Machine Learning in Modern Well Testing

Well testing is a crucial stage in the decision of setting up new wells on oil field. Decision makers rely on the metrics to evaluate the candidate wells’ potential. One important metric is permeability, measuring the ability of porous material to transmit fluids. High permeability often leads to high yielding. In a conventional well test, the well is controlled to produce at a constant flow rate, and the pressure is measured for a couple of hours (Figure 1). This pressure curve will be used to interpret the reservoir parameters, including the permeability k and initial pressure Pi. To interpret the pressure curve, a radial flow with infinite boundary model is utilized, whose mathematical solution may be simply written in the Equation 1. Key parameters in Equation 1 are: pwf , the measured bottom hole pressure, Pi, the initial pressure; q, the constant flow rate; k, the reservoir permeability. Traditionally the permeability may be interpreted by comparing the observed pressure curve with the calculated overlay template (Figure 2).