Journal of Constructional Steel Research 66 (2010) 748–754

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

journal homepage: www.elsevier.com/locate/jcsr

Strength and ductility of headed stud shear connectors in profiled steel sheetingA.L. Smith ∗, G.H. CouchmanThe Steel Construction Institute, United Kingdom

a r t i c l e i n f o

Article history:Received 22 October 2009Accepted 17 January 2010

Keywords:Composite beamsHeaded stud shear connectorsProfiled steel deck

a b s t r a c t

This paper details the results and subsequent analysis of 27 push tests performed using a new push rig,which investigate the effect of variables such as mesh position, transverse spacing of shear connectors,number of shear connectors per trough and the slab depth on the resistance of headed stud shearconnectors through-deck welded into a transverse deck.The analysis of these tests not only allowed characteristic resistances to be determined for the headed

stud shear connectors in each case, but also enabled comparisons to be made to determine the effect ofthe different variables on the resistance. It found that within the limits tested the transverse spacing ofthe shear connectors has little effect on the resistance, and that including a third shear connector givesno benefit over using shear connectors in pairs. Locating the mesh at the top of the slab, as is commonpractice for crack control, gives sufficient ductility for design using the minimum shear connection rulesin BS EN 1994-1-1, 6.6.1.2 (i.e. the characteristic slip capacity is greater than 6 mm, as required by BSEN 1994-1-1, 6.6.1.1(5)), but a strength enhancement of approximately 30% can be found by locating themesh directly on top of the profiled steel sheeting. The results also indicated that the resistance increaseswith slab depth, but it is not clear if this is an effect of the push test or is a genuine effect of compositeconstruction. Design rules based on these tests are proposed.

© 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Simms and Smith [1] gave details of a modification to thestandard push-off test that improves the correlation between theperformance of shear connectors in the standard test specimenand behaviour observed in beams tests, as described by Hicks [2].The modification applies a transverse load to the faces of the testspecimen to prevent the specimen from splitting or spreadingaway from the beam; this approach has previously been adoptedby Gibbings et al. [3]. This paper investigates the effect of anumber of key variables on the performance of the headed shearconnectors, namely,• slab depth;• the number of shear connectors per group, nr ;• mesh position relative to the head of the shear connector;• transverse spacing of pairs of shear connectors.

To establish the impact these variables have on the shear con-nector performance, nine sets of three tests have been performed.These are detailed in Table 1.Mesh located at the top is nominally 25 mm below the top of

the slab. Mesh located at the bottom rests on the flange stiffenersof the deck.

∗ Corresponding address: The Steel Construction Institute, Silwood Park, Ascot,SL5 7QN, United Kingdom. Tel.: +44 01344 636525; fax: +44 01344 636570.E-mail address: a.smith@steel-sci.com (A.L. Smith).

0143-974X/$ – see front matter© 2010 Elsevier Ltd. All rights reserved.doi:10.1016/j.jcsr.2010.01.005

2. Push tests

2.1. Test setup

The testswere performedon slabs constructed usingMD60pro-filed steel sheeting in 350 N/mm2 material, using 100 × 19 mmheaded stud shear connectors through deck welded (length aswelded of approximately 95 mm). The shear connectors were po-sitioned in the favourable position (the MD60 has a central troughstiffener in the troughs), and the deck trim was removed prior totesting to avoid any unconservative containment of the concrete.Structural Tees were used rather than UC sections to enable bothsides of the push specimen to be cast at the same time, thus ensur-ing consistent concrete properties within each specimen.In each case the test specimenwas loaded to failure by applying

a hydraulic jack to a plate on top of the steel Tees. An additionalload of 12% of the vertical load was applied horizontally to theexposed faces of the concrete specimens. A typical failure is shownin Fig. 1.

