Roughness Related to Cooling Performance of Channels Made Through Additive Manufacturing

Alexander J. Wildgoose, Karen A. Thole, Erika Tuneskog, Lieke Wang

Research output: Contribution to journalArticlepeer-review

Abstract

The complex surface morphology and multiscale surface features inherent in additively manufactured (AM) components contribute to the overall flow characteristics and heat transfer of cooling passages. As the AM process and cooling data in the literature continue to evolve, so does the need for more accurate heat transfer and pressure loss correlations for AM cooling schemes. This study improves the predictability of pressure loss and heat transfer for AM cooling passages by fabricating a range of coupons and investigating samples in the literature. Twenty-seven test coupons were manufactured using direct metal laser sintering in an assortment of build directions and build locations that produced a variety of surface morphologies. Nondestructive evaluation, computed tomography scanning, was used to quantify the surface morphology as well as capture the as-built geometric dimensions of the cooling schemes. The friction factor and bulk Nusselt number of the coupons were measured using an experimental rig. Pressure loss and heat transfer correlations in the literature were compared with the experimental results from the current coupons and datasets from the literature. Arithmetic mean roughness correlations in the literature struggled to predict the cooling performance of AM channels since the bulk roughness statistic did not capture the overall form of the surface morphology. A combination of root mean square roughness and skewness of the roughness was able to best predict pressure loss and heat transfer for the present samples and those in the literature while being independent of build location, build direction, material, machine, and laser parameters. The maximum absolute error was 25% and the average absolute error was 12% for the friction factor correlation. The maximum absolute error was 39% and the average absolute error was 8% for the Nusselt Number correlation.

Original languageEnglish (US)
Article number051008
JournalJournal of Turbomachinery
Volume146
Issue number5
DOIs
StatePublished - May 1 2024

All Science Journal Classification (ASJC) codes

  • Mechanical Engineering

Cite this