ANALYSIS OF CORRUGATED WEB BEAM TO COLUMN EXTENDED END PLATE CONNECTION USING LUSAS SOFTWARE

ANIZAHYATI BINTI ALISIBRAMULISI

UNIVERSITI TEKNOLOGI MALAYSIA

PSZ 19:16 (Pind. 1/97)

UNIVERSITI TEKNOLOGI MALAYSIA

ANALYSIS OF CORRUGATED WEB BEAM TO COLUMN EXTENDED END PLATE CONNECTION USING LUSAS SOFTWARE 2005/2006 ANIZAHYATI BINTI ALISIBRAMULISI

NO. 6, JALAN BUNGA KEMUNTING 2/10, 40000, SHAH ALAM, SELANGOR 19 MEI 2006

P.M DR SARIFFUDDIN SAAD

19 MEI 2006

CATATAN: * Potong yang tidak berkenaan. ** Jika tesis ini SULIT atau TERHAD, sila lampirkan surat daripada pihak berkuasa/organisasi berkenaan dengan menyatakan sekali sebab dan tempoh tesis ini perlu dikelaskan sebagai SULIT atau TERHAD. Tesis dimaksudkan sebagai tesis bagi Ijazah Doktor Falsafah dan Sarjana secara penyelidikan, atau disertasi bagi pengajian secara kerja kursus dan penyelidikan, atau Laporan Projek Sarjana Muda (PSM).

I hereby declare that I have read this project report and in my opinion this report is sufficient in terms of scope and quality for the award of the degree of Master of Engineering (Civil Structure)

Signature:

....................................................

PROF. DR SARIFFUDDIN SAAD Name of Supervisor: ASSOC. .................................................... 19 MAY 2006 Date: ....................................................

ANALYSIS OF CORRUGATED WEB BEAM TO COLUMN EXTENDED END PLATE CONNECTION USING LUSAS SOFTWARE

ANIZAHYATI BINTI ALISIBRAMULISI

A project report submitted in partial fulfillment of the requirement for the award of the degree of Master of Engineering (Civil Structure)

Faculty of Civil Engineering Universiti Teknologi Malaysia

MEI 2006

I declare that this project report entitled Analysis of Corrugated Web Beam to Column Extended End Plate Connection Using LUSAS Software is the result of my own research except as cited in the references. The report has not been accepted for any degree and is not concurrently submitted in candidature of any other degree.

Signature: Name: Date:

.. ANIZAHYATI BINTI ALISIBRAMULISI .. 19 MEI 2006 ..

iii

ACKNOWLEDGEMENT

In the name of ALLAH, The Most gracious, Most merciful, with His permission, Alhamdulillah this proposal report has completed. Praises to Prophet Muhammad, his companies and those on the path as what he preached upon, may ALLAH The All Mighty keep us in his blessings and tender care.

I would like to convey my highest appreciation to those who had sincerely, without hesitation helped to make this report a possible success. My highest level of appreciation to Associate Professor Dr Sariffuddin Saad for his guidance, without his corporation, I would not be able to complete this proposal report. A special thanks to Mr Arizu Sulaiman (PhD candidate UTM), Mr Anis Sagaff (PhD candidate UTM), and Mr Che Husni for giving me the required information and guidance for the completion of this study.

I would like to express my heartfelt appreciation to my husband (Ahmad Saifuddin bin Abdul) and my children (Amiratul Soffiya and Amiratul Syuhada), for their patient, love, prayers, support and also for understanding the sacrifices required in completing this study. My sincere and special thanks also go to my beloved friends and classmates for being supportive and for their contributions and understanding.

Lastly but not least, thank you to all that have contributed either directly or indirectly in making this study a success.

iv

ABSTRACT

Bolted extended end plate connections are commonly used in rigid steel frame. Inappropriate or inadequate connections of beam and column are hazardous and can lead to collapses and fatalities. Although laboratory testing is more accurate in analyzing the connection, but unfortunately it is time consuming and quite expensive. Thus, this project is intended to develop a Finite Element Analysis (FEA) approach as an alternative method in studying the behavior of such connections. The software being used is LUSAS 13.5 and the model used was an extended end plate, welded to the end of a corrugated web beam and then bolted to the column-flange. This type of connection will cause the column to bend about its major axis, and affect the end plate, bolts and corrugated web beam. Therefore, the analysis will be much more difficult as compared to plain web beam. The moment-rotation (M-) response of the joint was plotted in term of a M- curve, and then, it was superimposed with the curve taken from an existing experimental result. It was found that the two curves shared the same stiffness at the elastic stage of the loading and they started to diverge as the connection became plastic. However, the LUSAS moment of resistance is 50% more than that obtained in the experiment. Further investigations are necessary to improve the finite element prediction.

ABSTRAK

Sambungan rasuk kepada tiang dengan menggunakan skrew dan plat hujung adalah satu perkara biasa dalam sambungan kerangka besi. Ketidaksesuaian dan kelemahan sambungan rasuk dan tiang adalah berbahaya, dan boleh mengakibatkan keruntuhan kerangka dan kemalangan jiwa. Walaupun ujikaji makmal merupakan kaedah yang tepat untuk menganalisa jenis sambungan tersebut, tetapi ia memakan masa yang lama dan memerlukan kos yang lebih tinggi. Oleh itu, projek ini bertujuan untuk membangunkan analis unsur terhingga sebagai salah satu alternatif dalam mengkaji kelakuan sebenar sambungan tersebut. Perisian yang digunakan bagi analisis unsur terhingga ini adalah LUSAS 13.5 dan komponen-komponen ynag terlibat dalam sambungan tersebut adalah; plat hujung yang dikimpal kepada hujung rasuk yang corrugated dan kemudiannya diskrewkan pada bebibir tiang. Sambungan jenis ini akan menyebabkan tiang melentur pada paksi major dan memberi kesan kepada plat hujung, skrew dan rasuk yang corrugated tersebut. Analisis ini adalah lebih kompleks berbanding dengan rasuk biasa. Tindakbalas momen-putaran(M-) sambungan tersebut diplotkan dalam bentuk lengkungan M-, yang kemudiannya disuper-impose dengan lengkungan M- ujikaji. Hasilnya didapati, 2 lengkungan tersebut berkongsi nilai kekuatan yang sama pada tahap elastik beban dan kemudiannya berpecah apabila sambungan mula bersifat plastik. Walaubagaimanapun, keputusan momen kapasiti LUSAS adalah 50% melebihi momen kapasati ujikaji. Oleh itu, lebih banyak penyelidikan diperlukan di masa hadapan untuk memperbaiki keputusan analisis unsur terhingga ini.

vi

TABLE OF CONTENTS

CHAPTER

TITLE

PAGE

INTRODUCTION 1.1 1.2 1.3 1.4 1.5 PROBLEM BACKGROUND PROBLEM STATEMENT OBJECTIVES OF THE STUDY SCOPE OF THE STUDY SIGNIFICANCE OF RESEARCH

1 1 1 2 2 3

LITERATURE REVIEW 2.1 2.2 INTRODUCTION CORRUGATED WEB BEAM AND EXTENDED END PLATE 2.3 2.4 CLASSIFICATION OF CONNECTIONS MOMENT-ROTATION (M-) CHARACTERISTICS 2.5 ANALYSIS OF CONNECTIONS

4 4 12

15 19

22 22

2.5.1 EXPERIMENTAL SET-UP

RESEARCH METHODOLOGY 3.1 LUSAS SOFTWARE

30 30 30 31 35 36 38

3.1.1 FINITE ELEMENT MODEL 3.1.2 ELEMENT TYPES 3.1.3 NON-LINEAR ANALYSIS 3.1.4 BOUNDARY CONDITIONS 3.1.5 SUMMARY OF LUSAS FINITE ELEMENT SYSTEM

vii

RESULTS AND DISCUSSIONS

39

CONCLUSION AND RECOMMENDATION FOR FUTURE WORK 5.1 5.2 CONCLUSIONS RECOMMENDATION FOR FUTURE WORK

46

46 46

REFERENCES

48

APPENDIX A (LUSAS Element Types)

51

APPENDIX B (Variations in K)

61

viii

LIST OF FIGURES

FIGURE 2.1 2.2 2.3 2.4 2.5 2.6 Corrugated plate Corrugated web beam

TITLE

PAGE 12 12 14 14 14 15

Extended End Plate Connection Connection Loading Components of beam to column connection Strength, stiffness and deformation capacity of steel and connections

2.7 2.8(a) 2.8(b) 2.8(c) 2.9 2.10 2.11 2.12 2.13 2.14 2.15 2.16 3.1 3.2 3.3(a) 3.3(b) 3.3(c) 3.3(d) 3.3(e) 3.4

Failure modes for bolted T-stub connections Moment-rotation curves of beam to column connections Moment-rotation curves of beam to column connections Moment-rotation curves of beam to column connections Nominally pinned connections Flush end plate connection Rigid Connections Semi rigid (semi flexible) connections Characteristics of beam-to-column connections Moment-rotation diagram of beam-to-column connections Experimental set-up Details of beam-to-column connections Element types Enlarged FEA Bolt Arrangement Line mesh Mesh Discretisation Mesh BRS2 and JNT4 Attribute forms Attribute forms Attribute forms

15 16 16 16 17 18 18 19 20 21 22 23 32 32 32 33 33 34 35 36

ix 3.5 3.6 4.1 Attribute forms FEA Supports and Loading Comparison of M- Curve between Experimental and FEA results (Nonlinear Analysis) 4.2 Experimental graph for Moment-Rotation Curve (N5 specimen) in determining Moment capacity, MR of the connection. 4.3 Finite Element Analysis graph for Moment-Rotation Curve (N5 specimen) in determining Moment capacity, MR of the connection. 4.4(a) 4.4(b) 4.5 4.6 Position of nodes selected in determining the displacement Position of nodes selected in determining the displacement N5 specimen after failure FEA deformed mesh for N5 43 43 45 45 43 41 37 37 41

CHAPTER 1

INTRODUCTION

1.1

PROBLEM BACKGROUND To date, the experimental approach to study the behaviour of connection in steel structures will certainly remain the most popular for still some years but because of the highly cost involved, researchers are increasingly looking for less costly but acceptable alternatives. The most obvious alternative is modeling by the finite element method. Due to the highly complex nature of connections and the large number of parameters involved, numerous tests are required before an adequate set of empirical formulae is developed for the design of a specific type of connection. It appears to be more rational and more economical to develop numerical models to play with the various parameters and to check the accuracy of the numerical models against the results of an appropriate number of experimental tests. Not only are experimental tests needed to validate the models but they are also required for calibration purposes.

