This action might not be possible to undo. Are you sure you want to continue?

Paper

**Junction effects In St Venant's torsional stiffness
**

C. J. Burgoyne,

Engineering

**MA, MSc, PhD, CEng, MIStructE,
**

Cambridge University

MICE

Department,

Synopsis A method is presented for dividing an open cross-section into components, Computer analyses have been undertaken to determine the effects of the junctions between components, which are significant in many sections, The results of these analyses are presented in the form of charts, which can be used in design offices to find equivalent sections whose inertia can be calculated by simple formulae, No computer analysis is needed by users of the charts, The validity of the method is demonstrated by calculating the torsional stiffness of a Y-beam and its flange, The result is compared with accurate analyses, which have been used to calculate the torsional stiffness of the complete range of Y-beam sections. Introduction

The calculation of St Venant's torsional stiffness for a complex open crosssection is not straightforward, Unlike the flexural stiffnesses, which are functions of simple integrals over the area of the section, the torsional stiffness depends on the solution of a differential equation which is valid over the whole cross-section, It is thus not normally possible to divide the section into a number of regions, for each of which the stiffness is known, and then to combine the results to accurately determine the properties of the overall section, I fit is required to calculate the torsional stiffness at present, the engineer must either solve the governing differential equation (which usually involves a complex numerical calculation) or make a number of simplifying assumptions which introduce errors of an unknown magnitude. The commonest approximations make use of the well-known analytical solution for a rectangular section I. This shows that the stiffness is a linear function of the shear modulus of the material (G), the largest dimension of thc section (b), and the cube of the smallest dimension (I). The multiplying factor is a function of the aspect ratio, varying from 0.14 when b = t to Y\ when b » t. Estimates of the stiffness of more complex sections can be obtained by assuming that they are built up from a number of rectangles; the stiffness of each rectangle is determined taking account of its aspect ratio, and these stiffnesses are summed to give that of the whole section. A recent review paper by Johnson' studied most of the available methods. There are two disadvantages of this method that apply most significantly to concrete sections. It is not always obvious how best to divide the section into rectangles since sections may not have parallel sides and, with low aspect ratios, the choice of which is the 'long' side and which the 'short' side can cause significant changes in the results. Secondly, the torsional stiffness contributed by the junction areas can be much larger than that from the rectangular regions; Johnson' concluded that this was the most significant cause of error when estimating the torsional stiffness of concrete elements. The junction effect is particularly important when sections have chamfers and can cause serious underestimates of the torsional stiffness. If individual rectangles have high aspect ratios, the error in the total stiffness is small; this is frequently the case with steel sections in which the breadth/thicknesses ratio for flanges and depth/thicknesses ratio of webs are usually of the order of 10: I and 100: I, respectively. The thickening at junctions is not usually significant, so both errors mentioned above tend to be negligible. In addition, most steel members belong to standard ranges, so that an accurate 'once and for all' numerical calculation is carried out; the results are published along with other section properties. Concrete sections, on the other hand, cause more problems. The aspect ratios of the components of the members are frequently less than 5: I, so that approximations which rely solely on dividing the section into a number of rectangles are not satisfactory. It is also much less likely that they are standard sections. This paper presents a method by which the torsional stiffness of a complex section can be calculated without recourse to solving a complicated differential equation. All the numerical coefficients required will be provided

in the form of design charts for which all the complicated calculations have already been performed.

**St Venant's theory
**

The derivation of the differential governing equation is well known and will not be repeated here. A stress function .p must satisfy

a'.p ax'

+-

a'.p ay'

2

· ... (I)

**over the cross-section, inertia J is given by J
**

=

with .p set to zero on the perimeter.

The torsional

2j.pdA

· ... (2)

and the maximum shear stress occurs where the slope of the stress function is a maximum. Hence

T max.

