嵌入地下水动力模型的井灌区机井布局耦合优化模型

为探索井灌区适宜的机井布局方法,降低井灌区提水能耗,构建了以取水费用最小为目标的机井布局优化模型并把地下水动力模型(Modflow-2000)嵌入其中,通过2模型中的公共参数进行数据交换,从而形成机井布局耦合模型。利用实测资料对模型中相关参数进行率定,最后以通辽井灌区为研究对象,采用黄金分割搜索法对灌区机井位置进行寻优。应用结果表明,利用该耦合模型可以精准模拟灌区地下水位,模拟值与观测值的相对误差绝对值的平均值为0.88%,RMSE值为0.135 3 m;与灌区目前机井布局方式相比,优化后的机井布局方式可以有效减小区域机井水位降深,单次提水灌区井群地下水位累积降深值可减少5.96 m(降深平均值减少0.134 m);构建的耦合模型在确保模拟精度的前提下寻优速度快(单次运行时间仅需40 s)。为基于灌区地下水采补协调的机井空间布局优化提供了新的思路。

Coupling optimization model embedded groundwater dynamic model for distribution of pumping wells in well irrigation area

Abstract: It is important to study the method of well layout to reduce the irrigation energy consumption and to promote the sustainable utilization of water resources. This paper aimed to explore the suitable layout method of well to reduce the water consumption in well-irrigation area. The optimization model of well layout was constructed with the minimum cost of water intake, in which the groundwater dynamic model (Modflow-2000) was embedded. And the data were exchanged through the common parameters of the 2 models, thus forming the coupling model of well layout. Based on the measured data, the trend of the observation value of the water level was consistent with that of the calculated value at different time after the corresponding parameters of the model were determined. And the mean value of the absolute relative error was 0.88%, and RMSE (root mean square error) value was 0.135 3 m, which could meet the accuracy requirements of the optimal allocation of groundwater resources. It showed that the model could reflect the actual change trend of groundwater level in irrigation area. Finally, taking Tongliao well-irrigation area in Inner Mongolia as the research object, the golden section search method was used to find the optimal positions of the pumping wells. The fluctuation of the water level in wells showed that the irrigation quota in the research area was too large and the irrigation water should be reduced to keep the water resource balance. The results of single well optimization were different from those of the existing well-spacing method. It was found that there were no irrigation wells in the northeastern uncultivated land and villages, and the optimal well location was not in the center position of adjacent 2 wells in the single well optimization results. From the results of global optimization, there was a great change in the distribution of wells compared with the current method of well spacing. And the cumulative value of drawdown in single water extraction could reduce by 5.96 m, and the average value of drawdown could reduce by 0.134 m, which had a certain effect on the decrease of energy consumption in the community. The groundwater movement is a complex system and the model generalization is simpler. Moreover, there is a certain gap with the real flow field. But the application of the coupling model to the optimization of the distribution of well group can also be explained to some extent. The optimization process of single well has reached automatic optimization, and the Modflow-2000 model averagely needs to be operated 10 times in single operation. Meanwhile, the time for single well is about 8 min long. But the process of multi-well optimization is cumbersome. If the model scenario is set to the foundation pit dewatering and the well is set around the foundation pit, the coupling model can optimize the dewatering of the foundation pit; if the water injection is adjusted, the coupling model can also optimize the layout of the artificial recharge well.