Detector configuration optimization based on wind tunnel tests using normalized adjoint concentration gradient for urban spatial source parameters estimation

Abstract

When hazardous gas leaks occur, accurate and timely identification of the leakage location and rate is crucial to ensure environmental quality and public safety. However, obtaining accurate source parameter estimations becomes challenging if detectors are not appropriately arranged, regardless of estimation algorithms used. Therefore, this paper proposes a method for detector configuration optimization in three-dimensional space. Based on the flow field and with clear physical meaning, this method defines source parameters sensitive areas through the normalized gradient magnitude of the adjoint concentration field. Genetic algorithm is utilized to search for the detector combination that maximizes source parameters sensitive areas, i.e., the optimum configuration. Two types of urban environmental wind tunnel tests are designed as research objects. The rationality of obtained optimum configurations is analyzed combined with the flow field characteristics. Using optimum configurations increase estimation accuracy by 186% (S4 in urban type Ⅰ) and 398% (S5 in urban type Ⅱ) compared to the uniform and random configurations, respectively. Additionally, variation patterns of estimation accuracies with different numbers of optimum detectors are analyzed to determine the proper numbers of detectors. For urban type Ⅰ, 20 detectors are ideal, but 10 is sufficient due to space and cost limitations. For urban type Ⅱ, 22 detectors are recommended, while 12 is also decent for estimation. Although some factors may affect the estimations, the proposed method exhibits high robustness; moreover, the research scenario resembles what happens in reality, indicating its strong potential for practical application.