· ... (3)

Analytical solutions for this problem are not usually available for sections of practical interest; most solutions must be found numerically. Fig 1 shows contours of.p for a precast Y-beam section with an in situ flange, calculated by an accurate finite difference method which is described below. The contours are at equal intervals in the value of .p; the particular values are of no concern here, but the complexity of the shape of the function is immediately apparent. The actual properties of these particular sections will be used as an example later in the paper.

Fig I. Contours of the stress-function

.p over a Y-5 beam with a 200 mm

flange

47

The Structural

Engineer/Volume

71/No.3/2

February

1993

node rectangular at every which four function relates uses finite differences (I) subject within to the and successive appropriate covers at that a finite boundary conditions.3. n = 1. conditions . the stress the end effects is unchanged. against an asymptote how rapidly (. (1). to be considered I 7r ~2 _L E b n (J. Fig 2.~~-. would what formulae. 32 c. to solve making elements and two outliers. but clearly does not take into account becomes becomes which indicates end of the taper.315124) . alternatively.~~-program is given.5 n 21 b)] with b which All of the results . to the such to be is not which and was written l t b The program to solve eqn. along the edges. this function to be useful. elements but this five non-zero diagonal. the factor at the outer be ignored. It can be seen that A. I (Fig 3(a». section. difference is set up D '" '" ~ C a w ::l c: 30 - of the function At points which allows neighbouring on the boundary. the aspect as b/t is bit. Stress function for an infinitely so that it satisfied use of later were analysed effect condition sections one end that if. indeed. the asymptote b/: > 1. It should is less than of this result of a rectangular is reduced inertia When implication constar. will be needed sections. which which that that result form. eqn. Tirnoshenko the torsional r fl7 significant. but this conditions of thickness with can be of the half-angle that due to the boundary if. and even at the St t214 -x2 ljJ= b= The I the difference It is possible with great by 0. function than a certain represent (but as yet undetermined) It would be convenient member by 'lA. the inertias A. (6) n5 The ratio series converges rapidly.Paper: Burgoyne Computer A computer ---.3. rectangular section for section from which an analytical in the inset he obtains solution does exist inertia is the simple the shown which in Fig 2. in matrix from this finite difference have many elements on each allows technique in terms been iterative row. here that the end effect the real rectangular could be ignored. Rectangular section: reduction in length due 10 end effect in computer A penalty Rectangular One stress section function. will be made to be analysed obtained a procedure section. tapers has a thickness I. without modifying This allows the program. section ratios the value of the stress localiy parabolically. points. (J-ton2«) the end effect completely. which to be increased section.. effects the way to of a crosssection' analytical a numerical of 2(1 for obtaining take can the various for other which stiffness most account of the stiffening by determining by simple be possible for the the torsion of the interaction components be determined it will to that not used sections analogous rectangle. approaches as expected. Tapering section analysed 48 The Structural Engineer/Volume 71 /No.. to find suitable Thus.3151 to calculate Venant's that calculating When from accuracy by assuming There the cross-section the torsional the aspect at each end. (a) Rectangular section (b) Tapering section b/t is 3.51 long at This need result. with the boundary distribution to decide trivial. has been kept very flexible.. the boundary 29 l I _l Aspect ratio (bit) can be specified. varies The first term term but represents is a measure the inertia of the at high of a rectangle 'end effect'. ' . There would technique. c: E c: '" 0 if> 31 -- program program was already available to solve in ref. procedure standard numerically..5 ~ """ (1 -tan- h mrb) 2t Fig 2 shows be noted 9%. is potentially The format a wide range program . ' gives No problems a problem of sections for the data presented non-convergence. the slope the function the slope This allows must is zero). reached. to analyse (1) over a range Fig 4.tanh 1. extremely of execution found with techniques. boundary be expected end become a dominant large call 96 TIs ex and cannot However. the equations the solution is paid with have input below The use of an iterative the non-zero efficient time. if.(4) have been obtained with the use of this Tapering section analytical solution is available solution for a tapering section. gets larger. the value of A. by the factor!' by an equivalent is reduced by (Fig 3(b i). with a taper a) over satisfies dominant infinite. larger is present for aspect will be seen not to be a complete at the open end of the taper. This factor this the boundary for large values that for factor of the equivalent sections will be given identically at".7. is a function It will be assumed dimension which ratio but that the larger of the aspect rectangular conditions the open we wish to determine. specified boundaries the value of the slope can be fixed. The for the work described where a full description over-relaxation (SOR) A and at the of the in either to be also (when would.t. of Ol is negligible For a long thin rectangular function For a section = I1(I-tan' function It is postulated if we could in which dimension where Thus. is within IOJo of the asymptote. is presumed of the function be zero) to be prismatic representation on or adjacent the x or the y direction (across across axes of symmetry necessarily it will be specified The equations if expressed be only leading as SOR. A. will be constant value. by itself is relatively a well-known similar results equation. while for the all A partial second rectangles.. is only trivially long section. stored. and then complex if a section of if. the equations. Fig 3. grid of nodes the value is set up which the area of interest. which gives (5) of the Ol. = 45°. memory.3. is clear.3 / 2 February 1993 . only three that were zero. is no need to solve a complex ratio the result equation or to remember Thus. for low values of ex. If the section to a point. the answer is easily But it points components an 'equivalent formulae. between whose For solutions. eqn. different 1. = 0 to be specified. expression node to the values the value condition the section. section in length by ignoring values of A. that the smaller a peak value of shown t' I4 on the centreline. the correct in the paper The determination by rearranging of A. aspect » I.. is prismatic. ratios.

