OTC 4067 MOVEMENTS OF SUBMARINE PIPELINES CLOSE TO PLATFORMS by Andrew C. Palmer, University of Manchester; Michael T.S. Ling, Total Oil Marine Limited ©Copyrlght 1981 Offshore Technology Conference This paper was presented at the 13th Annual OTC in Houston, TX, May 4-7, 1981. The material is subject to correction by the author. Permission to copy Is restricted to an abstract of not more than 300 words. ABSTRACT and move towards the piatform: An analytical model of expansion movements at the ends of pipelines is developed. A comparison with measurements on two North Sea pipelines shows that the analysis is consistent with observed behaviour, and can be used to assess the results of corrective action. An alternative mechanism, that of creep deformation in corrosion coating, is analysed briefly. Alterations of pressure also cause movements. Close to the elbow, in the horizontal leg, the longitudinal stress is tensile, am the combination of circumferential and longitudinal stress induces a longitudinal tensile strain, and therefore a longitudinal movement. Far from the platform, on the other hand, longitudinal movement is prevented by friction on the bottom: there the strain is zero and the longitudinal stress is not the same as it is close to the elbow. INTRODUCTION Expansion due to changes in temperature and internal pressure can produce substantial movements at the ends of submarine pipelines l • At platforms, these movements are important beca~se they can overstress risers and elbows, and bring the pipe into contact with the platform itself. The paper begins by describing the mechanisms that give rise to expansion movements, and goes on to an analysis that predicts how much movement is to be expected. The results are compared with measurements on two North Sea pipelines. In a few instances, another mechanism may occur, and the movement may be due to creep deformation in the corrosi9n coating : this will be analysed briefly. MOVEMENTS AT THE END OFA PIPELINE Consider a straight submarine pipeline connected to a platform riser (Fig.la). The riser passes through clamps on the platform, and then has a 900 elbow. At a short distance from the platform, the pipeline reaches the bottom, and from then on is continuously in contact with it. It is helpful to begin by considering why the pipeline should tend to move. The operating temperature and pressure are higher than the temperature and pressure when the pipe was tied in. Because the temperature is higher, the pipeline tends to expand. Far from the platform, the expansion is constrained by friction between the pipeline and the sea bottom, am longitudinal expansion stresses are set up. At the platform, however, the pipeline is only slightly constrained (by the vertical leg of the riser, which is relatively flexible) , and there it can expand freely References and illustrations at end of paper 17 It follows that both temperature and pressure changes induce movements. At a distance from the pJ.atform, friction prevents these movements, but it does not do so close to the platform. The movements occur within a transition region whose length depends on the limiting frictional force between the bottom and the pipeline: if friction is large, the transitior region is short and the movements are small, but if friction is small the movements are larger. If the operating temperature and pressure are reduced, the movement towards the platform is reversed. only part of the original movement returns, and there remains a residual movement towards the platform, even if the pressure and temperature are returned to their tie-in values. This is because friction always opposes motion, so that when the temperature is reduced the frictional forces do not return to zero, but partially reverse, holding the pipeline in its extended position and preventing it from slipping back. ANALYSIS The idealizations used in the analysis are those customary in pipeline engineering, and the errors they introduce will almost always be negligible in practice. They are : (1) that the pipe remains elastic, am that its material properties are described by Young 1 s modulus E, Poisson's ratio v and linear thermal expansion coefficient ll. (2) that the pipe can be treated as a straight thinwalled circular tube of thickness t and mean radius R (defined as ~ (outside diameter - t». (3) that the limiting longitudinal force f per unit length, between the pipeline and the bottom, is uniform along the length, independent of the distance moved, and the same for either direction of motion. The length Z over which movanent occurs can be found .. f.rom the condition that cr L is continuous at z, and so, by equating the values of crL in equations (5) and (7), z is the solution of (4) that when the line was tied in, its temperature was the same as that of the sea water during subsequent operation, that its internal pressure was negligible, and there was no cold spring. 