You are on page 1of 6
37/9 37/9 The Effect of Overburden Stress on Geopressure Prediction from Well Logs Ben A. Eaton, SPE-AIME, Universal Driling and Engineering Consultants, Ine. Introduction In 1965, Hottman and Johnson’ presented a method for predicting geopressure magnitudes by using resis- tivity and sonic log data. This technique has received ‘wide acceptance even though the prediction charts were based only on data concerning Tertiary Age sediments in the Gulf Coast area. It was specifically pointed out that these techniques were applicable only in areas where the generation of geopressures is primarily the result of compaction in response to the stress of overburden, ‘Compaction caused by overburden stress was de- seribed classically in a soil mechanies book by Ter- zaghi and Peck in 1948." With a vessel containing fa spring and a fluid, they simulated the compac- tion of clay that contained water. Overburden stress ‘was simulated by a piston, as in Fig. 1. [twas shown that the overburden stress, 5, was supported by the stress in the spring, g, and the fluid pressure, p. Thus, the long-accepted equation of equilibrium was es- tablished. Saoto a If Fig. 1 and Eq. 1 are studied, it is obvious that if Sis increased and the fluid is allowed to escape, ¢ ‘must increase while p remains as hydrostatic pres- sure, However, if the fluid cannot escape, p must also increase as S is increased, Hubbert and Rubey* published a comprehensive treatment of this theory as related to sedimentary rock compaction. They showed that as the overbur- den stress is increased as a result of burial, the poros- ity of a given rock is decreased. Therefore, some fluid that was once in the pores of a given formation was later squeezed out by compaction. In many such ceases, there is no escape route for the fluid, and thus. the fluid becomes overpressured according to Eq. 1. This happens in many areas, and such generated overpressured zones are often called “abnormal” pressure zones or “geopressure” zones. Holtman and Johnson recognized the main sig- nificance of the preceding theory and developed a very useful relationship between electrical log pro- pperties and geopressures. They reasoned that since rocks are more resistive to electrical current than is formation water, a well compacted shale containing less water (because the water has escaped) is more resistive than a less compacted shale containing more water (one in which the water has not escaped to the same degree), Also, they reasoned that a sequence of normally compacted sediments (in which water is free to escape) should have a normally increasing re- sistivity trend. They substantiated this when they plotted resistivities from actual well logs. Any re- sistivity decrease from the well established normal trend indicates the presence of abnormally high- pressured zones, Empirical data from well tests and logs were used to develop a correlation of the pore pressure gradient as a function of the resistivity departure ratio (see Fig. 2). A similar idea was used in conjunction with Think of the money that could be put to better use if we could predict the depth below which commercial production will not be found. It has been suggested that the magic level in geopressured areas is where log resistivity ratios exceed 5.50. The theory offered here, with the hope that it will be carried further, is that the limiting ratio is a function | also of overburden stress gradient, auoust, 92 7 PT 229 similar empirical data to develop a correlation of the pore pressure gradient with sonic-og travel-time de- parture from normal travel times (see Fig. 3). Note {hatin each correlation, only one line is drawn and the data points scatter considerably. Such’a scatter indi- ‘cates that the chances of error in the pressure predic- tion magnitude are high when values are read from the line, Timko and Fert have shown a different correlating curve that more accurately predicts the magnitudes of gcopressures in one localized area (see Fig. 4). They have gone one step further. They indi- cate that commercial production is never found below the depth at which the resistivity ratio reaches a value of 3.50, regardless of the magnitude of the pore pres sure gradients that actually exist. Considerable data aze shown to verify this contention. They theorize that as the resistivity ratio increases, the size of the hydrocarbon reservoir decreases. The theory is more of less substantiated statistically, However, Fowler’ pointed out several large reservoirs with original pore Pressure gradients greater than 0.85 to 0.90 psi/ft Such high gradients are well above the 3.50 resistivity ratio on the Hottman and Johnson correlation shown in Fig. 2, but not on the Timko and Fert! correlation of Fig. 4. Fowir suggested that a more complex set ‘of conditions must be met before one could be sure before driling and testing a reservoir whether or not it was commercial. Its postulated here that the over- burden stress gradient is the other limiting variable Jn other words, a resistivity ratio limit and an over- burden stress gradient limit will determine the depths below which commercial production is not found. ‘The Theory ‘The log