2.2. Test results

A typical load–slip relationship is presented in Fig. 2.A summary of the maximum load per shear connector obtained

in each test is presented in Table 2.The compressive strength of the concrete was determined

for each set of specimens using a series of 100 mm cubes. The

A.L. Smith, G.H. Couchman / Journal of Constructional Steel Research 66 (2010) 748–754 749

Table 1Tests performed using the new push rig.

Test Ref nr Slab depth (mm) Mesh location Transversespacing (mm)

A1U 1 140 Top –A1D 1 140 Bottom –B1U 1 225 Top –A2DX 2 140 Bottom 75A2UY 2 140 Top 105A2DY 2 140 Bottom 105A2DZ 2 140 Bottom 140B2U 2 225 Top 105A3D 3 140 Bottom 120

Table 2Maximum load per shear connector obtained in each push test.

Test Ref Test 1 result(kN)

Test 2 result(kN)

Test 3 result(kN)

Mean(kN)

A1U 94.56 83.09 95.98 91.21A1D 115.28 112.49 121.49 116.42B1U 121.32 107.87 107.14 112.11A2DX 57.85 62.93 66.60 62.46A2UY 55.10 47.90 65.10 56.03A2DY 76.15 70.88 75.04 74.02A2DZ 54.40 64.90 66.50 61.93B2U 76.50 69.40 78.20 74.70A3D 52.90 47.30 39.00 46.40

Fig. 1. Typical failure of test specimen with horizontal loads removed.

Fig. 2. Load–slip plot for test A1D-3.

compressive cube strengths are converted to equivalent 150 mmcubes by using the conversion factor of 0.9 specified in CEB-FIPModel Code 1990 [4]. These modified cube strengths are thenconverted to cylinder strengths using the following curve fit of data

from Table 3.1 of BS EN 1992-1-1 [5]:

fc =fc,cube − 2.00291.1716

. (1)

The secant modulus of elasticity was determined for some of thespecimens using cylinder tests. However, it is felt that these maybe inaccurate, and so the relationship detailed in Table 3.1 of BS EN1992-1-1 was used to calculate the secant modulus for use in theanalysis.A summary of the concrete properties is given in Table 3.

3. Characteristic resistance

The characteristic resistance for each layout was determinedusing the method given in BS EN 1990, Annex D8 [6]. This methodtakes into account the variability of the test results and considersthe variability of the component parts (be it the strengths or thedimensions) by making use of a valid theoretical model, which canin turn be calibrated to cover a wider range of design scenarios.The procedure was used to ensure that a design model satisfiesthe target level of reliability demanded in modern UK and Euro-pean Standards. According to the guidance given in Gulvanessianet al. [7], the analysis was undertaken for ‘VX known’ as the testmeasurements are compared with a theoretical model. The use ofthis method is illustrated for the case A1U.The theoreticalmodel used to predict the resistance of the shear

connectors was taken from BS EN 1994-1-1, 6.6.3.1 [8]. Two equa-tions are presented in this Eurocode, depending on the relativestrengths of the concrete and shear connector steel, but in all thetests the concrete was critical so Eq. (6.19) dominated, shown hereas Eq. (2):

PRk = 0.29αktd2√fckEcm (2)

where

PRd is the characteristic resistance of a shear connector.α is a function of the dimensions of the profile and shearconnectors— forMD60 the limiting value applies andα = 1.

kt is a reduction factor based on the dimensions of the profileand the number of shear connectors per trough. For MD60,the limiting values apply for 0.9 mm gauge deck and kt =0.85 for one shear connector per trough, kt = 0.7 for twoshear connectors per trough and kt = 0.7 has been assumedfor three shear connectors per trough.

d is the diameter of the shear connectors.fck is the characteristic cylinder strength of the concrete.Ecm is the secant modulus of elasticity of the concrete.

