Está en la página 1de 6
The Drill-Stem Test: The Petroleum Industry’s Deep-Well Pumping Test by J. D. Bredehoeft® Abstract Drill-stem tests provi tion on three critical prope pressure head, permeability lc the petzoleum industry inform jes of subsurface formations — temistry ~that the in making pumping Tests of water wells, As it is increasingly necessary to and water pround-water hydrolopixe also seeks study the hydmulie and geochemical properties of deep- lying tocks in order to understand the behavior of ground water, data on drill-stem tests made by the petroleum indus ty become an important source of information which other is unobtainable because of the high cost. Data fom these = made by methods currently ia use ace highly ies. An obvious co h petcoleum engineering and ground-water hydrology seand to Profit substantially from hange of eas and techniques. m increase in the in Introduction As utilization of ground water and ground-water reservoirs increases the importance of having know!- edge on regional ground-water systems also increases. Understanding these systems often requires hydrologic data from great depths. In some areas such information may be available pumping ests made during the course of petrole tum exploration, The usual hydrologic test performed by the petroleum industry is the drill-stem test The oil industry developed the drill-stem test as a method of sampling the 4 tion during the course of deilling operations. modern drill-stem rests, however, yield chree types of (1) a sample of the formation fixid, (2) the undisturbed formation pressure, and (3) a co: efficient of permeability for the stratigraphic interval rested. from lid in a subsurface forma- Most hydrologic daca! Drill-Stem Testing During the drill-stem test the stratigraphic inter- val of incerest is isolated in the hole by the use of packers attached to the drill string and is allowed to yield Guid into the drilling pipe under the influence of the formation head. The arrangement of down-ole components in a typical drill-stem test stzingis shown in Figure 1 In the usual relatively shallow cest the drill pipe initially is completely empty and open to atmospheric pressure, By opening the tester valve in the vest string “Publication authorized by the Director, U. S. Goo. logical Survey BReseatch Geologist, U. S. Geological Survey, Water Resources Division, Washington, D. C Discussion open until September 1, 1965, DRILL PIPE AUXILIARY VALVE AND/OR CIRCULATING VALVE AIR CHAMBER GAUGE. (OPTIONAL) TESTER VALVE AND BY-PASS VALVE FLOW-STREAM GAUGE OPTIONAL TOOLS WARS, SAFETY JOINTS, ETC.) PACKER ASSEMBLY PERFORATED ANCHOR PRESSURE GAUGE Fig, 1. Main components of « typicol drillestem test string (alter vor Pesllen, 1960) (Eigure 1), the operator allows the formation to yield uid into the dell column for a relatively short period of time, The period usually ranges from 30 minutes to 2 hours. Following the production period che tester value is closed, shutting in the formation and causing the formation pressure to recover. Pressures are re corded throughour che rest by a bourdon-tube pressure recording gage contained within the drill string near the bottom: Two typical pressure graphs taken directly from the bottom-hole recorder are shown in Figure 2. The test can be subdivided into five phases, each of which is represented by a segment of the pressure graph: 31 <—$$ rs 2. Typical drillstom test pressure chorts (courtesy Halliburton Oil Well Cementing Co.) 1, The initial phase of the test, represented by segment AB on the lower pressure curve, involves lowering the drill string into place in the hole. The pressure gage records the increasing pressure of the drilling uid as the pipe is lowered into the hole. The “ appearance of the pressure curve is due Je procedure of adding joints of drill fo the stepw pipe used in making up the drill steing. Once the perforated section of pipe is in the desired position, the packer is expanded, of “set,” isolating the interval of interest, as well as the gage, from the columa of mud above the packer. The tester valve is momentacily opened (point B on the pressure ph), allowing the formation to produce for a few minutes reducing che pressure in the isolated zone 10 ric (point C), and then closed. The pressuse builds up again (segment CD of the curve) to some point (D) which asymprotically ap- proaches the undisturbed formation pressure. The initial shut-in period commonly ranges from 15 t0 45 very nearly atmosph 3, After the initial production and shut-in period ve is again opened and the formation fiuid ows into the drill pipe, Within che isolated interval the pressure reduces to essentially atmos pheric (point E on the pressure graph), but as the columa of produced fiuid builds up in the drill pipe, 32 the tester va the pressure increases (segment EF of raph). Commonly minutes to 2 hours