The Evaluation Method Of Pore Structure And Physical Property In Carbonate Rock Reservoir With High Porosity And Low Permeability

The A oilfield,located in the western collapse of the Sinjar uplift in Syria,is one of the major investment fields for Chinese enterprises.As a carbonate oil field,A oilfield is characterized by high porosity and low permeability of its reservoir matrix and well-developed fractures.However,limited studies on this type of reservoir can be referred to due to the fact that most fractured carbonate reservoirs in China have low matrix porosity and permeability.Suffering from the high matrix heterogeneity,previous studies of A oilfield still have some remaining issues related to insufficient information from rock physics data,inadequate quantitative characterization of reservoir pore structure and physical property(porosity and permeability)assessment of reservoir matrix.Those remaining issues restrict the rational and efficient development of carbonates reservoirs with high porosity and low permeability.In this paper,we studied the high porosity and low permeability carbonate rocks in A oilfield of Syria.Based on fractal theory,a research on pore structure and physical characterization of high porosity and low permeability carbonate rocks was carried out using the data collected from rock physics experiments,logging,oil testing and initial production.The main steps of the research are summarized as follows:? Since the two-dimension pore structure information can be represented quantitatively and accurately using image processing technology,combining the information with fractal theory,the heterogeneity of two-dimension pore characteristics was characterized and the relationship between fractal dimension and the morphology,quantity and distribution of two-dimension pores was discussed.?The morphology and characteristic parameters of mercury injection curves were analyzed,and the pore-throat structure was classified;the subclass division of core pore-throat structure was further conducted based on fractal theory.? Multiple nuclear magnetic T2 experiments and T2 cutoff experiments were carried out,and the estimation method of T2 cutoff value was discussed;the influence of mineral composition on the development of movable pores was analyzed;A permeability model based on movable porosity information was proposed;with the two-dimension casting thin sections data,the formula to calculate fractal dimension of nuclear magnetic resonance was improved and the fractal dimensions of bound pores and movable pores were determined;the relationship between fractal dimension and mineral composition and pore structure was studied,and the fractal dimension was used to quantitatively characterize the heterogeneity of nuclear magnetic resonance pore structure.?Based on fractal theory,the fractal dimensions of logging curves were calculated using cores,electrical imaging and conventional logging data,and the fractal characteristics of logging curves were used to identify fracture development intervals.?With conventional well logging data,support vector machine(SVM)was used to identify reservoir fluid properties by multiple parameter optimization methods;the calculation of logging parameters such as total porosity,movable porosity and matrix permeability was realized by the regression calculation of SVM and the calculation results were verified by core experimental data.The main observations and conclusions of the research are as follows:(1)The study on two-dimension pore structure based on casting thin sections analysis:(a)the result with image digital processing method is different from that with the ocular estimation method,which is used to determine the areal porosity.The areal porosity from ocular estimation is generally smaller and there is a higher correlation between the areal porosity from image digital processing and that from core gas measurements.The aspect ratio decreases as the pore shape factor increases.The pore equivalent diameter increases as the specific surface decreases.The pores of carbonate rocks in the study area are well developed,and overall the pore morphology is relatively regular.However,the pore size is generally smaller,with a large number of intercrystalline micropores existed.(b)As the fractal dimension increases,the pore equivalent diameter decreases,the aspect ratio increases,and the pore structure becomes more complex.The value of fractal dimension can indicate the complexity level of pore structure.Dpa can better indicate the change of pore morphology and the box dimension can better indicate the change of pore quantity and pore distribution.(2)The study on pore-throat structure based on mercury injection experiments:(a)based on feature parameters and curves sharp,the pore-throat structures of the samples in the study area can be divided into four major types,which are ? pore throat,? pore throat,?pore throat and the non-reservoir rocks in the order from good to bad.(b)Combining the classification of pore-throat structure with the difference of number of characteristic segments of fractal curves,the subclass division of pore-throat structure is realized.Fractal dimension has complex relation with mineral composition and pore structure parameters.In general,pore-throat structure becomes worse and fractal dimension increases when clay content increases;how fractal dimension changes with mineral composition and pore structure parameters are different among different subclasses of pore-throat structure.(3)The study on pore structure based on nuclear magnetic resonance experiments:(a)the oil content is generally good when the core contains biological body cavity pores.However,some mudstones,without biological body cavity pores,are very poor in oil content.This kind of cores mainly contain bound pores,which are formed form intercrystalline micropores,and they have a few movable pores.Some cores,with low total clay mineral content and low proportion of montmorillonite in clay minerals,show the absence of nuclear magnetic signal at 3ms of the nuclear magnetic T2 spectrum.(b)Combining fractal theory with nuclear magnetic pore volume physical model,the T2 cutoff values are estimated and the estimated values have small error and good correlation with the measurement.Clay and feldspar have no obvious effect on the total porosity,while calcite,dolomite and quartz show double effects on it.The penneability calculation model based on movable porosity information has better applicability than SDR model,Coates model and the model derived from mercury injection experiments.Movable porosity is the key parameter.(c)Combining with the fractal characteristics of the casting thin sections,the formula to calculate the fractal dimension when T2 and r have linear or non-linear relation is improved.The fractal dimensions of bound pores and movable pores are calculated.The increase of clay content makes the pore structure of bound pores and movable pores more complicated,especially for the latter.Calcite mainly affects the pore structure of movable pores.The increase of calcite content makes the pore structure of movable pores simpler.Dolomite mainly affects the pore structure of the bound pores.The higher the calcite content,the lower the clay mineral content,the lower the dolomite content,and the better the pore structure of the reservoir.The fractal dimensions of movable pores are closely related to the total porosity and matrix permeability of cores.When the physical properties of cores become worse,fractal dimension increases and pore structure becomes worse,which proves that fractal dimension can characterize pore structure.(d)The reservoir matrix in the study area is divided into three categories:the high quality reservoirs having movable porosity larger than 15%and fractal dimension of movable pore smaller than 2.8;the medium quality reservoirs having movable porosity larger than 5%and fractal dimension of movable pore smaller than 2.9;the poor quality reservoirs having few movable porosity and fractal dimension of the movable pores close to 3.(4)The study on fracture identification based on curve fractal feature:the single fractal dimension of the three porosity curve and the deep lateral resistivity curve cannot be used to identify the fracture development section because of the double effects of fracture and lithology.By overlapping the fractal dimension of the deep lateral resistivity curve with the fractal dimension of the density curve,there is a good agreement between the position and the effective fracture development position of imaging logging.It is shown that the fractal dimension overlap method is able to identify the effective fracture development section.(5)The study on pattern recognition and regression calculation based on SVM:(a)The methods of grid search optimization,genetic algorithm optimization and particle swarm optimization have good adaptability in the identification of fluid for carbonate reservoirs with high porosity and low matrix permeability.In the training set of 65 layers,the coincidence rate of all three methods is above 76%.In the test set of 35 layers,the coincidence rate of SVM fluid recognition in grid search is 74.3%,which shows improvement compared to the fluid recognition coincidence rate of 65%using conventional methods.Support vector machine classification achieves the direct fluid recognition without discrimination of lithology and reservoir type.(b)GA-SVR calculation model has good learning effect and strong generalization ability.The test from GA-SVR calculation model shows that the mean squared error is 0.0075 and the squared correlation coefficient is 0.92.GA-SVR calculation model of movable porosity has stronger generalization ability than GA-SVR calculation model of total porosity.(c)The GA-SVR based matrix permeability model has good learning and generalization ability and there is no over-learning phenomenon.The squared correlation coefficient between the predicted value and the actual value in the test set reaches 0.62.By analyzing the relationship between movable porosity and matrix permeability,the movable porosity is calculated by GA-SVR calculation model,after which the matrix permeability is estimated.The matrix permeability calculated by logging has good correlation with the measured permeability where the squared correlation coefficient is 0.8.The calculation of matrix permeability is well realized.