For the A1U tests, using the concrete properties from Table 3 anda mean shear connector diameter of d = 18.8 mm, the predictedresistance PR = 47.29 kN.The overall coefficient of variation used for determining the

characteristic resistance is composed of two parts: the coefficientof variation of the errors, Vδ , which is based on the test results, andthe coefficient of variation of the resistance function, Vrt .

3.1. Calculating the coefficient of variation of the errors

The first of these is calculated by first calculating the least-squares best-fit to the slope that compares the test results to thetheoretical model. This is calculated from

b =∑rert∑r2t=1692211473

= 1.475. (3)

The numerator and denominator of this equation are calculated inTable 4. re is the experimental resistance and rt is the theoreticalresistance.

750 A.L. Smith, G.H. Couchman / Journal of Constructional Steel Research 66 (2010) 748–754

Table 3Concrete properties corresponding to each test setup.

Test Ref fcm, cube, 100 (N/mm2) fcm, cube (N/mm2) fcm (N/mm2) Ecm, test (kN/mm2) Ecm (kN/mm2)

A1D 27.01 24.31 19.04 – 26.69A2DY 29.66 26.69 21.07 – 27.51A1U, B2U, A3D 26.85 24.17 18.92 24.11 26.64B1U 27.03 24.32 19.05 23.92 26.69A2DX, A2UY, A2DZ 27.38 24.64 19.32 23.55 26.81

Table 4Determining the coefficient of variation of the errors.

Test Ref re (kN) rt (kN) rert r2t δi ∆i (∆i − ∆̄)

A1U-1 94.56 61.84 5848 3824 1.037 0.036 0.0015A1U-2 83.09 61.84 5138 3824 0.911 −0.093 0.0083A1U-3 95.98 61.84 5936 3824 1.052 0.051 0.0028SUM 16922 11473 −0.006 0.0126

Table 5Coefficient of variation components.

Variable, Xj∂grt∂Xj

∂grt∂Xj

1grt (Xm)

d 2× 0.29αd√fcEcm 2

d

fck 0.5× 0.29αd2√Ecmfc

12fc

Ecm 0.5× 0.29αd2√

fcEcm

12Ecm

The error term, δi, is calculated using Eq. (4), and it is the naturallogarithm of this term,∆i, that is used to determine Vδ:

δi =reibrti. (4)

The estimated standard deviation, s∆, of the error terms is calcu-lated from

∆̄ =1n

n∑i=1

∆i = −0.002,

s∆ =

√√√√ 1n− 1

n∑i=1

(∆i − ∆̄

)= 0.0793. (5)

The coefficient of variation of the errors is then calculated from

Vδ =√exp

(s2∆)− 1 = 0.0795. (6)

3.2. Calculating the coefficient of variation of the resistance function

The coefficient of variation of the resistance function is calcu-lated by establishing the effect the variation of each of the variableshas on the resistance function, as shown here:

V 2rt =1

g2rt (Xm)×

J∑j=1

(∂grt∂Xj

σj

)2=

J∑j=1

(∂grt∂Xj

σj

grt (Xm)

)2. (7)

For each variable, Xj, this relationship is presented in Table 5.The coefficient of variation of the resistance function is

therefore

V 2rt =(2σdd

)2+

(σfc

2fc

)2+

(σEcm

2Ecm

)2= (2Vd)2 +

(Vfc2

)2+

(VEcm2

)2. (8)

The coefficients of variation themselves are taken from manufac-turers’ data in the case of the headed stud shear connectors, andfrom BS EN 1992-1-1 in the case of the concrete properties.