1.2

PROBLEM STATEMENT Accurate analysis of the connection is difficult due to the number of connection components and their inherit non-linear behaviour. The bolts, welds, beam and

2 column sections, connection geometry and the end plate itself can all have a significant effect on connection performance. Any one of these can cause connection failure and some interact. The most accurate method of analysis is of course to fabricate full scale connections and test these to destruction. Unfortunately this is time consuming, expensive to undertake and has the disadvantage of only recording strain readings at pre-defined gauge locations on the test connection. A three dimensional materially static non-linear finite element analysis approach has therefore been developed as an alternative method of connection appraisal. For this research, extended end plate and corrugated web beam will be used, since not much research is done on such connections.

1.3

OBJECTIVES OF THE STUDY The main objective of this research is to study the moment-rotation behaviour of corrugated web beam to column connections. A static non-linear finite element analysis will be used to model and analyze the bolted connection. Extended end plate and non linear elastic-plastic behaviour will be considered in the analysis. The moment-rotation curve plotted from the result will be compared with the relevant data available from experimental testing.

1.4

SCOPE OF THE STUDY There are various types and shape of connection in structural steelwork. This study focused mainly on extended end plate bolted connection and corrugated web beam, particularly, trapezoidal web beam. The plate has 8 holes and M20 bolts will be used. The column size is 305x305x118 UC (S275) and its length is 3 m and the beam size is 400x140x39.7/12/4 1.5m, Flange S355, Web - S275. A static point load was applied incrementally at the end of the cantilever beam. LUSAS software [1] will be used to model the connections. The result from the finite element analysis, mainly moment-rotation curve, will be compared with the existing experimental result.

3 1.5 SIGNIFICANCE OF RESEARCH Research significance to be obtained from this study will be the results and analysis of the behavior of beam to column connection, when extended end plate and corrugated web beam is used. It is necessary to compare the moment rotation curve of the result from the finite element analysis and experimental testing. The aim was to determine the accuracy of the analytical method and to verify the strength of the corrugated web beam as compared to a plane web. Corrugated web beam is still new in the industry, so if much research is done on it, more application of it can vary our steel industry products.

CHAPTER 2

LITERATURE REVIEW

2.1

INTRODUCTION Despite numerous years of extensive research, particular in the 1970s, no fully agreed design method exists. Many areas of connection behaviour still require investigation. More recently Bose, Sarkar and Bahrami [2] used FEA to produce moment rotation curves, Bose, Youngson and Wang [3] reported on 18 full scale tests to compare moment resistance, rotational stiffness and capacity. The latest design method utilizes plastic bolt force distribution to create an increased moment connection capacity and reduced column stiffening. In 1995 when the SCI and the BCSA produced the Green Book guide, based on the EC3 design model, the editorial committee felt a number of areas, particularly bolt force distribution and compression flange overstress required further investigation. Krishnamurthy (1979) [4] conducted early finite element analysis of moment endplate connections. This study included thirteen finite element models of benchmark connections, with dimensions spanning values commonly used in the industry. The study was limited greatly by the technology at that time. A 2-dimensional/3-dimensional finite element analysis was conducted to determine adequate correlation between results. If such correlation could be found for the thirteen connections considered, then two dimensional analyses could be used

5 with factors that correlated to the three-dimensional model of the same connection. According to this paper, it would have been impossible to feasibly conduct an exhaustive three-dimensional analysis of end-plate connections because of the time involved in the programming, as well as the computational time required. Additionally, the task of creating a threedimensional model with every detail and adequate proportions was deemed impossible at that time. Therefore some simplifying assumptions were made, and a three-dimensional model was created based on a constant strain triangle and eight-node sub parametric brick elements. Bolt heads were omitted, and the bolts were modeled as rectangular shanks, having the same crosssectional area as round bolts. No contact elements were used. The bolts were assumed to be in tension, and an effective square area at the compression flange was assumed to be compressed against the column flange. This study produced stress distribution plots at different loading magnitudes on the plate. Some correlation between the two-dimensional and three-dimensional models was observed. The three-dimensional model had less stiffness than the two-dimensional model because of the prevention of the transverse variation of deformations and stresses. Seven correlation factors, relating the two different models, were tabulated for each of the thirteen benchmark connections. Krishnamurthy (1979) [4] concluded that prying forces do not exist in moment end-plate connections based on this study. Bursi and Leonelli (1994) [5] presented some additional results that had not been discussed by Bursi and Jaspart (1997b) [6]. Twenty-node brick elements were used to model the beam and plate material. Contact elements were used to represent the end-plate/column-flange interaction problem. Once again, beam elements were used to model the bolts, but here the bolts were pretensioned to a snug tight condition. The column flange was considered rigid. End-plate rotation and bolt loads were examined using the finite element model. Fairly good correlation with experimental results wass obtained. A direct application of the model suggested by Richard and Abbott (1975) [7] is used to describe the analytical results obtained from the finite element model. Using the finite

6 element method to obtain the elastic stiffness Ke,th, the inelastic stiffness Kp,th, the plastic failure moment Mp,th, and the ultimate applied moment Mu,th for the connection, the moment rotation plot or M- relationship can be described by;

Where n is the shape factor. Gebbeken et al. (1994) [8] investigated the different finite element modeling techniques to uncover the important criteria for describing moment end-plate connection behavior. Also, the authors discussed the results of a parametric study to determine which elements of the connection provided significant amounts of connection flexibility. The four-bolt unstiffened extended endplate connection was considered. First, a two-dimensional model was used. The material stress/strain relationship was represented as a bilinear function. Friction between the column flange and the end-plate was neglected. The results from this analysis were poor since strength predictions were found to be very unconservative when compared to the experimental results. The three-dimensional model used by the authors provides some limited success in predicting the moment-rotation characteristics of the connection. The description of the finite element model was vague, yet it was mentioned that brick elements were used. Also, the figures in the paper made it to appear that a tee stub and not an actual end-plate was considered. In some cases the results were accurate, but in others the strength were off by 50% or more, possibly suggesting inadequate modeling assumptions. Rothert et al. (1992) [9] presents similar results and findings based on the same research. Sherbourne and Bahaari (1997) [10] developed a methodology based on three dimensional finite element designs, to analytically evaluate the moment rotation

7 relationships for moment end-plate connections. ANSYS 4.4 was the software package used. The purpose for this research was to provide designers with a method of determining stiffness for these connections. It was apparent at the time that the ability of designers to produce a moment-rotation curve for moment end-plate connections was limited. Because of advancements in computer technology, Sherbourne and Bahaaris models included plate elements for the flange, webs, and stiffeners of the column and beam, as well as taking into account the bolt shank, nut, head of the bolt, and contact regions. However, bolt pre-stressing was not included. It was determined that the behavior of a moment rotation curve for an end-plate connection throughout an entire loading history, up to and including failure, can be feasibly and accurately modeled by performing a three-dimensional finite element analysis. This is particularly useful when one of the plates in contact, either the column flange or the end plate, is thin. The analysis of such a plate is inaccurate when using two-dimensional models. An additional advantage to the use of the three-dimensional model is the separation of the column, bolt, plate, and beam stiffness contributions to the overall behavior of the connection. Bahaari and Sherbourne (1997) [10] presented part two of their finite element study on moment end-plate connections. Based on the parametric study found in Sherbourne and Bahaari (1997) [10], this paper uses the Richard-Abbott power function (similar to that suggested by Bursi and Leonelli (1994) [5]) to describe the moment-rotation behavior of four-bolt unstiffened extended moment end-plate connections of known geometrical configuration. The proposed moment-rotation relationship is;

where the elastic stiffness Ki, the inelastic stiffness Kp, and the plastic failure moment Mp were all obtained from the results of a finite element analysis. Mo

8 and n are the connection-dependent reference moment and shape factor, respectively. If Ki and Kp are equal, the function becomes linear. Likewise, if Kp is zero, the curve becomes an elastic plastic model of the connections behavior. For large values of n, the model approaches a bilinear model of behavior. A curve fitting technique is used to determine the best set of values for the variables of numerous connection configurations. Using these results, an empirical equation was developed to describe the moment-rotation characteristics based on the end-plate configuration, bolt size, beam dimensions, and column dimensions. The results of this paper are eminent for the application of four-bolt unstiffened extended end-plates to semi-rigid connection philosophy. Although the moment-rotation plots given in the application examples included in the paper have decent correlation, the connection strength predicted by the method is off by as much as 75% in some cases. Bursi and Jaspart (1997a) [11] presented part one of a two-part investigation of finite element modeling of bolted connections. Unlike its companion paper (Bursi and Jaspart, 1997b) [6], this paper did not consider moment end-plate connections themselves. It did, however, present the results of which showed that finite element programs could be used to accurately predict the behavior of moment endplate connections. Hence, it is included here. Tee stub connections were first modeled to determine the accuracy and/or calibration required when using finite elements to model connection behavior. Using the LAGAMINE software package, the models were constructed using both hexahedron (more commonly called brick) and contact elements. The contact elements utilize what is called a penalty technique. Here, a value was chosen as a penalty parameter and this is similar to placing a spring between two bodies. Contact is simulated only for displacements within this given penalty value. Friction caused by the sliding and sticking between bodies is modeled with an isotropic Coulomb friction law. Nonlinear finite element analysis that considers large displacements, large rotations, and large deformations is used. Loads were applied using displacement as the controlling parameter. When considering the bolts, the

9 additional flexibility provided by the nut and threaded region of the bolt were taken into account by using an effective length of the bolt. Due to the symmetry of the tee stub connection, only a quarter of the connection was modeled. Preloading forces in the bolts were taken into account by using applied initial stresses. The material properties are modeled using piece-wise linear constitutive laws for the material from experimentally tested connections. For several of these experimentally tested connections, a finite element analyses were performed. The finite element results compared quite nicely to experimental results. There was a slight difference in deflection values at the onset of yielding, which was primarily due to the presence of residual stresses in the actual tee stubs which was neglected in the finite element models of these members. Bursi and Jaspart (1997b) [6] presented the second part of the two-part investigation by the authors. They used ABAQUS finite element code to analyze four-bolt unstiffened extended moment end-plate connections under static loading. The purpose of the study was to examine the stiffness and strength behavior of these connections. The finite element results were compared with those from an experimental study. End-plate rotation and bolt forces were both considered. The authors intent was to show the feasibility of using the finite element method via commercial codes to determine momentrotation characteristics of semi-rigid connections. Although dynamic characteristics of these connections was not considered, the authors did consider thin endplates mainly for their ability to behave in a ductile manner when plate yielding occurs. The finite element model considered by the authors was quite complex. The bolt and bolt head were modeled using beam elements. Both preloaded and non-preloaded bolts are considered, but only bolts in the tension region were included. The end-plate and beam elements were generated using eight-node brick elements that allow plasticity. Contact elements were used to describe the interaction between the end-plate and the rigid column flange. Around the bolt holes, nodes were constrained in the direction perpendicular to the face of the endplate. This assumption was used, as tests and other finite element studies

10 have shown that end-plates tended to pull away from the column flange even at the bolt locations. Other than friction forces taken care of by the contact elements, there were no lateral constraints mentioned in the paper. However, results were obtained even for the zero friction case, which should result in divergence due to a singular stiffness matrix. Thus it was assumed that some other boundary conditions were provided, but this was not discussed. By comparison with experimental results, the results indicated that the model predicts the end-plate moment-rotation characteristics quite accurately. However, the bolt forces were not recorded experimentally and no comparison is made. The bolt axial force versus beam flange force seems reasonable in the plots provided. Bursi and Jaspart (1998) [12] presented basically the same results as the paper discussed in this section and is not considered separately. Ribeiro et al. (1998) [13] discussed results of an experimental study of beamto-column moment end-plate connections. This study included testing of twelve cruciform built-up sections to validate design criteria used for rolled shapes for the design of built-up sections. Specimens were designed, specifically to check the method proposed by Krishnamurthy (1979) [4]. The following observations among others were made: (a) Applied moments were about 20% greater than the plastic moment capacities predicted. (b) The greater the bolt diameter, the greater was the influence of end-plate thickness. (c) Krishnamurthys method was found to be non-conservative. (d) Bolt rupture occurred in the tests in which the Krishnamurthy method Predicted otherwise. (e) Results involving the collapse modes of the specimen led to a hypothesis concerning prying forces which was not accepted by Krishnamurthy.

11 In 1998, Troup et al. (1998) [14] presented a paper describing finite element modeling of bolted steel connections. The ANSYS software was used for this study, which included an extended moment end plate model as well as a teestub model. The model utilized a bilinear stress-strain relationship for the bolts. Also, special contact elements were used between the end-plate and the column flange for the extended end-plate model, and between the tees for the tee model. By using the contact elements between the contact surfaces of the models, the geometric non-linearities that are present between the surfaces as separation occurs due to increased load can be realistically modeled. Both models were calibrated with experimental test data to show excellent correlation between analytical and experimental stiffness. Bolt forces were also analyzed. It was found that for the simple four-bolt arrangement about the tension flange, the tee design prediction was accurate. However, for more complex bolt patterns, the distribution of prying forces was not as clear. Troup, et al. (1998) [14] concluded the following: (a) Tee-stub analogy was a useful benchmark to provide an indication of the performance of analysis techniques. (b) Shell elements are more accurate for modeling beam and column sections. Thick endplate design could provide additional rotational stiffness and moment capacity but may result in bolt fracture. (c) Thin end plates could provide enough deformation capacity to allow semirigid connection design, but may result in excessive deflection. (d) The moment capacity prediction of Eurocode 3 had been shown to be reasonable, but conservative, for simple end-plate bolt configurations. However, the code is inaccurate when analyzing more complicated bolt arrangements. If these inaccuracies did not lead to bolt failure, then they might be acceptable. Maggi et al. (2004) [15] focused on the behavioral variations of bolted extended end plate connections due to changes in plate thickness and bolt diameter. It also discussed the application of FE model as tools to perform parametric analysis in

12 order to assess the accuracy of commonly used design procedures and to provide data for development of new analytical models.

2.2

CORRUGATED WEB BEAM AND EXTENDED END PLATE In this research, the focus is to obtain the M- characteristics of an extended end plate connection involving a corrugated web beam. A corrugated web beam is a built-up girder with a thin-walled, corrugated web and plate flanges. In this case, the profiling of the web attributes to its high load-bearing capacity at low design weight, which represents a particular economical solution for wider spans. See Figure 2.1 and 2.2.

Figure 2.1: Corrugated plate

Figure 2.2: Corrugated web beam

13 Nowadays, corrugated webs are used to allow the use of thin plates without stiffener for use in building and bridges. Thus, resulted in the reduction of the beam weight and cost. The early investigation of such beam is carried out by Elgaaly [16] and has been further developed to the practical stage. Most of these analytical and experimental studies concentrated on the trapezoidal vertically corrugated webs. Elgaaly et al [16] investigated the failure mechanism of these beams under shear, bending and compressive patch loads. It was found that the failure of beams under shear loading is due to the buckling on the web, where local buckling and global buckling occurred for coarse and dense corrugation respectively. Similarly under bending, the compression flange vertically buckled into the crippled web when the yield stress was reached. It was also found that the ultimate moment capacity could be calculated considering the flange and neglecting the web as its contribution to the beams moment carrying capacity was considered to be insignificant. Nevertheless, under compressive patch loads, two distinct modes of failure were observed. These involved the formation of collapse mechanism on the flange followed by the web crippling or yielded web crippled followed by vertical bending of the flange into the crippled web. The failure of these beams was found to be dependent on the loading position and also on the corrugation parameters where it can be a combination of the aforementioned modes. Zhang et al. [17] and Li et al [18] studied the influence of the corrugation parameters and developed a set of optimized parameters for the wholly corrugated web beams based on the basic optimization on the plane web beams. It was also found that the corrugated web beam had 1.5 2 times higher buckling resistance than the plane web beam. An extended end plate connection consists of a plate welded in the fabrication shop to the end of the steel beam as shown in Figure 2.3. The end

14 plate is pre-drilled and then bolted at site through corresponding holes in the column flange. The plate extends above the tension flange in order to increase the lever arm of the bolt group and subsequently the load carrying capacity. The connection is usually loaded by a combination of vertical shear force, axial force in the beam member and a moment as shown in Figure 2.4. The overall components of the beam to column connection are shown in Figure 2.5.

Figure 2.3: Extended End Plate Connection

Figure 2.4: Connection Loading

Figure 2.5: Components of beam to column connection The beam to column connection should have comparable properties to the structural steel. Relevant properties of steel are its strength, its stiffness and its ductility or deformation capacity. These properties can be demonstrated in a tensile test (see Figure 2.6). A well designed steel structure should possess the same good properties.

15

Figure 2.6: Strength, stiffness and deformation capacity of steel and connections The failure of such connection is shown in Figure 2.7 below.

Mode I: Complete flange yielding

Mode II: Bolt failure with flange yielding

Mode III: Bolt failure

Figure 2.7: Failure modes for bolted T-stub connections

2.3

CLASSIFICATION OF CONNECTIONS The structural properties of connection can also be presented in a M- diagram. In Figure 2.8(a), 2.8(b) & 2.8(c) below, shows a set of M- curves for connections with different types of behavior.

16

Figure 2.8(a): Moment-rotation curves of beam to column connections

Figure 2.8(b): Moment-rotation curves of beam to column connections

Figure 2.8(c): Moment-rotation curves of beam to column connections

17 For the use in plastic design, the connections can be classified in the following categories, namely; nominally pinned connections, full strength partial strength connections. connections and

Nominally pinned connections This type of connection is designed to transfer shear and normal force only. The rotation capacity of the hinge should be sufficient to enable all the plastic hinges necessary for the collapse mechanism to develop. Full strength connections (Connections A and B in Figure 2.8(c)) The moment capacity is greater than that of the member. A plastic hinge will not be formed in the connection but in the member adjacent to the connection. In theory, no rotation capacity is required for the connection. Partial strength connections (Connections C, D and E in Figure 2.8(c)) The moment capacity is less than that of the member. A plastic hinge will be formed in the connection, so sufficient rotation capacity is required. In elastic design, traditionally two categories of connections were considered: Nominally pinned connections (Figure 2.9) The connections are assumed to transfer only the end reaction of the beam (vertical shear force and eventually normal force) to the column.

