You are on page 1of 8
SPE 37443 Socely of Petroleum Engineers Multi-String Casing Design with Wellhead Movement ASS. Halal, SPE, Shell E&P Technology Co.. RF. Mitchell, SPE, and R.R. Wagner, Enertech Eng. & Research Co, cnr 07, Sey of etna Eg, {is pane was pad fr pesaraton st 6 3061 SPE Potion Opetont ‘Sym, fst Chana Cy one, 1 087 Abstract Multi-string axial interactions at points of joined tubular ‘movement such as the wellhead are necessary for accurate cetermination of axial loads and displacements. In conventional single-string stress analysis, a zero-displacement boundary condition is usually assumed at end poinis. In the actual well ‘completion, these boundary conditions are not fixed, but instead. represent connections between multiple strings or with the formation, which are all elastic to some degree. In addition, single-string predictions of burst and collapse loads generated by thermal expansion or contraction of trapped annular fluids can be exceedingly conservative, Ths is so because the radial interaction across uncemented intervals and the composite elasticity of cemented sections are not considered. This paper presents the required theoretical framework for addressing the mult-string problem, together with an efficient solution technique for solving the governing non-linear equations. Application of the technique to sample wells shows that while conventional design may be adequate for casing strings landed in tension, it could result in ‘substantial eror for free-standing structural casings. Introduction Muli-string stress analysis can be classified as axial, radial, ‘or coupled axial-radial according tothe well completion type. In onshore and platform wells, the readily accessible wellhead allows all annuli to be monitored and bled-off to relieve any annular heat-up pressures. However, the wellhead, downhole hangers, and packers represent points of joined tubular movement and the conventional single-string assumption of fixed ends may be inadequate for determining, ‘axial loads and displacements, This problem is considered 477 rmult-stsing axial one where the radial interaction is not of significant concern, However, this is not the case with subsea wells where limited access to the wellhead requires carefil consideration of annular heat-up and cool-down pressures during casing design’. Subsea high-pressure, high-temperature wells can ‘experience significant casing heat-up, not only during production, but also during testing and even during driling®. ‘Tubulars will elongate and trapped annular fluids will expand causing severe loads that must be considered for casing design. Burst and collapse are not the only concern. Casing. axial loads due to constrained thermal elongation together with “reverse ballooning” from high annular pressures can generate sufficient compression to relieve all hanging weight and cause upward forces at the mudline hanger. For simple subsea completions with uniform casing weights, good conductor cement bond, and packer set in a cemented interval, the problem may be considered a multi-string radial cone. This is so because the effect of the pressure induced change in axial stress (through Poisson's ratio) on the annulus volume change is small compared to that resulting from the corresponding hoop strain. However, for most practical applications, the interaction between axial stress and annular pressures may be significant which requires a simultaneous solution to the coupled axial and radial problems. ‘The methodology presented in this paper addresses the multi-string problem by decomposing it into two elastic problems: 1) the radial stresses and displacements, and 2) the axial stresses and displacements. These are solved successively and iterated until converged. ‘The radial stresses and cisplacements are determined through applications of Hook’s law and Lamé’s thick-wall cylinder equations’. For composite systems, such as cemented. concentric strings, boundary conditions of stress and displacement continuity are applied between different materials'®. For sealed annuli, fluid volume changes are obtained by numerical integration of fluid elements using constitutive PVT relations? so as to preserve variations in fluid compressibility and thermal expansion with temperature and pressure’. This is particularly important in the presence of annular gas The axial stress problem is Solved through