Вы находитесь на странице: 1из 26
DESIGN OF TAPERED HAUNCHED UNIVERSAL SECTION MEMBERS IN PORTAL FRAME RAFTERS by T.J. Hogan’ and A.A. Syam? 1. INTRODUCTION Haunching, that is the localised strengthening of members fabricated from universal sections, is an engineering design solution commonly used in the proportioning of portal frame structures with rigid connections. The haunch is typically introduced into the rafter members at their connections to the columns in order to maich the rafter design capacity to the design bending moments which often peak sharply at such locations. A balanced design results in a haunched rafter (see Figure 1) that: + is locally strengthened at the column connection to cater for the larger design bending moments in that region + has a uniform section towards the centre of the span to match the smaller design bending moments in that region. The net result allows the use of @ haunched rafter of smaller total mass than that of a uniform rafter with a larger total mass over the span. Such a design solution becomes economic when the cost saving due to the decreased mass of the rafter is greater than the additional cost of fabricating the haunch [1,2]. Haunching is generally economic only where high bending moments are very localised. This often occurs in longer span (generally > 20 metres) portal frame structures, where the calculated bending moment at the knee joint may be 2.5 to 3.0 times that at the centre of the span. However, this does not preclude haunching from being economic in other situations. Using elastic analysis, a haunched section also tends 10 attract additional bending moment to itself [3], though the degree to which this occurs varies according to frame geometry, Haunched sections also have a significant effect on deflections due to the increased section properties in regions of high moment [4,19,21]. The length of a haunch is typically 10-15% of the span [1] although longer haunches are sometimes employed specifically to reduce vertical deflections of the rafter or horizontal deflections of the overall frame. This increased length of haunch may allow Director, SCP Consulting Pty Li. Manager Technical Services, AISC. the use of smaller sized structural members which might otherwise need to be increased solely in order to control deflections. ‘An additional benefit of haunching a rafter is that the flange forces being transferred to the column section (which is usually is not haunched) are reduced because of the increased lever arm, due to the increased depth at the connection [5]. This is particularly beneficial with the most common connection, the bolted moment end-plate, as it can keep the bolt size at or below M24 for a wide range of sections. Such a situation is desirable when fully tensioned bolts are used [1,5]. Members are sometimes locally haunched at bolted moment end-plate connections purely to reduce the calculated tensile forces in the bolts. In fact, conventional haunching ‘of members for strength grew out of localised ENON MOMENTS Eston FESION SECTION COPAY C 1 FIG. 1 Basis of Haunching in Rafters ‘STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 haunching at connections to reduce connection forces (4. ‘A disadvantage of haunching is that the column section, which is of uniform section, has to resist the same design bending moment as the deeper haunched rafter. This is of particular relevance for the external columns in portal frames. Thus, any increase in column size must be considered when evaluating the economic. benefits of haunching. Although haunching is a relatively common practice, little technical information is available on: + economical dimensioning and detalling of haunches + caloulation of section properties of haunches + computer modelling of the haunch length + haunch design to AS 4100: Steel Structures [6] + flexural-torsional buckling of haunched members. This paper considers the above items. However, due to the lack of other reference material on some of the above aspects, the authors are presenting initial guidance which seeks to stimulate discussion in these areas. As with many evolving design methodologies, the authors would be grateful for practical feedback on these issues. Unless noted otherwise, the notation used for the equations and figures in this paper are the same as that in AS 4100 [6]. 2. DETAILING AND FABRICATION 2.1 Costing of Haunches References [1,2] provide some guidance on the relative fabricated costs of haunched and unhaunched rafters. The relevant aspects of fabricated costs include: (a) the cost of cutting out the haunch (b) the cost of welding on the haunch (0) the cost of any treatment at the end of the haunch (see Sections 2.2 to 2.4) (d) the cost of the lighter basic rafter (to which the haunch is attached) compared to the cost of a heavier uniform rafter (©) the cost of any increase in column mass due to higher design bending moments at the haunch (f) the difference in cost of any local stiffening or column strengthening in the vicinity of the connection — which is likely to be cheaper in the case of the haunched rafter due to the decreased flange forces resulting from the increased depth. A detailed cost analysis evaluating all of the above parameters Is usually beyond the typical role of a design engineer. If possible, discussions with fabricators, should be held to evaluate alternative designs of haunched and unhaunched sections. As an approximate rule, if a rafter is haunched 15% (of span) at each end a 20-30% saving in mass per metre is required to make haunching economic STEEL CONSTRUCTION VOLUME 31 NUMBER SEPTEMBER 1997 2.2 Types of Haunches ‘A number of different methods of haunching universal beam sections at the columns of portal frames have been proposed [1,3] with four of the most common configurations shown in Figure 2. Further variations on these four configurations are also possible. The emphasis in all haunch details should be on simplicity, to assist in keeping fabrication costs down. This paper principally considers tapered haunches of the type shown in details (a) and (b) of Figure 2. In this configuration the haunch itself is fabricated out of a universal section (usually the same size as the basic rafter section itself) and fillet welded to o Ch FIG. 2 Types of Haunch the rafter bottom flange. The common procedure is to make the haunch depth at the end connection around twice the depth of the basic rafter section. These details are generally preferred on the grounds of appearance and economy. Details (a) and (b) in Figure 2 result in a cross-section as shown in Figure 8 and can be economically detailed if certain basic factors are observed (see Section 2.3). FIG. 3 Cross-section of Tapered Tee Type Haunched Rafter The method of haunching shown in Figure 2(c) is essentially a tapered section, fabricated from three plates, which is suitably spliced ~ generally by welding ~ to a three plate or universal beam section rafter. The fabrication cost of this detail usually restricts its use in portal frames using universal sections, but it is common in portal frames composed entirely of three plate sections, particularly in standard industrial building frames. The haunch detail shown in Figure 2(d) — and variations of it - are types of stepped prismatic haunches rather than tapered haunches, with the transition point being generally located in a region of low design bending moment. Essentially, the haunch is achieved by welding together two universal beam sections of different depths. with the fabrication cost of such a detail being approximately equal to that of the tapered haunch in dotails (a) and (b) of Figure 2. ‘The critical aspect of the haunch type in Figure 2(d) is the transition point whore the flange force from the smaller section must be transferred into the web of the deeper section. The transition point should be in a location where flange forces are as low as practicable and the transfer length must be sufficient to allow the transfer of forces without local failure of either member. No testing of this detail is known to the authors at this time. Consequently, it is not possible to give any specific guidance for the member design at the intersection point where the local flow of forces is likely to be critical. ‘The main focus of the paper is the tapered tee types, of haunch (i. detail (a) and (b) of Figure 2) as they are the more common forms of haunched rafter, 4 members. Other variations of haunch configurations may be specified where there are further over-ri Considerations for thelr use and, therefore, should not be precluded as a design option. Some portions of the material provided in this paper may also be applicable to these other haunch types. 2.3 Detailing The detailing of the tapered haunch must have due consideration of: + aesthetics — particularly in applications of architectural importance + structural behaviour + fabrication cost * clearance ~ e.g. for applications involving cranes or racking. * corrosion protection Feedback from structural engineers, steel detailers and fabricators has indicated to the authors that both details (a) and (b) of Figure 2 are commonly specified in Australia and New Zealand. It appears that both types of haunches can be justified for particular — though dissimilar — situations. In considering the application best suited for each type of haunch, it, would be best to assess their positive and negative attributes. For the purpose of comparison, Figure 2{a) Is referred to as the “fully tapered haunch” and Figure 2(b) is referred to as the “curtailed tapered haunch*. 