Escolar Documentos
Profissional Documentos
Cultura Documentos
Biegler
\
Chapter 4. Equipment Sizing And Costing
In the previous chapter we developed the tools for a preliminary mass and energy balance
of our candidate flowsheet. This task provided us with important data for economic
evaluation of the process. In this chapter we will build on these concepts and pursue the
next step of determining equipment sizes, capacities and costs. As in the previous
chapter, we will use approximations in order to perform the calculations quickly and
establish qualitative trends for screening process alternatives. In particular, direct,
noniterative correlations will be applied for equipment sizing and a well established
method developed by Guthrie (1969) will then be used for costing this equipment. With
this information, we are then able to complete an economic analysis, which will be
discussed in the next chapter.
4.1 Introduction
• Physical sizing of equipment units. - This includes the calculation of all physical
attributes (capacity, height, cross sectional area, pressure requirements, materials of
construction, etc.) that allow a unique costing of this unit.
1
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
• Cost estimation of the unit - Here the sized equipment will be costed using power
law correlations developed in Guthrie's article (Chemical Engineering, March, 1969,
p. 114). In addition to unit capital costs we will also consider the operating costs such
as utility charges. This information, together with the feedstock costs and product
sales will be used in the subsequent economic evaluation.
In the remainder of the chapter we will consider sizing and cost models for all of the
process units analyzed so far. The next section will develop shortcut correlations for the
sizing of these units. Section 3 will then describe Guthrie's cost estimation as applied to
these units. Both sizing and costing will be illustrated by numerous examples. Finally, the
last section will summarize the chapter and set the stage for the economic analysis in
Chapter 5.
This section presents an overview of quick calculations for equipment sizing. Basic
procedures will cover the following units:
• vessels
• heat transfer equipment
• columns, distillation and absorption
• compressors, pumps, refrigeration
All of these calculations require flowrates, temperatures, pressures and heat duties from
the flowsheet mass balance and these sizing calculations will determine the capacities
needed for the cost correlations developed in the next section. In addition, we will
develop the concept of material and pressure factors (MPF) used to evaluate particular
instances of equipment beyond a basic configuration. This concept is an empirical factor
developed by Guthrie as part of the costing process. As shown in section 3, the MPF
multiplies the base cost in the evaluation of the final equipment cost.
Vessel Sizing
Vessels include flash drums, storage tanks, decanters and some reactors. Unless specified
otherwise by particular unit requirements, these will be sized by the following criteria.
a) Select vessel volume (V) based on a five minute liquid holdup time with an equal
volume added for vapor flows. Thus the formula is given by:
V = 2 [FL τ /ρL]
where FL is the liquid flowrate leaving the vessel (as in a flash drum), ρL is the liquid
density and τ is a residence time, typically set to five minutes.
2
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
• For general costing purposes, the aspect ratio, L/D, will be assumed to be four. (This
is the optimal ratio if the bottom and top caps are four times as expensive as sides).
• If diameter is greater than four feet (1.2 m), size the unit as a horizontal vessel (This
requires more space but less cost for structural support.)
• As a safety factor choose the vessel (gauge) pressure to be 50% higher than the actual
process pressure from the mass and energy balance. From this we also observe the
appropriate pressure factors in Guthrie's method when costing the vessel.
• For the desired temperature range, consider the required materials of construction as
shown in Table 1. Observe the appropriate material factors in Guthrie's method
when costing the vessel.
In Guthrie, the basic configuration for pressure vessels is given by a carbon steel vessel
with a 50 psi design pressure, and average nozzles and manways. For vertical
construction this inludes the shell and two heads, the skirt, base ring and lugs and
possible tray supports. For horizontal construction this includes the shell and two heads
and two saddles. The material and pressure factor for various types of vessels is given in
Table 4.2. In addition, various types of vessel linings are costed in Guthrie (1969, Figure
5).
3
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Table 4.2 Guthrie Material and Pressure Factors for Pressure Vessels
MPF = Fm Fp
Consider the countercurrent, shell and tube heat exchanger shown below:
T1
T1 T2
Heat Exchanger T2
T
Q (Btu/hr) t2
t2 t1
t1
Q, heat exchanged
Figure 4.1. Heat Exchanger Temperatures
Sizing equations for these heat exchangers can be found from the following equation:
Q = UA ∆Tlm
where Q is the heat duty, known from the energy balance, A is the required area, the log
mean temperature (∆Tlm) is given by :
∆Tlm = [(T1 - t2) - (T2 - t1)]/ ln{(T1 - t2)/(T2 - t1)}
and the overall heat transfer coefficients can be estimated from Table 4.3. Again for
sizing and costing, we need to observe the design criteria for temperature and pressure
(Prated = 1.5 Pactual) and observe the appropriate pressure and material factors in
costing the exchanger.