when It exceeds I.) of prismatic rectangular section. Instead. the reducing section (Fig 7(a)) is replaced with an equivalent length of the wider section At' as shown in Fig 7(b). and are shown inset in Fig 6. values of of never follow the f'r / 4 curve predicted by the partial analytical solutv n However. for the narrowing section there will be a reduction in length by \. The value of of at the outer end. Curve A in Fig 5 shows the value of of along the centreline of a taper section with a = 20°. In the first case. For small values of ce. of is defined as being prismatic. values of \. as though the section continued indefinitely beyond this point. (11) The Structural Engineer/Volume 71/No. the St Venant's torsion constant can be found from 1. is very close to that expected in a prismatic section (even though only its normal slope was specified there). consider a section which tapers from a thickness I. = 0. An equivalent section is sought for which the torsional stiffness can be calculated by simple formulae and which has the same torsional stiffness as the real section when calculated by the program. but of opposite sign. to zero over a length I.. -2 _ + At) I. (9) Actual section 4 These two expressions can be used with the results of the computer analysis of typical sections to calculate values of \.153. for smaller cc.51. while for the widening section there will be an increase in length by the same amount. the other widens from the prismatic section with the same angle. will be quite large.. gives J= Values of At' derived from eqn.-------. away from the tapering section.51. and 1. and A". which can now be used as a design chart. a positive value corresponds to an increase in length. o 30 Posit ion along section Fig 5. one narrows from a prismatic section of width I with angle o . Tapering and reducing sections with the centreline 0< 20°: values of of along A" will be negative for large values of a.. Taper section: alternative equivalent sections To study these effects in more detail. The section extends for a length of 1.3/. .orsional stiffness can then be calculated from J = (/. to I" with a taper angle « as before. the calculated value of of is close to this higher value. Integrating the analytical stress function discussed above. always corresponds to decreases in length._-----. 12(1 . These ideas can be extended to a reducing section of length It' where the width of the section tapers from I. are shown in Fig 8. An alternative for small a is to calculate J for the taper using eqn.tarr'«) + 3 · . (8) o+-------./~ fr~-~-J---_ t . over the taper area. + I" the end effects would become independent.51.) .tan'". but to take account of the junction effect between the taper and the prismatic section by changing the length of the prismatic section by \. for very Iowa. Care should be taken with the signs of these two correction factors.l. Extending the section in this way means that the results used here can be applied when the taper section forms part of a more complex shape. for which J A . (Fig 4). depending on a. The length of the reducing section is too short for the end effects at the start and finish of the reducing section to be considered separately.) + 1.Paper: Burgoyne Reducing section (curve 8) Tapering section (curve A) t214 \ \ -- \ taper.(10) -4 Fig 6. Near the point of the taper.51. Thus. A" can correspond to either an increase or decrease in length of the prismatic section. (10) and the computer analyses. Consider two sections.-------. for this section. which can be used for design purposes. f' = 1. . however. Curve B in Fig 5 shows peak values of of for a section with o: = 20° and 1/1. This will be satisfactory for most values of a but. but positive for small values. As before. + A"I.7. the taper is replaced by an equivalent length (A"I. For most sections. Also shown on the figure are the value rl4 (the peak value to be expected in a rectangular section of thickness I) and f'r I 4 (the peak value expected in a taper section)..Al.3 3 · . computer analyses have been carried out with the prismatic sections extending 1. Thus. derived earlier. as in Fig 7(c). use can be made of the values of \. but soon becomes affected by the boundary conditions at the outer end.) 1. The . at this end of the section. For small 0< the effect of the junction between the straight and tapered sections will be the same. Reducing section . achieves the desired purpose.' (/. which are indicated by positive values in the chart.Tapering section (/.-------. Two alternatives seem logical. \.(7) A number of sections of this type have been analysed using the computer program.) I. beyond the end of the .3/2 February 1993 49 . indicating that extending the section by 1.t.) 6 24 tan« (1 .--~--_.1/ 3 3 -3 Equivclent section 1 · .. These are shown in Fig 6. (7). to I.I 1-----------+------:::0':> Equivalent section 2 g-·I-------1T5------3'0--~--4'5------6~O------~75------~90 Taper ang te -1 Ct. for the section reducing from I. J I. this is not entirely satisfactory.

