British Engine Technical Report (941 Volume NT Non-Circular Pressure Vessels —Some guidance notes for designers M. Starczewski BSc.Eng. Stress Engineer, Research Dept. Introduction The aim of this article is to set down, in readily available fashion, the fundamental theory needed for the design of some typical pressure vessels of non-circular cross- section. Of these the most common arc the rectangular section tanks. ‘They are often used as bulk storage containers oF as baths in the treatment of metals and fibres and surface coating processes etc. For this reason vessels of this type have been given special attention. Other shapes are also included by reference rather than by a worked example, In explanation of the underlying theory a number of fully worked out examples are given showing the procedure which may be adopted when preparing the necessa: calculation sheets for different types of vessels. Such calculation sheets may frequently be required for approval by Inspection and Insurance companies as well as Certification Authorities. ‘There are no national or international standards or codes of practice that will caver all of the types. Here ASME VIII, Div.1, Appendix 13" probably offers the best guidance on a number of diferent designs. Unfortunately, in its present for is rather cumbersome and requires considerable time for pri understanding and assimilation. Another useful source of information on rectangular tanks can be found in the Theory and Practical Design of Bunkers?. Rectangular section headers are also covered by the Swedish Pressure Vessels Code®, British Standard BS 1113: 1969* and the Tealian Standard ANCC—VSR Collection Section VSR IS: 1978°. The last two references need to be viewed with considerable reservations as they appear to contain a number of discrepancies which are inconsistent with the fundamental theory. Where there is no relevant code, the procedure outlined in this article follows the same logic, based on fundamental engineering theory as used in the codes and should therefore be equally acceptable. Such procedure should be regarded as evidence of good modern day general ‘engineering practice in this field. Is hoped that it will promote a better understanding of the problems associated with such vessels which are often either ignored or not given the consideration and attention which they deserve. Such tanks can be ite complex in their detailed design and unawarenes on the Partofthe designer and/or fabricator, to appreciate the various aspects can lead to costly ramifications later on. These tanks although they appear to be very simple indeed. can nonetheless cause considerable embarrassment ifmot assessed adequately at the outset. Additional guidance is given on juare/rectangular ducting. lormally such ducting restricted to 20 psig (0-138 N/mm?). However, the rocedure outlined in this article no limitation per se. ‘Comparison of the rectangular vessels with the equivalent size cylindrical (circular cross-section) vessels indicate that the former are rather inefficient. Cylindrical vessels will sustain considerably higher pressures, for the same wall thicknesses and size, see Fig. 21 However, practical consideration will often ‘bree the designer to select a rectangular shape as the best available option. The fundamental theory is applicable to both external and internal pressure. Worked examples given in the text refer internal pressure for the simple reason that, for the external pressure application, considerable gaps still exist in the knowledge of the allowable compressive stress levels which will not cause buckling or plastic collapse in rectangular and other non- circular tanks. In such instances it should be possible to use the design data contained in the British Standard BS 449: 19698 for checking the main stiffeners and beams. Fundamental theory for rectangular section pressure vessels Figure | shows the basic geometry of the rectangular vessel with sharp corners and which is subjected to a uniform pressure ofp. where L = the longer span A = the shorter span +A = second mument of area of the beam BCB about its neutral axis 4, = second moment of area of the beam BAB abou its neutral axis Due to symmetry about axes AA and CCit will be convenient toNiw-canutay Previne Verses CITTT ttt > Fig I. analyse one quadrant only of the cross-section shown. This. juadrant is in equilibrium under the action of the loads and moments indicated in Fig. 2. Clearly from the balance of the horizontal and vertical forces acting on the quadrant we obtain which represents the direct tensile load in member BC, and pl. ™= >, ame the tensile load in member AB respectively. In evaluating these tensile forces the thicknesses /, and {, are considered to be negligible in comparison with dimensions such as Land h, ie. 2 L 2) effectively equals to = In beams and frames having igid joints, as in this particular ase, the strain energy due to the direct and shear forces is so small in comparison with that due to ending that only the latter need to be considered when evaluating statically indeterminate moments. In any member of a structure subjected to bending the total strain energy is given by 2 v= (ae veseeerereeee Al) lo 2EI where M is the bending moment at any point on the member caused by the combined effect of the imposed loads and the supporting forces and moments, whether statically determinate or not. The integration must be taken over the entire lengih of each member, of which dr is an Some Guidance Notes Jor designers element of length. ‘Two further postulates (Castigtiano's theory) help to solve the problem. These are: (i) |The partial differential Coefficient of the strain energy in a structure with reapect to a load F acting ona structure, it equivalent to the displacement of F along its line of action, i.e. ou _ fizMam OF Jo 2E1GF MOM Sere |, E1aF = @ (it) The partial diferential coefficient of the strain energy with respect toa moment acting ona structure is equivalent to the angle through which that portion of the structure rotates when the moment is applied au _{' Mem Oa, ~ Jy ETOM, When, asin this case, the support of the structure (point 4) does not give way under the action ofthe loading, then there is no deformation of the structure at this int of support and the two Expressions just quoted can be equated to zero. Frag & = 9% +--+ -B)British Engine Technical Report 1981 Volume SU By setting down the equation for moments along A ond BC and by considering the strain energy due to bending (by integrating. along AB and BC respectively) it can be shown that the moments at the diree important points 4, 3 and C become for a general case 2 My ix 1/(K+3-267 [#- 1 (BS )]..0 Mate x [: - (*) bepeccer 6) pe (*) Me = 34\ Fat where Lh A Rr 4 baz Notice that pl? (K+3-287 8 MY K+ where the first term denotes the bending moment at mid span for a simply supported beam BCB under the action of uniform load p. For a uniform wall thickness throughout, the parameter A K=a, which is the same as f and the three moment expressions simplify to the following 2 My PE x rs 64 B+t Jeo 1/R4+3—26? pi }: - 8) pL? (B+3—26" Me eG a where once again h B the ratio of shorter to longer spans. By substituting specific values for the parameter B =0,0-1, 0-2, ...etc. up toB = 1) ‘we can express the three moments ina very much simplified form My, = 4,92. ++(10) My = cpl? qa) Me = tpl? . (12) where a4, @g and a are the three new parameters which, for uniform wall thickness throughout, are dependent on * ratio onl L 2 The plots for these three. parameters are shown in Fig. 3, where after simplification these can be written as 2 +2) a, = EHR? 7 — B+ 1 oe ee “9 2p? +28 +1 nr 45 Identical plots to those shown in Figure 3 were obtained from the equations given in ASME VIII, Appendix 13! and the Swedish Pressure Vessels code? indicating that these are also based on fundamental engineering theory. When comparing the abov ‘equations with the correspon parameters given in these two codes it must be borne in mind that the latter? specifies the two spans as 2m x 2nso that the relevant constants & will diffs thy a factor of four, since M = a, pl? reference (1) and M = aspm? in reference (3). Thus for consistency aL? must be equal to. 23m? AS 2? = 4m? hence this factor of 1 is contained in the parameter a in all the expressions for moment given as M = apm? Hence it can be seen that the approach presented in this article will satisfy both ASME VIII, Div.1 and the Swedish Pressuire Vessels code for the plain rectangular vessels but reference: has still to be made to these codes for the allowable design stress level and the weld factor where necessary. From the plots shown in Figure 3 the moment distribution curve along each member can be quite easily obtained by the following method (a)_ For members BCB, span /. First draw to a suitable scale the free end moment distrib curve BCB which is given by the standard equation My =f. —Z} where x is the distance from point B. (Distance B~B to represent the n length L.) Refer back to ig. 3 to obtain the relevant bending moment at the corners, My = appL?, Draw a new “zero bending moment” axis 0-0 at a distance equal to My below the original datum line BB (as shown in Fig. 4a). The resultant sketch will give the complete bending moment distribution diagram for the longer span length, Moment al any point along BB is then simply given by the vertical intercept, either above or below = (16Non Circular Pressure Cessels ‘Some Guidance Notes for designers sper unit depth ECCT Re 0041; 90699 pot ~o70 0-08 =O08 t Fig. 3. the new datum line 0-0 as the case may be, In this particular instance Fig. 4 shows the moment bution curve while is applicable to the following three geometries: (i) datum line 0-0 for arectangular header whose h/L ratio equals 0-5, (ii) datum line 0-0" for square header, ic. k= £ and (ii) for built in beam where h = 0, The points of contraflexure are also shown for these cases. This sort of information could prove useful when che decision has to be made on the best location of the welded seam or any other outside attachment. (b)_ Formembers BAB, spank Simitar procedure to that dlescribed above can be used to obtain the moment distribution diagram for the shorter span. The only difference in procedure is that the initial free end bending a) ‘308 Vales of (for I, = 13) moment curve is now given by the equation 1 a