4
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
5
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Gas-liquid media
Vaporizers
Note that phase changes in heat exchangers lead to changes in U and need to be
considered more carefully. In this case, we split the exchanger into serial units and, as
shown in Figure 4.2, calculate U and A for vapor media and for condensing media
separately. Thus the total area is given by:
Avap = Qvap / {Uvap [(T1 - t2) - (Tc - t3)]/ ln{(T1 - t2)/(Tc - t3)}}
Acon = Qcon / {Ucon [(Tc - t3) - (T2 - t1)]/ ln{(Tc - t3)/(T2 - t1)}}
and Atotal = Avap + Acon
Finally we choose 5000 ft2 (or ~ 500 m2) as the maximum exchanger area. If more heat
exchange area is required, we simply use multiple heat exchangers in parallel.
6
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
T1
Tc
T T2
t2
t1
t3
Qvap Qcon
Uvap Ucon
A vap Acon
Figure 4.2. Sizing Heat Exchangers with Intermediate Phase Changes
While this simplified method is adequate for preliminary designs, we note that detailed
sizing of heat exchangers is much more complicated. See Welty, Wicks and Wilson
(1984) and Peters and Timmerhaus (1980) for a more detailed treatment on the sizing of
heat exchangers. In Guthrie, the basic configuration for heat exchangers is given by a
carbon steel floating head exchanger with a 150 psi design pressure, and this includes
complete fabrication. The material and pressure factor for various types of heat
exchangers is given in Table 4.4.
Table 4.4 Guthrie Material and Pressure Factors for Heat Exchangers
MPF = Fm (Fp + Fd)
Design Type Fd
Kettle Reboiler 1.35
Floating Head 1.00
U tube 0.85
Fixed tube sheet 0.80
Shell/Tube Materials, Fm
Surface Area (ft2) CS/ CS/ CS/ SS/ CS/ Monel/ CS/ Ti/
CS Brass SS SS Monel Monel Ti Ti
7
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Capacities and sizes for furnaces and direct fired heaters will not be obtained directly for
a preliminary design. Instead we follow Guthrie and base the cost of these units on the
heat duty. Observe that pressure and material factors, as well as design types, still need to
be considered for costing. Here the basic configuration for furnaces is given by a process
heater with a box or A-frame construction, carbon steel tubes and a 500 psi design
pressure. This includes complete field erection. The material and pressure factor for
various types of furnaces is given in Table 4.5.
Design Type Fd
Process Heater 1.00
Pyrolysis 1.10
Reformer 1.35
(without catalyst)
Similarly, in Guthrie the basic configuration for direct fired heaters is given by a process
heater with cylindrical construction, carbon steel tubes and a 500 psi design pressure.
This also includes complete field erection. The material and pressure factor for various
types of direct fired heaters is given in Table 4.6.
Table 4.6 Guthrie Material and Pressure Factors for Direct Fired Heaters
MPF = Fm + Fp + Fd
Design Type Fd
Cylindrical 1.00
8
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Dowtherm 1.33
Reactors
For reactor sizing we assume a given space velocity (s in hr-1) based on a liquid or gas
molar flowrate, µ. Then we have:
s = 1/τ = µ/ (ρ Vcat)
where ρ is the molar density at standard temperature and pressure (1 atm, 273 K) and
Vcat is the volume of catalyst. The total volume, V, is then calculated based on the void
fraction, ε, of the catalyst (assume 50%). In this case, we have:
V = Vcat /(1-ε) = 2 Vcat
Depending on reactor conditions, we can then cost the reactor as a pressure vessel, heat
exchanger or furnace. Also, for these units use the appropriate material and pressure
factors in Guthrie's method.
Distillation Columns
To apply costing for distillation columns, we need to calculate the height, diameter and
number of trays in the tray stack. In particular, tray stacks are defined in Guthrie with the
basic configuration given by a 24" tray with carbon steel of either plate, sieve or grid
type. This includes all fittings and supports. The material and pressure factors for various
types of tray stacks are given in Table 4.7.
Table 4.7 Guthrie Material and Pressure Factors for Tray Stacks
MPF = Fm + Fs + Ft
Tray Type Ft
Grid (no downcomer) 0.0
Plate 0.0
Sieve 0.0
Valve or Trough 0.4
Bubble Cap 1.8
Koch Kascade 3.9
9
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Tray Spacing,Fs
24" 18" 12"
Fs 1.0 1.4 2.2
Tray Material Fm
Carbon Steel 0.0
Stainless Steel 1.7
Monel 8.9
However, in order to cost the vessel, tray stack and heat exchangers, we first need to
calculate the number of theoretical trays and the reflux ratio. As discussed in Chapter 3,
shortcut calculations for these can be performed through the Fenske equation (for
minimum number of theoretical trays), the Underwood equation (for minimum reflux
ratio) and the Gilliland correlation that allows us to obtain the actual reflux ratio and tray
number. Using the Underwood equation, however, involves an iterative procedure and
will be deferred to Part III. Instead, we apply a simple, direct correlation that has been
developed for (nearly) ideal systems. This allows us to determine qualitative trends
rapidly for preliminary design.
