Experimental Report

(1)

Technical University of Denmark

Master Thesis

Development of an efficient structural system for buildings of glass fibre reinforced

polymer

Experimental Report

Author:

Aslak ClarkeJensen

Supervisor:

Henrik Almegaard

Co-supervisors:

JanSøndergaard VickiThake

A thesis submitted in fulfilment of the requirements for the degree of MSc

in

Civil Engineering

June 2015

(2)

Contents i

List of Figures iii

List of Tables iv

1 Introduction 1

2 Theory 2

2.1 Estimation of mean material strength . . . 2

3 Experimental preparations 5 3.1 Angle profiles . . . 5

3.1.1 Test setup . . . 5

3.1.2 Creating the test specimens . . . 6

3.1.3 Expected capacity . . . 9

3.2 Bolted corner connection . . . 11

3.2.1 Test setup . . . 11

3.2.2 Creating the test specimen . . . 12

3.3 Adhesive corner connection . . . 16

3.3.1 Test setup . . . 16

3.3.2 Creating the test specimen . . . 16

4 Conduction of experiment 19 4.1 Angle profiles . . . 19

5 Test results 23 5.1 Angle profiles . . . 23

6 Discussion and Conclusions 31 7 Bibliography 33 7.1 Books . . . 33

i

(3)

Contents ii 7.2 Articles . . . 33 7.3 Reports and Theses . . . 33 7.4 Codes and Standards . . . 33

(4)

3.1 Principle for test setup for angle profile . . . 6

3.2 Drawing specifications of angle profile . . . 7

3.3 Picture of minor damages of angles . . . 8

3.4 Picture of the actual angle profiles . . . 8

3.5 Expected displacement shape of Angle profile, from STAAD.Pro . . . 10

3.6 Isometric view of test setup for test of corner connection . . . 12

3.7 Section with dimensions of corner connection . . . 12

3.8 Overall layout and dimensions of test specimens for corner connection . . . 13

3.9 Bolt layout and dimensions of the internal angle of the bolted connection . . . . 13

3.10 Overall bolt layout of bolted corner connection test specimen . . . 14

3.11 Expected displacements at failure for corner connection specimen . . . 15

3.12 Picture of the fit of the angle profiles within the MD Plank . . . 16

3.13 Picture of the angle profiles in place and glue distributed on the surface . . . 17

4.1 Actual test setup for angle profile . . . 19

4.2 Yielding of the steel fork profile and strengthening welding of steel fork profile . . 20

4.3 Reinforced test setup for bolted corner connection . . . 21

4.4 Rotation at failure of the bolted connection . . . 21

4.5 Adhesive corner connection test specimen . . . 22

5.1 Plot of results from test of angle profiles . . . 23

5.2 Failure of angle profile 1 . . . 26

5.3 Plot of the direct measurements for the bolted connection . . . 26

5.4 Plot of the measurements for the bolted connection corrected for support displacement . . . 27

5.5 Failure of the bolted connection . . . 28

5.6 Plot of the direct measurements for the bolted connection . . . 29

5.7 Plot of the measurements for the bolted connection corrected for support displacement . . . 30

5.8 Failure of the adhesive connection . . . 30

6.1 Plot of the measurements for the bolted connection and adhesive connection for displacements at end of beam corrected for support displacement . . . 32

iii

(5)

List of Tables

2.1 Relation between design, characteristic and estimated mean strength . . . 3 3.1 Design and actual plan view dimensions and mass of angle profile . . . 7 3.2 Design and actual cross-section view dimensions of angle profile . . . 7 3.3 Expected applied failure load for angle profiles dependent on capacity assumptions 10 3.4 Expected applied failure load for bolted connection dependent on capacity as-

sumptions and element of failure . . . 14 3.5 Expected applied failure load for adhesive connection dependent on capacity as-

sumptions and element of failure . . . 18 4.1 Expected applied failure load for angle profiles dependent on capacity assumptions 20 5.1 Applied failure load for angle profiles . . . 25 5.2 Applied failure load for bolted connection . . . 27 5.3 Applied failure load for adhesive connection . . . 29

iv

(6)

Introduction

Structural design of glass fibre reinforced polymer (GFRP) is a relatively new discipline, and while verification of the capacity of the structural members is well established, the connection design is less developed. This especially goes for rigid connections, both for bolted and adhesive joints. For rigid bolted connections, some researchers believe that the inherently low stiffness of the material will result in the connections only acting as semi-rigid connections ([A1]). For adhesive connection design, analytical formulations only represents simple single and double lap joints, and even for these, it is advised that results are verified by experiments ([C1], [D1]).

This Experimental Report is to be read in connection with the Main Report, where a rigid frame is designed based on presented theory. A major part of the structural design of the rigid frame was the connections, which were designed both as bolted connections and adhesive connections.

Due to the uncertainties presented above, it is decided to perform an experimental verification of the capacity of the designed connections. These experiments of the bolted corner connection and the adhesive corner connection are presented in this report. In addition to this, three of the internal angle profiles are tested separately, to ensure that these have the expected capacity.