pressure 's production period lasts from 30 4. After the production period the tester valve is again closed (point F), isolating che forn allowing the pressure to recover. During this final shut-in (cecovery) period the pressure gage records the pressure buildup in (Segment FG on the pressure curve). The final shut-in time is often made equal to the time of production ¢ isolated section of hole 5, After the final shut period the packer is unseated (point G), and the pressure returns to that produced by the columa of drilling mud in the hole (point H), the so-called by drill pipe is then pulled ous of the hole. The pressute- recording gage records the decrease in pressure as the pipe pressure curve) uid remains in the drill pipe until the pipe reaches the surface. This affords a sample of fluid produced from the formation tested, rement of the length of drill pipe filled with uid the volume produced can be calculated irostatic mud pressure. The removed from che hole (segment HJ of che The column of produced formation From the me A wide variation is possible in the combination of operations as well as in the components that make up the test string. Tests in recent years record both fan initial shue-in pressure and final shue-in as described above, whereas most older tes: without recording the inivial shut-in pressuce. Theoretical Analysis Mathematical analysis of the cest results is based upon the diffusion equation fiest applied co problems, of heat flow. Theis (1935) showed the usefuiness of this equation for analyzing fuid fow to a producing Muskar (1937) suggested the following form of the differential equation for petroleum prob: water well Tems which applies to horizontal radial flow through a anit thickness of the producing formation 1» fen a +2 a ae ae a where r= the distance from the center of the well (L) p= the formation pressure (F/L? f= the porosity of the producing formation = the compressibility of the fuid (J.*/F) j= the viscosity of the Guid (FT/L4) k= the p meability (1.") ‘The concept of storage as expressed in equation IL is somewhat different from the concept of storage used in ground-water theory. Storage is express: equation 1 by the factors / and ¢, which include oly, the compressibility of uid within the producing forma- tion, Muskat in a later publication (1949), following. both Theis (1935) and Jacob (1940), modified the con- cept to take into account not only the compressibility of the fuid bur also deformation of the aquifer skele- Homer (1951) suggested a method to analyze che pessure recovery during a drill-stem test-based upon the following solution to the basic equation te At ae 2354p ‘aakhs bog Q Pu = Po where y= pressure at the well bore (F/L?) Po = undisturbed formation pressure (F/L*) = the tate of produetion (L°/T) = viscosity of ehe fuid (FT/L*) = permeability of the producing formation (L*) b= thickness of formation being tested (1.) , = interval of time of produetion (T) At time elapsed since closing in the well afeer period of production (7) The analysis is equivalent to that suggested by Theis, (1935) for analysis of recovery tests in water wells. Homer's solution like Theis's, is based on che fact that with sufficient time che Fi function (the ex ponential integral) may be closely approximated by a logarithmic function. Horner (1951) states that the criteria for applying the equation is, 2575 fee oat k where ry = radius of the well, For che usval situation, the cheoretical error in using equation 2 would be very small in a matter of a few seconds to a few minutes. ‘The coefficient of transmissibility as the vermis, used in ground-water terminology is comparable to the terms bh/ in equation 2. The viscosity, which is neglected in the usual ground-water problem, becomes significane when considering many regional ground: water systems. For example, in the Big Horn Basia, Wyoming, the temperature of fuids within the Tensleep Sandstone varies from approximately 48° F near the outerop area to something in excess of 300° F in the deeper parts of the basin. This change in temperature produces a change in viscosity from approximately 1.4 centipoises to less than 0.2 centipoise, affecting the ‘ease with which fluid may move in the system by @ factor of approximately 7. ‘The practical solution to equation 2 is made by plotting log (fg + AD/At against Py. Pressure-versus- time data are taken from the pressuse chart (Figure 2) ‘As suggested above, the theory is developed to ana- lyze the recovery in the shut-in pressure following a period of production. An example of a plot of log (i, + AU/At versus py is shown in Figure 3, The points should fall on a straight line if the assumptions ‘on which the mathematical model is based are closely od ghaptenn Wein i 3 asco | : | en I "a3 a6 _ \ “a 3, Drill-stom test pressure buildup graph; versus log (io 4 AQ/A approximated in