Session: 10-09-01: Industrial Flows

Paper Number: 150912

150912 - Accelerating Multiphase Flow Simulations With the Volume of Fluid (Vof) Model in the Ansys Fluent Native Multi-Gpu Solver 

The Ansys Fluent GPU solver was introduced in 2022 as a cutting-edge computational fluid dynamics (CFD) tool that harnesses the power of graphical processing units (GPU) to dramatically accelerate simulation times.  This native multi-GPU solver is designed to run entirely on GPUs, and is compatible across various hardware configurations, from laptop-level GPUs to multi-GPU servers.  It offers several significant advantages over traditional CPU-based (central processing unit) CFD solvers such as exceptional runtime reduction for large-scale simulations, while simultaneously reducing hardware expenses and energy consumption compared to CPU systems that provide equivalent performance.  Over the last two years, the Fluent multi-GPU solver has been developed in a staged manner, initially implementing fundamental physics to enable single-phase flow and heat transfer simulations, followed by progressively adding more complex capabilities such as scale-resolving turbulence models like Large Eddy Simulations (LES), sliding mesh capability to model rotating machinery, compressible and reacting flows for combustion, and aeroacoustics.  In a further step-jump enhancement of the solver, multiphase flow capability will be enabled in the form of the Volume of Fluid (VOF) model.  The VOF method is primarily used for modeling free-surface flows where there is a clearly-defined interface between two or more immiscible fluids.  This model is crucial for studying free-surface flow applications such as wave dynamics in oceans or droplet formation in lab-on-chip devices, liquid-gas flows in spray atomization, and liquid-liquid flows in oil pipelines, to name a few.   It is well-known that industrial-scale multiphase flows simulations are time-consuming on traditional CPU-based systems, and impractical in several instances owing to their computational demand.  These simulations typically require high spatial and temporal resolution to adequately capture the interfacial dynamics and complex physics such as surface tension, often necessitating days or even weeks of runtime on traditional CPU systems.

In this presentation, we will showcase several applications solved with the new Volume of Fluid (VOF) model in the Ansys Fluent GPU solver.  The focus of this effort is two-fold: first, we will assess the accuracy of the solver by examining fundamental use cases such as dam break, fluid sloshing in a storage tank and steady-state vortex formation in a mixing tank, by validating the CFD solution with experimental data (wherever available) and by comparing the results with the well-established Fluent CPU solver solution.  Then, we will focus on industrial-scale problems including automotive applications like engine coolant jacket filling, and filling processes in the consumer products and pharmaceutical industries, where we will emphasize the performance and speed-up achieved compared to traditional CPU-based approaches.  In these examples, we will highlight the significant reduction in solution times with the GPU compute capability, scalability with multiple GPUs, and the potential impact on design iteration cycles.  Through the presentation, we will demonstrate how the Ansys Fluent GPU solver with VOF capability can maintain high accuracy while  dramatically accelerating simulation times for industrial applications, making it feasible to perform complex multiphase flow simulations that were previously impractical on CPUs,  thereby potentially transforming CFD workflows in product development.  By significantly reducing the turnaround time of simulation, this opens the door not just for product optimization, but also for generating enough data that can be used as the building blocks for AI/ML-based models.

Presenting Author: Santosh Konangi Ansys Inc.

Presenting Author Biography: Santosh Konangi works as a Senior Application Engineer at the Ansys office in Minneapolis. His expertise is in the computational fluid dynamics (CFD), particularly free-surface and multiphase flow applications. Prior to joining Ansys he received his PhD in Mechanical Engineering from the University of Cincinnati, Ohio, USA.

Authors:

Santosh Konangi Ansys Inc.
Sreenivas Viyyuri Ansys Inc.
Hossam Metwally Ansys Inc.
Palash Waghmare Ansys Inc.