30000 750S0。81                                                                    (52) 

proach compared to GA approach considered by Ponce-Ortega et al。 [16]。 

Various alternatives of exchangers 

The solution space of  cost  objective  function  with multiple constraints is very  much  complicated  with multiple local minima。 Cost wise these local mi- nima may be very near to each other but geometri- cally represent complete different sets of exchangers。 To assess these multiple local minima, the SA pro- gram was run 100 times with new starting guess every time。 For case study 2, i。e。, different sets of solutions vectors x is assumed in every run。 Most of the  times  SA converged to global minima but sometimes it were 

found that it got stuck to local minima depending upon 

m

Capital cost of pump ($) = (2000 5(   t  ΔP

t

)0。68 )  

the complexity of solution space。 All these feasible solutions were collected and solution within 5% of mi-

(2000 5(   s  ΔP )0。68 )                                   (53) 

nimum cost is presented in Table 4 for case study   2。 

s From this table, it is clear that multiple heat   exchanger configuration is possible with practically same cost  or 

Cost of power ($/W Hr): 

Cpow  0。000045                                            (54) Pump efficiency: 

70%                                                             (55) 

Plant operation (h/year): 

H = 8000 (56) 

Annualization factor (/year): 

Af   0。322                                                          (57) 

The design proposed by  Ponce-Ortega  et  al。 [16] using GA approach assumed an exchanger with four tube side passages (with square pitch pattern)  and one shell side passage。 The same   configuration  is not retained in the present approach and considers the tube side passes a free variable。 Results show  increment of tube side and shell side heat transfer  coefficient (5。26% on tube side and 88。7% on shell side) resulted in 23。85% increment in overall heat  transfer coefficient in the present approach。 The higher overall heat transfer coefficient results in 19。52% re- duction in heat exchanger area and 20。83% reduction in heat exchanger length in the present approach  compared to GA approach。 The capital investment is decreased by 3。69%。 Less pressure drop in tube side (due to shorter tube length) and less pressure drop in shell side (due to less number of baffles) resulted in 66。45% reduction in pumping  cost  as  compared to GA approach。 In totality, the combined effect of  capi- tal investment and operating costs results in 11。16%  reduction  in  the  total  cost/year  in  the  present   ap-

with little cost difference。 All these solutions are feasible and the user has flexibility to choose any one of them based on his requirement and engineering  judgment。 For example, some users have very less  space available in his company, so they may choose the lowest length heat  exchanger  and  fixed  tube  head, as they have no space available to use a pull  through heat exchanger。 Selecting the best exchanger design from Table 4 is a combination of science and arts。 Decision of best exchanger selection for a parti- cular service and industry is based on multiple criteria including costs。 These  criteria  sometimes influence  the best selection decision much more than the simple lowest cost criteria。 Maintainability, ease of    cleaning  of tubes and shells, less fouling tendency, flow in- duced vibrations, less floor space requirement, com- pactness of design, etc。, are some of these criteria which must be considered in an industrial scenario。 The lowest cost exchangers are not always perform- ing best in actual shop  floor。  These  criteria though very influential for final selection of  exchanger  are often qualitative and difficult to express  quantitatively。 It requires designer experience, engineering  judg- ment, customer requirements and normally very prob- lem specific。 Following section describes some of the criteria which can facilitate the user to select the best exchanger for his case studies。 These criteria are col- lected from literatures and based on experience of designers。 The final decision is dedicated to the user。