| dc.description.abstract | The oil and gas industry has been subject to a series of challenges during the last couple of decades, including increased water depths, step-out distances and harsher environmental and operational conditions. This has caused a need for new innovative ideas and solutions, which has resulted in increased attention to the development of subsea processing. The main objective of this thesis has been to develop an mixed integer nonlinear programming (MINLP)-model for optimizing the planning and development of offshore oil field infrastructure with an integrated subsea separation system for each field.
First, a MINLP-model was developed for a single field connected to a floating production storage and offloading (FPSO) unit with subsea processing. The objective of the optimization was to identify the set of subsea processing units that maximized the net present value over the given time horizon. A superstructure including all potentially useful units and interconnections was generated and used as a basis for the optimization model. Binary variables were assigned to each of the potential units to represent the installation of the unit in the optimal solution, while continuous variables were used for the mass flows, pressures, cost and sizing variables. The proposed constraints of the model included mass balances, logical conditions related to the binary variables, as well as sizing and cost estimation equations.
The single-field model was extended to include multiple fields and FPSOs, as well as time scheduling of the installations and drilling over the time horizon. The objectives of this multi-field model included determining the FPSOs to be installed, the fields to be developed, the number of wells to drilled in each field in each time step, the production rates of the fields, as well as the installation of subsea equipment in each of the fields developed. The superstructure that was developed for the single-field model, was used to optimize the configuration of the subsea system of each of the developed fields in the multi-field model.
Both the single-field and the multi-field model were implemented in the high-level mathematical optimization system GAMS (General Algebraic Modeling System). The models were then solved for different cases on the NEOS-server for numerical optimization problems with two different MINLP-solvers (DICOPT and BARON). The results showed that the single-field model was solved to global optimality by BARON within reasonable computational time limits. The multi-field model was only solved to global optimality by BARON when the number of time steps were low. For higher number of time steps, the optimality gap was too large to conclude that the global optimal solution was obtained. | en |
