Hydraulic Fracturing Candidate-Well Selection Using Artificial Intelligence Approach

Abstract

<p><strong>.</strong><em> </em>Hydraulic fracturing is one of the stimulation method that aimed to increase productivity of well by creating a high conductive conduit in reservoir connecting it to the wellbore. This high conductivity zone is created by injecting fluid into matrix formation with enough rate and pressure. After crack initiate and propagate, the process continue with pumping slurry consist of fracturing fluid and sand. This slurry continues to extend the fracture and concurrently carries sand deeply into formation. After the materials pumped, carrier fluid will leak off to the formation and leave the sand holds the fracture created. TLS Formation in X and Y Field is widely known as a formation that have low productivity since it has low permeability around 5 md and low resistivity 3 Ohm-m. Oil from TLS formation could not be produced without fracturing. This formation also have high clay content, 20 – 40 % clay. Mineralogy analysis also shown that this formation contains water sensitive clay such as smectite and kaolinite. Hydraulic fracturing has been done in this field since 2002 on around 130 wells. At the beginning of hydraulic fracturing campaign, the success parameter is only to make the wells produce hydrocarbon in economical rate. As the fractured wells become larger in number, several optimization is also been done to increase oil gain. Later on, the needs of conclusive analysis to evaluate well performance after hydraulic fracturing rise up due to sharp decrement of crude oil price. Accurate analysis and recommendation need to be conducted to assess the best candidate for hydraulic fracturing to maximize success ratio. Even though a common practice, candidate-well selection is not a straightforward process and up to now, there has not been a well-defined approach to address this process. Conventional methods are not easy to use for nonlinear process, such as candidate-well selection that goes through a group of parameters having different attributes and features such as geological aspect, reservoir and fluid characteristics, production details, etc. and that’s because it is difficult to describe properly all their nonlinearities. In that matter, Artificial Intelligence approach is expected to be an alternative solution for this condition.</p>

Machine Learning in Reservoir Engineering: A Review

“…For instance, during the design of chemical flooding optimization schemes, oil production or another key parameter is taken as the optimization objective, and an ANN is employed to obtain the corresponding optimization combination of design steps or parameters [41]. The selection of candidate wells is actually determined by predicting the income (oil production) of each potential oil well [28]. Therefore, the primary focus is on introducing numerical prediction methods.…”

“…(1) In the case of a lack of historical data in the pool, machine learning methods are mainly applied to select from geological aspects, as well as reservoir and fluid characteristics. For example, Aryanto A. et al [28] applied Adaptive Neuro-Fuzzy Inference System (ANFIS) to optimize the identification of fracturing wells in a pool of geological aspects to improve the success rate of well selection.…”

Well production after hydraulic fracturing in sandstone rocks in the north of the Perm region

“…The minimal correlation density for the productivity index after HF is observed with the flow rate of proppant. In [22], it is noted that the clay content of the rock skeleton also has a major impact on the efficiency of HF, as it increases, the pore permeability of the reservoir space decreases.…”

The Application of Neural Networks to Forecast Radial Jet Drilling Effectiveness

“…The problems of forecasting the effectiveness of oil recovery enhancement technologies, production stimulation and well stimulation are considered in References [36][37][38][39][40][41][42][43][44], where the following algorithms are mainly used: Shallow and Deep Artificial Neural Networks (ANN), Naïve Bayes (NB), Decision Tree (DT), Random Forest (RF) and Dimension Reduction using Principal Component Analysis (PCA); the model's correlation coefficients vary significantly between 0.6 and 0.9.…”