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

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

Editors' Picks Books

Hand-picked favorites from

our editors

our editors

Editors' Picks Audiobooks

Hand-picked favorites from

our editors

our editors

Editors' Picks Comics

Hand-picked favorites from

our editors

our editors

Editors' Picks Sheet Music

Hand-picked favorites from

our editors

our editors

Top Books

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Audiobooks

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Comics

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Sheet Music

What's trending, bestsellers,

award-winners & more

award-winners & more

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

OTH 354

STRESS CONCENTRATION FACTORS FOR SIMPLE TUBULAR JOINTS Assessment of Existing and Development of New Parametric Formulae

**Prepared by Lloyd’s Register of Shipping 71 Fenchurch Street London EC3M 4BS
**

HSE BOOKS Health and Safety Executive - Offshore Technology Report

© Crown copyright 1997 Applications for reproduction should be made in writing to: Copyright Unit, Her Majesty's Stationery Office, St Clements House, 2-16 Colegate, Norwich NR3 1BQ First published 1997 ISBN 0-7176-1418-2

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means (electronic, mechanical, photocopying, recording, or otherwise) without the prior written permission of the copyright owner.

This report is published by the Health and Safety Executive as part of a series of reports of work which has been supported by funds provided by the Executive. Neither the Executive, or the contractors concerned assume any liability for the report nor do they necessarily reflect the views or policy of the Executive. Results, including detailed evaluation and, where relevant, recommendations stemming from their research projects are published in the OTH/OTI series of reports.

6 Lloyd’s Register equations (1991) 5.5.1 Supported chord effects at the crown under axial load 2.2 Short chord effects at the saddle 3.0 SIMPLE JOINT SCF EQUATIONS 4.0 REVIEW OF EXPERIMENTAL TECHNIQUES 2.2 Criteria for assessment of parametic equations 5.0 CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES TABLES iv v 1 3 3 3 4 6 7 7 9 10 11 12 12 14 14 15 17 17 18 19 19 20 21 23 23 26 27 28 28 29 31 31 32 35 1 .1 The inclusion of a weld fillet 2.0 INTRODUCTION 2.0 DATABASE 3.5.2 Strain gauge locations and stress sampling positions 2.0 joints 6.2 Weld cut-back at the saddle locations on ß = 1 joints 2.1 Criteria for the acceptance of SCF data 3.2 Wordsworth/Smedley equations (1978 and 1981) 4.6.4 Extrapolation procedure 2.0 ASSESSMENT OF SIMPLE JOINT SCF EQUATIONS 5.5.6.CONTENTS PAGE EXECUTIVE SUMMARY NOMENCLATURE 1.4 Efthymiou/Durkin equations (1985 and 1998) 4.2 Factors applied to the SCF database 4.1 Kuang equations (1975 and 1977) 4.3 Weld fillet profile 2.4 Assessment of parametric equations 5.1 Assessment database 5.5 Data used in SCF assessment 5.6 Beta = 1.3 UEG equations (1985) 4.3 Measurement of stress and strains 2.5 The influence of a weld fillet 2. Connolly and Dover equations (1990) 4.5 Hellier.6 Chord length and chord end effects 2.1 Methods of modelling tubular joints 2.3 Assessment of parametric equations 5.

APPENDIX A Parametric equations used in this study APPENDIX B Simple joint SCF assessment database APPENDIX C Extrapolation procedures 49 70 101 2 .

The assessment criteria which was agreed with the RPFG is given in Section 5. This latter work was carried out under the auspices of the HSE Review Panel for Fatigue Guidance (RPFG). which is given in Appendix B. the work being largely funded by the HSE. were developed. For the assessment of SCF parametric formulae for simple tubular joints the database above was refined. The joint acceptance criteria for this database was agreed with representatives from the Industry. The finalised database.2 and the equations assessed are given in Appendix A. which are given in Appendix A. was used to assess the existing commonly used parametric formulae and the new LR equations. Additionally. a comprehensive database of measured SCFs for full-scale steel joints and acrylic models was created. the report covers as assessment of the commonly used SCF equations for simple tubular joints including the new LR equations. ie the Lloyd’s Register equations.0 Introduction 3 . In the development of the LR equations.1 1 See footnote to Section 1. From this database the new LR equations.EXECUTIVE SUMMARY This report covers the development of a new set of SCF parametric formulae for simple tubular joints.

4 .

In the second study the database above was refined under the auspices of the Review Panel for Fatigue Guidance. FE analysis. The finalised database. These differences led to inconsistencies both in the measured SCFs themselves. The acceptance criteria for this database was agreed with representatives of the Industry. and also in the SCF parametric formulae based on these measured SCF values. many of the anomalies highlighted in the UKOSRP II project remain unresolved. Similar expressions also exist which give the relief obtained by grouting tubular joints. However. or in deriving parametric equations. A test programme on unstiffened tubular joints performed by Lloyd’s Register (11). reported in 1987 (10). INTRODUCTION In the 1970s with the increasing development of the hot-spot stress S-N approach to nodal joint fatigue life estimation. Despite considerable differences that exist between parametric SCF formulae.9) employing 3-D shell FE analyses.5) to use finite element (FE) analyses based on analytical models of cylindrical shells. are presented. it is the intention of future guidance (13) to give more specific directions on which parametric SCFs may be employed in fatigue design. a limited programme of work investigated anomalies between the existing simple joint parametric SCF equations.1. Subsequent equations by Wordsworth and Smedley (6. further highlighting the need for accurate simple joint SCF equations. These inconsistencies in SCF derivation are reflected in the hot-spot S-N curves used to estimate fatigue lives for simple tubular joints. While current guidance merely states that ‘the appropriate SCF’ should be used. In this report. titled the SCF assessment database and given in Appendix B. Visser (3) and Kuang et al (4. In the UKOSRP II project. The first parametric SCF equations covering simple tubular joints were derived by Toprac and Beale in 1967 (1) using a limited steel joint database. These new equations are given in Appendix A. differences arose between the experimental procedures used to derive stress concentration factors for simple tubular joints. have made considerable advances both in the accuracy of parametric equations and in the range of joints covered. titled the LR SCF derivation database and described in Section 3.7) using acrylic model specimens and by Efthymiou and Durkin (8. 2 studies. and the procedures that should be considered prior to performing either experimental tests. In the first study a comprehensive database of steel and acrylic simple joint SCFs was created. In the Lloyd’s Register (LR) group-sponsored project ‘SCFs for tubular complex joints’. has further emphasised inconsistencies between test results and the more commonly used parametric equations. it became clear that the determination of reliable stress concentration factors (SCFs) for tubular joints was fundamental to this concept. was 5 . From this database a new set of SCF formulae for simple joints (ie the LR equations) has been derived. Over this period. the current offshore installations guidance document (12). In both cases a relief factor is applied to the unreinforced joint SCF. largely funded by the HSE. parametric expressions have been derived to calculate the relief obtained by adding ring-stiffeners into a simple unstiffened tubular joint. accepts the use of parametric equations to determine hot-spot stresses in tubular joints. The prohibitive cost of testing scaled steel models led Reber (2).

6 . in the new HSE fatigue guidance only 2 sets are considered for recommendation. are given in Appendix A. However.used to assess existing simple joint SCF parametric formulae including the new LR equations. Therefore. 5 sets of parametric equations were considered. only these 2 sets. ie the new LR and Eftymiou equations. The equations assessed are given in Appendix A and the assessment criteria in Section 5.22 2 In the SCF assessment.

2. acrylic models and FE results have been utilised to develop parametric formulae which are verified against results from steel models. This assessment concluded that differences observed between SCFs measured on steel models. thus enabling the derivation of the ‘T’ S-N curve.2. differences in the definition of hot-spot stress and between the type of experimental technique employed has led to considerable variation in measured stresses between nominally identical joints. However.2Å rt (and not less than 4 mm). where r and t are the brace outside radius and thickness respectively. In this project the HSE definition of hot-spot stress and design guidance has been adopted. gave very similar results once differences in the local joint geometry and the experimental procedure were taken into account. The maximum extent of this local notch region is defined as 0. it is recommended that strain gauges should be located outside this notch region. to the weld toe. The dependence upon Årt has been derived from 7 . In previous studies. although alternative approaches are discussed and assessed. depending upon the source quoted. In this study the limited range of geometrics for steel joints led to the consideration of alternative methods into the database against which parametric methods would be benchmarked. proposed linear extrapolation of the maximum principal stresses. REVIEW OF EXPERIMENTAL TECHNIQUES The HSE fatigue guidance background document (14). studies of the measured stress around the brace/chord periphery indicate that under IPB and axial load the hot-spot stress may be located at an interim position for some geometrics. During early UKOSRP studies the HSE sought a clarification in the definition of the hot-spot stress and its calculation. This decision is discussed in detail in Section 3. The definition of hot-spot stress (14). However. acrylic models and photoelastic models or derived using a finite element analysis method. For this factor. as the region where fatigue cracking is most likely to initiate due to a stress concentration. 2. The first of these factors describes the overall geometry of the tubular joint and the detailed geometry of its welds. For most simple joint geometries and loadings the hot-spot will be located at either the saddle or crown.2 STRAIN POSITIONS GAUGE LOCATIONS AND STRESS SAMPLING Due to the rapid increase and variation in stress around the region of the weld toe resulting from the local weld geometry. and hence determine the joint fatigue life.1 METHODS OF MODELLING TUBULAR JOINTS In the UKOSRP I project an assessment of the performance of different modelling techniques was performed (15). The definition of the hot-spot stress varies considerably from a very general philosophy to a detailed description of its formulation. highlights 5 factors that influence the development of fatigue failure in welded tubular joints. outside the region of weld toe influence. the ‘hot-spot’ is described as the region where fatigue cracking is likely to initiate due to a stress concentration.

An alternative approach developed by Gurney (18). Swensson et al (19) tested 3 X joints (ß = 0.6 irrespective of the method used. with the least difference between the methods occurring when ß = 0.studies of the bending stress in tubes (16). 8 .65Å(rt) (It should be noted that this definition does not give guidance as to the gauge location on the chordside between the saddle and the crown locations). and this parameter has been empirically modified following detailed analyses of large scale tubular joint intersections in the UKOSRP and ECSC projects. In comparing the variation in SCF between the ECSC and the Gurney recommended gauge locations.12%. recommended a minimum strain gauge distance of 0. However more recent work on photoelastic joints (17) suggests that for some K joint configurations this notch region extends beyond 0. where T is the chord thickness. which indicate that the region of the notch stress is a function of the thickness of the plate upon which stresses are being measured. from the weld toe.67.10Årt). are located according to the location around the brace/chord intersection: chord saddle chord crown braceside = = = 5°arc 0.6%. The ECSC recommendations for the extent of the notch region to 0.35.2Årt. eg ß = 1. Gurney’s recommendation results from FE analyses of simple fillet welded joints in plates.4T. Gurney’s method describes the notch region better than the ECSC method.2Å rt have been adopted by the HSE as the suggested location for the gauges nearest to the weld toe or brace/chord intersection. A second set of gauges. which states that the hot-spot stress will be ‘nearly the same’ for joints with ß = 0.67 and 1. These tests showed that on average the SCF derived by the Gurney method exceeds the ECSC method by 7. 0.00). This result corresponds with the findings of another analysis by Wardenier (20).0. Where non-linear extrapolation is required a third gauge set is placed equidistant from the second gauge set (eg braceside = 1.44Å(rtRT) 0. although for joints with pronounced 3-dimensional effects. giving a maximum error in the order of 10% . enabling linear extrapolation to the weld toe.

the difference between SCF and SNCF will tend to increase).SNCF90 (1 − u 2 ) 9 . that for vertical braces the angle of the maximum principal stress (Y) tends to lie almost perpendicular to the joint intersection (ie Y = 0°). As the angle of the maximum principal stress increases.2. The methodology for determining maximum principal stresses from measured strains is shown below: Maximum principal stress = r max Maximum principal stress = r min r maxÂ r min = E 2 (e a + e c ) 1−v ! Å2 (1 + v ) Å(e b − e a ) 2 + (e b − e c ) 2 Where E = Young’s modules of the material u = Poisson’s ratio of the material For angle of maximum principal stress (a) tan 2a = 2e b − ea −e c e c −e a It has been suggested by Lalaini (21). Conversion of the strain concentration factor calculated in method (b) to a biaxial stress using: SCF = SNCF + u. Extrapolation of the strains perpendicular to the weld toe. Consequently the difference between the SCF obtained from principal stresses and from strain concentration factors (SNCFs) is minimised. possibly due to the brace angle decreasing. Three ways of utilising the strain gauge results to calculate the SCF or SNCF were assessed by Lloyd’s Register in this study: a) b) c) Extrapolation of the maximum principal stresses.3 MEASUREMENT OF STRESSES AND STRAINS Strains are measured on physical models using either strain gauge rosettes or via sets of single strain gauges placed perpendicular and parallel to the joint intersection or weld toe.

However. In this study. while examination of a K joint by Lalani (21) suggest a lower factor of 1. 2.30 steel. These results correspond well with the UEG design guide (22) which recommends an overall factor of 1. larger differences of around 10% can be found. for both steel and acrylic 90° joint specimens.16.20. However. For all joint configurations. the hot-spot stress S-N approach for simple tubular joints is currently based on linear extrapolation.36 acrylic) and SNCF90 is the strain concentration factor from the nearest parallel gauge to the weld toe or joint intersection].[where u is the Poisson’s ratio of the model material (u = 0.1. This inconsistency between the limited number of steel and acrylic joint specimens reviewed requires investigation in future testing programmes. while non-linear extrapolation. it was found that for 90° joints. However. the maximum principal stress (method (a)) was approximately 20% . the SCF derived from maximum principal stress (method (a)) is consistently 2% .4 EXTRAPOLATION PROCEDURE The HSE recommendations regarding extrapolation technique requires linear extrapolation to be adopted for 90°T and X joints (ß<1). although the degree of brace inclination does not appear to be a significant factor. the extrapolation procedures adopted by Lloyd’s Register in conjunction with acrylic model testing are given in Appendix C. Furthermore. X and K joints that employed both linear and non-linear extrapolation of maximum principal stresses to the brace/chord intersection (23). particularly at the brace crown under IPB. while for steel models the difference was nearer to 15%.30%) despite the aforementioned difference in the Poisson’s ratio. It was concluded that non-linear extrapolation exceeds linear extrapolation on the chordside for all loadcases by less than 5% irrespective of brace angle. irrespective of the joint configuration.3% larger than the SCF derived from biaxial stress (method (c)). employing a quadratic fit or Lagrange polynomial. For both steel and acrylic joints with inclined braces. the angle of the maximum principal stress increased as the brace angle decreased. it should be noted that to give a realistic 10 . however. Lloyd’s Register have reviewed 67 simple acrylic T/Y. Linear extrapolation was performed through the 2 rosettes nearest to the brace/chord junction. it was confirmed that the angle of the maximum principal stress varied by less than 10% from the perpendicular to the joint intersection. This can be accounted for by the different Poisson’s ratio for the 2 materials.50%) than on similar acrylic model specimens (23% . see Table 2. far more significant differences occur between the maximum principal stress and the perpendicular strain on steel joints (30% . while for inclined Y and K joints it is stated that in some cases there may be a non-linear geometric stress distribution. On the braceside. The HSE does not give a method for extrapolating stresses. u = 0. utilised these 2 rosettes and a third rosette positioned on equal distance from the second rosette.25% larger than the perpendicular strain (method (b)) for acrylic models.

based on FE analysis of K joints. Smedley also considered this expression to be acceptable for 90° joints on the braceside.76 non-linear).5 were excluded.0 + [SCF Brace mid−surface − 1.0 ] x exp − 0. and is generally used in conjunction with the Kuang parametric equations (5). The addition of a weld fillet to a tubular connection results in a reduced SCF which is measured at the weld toe rather than at the joint intersection. and SCF = 0. SCF Brace weld toe = 1. However. uses an exponential decay function to approximate the SCF at the braceside weld toe. For ß = 1 joints (particularly ß = 1 X joints) where the stresses tend to be relatively small.5.5 THE INFLUENCE OF A WELD FILLET Acrylic model tests and some FE analyses do not include a weld fillet on the tubular joint intersection.1 Increase in SCF using non-linear extrapolation Joint config T Y X K No of joints 20 6 9 22 Axial Load OPB IPB Chord Chord Brace Brace Chord Brace Chord Brace Saddle Crown Saddle Crown Saddle Saddle Crown Crown 4% 0% 6% 6% 2% 5% 4% 10% 4% -1% 2% -2% 3% 5% 6% 6% 3% 6% 5% 12% 4% -1% 1% 8% 2% 0% 3% 7% 2. and therefore require some modification to their measured hot-spot stress if they are to be compared to steel joint test results. SCFs using non-linear extrapolation could be twice those using linear extrapolation (eg SCF = 0.1 The inclusion of a weld fillet Employing acrylic model T and 90° X joints. Alternatively. joints where ß = 1 or where the measured SCF was less than 1. Smedley (24) produced a weld fillet correction factor based on the chordside weld fillet leg length. Marshall (25) proposed a design equation.5T+t írt This equation. 11 . for inclined braces it was considered that this factor was not applicable.41 linear. applied to the braceside mid-section stress. Table 2. SCF Weld toe = SCF No weld 3í 1 + X T (where x is the weld fillet leg length on the chordside) While being specifically designed for the chordside stress on 90° joints. 2.assessment of the differences between linear and non-linear extrapolation. It is important to note that the extrapolation of stresses on a model with no weld fillet should not be foreshortened to where the weld toe would be in an attempt to reflect the influence of the weld.

