Use this URL to cite or link to this record in EThOS:
Title: Estimation of distribution algorithms for reservoir history-matching optimisation
Author: Petrovska, Iryna
ISNI:       0000 0004 2678 240X
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2009
Availability of Full Text:
Access from EThOS:
Access from Institution:
Reservoir modelling is widely used in the oil and gas industry to quantify the riskassociated with alternative production scenarios. However, reservoir models themselvesstill contain a high level of uncertainty because of the typically very limited,sparse and multiscale field knowledge available. History-matching (HM) reducesthis uncertainty by constraining the reservoir model to the available dynamic field data. History-matching is an example of a typical non-linear inverse problem whichyields the existence of not one but multiple solutions, which all satisfy available dataconstraints. In inverse problem theory Monte Carlo methods are regarded as themost accurate methods for generating a family of problem solutions and capturingposterior distributions of model parameters by exhaustive exploration of parameterspace. However these methods are very rarely applicable to HM problems becausethey are too time and cost consuming. While other stochastic inversion techniques have successfully overcome the runtimeissue Monte Carlo methods have, none of them has provided a deliberate estimationof the posterior probabilities one would expect from Monte Carlo methods. This thesis introduces an innovative application of a member of a class of Estimationof Distribution Algorithms - a histogram-based Population-Based IncrementalLearning algorithm, to the problem of reservoir history-matching optimisation. It is shown that while avoiding an exhaustive exploration of parameter space the proposedalgorithm is capable of producing the approximations of the marginal posteriordistributions of model parameters which can be interpreted as their uncertaintyestimates. We also suggest a new extension of histogram-based PBIL for pair-wise conditionalprobabilities sampling. The developed extended version of the histogrambasedPBIL is the first attempt to explicitly capture possible dependencies betweenreservoir model parameters and use them to perform conditional sampling of thesolution space. None of the currently used algorithms and techniques for reservoirhistory-matching optimisation explicitly utilizes this dependency information.
Supervisor: Carter, Jonathan Sponsor: Schlumberger AbTC
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral