Professional Documents
Culture Documents
and
E. E. ABDEL GHAFOUR
Atomic Energy Establishment, Cairo, U.A.R.
Abstract-A new approach to the optimization of multi-stage flash evaporation plants is presented.
While retaining the basic features of current theories, the approach presented involves the use of
relative values of economic parameters rather than their absolute values. The paper deals with a
3-free-design variable problem of constant output while two cost parameters may vary according to
prescribed schedules. Optimization procedures are presented and tested using an illustrative example.
CURRENT theories [4,6,7] concerning optimiza- 1.1 Thermal energy cost term
tion of multi-stage flash evaporation plants are This is given by EpoQ, where Q is taken as the
aimed at clarifying the interrelations between the thermal energy supplied in the brine heater.
major variables involved. The main objective of
the present work, however, is to develop optimi- 1.2 Surface area cost term
zation procedures through a new approach which Among cost items which are proportional to
retains as much as possible the basic features the heating surface area (A) are; installed cost of
of the current theories. Principal optimization heating surface plus its maintenance cost. These
variables considered are; number of stages N. will be allocated to a cost term, C,J. There are
terminal temperature difference TTD, perform- other constructional materials such as walls,
ance ratio R and concentration ratio CR. The foundations and auxiliaries of flashing chambers.
paper deals with a 3-free-design variable prob- In a plant comprising N stages, the cost of such
lem of constant output while two cost parameters items can be incorporated in a cost term;
may vary according to prescribed schedules. FIVpo+J,O (N/F)1’2] [4].
Silver[4] introduced the idea that the unit cost
of fresh water is governed by certain economic 1.3 Sea water make-up cost term
parameters, namely, volume parameter Vpo, Cost of items which are proportional to the sea
Staging parameter Jpo, Surface parameter SD0and water make-up, F,, such as pumping, deaeration
Energy parameter Epo. The present work, though and scale prevention chemicals will be allocated
retaining this basic concept in its early stages of to a cost term, C,F,.
development, it will finally involve the use of the
relative values of such economic parameters 1.4 Remainder cost term
rather than their absolute values. All of the remaining cost items of sea water
conversion are incorporated into a remainder
1. MANUFACTURING COST TERMS cost term, C,F.
All of the items contributing to the manufactur-
ing cost of fresh water are distributed over a set 2. UNIT COST EQUATION
of cost terms, so chosen to cope with optimiza- From the above discussion, the total plant
tion purposes. production cost, C. in cents per day is given by:
611
M. A. MANDIL and E. E. ABDZL GHAFOUR
E= & (GJ),,(~PE)o.
and
9_ r
- 1000 (C,)mA TBH.
0F
A
F =a,.Nln@ (5)
0
where “tloou,
-r
L ‘.
’ &t
ao =NUL\t, -.
.
‘.
(9 =$[a.+:+N(;- 4 (6)
where
1000 (4, A&r (BWo Fig. 1. Temperature changes in stage 1 ATBH= T,--t, =
a1 = L, (c~)~~ At,’ u2 = NAr, ’ Dt,,+ BPE+ TTD.
EC_
a2
a1 C* = Epoe+Epoa-+(q+&)+S;N
N
(UPE) =
lnf3+ VPo+JPo (N)“2+C.?(&)+cr
F (12)
is the general practice so far, a certain evaporat- cost ratio, respectively. Recalling that JPo, S,O,
ing area is needed per unit of product. This will F, EPo and a1 have the respective dimensions
lead to the conclusion that VP0is constant. Also, c/d(lb hr), c/lb, Ib/hr, c/B.t.u. and Ib/B.t.u., it
the unit remainder cost C, is constant by defini- could be easily shown that X, and X, are dimen-
tion since it is not proportional to any of the sionless groups.
variables involved. The cost terms E,“( l/alN) X In order that X, might be given in more plau-
( a3 + l/0 - 1) , &ON In 19and JPo (N/F) ‘I2are func- sible representatton, a, and SPo (Sections 4 and 5)
tions of the design variables (0 and N). In are substituted for in the definition of X,.