2.3.1 Fully tapered haunch (Figure 2(a)) Positives (a) Though somewhat subjective, the detail is, aesthetically pleasing as it indicates flow or continuity of the rafter outline. (b) Preferred for structural detailing as a direct load path exists for the flow of flange forces from the rafter bottom flange to the haunch flange for the design bending moment in that region (Figure 4(@)). This is achieved by a butt or seal weld across the rafter flange (Figure 5). The haunch can be terminated at the theoretical cut-off point if desired although this is rarely done as the actual haunch length is rounded up by at least 100 mm. (©) The haunch flange is restrained at both ends if itis the critical (compression) flange. Based on these end conditions, there are adequate provisions in AS 4100 [6] to design for members subject to flexure and axial compression loadings. (d) The performance of this configuration has been validated by tests (see Section 6). Negatives ()) Potentially a more expensive detail than Figure 2(b) — due to the deposition of a butt or seal weld across the rafter flange, cutting in the section jet radii (i.e. the web to flange transition with STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 (a) [nl Pare TaLTeRwanve) to) FIG. 4 Bottom Flange Load Paths eee ee pea Ney pew Nur Wein oR SEAL WELD oN POOR ACCESS FOR WELOING FIG. 5 End Detail for Fully Tapered Haunches variable thickness) and additional grinding for fitup. (li) Due to the acute angle, there may be problems with the placement and quality of weld at the haunch web to rafter bottom flange connection in the region of the haunch termination. Consequently, welding is often omitted in this, region (Figure 5). (iil) Inadequate detail for application of corrosion protection systems (e.g. galvanising, etc) in the region noted in (ii) above. ‘STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 2.3.2 Curtailed tapered haunch (Figure 2(b)) Positives (a) Generally a less costly detail than Figure 2(a) due to omission of a butt or seal weld and reduced grinding, (b) Depending on curtailment depth (Figure 4(b)), there is better placement of the fillet weld along the haunch stem to rafter connection, (c) Detail at curtailment is better for application of corrosion protection systems. Negatives (i) Aesthetics appear to be adversely affected as the curtailed taper indicates an abrupt change in section ~ this would be more pronounced with a longer stem at the haunch curtailment. (li) The haunch detail requires the flange forces to flow through the web of the haunching section, as shown in Figure 4(b) which may not be a problem if the curtailment of the haunch is at point of zero moment (contraflexure). This is generally the situation for particular, though not all, load cases and if a long haunch is used to minimise peak shear flows along the haunch stem to rafter bottom flange junction. (ii) In most situations, the web is relatively thin (6 to 9 mm typically) for universal beam sections which are haunched. Due to the connection geometry at the haunch curtailment, the rafter bottom flange forces are constrained to spread into the orthogonally orientated haunch web — this load transter requiring a reasonable transfer length if flange loads are large. The stability of the haunch stem (i.e. web) for the above shear transfer cannot be readily assessed during design as no methods are available. This applies particularly to configurations with large curtailment depths (Figure 4(b)) and high rafter flange forces. (iv) The buckling behaviour of the haunch bottom flange under compression could be regarded as ‘complex because the flange is restrained through ‘a weldment at the column side and is partially restrained ~ if not unrestrained — by the web stem at the haunch termination. The restraint stiffness of the haunch flange at the haunch curtailment cannot be readily evaluated. (v) The authors know of no experimental studies of the detail in Figure 2(b) which have investigated the ability of the web to accept this load without failure. Until such testing is reported, the use of this detail should proceed with some caution. 2.3.3 Fully tapered versus curtailed haunches Both fully tapered and curtailed haunches are ‘commonly used in Australia with the former alternative being more common in New Zealand. However, the applications best sulted to each alternative generally have differing criteria. Based on feedback received by the authors from design engineers and fabricators, the use of these two haunch types can be summarised as follows: 2.3.3.1 Fully tapered haunch The fully tapered haunch is generally regarded as the alternative to use when the total haunched segment is required to increase the rafter capacity for bending moment. This option is considered to be the structurally optimal alternative due to the continuity of the rafter, bottom flange into the haunch flange. This allows ‘or a continuous load path of force in the rafter bottom flange. Though it can be argued that the haunch flange is generally welded to the rafter bottom flange at or near a point of zero bending moment, it must be noted that this may only occur for a particular load case. The configuration has been tested (see Section 6) and is readily designed to AS 4100 [6] The most significant aspect about this configuration is the buckling behaviour of the haunch bottom flange. This flange is welded at both ends and the haunch portion of the rafter can be designed as a tapered segment with the critical (buckling) flange being fully restrained at these ends provided there is judicious placement of fly braces. Consequently, due to the combination of adequate assessment of buckling behaviour and appropriate detailing for transfer of flange forces, the fully tapered haunch is specified when the total haunch length is designed for strength, due to high design bending moments. However, when compared to the curtailed ‘tapered haunch, the most significant negative aspects of the fully tapered haunch are the slight increase in fabrication costs and the awkward welding detail in the region of the haunch and rafter flange connection. The design methods contained in this paper can be readily applied to fully tapered haunches. 2.3.3.2 Curtailed tapered haunch Curtailed tapered haunches are also commonly specified for portal frame rafters though not specifically ‘or haunches which are working close to their strength limits. The main teason for this is fabrication economy. It appears to be generally accepted that this haunching alternative may be used in the following situations: + Lightly loaded portal frame (shed) structures e.g. no heavy roof loads, + To increase the lever arm length between the top and bottom rows of bolts to reduce the bolt forces (in bolted moment end plate connections) + To increase the distance between top and bottom flanges to avoid column web stiffeners * To reduce frame deflections ~ where the haunch may not be required for strength reasons. For the above situations, the strength limit state design of the rafter in the haunched region may not be critical. This may alleviate the need to have suitable detailing and design of load paths for flange forces at the curtailed end. Additionally, a rigorous method for assessing the buckling capacity of the haunch segment = particularly due to the lower restraint stiffness at the curtailed end of the haunch ~ may not be required. However, in lieu of testing, appropriate caution and ( engineering judgment by designers will be required to assess whether buckling design calculations are required, and if so, the level of restraint offered by the web of the haunch to the haunch (critical) flange and its impact on design. Some methods used include keeping the curtailment depth (Figure 4(b)) between 30 to 60 mm, or the use of a transversely placed small plate which is fillet welded at the haunch curtailment (Figure 4(b)) to restrain the haunch flange. ‘The design methods contained in this paper can be used for curtailed tapered haunches though engineering judgement is required in the relevant provisions dealing with the buckling design of the haunch flange and the ability of the haunch web to accommodate the additional shear load in transferring flange forces from the rafter to the haunch flanges. 2.4 Fabrication The tapered haunch is readily fabricated by taking the required haunch length of universal beam section and cutting it diagonally in the manner shown in Figure 6. Two tee sections of tapered depth result. Methods of cutting universal sections to form tee sections have boon dlaoussed eleenhars (7. us tothe shor lange ¢~) of haunches (1600 ~ 5000 mm) straight cutting is usually employed, followed by any required straightening. In order to avoid any wastage, the cut in Figure 6 should result in two Identical sections and the haunches should be detailed to permit this situation semusve a F at sae FA *H FIG. 6 Cutting of Tapered Haunches Standardisation of haunch depths has yet to be advanced though [8] gives some guidance in this area. Industry feedback has indicated that the standardised haunch dimensions in [8] are generally not used. This is being reviewed by the authors and will be revised in the next edition of [8]. If a haunch is to be employed it seems logical to haunch to the (_) maximum extent possible as shallower haunches are generally no less costly. Standardisation of haunch dimensions permits the calculation of section STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 properties and capacities by a design office to speed up the design process, and so standardise its own procedures. itis also possible to use a different section size for the haunch to that used for the rafter. However, standardisation of these multiple combinations is not ractical he fully tapered haunch detail in Figure 2(a) can be diagonally cut from the inside flange on one side to the inside flange on the diagonally opposite side (Figure 6). Consequently, the standard depth for this haunch type is readily set at (d,.