary planning to investigate the feasibility of a

structural system option, the exact details for when and where to install dry wall panels

might not be necessary to know.

Known uses

Schedules are used in all construction projects, and especially in the construction pha-

ses. Two common formats for showing scheduling information are the Gantt chart (Fig-

ure 4.9) and the Line of Balance (LOB) method. The LOB method was developed by the

U.S. Navy during the 1950s (United States Navy, 1962; Al Sarraj, 1990). A schedule that

uses the line of balance method typically emphasizes spatial information in addition to the

time information. The line of balance is especially useful for repetitive work that is per-

formed in multiple locations on the project site. For example, multiple floors of high-rise

buildings, and highway pavement.

Bridge loss modeling. Schedules for repair of highway bridges are used in various loss

estimation methodologies. Mackie et al. (2008c) uses estimated schedules for bridge repair
91

as data for repair duration and repair effort on specific reinforced concrete bridges. Werner

et al. (2006) also estimates scheduling data in order to determine probabilistic bridge down-

time throughout a highway network after earthquakes.

4.3 Application to highway bridge loss estimation

This example of highway bridge loss estimation methodology Mackie et al. (2008a)

illustrates the importance of using the communication patterns in describing preliminary

information. Understanding pattern usage facilitates conversation with the various partici-

pants involved in collecting data to support this methodology: structural engineers, bridge

estimators, and Caltrans decision-makers. Failure to communicate preliminary information

clearly among the participants can lead to confusion and lack of progress. This is impor-

tant because the implementation of the PBEE methodology required specifying enough

information to estimate future project repairs, yet at a very early stage during preliminary

design.

In some cases, repair estimation can only be done at stages of design where enough

information exists to construct the bridge. But, for bridges of "typical" construction, repair

estimation can be done early in design because many details and sub-structures can be

selected during an immature design state. This is the situation with conventional highway

overpass bridges which have similar details and similar superstructure design (Ketchum

et al., 2004). The opposite situation occurs for unique signature bridges, such as the self-
92

anchoring suspension span of the new San Francisco-Oakland Bay Bridge, which has very

few details that can be known during early stages of design (Bay Bridge Public Information

Office, 2008; Sun et al., 2004).

Understanding the different categories and qualities of design information is also im-

portant because much of the information necessary to determine post-earthquake repairs

involves material, spatial, and process information that is not usually part of preliminary

structural design. For example, identifying the materials and processes involved in in-

stalling temporary supports and installing column jackets are required for computing the

repair time and repair cost, but are not needed for completing the structural design on the

bridge.

The local linearization repair cost and time (LLRCAT) loss estimation methodology

implementation and its application to bridges are presented in Mackie et al. (2008a). The

example presented here adopts that methodology and focuses on the process used to gener-

ate the preliminary design information and on the communication patterns used to express

that design data at the appropriate level of detail. Table 4.2 shows the procedure broken

down into individual steps along with the communication patterns used in each step.

Step 1: Define bridge design information.

One of the challenges to defining the bridge design information for the loss methodol-

ogy is the fact that this is done at an early stage of design where alternatives are still being

considered and it is not desirable to spend lots of time detailing any particular option in or-

der to reduce rework. So, this challenge of defining bridge information is to define enough
93

Table 4.2: Loss estimation procedure for reinforced concrete bridges.

# Step description Communication Patterns applied

1 Define bridge design information Spreadsheet (4.2.4),

Sketch (4.2.2)

2 Define performance groups and damage states Narrative (4.2.1)

3 Compute EDPs from a mechanics model Mechanics Model (4.2.5),

Spreadsheet (4.2.4)

4 Create damage scenarios Sketch (4.2.2), Annotation (4.2.3)

5 Estimate repair cost and repair schedule for Spreadsheet (4.2.4),

damage scenarios Schedule (4.2.8)

6 Obtain repair cost and repair effort (production Spreadsheet (4.2.4)

rate) for each repair quantity

7 Run LLRCAT program to produce results Spreadsheet (4.2.4)


94

of the bridge so that losses can be estimated, but not to require too much such that the in-

formation would not be available this early in the design. This means that the information

presented cannot be at the level of precision required for construction or purchasing, but

should be at the level of precision needed for defining a mechanics model and ensuring that

the structure will be able to perform its basic functions.

The bridge information at this stage is defined using two different communication pat-

terns: the Spreadsheet (4.2.4) and the Sketch (4.2.2). A spreadsheet gives the numeric

values needed for the bridge. Most of the values presented in this spreadsheet refer to the

external dimensions of the bridge superstructure, roadway deck, abutments, and founda-

tions. This allows the computation of volumes and weights so that the amounts of quanti-

ties can be estimated based on size. Some basic material properties must also be defined

such as unit weights for steel and concrete, and the compressive strength of concrete f'c,

and yield strength of reinforcing steel fy.

Only a few pieces of information are needed beyond the basic dimensions. This means

that the rest of the procedure can generally continue even before the state of the design

reaches the level of maturity needed for construction. So, the loss estimation method can be

performed at early stages of design, which is when performance information can have the

most influence on the final design without requiring substantial rework. Basic dimensions,

sizes, and material counts can be estimated based on previous designs and preliminary

design tools.
95

Step 2: Define performance groups and damage states.

A performance group (PG) is a group of one or more bridge parts that contribute signifi-

cantly to repair-level decisions; are repaired independently from other performance groups;

and, have damage which can be characterized by an EDP from the mechanics model. Per-

formance groups are not necessarily the same as load-resisting structural elements, and can

include "non-structural" components because they also suffer damage and contribute to

repair costs.

The unit of analysis needed for design is often different from that which is needed

for repairs. Grouping bridge parts into a performance group allows for groupings that are

specifically suited for analyzing damage and repair. In the analysis for design of seat-type

abutments, several distinct functional components: shear keys to limit transverse motion,

back walls to limit longitudinal motion, and bearings to connect superstructure and abut-

ment. Each components must have its own material properties and force-displacement

behavior. But, understanding the repair of seat-type abutments, such distinctions are un-

necessary, and it becomes necessary to treat all these parts as a whole system. Bearings

need to be replaced when their strain capacity is exceeded, and this only occurs when there

is damage to the shear key or back wall which allowed this excess displacement to occur at

all. Thus, it is better to handle these repairs as always linked to one another.

These different units of analysis needed to characterize repair are the performance

groups for the structure. The performance groups are connected to the mechanics model of

the structure via the EDPs that characterize the damage of the performance group.
96

Steps involved in defining performance groups:

(a) Use best judgment to enumerate the different types of possible post-earthquake dam-

age and repairs. Data on damage and repairs from previous earthquakes may be un-

available because new designs may have never experienced a major earthquake, and

bridge replacement may have been preferred over bridge component repair. Group

discrete repairs if they are performed together.

(b) Create the PGs by grouping the physical portions of the bridge where these repairs

are located.

(c) Choose the EDPs that will characterize the damage of these individual performance

groups.

(d) Define one or more damage states (DSn) for each PG by ranking the damage to

each PG in order of ascending severity. Also define another damage state (DS°°)

corresponding to the maximum possible severity, which can be equal to the highest.

(e) Use prior data on earthquake damage to assign EDP values corresponding to the

different damage states.

(f) Check for overlapping repair methods among the performance groups. If there are

overlaps, then either consolidate multiple performance groups, or split up into smaller

groups so that there is no longer any overlap.

Once the performance groups with their corresponding repair methods and damage

states are defined, they need to be described using the appropriate level of detail. This is

a thorny issue because we only have preliminary design data to deal with and are working
97

with a design that does not have enough maturity to be constructed. Because of this relative

lack of maturity in the design, it is not possible to describe things at a Describe the PGs

using a Narrative (4.2.1), and input the numerical data into the Spreadsheets (4.2.4) of the

methodology.

Step 3: Compute EDPs from a mechanics model.

After the dimensions of the bridge are determined and the EDP responses are known,

then a mechanics model can be created that accurately represents the bridge.

The information needed to create the mechanics model might be different than that

needed for the bridge design information. This is because the model requires selection of

such things as material models and numerical analysis routines, which are necessary for

computing the model, but are unnecessary for constructing the bridge. In the creation of

the mechanics model, there may be the need to create additional Abstract (5.1.1) variables

to represent the needed information.

The mechanics model must be very specific to the individual structure, because differ-

ences in site conditions and bridge geometry might not affect the definition of performance

groups and repair methods, but could have an impact on bridge response during earth-

quakes.

The mechanics model should be analyzed with a suite of ground motions for the site at

different levels of the chosen intensity measure. This will produce a range of data that can

be used to interpolate the performance of the bridge as a function of earthquake intensity.

For each ground motion, the EDPs for each performance must be recorded. These values
98

will produce a Spreadsheet containing all of the numerical data in a summary form. The

time histories are not needed, only the value of the corresponding EDP. The types of EDPs

could be the maximum values throughout the time history (e.g. maximum column drift),

or the value at the end of the time history (e.g. residual column drift).

The procedure for running a bridge model and selecting analysis methods and IMs are

described elsewhere (Mackie, 2004; Mackie et al., 2008a).

Step 4: Create damage scenarios.

A damage scenario is a correlated assembly of performance group damage states. The

scenarios are to be used for estimating the cost and durations of repair. The scenarios

should also be representative of the different damage states and should cover most if not

all of the different damage states so that it will be possible to understand the cost and time

involved in all the different repair methods.

The scenarios do not necessarily represent real instances of earthquake damage, but

are intended to reveal dependencies in the different repair methods and provide a realistic

indication of how much various repairs might cost in combination with other repairs.

The steps for creating damage scenarios are:

(a) Decide on the number of damage scenarios to create based on the bridge design

information and the nature of the expected bridge damage.

(b) For each damage scenario, group the damage states for each performance group to-

gether in a way that it is feasible for all the chosen damage states to be achieved

together in the bridge.


99

(c) Decide on the repair methods to be used in each damage state.

(d) Create drawings with Annotations (5.1.1) showing the extent of the bridge repairs to

perform, and the quantities of each repair item.

Good damage scenarios represent the correlations among the different damage states in

each performance group. The mechanics model provides these correlations through defor-

mation compatibility for the deformation-based EDPs, and force equilibrium for the force-

based EDPs. For example, good scenarios will have similar damage states for elements that

undergo similar displacement. A bad scenario would have vastly different damage states

for those same elements. Intuition and judgment are needed to design the repair methods

and repair quantities in a manner that suits these correlations.

Assigning repair quantities can be challenging because specific details are not yet fully

available early in the design process. Usually, full estimation of repairs does not happen

until after making a commitment to details down to the level of individual rebar. But, this

information on repair is most useful during these earlier stages because then the analysis

can have greater influence on overall project delivery. This challenge can be overcome by

using case study repair data from similar bridges after previous major earthquakes, labora-

tory testing data, or mechanics model results.

The most important part of this step is the creation of drawings that communicate the

damage scenarios in a clear and spatial way. This enables the estimation step to proceed,

and provides enough information so that the damage scenarios can be critiqued and checked

by the bridge engineer who estimates the scenario repairs.


100

Step 5: Estimate repair cost and repair schedule for damage scenarios.

Each of the damage scenarios should have enough information to estimate the cost and

schedule for the repairs.

The cost estimates should include a total amount for each separate type of item item. It

should provide enough information to determine a unit cost for each repair item. Obtaining

this unit cost allows the information to be reused for all the different damage states that

might be encountered in the structure.

The estimating process also serves as a check on whether the estimating quantities for

the repair methods were correct and whether there is enough information in the structure

design in order to estimate. This is a difficult issue because there might be insufficient

information to estimate quantities if the design lacks sufficient maturity. The estimating

process will reveal any missing quantities. For example, an item for installation of a steel

jacket column casing, might be missing the associated excavation and backfill needed at

the column base.

The schedule estimate also reveals where certain types of work can be performed con-

currently with other types of work. This would affect the repair duration, but not the repair

effort, which is a measure of the number of crew working days required to complete the re-

pair. It is assumed that some durations can be shortened if additional resources are allocated

to a project.

Step 6: Obtain repair cost and duration for each repair quantity.

The unit costs and durations for each repair item can be derived from the estimating data
101

provided in the previous step. The variations in the unit costs between the different scenar-

ios provide some statistical data that can be used to estimate a mean value and dispersion

on the costs and durations.

The method for obtaining unit costs involves:

(a) For each repair quantity line item in the cost estimate, divide the cost by the total

quantity.

(b) Choose the most representative value among the different scenarios to use for the

unit costs in the methodology

(c) Create a Spreadsheet summarizing this information

The method for obtaining durations is more complicated because it involves the sum-

mation of several sub-tasks that are reflected in the schedule estimates, but that are not

directly tied into a repair quantity. This methodology is based on using repair quantities as

the interim variable needed to compute the two decision variables of repair cost and repair

time.

The procedure for obtaining production rates for repair effort involves:

(a) Group the schedule activities by the damage states for each performance group.

(b) Sum the durations of time for each activity for each damage scenario

(c) Create a spreadsheet containing the minimum, average, and maximum amount of

effort needed to perform each repair item.


102

Step 7: Run LLRCAT program to produce results.

The final step is to ran the LLRCAT Matlab program that reads all of the design and

model information from Excel spreadsheets and computes the probabilistic repair costs and

times (Mackie et al., 2008a).

4.4 Application to building information modeling

New advances in technology for building information modeling (BIM) open up new

possibilities for work flow in the architecture, engineering, and construction (AEC) indus-

try. Tools for BIM allow the storage and representation of a much greater precision and

range of information than available before. Additionally, the implementation of new data

exchange standards allows for greater electronic interchange of information among dif-

ferent software applications. These advances provide greater opportunity to communicate

design information in a way that is more clear and complete with less errors.

Nevertheless, the limitations of computer models must be understood in order to avoid

the pitfall of poor design communication. Although BIM technologies have their advan-

tages, they also have disadvantages in communicating design information which must be

understood. For example, Day (1997, p. 156) points out a conceptual issue that is en-

countered when using today's general-purpose BIM tools for communicating spatial design

information:

One of the problems encountered when trying to construct a full 3D model


rather than a set of drawings is that the model lacks the abstraction of the draw-
ings and there is a tendency to keep adding detail, thus making it unwieldy.
103

When working in 3D at the detailed design stage this lack of abstraction can
also lead to frustration as it is impossible to construct many parts of the build-
ing because of decisions that have not yet been made. This either results in
time being spent trying to make these decisions, perhaps at a time which is
earlier than appropriate, or the decision is not made leaving the model incom-
plete. This incompleteness then becomes a problem itself as it has a knock-on
effect on other parts of the building.

This difficulty comes into sharper focus when examining the definition of BIM technol-

ogy. Eastman et al. (2008, p. 13) defines BIM as "a modeling technology and associated

set of processes to produce, communicate, and analyze building models", which is charac-

terized by the following four properties:

Building components that are represented with intelligent digital represen-


tations (objects) that 'know' what they are, and can be associated with com-
putable graphic and data attributes and parametric rules.
Components that include data that describe how they behave, as needed for
analyses and work processes, e.g., take off, specification, and energy analysis.
Consistent and non-redundant data such that changes to component data are
represented in all views of the component.
Coordinated data such that all views of a model are represented in a coordi-
nated way.

Eastman et al. (2008, p. 15-16) also contrast BIM with four aspects of models that are not

BIM: models that contain 3D data only without object attributes, models with no support

for behavior, models composed of combined 2D CAD reference files, models that allow

changes in one view that are not automatically reflected in the other views.

Based on Eastman's definition on what is not BIM, there are some design communi-

cation aspects that are not suited for BIM tools. Early in the design process, the object

attributes might not be well known or determined. This is especially true when commu-
104

nicating abstract functional information, which might not yet be related to individual el-

ements or material characteristics. Also, some parts of the structure might not yet have

a well-defined and understood behavior until later in the design process. In these early

stages, the ability to communicate and define information without full object behavior is

essential. This is the reason why tools such as the Narrative (4.2.1), Sketch (4.2.2), and An-

notation (4.2.3) patterns are needed for communicating preliminary design information at

the appropriate level of information, without implying too much information prematurely.

Eastman et al. (2008, p. 14) also define parametric BIM objects by their properties:

• [Cjonsist of geometric definitions and associated data and rules.


• [Gjeometry is integrated non-redundantly, and allows for no inconsis-
tencies. When an object is shown in 3D, the shape cannot be represented
internally redundantly, for example as multiple 2D views. A plan and el-
evation of a given object must always be consistent. Dimensions cannot
be 'fudged'.
• Parametric rules for objects automatically modify associated geome-
tries when inserted into a building model or when changes are made to
associated objects. For example, a door will fit automatically into a wall,
a light switch will automatically locate next to the proper side of the door,
a wall will automatically resizes itself to automatically butt to a ceiling or
roof, etc.
• Objects can be defined at different levels of aggregation, so we can de-
fine a wall as well as its related components. Objects can be defined and
managed at any number of hierarchy levels. For example, if the weight of
a wall subcomponent changes, the weight of the wall should also change.
• Objects rules can identify when a particular change violates object feasi-
bility regarding size, manufacturability, etc.
• [Ojbjects have the ability to link to or receive, broadcast or export sets
of attributes, e.g., structural materials, acoustic data, energy data, etc. to
other applications and models.