the field. A straight line is fitted trough the points on the graph (Figure 3). By substi tuting the change in pressure of the fitted straight line lover one log cycle in equation 2, the equation is simplified to Ab 2.34 o dnp where gq = the average production rate, and Ap = the change in pressure over one log cycle on the graph. This relationship is used to solve for the quantity bie. ‘The graph of pressure versus log te + AU/At (Figure 3) is also used to indicate the undistusbed formation pressure. As the shut-in time, At, becomes increasingly large the pressure approaches the undis- tucbed formation pressure, and the quantity (te + At/ Ar approaches 1. Extcapolation 0 this undisturbed formation pressure is made graphically by extending the Steed straight line to intersect the ordinate of (i, + NU/At = 1, The abscissa value of py at this point is taken to represent undisturbed formation pressure, py (Figure 3) The Effect of Drilling Mud Drilling mud is engineered to form an impervious layer along the wall of the hole. One of its principal functions is to ceduce the loss of fluids from the hole. ‘The so-called fileer-cake built up on the well-bore face may penetrate the formation for short distances, and thereby reduce the formation permeability around the borehole, This zone of reduced permeability af- fects the rate of pressure buildup in the drill-stem test in che same manner as a partly plugged screen in a water well. A typical drill-stem test illustrating the effect of the mud cake is shown in Figuee 4. As the time of pressure recovery increases, the effect of the reduced permeability zone near the well bore dimin- ishes and che points approach a straight line which is largely a function of the formation permeability beyond the mud-affected zone. The straight-line por 33 2109 Ea a Fig. & Type soffect of drlting mud damage. tion of the curve of pressure versus log (fy + Ag/At is thae of most ineerest io determining che transmissibil- ity of the formation. ‘The reduction of permeability by the mud, if large, may greatly reduce the quantity of fluid produced during @ short-term production test. The concept of “skin effect” (damage produced by the mud) was introduced by A. F. Van Eveedingen (1953) in order to evaluate the effect of the drilling mud in reducing production. He states (1953, p. 173): “The volume of the fluids contained in such a cylinder (a cylinder around the well-bore in which permeability is ceduced by mud invasion) is small ‘compared to the volume of Quids within che drainage area of the well. It may therefore be concluded that ‘aay tansient conditions set up in this cylinder are of ‘short duration and can be neglected in the analysis. Hence the effect of a reduction in permeability in this cylinder can be taken-into account as an additional pressure drop, proportional at all times to the rate of production from the formation. For this reason che additional pressure drop (per unit rate of flow) near the well bore is considered to be caused by a skin and denoted by S.”” Dolan and others (1957) introduced another ex- pression, the “damage ratio,"" which relates the quan- tity of uid that might have been produced if no mud damage had occurred to the quantity of uid actually produced: 1 = 4a (DR) where 4" = the theoretical average rate of production withoue mad damage. Both the skin effect and the damage ratio ase discussed theoretically in the following quotation from Dolan and others (1957, p. 323): “The equation for total pressure drop across the wellbore is MProtal (where § = skin effect) cy where B(x) =~ [= oF du xe which may be approximaced as Fi(—x) nx + .5772 if x <.01 “Damage ratio, D. R., is defined as the dimen- sionless ratio of &b/j to instantaneous P. I. (produc~ tivity index), where P.I.is taken to mean che measured ratio of production rate to differential pressure, PL1 = ay/lPe > Po (where 4, = average production rate during test; 2 — final flowing pressure) Ab J no be 195 “The D.R. is measurable from the flowing and shut-in curves on a DST chart, Replacing &h/jq with Inl0{4zNp) from the shut-in curve, the following equation is obtained, DR. = (.183) a 5 @ The problem of mud damage to the producing formation is very similac to the problem of well losses, in inefficient water wells. The productivity index de- fined above relates quantity of flow to pressure drop at the well and is analogous to che specific capacity yield per unit drawdown used by ground-water investi= gators. Units ‘The usual units used in drill-stem ceses are of course suited co the petroleum industry. Table 1 lists these units along with the corresponding units gener- ally used in ground-water studies. Petroleum produc- tion rates are usually expressed in barrels per day, permeability in millidarcies, and depth and formation thickness in feet. The kh/u terms in the equations, which are analogous to