Scribd Logo
Hochladen

Willkommen bei Scribd!

  • Men 0593

    Hochgeladen von

    fileseeker
    18 Ansichten28 Seiten
    MEN 0593

    Originaltitel

    MEN 0593

    Copyright

    © © All Rights Reserved

    Verfügbare Formate

    PDF oder online auf Scribd lesen

    Stufen Sie dieses Dokument als nützlich ein?

    Sind diese Inhalte unangemessen?

    Dieses Dokument melden
    MEN 0593

    Copyright:

    © All Rights Reserved
    Als PDF herunterladen oder online auf Scribd lesen
    Markieren Sie unangemessene Inhalte
    18 Ansichten28 Seiten

    Men 0593

    Hochgeladen von

    fileseeker
    MEN 0593

    Copyright:

    © All Rights Reserved
    Als PDF herunterladen oder online auf Scribd lesen
    Markieren Sie unangemessene Inhalte
    von 28
    Mechanical Engineering IN Kei) For the Power, Petrochemical and Related Industries "The COADE Mechanical Engineering News Bulletin ix published periodically from the COADE offices in Houston, Texas. The Bulletin is intended to provide information about software applications and development for Mechanical Engineers serving the power. potrachemical. andelated industries. Additionally the Bulletin will serve a the official notification vehicle fr software emorsdiscoveredinthose Mechanical Engineeringprogranms offered by COADE, (Please note. this bulletin is published only two to Lee times per year. As ofthis stu the issue number will appear inthe banner block athe top ofthe First page.) ‘CAESAR I Version 3.19 has recently shipped to users: in our effort to mintsin COADE as the leader in the Geld of piping engineering software, we have made over 60 modifications 10 the ‘rogram, Several new features. suchas the missing mass comretion «¢p.8), the B31G code (p. 27), and virtual memory support (p.2) are discussed in detail in this bulletin, Several usoful features, added to Version 3.19 t00 late 10 document in the release notes, are Seseribed on page 3 {In adivion tour usual tchnical aticles, we also are using this bulletin to discuss recent developments at COADE, as well 3810 prowide a preview of the enhancements we have "inthe works” for ‘our piping and vessel software. We continue to welcome any articles hy asers af cur software for inclusion in future isses of Mechanical Engineering News, New Features of CAESAR II Version 3.19, * Dynamics modules provide Missing Mass and Scis-| mie Anchor movements. BIG eriteria for corroded pipelines added RCCMC piping code added, High Resolution Animated Graphics. All dynamics modules are now in 32 bit “protected” ‘mode, yielding substantial speed increases. ‘Thermal bowing loads can be analyzed. “Muliple analysis via Batch Stream Processing. * Additional expansion joint and spring hanger manu facturcrs data bases added, ~ French and Spanish language support. Additional output reports. COADE, Inc. ola} May, 1993 TABLE OF CONTENTS PC Hardware forthe Engineering User (Part 16) CAESAR II Version 3.19 Undocumented Features What's new at COADE. COADE Bulletin Board _ CAESAR I Quality Assurance Mana. AO? Pressure Vessel Cade Updates CAESAR II Development Schedule Fe ipe Version 2.7 Announcement. COADE Seminars Technology You Can Use CABSAR IL, Use ofthe Missing Mass Correction Option in Spectral AMAIYSIS oo... a ‘The Role of Sess Analysis in the Life Extension of Piping Systems - Evaluation of Creep Stresses... 18 “The ASME B3I-G Criteria a) ‘The CAESAR I Expansion Joint Moder o...-29 Expansion Case for Temperatures Below Ambient... 32 CARSAR II Specifications smn PC Hardware for the Engineering User (Part 16) AS many users are aware, CAESAR II is being converted {rom 16 bitoperationto 32bitoperation. There are two main advantages to this conversion: first, the 32 bit modules can e the extended memory found on the computer, and second, there is significant speed increase due tothe 32 bit addressing, This conversion of the CAESAR II source code involves switching from the Microsoft FORTRAN 5.1 compiler to the WATCOM FORTRAN 9.0 compiler. The WATCOM compiler utilizes the 32 bit DOS extender from Rational Systems. An item of importanceto CAESAR Husersisthe fact that this DOS extender has the potential to support “ital memory”. By activating this virtual memory op- tion, the computer system can use the hard disk as main memory, just like RAM, only slower. This gives all, CAESAR Tl users the ability to analyze a system requiring COADE Mechanical Engine ing News May, 1993 CAESAR II, Use of the Missing Mass Correction Option in Spectral Analysis During spectral (ether seismic or force spectrum) analysis, the response of asystem undera dynamic load is determined by superposition of modal results. One ofthe advantages of this type of modal analysis is that usually only a limited ‘number of modes are excited and need to be included in the analysis. The drawback to this method is that although displacements may be obtained with good accuracy using. only a few of the lowest frequency modes, the force, reac tion, and stress results may require extraction of far more ‘modes (possibly far into the rigid range) before acceptable accuracy is attained. CAESAR II permits the user to specify — either through a ‘mode number cutoff ora frequency cutoff — the number of ‘modal responses to be included in the system results. Ifthe analyst is more interested in providing an accurate represen- tation ofthe system displacements, it ay only be necessary to request the extraction of a few modes, allowing a rapid calculation time. However, if an accurate estimate of the forces, stresses, etc. in the system is the objective, calcula- tion time grows as it becomes necessary to extract far more ‘modes. This is particularly true inthe case when solving a fluid hammer problem in the presence of axial restraints; ‘often modes with natural frequencies of up to 300 Hz must be extracted. How does the analystknow how many modes are sufficient? ‘The usual procedure is to extract a certain number of modes and review the results; then to repeat the analysis while extracting 5 to 10 additional modes, comparing the new results othe old. Ifthere isa significantchange between the results, anew analysis ismade, again extracting 5to 10more modes above those that were extracted for the second analysis, This iterative process continues until the results taper off, becoming asymptotic, as shown in Figure 1. ed Figure 1 -- Maximum Stress vs. Extracted Modes ‘This procedure has two drawbacks, the first one obvious — the time involved in making the multiple analyses, as well as the time involved in extracting the potentially large number of modes. The second drawback i less obvious — degree of conservatismis introduced when combining the contributions of the higher order modes. Possible modal summation methods include RSS, ABSOLUTE, and GROUP —all methods that combine modal resultsas same sign (positive) values. In reality, theory states thatthe rigid modes actually act in phase with each other, and should therefore be combined algebraically, thus permitting the response of some rigid modes to cancel the effect of other rigid modes (ths is exactly what occurs in atime history analysis). Because of this conservatism, it is actually possible to get results which exceed twice the applied load, despite the fact that the Dynamic Load Factor (DLF) of an impulse load cannot be ereater than 2.0. With Version 3.19, CAESAR IT has introduced a feature, called the “Missing Mass Correction”, which helps solve these problems. This feature offers the ability to include a ‘correction which represents the contribution of the higher ‘order modes not explicitly extracted for the modal/dynamic response, thereby providing greater accuracy with reduced calculation time. When the option is activated (by entering YES for the appropriate dynamic control parameter), the program automatically calculates the net (in-phase) contri- bation ofall non-extracted modes and combines it with the ‘modal contributions — avoiding the long calculation time associated with the extraction ofthe high order modes and the excessive conservatism of the summation methods. This article explains this feature in detail, describing the ‘mathematics behind it, providing instructions on its use, ‘demonstrating its effect through an example problem, and finally discussing the Included Mass Report, a new report added to the Dynamic Output processor. Missing Mass Theory ‘The dynamic response of a linear multi-degree-of-freedom system, such asa piping system, is described by the follow- ing equation: Mg++ Kx) =FO Where: M = =nXxnmass matrix of system © =x damping matrix of system K_ =nxn stiffness matrix of system COADE Mechanical Engineering News May, 1993 {700 =n 1 time-dependent acceleration vector (0 =n 1, time-dependent velocity vector (0 =x Iy time-dependent displacement vector F(O = nx 1, time-dependent