and Aw reasonably small. The middle term of eqn. the triangular region defined by the chamfer becomes the dominant element in the torsional stiffness. I. and Iw' For larger chamfers. > tw' for small Ar 50 The Structural Engineer/Volume 71/No. Lines are plotted corresponding to different chamfer sizes (values of I/Iw)' Two of the figures show more accurately the smaller values of Ar and Aw' The figures give results for sections with chamfer dimensions (I) less than the smallest of I. if I. Similarly. A'2 -0.0 (i. Lines of cons1ant c (1 In degrees) o 4 Ratio of ttucknes s t2Jt. {2 . the web length will be increased by Awlw (Fig 10). and a length of the flange and web (I. Equivalent sections for reducing section (the equivalent section 1 would normally be used for sections with a large taper angle. if the web is thicker than the flange. twit! Fig 9(a).065. plotted against the ratios of the flange and web thicknesses.0 and I.Paper: Burgoyne {I a 112 Actual Section I. Figs 9 and 10 show values of Af and Aw. I. There will be occasions when it is convenient to include these terms with adjacent regions.83928. = 0. (11) relates to the tapering part of Fig 7(c).7189). and it is unreasonable to idealise the section as a simple rectangle. The accuracy of the method can be seen by considering the section with ex 8°. thus.) Iw' the length of the flange will be increased by Arlr (Fig 9). the 45° chamfer (if it is present). Equivalent section and design chari [or T-junclions when tf > two Lines of constant [r/t.e. Design chart for t. while the first and last term relate to the two parallel sections. t. Design chart [or reducing section.5 times the appropriate thickness). would give the value that has been computed for the real section. particularly if there are chamfers or radii at the resultant corners. From Fig 6. if its inertia were calculated by the simple ('!J)bi' expression. 1.3/2 February 1993 . = 1. the error is negligible. .1 -c . while equivalent section 2 would be used for sections with a shallow taper angle) Actual sect-en Equivolent section 8. using the equivalent section 1 (Fig 7) Fig 9(b).83934.and T-sections.1 where "" is found from Fig 6. 11 I~ p1 ~1 Web-flange junction The junction between a flange and a web is a region which contributes significantly to the torsional stiffness of 1. The equivalent sections chosen are shown inset in Figs 9 and 10. respectively. Equivalent Section 1 ~1"rt11 t2 {I~ +===~~I I -I l]t2 Equivalent Section 2 Fig 7. 10 Fig 8. if it exists. I. (II) gives J 0. It is important to note that no chamfer is included in the equivalent cross-section. while an accurate finite difference analysis gives J 0. The length of the thicker component of the cross-section will be modified. The finite difference program has been used to determine the torsional inertia of a section which includes the junction. An equivalent section is sought which. its effects will be included in the values of the A factors. . Eqn. This has the effect of keeping the modification factors A.