M,, = Pee - i} where xis the distance from point B (towards A this time). The new distance RB should now represent, to the same scale as above, the shorter span A, The basic engineering theory and the above procedure indicate that each member of a rectangular section vessel can be treated as an iniially simply supported (ree end) beam uniformly loaded along its entire length which is then subjected to the end moments Mz, the latter determined from Fig. 3. This approach will be useful for calculating the central deftection of the members. This is illustrated in Figs. 4(b) and (c) and the plot for the central deflection of the longer span L is given in Fig. 5. (17) 10 So far we have dealt essentially with a uniform wall rectangular vessels. The preceding basic Uieury is equally applicable w rectangular vessels which have peripheral stiffeners spaced along the length of the vessel as shown in Fig. 6. In such cases we have to check not only the strength of the ters but also the stress levels in the wall panels between these stiffeners. The strength of the peripheral feners can be determined by the method described above, as for the plain rectangular vessels, by substituting ps for the uniform pressure load p used in the receding analysis. Equations (4), (5) and (6) can be used directly for a general case where the second moments of area of the stiffeners /, and J, and the wall thicknesses 6.4 of the two main sides are 65British Engine Technical Report 1981 Volume X11 Fig. 4. Central deflections yt pe sei ee Meme different. For uniform wall and stiffener sections Equations (10), (11) and (12) and Fig. 3 become once more applicable provide ps is substituted for p in the relevant equations. ‘The wall panels between the stiffeners can be treated as rectangular panels fixed (built-in) at al) four edges and subjected to a uniform pressure load p over the entire area. Reference (7) covers this particular case and gives the maximum bending stress, which ‘occurs at the centre of the long edges, as, where the value of f depends solely on the ratio of the two sides a/b, b is the width or the shorter span and fis the panel plateNow Giveular Pressure Vessels All edges xe, uniformly ddarbuted load p over the [ a NO 12 | I ‘95 OF os Val o 8 ‘Some Guidance Notes for designers thickness, Fig. 7 gives the plot iarihe varabie Bor various a/b ratios. Notice that for a/b values above 2-15 the parameter f = 0.5, giving vn 05th wont, os ost This represents the same situation that occurs for a built-in beam of span 6, Here the end moment pl 2 and the plate section modules for a unit wideh strip e i-5- Hence the bending stress at the built in edge M__ 6pl? pe o= 71g OO i.e. the same as above. This confirms that for walll panels whose a/b ratio exceeds 2-15 we can treat the central portion of such panels as a fixed-in beam of span equal to the width of the panel. Ouse farther devail whieh will require consideration is the solution for the corner wall panels, whether the corner occurs between the main side panels or between the side panels and the flat ends which may have transverse stiffeners. Such detail can be dealt with by evaluating the bending moments and tensile loads shown in Fig. 8. The information presented here is based on the basic theory contained in Reference (8) by combining the two separate loading conditions for panels L and h respectively. Basic information on critical moments and tensile loads is also iven for Gi tectangutar vessels with radiused corners—see Fig. (b)elliptical vessels—see Fig’ 10; 7 MsBritish Engine Technical Report 1981 Volume XIV (0) oblong vessels of uniform thickness—tee Figs. 11 and 12 Table | in the Appendix gives some basic equations for the simple geometries and loading systems considered in this article. WORKED EXAMPLES 1, Open top rectangular tank with continuous horizontal wall stiffeners Figure 13 shows the essential details of one such tank measuring 5500 x 2000 x 2500 mm deep. The tank is to contain liquid of specific thickness is considered to be weefial life of It would normally require several attempts to establish the optimum size of the stiffeners and \eir respective spacing. The following check will deal with the tank as shown in Fig. 13 in order to demonstrate the design method rather than the final choice. The pressure distribution on the tank walls will be linear and as shown in Fig. 14. The pressure at the bottom of the tank due to equivalent to | atii or 1 kg/cm? Presaure), or in Newtons per mm? this pressure is equivalent to 0759.81 100 = 0.0368 N/mm? or 3-68 x 10°? N/mm?, {A) Check on Stiffeners () Considering the first stiffener from the bottom, namely $1. It is fabricated from 300 x 100x8 mm rectangular hollow section whose properties are as follows density 1-5. Itistobesupported the zg mend ofliquid ofspecific =x = moment of inertia on beam members forming part of Gensity of |S will by = 2269 cm* the eral plant {structure x25 Zax = elastic modulus tank is to be built from 5x2! 