applied force vector Assuming harmonic motion and neglecting damping, the free vibration eigenvalue problem for this system is: K® -M@w2=0 Where: © =nxnmode shape matrix 2 =m xm matrix where each diagonal entry is the frequency squared of the corresponding mode ‘Themodal matrix © maybe normalizedsuch that OTM © =1 (where Tis then xn identity matrix)and ®T K @= 2, Fach modal response represents the response of a single degree-of-freedom oscillator, with tsown natural frequency (co}) and share ofthe total system mass, and acted upon by ‘a share of the total external load, Each mode’s share of the total system mass and external load (exclusive of DLF) are calculated as: m,=MO,O7M j= MO OTF of modal mass (systemmassactive in mode) for mode i ©; = mass normalized mode shape for mode i = vector of modal force (force exciting the mode) for mode i This can be verified by summing the modal masses and forces over all modes. which shows that they equal the total system mass and external force, respectively: Em; =M OO! M=@TM@M=IM=M Bf =MOOTF=0TMOF= =F Rarely are all modes extracted during a spectral analysis; therefore the mass and applied force actually considered ‘during the spectral analysis can rarely be identical tothe true system mass and applied force. Furthermore, as differing numbers of modes are extracted forthe analysis, the amount of mass and force included in the solution will vary. The accuracy of the solution will likewise vary as modes with large modal masses and forces are included or omitted. ‘The matrix of all system modes may be partitioned into two submatrices — those which are extracted forthe analy- sis and those which are not: =[0, ] Where: @, = mode shapes extracted for dynamic analysis lowest frequency modes) , = residual (non-extracted) mode shapes (correspond- ing to the “missing mass” contribution) ‘The system mass is equal to the sum of the modal masses allocated to the extracted modes plus the sum of the modal ‘masses allocated to the residual modes. Likewise, the applied force is equal to the sum of the applied forces allocated to the extracted modes plus the sum of the applied forces allocated to the residual modes: M=Mg+M,=M_ TM +M0, 7M FaFe+Frp=M@e 1 F+M®, 0,0 F Ideally, the effective mass included in the extracted modes should be as close to 100% as possible, indicating that as ‘much of the modal response as possible has been considered. Effective mass is calculated as: effective mass = Me /M Normally, modal analysis completely neglects the rigid response — the response ofthe non-extracted modes, due to the load F,, However, since M, ®,, and Fare all known, Fr ‘can be easily calculated, without actually extracting the high-order modes, as: F,=F-Fp=F-M0,0,1F ‘This residual force can then be applied statically to the structure, with the resulting displacements, forces, stresses, and reactions representing the net response of all rigi modes: COADE Mechanical Engineering News May, 1993 Kx=(LP)F; Where: x = maximum displacements of residual (non-ex- tracted) modes LF dynamicload factorfornon-extracted modes (maxi- ‘mum DLF for any frequency above the highest ‘extracted frequency) “This responses then added to the responses ofthe extracted. Jow-order modes in order to get the total system response — effectively including 100% ofthe system mass and 100% of the external force while obviating the need toextracta large number of modes. Note that although this procedure is called a “missing mass” correction, it might possibly be better called a “missing force” correction, since itis the force acting an the rigid ‘modes which is added back into the solution, ‘The term “missing mass” is sed since this correction has traditionally bbeenused forimproving seismic analyses, where the applied load is an inertial load —i.e., MA. Inthis case. the applied missing force is the missing mass (i.e., the difference be- tween the total system mass and the sum of the mass associated with all ofthe extracted modes) times the ground acceleration, ot ZPA: F,=F-M@, 0,TF=M (ZPA)-M@,@,TM (ZPA) F,=M,(ZPA) ‘When considering force spectrum analyses forloads such as fluid hammer, relief valve, slug flow. etc.. the applied load isnot function ofthe system mass, so the corrections truly “missing force” correction. Recommendations For Use of the Mis tion 1g Mass Corree- As described above, CAESAR II implements the missing. ‘mass correction according to the following procedure: 1) The missing mass load iscaleulated foreach individual shock load as: -M®, OTF ‘Note that the load vector F represents the product ofthe force set vector and the rigid DLF for force spectrum loading. For seismic loads from uniform support motion, F represents the product of the mass matrix, ZPA, and directional vector for the seismic load. For seismic loads from Independent Support Motion (ISM) and seismic anchor movements, F represents the product of the mass matrix. ZPA, and displacement ‘matrix under unit ISM support displacement. Note that the “rigid” freauency , for the purposes of determining the rigid DLF, or the ZPA, is taken by CAESAR I to be the frequency of the last extracted frequency (this will give a conservative value for the missing mass load, since many of the rigid modes would actually be excited ata lower DLF; however the altemative — to apply the ZPA to the non-extracted modes — would be non-conservative if the frequency cutoff actually falls within the resonant range of the response spectrum). Note that the missing mass load will vary, depending upon the number of modes ex- tracted by the user and the cutoff frequency selected (or more specifically, the DLF or acceleration correspond- ing to the cutoff frequency). 2)The missing mass load is applied tothe structure as a static Toad, The static structural response is then com- bined (according to the user-specified combination method) with the dynamically amplified modal re- sponses as if it were a modal response. In reality this static response is the algebraic sum of the responses of all non-extracted modes — representing in-phase re- sponse, as would be expected from rigid modes. Since CAESAR II's procedure assumes that the missing ‘mass correction represent the contribution of rigid modes, and that the ZPA is based upon the spectral ordinate value at the frequency of the last extracted mode, itis recom- ‘mended thatthe user extract modes upto, butnot farbeyond, a recognized “rigid” frequency. Choosing a cutoff fre- ‘quency to the left of (and therefore below) the spectrum's resonant peak will provide a non-conservative result, since resonant responses may be missed. Using a cutoff fre- quency tothe right ofthe peak, butstllintheresonantrange, will yield conservative results, since the ZPA/rigid DLF will be overestimated. Extracting a large numberof rigid modes for calculation of the dynamic response may be conservative, since all avilable modal combination meth- ‘ods (SRSS. GROUP, ABS, etc) give conservative results versus the algebraic combination method (implicit in the missing mass correction) which more accurately represents, the net response of the rigid modes. Figure 2 shows one appropriate and several inappropriate cutoff points for a typical response spectrum. 10 COADEF Mechanical Engineering News May. 1993 2a} 1s | tea Proper 1, Non-conservative cutoff (Misses amplification of ‘any modes in resonant range) 2. Conservative cutoff (Multiplies missing mass con- tcibution by excessive DLF 1.6) 3. Optimal cutoff (Includes all modes in resonant range, uses low DLF--1.05--for missing mass con- tribution, minimizes combination of rigid modes) 4, Conservative Cutoff (Too many rigid modes com- bined using non-conservative summation meth- ods) Figure 2 - Dynamic Load Factor Response Spectrum CAESAR II provides two options for combining the miss- ing mass correction with the extracted modal results — SRSS and Absolute. The Absolute combination method of Course provides the more conservative result, and is based ‘upon the assumption thatthe peak dynamic amplification of| the extracted modes occurs simultaneously with the peak absolute ground acceleration or force load. Research sug- {gests that the modal and the rigid portions of the response to typical dynamic loads are actually statistically independent, so dhat an SRSS combination wediod is a better vepresentar tion of the total system response. For this reason, (CABSAR 11's default missing mays combination method is SRSS. Example of the Use of the Missing Mass Correction ‘The benefits of this feature may be illustrated through the uscof an example, The system shown in Figure 3 was found to have undergone a steam hammer loading following a turbine trip. During the transient, the axial snubber at node point 200, originally sized with a 10,000 pound capacity, was found tohave filed under the actual shock loading. An ‘analyst was asked to re-review the dynamic analysis of the system forthe transient oad and determine why the snubber was undersized. Figure 3 Review of the thermal/hydraulic properties of the system indicated that the critical loading case for this snubber ‘occurred when the steam hammer pressure pulse was present in the 150-210 leg. Analysis of the valve and system characteristics indicated that the unbalanced pressure load profile at node point 150 for this case was approximately as shown in Table 1 When processed through CAESAR II's Pulse Table Gen- ‘erator, this corresponded to a DLF response spectrum as shown in Table 2 * FORCE SPECTRUM DATA FILE * Jobname = STMHAM * USER ENTERED TIME HISTORY * —TIME(milliseconds) FORCE(Ib) . 