Due to the limited amount of test specimens, the verification done here can only be regarded as indicative, as variations in material properties, assembly of the connection and uncertainty in measurements could give variation from connection to connection, possibly making the test results unrepresentative. In order to determine a reliable capacity of the connections and un- derstand the variations that occur, more test should be performed.

1

(7)

Chapter 2

Theory

As this Experimental Report is to be read in connection with the Main Report, the work within this report relies on the theory presented in the Main Report. Theory concerning GFRP or the approaches used to determine the capacity will therefore not be given here. The only theory that will be presented is connected to the experiments and the expected capacities. As the characteristic material properties given for GFRP are based on the lower 5% fractile, it is likely that failure will happen above the characteristic value of the material. In the following, a method will be presented to estimate approximately how much larger the mean capacity will be compared to the characteristic values.

2.1 Estimation of mean material strength

The strength properties used in the work of this thesis are based on values from Fiberline’s design manual ([D4]) and are characteristics values. These value are, as in Eurocode, based on the lower 5% fractile of the material strength. This is illustrated in figure 2.1. This plot is for concrete specimens and assumes a normal distribution. Assuming that the properties of GFRP likewise follow a normal distribution, the relation between the characteristic value and the mean value can, in accordance with [D2], be determined by:

r_k=rm−1.64SD (2.1)

Where, rk is the characteristic strength property,rm is the mean of strength property and SD is the standard deviation. The determination of the partial safety factor should according to [B1] be based on an aim of providing a probability of failure below the design strength of 10⁻³. This is fulfilled when the design resistance is 3.04 times the standard deviation, SD, away from

2

(8)

Figure 2.1: Illustration of material strength distribution and characteristic strength (from [D2]).

the mean strength,rm, in the normal distribution:

rd=rm−3.04SD (2.2)

The relation between the characteristic value and the design value is known, being the safety coefficient of γGF RP = 1.3 for the GFRP elements, as all tests will be short term loading, and γ_Adhesive = 3.0 for the adhesive. The magnitude of the mean strength can thus be estimated based on the characteristic strength by solving equation (2.1) and (2.2). This gives the estimated mean strength values shown in table 2.1

Table 2.1: Relation between design, characteristic and estimated mean strength

Design strength Characteristic strength Estimated mean strength

fd fk fm

GFRP f_k/1.3 1.0f_k 1.27f_k

Adhesive f_k/3.0 1.0f_k 1.78f_k

The values in the table give an indication of how much larger the mean strength will be compared to the characteristic strength. In principal, it is most probable that the test specimens can withstand the applied load that corresponds to the hardest loaded part of the test specimen being loaded until its estimated mean strength. However, the estimation of the mean strength here is somewhat theoretical, assuming a perfect normal distribution and that the safety coefficient is determined to give a failure below the design value with a probability of 10⁻³. The estimated mean strength shall therefore only be seen as indicative, in order to give an under- standing and quantification of why it is expected that failure is experienced significantly above

(9)

Chapter 1. Introduction 4 the characteristic or design strength. It is also likely to obtain higher capacities due to the design being carried out based on a lower bound theorem, striving to make simplification to result in a safer design.

In the following chapter the preparations done for the experiments is presented, including the test setup, creation of the test specimens, as well as a calculation of the expected capacities from the test. These expected capacities will result in an estimation of the applied failure load when using the design strength, characteristic strength and estimated mean strength, presented in this section. Hereby, it can be evaluated to what extent the actual failure loads from the experiments vary from the expectancy and whether these failure loads are unexpectedly low or high. The capacity is also calculated here due to some minor changes in the geometry of the actual test specimen compared to the designs presented and calculated in the Main Report, resulting in some variation in the expected capacity.

(10)

Experimental preparations

In this chapter the preparations done for the experiments are presented. Three different elements have been tested, being the internal angle profiles used in the corner connections, the bolted corner connection and the adhesive corner connection. For each of these three types of test specimens, the test setup is explained, followed by the creation and final geometry of the test specimens, as well as the expected capacity of the actual test specimens, calculated based on the design, characteristic and estimated mean capacity.

3.1 Angle profiles

Three separate angle profiles, identical to the ones used in the corner connections, were created to be tested separately. This was to ensure that the capacity of these internal angles was as expected, whereby a possible weakness in the connection could be identified.

3.1.1 Test setup

The angle profiles are tested through a tension test, pulling in the ends of the profile, creating a combined moment, axial and shear force in the corner of the angle profile where failure is expected to occur.

The angle profile is attached to a fork profile of steel, connected to the test machine and bolted to the angle profile, as illustrated in figure 3.1. The force is applied by a test machine with a capacity of 100 kN.

5

(11)

Chapter 2. Experimental preparations 6

PRODUCED BY AN AUTODESK EDUCATIONAL PRODUCT

Figure 3.1: Principle for test setup for angle profile.

