Sei sulla pagina 1di 10

LA.

SPEASRM

47582 in Carbonates: Dependence on Stress Path and Porosity

Plasticity and Permeability

David P. Yale, SPE, Mobil Technology Company; and Brian Crawford, Heriot-Watt University

QwWht

f~.

SocieIy C4Petroleum Enginee$s, Inc Eurock 9S held m Tnmdheim.

7h!s paper was prepared fw presenlaiion ar me SPETSRM Nciway, S-10 July 1999

permeability enhancement due to microcracking depending on stress path and differential stress.
Introduction

This paper was sekcfed for pfesentatiw by an SPE Prcgram Cc+rmilke followng retiew C4 mformwcm contam+d m an abstract wdxmitted by (he author(s) Ccotents of the paper, as by the S.XMty of Petroleum Engineers and are subjed to presemed, haw not bem rewwd cwrecimn by the auffw(s). The material. as presented, does not necessarily reflecf any POSIIIC+I of tfw Soaefy of Pelrdeum Engmews, ds dfwrs, w membem. Pap.ws presented at SPE meebngs w? subjecl to @lIcalIcm mwew by EdNoi!al Cc+nmIftees of Lhe Soaely of Petrdwm Engineers Elecbwic rcwcduction, disbibidion, w stmage C4any part .S this paper fcf 1 WPOSISS wIthud tfWJ wfflen -sent of lhe .Scae(y of Petroleum Engmers is p+ciubded PemwssIon to repmd.ca k! Wt is resmcted !0 an abstracl of not more than 300 words, dlustrwons may !wf be c@ed. The abstract must contain ccmsplcucus acknowledgment of wtw?reand by whom the paper was presented Wnle Llbranan. SPE, P.0, Sax 833S36. FGchardxm. TX 750S3.3&3S. U.S.A.. fax 01-972.952.9435

Abstract

Pore collapse and the plastic behavior of high porosity chalks is well known but the extent to which harder, more moderate porosity carbonates are plastic is not as clear. We have experimentally determined the elastic and plast]c deformation and effect of that deformation on permeability for a suite of carbonate rocks spanning the porosity range of 14/0 to 42/0. The depositional environment of the carbonates was a shallow marine reef with some secondary porosity development. Biaxial compression tests were carried out along various stress paths where the ratio of axial to radial effective stress change was kept constant. Yield and failure surfaces and strain hardening response were mapped as a timction of porosity. Variation in permeability and compressibility as a function of porosity, lithology, and stress path were determined. The results were modeled with a critical state-cap plasticity model. The deformation of the carbonates generally followed a critical state cap model. The yield cap and hardening fimction were strongly dependent on porosity. Even moderate porosity samples ( 15-22/0) showed significant plastic strain under stress conditions attainable in many petroleum reservoirs. The compressibility of the samples in the plastic regime was a strong function of both porosity and stress path. One aspect of the observed deformation that did not fit critical state theory was the long transition from elastic to fully plastic behavior. We found we could represent this behavior with two cap surfaces: an initial yield surface and final hardening surface. Permeability reduction was a strong fimction of stress path. Permeability reduced less for a given increase in mean stress along stress paths with higher differential stress increments. Permeability reduction due to compaction was offset by

Large deformations associated with pore collapse have been well documented for high porosity chalks l234. The extent to which pore collapse and the subsequent deformation and fluid flow effect are relevant to harder carbonates (reef structures, lagoonal and oolitic deposits, dolomitized limestones) are not as well documented. The effects of plastic yielding and pore collapse on fluid flow in carbonates in particular have been studied only under a narrow range of stress states and lithologies. For this study, core samples were taken from a shallow marine carbonate reef structure at depth, Porosities ranged and although most samples were from 14~o to 42~0 predominantly limestone, several samples had undergone varying degrees of dolomitization. Part of the porosity was secondary due to subareal exposure in the past, but the sections from which the core samples were taken were not Vuggy. Stress Path Theory. The study included deformation testing of the samples along a variety of stress paths to define the yield envelope and post-yield behavior (see Fig. 1). We define the stress path, K, for laboratory tests as the ratio of change in effective radial stress, Au,, to the effective axial stress, Ar_r~, from initial hydrostatic stress conditions:
K = ArJ,lAo.

(1)

The effective stress, cr, is defined ass: o~=~-~p (2)

where S is the total stress, P is the pore pressure, and a is the Biot poroelastic parameter. The stress path that any given volume of reservoir undergoes during drawdown can be given by:
Km = ACrhfAGv

(3)

where Acrvis the change in effective vertical stress and Auh is the change in effective horizontal stress as measured from

485

..

