This article is devoted to the problem of automation of the stage of combining wells into clusters, considered as part of the process of designing the development of oil fields. The solution to the problem of combining wells into clusters is to determine the best location of well pads and the distribution of wells into clusters, in which the costs of developing and maintaining an oil field will be minimized, and the expected flow rate will be maximized. One of the currently used approaches to solving this problem is the use of optimization algorithms. At the same time, this task entails taking into account technological limitations when searching for the optimal option for the development of an oil field, justified, among other things, by the regulations in force in the industry, namely, the minimum and maximum allowable number of wells in a pad, as well as the minimum allowable distance between two well pads. The use of optimization algorithms does not always guarantee an optimal result, in which all specified constraints are met. Within the framework of this study, an algorithm is proposed that allows us to work out the resulting design solutions in order to eliminate the violated restrictions at the optimization stage. The algorithm consistently solves the following problems: violation of restrictions on the ultra-small and ultra-large number of wells in a pad; discrepancy between the number of pads with a given one; violation of the restriction of the ultra-close arrangement of pads. To study the effectiveness of the developed approach, a computational experiment was conducted on three generated synthetic oil fields with different geometries. As part of the experiment, the quality of the optimization method and the proposed algorithm, which is a raise to the optimization method, were compared. The comparison was carried out on different values of optimization power, which denotes the maximum number of runs of the target function. The evaluation of the quality of the work of the compared approaches is determined by the amount of the fine, which indicates the degree of violation of the values of the main restrictions. The efficiency criteria in this work are: the average value, the standard deviation, the median, and the minimum and maximum values of the penalty. Due to the use of this algorithm, the value of the penalty for the first and third oil fields is reduced on average to 0.04 and 0.03 respectively, and for the second oil field, the algorithm allowed to obtain design solutions without violating restrictions. Based on the results of the study, a conclusion was made regarding the effectiveness of the developed approach in solving the problem of oil field development.
