Computer modeling of heat transfer surfaces allows the designer to rapidly determine the performance of a particular heat exchanger configuration. For example, a portion of a heat pump system computer program can be used to model air-to-refrigerant heat exchangers. Using this evaporator heat exchanger model,a designer can iteratively determine the heat exchanger configuration required for a given application. By combining Artificial Intelligence (AI) and the evaporator computer model, an experienced designer can determine a heat exchanger configuration fasterand more accurately. A novice designer can produce better designs and learn the proper design methods. This paper will describe how to integrate the air-to-refrigerant heat exchanger model and AI using a PROLOG knowledge base. Applicable portions of the heat pump c mputer model and required modifications will be discussed. An overview of PROLOG will be presented along with an example of knowledge capture.