SPE 97269

Risk and Uncertainty ManagementBest Practices and Misapplications for Cost and
Schedule Estimates
S.K. Peterson, SPE, J Murtha & Assocs.; J. De Wardt, SPE, De Wardt and Co. Inc.; and J.A. Murtha, SPE,
J Murtha & Assocs.

Copyright 2005, Society of Petroleum Engineers

This paper was prepared for presentation at the 2005 SPE Annual Technical Conference and
Exhibition held in Dallas, Texas, U.S.A., 912 October 2005.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300
words; illustrations may not be copied. The proposal must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
Risk management, risk analysis, and uncertainty analysis are
still-growing trends in cost and schedule estimating.
Engineers and managers alike have been lead to believe that
correct application of best practices will ensure that operations
achieve their objectives on time and within budget.
Unfortunately,
a
number
of
misapplications,
misunderstandings, and mistakes have threatened to endanger
the continued useful growth of this trend. Insufficient tools
and / or incorrect use of the available tools have allowed
creation of a false sense of security which is shattered by loss
of objectives and time / cost overruns. It is very important that
the industry understands and chooses the correct applications,
and has realistic expectations.
This paper will present best practices for applying risk
management, risk analysis and uncertainty analysis to capital
expenditure cost and schedule estimates. In addition to
outlining our recommended process, we will highlight current
misapplications that, in our opinion, are potential barriers to
the continued growth of this valuable management tool.
Introduction
The oilfield industry is, by its very nature, an industry that
contains many uncertainties (unknown variables) and risks
(things that can go wrong). Many companies, albeit to
varying degrees and with varying degrees of success, are
currently applying risk management, risk analysis, and
uncertainty analysis for their cost and schedule estimating,
particularly for drilling and completion operations and
facilities in upstream oil and gas operations.
The paper will begin with a brief review of the background
to risk management, risk analysis, and uncertainty analysis. It
will then detail best practices covering risk and uncertainty for
cost and schedule:

Identification the who and the how

Rating and ranking what is important, how to carry
it into quantitative analysis
Probabilistic estimating for times and cost the
ranges that can be expected
Mitigation methodologies how to reduce the impact
of unwanted events.

Background of Risk Management, Risk Analysis and

Uncertainty Analysis
Risk and uncertainty identification, risk and uncertainty
analysis, and decision analysis are all integral to any
responsible and respectable risk management process. A
thorough risk management process combines two defined
forms - qualitative analysis and quantitative analysis.
Qualitative risk analysis defines and frames the scope of
further analysis by providing context of a scenario or story.
Furthermore, qualitative risk analysis includes methods for
prioritizing the risks for action. It is a necessary pre-requisite
for quantitative risk analysis.
Quantitative risk and uncertainty analysis can further be
subdivided into deterministic methods, such as decision trees
or risk matrices, and stochastic methods, such as Monte Carlo
simulation. Quantitative risk analysis analyzes the effects of
the prioritized risks and values assigned to them. As reported
and illustrated in the literature1,2, the preferred method of
Monte Carlo simulation helps to:
assess the probability to achieve project
objectives,
identify realistic and achievable cost, schedule
and scope targets,
determine the apparent best project decision when
some conditions or outcomes are uncertain.
Best practices require the earliest possible identification of
risk / uncertainty and their cataloging in a risk register. This
register is then continuously added to as the project
commences through the concept to definition and finally
planning phase. Simple registers, built in spreadsheet format,
can be readily assessed and ranked. In this manner the design
and planning decisions as well as quantitative analysis can
focus on the ones that have the most impact.
Best practices also require the quantitative model to be
built (in a granularity appropriate for the project phase) and
tested as early as possible. This quantitative model, also built
in a spreadsheet format, should be transparent and adaptable.

Disillusionment and Discontent

Engineers and managers alike have been lead to believe that
correct application of risk management best practices will
ensure that operations achieve their objectives on time and
within budget. This is a noble goal, but not a guarantee.
In reality it is the mitigation actions that improve the
probability of achieving a target cost and schedule and it is the
quantitative analysis that provides the range of expectations
for the outcome. Correct understanding of this range can better
prepare a company for high quality decision-making and
preparedness for the actual project outcome.
Unfortunately, a number of misunderstandings,
misapplications, and mistakes have threatened to endanger the
continued growth of the risk management, risk analysis, and
uncertainty analysis trend. Insufficient tools and / or incorrect
use of the available tools have allowed creation of a false
sense of security which has often been shattered by loss of
objectives and time / cost overruns.
What are some of the misunderstandings?
Quite a few misunderstandings continue to exist in the oil
industry today, in spite of all the publications and training that
are available on this issue. These misunderstandings include:
There can be enough information and/or data to do a
deterministic analysis, but not enough to do
uncertainty analysis.
There must be an agreed deterministic estimate prior
to creating a probabilistic estimate, and that the
deterministic estimate has some statistical relevance.
Uncertainty and risk analysis are only useful once the
program or project is underway and well-defined.
Risk analysis will lead to more accurate estimates.
All risks can, or should, be explicitly listed.
Uncertainty and risk always decrease as project
progress.
A clearer understanding is required across the industry
before the correct application can be consistently achieved.
What are some of the misapplications?
Misapplications appear to occur, or be an outgrowth, from
lack of understanding. If companies are going to reap the
rewards in terms of project delivery, they must rectify this
situation since the following misapplications can become a
cause of economic project underperformance or even failure.
The misapplications include:
Risk registers developed, but then not linked to or
carried over into the quantitative analysis. The
quantitative analysis, consequently, fails to model the
identified situation.
Probability distributions are assigned to every one of
hundreds of line items, just because the software
enables the modeler to do so. This creates a highly
inaccurate analysis that shows far more certainty in
the outcome than is realistic, due to the effect of the
central limit theorem.
Indiscriminate modeling not knowing what
questions the model is meant to answer or what
business decision(s) the answer(s) will be used for.
Such models at best are time-wasters for

SPE 97269

management, and at worst create inaccurate results

with consequent poor decisions.
Performing a quantitative risk analysis once,
checking off a requirement, and then shelving it
rather than using it.
Abandoning an uncertainty and risk analysis because
the desired deterministic estimate falls at an
embarrassing point in the probabilistic distribution.
Determining cost and schedule risk isolated from
production and/or market risk (i.e., only on cost or
time rather than on economics). This sort of analysis
misses the real understanding and value since the
revenue side of the model has been ignored.

What are some of the mistakes?

Risk registers are developed, risks qualified and then
no, or insufficient, action is taken in design and
planning decisions. The result is that these projects
suffer from foreseen problems.
Uncertainties are not acknowledged in the
everything goes right idealization.
Line items are added up incorrectly, i.e. not correctly
probabilistically. One example is combining worstcase or best-case scenarios.
The effects of the central limit theorem are not
accounted for or compensated.
Risk events are not included.
Risk events are included, but erroneously. One error
occurs when the uncertainty ranges on line-item
distributions are inappropriately extended to account
for the events. Another error is when the risk events
are assigned unrealistic probabilities.
Not incorporating correlation, for uncertainties as
well as for risks.
Not using quantitative risk analysis for bid
evaluations.
Best Practices Qualitative Risk Analysis
A story has to have a setting, a theme and a conclusion. So
does the scenario for which we are about to embark on a risk
management exercise that includes risk analysis and
uncertainty analysis. Qualitative risk analysis defines and
frames the scope of further analysis and action by providing
an agreed list of risks that are ranked in order of their impact
within the contextual constraints of the scenario. It is
necessary before quantitative risk analysis is initiated.
When do we start?
The qualitative description must be as detailed and explicit
as possible for the stage of the project. Does this mean that
we dont do risk analysis during the conceptual or scoping
phase? No! Risk analysis is vital to making informed decisions
between alternative approaches or designs. But the level of
detail is different than during the execution phase.
Start early, but not too early. You would like to have the
projects risks identified as input to the project planning, since
they certainly will affect the budget, schedule, and other plans.
In order to do that, you must be able to adequately describe the
project. Smith and Merritt3 presented the Product 5Ws and