2. It should be noted that for many of the steel test specimens. showed a consistent stress reduction factor of 0. the weld leg length on the braceside exceeded the brace thickness (t). suggested that for a specimen with no weld fillet. see Figure 2. the extrapolation to the brace/chord intersection should be foreshortened by half the standard fillet leg length to simulate the inclusion of a weld fillet.88 on the braceside. when a weld fillet leg length = t/2 on the chordside and = t on the braceside was described in accordance with API recommendations for controlled weld profiles (27). 12 . Further work by Lloyd’s Register using this approach on joints with ß<1.A more recent approach (26). particularly those of small scale.95 on the chordside and 0.

13 .

958 19.71 10.856 0. For example.95. using the Wordsworth/Smedley equation (6) on a T joint.52 OPB 20. At the saddle location.3. It was suggested by Wordsworth that for joints with equal chord and 14 .69 Wordsworth factor 0.2 Weld cut-back at the saddle location on ß = 1 joints Following a series of steel joint tests on ß = 1 T joints under OPB. Wylde (29) described the variation in measured SCFs between tubular members of different diameter. Some of these inconsistencies in measured SCF values may be attributed to the varying degrees of weld cut-back used in the test specimens.856 (Joint parameters: ß = 0.95 0. t = 1.95 12. It was concluded that caution should be applied to ß = 1 joints since no parametric equation takes into account the weld profile.850 0.63 Mean reduction factor No weld SCF 14.54 0.38 No weld SCF Axial 24. the Wordsworth weld reduction factor gives the closest agreement to the measured weld fillet reduction factor.87 0.6 and a = 5.5.951 Braceside Inc weld SCF 12.50 0.29 IPB 6.40 5.0. the separation between the weld toes at the saddle locations may be considerably less than the chord/brace diameter.2 FE analysis of a joint with and without a weld fillet (linear extrapolation) Loading Chordside Inc weld No weld ÷ SCF inc weld 23.Table 2. that for the joint geometry modelled by Efthymiou. see Figure 2. that the largest differences between measured SCFs and predicted SCFs occur when the brace diameter approaches the chord diameter.95 0. Wylde felt that the difference between the 2 types of joint was most likely due to the differences in the weld profiles rather than the joint size. Table 2. c = 28.5.86 Smedley factor 0.861 0. On the chordside.3 Reduction in SCF due to the inclusion of a weld fillet Measured factor 0.87 Marshall factor 1. the SCF increases by 38% under axial load and by 50% under out-of-plane bending by altering ß from ß = 1. For ß = 1 joints with ‘extreme’ weld cut-back.09 No weld ÷ Inc weld 0.932 0. the SCFs on 168 mm diameter joints were more than twice the SCF values on 457 mm diameter joints.3.0 to ß = 0.0) It can be seen in Table 2.963 6.18 0.88 Chordside Braceside 2.00 0. the SCF decreases rapidly as ß approaches 1. It was noted by Wordsworth (26) and confirmed in this project.62 4.

By employing this expression for ß = 1 joints. This expression has been included in the new Lloyd’s Register parametric equations for simple tubular joints. Lloyd’s Register found that ß’ may be reasonably estimated by: ¾’ = 1 . see Figure 2. ß should be defined as the weld toe separation relative to the chord diameter (ie d’/D). 2.65 (Y o ) ) Where Y° is the degree of weld cut-back (If Y is not defined.( yt X sin 0. Following a detailed analysis of the parameter ß’ = d’/D. straight and convex weld profiles and concluded that outside the notch region the weld profile was not a highly significant 15 . This study compared concave. a default value of Y = 20° is suggested). However. it was noted that the weld toe separation d’ is neither easily calculated nor measured.5.brace diameters. and consequently more accurate fatigue lives may be estimated. SCFs may be derived to a greater accuracy for these complex joint configurations.4.3 Weld fillet profile The profile of the weld fillet was examined in an FE analysis of X joints by Lieurade (30).

The problem is further complicated in the test rig by variations in the clamping of the chord (ie pinning or fixing the chord ends). In the literature study performed at the beginning of this project. 16 . although Wordsworth/Smedley (6) performed tests on acrylic models with a values up to a = 40.6 CHORD LENGTH AND CHORD END EFFECTS Most test specimens are constrained by their physical size to an a (= 2L/D) value in the range a = 5 to a = 15. although a concave weld will always give the lowest stress. In practice over 60% of nodes have a values in excess of a = 20. and the inclusion of rigid diaphragms which act in a similar manner to ring-stiffeners. it was frequently found that there was very little if any information given about the chord length.factor. Since most joints are generally modelled to a concave weld profile there will be negligible influence on the hot-spot stress due to marginal differences in weld profile. 2.5 and 35% of nodes exceed the (a = 40) upper bound for a recommended by most of the parametric SCF equations. confusion often arises over the somewhat ambiguous definition of the chord length and a parameters in both test specimens and in structural nodes. Furthermore. see Figure 2. Therefore all joints in this project were assumed to be pinned at the chord ends when assessed against the parametric equations. chord end fixity or the test rig structure.

This fact is not recognised by the Kuang (5) and HCD (31) equations which will eventually lead to underprediction in the crown SCFs as a increases. then a simpler and better analysis would result from utilising these known FE values rather than estimating these effects from simply supported joint approximations. For some steel joints with relatively short chord lengths the quoted value of a increased by around 30% once the pinned supports were also considered.6. Since the overall structural FE analysis readily gives the chord in-plane bending and axial stresses.5 to a = 40. These correction factors were found to depend not only on 17 . act in a similar manner to ring-stiffeners and restrict the degree of chord ovalisation.0. the quoted a value should be based on the separation of the chord end diaphragms (ie the physical length of the chord specimen).2. Therefore at the saddle. Wordsworth/Smedley tested joints with a varied from a = 13. 2. it was found that there were considerably more SCF underpredictions for the crown under axial load than for other loadcases/locations. Consequently. it was found that this approach could lead to significant differences in the overall hot-spot stress for some joint configurations when compared to the approximation described above. it is almost certain that a poor SCF result will be obtained. An analysis of these results concluded that the predicted SCFs were underestimating the measured SCFs due to an incorrect interpretation of a in the chord in-plane bending term. the stresses in the chord wall in an offshore structure are not simply related to the brace axial force. whereas for the chord in-plane bending at the chord crown a should be based on the separation of the pinned supports.1 Supported chord effects at the crown under axial load The SCF at the crown location under axial load in T/Y joints is composed of a local shell deformation effect and an overall chord in-plane bending effect that is linear in a. However Efthymiou and Durkin (8) found that joints with short chord lengths (a<12) exhibit a significant reduction in the SCF at the saddle. In the Lloyd’s Register fatigue analysis using the PALS software package. any chord length effects at the saddle will be due to the restriction on chord ovalisation imposed by the chord end diaphragms. Efthymiou and Durkin proposed correction factors to be able at the saddle location on joints with short chord lengths.6. Consequently. and concluded that chord length was not a significant factor at the saddle. In this assessment of the T/Y joint SCF equations. due to the restrictions on chord ovalisation caused by the locality of the chord end supports. In some cases the quoted value of a is based on the physical chord length. Chord end diaphragms used to support the test specimen. The chord in-plane bending term (B0) has been derived from the brace axial force applied to a simply supported centrally located brace where: B0 = C t (b − t/2y)) (a/2 − b/ sin h) sin h (1 − 3/(2Y)) l C t b (a/2) sin h While B0 gives a good estimate of the bending stresses in a test specimen.2 Short chord effects at the saddle Unlike the crown where the chord length influences the degree of bending in the chord.

Therefore. At the outset of the project lengthy discussion took place with both the HSE and with many representatives from the oil industry regarding the composition of this database. generally gives good correlation to similar steel joint configurations. it is important that end effects are taken into account prior to any assessment. To meet this objective. it was felt that published data was of variable quality and often the analysis methodology was insufficiently documented. It should be emphasised that this is a chord end ovalisation effect rather than a chord support effect. and as a consequence the SCFs measured on test specimens may be significantly less than those observed on more realistic joints. 18 .1 CRITERIA FOR THE ACCEPTANCE OF SCF DATA The initial objective of this project was to gather measured SCF data relating to simple tubular joints. chord length effects are far less pronounced. With regard to FE results. 3. As steel joint test specimens are used as a benchmark for testing the validity of parametric equations. K and KT joints. It was considered that acrylic models gave a good representation of the stress distribution observed in equivalent steel joints. This database of steel and acrylic joint SCFs was created for the development of a new set of parametric equations for simple tubular joints. SCF or fatigue endurance. particularly for X. This conclusion was supported by assessments of experimental techniques performed in the UKOSRP I project (15). Therefore. 3. nearly all the SCFs derived from steel joint test results are influenced by the locality of the end restraint. So that for a T joint under axial load with a relatively thin chord wall and a = 5. the short chord correction factor can halve the saddle SCF compared to a similar node with a = 10. Most steel joint test specimens are constrained to having a values in the range a = 5 to a = 10. but also on ß and c.a. The database of photoelastic model tests was considered too small to be validated independently. it is recommended that consideration should be given to short chord effects before any physical model or FE analyses are performed. with the larger diameter joints tending to have the smaller a values. it was decided to exclude SCFs derived by FE analysis from the database. and that acrylic joints should be included in the database provided allowance was made for differences in experimental techniques. The work was largely funded by the HSE. this literature search has provided a substantial database for this project. In future. whether for static strength. for most X joints where the joint is supported by the brace ends and no chord end diaphragms are fitted. It was felt that steel data alone was insufficient to enable a new set of parametric equations for simple joints to be developed. particularly using 3-D shell elements. a literature search was undertaken to locate experimental data relevant to this project. In association with SCF results from work performed in-house. while accepting that FE analysis. Therefore. DATABASE This section describes the Lloyd’s Register SCF derivation database.

The separation between the chord end diaphragms is required to derive ovalisation effects. authors. however the necessity of this cross-checking procedure was proved with several inconsistencies in SCF results occurring between published papers. non-overlapped. The measured SCFs and their location on the joint . Tension/compression . balanced. and the following subset of data was felt to be typical of offshore platforms: t[1. crown or an interim location).(chordside or braceside .(axial. Only joints with parameters typical of those noted in practice were included. the danger of bias in the analysis was noted. stress/strain measurement. However. and the separation between the chord end supports is required to calculate the degree of chord bending in the test specimen. h. the results of each test were included. (i) Single-plane. Experimental details . L.(D. location of the nearest gauge. Number and configuration of the joints tested. IPB . and this was the value recorded in the database.0 (only b = 1 joints at the saddle where the degree of weld cut-back is recorded). unstiffened and ungrouted tubular joints. d.5. extrapolation type. the average of the 2 results was taken.single. Several saddle results . Geometric and parametric details of the joint .title of paper. T.(i) The following data items are included: a) Source of the SCF data . The chord length quoted in most papers was the physical length of the test specimen. were recorded on the database.where a joint is axially loaded in both tension and compression. c. t. it would have been desirable to record 2 ‘chord length’ values if these had been quoted. b.05. with a chord diameter m150 mm and the strain gauge nearest to the brace/chord intersection being outside the notch region. This exercise was time consuming and not always feasible. where the data has been located. b[1. 19 . a).where several configurations were tested with the same joint parameters. b) c) d) e) f) g) h) i) Verifying the accuracy of the data involved locating the source paper or experimental reports where possible and comparing the given results to a second publication. the maximum SCF result was taken.where 2 or 4 individual SCFs are quoted for one joint. the date of publication and any confidentiality restrictions. OPB. unbalanced). c[40. However.specimen material.saddle. weld fillet inclusion and chord end fixity. Repetitions . Applied loadcases . t. SCFsm1.

but would appear to give a better estimate of the difference between linear and non-linear stress since this value is based on a database of 57 joints analysed in the LR project (23). These factors were applied to all braces despite these limitations. SCF = SNCF X 1. however there is insufficient data on strain concentration factors from inclined braces to derive an estimate of this difference.65 (Y o ) ) Where Y° is the degree of weld cut-back w w w (If Y is not defined.95 on the chordside and of 0.23 for T joints (acrylic joints). 20 .95 (including b= 1 joints). Maximum principal stress or biaxial stress derived from strain concentration factors: it is considered that no factor is required to differentiate between SCFs derived from either biaxial stress or maximum principal stress. see section 3. since some papers appear to round strain gauge location to the nearest mm.2 FACTORS APPLIED TO THE SCF DATABASE w Strain gauge locations: the nearest strain gauge to the weld toe or to the brace/chord intersection must be in excess of a distance of 0.2. this has been achieved by the use of a number of correction factors covering differences in the experimental derivation of the hot-spot stress.15 for T joints (steel joints). 3.86 on the braceside have been adopted. This small allowance did not result in a significant increase in the extrapolated SCF value. Stress and strain measurement: the recommended factors for converting perpendicular strain (SNCFs) to maximum principal stress (SCFs) are: SCF = SNCF X 1. This factor is less than the 10% difference proposed by UEG (22). w Extrapolation procedure: the recommended factor for converting non-linear SCFs to linear SCFs is: Linear = non-linear X 0.2írt. Weld cut-back at the saddle location on b = 1 joints: following a detailed analysis of the parameter b = d’D. For inclined braces larger differences between SNCFs and SCFs would be expected.18írt from the weld toe or the brace/chord intersection. This is based on a 10% reduction in the recommended value of 0. w The absence of a weld fillet: weld fillet correction factors of 0. a default value of Y = 20° is suggested). these values are based on the only measured weld fillet comparison available (ie the KSEPL joint) (28). Lloyd’s Register found that b’ may be reasonably estimated by: t bÂ = 1 − ( c X sin0.(ii) The initial analysis of the SCF data involves the standardisation of the measured SCFs to conform with the HSE recommended method.

The Kuang equations are based on a mean fit to the database of FE joints examined. cover T/Y. A substantial number of steel T joints were tested with a values in the range a = 4. have a values based on the separation of pinned end supports rather than chord end diaphragm separation. however at the present time no equations are recommended by the HSE. but are merely expressed as chordside or braceside.0. The Kuang equations were not designed to cover joints with b>0. w It should be noted that many steel joints in particular. the Kuang equations have the most restricted validity range. while the reduction in SCF due to the restriction on chord ovalisation noted at the saddle on joints with short chord lengths tended to be underestimated.8 and do not cover X joints or the unbalanced OPB loadcase for K and KT joint configurations.3 to 12. Chord length effects due to the restriction of ovalisation caused by the diaphragms at the chord ends leads to considerable reduction in SCFs at the saddle for short chord lengths. the following points were noted: (i) Of the equations reviewed. Even between the more commonly used parametric equations. The tubular connections were modelled without a weld fillet. no account is given to the beam bending effect. (ii) For T/Y joints under axial load. and in their recommended ranges of applicability. K and KT joint configurations and utilise a modified thin-shell finite element program specifically designed to analyse tubular connections. 4.1 KUANG EQUATIONS (1975 AND 1977) The Kuang equations (5). Therefore the stresses calculated using the Kuang finite element models are considerably different from the HSE definition of hot-spot stress. With regard to the Kuang equations. SIMPLE JOINT SCF EQUATIONS The parametric equation gives the designer the most convenient way of estimating the hot-spot stress in simple tubular joints. To eliminate these joints from the database would have had a substantial effect on the number of steel joints that could be considered. and stresses were measured at the mid-section of the member wall. listed in Appendix A. calculated.0. and do not indicate the location of the hot-spot around the periphery. and consequently underpredictions at the crown for high a values may be anticipated. there are significant differences in the way the hot-spot stress is defined. 4. This gives a more accurate assessment of the SCF at the crown which is dependent upon the level of bending in the chord.This expression has been included in the new Lloyd’s Register parametric equations for simple tubular joints. 21 . and consequently cover the fewest joints in the database. This problem is described by Efthymiou (8) who published short chord correction factors to be used in accordance with the Efthymiou SCF equations when a (=2L/D) <12. It should be noted that some of the comments made here are based on the assessment covered in Section 5.