addition, the cost terms Epoe and C,(CR/(CZ?-
1) ) are functions of the operating variable (CR), &=J$= Ep”/l~ _
(Cd,, &c 2
since E is in turn a function of CR. Therefore, the
cost terms included in Eq. (12) can be divided
1 PO Calu (Cp), A&t (> (FTR)
’
(13)
into three distinct groups:
Since the rate of recycle flowing in the con-
(a) group of constant terms +1 = VP0+ C, denser tubes is the same as the sum of the flash-
(b) group of terms which are a function of CR ing brine and the distillate, the temperature rise
only. of the brine in the tubes is the same as the
temperature decrease due to the flashing. Thus
At, could be assumed equal to AtsP Hence,
(MTD) is 1°F.
Minimum unit cost of fresh water will be
It is clear now that, X, is the ratio between the
realized when:
cost of producing and transmitting one B.t.u. for
(MT@ and (FTR) of 1°F.
I
@W= 1+ (l+4N2,XE)1/2_ 1’ (16) I50
loo
In the limiting case where the desalination (TTD),, = [r+)‘+$$& (FTR)"' -%.
plant contains an infinite number of stages and
infinite heating surface, a hypothesis which (2lb)
could only correspond to a case where the unit
cost of energy is prohibitively high, S,O/E,O and 6.1.6 Optimum brine heater temperature rise.
hence X, would be quite close to zero. If this (ATBH)OPtwith reference to Fig. 1, (ATBH),,ptfor
finding is inserted in Eq. (19) and remarking a certain N is given by:
615
M. A. MANDIL and E. E. ABDEL GHAFOUR
L
H
0. 2.4 -
S
‘Z
c”
2 2.2 -
8
$
(r 2.0 -
.c
.z
c
; 1.6-
E
P
- I.6 -
E
‘E
._
= I.4 -
0
0 5 1015202530354045505560665107560659095 0
Optimumnumberof stages, NW
gives all possible optimum concentration ratios needed, which gives the pattern of variation of
for different values of E,O/C,. Hence, CR,,, an adjusted energy parameter (Epo)a with
could be obtained by superimposing the newly (PEF). The adjusted energy parameter which
constructed curve on that given in Fig. 2 for the corresponds to the productive energy require-
same T,, adopted. The optimum concentration ment is defined as follows:
ratio being given by the point of intersection of
the two curves. Q(J%~)Q-~F(&")~
6.2.2 Opt&num design features. In this case,
tEpOja =
Q--EF *
(26)
the adaptation of theory developed to variable
economic parameters needs the introduction of In addition, a modified area cost schedule
new design functions, namely, productive energy would represent the pattern of variation of SPo
function (PEF) and surface function (SF) which with (SF).
are given by: Adopting the use of the term JPo(N/F)‘/’ to
represent the amortized cost of constructional
(PEF) = a, ( UPE) = + (0993 + A) (24) material relating to staging per unit product[4],
the staging parameter JPo for a particular plant is
obtained by dividing the amortized cost of con-
(SF) =$g) =Nln8. (25) structional material relating to staging in the
whole plant by (NF)‘/* where N is any arbit-
Figure 4 gives (PEF)O,, and (SF),,, for each rarily chosen value of number of stages.
pair of unit cost ratios X, and X,. In order to evaluate EPo and S,O that prevail at
A modified energy cost schedule would be optimum conditions, a wide range of (SF) values
617
M. A. MANDIL and E. E. ABDEL GHAFOUR
is considered. For each value of (SF) within from Fig. 4, the unit cost ratios X, and X_, can
such a range, corresponding S,O could be ob- be determined.
tained using the modified area cost schedule. From thereon the same procedure given for
For X, = JPo/SPo(F)‘/*, a set of X, values is constant economic parameters could be
obtained. followed.
0.200
I
NOTATION
heating surface area
(BP; elevation in boiling point
A(BPE) change in elevation of boiling point
with respect to change in brine
temperature in the evaporator
C fresh water production cost per unit
time
c* unit cost of fresh water
Cl2 installed cost and maintenance cost of
a unit surface area.
C, unit remainder cost
ohIO ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ 1 unit cost of make-up sea water
II 12 I3 14 I5 16 17 I6 192021 222324252627262930
Optimum surface function, SF,, (C2 mean specific heat of brine in the
evaporator
Fig. 4. Optimum surface function and optimum productive mean specific heat of brine in brine
energy function using X, and X, as parameters.
(C,)m
heater
CR concentration ratio
Using Fig. 4, each value of (SF) and its corre- EIJO energy parameter, cl1000 B.t.u.
sponding value of X., would give: F fresh water production rate
F, make-up sea water rate
(a) Value of X, that must prevail if (SF) were FTR flashing temperature range
optimum, JIJO staging parameter, c/d(lb.hr)
(b) Value of (PEF),,, that corresponds to L, mean latent heat of vaporization of the
(SF). flashed vapour in the whole evapor-
ator
From the relationship Epo = al &OX,, one could MTD mean temperature difference
calculate a corresponding value of Epo. On the N number of flash stages
other hand, from the modified energy cost sched- Q total thermal energy required
ule (PEF),,, would give another value of EPo. R performance ratio of the plant
These two values of Epo are not necessarily recirculation ratio
equal except when the design is optimum. SJ surface parameter, c/lb
Now, having two sets of values of Epo, these T temperature of brine in the evapora-
are plotted separately versus (SF), giving two tors
curves which if superimposed the point of their t temperature of brine in the condensers
intersection will give (SF),,, and the correspond- T max maximum brine temperature
ing value of Epo that prevails at optimum condi- TBD brine blowdown temperature
tions. From the modified energy cost schedule, to temperature of flashed vapour
the value of E,” just obtained, determines the Ate stage temperature rise in the conden-
optimum productive energy function. While ser
618
Optimization of multi-stage flash evaporation plants
REFERENCES
[ 1] DODGE B. F. and ESHAYA A. M., Thermodynamics of Some Desalting Processes, Advances in Chemistry, Series No.
27. American Chemical Society 1960.
[2] FLUOR CORPORATION LTD., Preliminary Study of an Optimum Nuclear Reactor-Saline Water Evaporator
Process, Office of Saline Water, Research and Development Progress, Rept. No. 34, PB 161010, U.S., Dept. of Interior,
Washington, D.C. 1959.
[3] MURPHY G. W. et al., Minimum energy requirements for sea water conversion processes, O&e of Saline Water
Research and Development Program, Rep. No. 9 1956.
[4] SILVER R. S., Fresh Waterfrom the Sea, Paper Presented during the Symp on Nuclear Energyfor Water Desalination,
International Atomic Energy Agency, Vienna, 1966; Tech. Rep. Series No. 5 1, P. 29.
[5] SPIEGLER K. S., Salt Water Purification 1962.
[6] TRIBUS M. and EVANS R. B., The Thermoeconomics of Sea Water Conversion, University of California, Los Angeles,
Water Resources Center Contribution NO. 61, Rep. NO. 62-53 1963.
171 TRIBUS M. and EVANS R. B., Thermoeconomic Design Under Conditions of Variable Price Structure, Paper Pre-
sented during the First Int. Symp. on Water Desalination, Washington (1965).
APPENDIX 0.12
ILLUSTRATIVE EXAMPLE I
6 O.ll-
For the purpose of illustrating the use of suggested
optimization procedure a numerical example will be taken
g
up. Cost data to be used here are those given by Tribus[7] 7
_ 0.10 -
and Silver[4]. In order that such data might be used without
subjecting the results to be obtained to undue errors, it was .E
decided to use the same plant capacity quoted by Silver, OP
w 0.09 -
namely one million Imperial gallons per day. 8
t
E
0 0.06 -
1. OPTIMUM CONCENTRATION RATIO b
0
1: 1 Basic data
8
(a) Tmax= 250”F, (b) ZBD= lOO“F, & 0.0-r -
tc) C, = 4.08 x lO-4 c/lb[6] I:
(d) Energy Cost Schedule: based upon data given by
Tribus [7] and shown in Fig. A 1.