use ofthe finite 2 AS. HALAL, RF. MITCHELL, RR. WAGNER SPE 37443 clement method", The principal difficulty presented by the multistring axial problem is the efficient simultaneous solution of several strings with coupled displacements Formulation Radial Model. A detailed presentation ofthe theoretical basis and solution method for the multi-string radial problem can be found in Reference 1. In brief, the formulation is based on three sub-models. The first sub-model addresses changes in ‘wellbore temperatures from the setting conditions for the load ‘ease in question, In the second sub-model, the net fluid volume change in each annulus, AV, is determined from 10 temperature and pressure changes as follows: AV (= A, Bpr/ (Pr Ad de - AVS, wo ‘where the integration is over the annulus fluid length, pris ‘the initial fluid density, A, is the initial annular cross-sectional arcaat depth z, Apis the change in fuid density from the inital ‘state to the final loaded state, and AV, is the bleed-off volume, ‘Note that Equation (1) is derived from a mass conservation principle and can be easily adapted to cases where annular gas or time-dependent mud deterioration might be of interest. In the third sub-model. the volume change of cach annulus, AV,, is determined from the radial displacements sing the following expression AV, = adr.’ + 21,Ar,) - (Ar + 24,ar de + AV-(2) ‘where is the annulus inside radius, is the annulus outside radius, Ar, and Ar. are the incremental radial displacements at r= r, and r,, respectively, and AV, is the volume change resulting from change in annulus axial dimensions. The incremental radial displacement, Ar is determined though: AT thee ° where Ace is the change in hoop strain given in terms of the corresponding changes in radial, hoop, and axial stresses, 46%, 80s, and Act, through Hook's stress-strain relationship for axisymmetric behavior Ace = [Ao - v(Ao, + Ao] E+ aT “ where E is Young's Modulus, v is Poisson's ratio, ois the coefficient of linear thermal expansion, and AT is the temperature change For uncemented intervals, Acs and Ac, are calculated from Lamé’s thick-wall cylinder equations’ and Ao, is determined from the axial model including the effects of buckling, friction, thermal stresses, ballooning, boundary displacements, et. For composite cemented casings, boundary conditions of stress and displacement continuity are applied between different materials to determine the radial displacement at the casing inside diameter (see Reference 1 for details) ‘The Secant iteration method is employed to converge the 478 ‘heat-up oF cool-down pressure in each annulus and the Gauss- Seidel method is used for mult-string radial equilibrium. ‘Axial Model. The major theoretical considerations for multi- siring axial calculations can be broken down into two categories: 1) tubular forces and displacements and 2) ‘boundary conditions and loads, and 3) solution technique. Tubular Forces and Displacements. Tubular forces are determined by pressures, temperatures, tubular weight, external mechanical forces, and fiction, The axial force ‘varies with depth due to the tubular weight and friction: FA@) = 0054 + EW, © ‘where F; isthe axial force (tensile force is positive, * denotes the derivative d/dz, w, is the tubular weight per foot in air, 6 is the angle of inclination of the wellbore with the vertical, . iis the friction coefficient, and W, is the contact force between tubulars. The friction is positive for incremental upward movement and negative for downward movement. The contact force depends on the buoyant weight plus the effect of buckling (in this development, z is measured from the surface). The frictional force is not casy to calculate because it depends on the load and displacement history of the pipe. Special cases of the friction force have been developed for landing pipe and for loading pipe with no previous history'* Using elastic relationships, Equation (5) can be expressed as TEA, (u’ - er - ey - &p)]' = wcos¢ + HWa © where u is the axial displacement, A, is the pipe cross-sectional area, 6 is the thermal strain equal to GAT, ey is the axial strain duc to hoop stresses equal to 27v(p- Pale WEA, prand pare ‘intemal and external pressures, respectively, and ep is the ‘buckling “strain” in the sense of Lubinski'* Equation (6) is solved using single-string stress analysis with a specified displacement boundary