The manufacturing tolerance for the diameter of 100× 19 mmshear connectors is +0 mm, −0.4 mm (information provided byNelson Stud Welding UK). This translates as a mean diameter of18.8 mm with a variation of ±0.2 mm, which can be assumedto represent 1.64 standard deviations (i.e. this is a characteristicvariation). The coefficient of variation is therefore

Vd =0.2

1.64× 18.8= 0.65%. (9)

The concrete cylinder strength and secant modulus are related bythe following equation, according to BS EN 1992-1-1:

Ecm = 22[fcm10

]0.3= 22

[fck − 810

]0.3, (10)

where the subscript k refers to the characteristic strength and thesubscript m to the mean. The variability of cylinder strength isdefined by BS EN 1992-1-1 as an 8 N/mm2 difference betweenthe mean and characteristic. For an infinite set, this is equivalentto 1.64 standard deviations, so the coefficient of variation of theconcrete strength is

Vfc =8

1.64fcm. (11)

As the secant modulus of elasticity is related to the cylinderstrength, the coefficient of variation can be taken as

VEcm =(1+ Vfc

)0.3− 1. (12)

For the A1U set of tests, using the concrete properties given inTable 3, the coefficients of variation are Vfc = 25.79% and VEcm =7.13%. This results in an overall coefficient of variation of Vrt =13.44%, using Eq. (8).The coefficient of variation of the errors represents a sample

size of 3, whereas the coefficient of variation of the resistance func-tion represents an infinite sample size. The statistical uncertaintiesfrom the small sample size are taken into account using themethodin BS EN 1990, Annex D.8.2.2.7(5). This gives a characteristic resis-tance of

rk = bgrt (Xm) exp(−k∞Qrt − knαδQδ − 0.5Q 2

)rk = 1.475grt (Xm) exp (−1.64× 0.134

− 1.90× 0.511× 0.079− 0.5× 0.1552)

(13)rk = 1.117grt (Xm) ,

where grt(Xm) is the predicted resistance using the mean testvalues with Eq. (2). This yields a characteristic resistance of 69.06kN for the A1U case with the concrete strength as tested.Each set of tests has been analysed using this method, and the

characteristic and design values are given in Table 6, together withthe gamma factor that relates the two. The characteristic valueof an infinite set of data that is normally distributed is calculatedas the mean value minus 1.64 standard deviations, while the de-sign value is calculated as the mean value minus 3.04 standard de-viations. These calculations vary slightly when using non-infinitesets (such as in this case), but the principle remains the same. Thegamma factor is the ratio of the characteristic value to the designvalue, and if this is approximately equal to the gamma value used

A.L. Smith, G.H. Couchman / Journal of Constructional Steel Research 66 (2010) 748–754 751

Table 6Headed stud shear connector calculated characteristic and design resistances.

Test Ref PRk (kN/stud) PRd (kN/stud) γ = PRk/PRd

A1U 69.06 54.91 1.26A1D 91.58 75.19 1.22B1U 86.62 70.03 1.24

A2DX 47.94 38.54 1.24A2UY 38.18 27.85 1.37A2DY 58.93 48.89 1.21A2DZ 45.23 34.91 1.30

B2U 57.52 46.39 1.24

A3D 31.47 22.88 1.38

in the Eurocodes, it indicates that the coefficient of variation of thesample is similar to that used to determine the design equations inthe Eurocodes, which supports the use of the ‘Vx known’ criteria.The gamma factors in most cases are within a reasonable

amount of the value used in BS EN1994-1-1, i.e. γv = 1.25, and thisis an indication that the combined coefficient of variation from thetest results and the resistance function is appropriate, as the spreadof these results is similar to the sample used in determining theEurocode safety factors. The three exceptions to this are also thecases where the analysis method from BS EN 1994-1-1, Annex Bcannot be used (because the test results are too inconsistent). Thisindicates that the spread of the test results was higher than thatwould normally be expected for these cases, and perhaps theseresistances could be improved by performing more tests and soreducing the coefficient of variation.For comparisonpurposes, a further analysis has beenperformed

that groups the nine tests with two shear connectors, mesh atthe bottom of the slab, and a typical slab depth (i.e. A2DX, A2DYand A2DZ). This analysis resulted in a characteristic resistance ofPRk = 50.33, a design resistance of PRd = 40.15 and a gammafactor of γ = 1.25. If the effect of the transverse spacing is foundto be small, these values could be used instead as a basis for design.

