Using lithology to average the properties of rock is a general approach in the construction of mechanical properties models. At the same time, lithotype is a kind of classification that does not use elastic properties of rocks directly. In this paper, clustering algorithms are considered for constructing mechanical facies models. Acoustic logging data is used as a basis for clustering. The clustering procedure is performed in the space of dynamic elastic modules and leads to the vertical stratification of the formation onto layers with similar acoustic properties. As the basis, authors use popular machine learning algorithms, which provide the necessary control (selection of the number of clusters, automatic calculation of the number of clusters) and the determinism of the solution. In addition, a voting method based on all the standard algorithms used is implemented. The clustering algorithms are implemented in the RN-SIGMA software as a separate module. One practical example shows the effect of the minimum allowable thickness of the interlayers on the clustering result. The results of clustering by mechanical facies are compared with the results of constructing a mechanical model based on lithology, general properties and differences are shown. Based on the constructed clustering model, calculations of hydraulic fracturing design for Domanic deposits were carried out. The influence of discretization method on the result of the design is shown. In addition, it is shown that the design calculation time is significantly reduced when clustering is used compared to the uniform discretization of the calculated grid by height. Using the clustering tool allows the engineer to either reduce the calculation time of a specific hydraulic fracturing design, or increase its accuracy by increasing the sampling without increasing the calculation time.
1. Nikolenko S.I., Tulup'ev A.L., Samoobuchayushchiesya sistemy (Self-learning systems), Moscow: Publ. of MTsNMO, 2009, 288 p.
2. Mandel' I.D., Klasternyy analiz (Cluster analysis), Moscow: Finansy i Statistika Publ., 1988, 176 p.
3. Lance G.N., Williams W.T., A general theory of classificatory sorting strategies: 1. Hierarchical systems, Comp. J., 1967, no. 9, pp. 373–380, DOI: https://doi.org/10.1093/comjnl/9.4.373
4. Syakur M.A., Khotimah B.K. et al., Integration K-means clustering method and elbow method for identification of the best customer profile cluster, IOP Conf. Series: Materials Science and Engineering, 2018, no. 336, pp. 1–6, DOI: 10.1088/1757-899X/336/1/012017
5. Aldenderfer M.S., Blashfield R.K., Cluster analysis, SAGE Publications, 1984.
6. Rand W.M., Objective criteria for the evaluation of clustering methods, Journal of the American Statistical Association, 1971, no. 66(336), pp. 846–850, DOI:10.2307/2284239.
7. Novikov A., PyClustering: Data mining library, Journal of Open Source Software, 2019, no. 4(36), 1230 p., URL: https://pyclustering.github.io/docs/0.9.2/html/index.html
8. Kuhn H.W., The Hungarian Method for the assignment problem, Naval Research Logistics Quarterly, 1955, V. 2, Issue 1–2, pp. 83–97, DOI: https://doi.org/10.1002/nav.3800020109