A new method for determining where gas detectors should be installed in industrial facilities could substantially reduce the computational modelling required while maintaining the accuracy needed to identify hazardous gas clouds.
Researchers at the University of Campinas in Brazil developed the approach to address a drawback of existing gas-detector optimisation methods: their reliance on large numbers of computational fluid dynamics (CFD) simulations. These simulations model how gas released from a leak disperses around equipment and structures, but can require substantial computing resources and increase the time needed to design a detection system.
The researchers instead combined a mathematical model based partly on the kinetic theory of gases with a limited number of CFD simulations. The model estimates the volume of a flammable gas cloud under different leak and ventilation conditions, while Monte Carlo simulations and exceedance curves are used to account for the probability of different leak and wind scenarios.
The resulting information is then used to optimise detector positions, with the objective of providing the required coverage using the minimum number of detectors.
The method also incorporates “2-of-N” voting logic, under which at least two detectors must respond to a hazardous gas cloud. This provides redundancy and reduces the likelihood that a single detector response will determine whether an alarm is triggered.
Testing showed that the mathematical model estimated flammable cloud volumes with accuracy comparable to full CFD simulations, while substantially reducing the number of CFD runs required and therefore computational time. Independent CFD scenarios not used to develop the model were also employed to test its robustness.
The researchers said the combination of reduced modelling requirements and voting logic could make optimisation of industrial gas-detection networks more computationally efficient while maintaining system reliability.
The study, “A novel gas detector optimisation method combining 2-of-N voting systems, kinetic theory of gases, and exceedance curves to minimise CFD requirements”, was published in Scientific Reports.
