Gas Dispersion Simulator for Developing Gas-Source Localization Algorithms

This paper reports on an improvement we have made to our gas source localization simulator. Realization of robots that can localize gas sources is awaited in various fields including industrial plant safety and landfill methane emission monitoring. We have been building a simulator to promote the development of such robots. A highly complicated gas distribution generated by turbulent wind is the biggest challenge for gas source localization robots. Our simulator uses wind data recorded in a real outdoor environment to reproduce large wind fluctuations. However, the assumption of the wind uniformity made in calculating gas dispersion has often resulted in a gas distribution with a too simple shape. Here we propose to add an eddy-like noise field to represent the non-uniformity of the turbulent airflow field. The noise field is synthesized to satisfy the frequency characteristics of the recorded wind. As a result, a realistic and complex gas dispersion distribution can be successfully simulated. The results of gas source localization simulations show that the performance of a source localization algorithm is surely affected by the complex gas distribution, suggesting the importance to incorporate the turbulent eddy motion in gas dispersion simulations.