3.1.2 Creating the test specimens

The angle profiles were created at Fiberline Composites, by cutting out the angle profile of 6 mm thick strips in the complete shape of the angle profile and then gluing them together with alternating directions of pultrusion. The strips were glued together using Sikadur330, as for the adhesive connection. The properties of this glue are described in the Main Report, and the datasheet is attached there as an appendix. The result is a profile with equal mechanical properties around both its axes, symmetrical around the corner of the profile. One side of the angle profiles were cut to give a geometry that match the ribs of the MD Plank.

Due to the fabrication of the angle profiles, the dimensions were changed slightly to ease the production process. The angle profiles were thus made as shown in figure 3.1, with the corresponding values as shown in table 3.1 and 3.2. The dimensioned which the angle was designed for, is given as ’Design’, while the actual dimensions of the three different angles likewise are given.

It is seen that all the dimensions are made very closely to the specifications given by the design dimensions. Most of these dimensions were measure with a calliper, with an exception of the larger dimensions d₂, d₃, d₅, d₆ and d_a, which were measured with a ruler, which is why there is no decimal to these. The dimension da represents the moment arm and thus based on the drilled holes. The largest deviations are from Angle 2A d₉ and d₁₀, which are due to the notch

(12)

Figure 3.2: Drawing specifications of angle profile from Fiberline Composites (Based upon drawings from Fiberline).

Table 3.1: Design and actual plan view dimensions and mass of angle profile

d1 d2 d3 d4 d5 d6 da mass [mm] [mm] [mm] [mm] [mm] [mm] [mm] [kg]

Design 84.5 415.5 415.5 84.5 500 500 280 4.76

Angle 1 84.7 409 407 84.7 494 492 283 4.61

Angle 2 84.4 408 409 84.8 492 494 282 4.51

Angle 3 84.8 410 409 84.8 495 493 282 4.70

Table 3.2: Design and actual cross-section view dimensions of angle profile.

d7 d8 d9 d10 d11 d12

[mm] [mm] [mm] [mm] [mm] [mm]

Design 35.5 84.5 6.5 4.0 6.0 10.0

Angle 1A 34.5 84.7 6.4 3.2 5.0 9.6

Angle 1B 34.7 84.7 5.8 3.8 5.9 9.9

Angle 2A 34.5 84.8 9.8 4.2 5.8 10.2

Angle 2B 34.7 84.4 6.1 3.4 5.1 9.4

Angle 3A 34.5 84.8 6.7 3.8 5.4 9.7

Angle 3B 34.7 84.8 6.5 4.0 5.7 9.5

(13)

Chapter 2. Experimental preparations 8 being slightly damaged, as illustrated left in figure 3.3. This notch might partly be the reason of the lower mass of this Angle 2, seen in table 3.1. The reduced mass for Angle 1 could be due to some damage at the end of the profile, as shown right in figure 3.3. Despite these small variations and damages, it is not expected that this will result in significant variation in regards to the test results, and the expected values will therefore be calculated based on the design dimensions.

Figure 3.3: Picture of minor damages of angles: Left: Angle 2 with increased notch, Right:

Angle 1 with damaged end.

Pictures of the actual angle profiles are shown in figure 3.4. The picture to the right shows how bolts are connected through the hole drilled in the profile, which is how the profile is connected to the test machine.

Figure 3.4: Picture of the actual angle profiles. Left: Entire profile, Middle: Cross section, Right: Angle 1 with bolts

(14)

Due to the slight changes in the dimensions of the angle profiles, the expected capacity is also slightly reduced. However, this has little impact on the objective of the experiments, as it will still show whether the assumptions about the calculation of the capacity give reasonable values.

In this subsection the calculation of the expected applied failure loads will be described ending with a presentation of the expected values.

The actual test specimens will have the following cross-section properties:

A= 2914mm² Iy = 1.658·10⁶mm⁴ (3.1)

The characteristic capacity of the material with alternating direction of pultrusion will be:

f_b,k= f_b,k,90^◦+f_b,k,90^◦

2 = 240 MPa + 100 MPa

2 = 170 MPa, f_τ,k = 25 MPa (3.2)

The axial, moment and shear forces in the profile can be expressed as following, as a function of the applied force,F:

M_max =F·0.28 m, F_x=V = 1

√2F (3.3)

Here the 0.28 m are the distance from the line of action of the applied load and the centre of the angles corner, as illustrated in figure 3.1. The axial and shear force are based on the 45 degree inclination of the legs of the angle, relative to the direction of the load.

From the section forces, the largest stresses acting at the corner can be determined by:

σmax= Mmax

I_y h

2 +Fx,max

A , τmax = 3 2

V

A (3.4)

Hereh= 84.5 mm, being the height of the angles cross-section. The determination of the shear stress here is slightly simplified, as it assumes a rectangular cross-section.

Setting these stresses equal to the capacity and inserting the expressions for the section forces based on the applied load, the failure load based on the design, characteristic and mean capacity can be determined. The mean capacity is estimated based on the method presented in section 2.1. These applied failure loads are given in table 3.3, and do not take interaction between the shear and normal stresses into account. It is found that the normal stress is most critical, and this is determined assuming the entire normal stress being due to bending. In table 3.3 the expected displacements, corresponding to the failure loads are also given. These displacement are calculated using STAAD.Pro and using the average modulus of elasticity between the direction

(15)

Chapter 2. Experimental preparations 10 of pultrusion and the direction perpendicular to the direction of pultrusion, calculated by:

E = E₉₀^◦+E₀^◦

2 = 23 GPa + 8.5 GPa

2 = 18.3 GPa (3.5)

Table 3.3: Expected applied failure load for angle profiles dependent on capacity type

Expected failure load Expected displacement at failure load

F, [kN] u, [mm]

Design capacity 16.3 17.1

Characteristic capacity 21.2 22.2

Estimated mean capacity 26.9 28.2

It is expected that all the three individually tested angle profiles will be able to withstand a load higher than that corresponding to the design capacity, as it would otherwise indicate a lower material strength than that used in general design. Failure loads above that corresponding to the design capacity are increasingly likely until the failure load corresponding to the estimated mean capacity is met. This is under the condition that the assumptions for estimating this value are correct. According to the method used to determine the capacity of the angle profile, failure loads above that corresponding to the estimated mean capacity will be increasingly unlikely. However, there might be contributions from the glue within the angle profile, which are not taken into account, affecting the capacity. Likewise wrong assumption concerning the alternating direction of pultrusion might results in lower capacities. The displacements are expected to be similar to the values given in table 3.3, with a displacement shape as shown in figure 3.5, where the corner remains perpendicular.

Figure 3.5: Expected displacement shape of Angle profile, from STAAD.Pro (Exaggerated by a factor 2 from estimated mean capacity failure load).

(16)

point to the steel fork profile, or failure of the steel fork profile itself. To control that the failure will happen due to bending of the angle, strength verifications are performed for both the connection to the fork profile and for the steel fork profile itself. The following results are found, using steel calculations based on [D3]:

• A M20 bolt through the GFRP profile, with 35.5 mm thickness, with alternating direction of pultrusion, will have a capacity ofFdesign= 60.1kN. This is calculated based on methods presented in the Main Report.

• The bolt from the fork profile to the test machine is subject to tension, and will have a capacity of F_t,Rd = 0,765nA_sf_ub/γ_M2 = 62.5kN. Here n = 1 is the amount of bolts, As= 245 mm² is the stress area of the bolts,f_ub= 400 MPa is the design tensile strenght of the bolt and γ_M₂= 1.2 is the safety coefficient.

• With an underlying washer with a diameter of 40mm, on top of a plate of 10mm, the punching shear capacity is: Bp,Rd = 0.6nπdmtpfu/γM2= 84.8kN. Herdm= 40 mm is the washer diameter andt_p = 10 mm is the material thickness, while the rest are as described for the tensile capacity.

These are considered the likeliest failure modes and it is seen that all these failure modes have significantly more than thrice the design capacity of the angle profile due to bending. It shall be noted that the M20 bolt going through the angle profile does not meet the minimum distance requirements from [D4]. It is estimated that this will not be critical, as more than three times the needed capacity is present, disregarding the minimum distance. The risk is however considered and strengthening of the connection will be enforced, if failure occurs here in a test specimen.

3.2 Bolted corner connection

A single full scale bolted corner connection is planned to be tested. As for the angle profile, the test setup, specimen creation and calculations of expected capacities will be preformed in the following three subsections.

3.2.1 Test setup

The corner connection is planned to be tested in a custom designed test setup, with a rigid support and a point load. The load is positioned with a distance from the corner that gives the

(17)

Chapter 2. Experimental preparations 12 same relation between shear and moment load in the connection, as will occur in the frame of the Main Report from an evenly distributed load. Isometric views of how the test setup was planned are shown in figure 3.6. A section of the test setup with the main dimensions is shown

Testspecimen 200 kN press

Transverse support

Wood piece for light touch of specimen, to support again transverse deformations

Load distributing profile

Threaded rod to ensure rigid support and wood plates for transverse positioning.

Rigid support

Figure 3.6: Isometric view of test setup for test of corner connection.

in figure 3.7.

Figure 3.7: Section with dimensions of corner connection.

3.2.2 Creating the test specimen

The test specimen of the bolted corner connection was created at Fiberline Composites and consists of five of the internal angle profiles described in section 3.1. The overall dimensions of

(18)

Figure 3.8: Overall layout and dimensions of test specimens for corner connection (Drawing from Fiberline).

A total of 8 holes were drilled in each of the internal angle profiles and correspondingly in the main profiles of the test specimen. While the design was done for M18 bolts, the actual test specimen was created with M12 bolts with a 18 mm sleeve, which will result in a similar load on the GFRP profiles, and as the capacity of the bolts themselves not is critical, this should not influence the overall capacity of the connection. As it was found advantageous to use the same angle profiles as for the adhesive joint, the dimensions of the angle profiles are slightly different than the original design in the Main Report, as is the bolt distribution. The bolt distribution in each of the internal angle profiles is shown in figure 3.9.

Figure 3.9: Bolt layout and dimensions of the internal angle of the bolted connection.

(19)

Chapter 2. Experimental preparations 14 This bolt distribution in each of the internal angle profiles result in the overall bolt layout in the test specimen as shown in figure 3.10.

