Experimental, numerical and deep learning modeling study of heat transfer in turbulent pulsating pipe flow.

Publication Date

Spring 5-1-2024

Description

Pulsating heat transfer is important in many engineering applications but is not fully understood or easily modeled. This work helps understand the phenomenon and demonstrates the efficacy and utility of deep learning modeling for pulsating heat transfer. The convective heat transfer coefficient (HTC) of pulsating flow through a copper tube was experimentally investigated at seventy-seven different operating conditions (500,000 time-series data points were collected at each operating condition) corresponding to different pulsating waveforms, frequencies, mass flow rates and pressures. The intensity of pulsation ranged from mild to strong enough to cause reverse flow. The literature on pulsating HTC is mixed, with several researchers reporting increased heat transfer and several other reporting decreased heat transfer relative to steady flow. The experimental data showed that both could be true depending on the ratio of the amplitude to mean flow velocity (amplitude ratio AR). The enhancement ratio (ER) of pulsating HTC relative to steady-flow HTC was measured to vary between 0.48 and 2.18 for AR ranging from 0.64 to14.42. A numerical study showed that this was only a (positive) correlation; fundamentally the overall pulsating HTC could be approximately explained by the variation of instantaneous Reynold’s number (Re) during the pulsating cycle. The distribution of instantaneous Re and hence the ER is primarily determined by the pulsating waveform, and this explains the disparate observations by different researchers. Long Short-Term Memory (LSTM) networks have been used to model pulsating HTC as a function of pulsating waveforms occurring across a measurement orifice. The ‘Toy Model’ concept where physics-based variables are used to increase generalization was implemented with the LSTMs for increased model robustness. These models could be used to maximize pulsating heat transfer by optimizing the waveform.

Journal

Applied Thermal Engineering

Volume

244

Issue

122685

Department

Mechanical Engineering

Second Department

Mechanical Engineering

Comments

Non peer reviewed version here:

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4576633

DOI

https://doi.org/10.1016/j.applthermaleng.2024.122685

This document is currently not available here.

Share

COinS