Você está na página 1de 7
Pore Pressure Estimation From Velocity Data: Accounting for Overpressure Mechanisms Besides Undercompaction Glenn L. Bowers,” Exxon Production Research Co. Summary ‘A new method for estimating pore pressure from formation sonic ve- locity data is presented. Unlike previous techniques, this method ae counts for excess pressure generated by both undercompaction, and fluid expansion mechanisms such as aquathermal pressuring, hydro carbon maturation, clay diagenesis, and charging from other 20ne. ‘The method issn elective stress approach: the effective stress is ‘computed from the velocity and the result is subtracted from the over burden stress to obtain pore pressure. To include multiple sources of overpressure. a pair of velocity-s-effectivestress relations are ‘introduced. One elation accounts for normal pressure and overpres- sure caused by undercompaction. The second is applied inside veloc- ity reversal zones caused by fluid expansion mechanisms. "Example applications ofthe method are presented from the U.S. gulf coast, the Gulfof Mexico, and the Central North Sea. Some oth- pore pressure estimation approaches ae also examined todemon- strate how these techniques have unknowingly accounted for over- pressure mechanisms other than undercompaction tis also explained how velocity-vs -effective-stress data can be used to identify the general cause of overpressure in an area, For instance, the empirical correlation of Hottman and Johnson indi- cates that overpressure along the U.S. gulf coast cannot be due only to undercompaction [Numerous methods have been developed for estimating pore Muid pressure from geophysical data and heist continues to grow. Empir- seal approaches equate departures from the normal trend line of some porosity-dependent measurement to an equivalent pore pressure gra- dit. Recent methods have followed the more fundamental effective stress approach pioneered by Foster and Whalea.! Ham,2 and Eaton. Al current pore pressure estimation methods filo formally take Jno account the cause of overpressure. It wll be demonstrated that this can lead to significant errors. Fora given velocity at @ given, ‘dept, the pore pressure can vary by 4 Ibmvgal or more, depending, ‘upon how the excess pressure was generated ‘This paper presents a method for estimating pore pressure from sonic velocity date that systematically accounts forthe cause of pressure. When appliedto wireline sonic log, itis preferable 10 nly, ‘se shale data to minimize the effets of lithology changes. Howey er, the method is also applicable for pre-dil predictions from seis mic interval velocities. “The paper begins with review of fundamental aspects of shale com= ‘ection behavior that frm the Foundation of this method. This is fol- lowed by a discusion of how different causes of overpressure affect the soni velocity. Some curent pore pressure estimation methods are ‘then examined in light ofthese concepts. The new method is then de ‘senbed, and example applications are presented and discussed, Stress States. Under increasing effee- tive pressure, sediments compact, and their sonic velocity goes up. {tthe limit their porosity approaches zero, and ther sonic velocity approaches the value forthe sediment grains. Bortowing terminolo- “Now mt cpa Mechs Teowolpes enya 105 ect Pate Ege ‘SPE marascrpt nce wn Man 19,1964 Ravens menace ces (Sadsi tw 9 ADSSPE Ditty Canewca had vOsies Pos 8 SPE Dring & Completion, une 1995 {ay from soil mechanics, the velocity-effective stress relation for ron-decreasing effective sess states willbe referred toasthe virgin ‘curve. Fig, a plots shale vigin curve data derived from well log, ‘and RFT meastrements from the Gulf of Mexico slope, and shows, ‘an estimate ofthe complete virgin curve. Effective Stress Reductions. Much (but not all) ofthe porosity los! velocity gain that oeurs during compaction is permanent. Asaresult the sonic velocity will not go down the virgin curve when the effective stress reduced (unloading). The velocity wil rack a diferent ast cer velocty-vs.