Baseplate Column Connection under Bending

J. Construct. Steel Research 27 (1993) 37-54

Baseplate Column Connection under Bending: Experimental and Numerical Study

R. Targowski, D. Lamblin & G. Guerlement

Service de M6canique des Mat6riaux et des Structures, Facult6 Polytechnique de Mons, rue de Houdain, 9, B-7000, Mons, Belgium

A B S T R A C T

In the past baseplate column connections have not received much attention from

scientists. Only very simple elastic or plastic models are in common use for moment resisting bases. The aim of this paper is to present some experimental

and numerical study for such a connection. Unstiffened rectangular baseplates with constant shape have been welded to different types of columns with square, rectangular, circular, channel and I cross-section. The assembly was then

connected to a concrete foundation with four anchorage bolts and submitted

to pure bending till the baseplate failed. Experimental applied bending moment curves versus deflections were recorded and the plastic behaviour of the base-

plate was identified. The failure mechanism was approximated by yield-line analysis and the plastic moment was obtained. The kinematic study was com- pleted by a static study with a non-linear finite element program taking into account:

- - material and geometrical non-linear behaviour of the baseplate; - - non-linear contact with concrete foundation.

Comparison between theory and experiment is made and practical conclusions are obtained.

1. INTRODUCTION

Column bases are no doubt very important structural elements affecting the real structural behaviour of a very large category of structures including building frames and supports (especially in the piping industry).

37 J. Construct. Steel Research 0143-974X/93/$06.00 © 1993 Elsevier Science Publishers Ltd, England. Printed in Malta

38 R. Taroowski , D. Lamblin, G. Guerlement

The actual semi-rigidity of the bases, which is traditionally disregarded, in the same way as the semi-rigidity of beam-to-column joints, largely influences frame response. This includes cases in which the column bases classified as hinges often possess fairly significant rotational stiffness, and cases in which the column bases classified as rigid may exhibit consider- able deformation. Among the main different types of bases, the baseplates are extensively used in engineering practice. This is due to the relatively simple manufacture (especially if they are unstiffened) and assembly of these bases. However, the structural and technological simplicity in their use is in radical contradiction with the difficulties from their behaviour under loads. The baseplate welded to the end of the column, the bolts and the concrete foundation are deformable structural elements transmitting acting forces by the mutual and varying unilateral contact. Thus, it is quite obvious that the only hope for overcoming all complications concerning unilateral multi-body contact, geometrical non-linearity and metal plastifi- cation is to employ sophisticated numerical methods. On the other hand, we have to remember that, even with very efficient numerical techniques, the real state of a connection is always very difficult to analyse due to the variable conditions in th~: real structures. The initial geometrical imperfec- tions, non-uniform distribution of material properties in the elements of a real connection, initial and manufacture (welding) fields of residual stresses are the main difficulties in the way of all theoretical analyses. Due to these facts several simplified models which incorporate important assumptions without necessarily reflecting real behaviours of the structure are found in the not extensive literature.

The main groups of these assumptions may describe the linear (elastic) properties or the physically non-linear (plastic) properties of the connec- tion.Xa In the second option the help of simple or conventional yield-lines mechanisms 4'5 is appreciated.

Some similar problems are discussed in the much more extensive literature on the beam-to-column steel joints. In this case we also have many simplified models typical of the conventional design approach. 6 9 Recently, several numerical studies were carried out in this field in order to analyse more precisely the load distribution among bolts, the stress and strain field of elements and unilateral contact forces. 71° But despite a more realistic idealisation considered in these studies, the analysis is not always very representative because it is limited to a small part of a beam-to-column joint neglecting the influences of the all spatial effects. Moreover the results presented are generally restricted to some character- istic curve describing the external loads as a function of some generalised displacements.

Baseplate column connection under bending 39

Thus, in conclusion, an adequate analysis and a fundamental under- standing of the mechanical behaviour of the baseplate column connec- tions, as a basis for assuming engineering rational future simplifications, are not found today in the literature. Of particular interest are the nature of the interaction between connected members, the change in boundary conditions (area of bearing surfaces) as the load level increases and the identification of the elastic stiffness and of the limit and post limit state of the connection. When considering a slab base as a part of a real frame (or supporting structure), the answers to all the above mentioned prob- lems are a first requirement for predicting the joint actual behaviour which is necessary for an adequate and precise analysis of all the structures. The present study is not so ambitious and is just a first step in that general trend. It is restricted to monoaxial pure bending and focuses mainly on:

- - identification of the limit state of an unstiffened base plate connection for various column sizes and analysis of associated yield lines distribution;

- - analysis of unilateral contact between the baseplate and the concrete foundation.

For the second aim, non-linear sophisticated finite element analysis will be used with some assumptions. For the first aim and due to our special interest in biaxial bending of rather large baseplates under external monoaxial bending moments applied around the major axis of a small sized column, experimental information was judged necessary. Our inter- est in such a problem is justified by two practical uses:

- - One is in relation to realisation of a semi-rigid connection with a rather thick unstiffened baseplate. Economically such a connection may be much more interested than a very stiff connection obtained with expensive welded added stiffeners.

- - The second applies currently in piping technology. A very large number of pipes supporting structures are connected to slabs or foundations with special bolted anchorages localised in bore holes. The development of maximum performance, for a given connection, is considered dependent on some critical distance between two proximate or successive connections. So baseplate dimensions are generally rather large in comparison with the dimensions of column profiles.

These two kinds of connections are strongly influenced at the limit state by the plastic behaviour of the baseplate itself.

40 R. Tarfowski, D. Lamblin, G. Guerlement

2. Q U A L I T A T I V E D E S C R I P T I O N O F T H E

B A S E P L A T E B E H A V I O U R U N D E R T H E L O A D

In general, all of the connections studied behave in a similar manner. At the initial state, without loading, the lower surface of the baseplate is in contact with the concrete foundation and some parts of the upper surface are in contact with the heads of the bolts. The shanks of the bolts are usually separated from the baseplate material.

When external loading is initiated and the baseplate begins to bend, some regions of the lower surface of the plate lift up and lose contact with the foundation. The area and distribution of these regions are variable and depend on the load level. This yields the first source of initial non-linearity in the problem by changing the boundary conditions in the system.

When the load increases, the material linearity is being lost and some regions of plastic material are identified. These regions are localised near the tension bolts and theoretically predicted yield lines. At the next stages of loading the continuous bending and lifting up of the baseplate caused a changing load distribution in the system. The second order effects such as, for example, membrane forces due to a pressure between the bolts shanks and baseplate material, distortion of the bolts, prying effect and localised compressions appear in the plate. All these effects cannot be neglected when the load-carrying behaviour of the slab base is being analysed realistically.

3 E X P E R I M E N T A L S T U D Y

3.1 General description of test rig

The representative baseplate specimen analysed in the present study consists of a 400 × 300 mm steel plate welded to the end of variously sized steel columns and founded on a specially constructed concrete square foundation (with 1 m of side) clamped to the laboratory strong floor (Fig. 1). The plate was bolted to the foundation by four high quality steel bolts, 24 mm in diameter except in the neighbourhood of the plate where the diameter of threaded part was 16mm. The bolts were placed in reserved holes in the concrete with the internal diameter equal to 28 mm and were provided with suitable washers and nuts. The initial clearance between bolts and holes in concrete does not favour develop- ment of the membrane effect at beginning of loading. Before the tests, anchorages were manually bolted by the same person (only muscle power

Baseplate column connection under bending 41

" _' - I - ~' .=el ~ plates 1 .J.~cc~umn jac_k. ~.

• .~o!

. . . . , , i I 1 ' ¢ i:1 f

/ / / / 2 . 2 ~ / / / / / / 4 . ' , 2//)~. '?, / / i / /i / i / ~

F i g . 1. Schematic view of the test rig.

"arm

applied to a standard spanner) without any precise control of the installed prestressing load.

