Numerical investigation of non-premixed hydrogen flame dynamics using RANS models

Research output: Contribution to journalConference articlepeer-review

Abstract

Hydrogen has sparked major stakeholders as an alternative to fossil fuel in industrial gas turbines or aerospace applications, considering its clean energy to reduce greenhouse gas emissions. Furthermore, hydrogen is known to have higher gravimetric energy density than conventional fuel. However, understanding hydrogen flames is particularly challenging due to the intricate nature of the turbulence characteristics. Hence, non-premixed turbulent hydrogen flames are numerically investigated in this work using ANSYS Fluent with and without a user-defined function. This work aims to assess the predictability of various RANS turbulence models for an axisymmetric case, including standard, modified, realizable, and Pope correction models. Simulation of test cases has shown that the realizable model exhibits the best accuracy in the centerline region further downstream with high accuracy for the centerline mean temperature. The modified model demonstrates superior predictability near the injector exit plane in the radial direction. Overall, the findings suggest that RANS models are effective for initial data collection in the early stages of the design process, providing valuable insights into the behavior of hydrogen flames.

Original languageEnglish
Pages (from-to)657-664
Number of pages8
JournalTransportation Research Procedia
Volume84
DOIs
StatePublished - 2025
Event1st Internation Conference on Smart Mobility and Logistics Ecosystems, SMiLE 2024 - Dhahran, Saudi Arabia
Duration: 17 Sep 202419 Sep 2024

Bibliographical note

Publisher Copyright:
© 2024 The Authors. Published by ELSEVIER B.V.

Keywords

  • Hydrogen flames
  • Non-premixed turbulent flame
  • Numerical
  • RANS
  • User-defined function

ASJC Scopus subject areas

  • Transportation

Fingerprint

Dive into the research topics of 'Numerical investigation of non-premixed hydrogen flame dynamics using RANS models'. Together they form a unique fingerprint.

Cite this