Pressure Vessel Research Committee WRC—50 Years of Progrsss Through Cooperative Research Local Stresses in Vessels—Notes on the Application of WRC-107 and WRC-297 LAC, Peng! ‘The welding Research Council (WRC) Bulletin No. 107 {1] and Baitetin No. 297 [2} are two of the most important desiga, ‘guides ever published for the design of pressure Vessels. Wide- ly used in the design of attachments and nozls comaections, they have become indispensable tools fs recent years. WRC-107 and WRC-297 are lavaluable due to their come bined consideration of theory, experimental data, and engineering judgment. In sores instances the theoretical values have been adjusted several hundred percent upward to match the avallable experimental results. They are reliable tools ‘without which the so-called “Design by Analysis” eppreach would have been impractical, Unfortunately many designers have misapplied the data presented, thus resulting in inconsls- tent designs. This article describes the data availabie im these ‘wo bulletins acd explains the aatuse of inconsistency incusred in some designs. A supplemental formula is then developed for calculating the combined maximum stress intensity to be used i designs. WRC Bulletin No, 107, WRC-107 was frst published in August 1965 [1]. It was bared on Professor Bijicard’s Sheoretical work, with some adjustments, made based on avtilabie experimental data, A few revisions have been made ince is frst publication, The latest revision made in March labeled some"of the curves. WRC-107 has been used widely in the design of vessel nozzles and attachments. It was ene of the major driving forces in promoting the “Design by Analysis” philosophy. The bolletin covers bock spherical ang cvindricat vessels, In spherical vessels, the original theoretics! work was based on courd rigid inserts and round nozzle connections, Square jnsens of conneccions can be anelyzed using an equivalent round attackment having a diameter equal to 8/7 of the ar tachment width, On the other hand, the original theoretical work for the cylindrival vessel was Based om square and ree- ‘angular-shaped uniform loads acting on unpecforated vessels. Round autackments can be analyzed using an equivalent square having the width of 7/8 of the attachment diameter. Besause of the assumption of the unperforated shell, the ap- “Peng Engineering, Houstun, Tex 7083, CCootsbured bythe ressare Vessels nd Piping Dison for publication bs tke Jovnsat or Puzssuns Visas Tacaococr, Masur secaived by the PVP Divion, December 21, 1987. 108 / Vol. 140, FEBRUARY 1988 plicability of the cylindrical vesset portion of the bulletin is United to rigid inserts, The Bulletin does not recommend any specific method in analyzing an actual nozzle connection in the cylindrical vessel i is left largely to the designers to make heir own judgment. ‘WRC-107 presenis detalted tabular forms for calculating auresses et four major axis locations. Stresses at both inside and outside surfaces on these locations can be readily cateulated following the step-by-step procedure ovtlined in the form, The final results are the total skin stress intensities at shese four locations in the sbell, No separate membrane siress Intensity Is given, nor is the stress in the nozzle caicuiated. WRC Bulletin No, 297. WRC-297 was published in August 1984 (2). It is a supplement to WRC-107 end is specifically applicable to oslindrical nozzles in cylindrical vessels. This bullerin was based on Professor Stecle’s theoretical work, 41 gives data for larger D/T ratios than in WRC-107, and ‘also provides beiter readability for small values of d/D by ploning the curvesusing h=@/VBT" asthe abscissa. Most importantly, the new theory considers the opening on the shel together with the restraining effect of the nozzle wall. This is 2 betier model than the unperforated shell used in WRC.107 for simulating the nozzle connections ia ylindrical vessels. Because most pressure vessel and piping codes have ferent allowable stress criteria for differant siress categories, ‘WRC.297 emphasizes the separation of membrane stress end skin bending sttess, Two examples are given oudining the detailed procedures for calculating membrane aud total stress intensities at points located in the longitudinal and the ‘ransverse planes. Both shell and noze stresses are calculated, Loestion of Muxiaum Stress ‘The fact that the examples given in both WRC-107 and WRE-297 specifically outline the procedees for ealeuiating stresses at the four major axis comers, has led designers 10 ‘think thar the maximum stress inthe conection must be ome of those stresses. This presumption iatroduees inconsistency ‘aad nonconservatisan ia the desien of nozzle and arachment connections, The maximam stress fs got normally focated at these comers. Although the calculation involves only she secondary stress which itself involves various uncertaitties, & Fig.t Versel attachments Transactions of the ASMEa oo er Fig. 2 Stogs on sphestcal shell certain emount of deviation is always expected. For instance, as much as a 10-percent