for small A. replacing from will be used to compare I.- '\'-~ co " n. I rn that 't VI and also with a 200mm-thick another to the total is presumed a significant tlange stiffness. 40-t--. thickness as made up from: 200 mm..~ 5 x5 _. the results exercise. .5 nun. I 0 > 0 '" . in It is from shown has been proposed using the computer beams arc included constant top flange The as the new standard the contour in the V-beam for the eight of width clearly have been analysed Fig 1 is taken. Design chart for L-junction.-" '\'-~ " coo" tw1tt Fig 11 (b). Equivalent t" section and design chart for L-junclioll section and design chart for Tvjunctions when > I. co'" " ~" . of the thickest If) will be modified of A. by adding for different can be catered In this case.« < 4 I.. Design chart for tw for small Aw / L-junction Certain sections These will have L-shaped junctions between similar between (width values rectangular to that and used web 371 components. Fig 12. Eouivalent Fig IO(a).2 The section the small parallel with a short the top to the bottom. which the St Venant's torsional program that inertia described of the new range bridge earlier. top tlange.>/2 February The Structural Engineer/Volume 1993 51 .f (b) '~+]50 >t f. torsional Use of the method The Y . length for by a method the distinction section JI~/ .-~~ -1200 \If• Fig 11 shows ~T 342 313 ~-~. reducing in thickness from 393. and a web of 25 (a) two length (b) a T-junction thickness (c) a section a flange of thickness 200 mm 393. Y-beam sections (from ref. with the predictions in Fig 13.5 mm (0< = 8.-~-- B I .-" {-~ " co l. section. in this paper. is irrelevant.204"). will be slightly splays and Thus._/ Edge detent --" ~-~1400 ~ Y8 for the T-junction. 'I ---~1100 -~ Y5 ~~000 -+ Y4 1 900 800 700 Y -bea m sections As a separate of Y -beams. the Y-5 section as shown the curved with a 200 mrn flange. over a length of 615. and Table acting I gives alone to ~ V2 they are designed to be used with a 200mm-thick of the torsion beam.beams given in Table Consider idealised. accurate continue contribution values to Eight of one of these analyses plot of'. range beams makes beam'.' ~ II (Fig 12).1\0 . between can be regarded mm. ~0 ~. .Paper: Burgoyne 10 1 Actual section tw?:tf Equivalent section . at the throat the section 303. -.4) 71.~ " 0 o (0) 8 10 twltf Fig l Lta). calculated I --1300~~V7 ~--.5 !l1lll to 216 111m.I. flange the length an equivalent sections. parallel flanges. Fig IO(b). thus section the results of the method ignoring of an accurate described corner analysis.