condition'*"*. Multi- sting axial interactions are represented by changes to Equation (6) from the initial state to the final loaded siate, namely TEA, (Au’-Aer-Aep-AewI'= + nAW. Ignoring frictional contributions to the multi-string effects yields: IEA, (Au' - Ae)]’ = @) where ¢ is the sum of the various strains. Equation (8) is solved numerically, using the Galerkin method”®, In the Galerkin method, often called the finite element method. the axial displacement u is approximated by the expression: u=Dofle) forzin [2,22] Oy where c; are coefficients to be determined and f(z) are called interpolation functions. The coefficients c, are usually chosen to have physical significance. In this case: ‘Au(z) = Auf + dus fe for zin [2221 (10) SPESTAS where: f= @2-2l22-21) (1a) = @-2Ne-n) «by Notice that Au(z) equals du, i =1,2. In general, Equation (1D, does not solve Equation (8) exactly. However, the coefficients can be chosen to minimize the error in the solution. A technique that integrates the error times a ‘weighing function aver the interval [21,22] and then chooses the coefficients 10 minimize this integrated error, is called a method of weighted residuals”. In the Galerkin method, the interpolation functions are chosen as the weighing functions, so the integrated error is: JIEA (Au - Ay feeydz=0 jn. Equation (12) can be integrated by parts to give Fa + SEA Au'f'dz = J EAAe f'ds “AF + JEA,Au'f’de = | EA,se fds (3b) This method generates a set of 2 algebraic equations in unknowns Au, and Au; AF. -EAJAz{Au; - Auy] = -EAAL aay (13a) (4a) “AF + EAJAz{Au2- Au] = EA,AL (db) where AL is the total length change due to strain Ae and Az = 7277), The two equations imply thatthe force change at point 1 equals the force change at point 2, that is, the element is a spring element with spring constant equal to EAJAz. Boundary Conditions. For tubing sealed in a packer, the change in tubing axial force atthe bottom of the string is: AF = OL) (A,~ Ap) = ApsL) (Ay Ap) + AF, (15) ‘where AF, is the change in the axial force, Ap(L) and Ap.(L) are the change in the inside and outside fluid pressures at L, respectively, A, is the packer bore area, AF, is the change in the packer to tubing force, L is the tubing string length, A, is the cross-sectional area of the tubing flow area, and A, is the total cross-sectional area of the tbe (Le, stee! and flow area). For tapered strings. there may be a change in the cross- sectional area at 2, where two different sized strings are Joined. There is a discontinuous change in the axial force at ‘this area change because of the fhuid pressures: AF, = Ap(Z.) AA, - Ap.tz,) AA, 3) ‘where Ap(Z.) and Ap,(Z.) are the change in the inside and ‘outside fluid pressures at 2., and BAH A= AL BAL= AD =A (17) with the superscript + indicating 2 > 2,, and the superscript - indicating 2 <2, (17a) 479 MULTI-STRING CASING DESIGN WITH WELLHEAD MOVEMENT 3 For an intermediate packer with free motion, there is a discontinuous change in axial force: AF,~ Ap, AA,- Ap) (Ay - Ac) + Apo'(Ap= Ac’) (18) and for a fixed displacement boundary condition at point i: 4u=0 as Solution Method ‘The first step in solving for incremental axial displacements is to identify the degrees of freedom (dof) of the problem. ‘Typically, these are axial motion at slips, hangers, packers, cemented intervals, and cross-sectional area changes. Parts of Sirings that are cemented to the formation are assumed t0 be fixed, so degrees of freedom are removed from the initial list, ‘when identified with fixed points Once the N degrees of freedom have been established, ‘equilibrium equations are written for each degree of freedom i, namely ZR) +F.=0 20) ‘The force Fis the force of the jth string at the ith degree of freedom, which is represented by stiffness equations in terms of displacement, such as equation (14), and the force Fy: is, due to boundary loads such as equations (15), (16), and (18). The result of assembling N equations of the form of Equation (20) is the system of equations Kyay=d, an For example, assume that the wellhead is dof 1 with three strings in the slips. The force exerted on the slips by each siring is of the form: string 1: Fy, = ky( Au; - Au) - kiAh (22a) string 2: Fyz = kx( Aus - Au; )- kpAl, (22b) string 3: Fis= ky( Auly~ Auy )~KyAls (220) ‘Also given is an axial force due to pressure increments F.. The resulting stiffness equations and loads re: Ky == (6 +h tk) (23a) Kirk, (23a) (3a) (23a) )= kydh; + Keds + hal = Fp (23a) ‘Once the fll set of equations is assembled, the inverse of K is ‘determined using Gaussian elimination with pivoting, giving ‘the displacements Au, at cach degree of freedom: Au,=K "yd; (ay Equations (24) are non-linear because of the implicit dependence of the d, on buckling and on annulus expansion 4 AS. HALAL, RF MITCHELL, RR, WAGNER SPE 37443 pressures. Dependence on annulus expansion pressures is through volume changes associated with Poisson’s ratio and displacements au. The coupled system is solved by the method of successive substitutions: 1. Given an initial state, calculate inital values of d, 2. Solve Equations (21) for the displacements 3. Calculate the heat-up or cool-down pressures from the rulti-string radial mode! 4, Calculate new length changes and forces from the single-string axial model. '. Calculate new values of d, using the latest values from steps 2 and 3. 6. Repeat steps 2-5 until convergence. For most applications, this system usually converges after Jess than 5 iterations since d, is a weak function of annulus expansion pressures and Au, Similarly, convergence in the radial model in step 3 (where the fluid volumetric behavior is ‘a non-linear function of pressure) and the axial model in step 4 (where the buckling strain is a non-linear function of axial stress) is reached in a few iterations. Details of the solution ‘method for the multi-string radial model can be found in Reference 1 and forthe single-tring axial model in Reference 16. ‘Sample Application ‘The offshore platform well example included in this section clearly shows the magnitude of error in an axial load calculation resulting from 2 conventional analysis that does not include multi-string interactions. This error is very substantial for the short, stiff surface casing A conventional analysis results in error because the following multi-string axial effects are typically omitted: 1. Wellhead movement due to landing subsequent strings 2. Wellhead movement due to thermal elongation during production. The example well has a mudline depth of 532 ft RKB. Fig. 1 shows a well schematic of this moderately deep, deviated well. The bottombole temperature at 15,354 ft TVD is 310°F. The braiden head is welded tothe free standing 16” surface casing and the base plate is welded to the 30” drive pipe for additional axial support. ‘After the braiden head is installed, the subsequent landing of the 10-3/4” protective casing, 7” production casing, and 3-1/2” tubing results in a cumulative downward displacement of the wellhead of 1.6". This, in turn, causes a redistribution of loads for each string at the wellhead (see Table 1). Note ‘that the difference in the initial load and the redistributed Joad becomes more significant for the outer, stiffer strings. ‘When they were cemented in place, each string was assumed to have a temperature distribution over its length identical tothe undisturbed temperature gradient. This is not strictly correct, but serves the purposes of this study. A temperature simulation was then performed to establish what temperatures would be seen by each string during gas 480 ‘Production at a rate of 6 mmcf/d. The simulation indicates a wellhead temperature of 147°F after two years of production (Gee Fig. 2). Radial heat transfer outward from the flow stream resulled in different temperature distributions for each string, The temperature distribution for the 10-3/4” protective casing is shown in Fig. 3. Note the cooling effect in the riser region ‘Based on the changes in the temperature distribution of cach string, the wellhead moves upward during production. ‘Table 2 summarizes the effect of mult-string loading and wellhead movement on stress analysis. In this case, the effects are negligible for the inner three strings, but represent a significant portion of the total loading for the 16” surface casing. Conclusions Theory and solution method have been developed for determining the impact of multi-string interactions on burst, collapse, and axial loads in casing design. An accompanying ‘computer model was also developed to study these effects ‘through application to sample wells. A summary of the main, conclusions follows 1. In platform wells, wellhead movement due to subsequent landing of inner strings and thermal elongation can significantly affect the loading of free-standing structural casings. Subsea wells in weak shallow formation can exhibit similar effect. 