YALE AND CRAWFORD

SPE 47582

initial reservok conditions. This reservoir stress path is often approximated by that stress path which leads to zero lateral or horizontal strain. For an elastic material with constant Poissons ratio, v, this no-lateral-strain stress path leads to a ratio of vertical to horizontal stresses equal to:
Ko=
V/(I-V)

(4)

However, in elastoplastic materials, such as the carbomtes tested, the Poissons ratio of the material is not constan~ but stress dependenq leading to the no-lateral-strain stress path being highly nonlinear. For example, in Fig. 1 we show the no-lateral-strain stress path for one of the 20?40samples, In the early stages of the test, KO=O.15 but it increases to a value of nearly 0.6 beyond the yield cap. In addition, field measurements by several authors67 have shown that in many reservoirs, the actual reservoir stress path can deviate substantially from that calculated for no-lateral-strain via equation 4. Therefore in this study we have used a variety of stress paths both to span the range of probable in-siiu reservoir stress paths and to better define the yield cap. Permeability versus Stress. Permeability measurement on a subset of the cores was also conducted in order to understand how stress and yieIding affect the fluid flow properties of these carbonates. Several recent studies have sought to nonunder quantify stress-dependent permeabilities hydrostatic triaxially compressive stress configurations. FIo1t8 found that for a weak sandstone, permeability reduction was more pronounced under non-hydrostatic applied stress, compared with the slight decrease measured under hydrostatic loading, Permeability was significantly reduced at non-hydrostatic stress conditions (up to < 10/0 of its initial value) when the shear stress exceeded the yield level. The sharp decrease in permeability after yielding was interpreted as a change in specific surface area by the crushing of fine grains andfor a change in tortuosity resuhing from changes in the crack pattern during yielding. Rhett and Teufe19 conducted a series of stress path ( 1.02K 20) triaxial compression tests on two reservoir sandstones, in which they found that with decreasing K, the reduction in matrix permeability, with increase in effective stress, rapidly declined. Permeability actually increased at 0.52K and was greatest for the lowest stress path ratios. With radial strain monitoring indicating negligl%le dilation, the exact mechanism responsible for their observed permeability increases was not clearly understood, although the authors did suggest that matrix permeability may be related to shear stress, which increases with decreasing K. Ferfera et al*Oconducted permeability measurements on a Vosges sandstone (20%. porosity) under a variety of nonhydrostatic stress paths including: K =0, 0.1, and 0.2 and constant mean shess, p, loading. The results were analyzed in terms of the relative influence of the mean effective stress, p, and the differential stress, q. A permeability threshold was below which negligible defined in p-q stress-space, permeability impairment was observed, and above which

significant reduction occurred (as high as 600/0in some cases). This permeability criterion was below the damage or yield threshold (defined as the point where the Youngs and shear moduli decrease). Permeability reduction as a function of mean effective stress from the uniaxial compression tests, showed that the higher the confining pressure, the higher the final decrease in permeability. Plotting the same permeability reduction data as a fimction of differential stress leads to a single function curve. A unique permeability reduction rate (linear slope of permeability decline with increasing differential stress outside the threshold criterion) was obtained for different stress path values - the more differential the stress path the lower the reduction and conversely the more isotropic the stress path the higher the permeability decrease. Teufel et at and Teufel and Rhettl 1 have conducted triaxia] compression tests on high porosity (>34VO)reservoir chalk from the Ekotisk Field, North Sea, under stress paths of K = 0.17, 0.25 and 0.33 (representative of those seen in the field). Significantly, all tests resuked in macroscopic shear failure of the specimens, as differential stress increased with drawdown. Contrary to the assumption that permeability will decrease with reservoir compaction and porosity reduction, these authors contend that such shear failure has had a beneficial intluence on production, by increasing fracture density, reducing matrix block dimensions, and thereby maintaining producibility.
Critical State Theory. In addition to the experimental testing, numerical geomechanical modeling using a critical state theory was done to determine how well established models could represent the behavior seen. Fig. 1 shows the critical state representation of the stress states for yielding and hardening of one set of the carbonate samples (those with porosities in the 20% range). The stress states are given in p-q space where p is the mean effective stress (average of three principal effective stresses) and q is the differential stress (difference between the maximum and minimum principal stresses), As shown in Fig. 1, the classical critical state theory 1213 represents the deformation of elastoplastic material with a yield cap (initial yield cap in Fig. 1), within which the material behaves elastically and outside of which the material behaves plastically. In addition, a critical state or failure line is defined above which the material fails brittlely. Our experiments were modeled best by defining a second critical state cap. This hardening cap as shown in Fig. 1 delineates the onset of plastic hardening. In classic critical state theory, elastoplastic materials harden immediately upon exceeding the yield cap. In all of our tests, there was a significant gap in stress space between initial yielding of the material and the onset of volumetric strain hardening. Experimental Procedures

All testing was done on 1.0 or 1.5 inch diameter, vertical core plugs. Atler plugging, core ends were ground flat, the samples dried, and porosity, nitrogen 4Klinkenberg permeability and grain density measured. Samples were loaded into Hock-type pressure cells that were then placed in geomechanical load

486

SPE 47582

PLASTICITY AND PERMEABILITY IN CARBONATES: DEPENDENCE ON STRESS PATH AND POROSITY

frames for testing. Axial, radial, and pore stresses were brought up to the initial conditions for the test and allowed to equilibrate for 4-8 hours. For one subset of samples (deformation testing with no liquid permeability), axial strain, radial strain, and pore strain were measured continuously during sample loading. Tests were taken well beyond yielding in most cases. Axial stresses (or strains in some cases) were incremented at constant rate and radial stresses were increased via computer control as a fimction (stress path coefilcient, K) of the axial stress, For another subset of samples (deformation testing with liquid permeability), axial strain, vohunehic strain, and fluid pressure drop across the sample due to constant flow rate pore fluid injection were measured continuously. Again, axial stresses were increased and radial stresses were increased to follow the specified stress path, It is important to note that in all tests, all increments of load and measurement of strains and permeabilities were done in a continuous fashion so as to be able to accurately pick out the subtle features of initial yielding and post-yield hardening. In most of the tests, pore collapse was the observed form of yielding rather than dilatency and shear failure. With the exception of the shear failure tests, most of the samples were run under constant stress rates rather than tbe more classic coustant strain rate of failure testing. Effective strain rates for all tests were in the range of 2 to 6 x 104 see-. The constant stress rates allow more accurate determination of yielding, hardening, and post-yield property determination. In addition, fluid withdrawal from reservoirs generally leads to an approximately constant stress (pressure) rate environment rather than a constant strain (deformation/ compaction) rate environment. Analysis. In addition to calculating axial, radial, pore, and volumetric strain and axial and radial stresses, tangent moduli and permeability were also calculated along with initial yield point and onset of hardening. Tangent compressibilities were calculated from the slope of the stress-strain curve. In the pre-yield region, these moduli corresponded to the elastic moduli of the materials. We define the tangent pore compressibility as:

measurements were done by flowing a fluid of known viscosity at a constant flow rate through the cylindrical rock samples and measuring the resultant pressure drop across the sample. The carbonate samples tested for permeability were found to fall into two distinct groups depending on their initial permeability and therefore viscosity of the perrneant. High poroperm samples had sufficient permeability to warrant the use of a viscous (23-30 cp depending on temperature and pressure) refined mineral oil. Low poroperm samples required the use of a low viscosity (1.2- 1.3 cp depending on temperature and pressure) paraffin permearrt. The cutoff between the two groups was found to be a nitrogen permeability of 5 mD. Fluid injection rates varied between 15 and 100 @Anin.
Results

Sample properties and test parameters for the 39 tests run in the study are shown in Table 1. The majority of the samples were limestones with a few partially dolomitized samples and a couple of samples of nearly pure dolomite. Thin section analysis shows that those samples with grain densities in the 2,76 to 2.78 range have 25~0-35V0dolomite. Those with grain densities in the 2.69 to 2.73 range have very little dolomite and those with densities in the 2.84 range are nearly entirely dolomitized. As shown in TabIe 1, a wide variety of stress paths were tested for each porosity interval. The samples are categorized into four porosity ranges, 20/0, 25/0, 30?o,and 40%. All samples are from a shallow marine reef structure and none can be characterized as chalks. Effect of porosity, Iithology, and stress path on yielding. The initial yield point in p-q space for each sample (as defined by the change in slope of the pore strain and axial strain plots) is shown in Fig. 2. Note that the yield points for samples within a given porosity range can be well characterized by the classic elliptical yield surface of the critical state or Cam-Clay constitutive model for geomaterials. The slope of the best fit critical state line corresponds to a Mobr-Coulomb failure line with a 42 angle of internal friction. Four samples (labeled RS) which failed brittlely help define the critical state line. The stress state for the samples residual strength is shown in Fig. 2. Fig. 2 also shows that porosity and dolomite content are the main factors controlling initial yield in these carbonates. The 20%.porosity carbonates have yield stresses nearly four times higher than the 40~o porosity carbonates. In addition, small differences in porosity can yield large changes in All the 20% samples near the yield cap have strength. porosities between 18.5V0 and 21?4.. Two samples with porosities of 22V0and 23V0are significantly closer to the 25V0 porosity yield curve. Two samples with 16.9% and 15.1V. porosity yielded at higher stresses than the remainder of the samples. The scatter in the yield for a given porosity range is larger for the higher porosity samples than for the lower porosity samples due to heterogeneities in the samples.

