Abstract:

The scheme for well test is frequently made empirically, and there may be large differences between the design goal of testing and the actual outcome, making data interpretation difficult and unreliable. The problems are mainly derived from two aspects: first is the difficulty of accurately estimating the parameters of the well, fluids and formations, and second is the lack of method for systematic evaluation of the test design. The design can be evaluated through well test simulator which generates diagnose curve for specific suit of parameters. Monte Carlo stochastic simulation is applied to evaluate risk derived from parameter uncertainty in well test design, in which artificial intelligence method is used to automatically recognize the curve and evaluate successful rate, and then to guide scheme adjustment to reduce risk.

Key words: well testing, design, simulation, artificial intelligence