Precision Tailoring Cooling Systems using Linear Regression

Abstract: This bachelor thesis is written for KTH Royal Institute of Technology and Ymer Technology AB. It examines the friction factor associated with the pressure drop in wavy-fin heat exchangers. Previous studies have used a nonlinear approach for determining the regression coefficients. This study introduces a linearizing transform that solves the issues of finding non-optimal parameter values stemming from the current nonlinear approach. The linearization also drastically improves the computational time. A final model explaining 95.4\% of the variance in the data is presented. Methods for avoiding overfitting, detecting outliers and hidden extrapolation are presented and used to evaluate the model. The implications of the mathematical results on improving profit, competitiveness, and customer relationships are analyzed in the second part of the report with a qualitative approach.