Convolutional neural networks for fluid flow analysis: toward effective metamodeling and low-dimensionalization

TL;DR

This paper investigates how various CNN operations and configurations affect fluid flow analysis, emphasizing optimal input placement, autoencoder use, and boundary handling to improve metamodeling and low-dimensional representations.

Contribution

It provides insights into the influence of CNN design choices, such as input placement and padding, on fluid flow analysis performance, aiding better CNN architecture design.

Findings

Careful placement of scalar inputs improves estimation accuracy.

Zero padding performs well for boundary conditions in fluid data.

CNN robustness varies with dimensional reduction and extension methods.

Abstract

We focus on a convolutional neural network (CNN), which has recently been utilized for fluid flow analyses, from the perspective on the influence of various operations inside it by considering some canonical regression problems with fluid flow data. We consider two types of CNN-based fluid flow analyses; 1. CNN metamodeling and 2. CNN autoencoder. For the first type of CNN with additional scalar inputs, which is one of the common forms of CNN for fluid flow analysis, we investigate the influence of input placements in the CNN training pipeline. As an example, estimation of force coefficients of an inclined flat plate and two side-by-side cylinders in laminar flows is considered. We find that care should be taken for the placement of additional scalar inputs depending on the problem setting and the complexity of flows that users handle. We then discuss the influence of various parameters and operations on the CNN performance, with the utilization of autoencoder (AE). A two-dimensional turbulence is considered for the demonstration of AE. The results of AE highly rely on the decaying nature. Investigation on the influence of padding operation at a convolutional layer is also performed. The zero padding shows reasonable ability compared to other methods which account for the boundary conditions assumed in the numerical data. Moreover, the effect of the dimensional reduction/extension methods inside CNN is also examined. The CNN model is robust against the difference in dimension reduction operations, while it is sensitive to the dimensional extension methods. The findings of this paper will help us better design a CNN architecture for practical fluid flow analysis