PIPING VIBRATION AND STRESS by J.C. Wachel Manager of Engineering Engineering Dynamics Incorporated San Antonio, Texas J.C. Wachel holds an MSMB agree from the University of Texas. He has been with Southwest Research Institute Since 1961. His activities have centered in the fields ‘ of vibration, pulsation, @ynamic simulation, acous- tics, and fluid flow prob- lems. “He has developed procedures which are used to control piping vibration in systens Subjected to acous— tical pulsations. He also was instrumen- tal in the development of techniques for predicting and controlling compressor man- Told vibrations. In recent years, he has specialized in the analysis of vibration and failure problems in rotating machinery fle is a member of Tau Beta Pi and Pi Tau Sigma. ENTRODUCTION To even a casual observer, a most ob- vious effect of pulsations is'that it forces piping and other plant systems into Sustained vibrations and, under some condi- tions, the vibrations can cause fatigue failufes at critical, high bending stress regions in the mechanical systems. The éxistence of such pulsation~induced me- Ghanical vibrations suggest two obvious approaches to control and one approach “hich is perhaps less obvious. These are: 1. Supply mechanical restraints which will prevent movement of the pipe. 2. Eliminate tions. ¢ control the pulsa- 3. Bliminate the coupling of pul- Sations as forces into the piping. While each of these approaches are valuable, no one approach is optimum in all cases and any one by itself can prove excessively expensive. The cost of me- Ghanically restraining compressor piping Or overhead plant piping, for example, Soon causes the engineer to seek help’ from Other control approaches. A similar situ- Seion exists with pulsation control. If pulsation suppressors are designed to Eliminate “all” pulsations (i.e., to a level that any piping system could be utilized) then it is soon found that pres- gels of excessive size are re~ The concept of decoupling the pulsa~ tions from forcing the mechanical systen into vibration will be discussed in a later section, but basically it involves control- ling the location of bends, constrictions and piping discontinuities relative to the pulsation standing waves. It is difficult, for example, to excite an infinitely long, Straight, constant diameter pipe into vi- bration from internal pulsations. In more tic piping configurations, however, ere are also things that can be done to minimize pulsation shaking forces. These, foo, Will be described later in this chap- ter! While each of the above approaches are useful in controlling known piping vibra~ tion problens, two fundamental questions renain which can drastically reduce the tine and effort involved in field fixe: 1. Are you aure the vibrations are excessive and require reduction? 2. What can you do at the design stage to prevent the problem? again, these will be dealt with in subsequent sections involving "Criteria" “pield Testing” and "simulation Techniques for Predicting Pulsation Induced Vibra- PIPING VIBRATION AND STRESS CRITERIA one of the major reasons why pulsation trol alone should not be used to control flow-induced piping vibrations lies in the fact that there are no pulsation criteria which can be reliably used for preventing Vibrations. In spite of the fact that many Buch criteria have been evolved, it is not pulsations per se which are the problem, Bue rather the dynamic stress levels which result in the pipe wall. Whenever vibra~ tory stress exceeds the endurance level of the material, piping failure is imminent. By similar argument, it can be seen that vibration amplitude criteria for piping systens are likewise dangerous and, Egain, are fundamentally the wrong approachunless consideration is given to the con- figuration and dimensions of the piping systen being considered. ‘The technical literature is replete with vibration cri- teria for plant piping, machinery, and structural systens which specify ‘allow~ able” vibration amplitudes as a function Of frequency as shown in Figure 1. Such criteria are based largely upon the experi- ence of field personne! operating or work- ing with such equipment. While they may be applicable in a statistical sense to average or typical piping, they are funda- mentally incorrect because they do not con- fider the configuration involved. As such, they introduce considerable risk when used in evaluating any specific piping system as they may result in a degree of design confidence which is unwarranted by the de~ sign procedure used. Although the criteria are based on typical or average conditions, they do not normally contain such a warning or supply a definition of the linits of what constitutes average or typical. ‘The problem with any such criterion is a not so mich that it is not applicable to many plant systems but rather the cost of failure and downtime in those cases in which st does not work. While the statis~ tical data from which the criterion was generated proves it works in most cases, the risk that it may not work for the next design should often dictate a more thorough analysis. Note that the criterion as pre- sented does not differentiate between a Stiff compressor manifold system and a flexible scrubber lead line. If the crite- zion is sufficiently conservative to pro- tect the compressor manifold system, it will normally be overly conservative for the lead line. Tt should also be noted that the stress level in a pipe is a func~ tion of physical distortion only (i.e., strain}, and is not a function of frequency for the general case. If frequency is to be one of the controllable allowables in vibration amplitude, it must include cogni~ zance of the tyse of piping span involved and its resonant frequency and node shape, as discussed below. Figure 1 Allowable Piping Vibration Levels Note: Indicated vibration limits are for average piping system constructed in accordance with good engineering practices. Make additional allowances for critical applica~ tions, unreinforced branch connections, etc.Stress Predictions in Idealized Pipe Spans For any span configuration, it is fea- sible to calculate the stress which re- sults from a given deflection (stress per mil, of 8/y), providing the ond conditions Bnd vibratory mode shape (deflection pro- Eile) are known. ‘The general equation relating maximum stress in a pipe to the maximum deflection @iong the span is given below for the low- Gat vibratory resonance mode (note that the maxinum stress and maximum deflection are generally not at the same point): ae her: Stress at maximum stress point, psi (1b/in2) B= Blastic Modulus, 1b/in™ D = Diameter, in e®y/ax? evaluated at maximum stress point y = Deflection, mils, at maximum deflection ‘point span length, in Span length, ft Proquency factor for steel pipe, this becones 6 fee soe (earsrr Smell fy" 32 [ait 5 y De psi stress/mil uy Getiection = 104.17 Solution of this equation for several span configurations is given in Table 1. This table can be used either: 1, To determine the stress resulting fron a given deflection (at the naximin deflection point) or 2. To establish maximum allowable deflections. Non ideal Beans ‘Table 1 assunes idealized end condi- tions. As described in a later section, S typical straight continuous span with Strap and pier supports most nearly matches the resonant frequency prediction Of the £ixed/simply supported beam (A = {5.42). in field situations where the Table 1, Constant Factors for Calculating the Stress Por Mil (s/y) in various Pipe Spans Bem Type Fixed-Fixed sly = 366.3 cantilever — aly = 1028 Simply ‘Supported aly + 2128 2, ;——— u Fixed/Simply Supported 2 sly = 1731 hie L-tend 1/3 sly = 3600 Ls aeBend us v-tend |L/3 aly = 2836 7 ws lowest resonant frequency of the span can be measured (as by bumping with a cross~ tie), the stress per mil can be adjusted to Conpensate for nonideal supports by the following equation (for straight spans only)?ts/y) (s/y) actual ~ calculated where: resonant frequency SCF = stress concentration factor, as may be applicable to the point (fitting, etc.) wh maximum stress occurs For ideal simply supported spans, the above linear relationship between stress and frequency may be as much as 50% high (conservative), but for other end condi tions accuracy'is generally within about 58. Vibration Criteria Tt