2 = 207 cm 6mm thick plate material of P= = 0375 kg/cm’ 7 432 N/mm? ultimate strength. . = sectional area The corrosion effect on the plate {as 10 m head of water is = 45-1 cm’ = osidorn prowre oe > A, L being the longer span Fi 8 Value of 8 ICT] 3 Z tt : Zl 4 : L “ Li he ot 3 4 [TT TT Help‘New-Coveular Prevsune Vives — Some Guidance Notes for designers ety a 1 tee ord 0 2n18 ° 60:0, 800 400 90-0 500 0-0 200 z0-0 40-0 ° —T— ——- LELELI fo=uls'e"t | 2 +o (eols'z't ry + 20 (y-o)s'z't- NI = 1 5 | 7 | z leo 2 oN Me lui sity = wow0y saneSou sty sreonpuy x— (ones 7/y 4) x uonttod 1 y Jo anjeq,British Engine Technical Report 1981 Volume X11 Values of K, and Ky 0:36-——J 0-32 + 0-28| 7 | 0-24 Mtge a K. 4 Tension at point 4 + Ta = pea 0-20) ] =K] 016 NX \ 012 Ky (Negative) \ 0-08 04 0 02 04 bfa Ratio 06 LoNon-Gireular Pressure Vessels — Some Guidance Notes for designers : | (Wife thicken) | [4 fF bo e ” 2 3 % Za “Temion a A and B= pe Temion at C= Ala) 2 \ {| | o 1 2 3 4 5 6 7 6 10 W The val of the coeficien “a Fig. I. 1. t > i 3 Mae : Mea aed < i—-- Ps 20[—(Coitorm thicknes) ae 4 18 4 16 47 1 — |- | -+—4 4 r2p3 “re Tenia 3 4 Boe C= ree 6 4 2 o 1 — 10 20 wo 40 50 60 70 0 0 100 110° 120 “The vale of the coc“ Fig 22. 7British Engine Technical Report 1981 Volume XIV 3 3 8 8 = Plt thichnen = 6 mm. Specie deny of content = 1 ‘Al dimensioe ia mm ‘Al materia 432 Nimo? UTS Fig. 13. ‘The pressure load on this stiffener 29-87 x 10° (i) Ror iifener $2, abriented per mm oflength (span) will be, 8 = 59757999 from the same hollow section, the according ta Figure 14, Pr = $(25243-32) x 107? a ‘500 + 560" 2 = 15-476 Newtons per mm of span. From Fig. 3, for the A/L ratio of 2000 20 = 0364, 3500 7 056 the max. bending moment Mg is given as My = 0-0638p, 1 = 00638 x 15-476(5500)* = 2987 x 10% Numm Since the section modulus Z of this stiffener = 227 cm? (or 227 x 1000 mm’) hence the ‘ending stress at the corners of the stiffener will be Rn In addition there will be a direct tensile load acting on the stiffener. This tensile load can be calculated as follows PixL _ 15476%5500 2 2 = 42559 Newtons Hence the direct tensile stress at corners (or along the shorter spans) is given by 42559 = = 944 N/mm? 8 = aera 7 S44 Némm Thus the combined maximum tensile stress in the stiffener $1 is given by Oy = Oy + Op = 131-6 + 9-44 = 141.04 N/mm? Th corresponding moment, tensile ~ force and stress levels were found to be as follows ‘The pressure load pa = (1-64-4252) x 107? 560 +640 2 = 1248 Newtons per mm of span. x The max. bending moment My = 0-0638 x 12-48 x (5500)? = 24.086 x 10° N.mm, he bending stress og _ 24-086 x 106 “227 x 1000 The direct tensile force 1248x5500 2 = 34320 Newtons = 106-1 N/mm TsSo that the direct tensile stress 34320 = ———_ =7. 2 = 1x10 ~ 76 Nim Thus the combined max. tensile stress in stiffener 52 is given by Oy = 1061 +76 = 113.7 N/mm? Ttappears that a lighter sifeher seein could be used at this position, such as 200 x 100 x 6:3 mm hollow section which will result in the combined bending stress level of op = 2HO86x 10% | _ 34320 T= [85x 1000 ~ 36-0%100 = 130-2 +95 = 139-7 N/mm? (iii) _Stiffener $3, fabricated from 150 x 100 6-3 mm hollow section. ‘Nan Circular Pressure Vessels — Some Guidance Notes for designers Ds = $(0:58 + 1-64) x 107? ‘640-+800" 2 = 7.992 Newtons per mm span My = 0.0638 x 7.992 x (5500)? = 1542 x 106 Nmm 7.923500 7 —e_ = 21978 Newtons and 21978 : % = pe7xi09 ~ ** Nim Thus the combined max. tensile stress it 134.9 N/mm Stiffener S4, the top flange, fabricated from 100x506 mm channel section, Z,, = 41. cm? and A= 13-2cm?. Pa = $x 0-58 x 400 = 11-6 Newtons per mm span My = 0-0638 x 11-6 x (5500)? = 22.907 « 106 Nunn w_ 22:387x 106 7 oe = 411000 = 546 N/mm? Ty = LU6x5500 = 31900 Newtons, and 31900 : & = Teaya 7 242 Némm Hence the total combined tensile stress Oy = 78-8 N/mm?. ‘This appears to be rather lightly stressed by comparison with other stiffeners but normally a heavier section is required at the “top flange position” for handling purposes. 36810" Nia? Fig. It 3British Engine Technical Report 1981 Volume NTI i) The lowest wall panel (between the floor plate and the edge of stiffener $1) is loaded as shown in Figs. 14 and 15. The actual trapezoidal pressure distribution can be substituted by uniform load case shown in Fig. 15(a) and a triangular load case shown in Fig. 15(b). These two cases can now be evaluated by the basic engineering beam theory. From the equations contained in Table It of Reference (7) the bending moment at point B will be 2 My = PZ for case (a), and 2 PAE’ for case (b) M, = 20 giving a combined bending moment at this point ey Ba) 2 m= (+ +2B)e ‘Thus, substituting the relevant values we have 3.04 |, 064 A My= ( + 3) x10 x (450)? = 5778 N.mm per unit width. Now the section modulus of a unit width af plate of thicknese tic given by 2 cord = tat the bending stress at point a= 6M - 6x577-8 ee 963 N/m (i) Second wall panel, between a ‘Case a) Case b) Me 3.04- + HS Fig 15. ‘edges of stiffeners 51 and S2, L= 460mm. x (460)? = 459-9 N.mm per unit width. 