- f,,) where terms are as defined in Figure 7. Fabricators have reported that computer numerically controlled (CNC) cutting processes can adequately handle this, though manual methods may involve some minor grinding of the web near the end of the haunch in order to fair it to suit the beam slope. Stopping the fillet weld to the haunch web short of the end of the haunch is recommended 2s both an economical and practical step (as in Figure 5). Access in this region is poor and any weld is likely to be of doubtful quality. The butt weld at the end of the haunch across the rafter flange requires no special preparation as a natural V forms. It should be noted that even though this weld is notionally termed @ butt weld it may also be a comer (fillet) or seal weld (e.g. in New Zealand). Consequently, this weld may not necessarily be specified as a complete penetration butt weld. The curtailed tapered haunch detail in Figure 2(b) can be more readily cut in the manner noted above, as the cut may be made along the flat portion of @ie wed. Tis means thatthe cut should occur betwoen the “k’ dimensions (Figure 6) between each end of the diagonal cut. Consequently, the standard depth for this type of haunch can be set at anything less than (dyo — tie ~ fo) where terms are as defined in Figure 7. Based on this cutting procedure, the curtailment depth (Figure 4(b)) can be made to be between 30 to 60 mm. Due to the lack of contact between the haunch flange and the rafter bottom flange, the fillet weld along the haunch web should be continued to the end of the haunch. 3. CALCULATION OF SECTION PROPERTIES 3.1 General Parameters The cross-sectional shape of a haunched universal section is shown in Figure 7. The cross section is notated so that the dimensions of the rafter (subscript ‘o’) section can be differentiated from those of the attached tee section (subscript ‘’). In many cases, the basic section dimensions will be identical since the same section is often used for the haunch as for the rafter. The major variable is the depth of the jaunch (dj) which varies from a maximum at one ‘end to (approximately) zero at the other end. Thus, the overall depth of the haunched section varies from yo t0 d,, (= Oyo + d,) ~ see Figure 7. STEEL CONSTRUCTION VOLUME 31 NUMBER SEPTEMBER 1997 3.2 Elastic Section Properties The properties of the elastic section are obtained by combining: ()) the properties of the universal beam section with (li) the properties of the added tee section [7] using the parallel axis theorem to give the properties of the total haunched universal beam section, The necessary formulae are given in Table 1. Due to its generalised format, the formulae can be readily written into a computer program or a spreadsheet, to allow section properties to be calculated for any ‘combination of universal beam section and haunch section. Z cae 4 ~T WA, oF vara. Sicon 4 FIG. 7 Cross-section of Haunched Universal Section 3.3 Plastic Section Properties Properties of the plastic section are obtained by determining the location of the neutral axis at full section plastification - i.e. by equating compressive and tensile forces ~ and taking moments about the neutral axis at full plastification. The only difficulty lies in the unsymmetrical nature of the section and the fact that the plastic neutral axis may lie either in: (i) the web of the original universal beam (i) the bottom flange of the original universal beam (ii) the lower fillets on the original universal section. Due to equating the section tensile and compressive (continued on page 13) Table 1: Elastic Section Properties Terms as defined in AS 4100 [6] and Figure 7 (fillets included) Universal Section A® = Dbl + (Cho — Btyllyo + 4 x 0.21463 byt? = e+ 2b 3 2" ha ~ 2) . eT + oo = 2te) te fp vores + o2ras( Me 5 Me — oes, ) | 2igb2. , (boo Bo)fno ‘oso + o2ree( p+ o228) ] 2 7 2 I? = 2 both + Fao ~ Atle + 2c,D} ~ 0.42015, where a, and D, [9] are given by: Note: J“ may be readily approximated by omitting the last two terms in the above equation for J 48, ‘Tee Section Haunch AY = Dyly + (dy — tay + 2 x 021467? Pat. 6, — tal fe + ]+2xo2raeg(t, + 02257) . = 2 oe een a = distance from bottom flange of tee to neutral axis of tee + 2 x 0.00755," +2 x O.21a6r2 (oh - ty ~ 0.2287)" ta} y= 492 * 7B pe = ite , + 4p oo7ssrt + 2x owe + oes) | ‘STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 oO O Table 1: Elastic Section Properties (Cont'd) St = Louth + Lid, ~ tat + aD} - 0.21044 - 0.10544, Orrere a, and D, [9] are given by: 4, he hath fot ae bal i + 0.1955 jt — 0.0865 “as = ova) 3 a Un +P + balls + J Oh + ty Note: J+ may be readily approximated by omitting the last three terms in the above equation for Total) Haunched Section AM = AP + At 2 ala, + *) + Atop ee = distance from bottom flange to neutral axis of haunched section (Figure 7) 2 I = 4 ala, oo on) +e Ate, - w 4 Ie Zt top ange = a a,j rm 2! bottom flange = 1 Cn He Hs Wey iz Yar Ja Bet Q = first moment of area of tee (L) connected to universal section about neutral axis (N.A.) \ fies = tteOu = $B) + (= tn = = +2 x 0.214672(cy ~ ty ~ 0:228n) (contribution from bottom fillets of UB neglected) STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 i Table lastic Section Properties Unhaunched From Figure 8 and Table 1: oO B® = 2Daty + (Igy ~ Btligo + 4 x 0.2146r2 FIG. 8 fo Neutral axis at full plastification is at mid-depth with equal areas both sides results in a plastic modulus of: Apo — 2tr)'t 4 2 From Figure 9 and Table 1 bo & neutral ois oO ZY EL : be Total area = AM = 2bgty + (dgo — BoMyo +4 x O2146/2 + ety + (0; ~ teNy + 2 x 0.214607 at Date + (dap ~ 2p) 2 + 2x 0.21462 + Pale + (a, ~ ty) et + 021462 Half total area af Another expression for half total area in terms of c, (location of plastic neutral axis — Figure 9) is: a Halt total area = 4 be ~ (nq ~ 2h) + (pp — Zhellye + 4 x 0.214672 Dito + Boley 2 Equating expressions for half areas leads to 2 solution for cy AY — dati — (doo — 2teltyo — 4 x 0.214612 eer Saeeasic Ricans Dis esnnRNMMMMNNNE AS Gl which is only valid if: QO (0 ~ tho) $$ Abo 10 ‘STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 Table 2: Plastic Section Properties (Cont'd) For this case, the plastic section modulus of the haunched section is given by: }+ 2x 02146/2(6, - ty ~ 0.2287,) + (dy — A FB (ee ~ deo + to) + 2 x 021488 (6, ~ don + ty + 0.228,,) + 4 +(4- ten de 6, +45 ‘n) + ort do ee 5) +2 x O2146/2(d,, ~ Gy + dy ~ ty ~ 02287) ‘Haunched jersal Sex Neutral A) b liversal Section a As before, Half Total Area = 4° = & ; - neutral Another expression for half total area in terms of c, (location = | ~~ —}~ — — ‘xe of plastic neutral axis — Figure 10) is: Half Total Area = tra) tno + 2 x 0.2146 12 7 oo“ expressions for half areas leads to a solution for c,: as BAY — dole - 2x 0.214672 % ¢, +t FIG. 10 which is only valid if (e+ te) $6 5 (by be ~ fo) For this case, the plastic section modulus is given by:- St = bat ee ig) +2xo2tser3 + bho 0, +d, ~ Been 02048 9 (ty ~ 6, +4,~ ty ~0228n) Hi U Section, Neutral Axis in Bottom Fil al Section (i.e. Intermediate Fillets) The precise evaluation of c, and SH is complex for this situation. If the plastic section properties calculations indicate that the neutral axis position (c,) does not satisfy the c, criteria for the above two cases, this @.2 mean the neutral axis is in the intermediate fillet region. The values of o,and S!’ may be established by trial and error or by reassessing the above two cases with all fillets ignored as calculations allowing «for the presence of fillets become very complicated. STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 u ‘STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 Table 3: Warping Constant Assuming haunch cut from Universal Section. Unhaunched Universal Section From Figure 11 0 2 hI We =, op. (Clause H4, AS 4100 [6]) a where two {trom Table 1) (conservative alternative) aa (Figure 11) FIG. 11 Haunched Universal Section n 8.4 and Figure 12 bf . " i = er 2 Wn T = where Oo reer a | 0 it = Ge (neglect middle flange) | . = d= dy, ~ by two rae ie el FIG. 12 O 12 ‘STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 forces, the plastic neutral axis cannot lie in the web S10UB40.4 Basic Dimensions: of the tee section added to the bottom flange of the 4, = 16 = 10: original universal section since the area of the tes bo = S04 min, Big = 165 mm, fy = 10.2 mim is always less than that of the original universal soction. 