was noted in the first section above, that generalized piping vibration eriteria are fundamentally incorrect un- less configurations and dimensions are in eluded. This section will therefore in- Elude these considerations and generate 10? FOR UTS < 80 KSI VALUES OF 5, KSI new vibration criteria, at least for some piping configurations. API Standard 618, "Reciprocating Com- pressors for General Refinery Services", 2nd Edition, 1974, in Section 3.3.2.1.a., States that the vibration induced cyclic Stresses should be less than 26,000 psi peak-to-peak for steel pipe below 700 F This criterion is based upon the curve given in Figure 2, "Allowable Amplitude of Alternating Stress Intensity, Sa.", given in ANSI USA Standard 531.7, ‘Nuclear Power Piping” and other ASME codés. Extensive use of these curves has shown then to be conservative even when the combined steady State stresses introduced by pressure, thermal and weight loading are near the yield stress. Based upon some 25 years of experience with piping vibration and failures, SwRI has developed vibration amplitude versus frequency criteria (Figure 1) in lieu of a nore exact technique for estimating the vibratory dynamic stress in specific piping configurations. The disadvantage of cri- teria such as given in Figure 1 is that if they are conservative for stiff compressor manifold systens they can be overly con- servative for long flexible lead lines. INTERPOLATE FOR UTS a0 115 KSI €=20X 109 KS! [FoR urs 1s 120 xsi 2 0 2 10 ote oes ioe Oo ca t08) NUMBER OF CYCLES Figure 2. Allowable amplitude of Alternating Stress Intensity, Sa, for Carbon and Alloy Steels With Metal Temperatures Not Exceeding 700°FThe importance of configuration is i1- dustrated by comparing a cantilever pipe section with a fixed-fixed span or L-bend Of equal lengeh. Obviously, the stress generated in the cantilever span due to a {finch end deflection is different than in that generated in the other configurations by an equal deflection. Tt is also appar— rene that a lower stress will be generated ina long span than in a short one. An in- Yestigation into the dynamics of such spans Shows that the variation in stress per unit Geflection tracks rather directly with res~ Gnant frequency for a given span type; ew @ long flexible span has low stress per Unie deflection and low resonant frequency. This frequency variation may be used advan~ tageously to normalize (non-dimensionalize) allowable stress criteria. For exanple, the usual allowable deflection vs. fre~ Gueney plots could be made substantially fore accurate if the abscissa wore changed from vibration frequency to fundamental span resonant frequency. ‘the vibration allowable deflection cri- teria for L-bené piping spans is given in Figure 3 for first and sccond mode resonant Vibrations. Note that when the usual dis- placenent criterion is multiplied by fre- Guency, an almost flat, horizontal crite dion Curve results and’the product of vi- bration amplitude and frequency is, of course, vibrational velocity. The’approach Used in developing these criteria follow the analysis procedure described in the UNEQUAL LEG L-BemD STRESS, PST aa ky xk, 107 pst/iPs. v= Velocity, in/sec Ind Mode (Long Leg Fixed End) Ise Node (Short Leg Fixed End) Figure 3. Allowable Deflection Criterion for Bll Bends, For First and Second Mode Res- onant ‘Vibrations (Steel Pipe)preceding section, wherein the deflection Fequized to produce 13,000 psi bending stress is Used as the Standards of accept- ability (i.e., when stress equals the en- Gurance limit ‘of the steel.) Similar criterion curves are now being generated for other piping span configura- tions as a part of the SCA Research Pro- gram, and a nomograph is being prepared to compute stress as a function of deflection for a broad spectrum of span configura~ tions. DEVELOPMENT OF VIBRATION AMPLITUDE AND VELOCITY CRITERIA ‘The natural frequency of a uniform beam can be calculated by any of the fol- lowing equations: EI ue fe ( bin es = fe mz Vy there: waa y = Density 1b/in? A= Metal area, in? ke = 0.34 D for pipe (see Figure 4) pind Using the expression of 0.34 for the radi- us of gyration, Ys fae (v.40 fem Vi (2°) (note 4 is in inches.) For steel pipe this becones: ap. = if L is in feet. a= 76 4B, sek e Solving for D/L? Db i aft Fer Substituting this into the stress per mil equation: shy = Be Wy = iar Bat Ll s . z yy = Lane d Figure 4. Comparison of Approximations for Radius of Gyration Ver~ sus True Value for Various Pipe sizes For sinusoidal vibrations, vibrational ve- locity (v) is 2rfy, and the stress equation can be written in terms of stress per unit of vibrational velocity: s_ _ 21900 : ag - Baan Se z 2 as.2 $ In another form, s . x! 2, where x’ = —E y 7 KR gg where © tar ToS a. xis & Se asx ios E s « ig ataee oe ‘Table 2 gives the allowable stress per ve~ locity’ for straight beams and equal leg bends vibrating at their lowest resonant frequency. Note that the range of stress per velocity only ranges from 218 to 370 psi/ips. If the fixed end stress coeffi- Gients are used (275 psi/ips) then this Value would be within 30 percent for theTable 2. a ‘ ; tea 15.6 2128 co 1028 ceo 366 Ph, 20.4 3600 mininum and maximum values. If the stress fer velocity 1s equated to the maximum al- Yowable dynamic stress (Sall = 13,000 psi Oop), then the allowable velocity (Vall) is obtained from: sal. = 23,000 vall = qygyall = “275 = 47-9 Ses Configurational Corrections in order to apply the criteria to a real piping system, the stress concentra~ tion factor and other reduction factors Such as correction for concentrated eights, non-ideal end conditions, changes qn pipe diameters and effect of vibration node shape must be taken into considera~ tion. sall vall = qTeyqyall C, C2 C3 Gy CS Correction factor to compensate for the effect of concentrated weights along the span of the pipe. cy = Stress Concentration Factor. cy = A correction factor accounting for pipe contents and insulation i = 0+ E Wp = Weight of pipe contents per unit length. = Weight of pipe per unit length. Wing = weight of pipe insulation per unit length. 28 39.6 169 28 59. 14.9 ua 33.7 13.4 an 39.2 98 370 3 a8 cq = Correction factor for end condition different from fixed ends and for Configurations different from straight spans: cy = 1 for straight spans fixed at both ends. 0.75 for cantilever and simply supported beams 1.35 for equal leg 2-bend. 1.2 for equal leg U-bend. cy = Correction factor to compensate for Vibration mode shapes other than the first. Based upon SwRI experience, a stress concentration of 4 is appropriate for welds fh branch connection without being overly Conservative, Plots of the correction fac- for ¢] are given in Figure 5. In most Gases the contents and insulation weight is {ess than the pips weight itself so C3 is generally less than 1.5. Applying the stress concentration, the allowable velocity becomes: 13,000 vall = 375ca) 12 ips ‘This would apply to uninsulated pipe vibrating at resonance in its fundamental mode. If the maximum effect of concentrated weights (which is approximately 8), pipe Contents, anda safety factor of 2/are used: vall i 3 ies (002g cael2 Higher Mode Vib: A tabulation of the stress coefficients for higher modes and for some straight beams with concentrated weights equal to » the pipe span weight are given in Table 3. Table 3. Higher Node Stress Coefficients A and Deflection/Stress Relation sf ships for Various Span Config- i (where: concentrated Weight (P) = Pipe i, weight.) i i ten tan © Pn 5 RASS se he oe wm ee Be eel eee ts Ret me fee on oF The data can be used to det: piping stresses for the span by using the hhaxinum measured deflections or vel in the following equations a sex & v) °) e = Stress, psi Figure 6. Allowable Piping Vibration D = Pipe 0.D., inches Levels with Velocity CriteriaL = Pipe length, ft. y = Maximon deflection, mile v = Maximum velocity, ips = stress/deflection, psi/mi! 5 /aeflection, pai /mil § = stress/velocity, pai /ips REFERENCES 1. Wachel, 7.C., SGA-PCRC Seminar on Controlling Effects of Pulsations and Fluid Transients in Piping Systems, Report No. 160, Chapter VI, Novenber 7-9, 1979. Wachel, J.C., von Nimitz, Ws, “assuring the Reliability of of¢- shore Gas Compression Systems,” EUR205, European Offshore Petroleum Conference and Exhibition, 1980 Proceedings, Volume 1, pp. 559-570.