Figure 3.10: Overall bolt layout of bolted corner connection test specimen (Drawing from Fiberline).

3.2.3 Expected capacity

The expected applied failure loads are given in table 3.4 dependent on the assumed capacity.

The failure loads are shown for failure due to the contact force between the GFRP profile and the bolt, as well as for failure of the internal angle profile. The force distribution as well as the capacity determination is performed in the same way as in the Main Report, using the actual bolt position and geometry of the angle profile.

Table 3.4: Expected applied failure load for bolted connection dependent on capacity assumptions and element of failure

Expected failure load Displacements at failure

F, [kN] u, [mm]

Failure from Failure of Vertical Horizontal bolt load angle profile at end at corner

Design capacity 26.6 21.9 25.3 4.0

Characteristic capacity 34.6 28.5 32.9 5.1

Estimated mean capacity 43.9 36.2 41.7 6.5

(20)

which corresponds to an applied load of 19.5 kN. Because the bolts are moved slightly further from the centre of the connection due to the longer angle profile, the largest bolt force is slightly lower than in the original design, resulting on a slightly higher capacity based on the direct bolt contact. However, the increased length of the angle profile increases the moment at the corner of the angle profile, resulting in failure of the angle becoming more critical. The expected applied loads are estimated when taking the self-weight of the test specimens and the superimposed dead load caused by the force distribution profile into account. The contribution from the dead loads reduces the failure load by 0.3 kN.

For the critical failure loads of the angle profile, the corresponding expected displacements are given, which are estimated using STAAD.Pro. The deflection shape is shown in figure 3.11, where the location of the expected displacements are shown with red dots. These two loca- tions for measuring horizontal and vertical displacements respectively, are found to be the most interesting. The horizontal displacement at the corner together with another horizontal measurement along the column, will show whether the rigid support is sufficient. If this horizontal displacement is significantly larger than the expected value, it indicates that the support in the test setup has slipped. Having the additional horizontal measurement along the column, will also allow evaluation of whether the column deformation is linear along the height, indicating slipping of the support. The vertical displacement at the end shows whether or not the designed rigid connection acts truly rigid. If the vertical displacement is significantly larger than the expected value, it indicates that the connection only acts semi-rigidly. Apart from this, it will be of interest to observe how the test specimen acts in the corner, to see whether gaps occur in the tensile part, shear in the middle and compression in the bottom.

Figure 3.11: Expected displacements at failure for corner connection specimen (Exaggerated by a factor 10 from estimated mean load).

(21)

Chapter 2. Experimental preparations 16

3.3 Adhesive corner connection

The final experiment is the adhesive corner connection, which will be tested using the same test setup as for the bolted connection. As for the other experiments, a description of the creation of the test specimens will be given, as well as a presentation of the expected failure load and location.

3.3.1 Test setup

The test setup is identical to that for the bolted corner connection. Reference is made to subsection 3.2.1.

3.3.2 Creating the test specimen

The overall geometry is identical to that of the bolted connection, shown in figure 3.8. As seen to the left in figure 3.12, the angle profiles fit very well in one side of the MD Plank, while, as seen to the right, there is a gap to the other side. As a result, the angle profile can only effectively be glued to a reduced area of the MD Plank.

Figure 3.12: Picture of the fit of the angle profiles within the MD Plank.

The Sikadur330 glue was distributed with a paint roller. This method does not allow a accurate distribution or even adhesive layer thickness. To control the adhesive layer thickness one could use high precision glass beads, as was done in [C1]. However, this was not practically possible here, and would not represent the actual implementation of such an adhesive connection.

When glue was distributed to all inner surfaces of the MD Planks and the corresponding sides of the angle profiles, the angle profiles were positioned and MD Planks put together. It was attempted to avoid sliding the angle profiles into place, as this would scrape the glue away.

Instead the angle profiles were tilted into place, being possible due to the gap at one side between

(22)

Figure 3.13: Picture of the angle profiles in place and glue distributed on the surface.

The glue was likewise distributed to the 6 mm strip profiles after which it was carefully placed on the MD Plank and internal angles. When placed, pressure was applied during the hardening of the glue. Again, the applied pressure will influence the thickness of the adhesive layer, while evening out the distribution. The gluing of the profile was performed 20th of May 2015, and according to the Datasheet of Sikadur330, the subscribed mechanical properties are obtained after seven days hardening.

The design was done based on a thickness of the adhesive layer of 2 mm, which, after having worked practically with the adhesive, is found to be larger than what will be obtained through the used method. Applying pressure to the connection during hardening, will further decrease the adhesive layer thickness. In the following subsection the expected capacity will therefore be based on a reduced adhesive layer of 1 mm.

3.3.3 Expected capacity

The expected failure loads for the adhesive connection are shown in table 3.5. The failure loads are calculated based on the methods described in the Main Report, and calculated based on both the presented approaches for the adhesive connection, as well as for failure of the angle profile. The capacities here are based on the actual geometry of the angle profiles, reducing the adhesive area, and with a adhesive thickness of 1 mm.

