大气扩散计算流体力学模型综合评估方法研究

模型评估方法研究是模型研发工作的重要组成部分。基于科学性验证与统计性确认提出了大气扩散CFD模型的综合确认评估方法,通过示例的方式展示了方法的使用效果。综合确认评估方法可以提高模型筛选的速度,降低统计性评估对确认性试验数据的需求量,从总体上提高模型确认的效率。评估理论的研究有助于提高基于模型的大气扩散研究的准确性,也有利于高精度模型和试验的设计、开发与遴选。

A renovated comprehensive assessment method for CFD models for atmospheric dispersion

This paper would like to propose a renovated comprehensive assessment method for CFD models for atmospheric dispersion so as to develop some certain efficient and effective means for assessing performance of computational fluid.dynamics models through the following four steps.First of all,it is necessary to illustrate and discuss the principal procedures(i.e.define and determine as many typical dispersion characteristics as possible for the verification assessment basis) to be used for the verification assessment of atmospheric dispersion models.Any computational fluid dynamics models for dense gas atmospheric dispersion should be capable to provide visual physical results that can reflect at least the three above mentioned characteristics qualified enough to pass the model verification and validation assessment.Secondly,the model should propose the metric indexes of statistical validation,such as the geometric mean bias,the geometric mean variance and fraction of the predictions within a factor of two of observations that have been done,and the minimum demands for models to be accepted.And,thirdly,it is also necessary to define and determine the comprehensive validation assessment method and the flow chart of its working process.Finally,it is also to work out a performance assessment model for the dense gas dispersion as a sample and present the procedures of the new method and assessment outputs (including but not limited to the key physical phase reappearance,the diagram of the geometric mean bias vs the geometric mean variance,as well as a distribution chart for the fraction of the predictions within a factor of two observations).Thus,a comprehensive assessment method should not only serve mainly as the assessment tool for the computational fluid dynamics model and enhance the efficiency of the model selection,but also should help to reduce the amount of calculation in assessing the models through statistical validation.From what is mentioned above,the renovated method under study in this paper can improve the speed of model validation on the whole.And,in turn,the work we have done at the assessment theory can not only help to enhance the accuracy of atmospheric dispersion models,but can also heighten the selection and designing levels more precisely with the models and experiments brought up to a higher precision.