difference may be made just from the chart by different persons, He vever, just because ‘of is inherent unceriainty, effort is needed to meke it 2s con- sistent 2s possible. Anything that can be done to improve its certainty ‘should be done. In cases when deviation is ‘unavoidable, its preferred to deviate on the conservative side. To demonstrate when an incousistency may occur, the at- tachment on a spherical abell can be used 2s an example. Figure 2(¢) shows @ bending moment, M, acting on a nozzle conection at @ spherical vessel. By choosing & coordinate system as shown in Fig, 2(0) the soresses at the four mrajor axis Points ace SA=S, SB=S, SC=0, aud SD=0. However, if a designer happens to have the coordinate system set up as shown in Fig, 2(), the applied moment will be decomposed in- fo My=M,=0.7G71M two components. The stresses at the four major axis points, in this case, are SA=SB= SC=SD=0.7071S. This stress is about 40 percent below the expecied maximum stress. The same nazzle connection and the same applied moment, yet the calculated stresses are substantially different depending solely on how the coordicate system is setup. From the foregoing demonstration, itis clear that in Fig, 2(¢), the maximum stress is not located at the ma- Jor axis zoints, but at the off-axis points P and N. In general, Hf the moments acting sround both coordiaate axes are nonzero, the maximum stress is not located at the major axcs. ‘This warning was properly stated in Paragraph 3.3.5, WRC-107, which said, “However, in the general cast of ar- bitrary loading, one has no assurance that the absolure imax am sess intensity in che siell will be located at one of the eight poims considered in the above discussion.” The eight points mectioned are the inside and outside surfaces of the four major axis points. Strangely, this message has been large- 4y ignored. Maximum Stress Intensity ‘The calculation of the stresses at the four major axis points is helpful in understanding the stress distzibuden, However, for the design purposes itis essential to calculate the maximirm stress intensity occurring throughout the entire connection. In developing the calculation procedure, the following notations are used: P = radiat load Mg = circumferential moment (or moment in [-1 disection) ‘AM, = longitudinat moment (or moment in 2-2 dicection) Mg, = total bending moment ‘My = torsional moment Vz = circumferential force {or shear force in 2-2 direction) Journal of Pressure Vessel Technology ¥, = longitudinal force (or shear force in 14 direction} V = total shear force Sy(L) = 1 disection J category stress due to lead Zs ier, 8, for tadial and circumferential, respective. ly; /= Mb, for membrane and bending, respectively SS m shear stress ‘The applied ioading cun generally be divided inco four com- pouent groups. They are radial force, bending moment, tor- sional moment, and shear foree. Im the following discussion the stress contributed by ezci componeat group is described frst. They are then combined to become the maximum siress. The discussion follows the WRC-297 stress oriencation of radial and circumferential directions with respect to the n02- ile, Because of the different stress orientation edopied in the cylindrical shel! portion of the WRC-107, reorientation of the WRC-107 data is requited for the cylindrical vesse’ The pro- cedure shows the method for calculating the stress in the shell, ‘The same procedure can be used for calculating the siress in the nozzle. Stress due io P. ina spherical vessel, it is Obvious that the stresses created by the radial loed are uniform around the ex tire attachment circumference, In a cylindrical vessel, though the stresses differ from location to location, it can also be regarded as uniform taking the maximum around the attach- reat cirsumference as te uniform value, This asussptioe in troduces some conservatism but is not overly conservative, ‘WRC-297 has already adopred this approach in developing the ‘design curves. In reference to the aztachment orientation, the stress created by the radial load caa be written symbolicaly 2s Membrane sess: Sym (P), Soq(P) Bending stress: 5,,(P), Sy(P) ‘They are constant around the entire attachment ci- cumference. a Siress due to Mc and M,. Steesses created by Mf and M, are not independent os assumed by some daiigners, In a spherical vesel the M, and Mf, can be conveatently combined as MgaVHOTHD @ ‘The combined stress can then be calculated based on M, rather than on Mc and M, individually. However, to haves sommion methad appiicable ro both spherical and cylindrical vessels, @. more general approach is preferred. Professor Bijlaard has shown that the stress, due to bending moment, varies according to the casine function in a spherical shell. As showa in Fig. 3(o), when the attachment is loaded simultaneously with Mc and. Mf, each M, aad Af, commands a cosine shape stress distribution. With this type of cosine distribution, it can be shown that the niaximum combined [a2 Somericat Yesse! (03 Cyuaeient eoens Fig. 3 Combined stress dus to bending