a) From the mass and energy balance calculations we have the relative volatilities
and the top and bottom recoveries: αlk/hk , β lk = ξlk, β hk = 1- ξhk.
b) Calculate tray number and reflux ratio from the following correlation:
Ni = 12.3/{(αlk/hk - 1)2/3 (1 - β i )1/6} and Ri = 1.38/{(αlk/hk - 1)0.9 (1 - β i )0.1 }
for both i = lk, hk. Then the number of theoretical plates is:
NT = γN maxi(Ni) + (1 - γN) mini(Ni)
and the reflux ratio is:
R = γR maxi(Ri) + (1 - γR) mini(Ri)
where γR and γN are arbitrary weights (set to 0.8).
The number of actual trays is given by NT/η where the efficiency (η) is assumed to be
80%. Also, assuming a two foot (0.6 m) tray spacing, the tray stack height is easily
calculated. Finally, we choose a maximum height of 200 feet (60 m). A larger calculated
height will require that the column be split into two with liquid and vapor flows running
between them.
10
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
From the reflux ratio and the state of the feed, we can now calculate flow rates in the
distillation column. Based on these flowrates and relationships for flooding velocity we
can then calculate the diameters.
a) The figure below illustrates the flowrates for two feed conditions. For two phase feed
the flowrates can be calculated in an analogous manner.
D V=L+D D
L=RD L=RD
L
L
F F
V=L+D
L+F V-F
B
B
Bubble point feed Dew point feed
b) To determine the diameter we design the column to run at 80% of the flooding
velocity. At the flooding velocity, the vapor flowrate is so high that no net liquid flow
occurs and entrainment begins. Flooding relations are represented in Figure 4.4.
11
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
0.5
36"
0.4
24" Tray Spacing
0.3
18"
0.2 12"
9"
0.1
.07
Csb,f
.05
0.01 0.02 0.03 0.05 0.07 0.1 0.2 0.3 0.5 0.7 1.0 2.0
Here we define Unf as the linear flooding velocity (in ft/sec) and from Figure 4.4, Unf is
given by:
Unf = Csb,f ((ρ l - ρg)/ρg)0.5 (20/σ)0.2
ρ g and ρ l are the gas and liquid mass densities, respectively, and σ is the liquid surface
tension in dynes/cm. (For a rough guess, we use σ = 20 dynes/cm). Douglas (1988)
provides a simplification of the diameter calculation by noting that the Csb remains fairly
constant for Flv values of 0.01 to 0.2, a fairly wide range. Also by noting that σ = 20
dynes/cm for hydrocarbons and (ρl_ρ g) ~ ρl , the flooding velocity is given by:
Unf = Csb (ρ l/ρg)0.5
For a typical 24" tray spacing, Csb is about 0.33 ft/s in this range. The column diameter
is then given by the cross sectional area:
A = π D2/4 = V'/(0.8 Unf ε ρg)
where ε is the fraction of the area available for vapor flow (about 0.6 for bubble cap
trays, 0.75 for sieve trays). We allow the maximum column diamter to be 10 ft. (3 m).
Any larger calculated diameters require the column to be split into two columns run in
parallel.
12
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
For a total condenser, we know from the energy balance that Qcond = HV - HL, where HV
and HL are the total stream enthalpies for the vapor and liquid streams around the
condenser. The reboiler duty can then be calculated either directly from the vapor flow or
from a total energy balance around the column.
In order to apply Guthrie's method and obtain the costing information, we now need only
to group the capacities of the following components: the empty vessel, the tray stack and
the heat transfer equipment (condensers and reboilers). The overall sizing procedure is
best illustrated with a small example.
Example:
Consider the separation of acetone and water at 100 kPa as shown in the figure below:
T = 385 K
0.1 gmol/s Acetone
329.7 gmol/s Water
At T = 368 K we have a relative volatility, α A/W = 3.832 and key component overhead
recoveries of ξA = 0.995 and ξW = 0.001 (βA = 0.995 and βW = 0.999). From the tray
number correlations we have:
NA= 12.3/{(3.832 - 1)2/3 (1 - 0.995)1/6}= 14.85
NW= 12.3/{(3.832 - 1)2/3 (1 - 0.999)1/6 } = 19.42
and NT = 0.8(NW) + 0.2NA = 18.51
with the actual tray number N = NT = 18.51/0.8 = 23.13 ~ 24
The reflux ratio is calculated by:
RA = 1.38/{(3.831 - 1)0.9 (1 - 0.995)0.1 }= 0.92
RW = 1.38/{(3.831 - 1)0.9 (1 - 0.999)0.1 } = 1.08
with R = 0.8 RW + 0.2 RA = 1.047
13
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
A rough condenser and reboiler sizing can be done by noting the flowrates and, from
handbook data, the heats of vaprozation at 329 K.