00000 00 . 3.62000 38000.00 . 70.00000 38000.00 . 103.62000 00 Table 1 ul COADE Mechanical Engineering News May, 1993 . FREQ (HZ) MULTIPLIER 0001787 9003707 0028588 0011996 0144725, 0063606 10457403 0201114 1116706 0491111 2315600 1017906 4289935 1883361 [7318439 3201743 1.172720 “5085984 1,7867280 7612415, 2.139490 10743430 3,7049600 1.4182720 5.1030760, 1,7162570 6.8638980 1,8665350 90433170 1,8545330 117095100 1,7638710 14.9229400 16344420 18.7563700 1.4628940 23.2848300 1.2563820 285876600 1,0403550 34.7484800 1.1374020 41.8551800 1.216540 49.9999800 1.1595610 Table 2 ‘A dynamic analysis was performed, with a 38,000 pound load applied in the Z-direction, in conjunction with the above spectrum, at node point 150. Theanalystextracted all ‘modes up to 33 Hz, the nominal rigid frequency. This resulted in the extraction of 41 modes (total eigen solution time of 2 minutes, 11 seconds using a 80486-25 MHz ‘machine). Thedisplacements and reactionsin the vicinity of the 150-210 leg resulting from that analysis are shown in ‘Tables 3 and 4 ‘CAESAR BISPLACEMENT REPORT FIL :STMHAM (OCO'Stock Case #1 DATE:APR 14,1993 “Transations{in.} —— ——Rotations( deg) — NODE DX DY DZ RX RY RZ 150.0001 0.0032 0.0001 0.0022 00012 0012 160.0000 00035 0.0001 0.0023 0011 a.0012 170 0.0007 00051 0.0001 0.0027 0.0010 0011 180.0017 00082 9.0001 0.0028 0.0008 0.0010 182.0025 0115 0.0002 0.0023 0.0007 0.0000 185.0029 00138 0.0002 0.0015 00008 0.0008 187 0.0031 0014S 0.0002 0.0011 0.0006 0.0008 192.0029 00134 0.0001 0.0017 0.0005 0.0007 194 0.0019 0005 0.0001 0.0028 0.0002 0.0007 200 0am ons! nnn amet m1? oN 209 0.0013 0009 0011 0.0027 0.0013 00007 >i anak om oan a.em1 01? OONNK Table 3 costco DaTEAPRiGn noo oso 0 gars Tables 12 COADE Mechanical Engineering News May, 1993 ‘This shows the expected load on the axial snubber at node 200 to be only about 1550 pounds — far below the actual snubber capacity of 10,000 pounds. According to this, analysis, the snubber should not have failed. However, as stated earlier, the analyst, specifically in the case of axial fluid hammer, does notreally know whether the 41 extracted modes are sufficient to provide an accurate estimate of the forces, stresses, etc. in the system. One way to determine this is To continue to extract more modes (Getting the cutoff to 40, 50, 60 Hz, etc.) until the results of interest become asymptotic. Repeating the analysis inthis ‘manner several times, the snubber load suddenly increased substantially once frequencies above 70 Hz were included. Using a cutoff of 75 Hz (78 modes; otal eigen solution time ‘of 6 minutes, 2 seconds, in addition to the time required to ‘sum the responses ofthe additional 37 modes), the displace- ‘ments and reactions were as shown in Tables 5 and 6. [CAESAR DISPLACEMENTREPORT FILE-STMHAM (OCOSImkCae #1 DATEAPR 14,1993, rarseiontin,} ——— Reatonsieg} —| NODE DX = DY «bz RK RY RZ. 150 0.0002 0.0038 0.0003 0.0024 OI 00013 160 0.0000 0.0036 0.0023 0.0024 0013 0013 170 0.0008 0.0053 0.0022 0.0027 00012 0012 180 0.0021 0.0082 0.0021 0.0029 00010 0.00 12 00027 0.0116 0.0019 0.0025 00010 0010 185 0.0029 0.0138 0.0016 00017 00013 0000 187 0.0033 0.0145 0.0013 00013 00007 0.0008 192 0.0030 0.016 9.0010 0.0019 000130008 194 0.0025 0.0096 0.000 0.0029 0.0009 0009 200 0.0000 ©0063 9.0000 0.0032 0.0014 o.0ot0} 209 0.0014 0.0010 0.0012 0.0028 0.0014 0.0010] 210.0021 40000 0.0034 0.0022 0.001 0.007) Table S nan nan Table6 Itisnow clear why the snubber failed — the load under this transient would actually be expected to be on the order of 36,000 pounds, or more — it is possible that extracting additional modes might increase the snubber load even further. (Note that the displacements did not vary as much as did the restraint loads from the one analysis tothe other, since the displacements are largely dependent on the re- sponse of the lower order modes.) This iterative process can be time consuming; nor is there ‘any guarantee that sufficient modes have been extracted at any given cuust feyuency. Alsu, as uoted cai, futh {inaccuracy may be introduced duetothe combination method used when sung the rigid nodes, “The alternative wo this iterative med is o do the analysis ‘with the missing mass correction included. As explained above, the frequency cutoff should be w the rightof dre peak fof the DLF curve, but not too far into the rigid range. Reviewing the DLF curve, it appears Uhat any Frequency above 6.86 Hz would be an acceptable cutoff, with 33 Hz being appropriate. (A 33 Hz cutoff would be superior to a 6.86 Hz cutoff, since the missing mass contribution would be loaced with a much lower DF, while modes up to 33 Hz, ‘can still be extracted fairly quickly. Also, this frequency ignifcs “rigid” in the minds of many engineers.) 3 COADE Mechanical Engineering News May, 1993 Performing the analysis using all modes up to 33 Hz (still 41 ‘modes, with an extraction time of 2 minutes, 11 seconds), but including the missing mass effect, the results were as shown in Tables 7 and & ‘These results give fairly close agreement with the displace- ments achieved by extracting modes up to 75 Hz, but they ‘were delivered in one-third of the time, The restraint load at node point 200 is 38,952 pounds — enough to fail the snubber. This is far more accurate than the 1548 pounds |CAESAR II DISPLACEMENT REPORT FILE:STMHAM (OCO'Shock Case #1 DATEAPR 14,1993 Transation(in} —— ——Retations(deg) — NODE DX DY DZ RX RY RZ 150 0.0001 0.0035 00034 0.0028 0.0012 0012 160 0.0000 0.0037 00033 0.0028 0.0011 00012 170 0.0007 0.003 00031 0.0027 0.0010 aor 180 0.0017 0.0082 0.0028 0.0029 0.0008 0.0010 12 0.0025 011s comm 0.002 0.0007 0009 18S 0.0029 00138 00020 0.0015 0.0008 0008 167 00031 00185 00015 0.0012 0.0007 0008 192 0.0029 0.0134 90011 0.0018 0.0005 0.0007 194 0.0019 ones 0000S 0.0028 0.000 0007 200 0.0000 0.0081 0.0000 0.0031 00012 0.0007 208 00013 0.0009 0.0011 0.0027 00130007 210.0018 0.0000 0.0033 0.0022 0.0012 0.0006 Table7 em wos 220) moma aa) m0 m8 va Table culated forthe same number of modes, when the missing mass correction was not considered. This number also exceeds the 35,943 pounds found when modes upto 75 Hz were extracted — indicating that still more modes might have tobe extracted in arderto getan “accurate” load on this smubber through dynamic means only. Strangely enough, the loads calculated on the restraints at nodes 160, 200 (X- component), and 220 are much lower for this analysis than for the analysis using the 75 Hz frequency eutoff. This differences explained by the excessive conservatism intro duced by the modal summation method used in the 75 Hz analysis — the missing mass correction assumes that rigid modes respond in-phase, leading to these lower (and more accurate) loads. ‘Therefore, correct use of the missing mass feature provides the user with etter accuracy (comeeting hath non-conser- vative and overly-conservative results) with shorter calesiational times Discussion of the Included Mass Report In Version 3.19, CAESAR Ii has added a report called the Included Mass Report, which informs the user how much of the system “mass” was included in the analysis, and how much was added for the missing mass correction, if any. ‘This generally gives the analyst an indication of the degree to which the system response falls within the dynamic (amplified) vs. the rigid (non-amplified) range. ‘The percent of included “mass”. as provided in this report, is calculated forall shock cases, whether missing mass is to be included or not, as shown below: Mass Included = 1 - (ZF{i]/ EF) sunimed over i= I ton Where: Fy. = the vector of residual forces (missing mass feces), th F = the vector of applied system forces, Ib 1 = number of translational DOFs COADE Mechanical Engineering News May, 1993 ‘The percent of “missing mass” added, is also calculated for all shock cases (if not included. it is shown as 0.0%) as the converse, shown below: % Missing Mass Added = EF [i] / EFUi] summed over i= 1 ton (Note — if all possible modes are included in the analysis, the percent included will of course be 100%. In general, if fewer modes are included, the percent of included mass will usually be less than 100%. However, itis possible under certain circumstances — such as when localized forces coincide with modal high points — for this number to exceed 100%. This situation is discussed in some detail in the example below.) ‘As seen from the above formulas, this report is actually an included missing force report" not an “included/missing mass report” (note that for uniform inertial seismic loads, however. these terms are identical). The reason for this is that the former is much more useful to the analyst. For example, consider the planar cantilever modeled with three ‘mass points as shown in Figure 4, along with its first two ‘modes of vibration. ny Mode 1, 5.0705 hz. ‘other hand, mass point 2 displaces relatively less inthe first mode of vibration than it will in the total system response. ‘Therefore, the mass at point 2 is under-emphasized and the mass at point 3 is over-emphasized in Mode | How does this getcorected? Doinga statistical superposition ‘of Mode I and Mode 2, itis evident that Mode 2 brings up the response at point 2 significanly more than it does the response at point 3, thus equalizing the emphasis onthe mass at the two locations. This can be seen by noting the corresponding terms in the mass-normalized mode shapes for this cantilever, shown in Table 9, a ———— Mode 2, 25.8548 hz ee See ee L Figure 4 In the first mode of vibration, mass point 3 extends far from. itsinitia location —relatively further than the point actually ‘will displace in the total system response, onaverage. Onthe CAESAR II MODE SHAPES FILESTESTI MASS NORMALIZED DATE:APR 14,199: Translations — Rotaton— NODE Dx by Rz MODE 1 Frequency (Hz) = 5.070 ' 0.0000 0.0000 0.0000 2 0.0000 0.4839 0.0070 3 0.0000 1aT4s 0.0089 MODE 2 Frequency (Ha) = 25855, 1 0.0000 0.0000 0.0000 2 ‘0.0000 1.0826 oor 3 00000-06843, 0219 Table 9 ‘The implication of the “included force” calculation is that it ‘considers the impact of loading a mode at an over-emph sized, or under-emphasized point, whereas an “included ‘mass” calculation considers the average of the modal em- phasis at all points, regardless of the location or direction of the load. For example, consider the included mass and included force, respectively, for the three dynamically ap- plied loads shown in Figure 5, when only the first mode of the cantilever is extracted. 15 COADE Mechanical Engineering News May, 1993 i +. Figures For all three load cases, the mass matrix and the mass normalized first made shape are the same, while anly the applied force vector changes. The mode shape is shown above: the system mass matrix (assuming a 20-foot long. empty, S-inch diameter, standard schedule pipe) and the three force vectors are shown below: 4 OM Xz Ya Xs Ya 8 037100 09 00 09 00 09 09 00 00 03781 00 00 09 09 00 00 00 00 00 09 00 a9 09 09 09 00 00 00 00 0756200 09 00 00 00 [Mj=00 00 00 00 0756200 00 00 00 00 00 00 09 09 09 00 09 00 09 00 00 09 09 a0 037800 00 09 00 00 09 a0 00 00 o378100 09 99 99 99 09 anno on a0 00 00 09 1000.0 00 00 00 00 09 00 00 00 (R)= 00 (FJ= 1000 R= 00 00 00 00 00 00 a0 00 00 toon. 00 00 a0 Asexplained earlier, the included, or effective, mass for the first mode of the system is expressed as M,=M®, 7M Since the extracted mode and the mass matrix are un- changed for the three dynamic loads, the included mass ‘matrix is identical forall three loading cases as wel: 09.00 00 00 09 090000 00 0000 00 09 00 000000 00 0900 00 09 00 090000 09 9000 00 00 00 000000 00 00.00 00 00 0.1339 00 00 00 00 0000 00 00 00 000000 00 0900 00 09 09 090900 00 00.00 00 09 09 00 00 o310800 0000 00 09 00 090000 00 ‘The percent of mass included in each of these three load ccases is independent ofthe load, or: ‘Seinclmass=My/M =(0.1339+0.3108)/(4x0.3781 +2x0.7562) COADE Mechanical Engineering News May, 1993 It should be noted that 14.7% ofthe mass is not uniformly included forall degrees-of freedom — for the over-empha- sized point 3 in the Y-S, 6, and, S, - 8,/6,, oF: 9,=0,0,/S, for, <8, -8,%6, Where: 6, = elastic, steady stress level following shake- down, psi S, = yield stress, psi ©, = thermal expansion stress, psi 1, = sustained stress, psi For the maximum permissible sustained stress (approxi- ‘mately 0.667S,), and a typical expansion stress ofS, equates to an actual stress level of about 0.8455, — this difference is accommodated by reducing the thermal expan- sion allowable (1.25 times the sum of the cold and hot basic allowable stresses) for creep as well. Since creep is caused by a tr-axial stress it is necessary to ‘se the sustained stress intensity inthe calculations. This stress (3-D Stress Intensity) is readily available in the ‘CAESAR 11 132-column stress reports. ithe 132-column report is not available, this could present a minor inconve- nce, because the piping codes do not explicitly require COADE Mechanical Engineering News May, 1993 thecalculationof sustained stressintensity atany giventime. Suessinensiy inthe pipe is approximately calculated asthe S=(G,-SP +47)" S=((S,-S)+4Ty" S=(G,-S¥+4Ty" Where: S__= sess intensity, psi = circumferential normal stress, psi longitudinal normal stress, psi radial normal stress, psi T= shear stress, psi In most cases, both circumferential, longitudinal, and racial normal sustained stresses are predominantly due to pressure, and so the first two are both tension, or same-signed, while the third is compression, or opposite-signed. In these cases, the second or third expression will yield the largest value, and sothe first can be ignoréd. Mos piping codes ignore the radial stress, just requiting that stress approximately equal tothatcalculated using the second expression be used forthe sustained stress analysis case, while a stress approximately ‘equal to that ofthe third expression be used when selecting the minimum wall thickness. Therefore, i the absence of a true sustained stress intensity, a coarse estimate of the sustained stress for ereep analysis purposes would be the ‘maximum of either the CAESAR II sustained code stress ot the hoop stress. Effect of Creep on Determining Design Life: As noted above, the basic material allowable stresses limit the creep rate, not the total creep strain. Therefore, itis the ‘engineers responsibility to ensure thatthe creep strain rate permitted by these allowables is acceptable for the entire design life. For example, ifthe piping system can tolerate a creep strainof 1.5%andcan still performs function, andthe piping material ruptures at 2.0% creep strain (i... rupture stress does not limit), then the permissible design life could bbe set as high as 1000 x (1.5% / 0.01%) = 150,000 hours, or 17.1 years. Tis not always evident whether a basic allowable stress is, set based upon non-creep or creep criteria, or if the latter, Whether it was based upon the strain rate or rupture criteria. ‘Therefore, when wishing to extend a system's design life beyond the nominal 11.4 years, it may be necessary to refer to experimental creep data, (In the absence of this data it ‘may be reasonable to assume that creep criteria governs in the range of 800°F to 900°F and above. If the material temperature is below 800°F, creep is probably not a con- cem.) Typical steps for determining design life based on creep criteria are: 1) Determine the creep strain which constitutes failure for the system, Obviously, the strain rate must be less than that causing rupture. A higher strain would be permis sible for piping, vessels etc. than would be permissible for rotating equipment or other systems having delicate tolerances, 2) Estimate the magnitudes of sustained (non-thermal) Joadings expected, along with their expected durations. In general, potential loadings will be weight only, and weightpressure. The durations of each ofthese loads will be dependent upon how much time the system spends shut down vs. time in operation, and the length ofthe anticipated design life. Itmay be that one of these Joad cases may have such a short duration that it an be ignored. 