C*= (NJ V*)/ Scrm

(5)

where VP is the pore volume and cr~ is the effective mean stress. The onset of initial yielding was determined fkom the slope of the pore strain-mean stress and axial strain-axial stress plots, When the curve deviated more than 10/0 from linear, yielding was said to have occurred. We define our hardening cap as the point at which hardening started after initial yielding. Hardening was defied as an inflection point in the pore strain-mean stress or axial strain-axial stress curve after initial yielding. This inflection was most easily seen in the tangent compressl%ility plots as a Iocal maximum in compressibility after yielding. Following horn Darcys law, steady state permeability

487

YALE AND CRAWFORD

SPE 47582

Five of the samples with 25-95% dolomite are represented in Fig. 2 by labels showing their grain density, Dolomite contents of 25/0-350/0(grain densities of 2.76-2.78) sometimes show more strength than samples of similar porosity and sometimes not. Samples of nearly pure dolomite show very large increases in strength as compared with limestone samples of similar porosity. Fig. 1 shows the initial yield cap and hardening yield cap (as defined in the Procedures section) for the 20/0 porosity samples. The yieId and hardening points of most of the samples defiie well the two yield caps. Two samples shown with filled in symbols have slightly higher and slightly lower porosities than the remainder of the samples and have lower and higher yield points respectively than the other samples. The four samples lying near the critical state line represent samples that failed in shear or dilation. The lower two points failed in shear whereas the higher two points compacted but underwent dilation. Both these points are near the intersection of the critical state line with the yield and hardening caps and thus the material behavior is expected to represent the transition zone between brittle and plastic behavior, Standard criticaI state theory predicts that once the samples yield, strain hardening occurs (i.e. the initial and hardening yield caps should be the same). However, we note in all the samples in all our porosity ranges, that there is a elastoplastic transition period between initial yield and the onset of hardening. The difference between these two caps is greatest for the lowest porosity samples (hardening cap 8000 psi higher than the initial yield cap) and least for the higher porosity sampIes (hardening cap 1500 psi higher than initial yield cap), The importance of this secondary or hardening cap is significant for petroleum reservoirs. We have found that tangent pore compressibility increases dramatically between the initial yield and hardening cap before decreasing once volumetric hardening occurs. This process can occur over several thousand psi of drawdown and depending on depth, fluid pressure, porosity and lithology, the sample may or may not ever go into classic plastic hardening. In determining compressibility parameters to put into reservoir engineering modeIs, compressibility becomes a complex finction of effective stress, porosity, and lithology. Permeability reduction and differential stress. In Fig. 3 and Fig. 4, plots of differential stress, q, and permeability reduction, +dko> variations with axial strain, &axial, are shown for all test samples from the high and low poroperm Iithotypes respectively. TIigh poroperm specimens #1, #2, #4, #6 and #9 all show distinct yield points (where the stiffness of each sample changes sharply). Specimen #1, possessing the highest initial porosity, shows yielding at the lowest differential stress (= 725 psi), while specimen #6 with the highest dolomite content is seen to accommodate the greatest differential stress (= 14500 psi) before yielding. Sample #7, subjected to conventional triaxial compression (K

= O) under low initial mean effective stress, p, produces a peak load response, prior to a stress drop associated with deformation localization into a single open shear fracture. Permeability remains at a constant level of some 1S%o reduction during fi-ictional sliding on the induced macrofault. Sample #5, subjected to the largest stress path ratio, K = 0,7, does not show an obvious yield point due to the low level of differential stress achieved. Permeability decline is initially steep but flattens out at higher strain. Samples #2, #4, and #9 all show robust logarithmic negative comelations between permeability decline and axial strain, specimen #l showed a linear decrease tending to steepen at higher strain, and #6 a broadly linear permeability reduction over the elastic to plastic transition. Thus, in general, the high poroperm samples show little deviation in the trend of permeability decline over the elastic to plastic transition. The low poroperm specimens show markedly different permeability reduction responses from the high poroperrn sample suite. In Fig. 4 samples #11, #15, and #16 show distinct yield and hardening points with permeability reduction showing a shallowing of slope post hardening. In the case of sample #12 (K=O, 6000 psi initial mean stress), we see permeability recovery over the hardening phase associated with volumetric dilation (the sample was noticeably barreled on removal from the pressure vessel). Permeability shows an initially sharp decline during linear elastic deformation decreasing to a minimum value in the elastoplastic transition zone followed by permeability recovery of some 15t associated with strain hardening. Sample #3 exhibited localization of deformation into a network of open shear marcrofiactures with an attendant fhre-fold increase m permeability associated with the transition from matrix to fracture flow. In Fig. 5, normalized permeability is plotted as a function of differential stress for the high poroperm sample suite Following Ferfera et aI.0 a permeability reduction factor (PRF) is defined as the slope of a straight line fit to such data, Fig. 5 illustrates a fan-like spread of linear permeability reduction rates, with PRF decreasing with decreasing stress path ratio magnitude. It is also evident that the higher the porosity for a given value of stress path ratio, the greater the reduction in permeability (compare samples #1, #2, and #9). Likewise, sample #6, composed of 90%0 dolomite, shows an anomalously low PRF for the high poropenn carbonate Iithotype. Thus it is apparent that the mechanically weaker the carbonate material, (high porosity or low dolomite content) the correspondingly greater is its PRF value, In Fig. 6, linear permeability reduction factors are defined for the low poroperm carbonate suite (note that the linear fits appear as curves as the normalized permeability abscissa is on a logarithmic scale). Again PRF is observed to decrease with decreasing stress path ratio, however, this time permeability is seen to recover for K=O samples #12 and #3 (PRF for such cases being defined only for the permeability reduction part of each curve). The permeability recovery of sample #12 is associated with distributed damage leading to volumetric dilatency during strain hardening while the

