基于数理统计的海洋平台燃爆数值建模优化研究

海洋平台燃爆风险分析的关键基础之一是数值模型的建立,然而由于没有网格、障碍物拥塞度等的统一建模标准,难以保障超压载荷结果预测精度。基于数理统计理论离散超压及其影响因素的数学物理方程,构建超压后果集、模型集与工况条件集;进而借助拉丁超立方抽样,抽取不同工况条件组成的向量空间,以此进行模型建立的网格优化、物理模型优选方案及相似增补方案的分析;最后,以模型预测与理论模型结果的相关系数为优化结果考核指标,获取具有良好工程适用性的数值网格尺寸、障碍物删减尺寸等。结果表明:海洋平台燃爆数值模型核心区域的网格尺寸为0.2 m;对于在役阶段的平台,建模过程中可直接忽略等效直径小于0.06 m的障碍物,可采用均匀圆管替代等效直径小于0.08 m的障碍物;对概念设计阶段的平台,可将等效直径为0.08 m的圆管用于数值建模,代表平台拥塞度,从而实现在保障超压预测准确度的基础上较大程度地减轻建模工作量。

Optimized scheme of numerical modeling for explosion of offshore platforms based on mathematical statistics

The numerical modeling is one of the key foundations for the explosion risk assessment of offshore platforms. However, there is a lack of unified modelling standards for grid size, obstruction congestion etc., and thus the prediction accuracy of the explosive overpressure is difficult to guarantee. This paper aims to propose a procedure to derive the optimized numerical modelling scheme based on the statistics theory. Firstly, the numerical equations of the overpressure and its influencing factors are discretized to get the overpressure set, model set and condition set. Then, the analysis of the grid optimization, physical model optimization and similar supplementary plan is carried out based on the vector space derived from the Latin hypercube sampling of condition set. At last, the correlation coefficient between the results of numerical model and theoretical model is used to determine the suitable grid size, the equivalent diameter of deleted obstacle for engineering. The result shows that the size of the grid in the core area of the numerical model for explosion risk assessment of offshore platforms is recommended as 0.2 m. For the platform in service, the obstacles with equivalent diameter less than 0.06 m can be ignored during modelling, and the obstacles with equivalent diameter less than 0.08 m can be replaced by the equivalent uniform tubes.For the platform under design, the tube with diameter of 0.08 m can be used to represent the congestion of the platform. The conclusion can help to reduce the workload of modelling while ensuring the accuracy of overpressure prediction.