Development of constraint handling techniques for well placement optimization in Petroleum Field Development - Problem formulation and implementation for FieldOpt software including well index calculation for deviated wells

Abstract

Well placement optimization is an important part of Petroleum Field Development. However, in order to improve the optimization procedures, it can be important to incorporate considerations like knowledge about the geology of the reservoir or about existing or planned well paths. This leads to additional constraints that have to be satisfied during the optimization. In this thesis we concentrate in particular on constraints on the well lengths and the distance between the wells.

We suggest an alternating projections method to deal with both constraints at the same time, and develop an efficient numerical method for the solution. Although we cannot prove that the method is convergent, numerical results of our implementation indicate that the approach works as intended.

An additional important contribution from this work is the implementation of a well index calculator.

In reservoir simulation, the well index relates the flow rate and pressure of the wellbore to the pressure solution of the subsurface fluid flow system, and is therefore an essential part in computing resulting production volumes.

We also implement an algorithm that, given a slanted well and the physical state of a reservoir, calculates the well indices for the well blocks that are intersected by the well.

In particular the well index calculation for deviated wells is a nontrivial task that is important for well placement optimization research.

This task is already handled by some industry reservoir simulators, but the implementation is hidden from the end-user.

All of the implementations are meant to be an addition to FieldOpt, a petroleum field development optimization framework that is currently under development by the Petroleum Cybernetics Group at NTNU.