D = 20.2 gmol/s
L = R D = 1.047 (20.2) = 21.2 gmol/s
V = L + D = 41.4 gmol/s
∆HvapA = 29.1 kJ /gmol
∆HvapW = 43.0 kJ /gmol
As a result, the condenser duty is:
Qcond = (41.4/20.2)[19.9(29.1) + 0.3(43.0)] = 1215 kW
At the bottom we have V' = 41.4 gmol/s and at 385 K the heat of vaporization of water is
40 kJ/gmol. Assuming (almost) pure water in the bottom stream, the reboiler duty is:
Qreb = 41.4 (40) = 1657 kW.
Now the column diameter can be calculated separately for the top and bottom of the
column. For the bottom section below the feed we first compute liquid and vapor
densities. ρ L = 106 g/m3 and assuming ∆p = 50kPa then from the ideal gas law, we have
ρ g = PM/RT = (150 kPa)(18 g/gmol)/(8.314 J/gmol K)(385 K) = 845 g/m3
This does not satisfy the assumptions of the simplified correlation, so from the flooding
curve we calculate the abscissa, the dimensionless flow parameter:
Flv = (L'/V')(ρg/ρL )0.5 = 0.260
and for 24" tray spacing we obtain the ordinate, the capacity parameter C sb is about 0.25.
Here:
14
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Absorbers
Column sizes for absorbers are calculated in a similar manner as for distillation columns.
However, here NT is derived from the Kremser equation and we use a very low efficiency
(20%) as equilibrium on a tray is a very poor assumption. Height and diameter for the
vessel and the tray stack are costed in the same way as for distillation columns. Again,
we assume a 24 in tray spacing. Packed columns can also be costed using the data in
Guthrie. Information on the costs of various packings is given in Figure 5 of Guthrie
(1969).
Pumps
For pumping (pressure increases) in liquids we define the theoretical work as V ∆p, since
the specific volume remains (nearly) constant. The brake horsepower can be written as:
Wb = µ (p2-p1)/(ρ ηp ηm)
where η p is the pump efficiency (assume to be 0.5) and ηm is the motor efficiency
(assume to be 0.9).
In particular, centrifugal pumps are defined in Guthrie with the basic configuration given
cast iron unit operating below 250 oF and a suction pressure of 150 psig. This includes
the driver and coupling as well as the base plate. The material and pressure factor for
15
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
centrifugal pumps is given in Table 4.8. Cost correlations for reciprocating pumps are
also given in Guthrie (1969, Figure 7).
Material Type Fm
Cast Iron 1.00
Bronze 1.28
Stainless 1.93
Hastelloy C 2.89
Monel 3.23
Nickel 3.48
Titanium 8.98
Operating Limits,Fo
Max Suction Press. 150 500 1000 (psig)
Max Temperature 250 550 850 (oF)
Fo 1.0 1.5 2.9
Compressor Turbine
µ µ
W W
For an adiabatic compressor, the ideal compression work can be calculated from the
change in enthalpy:
W = µ [HV (P2, T2) - HV (P1, T1)]
16
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
where µ is the molar flowrate (e.g., gmol/s) and the gas enthalpy is given by HV.
Assuming an ideal system, this equation can be written as:
W = µ Cp (T2 - T1) = µ (γ/(γ-1))R (T2 - T1) (for ideal gas)
where Cp is the constant pressure heat capacity, γ = Cp/Cv is 1.4 for an ideal system and
the gas constant, R = 8.314 J /gmol K. Assuming an ideal, isentropic, adiabatic expansion
we can calculate T 2 from the pressure ratio (P 2/P1) using
T2 = T1 (P2/P1) (γ-1)/γ
and substituting back into the above expression gives the theoretical power for an ideal
system:
W = µ (γ/(γ-1))R T 1 [(P2/P1) (γ-1)/γ -1]
MPF = Fd
Design Type Fd
Centrifugal/Motor 1.00
Reciprocating/Steam 1.07
Centrifugal/Turbine 1.15
Reciprocating/Motor 1.29
Recip./Gas Engine 1.82
Staged Compressors
17
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Staged compressors are useful to perform a given service (a desired increase in gas
pressure) with less work. This is accomplished by allowing intercooling after each
compression stage. As a result, for ideal systems the compression work required per
p1
mole is illustrated in the figure below from W = ∫ V dp.
p2
p1 p p2
Figure 4.5. Work required ( ∫ V dp) for different compression sequences
Note that isothermal compression requires constant heat removal to keep the system at
T 0. While it is physically unrealistic, it is easy to see that isothermal compression is a
limiting case of staged compression (as the number of stages goes to infinity).
