A data-based real-time petrochemical gas diffusion simulation approach on virtual reality

Abstract

Background

Petrochemical products are flammable, explosive, and toxic, petrochemical accidents are generally extremely destructive. Therefore, disaster analysis and prediction and real-time simulation have become important means to control and reduce accident hazards.

Methods

In this study, a complete real-time simulation solution of gas diffusion with coordinate data and concentration data is proposed, which is mainly aimed at the simulation of the types of harmful gas leakage and diffusion accidents in the petrochemical industry. The rendering effect is more continuous and accurate through grid homogenization and trilinear interpolation. A data processing and rendering parallelization process is presented to improve simulation efficiency. Combines gas concentration and fragment transparency to synthesize transparent pixels in a scene. To ensure the approximate accuracy of the rendering effect, improve the efficiency of real-time rendering, and meet the requirement of intuitive perception using concentration data, a weighted blended order-independent transparency with enhanced alpha weight is presented, which can provide a more intuitive perception of hierarchical information of concentration data while preserving depth information. In this study, three order-independent transparency algorithms, depth peeling algorithm, weighted blended order-independent transparency and weighted blended order-independent transparency with enhanced alpha weight, are compared and analyzed for rendering image quality, rendering time, required memory, hierarchical information and so on.

Results

Using weighted blended order-independent transparency with enhanced alpha weight technique, the rendering time is shortened by 53.2% compared with the depth peeling algorithm, and the texture memory required is much smaller than the depth peeling algorithm. The rendering results of weighted blended order-independent transparency with enhanced alpha weight are approximately accurate compared with the depth peeling algorithm as ground truth, and there is no popping when surfaces pass through one another. At the same time, compared with weighted blended order-independent transparency, weighted blended OIT with enhanced alpha weight achieves an intuitive perception of hierarchical information of concentration data.