Linear Regression with Regularization for Well Flow Rate Back Allocation

Well flow rate back allocation is the estimation of how much individual wells contribute to the total flows out of a process. It is essential for the oil and gas industry. The allocated rates are used for production optimization, reservoir management, placement of new wells and distribution of revenue between stakeholders. To obtain good allocation is often difficult and production engineers spend much time on it. Well tests may lead to production losses, and needed instrumentation can be expensive. This article describes a new method for well flow rate back allocation. Available well sensors are used as input to a linear regression with regularization model. The model is trained to match the total flows out of the process and is constructed such that it is possible to extract the individual well flow rates. Model results are presented both for simulated and real field data. It is shown that for some cases results can be excellent without any well testing, whereas for others one needs far fewer well tests than with traditional methods. Other advantages are that it is cheap to implement and can succeed where traditional methods fail.