SPE 97269

Process 5Ws that would indicate you have sufficient

information to identify risks. Here they are adapted for our
application in upstream oil and gas cost and schedule
estimates.
Project 5Ws
What is the scope for the facility (wells) in terms
of features and functions?
Who needs the facility (wells)?
Why does the company want to build this facility
(drill these wells) now?
Where will the facility be located (the wells be
drilled)
When is the facility (wells) needed?
Process 5Ws
What will be developed or designed internally,
and what will external partners provide?
Who will be developing or designing each
portion?
Where will the facility be built (wells be drilled)?
When is it possible to deliver the facility (wells)?
Why are those alternatives for the above four
Process Ws being considered?
The original scenario and qualitative risk analyses should
be carried through as the project progresses, with some
options or scenarios being discarded, and those that are still
viable alternatives developed in increasing detail and
complexity as definition and scope are committed to.
Whenever there is significant new information, or a welldefined phase is completed, there should be a corresponding
post-mortem and plan-forward update. Likewise, any change
orders and changes of scope will then be highlighted and
documented, so that adjustments to the estimates are
categorized. Addressing change-orders and changes of scope
is of utmost importance in the overall risk management
process, as they cannot be compensated for or included in the
quantitative risk analysis yet they can have a significant
impact on both the uncertainties and the risks.4
Any respectable risk and uncertainty management program
includes a version of a risk register. While we object to recipe
books and detailed prescriptiveness, there are a few details
regarding this key element of the cost and schedule risking
that are worth addressing.
Identification the Who and the How
Listing all the potential risks is a prerequisite to ensuring
identification of the key aspects that can affect project
outcomes. On the one hand, exhaustive lists are viewed as
burdensome while on the other hand, insufficient listing can
miss critical risks that cripple the project. Listing by a broad
group of stakeholders (clients, designers, operators, project
managers, suppliers, etc.) provides the best opportunity for a
quality list. Listing by an individual will miss critical risks in
areas of unfamiliarity; for example, an engineer who is not
familiar with the operation, or a consultant who is not familiar
with the issues in a particular country or region.
It is important to assemble a diversity of opinions and
experience. Obviously this is easier to accomplish in a large

organization; in a smaller organization external resources may

need to be utilized for this important step.
The convention for risk identification is to hold a
workshop, with a brainstorming session. While it is
convention, it is not necessity. Risks can be identified through
individual interviews equally as effectively, and sometimes
more honestly given certain group dynamics. The issue is to
drive the input to ensure the broadest listing such that, when
sorted, everything possible has been considered.
Regardless of whether the risks are identified in a
workshop or through individual interviews, or some
combination, three things are necessary:
1. That each participant has a solid level of project
understanding before the session.
2. That the quest is to identify as many risks as possible.
3. That each risk identified is possible, but not certain.
Early on, the same session can serve to identify risks for
multiple scenarios or options. Often there is significant enough
overlap among alternatives to make a single forum most
efficient.
Rating and Ranking What is Important
We need to be able to differentiate amongst the risks, so as to
whittle the list to a reasonable number, the management of
which gives us the most benefit and the highest probability of
achieving our objectives within the allotted time and budget.
Processing the listed risks is key to managing the front end
of the risk management process. Simple spreadsheets available
to everyone provide sufficient computing power to readily
achieve this once a system has been set in place. Processing
involves assessment and ranking. Risk assessment includes
assigning a probability of occurrence and a distribution of
potential impact. Combining these through a suitable scoring
system provides a qualitative ranking, which can quickly be
sorted to highlight the critical risks for both mitigation and
qualitative analysis.
Critical aspects of risk identification and assessment
include:
maintaining and updating the risk register during
all phases of the project,
creating a matrix of probability and impact that
represents suitable ranges for the specific project
in order to attain a representative ranking.
Our list of critical components to capture in the risk
register differs slightly from those we have seen in the
literature, in that we include parameters that ease the transition
between the semi-quantitative risk register and the fully
quantitative probabilistic cost and schedule estimates.
Risk Identifier
Estimate of risk probability
Risk cost / schedule / functionality impact
Correlation to or with other risks
Risk category
Risk ownership
Mitigation strategies
Mitigation success probability
Cost/Benefit of mitigation
Timing of mitigation

