基于随机森林和粒子群算法(RF-PSO)的泸州区块页岩气压裂施工参数优化

Optimizing shale-gas fracturing parameters in Luzhou block based on random forests and particle swarm algorithm(RF-PSO)

为了充分认识四川盆地南部地区泸州区块深层页岩气主体压裂施工参数的适应性,为其深层页岩气压裂施工参数的进一步优化提供参考,以该区块80口页岩气生产井为研究对象,结合储层静态物性参数,分析了各因素与前三个月累计产量的相关性,并利用随机森林算法建立产能预测模型,再以区块地质参数均值作为基准指标,通过粒子群算法,寻找模型最优产量的压裂施工参数组合。研究结果表明:①与前期产量相关性最高的三个参数分别是:含气量、脆性矿物含量、用液强度;②有机碳含量、水平应力差、用液强度、液体携砂效率、平均簇间距是影响模型前三个月累计产量的主要影响因素;③模型给出的最优压裂施工参数组合中,平均段长、平均每段孔数和目前的设计情况较为一致,模型其余最优参数:施工排量15.6m^(3)/min、用液强度35m^(3)/m,加砂强度3.52t/m,簇间距7.3m,液体携砂效率11t/100m^(3),40/70支撑剂比例54%。结论认为:①泸州区块压裂施工参数优化方向主要是“控排量、提液量、增砂量、缩短簇间距、提高液体携砂效率及40/70支撑剂比例”;②给出了最优压裂施工参数组合,可以为泸州区块深层页岩气压裂方案设计提供借鉴和参考。

In order to fully understand the adaptability of main fracturing parameters to deep shale gas in Luzhou block and also provide reference for further optimizing these parameters,80 shale-gas production wells in this block were taken as objectives to figure out the correlation between each factor and total production in the first three months based on static physical parameters.Then,a model for production prediction was established by means of the random forests algorithm(RFA).Finally,with the mean geological parameters as reference index,a fracturing-parameter combination available for the best production was achieved through the particle swarm algorithm.Results show that(i)there are three parameters highly correlated with early stage of production,like gas content,brittle-mineral content and fracturing-fluid intensity;(ii)the TOC,horizontal stress difference,fracturing-fluid intensity,liquid sand-carrying efficiency and average cluster spacing are the main factors influencing on the cumulative production in the first three months;(iii)in the optimal fracturing-pa⁃rameter combination offered by the model,average segment length and average perforation number in each segment are rel⁃atively consistent with the current design parameters,and the other optimal parameters include construction displacement of 15.6 m^(3)/min,fracturing-fluid intensity of 35 m^(3)/m,sanding intensity of 3.52 t/m,cluster spacing pf 7.3 m,liquid sandcarrying efficiency of 11 t/100 m^(3),and 40/70 proppant proportion of 54%,respectively.In conclusion,optimization schemes for fracturing parameters in Luzhou block are exactly"controlling discharge,increasing liquid and sand contents,reducing cluster spacing,upgrading liquid sand-carrying efficiency and 40/70 proppant proportion".What's more,the optimal frac⁃turing parameter combination may provide guidance and reference for design plan on deep shale-gas fracturing in Luzhou block.