(vi) For KT joints under balanced axial load. the HSE recommendations for the derivation of the hot-spot stress were followed. the following points were noted: (i) On the braceside.2 írt) ‘notch’ zone. on the braceside. yet gives a reasonably good fit to the braceside SCFs under OPB. with the predicted SCFs for the KT joints in this database up to 4 times larger than the measured SCF values. However. given that the database of FE joints utilised by Kuang had generally short chord lengths. the Kuang equations show good agreement with the measured SCF values on the chordside. For the W/S and Wordsworth equations.(iii) No account is given to chord length effects at the saddle. The second concerns the number of sets of strain gauges adopted around the brace/chord intersection. Therefore. for all loadcases and measuring locations. there is a possibility of underestimation of SCFs for more realistic chord length (a) values. and the K and KT joint configurations were covered by Wordsworth in 1981 (7). For example. (v) For K joints. the W/S and Wordsworth equations estimate the SCF using a simple factor applied to the chordside SCF. the Kuang equation gives a very poor fit to chordside SCFs under OPB underpredicting 70% of joints in this database. However. the predicted SCFs on the braceside tend to be rather conservative relative to the measured braceside 22 . using maximum principal stresses from outside the (0. which appears to be generally linear except where the stress distribution was found to be particularly non-linear and therefore open to error. Generally. The Kuang equations are still widely used in the fatigue design of offshore tubular joints. Non-symmetric K joint configurations exhibit the largest difference between the measured SCFs and the predicted SCFs. some areas of uncertainty do exist. the Kuang KT joint SCF equations differs considerably from the corresponding equation for a K joint. but it is unclear whether interim sets of gauges were adopted.5 is generally poor. although there can be considerable variation in the degree of underprediction depending upon the loadcase considered. The equations covering T/Y and X joint configurations were published by Wordsworth/Smedley in 1978 (6). Consequently. since the Kuang K joint equations were specifically designed for joints with symmetric braces. The first surrounds the extrapolation technique adopted. The Wordsworth parametric equations specifically cover the saddle and crown.2 WORDSWORTH/SMEDLEY EQUATIONS (1978 AND 1981) The Wordsworth/Smedley (W/S) equations were derived using acrylic model test results on tubular joints modelled without a weld fillet. the Kuang equations are generally conservative for all values of b. due to the chord end restraints. particularly under IPB where for some configurations the hot-spot stress occurs between the saddle and crown. 4. (iv) The performance of the Kuang equations for T/Y joints for b values above 0.

The following points may be made with regard to the UEG equations: (i) All the comments for the W/S equations hold true for the UEG equations except for the weld-cut back simulation at b = 1. (ii) For joint configurations with equal chord and brace diameters (ie b = 1 joints). and joints with b = 1.0.0. and by using the W/S and Wordsworth equations a conservative assumption of R/T = 12 must be made in determining the SCF.6 for c<20 for cm20 These modification factors based on comparisons of predicted and measured results from both static and fatigue studies.0 Q’c = 480/c(40 . 4.3/b(1 . are based on the W/S and Wordsworth equations with a modification factor applied to configurations with high b(b>0. However.0 Q’b = 0. The Wordsworth equations generally give a good estimate of the measured chordside SCFs.6 for b>0.SCFs. Therefore the influence of adding further braces to a simple T joint can clearly be determined.833c) for b [0.833b) Q’c = 1. apply to all joint configurations and are designed to give a characteristic set of equations (ie underpredicting 5% of the measured data).8 and high c. Consequently.6) or high c (c>20) values.3 UEG EQUATIONS (1985) The UEG equations proposed in 1985 (22). A significant number of tubular joints are designed with relatively thick chords. (v) The W/S and Wordsworth equations are only valid for chord radius to thickness ratios R/Tm12. 23 . R/T ratios in the range 8 [R/T< 12 are not uncommon in practice. to avoid the use of ring-stiffeners. a b value of b = 0. (ii) The factors íQ’b and Q’c are both applied under axial load and OPB while only íQ’c is applied under IPB where: Q’b = 1. the Wordsworth equations tend to underpredict measured SCFs on joints with b = 0.0 where there is a significant degree of weld cut-back.98 was taken at the saddle location as recommended by Wordsworth to simulate the weld cut-back found at the saddle in typical steel b = joints. (iv) Under axial loading and OPB at the saddle location. (iii) The Wordsworth equations for K and KT joints utilise carry-over functions applied to the T joint expressions. which is accounted for in the Q’b term. it should be noted that the braceside SCF on a simple tubular joint rarely exceeds the chordside SCF.

but modelling of the stress distribution around the brace/chord intersection and the proportions of bending to axial stress through the member thickness. Efthymiou (9) published a comprehensive set of simple joint parametric equations covering T/Y. These equations were designed using influence functions to describe K. these equations tend to be nearest to a mean fit and consequently more underpredictions are frequently observed. (iv) Under unbalanced OPB. (iii) The Efthymiou equations give a comprehensive coverage of all the parametric variations and are designed to be mean fit equations. and the fewer conservative assumptions made. K and KT simple joint configurations. in accordance with the HSE recommendations. the overall programme included not only hot-spot stress estimates. Connolly and Dover (HCD) equations (31) were published in 1990 and were primarily developed to improve fracture mechanics estimates of remaining life for a joint rather than for tubular joint design.5 HELLIER. it is unclear why the chord end fixity should be a factor. Over 150 configurations were analysed via the PMBSHELL finite element program using 3-dimensional shell elements. KT and multi-planar joints in terms of simple T braces with carry-over effects from the additional loaded braces. and have a range of applicability for the b parameter limited to 24 . With respect to the Efthymiou/Durkin equations. CONNOLLY AND DOVER EQUATIONS (1990) The Hellier. The SCF equations themselves cover T/Y joint configurations alone. as would be expected. the Efthymiou equations give a good fit to symmetric K joints or the outer braces in KT joints. In 1988. with some consideration being given to boundary conditions (ie short chords and chord end fixity). the following points may be noted: (i) It has been shown by Efthymiou that the saddle SCF is reduced in joints with short chord lengths. The short chord effect at the saddle is due to the presence of chord end diaphragms. X. therefore. The hot-spot SCFs were based on maximum principal stresses linearly extrapolated to the modelled weld toe. Efthymiou and Durkin (8) published a series of parametric equations covering T/Y and gap/overlap K joints. Factors have been included in the Efthymiou parametric equations to cover short chords. The equation for X joints at the brace crown under axial load does not equal the corresponding T/Y joint equation excluding chord bending terms. due to the restriction in chord ovalisation caused by either the presence of chord end diaphragms or by the rigidity of the chord end fixing onto the test rig. Due to the greater correlation with steel models by the Efthymiou FE models. the measured saddle SCFs on joints with short chords may be less than for the equivalent joint with a more realistic chord length.4 EFTHYMIOU/DURKIN EQUATIONS (1985 AND 1988) In 1985. and the results were checked against the SATE finite element program for one T joint and 2 K joint configurations. (ii) The T/Y joint equation for the saddle under axial load includes a short chord correction factor for either fixed or pinned ends.4. but consistently appear to underestimate the SCF in the branch with hmax in non-symmetric K joints. Therefore. Consequently. 4.

Therefore. the majority of specimens modelled were influenced by chord end constraints. and in other cases the saddle equation SCF was larger than the hot-spot (saddle) SCF. the linear overall beam bending effect has not been accounted for in the expressions. The position of the hot-spot stress may be identified as being at the saddle or crown.b [ 0. these expressions give comprehensive coverage of realistic geometries observed on offshore structures. Furthermore. The expressions were then refined by the addition of further terms. where the hot-spot SCF (at the saddle) exceeds the saddle equation SCF by up to 20%. These expressions give a comprehensive description of the stress distribution around the periphery. but the hot-spot SCF equation (eg chordside) may not give the same result as the saddle/crown equations. some anomalies are associated with these expressions. the HCD and Kuang equations often gave similar predictions. Therefore. these expressions give the angle around the brace/chord intersection of the maximum SCF. the following points were noted: (i) For T/Y joints. the HCD equations were initially derived using the simple form: SCF = a 1 aa 2 ba 3 ca 4 ta 5 ha 6 where variables a1 to a6 are determined using a regression analysis method. The failure to consider chord length effects must have an influence on the estimation of SCFs for joints with more realistic chord lengths. which differed from the other equations reviewed.6 (iii) The derived expressions estimate the SCF at both the saddle and crown locations and also the maximum SCF around the brace/chord intersection.8. 25 . In addition. on both the chordside and braceside. In this study for b values near to b = 0.8. However. It has been noted in previous numerical modelling of tubular joints using the PAFEC semi-loof FE package (32) that poor results were obtained for large brace relative to chord diameters. and in a similar manner to the work by Kuang weld fillets were not modelled. and hence an upper recommended limit of b = 0. Cases were observed. at the chord crown under axial load. With regard to the HCD equations. In a similar manner to the Kuang equations. and it is particularly worth noting that the chord radius to thickness ratio (c) was studied to as low as c = 7. for all loadcases. consequently this upper bound test parameter was removed. A thin-shell finite element method was developed using the PAFEC package. further correlation studies between the PAFEC FE package and physical specimens would be beneficial for large b ratios. these expressions are currently limited in their application.21. however.1. (ii) With the exception of the b range. so that these expressions are applicable for all a values exceeding a = 6.8 was proposed. further work is intended for b = 1 joints and for X and K joint configurations. The chord length (a) effect was deemed to have little effect for a values greater than 13. but again no account of this effect was taken.

Therefore. in accordance with the LR SCF equations. This rather subjective view of the reliability of SCF prediction led LR to multiply the mean fit equation. No safety factor is suggested for SCF equations in the HSE guidance notes (12). however the 0º and 180º expressions for IPB do not give the same SCF. by minimising the percentage difference between the measured SCF values and the estimated SCF values. Therefore the HSE suggested that a consistent set of SCF equations should be developed. so that the expression for one brace on a K joint would be that for an equivalent Y joint with factors to describe the stiffening effect of the additional brace and loading given the magnitude of the load applied to this additional brace. although it was considered that the differences would be small. The LR SCF derivation database is almost identical to the SCF assessment database that was used in this project to assess SCF formulae.For symmetric 90º T joints the 2 crown positions would yield the same SCF. the following points should be noted: (i) These equations generally give the SCF at the saddle and crown locations (except for IPB). Short chord effects were also considered to be of particular significance. 4. consideration was to be given to b = 1 joints where particular problems had been identified due to the degree of cut-back of the weld at the saddle location. This is most likely to be the case for K/KT joints under axial load. by one standard deviation of this mean fit to the LR database. comparisons between the SCF assessment database and the LR equations should be considered in the knowledge that the equations themselves were largely developed from the dataset against which they are being compared. however it became clear that for the data available to LR derivation of short chord correction factors could not be achieved. This led to a design equation that underestimated around 15% to 20% of results. it was felt that SCF equations that are currently used in offshore tubular joint design have an appropriate level of safety. without any independent verification being possible. With regard to the LR equations. in 1991. and may underestimate a larger SCF if located between these locations. Furthermore. It includes both steel and acrylic joint data. An initial review of the performance of existing SCF equations with regard to the LR SCF database highlighted a number of anomalies in the goodness of fit of all these expressions. the Efthymiou short chord correction factors were used. and consequently some SCF equations are designed to be a mean fit to the associated experimental dataset. while some are designed to be a characteristic fit. and has the same parametric and geometric limitations. The design factor to be applied to these mean fit expressions was also a matter for some debate. Therefore. Overall. described in Section 3. The LR equations (33) were developed as mean fit equations to the LR SCF derivation database. The objective of these equations was to employ influence functions wherever possible.6 LLOYD’S REGISTER EQUATIONS (1991) The Lloyd’s Register (LR) equations were developed as part of the “SCFs for simple tubular joints” project which was largely funded by the HSE. 26 .

Consequently. generally by some form of least squares curve fit to these data points. UEG and HCD are based on either acrylic model test results or on the results of thin shell FE analyses. (iii) The LR equations are limited to c ratios greater than c = 12. since equations have been produced. The HSE Review Panel for Fatigue Guidance (RPFG) considered that the new fatigue guidance should give recommendations regarding the applicability of SCFs. (v) The form of the equations. The equations derived by Kuang. while a significant number of tubular joints are designed with c values below this limitation. gives a more logical influence function format.40% of measured SCF results depending upon the equation/joint configuration/loadcase/measuring location considered.(ii) The LR equations use the Efthymiou short chord correction factors. were based on a least square fit to FE results using 3-D shell elements including a realistic weld fillet. The LR SCF derivation database described in Section 3 was refined and used to assess the applicability of the more commonly used SCF parametric equations for simple joints. The panel was seeking to give a clear definition of the requirements of an acceptable parametric equation for simple tubular T/Y. the Efthymiou equations frequently give the closest agreement with steel joint test results in terms of scatter. no specific probability level is currently defined for SCF equations. From this assessment a series of matrices for each joint type and 27 . It was found that the parametric equations generally underpredict around 0% .5% probability of failure is predicted. The Efthymiou equations on the other hand. ASSESSMENT OF SIMPLE JOINT SCF EQUATIONS Throughout the duration and subsequent to the completion of this project. acceptable levels of underprediction have not been defined. W/S. Unlike the current tubular joint S-N fatigue design curve where a 2. the equations themselves do tend to be conservative relative to steel joint test results. while being more complex for ‘hand calculations’. chord bending effects and the weld influence have been considered in deriving these equations. the number of SCF underpredictions and to a lesser extent the number of significant SCF overpredictions. which have not been independently verified. no weld fillets were described and consequently the test results themselves tend to be conservative. While it is clear that an equation is superior if it exhibits less scatter with respect to test data for all combinations of parametric values. Therefore. a considerable amount of discussion has centred on defining the characteristics of a ‘good’ or ‘poor’ SCF parametric equation. which largely removes the problem of joint classification. X and K joints (13). This definition takes into account the scatter of the equation with respect to measured SCF results. 5. In both these methods of analysis. but are generally less conservative than other parametric equations. (iv) Short chord length effects.

c = 28. some anomalies had been noted in the design SCF equations when compared to acrylic model test results which would not be noted if only steel joint data was considered.loading case was produced. than remove acrylic models from the assessment of parametric equations.0 13.0 6.9 5.6 24. see Table 5.8 22.2 21. A direct comparison of one joint tested in steel.0 6. it was clear that a substantial number of steel models would be required given the combinations of joint configuration. Initially.9 6.3 (t = 1. It was felt to be more important to highlight potentially unsafe SCF design equations given the aforementioned limitations.6 23.9 5.0 6. Comparisons of the steel and acrylic model subsets within the database were performed to assess potential differences between the modelling techniques. applied loadcases and in location of strain gauges.0.7 21. a = 5. acrylic joints are not steel joints. steel data was known to be limited in geometric range. approximations had to be made to represent the effect of the weld fillet. b = 0. and often small differences in the parametric values are reflected in differences between measured SCF values. direct comparisons were sought for nominally identical joints tested in both steel and acrylic.1.5 21. KT joints.5 7. There was insufficient data to produce a matrix for 5. Table 5. very few acrylic and steel tests were performed under the same conditions. Additionally.8 6.1 14. Furthermore. the steel data was potentially biased due to the short chord lengths of the specimens and the proportionally large welds modelled.5 6.9 7. However.3 12.8 15.4 4.1 SCF comparison of experimental methods Loading Location on joint Steel Acrylic model Wimpey FE UKOSRP II KSEPL FE 23. With regard to the measurement of SCFs. acrylic and using 2 FE analyses (11) showed generally good agreement between all the methods employed.2 14. once differences in the experimental techniques and the inclusion of a weld fillet were considered.0 13. However.0 3. Acrylic models do not include weld fillets and consequently. thus leading to possible underestimation of tubular joint stresses in practice.2 8.9 2.1 1.1 13.3 3. The short chord length and large weld fillet effects would both tend to reduce the measured SCFs. Unfortunately.0 1. It was agreed that the ideal position would be to use acrylic models/finite element analyses to identify the areas of potential underestimation.6 4.2 Axial load OPB IPB Chord Saddle Chord Crown Brace Saddle Brace Crown Chord Saddle Brace Saddle Chord Crown Brace Crown Chordside Braceside 24. loadcase and measuring positions to be assessed.0.0 14.4 5. h = 90º) 28 . they merely represent steel models.5.6. and model these joints in steel.5 1.1 ASSESSMENT DATABASE The decision to compare parametric equations with a combined database of steel and acrylic joints led to a great deal of debate over the validity of this approach.4 2.