transmissibility, commonly have the dimension millidarcy-feet per centipoise, Using the petroleum-industry units reduces equa- tion 2 to: tt At ast pe te tog Pu = Bo 162.6 = tog ( ©) wy ‘TABLE 1. COMPARISON OF UNITS USED IN PETROLEUM INDUSTRY WITH THOSE NORMALLY USED IN GROUND-WATER STUDIES Peteolewn-Industry Unit Ground Water Units 1 barrel A2 gallons, 9,702 eubie inches 5.6146 cubie fect 1 datey 18.24 gallons per day per square foot (GE) 1 millidarey (11x10 darcy) 101824 gallons per day per square foot where ba and py = pounds per square inch 4g = bartels per day = centipoise k = millidarcies b feet te and At any convenient unit of time. Equation 3 in petroleum-induscry units is 4a bh/p = 1626 ‘e a © where Ap = pounds per square inch per log cyele. Solution of a Typical Problem Field data from a drill-stem test of the Tensleep ‘Sandstone within che Big Horn Basin are presented in Figure 4, Data from this test will be used to illustrate the calculations, ‘The test was made in the depth interval 5,838 to 5,851 feet, the recording gage being positioned at a epth of 5,845 feet, The formation was not shut ia initially; the tool was open for flow for 60 minutes and then closed in for a final shut-in period of 30 minutes. ‘The graphical plot of pressure versus log(f,+ AN)/ Aris used co determine fo, che undisturbed formation pressure, in this instance 2,072 psi, and Ap, the intercept of the curve per log cycle, is 53 psi. ‘This cese “recovered” 1,148 feet of fuid, of which approximately 95 percent was water and $ percent was drilling mud, The total quantity of fluid produced during a test is determined from the length of drilling pipe filled with uid, The string is usually composed of two different eypes of drill pipe: (1) drilling collars and (2) normal drill pipe. Drilling collars are simply ‘weighted sections of pipe used near the bottom of the drill string for addieional weight. The tocal quantity of fluid produced is calculated from the inside diameter for the capacity of both the drill pipe and the drill collars. For example, in this cest the deill string in- cluded only drill pipe. The quantity of fuid produced total Q = 1,148 x .01422 = 16.32 bbl. where 0.01422 = cay pet foot. The rate of flow varies throughout the duration of fa drill-stem vest as the head produced by the columa fof Suid within the drill pipe increases. The rate of flow at any time during the cest may be calculated from the flowing portion of the pressure-time graph (Figure 1), However, in the usual case only a small error results from assuming the rate of production to be constant and equal to the average sate of production (Dolan and others, 1957), In this example the average rate of production is: 16.32 x 1,440 oO ity of drill pipe used in barrels a= = 391.8 bbl/day Using the above information in equation 6 the transmissibility is calculated: 1.8 bb/y = 162.6( 2° ) 1,206 mittidacey-feet/ 53 centipoise Gq = 3918 bbl/day Ap = 33 psi The damage ratio may be evaluated from equation 2072 - 467 DR. = (183) ——— 33 = 5.6 where 467 = the final flowing pressure (pi). A damage ratio of 5.6 suggests that approximately 5 times as much fuid might have been produced if n0 skin damage had occurred, Computer Program Mathematical analysis of che data for a single test is nor difficult; however, a digital computer pro gram was developed to facilitate the analysis of a large number of tests, The program is based upon the mathematics given above, Mathematically, once a straight-line ft is select- ‘ed for the patt of the curve for pressure versus log (iy + AN/At that approaches the theoretical straight line, the solution of the problem is sttaigheforward "The computer method used to fit the straight-line por tion of the graph involves solution of ewo simultane ‘ous normal equations in which both the cransmissibili- ty &h/p and py are solved. The least-squares method is used to make a number of solutions ia which ini- tially all points are considered and then successive points are neglected until only the Jase four points remain for the final solution, Figure 5 graphically illustrates the method used. Figure 6 shows the change in the calculated py and Ap values as successively fewer data points are used to fit the regcession line. ‘As the data approach the theoretical straight-line portion of the (t + A‘)/Af plot, the calculated values should converge. Data from the two Tensleep tests 3 aF Fig. 5. Pressure buildup groph illustrating successive re- sted by the compute 35 Seo a So é Bl aaa : ; er ET ummm |b oh 4 E ee & som | Ecol rate en igen 3 eo me Fig. 6. Grophs illustrating the change in iy ond Ap values fas successively fewer points ore used to fit the regression Tine owach oF poivre seo (Figures 3 and 4) illusteate the tendency to converge; however, data from the Phosphoria test (Figure 5) does not. The period of ecovery for this Phosphoria test