488

SPE 47582

PLASTICITY AND PERMEABILITY IN CARBONATES: DEPENDENCE ON STRESS PATH AND POROSITY

permeability enhancement observed in sample #3 is attributed to fracturing as an end-product of brittle deformation. The lower the stress path ratio, and thus the greater the differential stress change for an increment of mean stress, the less permeability is seen to be impaired in both the high and low poroperm carbonates. Thus microcracking is advanced as the most likely differential stress controlled mechanism for This observed trend of permeability enhancement. permeability reduction becoming less (PRF decreasing) with decreasing stress path ratio is attributed to a progressive increase in the extent of dilatent microcracking over compactive processes. Effect of stress path on deformation and compressibility. The effect of stress path on the axial strain versus axial stress plots for the samples in the 30V0 porosity range is shown in Fig. 7. The K=O path tests show the classic flattening or strain hardening of the stress-strain plot after yielding. All the other stress path tests show an upward concave behavior afier yielding. This upward concave behavior of the stress-strain plot has not been discussed in the literature before. As the ratio of horizontal to vertical stress increases (K increasing), we note that the curvature of the stress-strain plots after yield increases. In addition, the higher the stress path ratio, the steeper the average slope of the curves after yield. This corresponds to an increase in the tangent moduhrs after yielding with increasing stress path coeftlcient. Considering that most critical state and Cam-Clay type models represent plastic strain hardening as a relationship between volumetric plastic strain and mean stress, this concave upward stress-strain behavior appears intuitively correct. As the stress path coeftlcient, K, increases there is an increasing component of mean stress in the total stress applied to the sample. As such, we see more of a volumetric or mean stress behavior in the axial stress versus axial strain plot. The volumetric strain-mean stress plot for a volumetric strain hardening material should be a concave upward curve in the plastic zone as the material stiffens or hardens, The effects of stress path on tangent pore compressibility in the hardening regime for the 25/0 porosity range samples All the samples showed a low are shown in F@. 8. compressibility (between 5 and 10 X 104 psi- or 5-10 microsips) in the elastic or pre-yield regime. After yielding, the tangent pore compressibility increased dramatically in the p-q stiess space between the yield and hardening caps. Fig. 8 shows the compressibility from slightly before the hardening We see the peak tangent cap to the end of the test. compressibility in the plastic regime vanes between 35 microsips for stress paths K=O.36-O.44 up to 110 microsips for a stress path of K=(3 for the limestones in this porosity range. dependence of have noted this strong 0thers9 compressibility on stress path in carbonates. However, it has not been shown previously that taken all together, the compressibility along various stress paths forms a single relationship with respect to mean stress as shown in Fig. 8. This relationship appears similar to the classic volumetric hardening one would expect from a single test taken over this

range of mean stress. Similar behavior as shown in Fig. 8 was found for the samples from different porosity ranges. Each porosity range, however, had a different pore compressibilitymean stress relationship. In addition, the maximum compressibility for each porosity range was a function of the porosity. It ranges from 350 microsips for the 40~o range, to 250 for the 30% range, to 110 for the 25% range, to 40 microsips for the 20A range.
Modeling and Discussion

As mentioned in the RESULTS section, the behavior of the carbonates studied follows the general trends one would expect from a volumetric strain hardening material under classic critical state theory. In order to test this, we ran a number of geomechanical simulations using a critical state model. The numerical geomechanical model used was the VISAGEm model from VIPS, Ltd. The VISAGEm code is a fully coupled geomechanics-fluid flow code that allows the user to run simulations using a number of different constitutive models. For these simulations, the critical state constitutive model was used, The simulations were run on a small, 16 element, axisymmetric FEM grid with boundary conditions similar to those experienced by the sample in the Laboratory. The stress pathways were simulated by applying loads radially to the model in proportion to the axially applied loads via the stress path coeftlcient, K. A yield cap similar to the SOYO carbonates was used along with a friction angle of 40. A hardening parameter of 70 was used. Stress paths of K+ 0.3, and 0.5 were simulated. All three stress paths started with an initial hydrostatic stress of 2000 psi. Fig. 9 and Fig. 10 show the results of these simulations. Fig. 9 plots axial strain versus axial stress in order to compare with Fig, 7. We see the same behavior. For K=O, we see a steep elastic response followed by the plastic strain hardening response (concave downwards). For the K=O.3 and 0.5 plots we see the same concave upwards appearance of the stressstrain line in the plastic region as we observed in the data. In addition, the average slope of the line and the degree of curvature are larger for the K=O.5 simulation than for the K=O.3 simulation. This is the same trend we observed in the experimental results. The modeled results show steeper slopes in the post-yield space than the experimental data suggesting a smaller hardening parameter is needed in the model to match the results more exactly. The only experimental observation not reproduced by the simulation is the slow turnover of the experimental stressstrain curves as compared with the abrupt transition from elastic to plastic strain hardening of the model results. Fig. 10 shows the tangent pore compressibility for the simulations as a timction of mean stress. We see the same increase in compressibility from an elastic value of 10 rnicrosips and the decrease in compressibility due to volumetric stiain hardening. We see that all compressibilities for the different stress paths fall along the same hardening curve. The difference in peak compressibility for each stress