6x4599 e = 7665 N/mm?. (i), Third pancl, between edges ( eneas and 53, L = 550mm. 124 UBT) My = CE wt )* x 107? x (650)* = 435-1 Numm per unit width and 6x435-1 =~ = 7252 N/mm}. C._Floor panels and stiffeners (8) _ Floor panels, between the horizontal stiffeners which are spaced at 500 mm pitch, see Fig. 13. Here b = 500-80 = 420mm. Using Fig. 7 for the builtin ‘plate panels, where a _ 2000 2M = 47 37 a9 = 478 which is. reatet than 2-15, we have the bending seat builtin - 2 a vba x 420 = 90-2 N/mm? {ii) Floor stiffeners, fabricated from 120%80 x63 hollow sections, Z = 746m? Treating the stiffencrs as built in beams, the load per beam = 3.68x 107? x500 = jewtons per mm of “Therelore the bending moment at fixed ends pL? _ 18-4 x 1800? TD 12 and the bending stress M _ 184 1800 ~ % ~ 12% 74-6x 1000 = 66-6 N/mm? My=The preceding analysis indicates that the stress values at various locations do not exceed the allowable stress levels confirming that all the details considered above are acceptable and no farther refinement is really necessary. 2, Enclosed Rectangular Tank with Flat Bolted Top Cover with Gasket Seal Fig. 16 shows the essential details of the proposed tank. The tank is to contain liquid of specific density of 0 up toa maximum depth of 1-2 m. The “gas space” pressure = 0-069 N/inn? (10 pig). Because of the presence of the flanged joint the cross-section of the tank cannot be considered as an integral entity. In this instance the cover and the tank have to be treated independently. (i) Consider the cross-section of the tank itself. The vertical ‘Non-Gircular Pressure Vessels — Some Guidance Notes for designers stiffeners on the side wall can be considered as beams builtin at the lower and simply suj ted athe lange acejorg ete and loaded as shown in Fig. 17. The pressure at the bottom of the tank will be given by fa = 0.069 + 0.012 = 0.081 N/mm? The pitch between the stiffeners = 440 mm. Thus the loading per unit length of span will be py = 0069 x 440 = 30.36 Newtons per unit length of span AB, and bs = 0012 x 440 = 5-28 Newtons per unit ofspan AB. The two loading conditions shown above will produce the following bending moments at base (point B) pe My = PS for case (a), and 2 M,= ae for case (b), assuming that the triangular load distribution is over the entire length of 1-5 m instead of 1-2 m. This will give slightly conservative results. Thus the combined bending moment at base will be M,=M.+M, = 9330750 N.mm Hence che bending stress at this location cidBritish Engine Technical Report 1981 Volume XIV M __9330750 Z 746x100 = 125.08 N/mm In addition there will be a tensile load equal to pak 2 acting on each stiffener. Therefore 0-081 x 440x 1500 223-4 x 100 = 11.42 N/mm? So that the total tensile stress or = 125-08 + 1142 = 136-50 N/mm?. (i) Consider the floor panels. space between edge of stiffeners, b = 440-80 = 360mm = 5. The length of these panels a= 1500, so that a _ 1500 Stam tl which again is greater than 2-15. Hence from Fig. 7 the bending stress at built-in edges is given by a=0(!) ‘Thus = op = 05x oon (= . = 145.8 N/mm? The longitudinal direct tensile load = 0.081 «1500 4 giving a direct tensile stress 0-081 x 1500, x6 = 5.06 N/mm?. Hence the combined tensile stress at the built-in edges of the floor panels is oy = 150-86 N/mm? 16 Op = |. Consider now the corner nels (see corner detail i 6). Here k = 200 mm and Pe 266 mm, p = 0-081 N/mm?, Therefore and from Fig. 8, fe = 0-161, 30 that the bending moment at point C, sce sketch ~ Fig. 18, will be: Mg = 0-091 x 0-081 x 260? = 498-3 N.mm per unit width Hence 6Me _ 6x 498-3 ware = 83.05 N/mm? ‘The tensile load at this point will be 7, = Ph 0081 x200 ne 2 = 81 Newtons. Hence — = 1-35 N/mm?, thus giving a combined tensile stress at Cequal to 84-4 N/mm? (iv), Consider the cover itself. “The horizontal stiffeners can be conveniently treated as simply supported beams under uniform pressure load. The bending Moment at mid span is then given by Maia = Here p = 0.089 x 440 = 30-36 Newtons per unit length of span. ‘Therefore Mig = 8538750 N.mm and the bending stress 8538750 7465x1000 = 114-61 Njmm? The direct tensile load (0-069 x 440) x 1500 Ce = 22770 Newtons, O,= which gives a direct tensile stress of‘Non-Gircular Preware Vessel — Some Guidance Notes for designers Fig. 22770 0 = 33.4% 100 = 9.73 N/mm, Hence the combined tensile stress level = or = 124-34 N/mm?. The above calculations indicate once again that the plate panels and the stiffener sections are adequate for the loading conditions specified. Now that the overall design check has been made on the main components the designer can carry out detailed calculations on other features particular to his case, such as flanged joint details and the adequacy of supporting sections. Note that if this enclosed tank had been of integral construction the max. combined tensile stress level would only be 101 N/mm? Compare this with the 136-50 N/mm? calculated at the base of the vertical stiffener under item (. (a) Uniform wall thickness throughout~no stiffeners In such cases we simply determine, from Fig. 3, the max. bending mament and the direet tensile load at the corners, ie. My = appl? and Ty = & for a given A/L rati The combined ma: stress is then given by Or = Oy +o “ell where Lis the larger span and tis the plane thickness By establishing the allowable design stress level for the plate material d,y, either from the appropriate standard or other sources, we can then directly calculate the max. design pressure for that particular geometry (h/L, L/tratios) by substituting ayy for 18) 6, in Equation (18) and re~ arranging to give the following relationship. p= (19) Equation (19) has been used to establish the plots shown in Figs, 19 and 20 for an allowable design stress level of 155 N/mm? (22475 psi) for square and 2:1 ratio rectangular ducting, the latter proving to be the optimum design ratio. The band between the two curves (shown in Figure 19) covers the whole range of rectangular vessels from AIL = StohjL = 1-0 or for Ljh = 1-0 (ic. square ducy) w Lih = infinity which refers to built-in beams under uniform pressure load. These plots can also be used to determine the max. design pressure for different ,y values or where the weld factor needs to be considered. In such instance thep value determined from Fig. 19 or 20 as the cate may be, where ayy is the appropriate allowable design stress and Eis the weld factor. Fig. 20 also indicates how inefficient the rectangular cross- section vessels are in comparison with the cylindrical (circular cross- section) ones, the latter being able to carry 20 to 120 times higher essures for equivalent size, for Ujtrange between 20 and 130 respectively. (b) Uniform wall thickness with uniform section stiffeners The approach is the same as for case (a) above with the exception that the uniform pressure p is replaced by px pitch, i.., ps in all the expressions containing the pressure term. ”British Engine Technical Report 1981 Vilume NIV 8 é — equal 2:1 (or 05) ducting Rectangular larger side +——}-— ; t= plate thicknes , + E Tnternal Pressure )—pal 2 s € 8 3s = 8&8 & 8 P—Newtons/mm?, & & z g Square ducting : SHH 01 ‘Lit Ratio 8Scale ‘A’ 4—psi—Tube Fig. 20. ‘Non-Cireular Pressure Vessels — Same Guidance Notes for designer 149 1 = plate thickness 140 60 (L/d) Ratio 80 120 Scale ‘B’ —psi—Square DuctBritish Engine Technical Report 1981 Volume XIV In addition the plating between the stiffeners needs checking by making use of the information contained in Fig. 7. Other Design Features (1) Corner joint where the main stiffeners are not continuous, ie. not welded at corner joints In such instances the stiffener sections can be checked by treating these as simply supported beam si span L and h (as before) under the action of a uniform pressure load of pxs, where sis the pitch or the distance between a ir of stiffeners. Hence the max. ing moments at mid span potitions will be M = 0:0833(ps) 1? for span L, and M = 0-0833(ps) A? for span A. ‘The direct loads will be wt ang HE T zy and respectively, faving checked these we then need to assess the strength of the “corner angles” themselves. Consider the construction detail shown below. (Such details do still occur on a number of hoppers or silos here as well as in other countries.) ‘The corner angle will be subjected to the direct loads pol bok Ee ©) as well as a uniform pressure load ‘over the two arms of length J. these forces create a bending moment M as shown which cause the rotation of the two arms shrough an angle 6. This angle @ can be determined by the following method. For a simply supported beam of length L loaded by a uniform pressure of ps the ends will rotate through an angle 8, given by O where E = Youngs modulus, usually 29—30 x 10° psi, for common steels (207 x 10° N/mm? and J, is the moment of inertia of the beam section of span L. ‘Thus the combined angle of rotation 6 = (0, +8) x4 =H (2 4B) ct o(F+F 9 2p aA (Eh). Eh," T, ‘Now for a cantilever of length /, the end deflection due to an edge moment M, is given by — Mt El, ane where ris the mean thickness of the angle arms. ‘Combining the two equations 6,,80 that Fig. 21. + (25) a ee(D M = 'se\1, * Th and the bending stress is given by 6M, oO pst (Bw =F (=+ 7). 96 \i, hy, The additional bending stress due to pressure load over the span twill normally be small by comparison with o above and can be safely ignored Exam of Equation (26) indicates that the angle mean wall thickness will need to be fairly substantial in order to keep the bending stress levels within acceptable limits. It is much safer and more economic to weld the two main stiffeners together so that they form an integral rectangular structure. (26) (2) Floor or supporting structure-local buckling loads Asan example let us assume that we have a rectangular vessel which has a number of floor stiffeners of a certain cross-section.The vessel is resting on a number of discrete points so that the max. load per support = 10 tonnes D = 257 mm overall depth d= 240 mm 6 = 127mm t=61 mm D_ 257 Ta”? as shown in ee Since D/t ratio for the stiffener section is over 35 suggest using the angle of 60° for 8 shown below, so that the effective width of the floor stiffener web taking the compressive load of 10 tonnes = 6 + 0-866d = 127 + 208 = 335mm Hence the cross-sectional area of the web acting as column under compression = 335 x 6-1 = 2043-5 mm? and the compressive stress. = 10x 1000 2043-5 = 4894 kg/mm? = 4894 X 981 = 48.0 N/mm? The length of the “column” 240 mm. Since the ends of this column are restrained in direction at both ends, the effective length = 0-7L = 0-7 x 240 = 168 mm. (See BS449, Appendix D.) ‘The slenderness ratio for this columa = L 168 of. From Table 17a of BS 449, Part 2, 1969 the allowable