was not sufficiently long to approach the straight- line portion of the (lq + Ad)/At plot. This Phosphoria test was selected as a problematical test for the computer. As indicated above, the program does not select the most probable values of undisturbed formation pressure and tansmissibility; ie simply calculates the results in successive steps using fewer and fewer of the early data points. It is necessary to select the nost probable value by examining the convergence of the calculated undisturbed pressure values and tans- missibility values along with the computed statistical measures of how well the data ft the regression line lited by the mathematical model used. It is designed to solve the case of an infinite, fully penetrated by the costing device. A number of field conditions violate the mathe matic model. If the assumptions upon which the model is based are seriously violated, the computed results will fail co converge or will show unusual results, such as a significant change in slope, with time, in an upward direction, in which case the results must be evaluated more critically. As fewer data points are ‘The program is homogeneous aqui considered for the normal equations random errors in measurement become more and more significant. One first step in a more critical evaluation of the test would be to plot the pressure versus log (tp + AD/At raph, Limitations of the Drill-Stem Test The most obvious difficulty ia the drill-stem test is that of accurately recording bottom-hole pressures over the range involved. However, with the present bourdon-tube recording device, calibrated frequently, and a microscopic micrometer chart reader the gage error can be reduced to approximately * 1 to * 2 psi at pressures as high as 4,000 to 3,000 psi Difficulties involved arise from the fact that actu al field conditions only very approximacely satisfy the assumptions on which the mathematical model is based. Factors such as (1) the increased pressure produced in a formation by the colunn of mud in the hole during the course of drilling, (2)an imperfect seat of one of the packers, and (3) uncertainties in inter pretation due to the short duration of the test, partial penetration of the formation tested, etc., make quanti- tative interpretation increasingly difficult. Our experience with tests in the Big Hom Basin suggests that a reasonable estimate of the ettor i volved in obtaining the true undisturbed formation pressure is often on the order of 4 25 psi under the best of conditions Summary A drill-stem cest is a very useful hydrologic tool for the study of deep ground-water systems. In many instances, because of economic limitations, the drill- stem test is the only feasible method for performing a deep production test. As the scope of hydrologic studies increases, the dtill-stem test should be added to the cepertoire of possible tools for ground-water investigations. References Dolan, J. P., CA. Binaesen, and G. A. Hill. 1957. Special ‘pplication of deill-stem test pressure data, Am. last Mining Metall, Petroleum Engineers Teans, v. 210, p. 318324, Homer, B. R. 1951. Pressure build-up in wells. Proc. Third World Petroteum Congress, Section Il. E. J. Baill, Leiden, Holland. p. 503-521 Jacob, CE. 1940. On the Aow of water in an elastic arte sian aguifer. Am. Geophys. Union Trans, v. 21, p. 574 586. Muskat, Morris. 1937. Tae flow of homogeneous fds potous media. MeGeaw-Hill, New York, 763 p. Musknt, Notts. 1949. Physical principles of oil production MeGraw-Hill, New York. 922 p. ‘Theis, C. V. 1935. The celation between the Lowering of the Plezometsie surface and the rate and duration of dis- Charge of @ well using ground-water storage. Am. Geo phys, Union Teans. v.16, p. 519-525. Van Everdiogen, A. F. 1953." The skin effect and ins inf ach ence on the productive capacity of a well. Am. Inst Mining Metall, Pesroleum Engineers. Trans. v. 198, p 171-176, Ven Poollea, Hl. K. 1960, Status of drillostem testing tech niques aad analysis, in formation evaluation, Am. last. Mining Metall, Pewoleum Engineers, p. 1V-21-1V-38, 36

También podría gustarte

  • 1
    1
    Documento224 páginas
    1
    Juanito Gutierrez
    Aún no hay calificaciones
  • 02 3688
    02 3688
    Documento143 páginas
    02 3688
    Juanito Gutierrez
    Aún no hay calificaciones
  • Manualdelconductor PDF
    Manualdelconductor PDF
    Documento42 páginas
    Manualdelconductor PDF
    Juanito Gutierrez
    Aún no hay calificaciones
  • UBD en Cuenca Neuquina
    UBD en Cuenca Neuquina
    Documento80 páginas
    UBD en Cuenca Neuquina
    Fernanda Sanchez
    Aún no hay calificaciones
  • Falla Mandeyapecua
    Falla Mandeyapecua
    Documento6 páginas
    Falla Mandeyapecua
    Juanito Gutierrez
    100% (1)
  • 3 Alcoholismo
    3 Alcoholismo
    Documento8 páginas
    3 Alcoholismo
    Gustavo Fernandez
    Aún no hay calificaciones