Analysis of Thermal Performance of Shell and Tube Heat Exchangers: A Correlation and CFD Based Approach

Abstract

Shell and tube heat exchangers are devices which are widely adopted in thermal systems for the transfer of thermal energy due to both performance and reliability factors. Given their application in systems that are energy intensive, the design and sizing of these devices have become a widely growing field. Traditionally, empirical correlations which were based on experimental results were used for thermal sizing and design. This was replaced by computational fluid dynamics (CFD) modelling given its ability to model and visualize flow, expanding the horizon of possibilities for design and performance optimization. As of recent, CFD has been combined with numerical methods such as non-linear least squares regression to develop correlations which can predict the thermal performance based on the input design parameters. However, the application of this integrated method for shell and tube heat exchangers is limited. This study will model a single pass TEMA E-type shell and tube heat exchanger using ANSYS Fluent ®. CFD simulations are used to explore the effect of turbulence on thermal performance by varying both the inlet mass flow rate and the central baffle spacing. As steady state simulations are conducted for four models of six, eight, ten and twelve baffles. The results of CFD modelling are then combined with non-linear least squares regression in the MATLAB Curve Fitter Toolbox ® to develop four sets of correlation in the form of 𝑁𝑢 = 𝐶. 𝑅𝑒𝑎. 𝑃𝑟𝑏. Reasonably confident results were obtained in the final fitted data, however, relatively high 95% confidence interval widths were evident for certain fitted coefficients leaving space for improvement in the model. The study highlights that combining CFD with tools such as nonlinear least squares regression aids both engineers and designers in the thermal design process of shell and tube heat exchangers eliminating the need to limit design based on empirical correlations.