11.4 mm The actual position of the neutral axis is found by trial and error or can be readily obtained using the Haunch ~ Basic Dimensions; formulae given in Table 2. The parameters aro the same as that used for the calculation of elastic section Seeman ea properties, bg = 185 mm, fy = 10.2 mm 1.4 mm tyo = 6-1 mm F, 3.4 Warping Constant Formulae for the warping constant (/,) are given in Table 3. This parameter is used in member stability calculations. It should be noted that the calculation of Jy, assumes that the tee section haunch flange width (b,) and thickness (t,) is the same as the unhaunched section flange width (b,,) and thickness (ig) respectively — Le. the haunch’is cut from the Universal Section Substituting into equations in Tables 1, 2 and 3 the 4d, = 210 and 280 mm parameters (where relevant) are shown respectively on each side of a slash for the section properties below: universal section. "A® = 5208 mme For unhaunched universal sections, /,, can be readily = oe calculated from Appendix H4 of AS 4160 [6}. A simpler, 18 = 86.4 x 108 mmé and more conservative expression may be obtained 1? = 7.65 x 108 mm4 i by neglecting the contribution of the fillets and web ¥ in the calculation of /,. J! = 187 x 109 mm4 The evaluation of /,, is more complex for haunched 18 = 165 x 109 mm® universal sections. This is due to the presence of the middle (i.e intermediate) flange. The expression presented in Table 3 for the calculation of /,, neglects checks with data in [17]. the middle flange. This is based on the particular situation where the middie flange is positioned (centrally) on the neutral axis. In this instance the Tee Section Haunch middle flange is neglected in the evaluation of /,, 2 as only the outer flanges would be subject to warping Aa accom normal/shear stresses. Although practical haunches of = 48.5/73.9 mm have the middle flange below the neutral axis (Figure a nee 7) the assumption of neglecting the middle flange lake .83/3.83 x 108 mm+ 82.8/88.1 x 108 mm+ is used again as it is regarded as being rationally based and conservative, Additionally, the contribution of the fillets and the web have been omitted from the J, term. otal Haunch Section Atl = 8166/8592 mm? oy = 248/291 mm 1H = 284/877 x 108 mm* If = 11.5/11.5 x 108 mmé 3.5 Example Calculation of Section Properties Refer to Figures 7 and 13, 310UB40.4 27? = 1068/1285 x 108 mm? | 2507 = 1142/1296 x 108 mm? i = 97.5188.5 mm s 4 = 240/245 x 10? mm* 3 %3100B40.4 u x 2 ff = 1261/1486 x 102 mm? ® 1 = 486/629 x 102 mm® I FIG. 13. Example Calculation of Section QM = 591/736 x 103 mm? |] Properties i STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 13 4, COMPUTER MODELLING OF THE HAUNCH REGION Prismatic haunches of the type shown in Figure 14(a) should present no difficulty when modelied in standard computer analysis programs using the stiffness method. In this instance, the appropriate section properties are used for each beam element at each side of a node at the haunch/member junction after making some allowance for a length to transfer the design actions involved. It should be noted that each beam element is commonly modelled with their respective centrelines being coincident (Figure 14(b)) even though the physical situation is otherwise (Figure 14(a)). This is reasonable as [10] indicates that the coincident centreline modelling method is adequate when the beam axial load is small which is generally the case for rafters in portal frames. ‘Tapered haunches require additional thought. Some alternative options include: C) FIG. 14 Modelling of Prismatic Haunches for Analysis {a) modelling the haunch as a single beam element type over the full length of the haunch with the element's section properties based on that of the average depth of the haunch (Figure 18(a)) (b) dividing the haunch length into a number of segments (say 2 to 4) with each segment having uniform section properties which are representative of that segment, either maximum or average or minimum (Figure 15(b)) (c) using a single beam element of equivalent stiness, calculated using [11,12] (4) obtaining a stitiness matrix and fixed end reaction formulation for a tapered member and use a computer program which allows a stifiness matrix os eee FIG. 15 Modelling of Tapered Haunches for Analysis for a member to be input. Such expressions may be found in [3,13]. In [3], @ study was made of the number of points necessary to accurately model a tapered haunch using a series of uniform members for the flexibility method of analysis. This study concluded that for quite large variations in second moment of area from one end to the other (up to a factor of 5 — which is representative of many haunching situations), It is sufficient to use three nodes along the haunch length = that is two subdivisions of the tapered member length in Figure 15(b). This conclusion is supported by a simple study done by the authors, in which a 89.0 metre span portal frame was analysed by first order elastic analysis under two separate loads of: () 1.0 kNim applied vertically on the rafter (ii) 1.0 N/m applied horizontally to the left column. The frame deflections and bending moments were calculated using a standard commercial stiffness analysis program. The haunch length was modelled in the following alternative manner: (A) four equal segments, the properties for each being the maximum in that segment (8) two equal segments, the properties for each being the maximum in that segment (C) two equal segments, the properties for each being the average for that segment (D) one segment of uniform equivalent stiffness calculated using [11}. The results are presented in Table 4. These results indicate that, within the bounds of engineering accuracy, there is no benefit in using more than two segments and that either assumption in (B) or (C) or (D) is not structurally critical to getting a reasonable engineering solution. Ref [10] also provides some reasonabie guidance and suggests that haunches may be satisfactorily 14 ‘STEEL CONSTRUCTION VOLUME 31. NUMBER 3, SEPTEMBER 1997 oO Table 4 Results of frame analysis Eo Pr qr ef | tm | ome A modelled by using two ‘rafter’ elements and one oxac sera swan unt column element — see Figure 16. The two ‘rafter’ elements are modelled with average section properties for lengths corresponding to 1/3 and 2/3 of the haunch as shown in Figure 16. Interestingly, [10] also suggests T that the portion of the modelled column within the depth of the total haunched rafter may be modelled using the same section properties as that of the adjacent modelled rafter member. This is under- standable as the column would also be locally stiffened by the presence of the haunching of the rafter. As this seems to be typical UK practice, it would seem reasonable to use it also in the Australian context. Ref [10] also notes that: (1) the assumption that the neutral axis remains at the centre-line of the rafter and does not descend towards the haunch is safe since the assumption tends to overestimate both the force in the bottom flange and the shear force (2) increased refinement is not justified by improved accuracy for most normally proportioned eaves haunches. (3) small haunches used locally at connections to reduce connection forces need not be modelled, Ref [14] concludes from a study that “it does not matter whether the input data of the frame analysis consists of a stiffness (or flexibility) matrix, or of the equivalent geometric data of a substitute multisegment model". Consequently, Kosko [14] would argue that options (C) and (D) above give identical results. Ref [14] also derives a stiffness matrix for a tapered member and presents methods of determining equivalent geometric data for use in multisegment models ~ replacing the tapered member with either ‘one, two or three uniform members, Expressions are given for the caloulation of the section properties to be used for each segment for full equivalence. ‘STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 15 ‘STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 FIG. 16 Modelling of Eaves Haunch [10] Ref [15] has employed non-linear finite element analysis using an eight-noded isoparametric shell element to model the behaviour of steel portal frames ‘which were also tested. Such modelling included plate elements along the full haunch length and depth and is the ultimate in modelling accuracy. Previous theoretical modelling of haunched rafters by [16] have demonstrated the importance of a number of parameters related to material non-linearity and geometric imperfections. The last two approaches [15,16] are not practical for normal design office procedures. 5. STRUCTURAL DESIGN TO AS 4100 5.1 Design Actions ‘The design actions on a typical haunched member are: N’ maximum design axial force in the member M, maximum design bending moment in the member which takes account of second-order effects due to design loads acting on the displaced and deformed member/structural configuration V" maximum design shear force on the member For the typical application of haunched members (portal frames with I-section columns and haunched rafters), design actions are calculated by either: (a) first-order elastic analysis and then modifying the calculated first-order bending moments by using appropriate moment amplification factors determined from Clause 4.4.2 of AS 4100 [6], or (©) second-order elastic analysis in which the design bending moment is obtained directly The modelling of the haunched section is discussed in Section 4. 