-ffective-stess relation that wil be called the unload. ing curve. Ifthe effective stress is subsequently increased, the veloc ity will follow the unloading curve back othe virgin curve Fig. Ib illustrates unloading behavior with laboratory velocity- effective stress data for Coton Valley shale (Tosaya*), The velo ties measured at effective stresses below the maximum in-situ stress state must be on an unloading curve. For-edmparison, the virgin curve forthe Gulf of Mexico sediments is replotted in Fig. 16 Overpressure Causes/Eftects Normal Pressure. During burial under normal pressure conditions, the effective stess continually inereases with depth. Consequeatl, normal trend velociy-vs-effectve-stess data follow a virgin curve. Undercompaction. Overpressure most commonly occurs when pore fluid rapped by low permeability is squeezed by the weight of newly deposited sediments. This overpressuring process i referred tos undercompaction or compaction disequilibrium. ‘Undercompaction cannot cause the effective stress to decreas. the viggin curve also applies for formations overpres "undercompaction. The most undercompaction can do is. the effective stress in time, which would cause the velocity to become fixed on the virgin curve. On a velocty/depth pls, this, ‘would appear asa velocity plateau "Fig. 2 illustrates undercompaction overpressure in the Gulf of Mexico, Pore pressure, sonic Velocity, and stess information are shown in Figs. 2a, 2b, and 2c respectively. Fig. 2d plots velocity ‘sceffective-siress data determined at RFT locations. Ican be seen ‘that all of daa in Fig. 2d appear to lie ona virgin curve, and that the points below 7200 ft are approaching 2 fixed point Fluid Expansion, Overpressure can also be generated by fluid ex pansion mechanisms suchas heating. *® hydrocarbon maturation ? Charging from other 2anes¥ and expulsion/expansion of iniergran lar water during clay diagenesis °° Here, excess pressure results| from the rock matrix constraining the pore fluid as the fuid ties 0 increase in volume. Unlike undercompaction, ud expansion can cause the pore pressure to increase ata faser rate than the overburden sre. This forces the effective stesso deerease as burial continues, which pro duces a velocity reversal (sce Fig 3b). Velocities inside the reversal will rack an unloading curve(s), while velocities outside the rever- sal will remain ona Fig. 3 illustrates this with data from an Indonesian wel, Pore pressure velocity and sess data are displayed in Fig. 3a, 36. and 5c, respectively, while Fig. 3d compares velocity-vseffective- stress data from inside (open circles) and outside (solid cieles) he ‘velocity reversal, The sta of the reversal coincides with the top of ‘overpressure at approximately 6350 ft Itcanbe seen in Fig 3d thatthe velocities from inside the reversal track a much faster wend. This suggests that fluid expansion meche 8 == Srey = " a i “meat i co , , © ut yen © Tomy stn etaee Tegnaeacsc, So ga 6 8 10 12 4 16 18 Soa ae a oe 8 8 ecm a ence Ss » Fit 1 nlnding cure. om Pes) veto) vc) tac nn ovens \ 2 owen Ome " — so vcum oe ren ‘a i 1 10 ~ 5 @ m 3 o Fig, 2-Underompaction overpressure—l of Mexico. on Pi 0) eta) tomo) Etec Sm or TT? 3 : omen 2 : g ee i, ve TT , vod Sn w » Fig. 3—Fluid expansion overpressure offshore Indonesia. 0 ‘SPE Drilling & Completion. June 1995, ver 9915560 01517 pea Sora stpevoteses eT ‘active Sone (a) ‘tacove Sees (as) a) » Fig. 4—Fluld expansion overpressure, Gulf of Mexico. nisms contributed tothe overpressure inside the reversal. Note the similarity between the unloading curve in Fig. 3d and the trend fol- lowed by the Coton Valley shale data in Fig. 1b. Fig. 3d also demonstrates the importance of accounting forthe ceause of overpressure when estimating pore pressures. The viegin, curve would overestimate the effective stress at 6950 ftby 1700 ps, Which means the pore pressure would be underestimated by the ‘sume amount. This would correspond to 4.7 Ibmvgal error inthe equivalent mudsieight prediction Fig 4 present futher evidence of unloading. Fig 4a shows veloc- ity-efective stress data derived from RFT and well log measure- :ments along the Gulf of Mexico slope. The data are divided into wo ‘groups according tothe magnitude of overpressure, Solid circles are inthe 9 to 15 Ibm/gal equivalent mudweight range, while the open circles are in “hard” overpressure. Contrary to what might be ex: pected. the higher pressure data tend to have faster velocities. As Fig. 4b indicates, a possible explanation i that most ofthese data are from formations that have undergone some unloading. Similar trends are evident in the Central Nomh Sea velocity-vs-effective: stress data plotted in Fig, 8 thers have also reported evidence of unloading caused by high pressure. Magara! found thatthe Equivalent Depth method under predicted the pore pressure data of Hottman and Johnson.!? The equivalent depth method? equates the effective stress in an over pressured zone to that in a normally pressure interval withthe same velocity (ee Fig. 6). Tis assumes that the overpressure data are on virgin curv. If uid expansion has driven the data onto an unloaé: ingeurve, asin Fig, 3, the effective stess willbe overestimated, and the pore pressure will be underestimated, as Magara observed. ‘pong | ey y fala EEE oA Leroi fr. : Pare Peware ? von | 801300 Tire | o text000 1 29 4 6 67 8 Etecav Sess) Fig. 5—Fluld expansion overpressure, central North Sea. Plumey!® discussed unloading caused by fuid expansion over: pressure, and presented a US, gulf coast example of its occurrence. He compared porosities from an overpressured and normally pres sured interval having the same effective stresses. Te porosity inthe ‘overpressured zone was half te value in normal pressure (17.65 vs. 38%), Phumley concluded this was because the high pressure zone hhad undergone unloading. Berg and Habeck! came up witha simi Jar conclusion using density log data from a south Texas well ‘Velocity Reversal Without Unloading. Not all velocity reversals, are caused by fluid expansion mechanisms. The velocity will also drop across the transition from a normally pressured sundishale se {quence to massive undercompacted shale. If Muid expansion snot the cause, the veloity-vs-effective-stress data fom the reversal will follow a virgin curve Determining the Cause of Overpressure In Velocity Reversals. ‘A reliable way has not been found to determine from velocity data alone whether a reversal was eaused by undercompuction or uid ‘expansion. However, some general guidelines can be offered on the ‘geologic conditions that are conducive to each of these causes of Dressure (Miller and Luk®) ‘The amount of overpressure generated by undercompaction de- pends upon te relative compressibility of the rock matrix and the pore fluid, They act like wo springs in parallel I the rock matrix is much more compressible, increases in overburden stess willbe cared primarily by the pore fluid. Ifthe rock matrix is much less ‘compressible, then it will hear most of the overburden. Therefore, ‘undercompaction wil typically generate the greatest overpressure at shallower depths, where formations ae soft. a \, \ omen \ saat, deus (FT) = 18.0500 aimed 157 09 ofS or Paar a 6050 ‘new FT) = 162009, Emmate = 117500 ecety ee ms oo a a ee Some Fig. 6—Case where equivalent depth method works—Gult of Mexico. ‘SPE Dring & Completion June 1995, any ‘Sees 6 Fig. 7—Case where equivalent depth method fals—otfshor donesia. 1 ae Pen) tectve S88 © Pren Toe a » Fig. 8 U.S. gulf coast application of the Hottman & Johnson and equivalent depth methods. (On the other hand, the activity of many fuid expansion mecha- nisms increases with temperature, and therefore depth. To be a strong source of overpressure, fluid expansion also requires a fairly rigid, well-compacted rock matrix that can adequately constrain the pore fluid. Consequently, uid expansion is more likely tobe anim Portant source of overpressure at deeper depths in stife rocks ‘The only sure way to determine the cause of overpressue in a ve locity reversal is with measured pore pressures. One way i o plot velocity-vs.effective-stress data from inside and outside the rever- sal, a8 in Fig. 34. The reversal data wil track a much faster tend if fluid expansion mechanisms were active. ‘Another approach is to compare measured pore pressures with those computed with the equivalent depth method. The equivalent depth method will underesimate pore pressure caused by fluid ex pansion. Figs. 6 and 7 present cases where the equivalent depth method works and fails Cementation. Unloading may not be the only eason that velocity, reversal data deviate from the virgin curve: cementation could also be a factor. From the standpoint of pore pressure estimation this is, inconsequential, What counts is that the separate trend tracked by the reversal data be recognized and accounted for. However, cementation does complicate diagnosing the cause of ‘overpressure within a reversal. Even with petrographic analyses it ‘can be dificult to sort out the relative effets of unloading and ce ‘mentation, What can be saidis that cementation is conducive to fuid ‘expansion overpressure, because it increases the rock matin’s ‘constraint of the pore fluid. Consequently, while local geologic conditions must be considered, there is reason to believe that under: ‘compaction is generally not the sole source of overpressure when velocity reversal data diverge significantly from the virgin curve, Volos vs) Por Prssu Go) ms ms 6 8 eM we Current Pore Prossure Estimation Methods “Most pore pressure estimation methods claim only be applicable for overpressure caused by undercompaction, However, it tums out that many current pore pressure estimation methods are (unknow- ingly) predicting fluid expansion overpressure. To illustrate this, ‘wo popular techniques are examined ‘Hottman and Johnson. The Hottman and Johnson (H&) method empirically correlates departures