as it includes '+' elements which are not analysed here.2°) and is best considered by the second method given for reducing sections. The total inertia (obtained by summing the subtotals in Fig 14) is . (b) The web corning in to the T-junction is thicker than the flange.01670 . and at the inner end. = 0. = 0. have to have their inertias calculated by methods that take account of the hole(s) in the section.3/2 February 1993 . where t in this case is taken as 202 mm. as there is clearly an error in Johnson's result.00860 .01235 .01812 .01133 . may be considered doubtful.01454 .02046 .. (a) The top tlanges are parallel-sided elements. so the length of the flange is adjusted by a factor A" found from Fig 9(b) with Iw/l.066.Paper: Burgoyne (d) a parallel section of length 60. so that A. he used a finite element formulation to solve the governing differential equation.t -~:.01460 . 1000 740 AI! dimensions In mm Fig 13. It must be emphasised that this is a very severe test for this method. AI t dimensions Units of In mm J ore rnrn« Fig 14. which has large chamfers. which gives A. (II).5. which have a much higher torsional stiffness. showing method of calculation of dimensions (see text for details of coefficients used) TABLE 1- Torsional inertia of Ysbeams Accurate torsional inertia (rn") Beam and flange . Closed sections. Shear flows have to be calculated around each hole.540.\2' where At2 is found from Fig 6 to be 0. Idealised Y-5 section 1032 I 16155 I 60.01931 Limitations of the method The method described here applies only to the calculation of the torsional inertia of open sections. Thus the length of the thicker section will be reduced by 393. Table 2 shows the results of this finite element analysis and those of the present analysis. but it is reasonable because skewing the section slightly will not significantly alter the values of '" from which J is calculated. The Y-beam section is exceptional in that respect because it is designed for production in factory conditions with moulds that are frequently reused.271.01812 m" given in Table l. With I/Iw = 0. (g) The end effects at the end of each flange are taken into account by subtracting from the inertia of the section an amount equivalent to A. with various methods of allowing for the junctions. (e) a T-junction between a web of thickness 216 mm and a flange of thickness 373. over a length of 262 rnrn.7 mm and thickness 216 mm. whose inertia can be calculated by a simple al'/3 calculation. (d) The throat is another short parallel-sided section. with no parallel-sided elements in the tlanges. which are the ones most commonly used in construction as they are easy to form. The contribution to the torsional stiffness of each of these components needs to be considered separately. and VI. (c) The reducing section has a relatively shallow taper angle (S. and I/Iw = 0. This use of the results for the symmetrically reducing section. = 0." and the length of the thinner section is increased by 216. which would be acceptable for all practical purposes and is much better than could be achieved by simply replacing the section by a series of equivalent rectangles that do not take account of the junction effects. His section 4 is also not included. Y-7 Y-8 52 The Structural Engineer/Volume 71/No.8 mm to 202 mm.7.2. with no end effeets.". (f) two flanges of reducing section.S mm.01618 . Johnson's section 2 has not been analysed. on which Fig 8 is based.00930 . the worst section is Jackson's (sic) exact I-section (quoted by Johnson'). There is no end effect at the outer ends.14/3 from the total inertia. The method would give exact results for sections with parallel-sided elements between junctions. The nomenclature of the sections. As a reference. (g) end sections at the extremes of each flange. thus spreading the cost of the formwork over many beams. from Fig 8 is 0.50S. which should be compared with the accurate result of . when applied to a section which is reducing on one side only. the section is an L-section.577.-216+ (0 22+3~Q274-1 x3738=500. Equivalent Y-5 section. and additional equilibrium equations must be satisfied'.8 Comparison with Johnson's results Johnson' analysed a number of sections by a variety of methods which involve idealising the cross-section by a series of rectangles. = 0. whereas it must be larger. The torsional inertia of the tapering part of the section is found from the middle term of eqn. are the same as in ref.02663 Beam Y-I Y-2 Y-3 Y-4 Y-5 Y-6 Beam alone .22.066x 216=750 L l J .02328 .01334 . so the length of the web is modified by Aw taken from Fig lO(b). 1/1. (I) The two bottom flanges are reducing sections with 1/1. from 373. The results can be seen to give very good agreement. of thickness 202 mm.7+ . and the units used. = 0. The error IS less than 4 "70.IS.018S rn". (e) The bottom junction is between a wider tlange and a thinner web. since the flange is assumed to continue into an adjaeent beam.8 \' 202 1 1373. the effect will be taken into account in (b).01019 . and the value he quotes is less than the inertia of the largest component of the L by itself. this gives Aw = .01277 . numerical values and subtotals are shown in Fig 14.