5.2Section Moment Capacity of Haunched Section — Bending about x-axis OM ge = OF Zee Clause 5.2.1, AS 4100 [6] whore ti design sodon momen capac % 0.9 Table 3.4, AS 4100 [6] Zu = Zoy for Ag Shep : ry —*s Vio, 2, =p fe for hep S Ag S hoy dy = 232) for hg > ey S, = plastic section modulus (Table 2) Z, = elastic section modulus (Table 1) Zu, = lesser of (S, and 1.5Z,] a hot rolled section flange d, hot rolled section flange The above values for 2,, and 2, apply to the flange elements which usually Control for hot-rolled universal sections subject to flexure. For deep sections, the web may be the more slender element but this would not be the case for the usual depths of universal sections that are haunched. The web element is supported on both longitudinal edges by the flanges, including the intermediate flange, but can be subject to design actions which result in a variety of stress distributions, according to the design actions and the location along the tapered haunch. Table 5.2 of AS 4100 [6] does not address these situations. In practice, the stress situation in the web will be ‘somewhere between compression at one edge and zero stress at the other, to compression at one edge some tension at the other. Accordingly, a web slenderness plasticity limit somewhat less than 82 and a yield limit somewhat less than 115 Is appropriate. Since, 2g/%py iS generally much larger for flanges than for webs, flange slendemess generally governs. Accordingly, for universal sections, only flange slenderness need be checked, For the basic universal sections, in Grade 300 stee! [81], all sections are considered compact except [17]: 360UB44.7 flange has excess slenderness 310UB32.0 _ flange has excess slenderness 250UB31.4 _flange has excess slenderness 200UB25.4 _flange has excess slenderness 200UB22.3 _flange has excess slenderness The following references to Grade 300 steel! are for universal sections complying with AS/NZS 3679.1 31]. Example: 310U840.4, Grade 300, Haunched 210 mm (as before) Section properties as previously calculated (Section 3.6) = [320 A, (lange) = 7.78/320 = 8.81 < 2, = 9 = [2205S < dy = 2, (web) = 465/360 = 526 < Ay ~ 62 Accept section as compact Section Capacity with 210 mm deep haunch, from Section 3.5 S, = 1261 x 10° mm? z 1068 x 10° mm (minimum) Zyy = 1261 x 10° mm (lesser of 1.5Z, and S,) 0.9 x 820 x 1261 x 10% lax = 108 363 kNm Section Capacity for various depths of haunch can be calculated as follows: Haunch Total May. Depth Depth («Nm) (mm) (mm) 280 584 428 210 514. 363 140 444 307 70 374 259 5.3 Member Moment Capacity of Haunched Member — Bending about x-axis AS 4100 [6] uses the concept of segments in its provisions for member capacity. Clause 5.6.1 of AS 4100 [6] considers a segment to be the portion of a member between adjacent cross-sections which are fully or partially restrained. 16 STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 In Clause 5.6.1.1 (b) of AS 4100 [6], provisions are given for the member capacity of segments with varying cross-section — such as a tapered haunch — without {ull lateral restraint along the segment. The way to use this clause is to provide a restraint at the ends of the segment as shown in Figure 17 which is a typical configuration in an industrial portal frame. This approach is also used in [21] for flexural design of the rafter. Ref [21] also suggests a fly-brace be placed on the haunch as near as possible to the column. Such a fly-brace option (not shown in Figure 17) allows the column to have full torsional restraint (92) at the knee but does not enhance the buckling stability of the rafter [32] as the rafter end connection provides sufficient restraint. ‘The design of the portal column is outside the scope of this paper. Another approach is considered in [18] but will not be dealt with further. AUN arte COLUMN BTM, FLANGE COMPRESSION Fig. 17 Restraint to Haunched Tapered Segment Subject to Bending Moment Distribution It should be noted that Clause 5.6.1.1 of AS 4100 [6] considers open sections with equal flanges. The clause is generally applicable for I- and channel sections though Clauses 5.6.1.3 and 5.6.1.4 of AS 4100 [6] allow angle and hollow sections with an appropriate warping contant term (I,) to be designed to Clause 5.6.1.1 [6]. The following design member moment capacity calculations for tapered haunched rafters assumes Clause 5.6.1.1(b) [6] is applicable. Due to the presence of a third flange this is considered to be conservative, Another assumption is the evaluation of the member effective length (L,) of the haunched section within the calculation of the amended elastic buckling moment, (M,q). For the haunch restraint condition noted in Figure 17, L, is assumed to be the actual length of the haunch. This is considered to be conservative {or typically loaded haunches as the lateral rotational restraint effects (due to member continuity) are not taken into account. For bending moments in the reverse direction to that shown in Figure 17 (i.e. top flange in compression), L, can be conservatively taken as the distance between purlins. Ref [21] uses a similar approach and reduces L, by 10-20% (due to member continuity) depending on which flange is in compression. For fully tapered haunches (Figure 2(a)), the use of Clause 5.6.1.1(b)(i [6] then proceeds in the following manner for the bending moment distribution and restraint condition shown in Figure 17: OMy = Ons0M, S OM, (Clause 5.6.1.1, AS 4100(6)) where: (M,, = design member moment capacity 17M, oa Oh Miy, My Mi, Mi 28 defined in Figure 17 $M, = design section moment capacity at the critical cross-section = 6Z;f, (Section 5.2) | Catticl +9] ~ Z,f, at the critical cross-section. [arly eR [ov 4 PE, | uy actual length of haunch with restraint at its end J = Jfor haunch section at critical cross section Jy = Ifo haunch section at critical cross section J, = Iyfor haunch section at critical cross section E 6 200 x 10° MPa 80 x 10° MPa 1.0 (1.27, (1-1) 0.5 (tapered member) . 4a [os . Osea) Z Ais q flange areas at minimum cross-section 2byty flange areas at critical cross-section Bb yt STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 7 section depth at minimum cross-section section depth at critical cross-section ‘As defined in Clause 5.3.3 of AS 4100 {6}, the critical section in a segment is taken as the cross-section which has the largest ratio of the design bending moment (M*) to the nominal section capacity in bending (M,) Example: Consider the bending moment distribution for the haunched rafter segment in Figure 18. Haunch depth (mm) 280 210 140 70 0 Mg, (kNm) 428 363 307 259 182 Me (kNm) 400 231 175 119 100 Me We 0.84 0.57 0.51 0.41 0.49 Hence, deepest section (haunch depth of 280 mm) is the critical cross section. $M, = 428 kNm at critical cross section 1.7 x 400 py, = ee 400-7 5 25 vast +175? + 119 1, = 05 (tapered member) = Z[os + 243904) - 0500 1.0 ~ [1.2 x 05 x (1 - 0.539)] = 0.728, 4280.9 476 kNm at critical cross section 1.5 x 10° mm* (Section 3.5) J = 245 x 10° mm* (Section 3.5) 629 x 10° mm® (Section 3.5) 3000 mm (Figure 18) 0.723 [r2x 200108 «11.5108 oo "108 Y 30007 fr 3 10°] _ [tone a9 Extensa. 400 23 1s 19 ny 100 ae 3000 1500 = 310UB40.4 — f 304 3100860. 20 FLANGE IN COMPRESSION FIG. 18 Example for Haunch Segment a, = 06 Ji.00? +3 - 1.044] = 0.687 (May = hg 0M gy = 2.17 X0.587(6Myy) 1.27(@M,) but < 6M, (= 428 kNm) = 428 kNm = 400 kN Section is satisfactory. Based on the restraint configuration in Figures 17 and 18, the above example indicates that the design member moment capacity (6M,) at least equals the design section moment capacity (M,) of the haunch at its maximum depth. This is due to the high values of a, arising from the varying bending moment distribution which more than offsets the slenderness effects. Designers of haunches sometimes use this situation as a first check for structural adequacy ~ i.e, ensuring that maximum design bending moment (M*) in the haunch region is less than 6M, at the maximum haunch depth. However, this should not be relied on for a complete check of the haunch region as further investigation may be required for differing moment distributions, rafter stitiness, restraints, etc. Alternative approaches to Clause 5.6.1.1(b)()[6] include: (i)_ using the properties of the minimum cross-section (conservative) as provided in Clause 5.6.1.1(b)(i)[6}, or (ii) a buckling analysis as provided in Clause 5.6.1.1(b) inte) IF () was adopted in the previous example, then $M, = 6M, = 182 kNm [17] for the minimum cross-section (unhaunched 310UB40.4), This approach is too OM yy = Moy | \ a conservative and should not be used Ee aein Buckling analysis could involve a full analysis using a special program to calculate M, or methods from Me _ 476 _ 4 044 the literature could be used. Such methods are My ~ 456 provided in [18,19] 18 STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 O O Clause 5.6.1.1(b)(i) [6] was based on limited tests of simply supported tapered and stepped members. As with general rafter design, this clause can be extrapolated to haunched segments in portal rafters configured in the manner shown in Figure 17. In Figure 17 the haunch flange is restrained at both ends. However, the application of Clause 5.6.1.1(b)(i) (6] to curtailed tapered haunches (Figure 2(b)) may not, be possible due to the haunch flange being somewhat unrestrained at the curtailed end. There is an obvious (flexible) restraint offered by the tee stem although the level of restraint is not easily assessed. Small curtailment depths (Figure 4(b)) may supply adequate lateral restraint though the limiting depth can not be readily evaluated. Modelling of the curtailed tapered haunch as a stepped member within the intention of Clause 5.6.1.1(b)(ii) [6] is not recommended 5.4 Shear Capacity of Haunched Section The shear stress distribution down the web of a symmetrical I-section is approximately uniform, as shown in Figure 19(a). The introduction of an additional flange with the haunched section results in a small discontinuity