from the velocity normal trend Tine to an equivalent pore pressure gradient.!? Empirical corela- tions have n0 inherent bias towards one particular overpressure ‘mechanism. They simply reflect whatever the dominant cause of ‘overpressure is in the area in which they were developed. For their USS. gulf coast corelation, H&J assumed this was undercompac: tion. IF this is true, then the Equivalent Depth and H&J methods should give similar results, As atest, both approaches were applied to shale sonic log data from HAs original paper! (Well °R°) Fig. 8a shows the velocities, while Fig 8b compares the estimated pore pressures with bottombole pressure measurements Itcan be seen thatthe equivalent depth method underpredicts the pressure, while the H&] correlation performs wel. This has two pos- sible implications. First because the Equivalent Depth method filed, this suggests that uid expansion isan important source of overpres- sure. Second, because the H&) method worked, H&s's Gulf Coast correlation appears 10 have fluid expansion “bull” into i ‘As further evidence, Fig. 8e plots velocity-vs-effective-stess ata from the well. Solid circles are from the normal trend line. Ea ton's US. gulf coast overburden stress curve!® and a normal pres sure gradient of 0.465 psifi were used to estimate effective stresses at these points. Open circles are from the three pressure tests inside hace Sess) © Passe Ton Original Eaton Equation o= Gem GY veces Revised Eaton Equation ‘ remare a = Srou( yt Se) cs) >» °) Fig, 8-U.S. gulf coast application of the Eaton and equivalent depth methods. 2 SPE Driling & Completion une 1998 the velocity reversal. The faster trend tracked by the reversal data is characteristic of unloading. “The Solid line in Fig. Be is the velocty-vs-effetive-stress path efined by HEIs pore pressure estimates inside the reversal. This ‘curve clearly deviates from the normal pressure data, which again suggest the H&t} Gulf Coast correlation is biased towards id ex- pansion overpressure. This also means that this correlation will overestimate the pore pressure at wells where undercompactiontu- Iy is the dominant cause of overpressure, as in Fig. 2. Eaton, Original Eaton Method. Eaton's method! is an effective sess approach, with the elective stress, 0, computed from the equation: o ‘and Onorm and Vaan ar the values the effective stess and velocity should have under normal pressure conditions. Eq | implies that normally pressured and overpressured formations both follow a virgin curve relation of the form: vecan, cocteee sees 2 Consequently, Eaton's method should underestimate uid expan- sion overpressure. ‘The velocity normal tend line is usually assumed to be a straight line on a pit of og( gor) ¥S. depth, However, according to Ea, 1, ‘and Yom should both satisfy Eq. 2. Therefore, to be consistent, Yom should actually be calculated from Ea. 2 using Oper Values determined from the overburden stress and normal pore pressure profiles. his were done, the normal rend line would norbe a semi log straight line. Fig. 9a compares H&s's semi-log norma trend for Well R (dot- dashed curve) with one analytically computed from Ea, 2 (sol Tine). The parameter C= 564 was obtained by fitting Eq 2 through the normal trend velocity-effective stress data in Fig, 9e, It can be seen that the semi-log normal tend line is faster than the analytical solution below the ip of overpressure. By overestimating Vponn the semi-log normal trend will make Eg. 1 predict lower effeciive stresses, and therefore higher pore pressures than Eq, 2. This will, ‘generally be true. However, unless unreasonably large values are as- Sumed for Varn, Ea. 1 wil still underestimate uid expansion over Pressure ‘This is the case with the Eaton pore pressure estimates for Well R, which are ploted a a sold line in Fig. 9, The dotdashed line isthe equivalent depth solution ean be sen that both methods fall well short of the measured pressures inthe velocity reversal. The ve- locity-effectve stress data in Fig. 9e explain why, Even with the semi-log normal trend line, the Eaton solution inside the reversal ‘mains close to the virgin curve defined by Eq. 2. Vototy vaweek [autor aa) I Fig. 10—The unloading parameter “U." ‘SPE Dring & Completion June 1995, Modified Eaton Method. Ifthe Baton pore pressure estimates ae too low, Eq. I can be adjusted 10 yield lower effective stresses. One ‘way it increase Vag By shifting and rotting the normal tend line (Weakley"®), Another, simpler alternative, isto raise the exponent in Eq. Ether way the net effect isthe same: it allows the original ton virgin curve relation to be transformed into an unloading cutve. For example, by