moments FEBRUARY 1988. Vol. 110/107siressis Sy6bty) = VB OHI + Sy @ ‘The foregoing maximum stress located somewhere iaside the 0-50-deg segment, There is another maximmurn stress, with @ reversed sign, located within the 180-270-deg segment. In aeylindtical vesset, the stization is complicated by the ir- regular distribution of the stress. The siress due to Mc is istributed in a shape close 10 a shifted cosine curve, but the siress due co Mf, is humped coward the neutral axis. Due to this off-axis peaking, it appears shat an absolute sum may have to be taken to caleulate the combined maximum stress. Nevertheless, in consiGering the fact that the stress file dhe to ‘Mo is considerably narrower than & cosine distribution, equa~ tion (2) can still be used for cylindrical shells with good representation, In fact, this equation has been used by the pip- ing code [3] since the 1950's. ‘Since the purpose of the calculation is to find the maximum siress intensity, the relative signs between the radial stress and the circumfereatial stress is important. Fortunately, this sign reversal only oevurs at some of the circumferential mem’ “ane forces in WRC-297, One way of maintaining the sign is to take the $M) in equation (3) the same sign a5 that of the greater Sp(Mc) and 5,(M;). Even with this sign-preserving arrange- ment, the maximum membrane stress intensity calculated may stil be smaller than the ones caleulated at the four major axis points, However, the difference is insignificant. The stresses calculated at the four major axis points stil] need to be considered. Combined Normal Stress, The combined mamum nor- mai stress js determined by P, Mc. and M,. Since the stress ‘due zo P is uniform all around the attachment cfrcumference, ‘we can simply write 5,55 ,(P) +54 (My) (a) SS (P)~SpMy) (ab) Equations (4a) and (4) represent the maximum normal ‘uresses at she two maximum points located on apposite sides of the attachment. Euch equation further represents two stresses one at the outer, abd the other the inner surface of the shell. These four locations are to be checked for the maximmura sucess intensity. ‘Shear Stress due to My. The shear stress duc to torsional moment is uniform all around the attachment circumference. ‘This stress can be expressed as SSM), Shear Stress due to Ve and V_, The shear siress due to Vg and ¥, can be combined by $8() NSS oI + (SSE o Toiat Shear Stress. "The ‘otal maxinnsn shear stess isthe absolute sum of the shear stress du to torsion anc the shear sicexsduc to combined shear force. That is, S5=S8(Mp} +85(FY 6 ‘This maximum shear stress generally does not occur at the same locaton as the maximum normal stress. However, ince she shear ses is insignificant in most of the cases, i¢ can be conservatively considered as occurring at the same location Where the maximum normal sess ceetrs, Maximum Stress Intensity. ‘The stress intensity can be caleulaied by the maximum shear stress theory using the nor- sal sivess and shear stress calculated by equaitons (4) and (6), respectively. The WRC bulletins have given detailed formulas for this cafeulaticon. A to:at of four stress intensities repre- senting the maximum and minimum stress points and both oucside and itsidé surfaces should be calculated. The miax- imum valueis then used for the design. To satisfy certain Code 198 Vol. 110, FEBRUARY 1988 {] requirements, the maximum membrane stress intensity and the total stress intensity may also need to be separated. ‘Conclusions Regardless of the warning given by the WRC Bulletin 107 that there is no assurance thatthe absolute maximum sicessia- tensity in the shell will be located at one of the eighe points (four major axis points each having outside and inside sur- aces) considered in the example calculations, many designers, stilfuse only the stresses calenlated thece for design. ‘This prac- ties creates inconsistencies in designs and may introduce as much a5 a 40-percent nonconservatism. The present article cutlined the procedures for caleulatiog the maximum stress ine tensites both at and off the major axis points. This maximum stress intensity should be used in the design evaluations References | Wichmas, K. R, Hooper, A. G., and Merton J Sonsial wed Cyintteal Shel Due to Emeenal Leading, 40%, Aug, 1965, reviyat Mar. 1979, 2' Merson, J. 1, Mothurias, K., Raman, G. V., and Rocabovgas B.C, Local Sues i Chilieal Sheus Due wo” Enceral Loacings on Nain Sepnenent 0 WRC Boia, 1h" WRC Buln Ade ANSI Code for PresuePipig,ANSU/ASME B3L.3 Chemical Plt and Pevoleum Refinery Piping, ASME, New York, 1584 “TASME Daler and Presure Vee! Code, Div. 3 akerasive Rules, ASME, New York, 183. ‘acl Sarees in RE Bulan Ne. tt, Pressure Vests, DISCUSSION BR. Natarajan ‘At the outset, I would like to congratulate the author for bringing owt certain important points which a designer sometimes forgets while using design chars. However, there are some points which are worth mentioning about this paper: 1 While discussing the inconsistency about the location of the maximum stress ia a nozzle-spherical sheet intersection, it is expected that the designer will define the geometry and the loading using the same coordinate system. The location of the ‘maximum stress, and hence the inconsistency in defining the maximum stress lovation, is due to the misunderstanding by the designer and not dus to the examples given in WRC-107 or WRC297. 