Intercooling in staged compression requires the following configuration:
. . .pN-1
p0 p1 p1 p2 p2 p3 pN
Ta T0 Ta T0 Ta T0 Ta
T0
Moreover, for a fixed number of compressors, N, it can be shown that the minimum work
occurs when all compression ratios are equal, i.e:
p1 / p0 = p2 / p1 = p3 / p2 = p4 / p3 = ... = pN / pN-1 = (pN / p0)1/N
and the work required is given by:
W = µ N (γ/(γ-1))R T0 [(PN/P0) (γ-1)/γΝ -1]
and as N --> ∞, we obtain the expression for isothermal work:
18
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
W = µ R T0 ln (PN/P0)
We now see that there is a trade-off that must be dealt with in staged compression.
Minimum work is obtained as the number of stages becomes large, but this leads to
unrealistic capital costs. On the other hand, maximum work occurs with a single
(adiabatic) compression stage. Rather than finding the optimum number of stages, which
is case specific, for preliminary designs we invoke a rule of thumb, that the compression
ratio will be (pi/pi-1) = 2.5 (A practical limit for centrifugal compressors is about five.)
Having established these relations, let's see how they work in a small example.
Find the work to compress 10 gmol/s of an ideal gas at 298 K from P0 = 100 k Pa to PN
= 1500 kPa using adiabatic compression, isothermal compression and staged
compression.
For adiabatic compression, we have
W = µ (γ/(γ-1))R T 0 [(PN/P0) (γ-1)/γ -1] = 101.26 kW
For isothermal compression we have:
W = µ R T0 ln (PN/P0) = 66.86 kW
And for staged compression we choose a compression ratio at approximately 2.5. Thus,
the number of stages is derived from:
(PN/P0)1/N = 2.5
So, N ~ 3 and (p3/p2)= (p2/p1) = (p1/p0) = 2.47. From the staged relation we have:
W = µ N (γ/(γ-1))R T0 [(PN/P0)(γ-1)/γΝ -1] = 76.53 kW.
Finally, for staged compression the outlet temperature from each compressor is:
Ta = (P1/P0) (γ-1)/γ T0 = 386 K
and the heat duty required for each exchanger is:
Qint = µ Cp (T1 - T0) = 25.6 kW.
Reciprocating Compressors
19
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
4 3
intake
1 - 2, open intake
P 2 - 3, compress
V 3 - 4, exhaust
1 2 4 - 1, expand
exhaust
V
Figure 4.7. Compression Cycle for Reciprocating Compressor
As can be seen from the compression cycle, theoretical power can be calculated from:
W = µ (γ/(γ-1))R T 0 [(PN/P0) (γ-1)/γ − 1] /[1 - (c(p2/p1)1/γ - 1)]
and we assume a compression efficiency of γc = 0.9. Selecting among centrifugal and
reciprocating compressors depends on the gas flowrate and the desired pressure increase.
A discussion on appropriate selection regimes is given in Perry's Handbook (1984).
Refrigeration
If a process stream needs to operate below about 300 K, some sort of refrigeration is
required and a refrigeration cycle needs to be considered. Often, refrigeration can be
"purchased" from an off-site facility, if available. Otherwise, a separate refrigeration
facility needs to be constructed. In either case, refrigerating a stream is far more
expensive (on an energy basis) than lowering the temperature with cooling water, or even
raising the temperature with steam. Consequently, refrigeration is generally not a
desirable option and other process alternatives should be considered first.
Qc. cooling water
2
Condenser
Qc'
Compressor
Valve
3 2
1 P 1
3
W.
4
Evaporator Q' W'
Tcold 4
Q. process stream ∆H
Given that a refrigeration system needs to be designed, we first consider the refrigeration
cycle and the pressure-enthalpy diagram pictured above. Here Q is the heat absorbed
from the process stream at a subambient temperature and Q' is the heat per unit mass of
20
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
refrigerant. As with staged compression, we see from the diagram that there is a tradeoff
between capital and operating costs in choosing the number of refrigeration cycles. Here
a single cycle requires the maximum work and Q c while a large number of cycles requires
minimum work and Qc. To relate the W and heat rejected for refrigeration (Q), we define
a coefficient of performance, CP = Q'/W'. As with staged compression, we apply a rule of
thumb and select CP ~4 for design purposes. Thus, in a typical cycle, we have:
W = Q/4, and Qc = W + Q ~ 5/4 Q
and for the compressor driven with an electric motor,
Wb = W / ηm ηc = W / 0.72
In order to analyze multiple cycles and choose refrigerants for each cycle, we first
consider the following temperature constraints:
a) Refrigerant (R) must remain below its critical pont in the condenser, say:
Tcond,max = 0.9T cR.
If cooling water is used here, then we also know that:
Tcond,max > TCW (~ 300 K) + ∆Tmin
b) In the evaporator, refrigerant and pressure should be chosen so that T evap>
Tboil,R. Also, the evaporator pressure should be chosen greater than 1 atm. to
avoid air leaks into the evaporator.
c) Finally, we choose ∆Tmin ~ 5K, for both the evaporator and condenser heat
exchangers.