489

YALE AND CRAWFORD

SPE 47582

path is simply due to the fact that the higher K stress paths intersect the yield cap at higher mean stresses and therefore do not intersect the hardening function at as high a compressibility value as the lower K stress paths. For the experimental results (see Fig. 8) we have the same effect, except that the peak compressl%ility for any given stress path is given by the mean stress at which the stress path intersects the hardening cap, not the initial yield cap. Thus the K=O experimental stress path reaches a peak compressibility at 5000 psi mean stress whereas the K=O.36 test does not intersect the hardening cap until 10000 psi mean stress. This shows the importance of these two caps in modeling actual reservoir behavior. Depending on the depth, initi~ reservoir pressure and porosity of the carbonate, a very significant portion of its stress path during reservoir depIetion could lie in between the yield and hardening cap. Yet this is the specific zone in stress space where the current constitutive models do not adequately represent the reformational behavior of the material. Others have noted the delay in onset of h~dening afler initial yielding in chalks, Addiasl*4 introduced the concept of collapse zone outside of the initial yield cap with a a pore return to normal consolidation (hardening) beyond &is pore collapse. Scott et al. 15~s also obse~ed this phenomena. is work on acoustic velocities through this pore collapse or elastoplastic transition zone clearly delineates the different behaviors. The variety of stress paths chosen for our tests show for the fwst time, however, that this pore collapse zone ends along an elliptical yield surface. This elastoplastic transition zone has also been seen in much harder rocks at very high mean stresses. Both Evans and Kohlstedt 16 and Edmond and Paterson] 7 observed this behavior in low-porosity marbles. In those materials, this transition zone is generally referred to as the semi-brittle region between the brittle (elastic) and purely plastic deformation regimes, Evans and KohlstedtiG point out that adequate constitutive laws do not exist for this semi-brittle zone and that the potential complexity of mixed deformation mechanisms will make it difficult to model this behavior with a single theory. Two of the deformation tests and two of the permeability tests mu showed dilatency, For samples 20C and 20E, pore collapse occurred once the initial yield cap was exceeded-and then dilatency occurred once they reached near the critical state line. The samples appear to be right at the brittle-ductile transition zone at the end of their tests, but it is unexpected to see both pore coIIapse and diIatency occur in the same test. The permeability tests, however, clearly show dilatent and compactive processes occurring at the same time, Several of the low poroperm samples show permeability-differential stress relationships that suggest that significant microcracking and dilatency is occurring in the samples. The strong relationship between stress path and permeability decrease is also indicative of a competition between compactive and dilatent forces. These moderate to high porosity carbonates, therefore, appear caught in the middle between rock mechanics and soil
490

mechanics theories. At the high porosity end, these carbonates fit very well within soil mechanics critical state theory and chalk behavior. The elastoplastic transition zone is small and they show very clear plastic hardening, At the lower porosity end, however, the elastoplastic transition zone is large, they show mixed modes of deformation (compaction and dilation) along a single stress path and behave more in the brittlesernibrittle-ductile mode of classic rock mechanics. However, moving from the lowest to highest porosity samples, we see a very smooth transition of phenomena. Further examination of these ~es of materials may help form a bridge between soil mechanics and rock mechanics types of constitutive models.
Conclusions 1.Yielding and pore collapse in carbonates is a strong tlmction

-.

