(342q) Oilfield Production Optimisation Via Mixed-Integer Nonlinear Programming (MINLP)

The integrated management of oil and gas operations is a challenging task involving short to long-term decisions. Some of these decisions include valve configurations, choke openings and complex well-manifold routings [1, 2]. Hence, novel advancements in mathematics, algorithm development [3, 4] and related scientific fields have found tremendous applications in the modelling, simulation and optimisation of complex phenomena for improved oilfield recovery. These methods hold strong potential for enhancing field profitability without additional facility/equipment installation costs or additional operations. In this study, it is demonstrated that a methodical application of simulation-based optimisation methods guarantees process enhancement by only finding the optimal well to manifold and pipeline to separator connections. This is achieved by developing explicit surrogate models, which are compatible with the adopted optimisation algorithms; thus, resulting in a Mixed-Integer Nonlinear Program (MINLP).

The novelty of this study lies in the combination of naturally flowing, artificially lifted (gas lift and Electrical Submersible Pumps, ESP) wells, which create complex pressure responses at the pipeline level; these are accounted for via routing constraints and embedded in a complex economic objective function [5]. Key attention is also paid to the bottom hole pressure of the well, which in turn is affected by the wellhead pressure at a certain production rate. This is done to avoid sand production, which could be detrimental to the overall system performance. Furthermore, the adaptability and flexibility of the proposed optimisation formulation to varying scenarios and practical operational difficulties (including the common water-coning problem) are demonstrated herein, with the complexities of varying wellbore geometries with different multiphase flow properties incorporated [5].

In order to enhance solution quality and the computational efficiency of the optimisation formulation, the developed MINLP models are reformulated to a Mixed-Integer Linear Program (MILP) via piecewise linear approximations [2, 6]. Typical sources of nonlinearity in a production system are the production/injection wells’ pressure-rate response, the pipeline and valve pressure drop and multiphase flow rate relationships [8]. These complex relationships are usually not explicitly known, and they are dependent on several operational parameters estimated via high fidelity simulators [7, 8]. In this work, the benefits of this method are applied to synthetic but realistic case studies of different sizes (different number of wells, valves, pipelines, and separators). We compare the obtained solutions with the solution of the original MINLP and evaluate the impact of the number of linearization breakpoints, and problem size, on the solution time, accuracy, robustness, modelling effort and ease of automation. It is ensured that a high accuracy level of the nonlinear models and piecewise approximations in comparison to the simulation data is maintained while carrying out optimisation computations.

The analysis presented herein enables quality assessment of the relative performance of the respective formulations and their impact on the overall oil production. Furthermore, the computational performance comparison is also evaluated in terms of the optimisation solvers applied (BONMIN (v.1.8.6),
CBC (v.2.9.8), SCIP (v.3.2.1) and CPLEX (v.12.8.0.0)). Our proposed formulation is observed to perform better than a recently developed industrial tool for production optimisation – PIPESIM Network Optimiser^®. The proposed optimisation formulation demonstrates good utilisation of separator capacity for routing produced fluids. It is thus useful for production network design purposes when decisions relating to the size of separation facilities are to be made. The adaptability of the proposed formulation to changing operational conditions is also strongly demonstrated; thus validating its applicability for real-time decision support.

LITERATURE REFERENCES

1. Epelle, E.I. and Gerogiorgis, D.I., Optimal Rate Allocation for Production and Injection Wells in an Oil and Gas Field for Enhanced Profitability. AIChE J., 65(6). (DOI: 10.1002/aic.16592).
2. Gunnerud, V. and Foss, B., 2010. Oil production optimisation—A piecewise linear model, solved with two decomposition strategies. Chem. Eng., 34(11), 1803-1812.
3. Wächter, A. and Biegler, L.T., 2006. On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming. Prog., 106(1), 25-57.
4. Kronqvist, J., Bernal, D.E., Lundell, A. and Grossmann, I.E., 2018. A review and comparison of solvers for convex MINLP. Optimisation Eng., 1-59.
5. Epelle, E.I. and Gerogiorgis, D.I., 2019. Mixed-Integer Nonlinear Programming (MINLP) for Production Optimisation of Naturally Flowing and Artificial Lift Wells with Routing Constraints. Eng. Res. Des., 152, 134-148.
6. Gerogiorgis, D.I., Georgiadis, M., Bowen, G., Pantelides, C.C. and Pistikopoulos, E.N., 2006. Dynamic oil and gas production optimisation via explicit reservoir simulation. Aided Chem. Eng.21, 179-184.
7. Codas, A., Campos, S., Camponogara, E., Gunnerud, V. and Sunjerga, S., 2012. Integrated production optimisation of oil fields with pressure and routing constraints: The Urucu field. Chem. Eng.,46, 178-189.
8. Silva, T.L. and Camponogara, E., 2014. A computational analysis of multidimensional piecewise-linear models with applications to oil production optimisation. J. Operational Res., 232(3), 630-642.