Derivative-free methods applied to daily production optimization of gas-lifted oil fields

• Derivative-free algorithm uses augmented Lagrangian to treat nonlinear constraints.
• Derivative-free trust-region optimizes the production of a simulated oil field.
• The proposed method has comparable results to Mixed-Integer Linear Programming.

In the oil industry, computer simulators are routinely applied on day to day operations and in what-if analyses. With the increasing complexity of operations, engineers are relying on simulation to synthesize models for mathematical optimization or else applying derivative-free methods for simulation-based optimization, which requires only the sampling of the objective function. This work adapts an algorithm based on augmented Lagrangian and derivative-free trust-region algorithms to handle hard constraints found in production optimization. The effectiveness of the proposed method is assessed in an oil production network and compared with mathematical optimization based on piecewise-linear models.