It is seen that when using the design capacity, failure will first occur in the adhesive layer, according to approach 1. However, as the adhesive capacity operates with a larger safety coefficient, failure is expected to occur in the angle profile when using the characteristic or estimated mean capacity. Using approach 2 for determining the capacity of the adhesive connection, the estimated failure load is much larger than when using approach 1, as already discussed in the

(23)

Chapter 2. Experimental preparations 18 Table 3.5: Expected applied failure load for adhesive connection dependent on capacity as-

sumptions and element of failure

Expected applied failure load Displacements at failure

F, [kN] u, [mm]

Adhesive failure; Adhesive failure; Failure of Vertical Horizontal approach 1 approach 2 angle profile at end at corner

Design capacity 14.2 74.4 16.3 16.4 2.6

Characteristic capacity 43.3 96.8 21.5 24.5 3.9

Estimated mean capacity 77.3 123.0 27.3 31.2 4.9

Main Report.

Most likely, failure will occur in the angle profile, unfortunately meaning that the precision of the two approaches relative to each other will be further discovered. As for the bolted connection, the expected applied loads are estimated including the self-weight of the test specimens and the superimposed dead load caused by the force distribution profile. The effect of these dead loads, reduces the load by 0.3 kN.

The expected displacements are also given in table 3.5, for the most critical load case for each assumed capacity. The same considerations as for the bolted connection are valid, and the deflection shape is likewise expected to be as shown in figure 3.11.

(24)

Conduction of experiment

4.1 Angle profiles

The angle profiles were tested in an Instron 8521 machine with a capacity of 100 kN. The tests were performed 29.05.2015 and 01.06.2015. The actual test setup is shown in figure 4.1, where the angle profile is seen fixed to the two steel fork profiles. The top bridge of the test machine remains fixed in the set position and the test specimen is pulled from below.

Figure 4.1: Actual test setup for angle profile.

19

(25)

Chapter 3. Conduction of experiment 20 The applied load was set to be displacement based. This displacement was measure through the Instron 8521 machine itself. As a consequence of this, contributions to the displacement from slipping of the grips on the fork profiles, yielding of these fork profiles and local deformations of the holes of the angle profiles could occur. To have a control of the magnitude of this displacement compared to the actual deformations of the angle profile, a ruler was attached to the angle profile directly, giving a indication of the magnitude of the displacement contributions.

A more precise setup could be made, using strain gauges and a data logger directly on the test specimen. However, as the bearing capacity was the main parameter of interest, it was decided that the used approach was satisfactory. The main test parameters are shown in table 4.1.

Table 4.1: Expected applied failure load for angle profiles dependent on capacity type

Sample rate 64 Hz Load rate 0.033 mm/s

During the test of the first angle profile, significant yielding of one of the fork profiles occurred, which resulted in visible influences on the measured displacements. To avoid the measurements of the other angle profiles being affected by the yielding, the fork profile was strengthened through welding after the first test. The yielded fork profile is seen left in figure 4.2, while the strengthening welding is shown right in figure 4.2.

Figure 4.2: Left: Yielding of the steel fork profile, Right: Strengthening welding of steel fork profile.

(26)

The bolted corner connection was tested on 10th of June 2015. In the prior to the experimental several weaknesses in the test setup was discovered, being the capacity of the fixed support at the ground and the stiffness of the transversal supports against buckling close to were the load is applied. The fixed support was therefore extended to reach a total height of 90 cm, while steel profiles were used for the transversal support, with bolts toughing the profile to constained the transversal position. Both these reinforcements are shown in figure 4.3.

Figure 4.3: Reinforced test setup for bolted corner connection, Left: Fixed support, Right:

Transverse support.

As load was applied significant rotation occurred in the bolted corner and large gaps appeared in the top of the connection, as shown in figure 4.4. This resulted in a large increase in the displacement at the end of the specimen, where the vertical displacement was measured. The displacement hereby exceeded the limit of the displacement transducer, resulting in the final displacements being measured manually.

Figure 4.4: Rotation at failure of the bolted connection.

(27)

Chapter 3. Conduction of experiment 22

4.3 Adhesive corner connection

As mentioned in section 3.3, the test setup for the adhesive connection is identical to that of the corner connection. Based on the experiences from the bolted connection the same reinforcements of the fixed support and transverse support was made. In addition, the vertical displacement transducer was placed on the centre of the beam, where the displacements will be smaller. The position of the displacement transducer is shown in figure 4.5.

Figure 4.5: Adhesive corner connection test specimen.

The adhesive connection was tested the 15th of June 2015.

(28)

Test results

5.1 Angle profiles

The results from the test of the three angle profiles are shown in figure 5.1. Here the applied load is plotted as a function of the displacement.

0 10 20 30 40 50 60 70

0 5 10 15 20 25 30 35

Displacement of test machine [mm]

Applied load [N]

Design capacity Characteristic capacity Estimated mean capacity Angle 1

Angle 2 Angle 3

Figure 5.1: Plot of results from test of angle profiles.