These properties of BIM objects force the combination of spatial information along

with functional information and the behavior of pieces in the structure. But, this might not
105

be desirable when the designs are immature and there is insufficient design information to

show the spatial relationships. On this concept, Doumont (2002b) comments that "Pictures

are indeed superior for conveying intuitive or global information. By contrast, they do a

poor job of expressing abstract concepts and lack the accuracy that words are endowed

with. Words, in a sense, are worth a thousand pictures." This is why the Narrative pattern

(4.2.1) is important in cases where the design information is not yet rich enough to support

accurate spatial information.

The ability to show different levels of aggregation is good only when detail is available

down to the level of no aggregation. For example, during preliminary design, only groups

of rebar in reinforced concrete might be known, but not the individual bar layouts. A good

tool for communication would indeed allow the different levels of aggregation, but still

allow the use of the less detailed levels without forcing the definition of the more detailed

ones until later in the design process.

These factors indicate specific properties that BIM processes need in order to support

performance-based structural engineering design:

First, the ability to include abstract information that is not specifically tied to spatial el-

ements in the structure. This is important for expressing intensity measure (IM). Structure-

independent IMs for earthquakes depend on site conditions alone, and are not related to

any structural objects. Structure-dependent IMs require both site condition information

and general structure properties, such as the fundamental period, T\. The value of T\ is

not tied to a specific part, but is a general property of the whole structure. This is easily
106

accomplished using freeform communication patterns, and should be included within BIM

tools as well.

Second, the ability to express abstract and functional information without requiring

a high degree of spatial precision. Determining engineering demand parameters (EDP)

only requires enough information to define an adequate mechanics model. Often this only

requires preliminary design information for basic structural system behavior, and can be

done before high precision details are available. For example, the preliminary specification

for a lateral force-resisting system might contain only basic abstract quantities representing

structural behavior, such as damping ratio, stiffness, yield strength, and displacement limits.

That kind of specification might not have highly precise geometric rules, but should still be

capable of integration into the building model.

Third, the ability to express a range of possible structure states. Describing damage

measures (DM) requires the ability to model a range of discrete or continuous damage

states. This definition requires consideration of multiple possible states for the structure

after an earthquake. Furthermore, each value of DM will also trigger a different value of

the decision variable (DV) according the severity of the damage. A range of possibilities

must be defined instead of merely the ideal as-built condition of the structure.

Fourth, the ability to define performance groups with freeform information and cus-

tomized grouping. Expressing damage scenarios and repair methods requires freeform

information because many repairs are specified conceptually rather than with specific de-

tails. Customized grouping is needed because repairs within a performance group may
107

involve many elements in addition to the actual piece needing repair. For example, repair-

ing a bridge abutment backwall may require approach slab replacement because of access

requirements.

Fifth, the ability to define probabilistically uncertain quantities. Evaluating the struc-

ture using a performance-based earthquake engineering methodology requires definition of

uncertain quantities. IM data cannot be expressed by a single number, but must include

a distribution and parameters, such as median and dispersion. EDP data is also uncertain

depending on the ground motions suite used for analysis. DM data is uncertain because

of variations in the actual as-built condition of the structure. All of these quantities need

proper expression within a BIM process.

Based on these requirements and communication inadequacies of current software tools,

the following improvements are recommended for future BIM software:

1. Integrated tools for freeform sketching and annotation.

Providing a specific function for creating freeform annotations inside a working

model view would overcome the current barrier of making annotations in a sepa-

rate drawing tool or window. A feature for taking a model snapshot and scribbling

on it would allow the communication of conceptual information and avoid the ten-

dency to allow software programs to generate the views without considering how

other project participants will understand it. The snapshot freezes a model view and

prevents unintended changes to the sketch. This feature could be combined with the

hyperlinks feature noted below.


108

2. Hyperlinks that connect model entities to freeform descriptions.

Hyperlinking from model entities to freeform narratives and spreadsheets would

overcome the inability to link BIM objects to freeform descriptions. This would

allow the expression of abstract and functional information without requiring a high

degree of spatial precision, because such information could be linked to a volume

or aggregation inside the model instead of a single specific piece within the model.

Linking a 3D model-based visualization of the structure to spreadsheet cells would

also reduce the amount of time needed for spreadsheet data entry. For example, the

geometric spreadsheet data needed for the LLRCAT methodology would be much

easier to navigate and understand if linked to a 3D model view.

3. Present conceptual mechanics models and sketches as another view layer in 2D or

3D space.

This feature would allow different conceptual presentations of the structure to be pre-

sented together. For example, a rendered model view could be presented atop a hand

sketch of the same location to communicate design intent in a freeform manner. Or, a

line element mechanics model drawing could also be displayed to illustrate how the

different components act together to perform their intended structural function. This

would violate the principle of non-redundant geometry and composition of multiple

2D references, which Eastman et al. (2008) use in their definition of BIM. But, it is

necessary for ensuring good design communication.

4. Capture design intent at the moment of model object creation.


109

A key ingredient in the design process is not only the final material and spatial speci-

fication of the building, but also the functional reasons for their selection and design.

Presenting an option for the designer to enter freeform information about why a par-

ticular choice was made would enable automatic annotation of this design intent. For

example, this feature would allow the designer to note that a steel floor beam must

be a W14 shape because of ventilation duct size. This additional metadata could be

shown back to remind the designer if the shape were later changed. This feature

would help overcome the lack of support for abstract design information in current

BIM software.

5. Assign probabilistic uncertainties to numeric values inside the model.

This feature would overcome the inadequacy of using a purely deterministic model

for evaluating probabilistic performance. Performance-based earthquake engineer-

ing methods are inherently probabilistic, and providing this information would aid

performance evaluation. This might be implemented by adding fields for naming a

probability distribution and for the values of the distribution's parameters. This ad-

ditional information could be used for probabilistic structural analysis, and ignored

for other purposes where the median value could be used deterministically instead.

The advances in BIM technology are indeed exciting for the field of structural design.

However, caution is needed in order to avoid neglecting the strengths of time-honored de-

sign communication principles and methods. Implementing these suggested features would

improve the ability of BIM software to use the patterns of good design communication.
110

5 Summary

What I have done in this chapter:

1. Provided patterns for communicating design information at the appropriate abstract

and concrete levels of detail.

2. Illustrated the necessity of annotation and text in order to describe building model

attributes. Annotation communicates the appropriate amount of conceptual informa-

tion, while the building model (and graphical views) communicate attribute data.

3. Provided an example for communicating abstract performance criteria without over-

specifying the level of detail required to check it.

4. Commented on the use of BIM technology for communicating design and suggested

recommendations for improved design communication.


Ill

Chapter 5

Analysis and decision patterns

The patterns of analysis and decision for structural design involve approaches for de-

termining the information about the structure. The patterns for analysis describe ways that

additional information about the structure are created. This analysis information is not

necessarily required for constructing the actual structure but are required for the design

process. The Abstraction pattern describes the creation of abstract quantities in order to

facilitate the design process. The Aggregation pattern describes the concept of grouping

for physical items, and the Agglomeration pattern describes the concept of grouping for ab-

stract items that relate to function rather than material composition. The Updating pattern

shows an example of how design information is changed and updated. The Engineering and

Social Performance patterns show the formation of criteria for judging whether a design is

correct, and provide measurements for choosing one option among many. Engineering Per-

formance mainly focuses on the product, while Social Performance focuses on the people
112

and process involved in making the product.

The patterns for decision making describe ways of choosing more detailed spatial and

material information based on abstract design information and the goal of fulfilling the

functional requirements. They describe different ways of arriving at more mature design

information, and use the information from analysis results to make decisions. The Pick Any

pattern simply chooses any permissible item among a set of feasible choices. The Minimum

and Maximum patterns describe methods of selection based on the numerical value of a

single parameter, assuming there are no other tradeoffs in the selection. The Optimization

pattern provides examples of how to make decisions when there are tradeoffs and multiple,

competing criteria involved in making the best design choice. The Prescription pattern is for

decisions that are governed by a set of strict rules rather than intuition. The Set Exploration

pattern is a description of the set-based design method and illustrates the ability to conceive

of novel ways to make decisions based on the same analysis capabilities.

The analysis and design decision patterns proposed here address the common relation-

ships between the design decisions to be made and the qualities of the variables at hand. It

will be shown that under some circumstances, the appropriate tools for solving the design

problem are rather different. The solution could require a detailed engineering calculation,

a conversation between the structural engineering and the general contractor, or a lookup

to a table inside the building code. The proper identification of these different patterns will

help in the understanding of how best to approach the problem in a manner that promotes

innovation.
113

This chapter does not emphasize the actual methods for doing structural analysis itself,

but rather is focused on the conceptual content needed to in the analysis process. Patterns

for executing and running mechanics models are described in Heng and Mackie (2008),

where the design pattern concept is used to to identify best practices in object-oriented

finite element software design. The design patterns for analysis here are not specific to

finite element methods, but rather reflect the kinds of analysis results that are required to

make design decisions.

The names of these patterns, once adopted, can be used as a metalanguage to describe

design intent when using the pattern. Being able to label how a decision was made and

what type of analysis was carried out provides the rational for the information. Gamma

et al. (1995, p. 3) note that using pattern names "increases our design vocabulary", makes

it easier to discuss and document design, simplifies communication about trade-offs, and

allows people to "design at a higher level of abstraction". This discussion of abstract design

enables better design of the design process itself.

5.1 Patterns for analysis

5.1.1 Abstraction pattern

Intent

Create a variable that is itself non-physical in order to make general analysis feasible.

Convert the real spatial and material information into another form to facilitate thinking.
114

Also known as

Mechanics model variable, simplification, modeling assumption.

Assumed background

Foundational + Materials science + Mechanics + Problem recognition and solving +

Design, knowledge of dynamics of structures (Chopra, 2007), and knowledge of basic

earthquake engineering concepts.

Motivation

In mechanics models, it is sometimes necessary to create objects that are not directly

linked to real spatial or material design information, but that are useful for the process

of thinking. These kinds of abstractions allow mathematical manipulation and intuitive

physical insight into a design problem, yet without dealing with how these abstractions

must be instantiated as real, physical objects. Sometimes these abstractions are created out

of convenience as simplifications because the exact nature of the real physical phenomenon

is too complicated to capture. In these cases, only some essential characteristics might be

necessary to express the phenomenon in a meaningful mechanics model.

Applicability

The Abstraction pattern is useful in situations where:

1. A mathematical analysis procedure can be converted into a form similar to something


115

more basic and familiar, such as simplifying analysis of a composite material into an

equivalent homogeneous material.

2. Quantities inside mechanics-models occur frequently enough to be named and as-

cribed physical characteristics, such as with factors and coefficients that occur fre-

quently in design equations.

3. The actual physical phenomena are too difficult to model exactly, but can be ap-

proximated using a simple model to achieve acceptable results, such as with using

simplified material models which are abstract entities.

Structure

The abstraction pattern starts with something that is difficult to handle in its original

form and transforms it into another things that is more familiar, or more general. They are

discussed as though they were real entities, although they are in reality concepts and not

necessarily directly based on physical objects.

Consequences and implementation

Abstractions are the objects that are necessary for design, but unnecessary for for con-

struction because they are non-physical. This pattern allows a high level of thinking and

analysis to characterize the essence of an engineering problem at the appropriate level of

detail, avoiding the need to specify details prematurely.

The potential pitfall of creating abstractions is that they require careful judgment to be
116

used correctly. A lack of judgment might lead to false results based on abstract reasoning.

For example, Ferguson (1992, p. 183) laments that "nearly all engineering failures result

from faulty judgments rather than faulty calculations". A design procedure that consists of

abstractions without giving due consideration to the realities of scale, form, and proportion

are more susceptible to lapses in judgment than one that is based on the "subtleties of real

world engineering" (Ferguson, 1992, p. 169). It is necessary to deal with abstractions when

creation conceptual and mechanical models, and good judgment is needed to complete

design beyond the level of precision where the abstractions are applicable.

Known uses

Modulus ratio, n, in transformed section analysis. For reinforced concrete and steel

composites made of materials with different elastic moduli E, use a single value of E

and modify the area A. When calculating an engineering quantity using multiple different

kinds of materials, it can become confusing to deal with each material's properties sep-

arately. Using a "transformed section" allows one to perform all the computations using

only one material's properties by using a transformation on the area. In reinforced concrete

elements (MacGregor and Wight, 2005, p. 388-389) and steel composite design (Segui,

2007, p. 554—555), the area of the transformed section is usually computed by scaling the

steel area by the modulus ratio n = Es/Ec.

Material models. Assuming a particular form of material model, such as a linear-elastic,

perfectly-plastic stress-strain behavior. This abstraction is based on some physical proper-


117

ties, but is itself simply a mental construction because such materials do not actually exist

and cannot be created. In such a material model, the parameters Gy and E are actually

abstractions that permit analysis and thinking.

r factor in structural dynamics. The factor Tn = §„s/Mn is a quantity that occurs

frequently in dynamic analysis of structures (Chopra, 2007). It is a quantity that is not

directly physically measured or that is necessary for construction. Rather, this quantity is

an artifact of the equations of motion needed to analyze dynamic structural behavior. It is

useful for the purposes of design and checking engineering performance, but will not help

when evaluating for social performance or construction.

Euler-Bernoulli beams. The Euler-Bernoulli beam theory assumes that plane sections

remain plane. This assumption greatly simplifies analysis, but neglects the effects of shear

deformation and warping. Using this beam theory creates abstractions that approximate

actual beam behavior.

PBEE Variables. The variables in the PEER performance-based earthquake engineering

(PBEE) methodology are examples of abstractions. The intensity measure (IM), engineer-

ing demand parameter (EDP), damage measure (DM), and decision variable (DV) represent

concepts for general analysis that can be realized using different physical information.

The IM could be structure-independent, based on the physical earthquake phenomena

alone, such as peak ground acceleration or peak ground velocity. Or, the IM could be

structure-dependent, such as the spectral acceleration at the fundamental period, Sa(7i).

The EDP could be based on forces, displacements, or ratios at their maximum, minimum,
118

or residual values. The DM could reflect a smooth continuum of increasing damage, or be

based on discrete damage states. The DV could be based on dollars, downtime, repair time,

human safety, or other concerns.

5.1.2 Aggregation pattern

Intent

Group similar items by their material or spatial information, and treat conceptually as a

single item. Simplifies a conceptual or mechanics model by reducing the number of pieces.

Assumed background

Foundational + Materials science + Mechanics + Problem recognition and solving +

Design + Communication.

Motivation

In the design of structural elements, the behavior of a single piece of material is not

as important as long as the entire group of material can be assumed to work together.

For example, inside a prestressing cable, the individual strands are usually not considered

separately for design as long as the entire cable's behavior can be understood. Similarly, in

reinforced concrete design, the behavior of each individual bar is usually not considered as

long as the the total amount of steel functions together as a unit.


119

Applicability

Use the Aggregation pattern when many small items perform together as a group and

have similar behavior; individual behavior of individual items is not needed, and only the

overall group behavior is needed; and the material and functional information is important,

but highly precise spatial information is not. The Aggregation pattern is not applicable in

design situations where the spatial location of each individual rebar were significant, as

might be the design of a highly congested element.

Structure

The Aggregation pattern a conceptual device that helps eliminate unnecessary complex-

ity during the design process.

Consequences and implementation

This pattern provides the ability to compute performance without requiring spatial

specificity. For example, the reinforcement ratio p combined with a quantity As and depth

to steel centroid d allow the consideration of rebar behavior in a reinforced concrete ele-

ment without needing to specify a particular bar layout. These quantities support set-based

design by allowing engineer to work with a set of acceptable rebar layout options by only

dealing with the quality needed for input into the mechanical model. Aggregation uses

abstract parameters in order to represent a group of different objects inside a mechanical

model.
120

Aggregation is a useful tool for specifying information at the correct level of precision.

In the design of reinforced concrete, specifying the exact location of each individual bar to

a high degree of precision might be unnecessary to achieve structural performance.

Aggregation might be disadvantageous when the exact spatial location is needed for

each individual piece. For example, aggregating multiple rebars together in a beam might

be useful for performing section analysis, but would make it difficult to check beam-column

joint congestion.

Known uses

Area of steel, rebar groups. The area of steel As in reinforced concrete design is an ex-

ample of aggregation. The individual reinforcing bars are treated as a single object in cross

section for checking certain abstract quantities inside a mechanics model. For example,

checking the nominal moment capacity requires the value of As and steel centroid location,

but not the precise spatial arrangement of the individual bars.

Rebar can also be defined by groups instead of individually. For example, a design

could specify the number of bars, bar size, and cover location (e.g. "5-#8 bar @ 2" cover"),

while leaving the exact location of each individual bar up to the fabricator, detailer, and

placer. This works well as long as there is no ambiguity in the specification and the design

is complete.

Reinforcement ratio. The reinforcement ratio p is another type of aggregation based on

the rebars inside a reinforced concrete element. The reinforcement ratio is computed as the
121

ratio between area of steel As and the concrete section gross area Ag. ACI 318, Sec. 10.3.2

defines three different reinforcement ratios, p for tension steel, p' for compression steel, and

Ph as the reinforcement ratio producing balanced strain conditions where the steel yields at

the same time that concrete crushes.

The value of p is important for checking certain conditions in the mechanics model.

It is used to verify that steel yields before concrete crushes for ductile behavior in flexure

elements, and can also be used to estimate rebar congestion for placing difficulty during

construction.