raising the exponent from 30 5 the revised Ea- ton solution (dotted lie in Fig. 96) sable to closely match the Well R pressure data. As Fig. 9c shows, this is because the higher expo- ‘ent has allowed Eaton's equation to simulate an unloading curve New Method Overview. The new methods an effective stress approach that er ploys virgin and unloading curve relations to account for both un: ‘dercompaction and fluid expansion overpressure. Effective stresses outside of velocity reversal zones are computed from the virgin curve. Inside a velocity reversal, offset wel data are used to decide ‘hich equation is appropriate, For rank wildcats, the pore pressure can be computed bath ways fo establish lower and upper bounds on the pressure. The unloading curve will always yield higher pore pressure estimates Virgin Curve. Over stress ranges of practical interest, it has been found that the virgin curve for shale, as shown in Fig, la, canbe ade: ‘quately represented by the following equation: v=5000+ Ao see @ where vis velocity (PU), cis effective stress (psi), and A and B are ‘parameters calibrated with offset velocity-vs-effective-stress data Unloading Curve. The unloading curves defined by the empirical relation ¥= $000 + AlOmax (O/0 pmax) VU) a vau~$000) nn ® eek oe acct es ° + ‘onoeang oe HH (nn) * Fa ae =e Jo |e a zy Vere Mel vaety ar (nae) Fig. 11Normalized unloading curve, 2 \ ant | “7 . Ef El ge og i ~ 4 as : emntees, 4) q . — © acter Fig. 12-U.S. gutt coast application of new method. ‘The unloading parameter U is a measure of how plastic the sedi- ment is (ee Fig. 10). U=1 implies no permanent deformation, be- ‘cause the unloading curve reduces tothe virgin curve. U= & comre- ‘ponds tocompletely reversible deformation since V= Vin fo all Values of effective stress less than Ogu. In practice, U typically ranges between 3 and 8 ‘Solving for U. While virgin curve data track a single curve, unload- ing data from multiple wells wil generally ie on multiple unloading curves (see Fig. 5). However, by substituting Eq. 3 into Eq, 4, the unloading curve can be recat into & form that normalizes multiple ‘well unloading data onto a single curve: (034) = (Dye!Feraxd”, 6 ose = (500) co Asthe inser in Fig, 1 illustrates, esthe stress at which thecur- rent velocity v intersects the vigin curve. Fig. 1] shows the normal= ied version of the unloading data in Fig. 5 Example Applications US. Gulf Coast. Fig. 12 shows an application ofthe new method to Hottman and Johnson's Well R. The virgin curve parameters (A= 4.4567, B=0.8168) were determined by fitting velocty-vs. effective-stress data from the normal pressure interval above 10000 ft(see Fig. 122). A normal pressure gradient of 0.465 pst and Ea ton's US. gulf coast overburden stress curve were sed to estimate effective stresses along the normal trend. The normal tend Hine in Pore pressures inside the velocity reversal were computed from the unloading curve relation, with U= 3.13, This isa regional un loading parameter determined from U.S. gulf coast and Gulf of Mexico data. Fig. 12 compares the pressure estimates with the ‘measured values. The accuracy achieved is similar to that for the Hottman & Johnson U.S. gulf coast correlation, Fig. 12c compares the virgin curve and unloading curve relations with velocity-effectve stress data from inside and outside the rever sal. Also shown is the velocty-vs.effective-sress path defined by Hs pore pressure estimates inside the reversal. As can be seen, the new method's unloading curveis very close to Hs stress path, ‘Their divergence is due tothe Hottman and Johnson normal trend Tine exponentially increasing with depth Deepwater Gulf of Mexico. The next exampleis froma well died in nearly 1400 fof water inthe Gulf of Mexico. Fig. 13a plots the sonic log data, and a normal trend Tine analytically computed from Eq.3. There are a numberof small velocity reversals above 9700, ‘and one major reversal between 9700 ft and TD. Because al of the shallower reversals are very weak, they were not considered to be ‘duc to fluid expansion Within the large reversal, the velocity at fist drops at arate similar tothatinthe smaller reversals, However, at 10200 the slope steep ens significantly. This slope break was interpreted tobe the onset of fluid expansion overpressure. Therefore, inthe unloading relation, x Was assumed tobe the velocity at 10200, not the peak veloc ity a 9700 f. Where the velocities near TD are above the value at 10200 ft, the virgin curve was used to compute effective stresses, ‘Asin the USS. gulf coast example, value of 3.13 was assumed {forthe unloading parameter U. Regional virgin curve parameters,

Você também pode gostar