Multiple-point Geostatistics Model And Applications In Sedimentary Facies Modeling

This paper mainly analyzes the influence factors of multi-point geostatistics, andthen gets the optimized result through preference pattern, and finally applies theMarkov chain model to evaluate spatial data structure quantitatively. The result ofsimulation is hard to control, because there are more influence factors in the processof geostatistics modeling. Pattern is the crucial part of simulation result. So it isneeded to resolve that how to choose the pattern coinciding with actual geologicalcondition in the multi-point simulation. This paper analyzes three factors whichincludes the size of pattern, the model of pattern and multilevel gridding, and thenstudies the influence of factors to simulation result. The size of pattern is decided withthe acreage of work area. When it is undersize, the randomness of simulation is severeand it is hard to describe the geological situation of work area. On the contrary, thedescription of detail is relatively poor. The model of pattern reflects spatial variability,and changing parameters can influencing continuity and variability in differentdirections. The size of gridding is related to the size of work area. Results ofsimulation well be disperse because the gridding is too little. But oversize griddingwill slow down the computation speed and will not obviously improve the results. Weselect the optimized pattern group through comparison and analysis of simulationresults about different pattern group.Markov chain model is used to solve the problem of evaluating result in themulti-point geostatistical simulation. Papers uses Markov chain model to evaluate theresults of multi-point simulation, which in details will use the transition probabilitycurves and transition probability matrix to represent the spatial structure of the image.The transition probability curves with itself represent the spatial continuity of the stateof one variables in one direction. The mutual transition probability curves representthe spatial structure and correlation of transferring with each other of different statesin one direction. The Transition probability matrix comprehensively describes thespatial continuity and variability in each state and in each direction. The paper canverify its applicability through analyzing the specific training images and then it can quantitatively represent the spatial structure of the image in details.The paper does the instance application of analysis through selecting work areawith similar geology and sedimentary background. It turns out that differentdepositional model has different spatial structure using Markov chain model applyingto the work area with abundant well information. In the process of modeling to theexperimental work area, the paper selects the sedimentary model corresponding to thegeological sedimentary type as the training images and then uses it in the multi-pointgeostatistical simulation. Then it can evaluate the simulation results quantitativelywith Markov chain model. The researcher can select the simulation results throughcomparing the actual geological conditions and it can help the researcher do theanalysis of geological sedimentary facies.