Prestressing strands. The most common type of prestressing strands are seven-wire

strands in a twisted bundle (Naaman, 2004, p. 46-49). Although the strand is made up of

seven individual wires, the entire strand is treated as a single item for structural analysis.

The tension force is assumed to be distributed across the entire area of the each wire, and

acts through the strand center.

Bolt groups. Groups of bolts in steel connection design can be treated for design as a

single reaction force on a member with a possible eccentricity (Segui, 2007, p. 443-^46).

Individual bolts need to be considered for limit states involving the bolts and the holes or

slots. But, once the connection is known to be designed properly, the individual bolts are

usually not necessary to consider to check other measures of engineering performance.


122

5.1.3 Agglomeration pattern

Intent

Group different types of items in a structure by their function in order to facilitate

analysis.

Assumed background

Foundational + Materials science + Mechanics + Problem recognition and solving +

Design + Communication.

Motivation

For structural analysis, the behavior and function of structural elements can be more

important than their actual material composition and exact spatial extent. Agglomeration

is the concept that deals with the creation of conceptual items such as reinforced concrete

"beams" that are physically composites of concrete and rebar intended to resist loads in

flexure. The beam concept itself is an agglomeration of several other materials in a partic-

ular spatial arrangement.

This concept is helpful because an agglomeration can be modeled using simpler meth-

ods than considering its real composition. For example, a reinforced concrete beam's non-

linear behavior can be characterized in a mechanics model using a M-0 moment-curvature

relationship instead of modeling the individual rebars and concrete matrix. The beam's be-

havior requires knowledge of the concrete and steel parts, but once agglomerated, knowl-
123

edge of neither is necessary to characterize the component in a model.

Applicability

Use functional agglomeration when there are several different components perform the

same function in the structure, and when accomplishing a particular structural function

requires multiple components that work together. This is not applicable when there is no

related functional information between the different structure components.

Structure

The agglomeration pattern involves grouping different kinds of structure parts by their

function. The process requires judgment and understanding how load-resisting systems

operate and general principles of building form and basic subsystems. This pattern permits

the design strategy described by Lin and Stotesbury (1988, p. 18):

The strategy is to emphasize only the more primary levels of functional


and aesthetic and economic requirements until a promising total system design
has emerged. In this way, consideration of the more detailed specifics of each
entity (the secondary-level issues) is postponed so that they may be influenced
by the total-system context. Such an approach enables the designer to identify
overall issues with clarity and to deal with the design of each spatial subsystem
in a hierarchic fashion.
But this is not to say that the designer should ignore the importance of
structural specifics at the formative stages of his spatial thinking. Indeed, if
the designer is experienced, he or she knows that the ultimate provision of
both overall and specific means for physical integrity is essential. But he or
she will need to have a suitable way of discriminating and then dealing with
only the overall issues of structural design at the formative stages of design
thinking. When this is not the case, it is not surprising that structural (and
other technical) concerns will be more or less overlooked at schematic levels.
124

Consequences and implementation

Agglomeration focuses the design around functional information, which describes how

different components work together. This type of grouping is useful in structural system

design when the interaction of all the components need to be analyzed together. Agglomer-

ation is also helpful for the creation of simplified mechanics models to capture the behavior

of a structure. The elements in a simple finite element model are the agglomerated compo-

nents of the structure.

Overlooking the connections between agglomerated components and critical details in-

side them are possible pitfalls to using this pattern in analysis. For example, looking mainly

at the function of a beam and column might cause a designer to overlook challenging de-

tails involved in the construction of the beam-column joint. Wisdom will dictate how to

tradeoff using agglomeration to simplify analysis and when to look at the real components

when the details become necessary.

Known uses

Performance groups. Performance groups are composed of several different pieces of

the structure, but are combined in a way that their behavior are treated as a single unit

(Mackie et al., 2008a). The groups represent parts of a structure that are repaired together

after an earthquake and whose structural performance can be characterized by a single

engineering demand parameter. The example in Sec. 4.3 defines performance groups and

their contents.
125

Load paths. Analyzing load paths within a building uses agglomeration. Load path

analysis checks for a complete path for loads to propagate through the structure down to

the foundation. Only the function of the structural components is important for checking

the load path, and the actual material used in the building components is often not essential

as long as components are properly designed.

Elements in mechanics models. Mechanics models often use some degree of agglomer-

ation in order to simplify the analysis. A reinforced concrete beam might be simply treated

as an elastic beam line element for some analysis. In this case, the behavior of individ-

ual rebars do not need to be considered, only understood that they exist and perform their

intended function.

But, whenever the individual behavior is needed, the agglomeration must be broken in

order to complete analysis. This may happen with non-linear behavior where a fiber model

is required to separate the material behavior of rebar and concrete, or if the section warps

and the assumptions of the Euler-Bernoulli beam theory become invalid.

5.1.4 Updating pattern

Intent

Use when an early initial guess needs to be updated after receiving additional design

information.
126

Also known as

Recompute, refine, "sharpen the pencil".

Assumed background

Foundational + Problem recognition and solving + Design + Risk and uncertainty.

Motivation

An initial guess might need to be updated once more information comes in. Design

information with less maturity will change after the designer determines more information.

Sometimes an initial estimate on a value is needed in cases where the material variables

of the object being designed are themselves variables in the problem solving method. In

the design of beams, the beam's self weight contributes to the load on the beam. The

estimated load is needed to estimate the beam size, then the beam size is needed to update

the estimated load. The updating pattern occurs when there is interdependency in a design

process (Steward, 1981; Browning, 2001).

Applicability

Use the Updating pattern:

1. When there are interdependencies among the design variables, and one variable has

to be assumed in order to get the problem solving algorithm started. Overcomes


127

Start with initial


values of variables » Use XQto determine •
Use y to update the
XKiXQ
values of y values of XTHX\

Complete analysis
using x\ and y\

Figure 5.1: Updating pattern.

the "chicken and egg" problem when beginning to make design choices based on

information from mechanics models.

2. The level of detail and precision increases during the stages of the problem solving

algorithm. A lower level of detail and precision can be used in the beginning, then

updated to a higher level later in the process.

3. When additional data will become available in the future, and you want to use that

data to influence decisions.

Structure

The general structure of the Updating pattern is illustrated in Figure 5.1. Initial values

of the quantities x ss XQ are assumed in order to compute other quantities y\, which are then

used to update the initial guess to a more accurate value x ~ x\.


128

Consequences and implementation

The Updating pattern is useful when it is not possible to continue analysis with abstrac-

tions. The abstract information needs to be instantiated into combinations of real material

and spatial information, closer to the information needed for construction. This allows the

designer to start with a rough approximation in order to get started on a design task. But,

this process will require design information to be revisited again, requiring some iteration

and possibly rework. Implementing this may be difficult if a high degree of precision is

needed in the initial guess and insufficient information is available to guide the selection,

similar to the problem of "starvation" noted by Terwiesch et al. (2002).

Known uses

Estimating dead load. When sizing structural elements that carry their own self-weight,

an initial value of the element's self-weight must be determined in order to determine the

loading on the element. Once the size of the element is chosen, then the actual self-weight

can be added to the original loading to check if the element still meets the design criteria.

Preliminary steel frame design. Lee and Goel (2001) updating moments after choos-

ing beam and column shapes. Before choosing beams and columns, design forces Fi are

estimated based on estimates of the fundamental period, yield drift, and a selected target

drift. These design forces are used to determine beam sizes using the AISC-LRFD method.

After the beam sizes are chosen, the design forces are updated F-m for the final selection of

beams and columns.


129

Structural health monitoring. Data collected from the actual performance of a building

can be used to update estimates of performance based on variables used in the design pro-

cess. This process is often referred to as structural health monitoring, where the condition

of the structure, "health", is monitored using sensors or human observations. The devel-

opment of sensors and methods for performing updating based on sensor data is a topic

of much current research given the recent innovations in sensor technology and wireless

communication (Lynch and Loh, 2006). Methods of using sensor data to update bridge

performance within the PEER PBEE framework have also been proposed by Wong et al.

(2006); Wong and Stojadinovic (2006).

5.1.5 Engineering Performance pattern

Intent

Measure structural design performance using the results of a mechanics model related

to the stress and strain, force and displacement of a structural model.

Also known as

Structural performance, engineering demand parameter (EDP), limit state, mechanics

model results.
130

Assumed background

Foundational + Mechanics + Materials science + Problem recognition and solving +

Design + Sustainability + Risk and uncertainty + Contemporary issues + Project manage-

ment + Communication + Public policy, and comprehension of performance-based earth-

quake engineering (Cornell and Krawinkler, 2000; Moehle et al., 2005).

Motivation

Engineering performance provides a way of transforming qualitative requirements of

building function into quantitative requirements that can be checked using engineering

tools. Engineering performance measures how well the materials in the structure are able to

perform their intended function to resist loads and provide serviceability. These measures

generally come from a mechanics-based model which gives information about stresses,

strains, forces, and displacements in the structure. Other analysis outcomes such as vi-

brations and aerodynamic effects are also important for characterization depending on the

unique aspects of each structure.

Applicability

This pattern is applicable when the performance requirements are based on the struc-

ture's intrinsic properties. This encompasses the physical behavior of the structure and

its response to loads. This pattern converts qualitative and descriptive requirements into a

quantitative form that can be evaluated using mechanics-based models and mathematical
131

methods.

This pattern is not applicable when the requirements are not fulfilled through the ma-

terial behavior of the structural components. In those cases, the descriptive requirements

cannot be translated into a quantitative form, and therefore a different type of performance

measure must be used to evaluate.

Structure

The Engineering Performance pattern involves most of the quantities related to struc-

tural engineering analysis. This includes such things as maximum material capacities like

plastic moment capacity Mp, global displacement measurements like drift ratios, and local

displacement measurements like maximum element deflections.

The pattern works by taking some functional requirement and translating that require-

ment into quantitative requirements that can be checked with an analysis method. For

example, a residential balcony that has cantilever beams needs avoid damage to finishes

and remain level enough for people to use it without feeling unsafe. This might translate

into a maximum deflection at the balcony end, which provides a minimum required beam

stiffness. The deflection and stiffness are the Engineering Performance parameters that can

be checked in a mechanics model in order to satisfy the requirements.


132

Consequences and implementation

Implementation depends on the type of performance measure considered. The concept

of engineering performance typically refers to quantities that can be measured and repre-

sented in a mechanics model of a structure.

Known uses

Almost all engineering analysis procedures use this pattern tacitly by aiming to produce

results for evaluation and comparison to a standard of acceptable performance.

The PEER PBEE methodology links specifically links engineering performance to

functional requirements by using the total probability theorem. This is accomplished by

defining functions that related earthquake hazard, structural performance, and the possible

consequences (damage, downtime, dollar cost) to the parameters that drive decisions about

performance requirements. The engineering demand parameter (EDP) in this methodology

is an explicit example of the Engineering Performance pattern.

5.1.6 Social Performance pattern

Intent

Measure structural design performance using information about social conditions.

Also known as

Decision variables, non-engineering factors, "soft" issues.


133

Assumed background

Foundational + Mechanics + Materials science + Problem recognition and solving +

Design + Sustainability + Risk and uncertainty + Contemporary issues + Project manage-

ment + Communication + Public policy + Teamwork, and comprehension of performance-

based earthquake engineering (Cornell and Krawinkler, 2000; Moehle et al., 2005).

Motivation

Mechanics models alone are insufficient to characterize all performance factors for de-

sign. Factors such as cost, construction time, construction phasing, material availability,

and estimated post-earthquake repair cost play a role in design decisions, but cannot be

estimated based on mechanics-based structural analysis models alone. They involve fac-

tors that are outside the usual realm of mechanics-based analysis. These situations call for

Social Performance criteria that involve this additional information beyond mechanics.

Applicability

This pattern is applicable when the performance requirements are based on external

factors other than the structure's intrinsic properties in as-built state. This includes factors

involving the construction process, cost, time, and issues related to labor and scheduling.

This pattern analyzes performance measures that are not based on mechanics models, but

rather on issues of a social nature that emphasize process information over material and

spatial design information.


134

Design Additional Social


social or market Performance
information information measure

Figure 5.2: Social performance.

Structure

The structure of this pattern generally includes design information plus additional social

information and a methodology for combining the two types of information (Figure 5.2).

The additional information is often tacit information based on experience, or particular

industry-specific data. For example, for estimating the cost of a reinforced concrete shear

walls, the additional information could include the prices of rebar at different sizes and

grades, the cost of concrete, the material availability of aggregates, and the optimal dimen-

sions for formwork erection.

Consequences and implementation

Including Social Performance into a design process is difficult because of the need for

additional information that is external to the design itself. The use of tacit, informal infor-

mation is also a challenge to implementing methods for incorporating Social Performance

into design processes. Factors such as labor productivity are often estimated in a qualita-

tive manner which makes it more difficult to convert to a quantitative value. Much of this

information is not traditionally part of the structural engineering design process. There-

fore, implementing this pattern might be very difficult because it forces collaboration and
135

communication in various ways that may only be available in a collaborative environment.

Applying this pattern early in the design process will allow much better decision making

because of greater knowledge about downstream conditions and the largest constraints on

executing the project. Social Performance measures are important for determining project

feasibility and profitability. But, Social performance measures are also more difficult to

evaluate when the design is still immature because there may be insufficient precision to

specify particular products and materials or specific processes for construction. This dif-

ficulty can be overcome using case study data from previous projects and general rules

of thumb, such as the suggested rules for limits on concrete reinforcement ratio p to aid

constructibility in ACI 315.

Known uses

The PEER PBEE method uses "decision variables" (DVs) in order to express Social

Performance (Cornell and Krawinkler, 2000). Mackie et al. (2008 a) describe an applica-

tion of the PEER PBEE framework for computing repair times and repair cost ratios for

post-earthquake damage of reinforced concrete bridges. The design information for the

bridges are based on Ketchum et al. (2004), and OpenSees-based mechanics models pro-

vide engineering performance data. The additional information needed to compute social

performance was obtained through multiple interviews with Caltrans bridge engineers and

cost estimators. The mechanics-based and social-based information are combined using

the local-linear repair cost and time methodology (Mackie et al., 2008c).
136

Parrish et al. (2008b) uses this pattern in describing the notion of "value propositions",

and illustrating with labor production rates for different reinforcing steel sizes and bending

types. The additional social-based information is obtained from interviews with rebar fab-

ricators and rebar placers. This information is combined with material and spatial design

information for rebar size and bending type, and with functional information describing the

type of structural element (e.g. wall, column, beam).

5.2 Patterns for decision making

5.2.1 Pick Any pattern

Intent

Choose any option from a set of options, then confirm whether it works.

Also known as

Guess and check, trial and error.

Assumed background

Foundational + Mechanics + Problem recognition and solving + Design.


137

Motivation

With a mechanics model that does not contain abstractions which are able to character-

ize entire sets of different real design options, it is often necessary to pick any one candidate

solution and check whether it satisfies the design criteria.

This pattern is also useful in situations where the mathematics is too difficult to carry

forward a symbolic solution, and numbers are needed. The Pick Any pattern allows the

selection of values for the design variables, allowing the design process to move forward.

Applicability

The Pick Any pattern is applicable when there is little information available to discern

any differentiating benefit between a set of options. If all the options are the same for all

measures of performance, then it is okay to pick any option.

Often, this pattern is misused when the options have very different performance. Many

decisions in common structural engineering problem solving algorithms implicity use the

Pick Any pattern. But, the only performance criteria evaluated for selection involve struc-

tural performance based on abstract variables and a mechanics model. These applications

rarely consider the downstream implications on the rest of the delivery supply chain. For

example, different sizes or grades of rebar might be unavailable, or the selection of a steel

shape might cause clashes with duct work.


138

Structure

The Pick Any pattern requires abstract design information and basic performance cri-

teria that define the design choice to be made (Figure 5.3). The choice is usually to define

more mature or more precise material and spatial information. For example, the choice

might be for the number and size of rebars in a concrete beam given a requirement of

As > 2.00in.2. The criteria for design involve the specification of an Aggregation of steel,

limitations on spacing and dimensions, a required cover, and beam dimensions b and d.

The Pick Any pattern is used to choose any particular instantiation of rebar that fulfills

those requirements.

Consequences and implementation

The Pick Any pattern will lead to a solution if the choice is good and satisfies all the

required performance criteria. If the choice is bad, then it will require iteration to determine

a workable design.

This pattern tends to overlook other options that might be far better than the initial

choice. This pattern also tends to eliminate options which may be better than the initial

option. It does not allow consideration of the downstream consequences.

Implementation is simple, because an arbitrary choice can be made among options that

already satisfy the acceptance criteria. In reality, choices are not completely arbitrary,

because tacit preferences inform a designer's selection.


139

Abstract design
requirements

Determine
performance
criteria for design

Pick Any

Evaluate
performance of
design choice

Satisfies \ N 0
performance
criteria?

Yes

Proceed
with design

Figure 5.3: Pick Any pattern.


140

Known uses

The pick any pattern is used in reinforced concrete design procedures for choosing rebar

layout (MacGregor and Wight, 2005). After the mechanics model constraints on rebar are

selected then any combination satisfying the constraints is selected and used in the rest of

the design process.

5.2.2 Minimum Weight pattern

Intent

Choose the design option that provides the least material weight.

Also known as

Lightest section.

Assumed background

Foundational + Mechanics + Problem recognition and solving + Design.

Motivation

Steel and concrete are priced by weight. So, reducing weight attempts to minimize cost.

But, this does not work when overall cost is controlled by other factors such as installation

labor.
141

A lighter structure might more efficient due to having less material, reduced gravity

loads, and reduced seismic mass. Choosing structural elements that have the least weight

will fulfill these goals as long as the choices meet all the other design criteria. This pattern

most commonly occurs in the selection of steel shapes because the AISC manual provides

design aids for choosing minimum weight and names steel shapes by their weight per foot.

Applicability

The Minimum Weight pattern can be applied in situations where the weights of each

option are easily computed and have a direct influence on the design decision. This would

be the most applicable in the selection of structural members for framing or support. In the

selection of rebar layouts, selection by weight is probably not useful because all choices

will have approximately the same weight since they must all provide a similar area of steel.

For rebar, congestion often controls cost more than material weight.

Using this pattern is better than the Pick Any pattern because it provides a more intel-

ligent basis for determining the material and spatial information needed to complete the

design.

Structure

The Minimum Weight pattern picks the design option that has the least weight and

fulfills all other structural criteria. This works best when there is a catalog from which

to choose, or some clearly understood set of options. When the design options are more
142

/ —, Table 3-2 (continued)


£ W Shapes /> = so ksi
X Selection by Zx

V«» m* fif "t/Or W«


Shape ** v\
kip-ft
%Mpx
kip-ft kip-ft kip-ft kips kips ^ L, '* kips kips

in.3 ASD H LRH'I" ASD LRFD ASD LRFD ft ft in. 4 ASD LRFD

W40x149v 598 1490 2240 898 1350 38.6 58.U 8.09 23.5 9800 432 650
W36x150 581 1450 2 , 8U 880 !•.'(: 34.5 M 8 8.72 25.2 9040 448 672
W27x178 570 1420 2140 882 21.7 .,.'7 11.5 36.3 7020 403 605
W33x152 559 1390 2100 851 • ; < • 32.0 4fl. 8.72 25.7 8160 425 638
W24x192 559 1390 858 • ' % . 18.7 28 U 10.8 39.6 6260 413 619
W18x234" 549 1370 814 12?C 10.8 :••/ 10.1 61.5 4900 489 733
W14x283h 542 1350 ..! '1 802 1200 5.53 14.7 114 3840 432 648
W12x305ft 537 1340 2011) 760 •1.'.l) 4 66 7 00 12.1 137 3550 530 796
W21x201 530 1320 1 VJl. 805 121u 14.6 .'• >> 10.7 46.1 5310 419 629
W27X161 515 1280 800 ".cU'J 20.8 31 3 11,4 34.7 6310 364 546
' 'M.'j

W33x141 514 1280 782 1180 30.4 45.8 8.58 25.0 7450 403 604
1930
W24x176 511 1270 786 1180 18.3 10.7 37.4 5680 379 568

Figure 5.4: Minimum weight design aid sorts steel W-shapes by weight and performance
(AISC Manual, Sec. 3).

freeform, more information might be required.

This pattern is best illustrated using the selection of a steel beam from the catalog of

standard AISC W-shapes. The AISC Manual, Sec. 3 contains tables of steel shapes sorted

by required plastic modulus Zx and arranged in subgroups by weight. The least weight

section is positioned at the top of each subgroup and indicated by bold text.

Consequences and implementation

Implementation of this pattern is typically straightforward. Determining total weight is

usually a simple computation that can be done using basic tools. Design aids for choos-

ing by minimum weight are readily available for structural steel. The AISC Manual pro-
143

vides tables for choosing least weight W-sections by minimum required engineering per-

formance, measured by either plastic strength Zx (Figure 5.4) or stiffness /.

When applied haphazardly, this pattern could lead to excessive variations in the total

design. For example, attempting to minimize the weight of every single steel beam in a

framing system would create a large variety of different shapes, which would be difficult

to procure, transport, and install on the job site.

This pattern is sometimes applied as a substitute for the Minimum Cost pattern, but this

is not preferred. The pattern may lead to economical designs when the cost of materials

dominates over the cost of labor, transportation, and other factors. However, simply assum-

ing that minimum weight will lead to low cost is short sighted and can lead to unintended

consequences.

Known uses

Selecting steel shapes by minimum weight is a commonly used algorithm for choosing

among the catalog of AISC shapes (Segui, 2007).

5.2.3 Minimum Cost pattern

Intent

Choose a design option that will produce the lowest overall cost.
144

Assumed background

Foundational + Mechanics + Problem recognition and solving + Design + Project man-

agement + Public policy + Business + Teamwork, and comprehension of production theory

(Koskela and Vrijhoef, 2001).

Motivation

Money is one of the most important resources on any project, and the motivation for

reducing overall cost is obvious. Yet, there are many different methods for estimating cost

and the method of arriving at minimum cost is not always clear.

Applicability

The Minimum Cost pattern can be used to choose among a set of structurally acceptable

design options that differ by their cost. The pattern is best used when the other differenti-

ating factors in the selection are less important than cost.

Structure

The Minimum Cost pattern simply chooses the design option with the lowest cost value.

Consequences and implementation

Implementation is difficult, because there are many different methods of estimating

project and component cost. Simple cost models involve multiplying quantities by unit
145

costs. Slightly more advanced models use a unit cost function, where the unit cost is

usually high for small quantities and lower for large quantities. Mackie et al. (2008a) use

a fixed unit cost for estimating bridge repair costs, and Naeim et al. (2007) use a variable

unit cost in the PACT software for building seismic damage cost estimation.

Depending on how cost is evaluated, many different decisions might be made using

this pattern. If cost is evaluated based on up-front component costs, then the pitfall would

be choosing an option that might actually be more expensive in the long run. If cost is

evaluated based on life-cycle costs, then the initial cost might be higher, and the estimate

will also depend on risk models used.

A good approach would be to blend consideration of costs during design, construction,

structure operation, and structure functional use. The costs for design and construction

are generally an order of magnitude less than operation and use costs. The estimates for

the cost ratio between construction, operation, and business use vary from 1:5:200 (Saxon,

2005), to a perhaps more realistic figure of 1:0.4:12 (Hughes et al., 2004). Design costs on

most projects are approximately 2-5% of construction costs. So, there is a clear incentive

to consider the total project cost instead of only design fees and construction expenses in

making decisions.

Making decisions purely on cost alone often neglects other factors that are important

for the structure's performance. For example, other measures like safety are not always

easily evaluated by money costs. Speed may be more critical for structure use than cost, as

in the case of many emergency infrastructure repairs for bridges and highways.
146

Known uses

Uses of this pattern are difficult to find in most design methods, because of the difficulty

involved in computing the costs of design options. Cost tends to be evaluated after decisions

are made instead of before initial decisions. The concept of value propositions attempt to

overcome this barrier by providing tools for estimating costs earlier in the design process

(Parrish et al., 2008b).

Mackie et al. (2007) proposes a method for choosing among different bridge column

designs using total highway bridge cost including initial construction and post-earthquake

repairs.

5.2.4 Optimization pattern

Intent

Choose a design option based on some optimal value of multiple performance measures

involving tradeoffs between them.

Assumed background

Foundational + Mechanics + Problem recognition and solving + Design + Teamwork.

Motivation

Many decisions have neither a single solution, nor a solution based on one single mea-

sure of performance. For these cases where tradeoffs need to be made among multiple En-
147

gineering Performance and Social Performance measures, some method needs to be used

to make rational choices among sets of design options. The Optimization pattern describes

methods for making these choices where there may not be a single clear option, or a single

decision variable. For example, the selection of a structural system might involve several

different factors ranging from engineering criteria like drifts, accelerations to social criteria

like cost, labor rates, and material availability.

Applicability

This pattern is applicable in situations where there is no single "best" solution, but all

options involve a series of tradeoffs. This occurs most commonly the earlier phases of

design, and less so in the more detailed stages of design when most of the functional deci-

sions have already been made. For example, the selection criteria for choosing a structural

system are not as clear cut as compared to choosing a beam size or shape.

Structure

This pattern involves the evaluation of multiple performance objectives and comparing

them to one another. This is typically accomplished in one of two ways: converting the

design problem into a computable form to be optimized using numerical tools, or using a

qualitative method of comparing the differences and advantages within a set options. The

selection of the best design option will attempt to maximize the value of each performance

objective with a higher or lower weight assigned to each objective depending on their rela-
148

tive importance.

The Minimum Weight and Minimum Cost patterns are examples of Optimization for a

single variable.

Consequences and implementation

Implementing this pattern is challenging because of the difficulty in assigning weights

to each performance objective. Assigning relative importance to each objective is a subjec-

tive process, and the results of the weighting can vary from person to person. Evaluating

multiple options requires a clear understanding of the functional design requirements so

that options are considered in an equivalent manner.

Numerical methods are more suited to fulfilling multiple Engineering Performance ob-

jectives, usually related to a mechanics model. Qualitative procedures are usually better at

comparing Social Performance objectives or a combination of both.

Known uses

At the element level, Aschheim et al. (2008) presents an optimization scheme for de-

signing reinforcement positions for rectangular reinforced concrete beams, columns, and

walls under biaxial bending and axial load. This scheme uses a conjugate gradient search

method and defines a number of spatial variables related to the position and size of reinforc-

ing inside the rectangular section. This method optimizes for Engineering Performance and

uses an Aggregation to represent individual rebars by considering them to be "distributed


149

uniformly, or smeared, into a single, continuous plate that is inset from the edge of the

section." Additional design is still needed to choose bar layouts and bar sizes.

At the system level, Austin et al. (1987a,b) use the DELIGHT system (Nye and Tits,

1986) to optimize steel frame design for three criteria: minimum volume (weight), mini-

mum story drift, and maximum energy dissipation. A single variable, moment of inertia

/, characterized the section properties of the frame elements, and section area A for truss

elements. This method produces a well-proportioned frame system, but still requires addi-

tional design selection to choose actual steel shapes to implement the abstract design.

For qualitative comparisons, the Choosing by Advantages method (Suhr, 1999) which

assigns weights to different decision criteria and scores to each option with the advantage.

The Analytic Hierarchy Process (Saaty, 1990) uses pairwise comparisons in order to weight

and score decision criteria. The option with the highest score is considered the optimal

choice.

5.2.5 Prescription pattern

Intent

Follow an established rule for making a design decision, or confirming that a candidate

design choice is valid.

Also known as

Code-based design, "cookie cutter" procedure.


150

Assumed background

Foundational + Materials science + Mechanics + Experiments + Problem recognition

and solving + Design, and specialty knowledge for applying seismic design criteria.

Motivation

Sometimes there is only one acceptable design decision for a particular decision unit

because of regulatory constraints, or the lack of other acceptable options. These kinds of

decisions are based on an accepted method or a list of acceptable choices that are con-

strained by factors outside of generalized performance measures.

Prescriptions also solve problems by providing maximum or minimum ranges of quan-

tities to be used in design. This is a good way of encapsulating general constraints based

on prior experience in industry and consensus of engineering judgment.

In the design of steel moment frames, there is a prescriptive method in FEMA 350

for the design of the moment-resisting connections. This method promotes the use of

prequalified connections with specific details, because they have been tested and proven

to achieve sufficient ductility. In this case, no non-tested connections should be used in

order to prevent fracture. The allowable design choices are limited because of the limited

amount of available laboratory testing data. These constraints are due more to external

factors than ones that are intrinsic to the design itself.


151

Applicability

This pattern addresses situations where Engineering Performance is difficult to quantify,

as in the case with stress concentrations in steel moment-resisting connections. Prescrip-

tions are also a common method of encapsulating engineering judgment and experience

from past projects. This is useful when the solutions to design problems are generally

understood and there is a good reason for limiting the range of possible solutions.

Structure

The general use of this pattern involves following a step-by-step procedure that almost

always leads to an acceptable solution. The use could also be simply to adopt existing

design information.

Consequences and implementation

The prescription pattern fulfills its objectives by providing a simple procedure that is

easy to follow and suited for most conditions. But, this simplicity is also its greatest short-

coming. Prescriptions might not be applicable in all situations, and certain edge cases

might make the prescription invalid or, in the worst case, dangerous for design.

Many prescriptions for reinforced concrete design are invalid for high strength or light

weight concrete, and need to be specifically modified or not used at all. For example, the

common approximation for concrete elastic modulus (Ec — 57,000^/^T) is not valid for

light weight concrete, in which case (Ec = w,!.'533-y/^) must be used, and even that is
152

only valid for concrete weights of 90-155 lb/ft3. Another example is the minimum steel

requirement for flexural elements, which needs to be altered to accommodate high strength

concrete with f'c > 5,000 psi.

Known uses

ACI 318, Sec. 10.5 contains rules for flexural elements that specify a maximum and

minimum amount of steel in a cross section, As^m = 3^jJ-bwd > 200bwd/fy. This min-

imum amount of steel is intended to encapsulate good judgment in order to prevent the

dangerous condition of sudden failure when the uncracked moment capacity is greater than

the cracked moment capacity. The designer can prevent this type of failure by simply fol-

lowing the prescription, instead of needing to check this condition specifically based on a

mechanics model and definition of specific engineering performance criteria.

FEMA 350 gives a prescriptive design method for prequalified steel moment-resisting

connections. The step-by-step procedures will specify all the information needed to detail

the connection for fabrication. Staying within the bounds of these prescriptions allows

the designer to complete the task, knowing that the connections will not fracture based on

confirmation by laboratory tests.


153

5.2.6 Set Exploration pattern

Intent

Investigate the consequences of multiple different options before continuing with a sin-

gle option.

Also known as

Set-based design, set-based iteration.

Assumed background

Foundational + Mechanics + Problem recognition and solving + Design + Commu-

nication + Teamwork, knowledge of set-based design concepts for structural engineering

applications (Parrish et al., 2007), and comprehension of structural system behavior (Lin

and Stotesbury, 1988).

Motivation

This decision making pattern considers several design options before making a deci-

sion. Instead of choosing a single option first then evaluating it, the set exploration uses

evaluation first, then the selection of the best option.

Point-based design involves selecting a single feasible design option that satisfies En-

gineering Performance (5.1.5) criteria at each step in the design process and then refining

that single design (or point) while developing more details during the design process. This
154

single design is then re-worked until a solution is found that is feasible. The first design

thus selected by a structural engineer tends to be uninformed by the expertise of rebar fab-

ricators, placers, and concrete suppliers who will perform the actual rebar detailing, rebar

placement, and concrete placement.

In contrast to point-based design, set-based design focuses on keeping the design space

as open as possible for as long as possible. A key to the success of set-based design is

knowledge sharing; whenever the feasible design space is reduced, the reason for eliminat-

ing any part of it needs to be documented and made accessible to all relevant stakehold-

ers. Preserving the maximum number of feasible designs as long as possible reduces the

likelihood that rework will be necessary and allows consideration of Social Performance

objectives (5.1.6) by creating a process for all project participants to leverage their unique,

individual, and team-based expertise to make the project successful.

Applicability

Set Exploration can be applied in situations with low ambiguity in the decision making

process, and is usually desirable when the costs of exploring options is less than postpon-

ing early commitment, as in point-based design (Terwiesch et al., 2002). It addresses the

shortcoming of other decision making patterns by allowing several options to be carried

forward instead of having to backtrack and perform rework later in the process.

Set Exploration works well in cases where it is possible to explore different design

alternatives. When Engineering Performance criteria are unable to distinguish between


155

(18")
I 47cm-

iwTv *8b" w ^ WPi * P v J O |


^ ^ ^ ^ ^ I V u ^ T 3#7bars VlU^T \ ^ ) r *H&T »fefe^K

^ ^'' * ' ' * * a' w* ' ' ' ' * * * I LJ- 1 -^


Beam I Beam 2 Beam 3 Beam 4 Beam 5 Beam 6 .

Hard Constraints:
ACI Code

Soft Constraints:
Design Preferences

Column Column 2 Column 3

Figure 5.5: Set Exploration involves evaluating multiple options for design preferences in
addition to checking structural engineering criteria (Parrish et al., 2007).

multiple options, set exploration provides an opportunity to check downstream design cri-

teria before making a commitment. Set Exploration does not make sense when evaluating

multiple options is too difficult because of unclear Engineering Performance criteria.

Structure

The structure of set-based design involves enumerating several options that all meet the

required Engineering Performance criteria and then evaluating them for other selection cri-

teria. Once the options are evaluated, the optimal choice is made. An example illustrating

the selection of reinforcing steel for a beam-column joint is shown in Figure 5.5 (Parrish

et al., 2007).
156

Consequences and implementation

Exploring the set of options accomplishes a better design because it allows better con-

sideration of downstream factors for the construction process. Using set exploration allows

better evaluation of Social Performance criteria by developing multiple options instead of

just one. This provides more information for tradeoffs and provides more information dur-

ing the early design phases of a project.

Set exploration requires additional up-front work in order to develop the set of options,

and additional analysis needed to evaluate the multiple options. The implementation of set-

based design is difficult because most existing design procedures depend on point-based

decision making patterns.

Known uses

Parrish et al. (2007) explains the application of set-based design strategy for structural

engineering in reinforced concrete. Set-based design is described by Ward et al. (1995)

in the context of set-based concurrent engineering. Set-based design has also been doc-

umented as a key component of Toyota's success in the automotive industry, where it is

used internally to evaluate design alternatives, and also by its suppliers (Liker et al., 1996).