1
0 0~5XlD6 I06 I.5xlO6 2x106 2.5~10’
619
M. A. MANDIL and E. E. ABDEL GHAFOUR
Table A 1
.E
oa660
0.0620
I\
;,,,\-----,_
-
then plotted against (SF) to give two separate curves as
shown in Fig. A4. The point of intersection gives (SF),,
of 18.06 and (E,“) of 0.0785 in c per thousand B.t.u. From
Fig. 4, (PEF),, is then found to be 0.0710 in thousands
“w”
B.t.u.‘s per pound.
E 0.0790 2.6 The remaining design features are then obtained as
t follows:
e
0 0.0740
0 X, = 332, from Fig. 4 or computer results.
6 X, = 1.833, from Fig. 4 or computer results.
$ o~07000
c 0.02 0.04 006 0.06 0.10 0.12 0.14 0.16 N,,,, = 35 stages, from Fig. 3
W
Productive energy function, PEF
(At&,t = v = 4.286”F per stage
Fig. A2. Modified energy cost schedule. I&, = 1.6738, Eq. (16)
Table A2
From Fig. 4
or computer results
Assumed
values Staging Productive Calculated
of Surfacet surface Energy energy Energy+ energy
surface parameter unit surface function parameter parameter
function S,O in cost unit cost (PEF) in Epo in Epo in
(SF) 1O-4c/lb ratio X, ratioXE lOOOB.t.u./lb cllOO0B.t.u. c/lOOOB.t.u.
Walues of SD0 given in this column are obtained from Fig. A3 corresponding to
values of (SF) in the first column of table.
SValues of Epo given in this column are obtained from Fig. A2 corresponding to values
of (PEF) in the fifth column of table.
620
Optimization of multi-stage flash evaporation plants
0.16
0.15
0.14
0.13
0.12
0.11
0.10
0.09
0.08
0.07
0.06
0.05
0.04
0.03
IO II 12 I3 14 I5 16 17 18 19 20 21 22 23 24 25 26 27 28 29
I
'0
Surface fundon, SF
Unit productive energy requirement QoDt= 10’ (089157) = 9.157 x 105thousands B.t.u./day
A.,, = aoF(S = 0.018
(PEF),,, = ($-c)~,, = s = 0.07667thousand B.t.u./lb.
= 135450ft*, Eq. (25).
Resume- Une nouvelle approche est presentee en ce qui conceme I’optimisation eclair Z+plusieurs
&tapes des plantes d’evaporation. Tout en retenant les caracttristiques de base des theories courantes,
I’approche qui est present&e traite de I’utilisation des valeurs relatives des parambtres tconomiques
plutbt que de ieurs valeurs absolues. Cet expose traite d’un probleme variable de 3 modtles libres
de rendement constant, alors que deux parametres de cout peuvent varier selon les programmes
ttablis. Les procedures d’optimisation sont presentees et test&es au moyen d’un exemple illustratif.
Zusammenfassung- Es wird ein neuer Weg zur Optimierung von mehrstufigen Flash-Verdampfungs-
anlagen dargelegt. WPhrend dieser neue Weg die grundlegenden Merkmale der gegenwartigen
Theorien beibehalt, werden hier jedoch mehr die relativen Werte der wirtschaftlichen Parameter
verwendet als die absoluten Werte derselben. In dem Artikel wird ein Problem mit 3 freien Konstruk-
tionsvariablen bei konstanter Leistung beschrieben, wobei zwei Kostenparameter gemlss vorgeschrie-
benen Planen variieren konnen. Die Optimierungsverfahren werden an einem erlautemden Beispiele
geprtift.
621