Turboshaft Engine Weight Estimation Model Development Based on Statistical Data

This study is conducted to address the inability to predict design objectives, such as weight and cost, that cannot be analytically determined during the preliminary design phase of jet engines. For this purpose, a model establishing the relationship between the design parameters used in the preliminary design stage and engine weight is developed. The engine design parameters affecting weight, along with the corresponding weight data, are obtained from engine data available in the literature, and a database is formed. Using this database, a model is developed in the MATLAB® environment through a regression method. During model development, two criteria reported in the literature—the minimum error criterion and the weight trend reflection criterion—are employed. The distinguishing feature of this study and its contribution to the literature lies in the application of weight model development criteria to the development of a turboshaft engine weight model, thereby introducing a turboshaft engine weight model to support the turboshaft engine design process. To increase model accuracy, turboshaft engines are classified into two categories: light-weight and heavy-weight engines. For the light-weight turboshaft engine category, a model is developed based on 48 engine cases, yielding a maximum error (E) of approximately ±40%, a Root Mean Squared Error (RMSE) of 35, an R-Squared (R^2) value of 0.255, and an Adjusted R-Squared (〖R^2〗_adj) value of 0.238. For the heavy-weight turboshaft engine category, a model is developed based on 16 engine cases, resulting in a maximum E of approximately ±30%, an RMSE of 36.1, an R^2 value of 0.306, and an 〖R^2〗_adj value of 0.256. Examination of the statistical parameters indicates that the model’s ability to represent reality is limited; this outcome is a natural consequence of developing a model based on a relatively small sample size, and differences in the technological levels of the engines included in the database constitute an additional contributing factor.