Strain gauges on the bolts recorded during tests only to control the elasticity of bolts. No mortar layer was prepared between the steelbase plate and footing concrete. The loading was restricted, as previously explained, to a one dimensional pure bending moment applied around the major axis of the column in the longitudinal plane of symmetry of the baseplate. Generally (except for a ] profile) the plane of bending is also a plane of symmetry of the column. The loading moment is transferred to the column and the baseplate by a rigid horizontal girder fixed to the free end of the column. The connection is made with the help of bolted circular plates.

Small hydraulic jacks with a common oil alimentation pump and built in long hinged arms of equal length were used to introduce the loading moment. The eventual rotation of the concrete foundation during tests was not recorded but our feeling is that such a rotation is very negligible.

Approximate dimensions are indicated for the girder and the arms showing that the usual rotation of a column axis does not give unaccept- able errors to the value of the applied bending moment calculated with initial geometry and recorded loads of the jacks.

Let us remark that this testing system permits the following:

- - to test a sufficient number of plates for a given sized column. After each test the deformed baseplate is cut from the column for which

42 R. Targowski D. Lamblin. G. Guerlement

the end is prepared before welding of a new plate. The packing under the arms of the jacks makes it easy to insert modifications before a new test; to apply biaxial pure bending with relative rotation of the two bolted circular plates. Such bending will be examined in the future.

3.2 Baseplates and columns

Detailed geometry of the twelve baseplates (6 of 6 mm and 6 of 10 mm thickness) used for the tests is given in Fig. 2 (dimensions in ram). All specimens for a set of tests, were cut from the same commercially graded steel plate. The material properties needed for theoretical analysis or interpretations were obtained from 22 traction tests on samples cut from several free regions remaining in the steel plate after cutting the tested specimens. Finally, it was decided to use only a mean value of the yield stress for calculations, namely, 275 MPa for 6 m m thick baseplates, 311 MPa for 10mm thick baseplates. The maximum difference between a given experimental value and the used mean value may be +8°/,,. Elongation before fracture was +25%, with +25% just before strain hardening. The real thickness of tested specimens was found to be a maximum difference of 0"1 mm less than the nominal value. So no correction was applied to nominal values for theoretical calculations. The cross-sections of the various sized columns are also given in Fig. 2. After welding of a baseplate, the two axes of symmetry of the cross-section are coincident with those of the baseplate. For profile, the weak axis of the profile was in coincidence with the longitudinal axis of symmetry of the plate. The maximum dimension of the cross-section of the profile was always parallel to the maximum dimension of the baseplate. The vector associated with the applied bending moment was always parallel to the small dimension (300 mm) of the baseplate.

< 50, 5(

t

J _..(~..

Fi-1 2__L

4OO

A 1

L _x~.,I.~J I 200/20o/6 193,7 / 6,5 180/100/ /10

i

I

TPE 200 H E A 2 0 0 UPN 20.0

Fig. 2. Geometry of the baseplate and cross-sections of columns under consideration.

Baseplate column connection under bendin9 43

For a future easy definition of a yield mechanism, it is convenient to associate with each cross-section a closed polygonal line (rectangular or square) generated by some lines belonging to the real cross-section and by some added dashed lines (see Fig. 2). The longitudinal dimension of this polygonal line was for each case between 180 and 200 mm (70 or 66% of the longitudinal distance between the bolts) but the transverse dimension was between 75 and 200mm (37.5 or 100% of the transverse distance between the bolts). These ratios are of interest for the biaxial bending of the baseplate. The choice of thicknesses equal to 6 and 10 mm for the baseplates is justified below. We intend to find yield line mechanisms in the baseplate for which, roughly speaking, clear spans are 200 (300-100) and 300 (400-100) ram. With a thickness equal to 10ram, the ratios span/thickness are between 20 and 30. It is normally a correct field for applying limit analysis. In some experiments, 6 mm was used with the hope of seeing more clearly the scheme of the yield lines.

The baseplates were fillet welded to the end of the stub column with continuous welds with throats equal to 6 mm. The welding work was done by a qualified welder applying good field practice but no special pre- caution was applied for avoiding deformation of the baseplate during welding. So the flatness of the baseplate was affected. Such a deformation of variable magnitude was not measured and its influence on the contact area between the baseplate and the concrete foundation was totally neglected. Our constant need was to make experimental evaluations in very usual conditions totally similar to those in practice. Of course very fine agreement between theory and experiments is not necessarily ex- pected.

3.3 Sensors

Loads developed by the jacks and the bolts and dial gauges measuring the baseplate uplift at several locations were continuously recorded during tests. Such measurements are sufficient to control the elasticity of the bolts and to plot, with minimal interpretation or transformation, graphics for the applied bending moment versus uplift displacement. It was also attempted to record rotation of the column axis but such measurements were not necessary for our limited interpretation.

4 R E S U L T S A N D C O M M E N T S

We give here only extended results for baseplates with thicknesses equal to 10 mm. For the six plates 6 mm thick the results were similar but second

44 R. Taryowski, D. Lamblin, G. Guerlement

order effects were much more pronounced. The general shape of the graphics shows the complexity of all interacting phenomena (see Figs 3 and 4).

Our feeling is that the initial deformation due to welding (more important for HEA and tube profiles) and varying contact between baseplate and concrete are responsible of unexpected graphics such as Figs 5 and 6. Nevertheless it seems that the existence of a limit state

M

2 0 lexp. I / . . . . . . ~ /

l

I c m

11 3 0

10

V

Fig. 3. Loading moment versus vertical displacements of indicated points on the plate and the limit load definition: experimental results.

M

2 0

$

f11111 V

o i 2 3 4 ~ ~- Fig. 4. Loading moment versus vertical displacements of indicated points on the plate:

experimental results.

Baseplate column connection under bending 45

M

kN exp

20_

I0 .

-1

3 1 ~ 2

I

nlnl V

Fig. 5. Loading moment versus vertical displacements of indicated points on the plate: experimental results.

M

exJ:

2O

10

3 I 2

mm V

' ~ ~ ~ ~ e Fig. 6. Loading moment versus vertical displacements of indicated points on the plate:

experimental results.

associated with a deep modification in the slopes may be accepted. Definition of the level of loading corresponding to the limit state is always a more or less conventional problem to solve.

Generally the intersection point of straight lines generated as ap- proximations of some parts (elastic and post limit states) of the real curve may be adopted. This definition has been used to obtain the experimental loads indicated on the graphics and in table 1 (see also Fig. 3).

After the experiments, the deformed baseplates were carefully exam- ined with a view to defining a scheme of yield lines. The 6 mm thick

46 R. Targowski, D. Lamblin, G. Guerlement

Table 1 The limit loads for the base plates

Column Baseplates (10 ram) Baseplates (6 ram) Type of size mechanism

Experiment Theory Error Theory Experiment Error M(kN m) M(kN m) (0/o) M(kN m) M(kN m) (o~,)

180x 100 18.3 15.23 -16.8 4.85 6.47 -25 7v 200 × 200 23-6 23.32 - 1-2 7-42 10.12 -. 26.7 3 YL

IPE 21.3 19.92 -6.5 6.34 6.53 -2.9 7V HEA 21.7 20.78 -4.2 6'62 7.9 - 16.2 3 YI. UPN 19.7 18.48 -6.2 5.88 555 +6 7V

O1937 20.8 18.79 -9.7 5.98 6.39 -6.4 3 YL

baseplates were preferably used for this aim being more easily inter- preted.

In the theoretical approach, two mechanisms suggested by experiment will be used. The first is a very classical one with three yield-lines (3YL) along the whole of the width of the baseplate and is expected for monoaxial bending of the baseplate welded to a profile of similar width (Fig. 7). The positions of the yield-lines are imposed by the column profile and the bolts in tension. The other mechanism, much more complicated, is used to express the biaxial bending of the baseplates welded to narrow profiles (Figure 8). Despite the help given by experi- ments this mechanism may not be defined very precisely and has to be considered with parameters to be optimised (see the theoretical definition below).

Fig. 7. Deformation of the plate with a three yield-line pattern.

Baseplate column connection under bending 47

Fig. 8. Deformation of the plate with a complex yield-line pattern.

5 T H E O R E T I C A L A P P R O A C H W i T H L I M I T ANALYSIS

5.1 Definition of the mechanism for biaxial bending of the baseplate

The kinematical theorem of limit analysis 11 gives a powerful and simple method of obtaining an upper bound of limit bending moment applied to the column and producing failure of the baseplate. Such a theorem may be applied with a mechanism defined by a scheme of straight yield-lines (hinge lines) as is customary in reinforced concrete.

It is well known that yield-lines between two rigid moving areas in the mechanism go through the intersection point of the axis of rotation of these areas. Another possibility is to have yield-lines as parts of the rotation axis.

Figure 9 shows a seven-variable mechanism (7 V). Such a mechanism has six axes of rotation noted from (i), i= 1, 6 for which positions are defined by the positions of the bolts and the parameters ~i, i= 1, 6. The seventh parameter ~7 is related to the angular position of the yield line between parts 2 and 3 of baseplate.

Axes 4, 5 and 6 were added to the first mechanism to permit a more refined distribution of the strain around the bolts simultaneously with a rotation of the strain triangular part 6 in the neighbourhood of the column profile around axis 6. Areas denoted by O remain undeformed and are in contact with the concrete foundation.

5.2 Numerical results

First let us remark that the displacement of an arbitrary point of the baseplate or the relative rotation in a given yield-line may be determined

48 R. Targowski , D. Lamblin, G. Guerlement

~3

®

+_ ® I /

I 1'

1117

OL 6

(6)

Fig. 9. The 7 V yield-line pattern for the baseplate.

M~ kN

ex t ) .

2 0

f c In

/ , /

r l l l l l

Fig. 10. Loading moment versus vertical displacements of indicated points on the plate: experimental results.

versus the angle of rotation of the column axis arbitrarily taken equal to unity and the parameters ~i identifying the mechanism. The best upper bound M of the bending moment applied to the column and producing the failure of the baseplate is obtained as the result of the

Baseplate column connection under bending 49

M

kN

lolo + 4 " ~

V

2

Fig. 11. Loading moment versus vertical displacements of indicated points on the plate: experimental results.

minimisation versus the parameters ~i of the power of dissipation in the baseplate, t 1

The two mechanisms (seven variable and three yield-line mechanisms) were used with a view to defining limit loads in relation to tested baseplates. The mechanism with seven variables was the most efficient (giving minimum values to the external bending moment) for narrow column profiles as expected. Theoretical results are reported in Table 1 from which comparison with experimental loads is now possible.

Figures 10 and 11 also show optimal corresponding schemes with yield-lines. For a ] profile, the polygonal dosed shape of which has the axis of the baseplate as the symmetrical axis, only an approximate theoretical result has been determined.

5.3 Discussion

Comparison of theoretical and experimental results may be made from Table 1. The error is defined as the difference between the theoretical and experimental results divided by the experimental result. Limit analysis seems to appear as a safe guide for calculating limit bending moments. Poor agreement is obtained for the first column profile without any possibility of special justification.

Of particular interest may be the comparison between some experimen- tal bending moments associated with the failure of the baseplate and the calculated maximum bending moment obtained with a more classical method largely used in engineering offices. 3 In the last method, the criteria

50 R. Targowski, D. Lamblin, G. Guerlement

of failure is the equality of the maximum longitudinal bending stress in the baseplate to the yield stress of the material. For column profiles like HEA or IPE 200 (constant height of 200 mm--variable width) only a bending moment of 7.4 kN m may be accepted. Moreover, even for such a small moment, transverse bending of the baseplate is difficult to verify for an IPE profile (narrow profile) giving most probably with an approximate and may be unjustified method a new reduction factor. This brief compari- son shows how important it may be to study more clearly in the future the real baseplate behaviour.

6 F I N I T E E L E M E N T A N A L Y S I S

Preliminary numerical results for the baseplates welded to the square cross-section column are contained in this section. The primary objectives of this study were to assess the performance of the plate under the load conditions, especially in a highly non-linear state, and predict the contact force distribution between the lower surface of the plate and concrete foundation.

Applying a fully non-linear description (materially and geometrically non-linear formulation with unilateral constraints) and relatively exact discretisation of the plate with the welded column, require a considerable amount of both memory capacity and computing time. Due to these reasons certain restrictions have to be made.

