Sunteți pe pagina 1din 7

970 Ind. Eng. Chem. Process Des. Dev.

1985, 2 4 , 970-976

Nomenclature 4 = volume fraction solids


D = particle size, m Registry No. Calcium carbonate, 471-34-1.
g = acceleration due to gravity, m/s2 Literature Cited
U = settling velocity, m/s Fessas, Y. P. Ph.D. Thesis, Clarkson University, Potsdam, NY, 1983.
X = distance from the top of the buoyant sediment, mm Fessas, Y. P.; Weiland, R. H. AIChE J . 1981, 2 7 , 588.
Fessas, Y. P.; Weiland, R. H. Resour. Conserv. 1982, 9 , 87.
Subscripts Fessas, Y. P.; Weiland, R. H. I n t . J. MuMphase Flow 1984, 10, 485.
Michaels, A. S.;Bolger, J. C. Ind. Eng. Chem. Fundam. 1962, 1 , 24.
f = refers to flocs Stagle, D. S.;Shah, Y. T.; Klinzing, G. E.; Waiters, J. G. Ind. Eng. Chem.
p = refers to primary particles Process D e s . Dev. 1978, 17, 500.
Thomas, D. G. AIChE J . 1963, 9 , 310.
Superscript Vilambi, N. R. K. M.S. Thesis, Clarkson University, Potsdam, NY, 1982.
O = at infinite dilution
Weiland, R. H.;Fessas. Y. P.; Ramarao. B. V. J . Fluid Mech. 1984, 142,
383.
Weiland, R. H.; McPherson, R. R. Ind. Eng. Chem. Fundam. 1979, 18, 45.
Greek Letters Whitmore, R. L. Br. J . Appl. Phys. 1955, 6 , 239.
(Y = fluid to solid volume ratio in the floc
p = fluid viscosity, kg/(m s) Received for reuiew May 31, 1984
p = density (of fluid if not subscripted), kg/m3 Accepted November 29, 1984

Cost Dtagrams and the Quick Screening of Process Alternatives


James M. Douglas'
Chemical Engineerlng Department, Universm of Massachusetts, Amherst, Massachusetts 0 1003

Duncan C. Woodcock
Imperial Chemical Industries, Runcorn. Cheshire WA 7 4QE. England

Cost diagrams provide a useful way of summarizing total processing cost information for preliminary process designs.
I n addition, they are often useful for checking rules of thumb, for obtaining quick estimates of the economics of
process alternatives, and for establishing a hierarchy of optimization variables. Thus, they help to establish priorities
for more detailed design studies.

A t the initial stage of designing a new process, it is all the steam or cooling water costs usually are added
possible to generate more than 1 million flow sheets together and reported as a single item. Separate tables
(Douglas, 1985) and there are about 10-20 optimizations of manufacturing and capital costs would then be prepared
of the design variables required for each flow sheet for each process alternative.
(Westerberg, 1981). Heuristics can often be used to elim-
inate some of the alternate flow sheets and to provide first Cost Diagrams
estimates of some of the design variables, but there are still As an alternate approach to summarizing cost infor-
a very large number of process alternatives and optimi- mation at the preliminary stages of a process design, we
zations that need to be considered. can prepare a cost diagram. The annualized, installed
Since flow sheets normally are dropped from further capital cost of each piece of equipment is listed inside of
consideration based on the total processing costs (safety, the equipment box on a flow sheet, and the annual oper-
operability, and pollution may play a part), it is useful to ating costs are attached to the stream arrows. Figure 1
have a simple and efficient way of summarizing cost in- shows a cost diagram for the production of acetone by the
formation. Cost diagrams can be used for this purpose. dehydrogenation of isopropyl alcohol, which is a modified
In addition, cost diagrams sometimes are useful for version of the 1948 AIChE Student Contest Problem
checking rules of thumb, for obtaining quick estimates of (McKetta, 1976). The values in Figure 1are reported in
the economics of process alternatives, and for identifying terms of thousands of dollars per year.
a hierarchy of optimization problems. Of course, we could also divide both the annualized
capital costa and the annual operating costs by the annual
Previous Work production rate of the process. Then, the cost diagram
From an examination of published design case studies would indicate the dollars per pound of product that each
(Washington University Design Case Study Series; AIChE item on the flow sheet contributed to the final product
Student Problems; Stanford Research Institute Reports; price. This type of an approach is sometimes used in
Peters and Timmerhaus, 1980; Happel and Jordan, 1975; industry for the design of batch processes.
Baasel, 1977; Ulrich, 1984),it seems to be common practice
to tabulate the manufacturing costs and the capital costs Use of Cost Diagrams To Check Rules of T h u m b
separately. The list of capital costs normally is itemized It is natural to assume that the reflux ratios for the two
to correspond to names or tags shown on a flow sheet, but distillation columns shown in Figure 1 were fixed by the
0196-4305/85/1124-0970$01.50/0 0 1985 American Chemical Society
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985 971