2 Z = (1TR p/f){1- 2v+ 2E:lt exp(-Z!A)} (S) that the force differences associated with the longitudinal pressure gradient are negligible over the length of pipeline that takes part in the movement (6) that the temperature of the line is not necessarily uniform, but can be represented by an exponential function of distance from the platform, so that 8 (x) = 8 exp (-x/A) • • • • • • • • • • (1) 1 where 8 (x) is the temperature difference between the pipeline and the water, at a distance x from the platform, 8 is the difference at the platform, and A is a decay length over which the temperature difference falls to lie (0.369) of its initial value. This assumed distribution corresponds to the steady state reached if fluid flows along the pipeline away from the platform at a uniform rate, and the overall heat transfer coefficient is independent of time and temperature. A negative value of A represents flow towards the platform, and a zero value represents uniform temperature. (7) that the shear force in the vertical riser leg is negligible by comparison with other forces in the system. and the change in circumferential stress to the pressure p·.bY • • • • • (3) vpR/t Ea8 vpR/t Eet8 l exp(-x/A) E:L '" du/dx • • • • • • • • • • • • • • • • (1.0) and can be determined by substituting (7) into (2) and then integrating (10)'. At the platform, the movement !::. is !::. = ~:E:L(X) dx a8 l A{l-exp (-z/A) )+~{ {~-V)pRz/t - f z 2/41TRt} (11) and if the temperature is uniform !::. = 1TRE (a8 l )2t / f { 1 + E:8l r'~-V)}2 •• (12) It should be noted that the temperature effect and the pressure effect interact in a nonlinear manner the total expansion movement is not the sum of the movement that would be induced by pressure alone and the movement that would be induced by temperature alone. 2 lI sd = ~y f/1TRtE (4) in x ~ Z If the temperature is uniform, this reduces to Z = (1TR2 p/f)(I - 2v + 2Ea8 l t/pR) • (9) The displacement u, positive away from the platform, is related to the longitudinal strain by I t{ Longitudinal movements are confined to a length z, the distance from the platform to the (imaginary) 'anchor point' beyond which no movement occurs. Beyond this, EL is zero, and so cr L (8) If the temperature and pressure are reduced, a segment of the pipeline moves away from the platform, and on that segment the frictional force acts towards the platform. If the temperature is uniform both before and after a temperature reduction fran a 1. to a 2 and the pressure is simultaneous1.y raiuced from PI to P2, analysis by the method described above shows that reversed movement occurs over a distance y, 1.ess than z, given by y = (~-v) (Pl-P2)1TR2 + Eet1TRt(8 l -8 2 )} (13) that at the platform em the reverse movement !::.sd is The longitudinal strain E: and stress cr , the circumferential stress cr H and Lthe temperatu~e rise 8 are related by the stress-strain-temperature relation 1. "'L = E(O"L - VO"H) + eta •••• • • • (2) cr H = pR/t • • • • • • • • • • • (14) and that the longitudinal stress is ~P2R/t + (5) The longitudinal stress between the platform and the anchor point is statical1.y determinate. Fig. 1.b shows the forces that act on a segment of the pipeline and its contents between section across the riser just above the elbow and a vertical section at a distance x from the platform; x is less than z. At the right-hand end, 21TRtcr L is the longitudinal force in the pipe wall, and 1TR2p the longitudinal force on the contents. At the section above the e1.bow, the only horizontal force is the shear force S, which is negligible. OVer the length x, the pipeline is moving towards the platform, and so the bottom exerts on the pipe a force f per unit length, directed away from the platform. Since the segment and its contents are in equilibrium, the resultant horizontal force on it must be zero, and so o = fx + 21TRtcr L - 1TR 2p (6) (7) crL = ~R/t - fx/21TRt in x < Z 18 fx/21TRt in x < Y cr L = ~lR/t - v (PrP2)R/t + Ea{8 r 8 2 ) - fx/21TRt iny<x<z vP2R/t - Ea8 2 in x > z (15) Fig. 2 shows the distribltions of longitudinal stress, strain and movement before and after a reduction in operating temperature and pressure. If the pipeline has been in operation, and the pressure and temperature are then reduced to their initial tie-in values before start-up, the reversed movement away from the platform is half the maximum movement towards the platform, and so half the maximum movement ranains. It follows that a residual movement after a complete shutdown does not necessarily imply creep in the corrosion coating. A much more complex situation occurs if the temperature distributions before and after the shutdown are non-uniform, so that the temperature falls from used to find the force needed to move it. In this instance f is taken as 1500 N/m (153 kg/m), which corresponds to a coefficient of 0.9 and a gas-filled suhnerged weight of 1670 N/m (170 kg/m), but in order to check the influence of the choice of coefficient a Second set of calculations was made with f equal to 1330 N/m. 6 l exp(-x/Al) to 62exp(-x/A2)' for instance. A segment of pipeline close to the platform then moves away from it, but another intermediate segment moves towards the platform, because of interaction between the nonuniform distribution of thermal strain and a reduction in axial compressive force that follows from the reversal of movement. Analysis then leads to coupled differential equations that have to be integrated step by step. However, in many practical instances of temperature and pressure reductions during operation, the length of pipeline influenced by reversed movement is small enough for the simple idealization of uniform temperature to be a reasonable one. The temperature decay length A = mc /g p It is sometimes necessary to reduce expansion movements by increasing the resistance to movement after the pipeline has gone into operation. This can be done by backfilling gravel or crushed roock over the pipeline. An important practical case is the following sequence first pressure PI' temperature 6l exP(-x/Al) then anchoring backfill is placed over the line, and increases the limiting resistance to longitudinal movement by F(x) per unit length then the pressure is increased to P3 and the temperature to 6 3exp(-x/A3) > 6l exp(-x/Al) On September 30, conditions were as follows : 2 pressure l3l bars (12.8 MN/m ) 38 0c temperature flow rate 28 Mcm/day (990 MMscf/d) + 2~RtE{(63exP(-v/A3 -6 exp (-v/A )} (16 1 l and the additional movement at the platform is t. ad = (~-v) (P3-P J.)RV/Et + a(63A3(J.-exp(-v/A3)-61Al (l-exp(-v/A - 27f~tE1: s: F(~) d~ dx where m is the mass flow rate, c p the gas specific heat and g the rate of heat transfer :from the pipeline, per unit length per unit temperature difference from the surrounding water : it is assumed that the thermal resistance between the gas and the steel pipe is small by comparison with that between the steel and the water The estimated value of g is 100 W/m degC. The decay length A is found to be several kIn, even at low flow rates. The combination of pressure, temperature and flow rate that would be expected to induce the maximum movement occurred on one of three days in the fall of 1979, and each will be examined in turn. After the anchor is installed, the length v over which further movement occurs is the solution of the equation ~:F (x)dx is determined by • • • • • • • • (18) l» • • • • • • (17) The sea temperature.is taken as SoC, and so the temperature rise 61 is 33 dege. Fig. 3 is a graph of calculated movement at the platform as a function of , for the alternative assumed values of f. The estimated value of A is 8.7 kIn. The calculated and observed movements are as follows : z .movement distance over which movement at platform occurs (m) em) calculated f f COMPARISON BETWEEN FIELD MEASUREMENTS AND THEORY The Frigg gas pipeline 1 at treatment platform TPl in the North Sea provides an unusually favourable opportunity for comparison between observation and analysis. The riser is in a vertical shaft inside the concrete platform; the shaft is usually dry, but is normally flooded in winter. After an elbow at the bottom of the riser, the pipeline passes through a seal and leaves the platform through a horizontal tunnel. Access to the elbow through the shaft permits accurate measurements of pipeline movement, without the need to rely on divers. 1500 N/m, A = 1330 N/m, A 8.7 kIn 8.7 kIn observed 1.084 1.192 3430 3780 1.035 The agreement between observed and calculated values is good. On October 22, conditions w:ere as follows pressure 133. bars (13.0 MN/m 2 ) temperature flow rate 39.5 °c 25 Mcm/day and the estimated value of A was 7.6 kIn. The calculated movement is 1.114 m for 1500 N/m longitudinal friction, and the movement observed was 0.970 m. It is not clear why the observed movement was 0.065 m less than on The analysis requires data on the platform pressure, the gas temperature, and the pipe properties, September 30, but the most likely explanation is that all readily available and SUbject to little uncertainty. the system temperature had not fully returned to its steady-state distribution after a brief shutdown on It also requires the limiting longitudinal frictional October 21.. force f per unit length, and the temperature decay length A. These two quantities are more difficult to On November 16, conditions were as follows estimate, and it was therefore desirable to investigate 2 pressure 144 bars (14.4 MN/m ) the sensitivity of the results to the assumed values. temperature 40.3 °c Since the pipeline was not buried, its limiting flow rate 30 Mcm/day frictional resistance could be estimated by multiplying and the estimated value of A was 9.5 kIn. The calculated its suhnerged weight by a longitudinal friction movement was 1.272 m, again for 1500 N/m longitudinal coefficient. Experience in bottom pull installation of friction, and the observed movement was 1.090 m. indicates that if a pipeline has been in place for some time, a coefficient of between 0.8 and 1 should be 19 The analytical model can be tested further by comparison between observed and calculated movements during a shutdown. A brief shutdown lasting 12 hours took place on November 14 1979, and the observed reverse movement was 0.230 m after 12 hours. The pressure drop was not measured, but was estimated to be 4.0 MN/m2 (41 bars) after 12 hours. An estimate of the rate of temperature drop was made by assuming the heat transfer rate to be 100 W/m degC, the same as when gas is flowing (which is reasonable, since most of the thermal resistance is associated with the concrete, rather than with heat transfer at the inside wall); the thermal capacity of the pipeline and the gas is estimated to be 1.32 MJ/m deg C. After the 12 hours, the estimated fall in temperature ~1-e2 is 33.7 degC, and so the temperature had fallen from its initial value of 400 C almost to the sea temperature of SoC. In calculating the longitudinal movement, it was assumed that f was 1500 N/m, and that the effect of variation of temperature with distance from the platform was negligible, a reasonable assumption since the distance over which reversed movement occurs is only 1500 m. The calculated reverse movement was 0.337 m. The agreement with the measured value is quite gbod, in the light of the sensitivity of the result to the changes of temperature and pressure, neither of which was measured directly. After an evaluation of the movements observed at the platform, it was decided to take action to reduce the movements that would follow future increases in operating temperature and pressure. In the early months of 1980, crushed rock backfill was placed on a number of sections of the pipeline close to the platform, as part of a wider program of span correction. The specified cover above the pipe is 1 m. Taking the rock particle specific gravity as 2.7, and the in-place voids ratio as 0.6 (porosity 0.38), the submerged unit weight is 10.3 kN/m3 (1050 kg/m 3 ). The estimated add~t~onaL Long~tud~naL res~stance F ~s L4.L kN/m ~f the pipeline moves through the rock (against the extra friction generated by the rock's weight) and 12.7 kN/m if the rock above the pipeline is carried along with it. However, since the calculation of F involves a number of uncertain factors (among them the state of stress in the rock above the pipeline, and the extent to which arching can transfer the weight of the rock above the line to the rock on either side), it was decided to adopt a lower value of 7 kN/m for design purposes. A further shutdown on March 22 made it possible to confirm the effectiveness of the backfill. At that time, rock had been placed over two sections, one of 464 m (from pk 360.136 to 360.600) and one of 93 m (from pk 359.623 to 359.716); the platform tunnel entrance is at pk 361.068. Fig'. 4 shows the movements and temperature and pressure changes that occurred. The observed movements can be compared with those calculated under three alternative assumptions, that the additional longitudinal resistance F generated by the backfill is 14 kN/m, that it is 7 kN/m, and that it is zero. The comparison confirms that the presence of the backfill does reduce the movements significantly, but the results are not sufficiently sensitive to the value of F for it to be possible to make an independent estimate. Another comparison can be made by calculating the 'forward' movements after restart at midnight on March 22/23. The calculated movements are 0.114 m if F is 14 kN/m, 0.118 m if F is 7 kN/m, and 0.163 m if F is zero, in the first 5 hours of operation, while the observed movement was 0.125 m. 20 Loeken l has described a second instance of submarine pipeline movement, in the 36-inch (914.4 rom) Ekofisk-Emden gas line at platform R. Stephens and Rawlins2 describe work on the creep movement of an unspecified 'pipeline E I, but a comparison between the papers strongly suggests that they are describing the same line as Loeken. The