正交试验法在地下水数值模拟敏感性分析中的应用

Sensitivity analysis of groundwater numerical simulation using orthogonal experiment

地下水数值模型中参数的可靠性是决定模型可靠性的重要因素,进而也影响到利用地下水数值模型开展模拟和预报结果的可靠性,因而开展地下水数值模拟中参数的敏感性分析十分重要。但由于模型中参数众多,且各种参数间多彼此相互影响,使得敏感性分析十分复杂而困难。以往的数值模拟中进行敏感性分析的不多,方法单一且存在一些不足,结果的可靠性也无法验证。本文尝试性地将正交试验法用于地下水数值模拟的敏感性分析中,该法用部分试验代替全部试验,不仅大大简化了分析过程,而且得到了理想的敏感性分析结果。在本文的研究实例中,渗透系数对计算结果的影响最大,且远大于排在第二位的给水度,而贮水率的改变对计算结果几乎没有影响。

The reliability of parameters in groundwater numerical model is a key factor that not only plays an important role in determining the reliability of the model,but also influences the process of modeling and the reliability of prediction results.Thus,it is very important to carry out a sensitivity analysis for each groundwater numerical model during modeling process.However,the sensitivity analysis is a very complicated issue because there are so many parameters in each model,and each of them is possibly affected by the others.Consequently,there are a few models which ever carry out sensitivity analysis out of models established past.At the same time,the techniques used in these models are very simplex and have some deficiencies,and the reliability of results derived from these techniques cannot be verified.In the interest of full analyses replaced by partial analyses,and simplifying the process of sensitivity analysis,the orthogonal experiment method is firstly introduced to the sensitivity analysis by this paper,which is finally used to get the conclusion of sensitivity analysis.Results of the case of this paper indicate that,the ranking of parameters influencing the simulation results is： hydraulic conductivity,specified yield,and specified storativity,but the first parameter is far more important than the second one,and changes of the value of the third one only cause little changes of simulation results.