ATOMIC ENERGY WffR L`ÉNERGIE ATOMIQUE OF CANADA
Transcription
ATOMIC ENERGY WffR L`ÉNERGIE ATOMIQUE OF CANADA
AECL-6476 ATOMIC ENERGY OF CANADA UMITED WffR U B f L'ÉNERGIE ATOMIQUE DU CANADA LIMITÉE ELASTIC STRESSES IN U-SHAPED BELLOWS Contraintes élastiques dans les soufflets en forme de U P. JANZEN Chalk River Nuclear Laboratories Laboratoires nucléaires de Chalk River Chalk River, Ontario May 1980 mai ATOMIC ENERGY OF CANADA LIMITED ELASTIC STRESSES IN U-SHAPED BELLOWS by P. Janzen Chalk River Nuclear Laboratories Chalk River, Ontario KOJ 1JO 1980 May AECL-6476 L'Energie Atomique du Canada, Limitée Contraintes élastiques dans l e s s o u f f l e t s en forme de II par P. Janzen Résumé Ce rapport présente des r e l a t i o n s décrivant les niveaux de contraintes élastiques méridionales et c i r c o n f é r e n t i e l l e s à l'embase et à l a t ê t i è r e à cause de l a pression externe et de l a f l e x i o n a x i a l e des s o u f f l e t s en forme de U. La d é r i v a t i o n est basée sur l ' a n a l y s e s t a t i s t i q u e de données théoriques obtenues à p a r t i r d'une analyse des éléments f i n i s de c o n f i g u r a t i o n s choisies de s o u f f l e t . Les formules mathématiques et diverses représentations graphiques sont proposées comme aides pour l a conception et l ' a n a l y s e des s o u f f l e t s . Laboratoires nucléaires de Chalk River Chalk River, Ontario KOJ 1J0 Mai 1980 AECL-6476 ATOMIC ENERGY OF CANADA LIMITED ELASTIC STRESSES IN U-SHAPED BELLOWS by P. Janzen ABSTRACT This report presents relations describing the meridional and circumferential elastic stress levels at the root and crown due to external pressure and axial deflection of U-shaped bellows. The derivation is based on a statistical analysis of theoretical data obtained from a finite element analysis of selected bellows configurations. The mathematical formulations and various graphical representations are proposed as aids to bellows design and analysis. Chalk River Nuclear Laboratories Chalk River, Ontario KOJ 1J0 1980 May AECL-6476 iii TABLE OF CONTENTS page 1. INTRODUCTION 1 2. PERSPECTIVE 1 3. REFERENCE CONVOLUTION SHAPE 4 4. STRESS ANALYSIS 4 5. APPROACH TO STATISTICAL STUDY 6 6. REGRESSION ANALYSIS 7 7. RESULTS AND DISCUSSION 11 8. CONCLUSION 12 9. ACKNOWLEDGEMENT 13 10. REFERENCES 13 TABLES FIGURES iv LIST OF SYMBOLS e, exp exponential function (base of Napierian d Z nc n p logarithmic system) convolution depth bellows' active (convoluted) length bellows' number of convolutions bellows' number of material plies external pressure r convolution crown torus radius c r. r = r t E F K S S. S °S S, = _ S = m Y C r. + d/2 os " S is j S + S. os is T z bellows convolution inner radius bellows convolution mean radius convolution root torus radius material ply thickness material Young's modulus stress function correction factor bellows axial spring rate stress function inner surface tangential stress outer surface tangential stress S i — r i YF = t ~± a., i • 0,1,••',5 3.» i = 0,1,»••,4 bending stress component . membrane stress component non-dimensional configuration parameters exponents of configuration parameters exponents of non-dimensional configuration parameters ô axial deflection per convolution r À. = — m radius ratio X- = —p- thickness ratio y = /l2(l-v 2 )A 1 A 2 shape parameter v material Poisson's ratio <)> angle measured in the meridional plane A = n c<5 total bellows axial deflection ELASTIC STRESSES IN U-SHAPED BELLOWS 1. INTRODUCTION Although bellows have numerous applications, frcm large,low-pressure expansion joints in air ducts and pipelines to small, high-pressure bellows stem seals in valves, little information on design or analysis of bellows can be found in published literature. The information which is available is based on approximate solutions which frequently result in unacceptable discrepancies. This report derives equations describing the spring constant, and the meridional and circumferential elastic stress levels at the root and crown of Ushaped bellows due to pressurization and axial deflection. The derivation is based on a statistical analysis of theoretical data obtained from a finite element analysis of selected bellows configurations. These mathematical formulations and various graphical representations are proposed as aids to bellows design and analysis. 2. PERSPECTIVE The analysis of bellows to determine its response to various applied loads has received attention over many years. Selected contributions will be mentioned to place this subject in perspective. Theoretical treatments of this problem have exploited two basic mathematical techniques designated "approximate solutions" and "asymptotic solutions" (1) . Each approach has a limited range of validity related to some geometric parameter. Parameters recurring in published literature include a "radius ratio" \± = rc/rm, a "thickness ratio" \2 = rc/t and a combined shape parameter /12(1-V2) A ^ = À2(l-v 2 )r c 2 /r m t. Figure 1 defines the geometric variables and r m + d/2. - 2 - The 'approximate solution" employs a truncated infinite series expansion of the governing variables to solve the exact theory equations. Its applicability is limited to small values of X^(Al<0.1) and an approximate range of A 2 from 5 to 3 5 For large values of A2(^2>:1-5), the differential equations of the exact theory can be reduced to permit a closed form solution in terms of higher transcendental functions. This approach yields the "asymptotic solution". R.A. Clark has published asymptotic solutions for an "Omega" piping expansion joint subjected to axial load, a corrugated pipe subjected to axial displacement and internal pressure (2), and a U-shaped bellows subjected to axial load (3) . The same case of a complete torus slit at the inner edge and welded to a stiff cylindrical pipe subjected to an axial load was also considered by N.C. Dahl (4). However, he developed an approximate solution utilizing the principle of minimum complementary energy and a four-term series to approximate the change in the meridional tangent angles. Over a range of values of p his solution corresponds to that of R.A. Clark. Beyond u = 30 the solutions diverged, with the asymptotic solution possessing greater accuracy. U-shaped expansion joints have been the subject of more recent papers. A. Laupa and N.A. Weil (1) used an energy method for the toroidal sections, and the theory of symmetrical bending of circular plates augmented by the thick-walled cylinder analysis for the annular plate connecting the two toroidal sections. Their general solution permitted different radii to be assigned to the inner and outer toroidal sections and independent changes in the average thickness of the toroidal sections and the interconnecting annular plate. Axial and pressure loading were both considered. T. Ota, M. Hamada, T. Takezono, Y. Inove, T. Nakatani and M. Moriishi (5,6,7,8,9) extended this analysis slightly and included design charts to illustrate the behaviour of the maximum meridional bending and circumferential membrane stresses as functions of geometric parameters. While manufacturers have had some design formulas at their disposal for some time, either of a proprietary or published nature (10), the proliferation of thin-ply - 3 - formed bellows, especially in finite life applications, has revealed a need for more exact design criteria. A number of reports have been issued in an attempt to correct this deficiency. R. A. Winborne (11) and W.F. Anderson (12) have presented design procedures based on asymptotic shell theory and charts for "curvature corrections". A more comprehensive report by T.M. Trainer (13) summarized a three-year project whose objective was "to establish design procedures, stress analysis methods, and other factors essential to the successful design and fabrication of metallic bellows". Theoretical results of the bellows response to axial and pressure loads were obtained utilizing a multi-segment numerical-integration technique. A computer program, capable of linear elastic axisymmetric and non-symmetric deformation analysis and of non-linear axisymmetric deformation analysis was developed. Parameters pertinent to bellows design and fabrication were identified and some parametric curves to permit convolution shape optimization were presented. Although the report was extensive, its usefulness as a practical design aid is limited. No mathematical design formulas are presented and the parametric curves are not given in sufficient detail to aid in design. Many of the above publications, in addition to comparing theoretical results with previously published results, also reported some experimental results in support of their analysis. One of the more enlightening papers on experimental results was prepared by C E . Turner (14). He utilized an approximate solution for the range of y<5 as well as a numerical solution for an extended range of JJ from 0.4 to 35. Of most significance, he noted a number of sources for the discrepancies between theoretical and experimental results. Among these were the following: 1. The strain gage thickness is not negligible in comparison with the ply thickness; 2. The gage grid spans an appreciable arc length of the convolution torus. Since the gage measures an average strain over the grid length, it will not indicate the maximum strain in a region of a strain gradient; 3. Bellows generally possess a variable ply thickness along a convolution profile which is difficult to anticipate analytically; 4. Nominally similar convolutions exhibited different strain levels at corresponding locations. - 4- Even after attempting to account for all the known experimental errors, C.E. Turner (14) estimated potential errors between theory and experiment in the range of + 10%, in extreme cases + 20%. Similar limitations of experimental studies have been observed by the author. In addition, an effect due to ply interaction in multi-ply bellows has been noted. 3. REFERENCE CONVOLUTION SHAPE Although convolution shapes can be categorized, they are far from standardized. This is mainly due to the proprietary manner in which bellows have been developed: methods of fabrication and production machinery itself have often been developed in-house. Because other than shape parameters can affect bellows performance, optimum convolution configurations may vary from one manufacturer to another. Localized changes in ply thickness and material properties, for example, will influence bellows performance. These differences in convolution characteristics are especially apparent in small,highpressure multi-ply bellows. For the purpose of this study, an idealized,axisymmetric, U-shaped convolution with uniform material properties and thickness will be adopted as the reference configuration. The physical parameters describing this shape are the internal radius, r^, depth of convolution, d, root and crown torus radius, r r and r c , and thickness, t (Figure 1 ) . 4. STRESS ANALYSIS Stress states in bellows resulting from axial compression and external pressure were determined by the finite element method. Typical surface stress distributions are depicted in Figures 2 and 3. Thus, subjected to axial compression, a U-shaped bellows exhibits peak meridional surface stresses at root and crown. Resultant circumferential surface stresses are lower and more variable, not necessarily peaking at root and crown. All surface stresses approach zero at midspan. - 5 - External pressure causes peaks in all surface stresses at three locations: root, crown and midspan. Although of slightly different magnitudes, each principal stress has the same sign at root and crown which is opposite to that at midspan. Theoretical analyses of stress states in thin shells generally assume a linear strain variation with distance normal to the neutral surface. In bellows convolutions, the ratio of torus radius to thickness can be small enough to yield a significantly nonlinear strain distribution at root and crown. A good approximation of the resulting stress distribution is given by the Winkler-Bach formula for curved beams. The stress distribution has a hyperbolic pattern, attaining a maximum value at the concave surface in the case of pure bending. A study of the stress distribution obtained from a finite element analysis of a bellows bears this out. Due to the large circumferential radius of curvature the thickness ratio, r/t, is very large. The stress distribution resulting from loading should be practically linear and this is corroborated by the finite element analysis of bellows. While the maximum meridional stress in bellows due to axial deflection frequently occurs at mid-crown and mid-root (<{>=0) , this is not always the case. N.C. Dahl (4) found that for toroidal bellows the location of maximum stress moved away from <f>=0 for U>5, approaching <J>=ir/2 for large y. According to the results of the study of U-shaped bellows by T.M. Trainer (13), the location of maximum stress is at (J>=0 for y<0.75. For a transition range of y from 0.75 to 2 an approximately constant maximum stress level exists over a portion of the torus. For values of v>>2 the location of the maximum stress tends to <J>=7r/2 with a significant decrease in stress occurring as A similar situation exists for the case of bellows subjected to external pressure with the difference that the maximum meridional stress occurs at (j>=0 of the crown and root for p<5, the transition range is 5<y<10, and for \i>10 the location of maximum stress tends to (j)=ir/2, again with a significant decrease in stress occurring as <(>-*-0. The distribution of circumferential stress is more complex. In the case of axial compression, extreme - 6 - levels generally occur near <J>=0 in root and crown. Depending on the bellows geometry, however, the location of peak stress may shift to <j>>0. External pressure can cause the circumferential stress state to exhibit several locations of extreme levels. 5. APPROACH TO STATISTICAL STUDY The study of the elastic states of stress in bellows due to axial compression and external pressure reveals a complex picture: the stress distribution changes with change in any geometric parameter; the location of the extreme meridional stress levels changes from <|)=0 for small y to 0»ir/2 for large y; external pressure causes an extreme meridional stress level at midspan as well as the crown and root; except for small y the locations of extreme meridional stress levels due to axial deflection and external pressure do not coincide; the circumferential stress distribution is difficult to categorize. Some assumptions must be stated to reduce the problem to manageable proportions. It will be assumed that the meridional and circumferential stress levels at <j>=0 characterize the bellows performance capability. Bellows which exhibit extreme meridional stress levels due to axial deflection and external pressure at cj>=0 fall in the range of y<2. Such bellows have widespread application, generally at low pressure and where axial flexibility is important. Higher pressure may be readily accommodated by increasing the number of plies. The associated circumferential stress does not generally peak at <J>-0, but the resultant stress due to the combination of axial deflection and external pressure will be a maximum near cf>=0. The objective is to derive a response surface for the stresses at root and crown in terms of bellows geometrical parameters. Five geometrical parameters were considered essential to the statistical study: rc» *r> *i> d, t (Figure 1 ) . Selected bellows configurations were analysed by the finite element method to generate the relevant data. A "halfreplicate two level fractional factorial design" was chosen so that the number of test cases required would be minimized. Then, to permit an evaluation of the quadratic terms in the response function a "star design" (15) with center point was superposed on the fractional factorial. A frequent practice - 7 - is to make such a design "rotatable" by a proper choice of the length of the axis arm of the star design, thus reducing the complexity of mathematical computation. The preferred choice of the variable levels precluded this practice. Moreover, the least squares method was to be utilized in the regression analysis of the data and rotatability was not an essential characteristic. The result is a non-rotatable "central composite design" in five variables as shown in Table 1. Two such designs were considered in the study: one in a range of low values of all parameters; one in a range of higher values of all parameters. Table 2 gives the ranges of the parameters for each design. Taken together these arrays of tests permitted the inclusion of higher order terms in the mathematical model of the response surface. In the first design the values of y ranged from 0.09 to 2.64. In the second design y ranged from 0.29 to 2.28. 6. REGRESSION ANALYSIS Conventional formulas for the maximum meridional stress in a bellows are generally expressed as the product of operational (pressure or axial deflection), physical (number of plies or convolutions) and geometrical parameters, the last raised to a power. Such an expression was adopted as the mathematical model for the response surface: S a a a a a a = e ° r l r 2 d 3 r.* t 5 c r i ...(1) Here S is the stress, and a.,i=0,l,...,5 are unknown exponents to be evaluated for a best fit of the function to the data. In the case of axial deflection, stress is proportional to deflection and Young's Modulus. Thus _ (3 a a a a a f • ie • r ' i ! d > r.» t ! £i c r ...(2) l This expression is dimensionless. Then the sum of - 8 - oti, 1 • 1,...,5 must add to -1. A simplification may be introduced by dealing in the dimensionless ratios : Y Y B " ^ C Then „ S. Ci * = •^ _ r . 3 e B 0 y A 3 l 3 2 Y 3 3 V ii y l> ! & If the bellows under consideration possessed n c convolutions and experienced a total deflection A = n c 6, then n r S — — A E B 6 - e ° Y e X A B 1 2 Y Y B B Y Y C 3 6 Y " Y E Length and deflection are common design constraints. Further, the number of convolutions per length and hence the deflection per convolution, will depend on convolution pitch. It is therefore not immediately obvious which convolution configuration will have the more favourable stress state. Therefore, designers may find it advantageous to relate the bellows axial deflection to length rather than the number of convolutions. For U-shaped bellows the relation between length, i, and the number of convolutions is % = 2nc(rc + rr) .. .(6) The stress parameter can then be rewritten s*2 2 0 A(Y. + Y,)E A a s 1 e 3_ B 0 2 Y A Y 3 1 3 *B Y Y Y B3 B C I Y Y E - 9 - A similar line of reasoning yields the. relation for stress resulting from external pressure!, p. 3 3 6 Y A 3 Y B 3 Y C 3 Y i E i •••(»> where n n is the number of plies in the bellows. P I While the mathematical models, Eq.(7) ar>d (8) are not linear, they can be made so by taking their logarithm. A multiple linear regression! analysis utilizing the least-squares method is thjen employed. This analysis minimizes the logarithm of; the errors rather than the actual errors. These formulations were adopted for both membrane and bending stresses. Estimates of these stress components were based on the surface stresses obtained from the finite element method: S = m OS = + Si s 2 ...(9) S S b os - S is 2 where S o s , S±s, are the outer and inner surface stresses, S m , Sb the membrane and bending components of surface stresses. In view of the non-linear stress distribution, S m and Sb in the meridional direction are fictitious quantities. The circumferential components however should be accurate estimates. Design of structural components involving bellows often requires a knowledge of forces developed due to axial deflection. To meet this need a response surface for bellows spring rate was derived based on the mathematical model n K FT r i E oK = -2— r? (YA + Y n )E 3 3 3 3 e - ° V V V A B C - 10 K is the spring rate. While the response functions (7), (8) and (10) give acceptable accuracy for many applications, an improved fit will result through a subsequent derivation of a correction factor F = exp(f(Yi)) . . .(11) The function f(Yi) is a polynomial of products of the geometric parameters. Terms of f(Yi) significant in improving the fit of the response function were identified by a computer program using a multiple linear regression analysis with a forward stepwise algorithm. Within the range of geometrical parameter levels considered some of the circumferential stresses changed signs. This precluded a direct application of Equations (7) and (8). Instead, absolute values of stress were first introduced into Equations (7) and (8) to evaluate the unknown exponents. Subsequently a correction factor in the form of a polynomial in the geometrical parameters, F = f(Y^), was introduced to allow for the change in sign. The response functions then take the form - - 3. 33. 3 c AE 3 °V (Y A +Y f i )AE c for axial deflection; n_ S 3 3 3 3 3 o v lv 2 v^s Y^U ...(12) for external pressure; ,- 3 F 4 (YA+YB)E for the axial spring rate. e 3 Y A Y 3 Y B 3 3 C - 11 - 7. RESULTS AND DISCUSSION The regression analysis according to the mathematical models Equations (7), (8) and (10) yielded the results in Tables 3(a) and 3(b). These oversimplified models exhibit acceptable fit for many applications and indicate the effect of changes in the various geometrical parameters. According to the sum of 3., i=l,...4, all meridional stress components, S, due to axial compression and external pressure are practically independent of inner radius, rj.. This is also the case for circumferential bending stress components. The circumferential membrane stresses due to axial compression are approximately inversely proportional to inner radius. Under external pressure the circumferential stress components are approximately proportional to inner radius. All bending stress components decrease at root and crown for respective increases in torus radius. A change in root torus radius has little effect on bending stresses at the crown, and vice versa. This is also the case for all meridional membrane stress components. The circumferential membrane stress components are affected by changes in either torus radius. In the case of axial deflection, these relationships are valid for unit deflection per convolution. If axial deflection is set per unit length the effect is more complicated. Convolution depth is the most influential configuration parameter, closely followed by ply thickness. An increase in convolution depth or ply thickness results in a decrease and increase, respectively, of most stress components due to axial compression and vice versa for external pressure. Some circumferential membrane stress components decrease in magnitude for increase in both configuration parameters. The axial spring rate is approximately proportional to the inner radius and practically independent of torus radii when expressed in units per convolution. Ply thickness and convolution depth are the dominant configuration parameters. Tables 4(a) to 4(d) present formulas for the correction factor F, which includes the sign of the stress components. For the range of parameter levels considered, the magnitude of F is close to 1. - 12 - Because the regression analysis approximates the meridional non-linear stress distribution