submerged weight is not available, but can be estimated at 1860 N/m (190 kg/m) in operation. The first 400 m from the platform are unburied, and so in this section the longitudinal resistance is 1680 N/m if the friction coefficient is 0.9. Beyond that there is 2 to 3 m cover of sand. The additional resistance generated by this cover is hard to estimate, but Loeken suggests a value of 44 kN/m for 1.8 m cover and a soil friction angle of 400 • Within the period covered by the reported movement data, the maximum movement away from the platform should correspond to mid-April 1978, when the maximum pressure was 11.7 MN/m 2 (1700 Ib/in 2 ) and the temperature reached 42 0 c (l08 0 F). The calculated end movement is plotted in Fig. 5, as a function of the uncertain longitudinal resistance F in the buried section. The observed movement is somewhat less than the calculated movement, unless F is as low as 10 kN/m (1 tonne/m; 700 lb/ft). The result is sensitive to the amount of cover between 400 and 800 m from the platform, and 10 kN/m may be a reasonable value, particularly if arching is significant or if the cover is less than intended. CREEP BETWEEN A PIPELINE AND ITS WEIGHT COATING Some instances of thermal movement may be due to deformation of bituminous corrosion coating, which may soften when the pipeline temperature rises as it goes into operation, and will then deform as a viscous fluid. That would allow the pipe to expand even though its concrete coating remained statiQIlary, by shear deformat~on w~th~ the corros~on coat~ng, and th~s mechanism is one explanation of movements which continue to increase after the pipeline is in operation It is important to make clear that it is not the only or most likely explanation of such movements : in most pipelines, the operating pressure, temperature and flow rate are progressively increased after start-up, and quite modest increases can produce substantial additional movements. A complete analysis of creep deformation is complicated, because the corrosion coating has a complex rheological behavior close to its softening temperature, may be non-Newtonian, and is strongly temperature-sensitive. However, a simplified model throws light on the factors that govern creep, and allows one to determine whether or not it might be important. Imagine a pipeline subject to thermal expansion alone, so that the pressure effect is negligible. The temperature increase is e(x,~) at time ~ and distance x from the platform. The displacement of the pipe itse tself is u{x,~), and the displacement of the concrete coating is v(x,~), so u-v is the relative movement between the pipe and the coating. The thickness of the corrosion coating is h, and its material is idealised as a linear viscous fluid with viscosity n at the pipe temperature. The earlier assumption that the pipe is thin-walled is extended to include the coating, so that the radius of the coating is identified with the mean radius R. Longitudinal forces in the concrete are assumed to be negligible by comparison with those in the steel. Fig.6 shows schematically the forces that act on the different parts of an element dx. At the u(x,t)= - 2a6 { (~) ~exp (_x 2/4K~) - ~erfc (x 2/4K~)~} left-hand end of the element, the thermal strain ~ 2 component is a6 and the total longitudinal strain is 1 - ~ u(x' ,0) exp -(x-x' ) 2 - exp -" (x+x') au/ax, and so the longitudinal stress is E(au/ax - a6), ----- } + -(~K~)~ 0 4K~ 4K~ from (2), and the longitudinal force is this stress multiplied by the wall cross-section 2~Rt. At the right dx' • (26) end, the longitudinal force is different, because au/ax and a6 will in general have different values. The where (27) K shear force on the outer surface of the pipe is the (Eth/n) ~ • • • • • • • • • product of the viscosity n, the velocity gradient (au/a~ - av/a~)/h and the area 2~Rdx. Since inertia At the end, the displacement u (0 , ~) at time ~ is terms are negligible, longitudinal equilibrium of the pipe element gives the governing differential equation -2a6 (X) u(O,~)= au .- a:r av (a 2u a6 ) a:r = (Eth/n) ~ - a ax 6 1 .~. • • • • • • • (20) for ~ > 0 which is an idealization of rapid start-up at a high flow rate. The rise in temperature will be followed by an immediate expansion, because of slip between the concrete coating and the bottom. The amount of movement is governed by the analysis described earlier, because relative movement between the pipe and the concrete cannot occur instantaneously, since that would imply an infinite velocity gradient in the corrosion coating. It follows that, immediate after the temperature rise, v(x,O+) = u(x,O+) = where - ITRt~e)2E(1-x/z)2/f for x < z (21) o for x > z z = 2~Rta6E/f • • • • (22) Once the instantaneous motion has taken place, viscous creep begins. Intuition suggests that creep will relax the longitudinal forces, and that the force between the concrete coating and the bottom will tend to fall rather than rise (at least in the zone that slipped initially). If this is so, we can conjecture that there will be no further movement of the concrete, so that for ~ > 0 = 0 • • • (23) and then, since 6 is uniform, the governing equation (19) becomes the diffusion equation au/a~ (Eth/n)a 2u/ax 2 • (24) (28) for ~ > 0 This analysis can be generalised to include pressure, and to allow for temperature gradients. It has unfortunately not been possible to compare it with a case of movement that is known to be due to viscous creep in the corrosion coating. CONCLUSIONS Observed movements of submarine"~ipelines are shown to be consistent with an elastic/frictional expansion analysis (which is well known in a simple form). The analysis correctly predicts reversed movement during shutdowns, and the response to backfill intended to increase longitudinal resistance. Other effects, such as the relaxation of 'snaking' and relative movement in the corrosion coating, may sometimes be significant, but it is not necessary to appeal to them to explain the movement of the Frigg pipeline. NOTATION E f F h p R t \l V X Y z a t:. The initial condition is (21), and the boundary condition at the end x equal to zero is given by the condition that the longitudinal stress at the end be zero, and is a6 s: (1_~)2exp{_ ~:~2} d~ where A is the initial displacement at the end, from (21). The first term in this expression becanes dominant as ~ increases, and corresponds to the continued expansion of the pipe through the coating, while the second term decays from A to zero, and represents the redistribution of the initial movement. A general solution is complex, because of the strong dependence of viscosity on temperature. It is useful to examine a simplified problem. Imagine that the temperature 6 is rapidly raised to 61' and then held uniform along the pipeline and constant with time, so that o for ~ < 0 • (~K~~Z2)'i • • • • • • (19) The group Eth/n has the dimensions of (length)2/time, and is a diffusivity, analogous to thermal diffusivity in heat transfer theory and coefficient of consolidation in soil mechanics. Stephens and Rawlins 2 derived the same group, in a rather different way. 6 (x,~) (K~/~)I:! { e: n K (25) The solution of (24) subject to these initial and boundary conditons is elementary, and is A 6 'V ~ (J 21 Young's modulus limiting longitudinal friction per unit length additional limiting longitudinal friction provided by backfill coating thickness pressure mean radius wall thickness displacement of pipe displacement of concrete coating distance from platform distance over which reversed movement occurs during shutdo"W11 distance over which expansion movement occurs linear thermal expansion coefficient movement at platform strain viscosity (diffusia:ity)I:! decay length for exponential distribution of temperature Poisson's ratio integration variable stress ACKNOWLEDGEMENT REFERENCES The authors thank Total Oil Marine Limited for permission to publish this paper. 1. Loeken, P.A. 'The "creep" on the Ekofisk-Emden 36" gas pipeline', Proceedings, 12th Annual Offshore Technology Conference, Houston, 1980, paper OTC 3783, 393-40l. 2. Stephens, H.G. and Rawlins, C.E. 'Axial movement of warm buried pipelines', Proceedings, Interpipe 1980, Houston, 146-161. a: CLAMP w (j) a: -.8 fx • x Fig. 1 - DEFINITION SKETCH (a) geometry (b) forces on a segment. b...J Cf) Cf) ill a: I- Cf) ...J W z :;;: 0: I- Cf) I- z y (13) (9) y z x x m b.Sd] ::2 ill o> ::2 y z DISTANCE FROM PLATFORM Fig. 2 - STRESS, STRAIN AND MOVEMENT AT THE END OF A PIPELINE Solid lines represent condition after a reduction in temperature and pressure, dashed lines condition before; numbers refer to equations in text. Fig. 3 - RELATION BETWEEN CALCULATED MOVEMENT AND TEMPERATURE DECAY LENGTH. • bars w 140 a: ::) C/) C/) °c • •• • •• • 0 0 • 0 120 w a:: a... <.,0-'<;. ~e -. • ·\e~ 0 El 100 1200 MARCH 22 • ~'\.e ?r eSsure_ • MIN MARCH 23 0 • 100 I- Z w :2: w 200 > 300 0 :2: mm Fig. 4 - MOVEMENTS DURING A SHUTDOWN Solid symbols represent measured values, open symbols calculated values. 0.6 I- z OBSERVED _ m w 0.4 w ~ CALCULATED > 0 0.2 :2: 10 20 30 40 50 kN/m LIMITING LONGITUD INAL RESISTANCE IN BURIED SECTION Fig. 5 - RELATION BETWEEN CALCULATED MOVEMENT AND LIMITING RESISTANCE. dx . . LJ----,,;::..p--'iP=-e_-l' + E(~~ + BdX - "S -' ,,~! dX) 21fRt r-=-=:::!~=-....;(n(h)(~~ - ~n 21fRdx - coating I c oncrete b ottom Fig. 6 - DEFINITION SKETCH FORCES ON AN ELEMENT OF PIPELINE. 30 w a: ::) l- 20 « a:: w a.. 10 :2: w I-