Omre, H. and H. Tjelmeland (1996). Petroleum geostatistics. Invited lecture at the Fifth International Geostatistics Congress, Wollongong, Australia, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....not meaningful. The real object of interest is moreover complicated functionals of the reservoir like e.g. production profiles. Bayesian modelling, combined with MCMC computation of posterior distributions, seems a very appropriate approach to assessment of uncertainty in such situations, see e.g. Omre Tjelmeland (1996) and Syversveen Omre (1996) 1.2 Markov connected component fields 1.2.1 Definitions In practice images are represented digitally, i.e. as a matrix of pixel values indexed by a set I. An image parametrization commonly applied for image segmentation is based directly on the digital ....

....is moreover complicated functionals of the reservoir like e.g. production profiles. Bayesian modelling, combined with MCMC computation of posterior distributions, seems a very appropriate approach to assessment of uncertainty in such situations, see e.g. Omre Tjelmeland (1996) and Syversveen Omre (1996). 1.2 Markov connected component fields 1.2.1 Definitions In practice images are represented digitally, i.e. as a matrix of pixel values indexed by a set I. An image parametrization commonly applied for image segmentation is based directly on the digital representation, so that the unobserved ....

Omre, H. & Tjelmeland, H. (1996), `Petroleum geostatistics', Invited lecture at the Fifth Geostatistical Congress, Wollongong, Sept. 22.-27.

....biased, see Jones et al. 1992) By making models which are reasonable in a geological sense, it is more likely to extrapolate the results to a new production strategy. In the present approach small scale heterogeneity and large scale properties are modeled in a Bayesian framework, see also Omre and Tjelmeland (1996). In Bayesian models the model parameters are regarded as stochas 1 Preprint Statistics NO. 9 1996, Norwegian University of Science and Technology. 1 To appear in proceedings from the 5th. International Geostatistical Congress, Wollongong 1996. 2 BJRN K ARE HEGSTAD AND HENNING OMRE tic. ....

....the impact of the low dimensional model parameters, on production forecast is considerable. This may be used to reduce the dimensionality of the history matching problem in order to obtain predictions more efficiently. Exploration of normal models have also been performed, see Hegstad and Omre (1996), but it is not documented here. It shows a behavior more linear in the parameter , hence making the uncertainty assessment easier, and forecast uncertainties smaller. 5. Closing remarks A Bayesian formalism for stochastic reservoir modeling has been presented. It gives a formal way to assess ....

Omre, H. and H. Tjelmeland (1996). Petroleum geostatistics. Invited lecture at the Fifth International Geostatistics Congress, Wollongong, Australia, 1996.

....reservoir knowledge include geologic understanding, physically based models for uid ow and insight into the data acquisition procedures. The reservoir speci c observations include well observations, seismic amplitude data and production history collected from the reservoir under study. In Omre and Tjelmeland (1997) a Bayesian approach to integrated reservoir evaluation is presented. Stochastic reservoir modeling based on geologic knowledge and well observations only, has taken place in two decades. Inclusion of seismic amplitude data has been an active eld of research the last years, see Bortoli et al. ....

....last years, see Bortoli et al. 1993) and Eide et al. 1999) In order to represent the uncertainty, seismic inversion must be a part of the model. History matching of production data has only recently been phrased in a stochastic setting, see Oliver (1994) Wen 1 et al. 1997) and Hegstad and Omre (1997). This requires the use of a uid ow simulator which normally needs considerable computer resources to run. The current paper integrates all these sources of information in a framework de ned along the lines of Omre and Tjelmeland (1997) A graphical model is introduced to communicate the model ....

[Article contains additional citation context not shown here]

Omre, H. and H. Tjelmeland (1997). Petroleum geostatistics. In E. Y. Baa and N. A.

....analogues and (ii) reservoir specific observations such as well logs, seismic data and production history. Stochastic reservoir characterization in a Bayesian framework has proven useful for the integration of reservoir information and for assessing uncertainties, see Lia et al. 1996) and Omre and Tjelmeland (1996). Integration of seismic data is particularly important since it provides good spatial coverage over the reservoir. Doyen (1988) used a standard cokriging model to integrate well observations and seismic data. Later, several authors extended this model to include an indicator formalism, see ....

....= const Theta f(ojr)f(r) const Theta Z f(ojr; f( d Z f(rj )f( d : 1) For a few particular choices of pdf s this can be calculated analytically, for some choices of pdf s it can be sampled using current computer technology, while in the general case it is completely intractable. Omre and Tjelmeland (1996) give a more thorough discussion. Whenever sampling based approaches are used, it is more convenient to consider f(r; jo) const Theta f(ojr; f( f(rj )f( 2) By sampling the triplet (R; Theta; Psi) samples of R are implicitly obtained. 3. Operational Model There are two aspects ....

Omre, H. and Tjelmeland, H. (1996), Petroleum geostatistics, invited lecture at the Fifth International Geostatistics Congress, Wollongong, Australia, 22.--27. September 1996.

....are fairly exact measurements in wells drilled through the reservoir, other observations are indirectly obtained through non linear transfer functions as seismic data. This requires physically based assessment of the likelihood models. A framework for this in reservoir evaluation is presented in Omre Tjelmeland (1997). In Tarantola (1987) the seismic inverse problem is cast in a Bayesian setting, but no spatial model is used. Bortoli, Alabert, Haas Journel (1993) and Haas Dubrule (1994) introduce a spatial model, but sample from it using an ad hoc procedure. Abrahamsen et al. 1996) utilize a rejection ....

Omre, H. & Tjelmeland, H. (1997), Petroleum geostatistics, in E. Baafi & N. Schofield, eds, `Geostatistics Wollongong '96', Kluwer Academic Publishers, Dordrecht, pp. 41--52.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC