Ci 51507    49259 43250。34           19007 17599           16200。57 27311。7 25902。32 24945。11 

Cod 12973     5818 9679。63             8012 2704             2389。63 3874。9 3501。52 1174。59 

Ctot 64480    55077 52929。97           27020 20303           18590。21 31186。6 29403。84 26119。71 

of heat exchanger by 3。9, 8。4 and 11。1%, respecti- vely, as compare to previously reported cost。 This de- monstrates the effectiveness of present  approach over previously reported GA approach [6,16]。 

In commercial software (HTFS, HTRI), the heat exchanger cost function has been recently incurpo- rated and cost minimization is performed by applying mainly gradient based methods。 Heat exchanger so- lution domain with multiple constraints is  very com- plex and can have several local minima。 Depending  upon the degree of non-linearity and  initial  guess,  most of the traditional optimization techniques  based  on gradient methods have the possibility of getting trapped at local optimum。 Hence, these traditional op- timization techniques do not ensure global optimum  and also have limited applications。 As SA is a random algorithm, it has the advantage of not getting  trapped in local optima and has greater possibility to reach the global minima of cost function。 In this respect,    using  of SA algorithm is preferred over traditional optimi- zation technique adopted in commercial software。 

Case 1: 4。34 MW duty, methanol-brackish water   exchanger 

The original design as well as the design pro- posed by Caputo et al。 [6] using GA approach as- sumed an exchanger with two tube  side passages and one shell side passage。 The input parameters  for 

heat exchanger design are given in Table 2 and com- parisons of final results by different algorithm are shown in Table 3。 

The following observations are made by com- paring the optimum results of present approach and GA approach by Caputo et al。 [6]。 The total annual cost has reduced by 3。89% even the pumping costs (operating costs) has increased 66。3% with respect to GA approach considered by Caputo et al。 [6] 15。8%  reduction in surface area makes it possible to reduce heat exchanger capital investment by  12。19%。 This was achieved by 44% more heat transfer   co-efficient  in tube side by using smaller diameter tube and higher velocity in tube side。 There is no velocity constraints  adopted in either GA approach [6] and literature case。 However, tube side velocity has increase to 1    m/sec  to satisfy the constraint in present approach。 This higher velocity will help to prevent fouling inside the tube。 This is one major difference between the present ap- proach vs。 earlier approaches。 This comes as a   cost  of higher tube side pressure drop。 However shell side pressure drop decrease by 11% by decreasing shell  side velocity and using lesser number of baffles。 As a net effect 66。3% increment in  the  annual  pumping  cost is observed in the present case。 Another advan- tage of the present design that the final exchanger  conform TEMA standard and satisfies all the  geomet-