t. HZ

PREHEATER FEH E REACT CONDENSER FLASH


I 60 58
* 49 7t 76 -0
23 7 CAT

V
11.

!,
COND cw

c0 w7 COND
2 7
I t 12 5 33

ACET

WASTE WATER

Figure 1. Cost diagram for acetone process.

rule of thumb of 20% in excess of the minimum reflux from the flow sheet, it would be much better to undertake
ratio. However, a t the optimum design conditions ac- an energy integration analysis of the flow sheet (Linnhoff
cording to Happel and Jordan (1975), the distribution of et al., 1982). Recent experience indicates that 30-50%
annual costs should be 30% for the column, 35% for the energy savings are usually possible even in retrofit situa-
condenser and reboiler, and 35% for the steam and cooling tions (Boland and Hindmarsh, 1984; Linnhoff and Vre-
water. In a similar study, but using a 6.7 yr vs. a 2 yr time develd, 1984).
factor, Peters and Timmerhaus (1980) obtain 15% for the Use of Cost Diagrams To Identify the Significant
column, 10% for the condenser and reboiler, and 75% for Design Variables
steam and cooling water.
Even though there is a significant difference in the For the flow sheet shown in Figure 1,some of the design
treatment of the capital and operating costs between the variables of interest are as follows: the conversion, the
studies of Happel and Jordan and Peters and Timmerhaus, approach temperature between the Dowtherm furnace and
both of these studies seem to indicate that there is the reactor inlet, the approach temperature between the
something wrong with the column designs shown in Figure preheater exit temperature and the temperature of the
1;i.e., the column cost appears to be much too large. Thus, reactor products leaving the feed-effluent heat exchanger,
the cost diagram indicates that some additional column the fractional recovery of acetone in the compressor, the
design studies should be considered, where higher reflux reflux ratios in both distillation columns, the fractional
ratios would decrease the column cost but would increase recovery of acetone overhead in the product column (we
the annualized costa of the condenser and reboiler, as well assume that the product composition is specified), and the
as the annual cost of the steam and the cooling water. fractional recoveries of isopropyl alcohol-water azeotrope
and water in the IPA still.
Use of Cost Diagrams To Infer Structural Almost all of these optimizations involve only local
Modifications tradeoffs, Le., the column reflux ratio involves a tradeoff
The cost diagram shown in Figure 1 also makes it ap- between the number of plates and the costs of the con-
parent that only $5800/yr is being spent for energy inte- denser, reboiler, steam, cooling water, and column diam-
gration on the feed effluent heat exchanger around the eter. However, the optimum reactor conversion involves
reactor, but that a total of SO00 + 27 500 for the preheater, a tradeoff between an unbounded reactor cost as the
+
plus 29 300 18300 for the furnace (although some of this conversion approaches unity and unbounded recycle costs
as the conversion approaches zero. Thus, the optimum
energy supplies the endothermic heat of reaction), plus
+
7600 3900 for the condenser and cooling water is allo- conversion depends on all the costs in the recycle loop in
Figure 1,and it is apparent that it is much more important
cated for reactor heating and cooling. Thus, it seems as
if we should spend more for energy integration. Of course, to get the conversion close to the optimum than any of the
it might be difficult to recover more of the heat from the other design variables.
hot reactor products because the temperature of this Cost Diagrams for Complex Processes
stream will eventually approach the temperature of the The new synthesis procedures for heat-exchanger net-
preheater outlet. However, we can avoid this difficulty and works (Linnhoff et al., 1982) lead to highly coupled pro-
recover more of the heat if we simply eliminate the pre- cesses. Figure 2 shows a flow sheet for the production of
heater. benzene by the hydrodealkylation of toluene (McKetta,
In actual fact, rather than just eliminating the preheater 1977) before energy integration, and Figure 3 shows a cost
972 Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985

PURGE
4 an inverse proportion to their heat-transfer coefficients
(where the heat-transfer coefficient includes both the film
coefficient and the appropriate fouling coefficient) (eq 1).
cost allocated to stream -
l/heat-transfer coefficient of stream (1)

39' K r-
-.
* --1heatFrom Figure 3, we see that there are five stream-stream
exchangers in a row on the reactor effluent stream.
TOLUENE
The f i t exchanger (cost = 63 X lo3 dollars/yr) is a gas-gas
RECYCLE exchanger; we assume that the heat-transfer coefficients
of the two streams are identical, and therefore we split the
cost evenly. The toluene column reboiler (cost = 14 X lo3
dollars/yr) contacts a gas and a boiling liquid, and we
allocate a smaller cost (4 X lo3 dollars/yr) to the toluene
DIPHENYL c
reboiler because the heat-transfer coefficient is higher. The
Figure 2. Hydrodealkylation of toluene. allocation for the five exchangers is given in Table I.
Lump Costs Associated with Processing
Operations
diagram of the result for one alternative after an energy We lump the capital and operating costs of units: the
integration analysis. There are so many connections on compressor capital cost of 49 X lo3 dollars/yr and the
Figure 3 that it is difficult to use the cost diagram in the operating cost of 60 X lo3 dollars/yr; the toluene column
same way as we did Figure 1. Thus, we need to find a way shell cost of 52 X lo3dollars/yr, the allocated reboiler cost
to simplify the information presented. In particular, we of 4 X lo3 dollars/yr, and the condenser and cooling water
would like to be able to evaluate that fraction of the total costs of (12 + 8) X lo3 dollars/yr; and similarly for the
processing costs that is associated with the gas and liquid stabilizer and benzene columns. Also, we lump neighboring
recycle loops, as compared to the costs for processing the heat-exchanger costs on the same stream (reactor product
raw material streams. + + + +
cooling allocated costs of (31.5 10 10 10 29) X lo3
In order to simplify the cost diagram, we proceed dollars/yr, see Table I, plus the condenser 22 X lo3 dol-
through three stages: (1)allocate heat exchanger costs to lars/yr and cooling water 43 X lo3 dollars/yr = 155 X lo3
process streams; (2) lump costs associated with neighboring dollars/yr; feed and recycle heating allocated costs of (12
processing operations; and (3) allocate costs to the gas + 31.5) X lo3 dollars/yr, see Table I, plus the furnace 66
recycle loop, liquid recycle loop, and fresh feed rate of the x lo3 dollars/yr and fuel 262 X 103dollan/yr = 371 X lo3
raw materials. We discuss each of these steps below. dollars/ yr). With this lumping, we obtain the results
shown in Figure 4.
Allocate Heat Exchanger Costs to Process
Streams Allocate Costs to Gas Recycle, Liquid Recycle,
Following a suggestion of Townsend and Linnhoff and Fresh Feed
(1984),we allocate the cost of a heat exchanger, where two We allocate the reactor cost between the three categories
process streams exchange heat, to the process streams in based on the molar flow rates of the streams, see Table

Figure 3. Cost diagram for energy-integrated HDA process.


Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985 973

Table I. Allocation of Heat-Exchanger Costs


orig cost (X103dollars/yr) 63 14 14 14 41
stream reactor feed toluene reboiler stabilizer reboiler benzene reboiler reactor feed
hi,kw/(m2K) 0.69 1.57 1.57 1.57 1.57
l / h i allocated cost 31.5 4 4 4 12
stream reactor effluent reactor effluent reactor effluent reactor effluent reactor effluent
hi, kw/(m2K) 0.69 0.69 0.69 0.69 0.69
l / h i allocated cost 31.5 10 10 10 29

I-
RECYCLE COMPRESSOR Table 11. Reactor Cost Allocation

1
(09
molar flow
rate, kg cost (xi03
PRODUCT
stream mol/h dollars/ yr)
REACTOR PURGE
COOLING
198 453 6W gas recycle 3371 158
liquid recycle 91 4
I iI toluene feed 273
1 H2 feed 496
769 36

total 4231 198


DIPH‘NYL BENZENE

Figure 4. L u m p d cost diagram for HDA process. Table 111. Heating/Cooling Cost Allocation
prod cooler cost react feed heaters
11, since the reactor volume will be proportional to the flow 103 103
rate. We allocate the heating/cooling costs in proportion heat load, dollars/ heat load, dollars/
to the heat loads, see Table 111, for the product cooler and stream GJIh vr GJ/h vr
the feed and recycle heating, and we allocate the distilla- gas recycle 39.2 86 53.3 172
liquid recycle 6.1 13 12.1 39
tion costs proportional to the flow rates. fresh feed 25.5 56 49.6 160
Discussion of Complex Cost Diagram total 70.8 155 115.0 371
With the allocation assumption described above, we
obtain the cost diagram shown in Figure 5. Obviously,
the results are only approximate, but a t this stage of a by Douglas (1985). In some cases, heuristics are available
design (when we are screening process alternatives), a more to help in making these decisions, but in other cases, there
careful analysis might not be justified. It should also be are no guidelines available. Hence, the choice made to
noted that Figure 5 makes it appear as if the costs asso- develop a basecase design merely represents a guess. Of
ciated with the gas recycle stream, the liquid recycle course, if we change this choice, we are led to a process
stream, and the fresh feed streams are completely un- alternative. Since the choices a t early decision levels de-
coupled, which is not actually the case because the material pend on those made a t later levels, very different looking
balances are coupled. However, when we use the cost flow sheets might be obtained.
diagram to make first estimates of the economic incentive Thus, in order to determine what process alternatives
corresponding to various process alternatives, we use the should be considered, we merely have to review a process
process material balances to change the costs associated flow sheet in light of the decision procedure proposed by
with the relevant flows. Douglas (1985) and to recognize which choices were ar-
bitrary or were based on qualitative heuristics. These
Stream Costs choices are reviewed in Table IV.
We can also include the stream costs and the material Example-Hydrodealkylation of Toluene. If we
losses on the cost diagram. However, the most again refer to the process for the production of benzene
“reasonable”representation is to include only the amounts by the hydrodealkylation of toluene, where the reactions
of the feed streams that correspond to the excess above are
the stoichiometricrequirements of the specified production
+ H2 - + CH.,
-
rate, since these are the losses that might be avoidable by toluene benzene (2)
some other process alternative; see Figure 5. Our goal is
to use the cost diagram to help us develop a better process. 2benzene diphenyl + H2 (3)
Use of Cost Diagrams To Evaluate Process The level 2 decisions used to develop the base-case
Alternatives design (Douglas, 1985) were (1)do not purify the hydrogen
As was mentioned earlier, there are numerous process feed stream, (2) recover the diphenyl, rather than recycling
alternatives, and in a very large number of cases, there are this reversible byproduct, and (3) use a gas recycle and
no heuristics available for selecting the best alternative. purge stream.
One of the great advantages of cost diagrams is the help It would be desirable to estimate the economic incentive
they often provide in obtaining a quick estimate of the for changing these decisions and evaluating the process
economics associated with process alternatives without alternatives, without completely redesigning the process.
having to completely redesign the process. Some examples The decision not to purify the hydrogen-methane feed
of this type are discussed below for the benzene process. stream seems to be sound, since methane is produced by
the reactions. The other two decisions are discussed below.
Listing of Process Alternatives
A list of decisions that were required to develop a process Recycle of Diphenyl
flow sheet (for a limited class of processes) was presented In order to evaluate the possible incentive for recycling
974 Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985