of porosity and dolomite-limestone ratio. 2.Pore collapse under stress conditions achievable in petroleum reservoirs can occur in carbonates with porosities of 15%and possibly lower. 3. There can be a significant gap between the onset of yielding and the start of volumetric strain hardening in carbonates. This elastoplastic transition zone is similar to the semibrittle zone seen in hard rocks at very elevated pressures 4. Compressibility in carbonates is a strong fiction of the stress path under which the material is deformed. Stress paths with higher stress path coeftlcients, K, have lower compressibilities than those with lower values of K. 5. Critical state theory can model most of the experimental} observed reformational behavior of carbonates However, the delayed onset of hardening and its effect on the compressibility of the material in the plastic region IS not represented by classical critical state theory. 6. Permeability decline in carbonates is a strong fimction of the relative contributions of compaction and rnicrocracking with compaction decreasing permeability and microcracking enhancing permeability. 7. Permeability decrease for a given increase in mean stress is less along stress paths with higher differential stress increments, Stress paths with higher K (higher contributions of mean stress) show larger decreases in permeability than stress paths of lower K (higher contributions of differential stress). 8. Permeability enhancement due to microcracking and to shear failure is a fimction of initial permeability. Lower permeability rocks are more likely to show significant enhancement due to microcracking and shear failure than moderate to high permeability rocks. 9. Both high and low poroperm sample suites show linear negative relationships between normalized permeability and differential stress. Permeability reduction factor, PRY, is seen to increase with increasing stress path ratio, The permeability reduction factor increases with increasing porosity and decreasing dolomite content, that is the mechanically weaker the carbonate, the greater the PRF value.

u.

SPE 47582

PLASTICITY AND PERMEABILITY IN CARBONATES: DEPENDENCE ON STRESS PATH AND POROSITY

Acknowledgements

The authors would IiIce to gratefully acknowledge the laboratory assistance of J. Michael Rodriguez and Roger Hutcheon. Brian Crawford was supported by a Mobil PostWe also thank the Doctoral Fellowship for this work. management of Mobil Technology Company for permission to publish this study.
S1 Metric Conversion Factors

presented at 3ti North American Rock Mechanics Symposium, June, Cancun. 16. Evans, B. and Kohlstedt, D. L. (1995) Rheology of rocks, Rock Physics and Phase Relations: A Handbook of Physical Conslanls, ed. Thomas J. Ahrens, American Geophysical Union, Washington, D. C. 17. Edmond. J. M. and Paterson. M. S. (1972) Volume chatwes during deformation of rocks athigh pre;sures; Int. J. Rock Me;h
Min. Sci. Geomech. Abstr., 9, 161-182.

psi x 6.894757 md x 9.869233 psi- x 0.14503

E+OO=kPa E-04 =pmz E+OO =kPa-

Table 1-

Sample

properties

and test parameters

References 1. Addis, M. A., 1987, Material metastability in weakly cemented sedimentary rocks. Memoir of the Geological Society of Chirraj No. 9. 2. Johnson, J. P. and Rhett, D. W., 1986, Compaction behavior of Ekofisk chalk as a function of stress, presented at SPE European Petroleum Conference, SPE # 15872, London, October. 3. Blanton, T. L., 1981, Deformation of chalk under confining pressure and pore pressure. Sac. Pet. -&g Jour., February 1981, 43-50. 4. Jones, M., and Mathiesen, E., 1993, Pore pressure change and compaction in North Sea chalk hydrocarbon reservoirs. Int. J. Rock Mech. Min. Sci,, 30, 1204-1208. 5. Nur, A. and Byerlee, J. D. (1971) An exact effective stress law for elastic deformation of rock with fluids, J Geophys. Res., 76, 6414-6419. Teufel, L.W., Rhett, W. & Farrell, H.E., 199 I: Effect of Reservoir 0, Depletion and Pore Pressure Drawdown On In Situ Stress and Deformation In the Ekoflsk Field, North Sea, Rock Mechanics Roegiers (cd.), Balkem% as a Multidisciplinary Science, Rotterdam 7 Jones, M.E., Leddra, M. J. and Addis, M. A, ( 1987), Reservoir compaction and surface subsidence due to hydrocarbon extraction, Department of Energy Offshore Technology Report OTH87276, HMSO, London. 8, HoIt, R. E. (1990) Permeability reduction induced by a nonhydrostatic stress field, SPE Formaflon Evaluation, 5, 444448.