across the ply thickness by a linear stress distribution, the membrane and bending stress components, separately, will be slightly in error. Should an accurate estimate of the membrane stress be desired, Table 4{e) presents appropriate formulas for the effect of external pressure acting on the convoluted surface only. For axial deflection, the axial spring rate may be used. To help visualize the response functions of Equation (12), Figures 4 to 9 show the relationship of meridional stress components at the convolution root versus a configuration parameter for bellows subjected to axial compression and external pressure. Because these formulas are based on a statistical derivation, some error is likely to be present in most estimates. For the selected configurations of this study the predicted meridional stresses agreed with calculated stresses to within a maximum of five per cent. Errors for the circumferential stresses were higher, reaching a maximum of about seven per cent for bending and more for the membrane stresses. The latter errors were due to the low magnitude of the calculated stresses for the selected configuration. Results of the analysis were also compared with those of several sources for the example given in Â. Laupa and N.A. Weil (1). Although the convolution configuration possessed a characteristic 11-1.47, the dimensions were outside the range of parameter levels considered in this study. Nevertheless, as Tables 5(a) to 5(h) show, the statistically derived response functions yield results which approach those of the finite element method more closely than many design formulas proposed in literature or in current use. 8. CONCLUSION Response functions in terms of geometrical variables *i> rr» r c » ^ an<* *• a r e derived for U-shaped bellows with a characteristic shape parameter y<2 for bending and membrane stress components at root and crown in the meridional and circumferential directions due to axial compression and external pressure, for axial - 13 - spring rate and for the actual meridional membrane stress due to external pressure acting on the convoluted surface only. Suitable for design or analysis, these formulas represent an improvement in the accuracy of response estimates and in the simplicity of calculations over earlier design formulas. 9. ACKNOWLEDGEMENT The finite element computer program utilized in this study was developed by R. Shill of Atomic Energy of Canada Limited, Chalk River. 10. REFERENCES A. Laupa, N.A. Weil, Analysis of U-Shaped Expansion Joints, J. of Applied Mechanics, March 1962, pp 115-123. R.A. Clark, On the Theory of Thin Elastic Toroidal Shells, J. of Mathematics and Physics, vol. 29, 1950, pp 146-178. R.A. Clark, An Expansion Bellows Problem, J. of Applied Mechanics, March 1970, pp 61-69. N.C. Dahl, Toroidal-Shell Expansion Joints, J. of Applied Mechanics, December 1953, pp 497503. T. Ota, M. Hamada, On the Strength of Toroidal Shells, Bulletin of JSME, vol. 6, No. 24, (1963), p 638-654. M. Hamada, S. Takezono, Strength of U-Shaped Bellows, (1st. Report, Case of Axial Loading), Bulletin of JSME, vol. 8, No. 32, (1965), pp "=25-531. M. Hamada, S. Takezono, Strength of U-Shaped Bellows, (2nd Report, Case of Axial Loading Continued), Bulletin of JSME, vol. 9, No. 35, (1966), pp 502-513. - 14 - 8 - M. Hamada, S. Takeaona, Strength of U-Shaped Bellows, (3rd Report, Case of Loading of Internal Pressure), Bulletin of JSME, Vol. 9, No. 35, (1.966), pp 513-523. 9. M. Hamada, Y. Inove, T. Nakatani, M. Moriishi, Design. Diagrams and Formulae for U-Shaped Bellows, Int. J. of Près. Ves. and Piping, (4), 1976, pp 315-328. 10. The M.W. Kellogg Company, Design of Piping Systems, John Wiley & Sons Inc., Revised 2nd Edition, 1955, pp 214-230. 11. R.A. Winborne, Simplified Formulas and Curves for Bellows Analysis, NAA-SR-9848, Atomics International, California, USA, August 1964. 12. W.F. Anderson, Analysis of Stresses in Bellows, Part 1: Design Criteria and Test Results, Atomics International, NAA-SR-4527, October 1964. 13. T.M. Trainer, Final Report on the Development of Analytical Techniques for Bellows and Diaphragm Design, T.R. Ho. AFRPL-TR-68-22, Battelle Memorial Institute, Columbus, Ohio, March 1968. 14. C E . Turner, Stress and Deflection Studies of Flat-Plate and Toroidal Expansion Bellows Subjected to Axial, Eccentric or Internal Pressure Loading, J. of Mechanical Engineering Science, vol. 1, No. 2, 1959, pp 130-143. 15. J.S. Hunter, Determination of Optimum Operating Conditions by Experimental Methods, Part II-3, Models and Methods, Industrial Quality Control, vol. 15, No. 8, February 1959. 16. Standards of the Expansion Joint Manufacturers Association, Inc., Fourth Edition, 1975. -15 - TABtE 1 * CASE Design Matrix Used to Derive Computer Generated Data. CROWN TORUS ROOT TORUS RADIUS RADIUS r r c r A B CONVOLUTION DEPTH, d i | INTERNAL RADIUS, r i PLY THICKNESS, t C D E=ABCD -1 -1 -1 +1 -l I t 1 -1 2 +1 -1 -1 3 -1 +1 4 +1 +1 -l 5 -1 -1 +i 6 +1 -1 +i 7 -1 +1 +i 8 +1 +1 +i ! -1 -1 +1 -1 +1 +1 9 -1 -1 10 +1 -1 -l 11 -1 +1 12 +1 +1 -1 -i 13 -1 -1 +i 14 +1 -1 +i 15 -1 +1 +i 16 +1 +1 +i 0 0 0 0 18 -1 0 0 0 0 19 +1 0 0 0 0 +1 +1 +1 +1 +1 +1 Fractional Factorial -1 -l +1 25-l -1 +1 +1 -1 +1 -1 -1 +1 +1 17 . 0 20 0 -1 0 0 0 21 o +1 o 0 0 22 0 0 -1 0 0 0 o 23 24 25 0 o o ! : 0 +1 0 0 -1 0 0 0 +1 0 26 0 ; 0 0 27 o ; 0 0 > : 0 -2 0 +2 Center Point Modified Star Design - 16 - TABLE 2 : Variable Levels Corresponding to Elements of the Design Matrix DESIGN 2 DESIGN 1 VARIABLE SYMBOL, UNITS LOW (-1) HIGH (+1) LOW (-1) HIGH ( + 1) Crown Torus Radius, rc A» nun 0.762 1.524 2.540 4.445 Root Torus Radius, r r B, mm 0.762 1.524 2.540 4.445 Convolution Depth, d C> mm 3.175 6.350 8.890 17.78 D, mm 12.70 63.50 76.20 152.4 E j mm 0.203 0.305 0.356 0.457 Inner Radius, r i Ply Thickness, t TABLE 3(a) : VALUES OF EXPONENTS IN THE RESPONSE FUNCTIONS FOR MERIDIONAL STRESS COMPONENTS AND AXIAL SPRING RATE: r S "E i is ^ Axial Compression AE (Y.