GAS RECYC GAS RECYC

f
I
109

FEED HEAT REACTOR PROD COOL


TOL 1691
GR 86 PURGE
Hp 2463 NFF 56 - BEN2 8 TOL
c
LR LR LR i3
r 600 (FEEDSTOCK)
160 198 1 155 ( 5 mole/hr TOL)
95% H2 685 fuel

:1 ;:i
LR
r--
39
39
LR
~~~

76

76
NFF I or
129
43
172

DIPHENYL BENZENE
200 PRODUCT
(8mole/hr TOL)

Figure 5. Cost allocation diagram for HDA process.

the diphenyl, we can use the cost diagram to obtain most a requirement of a 5/1 H2/aromatics ratio a t the reactor
of the information we need. inlet in order to prevent coking. Thus, an extra 12 mol/h
(a) We save the 8 mol/h toluene that get converted to aromatics recycle will require an extra 12(5) = 60 mol/h
diphenyl (we still lose the equivalent of 5 mol/h toluene, hydrogen recycle, but since the hydrogen recycle stream
some as benzene, out the purge stream), but we lose the is only 40% H2,the gas recycle flow must be increased by
fuel value credit of the diphenyl. Thus, from the cost 150 mol/h. The costs associated with this increased gas
diagram, Figure 5, we can write recycle flow in the cost diagram, Figure 5, are, approxi-
mately,
150
savings in raw material = 1691 - - 200 =
(P3)
-(158
3371
+ 86 + 109 + 172) = 24 (X103dollars/yr)
840 (X103dollars/yr) (e) Combining the expressions above, we estimate the
(b) We can eliminate the toluene column order of magnitude of the saving as

savings = 76 (X103dollars/yr)
savings = 840 + 76 - 14 - 24 = 878 (X103dollars/yr)
Obviously, the calculations are only rough approxima-
On the other hand, the diphenyl flow in the liquid re- tions. However, they indicate that there H a large incentive
cycle recycle loop builds up to an equilibrium level, and for evaluating this alternative more rigorously.
all the equipment in this loop must be oversized to acco- Purification of the Gas Recycle Stream
modate this flow. In our original decision to use a gas recycle and a purge
(c) From an equilibrium calculation, see Hougen and stream with a purge composition of 40% H2,our increment
Watson (1974))the equilibrium diphenyl flow a t the re- raw material cost is 2163 X lo3 dollars/yr, while the fuel
actor exit is 12 mol/h, so that the total liquid recycle flow value of this hydrogen is only 685 X lo3 dollars/yr.
+
becomes 12 91 = 103 mol/h. We assume that the costs (a) Thus, the incentive for recovering and recycling the
associated with the liquid recycle streams are proportional hydrogen is very large, i.e.,
to the flow, and therefore the increased cost for the items
that depend on the liquid recycle flow in the cost diagram, raw material savings = 2163 - 685 =
Figure 5, would be 1478 (X103dollars/yr)
12 With the use of a hydrogen recovery system, such as a
-(4
91
+ 13 + 7 + 43 + 39) = 14 (X103dollars/yr) membrane process, the revised material balances are shown
in Figure 6. (b) The gas recycle flow is reduced from 3371
In fact, this cost should be an upper bound, because the to 1628 mol/h; so the cost savings associated with a de-
capital costs only increase with the flow raised to the 0.6 creased gas recycle flow from the cost diagram, Figure 5,
power, or some similar fractional exponent. However, at are
this point in the analysis, we are willing to sacrifice accu-
racy for speed in the first screening of alternatives. (1-3371)
1628 (172 + 109 + 158 + 86) =
(d) The increased liquid recycle of aromatics will require
that the gas recycle flow is also increased, since there is 271 (X103dollars/yr)
Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985 975

Table IV. Decisions


level 1: batch vs. continuous-we consider only continuous
processes
95 % H2 9 5 % H2 RECYCLE 95% H2/CH4
SEP

t
- C“4
2074

level 2: input-output structure of flow sheet REACTOR SEP -B


(1)‘should we purify the raw material streams before they are
fed to the reactor?”; if the impurities are inert, there are no
