Construction of pore network of mudstones using image analysis

Transcription

1 Construction of pore network of mudstones using image analysis Introduction One of the powerful technique for reservoir characterization at microscale is special core analysis (SCAL). However, many of them are destructive (capillary pressure/rel. perm. experiments) and core plugs are rendered useless once experiment is over. Several researchers (Petrovic et. al. (1985), Tollner & Verma (1989), Brown et. al. (1993), Hsieh et al. (1998), Vogel & Brown (2002), Akin & Kovscek (2003), Prodanovic et. al. (2006), Kelly et. al. (2015)) have used microscopic imaging for construction of 3-D pore throat networks & shapes. This provides an alternative way for non-destructive reservoir characterization at microscale. Advancement in imaging and computational methods are projected to make this technology more lucrative in future. The objective of this project is construction of 3-Dimensional pore network model from slices of 2-D images using focused ion beam SEM (FIB-SEM) specifically for mudstones (shales). It is observed that mudstones/shales are characterized by directional mechanical, elastic and transport properties (Sondergeld et al. (2010)). However, variation in clay content and presence of organic matter (kerogen with low density, anisotropic, oil-wet, adsorbed gas) changes the character of shales (Butcher&Lemmens (2011)). One of the nondestructive method for identifying unconnected porosity and characterizing amount of total organic carbon (TOC) at nanometer scale is using optical imaging techniques. FIB-SEM provides the best resolution for these characterization (Suhrer et al., Geocanada (2010)). Background Pore scale characterization and transport mechanism of flow between fracture and matrix is still an emerging topic in research (Kelly et al. (2015)). Further these image based models can be utilized to investigate representative elementary volume (REV) for shales (Kelly et al. (2015)). In general, it can be concluded that thermally mature shales have more TOC and hence more organic porosity (organic carbon). However, Milliken et al. (2013) and Loucks et al. (2009) have found negative correlation between TOC and organic porosity. Image analysis can be used for precise characterization of these samples. The quality of pore network construction from image processing depends on resolution and sampling conditions. Common technologies available for imaging are Micro/Nano Ct-scan and FIB-SEM. Figure-1 shows resolution of images from different techniques. Figure 1 : 3-D reconstructed images from Micro, Nano-CT and FIB-SEM. FIB-SEM provides the best resolution for proper characterization. (Source: Suhrer et al., Geocanada (2010)

2 Method Obtain stacks of image slices in 2-D space. Convert image slices to numerical values based on intensity of image. Apply binary transform based on median of the dataset. Use GAM to obtain variogram of the processed image. Also, obtain anisotropy direction using VARMAP. Select the representative variogram. Use VMODEL to model the variogram with experimental one. Use KT3D to populate properties in 3D using representative variogram

3 Implementation & Results Figure 2 : Histograms of raw images with threshold line in red used for Binary segmentation. Left pictures shows images obtained after binary transformation. Resolution of Image-1 is 3.3 Um, Image-2 is 2.9 Um, Image-3 is 4.8 Um. Pixel size for all images is 149X149 in order to fit GSLIB program.

4 Figure 3 : Processed Images and corresponding Variogram maps used for Generation of 3-Dimensional porosity network Figure 4 : Left figure shows Variogram for Image-1 in different directions. The variogram seems non-stationary in 0 & 135 degrees while range in 90 & 45 degrees is 25 Um approximately. Right figure shows variogram for Image-2 which is nonstationary only in 45 Degrees. All variogram are generated in GSLIB and plotted in Matlab.

5 Figure 5 : Left picture shows variogram for Image-3 seems stationary in all directions. With similar ranges as for Image-1 and Image-2. Right figure shows GSLIB output of representative experimental variogram in red and VMODEL variogram used for kriging in black. Kriged Output Image-1 Kriged Output Image-2 Figure 6 : 21*21 map of kriged output from KT3D program in GSLIB.

6 Discussion Using semi-variogram model in different directions. Small scale and large scale correlation length can be deduced, the stationarity of the sample provides estimation of representative elementary volume (REV). Using geostatistics on core images. The spatial structure and anisotropy of porous media can be investigated. Nested model of anisotropy may give a better fit to given experimental variogram as shales are considered highly heterogenous. References Akin, S., and A. R. Kovscek. "Computed tomography in petroleum engineering research." Geological Society, London, Special Publications 215, no. 1 (2003): Curtis, Mark Erman, Raymond Joseph Ambrose, and Carl H. Sondergeld. "Structural characterization of gas shales on the micro-and nano-scales." In Canadian unconventional resources and international petroleum conference. Society of Petroleum Engineers, HSIEH, H.T., BROWN, G.O. & STONE, M.L. 1998a.Quantification of porous media using computerized tomography and a statistical segregation threshold. Transactions of the American Society of Agricultural Engineers, 41, " Hurley, Neil Francis, Tuanfeng Zhang, Guangping Xu, Lili Xu, and Mirna Slim. "Method to quantify discrete pore shapes, volumes, and surface areas using confocal profilometry." U.S. Patent 8,311,788, issued November 13, Kelly, Shaina, Hesham El-Sobky, Carlos Torres-Verdín, and Matthew T. Balhoff. "Assessing the utility of FIB-SEM images for shale digital rock physics." Advances in Water Resources 95 (2016): Loucks, Robert G., Robert M. Reed, Stephen C. Ruppel, and Daniel M. Jarvie. "Morphology, genesis, and distribution of nanometer-scale pores in siliceous mudstones of the Mississippian Barnett Shale." Journal of sedimentary research 79, no. 12 (2009): Milliken, Kitty L., Mark Rudnicki, David N. Awwiller, and Tongwei Zhang. "Organic matter hosted pore system, Marcellus formation (Devonian), Pennsylvania." AAPG bulletin 97, no. 2 (2013): Petrovic, A. M., J. E. Siebert, and P. E. Rieke. "Soil bulk density analysis in three dimensions by computed tomographic scanning." Soil Science Society of America Journal 46, no. 3 (1982): Phogat, V. K., L. A. G. Aylmore, and R. D. Schuller. "Simultaneous measurement of the spatial distribution of soil water content and bulk density." Soil Science Society of America Journal 55, no. 4 (1991): TOLLNER, E.W. & VERMA, B.P X-ray CT of quantifying water content at points within a soil body. Transactions of the American Society of Agricultural Engineers, 32, Vogel, J. R., and G. O. Brown. "Geostatistics and the representative elementary volume of gamma ray tomography attenuation in rock cores." Geological Society, London, Special Publications 215, no. 1

7 (2003): Appendix Program for generating kriging output file: clc;clear all; close all ; im1 = im2double(rgb2gray(imread('a3_4.8um.jpg'))); im1=im1(1:149,1:149); thres=median(median(im1)); im1_new=double((im2bw(im1,thres))); varinput=im1;varinput=varinput(:); count=1;row=1;start=1; valrow=1; while valrow<=size(varinput,1) for i=1:149 for j=1:149 varinput(valrow,2)=i; varinput(valrow,3)=j; valrow=valrow+1; end end end nx=149;ny=50; nce=nx*ny; krigtry=[varinput(1:nce,1),varinput(1:nce,2),varinput(1:nce,3)]; %fileid=fopen('krigtry.dat','w');fprintf(fileid,'%f\n',krigtry); xlswrite('krigtry2.xls',krigtry); ims=im1_new(1:nx,1:ny); %ims=reshape(varinput(1:nce,1),nx, ny); Program for generating histogram and image plots: clc; clear all;close all; im1 = im2double(rgb2gray(imread('a1_3.3um_150.jpg'))); im1=im1(1:149,1:149);thres1=median(median(im1)); im1_new=double((im2bw(im1,thres1))); im2 = im2double(rgb2gray(imread('a2_2.9um.jpg'))); im2=im2(1:149,1:149);thres2=median(median(im2)); im2_new=double((im2bw(im2,thres2))); im3 = im2double(rgb2gray(imread('a3_4.8um.jpg'))); im3=im3(1:149,1:149);thres3=median(median(im3)); im3_new=double((im2bw(im3,thres3))); figure subplot(3,2,1);imshow(im1_new,[]);title('image 1'); subplot(3,2,3);imshow(im2_new,[]);title('image 2'); subplot(3,2,5);imshow(im3_new,[]);title('image 3'); subplot(3,2,2);histogram(im1,50);hold on plot([thres1,thres1],ylim,'r--','linewidth',2) hold off subplot(3,2,4);histogram(im2,50);hold on plot([thres2,thres2],ylim,'r--','linewidth',2) hold off subplot(3,2,6);histogram(im3,50);hold on plot([thres3,thres3],ylim,'r--','linewidth',2) hold off program for generating variogram plots from GSLIB output file: clc; clear all;close all; im1 = im2double(rgb2gray(imread('a1_3.3um_150.jpg'))); subplot(2,2,1);imshow(im1,[]);title('original Image 1'); max1= max(max(im1)); im1_new = ((abs(im1-max1))./max1)*1000; subplot(2,2,2);histogram(im1_new,50);title('original Histogram'); thres=median(median(im1_new)); %Overlay the thres hold on plot([thres,thres],ylim,'r--','linewidth',2) hold off %im_ind=im1_new>20&im1_new<980;im1_new(~im_ind)=-1;immod=im1_new(im_ind); subplot(2,2,3);imshow(im1_new,[]);title('rescaled Image 1'); %subplot(2,2,2);histogram(immod,50);title('removing Outliers');xlim([20 980]); binindex=im1_new>thres;subplot(2,2,4);imshow(binindex,[]);title('modified Image 1'); varinput=im1_new;varinput=varinput(:); xlswrite('varinput.xls',varinput,'sheet1','a4');

8 %fileid=fopen('test.dat','w');fprintf(fileid,'%4.2f\r\n',im1_new); % reading *.out file from gslib with lag distance of 100 [h y1]=importfile('v1.out',2,100);[h2 y2]=importfile('v1.out',103,202);[h3 y3]=importfile('v1.out',204,303);[h4 y4]=importfile('v1.out',305,404) figure subplot(2,2,1);plot(h,y1,'o');gridxy(get(gca,'xtick'),get(gca,'ytick'),'color',[.4.4.4],'linewidth',1,'linestyle',':'); title('0 Degree'); subplot(2,2,2);plot(h2,y2,'o');gridxy(get(gca,'xtick'),get(gca,'ytick'),'color',[.4.4.4],'linewidth',1,'linestyle',':'); title('90 Degree'); subplot(2,2,3);plot(h3,y3,'o');gridxy(get(gca,'xtick'),get(gca,'ytick'),'color',[.4.4.4],'linewidth',1,'linestyle',':'); title('45 Degree'); subplot(2,2,4);plot(h4,y4,'o');gridxy(get(gca,'xtick'),get(gca,'ytick'),'color',[.4.4.4],'linewidth',1,'linestyle',':'); title('135 Degree'); GSLIB gam file: Parameters for GAM ****************** START OF PARAMETERS: varinput3.dat -file with data number of variables, column numbers -1.0e21 1.0e21 - trimming limits v3.out -file for variogram output 1 -grid or realization number nx, xmn, xsiz ny, ymn, ysiz nz, zmn, zsiz number of directions, number of lags ixd(1),iyd(1),izd(1) standardize sill? (0=no, 1=yes) 1 -number of variograms tail variable, head variable, variogram type GSLIB KT3D file: Parameters for KT3D ******************* START OF PARAMETERS: krigtry3.dat -file with data columns for X, Y, Z, var, sec var -1.0e21 1.0e21 - trimming limits 0 -option: 0=grid, 1=cross, 2=jackknife xvk.dat -file with jackknife data columns for X,Y,Z,vr and sec var 3 -debugging level: 0,1,2,3 kt3d.dbg -file for debugging output kt3d.out -file for kriged output nx,xmn,xsiz ny,ymn,ysiz nz,zmn,zsiz x,y and z block discretization 1 8 -min, max data for kriging 0 -max per octant (0-> not used) maximum search radii angles for search ellipsoid =SK,1=OK,2=non-st SK,3=exdrift drift: x,y,z,xx,yy,zz,xy,xz,zy 0-0, variable; 1, estimate trend extdrift.dat -gridded file with drift/mean 4 - column number in gridded file nst, nugget effect it,cc,ang1,ang2,ang a_hmax, a_hmin, a_vert it,cc,ang1,ang2,ang a_hmax, a_hmin, a_vert