compressive stress p, = 84-5 N/mm?. As, > o, the floor stiffeners do not require reinforcing gussets. Note that the radius of | tion can be found from the following A = cross-section area = 2043.5 mm? {= moment of inertia ah - = 35X61 6336.553 mm* 12 But I= AP so that r= S310 = 1-761 mm (3) Flanged Connections Straightforward flanged connections are a common feature Ww pressure ducting, hoppers, silos and storage tanks. Sometimes these are introduced in order co facilitate handling, transport and/or erection at site. On Circular Pressure Vessels Guidance Notes, designers occasions the decision to introduce flanged connections is made following the release of the manufacturing drawing so that the designer is not really aware of their existence. On other occasions the designer or the manufacturer simply fail to appreciate the difference or the weakening effect of such an introduction on the strength of the component. Consider apical fange deal shown in Fig. 23. The effective width of the fange resisting the bending moment can be taken as b= 2tan60°xh = 3-464h. Note that if the two adjacent distances b so calculated overlap then b = 5 the pitch between bolts. Let F = the load per bolt pitch acting on the wall of the tank, For rectangular tank this force would be equal to 1 se prov example The moment at points B and C (see Case 9, Table I) M=Fxh. The section modulus of the vertical leg of the angle is given by oP _ 346482 wee; and the bending stress then becomes M__6Fh oe eo ggg (27) For a close pitched bolt the weakening effect of the bolt hole has to be considered. In-this case the section modulus = (s-ae So where d = bolt hole diam. In calculating the bending moment M = Fxhabove we have ij the effect of the pressure acting on BCas the flange opens. This effect alBritish Engine Technical Report 1981 Volume XIV is negligible compared to the product Fh. In addition to the above we need to check the stress level in the bolts. ‘The bolt load can be determined by taking the moment of forces about point A, the tip of the vertical leg of the flange, i. Fx (AC) = Lyx AB where Ly = force in bolt. Therefore FxAC AB tis clear that depending on the two distances AC and AB the load experienced by each bolt can be considerably higher than F the nominal load per bolt pitch. Hence the distance h should be t as small as possible tively making distance AB oy ‘This will also result in Tower bending stress in the flange as given by Equation (27). (4)_ Differential stiffener deflection For the rectangular vesels of the type shown in. ig! 13 it may be wBrch while to check the central deflection of each stiffener by using the information contained in Fig. 5. This may be expecially important for cases where section properties(moment of inertia) vary for each stiffener. ‘Where the central deftection for adjacent stiffeners differ considerably we need to check the additional bending moments imposed on the plate pancls between the two stiffeners. This can be accomplished in the following manner. (Fig. 24.) st calculate the central deflection for each stiffener using information given in Fig. 5. Then calculate the differential deflection between each adjacent ir of stiffeners. Let us call this differential deflection y. The additional bending moments imposed at the built-in ed; the plate panels are then given by where E = Young's modulus ° 1 = moment of inertia = 3 1 = distance between stiffeners 1 -= plate panel thickness The above “correction” procedure is particularly Important for plastic, or GRP tanks where the deflections can amount to several inches. (5) Weld Factor (E) Most of the codes tell us the weld factors we can use in the design calculation without specifying what this quantity really means. We know that the value of E (the weld factor) depends on what we do prior to, during and after fabrication, i.c., weld procedurc, weld preparation detail, welding method and X-ray or ultrasonic examination. The more NDT employed the higher the weld factor. ‘One rational definition of the weld factor would be to imply that the quantity (1 - E) represents the size of the defect situated somewhere within the plate thickness, as shown by an insert in Fig. 25, The various equations given in the codes are (with minor correction factors) variations of the following: Direct stress pd op =o QE - (20) and Bending etrece my = pe (2) E or in the more usual form‘Non-Ginwlar Pressuir Vessels ~ Same Guidance Notes Jor designers ‘The above clearly indicates that the correction procedure is the same for the direct and bending stress conditions. On closer examination this appears tobe rather strange. Yo check the validity of this approach to predominately bending stress situations the writer has calculated the section modulus Z for a plate thickness { which contains a planar defect of size (I— E) with the tip of this defect at a distance x from the free surface. The resultant plots for two weld factors, E = 0-75 and E = 085, are shown in Fig. 25. ‘The evidence shown here indicates that the present code Defect (1~£) 0.10 0.20 0.30 0.40 0.50 Distance from surface (x4) &) Fig.British Engine Technical Report 1981 Volume XIV method of using the weld factor, as in