Jet engine performance simulation with compressor stage stacking models

A computer model of the J-85 gas turbine engine has been used in a investigation for potential benefits in performance simulation, arising from the adoption of compressor stage stacking models. The eight-stage, axial flow compressor of the engine was simulated by means of two separate stage stacking methods: the Howell program and the projects code, the latter being developed on the basis of the pre-existing J-85 complete engine simulation model. Results obtained from the use of the Howell program generally support its suitability for the prediction of single stage and overall compressor performance. This includes the capability of the code to render the effect from the Variation of specific design parameters and from the incorporation of variable geometry. Certain weaknesses originating from the empirical nature and the logic of the program are identified. Recommendations for the assessment of the validity of produced results, a well as modifications for the improvement of the code, are proposed. The projects stage stacking code was tested with sets of stage characteristics derived from different techniques. The produced results depend on the topology and shape of the utilised characteristics. A special feature embedded in the program's logic denotes the tendency of individual stages to work within their stall areas or below the hypothetical choke points of the corresponding characteristics. Although in certain instances the predicted overall compressor performance appears satisfactory, analysis of results in individual stages indicates the need for improvement of the code, in order to obtain a closer approach to the physical mechanism and the associated limitations of stage matching. This is also true for the prediction of variable geometry effects on the overall compressor performance. The complete engine simulation model incorporating the projects stage stacking code provides generally satisfactory results, a compared to those produced by a equivalent model of the Cranfield's Turbomatch Scheme. Specific problems encountered are attributed to certain weaknesses of the embedded stage stacking code, which is susceptible of improvement as already reported