quantitative heuristics
(2) “how many product streams will there be?”; reasonable
heuristics seem to be available, except for the case of a
TO L
273
TOL RECYCLE
91
I
reversible byproduct Figure 6. Material balances for alternate HDA process.
(3) “should a reversible byproduct be recovered or recycled to
extinction?”; no quantitative heuristic is available
(4) “do we need a gas recycle and a purge stream?”; a These savings must be greater than the annualized in-
quantitative heuristic seemed to be available before the recent stalled cost of a membrane process for this alternative to
invention of membrane separation processes to separate be attractive. However, it should also be noted that the
gaseous mixtures.
large recycle flow of methane acts as a heat carrier (diluent)
level 3: recycle structure and limits the adiabatic temperature rise in the reactor to
(1)“how many reactor systems are required?”; the heuristics less than 1300 O F . Thus, eliminating the methane in the
seem to be reasonable; “is there any separation between the
reactors?”; usually a decision can be made based on the recycle might cause this constaint to be violated. If this
chemist’s data is the case, we would design the hydrogen purification unit
(2) “how many recycle streams are there?“; heuristics are so that enough methane could be recycled to satisfy the
available constraint, and we would need to adjust our estimates of
(3) “should we use an excess of one reactant?”; normally, the savings.
chemist’s data will indicate the answer
(4) “is a gas recycle compressor required?”; a heuristic is Other Decision Levels
available We can estimate the economic incentives for changing
(5) “should the reactor be operated adiabatically, with direct the decisions made a t other levels in a similar way. The
heating (or cooling), or is a diluent (heat carrier) needed?”;
some calculations are needed to use the heuristic other decisions are listed in Table V, and changing these
(6) “do we want to shift the equilibrium conversion?”; decisions will lead to other process alternatives (in most
calculations and judgment are required cases). We can also use the cost diagram to assess the
level 4: separation system economic incentive for changing design constraints, such
(1) “what is the structure of the vapor and liquid recovery as the 5 / 1 hydrogen/aromatics ratio a t the reactor inlet.
system?”; heuristics are available
level 4a: vapor recovery system Reducing the Hydrogen/Aromatics Ratio
(1)“what is the best location of the vapor recovery system?”; a The specification of a 5/ 1hydrogen/aromatics ratio at
heuristic is available the reactor inlet in order to prevent coking is probably an
(2) “what is the best type of vapor recovery system to use?”; uncertain constraint; it seems unlikely that a chemist
no heuristics are available would undertake experiments to find the exact coking
level 4b: liquid separation system
(1)“what separations can be made by distillation?”; a heuristic limitations for a significant range of design variables-
that usually works is available reactor temperature, pressure, conversion, etc. Instead,
(2) “what sequence of distillation columns should be used?”; a chemist would set the ratio to a sufficiently high value
the published heuristics are limited to sharp splits of ideal so that coking was never encountered (which is perfectly
mixtures for a single feed, but in many cases they do not lead reasonable). However, it is desirable to estimate the
to the best sequence economic incentive for reducing this ratio very early in the
(3) “how should the light ends be removed?”; calculations and
judgment are required development of a project, so that a chemist can make
(4) ”should the light ends be vented, sent to fuel, or recycled additional experiments in the range of the optimum design
to the vapor recovery system?”; calculations and judgment are conditions. The cost diagram can be used for this purpose.
required (a) If we evaluate the effect of a 311 rato (Rase (1977)
(5) “how should we accomplish the other Separations?”; no suggests a 2/1 ratio), the gas recycle flow is approximately
heuristics are available
level 5: heat exchanger network-a design procedure is available
+ +
cut in half, and the savings would be 0.5 (158 85 109
(Linnhoff et al., 1982) for developing alternative designs for + 172) = 260 (X103 dollars/yr).
heat-exchanger networks; also, a procedure described by (b) However, a decrease in the partial pressure of hy-
Androkovich and Westerberg can be used for energy drogen in the reactor will affect the rate of the reverse
integrating distillation columns reaction given by eq 3 and, therefore, might affect the
amount of diphenyl formed; i.e., the correlation of selec-
(c) The reduction in the hydrogen feed rate from 496 to tivity vs. conversion used for the initial design might
287.4 mol/h also reduces the costs associated with the fresh change.
feed processing costs on the cost diagram, Figure 5, to It is essential to verify experimentally that operation
with a 3 / 1 H,/aromatics ratio is possible before any de-
(1 - 2 ~ ~ ~ (160 +
~ ) 56) =
+ + 36 + tailed design studies are undertaken. However, the in-
formation obtained from the cost diagram does provide
68 (X103 dollars/yr) a rough measure of the economic incentive for carrying out
these experiments.
(note that we do not include the stabilizer and benzene
column fresh feed costs in this calculation because they Energy Integration Alternatives
are not affected by the hydrogen flow.) The cost diagram clearly indicates that the heating and
(d) The total potential savings of the hydrogen purifi- cooling costs are significant. Therefore, we want to con-
cation process is sider other energy integration alternatives. For example,
we could pressure-shift the toluene column, so that part
total savings = 1478 + 271 + 68 = of the benzene reboiler heat load could be supplied by the
1817 (X103 dollars/yr) toluene condenser. Similarly, we could use multiple levels
976 Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 4, 1985

Table V. Process A l t e r n a t i v e s f o r t h e HDA P r o c e s s obtaining a first guess of these design variables. However,
level 2: decisions-input-output structure it should be noted that the rule of thumb for choosing the
(1) d o n o t purify t h e hydrogen feed stream reflux ratio in a distillation column of 20% greater than
(2) recover, r a t h e r t h a n recycle, d i p h e n y l so t h a t there are the minimum is not valid for energy integrated columns,
three p r o d u c t streams (purge, benzene product, d i p h e n y l so that an optimization analysis might be useful.
byproduct)
(3) use a gas recycle a n d purge stream Evaluating Alternatives vs. Optimization
level 3: decisions-recycle structure Experience seems to indicate that greater savings nor-
(1) use a single reactor mally are possible by evaluating a new process alternative
(2) use a gas (H, a n d CH,) a n d a liquid (toluene) recycle rather than optimizing a process. However, since there are
stream
(3) Use a 5 / 1 H2/aromatics r a t i o t o prevent coking-assuming no rules of thumb available for selecting the reactor con-
t h i s t o b e a design constraint (although it could b e formulated version (which usually fixes the product distribution, raw
as a n o p t i m i z a t i o n problem) material costs, and recycle costs) or the purge composition
(4) a gas recycle compressor is needed of reactants (which usually fixes raw material costs and
(5) operate t h e reactor adiabatically gas recycle costs), some case studies or an optimization