4. Characteristic slip capacity

The slip capacity of each test is taken from the falling branch ofthe test data at the calculated characteristic load level. These arepresented in Table 7.One of these results A2DZ-1 is significantly smaller than the

next lowest result, and so it is felt that this result should bediscounted to avoid distorting the analysis.The entire group of test results can be used to determine a

characteristic slip capacity that will cover all the setups that havebeen considered for 95 mm × 19 mm shear connectors weldedin the favourable position of a transverse 60 mm profile. Thereare two statistical distributions that can be considered—a normaldistribution and a lognormal distribution; the latter will not allowvalues of zero or below. As the slip capacity cannot be negative,it is appropriate to use the lognormal distribution. For the data inTable 7, the lognormal mean is 2.64 and the lognormal standarddeviation is 0.20.From BS EN 1990, Table D1, the characteristic slip capacity is

calculated from

δk = emy−knsy (14)

where

δk is the characteristic slip capacity (mm);my is the lognormal mean;kn is a value taken from Table D1, relating to the number of testresults;

sy is the lognormal standard deviation.

Table 7Slip capacity at characteristic resistance.

Test Ref Test 1 result (mm) Test 2 result (mm) Test 3 result (mm)

A1U 15.38 11.11 13.24A1D 16.78 13.62 19.18B1U 19.33 14.23 15.49A2DX 15.30 9.99 15.42A2UY 11.28 14.69 15.25A2DY 13.34 11.41 12.81A2DZ 6.15 10.30 11.45B2U 14.18 10.87 16.65A3D 20.16 17.51 11.17

In this case 26 tests are included in the sample, and so kn = 1.742(using Vx unknown, as the standard deviation is calculated fromthe test data without a theoretical model). The characteristic slipcapacity is therefore

δk = e2.64−1.742×0.20 = 9.83 mm. (15)

In BS EN1994-1-1, 6.6.1.1(5), a shear connector is defined as ductileif the characteristic slip capacity is at least 6mm, and theminimumdegree of shear connection rules in the Standard are calibratedfor this ductility. The results here show that the characteristic slipcapacity of headed stud shear connectors in transverse trapezoidaldecking exceeds this minimum by a significant amount.

5. Design equation

The method in BS EN 1990 can be used to calibrate a resis-tance model, rather than just producing specific characteristicvalues. The values that have already been calculated, and thatare presented in Table 6, correspond only to the specific concretestrength that was used in the test. By using this characteristic re-sistance in combination with the concrete properties at the timeof testing, the theoretical model used in the analysis (in this casethe model from BS EN 1994-1-1, 6.6.3.1) can be calibrated. Thecalibrated equation will then apply over a wide range of concretegrades, rather than just for the grade under consideration.The design resistance of a headed stud shear connector should

be taken as the minimum of Eq. (6.18) in BS EN 1994-1-1 (whichconsiders the failure of the steel of the shear connector in pureshear) and Eq. (16) (which considers concrete failure):

PRd =0.29βktαd2

√fckEcm

γV(16)

where

PRd is the design resistance of a headed stud shear connector;β is a coefficient taken from Table 8 which depends on theform of construction;

γV is the partial factor.

Table 8 presents the values of the coefficient β along with thedesign resistance, PRd, of a 19 mm × 100 mm headed studshear connector in grade C25/30 concrete. These values are onlyappropriate for transverse 60mmdeep trapezoidal decks forwhichkt = kt,max, as defined in BS EN 1994-1-1.

6. Comparisons

6.1. Comparison to standard predictions

The values of β in Table 8 represent the conservatism of theresistance defined by BS EN 1994-1-1 in each case. These indicatethat for single shear connectors the code is conservative in all cases,and for pairs of shear connectors there is some unconservatismwhen using typical slab depths (designated with an ‘‘A’’ in the test

752 A.L. Smith, G.H. Couchman / Journal of Constructional Steel Research 66 (2010) 748–754

Table 8Summary of design data.