2 While calculating the combined stress due to bending moments, mention should be made that the flexibility of the nozzle has not been completely considered. Further, the boun- dary coaditions at the nozzle and cylinder ends also affect the value and location of these maximum values. K, Mokbtarian? T have the following general comments to make on Peng's Paper: 1 We have found that generally the maximum stress due to longitudinal moment occurs at the O-deg azimuth. We do ‘not agree with the shape of the viress curve due to M, im Fig. 316). 2 The last three sentences of the fast paragraph in the subsection ‘Stresses due ta Mc and Mf," are not clear and ap- ear to contain conflicting statements, . 3 Normally, the designer has to face the question of com- Dining the stresses due to presture with these due to “Fbankato Sint Uoiversiy, Mechanical Engineering Deparment CBE Ne-Con, in.. Oak rook Engineering Transactions of the ASMEinéchanical loads. 1 do oot know of any simple way of pro- Vicing those guidefines now, but eventually this question will Save to be adaressea ZF. Saag! ‘AS stated in the paper’by L. C. Peng, WRC-107 and WRC.297 published by PYRC are excellent references for calculating fecal stresces in nozzles and attachments, indeed, they are widely used im the design of pressure vessels and have become incispensable tools. ‘The author summarizes inconsistencies occurring in some Gesigns due to the designers misepolying the data presented in the aforementioned two documents. He also presents a ‘method and procedure for calculating the maximum stress 3 tensity. This is of importance and needs to be understood by Cesigners. It should prove to be an aid in applying the two documents correctly. {asm in agreement with Dr. Peng’s opiaion about the incon sistency end nonconservation, which will be created in the desigr procedure if a designer caunot determine the maximum -suress intensity. [n the paper, the formula which is developed for caiculating intensity seems to hold only for round radial nozzles and attachments on spherical shells. Only in this case are the stresses due to radial load P and torsioaal moment My uniform. For other shapes, particularly in the case of a ree- tangular attachment on a cylinder, the szesses are not uniform long the perimeter of the attactment. In the section “Location of Maximum Stress" the author states that “the calculations involve only the secondary scess.” From a stress classification point of view stresses due to external foad on an attachment include not oniy secondary siresses, but also primary ones. This is important, because there ure different allowable stresses associated with different stress careyories. “TJeieruty of tlesi ar Chiago, Mechanisat Eaginceing Departnear Whe above and dhe pressara Unconporeed 0s ‘Wish reference to the calculation of Ue maxitaum stress in- tensity, it is noted chat the maximum shear stress generally rot located at the same poist where the maximua normal stress occurs, But the author assumes that they do cceur at the same iocations, Is this a conservative assumption? AUTHOR'S CLOSURE In thanking Messrs. R. Natarajan, K. Moktuarian, and Z. F. Sang for the valuable discussions, the author would ike ro make a brief closure. This paper's main concern is the misapplication of the bulletins, not the validity of the bulletins which are excellent works. The nozzle flexibility and the vessel end condition, just as other geometsical parameters, have definite effects on the stress shape. The main point is If the interaction exists be- ‘owen the Lwo moment components, ‘The off-axis peak stress due to M, may not exist on sail AD vessels, bur it does exist on other vessels, as demonstrated by Prof. Steel, and various pipe brauch tests. There is indeed some confusion in the last threr sentences concerning the stresses due to Mg and M,. Because of the combination method proposed, the stress loves the orientation after tne calculation. With’ the proposed sign tracking method, the maximum calculated membrane stress Intensity may be occa- sionally smatier than the siress calculated at the four major corners. One way to correct the problem is to reverse one of the stress signs when the situation is detected. The author ‘agrees that there is no simple way to combine the pressure and. the mechanical load effects. Publication of some of the NRC upproved methods, for instance, should be encouraged. ‘The secondary stress mentioned by Dr. Sang should have ‘been more accurately stated as locsl stress. The inclusion of higher shear stress is always conservative in the calculation of the stress intensity when it is taken as twice the maximum shear stress. meubvane Strestes have beer Simelex and LOcAls aud hocale,