data and measured pressure data correlations discussed above show that there is definitely a rela- tionship between the two. Equations of these correla tions must be of the form: p/D = f (Normal R.s/Observed Ra), (2) or Al p/D = { (Observed At. ~ Normal Mia), 3) and Normal Ra _ 5 ‘Obsned Ry 7 F(/P)» oy and Observed 41.» ~ Normal Ata = f(p/D)- - (5) Now, if we incorporate the theories of Terzaghi and Peck,’ and Hubbert and Rubey,* whereby Eq. | is used, p/D is shown as follows Solve Eq, 1(S = 0 + p) for p and divide all quan- tities by depth, D, obtaining: p/D = s/D ~ 0/D © p/D = {(s/D, «/D) o If we combine Eq. 7 first with Eq, 4 and separately with Eq, 5, we find thatthe log parameters are func- tions of p/D, which in turn is a function of s/D and o/D. One empirical relationship was developed by trial- and-error fiting of data, and it prediets the abnormal pressure behavior data of Hottman and Johnson fair- Iy well The equation is p/D =s/d — 0535 (ene ge)” Normal Ray @) Eq, & reduces to the theoretical Eq, 6 when we assume that the overburden stress gradient equals 1.0 psi/ft, the resistivity ratio equals 1.0 (normal pressure), and the normal pore pressure gradient equals 0.465 psi/ft. p/D = 1.0 ~ 0.535 (1)* = 0.468 poi/ft 0 «/D is represented by the 0.535 term. ast Aei> 046s Aroses stace A srace srase ¢ Fig 1—Schematic representation of shale compaction ? wl [ele L NORMAL-PRESSURED gh OBSERVED Ryn Fig. 2—Shale resistivity parameter vs reservoir fluid ressure gradient RESERVOIR EQUIVALENT pre. "ae eat It is postulated here that p/D and s/D are the variables that control the log-derived groups. In other words, the parameters derived from log data are dependent variables primarily controlled by the existing pore pressure gradients and overburden stress gradients. If this is the case, correlations such as those developed by Hottman and Johnson should be ex- panded to include the effect of overburden stress ‘gradients. It could be argued that the overburden stress gradient is constant for a given area and there- fore of no significance. However, this is not true in areas where compaction and geopressures are caused by increasing overburden loads with deeper burial Overburden stress gradients are functions of burial depth in areas such as the Gulf Coast. ‘Overburden stresses and overburden stress grad- ients may be determined by any means whereby the bulk densities of sediments from the surface to total depth are measured, Overburden stress is related to bulk density by the following equation: S= rgd a0) Bulk densities are determined easily by the use of density logs. A cumulative averaging scheme* may be used 0 convert Jog bulk density data to curves of overburden stress gradient vs depth. Two such curves are shown in Figs. 5 and 6. Note that the overburden ‘gradient is not constant with depth. To develop the empirical relationship between the log resistivity parameter, the pore pressure gradient, and the overburden stress gradient, we need con- siderable data of the following form: (1) log resis- tivity data, (2) measured pore pressures, and (3) log buik density data, Similarly, we need sonic log data to develop the shale acoustic parameter correlation as a function of pore pressure gradients and overbur- den stress gradients. Such data are not available to us, but are thought to be available to several individuals within large oi companies. If such data are available and are plotted ‘on charts such as those shown in Figs. 7 and 8 we believe that the points will fall along lines signifying different overburden stress gradients similar to those shown, eT) yon im Fig. 3—Shate acoustic parameter vs reservoir fiuid Pressure gradient, AUGUST, 1972 pod iib Fig. 4—Shale resistivity parameter vs reservoir flld pressure gradient? ou iy compacted posite overburden stress gi for all ‘Gulf Coast formations ‘The lines shown in Figs. 7 and 8 should be nearly correct, but this can be determined only with experi~ mental data, However. two points on cach curve are fixed. One point is the end representing normial pres- sure gradients (see Eq. 9). The other end point of cach curve should be the extreme case in which the pore pressure gradient equals the overburden stress ‘gradient. In this latter extreme, the matrix stress ap- proaches zero, Therefore, the curve must approach the value of the overburden at some limiting log parameter value. The Relationship Between Commercial Production and Geopressure ‘Timko and Fert! state that their statistics show that commercial production is never found below the ddepth where the shale resistivity parameter (as shown in Fig. 7) exceeds 3.50. The Hottman and Johnson curve shows that the value 3.50 corresponds toa pore pressure gradient less than 0.90 psi/ft. Fowler,” how- ‘ver, lsted data for commercial reservoirs having orig- inal pressure gradients greater than 0.90 psi/f. ‘Close examination of Figs. 4 and 7 reveals that at the eutoll resistivity ratio value of 3.50, the following ‘cutoff gradients could be obtained 1, 0.88 psi/ft from the Hottman and Johnson 2, 0.93 psi/ft from the localized curve of Timko and Fert, 3. 