1. Criteria for a successful heat exchanger design

Any discussion of the design process for heat exchangers must be based on a clear understanding of the criteria by which the heat exchanger will be judged. These criteria are simple to state in broad principle, but the apparent simplicity can be very misleading when the designer or customer actually attempts to apply them to a specific case. The principles are given below in an approximate order of importance, with a few examples of how the principles must be elaborated in practice.

First, it may be pointed out that there are two broad classifications of heat exchanger application, and they call into being quite different philosophies and procedures for their design and production. Undoubtedly the greater amount of heat transfer surface in existence is in units of which there are many replicates — automotive radiators, domestic and commercial air conditioners and furnaces, engine oil coolers, and so on. For these units, production runs of thousands and even millions of nominally identical units are common. The design process for this class is modified by the opportunity — indeed, the economic compulsion — to build a number of variant candidate units and test these extensively over the range of conditions expected, ultimately selecting the best design and thenceforth attempting to build myriads more as close to the prototype as possible.

Contrast this with the typical heat exchanger found in a chemical or petroleum plant: a "one-off" design (duplicate items are usually intended to be used in series or parallel with one another), with no opportunity for testing until the plant becomes operational. Frequently these heat exchangers are to be used with streams whose compositions, properties, and fouling characteristics are poorly known and which, together with the flow rates and process specifications, may vary from day to day. Clearly this situation places greater demands on the design process if there is to be a high probability of success. This part of the handbook tends to emphasize design for the second case.

The first criterion that a heat exchanger should satisfy is the fulfilment of the process requirements: to accomplish the specified thermal change on the streams within the allowable pressure drops, and to retain the capability to do this in the presence of fouling until the next scheduled maintenance period However, it must be recognized that the design process is fraught with uncertainties: The physical properties are seldom known to a high degree of precision, the design methods incorporate basic correlations of data with various degrees of experimental scatter, the exchangers are constructed only within certain dimensional limits, the actual operating conditions and process stream characteristics vary from day to day, and the fouling characteristics are little more than guesses and vary with time in any case. Therefore, meeting process requirements is at best only a matter of statistical probability. At this point, too little is known about either the statistics of individual units or their interaction in heat exchanger trains and with oilier process plant components to allow for a quantitative treatment. [Some indication of a possible approach to the problem is given in Al-Zakri and Bell (1981).] Hence, the designer must assure himself of a reasonable probability of success by judicious overdesign and by taking advantage of the operational flexibilities and reserve capacity inherent in the rest of the plant. It is highly desirable but not often possible for the heat exchanger designer to work closely with the system designer to achieve a true optimal design.