Objective
Mathematical models are commonly used to predict the future behaviour of oil and gas fields. A key element in the process is calibration of the mathematical model, known as a reservoir simulator, so that it honours historical data. This is known as history matching.
History matching is a highly interactive and time consuming procedure. At each stage, small changes are made to the model which is then re-run. The results are inspected to see if the match between the predictions of the model and the observed history has been improved.
The current project is intended to develop an approach to history matching based on the method of simulation coefficients. This will allow the level of improvement in the calibration which would normally be obtained from multiple simulation runs to be derived from a single run.L%
The project is now somewhat more than half way through. Progress is on schedule for release of the first commercial product by the end of 1997.
Results of testing of the program to date by GeoQuest and by the project sponsors have been encouraging. A wide range of parameters with respect to which sensitivities can be calculated have been implemented and work on this area is continuing. The most appropriate way of handling information known prior to the start of the project is currently being evaluated.
The techniques being developed during this project use the Method of Sensitivity Coefficients. It involves calculation of the sensitivity coefficients associated with a set of prescribed parameters during a normal run of the reservoir simulator. As a consequence, rather than submitting one additional run for each sensitivity required, it is possible to obtain the instantaneous sensitivities for all parameters at once. Not only is this more convenient than submitting lots of separate runs of the simulator, it is also much faster. Typically, the calculation of one parameter's gradients over the entire simulation time period increases the CPU time by approximately 10 per cent. With one extra run the increase would be 100 per cent. The method also includes the use of a non-linear regression algorithm which uses the sensitivities to direct the change of model parameters of interest semi-automatically.
Application of this technique offers the prospect of streamlining the history matching of reservoir simulation models dramatically. Any simulation model used to predict the future behaviour of an existing field already on production has to be matched to the existing production data. In the case of complex models with years of history, this can take six months. The History Match module will allow the same level of match to be achieved in much less time. We conservatively estimate that the time taken to history match a moderately complex model with a number of years of history is likely to be reduced by 50 per cent. This represents a major saving in the time of the professionals involved in the exercise.
Fields of science (EuroSciVoc)
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques.
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques.
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringcomputer hardwarecomputer processors
- humanitieshistory and archaeologyhistory
- engineering and technologyenvironmental engineeringenergy and fuelsfossil energynatural gas
- natural sciencesmathematicsapplied mathematicsmathematical model
You need to log in or register to use this function
Topic(s)
Call for proposal
Data not availableFunding Scheme
DEM - Demonstration contractsCoordinator
OX14 1DZ Abingdon
United Kingdom