0.92 from the curve marked s/D = 0.95, ‘4, 0.94 from the curve marked s/D = 1.00. ‘Therefore, itis concluded that such a set of curves could be developed that would accurately predict the best cutoff point. Iti believed that the pore pressure i i z Fig. 7—Shale resistivity parameter vs reservoir fluid ‘pressure gradient. gradient cutoff point will increase as the overburden stress increases. Several resistivity logs from Gulf Coast wells have been analyzed and plotted, Fig. 9 shows the most typical plot shape. Note that the resistivity continues to decrease with depth after the departure from nor- mal. Of more than 120 plots, 80 percent exhibited this type of profile. In these cases, a plot of the resis- tivity ratio parameter is nearly linear with depth as shown in Fig, 10, Therefore, with such a plot we should be able to predict the depth below which com- mercial production will not be found for a given area. ‘The result of such an effort would be the saving of ‘vast quantities of money spent in the fruitless search for hydrocarbons below certain depths. Production from Reservoirs of High Resistivity Ratio Many hydrocarbon-bearing reservoirs exhibit resis tivity ratios that exceed 3.50, However, none of these are known to be commercial. The obvious question is “Why are they not commercial reservoirs?” When we examine the environment of such reservoirs we will find that: 1. Pore pressure gradients approach the same values as the respective overburden gradients. 2. Rock matrix stresses approach zero and there is no “rock strength.” 3. Reservoir fluid pressures support most of the overburden load. ‘Recent experience with six such reservoirs shows the same very severe production problems. First of all, the production casing must have a collapse resistance at least equal to the entire overburden stress at the Fig. 9 —Short-normal og resistivity date, South Louisiane Fig. 10—Ratio plot of data from Fig. 9 933, producing depths, or it will collapse, After such a re- servoir is perforated, production is normally estab- lished at a high rate for a short time. Then the reser~ voir collapses around the wellbore. Plastic shale and sand are then extruded through the perforations, fill. ing the tubing string. Reservoirs of this sort are nor- mally abandoned only after repeated attempts have been made to wash out the tubing and re-establish production, The result is usually the same. The shale refills the tubing string immediately. For that reason such reservoirs are noncommercial regardless of the size of the hydrocarbon volume. Such has been the experience with reservoirs in the following areas: ‘One in Ship Shoal, La. — offshore, One in East Cameron, La. — offshore, One in West Cameron, La. — offshore, One in Terrebonne Parish, La. — onshore, Two in South Texas — onshore. No doubt there are many others that have exhibited a similar behavior, A theoretical analysis of such plas- tic deformation has recently been published by Gnirk.* His theory agrees perfectly with the observed plastic flow behavior just described. Conclusions This study has resulted in the presentation of several ideas that are stil in need of additional verification. However, we have drawn several conclusions, and we hope that they will be verified or denied by others, who have the necessary data. These are: 1, Correlations such as those indicated in Figs. 7 and 8 will more accurately predict the magnitude of abnormal pressure gradients. 2. The effect of the variable overburden stress gradient vs depth must be included in such correla tions 3. The overburden stress gradient may be deter- ‘mined from density logs. othe Epes ain Se rea pe 4, A correlation as suggested in Fig. 7 will more accurately establish reliable depths below which com- ‘mercial production cannot be found. Both 2 limiting resistivity ratio and a limiting overburden stress grad- ient must be established. 5. A plot such as that shown in Fig. 10 can be used advantageously as an exploration tool, 6. There is a need for a great deal more work in this area. Nomenclature acceleration of gravity, cm/sec* formation pressure, psi shale resistivity, ohmem overburden stress, psi sonic travel time, microsec/ft average bulk density of sediments, gm/ce ‘rock matrix stress, psi shale bulk References 1.Hottman, C. E. and Johnson, RK. “Estimation of Formation Pressures trom Lop! - J°Pet Teck: (une, 1968) T7722 2,Timko, D. J. and Fer, W. H. Hydrocarbon” Accumulaijon” and Geopressure_and ts Economie Sigifcance,” J. Per" Tech. (Aug, 1971) 923> 938 Terzaghi, Karl and Peck, RB. Soll Mechanics in En- Aineering Practice. John Wiley & Sons, Inc., New York C1988)" S66. Hibbert, M. King and Rubey, W. Wz “Role of Fluid Pressure’ jn Mechanics of Overthrast Faulting, Part Bull, GSA (Feb, 1988) 70. 5. Fowler, W. A, Je: Diseusion of Ref. 2 6.Eston, Ben A.: “Fracture Gradient Predietion and ts Application in Oilfeld Operations," J. Pet. Teck. (Oct, 1590) 1383-1360, 7.Eaton, Ben A.: “Problems Encountered Ip Drilling and per SPE 3662 presnied at Completing Deep Well” ‘SPE Eastern Regional Meeting, Charleston, W. a5, 1971 Griek, Paul F.: “The Mecha Wellbgrs Situaied in Elastic Plastic Media Under Hydeo- Static Stress" Soe. Per. Eng. J. (Feb, 1972) 49.59. SPT JOURNAL OF PETROLEUM TECHNOLOGY

You might also like