Equations (20), (21), is safe for situations where the defects are entirely within the mid plate thickness, i.e. when x is within the limits of 0-254 and 0-75s. The approach becomes increasingly unsafe when () x becomes less than 0-2234 for E = 085 case, and (ii) x becomes less than 0+185¢ for E = 0-75 condition respectively. ‘One clear conclusion from the above is that for details subjected to predominately bending stresex the unfused land, lack of penetration, or lack of fusion, should occur within the central portion of the section. The local weld preparation should therefore be designed accordingly. (6) Ligament Efficitacy {E,)-Rectangular Vessels References 1, 3, 4 and 5 give various examples of how to calculate the ligament efficiency for various conbgeration of tube holes. All the efficiencies quoted seem to be based on the fallowing standard equation pod, E= >? where p = pitch between uniformly spaced holes, d, = mean effective tube hole diameter. For plain holes the method of calculating the ligament efficiency is the same for both the direct, or membrane, and bending stress situations. The procedure is strictly applicable to flat plates which contain a series of holes. The stiffening effect of the nozzles, themselves is completely ignored. Such vessels, with perforated side walls, probably do not exist, In practice the tube holes will be reinforced by the tube stubs as shown in Fig. 26. The height of the stub and reinforcement is given by a= 064 Jai, where d, is the mean nozzle dia.,and ¢ is che nominal nozzle wall thickness, From these two diagrams, the side containing the nozzle openings could in fact rove to be stronger than the plain (inperforated) wall since the former is effectively reinforced by a number of stiffeners. Thus the proposed method of calculating the ligament efficiency can be too conservative, especially for larger ‘openings. Limited strain gaug roof tests carried out on similar eat exchangers seem to confirm this conservatism and the strengthening effects of nozzle stub ends on side wall panels. Conclusions The basic engineering theory outlined in this article shows how we can check the design of a number of non-circular cross- section pressure vessels. ‘The worked examples demonstrate how we can represent various details of a rectangular tank by replacing them with simplified geometries which can peel be! evaluated ee the fundamen ineering theory. We mutt endeavour to make exch theoretical representation as close to the real component as possible. ‘The closer the approximation between model and actual detail the higher the allowable design ttres levels we can adopt. The simplification procedure and the degree of sophistication needed will depend on how arduous will be the intended duty, on the confidence of our knowledge of the material properties and other factors. Itis important to realise the implications of the simplifications and assumptions which have been made. If the theoretical model, or the simplified geometry, is too far detached from the real detail the design calculations may become In other cases we can compensate for any simplifications by using much lower design stress levels or by ensuring that the theoretical representation is conservative. ‘ireular Pressure Vessels — Some Guidance Notes for designers Personal experience and knowledge of the fundamental ‘engineering theory will dictate the course of the appropriate action. This approach is certainly not recommended for the beginners. {you are one then seek advice. (is hoped that by outlining some of the salient features of the non-circular pressure vessels this article would prove useful to the designers and fabricators alike, and that it would, in some small way, lead to fewer failures of the type normally classified as duc to oor or inadequate design. Retornene 1, ASME VILE, Division 1, 1980 Appendie 13, 2 TheThertpd Pagal Desgnt Bunkers 3. TheSwedish Presure Vesls Code—1973. 4 Beis Sandard 1113: 1969 Amendment Nort ove 20.3 grew Fae Reocanguia Set 6. British Standard 449: 1969—The eof ‘Stractura Statin Building. 7, Foemilas foe Stress and Strain by Raymond SioutoMcGree ta” 6, Rigid Frame Fe Right rane Formulas by A Appendix The Basic equations for moments, deflections and loads for the simple geometries considered in this article, are given overleafin Table | for quick reference.* Butch Engin: Technval Report (01) Volume XIV Loading system Moment, Deflection, Bending Moment ‘and geometry Rotation Diagram E 2 pe Max Mf, = ’ ic B 4 8 Me ions - i} 2 Me M Se a al 4 B 4 ® 0, = Met L 3. pl? Max My = aa pl? 48ETNun-Cinatar Pusure Vessels — Some Guidance Notes for devigners Loading system Moment, Deflection, Bending Moment ‘and geometry Diagram = A B = Mel % 4El Max My = My = a ott we ne lt 7 RAED Max A, =British Engine Technical Report 1981 Volume Loading system and geometry Moment, Deflection, Bending Moment Rotation i Diagram wt Lge (E482 Ms Help ola pe 1/k+3—26? me (Rae) pL fk+3—26? wat Fel 1 [eeeeere| oh L Tee= 5. Toa = Note: For cases Nos. 11, 12 and 13, moment of inertia J, refer ta the longer spans L and J; refer to the shorter spans A respectively; Ink . A bet Nekth Baz. Eagland by Waadfiold Printing Ca. Limited. Raruey