Structural modelling approach has been developed to support systematic, domain model–based knowledge acquisition for model–based diagnosis problem solving. Structural models capture declarative, deep knowledge about system morphology and operation in normal conditions as well under faults. Models of morphological and functional structure are built mainly using knowledge of problem domain experts. Formal methods and algorithms for building model of a morphological structure and its transformation into a model of a functional structure have been developed to reduce the workload of human experts. The paper focuses on model transformation issues and represents new results allowing to extend previously developed method of formal transformations. Analysis of different cases of logic of input and output flows of system components is carried out and corresponding transformations defined. The developed formal method is demonstrated for functional model building of cooling system of internal combustion engine.