Risks are ranked based on a rating that results from a

combination of the probability of their occurrence and the
potential impact when they occur. The rating is displayed in a
matrix with one axis showing discrete steps in a range of
probabilities that match the project and the other axis a range
of impacts that match the project.
The scale that we use for the probabilities and impacts may
well vary as the project progresses. During screening and
scoping, a resulting 3X3 matrix may be sufficient, whereas
during the construction phase we may want as much detail as a
9X9 matrix. One critical aspect of these matrices is that the
consequence of very high impact events, such as death or loss
of reputation, with a low probability of occurrence are
typically unacceptable. This policy must be reflected through
the use of an asymmetric matrix.
At this stage we must be wary of some common mistakes
with probability and impact estimating. There is a tendency to
overestimate the probability of very low probability events
(rare events) and to underestimate the probability of high
probability events.
Impacts, and their corresponding
uncertainty ranges, are often underestimated by assigning
ranges that are too narrow.
Risks to address should be reduced to a reasonable
number, in line with resources and personnel available. While
a first-level priority assignment can be made from the semiquantitative risk register, the final cut should be made only
after some consideration is given to the quantitative risk
analysis. Some risks will not lend themselves to usual
quantitative methods. The danger, however, in ranking or
prioritizing risks without quantitative assessment is that it is
too easy to ignore low-lows without proper consideration of
correlation, joint probabilities, and mutual exclusivities. Thus
a low probability low impact risk will often be dropped from
the list when in fact it acts as the first domino in a line of
dominos.
Best Practices Quantitative Risk Analysis
It has been recognized for some time that there is a need for
companies to establish guidelines for quantitative risk
analysis.5 Companies still need to address issues of model
design, detail, structure, and sensitivity analysis.
Model Design
Probabilistic cost and schedule models benefit from good
design. A Monte Carlo model is in principle just a worksheet
in which some cells contain probability distributions rather
than values. Thus, one can build a Monte Carlo model by
converting a deterministic worksheet with the help of
commercial add-in software.
Overall, the probabilistic cost model is more complicated
than its deterministic counterpart. It demands more
documentation and more attention to logical structure. Each
iteration of a Monte Carlo model should be a plausible
realization. The purpose of using a range of values for each
input is to acknowledge the realm of possibilities and the risk
events. Thus, once each of the input distributions is sampled,
the resulting case should be sensible, something an expert
would agree is possible.

SPE 97269

Model Detail
The probabilistic model needs to be built to the appropriate
level of detail for the understanding of the project and to meet
the objectives of risk and uncertainty management - to
encompass the range of uncertainty in order to make better
business decisions and to identify risks so that we can manage
and/or mitigate them. The risk analysis model, however, is
not meant to replicate step by step procedures for operations
(such as drill well on paper exercises).
It is critical that the industry addresses the fact that the
appropriate level of detail in the probabilistic model is NOT
the same as the appropriate level of detail for operational
project management. Why not? With more and more line
items as distributions, we are exposed to the effects of Central
Limit Theorem which fairly quickly reduces the spread of the
result until the results are clearly in conflict with our
experience of what a reasonable spread on the result should
be.6 We can counter the effects by (1) reducing the number of
distributions (2) introducing correlation between line items,
and/or (3) introducing risk events and their impacts.
Risk analysis will not ensure that every risk encountered
will have been identified beforehand. On the other hand, we
can only plan our mitigation strategies for the risks that have
been identified. In this aspect, model detail requires judicious
sprinkling of risks, sometimes combined from independent
risks in the same category or from risks resulting from a
common cause. Contingencies are then built up methodically
and allocated to categories just as the risks are categorized.
Model Structure
Specify all key equations
Some models for cost estimates, where total cost is just an
aggregation of line items, are very simple. The line items
themselves may have underlying calculations that are
important. Other models have complex structure, such as
alternative development plan models, or intricate timing
issues. While some aspects are routine, features unique to the
problem at hand should be stressed.
List existing alternative models
Sometimes there are two or more models that achieve
much the same objective. Comparing the model at hand with
other familiar models is often useful.
List other projects that use or have used this model
Knowing that other projects have used a model adds
credibility and opens the opportunity to learn methods and
applications. While there may be dozens or even hundreds of
analogous models, the prototype need be mentioned only
once.
List all assumptions and key deterministic inputs
For example:
Two successful wells are necessary before field is proved;
Reservoir fluid data will not be available for x months;
If field size exceeds 100 Bcf, then a second platform is
needed;
Steel price is locked according to contract;
Pipeline has maximum capacity of 50,000 Bbl/day;
All reserves must be produced within 15 years.
Other key deterministic values should be highlighted, such
as interest rate = 10.5%, start time = 01 Jan 2006, facility
design life = 10 yr.