On the basis of the geometrical and load conditions of the specimen there is a plane of symmetry which is perpendicular to the vector of the external moment. So, the computat ion can be carried out on a half of the tested specimen with the actual size of the column. After a detailed numerical analysis of this specimen, the part of the column with uniform linear stress distribution which is characteristic for simple bending was selected. This part of the column was eliminated from the subsequent analysis. The baseplate and the remaining part of the column was rediscretised in the way which is shown in Fig. 12. The linear distribution of external forces in this model corresponds with external bending moment.

Additionally, the following assumptions have been made:

- - the concrete foundation is considered as a rigid plate; - - the bolts are considered in a simplified manner as a system of special

non-linear one dimensional elements and the thread and movement of the bolts are ignored;

- - perfect adhesion between the upper surface of the baseplate and the bolt head is assumed;

Baseplate column connection under bending 51

/ ' - 4 - - . / t , " ,<;

Z 'i Z,#--,~/--z-/- , I/?Z," " IU--J__L i / ? i / / i /

k " L , ~[__,i f - T x ÷ - V ' . , " / . / ,,q ,Id ,

// !_ . ; l ~ - - . J _ ,' " / ,' h' ,,~,1',,"'t _ ,' / / .' - 7 ! i / , ' ,

,D' / >' " / '--, -,< i t / ,? --4' ; I ' ~ '.' - - ~ i,; "t / ~ , - - / - - d " ' I " l ~ '

" . , ' - - - ~ Z ~ : - V 7 , / , L / / I , " ,' ,' : , [ / < : ~ . ' ! l 7--,'--t--,< ; , < / ;

' ; ; , - ~ - ~ .j ! ,, ~ ' - - , ' / i' ,/-.~.

F i g . 12. T h e f i n i t e e l e m e n t m o d e l o f t h e b a s e p l a t e .

- - the friction between the lower surface of the baseplate and the foundation is neglected;

- - an elastic perfectly plastic model of material properties is assumed.

The SAMNL module of the Samcef program was applied to the analysis with isoparametric eight-node solid elements. 12

The deformed state of the model, as a result of the analysis, is shown in Fig. 13. It is evident from this figure that only a part of the plate is in contact with the flat foundation. Very large longitudinal bending and smaller transverse bending correspond very well with the experimental results.

The distribution of the regions of contact between baseplate and the foundation, which is show in Fig. 14 for the specimen analysed in the limit state, explains the unexpected transverse deformation of the plate in the compressed region. A localised, very high pressure in this region was obtained. In our opinion, this is due to the transverse bending of the plate

L; z , Q / / , / ' ~ _ , '-7"~--,"/., '~-~ ; ,~ ,' ~< n / "-----,4. ' / " . . . . ,"

/ / / ' / , " d ~ l"x--.z ,, i ,</, ,,: <. .... t , i XW-- . ' - . 7 .' .' ~ .,

/ i , ," 7't'mZ._lf~,,~-','K--¢" 1 7 A,/ l? ?%/;.DL./7->d----/k/ ; / / ? ~;'

~ _ _ _ ~ I .,,{ ,/1 -?<-v-q. / 7 " ' " ' S:' -UZ. . . . . , , ."-/ I .... ' ."

~ . , ~ - ~ ~ j ! J , i ,~-.'~i-~Ti,: <"TL,--z .... l..z<.i.".z,l,.y -. t . . : . . s , - ~ . < z ~ 7 ~ X . , ~ . + ~ _ ~ _ . I ,< 1 ~ ,

"" ,----Z ~ - _ . c ~ ? ~ - - . > x . b ,','-:= ~-~ ~ - - ' L - - ~ 2 ~ / " l ~=~--:,

"~--:~'-,,%J, ,,~, --_:t: -

Fig. 13. Deformed state of the finite element model of the baseplate.

52 R. Targowski, D. Lamblin, G. Guerlement

1 MN/m=

• l l l l l l l l l .... i l l i l i

i i i j

Fig. 14. Distribution of a contact pressure between the baseplate and foundation.

as a result of contact forces between the upper surface of the plate and the heads of the bolts. The problem has to be reanalysed with a more precisely defined finite element model and unilateral constraints for bolts and the neighbouring region of the plate.

