Heat Transfer Enhancement in Pulsating Flows: A Bayesian Approach to Experimental Correlations

Abstract

Abstract This research focuses on improving the modeling of heat transfer effects in pulsating exhaust flows. We address the challenges of understanding oscillating flow by employing the Metropolis–Hastings Markov Chain Monte Carlo sampling method for parameter estimation, accounting for measurement uncertainties. The knowledge can be applied to waste heat from reciprocating devices, pulsating turbocharger performance, and flow fields with significant cyclic variations. We demonstrate the feasibility of characterizing heat transfer capacity in pulsating flows using Bayesian inference and polynomial regression for experimental data correlation. This methodology is furthermore applied to identify heat transfer patterns in cold gas flow through a heated pipe across a range of mass flowrates and pulsating frequencies. To achieve this, the thermal performance variations across the length of the pipe through temperature and pressure changes are quantified. The model developed exhibits robust performance and high data efficiency (R2∈[0.83,0.89]) and notable extrapolation capacity in predicting mean heat transfer behavior based on boundary measurements. The results address the lack of experimental insights into pulsating flows encountered in heavy-duty transport applications and can be extended for heat recovery in systems such as exhaust manifolds and organic Rankine cycle gas turbines.