Example
Suppose we want to cool air as a process stream to 180K. Consider the refrigerants:
R Tboil(K) 0.9Tc(K)
Ethylene 169 254
Propane 231 332
We know that ethylene will go down to 180 K but not up to 300 K. The opposite holds
for propane. Therefore, we need at least two stages; one propane, one ethylene.
Stage 1
Here R = propane and we obtain the following cooling curves:
21
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
300 K
W.
Evaporator
Q. Ethylene
< 254 K
240 K
240 K ethylene
Propane 235 K
Stage 2
Here R = ethylene and we obtain the following cooling curves:
Qc. propane
Condenser
Propane 235 K
W.
Evaporator
Q.
Air
180 K
22
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
With this rule of thumb,we generally have ∆T = 30K/cycle for an ideal system. Thus if
Tcold = 180, need about (300-180)/30 = 4 cycles. This method gives us a quick way of
estimating utilities Qc (cooling water duty) and Wb (electricity). Similarly, capital costs
for the refrigeration cycle (heat exchangers and compressor) could be determined using
Guthrie's method.
MPF = Ft
Evaporator Temperature Ft
40oF/ 278 K 1.0
20oF/ 266 K 1.95
0oF/ 255 K 2.25
-20oF/ 244 K 3.95
-40oF/ 233 K 4.54
Example
From the handbook we have ∆Hvap = 9.336 kJ/gmol for ethylene so that the total duty is
Q = 4668 kW. The number of stages is:
N = [300-240]/30 = 2 stages
W = [(5/4)2 - 1] Q = 2626 kW
Wb = [W/0.72] = 3647 kW
Qc = [(5/4)2] Q = 7294 kW
Assuming the following data and prices:
23
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
For preliminary design calculations, we note that equipment costs (C) increase
nonlinearly with equipment size (S) or capacity. This behavior can often be captured with
a power law expression: C = C0(S/S0)α, where the exponent is less than one, often about
0.6 to 0.7, and S0 and C0 are the base capacities and costs, respectively. This nonlinear
cost behavior is reflected in an economy of scale, where the incremental costs decrease
with larger capacities.
Why is this the case? For pressure vessels, for instance, the service capacity depends on
volume, while the cost depends on the weight of the metal (surface area). For examples,
for a spherical vessel, we have:
V = π/6 D3 and W = ρM t (π D2)
where t is the vessel thickness and ρM is the metal density. In terms of volume, we have:
D = (6V/π)1/3 and W = ρM t (π1/3 (6V)2/3)
with the vessel cost as: C ∝ W = k V2/3.
For cylindrical pressure vessels, we adopt a more general form used by Guthrie: C = C0
(L/L0)α (D/D0)β . Correlations for pressure vessels are given in Table 4.11. Guthrie also
considers separate correlations for storage vessels of various geometries.
In Guthrie (1969), costs are plotted on charts with log - log scales so that C = C0(S/S0) α
is represented as log C = log [C0/S0α ] + α log S. Note that the slope is given by the
exponent α. Deviations on costs on some units vary by about 20% in Guthrie. However,
for preliminary design, we will only use the median data. Data for the correlations taken
from Guthrie are given in Table 4.12.
24
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
1 ≤ D ≤ 10 ft, 4 ≤ L ≤ 100 ft
The cost data in Tables 4.11 and 4.12 are given in terms of mid 1968 prices. In order to
update these costs, we apply an update factor to account for inflation. The update factor is
defined by:
UF = present cost index
base cost index
25
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
For cost updating we will use the Chemical Engineering (CE) Plant Index reported in
Chemical Engineering magazine. Representative indices are given below:
Year CI
1957-59 100
1968 1/2 115 (Guthrie's article)
1970 1/2 126 (Guthrie's book)
1983 316
1993 359
1994 366
To account for numerous direct and indirect costs associated with the cost of equipment,
Guthrie proposed a simple factoring method for add-on costs. A typical cost module
(with representative numbers) is given below.
2) Installation
a) Piping instruments, etc. 62.2
b) Labor (L) 58.0
As a result we define the Bare Module Cost = BC x MF. Here the module factor (MF) is
2.95 (a typical value); i.e., the equipment cost is almost three times the base cost. This
module factor is also affected by the base cost. Consequently in Tables 4.11 and 4.12 we
give module factors for the following base costs (BC in 1968 prices):
MF2 Up to $200,000
MF4 $200,000 to $400,000
MF6 $400,000 to $600,000
MF8 $600,000 to $800,000
MF10 $800,000 to $1,000,000
Moreover, for special materials and high pressures, we have already defined materials
and pressure correction factors (MPF) for various types of equipment. Here the bare
module cost is modified by the following factors:
26
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Finally, we do not treat contingency costs and indirect capital costs as Guthrie does.
Instead, as discussed in the next chapter, for preliminary designs we apply overall indirect
cost factors and a flat 25% contingency rate after all the equipment is costed. From this
description, let's consider the above examples again.