23

(29)

Chapter 4. Test results 24 As mentioned in section 4.1, the displacement is measured by the Instron 8521 testing machine itself, and is thus the relative displacement of the press. Based on the measurements from the ruler attached to the angle profile, superposed to the location of the holes, assuming the profiles remain straight, it is found that the measured displacements of Angle 2 and Angle 3 corresponds well with the actual displacements of the angle profile. This is valid when all initial measurements are discarded until a consequence load increase occurs, as the measurements until this displays local deformations around the connection. These first measurement were discarded for all the three tested angles, until a sustained load increase of 0.05 Newton was acquired. For Angle 3 it is found that the measured displacement around failure is significantly overestimated compared to what can be estimated based on the ruler. This is largely due to the local yielding of the steel fork profile, mentioned in section 4.1. When examining the plot in figure 5.1 it is clearly seen that the yielding occurs at a load of about 5 kN, resulting in a flat plateau, where the displacement increases but the load remains constant. Generally the curve of Angle 1 is not as straight as the plots for the other angles.

From the plot in figure 5.1, it is seen that all the angles were able to sustain a load larger than the characteristic capacity, while Angle 3 could sustain a significantly higher load than the others, exceeding the estimated mean capacity. All the results are therefore within the expected range, while the average is slightly below what was expected. The plot is made with crosses, circles or triangles representing the actual measurement points. Due to the high sample rate, the plots mainly appear as thick lines. However, immediately after the failure, some measurement points can be distinguished. This shows a very sudden drop in capacity after failure, as the capacity is more than halved for all the angles within a few measurements. With a sample rate of 64 Hz, this means that this large reduction of capacity happens within less than a second. This corresponds to the theory given in the Main Report, stating an elastic material behaviour with a drastic decrease in stiffness after failure occurs. The failure is thus shown to be very brittle, being a problem when implemented in building, as no warning is given before failure occurs.

Apart from Angle 1, it is seen that the behaviour is almost perfectly elastic until failure.

With a star, square and diamond, in figure 5.1, the expected displacements are shown corresponding to the expected design, characteristic and estimated mean capacity, as shown in table 3.3. It is seen that the displacements of Angle 2 and Angle 3 correspond very well with these expected values.

The maximum sustained load for each of the angle profiles are shown in table 5.1, together with the corresponding displacement. Setting the results in increasing order after the maximum sustained loads, it is noticed that it corresponds with the mass of the angle profiles, shown in table 3.1. Angle 2 is thus the lightest of the profiles, and also the one which could sustain the lowest load, while Angle 3 is the heaviest and the one sustaining the highest load. As the measured dimensions are very similar it is not found likely that the small variation that

(30)

material. It is seen that the average maximum sustained load is 24.5 kN, being slightly lower than the estimated load based on the mean capacity of 26.9 kN, as given in table 3.3. The average displacement is on the other hand significantly higher, than that corresponding to the estimated mean capacity. This is due to the large displacements occurring in Angle 1 due to the yielding of the steel.

Table 5.1: Applied failure load for angle profiles.

Max. load Displacement at max. load

F u

[kN] [mm]

Angle 1 23.88 37.4

Angle 2 22.06 24.7

Angle 3 27.70 29.1

Average 24.5 31.5

In figure 5.2, the failure of angle profile 2 is shown. Here it is seen that the failure happens in the compression loaded part of the angle, which is unexpected, as the normal force in the angle is tensile, meaning that the stresses in the tensile part of the cross-section will be largest.

The is even more surprising, because the tensile strength according to [D4] is lower than the compressive perpendicular to the direction of pultrusion, while they are identical parallel to the direction of pultrusion. Despite this, the failure is clearly compressive, and looks like buckling of the outer two plates and splitting of the inner plates, resulting in delamination of the GFRP.

This could indicate, that the adhesive connection from the outer plates failed, making these plates buckle, leaving the remaining cross-section reduced, resulting in the splitting failure.

Apart from being due to compression rather than tension, the failure occurred in the expected location, where the moment is largest while the cross-section remains unchanged. The same failure mode occurred in all the angle profiles.

Since all the angle profiles could sustain loads larger than that corresponding to the characteristic strength, it is expected that the angle profiles in the corner connection likewise will have sufficient capacity to avoid failure at lower loads than expected due to failure of the angle profiles.

(31)

Chapter 4. Test results 26

Figure 5.2: Failure of angle profile 1.

5.2 Bolted corner connection

As mentioned in section 4.2, the vertical displacement of the bolted connected exceeded the limit of the displacement transducer, why the displacement at failure was measured directly.

As a result few data point exist for a large part of the data range. The plot based on the direct measurements are shown in figure ??.

Figure 5.3: Plot of the direct measurements for the bolted connection.

(32)

can be corrected appropriately. When this is done, the plot shown in figure 5.4.

Figure 5.4: Plot of the measurements for the bolted connection corrected for support displacement.

With a star, square and diamond, the estimated capacity is shown as a function as the estimated displacement at this load. It is seen that the bolted connection could sustain a much larger load then expected, but that the corresponding displacement likewise was much larger than expected. This indicated that the bolted connection is not completely rigid, but rather acts as a semi-rigid connection with some rotation occurring.