Figure 2.9: Nominally pinned connections They should be capable of accepting the resulting rotation without developing significant moments, which might adversely affect the stability of the column. It is a common practice to design structures on a simply supported basis

18 and then to provide connections which are in effect semi-rigid. A typical example is the flush end plate as shown in Figure 2.10. This may be unsafe due to insufficient rotation capacity of the connection.

Figure 2.10: Flush end plate connection Rigid connection (Figure 2.11) Rigid connections are used to transfer moments as well as end reactions. Design assumes joint deformation to be sufficiently small that may influence the moment distribution and the structures deformation may be neglected.

Figure 2.11: Rigid Connections To fill the gap between pinned and rigid connections, a third category is defined and accepted in most modern codes. Semi rigid (semi flexible) connections (Figure 2.12) These connections are designed to provide a predictable degree of interaction between members based on actual or standardized design M- Characteristics of the joints.

19

Figure 2.12: Semi rigid (semi flexible) connections

2.4

MOMENT-ROTATION (M-) CHARACTERISTICS Because the flexural rigidity of each connection plays an important role in the behavior of the entire structural steel frame, most of the research on various connection types are focused on the investigation of moment-rotation relationships. For this purpose, many experimental tests have been conducted to obtain moment-rotation curves. Considering the moment-rotation curves obtained from experimental tests are available, a simplified analytical model is proposed in this project to predict the behavior of the connection by the application of Finite Element Analysis (FEA). The main structural elements of steel framed multi-storey structures are the columns, the beams and their connections. Conventionally the beam-tocolumn connections are considered to be either pinned or rigid. In the case of pinned or 'simple' connections, the frames have to be stabilized by appropriate bracing systems. Such frames are named braced frames by Eurocode 3. The term 'rigid' in this context implies that the connection is capable of resisting moments with a high stiffness, i.e., the connection flexibility has a negligible influence on the distribution of movements in the frame connections. When the connections are rigid, the overall stability may be provided by the frame itself without the inclusion of specific bracing systems. Although the idealisation of connection stiffness as pinned or rigid has been applied exclusively in the past it is generally recognized that the real behaviour

20 of the connections is never as ideal as assumed in the analysis (Figure 2.13). The two cases, pinned and fully rigid, actually represent extremes of connection behaviour. In reality, the connections behave somewhere between those limits, that is they behave as semi-rigid.

Figure 2.13: Characteristics of beam-to-column connections

A further classification of moment resisting connections relates to their strength. A 'full-strength' connection is a connection that can at least develop the bending strength of the elements it connects. A 'partial-strength' connection has a lower design strength than that of the elements it connects. The rotation capacity of a moment-resisting connection can also be important. For example a beam with partial-strength end connections can be designed plastically if the connection rotation capacity is sufficient to ensure the development of an effective hinge at midspan. For practical design situations the actual non-linear connection behaviour has to be approximated. The connection behaviour is characterised by its moment resistance MRd, its rotational capacity cd and its rigidity s = M/.

21 Figure 2.14 shows the moment/rotation diagram of a beam to column connection. For design purposes, the real connection behavior can be represented by a bi-linear diagram in which the following properties can be distinguished.

Figure 2.14: Moment-rotation diagram of beam-to-column connections

(a) The design resistance of the connection (b) The stiffness of the connection when subjected to small moments (c)The stiffness of the connection when subject to ultimate moments (d) The rotation capacity

22 2.5 ANALYSIS OF CONNECTIONS 2.5.1 EXPERIMENTAL SET-UP

Figure 2.15: Experimental set-up

23 The arrangement of the experimental set-up is as shown in the Figure 2.15 above. Load was applied near end of the cantilever beam and added progressively. Clinometer is attached at the beam and column as shown. It gives rotation reading for each beam and column. The differences between these two rotation values, will give the value of rotation, needed for plotting the M- curve. The details of the connection are shown in Figure 2.16 below.

Figure 2.16: Details of beam-to-column connections

24 2.5.2 FINITE ELEMENT METHOD (FEM) In the finite element method, the response of a complex shape to any external loading, can be calculated by dividing the complex shape into lots of simpler shapes. These are the finite elements that give the method its name. The shape of each finite element is defined by the coordinates of its nodes. Adjoining elements with common nodes will interact.

(a) Definition (i) FEM is a numerical procedure of finding solution to a complicated problem establishing the response of interconnected elements of finite dimensions with continuity and equilibrium considerations (Desai 1985) (ii) FEM is a computer-aided mathematical technique for obtaining approximate numerical solutions to the abstract equations of calculus that predict the response of physical systems subjected to external influences (Burnet 1998) (iii) FEM is a numerical method for solving problems of engineering and mathematical physics which include structural analysis, heat transfer, fluid flow, mass transport and electromagnetic potential (Logan 1981)

(b) The basic concept In the finite element method, the structure under consideration is divided into smaller zones, known as elements. The elements are assumed to be connected to each other at certain points (usually at the corners) called nodes. It is at the nodes that we compute the displacements. Thus the body with infinite number of degrees of freedom is approximated by a body having degrees of freedom equal to two or three times the number of nodes. It is obvious, though there are rigorous mathematical proofs available, that as the number of nodes is increased, a better (closer to the exact) solution is obtained. The displacements at any point within an element are related to the

25 displacements at the nodes by making certain assumptions. Displacements are fundamental variables. From the displacements, the strains can be obtained, and then using the stress-strains relationships, the stresses can be calculated.

(c) Solution process The process for solving a problem using the finite element method involves six major steps: Step 1. Establish governing equations and boundary conditions. In order to generate a valid approximate solution to a problem, the differential equation that governs the behavior and the corresponding boundary conditions for the problem must be determined. Once this is done the appropriate finite element formulation can be used to generate the solution. Step 2. Divide solution domain into elements. In this step, the entire solution domain is subdivided into small elements. Care is taken to make sure that enough elements are included to capture the behavior of the solution over the entire domain. Areas of particular interest and care are locations where critical values are expected, locations with large stress gradients, locations where the geometry changes suddenly, locations where boundary conditions and loads are applied. Typically, the larger the number of elements the better the approximation of the solution to the differential equation. Step 3. Determine element equations. Once the elements are formed, the algebraic equations to be solved are developed for each individual element. The form of the algebraic equations for every element will be the same. Differences

26 from one element to the next will be due to changes in element size and properties. This is the power of the finite element method, the equations can be written once for a general element then they only need to be modified to reflect particular elements geometry and properties. Step 4. Assembly of global equations. Once all the element equations are generated, they are put together to form a system of equations for the entire solution domain. Step 5. Solution of global equations. This system of equations is solved for the value of the dependent variable in the original differential equation at discreet points throughout the solution domain. Depending on the problem types there may be hundreds, thousands, tens of thousands, or even hundreds of thousands of points at which the solution to the differential equation is approximated. Step 6. Solution verification. The accuracy of the solution must be verified before the results can be considered valid. One way to do this is to refine the mesh (increase the number of elements) and rerun the analysis. If the value of the dependent variable at the discreet points in the mesh does not change significantly as the mesh is refined, the solution is deemed to be accurate. (d) Joint Element Joint elements may be introduced into the structural idealisation in order to model releases, springs or restraints between any two nodes in arbitrary directions. Joint elements are available for use in two and three dimensions and comprise a range of nonlinear material and boundary condition models. These nonlinear models enable the realistic modelling of hardening elastoplastic compressive and tensile joint behaviour as well as contact and friction types of nonlinear boundary condition.

27

k - spring constant

node node

local coordinate direction

local nodal force f 2x d 2x deg ree of freedom

Uniaxial Bar

k = AE/L

(e) Decision in finite element modelling Finite element analysis in the forms of computer software packages has now been made available. These products are presented and displayed very impressively and allow interactive modelling and checking, with colour graphics, windowing, etc. With the development of the sophisticated software plus a decrease in the price of hardware, it has now become a competitive business amongst the software suppliers. Some of the factors that govern the choice of this software are namely: (a) (b) (c) (d) (e) (f) (g) Availability Degree of sophistication Limitations Ease of use Accuracy Special features Costs

Currently, the complete FEA software packages available are; LUSAS, ANSYS, COSMOS-M, PAFEC, IMAGES-3D, GTSTRUDL, SAP80/90, FESDEC, SUPERSAP, GIFTS, ESDUFINE, etc.

28 In this research, LUSAS 13.5 [1] software is being used for the numerical analysis. This is mainly due to its availability, ease of use, accuracy and cost.

(f) LUSAS Version 13.5 A complete finite element analysis of LUSAS involves three stages: i) ii) iii) Pre-Processing Finite Element Solver Results-Processing

i) Pre-Processing Pre-processing involves creating a geometric representation of the structure, then assigning properties, then outputting the information as a formatted data file (.dat) suitable for processing by LUSAS.

Creating a Model A created model in LUSAS is a graphical representation consisting of Geometry (Points, Lines, Combined Lines, Surfaces and Volumes) and Attributes (Mesh, Geometric, Materials, Support, Loading). Each part of the model is created in two steps: First, Define the feature or attribute, and second, Assign the attribute or attributes. ii) Finite Element Solver Once a model has been created, on the solve button is clicked to begin the solution stage. LUSAS creates a data file from the model, solves the stiffness matrix, and produces a result file (.mys). The results file will contain some or all of the following data: Stresses, Strains, Displacements, Velocities, Accelerations, Residuals, Reactions, Yield flags, Potentials, Fluxes, Gradients, Named variables, Combination datasets, Envelope definitions, Fatigue datasets and Strain energy.

29 iii) Results-Processing Results-processing involves using a selection of tools for viewing and analyzing the result file produced by the Solver. Many different ways of viewing the results are available: Contour plots (averaged/smoothed), Contour plots (unaveraged/unsmoothed), Undeformed/Deformed Mesh Plots, Wood-Armer Reinforcement Calculations, Animated Display of Modes/Load Increments, Yield Flag Plots, Graph Plotting, Vector Plots.

30

CHAPTER 3

RESEARCH METHODOLOGY

3.1

LUSAS SOFTWARE LUSAS is an associative feature-based Modeller. The model geometry is entered in terms of features which are sub-divided (discretised) into finite elements in order to perform the analysis. Increasing the discretisation of the features will usually result in an increase in the accuracy of the solution, but with a corresponding increase in solution time and disk space required. The features in LUSAS form a hierarchy that is Volumes are comprised of Surfaces, which in turn are made up of Lines or Combined Lines, which are defined by Points.

3.1.1 FINITE ELEMENT MODEL LUSAS FEA software was used for the finite element analysis. The FEA models were created using command files rather than the CAD interface tools even though this method was longer and initially tedious. The command file could simply be copied and edited. The command file also was more logical in order than command files produced by the software after a model has been created. The command file was also well described by 6 comments within the file to provide a complete history of the model creation. FEA models can often be a black box that provides answers without the user being fully aware of what the model exactly entails. The extra work in creating the command files has been well worth the effort and allowed

31 the subsequent models to be created quickly. The technique of FEA lies in the development of a suitable mesh arrangement. The mesh discretisation must balance the need for a fine mesh to give an accurate stress distribution and reasonable analysis time. The optimal solution is to use a fine mesh in areas of high stress gradients and a coarser mesh in the remaining areas.

3.1.2 ELEMENT TYPES Four element types were used as shown in Figure 3.1 (as used by Jim Butterworth [19]). HX8M elements are three dimensional solid hexahedral elements comprising 8 nodes each with 3 degrees of freedom. Although the HX8M elements are linear with respect to geometry, they employ an assumed internal strain field which gives them the ability to perform as well as 20 noded quadratic isoparametric elements. These elements are used to model the beam flanges, end plate and connecting column flange. QTS4 elements are three dimensional flat facet thick shell elements comprising either 3 or 4 nodes each with 5 degrees of freedom and are used to model the beam (web and flange) and column (web and column back flange). JNT4 elements are non-linear contact gap joint elements and are used to model the interface between the end plate and the column flange. The bolts will be modeled by using BRS2 elements for the bolt shank and HX8M elements for the head and nut as shown in Figure 3.2. BRS2 are three dimensional bar elements comprising 2 nodes each with 3 degrees of freedom. Each BRS2 element is connected to the appropriate HX8M bolt head and nut to comprise the complete bolt assembly. All bolts used were M20 grade 8.8 and were assigned an area of 245mm2 which is equal to the tensile stress area. The bolt holes were modeled as a square cut-out in the end plate and column flange. Figure 3.3(a), 3.3(b) and 3.3(c) shows the FEA model with the arrangement of mesh discretisation, whereas the following Figures 3.3(d) and 3.3(e) show the relevant attributes forms.

32

Figure 3.1: Element types

Figure 3.2: Enlarged FEA Bolt Arrangement

Appendix A shows the properties of the element types used in this study.

Figure 3.3 (a): Line Mesh

33

Figure 3.3 (b): Mesh Discretisation

Figure 3.3(c): Mesh BRS2 and JNT4

34

Figure 3.3(d): Attribute forms

35

Figure 3.3(e): Attribute forms

3.1.3 NON-LINEAR ANALYSIS Material non-linearity occurs when the stress-strain relationship ceases to be linear and the steel yields and becomes plastic. The three sets of material data will be as follows: For the elastic dataset all elements are defined as elastic isotropic with a Youngs Modulus of Elasticity of 2.09 x 105 N/mm2 and Poissons lateral to longitudinal strain ratio of 0.3. The actual materials test certificates were obtained for all steel and enabled stress/strain curves to be based on actual values rather than theoretical Tensile tests records on a selection of bolts were available to enable the material properties used to be as accurate as possible. Von Mises yield criteria was used for all material. Figure 3.4 shows the relevant plastic attributes used in the model.

36

Figure 3.4: Attribute form

3.1.4 BOUNDARY CONDITIONS Displacements of the nodes in the X, Y and Z directions were restrained at the bottom of the column. Whereas at the top of the column the displacement of all nodes in the X and Z direction were restrained. The FEA model would have problems converging when the beam end plate had no supports restraining movement in the Y direction due to the lack of bending resistance in the bolt BRS2 elements. Therefore supports were added to the underside of the end plate. This removed the shear force from the bolts but not of course from the remaining connection elements. Shear in moment connections is usually of minor importance but it is felt that the supports are a compromise. The column flange to end plate interface was modeled by using JNT4 joint elements with a contact spring stiffness K of 0.1 kN/mm whereas, the bolt to end plate interface used JNT4 with contact spring stiffness K of 1 kN/mm. An initial point load of -20 kN was placed at 1300 mm from the column face. The load was then factored in the control file to achieve the required range of connection bending moments. Figure 3.5 shows the attribute form for loads, whereas Figure 3.6 shows FEA supports and loading.

37

Figure 3.5: Attribute form

Figure 3.6: FEA Supports and Loading

38

3.1.5 SUMMARY OF LUSAS FINITE ELEMENT SYSTEM The following chart shows the analysis processes involved in LUSAS; Initialize Model

Feature Geometry

Preparing Model Attributes, Define mesh, geometry, material properties, supports and loading

PreProcessing (MYSTRO)

Assign to Features Check Data Input

Analysis Create data file Save Model Run Analysis

NO

Successful

Post Processing

39

CHAPTER 4

RESULTS AND DISCUSSIONS

This chapter contains the results of the non-linear as well as linear analyses involving connections containing corrugated web as well as plain web (for the sake of comparison). LUSAS does not provide moment as well as rotation values. So, steps were taken to calculate these values using Excell Spreadsheet and using the loads and node displacements data extracted from LUSAS.

Moment-Rotation Curve Calculation For displacement analysis, the selected nodes are; CORRUGATED WEB BEAM Nonlinear analysis 29054 (0,300,-183.2) dz 35037 (0,200,100) dy Linear analysis 43153 dy PLAIN WEB BEAM Nonlinear analysis 37262 (0,300,-183.2) dz 42890 (0,200,100) dy Linear analysis 56670 dy

Node 29054 and node 37262 (See Figure 4.4 (a)) is located along the column centre line 100 mm above the intersection point between the beam and column centroidal axes. Whereas, Node 35037 and node 42890 (See Figure 4.4 (b)) is located along beam centre line 100 mm from the column face. All the 4 nodes are used for

40 nonlinear analysis. For linear analysis, nodes 43153 and node 56670 selected, were located at the cantilever end (to plot load-deflection curve). The LUSAS software doesnt have the capability to produce moment-rotation curve numerically, thus it has to be done manually. For the applied moment, it is calculated by using the following formulae; M = Total load factor * 20 * 1.3 20 kN is the initial point load and it is located 1300 mm @ 1.3m from the column face. The joint connection, rotation j is the difference of beam rotation b and column rotation c. It can be shown by the formulae below; j = b - c The unit of is in radian and the displacement is in milimetres. b = Tan-1 (y / 100 mm) c = Tan-1 (z / 100 mm) y is the vertical displacement of the node selected from the centre of rotation, whereas z is the horizontal displacement from the centre of rotation. 100 mm is the distance of the node (inclinometer position) from the centre of rotation. Table 4.1 shows the results of moment-rotation for the extended end plate connection containing corrugated web beam with increasing loads. Table 4.2 shows the moment-rotation for the connection using plain web beam. These results are plotted as shown in Figure 4.1. The M- graph obtained from the experimental testing is also plotted for the purpose of comparison.

41

Moment-Rotation Curve
300.000 250.000
Moment (kNm)

200.000 150.000 100.000 50.000 0.000 -10 -50.000 Rotation (mRad) 0 10 20 30 40 50

Experimental results

Element Analysis (FEA) for results Finite Graph 1: Experimental results M-corrugated curve web beam

Finite Element Analysis (FEA) plain web beam

Figure 4.1: Comparison of M- Curve between Experimental and FEA results (Nonlinear Analysis)

Moment-Rotation Curve for Test EEP-1

200.00

150.00

Moment, (kNm)

100.00

50.00

Reading 1

Reading 2

-5.000

0.00 0.000

5.000

10.000

15.000

20.000

25.000

30.000

-50.00

Rotation, miliradians(mRad)

Figure 4.2: Experimental graph for Moment-Rotation Curve (N5 specimen) in determining Moment capacity, MR of the connection.

42 Table 4.1: Corrugated web beam data (nonlinear analysis)

teta dy 0 -0.12868 -0.3861 -0.78267 -1.84377 -2.03647 -2.20023 -2.41193 -2.63961 -2.86489 -3.10805 -3.38192 -3.76746 -4.31824 -5.11394 -6.28785 t l factor 0 1 3 5.828427 8.656854 8.944215 9.142224 9.346859 9.516247 9.651213 9.773046 9.891128 10.032 10.2067 10.43211 10.72996 dz 0 0.052729 0.158208 0.309772 0.463158 0.49033 0.517218 0.555156 0.59956 0.647102 0.701902 0.773224 0.871815 1.012284 1.220651 1.536855 beam 0 0.001287 0.003861 0.007827 0.01844 0.020368 0.022006 0.024124 0.026402 0.028657 0.03109 0.033832 0.037692 0.043209 0.051184 0.062962 teta column 0 0.0005273 0.0015821 0.0030977 0.0046316 0.0049033 0.0051722 0.0055516 0.0059957 0.0064711 0.0070191 0.0077324 0.0087184 0.0101232 0.0122071 0.0153698 rotation mrad 0 0.75954979 2.27893643 4.72911635 13.8081448 15.4642112 16.8336702 18.5723162 20.4065524 22.1856095 24.0713575 26.0996631 28.9740984 33.0860943 38.9768944 47.59176 moment kNm 0 26 78 151.5391052 225.0782105 232.5496021 237.6978271 243.0183356 247.4224099 250.9315271 254.0992023 257.1693346 260.8320711 265.374231 271.2347509 278.9788722

Table 4.2: Plain web beam data (nonlinear analysis)

teta dy 0 -0.30643 -0.94769 -1.21765 -1.7556 -2.35666 -2.80337 -3.21119 -3.58685 -4.65843 -4.89024 t l factor 0 1 3 3.5 4 4.183545 4.242733 4.290365 4.332549 4.447822 4.472146 dz 0 0.127602 0.38794 0.4591866 0.7571551 1.0377382 1.150645 1.2497375 1.3432876 1.628811 1.693724 beam 0 0.003064 0.009477 0.012177 0.017558 0.023571 0.028041 0.032123 0.035884 0.046618 0.048941 teta column 0 0.001276 0.003879 0.004592 0.007572 0.010378 0.011507 0.012498 0.013434 0.01629 0.016939 rotation mrad 0 1.7882848 5.5977875 7.5851813 9.986142 13.193229 16.534047 19.624951 22.45021 30.328473 32.002575 moment kNm 0 26 78 91 104 108.7721613 110.3110558 111.5494839 112.6462763 115.643377 116.2757908

43

Moment-Rotation Curve
300

250

moment (kNm)

200

150

100

50

0 0 5 10 15 20 25 30 35 40 45 50 Rotation (mRad)

Figure 4.3: Finite Element Analysis graph for Moment-Rotation Curve (N5 specimen) in determining Moment capacity, MR of the connection.

Figure 4.4(a): Position of nodes selected in determining the displacement

Figure 4.4(b): Position of nodes selected in determining the displacement

44 These M- curves plotted are then used to determine the moment capacity, MR of the connection. It is obtained by first, drawing the tangent line to the graph, second, the angle of the curve is divided into two. From that, another line is drawn parallel to that line. The intersection of vertical and horizontal tangent line is the moment capacity of the connection. Figures 4.2 and 4.3 shows the way to calculate the moment capacities, MR of test and LUSAS data of the extended end plate connection containing corrugated web beam. The experimental and LUSAS value of MR are 100 kNm and 150 kNm respectively. A difference of 50%. From Figure 4.1, it can be observed that all three graphs behaves in a similar manner with increasing load. Initially, at small loading, the three graphs are straight (indicating that the connections are elastic). At a rotation of about 3 mrad, the three graphs curve showing that they become plastic. As indicated earlier, the difference between moment of resistance of LUSAS and test is 50%. The use of plain web beam, however, produces a moment of resistance MR of 75 kNm which is 25% lower than the experimental value. But, from the LUSAS moment of resistance result shown, the MomentRotation curve is 1.5 times the value of the experimental Moment-Rotation curve as well as its moment capacity (150 kNm vs 100 kNm). Thus, the model may not be sufficient to validate the experimental result or being the alternative method in replacing the actual case. It is believed, that the input data may not accurate in analyzing the model. Proper material testing should be carried out to get the actual data. The stiffness value for the contact elements, which plays an important role in the semi rigid connection behavior, was obtained by trial and error to get to the best moment-rotation curve as closely as possible to the experimental curve. And it was found that to get the closest graph of moment-rotation curve as the experimental, the value of K = 1.0 kN/mm for the contact interface between the bolt and the end plate, whereas, between the column face and the end plate, a spring stiffness value of 0.1 kN/mm should be used. It took 1.5hrs 2 hrs to run each model.

45 Also, the three M- curves from Figure 4.1 shows that the use of corrugated web beam produces a stiffer connection compared to that of using plain web beam. All the three curves indicate that they can be classified as semi rigid connections. Figure 4.5 and Figure 4.6 shows the test and LUSAS deformed shape of the connection respectively. The effect of the various combinations of K values for the contact interface between the bolt and the end plate and between the column face and the end plate on the shape of the M- curves can be seen in Appendix B (Variations in K).

Figure 4.5: N5 specimen after failure N5

Figure 4.6: FEA deformed mesh for

46

CHAPTER 5

CONCLUSION AND RECOMMENDATION FOR FUTURE WORK

5.1

CONCLUSIONS The M- curve obtained from the finite element results are in accordance with the experimental results. But the moment of resistance MR of LUSAS is 1.5 times the value of the experimental moment of resistance MR. Thus it shows that the model may not sufficiently accurate to obtain a good MR result for the extended end plate connection.

5.2

RECOMMENDATION FOR FUTURE WORK The following recommendation can be useful for future investigations: a) Proper material testing should be carried out to determine the actual Youngs Modulus, uniaxial yield stress, hardening gradient slope, plastic strain, etc. Thus, more accurate information can be input into the software. b) To get better results, whenever possible, the finite element mesh should be relatively uniform. Special caution should be exercised in transition from coarse to finer mesh. The aspect ratio between the elements longest and shortest dimensions should not be excessive. The optimum aspect ratio is close to unity. Illegal element shapes must be avoided. For triangular elements angles less than 30% are not desirable.

47 c) Since large stiffness variations between elements can lead to an illconditioned stiffness matrix of the total structure, rendering meaningless result, such conditions must be avoided by all means. d) The beam can be modeled with other type of corrugation, like; horizontal one arc corrugation, horizontal two arcs corrugation and vertical arcs corrugation, instead of trapezoidal corrugation. The corrugated beam can then be compared with the plain web beam. e) Different types of meshing can be compared for the same type of model to see the differences. f) Different type of end plate thickness can be modeled to see the connection behaviour.

48

REFERENCES

1. 2.

FEA Ltd, LUSAS: Modeller User Manual, Version 13, United Kingdom. Bose B, Sarkar S, and Bahrami M, Finite Element Analysis of unstiffened extended end plate connections, Structural Engineering Review, 3, 211-224, 1991.

3.

Bose B, Youngson G K, and Wang Z M, An appraisal of the design rules in Eurocode 3 for bolted end plate joints by comparison with experimental results, Proceedings from the Institute of Civil Engineers Structures and Buildings, 1996.

4.

Krishnamurty, N. (1976), Correlation between 2 & 3-D Finite Element Analysis of Steel Bolted End-Plate Connections, Computers and Structures, 6, 381-389.

5.

Bursi, O.S. and Lionelli, L. (1994), A Finite Element Model for the Rotational Behavior of End Plate Steel Connections, Proceedings of the SSRC Annual Technical Session, Structural Stability Research Council, Bethlehem P.A, 163-175.

6.

Bursi, O.S., and Jaspart, J.P. (1997b), Calibration of a Finite Element Model for Isolated Bolted End-Plate Steel Connections, Journal of Constructional Steel Rersearch, 42, 225-262.

7.

Richard, R.M., and Abbott, B.J., (1975), Versatile Elastic-Plastic StressStrain Formula, Journal of the Engineering Mechanics Division, ASCE, 101, 511-515.

49 8. Gebbeken, N. Rothert, H. and Binder, B. (1994), On the Numerical Analysisi of Endplate Connections, Journal of Constructional Steel Research, 30, 177-196. 9. Rothert, H., Gebbeken, N. and Binder, B. (1992), Nonlinear ThreeDimensional Finite Element Contact Analysis of Bolted Connections in Steel Frames, International Journal for Numerical methods in Engineering, 34, 303-318. 10. Sherbourne, A.N and Bahaari, M.R. 91997), Finite element Prediction of End Plate Bolted Connection Behavior. I: Parametric study, Journal of Structural Engineering, ASCE, 123, 157-164. 11. Bursi, O.S. and Jaspart, J.P. (1997a), Benchmarks for Finite Element Modeling of Bolted Steel Connections, Journal of Constructional Steel Research, 42, 17-42. 12. Bursi, O.S, and Jaspart, J.P. (1998), basic Issues in the Finite Element Simulation of Extended End Plate Connections, Computers and Structures, 69, 361-382. 13. Ribeiro, L.Calado, C.A. Castiglioni, C. Bernuzzi, Stability and Ductility of Steel Structures Edited by T.Usami and Y.Itoh, Elsevier Science Ltd. 1998, pp279-292, Behavior of Steel Beam-To-Column Joints Under Cyclic Reversal Loading: An Experimental Study 14. Troup, S. Xiao, R.Y. and Moy, S.s.J. (1998), Numerical Modeling of Bolted Steel Connections Journal of Constructional Steel Research, 46, Paper No. 362. 15. Y.I.Maggi, R.M.Goncalves, R.T.Leon, L.F.L.Ribeiro, Parametric analysis of steel bolted end plate connections using finite element modeling, Journal of Constructional Steel Research 61, Elsevier, 2005, pg 689-708

50 16. Elgaaly M, Hamilton RW, Seshadri A., Shear Strength of beams with corrugated webs, Journal of Structural Engineering ASCE 1996; 122(4):3908. 17. Zhang W, Li Y, Zhou Q, Qi X, Widera GEO. Optimization of the structure of an H-beam, with either a flat or a corrugated web. Part 3: Development and research on H-beams with wholly corrugated webs, Journal of Materials Processing Technology 2000;101(1):119-23. 18. Li Y, Zhang W, Zhou Q, Qi X, Widera GEO, Buckling strength analysis of the web of a WCW H-beam: Part 2. Development and research on H-beams with wholly corrugated webs (WCW), Journal of Materials Processing Technology 2000;101(1):115-8. 19. Jim Butterworth, Finite Element Analysis of Structural Steelwork Beam to Column Bolted Connections, Constructional Research Unit, School of Science & Technology, University of Teeside, UK

51

APPENDIX A (LUSAS Element Types)

Element Name Element Group Element Subgroup JNT4 Joints 3D Joints

Element Description A 3D joint element which connects two nodes by three springs in the local x, y and z-directions. Use JL43 for semiloof shell corner nodes. Number Of Nodes and local xy-plane. Freedoms U, V, W: at nodes 1 and 2 (active nodes). X, Y, Z: at each node. Node Coordinates Geometric Properties Not applicable. Material Properties Linear Not applicable. Matrix Stiffness: MATRIX PROPERTIES STIFFNESS 6 K1,..., K21 element stiffness matrix (Not supported in LUSAS Modeller) Mass: MATRIX PROPERTIES MASS 6 M1,..., M21 element mass matrix (Not supported in LUSAS Modeller) Damping: Joint MATRIX PROPERTIES DAMPING 6 C1,..., C21 element damping JOINT PROPERTIES 3 (Joint: 3/Stiffness) JOINT PROPERTIES GENERAL 3 (Joint: 3/General) matrix (Not supported in LUSAS Modeller) Standard: Dynamic general: 4. The 3rd and 4th nodes are used to define the local x-axis

Elasto-plastic: JOINT PROPERTIES NONLINEAR 31 3 (Joint: 3/Elasto-Plastic) Elasto-plastic: JOINT PROPERTIES NONLINEAR 32 3 (Joint: 3/Asymmetric) Nonlinear contact: Gap) JOINT PROPERTIES NONLINEAR 33 3 (Joint: 3/Initial

52 Nonlinear friction: Concrete JOINT PROPERTIES NONLINEAR 34 3 (Joint: 3/Frictional)

Not applicable.

Elasto-Plastic Not applicable. Rubber Not applicable. Composite Not applicable. Not applicable. Field Not applicable. Stress Potential Damage Viscoelastic Loading Prescribed Value Element Loads Distributed Loads Body Forces CBF Wz, ax, ay, az BFP, BFPE Velocities Not applicable. VELO Velocities. Vx, Vy, Vz: at nodes. Initial stresses/strains at nodes/for element. Fx, PDSP, TPDSP Prescribed variable. U, V, W: at active nodes. Concentrated loads. Px, Py, Pz: at active nodes. Not applicable. Not applicable. Constant body forces for element. Xcbf, Ycbf, Zcbf, Wx, Wy, Concentrated Loads CL Creep Not applicable. Not applicable. Not applicable.

Accelerations ACCE Accelerations. Ax, Ay, Az: at nodes. Initial Stress/Strains SSI, SSIE SSIG Not applicable. Residual Stresses Not applicable. Temperatures TEMP, TMPE Temperatures at nodes/for element. T1, T2, T3, T1o, T2o, T3o: actual and initial spring temperatures. Field Loads Not applicable. Not applicable. Temp DependentLoads Fy, Fz: spring forces in local directions. ex, ey, yz: spring strains in local directions.

53 Element Name Element Group Element Subgroup BRS2 Bars Structural Bars

Element Description Straight and curved isoparametric bar elements in 3D which can accommodate varying cross sectional area. Number Of Nodes Freedoms Node Coordinates Geometric Properties A1 ... An Cross sectional area at each node. 2 or 3. X, Y, Z at each node. U, V, W at each node.

Material Properties Linear Isotropic Joint MATERIAL PROPERTIES (Elastic: Isotropic) Matrix Not applicable. Not applicable. Not applicable. Not applicable. Concrete

Elasto-Plastic Stress resultant

Tresca: MATERIAL PROPERTIES NONLINEAR 61 (Elastic: Isotropic, Plastic: Tresca, Hardening: Isotropic Hardening Gradient, Isotropic Plastic Strain or Isotropic Total Strain) Drucker-Prager: Mohr-Coulomb: MATERIAL PROPERTIES NONLINEAR 64 (Elastic: MATERIAL PROPERTIES NONLINEAR 65 (Elastic: Isotropic, Plastic: Drucker-Prager, Hardening: Granular) Isotropic, Plastic: Mohr-Coulomb, Hardening: Granular with Dilation) Von Mises (B/Euler): MATERIAL PROPERTIES NONLINEAR 75 (Elastic: Isotropic, Plastic: Von Mises, Hardening: Isotropic & Kinematic) Volumetric Crushing: Not applicable. Rubber Not applicable. Composite Not applicable. Field Not applicable.

54 Stress Potential Creep Damage Viscoelastic Loading Prescribed Value Element Loads Distributed Loads Body Forces CBF Wz, ax, ay, az BFP, BFPE Zcbf Velocities VELO Velocities. Vx, Vy, Vz at nodes. Initial stresses/strains at nodes/for element. Accelerations ACCE Acceleration Ax, Ay, Az at nodes. Initial Stress/Strains SSI, SSIE (1) Resultants (linear material models): Fx , ex (2) Components (nonlinear material models): 0, 0, sx , ex SSIG Initial stresses/strains at Gauss points. (1) Resultants (linear material models): Fx , ex (2) Components (nonlinear model): 0, 0, sx , ex Residual Stresses SSR, SSRE Not applicable. SSRG Residual stresses at Gauss points. Components (nonlinear material models): 0, 0, sx Temperatures TEMP, TMPE Temperatures at nodes/for element. T, 0, 0, 0, To, 0, 0, 0 in local directions. Field Loads Not applicable. Not applicable. Temp DependentLoads Body force potentials at nodes/for element. 0, 0, 0, 0, Xcbf, Ycbf, PDSP, TPDSP Prescribed variable. U, V, W at each node. Concentrated loads. Px, Py, Pz at each node. Not applicable. Not applicable. Constant body forces for element. Xcbf, Ycbf, Zcbf, Wx, Wy, Concentrated Loads CL STRESS POTENTIAL VON_MISES

(Isotropic: von Mises, Modified von Mises) CREEP PROPERTIES (Creep) DAMAGE PROPERTIES SIMO, OLIVER (Damage) VISCO ELASTIC PROPERTIES

55 Element Name Element Group Element Subgroup HX8M 3D Continuum Solid Continuum

Element Description A 3D isoparametric solid element with an incompatible strain field. This mixed assumed strain element demonstrates a much superior performance to that of the HX8 element. Number Of Nodes Freedoms 8. The element is numbered according to a right-hand screw rule in the local z-direction. U, V, W: at each node. X, Y, Z: at each node. Node Coordinates Geometric Properties Not applicable. Material Properties Linear Isotropic: Orthotropic: Orthotropic Solid) Anisotropic: MATERIAL PROPERTIES ANISOTROPIC SOLID (Elastic: Anisotropic Solid) Rigidities. Joint Not applicable. Matrix Not applicable. Not applicable. MATERIAL PROPERTIES NONLINEAR 82 (Elastic: Concrete MATERIAL PROPERTIES (Elastic: Isotropic) MATERIAL PROPERTIES ORTHOTROPIC SOLID (Elastic:

Isotropic, Plastic: Cracking concrete)MATERIAL PROPERTIES NONLINEAR 84 (Elastic: Isotropic, Plastic: Cracking concrete with crushing) Elasto-Plastic Stress resultant: Not applicable. Tresca: MATERIAL PROPERTIES NONLINEAR 61 (Elastic: Isotropic, Plastic: Tresca, Hardening: Isotropic Hardening Gradient, Isotropic Plastic Strain or Isotropic Total Strain)

56 Drucker-Prager: Mohr-Coulomb: MATERIAL PROPERTIES NONLINEAR 64 (Elastic: MATERIAL PROPERTIES NONLINEAR 65 (Elastic:

Isotropic, Plastic: Drucker-Prager, Hardening: Granular) Isotropic, Plastic: Mohr-Coulomb, Hardening: Granular with Dilation) Von Mises (B/Euler): MATERIAL PROPERTIES NONLINEAR 75 (Elastic: Isotropic, Plastic: Von Mises, Hardening: Isotropic & Kinematic) Volumetric Crushing: MATERIAL PROPERTIES NONLINEAR 81 (Volumetric Crushing or Crushable Foam) Rubber Ogden: MATERIAL PROPERTIES RUBBER OGDEN (Rubber: Ogden) Mooney-Rivlin: MATERIAL PROPERTIES RUBBER MOONEY_RIVLIN (Rubber: Mooney-Rivlin) Neo-Hookean: MATERIAL PROPERTIES RUBBER NEO_HOOKEAN (Rubber: Neo-Hookean) Hencky: MATERIAL PROPERTIES RUBBER HENCKY (Rubber: Hencky) Isotropic MATERIAL PROPERTIES NONLINEAR 87 Generic Polymer Composite

(Generic Polymer Model) Not applicable. STRESS POTENTIAL VON_MISES, HILL, Field Not applicable. Stress Potential HOFFMAN (Isotropic: von Mises, Modified von Mises Orthotropic: Hill, Hoffman) Creep Damage Viscoelastic Loading Prescribed Value Element Loads Distributed Loads FLD PDSP, TPDSP Prescribed variable. U, V, W: at each node. Concentrated loads. Px, Py, Pz: at each node. Not applicable. UDL Not applicable. Concentrated Loads CL CREEP PROPERTIES (Creep) DAMAGE PROPERTIES SIMO, OLIVER (Damage) VISCO ELASTIC PROPERTIES

Face Loads. Px, Py, Pz: local face pressures at nodes.

57 Body Forces CBF Wz, ax, ay, az BFP, BFPE Zcbf Velocities VELO Velocities. Vx, Vy, Vz: at nodes. Initial stresses/strains at nodes/for element. sx, Accelerations ACCE Acceleration Ax, Ay, Az: at nodes. Initial Stress/Strains SSI, SSIE sy, sz, sxy, syz, sxz: global stresses. ex, ey, ez, gxy, gyz, gxz: global strains. SSIG Initial stresses/strains at Gauss points sx, sy, sz, sxy, syz, sxz: global stresses. ex, ey, ez, gxy, gyz, gxz: global strains. Residual Stresses SSR, SSRE Residual stresses at nodes/for element. sx, sy, sz, sxy, syz, sxz: global stresses. SSRG Residual stresses at Gauss points. sx, sy, sz, sxy, syz, sxz global stresses. Temperatures TEMP, TMPE Temperatures at nodes/for element. T, 0, 0, 0, To, 0, 0, 0 Field Loads Not applicable. Not applicable. Temp DependentLoads Body force potentials at nodes/for element. 0, 0, 0, 0, Xcbf, Ycbf, Constant body forces for element. Xcbf, Ycbf, Zcbf, Wx, Wy,

Element Name Element Group Element Subgroup

QTS4 Shells Thick Shells

Element Description A family of shell elements for the analysis of arbitrarily thick and thin curved shell geometries, including multiple branched junctions. The quadratic elements can accommodate generally curved geometry while all elements account for varying thickness. Anisotropic and composite material properties can be defined. These degenerate continuum elements are also capable of modelling warped configurations. The element formulation takes account of membrane, shear and flexural deformations. The quadrilateral elements use an assumed strain field to define transverse shear which ensures that the element does not lock when it is thin (see Notes).

58 Number Of Nodes Freedoms 3, 4, 6 or 8 numbered anticlockwise.

Default: 5 degrees of freedom are associated with each node U, V, W,

qa, qb. To avoid singularities, the rotations qa and qb relate to axes defined by the orientation of the normal at a node, see Thick Shell Nodal Rotation. These rotations may be transformed to relate to the global axes in some instances (see Notes). Degrees of freedom relating to global axes: U, V, W, qx, qy, qz may be enforced using the Nodal Freedom data input, or for all shell nodes by using option 278 (see Notes). Node Coordinates Nodal Freedoms Geometric Properties ez, t1... tn Eccentricity and thickness at each node. X, Y, Z: at each node. 5 or 6.

Material Properties Linear Isotropic: Orthotropic: MATERIAL PROPERTIES (Elastic: Isotropic) MATERIAL PROPERTIES ORTHOTROPIC THICK (Elastic:

Orthotropic Thick) Anisotropic: MATERIAL PROPERTIES ANISOTROPIC 5 (Elastic: Anisotropic Thick Plate) Rigidities. Joint Not applicable. Matrix Not applicable. Not applicable. MATERIAL PROPERTIES NONLINEAR 82 (Elastic: Isotropic, Plastic: Cracking concrete)MATERIAL PROPERTIES NONLINEAR 84 (Elastic: Isotropic, Plastic: Cracking concrete with crushing) Elasto-Plastic Stress resultant: Not applicable. Tresca: MATERIAL PROPERTIES NONLINEAR 61 (Elastic: Isotropic, Plastic: Tresca, Hardening: Isotropic Hardening Gradient, Isotropic Plastic Strain or Isotropic Total Strain) Concrete

59 Drucker-Prager: Mohr-Coulomb: MATERIAL PROPERTIES NONLINEAR 64 (Elastic: MATERIAL PROPERTIES NONLINEAR 65 (Elastic:

Isotropic, Plastic: Drucker-Prager, Hardening: Granular) Isotropic, Plastic: Mohr-Coulomb, Hardening: Granular with Dilation) Volumetric Crushing: Not applicable. Rubber Not applicable. Composite Composite shell: COMPOSITE PROPERTIES Field Not applicable. Stress Potential HOFFMAN (Isotropic: von Mises, Modified von Mises Orthotropic: Hill, Hoffman) Creep Damage Viscoelastic Loading Prescribed Value PDSP, TPDSP Prescribed variable. 5 degrees of freedom: U, Concentrated loads. 5 degrees of freedom: Px, Py, Pz, V, W, qa, qb or 6 degrees of freedom: U, V, W, qx, qy, qz Concentrated Loads CL Ma, Mb, where Ma and Mb relate to axes defined by qa and qb respectively. 6 degrees of freedom: Px, Py, Pz, Mx, My, Mz. Element Loads Distributed Loads FLD Not applicable. UDL Uniformly distributed loads. Wx, Wy, Wz: mid-surface CREEP PROPERTIES (Creep) DAMAGE PROPERTIES SIMO, OLIVER (Damage) Not applicable. STRESS POTENTIAL VON_MISES, HILL,

local pressures for element. Not applicable. Constant body forces for element. Xcbf, Ycbf, Zcbf, Wx, Wy, Wz, ax, ay, az BFP, BFPE Velocities Body force potentials at nodes/for element. j1, j2, j3, 0, Xcbf, Ycbf, VELO Velocities. Vx, Vy, Vz: at nodes. Zcbf, where j1, j2, j3 are the face loads in the local coordinate system. Accelerations ACCE Accelerations. Ax, Ay, Az: at nodes. Body Forces CBF

60 Initial Stress/Strains SSI, SSIE Not applicable.

SSIG Initial stresses/strains at Gauss points. Stress/strain components relating to local axes at Gauss points: sx, sy, sxy, syz, sxz, ex, ey, gxy, gyz, gxz. All of these 10 terms are repeated for each fibre integration point through the thickness (see Notes). Residual Stresses SSR, SSRE Not applicable. SSRG Residual stresses at Gauss points. Stress components relating to local axes at Gauss points: sx, sy, sxy, syz, sxz all of these 5 terms are repeated for each fibre integration point through the thickness (see Notes). Temperatures TEMP, TMPE Temperatures at nodes/for element. T, 0, 0, dT/dz, To, 0, 0, dTo/dz Field Loads Not applicable. Not applicable. Temp Dependent Loads

61

APPENDIX B (Variations in K)
Variations in K (Spring stiffness for both column-endplate and bolt-endplate) Unit of K is in kN/mm, Hardening gradient; Slope = 1, Plastic strain = 100

K=1 (C-E), K=10 (B-E)

Moment-Rotation Curve
600
350 300 250
Moment (kNm)

K=0.5 (C-E), K=5 (B-E)

Moment-Rotation Curve

500

moment (kNm)

400

200 150 100

300

200

100

50 0

0 -50 0 50 100 150 200 250 300 350 400 450 500 Rotation (mRad)

10

20

30

40

50

60

70

80

90

Rotation (mRad)

62 K=0.1 (C-E), K=1 (B-E)

Moment-Rotation Curve
300
350 300 250
moment (kNm)
moment (kNm)

K=0.01 (C-E), K=0.1 (B-E)

Moment-Rotation Curve

250

200

200 150 100

150

100

50

50 0

0 0 5 10 15 20 25 30 35 40 45 50 Rotation (mRad)

20

40

60

80

100

120

140

Rotation (mRad)

K=0.001 (C-E), K=0.01 (B-E)

Moment-Rotation Curve
300

K=0.5 (C-E), K=1.0 (B-E)

Moment-Rotation Curve
350 300 250

250

moment (kNm)

moment (kNm)

200

200 150 100

150

100

50

50 0
0 50 100 150 200 250 300 350 400 450

0 Rotation (mRad)

10

20

30

40

50

60

70

80

90

100

Rotation (mRad)

63 K=0.01 (C-E), K=1.0 (B-E)

Moment-Rotation Curve
350 300 250
moment (kNm)

K=0.25 (C-E), K=7.5 (B-E)

Moment-Rotation Curve
350 300 250
moment (kNm)

200 150 100 50 0 0 20 40 60 80 Rotation (mRad) 100 120 140 160

200 150 100 50 0 0 10 20 30 40 50 60 70 80 90 Rotation (mRad)

K=0.05 (C-E), K=5.0 (B-E)

Moment-Rotation Curve
400 350 300
moment (kNm)

K=0.15 (C-E), K=1.5 (B-E)

Moment-Rotation Curve
350 300 250
moment (kNm)

250 200 150 100 50 0 0 20 40 60 80 Rotation (mRad) 100 120 140 160

200 150 100 50 0 0 10 20 30 40 50 60 70 Rotation (mRad)