SPE 97269

List all input distributions: type, defining parameters,

basis for choosing this distribution, and correlations.
Cost and schedule models often have a new distribution for
each line item of prices and expenses. While there may be
dozens or even hundreds of distributions, the prototype need
be mentioned only once.
Each distribution should be identified by type (e.g.,
normal, lognormal, triangular, pert) and by defining
parameters (mean and standard deviation, or min, mode, max).
Moreover, a notation should explain why the particular
distribution was chosen: empirical data that was fit by
software, experience, or fundamental principal. The
justifications should be brief, especially when the particular
choice of distribution is not critical to the results. In case
other distributions were tested, there should be a comparison
between the results available if needed.
Each distribution should also be defined as uncorrelated
(independent) or correlated. If a distribution is correlated,
denote to which other variable, whether the correlation is
positive or negative, and how strong the correlation is. Again,
a notation should briefly justify why that correlation and
strength was chosen for this particular model data analysis,
expert opinions, fundamental principals, or some combination
of those reasons.
If data is used, that data should be accessible for review. In
the cases where expert opinions dominated the choice of data,
the interviewing technique (individual or workshop
environment) and the experts involved should be documented.
Finally, risk probabilities and uncertainty distributions
should be further documented and supported by direct
reference to the then-current risk register.
Model Sensitivity Analysis
Once the model is built, it should be run and re-run many
times with options to check the sensitivity. Tornado charts and
overlays of the model results are essential. The model results
will help direct the decision-making for risk mitigation and
risk management strategies.
1. First, consider the base model as the model with no
risk events, only the uncertainty that still exists if
everything goes right. We first interrogate this model
to find the stages or phases where we could focus our
engineering or contractual personpower to affect the
greatest benefit.
2. Next, we run the model with the risk events that are
thought to be manageable, even if they cannot be
mitigated 100%. This risk event sensitivity will
highlight and rank those events that most significantly
impact our cost or schedule.
3. The following runs of the model should layer on the
risk mitigation strategies, with their respective
probabilities of success and mitigation effectiveness.
In so doing, we can further highgrade our list of where
to focus attention that will have the highest payoff.
4. Repeat steps 2 and 3 using the base model with the
risk events that are thought to be beyond our
influence. Determine if the hierarchy of uncertainties
or mitigatable risks changes.
Finally, from the above sensitivities and the time and cost
percentiles, we will have a consistent and defensible way to

provide contingency allocation and accountability. We will

not, however, have accounted for all risks; there will still be an
additional contingency component of the non-explicit risks
that is necessary to consider.
Best Practices The Life of the Project
From the time of project scoping, risks are identified,
uncertainties are quantified, and a comprehensive analysis is
performed. The level of detail may be quite broad, and the
resulting ranges of times and costs may be quite broad. But
they should reflect the professionals understanding and
perceptions to date.
As a project moves forward, the responsible and
respectable risk management process insists that the
qualitative and quantitative (probabilistic) risk analyses are
revisited and updated. New risks and uncertainties will be
identified as awareness increases or changes to scope are
ordered. In our quantitative analysis, this will affect the
distributions of cost and/or time, thus broadening our
confidence interval.
Likewise, as a project moves forward, some of the old
risks and uncertainties will become obsolete, either because
they were managed out of the system or because they were
hobgoblins. In our quantitative analysis this will also affect
our distributions for cost and/or time, reducing the range of
our confidence interval.
There is not a simple plus or minus percentage for the
ranges that should be expected at a given stage of a project.
The appropriate range is a reflection of the uncertainty and
riskiness of the project.
An assessment on the quality of our estimates can be done
after enough projects (ie, enough for statistical significance)
that we can tell if 10% of the projects are coming in less than
their respective p10 estimate and 10% of the projects are
coming in more than their p90 estimate.
At the conclusion of a single project, it is difficult to
comment on the quality of the estimate. Risk and uncertainty
management may increase our accuracy by removing biases
from the process, but the true intent is to encompass the range
of uncertainty in order to make better business decisions and
to identify risks so that we can manage and/or mitigate them.
Success Stories
Deepwater projects in the Gulf of Mexico have presented new
challenges for the industry and an even stronger focus on
achieving predictable performance due to the large cost
involved. De Wardt7 describes the success achieved with one
of these projects. Risk and uncertainty management was an
integral part of the planning and execution process from an
early stage. In spite of significant challenges, the well
achieved best in class performance for deepwater Gulf of
Mexico.
Mature fields often present significant risks for infill
drilling. In one such example8, the full application of risk and
uncertainty management realized a reduction in cost from $2.5
to $1.8 per bbl of oil. This North Sea example encountered
many challenges ranging from active faults with gas
communication in the over burden, depleted pressure regions
along extensive horizontal intervals, active shales in the
reservoir seal as well as active faulting in the chalk reservoir.