The failure load and displacement, after correction for the support deformations, are shown in table 5.2.

Table 5.2: Applied failure load for bolted connection.

Max. load Displacement at max. load at end point

F u

[kN] [mm]

Bolted connection 53.1 212.8

(33)

Chapter 4. Test results 28 The much larger capacity is to a large extent believe to be due to the capacity being estimated based on a lower bound approach, where the calculations are simplified by not taken certain effects into account. As a result of this, the contact pressure between the beam and column profile was conservatively disregarded. However, as seen on the failure mode in figure 5.5 the failure is dominated by this contact pressure, showing that this has a large contribution.

Figure 5.5: Failure of the bolted connection.

5.3 Adhesive corner connection

As mentioned in section 4.2, the vertical displacement for the adhesive connection was measured at the mid of the beam rather than at the end, to be sure that the displacement would not exceed the limit of the displacement transducer, as was the case for the bolted connection. The measured load as a function as vertical displacement is shown in figure 5.6.

As for the bolted connection, this plot is corrected for the deformations at the support, resulting in the plot of figure 5.7.

As for the bolted connection, it is seen that the adhesive connection sustains a much larger load than expected, but unlike the bolted connection, the relation between that load and the displacement is as expected meaning that the connection acts as a truly rigid connection. As seen in figure 5.8, the adhesive connection fails due to the contact pressure between the column and the beam, and as this contribution was not taken into account in the calculations, it explains why the capacity was larger than expected. In the figure, it is likewise seen that there are no gaps at all along the connection line, explaining why the displacements are as for a truly rigid connection, in contrast to the bolted connection.

While the vertical displacement was measured at mid of the beam, the measurements can be used to estimate the displacement at the end. In table 5.3, the load occurring the first failure is given together with the corresponding displacement, estimated for the end of the beam.

(34)

Figure 5.6: Plot of the direct measurements for the bolted connection.

Table 5.3: Applied failure load for adhesive connection

Load at first failure Displacement at load at first failure at end point

F u

[kN] [mm]

Bolted connection 42.1 52.6

In following chapter a brief discussion of the found results are given, as well as plot showing the results from the bolted and adhesive connection together.

(35)

Chapter 4. Test results 30

Figure 5.7: Plot of the measurements for the bolted connection corrected for support displacement.

Figure 5.8: Failure of the adhesive connection.

(36)

Discussion and Conclusions

In this chapter the results found from the experimental work will be discussed and conclusion will be made.

During the creation of the test specimens, some changed were made to the design presented in the Main Report. This was partly a decrease of the cross-section size of the internal angle. This obviously resulted in a lower strength of the internal angle profiles, than that used in the Main Report. The capacity was calculated based on the actual dimensions of the angle profiles, and the test results showed good compliance with these calculations, both in regards to capacity and deflection. All the capacities of the test specimens were thus above that calculated when using the characteristic value, while the average was slightly lower than the estimated mean capacity.

It is therefore expected that the calculations done for the angle profiles in the Main Report are valid and are considered verified by the experiments.

Both the bolted connection and the adhesive connection showed capacities significantly above that expected, which is due to the failure being a result of the direct contact pressure between the beam and column profile, which was a failure mode which was a contribution to the capacity which was not taken into account. The bolted connection showed much larger displacements than expected, showing a semi-rigid behaviour, while the adhesive connection showed expected displacements, showing a truly rigid behaviour. The behaviour can be compared through the plot in figure 6.1.

As a result of these experimental results, it can be concluded that both the internal angle profiles, bolted and adhesive connection have a equal or higher capacity than expected. The only main difference between the expected and the actual results was the displacements of the bolted connection, which were much larger than expected. As a results of this, the bolted connection should be recalculated as a semi-rigid connection. Both the bolted and adhesive

31

(37)

Chapter 5. Discussion and Conclusions 32

Figure 6.1: Plot of the measurements for the bolted connection and adhesive connection for displacements at end of beam corrected for support displacement.

connection could furthermore be recalculated taking the contact pressure into account, which would result in a higher verified capacity, closer to those found in the experiments.

(38)

Bibliography

7.1 Books

[A1] Lawrence C. Bank. COMPOSITES FOR CONSTRUCTION: Structural Design with FRP Materials. JOHN WILEY & SONS, INC., 2006.

7.2 Articles

[B1] Mike Byfield & David Nethercot. Safety variations in steel designed using Eurocode 3.

JCSS, 2001.

7.3 Reports and Theses

[C1] Till Vall´ee. Adhesively bonded lap joints of pultruded GFRP shapes. Technische Hochschule Darmstadt, 2004.

7.4 Codes and Standards

[D1] John L. Clarke. Structural design of Polymer Composites, EUROCOMP Design Code and Handbook. E & FN SPON, 2005.

[D2] P.Bamforth et. al. Properties of Concrete for use in Eurocode 2. The Concrete Centre, 2008.

[D3] Jesper Gath. Bolted Connections. Alectia, 2009.

33

(39)

Bibliography 34 [D4] Henrik Thorning. Fiberline Design Manual. Fiberline Composites, 2003.