Reimundin . c. 23 4 December 1990. F. E.: 'Prestressed concrete I-beams under combined mixed torsion.: 'Non-linear behaviour of beams and beam-columns'. c. 91. Goodier. J.9 The present analysis also applies only to uncracked sections. are accounted for by adjusting tre 'engths of the parallel-sided elements by an amount that has been determined by an accurate computer analysis. H. D. It has been shown that the method gives good results. chapter 10. P. 24. elcm. Regan. The Structural Engineer. and Danesi. Johnson. 69. New York.. J. McGraw Hill 2.71 9.: 'Sectorial properties of straight thin-walled beams'. the torsional warping stiffness also plays a significant part in resisting torsional loads. The Structural Engineer. and more detailed results for prestressed l-beams were obtained by Luccioni et al6 while testing beams in combined torsion. N. pp207-210 3.: 'Behaviour of precast. R. prestressed V-beams in shear. No. PhD thesis. which significantly affect the torsional stiffness. Burgoyne. 68. 68. The Structural Engineer. including thick-walled concrete sections. C. Computers & Structures. 'Torsion'.5 2.46 3974 4976 analysis Error % Johnson's section -~--""-'--' Johnson's fin. Thus it is probably relevant for prestressed concrete beams at the working load condition. flexure and shear'.1991.3/2 February 1993 .lABLE 2- Comparison with Johnson'S' Present analysis 13. J.4 74. These length adjustments have been presented in the form of nondimensional charts. S.. torsion and negative bending'.4 1. No. and this needs to be calculated by other means 7• Conclusion A method has been presented which allows the St Venant's torsional inertia of a section to be determined by breaking the section up into a number of equivalent rectangles and tapering sections. For many concrete sections.: 'The V-beam: a replacement for the M-beam in beam and slab bridges. without the necessity of solving the stress function differential equation over the whole section. 23.: 'The elastic torsional stiffness of concrete sections'.ICE. M. University of London.7 (2'Ls' & 3'Ts') 2'Ts' with no chamfers 2'Ts' with chamfers 2'Ts' with chamfers 72.50 3920 4705 Notes I (Lvbeam) 3!TTII Jackson I (idealised) Jackson I (exact) Hambly I (exact) ° 0.. P. and Banks. P. It is in this type of analysis that an accurate knowledge of the torsional stiffness is required. pp147-156 The Structural Engineer/Volume 71/No. pp459-465 5. result 13. J. Proc. and it performed well when used on the examples given by Johnson. B. pp577-592 7.4 5. Part 2. The effects of junctions between elements of the cross-section. pp466-473 6. P. 4 December 1990. L. 11. flexure and shear. Timoshenko. Taylor. C. which can then be used in design offices to find the inertia of most open sections.. 1982 4. even for a section as complicated as a Y-bearn. but this covers the majority of cases when the transverse distribution of load is normally required. 1986.4 June 1991.: Theory of elasticity. No. References I. Regan estimated that the torsional stiffness of a Y-bearn cracked in shear was about half the uncracked value. Clark. A..7 9. Cracking will clearly reduce the torsional stiffness. Luccioni. Waldron.

Sign up to vote on this title

UsefulNot useful- Cracking the Cube
- In a Dark, Dark Wood
- Rust
- The Rosie Project
- On Writing
- The Great Bridge
- I Am Having So Much Fun Here Without You
- The Blazing World
- No One Belongs Here More Than You
- Tools of Titans
- Mr. Penumbra's 24-Hour Bookstore
- The Unwinding
- Red Queen
- The Kind Worth Killing
- The Rest of Us Just Live Here
- Bad Feminist
- Yes Please
- This Is the Story of a Happy Marriage
- The Handmaid's Tale
- Fourth of July Creek
- Mislaid
- Annihilation
- The Man in the High Castle
- Bobcat and Other Stories
- Field Notes from a Catastrophe
- White Girls
- After Birth
- Medium Raw
- Lucky Jim
- If on a winter's night a traveler

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd