2004, 66th EAGE Conference, Pars Quanttatve Dscrmnaton of Effectve Porosty Usng Dgtal Image Analyss - Implcatons for Porosty-Permeablty Transforms Gregor P. Eberl 1, Gregor T. Baechle 1, Ralf Weger 1, and Jose-Lus. Massaferro 2 1 Comparatve Sedmentology Laboratory, Unversty of Mam, 4600 Rckenbacker Cswy, Mam, FL 33149 2 Shell Internatonal E&P, Carbonate Development Team, Volmerlaan 8, 2280 AB Rjswjk, The Netherlands Key Fndngs Fve relevant dgtal mage analyss parameters for flud flow are obtaned from 2-D mage analyss of carbonate rocks: 2-D mage porosty, amount of pores, pore shape, total permeter per area and domnant pore sze range. Mcro-porosty leads to low permeable, hghly porous rocks. 2-D mage porosty represents the macro-porosty because the resoluton subtracts the small pores from the porosty. Compared to the total porosty, t mproves the predcton of permeablty by an order of magntude n the hgh porosty range. Pore shape factor γ reduces the uncertanty n permeablty predcton to 2 orders of magntude. Pore shape factor analyss s restrcted to samples wth more than 4000 pores/cm 2. CT scans of plugs reveals that the pore shape factor γ s a relatvely constant measure of the pore shape (+/- 0.2) throughout the sample plug. The larger the average pore sze, the hgher the velocty at a gven porosty. The average pore sze can be quantfed by the upper lmt of the domnant pore sze range. Introducton Carbonate rocks show often a lack of correlaton between porosty and other physcal propertes, n partcular permeablty and porosty. The hgh dagenetc potental of carbonates result n ntense alteraton of the pore structure, whch can lead to a decrease of effectve porosty for flow and wave propagaton. Permeablty and elastc propertes are strongly related to the rocks pore structure. As a result, samples of equal porosty can exhbt a wde varaton of permeablty and velocty. Two-dmensonal dgtal mage analyss can be used to quantfy the pore structure, not requrng any knowledge of pore type or dageness of the sample. Although dgtal mage analyss technque s performed on thn sectons, physcal propertes are related to the three dmensonal structure of the rock. Nevertheless our study shows that ncorporatng two-dmensonal nformaton from thn sectons provdes mproved estmates for permeablty and velocty predcton of the
2004, 66th EAGE Conference, Pars strata. The technque has the potental to be appled to very small samples, such as cuttngs, and thus mprove the characterzaton of reservor rocks wthout complete core recovery. Method We developed a dgtal mage analyss toolbox and desgned a workflow that combnes hgh resoluton full thn secton mage acquston and the use of XPL varaton nto a streamlned dgtal mage analyss process. Images are acqured at 5µm/pxel resoluton and full thn secton photomcrographs are constructed from plan-polarzed lght and XPL at dfferent angles. A XPL varaton attrbute s calculated from several mages acqured at dfferent angles. Ths attrbute s used as addtonal nformaton durng mage segmentaton. In partcular, the XPL attrbute s used to dstngush ar bubbles n pore spaces not flled wth dyed resgn (Fg. 1). In addton, the use of fully ntegrated relatonal database facltates the analyss of both physcal and mage derved parameters from wthn a sngle software applcaton. The XPL method produces a more accurate segmentaton nto bnary mages and allows for more relable porosty estmates and shape parameters. In addton, the use of open scrptng language (Matlab or Sclab ) and the resultng batch capablty provdes a very effcent method for segmentaton of large numbers of mages. As a result, statstcally more meanngful results can be derved quckly from a group of rock samples. Vsual parameter analyss through the use of mage- parameter cross plots elmnates the ambgutes ntroduced by verbal texture and pore type classfcaton. Fgure 1. Image (A) acqured usng plan polarzed lght shows a thn secton photomcrograph of a carbonate sample from the Bahamas that s mpregnated wth blue epoxy resn. Mnerals and grans are bege whle pore space s blue except for an ar bubble that s dentcal n color to the matrx mnerals. XPL s capable of destngushng ar bubbles (a) from mnerals (b). Pore shape factor γ, number of pores, 2-D mage porosty, total permeter and pore szes at 10, 50 and 90% coverage of cumulatve 2-D porosty have been extracted from over
2004, 66th EAGE Conference, Pars 180 bnary mages acqured from thn sectons of carbonate plug. Velocty, porosty and permeablty values have been measured on the correspondng plugs. Results 2-D macro-porosty from mage analyss: Macro-porosty s here defned by pores, whch are larger than 30 mcrons,.e., the thckness of the thn secton. In combnaton wth total porosty, t s used as an ndcator of effectve porosty wth respect to flud flow. Large devaton between mage analyss macro-porosty and total porosty ndcates sgnfcant amount of n-effectve mcropores, whch generally do not contrbute to flud flow. Those samples tend to plot at the hgh porosty margnal spectrum. Usng mage-porosty nstead of total porosty reduces the uncertanty n predcton of permeablty from 4 to 3 orders of magntude (Fg.2). Fgure 2: Permeablty versus porosty correlaton s characterzed by large uncertantes. The mage porosty that calculates porosty from dgtal mage analyss has a better correlaton manly because t elmnates the mcroporosty. Pore shape factor: Anselmett et al. 1998 establshed a correlaton of permeablty and pore shape factor γ for carbonate samples (Anselmett et al. 1998). Ths value s defned as: P γ = 2 πa where P = Permeter [mm], A = Area [mm 2 ] and γ = pore shape parameter [-]. By usng the square root of the area, the pore shape γ becomes a dmensonless parameter. For a perfectly round shape, the γ-value s one and any dervaton from ths form results n an ncrease of the γ-value. The average value of γ of one sample s calculated by weghng the ndvdual γ by the pore sze:
2004, 66th EAGE Conference, Pars γ = ( A γ ) A Ths procedure avods a hgh number of nsgnfcant, small pores wth specfc geometres domnatng a few larger pores wth a dfferent geometry, whch, due to ther sze, are much more mportant for the propertes of the whole rock. Compared to the porosty-permeablty trend, pore shape factor γ reduces uncertanty n permeablty predcton from 3 to 2 orders of magntude (Fg. 3). 10000 1000 Permeablty (md 100 10 Buhasa Malampaya 1 10 20 30 40 Porosty Fgure 3: The pore-shape factor γ has a better correlaton to the permeablty than porosty, despte the fact that only the pore shape s taken nto account. Total permeter: The measure of total permeter normalzed by the analyzed area can be used to separate n-effectve porosty from effectve porosty of the pore network. It s partcularly useful n separatng hgh and low permeablty samples. Samples wth a total permeter of more than 700 cm/cm 2 have permeabltes below 200 md, and rocks wth total permeter over 1500 cm/cm 2 show no hgher permeabltes than 10 md (Fg. 4). Ths mples that the more complex and tortuous the pore system wth the hgher the total permeter value are less permeable. Cumulatve 2-D porosty: The domnant pore sze s taken as the pore sze, at whch 50% of the thn sectons 2-D porosty s covered. Wthn a gven porosty range, plugs domnated by small pores tend to have a lower permeablty than rocks domnated by large pores.
2004, 66th EAGE Conference, Pars Fgure 4: Small numbers of Permeter/PoreArea values ndcate small and tortuous pore systems. Hgh permeablty s only found n samples wth small values. Fgure 5: The best correlaton s acheved when several dgtal mage analyss parameters are combned, as llustrated n ths combnaton or domnant pore sze versus permeter/pore area. References: Anselmett, F.S., Lüth, S. and Eberl, G. (1998). Quanttatve characterzaton of carbonate pore systems by dgtal mage analyss, AAPG Bulletn, Vol. 82, No. 10, p. 1815-1836.