3). Perform stress analysis, calculating sustained stresses for each of the loadings determined above. Determine ‘whether the system reaches the failure point, with these stresses, within the expected length ofthe load duration. ‘This can either be a simple task or a quite difficult one, as described below: a) Ifthedesignifeis 11.4 years orless and the failure criteria is 1% total creep strain or les, then the creep stress criteria can be considered to be met if the maximum sustained stress of any of the load casesis within the basic allowable stress forthe hot temperature (S,). by Ifthe design life is other than 11.4, and the failure criteria is other than 1% total ereep strain, then the creep stres criteria can be considered to be met if the maximum sustained stress of any of the load ceases is within Sand the design life is less than or equal to 11.4e, Years, where €, is the allowable exeep strain in percent ©) Ifthe maximum sustained stress of any ofthe load cases is within S,, but the desired design life exceeds 11.4e, then the creep contribution of the individual loads must be reviewed. For example, ifthe system spends one-third of its life shutdown (at a temperature outside of the creep range), this B COADE Mechanical Engineering News May, 1993 time may be reduced from the design life when ‘comparing to 11.4e,. Ifthe system spends h hours, ata temperature T, (below the maximum tempera- ture, but still in the creep range), then the Larson- Miller relationship can be used to convert these h hours to an equivalent number of hours at the ‘maximum temperature: tig, = 10° = 00 If the system spends h hours at a stress level 0, (below the maximum stress), then the expression. derived earlier for constant temperature can be used toconvert these h hours ton equivalent number of hours at the maximum temperature: Where n usually ranges for normal engineering ‘materials from 309, or can be calculated explicitly as described above. ‘The equivalent time for each load should then be combined: if less than 11.4, then the creep stress criteria can be considered to be met. @) For a more complex case, or when rupture is the limiting criteria, it may be necessary to use experi- ‘mental creep data (both strain and rupture) to de- ‘velop an explicit equation to describe the material creep performance. The creep strain should be calculated for each sustained stress/temperature/ time combination, summed, and compared to the allowable strain. The percent of rupture life used up should also be summed, added to the cumulative fatigue damage, and verified to be less than 1.0. If these two criteria are met, then the creep stress criteria is acceptable (Creep Evaluation for Life Extension: Piping life may be extended if the estimates used when determining the initial design life were overly conservative, thatis, the duration ofthe load, the operating temperature, of the stress level was over-estimated (orf any of these did not take full advantage of the allowable). In certain cases, ‘modifications to the system or its operating procedure 10 reduce temperature or stress may alter the creep strain rate sufficiently to add years to the operating life. Looking at these situations — overestimation of the load duration, operating temperature, or stress level — demon- strates the various techniques which may be used for life extension, ‘Assume that the system shown in Figure 3 has been operat- ing for 10 years, and the engineer is now interested in justifying the extension of its design life for 11 more years bbeyondits original 1-year design life, This original design life was based upon a limitation of creep strain to 1 percent, ‘and a requirement that there be no rupture. o~ Figure 3 ‘The piping material in this system is A 335, Grade P22, ‘which has an allowable stress of 5800 psi at 10S0°F, as per Appendis A of the B31.3Code. Based upon the rapid drop- off of allowable stress above 950°F (11000, 7800, 5800, and 4200 psi at 950°, 1000°F, 10S0°F, and 1100°F, respec- tively), the allowable stress definitely appears to be creep limited at this temperature. A.useful first step may be to develop the equation describing the creep behavior of the material. Thisis done as described ‘earlier, by fitting the generic creep equation to any available experimental data. Since A 335, Grade P22 is a 2-1/4% ‘chromium, 1% molybdenum steel, the equation developed earlier (from Figure 2) can be used (readjusting the mand n parameters for a temperature of 1050°F) to calculate the total creep strain: = IET1, (0,88) Since rupture does not always occur at a specific creep strain, it will probably also be useful to fit this general creep equation to the rupture data shown in Figure 4, Normalizing the ereep strain to 1.0 yields the cumulative damage equ: tion: Point A. 0,6,=11200psi,T, Point B 10,6, =28000 psi, ~ COADE Mechanical Engineering News May, 1993 1.0,6, = 20000 psi, T, 0E-51, (0.07 $1.0 m = -9,001275 48798 2 / Figure 4 Review of the duration of the load: Review of the actual duration of the design load versus the expected duration may be an easy means of extending piping life. This is often due tothe fact that initial estimates ofthe time that a system spends in operation, as opposed to shut down, may have been conservative. Aftera number of years (when life extension studies are usually undertaken), abetter understanding ofthe pattern of system operation available, and a more accurate estimate can be made. For example, a review of operating records may show that the actual length Of time that the system shown in Figure 3 spent in its operating condition was only six out of its first ten years This would immediately offer a minimum additional creep life of four years (ten minus six) or, assuming the operating Patterns continue, i, that the system operates only 60% of. the time, this would project a design life closer to 19 years (11.4 years divided by 0.6). Review of the actual vs. estimated temperature: ‘The desired life extension cannot always be achieved sim- ply by comparing the actual time in operation to that origi- nally projected. For example, assume thatthe system has actually been operating virtually without interruption forthe entire 10 years. In this case, itis necessary to examine other Parameters in order to increase the system's creep life. The review of operating records may indicate that the system. rather than operating continuously at 1050°F, actually spends about 40% of its time (40,000 hours out ofthe 100,000 hour design life) operating at only 975°F, 7% of ts time (7,000 out of 100,000 hours) operating at 900°F, and 3% of its time (3,000 hours) shut down, at 70°F. From previously derived relationships, the time spent at other temperatures can be converted to equivalent hours at 1050: Teme He Eau ic 080,000 sao 975 a0 Loman 2860 00 7.000 Lonesminiam et = he m0 3.000 Onrnotnereeprnse) Toal—> 524201 ‘This demonstrates thatthe system experiences only 52,429 hours at an equivalent of 1050°F per every 11.4 years of operation. Since the use of the code stress allowables implies a creep life of 100,000 hours, this would correspond to. design life of 11.4 x 100,000 / $2,429, or 21.2 years. Review of the actual vs. allowable stress level: Itmay be thatthe system has not only operated uninterrupted. forts entre life, but it may have operated at the anticipated temperature as well. In this case, the last remaining possi- bility is to determine whether the design level can be extended based upon stress levels falling below the allow- able. Itmay be that are-evaluation ofthe operating circumstances ‘may show that the original stress estimates were not realistic — design pressure may far exceed actual operating pres- sures, spring hangers may have been installed with different pre-loads than intended, or as-built deviations may have led to a significantly different sustained stress distribution. However, the most likely source of extended life due to reduced stress is found in the difference between the actual sustained stress and the code allowable — it is rare that 100% of the allowable is used in a piping system. For example, the Maximum 3-D Stress Intensity for the system shown in Figure 3, is $419 psi, found at nodes 10 and 15: 25 COADE Mechanical Engincering Nows Ac noted earlier, the allowable stress, upon which the creep. design life is based, is $800 psi. Since the actual tri-axial stress was only 5419 psi, this translates into a corresponding, increase in design life, which can be calculated from the relationship for constant temperature: 4 ($800") 100,000 = j (5419), Ifthe controlling creep criteria for the allowable stress was based upon elongation, n (as seen above) is 9.22031, and, the expected life at this stress level, ic 187,103 hours, or 21.4 years. If the controlling creep criteria was based upon ‘rupture, n was calculated to be 6.48798, sa the expected life is 155,401 hours, or 17.4 years. The system design life could be extended still further by reconsidering the pressure stress —creepis caused by along, term, sustained stress, while design pressure is usually chosen tobe the maximum pressure that could be postulated tooccur, usually due toa transient situation, Therefore, itis ‘more reasonable to base the long term stress on the normal ‘operating pressure, For example, if operating records reveal that the system pressure is normally 950 psi, rather than the 1000 psi design pressure, the sustained stress results would be: May, 1993 ‘With a stress intensity of 5149 psi, the design life can be calculated as: 5800/5149)" 100,006 09,735 hours (fora =9.22031).r 16,502 hours (forn=6.48798) This represents a minimum life of 24.7 years. Calculation of design life