SPE 97269

These wells typically carry some seven contingency plans

ready to address the uncertainties as they arise. The example
described in the reference realized the improved results
through a reduction in drilling time as well as a reduction in
well cost.
Conclusions
Applying the best practices for risk and uncertainty
management ensures that the cost and schedule estimates are
consistent (reproducible) and unbiased, have the appropriate
level of detail for the level of understanding of the project,
encompass the full interactions and interdependencies among
the disciplines / components involved, and are welldocumented.
Furthermore, correct application enables the team to
identify the critical risks and address them through re-design,
alternative plans or contingencies. Engineers and managers
with expectations aligned with the capabilities of uncertainty
and risk analysis will not be disappointed, and will be
equipped for decision-making.
The projects that have
correctly applied risk and uncertainty management have
consistently out performed those that have not, leading to
reduced costs and increased value for the oil company.
Unfortunately, misunderstandings and misapplications are
creating a false sense of surety, which is dashed as projects
fail to deliver to expectations. In the oil business today, it is
critical that projects deliver to expectations in order that
companies can meet their financial projects and can grow. It
is, consequently, critical that companies understand the correct
application of risk and uncertainty management and apply it
rigorously.
Acknowledgements
The authors wish to acknowledge J. Murtha and Associates
and DE WARDT AND COMPANY INC for their permission to
publish this paper.
The authors also wish to recognize the many team
members who have willingly taken on the risk and uncertainty
management process since it is these team members who
ultimately make the difference.
References
1.

Peterson, S.K., Murtha, J. A., Schneider, F.F.; "Risk Analysis

and Monte Carlo Simulation Applied to the Generation of
Drilling AFE Estimates," paper SPE 26339, presented at the
68th Annual Technical Conference and Exhibition, Houston, 3-6
October 1993.

2.

Peterson, S.K., Murtha, J. A., and Roberts, R.W.: Drilling

Performance Predictions: Case Studies Illustrating the Use of
Risk Analysis, paper SPE 29364 presented at the SPE/IADC
Drilling Conference, Amsterdam, The Netherlands, 28 February
- 02 March 1995, JPT, June 1995.

3.

Smith, P.G., and Merritt, G.M., Proactive Risk Mangement:

Controlling Uncertainty In Product Development, Productivity
Press, 2002.

4.

Peterson, S.K., and Pearce, D.W.; The Effect of Unplanned

Operations on Drilling Performance Evaluation, paper
SPE/IADC 25761 presented at the SPE/IADC Drilling

Conference, Amsterdam, The Netherlands, 23 25 February

1993.
5.

Murtha, J.A.; Monte Carlo Simulation: Its Status and Future,

paper SPE 37932 presented at the 1997 SPE Annual Technical
Conference and Exhibition, 05 08 October 1997, San
Antonio, Texas.

6.

Murtha, J. A.; Sums and Products of Distributions: Rules of

Thumb and Applications, paper SPE 77422 presented at SPE
Annual Technical Conference and Exhibition, 29 September 02 October 2002, San Antonio, Texas.

7.

de Wardt, J.P.; Deepwater Success Through Predictable and

Distinctive Drilling and Completion Performance, paper SPE
87117, presented at the IADC/SPE Drilling Conference, March
2004, Dallas, Texas.

8.

de Wardt, J.P., Cook, A., Smook, R.W.; Step Change

Improvement in Drilling Performance, Repeatable Worldclass
Performance is Possible, paper IADC/SPE 59203, presented at
the IADC/SPE Drilling Conference, February 2000, New
Orleans, Louisiana.