Test Ref Description β PRd for grade C25/30 (kN)

A1D Typical slab depth with one shear connector per trough and mesh resting on top of the deck 1.475 92.41A1U Typical slab depth with one shear connector per trough and mesh at the top of the slab with nominal cover 1.117 69.98B1U Deep slab with one shear connector per trough and mesh at the top of the slab with nominal cover 1.394 87.38A2D Typical slab depth with two shear connectors per trough and mesh resting on top of the deck 0.957 49.37A2U Typical slab depth with two shear connectors per trough and mesh at the top of the slab with nominal cover 0.739 38.16B2U Deep slab with two shear connectors per trough and mesh at the top of the slab with nominal cover 1.129 58.29A3D Typical slab depth with three shear connectors per trough and mesh at the top of the slab with nominal cover 0.618 31.90

Table 9Comparative resistances.

Test Ref A1D A1U B1U A2D A2U B2U A3D

ktβ 1.254 0.949 1.185 0.670 0.517 0.790 0.433

ID)—note that slab depth (or the mesh position) does not featureexplicitly in the BS EN 1994-1-1 equations, but the code doescontain associated detailing requirements. Three shear connectorsare beyond the scope of the code, but assuming kt of 0.7 gives aresult that is very unconservative.

6.2. Effect of variables

Other than slab depth, number of shear connectors, shear con-nector spacing and mesh position, which were the variables to beinvestigated, the tests were intended to be identical in dimensionsandmaterial strengths. The product of the reduction factor, kt , andtheβ coefficient defines the effect of all of these variables, as the re-mainder of the design equation is constant in each case. This prod-uct is shown in Table 9.

6.2.1. Number of shear connectorsThe relationship between the individual shear connector resis-

tance and the number of shear connectors per trough is defined inthe reduction factor, kt , which also includes the dimensions of thedeck. The equation for kt in BS EN 1994-1-1 implies the followingrelationship:

PR ∝1√nr. (17)

This indicates that the resistance of a shear connectorwhen used ina set of two should be 71% of that when used individually and (al-though the equation only applies to one or two shear connectors)implies that for sets of three shear connectors, each shear connec-tor should have 58% of the resistance of a single shear connector.Using the results presented in Table 9, the accuracy of this

relationship can be investigated. The values are presented inTable 10.The results presented above can be plotted as reduction against

the number of shear connectors and compared to the relationshipproposed in BS EN 1994-1-1. This is shown in Fig. 3.Clearly there is very little correlation between the experimental

resistances and the theoretical values. The results imply that a re-duction of 0.58 should be applied to pairs of shear connectors (i.e. atroughwith two shear connectorswill only give 16%more strengththan a trough with a single shear connector) and that a reduction

Fig. 3. Effect of number of shear connectors on resistance.

of 0.35 should be applied to sets of three shear connectors (i.e. atrough with three shear connectors has a similar performance to atrough with two shear connectors—this is to be expected as it is afailure of the concrete, not the steel, that dominates the resistance).

6.2.2. Shear connector spacingThe normalised resistance for each of the shear connector

spacings is presented in Table 11.These results seem to indicate that there is an increase in the

resistance at the middle spacing, and a similar resistance for thelarger and smaller spacings. However, this could also be an effectof the stronger concrete used in the A2DY tests. The coefficient ofvariation of the results, although high, can be reasonably assumedto be caused by variations in the testing, and so without furtherinformation it must be assumed that the spacing of the shearconnectors (within practical limits) has very little effect on theirresistance.The push tests showed that the resistance is dominated by

failure of the concrete around the shear connectors, rather thanshearing of the shear connector material itself. The typical failuresurface for single shear connectors is a cone of concrete startingunderneath the head of the shear connector and growing indiameter down the length of the shear connector, although theshape is restricted by the shape of the decking. For pairs of shearconnectors the failure surface is similar, but the cones around eachshear connector are joined. A typical failure surface for a test onpairs of shear connectors is shown in Fig. 4.From Fig. 4 it is clear that increasing the spacing of the shear

connectors will only have a small effect on the total area of the fail-ure surface, so it is expected that therewill only be a small increasein the resistance as the spacing increases. Comparing the results

Table 10Effect of the number of shear connectors on the resistance.