for more complex situations: Under certain circumstances, there may be many different combinations of stress level, temperature, and load dura tion, rather than just the simple changes described above. In this case, it may be easier to derive the design life fram scratch using the generalized creep formula and the code creep criteria For example, assume that during the first 100,000 hours of the system's life, the operating pattern has been established ‘as comprising the following combinations of temperature and stress: Operating State Hr Temp Stress 1) Transient 1,000 1050 S419 2)Nocmal Operation 49,000 1050 S149 3)Reduced Operation 40,000 975 S097 4) Standby 7000 900 ad 5) Shutdown, 3.000 702239 As stated previously, the B31.3 Code criteria imply that 1) the creep strain should not exceed 1%, and 2) the stress COADE Mechanical Engineering News May, 1983 ‘cannot exceed 67% of the average stress causing rupture at the end of 100,000 hours. Therefore, both the creep rate and, the rupture stress mustbe calculated for each ofthe operating, conditions described above. First, calculating the creep strain: IETLe"Y Operating State m a , 1) Transient -871765 9.22031 0.000033 2)Normal Operation 8.71765 9.22031 0.001012 3)Reduced Operation 9.12904 11.2831 0.000005 4) Standby 9.58581 14,8091S 0.000000 5) Shutdown, = — 0.000000 Total 0.001050 Aneffective creep strain of 0.00105 through 100,000 hours indicates a design life, based on a 1% creep strain, of 100,000(0.01/0,00105) = 952,381 hours, or 108.7 years. Next, rupture criteria can be checked by increasing the stresses (hy dividing them by the 67% factor of safety) and calculating the cumulative damage caused by each of the ‘operating states: cd = 10B-5t, (ee cumulative Operating State Stess0.67 =m —n_—damage 1) Transient 8088 9.00128 6.48798 0.009798 2)Normal Operation 7685 9.00128 6.48708 0.344607 3)Reduced Operation 7607 9.49332 8.24211 0.004075, 4) Standby 66618 -10.03063 11.78190 0.000000 5) Shutdown 342 — = — 0.000000 Total 0.358480 ‘This demonstrates that 35.8% of the code’s rupture life is used up for every 100,000 hours af operation, so the entire rupture life will be exhausted in 100,000/0.358 = 279,330 hhours (31.9 years). However, it must be remembered that fatigue and rupture ae interactive, so the following formula must he considered as well: eds LUN, 8,9) E85 N)] + (0.388/100,000), j Where N, = allowable number of cycles for maximum ex- pected cyclic stress range S, = maximum expected cyclic stress range. psi ‘= number of eyeles expected for eyelic load j, per 100,000 hr S, = expected cyclic stress range for cyclic load j, psi 4 \, = proposed design life, hr Simplifying matters — assuming that there is only one cyclic load, with a cyclic rate of 3000 eycles per 100,000 hours, and a maximum expansion stress in the system of 19408 psi — the allowable number of eyeles (a a function ‘oftime) permitedforthis expansion stesscan be calculated (as described in Part 1 of this anicle) = 60(N}*? x (1.258, + 0.255,] 19,408 ~ 6.0(N,]4? x [1.25 x 20,000 + 0.25 x $800} Ny = 36,558, 50: ed ~ (14(36,558 x 19,408%))[19,408° x 3000/100,000} + (0.358/100,000), Setting the cumulative damage to 1.0 (representing failure), and solving for t: = 227,241 hours ‘Therefore this system could have a design life of 227,241 hours, or 25.9 years — from a perspective of creep and/or rupture failure. It must be stressed that these analyses only demonstrate the creep life remaining in a piping system, ‘The degree to which the design life of axystem may truly be extended is in fact the minimum of the demonstrable remaining life — based upon evaluation of all factors contributing to piping aging and failure, including other stress and metallurgical considerations. The ASME B31-G Criteria The following urticle hus been submited for publication by ‘Mr. Darrin Yin of Clear Lake Engineering, 17100! Camino Real, Suite 100, Houston, Texas, 77056, 713-480-6845. Among the new featuses unveiled in CAESAR II Version 3.19 is an algorithm implementing the B31-G procedures. ‘The objective ofthis module isto provide pipeline engineers with computerized access to this criterion, but equally relevant, many non-pipeline stress analysts (ic. power and petrochemical applications) can also be introduced to this invaluable yet litle publicized Piping Code. 2 COADE Mechanical Engineering News May, 1993, As the ttle of B31-G, "Manual for Determining the Remain- ing Strength of Carroted Pipelines", illustrates, this manval presents a criterion of acceptance or rejection for some corroded pipelines upon certain visual examinations. For the benefit of users who donot own the ASME B31-G Code, the following B31-G Foreword should serve as a proper introduction: "It is recognized by pipeline companies that some sections of high pressure pipelines. particularly those installed a number of years ago, have experienced some corrosion. Where corrosion is found. pipeline operators have been deeply concerned about the need for a method of determin- ing the remaining strength of these corroded areas. If the corrosion does not penetrate the pipe wall, what is the Pressure containing capability of the remaining pipe metal in terms of its ability 10 continue to operate safely at the ‘maximum allowable operating pressure (MAOP) of th Pipeline system? Thus, one of the needs of the pipeline industry has been a procedure that will help operators, particularly field personnel, make decisions on existing pipelines. when exposed for any purpate, as to whether any corroded region may be leftin service or whether itneeds 10 be repaired or replaced. Suchdeterminationsmusthe haced upon sound research and extensive testing in order 10 provide safe and conservative guidelines on which to base field decisions. The Manual provides procedures to assistin this determination." Parts 2,3, and 4 are based on Appendices G-6. G-7. and G- 8 ofthe ASME Guide for Gas Transmission and Distribution Piping Systems, 1983 Edition. They are included in this Manual foruse by field operators to determine the remaining, strength of corroded pipe. The technology is based on research done in the Columbus laboratories of the Battelle Memorial Institute; specifically, their report “Summary of Research to Determine the Strength of Corroded Areas in. Line Pipe”, July 10, 1971. Before putting the criteria to use, one needs to acknowledge that the B31-G Manual has afew scope limitations, namely: 1) The Manual is not applicable for any new pipeline construction. Thats. itshould not serve asan engineer- ing design reference. 2) Its applications are confined to corrosion flaws solely (on the BODY of the line pipes made of ASTM AS3, ‘A106, A381, and APISL materials, thus excluding any corroded girth or longitudinal welds. 3) The procedures specified in the Manual ascribe only to the pressure loading in the pipeline; consequently, all, ‘other loads of discernible magnitude (such as external ‘bending moments) on the line can render this criterion void. (One problem some stress analysts may haveis orealize that ‘on occasion, even if a corrosion pit consumes up to 80% of | the pipe wall, the pipeline may still be considered safe by the B31-G criterion. Is 20% of the calculated wall thickness, ‘based upon the full measure of hoop stress requirements all that is needed for pressure containing purposes? ‘The answer to this question is of course no; the disparate representations of stress allowables in each Piping Code contribute to this ambivalence. Be reminded, central to the B31-G criterion is a semi-empirical fracture mechanice study made on 47 burst test results. What sets apart the B31- Griterion from any other Code stress criterion, therefore, the interpretation of the resulting leak or rupture failure of these burst tests. By contrast, all other Code etrese allow. ables simply apportion the SMYS (Specified Minimum. Yield Strength) or Tensile Strength of the pipe materials involved. In reading the B31-G, 1991 edition, one may notice that inlike all ather Piping Codes which expand applications in cach of their subsequent releases, this B31-G Manual essen- tially keeps the same criterion, scope, and limitations be- ‘ween the initial 1984 issue and its current 1991 edition. In short, B31-G 1901 is an affirmation of B31-G 1984, However, this status quo decision (as far at we can disor), ‘comes ostensibly from some policy deliberation ofthe B31- G 1991 Committee, rather than due to a lack of the pipe industry's contention for changes during the intervening years. One of those organizations that has become increasingly strident about the B31-G criterion isthe same distinguished Battelle, Columbus Division. Under the auspices of the American Gas Association, Battelle published a December 22, 1989 report entitled: Project PR 3-805, A Modified Criterion for Evaluating the