The basic principles behind Toyota's practice of set-based concurrent engineering involve

mapping the design space, integrating by intersection, and establishing feasibility before

commitment (Sobek et al., 1999).

The viscous damping wall structural system for the Cathedral Hill Hospital in San Fran-
157

cisco, California was chosen based on a set-based evaluation of multiple different options

(Parrish et al., 2008a). The driving factors for choosing the system involved constructibility,

cost, seismic performance, and site conditions.

5.3 Metadata for design process and design product

The patterns described above can be used to define a metalanguage to describe the tacit

knowledge and heuristics of structural engineers, and the information about the design itself

using the definitions from Chapter 3. This language can carry data and information over

from one design step to the next, beyond what is usually transferred. This metalanguage

defines the metadata for both the design process and the design product. This complements

the language developed to store structural-state data (Stojadinovic et al., 2004; Wong and

Stojadinovic, 2004) from sensors and inspector observations. The new metalanguage and

the state data language combined allow a way to capture information about the structure

through its entire life cycle from initial concept to ongoing operation with greater depth

than previously possible.

The metalanguage consists of terms that describe the characteristics of structural de-

sign information and indicate how things are related to each other. It defines specifically

the level of intentional precision (detail) by providing a slot for the expression of the ab-

stractions that led the engineer to design those pieces of information. When the set of ab-

stractions are given, and the problem solving method, and choice rationalization are made
158

explicit, then the level of intentional precision is more clear.

Describing the process for arriving at a conclusion may be just as important as the

conclusion itself. Understanding why decisions are made provides the necessary context to

the decision itself. What might seem like an arbitrary choice, might be understood as the

best possible decision once the context is known. For example, a material shortage may

have driven the decision to choose a steel shape that otherwise would be considered less

desirable.

The metalanguage also reveals specific details that are not considered in this particular

selection of information. This can be done by specifying the conceptual scale and level of

grouping (agglomeration and aggregation) that the information depends upon.

The concepts behind the metalanguage are not at all new. These things have been part

of engineering design. What is new is the formalism and explicitly-named categories of

information. This formalism allows computer software to interact with design information

in ways that match the structure of how structural design really happens.

Another goal of the metalanguage should be to make it minimally burdensome for an

engineer to think in terms of it and to express knowledge either through a computer or

through a conversation. The metalanguage does not perform any thinking, but rather simply

provides slots to ensure that critical information is not inadvertently left out of the picture.

This is similar to the way template forms are. Fields in the forms indicate the information

necessary to support a process. Empty fields on the forms indicate incomplete information,

but the person filling out the form is not obligated to fill out all the information. Some
159

of it might be not applicable, or the person might simply choose not to provide all the

information. Depending on the process, fields with "N/A" inside, or blank fields might be

acceptable.

The metalanguage provides a system for capturing the tacit information that goes into

the design process. The examples that follow illustrate ways that the metalanguage can

capture decisions which must be made during the design process. In a sense, the metalan-

guage provides a mechanism for built-in commenting and documentation during a design

process. It allows the designer to leave "breadcrumbs" that allow him to retrace steps and

identify places where the solution can be revisited.

5.4 Examples of design and decision patterns

Using the patterns identifies the locations where cost considerations can be inserted

along with existing textbook design methods. Where design decisions are made, the addi-

tional selection criteria need to be evaluated and considered during the selection process.

Structural design requires the manipulation and thought about abstract entities, based on

mechanics theory, in order to check physical designs for suitability in their environmental

conditions (e.g. loads). These abstract entities are based on physical objects, but only on a

specialized set of their physical properties. The description needed of the physical is special

to the formation of abstractions for engineering computation and design integration, and not

al all the same as needed for construction. For example, the field iron worker would not be
160

concerned about the value of p^ or some quantity such as the ductility p.

Identifying the needs of the designer apart from the needs of physical space gives rise to

the connection between the notion of design abstractions and other related concepts. These

abstract design quantities are needed to describe the relationship between "performance"

and the physical nature of the building or structure. Theories of mechanics, analysis, and

design are only able to accomplish operations on the parameters that it employs. Detail

beyond the amount capable of handling by the theory of mechanics used is unable to be

accounted for, and really unnecessary for the theory to be employed.

Mechanics models operate at a certain minimum and maximum level of detail. The

level of detail can be chosen, and generally it is preferred to keep model at the simplest

level of detail needed to answer the design or analysis question at hand. The level of detail

needed by the model's abstractions defines the set space for design. As long as the abstract

criteria are satisfied in the model, then any detail can work properly in the design.

The model to be used depends on the decision that needs to be made. The decision unit

is a question of performance that must be answered by specifying information about the

building. The type of decision being made will affect the information that is required to

know how to answer, and the form of the answer. Some decision units will require more

information on the functional level, while some will require more information on the detail

level. It might require the resolution of ambiguities.

The problem solving method leaves room for set-based exploration, or new problem

solving methods can be developed in order to enable set-based design. Existing problem
161

solving methods based on mechanics theory already permit much freedom in set explo-

ration.

5.4.1 Scope of examples

The scope of these examples is limited to a class of low ambiguity structural engi-

neering design situations, corresponding to high uncertainty/low ambiguity on the uncer-

tainty/ambiguity matrix in Schrader et al. (1993). These examples represent cases where

the variables and the functional relationships between the variables are known, but the val-

ues of the variables are unknown. High ambiguity structural engineering decision situations

are characterized by an unknown problem-solving algorithm, unknown functional relation-

ships between the variables, or both, and are outside of the scope of these examples. These

examples are suitable for solution by either iterative problem solving methods, or set-based

strategies, because they are unambiguous and lack only maturity (Terwiesch et al., 2002;

Giletal.,2008) 1 .

5.4.2 Reinforced concrete beam design

Decision unit

What reinforcing steel is needed for a rectangular concrete beam with known load-

ing?

'Terwiesch et al. (2002) refers to maturity as "uncertainty", a concept discussed in Chapter 3.


162

Problem solving algorithm

Reinforced concrete design using LRFD, MacGregor and Wight (2005, p. 148-155)

Lesson

The problem solving algorithm provides a method for checking a solution but not

for choosing among options. Therefore the Pick Any decision pattern is used for the

selection of bar layout.

MacGregor and Wight (2005) suggests one design procedure for beams having known

dimensions b and h. Table 5.1 presents the steps implied by this this design procedure. The

reinforced concrete beam is designed using designed using LRFD, satisfying the design

equation

<$>Mn>Mu (5.1)

Supposing that the overall height h and width b of the rectangular beam are known, then

the design example requires the following detail and prerequisite abstractions. No other

parameters are included in the mechanics design models used for this design decision.

1. Mu: factored moment

2. d: effective depth

3. As,ASjmin: cross-sectional area of rebar

4. fy: steel yield stress

5. f'c: concrete compressive strength

6. <j>: resistance factor


Table 5.1: Design procedure for known b and h.

# Step description Patterns applied

1 Estimate the factored moment, Mu. Eng. Performance (5.1.5)

2 Compute the effective depth, d. Abstraction (5.1.1)

3 Compute the area of reinforcement, As. Aggregation (5.1.2)

4 Try a specific configuration for As. Pick Any (5.2.1)

5 Check whether As > A,s-,min. Prescription (5.2.5)

6 Check whether the section is tension-controlled. Eng. Performance (5.1.5)

7 Recompute Mn and §Mn. Updating (5.1.4),

Eng. Performance (5.1.5)

8 Recheck the area of steel required. Updating (5.1.4),

Eng. Performance (5.1.5)


164

Here, we choose to work within a set of details needed to compute the value of Mn.

Another set of details could have been used to generate the same abstraction. Thus there is

some design flexibility to choose how open the design will be at the point where abstract

design targets are defined. Recognizing these decision implicit in the selection of theory

would avoid fixation.

Using the example in MacGregor and Wight (2005, p. 148-151) we have the following

design parameters for reinforcement selection: fy = 60,000 ksi, f'c = 3,000 ksi, b = 24 in.,

h = 24 in., and Mu = 541 kip-ft.

Step 1: Estimate factored moment demand, Mu.

The factored moment demand is estimated based on superimposed service dead loads

and live loads. In the case of known b and h, the dead load can be estimated using an

average unit weight of reinforced concrete. This estimate would need to be updated for a

design case with unknown b and h. The example uses a value of Mu = 493 kip-ft.

Step 2: Compute the effective depth, d.

The effective depth d, the distance from the top of the beam to the centroid of the

reinforcing steel, can only be estimated because the dimensions of the rebar have not yet

been chosen. The example makes a choice to design for a single layer of rebar, which

allows for a rough estimate of

drah-2.5" = 21.5" (5.2)


165

Step 3: Compute the area of reinforcement, As.

Moment equilibrium is used to estimate the required area of reinforcement As. For this

computation, the distance jd between the tension and compression resultants is needed. A

decision has to be made to estimate j PS 0.875. The required area of steel is then

M
A, = — ^ - = 5.83in.2 (5.3)

Step 4: Try a specific configuration for As.

The example enumerates three possible choices, and uses the Pick Any pattern and

selects the second option:

1. Ten No. 7 bars

2. Six No. 9 bars

3. Four No. 9 bars + Three No. 8 bars

At this point, the mechanics model and design procedure is not sharp enough to discern

the consequences of choosing #7 bar, #8 bar, or #9 bar. Choices like this have little impact

on engineering performance and are not reflected in the model, but could have significant

impact on social performance measures. Other information such as rules of thumb, value

propositions Parrish et al. (2008b), or rebar fabricator comments are needed to make this

decision.

At this stage in the design process, before the selection of an actual rebar configuration,

the communication of preliminary information must be done in a way that does not imply
166

a level of spatial or material detail greater than what is actually present. The functional and

abstract detail is complete to the extent allowed by the problem solving algorithm.

Step 5: Check whether As > As m i n .

The selected value for As must be greater than the minimum steel area A.vmjn given by

the ACI code

3
V / L ^ 200bwd
As,min = -^bwd > — ^ - = 1.72 i n / (5.4)

Step 6: Check whether section is tension controlled.

The selection of As must also satisfy the tension controlled criteria fs — fy, which places

an upper limit on reinforcement. One method of checking this is to compute pt,, the rein-

forcement ratio at balanced failure where concrete crushes and steel yields at the same time.

This ratio can be computed using

For typical values of ecu = 0.003 and Es = 29,000ksi the expression can be simplified

with fy in units of ksi to

085p,/; / 87,000 \
pt = <5 6)
-^ll77JooTAJ '
This can also be expressed as the depth of the equivalent rectangular stress block in a

rectangular beam at the balanced condition


167

ab ( 87.000 \
= pl (5J)
7 U,ooo+/J
Another determination of the tension controlled limit is the ratio between the depth a

and the distance dt from top of beam to extreme tension layer

^ = 0.3750! (5.8)

So, with these two constraints the example gives § < ^ = 0.503 and | < ?f- = 0.319.

This indicates that the section is tension-controlled for the choice of As.

Furthermore, the ACI code limits p < 0.75p/,, and a common rule of thumb says that

rebar is easier to place when p < 0.50p^.

Step 7: Compute Mn and <|>Afn.

The final rebar configuration must be checked to confirm that §Mn > Mu using the

tension-controlled reduction factor value § = 0.90. These constraints provide enough in-

formation for the selection of bar sizes and bar count to answer what rebar is needed for

this reinforced concrete beam.

Step 8: Check area of steel required. A final evaluation is needed to verify whether the

final design meets all the required performance criteria. Only a check on general behavior

is needed, because the minor changes in yield displacement associated with final detailing

are usually not of significance (Aschheim, 2002).


168

5.4.3 Preliminary frame design

Decision unit

What are the approximate sizes of members needed for a particular building config-

uration with a steel moment-resisting frame?

Problem solving algorithm

Target drift and yield mechanism (Lee and Goel, 2001).

Lesson

The frame design method uses the Updating pattern to improve initial estimates of

beam and column moment demand. This type of strategy arrives at a workable so-

lution, but might miss a better solution. The Minimum Weight pattern is implied by

method for choosing steel shapes.

The steps in the target drift and yield mechanism method (Lee and Goel, 2001, p. 168)

are outlined in Table 5.2.

Table 5.2: Target drift and yield mechanism method.

# Step description Patterns applied

1 Setup—

la Estimate fundamental period Eng. Performance (5.1.5)

lb Estimate yield drift Eng. Performance (5.1.5)

lc Select target drift Eng. Performance (5.1.5)


169

Table 5.2: (continued)

# Step description Patterns applied

Id Calculate y Abstraction (5.1.1)

le Calculate p\-, A,, Abstraction (5.1.1)

2 Design forces—

2a Calculate design base shear, V Eng. Performance (5.1.5)

2b Design forces, Fi Abstraction (5.1.1)

3 Beam design—

3a Select Mpc Abstraction (5.1.1)

3b Calculate required beam strengths Abstraction (5.1.1)

3c Determine beam sizes Minimum Weight (5.2.2)

4 Column design—

4a Select overstrength factors, %-t Abstraction (5.1.1)

4b Calculate updated forces F-m Updating (5.1.4)

4c Calculate design moment Mc(h) and axial forces Eng. Performance (5.1.5)

Pcih)

4d Determine column sizes Minimum Weight (5.2.2)

5 Verify with nonlinear static and nonlinear dy- Eng. Performance (5.1.5)

namic analyses
170

This example involves determining preliminary steel member sizes for a steel moment-

resisting frame in a building. The column spacing and floor heights are already established.

The next step in the design is to consider the possible sizes of steel elements that can be

used for this building. The loading is already known because of the site conditions and the

building occupancy type gives the loads. The decision is to determine approximate sizes of

a steel moment-resisting frame for a 6-story office building.

Step 1: Setup.

This step involves the use of the Engineering Performance (5.1.5) and Abstraction

(5.1.1) patterns. The engineering performance variables determined here are the build-

ing fundamental period T, yield drift 0^, and target drift 0f. The creation of two abstract

parameters y and (3/ is also used to facilitate the design thought process.

The building has five 30' bays and floor heights of 14', except for the ground floor which

is 18' high. The loads for each floor are DL = 100 psf and LL = 40 psf, and at the roof DL

= 75 psf and LL = 40 psf. The floor area is 22,500 ft2. The live load seismic contribution

a = 25%, which gives a seismic story weight w; per bay of 459 kips and 360 kips at the

roof. The structure period is estimated as T = 0.035H015 « 1.006 sec. The target drifts

are total drift 6, = 3%, yield drift Qy = 1%, and plastic drift Qp = 6f - 0y = 2%. Because

T > 0.7sec, then the structure ductility is estimated to be fis = 1 + -^ = 3, and R/j = /JS. The

shear proportioning modification factor is then y = ^ ~ 0.556.


171

The lateral load distribution on the building is estimated using the mode shape factor b

and shear proportioning factors (3/ and A,,- for computing the base shear.

A=I'^£*V

^=(Pi-p'-')|i^1 (5 io>
'
Step 2. Design forces.

The total base shear Vi, can be computed using

Vb _ ~a+yJa2 + 4yC^
(5.12)
W ~ 2

The Ce factor depends on the building's importance factor /, seismic zone coefficient Z

and soil factor S

71S
Ce = 1 . 2 5 ^ 3 (5.13)

The total base shear for this example building is VJ, = 246kips. The base shear is dis-

tributed as Fj = XiVt,.
172

Step 3. Beam design.

Once the lateral force distribution is known, the next step in the problem solving method

is to begin designing the beams. The beams are specified by a required value of plastic

modulus Z based on story moment demands. The moment demands are computed using

MPbi= • (5.14)
§b

where the reference beam plastic moment is

(rj=i Fjhj)-m
Mpbr = ~ yn ;R (5.15)

and bottom story column demand

MP« = H ^ 1 ( 5,6)

At this point, the set of possible steel beams is narrowed to the ones that satisfy the

moment demands. To further narrow the set, ASTM A992 grade steel is chosen. A992 is

the best grade of structural steel for seismic performance, and there are no other feasible

market alternatives. This gives a yield stress of Fy — 50ksi and required plastic modulus

of

Zmin = Mpb/Fy. The beams must also meet the seismic slenderness requirements for the

flange Xpf and web Xpw.

Step 4. Column design.

Once actual beam sizes are selected, then the story moment demands Mpt are updated

using the the actual shapes Mpi, = FyZ. Then, the design procedure goes through the se-
173

lection of the columns. The first step is to select the overstrength factors £,- at each story.

For this building, the overstrength factor is £,• = 1.05 except at the roof where %n = 1. The

roof level force Fnu (n is for the roof level, u means updated with the selected beams) is

determined using

2Mpc + 2(ZU^jMpbj/
nu =
x^n Ta 7~a T\ (5-17)

The individual lateral story forces using the updated values are

Fiu = (ai-^+l)Fnu (5.18)

The columns can be treated as free body cantilevers and their moments computed using

equilibrium equations

McMQ = [ ZSkMpbA - ^ (j^SiFiuihi-h) ) (5.19)

McM(h) = 2 I £ bfaMpbi I " f E Wiu (hi - h) J (5.20)

The axial forces are also computed using equilibrium and the controlling LRFD load

combination by

Pcinti = t,0.5L2 (l.2DLj + l.6LLj) (5.21)


174

S; J
Pc,^ = 0.5PcMi + £ / (5.22)
L
j=i

This provides all the loading information necessary to choose shapes for the columns

and the beams. A final check whether each joint satisfies Y,Mpc/Y,Mpt, > 1.0 ensures

strong column, weak beam behavior.

