Application of multivariate linear regression statistical and committee models to calculate the minimum water depth over a horizontal outlet pump at the bottom of urban open tanks

Important and abundant uses of open tanks (such as firefighting water tanks) in the municipal services complex have been defined, especially in critical situations. Inadequate and low submergence depth on the outlet of the tanks reduces the pumping efficiency (by forming vortices and air bubbles). Experimental terms have been performed based on limited laboratory studies which are not necessarily accurate enough. Statistical methods and soft computing tools have made significant progress in many engineering sciences due to their ability to model, analyze and control complex systems. In this study, multivariate linear regression statistical methods were used to calculate the minimum depth of a horizontal outlet pump immersed in open floor tanks. To apply the computation model and usability in a wide range of data, dimensionless input parameters have been used as training and validation data for the models. The lower errors and higher correlation coefficients of the results indicate that the multivariate linear regression model performs better than the previous well-known empirical and experimental methods. Then, the applicability and generalizability of the models were compared with previous studies, indicating the appropriate skill of this research. In the following, using sensitivity analysis of input parameters, the dimensionless parameter (the output speed to the input speed) is introduced as the most important and influential input parameter of the model and process. Also, the committee model derived from a geometric mean of the results of previous studies and this research shows a more efficient and accurate model to calculate the minimum submergence depth.

Keywords:

Open Tanks , Multivariate Linear Regression , Empirical Method , Committee Model