Optimizing automatic history matching for field application using genetic algorithm and particle swarm optimization

Lee Sangyhun, Karl Dunbar Stephen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

13 Citations (Scopus)


History matching is a commonly used process that integrates a static model with dynamic data to obtain an accurate tool for predicting reservoir performance. Automatic History Matching (AHM) assisted by computers helps engineers control a large number of parameters efficiently to minimize a misfit compared to the traditional trial and error approach. However, effort is necessary to optimize the process of AHM to reduce simulation running time. Stochastic algorithms such as the Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) can be combined with the least squares single objective function to explore the parameter space. Parameterization and realization methods are employed to enhance the effectiveness of history matching. In this field study history matching was conducted while identifying the feasibility of each method. A number of realisations of various properties such as porosity, net to gross, relative permeability, horizontal and vertical permeability, and aquifer size were selected for controlling throughout the AHM. The optimized process is discussed to provide guidelines. We found that stochastic algorithms are efficient at handling a large number of control parameters in heterogeneous reservoirs to improve the match. It was observed that GA was able to update a large number of control parameters. However, GA is a costly algorithm that requires more computing time. The realization and parameterization methods improved the accuracy of a full-field application although errors remained in predictions of the bottomhole pressure. While simple relatively homogenous fields may be manually history matched, more complex fields require a good parameterization scheme. In addition we observed that selecting too many parameters makes the problem difficult to solve while selecting too few leads to false convergence. In this study, we observed that the PSO had a shorter convergence time compared to the GA, Using the GA as a follow up method helped find better results. Overall this study can be used as a guideline in selecting an appropriate history matching method.
Copyright 2018, Offshore Technology Conference
Original languageEnglish
Title of host publicationOffshore Technology Conference Asia, 20-23 March, Kuala Lumpur, Malaysia
PublisherOffshore Technology Conference
Number of pages28
ISBN (Print)9781510862159
Publication statusPublished - 2018
EventOffshore Technology Conference Asia 2018 - Kuala Lumpur, Malaysia
Duration: 20 Mar 201823 Mar 2018


ConferenceOffshore Technology Conference Asia 2018
Abbreviated titleOTCA 2018
CityKuala Lumpur


  • Automatic history matching
  • Genetic algorithm
  • Parameterization
  • Particle swarm optimization
  • Realization

ASJC Scopus subject areas

  • Ocean Engineering
  • Mechanical Engineering
  • Safety, Risk, Reliability and Quality
  • Energy Engineering and Power Technology


Dive into the research topics of 'Optimizing automatic history matching for field application using genetic algorithm and particle swarm optimization'. Together they form a unique fingerprint.

Cite this