5% difference in SCF between the steel and acrylic joints. (ii) Steel joint data tends to cover a more limited number of joint parameters.9 18.Axial loading (SCF parametric equations .0 21. However. in comparison to the parametric equations considered.7 % Joints Underpredicting 9.4 24.9 SCF>1. Although this pattern is repeated for all the equations except Kuang in this example it should be noted that for other joint types/loadcases the situation could be reversed ie the average steel joint P/R exceedence being greater than the average acrylic joint P/R exceedence.2 7. Table 5.1 18.6 10.2 Steel vs Acrylic .19 % Standard deviation 9.6 20.3 70.0 Efthymiou SCCF>0. Given the fact that the steel and acrylic joint SCF databases contain difference joint geometries.07 Acrylic 1. it was reassuring to find so little difference between the steel and acrylic joints in their average level of SCF.T/Y joints . there is less than 3.6 12.2 16.7 17. It can be seen that the average level of SCF overprediction using the equations varies between the steel joints and acrylic joints with the largest difference occurring.1 17.15 Steel 1. and is very sparse form some joint configurations and loadcases. One example for the chordside of T/Y joints under axial loading is presented in Table 5.12 Acrylic 1.8 28.1 Loading Equation No of joints 22 58 28 58 28 57 14 20 28 50 28 59 Wordsworth Chord side ß < 1.16 Steel 1.5 9.5 28 steel 59 acrylic Kuang HCD LR UEG The comparison of steel and acrylic joint SCF results can be summarised as follows: (i) Steel data models alone must give the best comparison with full scale nodal joints.15 Acrylic 1. it is understandable why differences were noted for given equations. for this example. Taking the average steel joint exceedence and the average acrylic joint exceedence for all equations.90 Acrylic 0. comparisons were made between the steel and acrylic joint SCF databases. and for all loadcases.1 5.87 Steel 1.13 Acrylic 1.2 13.5 9. with the Efthymiou equation.17 Steel 0.3 64. The SCFs using the Efthymiou equation exceed the steel joint database SCFs by an average of 7% and the acrylic joint database SCFs by an average of 17% for this particular joint type and loadcase.During the assessment.0 7.range of applicability) Average Predicted Steel/ SCF Acrylic Recorded SCF Steel 1.09 Acrylic 1.2. and that these geometries are not evenly distributed across the range of parameters influencing the level of stress concentration in the joint.15 Steel 1.3 19.1 24.1 13. 29 .

factors have to be employed to account for the lack of a weld fillet. (ii) Use the mean and standard deviation of the data to fit probability distributions to the data. which on average balance out. The refined database.2 CRITERIA FOR ASSESSMENT OF PARAMETRIC EQUATIONS In describing the criteria used in this project. the LR SCF derivation database described in Section 3 was accepted for assessment of SCFs with only minor modification.In addition. However. Ultimately option (i) was favoured for the more direct comparison between the measured data values and the parametric equations because it was difficult to identify specific probability distributions using the limited data available. (iii) Acrylic data generally gives good coverage of parametric ranges for most joint configurations. however these generally fell into 2 forms: (i) Compare the equations to the database to see if predetermined levels of underprediction/overprediction are met. Global comparisons of the acrylic database and steel databases do show differences in the percentage of joints underpredicting.3 Simple joint SCF assessment database Joint type T/Y X K Total Steel ß<1 85 24 34 143 ß=1 26 8 1 35 All ß 111 32 35 178 ß<1 64 19 29 112 Acrylic ß=1 16 13 3 32 All ß 80 32 32 144 ß<1 149 43 63 255 Total ß=1 42 21 4 67 All ß 191 64 67 322 5. The total numbers of steel and acrylic joint specimens in the database are given in Table 5. The main differences were the inclusion of some steel joints tested by UCL and the exclusion of some steel joints tested by Wimpey Offshore which were found to have strain gauges located within the notch zone. 30 . thus reducing the significance of individual points that do not fit the norm. however. steel data has intended to be based on joints with relatively short chord lengths and is more prone to the effects of the end diaphragms. In conclusion. is given in Appendix B. some differences have not yet been fully understood and explained. titled the SCF assessment database. it should be noted that many attempts to find satisfactory criteria were made. and reject or flag equations that fail to meet these requirements.3. Then compare the equations to the probability distributions to see if predetermined levels of underprediction/ overprediction are met. (iv) Acrylic joints gives correlation with steel joints in the few cases where identical specimens have been tested. These differences are generally explained by different joint configurations held in the databases. Table 5.

0 [ 25%) and if percentage SCFs considerably underpredicting [ 5% (%P/R < 0. It was felt that it would be useful to know when an equation was generally overconservative. however. however an equation should not be considered unacceptable because it is overconservative. If the number of steel joint SCFs in the database < 20. If. After considerable investigation the following criteria were felt to give the best guideline and have been employed in this study: Ÿ If the number of steel joint SCFs in the database [ 20 then assess the steel database alone. If the number of pooled joint SCFs in the database < 15. in addition. ie % of joints where P/R < 1 where P = predicted SCF and R = recorded SCF.5%) then the equation is regarded as borderline and engineering judgement has been used. then note that the equation is generally conservative. if percentage SCFs underpredicting [ 25% (%P/R < 1. these values do meet the objectives of the project. For the given dataset. the following aspects were felt to be the most significant: (i) The percentage of joints underpredicted by a given equation. (iv) Steel joints should be given priority where sufficient data exists. then the equation cannot be assessed. If the acceptance criteria as used in this study is nearly met (ie 25% < [%P/R < 1.5 m 50%).5 was chosen as a guideline). the percentage SCFs considerably overpredicting m 50%) (%P/R > 1. (ii) The percentage of joints that are significantly underpredicted by a given equation (for this purpose percentage of joints where P/R < 0.8 was chosen as a guideline). and generally support the equations that are currently in use while identifying some significant anomalies in the parametric equations which should be avoided. and the number of pooled joint SCFs (steel joints and acrylic joints) m 15 then assess the pooled database.For determining the performance of an SCF equation against an SCF database.0] [ 30% and/or 5% < {%P/R < 0. 31 . Ÿ Ÿ It is obvious that the values taken in this assessment are subjective. (iii) The percentage of joints that are significantly overpredicted by a given equation (for this purpose percentage of joints where P/R > 1.8] [ 7.8 [ 5%) then the equation meets the acceptance criteria used in this study.

5. since b = 1 joints generally give very variable results primarily due to the degree of weld cut-back employed. For the acceptance criteria used.8. It was concluded that no single a cut-off value could account for the effect of the chord end diaphragms. %P/R < 1. (iv) Only joints that have all parameters inside the equation’s recommended range of applicability were included in the analysis.6. the standard deviation of the variable P/R. (ii) All joints were assumed to be pinned at the chord ends.22 the performance of each equation is described for joint geometries (T/Y. This action had the effect of considerably reducing the number of joints available.4 PERFORMANCE OF PARAMETRIC EQUATIONS In Tables 1 . the Efthymiou and new LR equations are the most consistent giving the best performance overall. chord crown.5 were excluded from the assessment. loadcases (axial. % P/R < 0. X and K). particularly for the Kuang equations.4 to 5.5 DATA USED IN SCF ASSESSMENT Stress concentration factor matrices for T/Y. OPB and IPB) and measuring positions (chord saddle. (vi) The variable P/R (= Predicted SCF/Recorded SCF) was analysed and found to be approximately lognormally distributed. There was insufficient data available to produce a matrix for KT joints. the short chord correction factors proposed by Efthymiou were employed. It should be noted that the HCD equations only cover T/Y joints and the Kuang equations T/Y and K joints. If an SCF were influenced by more than 10% due to the end conditions (ie SCCF < 0.9) then the SCF was excluded from the assessment.3 ASSESSMENT OF PARAMETRIC EQUATIONS In assessing the parametric equations the following methodology was adopted: (i) To assess the effect of chord length on saddle SCFs for axial loading and OPB a considerable amount of work was performed.5. X and K joints have been produced in this project. Consequently. 32 . (v) Joints with SCFs less than 1.5 are assessed for each combination of configuration/loadcase/measuring position/equation. 5. b = 1 joints were excluded from analyses at the saddle locations.0 and the %P/R > 1. brace saddle and brace crown). The mean P/R value. see Tables 5. (iii) Initially.

5 X Joints Loading Balanced axial Position Chord Saddle Chord Crown Brace Saddle Brace Crown Chordside Braceside Chordside Braceside Words (25 P) (12 P) (16 P) (16 P) (23 P) (16 P) (33 P) (27 P) UEG (27 P) (12 P) (16 P) (16 P) (23 P) (16 P) (33 P) (27 P) Efthy (32 P) (12 P) (19 P) (16 P) (23 P) (16 P) (36 P) (29 P) LR (32 P) (12 P) (18 P) (16 P) (23 P) (16 P) (36 P) (28 P) Balanced OPB Balanced IPB Table 5.4 T/Y Joints Loading Axial Position Chord Saddle Chord Crown Brace Saddle Brace Crown Chordside Braceside Chordside Braceside Kuang (34 P) N/A (25 P) N/A (18 P) (13 P) (19 P) (19 P) Words (22 S) (43 P) (46 P) (30 P) (70 P) (44 P) (61 P) (44 P) UEG (28 S) (43 P) (47 P) (30 P) (74 P) (47 P) (21 S) (24 S) Efthy (28 S) (48 P) (47 P) (35 P) (73 P) (48 P) (21 S) (24 S) HCD (28 S) (29 P) (42 P) (19 P) (65 P) (42 P) (44 P) (35 P) LR (28 S) (48 P) (48 P) (35 P) (75 P) (48 P) (23 S) (26 S) OPB IPB Table 5.2: () S P A N/A Number of joints in the database and SCF database used. The equation has been assessed against the acrylic joint database. The equation has been assessed against the steel joint database alone (ie nsteel>20) The equation has been assessed against the ‘pooled’ steel and acrylic joint database. There is no parametric equation for this loadcase. 33 .Table 5.6 K Joints Position Chordside Balanced Axial Braceside Chordside Unbalanced OPB Braceside Chordside Balanced IPB Braceside Loading Kuang (16 P) (13 P) N/A N/A (6 A) (7 A) Words (54 P) (46 P) (50 P) (39 P) (32 A) (34 A) UEG (54 P) (46 P) (50 P) (39 P) (32 A) (34 A) Efthy (25 S) (23 S) (50 P) (39 P) (32 A) (34 A) LR (24 S) (21 S) (50 P) (39 P) (32 A) (34 A) Key to SCF TABLES: Basis of the assessment in accordance with the criteria given in Section 5.

conversion from SNCF to SCF and weld fillet effects were very different from those for b ! 1. the minimum SCF from these acceptable equations could be taken. have been described. titled the SCF assessment database and given in Appendix B. The factors were also very inconsistent within the b = 1.0 dataset itself. These factors should be applied to those equations which meet the criteria in Tables 1 to 22 for b = 1.0 joints.0 X joints there were significant differences in quoted SCFs resulting from linear and non-linear extrapolation. see Appendix A. For b = 1. However. Firstly. In the assessment of existing SCF parametric equations. The assessment criteria was agreed by the Review Panel for Fatigue Guidance and the performance of the equations is presented in Tables 1 to 22. used in this study. For T/Y joints the b modification factors are given in Table 23 and for X joints in Table 24. The database was screened and for the assessment all joints failing to meet the specified criteria were excluded. with some justification.0 joints to meet the criteria used in this study. 6. contains 191 T/Y. at the saddle location. 2 studies largely funded by the HSE.0) were excluded from the main assessment at the saddle location.0 JOINTS Joints with equal brace and chord diameters (ie b = 1.0. No modification factors can be derived for the K joints because of the lack of suitable data (less than 10 joints of this type). In the second study the database was refined and used to assess existing simple joint SCF parametric forumulae including the new LR equations.0 joints the SCF is very significantly influenced by the degree of cut-back. b = 1. A number of b modification factors have been derived which should be used in conjunction with equations which meet the acceptance criteria. the objective was to produce SCF matrices giving recommendations regarding the use of the equations for new HSE fatigue guidance. The finalised database. that since a number of SCF equations are acceptable for a given loadcase. 34 . CONCLUSIONS In this report.0 acrylic models the factors applied with respect to extrapolation.0 range can be applied to b = 1. the allowable equations should be treated with caution for 2 reasons. In determining which SCF equations should be used.6 BETA = 1. From this database a new set of SCF parametric equations was derived from simple tubular nodal joints. These factors have been selected to ensure that the equations accepted for the b = 1. In the JISSP study for b = 1. the designer may feel. A one standard deviation safety factor was included to give the design equations. 64 X and 67 K joints.5. These equations are referred to as the new LR equations and were derived as a mean fit to the database. The first study led to the creation of a comprehensive database of steel and acrylic joint SCFs titled the Lloyd’s Register SCF derivation database and described in Section 3. see paragraph 2.0 joints. At the saddle location for b = 1.2.0 joints have been analysed separately to account for the wide variation in the degree of cut-back at the saddle.5.

‘mixing’ equations to give one allowable set of equations may lead to problems if in future the influence function approach is adopted rather than the current rather vague joint classification approach. 1985 (9) EFTHYMIOU M Development of SCF formulae and generalised influence functions for use in fatigue analysis Offshore Tubular Joints. Surrey 1988 (10) DEPARTMENT OF ENERGY 35 . 1975 (5) KUANG J G. POTVIN A B and LEICK R D Stress concentration in tubular joints OTC 2205.Secondly. T. Delft. ICE. Dr J Sharp of the HSE and N Nichols of MaTSU for their valuable contributions during this project. Houston. Houston. November 1978 (7) WORDSWORTH A C Stress concentration factors at K and KT tubular joints Fatigue in Offshore Structural Steels Conference. February 1981 (8) EFTHYMIOU M and DURKIN S Stress concentrations in T/Y and gap/overlap K joints Behaviour of Offshore Steel Structures. Texas. LEICK R D and KAHLICK J L Stress concentration in tubular joints Society of Petroleum Engineers Journal. and to the HSE for providing the major funding for the work described in this report. Taxas. Houston (1974) (4) KUANG J G. 1967 (2) REBER J B Ultimate strength design of tubular joints OTC 1664. Y and K welded tubular connections Welding Research Council 125. ACKNOWLEDGEMENTS The authors wish to thank their colleagues at Lloyd’s Register. August 1977 (6) WORDSWORTH A C and SMEDLEY G P Stress concentrations at unstiffened tubular joints European Offshore Steels Research Seminar. REFERENCES (1) TOPRAC A A and BEALE L A Analysis of in-plane. The influence function approach relies on an understanding of the effect of the presence and loading of individual braces and the change in SCF that results. POTVIN A B. 1972 (3) VISSER W On the structural design of tubular joints” OTC 2117.

an overview of recent work to revise Department of Energy Guidance Integrity of Offshore Structures . 1986 (20) WARDENIER J Hollow section joints Delft University Press.United Kingdom Offshore Research Project . 4th Edition. 1990 (13) REYNOLDS A and SHARP J The fatigue performance of tubular joints .1. 1990 (14) DEPARTMENT OF ENERGY Background to new fatigue design guidance for steel welded joints in offshore structures HMSO 1984 (15) DEPARTMENT OF ENERGY United Kingdom Offshore Research Project . 1982 (21) LALANI M. May 1986 36 .Final Report HMSO OTH-88-282 (16) CLAYTON A M and IRVINE N M Stress analysis method for tubular connections European Offshore Steels Research Seminar. Glasgow 1990 (12) DEPARTMENT OF ENERGY Offshore installations: Guidance on design. construction and certification”. Houston. OTH-87-265. 1979. HOADLEY P W. TEBBETT I E and CHOO B S Improved fatigue life estimation of tubular joints Offshore Technology Conference (OTC 5306).189 (19) SWENSSON K D. YURA J A and SANDERS D H Stress concentration factors in Double-Tee joints Phil M Ferguson Structural Engineering Laboratory Report 86 . pp 176 . Cambridge University Press.4. Glasgow.Phase I (UKOSRP I) . 1987 (11) SMEDLEY P A and FISHER P J A review of stress concentration factors for tubular complex joints Integrity of Offshore Structures-4. HMSO. IMechE. November 1978 (17) ELLIOTT K S and FESSLER H Stress at weld toes in non-overlapped tubular joints Fatigue and crack growth in offshore structures.Phase II (UKOSRP II) Summary Report HMSO. London 1986 (18) GURNEY T R Fatigue of welded structures 2nd Ed.

April 1985 (23) DEPARTMENT OF ENERGY Investigation into the differences between the measured hot-spot stress when derived by either linear or non-linear extrapolation techniques Prepared by Lloyd’s Register for the Den. Houston. May 1980 (31) HELLIER A K. September 1984 (33) SMEDLEY P A and FISHER P J Stress concentration factors for simple tubular joints The First International Offshore and Polar Engineering Conference.Interim report 1 Report No WOL 215/84A. GERALD J P and PUTOT C J Fatigue life prediction of tubular nodes Offshore Technology Conference (OTC 3699). CONNOLLY M P and DOVER W D Stress concentration factors for tubular Y .joints Int J Fatigue 12 No 1 (1990) pp 13 . February 1988.Stress concentration factors for tubular complex joints . UK. Houston.23 (32) WIMPEY OFFSHORE Lloyd’s Register of Shipping . September 1977 (25) MARSHALL P W A review of stress concentration factors in tubular joints CE-32 report. 1991 37 . Texas (26) WORDSWORTH A C Aspects of stress concentration factors at tubular joints Steels in Marine Structures (SIMS) Conference.(22) UNDERWATER ENGINEERING GROUP Design guidance on tubular joints in steel offshore structures UR33. Houston.Stress concentration factors in tubular complex joints Shell report RKER.Finite element analysis in tasks 4A and 4B . Shell Oil Co. May 1983 (30) LIEURADE H-P.and T . August 1 1991 (28) EFTHYMIOU M Finite element analysis for Task 4A of Lloyd’s project . 1987 (27) API RECOMMENDED PRACTICE 2A (RP 2A) Recommended practice for planning. (ISOPE). Rev 2 (24) SMEDLEY G P Peak strains at tubular joints I Mech E.84.151 sponsored by LRS August 1984 (29) WYLDE J G Fatigue tests on welded tubular T-joints with equal brace and chord diameters Offshore Technology Conference (OTC 4527). Edinburgh. designing and constructing fixed offshore platforms Nineteenth edition.

TABLES 38 .

3% 8.5 22.8% 0.5% 8.0% 0.19 32.0% 3.0% 9.2% 33.0% 35.6% 0.7% 0.0% 20.0 >1.4% 1.6% 1.3% 1.9% 4.0% 8.31 27.8% 1.0% 40.6% 0.5 28.13 13.8 <1.2% 22.6% 5.8% 0.9% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.3% 0.9% 4.0% 7.15 19.05 23.40 20.9% 1.3% 2.4% 12.1% 0.5% 1.0% 0.0% 26.0% 7.0% 0.6% Steel Acrylic Pooled Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled HCD Steel Acrylic Pooled LR Steel Acrylic Pooled Table 2 T/Y joints .6% 0.09 9.0% 0.2% 0.13 20.0% 0.0% 15.0% 12.(C = 1.1% 1.Table 1 T/Y joints .07 19.6% 1.0% 11.4% 1.0% 0.0% 0.Axial .07 10.0% 2.21 17.1% 0.6% 64.0% 4.05 23.2% 33.7% 1.12 25.9.0% 0.3% 5.0% 0.1% 10.12 9.89 18.1% 1.1% 25.3% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.1% 1.96 14.9% 0.4% 1.0% 0.0% 50.12 20.3% 0.5% 0.8% 0.19 23.1% 1.2% 0.7% 1.0% 0.3% Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled HCD Steel Acrylic Pooled LR Steel Acrylic Pooled 39 .16 13.15 12.6% 2.9.0% 10.4% 0.0% 0.Chord saddle .0 >1.0% 0.75% 1.0% 22.1% 0.8 <1.0% 4.8% 0.3% 1.18 17.0% 7.3% 0.(b<1.90 16.13 9.7% 1.4% 41.5% 6.8% 1.88 19.42 19.1% 0. SCCFm0.0% 0.3% 22.21 25.0% 25.0% 0.0% 9. SCFm1.1% 1.Chord Crown .0% 70.0% 11.5) Equation Kuang Steel/ acrylic No of Pts 14 20 34 22 58 80 28 58 86 28 57 85 28 50 78 28 59 87 Database %st dev of Mean Equn 0.8% 1.16 15.0% 1.3% 2.0% 13.0% 17.19 19.3% 0.0% 0.14 19.12 12.Axial .6% 1.0% 12.0% 28.2% 0.5% 67.8% 1.0% 0.0% 16.6% 2.9% 7.3% 0.0% 24.3% 0. SCCFm0.09 19.7% 23.0% 20.0% 17.15 21.0% 24.0% 8.2% 1.1% 33.1% 22.2% 1.5) Equation Steel/ acrylic No of Pts 9 34 43 9 34 43 9 39 48 4 25 29 9 39 48 Database %st dev of Mean Equn 1.7% 23.20 29.6% 1. SCFm1.9% 1.

**Table 3 T/Y joints - Axial - Brace Saddle - (b<1; SCCFm0.9; SCFm1.5)
**

Steel/ acrylic No of Pts 5 20 25 7 39 46 8 39 47 8 39 47 8 34 42 8 40 48 Database Mean %st dev of Equn 1.41 29.0% 1.32 46.7% 1.34 43.4% 1.09 26.0% 1.13 22.7% 1.12 23.0% 1.12 25.6% 1.16 23.0% 1.15 23.2% 1.29 25.1% 1.14 24.0% 1.17 24.6% 1.58 44.3% 1.38 34.1% 1.42 36.5% 1.23 20.5% 1.17 21.0% 1.18 20.7% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.8 <1.0 >1.5 0.0% 0.0% 40.0% 20.0% 30.0% 40.0% 16.0% 24.0% 40.0% 14.3% 42.9% 0.0% 5.1% 41.0% 5.1% 6.5% 41.3% 4.3% 12.5% 37.5% 0.0% 2.6% 41.0% 7.7% 4.3% 40.4% 6.4% 0.0% 12.5% 12.5% 5.1% 25.6% 7.7% 4.3% 23.4% 8.5% 0.0% 0.0% 50.0% 5.9% 11.8% 35.3% 4.8% 9.6% 38.1% 0.0% 25.0% 12.5% 0.0% 20.0% 12.5% 0.0% 20.8% 12.5%

Equation Kuang

Steel Acrylic Pooled Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled HCD Steel Acrylic Pooled LR Steel Acrylic Pooled

40

**Table 4 T/Y joints - Axial - Brace Crown - (C = 1; SCCFm0.9; SCFm1.5)
**

Steel/ acrylic No of Pts 4 26 30 4 26 30 4 31 35 1 18 19 4 31 35 Database Mean %st dev of Equn 1.92 48.2% 2.38 68.7% 2.32 67.5% 1.92 48.2% 2.38 68.3% 2.32 67.2% 1.55 19.6% 1.62 34.2% 1.61 32.7% 1.54 1.86 62.1% 1.84 60.8% 0.98 16.8% 1.29 24.9% 1.25 25.9% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.8 <1.0 >1.5 0.0% 0.0% 75.0% 0.0% 0.0% 92.3% 0.0% 0.0% 90.0% 0.0% 0.0% 75.0% 0.0% 0.0% 92.3% 0.0% 0.0% 90.0% 0.0% 0.0% 75.0% 0.0% 3.2% 64.5% 0.0% 2.9% 65.7% 0.0% 0.0% 100% 0.0% 5.6% 66.7% 0.0% 5.3% 68.4% 25.0% 50.0% 0.0% 0.0% 19.4% 19.4% 2.9% 22.9% 17.1%

Equation

Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled HCD Steel Acrylic Pooled LR Steel Acrylic Pooled

**Table 5 T/Y joints - OPB - Chordside - (b<1; SCCFm0.9; SCFm1.5)
**

Equation Kuang Steel/ acrylic No of Pts 6 12 18 14 56 70 18 56 74 18 55 73 16 49 65 18 57 75 Database %st dev of Mean Equn 0.87 8.0% 0.91 7.5% 0.90 7.7% 1.05 14.2% 1.13 17.1% 1.11 16.8% 1.11 18.7% 1.16 17.3% 1.15 17.7% 1.10 13.4% 1.11 15.5% 1.11 14.9% 1.21 13.9% 1.23 24.7% 1.23 22.4% 1.16 16.7% 1.19 17.9% 1.18 17.6% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.8 <1.0 >1.5 33.3% 100% 0.0% 16.7% 100% 0.0% 22.2% 100% 0.0% 7.1% 28.6% 0.0% 0.0% 21.4% 5.4% 1.4% 22.8% 4.3% 5.6% 22.2% 0.0% 0.0% 10.7% 7.1% 1.4% 13.5% 5.4% 0.0% 22.2% 0.0% 0.0% 20.0% 1.8% 0.0% 20.5% 1.4% 0.0% 6.3% 0.0% 2.0% 18.4% 18.4% 1.5% 15.4% 13.8% 0.0% 16.7% 5.6% 0.0% 8.8% 8.8% 0.0% 10.7% 8.0%

Steel Acrylic Pooled Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled HCD Steel Acrylic Pooled LR Steel Acrylic Pooled

41

**Table 6 T/Y joints - OPB - Braceside - (b<1; SCCFm0.9; SCFm1.5)
**

Equation Kuang Steel/ acrylic No of Pts 1 12 13 6 38 44 9 38 47 9 39 48 7 35 42 9 39 48 Database %st dev of Mean Equn 1.46 1.30 16.5% 1.31 16.4% 1.17 7.6% 1.19 26.7% 1.19 24.9% 1.57 91.9% 1.22 27.5% 1.29 47.7% 1.54 36.3% 1.17 20.4% 1.24 27.8% 1.54 12.1% 1.39 21.4% 1.42 20.8% 1.52 40.6% 1.17 18.6% 1.24 27.4% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.8 <1.0 >1.5 0.0% 0.0% 0.0% 0.0% 0.0% 16.7% 0.0% 0.0% 15.4% 0.0% 0.0% 15.4% 0.0% 28.9% 18.4% 0.0% 25.0% 15.9% 0.0% 0.0% 33.3% 0.0% 23.7% 18.4% 0.0% 19.2% 21.3% 0.0% 0.0% 44.4% 0.0% 17.9% 7.7% 0.0% 14.6% 14.6% 0.0% 0.0% 71.4% 0.0% 0.0% 31.4% 0.0% 0.0% 38.1% 0.0% 0.0% 33.3% 0.0% 12.8% 7.7% 0.0% 10.4% 12.5%

Steel Acrylic Pooled Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled HCD Steel Acrylic Pooled LR Steel Acrylic Pooled

42

0% 31.0% 25.3% 30.1% 4.0% 5.0% 16.7% 0.27 15.55 29.0% 1.60 14.15 14.36 29.9% 0.0% 0.0% 36.52 29.4% 0.4% 0.0% 71.8% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.0% 45.0% 0.0% 0.6% 5.5) Equation Kuang Steel/ acrylic No of Pts 3 16 19 6 55 61 21 58 79 21 60 81 6 38 44 23 63 86 Database %st dev of Mean Equn 0.3% 0.2% 1.0% 52.06 17.0% 0.6% 0.5% 1.0 >1.19 20.51 23.0% 14.25 17.13 13.8% 1.37 29.0% 0.0% 0.6% 1.8% 0.0% 1.2% 1.0% 0.54 22.8% 1.0% 50.1% 4.2% 1.0% 25.IPB .17 13.2% 17.5% 1.3% 4.0% 1.0% 66.2% 0.4% 0.39 19.Table 7 T/Y joints .7% 1.0% 31.0% 0.0 >1.3% 32.20 20.8 <1.2% 0.2% 1.Braceside .0% 42.7% 0.3% 3.03 10.0% 0.15 18.8% 0.5 0.0% 9.0% 10.8% 42.6% 0.09 16.0% 5.3% 1.5% 0.7% 0.IPB .9% 0.07 14.3% 29.0% 1.4% 1.5% 1.0% 7.9% 1.8% 1.(SCFm1.0% 6.2% 5.2% 1.5% 0.0% 0.3% 0.1% 1.9% 1.7% 1.5% 3.2% 44.0% 0.6% 1.5% 1.7% 4.4% 1.0% 66.2% 1.2% 1.7% 0.2% 0.3% Steel Acrylic Pooled Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled HCD Steel Acrylic Pooled LR Steel Acrylic Pooled 43 .7% 1.18 19.0% Steel Acrylic Pooled Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled HCD Steel Acrylic Pooled LR Steel Acrylic Pooled Table 8 T/Y joints .6% 5.8% 1.25 16.0% 2.22 19.5% 22.0% 7.3% 0.0% 16.0% 8.6% 4.0% 6.4% 0.0% 0.7% 1.0% 0.0% 12.8% 0.0% 0.(SCFm1.0% 0.52 22.8% 28.02 10.0% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.9% 0.55 28.71 26.5) Equation Kuang Steel/ acrylic No of Pts 3 16 19 8 36 44 24 39 63 24 43 67 7 28 35 26 44 70 Database %st dev of Mean Equn 1.0% 53.4% 4.4% 21.5% 5.97 13.5 0.0% 0.4% 1.05 17.0% 0.8 <1.8% 0.6% 0.0% 0.75 26.Chordside .26 16.0% 0.2% 1.15 12.5% 1.4% 1.48 29.0% 3.3% 4.0% 66.7% 0.0% 0.0% 50.0% 78.0% 0.2% 14.0% 0.3% 1.33 21.16 18.0% 1.21 16.

36 32.3% 1. SCFm1.Chord Saddle .14 6.5% 0.3% Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel Acrylic Pooled Table 10 X joints .0% 0.0% 0.3% 0.0% 25.8 <1.0% 12.1% 44.5 0.1% 44.7% 1.36 18.27 28.0% 0.09 15.0% 1.7% 1.0% 0.3% 12.3% 0.0% 0.3% 33.9% 1.12 6.0% 1.5 0.0% 4.8 <1.6% 1.8% 1.29 27.3% 1.0% 8.2% 11.(C = 1.7% 8.0% 0.0% 3.4% 1.1% 0.0% 0.0% 6.23 18.0% 8.14 24.3% 33.34 18.21 24.3% 0.0% 11.7% 0.2% 6.0% 0.13 6.0% 8.0% 0.(b<1. SCFm1.5 0.0% 0.3% 18.10 4.6% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.Table 9 X joints . SCFm1.0 >1.0% 16.0% 22.5% 0.8% 11.26 28.5) Steel/ acrylic Steel Acrylic Pooled Steel Acrylic Pooled No of Pts 4 12 16 4 12 16 Database Mean %st dev of Equn 1.0% 33.29 21.(b<1.3% 33.8% 1.12 38.9% 1.7% 1.15 18.7% 0.4% 0.3% 1.2% 12.5) Equation Steel/ acrylic No of Pts 3 9 12 3 9 12 3 9 12 3 9 12 Database %st dev of Mean Equn 1.1% 1.8% Equation Wordsworth & Smedley UEG 44 .0% 0.0 >1.7% 1.5% 0.1% 1.0% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.0% 0.Axial .0% 0.8% 1.0% 0.19 21.14 20.3% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.0% 11.9% 1.1% 0.0% 0.3% 16.9% 11.0% 0.3% 16.23 24.0% 12.07 5.0% 17.3% 22.2% 1.0% 0.0% 0.0% 28.6% 0.0% 0.23 22.24 20.9% 1.15 21.0% 33.26 22.6% 1.8 <1.Axial .2% 33.19 18.15 44.0% 16.5) Equation Steel/ acrylic No of Pts 8 17 25 10 17 27 16 16 32 15 17 32 Database %st dev of Mean Equn 1.3% 33.8% 0.0% 0.7% 1.0 >1.9% 1.3% Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel Acrylic Pooled Table 11 X joints .8% 0.0% 6.7% 0.0% 11.17 19.0% 46.0% 0.0% 0.7% 8.0% 5.2% 1.5% 0.5% 12.0% 1.0% 11.3% 0.4% 0.0% 0.3% 1.Axial .0% 8.04 5.3% 33.Chord Crown .Brace Saddle .15 25.0% 25.

0 >1.4% 1.6% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.6% 4.0% 14.0% 0.0% 16.7% 0.45 40.(C = 1.15 12.20 17.14 14.8% 4.0% 7.23 33.0% 6.8% 4.20 30.0% 0.30 40.7% 7.28 37.7% 0.0% 5.0% 1.(b<1. SCFm1.4% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.7% 16.0% 0.0% 1.0% 16.38 35.7% 16.7% 0.5) Equation Steel/ acrylic No of Pts 3 13 16 3 13 16 3 13 16 3 13 16 Database Mean %st dev of Equn 2.7% 1.25 34.3% 1.0% 16.12 15.3% 43.0% Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel Acrylic Pooled Table 13 X joints .51 44.7% 13.12 12.6% 1.0% 0.7% 38.5) Equation Steel/ acrylic No of Pts 6 17 23 6 17 23 6 16 23 6 17 23 Database %st dev of Mean Equn 1.0% 1.Equation Efthymiou Steel/ acrylic Steel Acrylic Pooled Steel Acrylic Pooled No of Pts 7 12 19 6 12 18 LR Database Mean %st dev of Equn 1.8% 0.0% 7.7% Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel Acrylic Pooled 45 .8 <1.0% 18.8 <1.52 44.3% 0.0% 25.0% 6.6% 11.4% 1.6% Table 12 X joints .3% 0.0% 0.32 17.2% 1.3% 0.7% 16.45 36.2% 6.3% 17.3% 26.0% 7.0% 0.0% 16.0% 0.7% 1.3% 21.3% 1.0% 1.5 0.8% 1.3% 13.0% 0.5% 0.0% 0.2% 1.0% 11.8 <1.1% 1.1% 1.4% 8.0% 100% 0.0% 0.7% 38.8% 17.5 0.7% 0.3% Pred SCF/Recorded SCF %P/R %P/R % P/R <0.5 16.22 13.4% 1.Axial .37 35.1% 2.OPB .0% 6.23 16.32 20.3% 7.0 >1.0% 8.7% 1.7% 16.47 9.0% 33.5% 1.9% 0.13 12.3% 6.0% 0.35 44.0% 0.0% 13.0% 0.3% 1.8% 1.7% 46.9% 5.7% 5. SCFm1.13 20.0% 25.5% 0.4% 1.0 >1.Chordside .0% 0.29 7.2% 1.0% 8.0% 0.1% 1.3% 50.7% 16.16 22.0% 17.3% 50.0% 0.7% 0.06 13.3% 33.0% 5.5% 1.3% 0.0% 0.15 13.16 21.25 17.Brace Crown .0% 1.7% 8.7% 0.0% 100% 0.

60 31.0% 0.0% 44.5 0.0% 6.57 10.3% 1.3% 1.0% 0.7% 1.35 28.1% 11.0% 63.0% 0.5) Equation Wordsworth & Smedley UEG Steel/ acrylic Steel Acrylic Pooled Steel Acrylic Pooled Steel Acrylic Pooled Steel Acrylic Pooled No of Pts 9 24 33 9 24 33 12 24 36 12 24 36 Database Mean %st dev of Equn 1.0% 6.5 0.0% 1.0% 60.33 42.5% 37.(SCFm1.8 <1.2% 0.32 39.7% 33.3% 0.0% 75.35 34.9% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.5% 43.2% 1.(SCFm1.0% 8.3% 0.0% 12.7% 0.Table 14 X joints .0% 75.3% 0.26 23.4% 1.1% 8.59 30.0% 0.33 37.0% 0.1% 1.5% 0.Braceside .9% 1.43 40.8 <1.0% 0.0% 12.0% Wordsworth Steel & Smedley Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel Acrylic Pooled Table 15 X joints .1% Efthymiou LR Table 16 X joints .28 28.5) Equation Wordsworth & Smedley UEG Steel/ acrylic Steel Acrylic Pooled Steel Acrylic Pooled No of Pts 4 23 27 4 23 27 Database Mean %st dev of Equn 1.0% 8.0% 0.3% 0.51 20.0% 44.5 0.7% 1.0% 16.0% 0.0% 12.0% 16.54 68.9% 1.0% 12.OPB .3% 25.0% 0.0% 16.1% 18.0% 0.0 >1.41 34.6% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.0% 50.0% 0.0% 16.0% 1.9% 0.0% 16.59 31.23 17.0 >1.9% 1.8 <1.5) Equation Steel/ acrylic No of Pts 4 12 16 4 12 16 4 12 16 4 12 16 Database %st dev of Mean Equn 1.0% 0.56 73.3% 0.4% 0.1% 1.7% 0.IPB .Braceside .51 20.0% 6.7% 1.7% 1.5% 25.0% 46 .0% 75.7% 33.0% 0.3% 8.0% 0.0% 8.0% 0.0% 0.7% 1.3% 0.0% 63.0% 60.3% 8.7% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.7% 8.5% 0.IPB .25 17. SCFm1.8% 1.2% 1.0% 0.47 37.9% 1.5% 31.3% 0.39 42.0% 0.7% 1.2% 1.2% 0.55 12.17 15.3% 0.0% 1.9% 0.24 18.0% 11.0% 75.1% 1.5% 1.2% 1.1% 18.3% 8.57 31.Chordside .0% 1.4% 0.31 27.0% 8.0% 0.0% 0.0% 75.0% 0.22 25.21 19.0% 12.50 21.0 >1.8% 0.7% 16.9% 1.0% 0.(b<1.52 18.

3% 10.0% 4.0 >1.7% 0.Braceside.29 22.30 36.0% 0.Bal axial load .2% 1.0% 6.0% 0.39 28.3% 0.0% 12.8% 10.5% 14.5% 36.14 20.0% 0.87 19.6% 5.0% 0.4% 1.0% 32.08 28.0% 8.6% 1.8 <1.0% 4.7% 1.25 16.15 24.8% 1.2% 0.6% 10.5% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.0% 0.6% 2.7% 8.3% 1.0 >1.(b<1.3% 1.(b<1.0% 2.4% 22.3% 3.8 <1.10 27.2% 20.41 43.8% 17.1% 0.4% 1.0% 0.4% 1.3% 17.7% 0.0% 43.9% 11.0% 16.0% 1.19 22.32 37.0% 5.0% 13.5% 25.0% 0.5) Equation Kuang Steel/ acrylic No of Pts 8 8 16 19 35 54 19 35 54 25 35 60 24 35 59 Database Mean %st dev of Equn 0. SCCF[0.5% 31.9% 1.0% 9.31 32.0% 45.3% 3.1% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.22 23.8% 1.6% 25.9% 17.33 25.39 44.8% 40.7% 1.0 >1.7% 5.0% 1.5% 0.6% 1.3% 33.34 37. SCCF[0.0% 20.7% 1.3% 33.4% 1.4% 29.0% 0.3% 1.5% 17.4% 0.15 11.8% 35.0% 0.0% 33.8% 1.Bal axial load .4% 3.0% 7.0% Table 17 K joints .5 50.9% 1.16 24.Equation Efthymiou Steel/ acrylic Steel Acrylic Pooled Steel Acrylic Pooled No of Pts 6 23 29 5 23 28 LR Database Mean %st dev of Equn 1.7% 0.3% 1.1% 42.3% 0.1% 37.8 <1. SCFm1.0% 1.5% 17.5% 1.2% 19.48 24.1% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.7% 14.2% 1.1% 8.3% 1.9.30 22.14 25.2% 19. SCFm1.7% 16.0% 20.7% 0.Chordside .3% Steel Acrylic Pooled Wordsworth Steel Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel 47 .2% 9.5 0.2% 19.06 26.20 41.4% 8.8% 1.31 33.0% 20.4% 10.4% 9.4% 8.5% 40.0% 87.9% 13.29 33.05 28.18 32.41 29.6% 1.6% 26.2% 19.8% 33.9% 5.0% 0.4% 1.8% 7.5 0.08 27.6% 14.7% 1.8% 6.0% 33.5) Equation Kuang Steel/ acrylic No of Pts 7 6 13 15 31 46 15 31 46 23 31 54 21 Database %st dev of Mean Equn 1.7% 34.15 12.0% 10.2% 17.7% Steel Acrylic Pooled Wordsworth Steel Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel Acrylic Pooled Table 18 K joints .7% 1.9.0% 0.15 12.0% 1.0% 2.33 10.4% 30.7% 1.36 20.0% 0.3% 8.

0 >1.0% 6.55 32.3% 1.3% 1.8% 1.0% 2.62 54.2% 35.6% 1. SCCF[0.0% 0.0% 0.08 22.0% 0.5) Equation Steel/ acrylic No of Pts 8 31 39 8 31 39 8 31 39 8 31 39 Database %st dev of Mean Equn 1.4% 18.2% 61.7% 5.0% Wordsworth Steel Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel Acrylic Pooled Table 20 K joints .8% 1.6% 0.0% 87.26 19.21 26.0% 0.29 37.0% 30.0% 2.9% 1.Unbal OPB .3% 0.0% 25.5 0.2% 1.4% 0.44 36.0% 2.5% 1.9.5) Equation Steel/ acrylic No of Pts 15 35 50 15 35 50 15 35 50 15 35 50 Database %st dev of Mean Equn 1.3% 0.0% 5.8 <1.30 24.0% 0. SCCF[0.7% 34.0% 12.9% 0.0% 0.1% 1.5% 58.7% 40.8% 1.0% 0.0% 3.0% 0.9% 11.7% 0.61 33.0% 0.0 >1.35 30.5% 50.7% 1.3% 19.0% 32.Chordside .0% 0.90 49.0% 12.8% 7.0% 33.4% 0.0% 0.9.0% 8.37 19.Braceside .0% 16.62 38.0% 1.5% 1.Equation Steel/ acrylic Acrylic Pooled No of Pts 31 52 Database %st dev of Mean Equn 1.5% 1.29 13.(b<1.8 <1.0% 17.9% 1.7% 26.95 44.8% 1.1% 0.26 15.0% 6.68 38.6% Table 19 K joints .5 0.39 25.7% 1.0% 2.0% Wordsworth Steel Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel Acrylic Pooled 48 .0% 6.18 16.1% 2.5% 0.0% 28.0% 15.5 0.7% 0. SCFm1.13 14.0% 8.5% 66.8% 1.2% 61.0% 18.0% 4.8% 1.0% 2.0% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.5% 3.1% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.9% 0.0 >1.21 14.6% 1.26 17.0% 75.16 23.0% 3.5% 1.9% 2.8 <1.0% 10.41 18.0% 0.31 17.2% 1. SCFm1.7% Pred SCF/Recorded SCF %P/R %P/R % P/R <0.0% 0.7% 0.(b<1.33 13.1% 9.7% 1.8% 25.7% 0.0% 50.Unbal OPB .6% 4.

0% 2.3% 1.27 26.6% 0.1% 1.(SCFm1.0% 12.3% 1.0% 2.77 43.27 26.5% 2.8% 1.0% 0.6% 0.Bal IPB .17 17.5% 0.9% 70.3% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.5% 12.8 <1.3% 28.26 26.0% 50.0% 2.1% 12.5) Equation Kuang Steel/ acrylic No of Pts 0 7 7 0 34 34 0 34 34 0 34 34 0 34 34 Database %st dev of Mean Equn 1.3% 1.9% 1.5% 2.9% 0.9% 5.Chordside .1% 1.5% 31.7% 1.3% 28.77 43.5% 12.0% 2.76 43.9% Steel Acrylic Pooled Wordsworth Steel Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel Acrylic Pooled 49 .0% 50.1% 12.5% 3.76 43.5% 12.57 31.9% 70.Bal IPB .0% 2.3% 0.9% 23.9% 52.5) Equation Kuang Steel/ acrylic No of Pts 0 6 6 0 32 32 0 32 32 0 32 32 0 32 32 Database %st dev of Mean Equn 0.07 20.3% 0.Table 21 K joints .9% 23.5% 1.0% 50.0% 50.0% 3.3% 0.9% 70.31 27.9% 1.5% 12.0% 2.8% 1.5 0.31 27.0% 0.7% 1.07 20.8 <1.5 0.3% Pred SCF/Recorded SCF % P/R % P/R % P/R <0.6% 6.3% 1.0% 14.9% 70.6% 6.6% 0.9% 5.17 17.Braceside .6% 0.5% 31.0% 14.6% 0.1% 12.0 >1.1% 12.3% 1.5% 3.10 18.0 >1.7% 1.3% Steel Acrylic Pooled Wordsworth Steel Acrylic Pooled UEG Steel Acrylic Pooled Efthymiou Steel Acrylic Pooled LR Steel Acrylic Pooled Table 22 K joints .3% 1.(SCFm1.0% 15.0% 15.5% 12.5% 3.94 8.6% 0.94 8.26 26.9% 52.57 31.7% 1.10 18.

98 0.95 0.98 0.98 0.Table 23 T/Y joints b = 1.65 (Y) ) Where: Y is the degree of cut-back at saddle. default value = 20º if cutback not known t = c = thickness of brace thickness of chord diameter of chord 2 X thickness of chord Table 24 X joints b = 1.0 0.98 LR Beta* mod 0.98 0.94 0.98 0.99 0.99 0.98 1.96 UEG Beta mod 0.96 UEG Beta mod 0.0 Loading Balanced Axial Balanced OPB Location No joints (P) Chordside (16) Braceside (16) Chordside (13) Braceside (11) Wordsworth/ Smedley Beta mod 0.0 0.( c sin 0.( c sin 0.98 Efthymiou Beta mod 1.98 0.99 *Note: for LR the Beta mod value quoted is that to be used in the LR equation t ie b = bmod .98 0.99 0.65 (Y) ) Where: Y is the degree of cut-back at saddle.94 0.98 0.98 LR Beta* mod 0.96 *Note: for LR the Beta mod value quoted is that to be used in the LR equation t ie b = bmod .97 0.98 0.98 Efthymiou Beta mod 0.94 0. default value = 20º if cutback not known t = c = thickness of brace thickness of chord diameter of chord 2 X thickness of chord 50 .98 0.99 0.97 0.0 Loading Axial OPB Location No joints (P) Chordside (9) Braceside (16) Chordside (25) Braceside (26) Wordsworth/ Smedley Beta mod 0.98 0.

APPENDIX A PARAMETRIC EQUATIONS USED IN THIS STUDY 51 .

1 A2.3 A2.INDEX A1 A1.3 A1.4 A1.5 A1.7 A2 A2.1 A1.6 A1.4 LLOYD’S REGISTER PARAMETRIC EQUATIONS Derivation of Equations and Safety Factors Derived Parametric Equations and Measured SCF Values Lloyd’s Reigster Equations for T/Y Joints Lloyd’s Register Equations for X Joints Lloyd’s Register K Joint Equations Lloyd’s Register KT Joint Equations Parametric Equation Expressions EFTHYMIOU PARAMETRIC EQUATIONS Efthymiou Equations for T/Y Joints Efthymiou Equations for X Joints Efthymiou Equations for K Joints Efthymiou Equations for KT Joints 52 .2 A1.2 A2.

which can be utilised to calculate the required safety factor. 53 . The quoted equations give an approximately mean fit to the database. It has been found that the LR equations are lognormally distributed about the mean fit line. SCF (design) = SCF (mean)† X (1 + nr/100) Where g = Chosen design curve weighting † Exclude the chord in place bending term. however characteristic equations can be derived by applying an appropriate safety factor. LLOYD’S REGISTER PARAMETRIC EQUATIONS A1. Beside each equation is the percentage standard deviation of the least squares fit to the database (r%). B0 X B1 and add in expression unfactored. Therefore. the design curve weighting may be calculated to give a estimated degree of joint underpredictions.A1.1 Derivation of Equations and Safety Factors These equations have been derived from the Lloyd’s Register (LR) simple joint SCF database using a least squares minimisation procedure.

86 X 1.2 Derived Parametric Equations and Measured SCF Values The SCF database was standardised to DEn fatigue recommendations prior to curve fitting (ie to results derived from linear extrapolation of maximum principal stresses outside the 0. Measured SCF results that do not meet this standard should be factored as follows: X 0.A1.95 X 0.95 X 0.23 to convert from non-linear extrapolation of stesses to simulate a weld fillet on the chordside to simulate a weld fillet on the braceside to convert perpendicular strain to principal stress (steel models) to convert perpendicular strains to principal stress (acrylic models) 54 .15 X 1.2í(rt) notch zone to the weld toe).

65 (Y ) ) ) (where Y° is the degree of weld cut-back (default value = 20°)).25 [ t [ 1.( c x sin 0.A1. t r=15% r=18% 55 .13 [ b [ 1.0 10 [ c [ 35 0.3 Lloyd’s Register Equations for T/Y Joints Notes: When a<12 the basic saddle SCF equation should be multiplied by the appropriate short chord correction factor F1.0 30° [ h [ 90° 4[a Note: b = 1 joints at the saddle: b = 1 . % Std Devn Axial load SCFCS = T1 x (F1 or F2) SCFCC = T2 + B0 X B1 SCFBS = T3 x (F1 or F2) SCFBC = T4 r=20% r=20% r=25% r=23% Out-of-plane bending SCFCS = T5 x (F3) SCFBS = T6 x (F3) r=22% r=28% In-plane bending SCFC = T7 SCFB = T8 Validity range The above equations for T/Y joints are generally valid for joint parameters within the following limits: 0. Apply the modified b value when predicting SCFs at the SADDLE on b = 1 joints under axial load or OPB. F2 etc.

Apply the modified b value when predicting SCFs at the SADDLE on b = 1 joints under axial load or OPB. F2 etc.0 10 [ c [ 35 0.( c x sin 0.13 [ b [ 1. % Std Devn Balanced axial load SCFCS = X1 x (F1 or F2) SCFCC = X2 SCFBS = X3 x (F1 or F2) SCFBC = X4 r=22% r=33% r=19% r=13% Balanced out-of-plane bending SCFCS = X5 x (F3) SCFBS = X6 x (F3) r=22% r=20% Balanced in-plane bending SCFC = X7 SCFB = X8 r=23% r=12% Validity range The above equations are generally valid for joint parameters within the following limits: 0.65 (Y ) ) ) (where Y° is the degree of weld cut-back (default value = 20°)).0 30° [ h [ 90° 4[a Note: b = 1 joints at the saddle: b = 1 .A1.4 Lloyd’s Register Equations for X Joints Notes: When a<12 the basic saddle SCF equation should be multiplied by the appropriate short chord correction factor F1.25 [ t [ 1. t 56 .

T4B S2BA IF4AB r=22% r=25% r=12% r=26% Single out-of-plane bending SCFCS = T5A S1AB x (F3A) SCFBS = T6A S1AB x (F3A) r=17% r=18% Unbalanced out-of-plane bending SCFCS = (T5A S1AB + T5B S1BA IF5AB) x (F3A) SCFBS = (T6A S1AB + T6B S1BA IF6AB) x (F3A) r=14% r=21% Single in-plane bending SCFC = T7A SCFB = T8A r=15% r=17% 57 . Apply the modified b value when predicting SCFs at the SADDLE on b = 1 joints under axial load or OPB. where brace A is always defined as the brace under consideration.A1.T3B S1BA IF3AB) x (F1A or F2A) SCFBC = T4A S2AB . The expressions should be calculated using the geometry associated with brace A.T1B S1BA IF1AB) x (F1A or F2A) SCFCC = (T2A S2AB .5 Lloyd’s Register K joint equations Notes: When a<12 the basic saddle SCF equation should be multiplied by the appropriate short chord correction factor F1. F2 etc. % Std Devn Single axial load SCFCS = T1A S1AB x (F1A or F2A) SCFCC = T2A S2AB + B0A x B1A SCFBS = T3A S1AB x (F1A or F2A) SCFBC = T4A S2AB r=18% r=13% r=20% r=23% Balanced axial load SCFCS = (T1A S1AB .T2B S2BA IF2AB) + B0A x B1A SCFBS = (T3A S1AB .

65 (Y ) ) ) (where Y° is the degree of weld cut-back (default value = 20°)).0 Note: b = 1 joints at the saddle: b = 1 .0 30° [ h [ 90° 4[a 0 < ζ < 1.0 10 [ c [ 35 0.25 [ t [ 1.( c x sin 0.13 [ b [ 1. t r=15% r=16% 58 .Balanced in-plane bending SCFC = T7A + T7B IF7AB SCFB = T8A + T8B IF8AB Validity range The above equations are generally valid for joint parameters within the following limits: 0.

(T1B S1BC S1BA . S2BC) + B0B x B1B B) SCFBS = T3A S1AB S1AC x (F1A or F2A) SCFBC = T4A S2AB A) SCFBC = T4B Max (S2BA. S2A = Max (S2AB. The expressions should be calculated using the geometry associated with brace A.A1.6 Lloyd’s Register KT joint equations Notes: When a<12 the basic saddle SCF equation should be multiplied by the appropriate short chord correction factor F1.T2A S2A IF2BA) + B0B x B1B.T1C S1CB S1CA IF1BC)] x (F1B or F2B) SCFCC = MAX [(T2B S2B .T1C S1CB S1CA IF1AC) x (FIA or F2A r=22% 59 . S2BC) B) (Outer Brace (Central Brace (Outer Brace (Central Brace r=18% r=13% r=13% r=20% r=23% r=23% Balanced axial load (only outer braces A and C loaded) Central brace (brace B) SCFCS = MAX [(T1B S1BA S1BC . T3B S1BC S1BA . F2 etc.T4C S2C IF4BC)] r=22% r=25% r=12% r=26% Where S2B = Max (S2BA. where brace A is always defined as the brace under consideration. S2AC) and S2C = Max (S2CB. % Std Devn Single axial load SCFCS = T1A S1AB S1AC x (F1A or F2A) SCFCC = T2A S2AB + B0A x B1A A) SCFCC = T2B Max (S2BA.T3C S1CB S1CA IF3BC)] x (F1B or F2B) SCFBC = MAX [(T4B S2B .T3A S1AB S1AC IF3BA). S2BC).T1A S1AB S1AC IF1BA). (T4B S2B . (T2B S2B T2C S2C IF2BC) + B0B x B1B] SCFBS = MAX [(T3B S1BA S1BC . except where otherwise stated. S2CA) Outer brace (brace A) SCFCS = (T1A S1AB S1AC . Apply the modified b value when predicting SCFs at the SADDLE on b = 1 joints under axial load or OPB.T4A S2A IF4BA).

T4A S2A IF4BA). (T3B S1BC S1BA .T3C S1CB S1CA IF3BC)] x (F1B or F2B) SCFBS = MAX [(T4 S2B .T3C S1CB S1CA IF3AC) x (F1A or F2A) SCFBC = (T4A S2AB -T4C S2CB IF4AC) r=22% r=25% r=12% r=26% 60 .T1A S1AB S1AC IF1BA). S2A = Max (S2AB. (T1B S1BC S1BA . S2AC) and S2C = Max (S2CB.T1C S1CB S1CA IF1AC) x (F1A or F2A) SCFCC = (T2A S1AB .T4C S2C IF4BC)] r=22% r=25% r=12% r=26% Where S2B = Max (S2BA.T2A S2A IF2BA) + B0B x B1B. (T4B S2B . S2CA) Outer brace (brace A) SCFCS = (T1A S1AB S1AC .T3A S1AB S1AC IF3BA).T2C S2CB IF2AC) +B0A x B1A SCFBS = (T3A S1AB S1AC .T2C S2C IF2BC)] + B0B x B1B] SCFBS = MAX [(T3B S1BA S1BC . (T2B S2BC .SCFCC = (T2 A S2AB -T2CS2CBIF2AC) + BOA x B1A SCFBS = (T3AS1ABS1AC -T3CS1CBS1CAIF3AC) x (F1A or F2A) SCFBC = (T4AS2AB-T4CS2CBIF4AC) r=25% r=12% r=26% Single out-of-plane bending SCFCS = T5AS1ABS1AC x (F3A) SCFBS = T6AS1ABS1AC x (F3A) r=17% r=18% Unbalanced out-of-plan bending (all braces loaded) SCFCS = (T5AS1ABS1AC + T5BS1BAS1BCIF5AB + T5CS1CBS1CAIF5AC) x (F3A) SCFBS = (T6AS1ABS1AC + T6BS1BAS1BCIF6AB + T6CS1CBS1CAIF6AC) x (F3A) r=14% r=21% Single in-plane bending SCFC = T7A SCFB = T8A r=15% r=17% Balanced in-plane bending (only outer braces A& C loaded) Central brace (brace B) SCFCS = MAX [(T1B S1BA S1BC .T1C S1CB S1CA IF1BC)] x (F1B or F2B) SCFCC = MAX [(T2B S2B . S2BC).

65 (Y ) ) ) (where Ψº is the degree of weld cut-back (default value = 20º)) t 61 .0 Note: β = 1 joints at the saddle: β = 1 −( c x sin 0.Single out-of-plane bending SCFCS = T5A S1AB S1AC x (F3A) SCFBS = T6A S1AB S1AC x (F3A) r=17% r=18% Unbalanced out-of-plane bending (all braces loaded) SCFCS = (T5A S1AB S1AC + T5B S1BA S1BC IF5AB + T5C S1CB S1CA IF5AC) x (F3A) SCFBS = (T6A S1AB S1AC + T6B S1BA S1BC IF6AB + T6C S1CB S1CA IF6AC) x (F3A) r=14% r=21% Single in-plane bending SCFC = T7A SCFB = T8A r=15% r=17% Balanced in-plane bending (only outer braces A& C loaded) Central brace (brace B) SCFC = MAX [T7B + T7A IF7BA).13 < β < 1. (T7B + T7C IF7BC)] SCFB = MAX [T8B + T8A IF8BA).0 30º < θ < 90º 4<α 0 < ζ < 1. (T8B + T8C IF8BC)] r=15% r=16% Outer brace (brace A) SCFC = T7A + T7C IF7AC SCFB = T8A + T8C IF8AC r=15% r=16% Validity range The above equations are generally valid for joint parameters within the following limits: 0.0 10 < ψ < 35 0.25 < τ < 1.

5)) sin-1 Θ Out-of-plane bending X5 = τβγ1.7β ) sin2.5 exp(-β1.1 T Factors .3 β (0.26-0.7 Θ T6 = 1 + τ0.3 θ T3 = 1 + τ0.3-0.4-β 5) sin1.5 γ0.2β 5) sin1.3 (0.27-0.2 Θ T4 = 2.6β 0.A1.65 γ(0.3 (1.6β 3) sin2 Θ X6 = 1 + τβγ1.2 β (2.5β) Out-of-plane bending T5 = τ γ β (1.68β) sin(1-β3) Θ T8 = 1 + τ0.3 β (0.7 Θ In-plane bending T7 = 1.6 γ1.22τ0.12-2β ) sin2 θ T2 = τ γ0.5-2.06 τγ exp (-β2 γ0.2 X Factors = X joint factors Note: Apply the modified β value when predicting SCFs at the SADDLE on β = 1 joints under axial load or OPB Axial load X1 = τβγ1.8 βγ(1-0.3 + 0.21β ) sin1.6 γ1.5 Θ A1.7 Parametric Equations Expressions A1.5)) (sin Θ + 3cos 2Θ) X3 = 1 + 0.7.7.2 (3.19-0.63-0.4β 2) sin2 Θ X2 = (0.4β ) sin0.T joint factors Note: Apply the modified β value when predicting SCFs at the SADDLE on β = 1 joints under axial load or OPB Axial load T1 = τ γ1.36 + 1.6 x X1 X4 = (1.9 τγ0.76-0.2 γβ (0.5 (0.185β 3) sin7(1-β2) Θ In-plane bending 62 .46-1.

4 IF Factors .7.8 βγ (0.2 Out-of-plane bending IF5ij = 0.7.X7 = τ0.13-2β i) γ0.5xij) where P = 2 if Θi > Θj P = 4 if Θi < Θj exp (-3xij) sin h i sin h j IF6ij = 0.25β ) sin1.5β-0.5 exp(−3x ij ) Where x ij = 1+ f ij sin h i bi 63 .5t i (−2) exp(−3x ij ) 2 IF8 ij = 40(b i − 0.5 ij x sin Where x ij = 1 + f ij sin h i bi ζij = Gap between weld toes of brace I and brace j/chord diameter A1.32-0.5 Θ A1.8 βγ(0.The stiffening effect of an additional brace Note: Apply the modified β value when predicting SCFs at the SADDLE on β = 1 joints under axial load or OPB (ie Eqn S1ij) S1 ij = 1 − 0.5 Θ X8 = 1 + τ0.3xij) where P = 5 if Θi < Θj exp (-0.5) (1-0.32β 5) sin0.14β i γ1.4x exp −30x 2 ij x bi bj 2 x sin h i Y (−2) S2 ij = 1 + exp − 2x 2 (h j xc (−0.75 ) − 2.6γ sin h i sin h j b min b max sin h i sin h j p P = 1 if Θi > Θj exp (-0.8β i) γ0.3 S Factors .2 sin Θi sin h j IF2ij = [20-8(β i+1)2] exp (-3xij) IF3ij = β i (2-1.Influence functions for K and KT joint expressions Note: Apply the modified β value when predicting SCFs at the SADDLE on β = 1 joints under axial load or OPB Axial load sin h i p IF1ij = β i (2.5 In-plane bending exp (-3xij) IF7 ij = 1.

05 + 30 t 1.8) 1.15 ) c Chord-end fixity parameter (C) 0.Short chord correction factors Note: Apply the modified β value when predicting SCFs on β = 1 joints F1 = { { { 1 .56β 2 .55β 1.97β 2 .89) α1.21γ(-1.0 1 .5) 1.2 − b )(cos 4 h + 0.5) 1.43β .16) α2.ζij = Gap between weld toes of brace I and brace j/chord diameter A1.7.00 B1 = 1.0 for pinned chord ends For a structural analysis a value of C = 0.5 (1.(1.02) γ0.04 exp(-0.16 exp(-0.(0.8 γ0.23 exp(-0.5 B Factors .38) α2.0 C = 0.71γ(-1.(0. A1.7 is normally assumed.83β .5 for fully fixed chord ends C = 1.0.0.7.0 1 .49γ(-0.03) γ0.0.0.Approximation of the chord in-plane bending B0 = Ct(B − t/ 2c )(a / 2 − b / sin h ) sin h (1 − 3 / (2c )) : for single axial load : for balanced axial load B0 = 0.5 < C < 1.6 F Factors .0 α < 12 α > 12 α < 12 α > 12 α < 12 α > 12 F2 = F3 = 64 .

12 exp(-4β ) + 0.2) Out-of-plane bending SCFCS = γ τ β (1.7 is normally assumed.23 exp(-0. Short cord correction factors (α<12) F1 = 1 .85 γ(1-0.55β 1.16) Θ Chord-end fixity parameter (C) 0.6 Θ SCFBS = τ(-0.6) τ β 2(1 .04 exp (-0.77β) sin(0.7-0.5) F3 = 1 .45 β τ0.0.7 .1)(0.96)} sin(2.99 .54) γ(-0.05) (0.83β .25β 1.0.1(β -0.0.β 2)0.7 and apply expression F2 if C>0.7 Θ SCFBC = 1 + 0.65 β τ0.5 for fully fixed chord ends C = 1.0.1 EFTHYMIOU PARAMETRIC EQUATIONS Efthymiou Equations for T/U Joints Note: When α<12 the basic saddle SCF equation should be multiplied by the appropriate short chord correction factor F1.71γ(-1.68β) sin0.187-1.06γ-1.65)2} + τ β (0.97β 2 .5 < C < 1.5C α-3) sin Θ SCFBS = 1.6 Θ + (2C .1 {1. F2 etc Short Cord Factor F1 or F2 Axial load SCFCS = γτ1.011β 2-0.(0.1{0.52 α0.A2.045} + τβ (0.16 exp(-0.8α .5 sin2 (2Θ) SCFCC = γ0.2 τ{2.05β 3) sin1.(1.3 + γ τ0.38) α2.02)γ0.89) α1.08β 4) SCFCS In-plane bending SCFCC = 1. A2.43β .56β 2 .4 γ(1.0.16) α2.49γ(-0.65 + 5(β -0.11 .03)γ0.7 65 .2C α-1.52)2} sin1.3(β .0.09-0.2 {0.1.47β + 0.01α) Θ SCFBC = 3 + γ1.0 C = 0.0.8) F1 or F2 Under axial load.8γ 0.21γ(-1. apply expression F1 if C<0.5) F2 = 1 .0 for pinned chord ends For a structural analysis a value of C = 0.

0 90° 40 66 .2 8 0.0 32 1.2 20° 4 < < < < < β γ τ Θ α < < < < < 1.Validity range The above equations for T/Y joints are generally valid for joint parameters within the following limits: 0.

23 exp(-0.97β 2 .55β 1.71γ(-1.12 exp(-4β ) + 0.0 C = 0.09-β 1.04 exp (-0.38) α2.2 {0.08β 4) SCFCS Balanced in-plane bending SCFCC = 1.8γ 0.5) F2 = 1 .6 Θ SCFBS = τ(-0.2 Efthymiou Equations for X Joints Note: When α<12 the basic saddle SCF equation should be multiplied by the appropriate short chord correction factor F1.011β 2-0.47β + 0.0.9{1.8) F1 or F2 F3 F3 Under axial load.56β 2 .9 γ τ0.16 exp(-0.65 + 5(β -0.β 1.5 for fully fixed chord ends C = 1.0 32 1.0 90° 40 67 . F2 etc Short Cord Factor F1 or F2 Axial load SCFCS = 3.8) sin1.5 Θ SCFBC = 3 + γ1.43β .45 β τ0.5 β 0.0.10 .0.7) sin2.65)2} .09-0.5 < C < 1.7 is normally assumed.49γ(-0.03)γ0.99 .16) α2.2 τ{2.21γ(-1.0 for pinned chord ends For a structural analysis a value of C = 0. apply expression F1 if C<0.7 and apply expression F2 if C>0.0.0.85 γ(1-0.0.77β) sin(0.89) α1.5) F3 = 1 .16) Θ Chord-end fixity parameter (C) 0.34β 4) sin1.68β) sin0.7 Validity range The above equations for X joints are generally valid for joint parameters within the following limits: 0.56 .2 8 0.1.A2.3τβ sin Θ SCFBS = 1 + 1.83β .(1.7 Θ SCFBC = 1 + 0.045} Balanced out-of-plane bending SCFCS = γ τ β (1.7 Θ SCFCC = γ0.05) (0.65 β τ0.87γτβ (1.4 γ(1.54) γ(-0.02)γ0.06γ-1. Short cord correction factors (α<12) F1 = 1 .(0.2 20° 4 < < < < < β γ τ Θ α < < < < < 1.

5 exp(-0.7ΘA (ATAN in radians) Single out-of-plane bending SCFCS = [γτΑ β A (1.16β A) sinΘA [1.045} + τΑβ Α (0.8x)] + [γτΒβ B (1.3 F1 or F2 x SCFB = 1 + [SCFC](1.1.7 .1)(0.6 Θ + (2C .8α .08β A4) SCFCS Unbalanced out-of-plan bending SCFCS = [γτAβ A (1.08β 4A) {SCFCS} F3 F3 F4 F4 68 .6) τΑ β 2A(1 .25)τA(-0.5 (0.1{0. F2 etc The expressions should be calculated using the geometry associated with brace A.5 sin2 (2ΘΑ) SCFCC = γ0.A2.14) sin0.8x)] SCFBS = τA(-0. where brace A is always defined as the brace under consideration.25β A1.3 + γτA0.2) Balanced axial load SCFC = τ0.99-0.47β A + 0.67-β 2A + 1.5max exp(-1.05β 3A sin1. For K joint equations.65 + 5(β A-0.6 ΘΑ] [1−0.2C α-1.0.05β 0.5 exp(-0.52 α0.05β A3) sin1.08(β Bγ)0.3 Efthymiou Equations for K Joints Note: When α<12 the basic saddle SCF equation should be multiplied by the appropriate short chord correction factor F1.3(β A .01α) ΘΑ SCFBC = 3 + γ1.1 {1.54) γ(-0.2 {0.8x)] x [2. the gap parameter.1.011β A2-0.99 .57β A0.96)} sin(2.05β 3B sin1.5 exp(-0.7 .7 .12 exp(-4β Α) + 0.05) (0.7-0.11 .0.5C α-3) sin ΘΑ SCFBS = 1.1(β A-0.08(β Bγ)0.97-1.6ΘB] [1-0.08(β Αγ)0.6Θ] [1-0.65)2} + τΑ β A (0.3x)] SCFBS = τA(-0.38) ATAN (8ζ)] sin h max sin h min 0.52)2} sin1.9A γ0.54) γ(-0.47β Α + 0.2τA{2.β 2A)0.29 β Α(-0.05) (0. x = 1 + z sin θA/β A Short Cord Factor F1 or F2 Single axial load SCFCS = γτA1.64 + 0.3 b max b min 0.1.187-1.

(0.0.7 and apply expression F2 if C>0.04 exp (-0.(1.4) Under axial load.2 8 0.5 for fully fixed chord ends C = 1.68β) sin0.16) ΘA Chord-end fixity parameter (C) 0.7 ΘA SCFBC = 1 + 0.0.85 γ(1-0.16) α2.7ΘA][1 + 0.65 β A τA0.8) F4 = 1 .4 γ(1. apply expression F1 if C<0.09-0.2 20° 4 0.77βΑ) sin(0.0 < < < < < < β γ τ Θ α ζ < < < < < < 1.0.46 β A1.107β 1.16) ΘA Balanced in-plane bending SCFCC = [1.43β .09-0.23 exp(-0.5) F2 = 1 .77βA) sin(0.0 C = 0.0 69 .7 Validity range The above equations for X joints are generally valid for joint parameters within the following limits: 0.0 32 1.03)γ0.06g-1.8γ 0.Single in-plane bending SCFCC = 1.45β A τA0.06) α2. Short cord correction factors (α<12) F1 = 1 .88exp(0.5) F3 = 1 .0 90° 40 1.56β 2 .21γ(-1.49γ(-0.0.16 exp(-0.97β 2 .65β A τA0.0 for pinned chord ends For a structural analysis a value of C = 0.5 < C < 1.02)γ0.4γ (1.89) α1.2 exp(-3ζ)] SCFBC = 1 + 0.38) α2.7 is normally assumed.85 γ(1-0.16γ(-1.68β) sin0.83β .0.45 β A τA0.71γ(-1.06γ-1.55β 1.

6ΘB][1-0.14) sin0.54) f AB sin h B bB and xBC = 1 + f BC sin h B bB γ(-0.5 (0.05β 0.6ΘB] 70 .5 max exp(-1.38) ATAN (8xMAX(ζAB.05β 3A)sin1.5 exp(-0.08(β Bγ)0.7-1. sin h C 0.8xAB)][2.7-1.87-1.57β Β0.08(β Cγ)0.7-1.08(β Bγ)0.64+0.7ΘB Outer brace (brace A) SCFC = τA0.16β A) sinΘA x [1.97-1.29 β B(-0. sin h C Min (sin h A.05) (0.A2.67-β 2B + 1.6ΘC] [1-0.3 Max (b A.67-β 2A + 1.05β 3A) sin1.3 (ATAN in radians) (ATAN in radians) Unbalanced out-of-plane bending Central brace (brace B) SCFC = {[γτBβ B(1.6ΘA] [1-0.7-1.64 + 0.6ΘA] x [1-0.05β 3B)sin1. b C 0.5 exp(-0.3 sin h max 0.38) ATAN(8ζAC)] SCFB = 1 + [SCFC](1.5 exp(-0.4 Efthymiou Equations for KT Joints Note: The expressions should be calculated using the geometry associated with the brace under consideration ie Central Brace B or Outer Brace A.9 γ0.29 β B(-0.8xAB)](b A /b B ) [1-0.05β 3C) sin1.5 exp(-0.3 b max 0.47β B + 0.3xBC)]] Where xAB = 1 + SCFB = t B (−0.08 b 4 B) SCFC Outer brace (brace A) SCFC = {[γτAβ A(1.0-0. ζBC)] SCFB = 1 + [SCFC](1.16β B) sinΘB sin h min b min x [1.8xBC)][2.3xAB)]] + [γτCβ C(1.8xBC)](b C /b B ) } 2 2 + [γτAβ A(1.08(β Cγ)0.97-1.5 exp(-0.05β 3B) sin1.8xAC)]} + [γτBβ B(1.7ΘA Where ζAC = ζAB + ζBC + bB sin h B Max (sin h A.5 exp(-0. Balanced axial load Central brace (brace B) SCFC = τB0.9 γ0.25)τA(-0.05β 0.08(β Bγ)0.08(β Aγ)0.99-0.57β A0.8xAB)][1.5 max exp(-1.14) sin0.25) τB(-0. b C Min (b A.5 (0.

3xAB)]] + [γτCβ C(1.5 max exp(-1.06γ-1.05β 3C)sin1.2A exp(-3ζAB)] SCFB = 1 + 0.0 71 .5 exp(-0.0 for pinned chord ends For a structural analysis a value of C = 0.ζBC)]] SCFB = 1 + 0.08 b 4 A) SCFC Balanced in-plane bending Central brace (brace B) SCFC = 1.0 < < < < < < β γ τ Θ α ζ < < < < < < 1.7 is normally assumed.[1-0.45β A τA0.4γ (1.46β 1.5 < C < 1.6ΘC] [1-0.7-1.5 for fully fixed chord ends C = 1.09-0.0 32 1.06γ-1.65β B τB0.8xAC)][2.68βA) sin0.2 8 0.7 ΘB [1+0.7ΘA [1+0.5 exp(-0.08(β Aγ)0.54) f AB sin h A bA and xAC = 1 + (f AB + f BC + b B / sin h B ) sin h A bA γ(-0.3xAC)]] Where xAB = 1 + SCFB = t A (−0.05β 0.0 90° 40 1.5 max exp(-1.09-0.85γ (1-0.2 20° 4 0.16) ΘA Outer brace (brace A) SCFC = 1.2B exp(-3Min(ζAB.77βB) sin(0.46β 1.0 C = 0.16) ΘA Chord-end fixity parameter (C) 0.47β A + 0.65β A τA0.45 β B τB0.85 γ(1-0.4γ (1.77βA) sin(0.68βB) sin0. Validity range The above equations for KT joints are generally valid for joint parameters within the following limits: 0.08(β Aγ)0.05) (0.05β 0.8xAB)][2.99-0.

APPENDIX B SIMPLE JOINT SCH ASSESSMENT DATABASE 72 .

1 .INDEX Tables B1.1 .1 .B1.B3.15 Tables B2.6 APPENDIX B REFERENCES T/Y Joint Geometries and SCFs X Joint Geometries and SCFs K Joint Geometries and SCFs 73 .B2.5 Tables B3.

74 .

75 .

76 .

77 .

78 .

79 .

80 .

81 .

82 .

83 .

84 .

85 .

86 .

87 .

88 .

89 .

90 .

91

92

93

94 .

95 .

96 .

97 .

98 .

99 .

1982/3 programmes Lloyd’s Register of Shipping. Stress concentration at tubular Y and oblique X joints Contract No. Comparisons of local stress and compliance in tubular welded T-joints Paper 4.. TAKIZAWA S. 12.3 SIMS.C. 1986 2. 1974 WORDSWORTH A. and DOVER W..K. 1979 YAMAMOTO N. 6. CONNOLLY M. 13.D. 7. and DHARMAVASANS Experimental and F.D. Stress concentration factors for tubular Y.P.S.E. 1979 Lloyd’s Register Complex Joint SCFs . UKOSRP I final report UKOSRPI I.and T-joints Int J Fatigue 12 No 1 (1990) pp 13-23 DOVER W. and KOMATSU M Static and fatigue tests on large size tubular T joints OTC 3424.. et al Experimental and analytical studies on fatigue of stiffened and unstiffened tubulat T joints OTC 5308.. 9. Studies of stress concentrations and tolerances for weld defects in full-scale tubular joints OTC 3692. T joint SCFs measured on acrylic models in 1974. YAMASAKI T. 5. 8. BHUYAN G. and KUDOH J. 1981 DIJKSTRA O. 100 . 1988 WORDSWORTH A.. 10. CHAUDHURY G.C. and DEBACK J.. 11. 62/04150343. Fatigue strength of tubular T and X joints OTC 3696. 1980 YAMASAKI T. 1980 UKOSRP II final report UKOSRP II 1986 Elastic stress concentration factor tests on tubular steel joints . 3.APPENDIX B REFERENCES 1. 4. 1982/3 HELLIER A.JISSP Wimpey Offshore 1986 Stress concentration factors for tubular complex joints Lloyd’s Register of Shipping. 1980 MUNASWANY K.D.K.

WILSON T. 1986 UCL 104 ZIRN R Fatigue behaviour of welded tubular joints in circular hollow sections Staatliche Materialprufung . Ref ST102/83A.Sansalt . 1986 15. et al Experimental results from X joints Paper 6.3.M. and MACDONALD A. 1981 MA A.A. 1981 123 IRVINE N. and TETSIYI F. 24. and MOE E. 21. 1983 DHARMAVASANS S and SENEVIRATNE L Stress analysis of overlapped K-joints ImechE. Fatigue fracture mechanics assessment of tubular welded T joints Dept of Energy contract OT/F/917. 20. Tubular joint fatigue data obtained on the National Engineering Laboratory UKOSRP Report 4/02 KRATZER et al Schwingfestigkeitsuntersunchungen as Rohrstruktukturelement-modellen von offshore plattformen Report BMFT-FB (M81). GIBSTEIN M.. Final report 1983-1985 MASAAHIRO S. 1979 LIEURADE H. BROWN G. 17.J.. 1975 Final report for the testing of K joints Wimpey Laboratories. 18. 16.D. 26. OTH 88 284 SWENSSON K. Numerical and experimental stress analysis of tubular joints with inclined braces SIMS 6. 22..B. 23.T. 101 . 25. YURA J.D.Stuttgard. 19. et al Stress concentration factors in double-tee tubular joints PMFSEL Report No 86-1. Experimental research on fatigue strength of tubular joints Mitsui Zosen Technical Review No 104.. Stress concentration factor data from large scale tubular joints Wimpey Offshore.14.5 SIMS. 1981 DOVER W. P.

the extrapolation would be carried out through -3 and -1.EXTRAPOLATION PROCEDURES The extrapolation methods used by Lloyd’s Register in conjunction with acrylic modelling to determine the brace to chord intersection SCFs at various positions are described below: METHOD 1 A linear extrapolation of the SCFs corresponding to the principal stresses at the two rosettes nearest the junction. with SCFs of -3 and 1 at the nearest rosette and -1 and 2 at the other one.APPENDIX C . At the rosette nearest the intersection the largest of the numerical principal stress SCFs is used and at the second rosette the SCF which is algebraically nearest to it. METHOD 5 Data given in this method is either the SNCF at an isolated single element gauge or the SCF corresponding to the greatest principal stress at an isolated rosette. SNCF + (v x SNCF90 ) 1 − v2 102 . SCF = Where v = Poisson’s Ratio METHOD 4 As for Method 3 but using a non-linear extrapolation on a quadratic curve through the three measuring points. METHOD 2 As for Method 1 except that a non-linear extrapolation is used on a quadratic curve through the three measuring points. This is then converted to an SCF using UNCF90 (ie the SNCF corresponding to the strain parallel to the intersection at the measuring point nearest to the intersection). Thus. METHOD 3 A linear extrapolation is carried out of the SNCFs based on the strains normal to the intersection at the two points nearest to the intersection.

each with one element aligned normal to the intersection Three 45° rosettes. 2.The methods of extrapolation which can be employed at the various gauge configurations at each position around the junction are therefore as follows: Method (a) (b) (c) (d) (e) Single isolated rosette or single element gauge Two single element gauges aligned normal to the intersection with one orthogonal gauge Three single element gauges aligned normal to the intersection with one orthogonal gauge Two 45° rosettes. each with one element aligned normal to the intersection 5 3 3&4 1&3 1. 3 & 4 103 .

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 listening from where you left off, or restart the preview.

scribd