+YÏAE B B B B * n S F e ° Y . 1 Y 2 Yv ' Yv " External Pressure A 'B C E ' P ncK Axial Spring Rate X.K 2ir r ± E STRESS COMPONENT LOCATION B r i(YA+YB)E 8i B2 B 0 B it CROWN -1.160 -1.082 -0.006 -1.980 2.033 ROOT -1.063 -0.032 -1.100 -1.932 2.053 CROWN -0.879 -0.490 -0.018 -1.699 1.082 ROOT -0.818 -0.029 -0.548 -1.655 1.111 CROWN -0.781 -0.532 -0.020 1.462 -0.918 ROOT -1.036 -0.038 -0.626 1.636 -0.988 CROWN -1.878 -0.531 0.028 2.255 -1.901 ROOT -1.760 0.019 0.519 2.353 -1.926 -0.850 -0.028 -0.001 -2.620 2.896 MEMBRANE AXIAL COMPRESSION BENDING MEMBRANE EXTERNAL PRESSURE BENDING AXIAL SPRING RATE * F determines the sign of stress TABLE 3(b) : VALUES OF EXPONENTS IN THE RESPONSE FUNCTIONS FOR CIRCUMFERENTIAL STRESS COMPONENTS: "c Axial Compression: r i AE S IS Fe B B 1 Y. A YB* B B Y C3 Y*E ** * External Pressure: STRESS COMPONENT LOCATION B B B2 B3 0 3It CROWN -1.647 1 .636 0 .452 -1.624 -0. 430 ROOT -0.702 0 .299 0 .746 -1.604 0.334 CROWN -2.143 -0 .496 -0 .026 -1.697 1. 086 ROOT -2.096 -0 .029 -0 .561 -1.655 1. 112 CROWN 1.566 0 .099 0 .509 -1.139 -0. 425 ROOT 1.538 0 .188 0 .478 -0.899 -0. 566 CROWN -3.091 -0 .534 0 .031 2.243 -1 891 ROOT -2.954 0 .025 -0 .522 2.333 -1 912 MEMBRANE AXIAL COMPRESSION I BENDING MEMBRANE EXTERNAL PRESSURE BENDING F determines the sign of stress. 05 - TABLE 4 (a) 1. : 19 - Formulas for Correction Factor F for Meridional Stresses due to Axial Compression, and Axial Spring Rate. Signs of Bending Stress Components are for Outer Surface. To obtain the Inner Surface Bending Stress Component, change the sign of F. MEMBRANE STRESS AT CROWN: F - -exp[0.711 x 10" 1 - 0.130 x 10<Y £ /Y c ) - 0.451 x 10 2 (Y A Yg 2 ) + 0.297 x 1O" 2 (Y E /Y B 2 ) + 0.344(Y A 2 /Y E ) + 0.363 x 10(Y B 2 /Y c ) + 0.188 x H f 1 <YC2/YE) - 0.207(Y A Y C /Y E ) - 0.340 x 10" 2 (Y B /Y 2. MEMBRANE STRESS AT ROOT: F « -exp[0.364 x 10" 1 + 0.139 x 10" 3 (Y c /Y A 2 ) - 0.104 + 0.566(Y A 2 /Y B ) + 0.571(Y B 2 /Y E ) + 0.318 x lO'1^ - 0.326(Y B Y C /Y E ) - 0.256 x 10" 1 (Y B /Y B Y c ) ] . 3. BENDING STRESS AT CROWN: F - exp[0.257-0.435 x 10" 2 (l/Y A ) + 0.820 x 10(YA Y c ) + 0.264 x lO^CY^Yg) - 0.454(Y E /Y A ) - 0.173 x 1 0 ' 4 ( l / Y c Y £ ) - 0.753 x 10" 2 (Y B /Y c 2 ) + 0.596(Y A 2 /Y B ) - 0.677(Y A 2 /Y E ) + 0.227 x 10(Y B 2 /Y c ) + 0.141 x 1 0 " 5 ( l / Y A 2 Y,,) - 0 . 1 8 0 x 10 3 (Y A Y c Y £ ) - 0.107(Y A YC/YE) + 0.141 x 10" 3 (Y c /Y A Y B ) ] . 4. BENDING STRESS AT ROOT: F - -exp[0.281-0.639 x 10~ 2 (l/Y_.) - 0.499 x 1 0 ~ 2 ( l / Y o ) 3 2 + 0.391 x ÎO^CY at,/Y ) - 0.211 x 10(Y_/Y n ) - 0.949 x 10 (Y. Y ) Ci (j A B - 0.920 x 10(YA Y c 2 ) - 0.703 x 104(YA Yg2) + 0.696 x U f 4 (Yj/Y^) + 0.185 x 10 2 <Y B 2 /Y c ) - 0.143 x lOttg 2 /^) + 0.201 x 10~ 5 (l/Y B 2 Yc) + 0.161 x 103(YA YB Yc) + 0.364 x 104(YA Yfl YE) - 0.623 x 10" 3 (Y B /Y c 5. AXIAL SPRING RATE: F - exp[-0.185 + 0.257 x 10' 4 (l/Y B 2 ) + 0.558 x 10" 2 (Y E /Y A 2 ) + 0.417(YA - 20 - TABLE 4(b) 1. : Formulas for Correction Factor F for Meridional Stresses due to External Fressure. Signs of Bending Stress Components are for Outer Surface. Inner Surface Bending Stress Components are obtained by a change of sign of F. MEMBRANE STRESS AT CROWN: F - - e x p [ - 0 . 3 4 5 + 0.870 x 10" 1 (Y c /Y A > - 0.727 x 10~ 2 (Yj/Yç 2 ) - 0.397 x 10" 3 (Y./Y. 2 ) + 0.129 x 10<Y. 2 /Y C A AC 2 ) - 0.662 x 10" 2 (Y,, Y./Y 2 ) B CA + 0 . 1 6 9 x 10" 6 (Y B /Y A Y E 2 ) - 0 . 2 5 8 x 10" 1 (Y ( , Y j / Y g 2 ) + O . 1 2 1 x l O 3 ( Y A Yg 2 - 0.1A2(Y B Y C 2 /Y E ) + 0 . 1 8 7 x 1 0 " 2. 3 2 MEMBRANE STRESS AT ROOT: F - exp[-0.782 x 10"2 - O.19O(Y-,/Y2) + 0.638 x lO'^Y^/Y-) be D & - 0.126 x 10~5(l/Y 2 Y.) + 0.104 x 10(Y 2 /Y. 2 ) - 0.998 x 10~3(Y 2/Y.2) a + 0.295 x 10~2(Y 2 I* o . \i LA \J / Y 2 ) + 0.927 x 10" 1 0 (l/Y. 2 Y_2) + 0.215 x 10"2 A D •(Y E /Y B Y c ) - 0.672 x 10~ 1 (Y B ?<?/*£+ L 0.213 x 10 3 (Y A 2 Yg Y c ) - 0. 338 x 10(Y A 2 Y B /Y E ) - 0.170 x 10~ 2 (Y A 2 /Y B Y £ ) + 0.143 x 10" 3 • (YC2/YB V 3. " °-189x 10 "2(YB Y C /Y A Y E ) ] - BENDING STRESS AT CROWN F - exp[0.596 x lO"1 - 0.245 x 10"4(l/YA2) + 0.167 x 102(YA Y ^ + 0.205 x 10"4(Yc2/YE2) + 0.190 x 10"7<l/YA2 Y^) - 0.394 x 10(Ytt YE/YA) - 0.226 x 10(YA2 YC/YB) - 0.226 x 10(YA2 Y C /Y E - 0.160 x 10(Yc2 Y E /Y A >], 4. BENDING STRESS AT ROOT: F - exp[0.117 + 0.268 x 10(Y_) + 0.151 x 10 3 (Y. Y^) - 0.304 x 10*(Y Y 2 ) - 0.103(Y_/Y O fit ti 2 ) - 0.250 x U f ^ Y ^ / Y 2 ) \f A if - 0.371 x 10(Y. YJV) - 0.131 x 10"1(Y_/Y. Y_) + 0.118 x 103. •(YA Y B Y c 2 ) - 0.337 x 103(YA Y E 2 /Y C ) - 0.260 x 10"5(Yc/YA2 Y B ) + 0.106 x 10~7(l/YA Y B 2 Y c ) - 0.278 x 10(Yg2 Yp/Yg)]. - 21 - TABLE 4(c) : 1. Formulas for Correction Factor F for Circumferential Stresses due to Axial Compression. Signs of Bending Stress Components are for Outer Surface. Inner Surface Bending Stress Components are obtained by a change of sign of F. MEMBRANE STRESS AT CROWN: F - [0.139 + 0.280 x lO^d/Y.) - 0.765 x 10"1(Y,,/Y_) U IS Ci - 0.182 x 10 2 (Y n Y r ) - 0.570 x 10"4(l/Y,. Y_) + 0.104 x 10 2 (Y. Y 2 5 2 + 0.375 x 10(Y B /Y A ) - 0.278 x 10" (l/Y A Y B ) - 0.916 x 10~J •<Y E /Y A Y B ) + 0.991 x 10' 6 (l/Y A Yg Yg)]. 2. MEMBRANE STRESS AT ROOT: F - - 1.00 3. BENDING STRESS AT CROWN: F - exp[0.318 - 0.679 x lo"2(l/YA) + 0.673 x 10"X(YA/YB) - 0.176 x 10(Y£/Yc) - 0.465 i^^/\) + 0.240 x 10(YB2/Yc> - 0.918 x 10~2(Yc2/YE) + 0.135 x 10"5(l/YA2 Yç) + 0.428 x 10~* •WE»' 4. BENDING STRESS AT ROOT: F - -exp[-0.628 x 10"1 - 0.320 x 10"1(l/Y(;) + 0.182 x 10 2 (YA Y ) 2 2 + 0.101 (Y_/Y,J - 0.925 x 103(Y.A Y_ ) - 0.100 x 105(Y. Y_ ) a t D At + 0.951 x 10~2(YA/Yc2) - 0.806 x 10"4(YB/YE2) - 0.226 x 10"? .(YC/YB2) + 0.407 x 10"7(l/Yc Y£2) - 0.453 x 10 3 (Y B 2 Y £ ) + 0.142 x 10(Y2/Y.) + 0.176 x 102(Y2/Y,,) - 0.201 x 10<Y 2/X ) DA 3 O 4 Li D IS + 0.112 x 10 (YA Y B Y c ) + 0.436 x 10 (YA Y B Y £ ) + 0.199 x 10"l (YE/YB Y c ) + 0.131 x 10~5(l/YA Y B Y c )]. 2 ) - 22 TABLE 4(d) : 1. Formulas for Correction Factor F for Circumferential Stress due to External Pressure. Signs of Bending Stress Components are for Outer Surface.Inner Surface Bending Stress Components are obtained by a change of sign of F. MEMBRANE STRESS AT CROWN: F - - [ - 0 . 1 7 5 x 10 + 0.521 x 10 2 (Y A ) + 0 . 1 2 8 ( l / Y c ) - 0.160 x U f 1 (1/Yg) + 0.145 x 10(Y c 2 ) + 0.331(Y c /Y E ) + 0.448 x 10(Y A 2 /Y E ) - 0.393 x 10" ? (Y c 2 /Y E 2 ) + 0.354 x 10 3 (Y A YR Y c ) - 0.534 x 10(Y A Y c / - 0.111 x 10 2 (Y A Y E /Y C 2 ) + 0.936 x 10" 2 (Yfi Yç/Yg 2 ) +0.236 x 10"? .(Y C /Y B Y E 2 ) + 0.374 x 10" 3 (Y B /Y c 2 Y £ ) - 0.230 x l o V / Yg/Yç) - 0.296 (Y B 2 /Y C Y £ ) - 0.543 x 10" 2 (Y c 2 /Y A Y^,)]. 2. MEMBRANE STRESS AT ROOT: F - - [ - 0 . 4 7 1 X 10 + 0.113 x 10 2 (Y o ) + 0.922 x 1 0 ~ 2 ( l / Y o ) + 0.191(Y c /Y E ) + 0.575 x 10 2 (Y E /Y c ) + 0.200 x 10" 2 (Y c /Y B 2 ) - 0.345(Y c 2 /Y A ) - O.729(Y C 2 /Y E ) - 0.134(Y A 2 /Y B 2 ) - 0.228 x 10" 2 (Y 2 /Y_ 2 ) + 0.757 x 10"1<Y_/Y. Y.) - 0.449 x 10"2. Vi B CM I» A •(YB/YA Y c 2 ) + 0.183 x 10" 5 (Y B /Y A Y £ 2 ) + 0.361 x l O ' V g / Y g Y,, 2 ) - 0.26 x 1 0 ~ V A / Y B 2 Y E ) - 0.519 x 10'2(YE/YB2 Yc> + 0.551 x 10~* •(YA YC/YB Y E ) - 0.346 x 103(YE2/Yc2)l. 3. BENDING STRESS AT CROWN: F - exp[0.461 x 10"2 - 0.105 x 10"*(l/YA2) + 0.650 x 102(YA Y,,) - 0.249 x 103(Y. Y_) - 0.943(Y_2/Y.)+ 0.104(Y. Y_/Y_ ) U £• V A A t> b + 0.674(YB Y E /Y C 2 ) - 0.869 x 10~6(Yc/YA Yj2) + 0.234 x 10 • .(Y Y 2/Y ) + 0.325 x 103(Y A 4. D L Y_2/Y ) - 0.664 x 10 (Y 2 Y /Y )]. U b D A U £ BENDING STRESS AT ROOT: F - e x p [ - 0.124 x 10" 1 + 0.674 x 10(YB> + 0.153 x 10~ 1 (Y (; /Y B ) - 0.468 x H f V / Z Y j 2 ) + 0.151 x 10 3 (Y B 2 Y c 2 ) - 0.574 x 10* •<YB Y c Y £ 2 ) - 0.390 x 10(Y B Yg/Yç 2 ) - 0.559 x 10" 1 (Y ( , Yg/Y A 2 ) - 0.341 x 10(Y B 2 Y C /Y E )J. - 23 - TABLE 4(e) : 1. Actual Meridional Membrane Stress due to External Pressure Acting on the Convoluted Surface Only. ROOT - ° - 0 6 8 Y R ° - W Y, 1 ' 112 Y -1'050 exp( - 0 .826)Y A 2. CROWN P p S mc /np -b L £• - 24 - TABLE 5 : Bellows Response to Loading in the Example of Laupa and Weil(l): r (a) = r = 1 3 . 7 mm, r r = 304.6 mm, d = 58.2 mm, t = 1.2 7 mm y = 1.47 v = 0.3 c Bellows Axial Spring Rate and Meridional S t r e s s Components a t Root due to Axial Compression. Spring Rate, Membrane K Stress,Sm SOURCE 10 n c K Surface Bending Stress, Sb 1 0 2 n c riSfe 10 2 n c r., S i m AE ri E AE Resultant Surface Stress, S R 10 2 n c r i S R AE Inner Outer Inner Outer 8 69 -9 17 8 .58 -9 28 6 82* -8 82* 8 .48 -9.17 Finite Element Analysis 2 .88 * Statistical Model 2.99 Laupa & Weil (1) 2.90 -0 11 8 80 -8 80 8 .6<) -8. 92 Salzmann (1) 3.06 -0 12 10 51 -10 51 10 .40' -10.63 Hamada 2 .64 * -0 10 9 91 -9.91 9 .81 -10.01 2 .92 -0 10 8 91 -8 91 8 81 -9.02 Kellogg (10) 2 .21 -0 08 t 12 48 -12. 48 12 40 -12.57 Anderson (12) 2 .62 -0 10 t 11. 73 -11. 73 -11 63 -11.83 EJMA 3 .05 -0. 09 (-0. 12) t 10.24 -10. 24 10 15 -10.33 (7) * (16) -0 .11 -0 .35* (-0 11) t Assumed linear stress distribution. ncK r i E = t Derived from K "cr iS m . 2*t ; AE r 4 Derived from V F --= KA i N * Sjjt, Membrane force per unit length. F * Axial Force. _' Bending moment per unit length. - 25 - TABLE 5 (Cont'd) (b) Bellows Axial Spring Rate and Meridional Stress Components at Crown due to Axial Compression. Spring Rate, Membrane K Stress, Sm SOURCE 105n r 10 K c iE Statistical Model ViSm 102 n AE 2.88 + Finite Element Analysis Surface Bending Stress, Sb 10 2 n c r± S R AE -0,09 -0.34* (-0.10)t 2.99 c ri Sb Resultant Surface Stress, S R AE Inner Outer Inner Outer -9.00 8.51 -9.09 8.42 -8.49* 8.49* -8.83 8.15 Laupa & Weil (1) 2.90 -0.09 -8.68 8.68 -8.77 8.59 Salzmann (1) 3.06 -0.10 -10.51 10.51 -10.61 10.41 Hamada 2.64 * 2.92 -0.09 -0.09 -9.91 -8.71 9.91 8.71 -10.00 -8.81 9.83 8.62 Kellogg (10) 2.21 -0.07t -12.48 12.48 -12.56 12.41 Anderson (12) 2.62 -0.08t -11.73 11.73 -11.81 11.64 EJMA 3.05 -0.09 (-0.10)t -10.24 10.24 -10.33 10.15 Y' (16) * t * Assumed linear stress distribution Derived from K Derived from Sm n r c K i E „ n rJ S i m c F - . Zirt r i (d + r ' *- ï i Membrane Force perUnit N M AE Sbt2 F - Length , Bending Moment per Unit Lengf 6 - Axial Force - 26 - TABLE 5 (Cont'd) (c) Bellows Meridional Stress Components at Root due to External Pressure Membrane Stress,Sm SOURCE Surface Bending S t r e s s , Sj, n S %Sm P F i n i t e Element Analysis S t a t i s t i c a l Model 23.4 (39.6)* 23.6 t R e s u l t a n t Surface S t r e s s , SR "p S R P P Inner Outer Inner Outer -588.4 620.8 -565.0 644.2 (600.4)* -560.8 640.0 (-600.4)* Laupa & Weil (1) 23.9 -582.? 582.9 -559.0 606.8 Hamada (9) 136.9 -625.4 625.4 -488.5 762.3 Kellogg (10) - -1050 1050 -1050 1050 Anderson (12) 120.5 -758.3 758.3 -637.8 878.8 EJMA (16) 22.9 -608.3 608.3 -585.4 631.2 * Assumed Linear Stress Distribution t Obtained from Axial Force Equilibrivan Root Axial Force F = S r . t r mi _ 27 - TABLE 5 (Cont'd) (d) Bellows Meridional Stress Components at Crown due to External Pressure Membrane Stress,Sm Surface Bending Stress, S], Resultant Surface Stress, S R SOURCE p pSm P Inner Finite Element Analysis Statistical Model "p S R P P -22.5 (-39.0)* -22.3 t Outer Inner Outer -588.2 555.4 -610.7 532.9 (-571.0)* (571.0)* -610.0 532.0 Laupa & Hell (1) -22.1 -591.3 591.3 -613.4 569.2 Hamada (9) -115.0 -625.4 625.4 -740.4 510.4 Kellogg (10) - -1050 1050 -1050 1050 Anderson (12) -120.5 -758.3 758.3 -878.8 637.8 EJMA (16) -22.9 -608.3 608.3 -631.2 585.4 * Assumed Linear Stress Distribution t Obtained from Axial Force Equilibrium Crown Axial Force F S m(ri + d ) t - 28 - TABLE 5 (Cont'd) (e) Bellows Circumferential Stress Components at Root due to Axial Compression Membrane Stress,Sm SOURCE 1Q2 Vi S m Surface Bending Stress, S^ 10 2 n c AE r iSb Resultant Surface Stress, S R 10 2 n C AE AE Outer Inner r. SR 1 Outer Inner Finite Element Analysis -4.59 2.67 -2.67 -1.92 -7.26 Statistical Model -4.43 2.56 -2.56 -1.87 -6.99 Laupa & Weil (1) -4.35 2.64 -2.64 -1.71 -6.99 Hamada (7) -4.56 2.73 -2.73 -1.83 -7.28 (f) Bellows Circumferential Stress Components at Crown due to Axial Compression Membrane Stress,Sm SOURCE 102n r.S c l m Surface Bending Stress S ' b 10 2 n AE c r. S, l b Resultant Surface Stress S ' R 102nc AE Inner r iSR AE Outer Inner Outer Finite Element Analysis 3.84 -2.62 2.62 1.22 6.46 Statistical Model 3.75 -2.47 2.47 1.28 6.21 Laupa & Weil (1) 4.04 -2.60 2.60 1.44 6.65 Hamada (7) 3.92 -2.67 2.67 1.25 6.58 - 29 - TABLE 5 (Cont'd) (g) Bellows Circumferential Stress Components at Root due to External Pressure Membrane Stress,S ' m Surface Bending Stress S • b Resultant Surface Stress, S., K SOURCE n pSm "pSR "p S b_ P P P Inner Outer Inner Outer Finite Element Analysis 87.2 -181.1 181.1 -93.9 268.3 Statistical Model 86.7 -175.1 175.1 -88.4 261.8 (1) 66.6 -174.9 174.9 -108.3 241.5 (16) 97.6 - - - Laupa & Weil EJMA 1 (h) Bellows Circumferential Stress Components at Crown due to External Pressure f t t Membrane Stress,Sm np Sm SOURCE P Surface Bending Stress , Sb Resultant Surface Stress S • R n _JL P "pSR P Inner Outer Inner Outer Finite Element Analysis . 47.7 -171.2 171.2 -123.5 218.9 Statistical Model 46.6 -166.6 166.6 -120.0 213.2 -177.4 177.4 -107.7 247.1 I i | Laupa & Weil (1) j 69.7 EJMA (16) , 97.6 i _ ! - CONVOLUTI 9N U HRnWN TORUS RADIUS t : PLY THICKNES: CROWN GAP CROWN WIDTH 1 1 CUFF I /SPAN ROOT WIDTH I \ ) CONVOLUTION DEPTH, d _ (OD-ID) I \ 2 r = / V / ROOT GAP V 1_ r r : ROOT TORUS RADIUS OD - OUTER D) AMETER ID - INNER D AMETER FIGURE 1 o Bellows Nomenclature 00 T i 1 ID 1 2 OUTSIDE MERIDIONAL OUTSSDE CIRCUMFERENCE INSIDE CIRCUMFERENCE — INSIDE MERIDIONAL I 0.5 1.0 1.5 2.0 CENTRELINE FIGURE 2 2.5 DISTANCE 3.0 3.5 4.0 4.5 (mm) Typical Bellows Stress Distribution Due to Axial Compression of 1 mm per Convolution. OUTSIDE 200h INSIDE MERIDIONAL OUTSIDE INSIDE CIRCUMFERENCE CIRCUMFERENCE MERIDIONAL i -150 0.0 05 1.0 1.5 2.0 CENTRELINE FIGURE 3 2.5 3.0 DISTANCE 3.5 4.0 (mm) Typical Bellows Stress Distribution Due to External Pressure of 1 MPa. a) Membrane Stress: Y, FIGURE 4 B 0.02 b) Bending Stress: Y = Y 0.02 B Compressive Meridional Membrane and Outer Surface Bending Stress Components at Convolution Roots Due to Axial Compression as a Function of Thickness t/r i" Parameter, Y, L 10 . - o.ooo c) 10 .rot .1 Membrane S t r e s s : 9.0» Y. = Y B 0.04 FIGURE 4 c o n t ' d d) .004 Bending Stress = YB = 0.04 I 10 0.000 e) Membrane S t r e s s : Y A f) Bending Stress: 0.06 FIGURE 4 cont'd 0.06 10", 1 o.ooo g) -m Membrane .a» .ou Stress: .016 .020 .021 .02s 0.000 Y . = Y_ = 0 . 0 8 n a FIGURE 4 c o n t ' d h) Bending S t r e s s : Y 0.08 10 0.000 1) Membrane Stress: .032 Y 0.10 0.000 j) Bending Stress: FIGURE 4 cont'd Y, 1 LO GO I o.ooo k) Membrane S t r e s s : 0.000 Y = 0.12 1) .004 .008 .012 .028 .016 Bending S t r e s s : Y. = Y_. = 0.12 A FIGURE 4 c o n t ' d .03? J5 10* 10* 10" 1 It" 0.000 a) Membrane S t r e s s : Y, = Y B A 0.02 b) .004 .00* Bending .012 .OU .020 Stress: Y A = Y = 0.02 1J FIGURE 5 Tensile Meridional Membrane and Outer Surface Bending Stress Components at Convolution Roots due to External Pressure as a Function of Thickness Parmeter, Y w = -40- IIIIIII i i—I» - —I SSXSSSRRK ses —hnrrn s s s -ao o 11 60 a •H c ai IttUJ.U J .'J J I11111 1 1 1 him IUUJJJ J 1 1 1 UiUU J U tur o o (0 to ai M 0) G M •8 10 -C- 0.000 e) 10"'. .004 Membrane S t r e s s : YA « YR = 0 . 0 6 f) Bending S t r e s s ; Y A Figure 5 cont'd = Y = 0.06 a .E- 0.000 g) .O0< .01» .012 .0» 0.000 .0» Membrane S t r e s s : Y = Y = 0 . 0 8 A B Figure 5 cont'd h) .004 .001 .012 .01* Bending S t r e s s : .020 Y .OU .ait .032 A ~YB» ° ' 0 8 «.0» i) .OU Membrane S t r e s s : Y. * Y_ = 0.10 A D Figure 5 cont'd j) Bending Stress: YA = Y - 0.10 10'. . I 10 k) Membrane Stress: t = "Ï = 0.12 Figure 5 cont'd 1) Bending Stress: Y A = Y B = 0.12 I O.D a) Membrane Stress: FIGURE 6 0.008 b) .1 Bending Stress: Yt 0.008 Compressive Meridional Membrane and Outer Surface Bending Stress Components at Convolution Roots Due to Axial Compression as a Function of Convolution Depth Parameter Y_ = d/r.. : 1 1—~ i — 1 1 1 ^^-Jo.02 -^*^O--0.04 0- 10*. i • Y I : '"'I 10' / 10" ON I 1 a) Membrane Stress: FIGURE 7 Y n = 0.008 Bending Stress: 1 1 1 Yv = 0.008 Tensile Meridional Membrane and Outer Surface Bending Stress Components at Convolution Roots due to External Pressure as a Function of Convolution Depth Parameter Y n = d/r., . 10' îo-l i) Membrane Stress: FIGURE 8 Y w = 0.008 E "r i L O.OO .02 .0* b) Bending Stress: ••= 0 . 0 0 8 Compressive Meridional Membrane and Outer Surface Bending Stress Components at Convolution Roots due to Axial Compression as a Function of Root Torus Radius Y B = r r /r ± . I 00 0.0 a) Membrane Stress Y_ = 0.008 ft FIGURE 9 b) .02 .01 Bending Stress: = 0.008 Tensile Meridional Membrane and Outer Surface Bending Stress Components at Convolution Roots due to External Pressure as a Function of Root Torus Radius Y B = rr/r.. ISSN 0067 - 0367 ISSN 0067 - 0367 To identify individual documents in the series we have assigned an AECL- number to each. Pour identifier les rapports individuels faisant partie de cette série nous avons assigné un numéro AECL- à chacun. Please refer to the AECL- number when requesting additional copies of this document Veuillez faire mention du numéro AECL- si vous demandez d'autres exemplaires de ce rapport from au Scientific Document Distribution Office Atomic Energy of Canada Limited Chalk River, Ontario, Canada KOJ 1J0 Service de Distribution des Documents Officiels L'Energie Atomique du Canada Limitée Chalk River, Ontario, Canada KOJ 1J0 Price $4.00 per copy Prix $4.00 par exemplaire 1112-80