Remaining Strength of Cor- roded Pipe. The report opens with an assertion that “Expe- rience has shown that the amountofconservatism embodied in the (R2I-G 1984) criterion is excessive, resulting in the removal or repair of more pipe than necessary to maintain adequate integrity.” This report accordingly calle for the relaxation of the B31-G 1984 criterion, (Note, the evalua- tion methods suggested in the Battelle report are also included in CAESAR II Version 3.19.) Another group performing research in this field is NOVA. Corporation of Alberta, Canada. Their position papers, entitled, “Standard Damage Assessment Approachis Overly Conservative” and “New Guidelines Promise More Accu- 28 COADE Mechanical Engineering News May, 1993 rate Damage Assesment”, were published inthe Aprit 9 and ‘April 16, 1990 issues of The Oil and Gas Jourmal. As these titles indicate, NOVA Corp. also advocates a lower thresh- ‘old of reectlon on corroded pipelines addressed by the B3I-G criterion. Perhaps there is a notion more basic than the issue of a “better, more accurate, oF less conservative” B31-G crite- rion. That is, who should make the decision to replace or retain a particular section of corroded line? B31-G states “This Manuals intended solely for the purpose of providing guldetine information forthe designer/owner/operator. Thus, the specific use of this Manual is the responsibility of the designer/owner/operator.”. We think B31-G is incorrect in the latter statement; the specifie application of the B31-G criterion should be the responsibilty of Professional Engi- neers engaging in pipeline stress analysis. The CAESAR II Expansion Joint Modeler Since Version 3.0 (March 1990) CAESAR IT has provided users with an expansion joint modeler. Users select the «expansion joint type andits characteristics, and CAESAR I creates the flexible element for the expansion joint along with any additional elements to fully define the joint assem- bly. This modeler was developed with the assistance of Pathway Bellows, Inc. and includes a database containing the information found in their Manual 191 G. With Version 3.19, CAESAR I also includes an expansion joint database from RM Engineered Products. Other databases are under construction, (Questions about this modeler come to mind. — What expan- sion joint data is required by CAESAR II for a proper analysis? Whatis the significance of each parameter held in the database? How does the program model the assemblies for each type of joint (i.e. hinged vs. tied universal)? How ‘can the modeler be used with other manufacturer's j data? How correct ig the model? This first article on the subject will start to answer some of the questions presented by this program capability, Before proceeding it must be said that neither the expansion Joint manufacturers nor COADE wish to create a situation ‘where the CAESAR IT modeler replaces the engineering ‘behind the proper selection of an expansion joint. What we offer is a convenient source for expansion joint data and consistency in model generation. This processor presents quick and easy models for analysis, but the expansion joint selection isnot trivial. Actual selection of the jointis not the. function of CAESAR IL. This important task is left to the individual engineer and the joint manufacturer. By the time CAESAR II selects information from the ‘expansion joint database, parameters such as pipe size, pressure rating, end type, and number of convolutions for the expansion joint have already been specified. (Try ton your machine or refer to the CAESAR I User Guide ‘Section 8 pp. 78-87.) With this dataand the manufacturer's catalog you can find the information that CAESAR IT presents to the user; thats, the flexible ength and effective inside diameter of the joint, and the axial, lateral, angular, and torsional stittess of the Joint, (See Figure 1.) What are these terms and how does CAESAR IT use them? iar ie o a Figure 1 Length: Catalogs listthe “shipped” length of the joint. This length includes the non-flextble en pleces to which the rest of the piping system is connected; for example, welding rings or flanges. A proper stiffness model of an expansion joint requires that the true flexible length of the joint be specified, not the longer, “shipped” length. This is because this length establishes the relationship between lateral and. bending stiffness for a proper model of te finite length expansion joint. For example, for Pathway Part Number 10, CS FF 150 12 (pg. 42 Pathway Manual 191 G) the overall length of this 12 convolution joints 16.125 inches, but that Includes the flanges. Knowing the length of the Manges ‘would help, but one can also back out the convolution length by referring othe length for another setofconvolutions. An ight convolution joint with flanged ends is 12.625 inches, leaving 3.5 inches (16.123" - 12,625") for the next four convolutions. Since all convolutions have the same length, ‘twelve convolutions will cover 10.5 inches (12 * 3.5°/ 4), ‘This length is valid forall end types. Effective inside diameter: This is not the minimum or ‘maximum inside dlameter of the bellows, nor is this value equal to the LD. ofthe attached pipe. The effective inside «diameter of the bellows s the mean diameter of the bellows. ‘The effective bellows diameter sets the effective inside area which is very important in calculating the total pressure thrust. Pressure thrust is significant because it may cause ‘gross deformation and failure of improperly restrained joints. 29 COADE Mechanical Engineering News May, 1993 ‘This diameter (along with the length) can also be used to establish the selationship between axial stiffness and the lateral and bending stiffnesses forthe joint. Expansion joint ‘users know to specify this data ifthe joint is untied so that CAESAR II will calculate the pressure thrust load acting around the joint. A simple model of a tied joint does not define the effective inside diameter, so this pressure thrusts not applied. Inatied joint, the pressure thrust foree loads up the tie rods and cannot extend the flexible bellows. The CAESAR II expansion joint modeler will always include the effective diameter since the program will also generate ‘amodel forthe tic rods which resist this pressure thrust. For the sample joint mentioned above, a note is listed in the upper right comer of the page under the category: “single unrestrained expansion joint”. The note states: “Pressure ‘Thrust equals 109 sq. in. times Pressure”. The value 109 sq in. is the effective area of the bellows and this is used to calculate the effective diameter used by CAESAR IL. Here, the effective diameter is 11.78 inches (2*(109 in’/n)"). See Figure 2 ‘Axial and lateral stiffness: Looking at an expansion joint, ‘one might first assume that it’s designed to absorb piping ‘growth along its centerline, Certainly there is a significant difference between the axial stiffness ofthe pipe and that of the joint. Axial deflection of the joint works well when the attached piping can be safely guided into the joint and thrust blocks incorporated into the design. However, many cases exist where this sort of protection from pressure thrust failures cannot be used — one example is tight piping around equipment. In these situations tie rods are used t0 ‘contain the pressure thrust. Typical tic rod applications remove the axial flexibility from the joint and keep the ends ‘of the joint parallelto onc another. With tic rods, then, lateral ‘and not axial stiffness isthe significant term. In these tight applications, expansion joint assemblies are usually aligned perpendicular to the plane of excessive piping expansion. While axial stiffness is set by the bellows geometry and ‘manufacture, lateral stiffness for the joint can be derived from the length, effective LD., and axial stiffness through the following equation: 3/2)*K, (Dy) eral stiffness of the joint (Ib/in) K,, - axial stiffness of the joint (Ib/in) Dj ~ effective inside diameter of the joint (in) 1- flexible length of the joint (in) For the example, axial stiffness and lateral stiffness may be read directly from page 42 - K,_= 761 Ib/in and K, = 1438, Ib/in, Lateral stiffness can also be derived from the equation above tobe (3/2)*(761 lbvin)*(11.78)*(10.5)*or 1437 Ibi. Bending stiffness: Fora finite length (non-zero) expansion joint, the bending stiffness is defined by the axial stiffness inthe following equation: K, = (2360)*K,*D yy” K, - bending stiffness ofthe joint (in-Ib/degree) Removing the reference to the effective inside diameter (whichisnot required input for CAESAR II), the lateral and bending terms are related to each other through the formul K, = (540)*K,"P With this association between lateral and bending stiffness, CAESAR II input for a finite length joint will permit the ‘entry of either bending ar lateral