大兴迭隆起地区地下水流数值模拟的不确定性分析
发布时间：2018-08-20 18:20
【摘要】：地下水数值模拟的不确定性分析,一直是水文地质学家和研究者研究的重点和难点。本论文旨在通过研究钻孔数量、钻孔分布不确定性对模拟结果的影响,探讨水文地质钻孔布设的依据；通过对水文地质参数场进行趋势化研究,提升钻孔稀缺地区地下水流数值模拟的精度；选用变差函数分析结合克里格插值的方法研究钻孔数量和分布对模拟的影响,；利用蒙特卡罗等方法实现对参数场的趋势化处理。将上述研究应用于大兴迭隆起地区区域地下水流数值模型,并对大兴迭隆起地下水流模型的断层性质和渗透系数场进行了不确定性分析。为研究样本数据对模型模拟结果的影响,针对大兴迭隆起地区第四系沉积条件,构建了由3种岩性(渗透性)分布组成的算例模型,三种岩性分别为粗砂、中粗砂和中细砂,其对应的渗透系数分别为40m/d的、30m/d和20m/d。构建的算例模型剖分为50行×50列,单元格大小为100m×100m的二维潜水稳定流模型。通过分析样本量与岩性场模拟结果可知,在本算例尺度下,模拟误差在样本量增加的前期有着快速明显的下降,在样本量增加至300后,误差下降速度迅速减缓,样本数达到1000后,误差基本达到稳定状态。通过分析样本位置对岩性场模拟的影响可知,在本算例中样本量相同的情况下,随机平均采样时的岩性模拟结果较为理想,模拟平均误差约为5.4%；以最邻近指数(NNI)表示样本的聚集程度,研究表明非均匀采样的模拟误差随最邻近指数(NNI)的增加而减小。当最邻近指数大于0.9后,模拟误差将小于10%。利用蒙特卡罗方法对渗透系数场进行趋势化处理,使渗透系数场符合二维线性分布。对比随机渗透系数场与趋势化渗透系数场的模拟结果显示：两种方法的水流模拟结果基本相同,而趋势化渗透系数场可降低模拟整体误差1/2左右,模拟误差较大的地方基本集中在抽水井附近。在渗透系数场趋势化处理的基础上,利用Ucode软件可以对渗透系数场进行进一步的优化,优化后模拟误差最多可下降1/3左右。大兴迭隆起地区主要由松散孔隙潜水含水层、松散孔隙承压含水层、奥陶系、寒武系及青白口系灰岩岩溶裂隙含水层组成,岩溶含水层含水介质以裂隙、溶隙为主,具有北方岩溶的特征地下水流动状态仍符合达西定律。本次研究建立了大兴迭隆起地区5层非均质、垂向各项异性、空间三维结构、非稳定的地下水流数值模型。在此基础上,通过改变模型边界参数以及趋势化处理渗透系数场的方法对模型存在的关键问题进行了不确定分析。结果表明：区内的南苑-通县断裂基本可以断定为阻水断裂,而模型西北的永定河断裂侧向隔水,而在垂向上形成第四系含水层补给岩溶含水层的补给通道。对第四系含水层渗透系数场进行处理后,在含水层颗粒较粗的地区,渗透系数变化范围较广,趋势化后模拟精度提升明显；而在岩性为细砂和粘土的区域模拟水位值变化十分微弱,趋势化处理对拟合结果影响较小。
[Abstract]:Uncertainty analysis of groundwater numerical simulation has always been the focus and difficulty of hydrogeologists and researchers. This paper aims to study the number of boreholes, the influence of uncertainty of borehole distribution on the simulation results, and to explore the basis for the layout of hydrogeological boreholes. The accuracy of numerical simulation of groundwater flow in the area with scarce holes is studied. Variation function analysis combined with Kriging interpolation is used to study the influence of the number and distribution of boreholes on the simulation. Monte Carlo method is used to realize the trending treatment of parameter field. Uncertainty analysis of fault property and permeability coefficient field of groundwater flow model in Daxing-stack uplift is carried out. In order to study the influence of sample data on the simulation results of the model, an example model consisting of three lithology (permeability) distributions is constructed for the Quaternary sedimentary conditions in Daxing-stack uplift area. The three lithologies are coarse sand, medium coarse sand and coarse sand. The numerical model is divided into 50 rows by 50 rows, and the cell size is 100 m by 100 m. The simulation results of sample size and lithologic field show that the simulation error is rapid and obvious in the early stage of sample size increase. When the sample size increases to 300, the error decreases rapidly, and the error reaches a stable state after the sample size reaches 1000. By analyzing the influence of sample location on lithology field simulation, we can see that the lithology simulation results are ideal when the sample size is the same, and the simulation average error is relatively good. When the nearest neighbor index is greater than 0.9, the simulation error will be less than 10%. Monte Carlo method is used to trend the permeability field to make the permeability field conform to two-dimensional. The simulation results of the random permeability field and the trend permeability field show that the simulation results of the two methods are basically the same, while the trend permeability field can reduce the overall simulation error by about 1/2. The larger simulation error is concentrated near the pumping well. Basically, the permeability field can be further optimized by Ucode software, and the simulation error can be reduced by about 1/3 after optimization. The Daxingtai uplift area is mainly composed of unconsolidated pore phreatic aquifer, unconsolidated pore confined aquifer, Ordovician, Cambrian and Qingbaikou limestone karst fissure aquifer, and karst aquifer aquifer aquifer aquifer aquifer aquifer aquifer aqui The groundwater flow state with the characteristics of karst in the north still conforms to Darcy's law. In this study, a five-layer numerical model of heterogeneous, vertical anisotropy, three-dimensional structure and unsteady groundwater flow in Daxing-Die uplift area was established. On this basis, the seepage was treated by changing the boundary parameters of the model and trending. The results show that the Nanyuan-Tongxian fault in the area can be basically regarded as a water-blocking fault, while the Yongdinghe fault in the northwest of the model is lateral water-blocking, and the Quaternary aquifer is vertically formed to recharge the karst aquifer. After the permeability field is processed, the variation range of permeability coefficient is wider in the area with coarser aquifer particles, and the simulation precision is improved obviously after trending, while the variation of simulated water level is very weak in the area with fine sand and clay, and the trending treatment has little influence on the fitting result.
【学位授予单位】：中国地质大学(北京)
【学位级别】：博士
【学位授予年份】：2016
【分类号】：P641
[Abstract]:Uncertainty analysis of groundwater numerical simulation has always been the focus and difficulty of hydrogeologists and researchers. This paper aims to study the number of boreholes, the influence of uncertainty of borehole distribution on the simulation results, and to explore the basis for the layout of hydrogeological boreholes. The accuracy of numerical simulation of groundwater flow in the area with scarce holes is studied. Variation function analysis combined with Kriging interpolation is used to study the influence of the number and distribution of boreholes on the simulation. Monte Carlo method is used to realize the trending treatment of parameter field. Uncertainty analysis of fault property and permeability coefficient field of groundwater flow model in Daxing-stack uplift is carried out. In order to study the influence of sample data on the simulation results of the model, an example model consisting of three lithology (permeability) distributions is constructed for the Quaternary sedimentary conditions in Daxing-stack uplift area. The three lithologies are coarse sand, medium coarse sand and coarse sand. The numerical model is divided into 50 rows by 50 rows, and the cell size is 100 m by 100 m. The simulation results of sample size and lithologic field show that the simulation error is rapid and obvious in the early stage of sample size increase. When the sample size increases to 300, the error decreases rapidly, and the error reaches a stable state after the sample size reaches 1000. By analyzing the influence of sample location on lithology field simulation, we can see that the lithology simulation results are ideal when the sample size is the same, and the simulation average error is relatively good. When the nearest neighbor index is greater than 0.9, the simulation error will be less than 10%. Monte Carlo method is used to trend the permeability field to make the permeability field conform to two-dimensional. The simulation results of the random permeability field and the trend permeability field show that the simulation results of the two methods are basically the same, while the trend permeability field can reduce the overall simulation error by about 1/2. The larger simulation error is concentrated near the pumping well. Basically, the permeability field can be further optimized by Ucode software, and the simulation error can be reduced by about 1/3 after optimization. The Daxingtai uplift area is mainly composed of unconsolidated pore phreatic aquifer, unconsolidated pore confined aquifer, Ordovician, Cambrian and Qingbaikou limestone karst fissure aquifer, and karst aquifer aquifer aquifer aquifer aquifer aquifer aquifer aqui The groundwater flow state with the characteristics of karst in the north still conforms to Darcy's law. In this study, a five-layer numerical model of heterogeneous, vertical anisotropy, three-dimensional structure and unsteady groundwater flow in Daxing-Die uplift area was established. On this basis, the seepage was treated by changing the boundary parameters of the model and trending. The results show that the Nanyuan-Tongxian fault in the area can be basically regarded as a water-blocking fault, while the Yongdinghe fault in the northwest of the model is lateral water-blocking, and the Quaternary aquifer is vertically formed to recharge the karst aquifer. After the permeability field is processed, the variation range of permeability coefficient is wider in the area with coarser aquifer particles, and the simulation precision is improved obviously after trending, while the variation of simulated water level is very weak in the area with fine sand and clay, and the trending treatment has little influence on the fitting result.
【学位授予单位】：中国地质大学(北京)
【学位级别】：博士
【学位授予年份】：2016
【分类号】：P641
【相似文献】
相关期刊论文 前10条
1 梁杏;张人权;牛宏;靳孟贵;孙蓉琳;;地下水流系统理论与研究方法的发展[J];地质科技情报;2012年05期
2 万力;蒋小伟;;“区域地下水流:理论、应用与发展”国际研讨会在西安召开[J];水文地质工程地质;2013年05期
3 ;地下水是怎样活动的[J];水文月刊;1959年08期
4 李俊平;地下水流的声发射检测[J];世界隧道;1996年03期
5 魏林宏,束龙仓,郝振纯;地下水流数值模拟的研究现状和发展趋势[J];重庆大学学报(自然科学版);2000年S1期
6 仵彦卿;估计地下水流系统分布型确定性-随机性参数的耦合算法[J];西安理工大学学报;2000年02期
7 罗振东,田向军,张桂芳,周礼;地下水流及其通量的数值模拟[J];首都师范大学学报(自然科学版);2001年01期
8 杜国平;智能化地下水动态参数测量仪[J];核技术;2002年01期
9 胡立堂,陈崇希;地下水流数值模拟前后处理软件的设计与实现[J];地球科学;2003年05期
10 钱会,王毅颖,宋秀玲;地下水流数值模拟中不应忽视的几个工作程序[J];勘察科学技术;2004年01期
相关会议论文 前10条
1 王旭升;万力;;“地下水运动方程”课程设计[A];城市空间结构理论与资源型城市转型研究——中国科协第224次青年科学家论坛论文集[C];2010年
2 黄s，
本文编号：2194569
本文链接：https://www.wllwen.com/shoufeilunwen/jckxbs/2194569.html