Single test ktβ Double test ktβ DoubleSingle Triple test ktβ

TripleTestSingle

A1D 1.254 A2D 0.670 0.534 A3D 0.433 0.345A1U 0.949 A2U 0.517 0.545B1U 1.185 B2U 0.790 0.667aMean 0.582 0.345a This result has not been included in the mean as the thicker slab appears to have an additional affect on the resistance.

A.L. Smith, G.H. Couchman / Journal of Constructional Steel Research 66 (2010) 748–754 753

Table 11Effect of shear connector spacing on the resistance.

Test Ref ktβ

A2DX 0.789A2DY 0.918A2DZ 0.745

Mean 0.817S.D. 0.0899CoV 11.0%

Fig. 4. Failure surface for pairs of shear connectors.

of the single shear connectors with the pairs of shear connectorsshows that there is only a small improvement, and this can be at-tributed to the small increase in the area of the failure surface. Therelationship between the area of the failure surface and the resis-tance also gives verification to the lack of improvement betweena set of two shear connectors and a set of three, where the mid-dle shear connector will be contributing very little. It is possiblethat if the spacing was increased significantly beyond those testedhere, the shear connectors would act independent of each otherand individual shear connector resistances closer to those foundfor a single shear connector per trough could be found. In addition,the resistance of staggered shear connectors may differ from theresults found here.

6.2.3. Slab depthIt was expected that there would be no effect on the resistance

of the shear connectors when the slab depthwas increased, so onlytwo verification testswere conducted at an extreme slab thickness.The comparison is presented in Table 12.Clearly the results are not as expected, and there is an increase

in the resistance with increasing slab depth. There is also amarkeddifference in the improvement between single shear connectorsand double shear connectors. As such, it is not possible to quantifythis improvement without further investigation by testing to look

Table 12Effect of the slab depth on the resistance.

140 mm depth ktβ 215 mm depth ktβ ThickTypical

A1U 0.949 B1U 1.185 1.248A2U 0.517 B2U 0.790 1.528Mean 1.388

Table 13Effect of the mesh position on the resistance.

Mesh at bottom ktβ Mesh at top ktβTopBottom

A1D 1.254 A1U 0.949 0.757A2D 0.670 A2U 0.517 0.772Mean 0.765

at intermediate slab depths. From the equation for PRd, there is arelationship with the elastic properties of the concrete, so it maybe that the increased inertia of the concrete slab has some effect onthe resistance of the shear connector. However, it is possible thatany effect on the strength caused by the slab depth is a trait of thetest setup rather than an effect seen in actual composite beams.

6.2.4. Mesh positionDetailing rules in BS EN 1994-1-1 require that lower reinforce-

ment be provided 30 mm under the head of the shear connector.Current practice within the decking industry is to install a singlelayer of mesh reinforcement at nominal cover to the top of the slabfor crack control and longitudinal shear, and it is not possible toposition the mesh reinforcement 30 mm below the head of theshear connectors using the vast majority of trapezoidal compositedecks in the UK. It is also unclear how the term ‘lower reinforce-ment’ applies to a slabwith only one layer. The results presented inTable 7 show that there is sufficient ductility in the shear connec-tors when themesh is positioned either directly on top of the deck,or at the top of the slabwith nominal cover. This indicates that cur-rent practice need not be altered. However, there is an improve-ment in the strength of the shear connectors when the mesh ispositioned lower, as shown in Table 13.The results in this instance are consistent and suggest that an

improvement in resistance of 31% (1 ÷ 0.765 = 1.31) is possibleby positioning themesh on top of the decking rather than the top ofthe slab. When positioned below the head of the shear connector,the mesh will be crossing the failure surfaces shown in Fig. 4,which explainswhy there is an increase in the strength of the shearconnector.