Step 5. Verify with nonlinear static and nonlinear dynamic analyses.

This final step in the design procedure verifies the final design by checking with more

detailed analysis as needed. This is intended to remind the designer to evaluate any remain-

ing Engineering Performance (5.1.5) criteria that have not been verified in the previous

steps.

5.4.4 FEMA 350 steel moment connections

Decision unit

How should the connections be detailed in order to accomplish a moment-resisting

steel frame?

Problem solving algorithm

Design method outlined in FEMA 350—Recommended Seismic Design Criteria for

New Steel Moment-Frame Buildings (SAC Joint Venture Partnership, 2000).


175

Table 5.3: Connection design procedure based on FEMA 350 basic design approach.

# Step description Patterns applied

1 Frame configuration Mechanics Model (4.2.5)

2 Choose connection configuration Pick Any (5.2.1)

3 Design connection using plastic hinge location Eng. Performance (5.1.5),

and connection demands Prescription (5.2.5)

4 Detail the connection Prescription (5.2.5)

Lesson

Different mechanics models are needed during different stages of steel moment frame

design. A model with highly precise spatial information is needed when designing

the connections, but another model is needed to design for the overall building behav-

ior. The method uses a combination of Pick Any and Prescription decision patterns

in order to arrive at a reliable design supported by laboratory experiments.

FEMA 350 explains a procedure for the design of new steel moment-frame buildings,

with particular emphasis on the design of the moment-resisting connections. The basic de-

sign approach is summarized in Table 5.3, and the following example is given for the design

of a reduced beam section (RBS) prequalified connection using the method summarized in

Table 5.4.
176

Undeformed
frame

thrift anqle - 0

Figure 5.6: Inelastic behavior of frames with hinges in beam span (FEMA 350, p. 3-3).

Step 1: Frame configuration.

The first step of the design procedure uses a simple Mechanics Model (4.2.5) to illus-

trate the design proper frame proportions. The performance criteria are primarily concep-

tual in this step. The mechanics model used here does not require a very high level of

maturity, and only needs basic spatial and material information in order to proceed. The

main requirements are having appropriate plastic hinge locations in the beam span (Fig-

ure 5.6) and the ability to achieve the require interstory drift angle.

Step 2: Select connection configuration.

This step in the procedure involves selecting a configuration that will be compatible

with the structural system, fit the element sizes, and be able to provide adequate inter-
177

story drift capacity. Some connections are only permissible for certain ranges of beam

and column size because of limited experimental data. If these criteria are met, then any

prequalified connection is considered acceptable, and the Pick Any decision pattern (5.2.1)

can be applied to choose among the catalog of prequalified connections. This example uses

the reduced beam section (RBS) prequalified connection.

Step 3: Design connection using plastic hinge location and connection demands.

The plastic hinge location s/, (Figure 5.7) is spatial design information based on the

selection of connection configuration from Step 2. This Engineering Performance parame-

ter also depends on the the amount of gravity load placed on the beam. When the gravity

load is low, the maximum moment is expected to occur at the column faces; when high,

the maximum moment could occur somewhere inside the beam span, requiring additional

computation. After sj, is determined, the other Engineering Performance (5.1.5) parame-

ters are computed: the probable plastic moment at hinges Mpr, shear at the plastic hinge

Vp, moment strength demands at the column face Mf and column centerline Mc, and the

yield moment at column face Myf.

Specifically for RBS design, the location 57, determines the position of the flange re-

duction. Then the rest of the prescriptive process outlined in Table 5.4 determines the

additional spatial information for connection details (Figure 5.8).


178

Connection
reinforcement
(if applicable)

Educed beam
section
( i f applicable)

Figure 5.7: Location of plastic hinge formation (FEMA 350, p. 3-6).

Table 5.4: Design procedure for Reduced Beam Section (RBS) steel moment-resisting
connection.

# Step description Patterns applied

1 Determine length and location of flange reduction Prescription (5.2.5)

2 Determine depth of flange reduction, c Prescription (5.2.5)

3 Compute Mf and Mc Eng. Performance (5.1.5)

4 Calculate shear at column face, Vf Eng. Performance (5.1.5)

5 Design shear connection of beam to column Prescription (5.2.5)

6 Design panel zone Prescription (5.2.5)

7 Check continuity plate requirements Prescription (5.2.5)

8 Detail the connection Prescription (5.2.5)


179

4c2+ b2

Figure 5.8: Reduced beam section connection detail (FEMA 350, p. 3-40).
180

5.5 Summary

What I have done in this chapter:

1. Documented existing methods for design analysis and decision making as patterns.

2. Defined a vocabulary of pattern names for describing structural engineering design

methods, which can be used as a metalanguage for better describing design intent.

3. Illustrated pattern usage for steel and reinforced concrete design at the system, ele-

ment, and detail scales.

4. Indicated places in design procedures where decision making can be improved.

5. Explained the reasons for using different decision patterns based on available infor-

mation and the information type to be determined.


181

Chapter 6

Extending performance-based design

Performance-based design has proven difficult to transfer into practice because con-

temporary design procedures are not suitable for supporting all aspects of a complete

performance-based methodology. Many design procedures have been proposed for im-

plementing performance-based methodology, but typically suffer from common flaws: a

narrow definition of performance that focuses only on the as-built condition but neglects

the other project phases, lack of guidance on how to select among multiple design options,

and an emphasis on design evaluation without a process for design creation. This chapter

extends performance-based design processes by using the patterns that solve these common

flaws.

Existing design methods can be augmented in order to consider performance criteria

for all phases, provide more guidance on design selection, and provide a method for eval-

uating and generating design options. These extensions to the performance-based design
182

process can be accomplished using the appropriate design patterns within structural engi-

neering design procedures. This chapter investigates and provides examples of three design

procedure extensions: (1) additional analysis for decision support, (2) changing decision

patterns, and (3) using set-based decision patterns.

6.1 Extension 1: Additional analysis for decision support

Problem

Performance is not considered for all project phases.

Solution

Add additional analysis that considers performance in the lacking phases.

Description

Adding additional analysis for decision support improves design by (1) providing per-

formance criteria for all the project phases instead of only considering the as-built structure,

and (2) providing additional information to discern benefits among multiple options (Fig-

ure 6.1). Many current design procedures only consider Engineering Performance (5.1.5),

which are insufficient to differentiate between design options that may have very different

Social Performance (5.1.6). These additional analysis steps may require additional work,

but ultimately allow better decisions in the design process. Adding additional analysis

might require a change in decision pattern in order to maximize or minimize a different

objective, or to attempt optimizing among a set of different tradeoff criteria.


183

Engineering Perf. Engineering Perf. Social Perf.


criteria criteria criteria

[Decision pattern] [Decision pattern]

Acceptable Better
final design final design

Figure 6.1: Additional analysis provides more performance criteria for making a better
final design choice.

Two examples illustrate this extension: (1) including construction phase performance

in reinforced concrete beam design, and (2) including construction and repair phase perfor-

mance in reinforced concrete bridge column design.

Example 1: Labor rates in reinforced concrete elements

Additional analysis is inserted into the reinforced concrete beam reinforcement pro-

cedure described in Sec. 5.4.2. The original design procedure did not consider any social

performance design criteria and only focused on engineering performance based on flexural

resistance. There was no additional information that could be used to discern the benefits

of the three different options considered for bar layout. The improvement adds a Social

Performance requirement, labor productivity rate, to refine the selection of the bar layout

(Figure 6.2).
184

The effect of this additional analysis will sharpen the selection of bar layouts. Sup-

posing that a minimum labor productivity rate of 2,000 lbs/man-day were required for the

beam rebar, only the options using #8 and #9 bar would be feasible. The labor productivity

of straight #7 bars is on the order of 1,700 lbs/man-day, while #8 and #9 bar are higher than

2,000 lbs/man-day (Bennion, 2007; Parrish et al., 2008b). The labor placing rates for shear

walls and beams are shown in Figure 6.3. This eliminates the feasibility of using ten #7

bars.

This additional analysis will ensure a selection that satisfies performance in the con-

struction phase of the project in addition to the engineering performance of the as-built

structure. The same Pick Any decision pattern can still be used to make the final decision

on rebar layout.

Example 2: Bridge repair costs and repair time

The existing method for designing reinforced concrete columns for highway bridges

emphasizes performance in the as-built condition. Adding additional analysis for evaluat-

ing bridge repair cost and repair time provide better performance-based design by consid-

ering performance during the bridge's construction and repair phases. This can be accom-

plished by integrating the LLRCAT methodology (Sec. 4.3) and rebar labor productivity

rates into the design procedure.

Some of the steps involved in the column design for a single column bent, 2-span box

girder bridge are summarized in Table 6.1. This procedure is based on information from
185

Existing method Improved method

3. Compute required 3. Compute required


reinforcement area, As reinforcement area, As

3a. Set a minimum


required labor <= Additional analysis
rate for installation

4. Try a specific 4. Try a specific


configuration for As configuration for As

Figure 6.2: Additional analysis considers construction phase performance by analyzing


labor productivity rates during reinforcement design.
186

1,800

1,600

1,400

1,200 4 - C

1,000

500 (.78 in2) 1000 (1.55 in2) 1500 (2.33 in2) 2000(3.10 in2) 2500(3.88 in2) 3000 (4.65 in2)

Figure 6.3: Labor productivity rates for shear walls and beams vary by bar size and bending
type (Parrish et al., 2008b).
187

Table 6.1: Original procedure for single column bent design.

# Step description Patterns applied

1 Determine column size limits based on super- Prescription (5.2.5)

structure design

2 Choose column size Pick Any (5.2.1)

3 Determine column reinforcement requirements Prescription (5.2.5),

Eng. Performance (5.1.5)

4 Choose longitudinal bar layout Pick Any (5.2.1)

5 Determine confinement requirements Eng. Performance (5.1.5)

6 Choose transverse bar layout Pick Any (5.2.1)

the column design procedure outlined in Caltrans Bridge Design Aids (BDA), Ch. 15 and

Ketchum et al. (2004), using Caltrans Seismic Design Criteria (SDC).

Step 1 determines column size limits based on the ratio between column cross sec-

tion and bent cap depth in order to meet joint shear requirements. This limits acceptable

columns to the ratio 0.7 < ^ < 1.0 (Caltrans SDC, Eq. 7.24). This provides basic pre-

scriptive criteria, but provides no additional guidance beyond this point. Performance for

the repair phase of the structure's life can be considered by limiting the maximum repair

cost ratio for the structure. Knowing the superstructure design and the basic estimates on

column size are sufficient to employ the LLRCAT methodology for estimating the cost of

repairing the bridge after a major earthquake.


188

Steps 3 and 5 establish requirements for longitudinal reinforcement and transverse rein-

forcement requirements. The required longitudinal reinforcement ratio can be determined

from design charts in Caltrans BDA, and the transverse reinforcement spacing and bar size

determined based required shear strength in the plastic load case. Beyond these minimum

limits, the procedure gives no additional guidance on rebar combinations that are the best

for construction phase performance. The construction phase performance can be consid-

ered by setting a limit on the minimum labor productivity rates for different types of rebar

type, size, and layout as in the previous beam example.

The improved procedure with these additional analysis steps is summarized in Ta-

ble 6.2. The extra analysis steps before decisions add performance consideration for the

construction and repair phases.

6.2 Extension 2: Change decision patterns

Problem

Design methods are unable to differentiate among several choices.

Solution

Change the decision pattern to one that optimizes performance. Add analysis if perfor-

mance measures are unavailable.

Description

A decision making pattern can be replaced with a better decision making pattern that
Table 6.2: Improved procedure for single column bent design.

# Step description Patterns applied

1 Determine column acceptable column size range Prescription (5.2.5)

based on superstructure design

la Limit column size range by maximum accept- Eng. Performance (5.1.5)

able repair cost ratio size range

2 Choose column size Pick Any (5.2.1)

3 Determine column reinforcement requirements Prescription (5.2.5),

Eng. Performance (5.1.5)

3a Limit reinforcement options by minimum re- Social Performance (5.1.6)

quired labor productivity rate

4 Choose longitudinal bar layout Pick Any (5.2.1)

5 Determine confinement requirements Eng. Performance (5.1.5)

5a Limit reinforcement options by minimum re- Social Performance (5.1.6)

quired labor productivity rate

6 Choose transverse bar layout Pick Any (5.2.1)


190

targets the desired measures of performance. These replacement decision patterns often

take advantage of the information from additional analysis (Figure 6.4).

When substituting the Pick Any pattern for another decision pattern, it is best to com-

bine with additional analysis that will enable the support of a different decision pattern.

Usually the additional analysis will be for a Social Performance analysis because these are

the types of performance criteria that are not usually inside most design procedures.

Example: Choosing steel connections

The example of selecting steel moment connection types in the example from p. 174

is improved by changing the decision pattern and adding additional analysis for the cost

of each connection. The design method suggested by FEMA 350 provides no guidance

on which moment-resisting connections should be chosen, beyond the limits on acceptable

engineering performance. The design method is unable to provide additional information

to choose, and therefore the Pick Any pattern could be used.

A better alternative would be to use the Minimize Cost pattern and perform additional

analysis to investigate the costs for each valid prequalified connection option (Figure 6.4).

This could be done by considering factors for cost-effective steel building design. Carter

et al. (2000) describe several criteria for bolted and welded connections, as well as gen-

eral suggestion for beam and column selection, material costs, fabrication labor costs, and

erection labor costs.

Market-specific information could be obtained from specialty connection suppliers as


191

Prequalified
connections

Pick Any

Evaluate cost 4
of connections

Minimum Cost

Figure 6.4: Replacing a Pick Any pattern with another decision pattern produces better
results and may require additional analysis.

well. For example, data from the supplier of the SlottedWeb™proprietary connection

shows that for an office tower with 750 moment connections cost over $340,000 less on

average than two other options for moment connections (Seismic Structural Design As-

sociates, 2005). Using this additional analysis to choose a minimum cost option before

proceeding with design requires additional work in design, but leads to better performance

during the construction phase.


192

6.3 Extension 3: Use set-based decision patterns

Problem

An evaluation method is available to check performance, but cannot be used adequately

until more details are chosen. The structure design is not mature enough to fully estimate

performance measures, yet it is also too early to commit to a particular design choice.

Solution

Use the Explore Set (5.2.6) decision pattern to evaluate performance of multiple can-

didate design options, then optimize among the options to make the final design choice.

These sets of design options can be further evaluated for other measures of performance,

allowing the exploration of more options, which would have otherwise been neglected (Fig-

ure 6.5).

Example: Beam-column joint reinforcing steel selection

The selection of beam-column joint reinforcing steel illustrates the use of the set-based

design concept (Parrish et al., 2007). Constructibility and structural performance are the

main performance factors for joint design.

Mechanics models analyze the loads and provide the minimum required steel area As

for the beams and the columns (Figure 6.6). Prescriptions from ACI 315 provide additional

limits on dimensions and rebar layout. These two factors take care of the requirements for

engineering performance on the joint. But, at this stage there is insufficient information

to characterize the constructibility of the joint. The mechanics model and prescriptions
Existing method Improved method

Engineering Perf. Engineering Perf.


criteria criteria

[Decision pattern] Explore Set

Acceptable Option 1 Option Option n


final design

Evaluate more
Eng. and Social
Perf. criteria

Optimization

Better
final design

Figure 6.5: Substituting a set-based decision patterns allows further evaluation of mu


options in order to find a design with better performance for all project phases.
194

are insufficient for selecting a rebar configuration that considers construction performance.

Additional preferences, rules of thumb, and feedback from rebar fabricators and placers are

needed to proceed with the best option.

But, the design lacks the information needed to evaluate construction performance. The

information is not yet available because the design is not detailed enough, using aggrega-

tions of bar instead of actual bar layouts. Using the Pick Any decision pattern to choose any

layout and go proceed with design might work, but would lead to suboptimal construction

performance.

Instead, the set-based approach can be used to evaluate construction performance. De-

tailing requirements can be determined by setting a ductility demand and then determining

strength (Aschheim, 2002), or by setting a deformation mechanism (Lee and Goel, 2001).

Several options can be chosen from the limits of acceptable strength and ductility pairs and

other prescriptive design limits (Figure 6.7).

Then, the combinations of beam and column reinforcement can be evaluated for con-

struction performance using preferences from the design team. The double layer reinforce-

ment option might be eliminated due to increased labor required to place the bars. Certain

sizes of rebar might be eliminated due to material availability or other market constraints.

The exact spacing of the bars inside the joint can be considered to check whether there

is enough space to avoid congestion. This would be difficult to check without evaluating

a specific layout of reinforcing bar including bar size, bar count, and each bar's position.

After evaluation of construction performance, the best option can be chosen. Since all
195

(18")
47cm 1

As>l9cm2 1
61cm
(24")
A's>l2cm2

Seam B
1-
(24")
I 61cm 1

As>6lcm2
1
61cm
(24")

Column C
??fe 7M7s

Figure 6.6: Required steel area based on mechanics model (Parrish et al., 2007).

the options were already satisfactory for as-built structural performance, the choice can be

made on construction performance alone.


(24")
| 61cm 1

I •
61 cm • I6#7bars



I2#8bars

(24") • • • •

1 Column I Column 2
• •

• 8# 10 bars •

Column C • •
Column 3

(18")
I 47cm-
(18")
•47cm- i • • • • •
5#7 bars
• • • •
4#8 bars

3#9 bars
• •

As>!9 cm2 I
61cm
6lcm
(24")

1 3#7 bars

Beam I
• •
3#7 bars

Beam 2
• •
3#7 bars

Beam 3
• •

(24")
A's > 12 cm2

1
• • • ••••••• • •
3# 10 bars 7# 10 bars • •
4#8 bars

3#7 bars 3#7 bars 3#7 bars


beam b • • • • • • • • •

Beam 4 Beam 5 Beam 6

Figure 6.7: Bar layout options for beam and column.


197

4 Summary

What I have done in this chapter:

1. Explained three methods for including performance-based criteria for all project pha-

ses into existing design procedures.

2. Provided examples of these three improvement patterns.

3. Demonstrated the applicability of using design and decision patterns to identify gen-

eral cases for improvement.

4. Developed examples for analyzing design methods and improving their behavior.
198

Chapter 7

Conclusion

7.1 What I have done here

In this thesis I have defined terms describing structural engineering design information,

explained patterns of communicating design information, described patterns of structural

engineering analysis, and patterns for making decisions within design structural engineer-

ing design procedures. Several examples from current design methods illustrate the use

of these patterns. I have also suggested general recommendations for extending design

methods in order to correct common problems.

Chapter 2 explained the phases in the structure life cycle. A fully performance-centric

design methodology should address performance in all of these phases.

Chapter 3 defined different aspects of the content and quality of structural design in-

formation. These definitions were applied to explain how different types of content and
199

quality influence structural analysis, and articulated several challenges to communicating

information quality.

Chapter 4 documented patterns for communicating design information at the appropri-

ate levels of detail. The examples illustrated the necessity of annotation and text in order to

describe structure attributes. Text and graphics combined excel at communicating concep-

tual physical design information. An example of performance criteria for the repair phases

was given for reinforced concrete bridges. Some comments on the use of BIM technology

indicate requirements necessary for implementing performance-based design.

Chapter 5 documented patterns for analysis and decisions. These patterns define a

vocabulary for describing structural engineering design methods, which can be used as a

metalanguage for better describing design intent. Pattern usage was illustrated for steel and

reinforced concrete design at the system, element, and detail scales. The use of patterns

helps identify the places in design procedures where decision making can be improved, and

explains the reasons for using different decision patterns based on available information.

Chapter 6 explained three methods for extending performance-based criteria in order

to avoid the common flaws of neglecting performance in some project phases, lack of

guidance for making decisions, and emphasis on evaluation alone. These methods identify

the reasons why performance-based design is difficult to implement and propose feasible

solutions.
200

7.2 Future work

The work in this thesis is never complete because there are many additional directions

to consider. A few suggestions for future work are described below:

7.2.1 Expanded pattern catalog

Forming a more comprehensive and detailed catalog of additional analysis patterns

would be helpful for understanding and teaching about the vast number of different ap-

proaches are in the literature. Specific methods for earthquake engineering analysis such

as the Yield Point Spectrum method (Aschheim, 2002) and the Performance-based Plas-

tic Design method (Lee and Goel, 2001) could be formulated as design patterns in order

to succinctly communicate when such methods are applicable and advantageous within a

design procedure. The six methods for estimating inelastic displacement compared in Mi-

randa and Ruiz-Garcia (2001) could also be written as patterns in order to list the specific

details involved in their implementation and when they are best applicable to use within

design procedures.

Future work could also develop additional communication patterns for the specifically

handling probabilistic uncertainty throughout the design process. Instead of handling vari-

ables in their single deterministic values, their probability distributions would need to be

understood. This would reflect the uncertainty between the real and ideal as-built states for

a fully mature design.

More patterns for extending performance-based design can help improve the methods
201

of today. Three examples were provided in Chapter 6, and many more could be developed

by analyzing where analysis is used for making decisions, and what criteria are used to

make those decisions.

7.2.2 Communication patterns with BIM

Future work could be done to extend the discussion of communication patterns into how

BIM technologies are influencing the way information is exchanged and expressed in de-

sign practice. There is great potential in using the recent advances of computer technology

for enhancing communication. But, much of the emphasis has been on the communication

of detailed spatial and material information, and not so much on developing these similar

tools for modeling at the preliminary phases of design. More work could be done to in-

vestigate how to incorporate these immature phases of design into a more unified set of

tools.

The use of BIM could also focus on modeling for the operation and repair phases of

a structure. There is great potential for merging building health monitoring data into a

cohesive framework with the building model. Instead of building information modeling,

during the operation phase, additional data could be merged with the building model to do

building information management as well. Building information management processes

might involve tracking the state of the structure after initial construction is over and the

building is placed into normal service. The effects of design and construction decisions

could be observed with the building in service. The merger of these pieces of information
202

could provide research engineers and practicing engineers with valuable data for improving

design.

7.2.3 Teaching and training engineers

Future work could explore using communication, analysis, and decision patterns as a

teaching tool that can help explain the reasons behind the steps involved in design proce-

dures. This would benefit students by providing them with more information about the use

of each pattern so that they can learn to practice good judgment for evaluating design and

communication procedures. Future work could expand this type of analysis of communi-

cation, analysis, and decision patterns into a more complete catalog for teaching students.

A masters-level class on structural systems might include a discussion of patterns for

both the structural systems (e.g., moment frames) and the methods for designing them (e.g.,

Lee and Goel, 2001). This would provide students with clear distinctions between the ben-

efits and challenges of using and designing each system. The pattern method could also

equip students with tools for choosing different analysis procedures available to them. For

example, there are many different numerical methods for performing nonlinear structural

analysis for earthquakes. But, many students tend to be confused about when and where

to use each one. Clear patterns that describe the strengths and weaknesses of each method

would make this topic much more clear, and enable students to make better informed deci-

sions in design practice. The pattern template used in the classroom could later become a

handy quick reference sheet in the design office. A blank pattern template could be used on
203

an exam for testing students' conceptual understanding of structural systems and methods

for analysis and design.

An undergraduate-level class on design communication might include specific topics

involving when and why to use different communication patterns. These topics could con-

nect to a capstone design course in order to give students an opportunity to communicate

a "real" design. For example, the project could begin with writing a conceptual narrative,

and making some preliminary sketches with annotation. Then, the level of detail would

increase throughout the course prompting the use of other communication patterns. Teach-

ing the patterns might also give students the tools to choose methods of communication,

instead of merely learning how to implement communication, such as merely learning how

to use CAD or BIM software without also learning why and when to use it. Simply requir-

ing students to turn in a "design report" with a list of required pieces would not be as rich

as providing the rationale for each piece using a pattern description.

The difference between teaching and training is clarified when the different patterns

are used to explain design methods. Teaching involves imparting knowledge and training

involved imparting skills (Gregory, 1886; Shafer, 1985). Thus, teaching can involve the

understanding and use of the patterns, while training can focus on the actual implementa-

tion and execution of these patterns. This distinction between the knowledge about design

and the actual work of design can benefit the way students understand and discuss what is

taught.

When van den Broek (1940) wrote on the use of plastic analysis, he noted that, "In
204

the training of engineers, the emphasis on the theory of elasticity leads to a respect for

the elastic limit which sometimes borders on fear." Spelling out the specific reasons why

certain analysis and decision patterns are used in a design procedure could help eliminate

this kind of unfounded fear that happens when changes are made. The use of the pattern

definitions and explanation of applicability and consequences can make these tacit fears

and assumptions more clear and considered.

7.2.4 Design with high ambiguity and unclear mechanics

The examples developed in this illustrate the process of set-based design under condi-

tions where the mechanics assumptions are well-known and there are an adequate number

of abstractions that can express the performance of generic structural configuration. One

area of future work is determining how to do set-based design in situations where the per-

formance of the system cannot be evaluated using established design methods.

The lack of ability to evaluate occurs commonly when using innovative designs. In

these cases, often laboratory tests are required to verify the performance of the new type of

structural device. It is in these situations that the tests may verify a particular instance of the

device, but not be able to verify an entire class of similar devices. The advantage of having

a well-established mechanics theory is that it allows one to make categorical assumptions

about behavior over an entire set of options even though the individual instantiations are

rare or have never been built before.


205

7.2.5 Reliability assessment during design

Future work could be done on describing patterns for reliability assessment during de-

sign. The probabilistic uncertainty of a future building's performance is categorized differ-

ently than that of an existing building. Der Kiureghian and Ditlevsen (2008) comment that

"the different categorization of uncertainties in an existing versus a future building dictates

a fundamental difference in the methods used for assessing their reliabilities." An exist-

ing building's reliability should be assessed using the known history of the building, but a

future building is known only from the design process and can be thought of as a random

sample taken from a population. For example, the concrete strength of a future building is

unknown and this aleatory uncertainty cannot be reduced. But, in an existing building, the

uncertainty in concrete strength is epistemic and could be reduced by obtaining data from

material samples obtained during construction.

Understanding this distinction leads to different patterns of analysis that are suitable

for existing or future structures. For example, a pattern could be described for Bayesian

techniques that are well-suited for information updating on existing structures, and another

pattern could account for the aleatory uncertainty between a design and the actual as-built

condition. Communication patterns could be explored for representing this distinction in

sources of uncertainty, which "has been missed in much of the literature on structural reli-

ability" (Der Kiureghian and Ditlevsen, 2008).


206

7.2.6 Development of computer tools for supporting set-based design

Computer tools can help support set-based design by making it easy to consider a range

of options that fulfill defined performance criteria. The framework for this comes out of

identifying appropriate abstract quantities and knowing how to select a range of instanti-

ations that satisfy the abstract constraint. This would be like making a tool to conduct an

easy "parameter" study.

For example, p is an abstract quantity derived from reinforced concrete mechanics the-

ory, and it is manifested in reality from a particular bar configuration inside a particular

concrete element. So, a tool that would allow iteration over many real rebar configurations

and values of As for a given b and d of a beam would be helpful in evaluating options.

This concept is important because the mechanics theory does not really provide any

guidance beyond the computation of parameters like p, As and checking whether a particu-

lar real configuration satisfies the required demand and capacity requirements. Mechanics

theory doesn't provide information about the constraints on bar availability, ease of placing,

or fabricator efficiency. Allowing mechanics theory to function on the level of mechanics

theory, and allowing other factors to influence the other design decisions will allow better

designs for overall project performance. Avoiding confusion between the categories and

types of design information can be helped with a computer tool.

I have already developed a prototype tool written in C# that integrates with the Tekla

Structures OpenAPI. This tool is able to evaluate Social Performance design criteria such

as: automatic machinability, rebar placing time, and to compute information on the level of
207

1 * ki-b.H S f t I I C I V M I D I ' M U I I Inul !ft' -••• .'38fe _ |n|x|


Rebar Labor Model

1 J-J button 1 | Refresh

Bar size d ashboard


T
Materia! A v a i l a b l y 3 -* 5 6 8 Q 10 >=n< i i 1

Bending tolerance 3 4 5 6 7 8 9 10 -n<—•*!*-••


Cost 3 4 5 6 ~ S p 10 •«--• i' W

Placing Time 3 4 5 6 " S 9 10 X--1-' l.fffsfe


#
Show Selected Obi Oass j 3nine tools
Gel Concrete Properties 1 dSaedPrqpeifiss i
8arsize
BE1 Wal BE2 eetbarproperties |
1"
Dimensions Dimenwn* Dimensions
J127D*312 8 J9144-508 j 1270*812 S
Concrete strenj-Si Concrete strength Concrete amngth Grabbing Tekla. Structures. Model SingleRebar
Moving to next object...
|4G:C (4000
Grabbing Tekte.Stajctures.Model.SingteRebar
Moving to neat object...
Grabbing Tdda.Structures.Modd.SngteRebar
Hoop size
J2-35bar
Honzortelbarst-e
J5S-S5bar
Hoop size
J2-S5bar
Moving to next object... A':
Vertical steel ^ze Vertical bar size Vertical steel size Bar Size I Bar Area j No. Bars 1 Total Seel Area !
|3C • =11 bar jSC-Pbbar J30-Pl1bar
Short dr ties Short dr. ties
]3-s5bar
Long * . ties Longdir ftes
|l-=Eoar )1-S5bar

Tsa cotorj by userfieki 3

Check machine bending J Reset color


1
Figure 7.1: Set-based design tools can integrate with existing BIM software. This prototype
tool uses data from a reinforced concrete shear wall model in Tekla Structures 14.

detail necessary for rebar fabrication time and cost estimates. The tool is able to extract de-

sign information from a Tekla Structures model for computation, and input the information

back into the model for visualization (Figure 7.1).


208

Bibliography

ACI Committee 315 (1994). ACI Detailing Manual. American Concrete Institute, Detroit.

ACI Committee 318 (2005). Building Code Requirements for Structural Concrete and

Commentary. American Concrete Institute.

Al Sarraj, Z. M. (1990). Formal development of line-of-balance technique. Journal of

Construction Engineering and Management, 116(4): 689-704.

Alexander, C , Ishikawa, S., Silverstein, M., Jacobson, M., Fiksdahl-King, I., and Angel, S.

(1977). A pattern language : towns, buildings, construction. Oxford University Press,

New York, xliv, 1171 pp.

American Institute of Steel Construction (2005). Steel construction manual. American

Institute of Steel Construction, Chicago, thirteenth edition.

American Society of Civil Engineers (2000). Quality in the constructed project: a guide

for owners, designers, and constructors. American Society of Civil Engineers, Reston,

VA.
209

American Society of Civil Engineers (2008). Civil Engineering Body of Knowledge for

the 21st Century : Preparing the Civil Engineer for the Future. American Society of

Civil Engineers, Reston, VA, second edition, http://www.asce.org/professional/

educ/bok2.cfm.

Anwar, N., Kanok-Nukulchai, W., and Batanov, D. N. (2005). Component-based, infor-

mation oriented structural engineering applications. Journal of Computing in Civil En-

gineering, 16(1):45—57.

Applied Technology Council (2008). Quantification of building seismic performance fac-

tors : ATC-63 project report, 90% draft. FEMA Report P695, Federal Emergency Man-

agement Agency.

Aschheim, M. (2002). Seismic design based on the yield displacement. Earthquake Spec-

tra, 18(4):581-600.

Aschheim, M., Hernandez-Montes, E., and Gil-Martin, L. M. (2008). Design of optimally

reinforced RC beam, column, and wall sections. Journal of Structural Engineering,

134(2):231-239.

Austin, M. A., Pister, K. S., and Mahin, S. A. (1987a). Probabilistic design of earthquake-

resistant structures. Journal of Structural Engineering, 113(8): 1642-1659.

Austin, M. A., Pister, K. S., and Mahin, S. A. (1987b). Probabilistic design of moment-
210

resistant frames under seismic loading. Journal of Structural Engineering, 113(8): 1660-

1677.

Ballard, G. (2000). The Last Planner System of Production Control. PhD thesis, The

University of Birmingham, Birmingham, United Kingdom.

Ballard, G. (2001). Lean Project Delivery System™. White Paper 8, Revision 1, Lean

Construction Institute, h t t p : //www.leanconstruction.org/pdf/WP8-LPDS.pdf.

Bay Bridge Public Information Office (2008). Fact sheet : Self-anchoring suspen-

sion span. Online, http://baybridgeinfo.org/sites/default/files/pdfs/

sasfactsheet.pdf.

Bennion, H. (2007). Rebar placing rates, personal communication. PCS Steel, Inc., Fair-

field, CA.

Birley, D. (2008). Reinforcement placing drawings are not shop drawings : requirements

for early submittal can be detrimental. Concrete International, 30(12):48-50.

Browning, T. R. (2001). Applying the Design Structure Matrix to system decomposition

and integration problems: A review and new directions. IEEE Transactions on Engi-

neering Management, 48(3):292-306.

California Department of Transportation (2004). Seismic design criteria, version 1.3.

Engineering services manuals, Division of Engineering Services, Sacramento, CA.


211

http://www.dot.ca.gov/hq/esc/techpubs/manual/othermanual/

other- engin-manual/ seismic- design- c r i t e r i a / sole. html.

California Department of Transportation (2005). Bridge design aids. Engineering services

manuals, Division of Engineering Services, Sacramento, CA.

http://www.dot.ca.gov/hq/esc/techpubs/manual/bridgemanuals/

bridge-design-aids/bda.html.

Carter, C. J., Murray, T. M., and Thornton, W. A. (2000). Cost-effective steel building

design. Progress in Structural Engineering and Materials, 2:16-25.

Chopra, A. K. (2007). Dynamics of structures : theory and applications to earthquake

engineering. Pearson-Prentice Hall, third edition, xxxiv, 876 pp.

Christy, C. T. (2006). Engineering with the spreadsheet: structural engineering templates

using Excel. ASCE Press, Reston, VA. xv, 317 pp.

Codd, E. F. (1990). The relational model for database management: version 2. Addison-

Wesley. xxii, 538 pp.

Concrete Reinforcing Steel Institute (2000). Reinforcing Bar Detailing. Concrete Rein-

forcing Steel Institute, Schaumberg, IL, fourth edition.

Concrete Reinforcing Steel Institute (2001). Manual of Standard Practice. Concrete Rein-

forcing Steel Institute, Chicago, IL, 27th edition.


212

Concrete Reinforcing Steel Institute (2002). Placing drawings—the detail drawings for

reinforcing bars in site-cast reinforced concrete construction. Engineering Data Report

EDR50, Concrete Reinforcing Steel Institute.

Concrete Reinforcing Steel Institute (2008). CRSI Design Handbook 2008. Concrete Re-

inforcing Steel Institute, Schaumberg, IL, tenth edition.

Cornell, C. A. and Krawinkler, H. (2000). Progress and challenges in seismic performance

assessment. PEER Center News, 3(2): 1-2.

Day, A. (1997). Digital Building. Laxton's, Oxford, xv, 176 pp.

Der Kiureghian, A. and Ditlevsen, O. (2008). Aleatory or epistemic? Does it matter?

Structural Safety, 31 (2): 105-112.

Ditlevsen, O. (1982). Model uncertainty in structural reliability. Structural Safety, 1(1):73-

86.

Ditlevsen, O. and Madsen, H. O. (1996). Structural reliability methods. Wiley, Chichester;

New York, xi, 372 pp.

Doumont, J.-L. (2002a). The three laws of professional communication. IEEE Transactions

on Professional Communication, 45(4):291-296.

Doumont, J.-L. (2002b). Verbal versus visual : A word is worth a thousand pictures, too.

Technical Communication, 49(2):219-224.


213

Eastman, C , Teicholz, P., Sacks, R., and Liston, K. (2008). BIM handbook : a guide to

building information modeling for owners, managers, designers, engineers, and con-

tractors. Wiley, Hoboken, NJ. xiv, 490 pp.

Fee, G. D. and Stuart, D. (2003). How to read the Bible for all its worth. Zondervan, Grand

Rapids, MI, third edition. 287 pp.

Ferguson, E. S. (1992). Engineering and the mind's eye. MIT Press, Cambridge, MA. xiv,

241 pp.

Gamma, E., Helm, R., Johnson, R., and Vlissides, J. (1995). Design patterns : elements of

reusable object-oriented software. Addison-Wesley, Reading, MA. xv, 395 pp.

Gil, N., Beckman, S., and Tommelein, I. D. (2008). Upstream problem solving under

uncertainty and ambiguity: evidence from airport expansion projects. IEEE Transactions

on Engineering Management, 55(3):508-522.

Gregory, J. M. (1886). The seven laws of teaching. Congregational Sunday School and

Publishing Society, Boston. 144 pp.

Harris, H. G. and Sabnis, G. M. (1999). Structural modeling and experimental techniques.

CRC Press, Boca Raton, FL, second edition. 789 pp.

Heng, B. and Mackie, R. (2008). Using design patterns in object-oriented finite element

programming. Computers and Structures, in press: 10 pp.


214

Hughes, W., Ancell, D., Gruneberg, S., and Hirst, L. (2004). Exposing the myth of the

1:5:200 ratio relating initial cost, maintenance and staffing costs of office buildings.

In Khosrowshahi, R, editor, Proc. 20th Annual ARCOM Conference, volume 1, pages

373-381, Heriot-Watt University, Edinburgh, Scotland. Association of Researchers in

Construction Management.

Imhof, E. and Steward, H. J. (1982). Cartographic relief presentation. De Gruyter, Berlin;

New York, xviii, 389 pp.

Jansson, D. G. and Smith, S. M. (1991). Design fixation. Design Studies, 12(1):3-11.

Kang, G. S. and Kren, A. (2006). Structural engineering strategies towards sustainable de-

sign. In Proc. Structural Engineers Association of Northern California 2006 Convention,

pages 473^190.

Kardon, J. B. (2003). The standard of care of structural engineers. PhD thesis, University

of California, Berkeley.

Ketchum, M., Chang, V., and Shantz, T. (2004). Influence of design ground motion level on

highway bridge costs. Lifelines Project 6D01, Pacific Earthquake Engineering Research

Center, Berkeley, CA.

Kneer, E. and Maclise, L. (2008). Consideration of building performance in sustainable de-

sign : a structural engineer's role. In Proc. Structural Engineers Association of Northern

California 2008 Convention. 15 pp.


215

Koskela, L. and Vrijhoef, R. (2001). Is the current theory of construction a hindrance to

innovation? Building Research and Information, 29:197-207.

Lee, S.-S. and Goel, S. C. (2001). Performance-based design of steel moment frames using

target drift and yield mechanism. Research Report UMCEE 01-17, Department of Civil

and Environmental Engineering, Univeristy of Michigan.

Liker, J. K., Sobek, II, D. K., Ward, A., and Cristiano, J. J. (1996). Involving suppliers in

product development in the United States and Japan : evidence for set-based concurrent

engineering. IEEE Transactions on Engineering Management, 43(2): 165-178.

Lin, T. Y. and Stotesbury, S. D. (1988). Structural concepts and systems for architects

and engineers. Van Nostrand Reinhold Co., New York, second edition, xiii, 507 pp.,

appendix 81 pp.

Lynch, J. R and Loh, K. J. (2006). A summary review of wireless sensors and sensor

networks for structural health monitoring. The Shock and Vibration Digest, 38(2):91-

128.

MacGregor, J. G. and Wight, J. K. (2005). Reinforced concrete : mechanics and design.

Prentice Hall, Upper Saddle River, NJ, fourth edition, xx, 1132 pp.

Mackie, K. R. (2004). Fragility-based seismic decision making for highway overpass

bridges. PhD thesis, University of California, Berkeley.

Mackie, K. R., Wong, J.-M., and Stojadinovic, B. (2007). Comparison of post-earthquake


216

highway bridge repair costs. In Proc. Structures Congress 2007, Long Beach, CA.

ASCE/SEI.

Mackie, K. R., Wong, J.-M., and Stojadinovic, B. (2008a). Integrated probabilistic per-

formance-based evaluation of benchmark reinforced concrete bridges. Technical Report

2007/09, University of California, Berkeley, Pacific Earthquake Engineering Research

Center, Berkeley CA.

Mackie, K. R., Wong, J.-M., and Stojadinovic, B. (2008b). Post-earthquake bridge repair

cost evaluation methodology. In Proc. Sixth National Seismic Conference on Bridges

and Highways (6NSC), Charleston, SC.

Mackie, K. R., Wong, J.-M., and Stojadinovic, B. (2008c). Probabilistic methodologies for

prediction of post-earthquake bridge repair costs and repair times. In Proc. 14th World

Conference on Earthquake Engineering (14WCEE), Beijing, China.

Miranda, E. and Ruiz-Garcia, J. (2001). Evaluation of approximate methods to estimate

maximum inelastic displacement demands. Earthquake Engineering and Structural Dy-

namics, 31(3):539-560.

Moehle, J., Stojadinovic, B., Der Kiureghian, A., and Yang, T. Y. (2005). An application of

PEER performance-based earthquake engineering methodology. Research Digest 2005-

1, Pacific Earthquake Engineering Research Center.


217

Naaman, A. E. (2004). Prestressed concrete analysis and design : fundamentals. Techno

Press 3000, Ann Arbor, MI, second edition. 1072 pp.

Naeim, E, Alimoradi, A., Hagie, S., and Comartin, C. (2007). PACT : Performance As-

sessment Calculation Tool user guide. Technical report, Applied Technology Council

ATC-58 Project, Federal Emergency Management Agency.

Nye, W. T. and Tits, A. L. (1986). An application-oriented, optimization-based method-

ology for interactive design of engineering systems. International Journal of Control,

43(6):1693-1721.

Parrish, K., Wong, J.-M., Tommelein, I. D., and Stojadinovic, B. (2007). Exploration of set-

based design for reinforced concrete structures. In Pasquire, C. L. and Tzortzopoulos, P.,

editors, Proc. 15th Annual Conference of the International Group for Lean Construction

(IGLC 15), East Lansing, MI.

Parrish, K., Wong, J.-M., Tommelein, I. D., and Stojadinovic, B. (2008a). Set-based de-

sign: Case study on innovative hospital design. In Tzortzopoulos, P. and Kagioglou, M.,

editors, Proc. 16th Annual Conference of the International Group for Lean Construction

(IGLC 16), Manchester, United Kingdom.

Parrish, K., Wong, J.-M., Tommelein, I. D., and Stojadinovic, B. (2008b). Value propo-

sitions for set-based design of reinforced concrete structures. In Tzortzopoulos, P. and

Kagioglou, M., editors, Proc. 16th Annual Conference of the International Group for

Lean Construction (IGLC 16), Manchester, United Kingdom.


218

Ramberg, B. and Gjesdal, K. (2005). Hermeneutics. In Zalta, E. N., editor, The Stanford

Encyclopedia of Philosophy. Stanford University, Winter 2005 edition, h t t p : / / p l a t o .

stanford.edu/archives/win2005/entries/hermeneutics/.

Saaty, T. L. (1990). How to make a decision : the Analytic Hierarchy Process. European

Journal of Operational Research, 48:9-26.

SAC Joint Venture Partnership (2000). Recommended seismic design criteria for new steel

moment-frame buildings. Technical Report FEMA-350, Federal Emergency Manage-

ment Agency.

Saxon, R. (2005). Be valuable : a guide to creating value in the built environment. Con-

structing Excellence, London. 50 pp.

Schon, D. A. (1983). The reflective practitioner. Basic Books, New York, x, 374 pp.

Schrader, S., Riggs, W. M., and Smith, R. P. (1993). Choice over uncertainty and ambiguity

in technical problem solving. Journal of Engineering and Technology Management,

10:73-99.

Segui, W. T. (2007). Steel design. Thomson Nelson, Toronto, fourth edition.

Seismic Structural Design Associates (2005). The SlottedWeb™steel moment connection.

http://www.slottedweb.com/ssda.pdf.

Shafer, C. (1985). Excellence in Teaching with the Seven Laws : A Contemporary Abridg-

ment of Gregory's Seven Laws of Teaching. Baker Book House, Grand Rapids, MI.
219

Sobek, II, D. K., Ward, A., and Liker, J. K. (1999). Toyota's principles of set-based con-

current engineering. Sloan Management Review, 40(2):67-83.

Sonin, A. A. (2004). A generalization of the ri-theorem and dimensional analysis. Proeed-

ings of the National Academy of Sciences of the United States of America (PNAS),

101(23):8525-8526.

Steward, D. V. (1981). The Design Structure System: A method for managing the design of

complex systems. IEEE Transactions on Engineering Management, EM-28(3):71-74.

Stojadinovic, B., Goethals, J., Wong, J.-M., Ohrui, S., Takahashi, M., Fukushima, I., and

Yamamoto, Y. (2004). Framework for integration and visualization of structural state

data. CUREE Publication CKV-03, CUREE-Kajima Joint Research Program Phase V.

Suhr, J. (1999). The Choosing by Advantages decisionmaking system. Quorum, Westport,

CT. x, 293 pp.

Sun, J., Manzanarez, R., and Nader, M. (2004). Suspension cable design of the new San

Francisco-Oakland Bay Bridge. Journal of Bridge Engineering, 9(1): 101-106.

Terry, M. S. (1974). Biblical hermeneutics : a treatise on the interpretation of the Old and

New Testaments. Zondervan, Grand Rapids, MI, second edition.

Terwiesch, C , Loch, C. H., and De Meyer, A. (2002). Exchanging preliminary information

in concurrent engineering: alternative coordination strategies. Organization Science,

13(4):402-419.
220

Thomas, R. L. (2002). Evangelical Hermeneutics. Kregel, Grand Rapids, MI.

Turk, Z. (2001). Phenomenologial foundations of conceptual product modelling in archi-

tecture, engineering and construction. Artificial Intelligence in Engineering, 15:83-92.

United States Navy (1962). Line of balance technology : a graphic method of industrial

programming. Technical report, Office of Naval Material, Government Printing Office,

Washington DC.

United States of America (2008). Part 100, "Reactor site criteria". In Code of Federal

Regulations, Title 10. United States Government Printing Office, Washington, DC.

van den Broek, J. A. (1940). Theory of limit design. Transactions of the American Society

of Civil Engineers, 105:638-730.

Ward, A., Liker, J. K., Cristiano, J. J., and Sobek, II, D. K. (1995). The second Toyota para-

dox: How delaying decisions can make better cars faster. Sloan Management Review,

36(3):43-61.

Werner, S., Taylor, C , Cho, S., Lavoie, J.-P, Huyck, C , Eitzel, C , Chung, H., and Eguchi,

R. (2006). REDARS 2: Methodology and software for seismic risk analysis of highway

systems. Technical Report MCEER-06-SP08, Multidisciplinary Center for Earthquake

Engineering Research (MCEER), Buffalo, NY.

Wing, J. M. (2006). Computational thinking. Communications of the ACM, 49(3):33-35.


221

Winograd, T. and Flores, F. (1986). Understanding computers and cognition : a new

foundation for design. Ablex, Norwood, NJ. xiv, 207 pp.

Wong, J.-M., Goethals, J., and Stojadinovic, B. (2005). Wireless sensor seismic response

monitoring system implemented on top of NEESgrid. Proceedings ofSPIE, 5768:74-84.

Wong, J.-M., Mackie, K. R., and Stojadinovic, B. (2006). Updating bridge fragility curves

for serviceability and extreme limit states using sensor data. In Proc. 5th National Seis-

mic Conference (5NSC), San Mateo, CA.

Wong, J.-M., Mackie, K. R., and Stojadinovic, B. (2008). Reinforced concrete bridge seis-

mic damage and loss scenarios. In Proc. Sixth National Seismic Conference on Bridges

and Highways, Charleston, SC.

Wong, J.-M., Parrish, K., Tommelein, I. D., and Stojadinovic, B. (2007). Communication

and process simulation of set-based design for concrete reinforcement. In Henderson,

S., Biller, B., Hsieh, M.-H., Shortle, J., Tew, J., and Barton, R., editors, Proc. of the 2007

Winter Simulation Conference (WSC07), pages 2057-2065.

Wong, J.-M. and Stojadinovic, B. (2004). Structural sensor data repository: metadata and

web-based user interface. In Proc. 13th World Conference on Earthquake Engineering,

Vancouver, B.C., Canada. Paper No. 956.

Wong, J.-M. and Stojadinovic, B. (2005). Metadata and network API aspects of a frame-
222

work for storing and retrieving civil infrastructure monitoring data. Proceedings ofSPIE,

5765:948-959.

Wong, J.-M. and Stojadinovic, B. (2006). Improving probabilistic seismic performance

evaluation using sensors and bayesian updating. In Proc. 8th National Conference on

Earthquake Engineering (8NCEE), San Francisco, CA.

Yu, L. and Kumar, A. V. (2001). An object-oriented modular framework for implementing

the finite element method. Computers and Structures, 79(9):919-928.

Yuan, M. (2008). Dynamics GIS: recognizing the dynamic nature of reality. ArcNews,

30(1): 1,4-5.
223

Appendix A

Pattern summary

A.l Communication patterns

• Narrative (4.2.1). Describe preliminary conceptual information using word descrip-

tions instead of using sketches and drawings. Using words alone focuses on commu-

nicating functional, material, and abstract information without prematurely commu-

nicating spatial information.

• Sketch (4.2.2). Describe preliminary conceptual information using sketches in order

to show spatial information and relationships between objects without implying an

artificially higher degree of precision.

• Annotation (4.2.3). Display information in a visual format consisting of a main

graphic element with text that explains the important components of the graphic.

• Spreadsheet (4.2.4). Present information in a tabular format in order to provide easy


224

access to data that is linked or sorted in a particular way using text and numbers.

• Mechanics Model (4.2.5). Use an abstract model of a whole structure, or structure

components, based on mechanics theory to show the structure's response to loads

and deformations.

• Rebar placing drawings (4.2.6). Provide drawings with information necessary for a

rebar placer to tie and place reinforcing bars within concrete formwork.

• Database (4.2.7). Store structure information in a database so that it can be reused by

a computer program for further analysis.

• Schedule (4.2.8). Display process information with an emphasis on showing actions

linked to time so that concurrency and dependency can be easily seen.

.2 Analysis patterns

• Abstraction (5.1.1). Create a variable that is itself non-physical in order to make gen-

eral analysis feasible. Convert the real spatial and material information into another

form to facilitate thinking.

• Aggregation (5.1.2). Group similar items by their material or spatial information,

and treat conceptually as a single item. Simplifies a conceptual or mechanics model

by reducing the number of pieces.

• Agglomeration (5.1.3). Group different types of items in a structure by their function

in order to facilitate analysis.


225

• Updating (5.1.4). Use when an early initial guess needs to be updated after receiving

additional design information.

• Engineering Performance (5.1.5). Measure structural design performance using the

results of a mechanics model related to the stress and strain, force and displacement

of a structural model.

• Social Performance (5.1.6). Measure structural design performance using informa-

tion about social conditions.

A.3 Decision patterns

• Pick Any (5.2.1). Choose any option from a set of options, then confirm whether it

works.

• Minimum Weight (5.2.2). Choose the design option that provides the least material

weight.

• Minimum Cost (5.2.3). Choose a design option that will produce the lowest overall

cost.

• Optimization (5.2.4). Choose a design option based on some optimal value of multi-

ple performance measures involving tradeoffs between them.

• Prescription (5.2.5). Follow an established rule for making a design decision, or

confirming that a candidate design choice is valid.


226

• Set Exploration (5.2.6). Investigate the consequences of multiple different options

before continuing with a single option.

Você também pode gostar