(6) d o n ' t consider e q u i l i b r i u m effects analysis is needed for these variables.
level 4a: decisions-vapor recovery system
(1) if a vapor recovery system is used, place it o n t h e purge
stream Conclusions
(2) don't use a vapor recovery system It should be apparent that the use of cost diagrams to
level 4b: decisions-liquid separation system evaluate process alternatives does not give results that are
(1)a l l separations by d i s t i l l a t i o n sufficiently accurate for final designs. Instead, we use this
(2) direct sequence o f simple columns-probably complex
columns should b e used
approach to get some "feeling" for the economic incentive
(3) remove l i g h t ends in stabilizer for evaluating various alternatives in more detail. Thus,
(4) l i g h t ends t o fuel-no vapor recovery system cost diagrams provide a useful tool for establishing pri-
level 5: decisions-energy integration-there are numerous orities for additional design studies.
alternatives
Literature Cited
Baasel, W. D. "Preliminary Chemical Engineering Plant Design"; Eisevier:
of utilities (e.g., LP steam instead of fuel). New York, 1977.
Bobnd, D.; Hindmarsh, E. Chem. Eng. Prog. 1984, 80(7), 47.
Optimization Variables Douglas, J. M. AIChE J . lg85, 3 1 , 353.
Happel, J.; Jordan, D. G. "Chemical Process Economics", 2nd ed.: Marcel
The cost diagram also helps us to identify which are the Dekker: New York, 1975.
most significant design variables. For the flow sheet shown Hougen, 0. A.; Watson, K. M. "Chemical Process Principles"; Wiley: New
in Figure 3, the factors that affect the raw material costs York, 1947; Vol. 111, p 875.
Linnhoff, B.; Townsend, D. W.; Boland, D.; Hewitt, G. F.; Thomas, 8. E. A.;
(reactor conversion as it affects the selectivity and the Guy, A. R.; Marsland, R. H. "A User Guide on Process Integration for the
purge composition of hydrogen) are the most important. Efficient Use of Energy"; Institution of Chemical Engineers: Rugby, 1982.
Linnhoff, B.; Vredeveld, D. Chem. Eng. Prog. 1984, 80(7), 33.
Factors that affect the gas and liquid recycle flows (again, McKetta, J. J. "Encyclopedia of Chemlcal Processing and Design"; Marcel
the purge composition of hydrogen as well as the 5/1 Dekker: New York, 1976; Vol. 1, p 314.
H,/aromatics ratio constraint and reactor conversion for McKetta, J. J. "Encyclopedia of Chemical Processing and Design"; Marcel
Dekker: New York, 1977; Vol. 4, p 182.
toluene recycle) are the next most important. The pressure Peters, M. S.; Timmerhaus, K. D. "Plant Design and Economics for Chemical
of the flash drum, which affects the purge loss of aromatics Engineers", 3rd ed.; McGraw-Hill: New York, 1980.
Rase, H. F. "Chemlcal Reactor Deslgn for Process Plant", Wiiey: New York,
balanced against the incremental cost of using higher 1977; Vol. 2.
pressure equipment in the gas recycle loop, might be next. Townsend, D. W.; Linnhoff, 8. Paper presented at the Annual Research
Note that there are no rules of thumb available to fix any Meeting of the Institute of Chemlcal Engineerings, Bath, U.K., April 1984.
Ulrich, G. D. "A Guide to Chemical Engineering Process Design and
of these variables. Economics", Why: New York, 1984.
The fractional recoveries in the distillation columns, the Westerberg, A. W. "Foundations of Computer Aided Chemical Process
Design"; Mah, R. S. H., Seider, W. D., Eds.; Engineering Foundation: Hen-
reflux ratios in these columns, the approach temperature niker, NH, 1981; Vol. 1, p 149.
in the heat-exchanger network, the pressure levels of the
intermediate stages of the gas compressor, etc., involve Received f o r review May 29, 1984
local tradeoffs. Also, rules of thumb are available for Accepted D e c e m b e r 17, 1984

S-ar putea să vă placă și