7. Conclusions

The results of the push tests show that the resistance ofheaded stud shear connectors when through-deck welded intotransverse trapezoidal sheeting is comparable with the resistancegiven in BS EN 1994-1-1 when mesh is provided directly on topof the profile. The results have also shown that there is sufficientductility of the shear connection when the mesh is positioned atnominal cover below the top of the slab, where mesh is generallypositioned for fire design, crack control, longitudinal shear andbuildability. However, there is a decrease in the resistance ofthe headed stud shear connector when the reinforcement doesnot pass through the concrete failure plane. Modification factorshave been proposed to improve the suitability of the equationin each scenario. However, it is expected that in due course themanufacturerswill eachperform testing on their ownproducts andproduce specific equations for use in design.The results show that there is sufficient ductility to use the

current minimum shear connection rules in BS EN 1994-1-1 forheaded stud shear connectors with conventional reinforcementeither resting on the profile or at nominal cover from the top ofthe slab when using transverse 60 mm deep trapezoidal sheeting.

754 A.L. Smith, G.H. Couchman / Journal of Constructional Steel Research 66 (2010) 748–754

A comparison between otherwise identical test setups suggeststhat using shear connectors in pairs will give an increase in resis-tance of approximately 16% over using a single shear connector, in-dependent of the spacing (within usual practice). The test resultsalso indicate that there is no further improvement by using shearconnectors in groups of three.The results of this study suggest that there is an improvement

in the resistance of approximately 30% by positioning mesh underthe head of the shear connector rather than at nominal cover fromthe surface of the slab, for both single and pairs of shear connectors.

Acknowledgements

The authors would like to gratefully acknowledge the BCSATask Group for Long-Span Composite Beams, including theMetal Cladding and Roofing Manufacturers’ Association, for theirtechnical and financial support, along with Dr. R.E. McConnel,Mr. M.R. Tohey and the technical staff of Cambridge UniversityEngineering Department for ensuring the success of the varioustesting programmes.

In addition, the authors would like to acknowledge theircolleagues at SCI Dr. S.J. Hicks (now of HERA, New Zealand), Dr J.W.Rackham andDr.W. I. Simms for their contributions to this project.

References

[1] Simms WI, Smith AL. Performance of headed stud shear connectors in profiledsteel sheeting. In: 9th international conference on steel concrete composite andhybrid structures. 2009.

[2] Hicks SJ. Strength and ductility of headed stud connectors welded in modernprofiled steel sheeting. The Structural Engineer 2007;(May):32–8.

[3] Gibbings DR, Easterling WS, Murray TM. Influence of steel deck on compositebeam strength. In: Easterling WS, Kim Roddis SM, editors. Compositeconstruction in steel and concrete II. New York: ASCE; 1993. p. 758–70.

[4] CEB-FIP Model Code 1990, Comite Euro-International du Beton, 1993.[5] BS EN 1992-1-1: 2004 Eurocode 2: Design of concrete structures — Part 1-1:General rules and rules for buildings. London: British Standards Institution.

[6] BS EN 1990: 2002 Eurocode — Basis of structural design. London: BritishStandards Institution.

[7] Gulvanessian H, Calgaro J-A. Holický: ‘Designers’ guide to EN 1990. Eurocode:Basis of structural design’. Thomas Telford, 2002.

[8] BS EN 1994-1-1: 2004 Eurocode 4 — Design of composite steel and concretestructures Part 1-1: General rules and rules for buildings. London: BritishStandards Institution.