stifiness: not both (an error condition). The program's expansion joint modeler uses the lateral stiffness and leaves the bending stiffness blank. Continuing the example, the bending stiffness is calculated to be (1u/540)*1437*(10.5)? of 922 in-Ib/deg. BOTH THE EXPANSION JOINT CATALOG AND DATABASELIST 231 IN-LB/DEG NOT THE 922 IN-LB/DEG ‘CALCULATED IIERE! WIIY NOT? ‘The angular stiffacss listed in the catalog is calculated independent of any lateral deflection. Thatis, it would take 231 in-Ib to impose a one degree net rotation on the joint without controlling any lateral offset. The beam stiffness formulation used by CAESAR TI uscs a bending stiffness ‘based on no corresponding lateral offset. It takes 922 in-Ib to impose a one degree net rotation on the same joint if lateral offset is held to 0. To produce the same net rotation, ‘a bending moment with no translation must be four times ‘greater than a bending moment where translation is allowed. See Figure 3, 30 COADE Mechanical Engineering Nows —) a) Figure 3 ‘The CAESAR Ii bonding stiffnessis fourtimesthe stiffness listed in the expansion joint catalogs. See example #1. ‘The bending term listed in the catalog and database is not used by CAESAR IT in the generation of an expansion joint assembly; instead, itis computed (internally) as afunction of the lateral stiffness (see Figure 2). If, however, the user wishes to develop a quick model of a hinged or gimbaled assembly, the catalog value is the number to use. If the expansion joint has no length, the relationship between ‘bending and lateral offeet is no longer maintained and the comrect bending stiffness is the value listed in the catalog (and database). For “zero length” expansion joints, ‘CAESAR Il requires both bending and lateral stiffinesses — for these models, use the catalog values directly. See Example #2. Torsional stiffness: Expansion joint failure is extremely sensitive to axial rotation, therefore accurate calculation of the net axial rotation is very important, The expansion joint database also provides ths torsional stiffness, The torsional stiffnesses contained in the Pathway database are tabulated in Manual 191 G starting on page 65. The 12 convolution, 10", 150 psig bellows example has @ torsional stiffness of (0.117E6 in-th/deg. Note that as ofthis writing, the tabulated. values shown starting on page 65 are mis-labelled -- the headings for torsional stiffnesses and allowed axial rotations are interchanged. The values in the database are correct. Example #1 - Use imposed rotation to calculate bending stiffness Here, where node 15 hasan imposed rotation and deflection is permitted, the load associated with a one degree imposed rotation is 231 in-b. The bending stiffness of this joint, then, is 231 in-tb/deg. This stiffness ignores the coupling, between bending and lateral offset and isthe value found in the catalog and CAESAR TI database, May, 1993 om oma ®) 8 = 0.00 in Sra Except for the Y restraint at node 15, this model is identical tothe previous example. Without allowing the Y deflection, the bending moment necessary to generate the one degree rotation is now 922 in-Ib. The bending stiffness is 922 in-Ib/ 4deg - four times the stiffness in the catalog. The Y support carries 1321b. The model above shows that node 15 would deflect 0916 inches if that load goes to zero — confirming the lateral stiffness of 1438 Ibvin (-132 Iby-.0916 in). Example #2; - Modeling a “zero length” expansion joint snout Qutput 8 15 231 n= ‘This model is equivalent to the hinged expansion joint ‘model produced by CAESAR IL. There are two elements ‘overlaying the expansion joint from 3to 13. These elements (5-10& 11-15) form the hinge by NODE/CNODE restraints ‘connecting 10 10 11. For thls load, this model's response is ‘identical to that of the first model in Example #1 This isa zero length model for the same hinged expansion Joint above. There are three elements in this model: 5-10, 10-11, & 11-15. The element 10-11 has no length and is defined as an expansion joint. All stifnesses for this joint, except for bending stiffness ae setto rigid (E12 Ib/inorin- Ib/deg). With a zero length element, the coupling between bending and lateral offset is eliminated and so the catalog value of 231 in-Ib/deg is entered for bending stiffness. The results of this simpler model match those of the detailed model above, Conclusion: This article defines the expansion joint length, effective diameter, axial, lateral and torsional stffnesses as straightforward modelling and analysis terms, Care must be used in applying bending stiffness as it can be tricky in its proper definition and use. Future articles on the expansion Joint modeler will discuss deflection limits, assembly 31 COADE Mechanical Engineering News May, 1993 models, and more detailed, custom-built models. Readers who wish to reviow the original ASME papers on this subject (PV&P Conference - 1989) can request reprints from COADE, Expansion Case for Temperatures Below Ambient Piping installations in northem regions that shut down in winter are sometimes subjected to temperatures as low as -50°F, well below the installation temperature, but at nor- ‘mal operating conditions, can be exposed to temperatures well above that of installation, The default load cases generated by CAESAR in multiple ‘operating temperature systems contain most, but not all, of| what needs to be considered in cases where one operating temperature is above, and one below the installation tem- perature. Recent discussions with CAESAR IL users suggest that there are dissenting opinions about how this should be done. This article presents our recommended approach to satisfy- ing load and code stress requirements in such cases. ‘THE PROBLEM Let us consider the following: Installation temperature = 70°F (CAESAR Il default) Normal operating temperature = 400°F ‘Winter shut-down condition temperature = -50°F By accepting the default CAESAR Il instalation tempera ture and setting TI and T2 as 400°F and -50°F respectively, typical program recommended load cases look like this: 1, WsP1+T1 (OPE) (operating load case T=T1=400°F.) 2, W+P14T2 (OPE) (operating load case T=12=-50°F.) 3.W4P1 (SUS) (sustained load case) 4.DI-D3_— (EXP) (expansionfrominstallationto 400°) 5.D2-D3__ (EXP) (expansion frominstallationto-50°F) These load cases allow us to determine the loads on equip- ‘ment and supports in three conditions: load cases 1, 2, and 3 (hot operating, cold operating, and as installed). ‘They allow us to compare sustained stresses to the allowable. ‘They also allow us to determine the expansion stressesas we heat up from installation to operating (70°F to 400°F) and the expansion stresses as we cool down (from 70°F to 50°F). ‘What these automatically generated load cases DONOT tell us is what the TOTAL expansion stress range is; from hhotteat to coldest. That is, we do not have a load case that describes the system behavior from 400°F to -50°F, even. ‘though this information is requiredto determine whether the allowable expansion stress range has been exceeded, Here are wo approaches tothe problemandan example that explains which of the two is more reliable. TWO SOLUTIONS? ‘The “Wrong” It has been suggested that we must first run the recom- mended load cases to determine the sustained condition and ‘equipment loads at both operating temperatures, and then perform a separate analysis based (using the above num- bers) on an installation temperature of -50°F, and an oper- ating temperature of 400°F, with no consideration of the true installation temperature. This second run, it has been argued, will provide the full expansion stress range of the system, Belief in this approach has prompted users to suggest that modifications to CAESAR IL are necessary to allow users to extract all required information using just one run, Although this first approach can sometimes provide correct resuls, it is also quite capable of providing completely incorrect results, depending on the piping and support configuration. Italso consumes time and unnecessary runs. ‘Some people have even assumed that it is valid to omit the ‘rue installation temperature altogether, setting tat -50° and operating at 400°F. In this case, not only can the expansion stress range be wrong, but the operating equipment and support loads will be wrong. Certainly, where loads are concerned, if the system is installed at 70°F, you want to see what the loads are when you cool to -50°F, shrinking the pipe, and you want to see ‘what happens when you heat from 70°F to 100°F, expand: ing the pipe. ‘The Right Approach To explain the correct procedure for this type of analysis, we'll first review the method that CAESAR II uses to determine expansion stress ranges. In a single operating temperature case, CAESAR IT first

    Das könnte Ihnen auch gefallen