The compression train sizing example is reconsidered in order to determine the costs of
the compressors and heat exchangers. For the three stage compression we assume that the
compressed air is desired at the inlet temperature, 298 K and therefore need to find the
costs of three identical compressors and heat exchangers:
The heat exchangers, on the other hand, each have a heat duty of 2.56 kW and from Table
4.2 with a water (shell) / air(tube) system, the overall heat transfer coefficient is U = 20-
40 Btu/hr-ft2-o F; we choose U = 20 Btu/hr-ft2-o F (5.678) = 114 W/m2 K (why?).
Assuming cooling water available at 300 K and an allowable discharge at 322 K we
calculate the log mean temperature difference to get:
∆Tlm = (386 - 300) - (322 - 298)/ ln(86/24) = 48.6 K
Aint = Qint / U ∆Tlm = 50 ft2
From Table 4.12 we obtain a base cost (BC) of 300(50/5.5)0.024 = $316. For a carbon
steel, floating head exchanger with a pressure factor of 0.25 (why?) we have a materials
and pressure factor (MPF) of 1.25 and a module factor (MF) of 1.85. The bare module
cost for the three exchangers is:
BMC = 3(Uf)(MPF + MF - 1)(BC) = 3(3.12)(1.25 + 1.85 - 1)(316) = $ 6210
Assuming a cooling water cost of 5.2¢/1000 gal. = $2.47·10 -7/gmol and a temperature
rise of 22 K, with a service factor of 0.904, the utility cost is about $ 14/yr.
27
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
T = 385 K
0.1 gmol/s Acetone
329.7 gmol/s Water
For the distillation example in the previous subsection, calculate the costs updated to
1993 prices. From the previous example we have the following data:
Column Diameter = 1.17 m (3.8 ft.)
Column Height = 20.4 m (67 ft.)
Tray Stack Height = 14.4 m (24 in. spacing)
First we find the cost of the column vessel itself. From Table 4.11 we have an fob cost
(BC) of about $12 600. Assuming carbon steel construction we have Fm and Fp as well
as the MPF equal to 1.0 (why?). The resulting module factor (MF) is 4.23 and the update
factor is Uf = 359/115 = 3.12. The bare module cost (BMC) is then obtained from:
BMC(vessel) = Uf (MF + MPF - 1) (BC) = $166 300
The tray stack is also calculated from Table 4.11 with H= 47.2 ft. (14.4 m) and D = 3.8 ft.
(1.17 m) we have BC = $2060. Assuming bubble cap trays with 24" spacing, we have
MPF (Fs + Fm + Ft) = 2.8. Note there is no module factor for tray stacks. As a result we
have the following cost:
Now the column condenser requires both utility costs and capital costs. The utilities can
be calculated first. From the above example,
Qcond = (41.4/20.2) [19.9(29.1) + 0.3(43.0)] = 1215 kW
and we assume the following for cooling water:
Cpw = 75.3 J/gmol K
Tout = 322 K
28
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Tin = 303 K
µw = Qcond/Cpw (Tout - Tin) = 849 gmol/s
Price = 5.2¢/1000 gal. = $2.47·10-7 J/gmol
Service factor = 0.904
Days of operation = 0.904 (365) = 330 days yr.
As a result, the annual cooling water utility cost is given by:
($2.47·10-7) (849) (3600) (24) (330) = $6000/yr.
The condenser can be sized and costed as follows. The overall heat transfer coefficients
can be estimated from Table 4.2, as well as Perry's handbook. For an acetone-water
(shell) / water (tube) system, we have U = 100 - 200 Btu/hr. ft2˚F and we select U = (100)
(5.678) = 567.8 W/m2 K. Also, from the example we have:
∆Tlm = [(329-303) - (329-322)]/ ln(26/7) = 14.5 K
A = Qc/(U ∆Tlm ) ~ 150 m2 = 1610 ft2 < 5000 ft2 (max.)
From Table 4.12, the base cost (BC) = $12 380 and for a floating head, carbon steel heat
exchanger MPF = 1.0 and the module factor (MF) = 3.29. Hence the bare module cost
is:
BMC= 3.12 (12 380) (3.29) = $134 400
Finally, the reboiler can be costed in a similar manner. First the utility costs are computed
from the heat duty in the above sizing example:
Qreb = 41.4 (40.) (water) = 1657 kW.
and we assume steam is available at 150 psig with the following characteristics:
Tsteam = 459 K and ∆Hvap = 35,871 J/gmol
so that µs = 46.2 gmol/s. If we are given a steam price of $4/ 1000 lbs and a condensate
credit of $1.2/1000 lbs, we apply a net price = $2.8/1000 lbs or $1.11·10-4/gmol. The
annual utility cost with a service factor of 0.904 is then:
steam cost = ($1.11·10-4) (46.2) (3600) (24) (330) = $1.464·105/yr.
Sizing the reboiler first requires an overall heat transfer coefficient. For a water (shell) /
steam (tube) system we have from Table 4.2, U = 250 - 400 Btu/hr ft2˚F and we select U
= 250 = 1420 W/m 2 K. Also, ∆Tlm = (459 - 385) = 74 K and so the area is:
Areb = Qreb/U ∆Tlm = 15.8 m2 = 170 ft2
From Table 4.12, we have BC = $2900, MPF = 1.45 (for a slightly higher pressure and
carbon steel kettle reboiler) and MF = 3.29 . The resulting bare module cost becomes:
BMC = (3.12) [(3.29+ 1.45 - 1) (2900)] = $33,840 ~ 34,000
In summary, the column has the following capital and utility costs.
29
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Capital Costs
Utility Costs
4.4 Summary
This chapter was devoted to straightforward sizing and costing calculations for
preliminary process design. Our goal is developing these calculations was to obtain rough
estimates quickly and to observe qualitative trends for candidate designs. As a result, our
estimated costs will be accurate within 20-40% and this is considered sufficient for
preliminary design.
In the next chapter the bare module costs and the operating costs will be combined into
an overall assessment of the plant economics. Here a key consideration will the
application of appropriate economic metrics in order to evaluate alternative designs.
References
Douglas, J. M., Conceptual Design of Chemical Processes, McGraw Hill, New York
(1988)
Fair, J., R., Petro. Chem. Eng., 33 (10), p. 45, Sept. 1961
Guthrie, K. M., "Capital Cost Estimating," Chemical Engineering, March 24, 1969, p.
114
Reid, Robert C., Prausnitz, J. M. and Poling, Bruce E., The properties of gases and
liquids, New York : McGraw-Hill (1987).
Perry, Robert H., Green, Don W., Maloney, James O.(eds.) Perry's Chemical engineers'
handbook. Chemical engineers', 6th ed., McGraw-Hill, (1984)
Peters, M. and K. Timmerhaus, Plant Design and Economics for Chemical Engineers,
Mc Graw Hill, New York (1980)
30
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
Welty, J., C. Wicks and R. Wilson, Fundamentals of Momentum, Heat and Mass
Transfer, Wiley, New York (1984)
Exercises
1) A gas stream of 1 kgmol/s consisting of 50% mol H2, 50% mol CH4 is available at
100kPa, 300°K. If the stream is to compressed up to a pressure of 3000kPa and delivered
at 350°K, determine the required investment and annual operating cost for the two
following cases:
Data Investment cost: bare module cost updated to Jan. 1994 prices
(Guthrie's method)
Assume overhead recoveries for IBH and IBA of 99% and 1%, respectively.
a) Find the overhead and bottoms compositions, the theoretical and actual
number of trays and the reflux ratio for this column
b) Find the column height and diameter. Determine the column cost from
Guthrie's article, for July 1968. Use carbon steel for all parts.
c) Using a very simple enthalpy balance, size the condenser in this column.
Assume cooling water entering at 310K with a 15K temperature rise.
Find the cost of the condenser from Guthrie's article.
3) 300 gmol/sec of a 70/30 mole % mixture of benzene and xylene are separated at 2
atm. The light and heavy recoveries are 0.99 and 0.01, respectively.
31
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
a) Find the theoretical number of trays, the reflux ratio and the column height if 24"
trays are used.
b) For a reflux ratio of 0.5, find the reboiler and condenser duties if the column in
part a) has bubble point feed
4) Starting from the relationship for work in a single centrifugal compressor derive the
equation for N compressors with intercooled stages. State all assumptions in the
derivation.
5) 160 gmol/s of propane requires a cooling load 200kW to cool the stream to 260K.
a) How many stages of refrigeration are required and which refrigerant should be
used in each cycle. (choose refrigerants from problem 6.)
c) What is the total compressor work and cooling water duty if the coefficients of
performance is four.
6) A stream of n-butane needs to be cooled from 300 K to 250 K. The change in heat
content for this stream is 300 kW.
a) How many stages of refrigeration are required? Which refrigerants among the
above should be used in each stage to maintain the lowest cycle pressures above 1
atm?
c) For a coefficient of performance of 5, find the compressor work and the cooling
water duty for this refrigeration system.
7) A 50 gmol/s stream of nitrogen needs to compress from 1 atm and 310 K to 35 atm.
The stream must also be delivered at 650 K. Assume a constant C p of 7 cal/gmol for
nitrogen.
32
Systematic Methods for Chemical Process Design (Draft August 23, 1996) Lorenz T. Biegler
a) For a staged compression system with intercooling (back to 310 K), calculate the
work and the amount of heating and cooling required. Assume an average com-
pression ratio of 2.5.
b) Your boss believes that you can deliver the stream at 650 K more cheaply by
avoiding intercooling in the compression. Is she right or wrong? How can you
make this argument quantitatively?
33