of the shear stress distribution as noted in Figure 19(b). For design purposes, itis proposed that a near uniform shear distribution be used again for the overall depth of the haunched section. Itis then valid to use Clause 5.11.2 of AS 4100 [6] to determine the design shear capacity of the haunched section. Clause 5.11.2 [6] is dependent on the web panel slenderness which is used to assess the stability of the web. The presence of the intermediate flange stiffens the total haunch web. The total web panel can be considered to be split into two panels by this stiffening flange. Consequently, the web panel depth to thickness ratio (4, / t,) check is controlled by the web panel of the umtiversal section since the depth of the haunch is always less than that of the universal section Hence for Se. (Coo ~ 2) < _82 by [hy 280 the design shear capacity (6V,) of the haunched section can be evaluated by: OV, = 90.61,A,, where Ay = (tho ~ 2eaw + (dh ~ fe) inci enleferearei ei ects ear depth of section is not used — i.e. total depth of web only) For dyty > | el Te 3 OV, = 00,0.6,A, STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 it Io (@) Shear Stress Distribution in I-Section +f two twee (b) Shear Stress Distribution in Haunched Section bo. dt FIG. 19 Shear Stress 2 were“ I7a,) 1% ty )V250 Example: 310UB40.4, Grade 300 Haunch Section as in example in Section 3.5 Ay, = 283.6 x 6.4 + (210 ~ 10.2) x 6.4 = 2949 mm? OV, = 0.9 x 06 x 320 x 2949/ 10° = 510 KN Ref [20] notes that the practice of assuming a uniform shear stress distribution with the total shear carried by the web is not strictly correct. The sloping flange provides a vertical force component that can either increase or decrease the design shear force depending on the direction of taper and the direction of force in the flange. 4 NA ” — Ag, =FLANGE AREA AT FLANGE 1 ~ Ag =FLANGE AREA AT FLANGE 2 FIG. 20 Calculation of Total Vertical Shear Force The simplified analysis proposed by [20] is to calculate the vertical component of the flange flexure force and add or subtract it from the total vertical shear force according to direction (Figure 20). The resulting shear force is still assumed to be carried by the web as a relatively uniform stress distribution. In Figure 20, My, apt An Ar Hence, Total Design Shear Force Nytan@ + No tang In this instance the intermediate flange is considered to be close to the elastic neutral axis and has negligible axial load (due to the bending moment) and is neglected. Additionally, the contribution of axial load in the total rafter section is also neglected due to its small magnitude with respect to section capacity. ‘Afinal shear design check should consider the interaction of shear and bending moment. AS 4100 [6] provides two methods (Clauses 5.12.2 and 6.12.3) for checking this interaction Clause 5.12.2 [6] considers a proportioning method in which the design bending moment and design shear force are considered to be resisted by the flanges and web respectively. The interaction check is satisfied if this simple proportioning method is observed. Clause 5.12.3 [6] considers an interaction method and specifies the design shear capacity (bV,) must be reduced to 6V,,, if the design bending moment (*) is greater than’75% of the design section moment capacity (6M,),where: for 0.75(6M,)s M" s(6M,) otherwise there is no reduction required for ¢V,. An alternative check should ensure that: V's 060%) where the interaction check is not required if the above shear force inequality is satisfied. This inequality can be determined by substituting M* /¢M, = (which is the maximum possible value for this ratio) into the above expression for Vim Hence from this expression, oV, can only be reduced to a maximum of 60% of its value — thereafter there is no reduction from interaction effects. It should be noted that the M"* /oM, = 1.0 ratio (similar to the ratio used to identify the. critical section of the bending moment segment in Section 5.3) appears in the above interaction expression for @Vjq This implies that an adequate haunch section to check for shear is the critical section (Section 5.3). Engineering judgement is required for practical design situations and some suggested section(s) to check may (at most) be the mid, quarter and end points of the haunch. However, design shear forces in haunched rafters are generally low with respect to design section shear capacity. Hence, the check of design shear capacity and the interaction check of shear and bending is usually not critical. 5.5 Axial Compression Capacity of Haunched Member 5.5.1 Section Capacity in Axial Compression ON, = Ok/Anf, (Clause 6.2.1, AS 4100 (6) where: 6N, = design section capacity 6 =09 ‘A, = net area of cross section as defined (_} in Clause 6.2.1 of AS 4100 [6] ky = form factor = A,/ Ay (Clause 6.2.2, AS 4100 [6)) 20 STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997. A, = effective area ‘Ag = gross area of the section The effective area (A,) is calculated by summing the effective areas of the individual elements (flanges © 372. wers), the tfective areas being determined by the effective width (b,) of an element multiplied by its thickness so that 5 D bath The effective width of an element (b,;) is given by (Clause 6.2.4, AS 4100 [6)): 45 2, (Coton in aunch) = 210102) BD = 871< 45 z. Deduction required from top web of original UB section can be determined by: A5" b, (web) = ras, os ) = 2426 Ay = A, (gross) = 8168 mm? A, = 8166 — (283.6 ~ 242.6) x 6.1 QN, = 0.9 x 0.969 x 8166 x 320/107 = 2279 KN STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 21 Section capacity in axial compression can then be calculated for various depths of haunch as follows: Haunch Total Ag ky ON, Depth Depth (mm?) (kN) (mm) (mm) 280 «584 = 85930971 2408 210 5148166 0.969 2279 140444739 0.967 2185 70 374 «7312 0.965 2032, 0 304 = 5208 0.952 1480 @N, for portal frame rafters is only relevant for calculations involving combined actions checks. Discussion on this aspect is left until Section 5.7. 5.5.2 Member Capacity in Axial Compression ON, = aoN, < ON, (Clause 6.3.3, AS 4100 [6)) design member capacity design section capacity (Section 5.5.1) @, = determined in accordance with Clause 6.3.8 of AS 4100 [6] using cr, = 0.5 (“other sections” in Table 6.3.3(1) [6)) where @N, = on, 0 The value to use for a, is arguable. The basic hot rolled universal section without welded haunch attached has an a, = 0 while the tee section which is added has an a, = 0.5 if used in isolation. When the tee is welded to the rolled universal section, Table 6.3.3(1) [6] provides no guidance except through the ‘catch- all” provision for anything not otherwise described (‘other sections") - for which a, = 0.5. Note that [21] uses a, = 0 (as for a hot-rolled section rafter) rather than 0.5, The haunched section varies along its length. Two approaches suggest themselves to handle this problem: (use the section properties at the minimum section along the tapered member within the effective length (as is done in (21]) (ii) use Clause 6.3.4 of AS 4100 [6] which states that Clause 6.3.3 [6] may be used by — (a) calculating @N, as the minimum value for all cross sections along the member (b) modifying the member slenderness (2,,) as calculated in Clause 6.3.3 [6] by using where Nominal section capacity, minimum value for all cross sections along the member Nom = elastic flexural buckling load of the member determined using a rational buckling analysis. Solutions for N.,,, ar@ difficult to obtain readily although computer programs are available which can do the necessary analysis. Hand calculation methods are available but are somewhat limited. @N, will need to be calculated for buckling about both major (x) and minor (y-) axes and critical to this calculation is the determination of the effective length for buckling about each principal axis. Since the member capacity is only relevant to calculations of the member capacity when subject to combined actions, discussion on this aspect Is left until Section 5.7 of the paper. 5.6 Axial Tension Capacity of Haunched Section ‘Assuming that the haunched section (complying with [81]) has no holes for connections and is welded to an end plate or column, then ON, = 4A, (Clause 7.2, AS 4100 [6)) where A,= gross area of haunched cross-section where nominal section capacity in tension is required Usually 6 is required for use with Section 8 of AS 4100 [6], for the analysis of a section subject to combined bending moment and axial tension. The value of A, will vary according to which check in Section 8 [8] is being carried out (see Section 5.7). If @ bolted connection as shown in Figure 21 is used, then: ON, = minimum of [ 6f,Ag, 60.85 A,f,] beat FIG. 21 Bolted Connection for Haunched Section through the connection, where A, = net area of haunched cross-section whose nominal section capacity in tension is required. Example: 310UB40.4, Grade 300, Haunched 210 mm No bolt holes at section Ay = 8166 mm? {from before) N= 0.9 x 320 x 816/103 = 2350 kN The design section capacity in axial tension can then be calculated for various depths of haunch as follows: Haunch Total Ay on, Depth Depth (mm?) (KN) (mm) (mm) 280 584 95932475 210 514 8166 2350 140 44g 7739 (2089 70 374 7312 2106 0 304 5208 ©1500 5.7 Design for Combined Actions 5.7.1 Section Capacity - Uniaxial bending about major principal axis Design requirement ML < OM, (Clause 8.3.2, AS 4100 {6) design bending moment 09 design section moment capacity, reduced by axial force e = oHa('- air) N* = design axial force (tension or compression) ON, = calculated in accordance with Section 55.1 $M,.= calculated in accordance with Section 5.2 Example: 310UB40.4, Grade 300, Haunched 210 mm, see Figure 18 Section as previously evaluated for gross section properties (Mg, = 363 kNm (Section 5.2) ON, = 2279kN (Section 5.5.1) M* = 231KN (as in Figure 18) Nt = 80KN compression oi, = 969 350 kN : > M’ = 231kNm_ Section is satisfactory 22 STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 ry an The ratios of M"/$M,, and N*/ 6N, vary with respect to each other along a haunch. Accordingly, the section capacity check should be carried out at every critical cross-section along the haunched member. This is not practical and requires engineering judgement which may, at most, assess the mid, quarter and end points ‘®or this design check. 5.7.2 In-Plane Capacity ~ Uniaxial bending about major principal axis (a) Compression members Clause 8.4.2.2 of AS 4100 [6] requires in-plane member failure to be checked. The basis of this check is to consider the in-plane behaviour of the member even ‘though it may (or may not) be fully restrained out of its bending plane (which is subject of another check - see Section 5.7.3). In its strictest sense, the portion of the rafter between in-plane restraints is considered to be relevant to the intent of the clause. The obvious in-plane restraints in this instance would be the column and ridge [21}. Notionally, the purlin/fly brace systems offer no restraint to in-plane buckling modes. Consequently, there is no in-plane capacity checks for the hauch itsetf as this design provision is essentially geared for the whole rafter (haunch included). ‘Assuming an elastic analysis has been carried out, as suggested in Sections 4 and 5.1, a possible design method for haunched portal rafters complying with Clause 8.4.2.2 [6] is: ® My = OM, (Clause 8.4.2.2, AS 4100 [6)) where: M, = maximum design bending moment in the rafter (Clause 8.2 [6]) 6 =09 4M, = design in-plane member moment capacity e m1 7 a) ¢Myy= design section moment capacity (Section 5.2) 'N* = maximimum design axial compression force (Clause 8.2, AS 4100 [6]) @Ney = design member capacity in axial compression calculated in accordance with Section 5.5.2 Clause 8.4.2.2 [6] can be readily used for uniform depth rafters. However, the application of this clause to haunched rafters is slightly more complex due to the localised stiffening at one end of the rafter. The use of @M,, for the minimum (unhaunched) section of the rafter may not be feasible as M; is generally en as the design bending moment at the column ‘which would be significantly more than the minimum Alternative A Considering that most of the rafter is of uniform section- (1) my Is the maximum design bending moment between the ridge and the smallest end of the haunch (away from the column) (2) @M,,i8 calculated on the base rafter (zero haunch) section (8) Nt is taken as the maximum design axial compression force between the ridge and the column (N* is generally small with respect to 4N, and NV, of the rafter) (4) ¢No,is calculated on the base rafter (zero haunch) section with an effective length equal to the distance between the column and the ridge (see Section 5.7.3 for the typical evaluation of ON.) Alternative B Ref [21] suggests that in calculating the design member moment capacity (9M, ) for a rafter segment, the haunch should be ignored and the design bending moments reduced by factoring the “moment at any section by the ratio of the elastic section modulus of the unhaunched rafter to the elastic section modulus of the section”. This suggestion can be further utilised in checking a haunched rafter to Clause 8.4.2.2 [6], ie. (1) My (the maximum design bending moment between the ridge and the column) is reduced by the above ratio of elastic section moduli (2) As noted in (2) of Alternative (A) (3) As noted in (3) of Alternative (A) (4) As noted in (4) of Aiternative (A). (b) Tension members An in-plane member capacity check for a member subject to axial tension (Clause 8.4.2.3 of AS 4100 {6)) is satisfied when the section capacity check (Clause 8.3 [6] and Section 5.7.1 of this paper) is satisfied 5.7.3 Out-of-Plane Capacity - Compression Members - Bending about major principal axis IN requirement MS Moy (Clause 8.4.4.1, AS 4100 [6)) where: My = maximum design bending moment in segment Ce) design out-of-plane member moment capacity 3 = Myf 1 A | section design section moment capacity. One of the Ne maximum design compression force following procedures may then be undertaken: in segment STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 23 calculated in accordance with Section 5.3 of this paper design member capacity in axial ‘compression calculated in accordance with Section 5.5 of this paper for buckling about the minor principal axis, determined from the minimum cross section in the segment with effective length taken as the purlin spacing [21] for the segment in consideration = a, ON, < ON, Example: 310UB40.4, Grade 300 Haunched as shown in Figure 18. Check the adequacy of the haunch segment between the column and first purlin restraint Bending moment diagram as shown in Figure 18 Design axial force = 80 KN Effective length y-axis = 1500 mm Haunch depth at minimum section in segment = 140 mm J, at mimimun section in segment = 11.5 x 108 mm* Aat mimimum section in segment = 7739 mm? 400 kKNm (Figure 18) 80 kN 428 kNm (Section 5.3) 1500 mm (Figure 18) 11.8% 10° _ 38.6 (Table 1) a ee Viena) minimum section GN, = 2155 KN. k, = 0.967 (Section 5.5.1) 1500 [320 2 = age O86 365, 05 0.847 ON, 43.2 (Table 6.3.3(3), AS 4100 [6}) 0.847 x 2155 = 1825 kN 80 e aaa( ) = 409 km 1825 2M* = 400 kNm Satisfactory 5.7.4 Out-of-Plane Capacity - Tension Members - Bending about major principal axis Design requirement M,, < 6M, (Clause 8.4.4.2, AS 4100 [6]) where M, = maximum design bending moment in segment 6 = 09 OM,,= design out-of-plane member moment capacity - - ) = oM,[1+ 4) < om, 1) = MoE gn) ) M*(kNm) 400 231 175 119 100\ ) M*IMg, 0.84 0.57 0.51 0.41 0.49 4N; (KN) (Section 5.6) 2475 2350 2289 2106 1500 N* (kN) 80 80 80 80 80 Critical cross-section is at haunch depth 280 mm My = 428 kim (Section 5:3) = 80.) = ply = 01+ Be) = estat ut @May My 80 OM, = 420(1 = 5Be) = 415 km > M'(= 400 kNm) Satisfactory 5.8Design of Weld Connecting Tee to Universal Section ‘The weld connecting the tee section to the universal section must be able to transmit the shear flow between the two section types. This design shear flow (v*) is given by: vee VO O where V*= design shear force at the section Q = first moment of area of connected tee about neutral axis 24 STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 second moment of area of haunched section about its neutral axis The design requirement in Clause 9.7.3.10 of AS 4100 (6] is that: © vi < vy, Where Vy = 90.8 fy ® = 0.8 SP weld category = 0.6 GP weld category fyy = nominal tensile strength of weld metal t, = design throat thickness k, = feduetion factor to account for the length of a welded lap connection = 1.0 since weld is not a welded lap connection 310UB40.4, Grade 300 = 100kN at section where haunch depth = 280 mm (end of haunch) Example: Q = 736x10% (Section 3.5) 1 = 877x108 (ection 3.5) 736 x 10° + = 100 x 36x 10- . 377 x 108 = 0.20 kN/imm For 6 mm fillet, both sides, GP weld category, fy = 480 MPa © ty = 08 x 0.6 x 48 x (0.707 x 6) x 1.0 x 2 sides = 1.47 kNimm sv Satisfactory Testing and measurement of shear stress indicated that apart from a peak stress at the end of the tee where it disappears, the magnitude of the shear stress in the longitudinal weld between the tee and the Universal section decreased along the haunch length as the end connection was approached, even reversing in sign. Ref (23,24] concludes that a 6 mm filet weld was “more than adequate”. Ref [23,24] have investigated the load diffusion at the tee/universal section intersections. They have noted that the diflusion of internal forces in this region is complex because it depends on factors such as, haunch geometry, web thickness, initial imperfections, the presence of any web stiffeners, the size of the weld used which in turn influences the level of residual stress. ‘The weld which joins the haunch flange to the rafter flange (transverse weld) needs to be considered critically. The use of oversize welds increase the cost Ind also the level of residual stress and the level of distortion, Previous British practice (1986-1987) ‘was apparently to either provide full strength welds (complete penetration butt welds) or no weld at all. No welding at all restricts the flow of forces through the web of the haunch section and also removes the inherent torsional restraint supplied by the weld at the intersection. Accordingly, Andrade and Morris [23] recommend using a transverse weld to join the haunch flange to the flange of the member. Testing by Andrade and Morris [23,24] confirmed a recommendation by Mortis [22] that the transverse weld be designed to transmit at least 75% of the force carried by the member flange. New Zealand practice [27] Is to use a seal weld. 6, REVIEW OF TESTING IN LITERATURE 6.1 Morris and Nakane [25] A study was conducted on the stability of haunched rafters in isolation from any other effect, the principal aim being to investigate the effect of the third flange. The tests were set up to force a plastic hinge to develop at the haunch/rafter intersection — where the transition occurs to the unhaunched member. Unloading was precipitated by local buckling of the compression flange in the basic rafter section, immediately adjacent to the haunch/rafter intersection. Due to the complex nature of the stresses in the Vicinity of the intersection, difficulty was experienced in forming a plastic hinge at the intersection before instability occurred. Other failures were web buckling failures, ‘These tests had a bottom flange restraint at the point of intersection of the haunch and basic rafter which is the arrangement advocated in this paper. 6.2 Andrade and Morris [16,23,24] ‘The study reported in these papers involved the use of the finite element method to investigate load diffusion In the haunched region in order to ascertain the flow of forces in the haunch elements. The results pertaining to the longitudinal welds between the tee and rafter sections are discussed in Section 5.8. Much of these references are concerned with correct finite element modelling assumptions and techniques then with the results. ‘Testing indicated the presence of large tensile stresses in the middle flange due to the presence of the longitudinal fillet welds, while the stress pattern in the inclined haunch flange varied according to its distance from the middle flange. The detrimental effect of residual stress was found to be more or less balanced by the beneficial effect of strain-hardening. The middle flange was generally always in tension because the applied bending moments were not large enough to cause compressive stresses which counter-balanced the locked-in residual tensile stress in that flange One interesting result was that the middle tlange away from the haunch-rafter intersection was in tension, probably due to the applied load being unable to cause compressive stresses large enough to counter STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 25 the locked-in residual tensile stress present in the middle flange. The investigation of [24] was primarily intended to investigate conditions which would ensure that a plastic hinge would develop at the haunch-rafter intersection. ‘Mathematical simulation indicated that for this to occur the stress at the deepest part of the haunched member, should be less than 0.8 times the yield moment of, the haunched member. 6.3 Morris and Packer [28] This paper attempted to study the lateral stability of haunched rafters. Guidance is given on methods of assessing lateral stabilty and the conclusion reached was that lateral braces must adequately restrain the compression flange at plastic hinge locations — an obvious requirement also provided in AS 4100 [6]. 6.4 Horne and Morris [29] This paper investigated the behaviour of haunched members restrained at intervals along the tension flange. Design proposals were advanced although these involved numerous approximations. The authors of [29] recommended that the haunch be designed to remain elastic during the formation of plastic hinges in a frame and noted that there remained a large number of questions regarding the behaviour and design of haunched members. 6.5 Horne, Shakir-Khalil and Akhtar [30] This paper is an attempt to address the question of stability of tapered and haunched beams by extending some earlier work by Home and Ajmani to this type of problem. The approach was to express the maximum stable length of a haunched or tapered I-section beam by way of closed form solutions which allowed for an arbitrary distribution of bending moment along the length. A testing program validated the method 7. CONCLUSIONS This paper considers various aspects of haunching universal section members in portal frame rafters. Differing haunch configurations are discussed and the tapered tee type haunches are principally considered. Methods of economical detailing, calculation of section properties, computer modelling, haunch design and flexural-torsional buckling of the haunched length are presented. The paper notes the lack of practical reference material available to designers in the Australian context on the above aspects. Consequently, the paper is also presented as an initial guide which should permit further consideration and discussion in industry and subsequent feedback. 8. nm 2) (3) (4) 5) Is) (7 (8) 9) [10] 1 (12) (13) 14] (15) (16) 7] 18) {19} REFERENCES AISC, “Economical Structural Steework, Fourth Eaton, ‘Austfalian institute of Steel Construction, 1997 Watson, K.B., Dallas, ., van der Kreek, N. and Main, T., "Costing of Stesiwork from Foasibilty through to Completion’, Steel Construction, Vol. 30 No. 2,-~ Australian institute of Steel Construction, June 1996! Monash University “Structural Design of Stee! Portal Frames’, Lecture Notes for an Ininsive Course, Monash University, 1979. Mois, L.A Commentary on Portal Frame Design”, The Siructural Engineer, The Institution of Structrel Engineers, Vol. 594, Dec. 1981, pp. 894-403 Hogan, TJ. and Thomas, I.R."Design of Structural Connections", Fourth Editon, Australian natitute of Steel Construction, 1994 SAAS 4100: 1990 Stoel Structures", Standards Austalia, 1990. Hogan, Td. and Thomas, Design of Tee Section Members to AS 1250", Steei Construction, Vol. 14 No. 2, Australian Insitute of Steol Construction, 1980, AISC,"Standardized Structural Connections”, Third Editon, Australian Instiue of Steel Constution, 1085, 1 Darwish, LA and Jonnston, 8.6. “Torsion of Situ Shapes", Journal ofthe Structural Division, American Society of Civil Engineers, Vol91 No. ST1, Feb 1965, pp. 203-227. SCI,"Modelling of Stee! Structures for Computer Analysis", Tho Stoo! Construction Inttute, Publzation Number P148, 1995 PCAHandbook of Frame Constants, Portand Comont Association, linols, 1986. Gere, J.M, "Moment Distribution Factors for Beam” ) of Tapered Sections", Department of Civil Engineering ‘Stanford University. (Distributed by American Institute of Steel Construction, undated). Rodsigues, J.S.-Beam idealization’, Journal of the Structural Division, American Society of Cv Engineers, Vol. 108 No. ST3, May 1982, p.1184 Kosko, E,"Uniform Element Modeling of Tapered Frame Members”, Journal of the Structural Division, American Socioty of Civil Engineers, Vol. 108 No. ST, January 1982, 9.245. Davies, JLM. and Mortis, L.*Realistie Modelling of Stoo! Portal Fame Behaviou, The Stuctural Engineer, Vol. 68 No. 1, January 1990, pp. 1-6 Andrade, S.A.L. and Mortis, L.,"Assessment of Parameters Aifecting the Behaviour of Haunched Rafters”, Pace Structural Stool Conforence, Auckland, Proceedings Vol. 2, August 1986, p.365. AISC,Design Capacity Tables for Structural Stoo, Volume 1: Open Sections", Second Edition, Australian Institute of Stee! Construction, 1094 81°BS 5950: Part 1:1990 Structural Use of Steelwork in Building, Pant 1: Code of Practice for Design in Simple and Continuous Construction: Hot Rolled Sections”, British Standarés Insiution, 990 (including Amendment No.1 1992) SC1Steo! Designors” Manual’, Fith edition, The Stel Construction Institute/Blackwell Scientific Publications” 1992. 8 26 STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 [20] Chong, K.P,, Swonson, W.D. and Matlock, R.B.,"Shear Analysis of Tapered Beams", Journal of Structural Division, American Sociaty of Civil Engineers, Vol. 102 No.ST9, September 1976, pp. 1781-1788. [21] Woolcock, S.T., Kitipornchai, S. and Bradford, M.A.,"Limit State Design of Portal Frame Buildings", Second Edition, Australian Institute of Steel Construction, 1993. [22] Morris, Lu.,"A Commentary on Portal Frame Design", Correspondence, The Structural Engineer, The Institution of Structural Engineers, Vol. 61, July 1983, pp.212- 221 [28] Andrade, 8. and Morris, LJ.,"Limited Studies of Non- Linear Behaviour of Haunched Members", Intemational Conference on Stee! and Aluminium Structures, Cordel, 1987. Edited by R. Narayanan as Vol 1 Steel Siructures: Advances, Design and Construction. Elsevier Applied Science. 1987. [24] Andrade, S. and Morris, L.,"The Influence of Stress Levels on the Behaviour of Steel Haunched Members Proceedings, 10th Australasian Conference on the Mechanics of Steel Structures and Materials, Adelaide, 1986, pp. 161-167. [25] Morris, LJ. and Nakane, K.,"Experimental Behaviour of Haunched Members”, Proceedings of Michael R. Hore Conference, Edited by LJ. Morris, Canada, 1983, ® [26] Andrade, S. and Mortis, La,"Assessment of Parameters Affecting the Behaviour of Haunched Rafters", Proceedings, Pacific Structural Steel Conference, Auckland, Vol.2, pp. 365-379. [27] HERA,"Manual of Standard Connection Details for Structural Steelwork", Second Edition, Heavy Engineering Research Association, Report R4-58, 1990. [28] Mortis, Lu. and Packer, J.A.,"Stability of Haunched Ratters", Prelim. Report, Stability of Steel Structures, Liege, April 1997, pp. 539-544, [29] Horne, M.A. and Mortis, L.v., "The Design against Lateral Instability of Heunched Members Restrained at Intervals along the Tension Flange’, Proc. Structural Stability Research Council, International Colloquium, May 1997, pp. 618-629, [30] Horne, M.R., Shakir-Khalil, H. and Akhtar, S., “The Stability of Tapered and Haunched Beams", Proc. Institution of Civil Engineers, Part 2, Vol. 67, Sept 1997, pp. 677-694. [31] SA/SNZ, "AS/NZS 9679.1: 1996 Structural Steel, Part 1: Hot-rolled bars and sections”, Standards Australia! Standards New Zealand, 1996. (32) Trahair, N.S., Hogan, TJ. and Syam, A.A., “Design of Unbraced Beams”, Stee! Construction, Vol. 27, No. 1, Australian Institute of Stee! Construction, 1993, STEEL CONSTRUCTION VOLUME 31 NUMBER 3, SEPTEMBER 1997 oA

Вам также может понравиться