The limit load defined on the basis of the numerical results, in the same way as in experimental analysis (intersection of straight lines), is in a good agreement with the tests. But, it has to be pointed out that the numerically obtained load capacity of the assumed model is higher than recorded during the test. This also needs additional consideration with more refined discretisation.

7 C O N C L U S I O N S

In this paper preliminary experimental and theoretical results for the baseplate connections have been presented. These elements are, equally as with a beam-to-column joints, very important components of real frame structures.

The study reported in this paper is a first investigation towards a better and more realistic understanding of baseplate behaviour. The theoretical analysis was based on the limit state description and finite element approach.

Different yield-line patterns, obtained from experiments, were used in a special minimisation procedure to detect the limit load. Here, emphasis has been on identification of the limit state on the basis of optima and comparison with theoretical results. It is important to note that usually it is difficult to identify the limit state from the experimentally obtained, characteristic curves in the presence of the large non-linear part with a small curvature. In this study the intersection of two straight lines which are tangent to the linear and non-linear parts of the characteristic curves was used as limit state identification. It is in good agreement with the

Baseplate column connection under bendinff 53

theoretical limit state results. But it has to be pointed out that it is not evident from a numerical point of view.

A comparatively large finite element model of the structure which was used in the numerical analysis gave quite interesting results. In this way we detect the non-uniform distribution of the contact forces between the baseplate and the foundation. It is not only the non-uniform prying forces distribution but also the highly non-uniform distribution of the contact forces in the compressed region of the plate. This was confirmed by the unexpected transverse deformation of the tested plate in this region. To study this problem more deeply it is necessary to change the finite element model of the plate, especially in the regions of contact with the bolts.

The overall results indicate that the presented experimental and theor- etical approaches for the baseplate connection analysis are viable methods for correcting the structural performance of computational models of these elements and that the approaches can be applied in a more detailed and extended study.

A C K N O W L E D G E M E N T

The authors acknowledge the support of FNRS (National Funds for Scientific Research) of Belgium.

REFERENCES

1. Ownes, G. W. & Cheal, B. D., Structural Steelwork Connections. Butterworths, London, 1989.

2. Dowling, P. J., Knowles, P. R. & Owens, G. W., Structural Steel Design. Butterworths, London, 1988.

3. Lescouarc'h, Y., Les Pieds de Poteaux Encastr~s en Acier. CTICM, 1988. 4. Hon, K. K. & Melchers, R. E., Experimental behaviour of steel column bases.

J. Construct. Steel Res., 9 (1988) 35-50. 5. Picard, A. & Beaulieu, D., Behaviour of a simple column base connection.

Can. J. Civil Enffng, 12 (1984) 125-36. 6. MAN, A. P. & Morris, L. J., Limit design of extended end-plate connections.

J. Struct. Enffnff, ASCE, 105 (3) (1979) 511-26. 7. Chasten, C. P., Le-Wu, Lu & Driscoll, G. C., Prying and shear in end-plate

connection design. J. Struct. Engn, ASCE, 118 (5) (1990) 1295-311. 8. Bernuzzi, C., Zandonini, R. & Zanon, P., Rotational behaviour of end plate

connections. Costruzioni Metaliche, 2 (1991) 3-32. 9. Jaspart, J. P., Analysis of the semi-rigid beam-to-column joints and its

influence on the strength and stability of steel frames. PhD thesis. University of Li6ge, 1991.

54 R. Targowski, D. Lamblin, G. Guerlement

10. Rothert, J., Gebbeken, N. & Binder, B., Non-linear three-dimensional finite element contact analysis of bolted connections in steel frames. Int. J. Numer. Meth. Engn#, 34 (1992) 303-18.

11. Massonnet, Ch. & Save, M., Calcul Plastique des Constructions, Vol. II. Centre Belgo Luxembourgeois d'information de l'acier, Bruxelles, 1967.

12. SamcefManual, Version 4.1. Samtech, Li6ge, 1991.