26,2 25C 25.2 25D 27.0 25E 25F <#6> ~ .6 25G [ 24.6

<2.71~ <2.73> <2.70> +.72>

<3700>

0.2 0.28 0.36

<1000> <2000> ~ <2500>

Dkformatlon Det%rmatmn Dcformalmn Delbfiaiion

9 Rhett. W. and TeufeL L. W. (1992) Effect of reservoir stress path on compressibility and per%eab~iity of sandstones, presented at SPE 67* Annual Technical Conference and Exhibition, SPE #24756, Washington, D. C., October 1992. 10. Ferfera, F. M. R., Sarda, J. P., Bouteca, M., and Vincke, O. (1997) Experimental study of monophasic permeability changes under various stress paths, Proceedings of 3ffh U. S. Rock Mechanics Symposium, Kim (cd.), Elsevier, Amsterdam. 11. Teufel, L.W., & Rhett, W., 199 I: Geomechanical Evidence for Shear Failure of Chalk During Production of the Ekofisk Fiel~ SPE #22755 presented at the 1991 SPE 66th Annual Technical Conference and Exhibition, Dallas, Oct. 6-9. 12. Schofield, A. N. and Wroth, C. P. (1968) Critical Slate Soil Mechanics, McGraw Hill, London. 13. Chen, W. F. and Mizuno, E. (1990) Nonlinear analysis in soil mechanics, Elsevier, Amsterdam. 14. Addis, M. A., 1989, The behavior and modeling of weak rocks,
Rock at Great Depth, Fortnintraux (cd.), Balkema, Rotterdam. 15. Scott, T. E., Azeemudin, M., Zammatr, M. M., Roegiers, J.C. ( 1998)Shear induced variation in acoustic velocities in chalk,

...
~

409 40C 40D <#l>

37.7 35.7

<2.71> +.71> <2.78>

0.28 0.44 0.5

<2000> <1000> <1000>

DeforiiatiOn Perm <801>

491

u.

YALE AND CRAWFORD

SPE 47582

Stress Paths and Yield Caps


Limestones

20% porosity G *

20000
K=O stress path

/
,. . ncimt%rstzitr
line _ K=O.5 stress
Stress

~
n

15000

paths

Yield caps

_Nmlateralstrain th la21E A sa Ies 20E2% I


G

20c 17%

5000

10000

15000

p - mean stress [psi]


Figure 1- Critical state representation of laboratory stress paths in pq space and initial yield and hardening caps in the 20% porosity samples. Stress path followed by no-lateral-strain test shown in bold.

Initial yield caps - All carbonate

samples
---. 24 2.83 6 0

10000

1
G

.
.

*
200

Cxiticalstate Iine cap 40% cap 3076 Yield cap


zs~o

Yield Yield

G .
.
. . . * , ,*

..

.. .

;; ----- +-Q&JB.. . . . ..Rg-O* .. .. * .. 0 RS *-

23 -. 00 i%. .. .> . . . . . . . .

RS 0=*

22

Yield cap 20!fo 40% range samples .3070 range samples 25% range sarrples 20% range sar@es

G
. ,

El

A
o

2000

4000

6000

8000

10000

p - mean siress [psi] Figure 2- Best fit initial yield caps for each porosity range and actual yield points for each sample. Samples with grain densities greater than 2.74 or porosities wtside ranges are labeled. RS samples failed in shear and residual stress state after failure is plotted.

492

SPE 47582

PLASTICITY AND PERMEABILITY IN CARBONATES: DEPENDENCE ON STRESS PATH AND POROSITY

20

#12

..

15

#11 x
e

#3 5 #16

0 -1

----

____

#5 % #2 02 4

#4

10

,oLtItIIz
o

10

&axial (A)
Fig. 3-High poroperrn carbonates Differential stress and normalised permeability versus axial strain plots. Fig. 4--Low poroperrn carbonates Differential stress and normalised DWMeabilkV veraus itXkIl strain plots. 1 I 1 1 I

1000
1

00

g
25

0.7
o o

!p 3.5

, 0.5 7.0 10.5

\ 14.0 17.5 21.0

10 0

3.5

q xl Oh3(psi)
Fig. 5-High- poropenn carbonates Normallsed permeability veraus differential Fig. 6--Low stress

7.0 10.5 14.0 17.5 21.0 (psi) q XIOA3


poroperrn carbonates versus differential stress

Normalised permeability

10

YALE AND CRAWFORD

SPE 47582

Axial Strain vs. Axial Stress 20000 .


Limestones 30A porosity
1

Axial Strain vs. Axial Stress


20000
\

Critioai State Model

K:

0.5

J
0

. .__
0.02

....
0.04 Axial strain 006

-.------- ~
008

---0.1

0.00

0,01

0.02

0.03

0,04

005

Axial strain Figure 9- Axial stress veraus axial strain from numerical model of critical state - volumetric hardening model.

Figure 7- Axial stress versus axial strain for various stress paths for samples in the 30% porosity range. Axial stress normalized for different initial radial stress values.

12E-04

Tangent _
o

Pore Compressibility
Limestones 25%
pOrOSity

Tangent
ioE-04

Pore Compressibility
.-

Critical

State Model

IoE-04 8,0E-05

( (

O. 8 0.2
:4

1 -

K: 0.44

??
2.OE-05

2.OE-05

.
!

O OE+OO . 400Q

aooo

.+ ?3300

mm

0.0Eu30 2UO0 4000 moo Qooo

ecnJo

p- AWanstre~Y[psi]

p -mean
Figure 8Tangent

stress

[psil Figure 10- Tangent pore mmpressibifily as a function of mean stress from numerical model of critical state - volumetric hardening model.

pore compressibility

for various stress paths for

samples in the 25% porosity range. Compressibihw values through the hardening cap and beyond are shown.

494

Potrebbero piacerti anche