For the issue of proppant embedment in hydraulic fracturing,a new calculation method of embedment depth considering elastic-plastic deformation was proposed based on the mechanism of proppant embedment into rocks by c...For the issue of proppant embedment in hydraulic fracturing,a new calculation method of embedment depth considering elastic-plastic deformation was proposed based on the mechanism of proppant embedment into rocks by combining proppant embedment constitutive equations and contact stresses on the rock-proppant system.And factors affecting embedment depth of proppant were analyzed using the new method.Compared with the elastic embedment model,the results calculated by the new method match well with the experimental data,proving the new method is more reliable and more convenient to make theoretical calculation and analysis.The simulation results show the process of proppant embedment into rocks is mainly elastic-plastic.The embedment depth of monolayer proppants decreases with higher proppant concentration.Under multi-layer distribution conditions,increasing the proppant concentration will not change its embedment depth.The larger the proppant embedment ratio,the more the stress-bearing proppants,and the smaller the embedment depth will be.The embedment depth under higher closure stress is more remarkable.The embedment depth increased with the drawdown of fluid pressure in the fracture.Increasing proppant radius or the ratio of proppant Young’s modulus to rock Young’s modulus can reduce the proppant embedment depth.展开更多
传统浅层模型不能有效提取FTIR光谱数据的潜在特征。提出一种基于栈式自编码器SAE(Stacked Auto Encoder)的光谱识别方法。通过堆叠稀疏自编码器构建深度网络,采用逐层贪婪训练学习光谱特征,根据学习到的特征有监督地训练softmax分类器...传统浅层模型不能有效提取FTIR光谱数据的潜在特征。提出一种基于栈式自编码器SAE(Stacked Auto Encoder)的光谱识别方法。通过堆叠稀疏自编码器构建深度网络,采用逐层贪婪训练学习光谱特征,根据学习到的特征有监督地训练softmax分类器,使用反向传播算法对网络进行微调。对麻花秦艽和大叶秦艽的FTIR光谱进行识别,基于SAE的分类准确率为96.67%,比偏最小二乘判别分析(PLSDA)和模型集群方法分别提高13.34%和10%。实验结果表明,该方法用于秦艽的快速、准确鉴别是可行和有效的。展开更多
文摘For the issue of proppant embedment in hydraulic fracturing,a new calculation method of embedment depth considering elastic-plastic deformation was proposed based on the mechanism of proppant embedment into rocks by combining proppant embedment constitutive equations and contact stresses on the rock-proppant system.And factors affecting embedment depth of proppant were analyzed using the new method.Compared with the elastic embedment model,the results calculated by the new method match well with the experimental data,proving the new method is more reliable and more convenient to make theoretical calculation and analysis.The simulation results show the process of proppant embedment into rocks is mainly elastic-plastic.The embedment depth of monolayer proppants decreases with higher proppant concentration.Under multi-layer distribution conditions,increasing the proppant concentration will not change its embedment depth.The larger the proppant embedment ratio,the more the stress-bearing proppants,and the smaller the embedment depth will be.The embedment depth under higher closure stress is more remarkable.The embedment depth increased with the drawdown of fluid pressure in the fracture.Increasing proppant radius or the ratio of proppant Young’s modulus to rock Young’s modulus can reduce the proppant embedment depth.
文摘传统浅层模型不能有效提取FTIR光谱数据的潜在特征。提出一种基于栈式自编码器SAE(Stacked Auto Encoder)的光谱识别方法。通过堆叠稀疏自编码器构建深度网络,采用逐层贪婪训练学习光谱特征,根据学习到的特征有监督地训练softmax分类器,使用反向传播算法对网络进行微调。对麻花秦艽和大叶秦艽的FTIR光谱进行识别,基于SAE的分类准确率为96.67%,比偏最小二乘判别分析(PLSDA)和模型集群方法分别提高13.34%和10%。实验结果表明,该方法用于秦艽的快速、准确鉴别是可行和有效的。
