Skip to main navigation Skip to search Skip to main content

Accelerating Accurate Prediction of Real-Fluid Thermodynamics in Ansys for GPU-based CFD

  • Navneeth Srinivasan
  • , Dan Williams
  • , Saurabh Ranjan
  • , Suo Yang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This work focuses on the development and implementation of the Ansys Fluid Properties (AFP) library module, designed to accelerate accurate real-fluid thermodynamics for GPU-based computational fluid dynamics (CFD). The module provides robust state equation capabilities across Python, Fortran, C, and C++ interfaces, enabling execution on both CPU and GPU architectures. The AFP library consists of two parts – the EoS library and the NIST REFPROP library. The EoS library provides access to cubic equations of state (EoS) such as the Peng-Robinson (PR) and Soave-Redlich-Kwong (SRK), to perform thermodynamic calculations for both pure components and multi-component mixtures. To address the predictive errors often associated with standard cubic EoS models, particularly for hydrocarbons, the library incorporates a volume translation (VT) method. The library also supports both ideal-and real-fluid mixture models, utilizing mixing rules to estimate EoS parameters and fugacity equivalence for saturation (i.e., vapor-liquid equilibrium, VLE) calculations. A comprehensive suite of VLE flash solvers – including isothermal-isobaric (TP), isenthalpic-isobaric (HP), isentropic-isobaric (SP) etc. – is implemented to handle the diverse thermodynamic queries required by CFD solvers. The second part of AFP called the NIST REFPROP library utilizes the FORTRAN-based REFPROP distribution to generate real-gas property (RGP) tables. This approach moves the expensive generation of state points to a pre-processing step. The tabulation is error-controlled and recursively bisects the temperature-pressure space to maintain local accuracy. A novel feature of this tabulation strategy is the handling of element cuts by spinodal or saturation lines, ensuring precise resolution of phase boundaries without over-resolving gradients near the critical point. Finally, the performance metrics determined by defined state point tests on different hardware (CPU and GPU) of the library are discussed.

Original languageEnglish (US)
Title of host publicationAIAA Science and Technology Forum and Exposition, AIAA SciTech Forum 2026
PublisherAmerican Institute of Aeronautics and Astronautics Inc, AIAA
ISBN (Print)9781624107658
DOIs
StatePublished - 2026
EventAIAA Science and Technology Forum and Exposition, AIAA SciTech Forum 2026 - Orlando, United States
Duration: Jan 12 2026Jan 16 2026

Publication series

NameAIAA Science and Technology Forum and Exposition, AIAA SciTech Forum 2026

Conference

ConferenceAIAA Science and Technology Forum and Exposition, AIAA SciTech Forum 2026
Country/TerritoryUnited States
CityOrlando
Period1/12/261/16/26

Bibliographical note

Publisher Copyright:
© 2025, American Institute of Aeronautics and Astronautics Inc, AIAA. All rights reserved.

Fingerprint

Dive into the research topics of 'Accelerating Accurate Prediction of Real-Fluid Thermodynamics in Ansys for GPU-based CFD'. Together they form a unique fingerprint.

Cite this