2. In subsea wells, substantial upward forces can occur at the wellhead during production as a consequence of annular heat-up pressures and constrained thermal elongation. 3. Compared to multi-string predictions, annular heat-up pressures. calculated from singlestring methods are always overpredicied (by as much as 30%) when using fully rigid outer casing, always underpredicted (by as much as 28°) when using flexible outer casing (ie., hydrostatic mud weight to shoe), and coverpredicied or underpredicted when using semi-flexble outer casing (ic, hydrostatic mud weight to cement top and rigid over ‘cemented section) depending on the cement top location. 5, For a fixed mud weight, predicted heat-up pressures associated with an oil based fluid could be higher than those of water based fluid, despite the lower compressibility of water. This behavior is a function of the density and concentration of the solid additives and can be readily observed from a fluid PVT analysis assuming incompressible solid particles. 6, In designing casing for cool-down loads (e.g. work- over or shut-in after production) in deep water wells, caution should be exercised since depletion of hydrostatic pressure at the wellhead is easily achievable. This problem may’ be further compounded by use of pressure relief materials such as syntactic foam” because of the increased annular volume provided by the crushed material. In such situations. a ‘moderate pressure buildup or depletion may be desirable for optimum design SPE 37403 MUCTLSTRING CASING DESIGN WITH WELLHEAD MOVEMENT s 7. This trade-off between pressure buildup and pressure depletion in trapped anmuli can be avoided by use of alternative operational practices (e.g. cementing below prior shoe, casing perforations, annular gas cap, etc.), mitigating wellhead hardware such as bleed-off valves, or wrapping materials that exhibit high elastic properties, low residual strain, and high thermal conductance Clearly, the effectiveness, economic feasibility, and risk associated with ch of these alternatives is site-dependent and a fit for ‘Purpose solution might be the best approach. Nomenclature A, = tubular total cross-sectional area, tubular inside flow area, L?, in? packer bore area, L?, in? AA, = change in total cross-sectional area, L?, in? thange in inside flow area, L?, in? -jth load for stiffness equations, mLit, tof ‘Young's modulus, m/L?’, psi ‘tubular axial force at bottom of string. tensile force is positive, mL/t,Ibt = force of th string at ith degree of freedom, mL, tof change in axial force at bottom of th section, mL, bf AAP = axial force discontinuity at z,, mL, Ib¢ AF, = change in packer to tubing force, mL/, lot f,~jth interpolation function clement of stiffness matrix, av’ Of stiffness matrix, nv, lb/ft stiffness of ith string, nV, Tot tubing string length, L, 1 = length of tubular element, L, ft ‘Al =ength change of tabular element due to thermal, hoop, and buckling strains, L, in, ‘P= internal pressure, nVLC, psi Po = external pressure, m/Lt’, psi 4p, = change in internal pressure, m/Lt’, psi 4p. = change in external pressure, m/LI’, psi += radial location, L, in = tubular inside radius, L, in 1, = tubular outside radius, L, in 47 = radial displacement, Lin ‘change in inside radius, Lin. ‘change in outside radius, Lin ‘AT = temperature change, T, °F 1 = axial displacement, L, in ‘incremental axial displacement for jth degree of freedom, L, in AV, = bleed-off volume, L’, bbl ‘AV; = change in annular fluid volume, L?, f° AV, = change in annular volume due to tubular axial displacement, L?, in? W, = contact force between tubulars, mL A, Ibf , = tubular weight per foot in ait, n/t t/t 12 = measured depth from the surface, L, ft 22. depth at cross-sectional area change, L, ft Ay = Zp1-F Lf ‘= coefficient of linear thermal expansion, IT. 1/°F ‘5 = buckling strain ‘64 ~axial strain due to radial pressures er ~ thermal strain ‘de= change in otal strain, dy + der + dep ‘deg = change in hoop strain > local wellbore angle of inclination with the vertical, degrees 1 fiction coefficient = Poisson's ratio ‘Ay annular fluid density, nvL*, ppg 4py = change in annular fluid density, nVL?, ppg, Ao, = change in radial stress, VL’, psi 4o, = change in axial stress, nL’, psi Ao = change in hoop stress, mL, psi Acknowledgment ‘We Thank Enertech Engineering and Research Co. and Shell EAP Technology Co. for permission to publish this work References 1. Halal, AS, and Mitchell, RF: “Casing Design for Trapped Annulas Pressure Buildup", Drilling and Completion Journal (Gune 1994), 2. Goodman, MA. and Halal, A'S: “Case Study: HPHT Casing Design Achieved with MultiString Analysis", paper SPE 26322 presented at the SPE Annual Technical Conference and Extibition, Houston, Texas (October 1993), 3. Halal, AS, Werling, DJ.,ané Wagner, R.R.: “Minimum Cost Casing Design”, paper SPE 36448. presented at the SPE ‘Annual Tectnical Conference and Exhibition, Denver, Colorado (October 1996). 4. MacEchran, A. and Adams, AJL: “Impact on Casing Design of hema Expansion of Fiuids in Confined Annuli”, paper SPEMADC 21911 presented atthe SPETADC Drilling Conference, ‘Amsterdam (March 1991). 5. Oudeman, P- and Bacarra, IJ: “Field Trial Results of Annular Pressure Behavior in a HP/T Well”, oper SPE 26738 preseed atthe OffShore European Conference, Aberin (September 1993) 6 Michell, RF. and Wedelich Tl, HF. “Prediction of Downhole Temperatures Can Be Key for Optimal Wellbore Design”, paper SPE 18900 presented at the SPE Production Operations Symposium, Oklahoma City (March 1989) 7. Timoshenko, SP. and Goodir, JN: Thory of Elasticity, MoGrase Hill Book Co,, New York (1970) 8. Mitchell, RF: “Dyramic SurgeSwab Pressue Predictions”, Drilling Engineering Journal Septersber 1988) 9, Sorelle, RR, Jandiolin, A, Buckley, P, and Barrios, IR “Mathematical Field Model Predicts Downhole Density Charge in ‘Static Drilling Funds”, paper SPE 11118 presented atthe 57th SPE ‘Annual Technical Conference and Exhibition, New Orleans (September 1982) 6 AS. HALAL, RF MITCHELL, RR. WAGNER 10. Meyer, CA, MiClinok, RB, Sides, G1, and Spencer Jr, RC: ASVE Steam Tables, Furth Eton (1979) 11. Kemp, NP. and Thomas, D.C: “Density Modeling for Pure and Miva Salt Brines as a Fonton of Composition, Temperate and Presse’ paper SPE 16079 pres atthe SPEAADC Diing Conference, New Orleans (March 1987) 12 Reid, RC, Prumitz, IM, and Poling B.E: The Properties of Gases and Liquids, McGraw Hill Book Co, Fourth Edition (1987) 13, Finlayson, BA: The Meta’ of Weighted Residuals and Variational Principles, Acaemic Press, New York (1972). 14 Mitchel, RF: “Simple Frchonal Anais of Helical Buckling of Tubing”, Drilling Engineering Journal Decerer 1986). 15. Labinsi, A, Althouse, WS, and Logan 1: ‘Seical Bucking of Tubing Sele in Packers" JPT ime 1962) 16, Mitchel, RE: “New Concepts fr Helical Buckling”, Dring Enginering Joural (September 1988). 17. Michel, RE: “Torque and Twist of Helially Buckled Casing”, ‘ASME Transaction, Journal of Enginering for Indusry, Vol 57, Series B No.2 (May 1974), 18, Davson,’R. and Pasay, PR: “Drilpipe Buckling in Inctined Hoes", PT (Cetaher 1984) 19. Leach CP. and Adams, AJ “A New Method for the Relie of Aula Heat Up Pressures" paper SPE 25497 SI Metric Conversion Factors op x 1o* E-03 = Pas Ax3048* — E-Ol=m A x 9.290 3044 fP x 2.831 685 in, «2.54% If 4.448 222 nd x 9,869 233 Psi 6.894.757 E+00= kPa *Comersion factor is exact 482 SPE 37443 SPE 374«3 MULTL-STRING CASING DESIGN WITH WELLHEAD MOVEMENT Incremental Total Distributed Loads: Pipe Description Initial Load ‘Wellhead Movement | Wellhead Movement | After Tubing Landed (tb) (in) (in) 30" Drive Pipe oO oO 16" Surface. 0 oO O 10-314" Proteclive 790,000 05 a5 7: Production 480.000. 08 HA 478,273 3-1/2" Tubing 112,440 02 1.6 112,363 Table 1 - Landing Conditions ‘2 Months Production: 0.8" Growth 2 Years Production: 0.9” Growth Wellhead Load (Ib) Welthead Load (Ib) Pipe Description |" wiMovement | wio Movement | % Error | _w/Movement | wio Movement | % Error 16" Surface =143,130 60,315 | 57.86 =140,867 -79,640 43.46, 10-814" Protediive 157,112 162,140 “3.20 113,791 718,160 “3.84 7" Production | __ 988,280 388,760 0.12 369,908 370,280 | -0.10_| 3/2" Tubing 94,661 94,571 0.10 ‘90,031 89,798 0.26 Table 2 - Axial Loads During production 19-374" a-as2" oat a-ara" Figure 4 - Well Schematic 483, 267. 3300. aise. a9aa. 3749. 7a38. 1140. 19800. 18943. et ft KOP ft, 4109. TUD ft, 4875. TUD ft, 5555. TUD ft, 6809. TUD fe ee fe, 25354. TuD AS. HALAL, RF. MITCHELL, R.R. WAGNER SPE 37443 7 eT 375 Figure 2 - Fluid Temperatures During Production = cs Tae ab. a ranprkivine, 3 Figure 3 - 10-3/4" Casing Temperatures During Production 484

You might also like