关于CFD验证确认中的不确定度和真值估算

The uncertainty and truth-value assessment in the verification and validation of CFD

本文回顾了CFD验证、确认中不确定度的概念和研究方法,CFD的不确定度尚无表达式可以使用。本文也讨论了现在正在进行的实验验证,对各个参加验证的软件,如何作出定量的精度评价也缺乏原则。针对这些情况,我们在不改变不确定度定义的前提下,对不确定度作了新的解读,即不确定度可解读为计算值或实验值与真值准确到前n位,从而可给出不确定度的表达式和真值估算的原则。并根据大样本数据的统计理论,对真值认为接近数学期望,从而给出准确到n位真值的计算方法。这个方法,可用于计算结果的检验,例如当模型一定时,可用此法寻求计算方法的真值,对算法进行检验;如算法一定模型改变时,也可用于检验模型的可靠性。利用这种方法,在没有实验结果的情况下,也可评价各计算软件的质量。这个方法当然也可以运用处理实验数据。因为CFD中计算模型是人为建立的,虽然可以检验它的解是否正确,但与物理情况是否一致,并未得到回答。因此,开展实验验证是必需的。

The idea and definitions for verification and validation of CFD or EFD are reviewed.The uncertainty is defined as the maximum error between counting datum of CFD or testing datum of EFD and its correspondent truth value.Since the truth value calculated or tested is unknown,there is no explicit expression for the uncertainty.In this paper,a new expression of uncertainty is given.We consider that the counting or testing value can be approximated to the first n digil number of truth values.Based on this consideration,the uncertainty of CFD or EFD and the approximate truth value can be assessed.If there are a great of counting or testing sample data,according to statistical theory and uniform design method,we can give the calculation method to get the values which accurate to first n bits of the truth values and the uncertainty.Then the verification and validation of CFD or EFD can be performed agreeably.