You are on page 1of 30

``````````````````````````````````````````````````````````````````````````````````````````````````

Scientific report 3

Beam to column
connections

EXPERIMENT
RESULTS
PhD Student:
Drd. Ing. Marius George MOLDOVAN

Scientific adviser:
Conf. Dr. Ing. Mihai NEDELCU

October 16, 2019


Beam to column connections – EXPERIMENT RESULTS v1.0

Table of Contents

1 General overview of the experiment .............................................................................................. 1


1.1 Short description of the setup ................................................................................................ 3
Description of the steel assemblies that were designed and produced for this test ..................... 4
Measuring equipment placement................................................................................................... 6
1.2 Number of test and variations in the assemblies ................................................................... 9
First monotonic test ........................................................................................................................ 9
First cyclic test ............................................................................................................................... 11
Second cyclic test .......................................................................................................................... 11
Third cyclic test ............................................................................................................................. 12
Fourth cyclic test ........................................................................................................................... 13
1.3 Testing assembly preparation ............................................................................................... 13
Step 1 ............................................................................................................................................ 13
Step 2 ............................................................................................................................................ 13
Step 3 ............................................................................................................................................ 14
Step 4 ............................................................................................................................................ 14
Step 5 ............................................................................................................................................ 15
Step 6 ............................................................................................................................................ 16
2 Test results and initial observations ............................................................................................. 16
2.1 Initial monotonic test ............................................................................................................ 16
2.2 First cyclic test ....................................................................................................................... 19
2.3 Second cyclic test .................................................................................................................. 22
3 Initial conclusions.......................................................................................................................... 28
Beam to column connections – EXPERIMENT RESULTS v1.0

1 General overview of the experiment

The purpose of this research is to offer a comprehensive and detailed behaviour of precast
concrete connections in industrial buildings during an earthquake. In order to achieve this goal
a series of laboratory test have been performed. The type of connection that have been
investigated during this research has at end of the column a fork that restrains the beam from
overturning, in which there are embedded mechanical connectors (steel dowels) (Figure 1). The
most common structural system of the precast industrial buildings in Europe consists of an
assemblage of cantilever columns tied together with beams. Typical beam-to-column
connection in these structures is constructed with steel dowels. Although this system has been
used for decades, its seismic response was poorly understood, which reflected in ambiguous
code requirements and conservative approach. Simply speaking, the analysed system consists of
an assemblage of cantilever columns tied together with beams. The key element in this system is
the beam-to-column connection. In spite of the frequent use of the analysed precast system, the
information about its behaviour during earthquakes has been sparse and sometimes
controversial. One significant difference between the actual connection and the one tested is the
absence of horizontal dowels that go through the column fork and the beam web. It has been
decided to remove these dowels because the test was performed horizontally and it was
impossible to insert these dowels during installation. Also these dowels are not injected with
mortar in actual practice and they are supposed to keep the elements in place during the
construction phase, thus they cannot be considered as mechanical connectors for horizontal
actions and this was another reason for eliminating them during the experiment.

Figure 1 Column end

In normal situations one or two dowels are used which extend from the end of the column and
are inserted in sleeves inside of the beam end-part (Figure 2Error! Reference source not
found.). The use of two dowels provides better stability for overturning moments in the
construction phase when the load distribution is uneven (Figure 2).

1
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 2 Beam end

The behaviour of this connection under seismic action is not completely understood, as it is
designed under normal vertical actions.

Beam-to-column connections are extremely important for the integrity and safety of the precast
industrial structure. The majority of collapses during earthquakes are due to the fall of the
beam. Nowadays it is obvious that the connection should not rely only on the friction and that
some kind of mechanical connection should be provided. The most common solution is the use
of steel dowels.

Nevertheless, the design (if done at all) was predominantly based on engineering feeling
and the requirements of non-seismic loading. However, the correct approach would be the
use of the capacity design, which is in fact required by Eurocode 8. (Toniolo 2007)

For this we obviously need to know capacity of the dowel connection and the demand imposed
during seismic action. Capacity design dictates that the connection should have an elastic
behaviour under seismic horizontal actions to allow for the dissipation mechanisms to develop
in the desired area. If this connection fails the entire structure is compromised and may lead to
a premature, partial or even total, collapse. The 2012 Emilia earthquake caused extensive
damage to precast industrial buildings resulting in economic losses in the order of billions, most
of the damage being the result of inadequate connection design. In some cases of pinned
connections with mechanical connectors, where failure should coincide with the yielding of the
dowels it was observed that the small concrete cover combined with insufficient stirrups
produced a premature collapse of the structures (Figure 3 Left). The breakdowns of even one
connection can cause a progressive collapse of the entire structure (Figure 3 Right).

2
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 3 (Left) Pinned beam-to-column connection failure and (Right) consequent loss of support of the beam
from column

When performing a traditional strength and behaviour research, it is necessary to perform at


least three tests on identical assemblies to obtain a statistically relevant result. In this case, in
order to maximize the information we obtain from this first set of test, we have included some
variations in the assemblies. A number of five assemblies were tested.

1.1 Short description of the setup

The tests were performed with the elements positioned horizontally. The benefits were that we
were no longer required to install the measuring equipment and the hydraulic press at heights
of over 8 m. However this has induced some additional friction on the sides of the column and
beam that were supported on the floor (Figure 4).

Figure 4 Testing setup

3
Beam to column connections – EXPERIMENT RESULTS v1.0

Description of the steel assemblies that were designed and produced for this test

The beam was supported on two steel assemblies that elevate it to the position required for it to
be positioned in the center of column fork (Figure 5).

Figure 5 Steel assemblies - Beam supports

The force was transmitted from the hydraulic jack to the concrete elements using a steel frame
(top right in Figure 4). This part was made up from three individual assemblies:

• Two frame supports – the same as in the case of the beam, the force transmitting frame
had to be elevated from the floor to the center of gravity of the beam to avoid any load
eccentricities (Figure 6).

Figure 6 Steel assemblies - Frame support.

• Force transmitting frame – this was used to push and pull the end of the beam (Figure

4
Beam to column connections – EXPERIMENT RESULTS v1.0

7). The hydraulic jack was moved (left/right) according to the desired loading direction.
In the middle of the frame a steel reaction element was placed for the hydraulic jack.

Figure 7 Steel assemblies - Force transmitting frame

• Frame to Beam end fixing – this was designed to be a hinged connection to avoid any
unwanted eccentricities and bending moments in the end of the beam (Figure 8).

Figure 8 Steel assemblies - Frame to Beam end fixing

The base of the column was fixed into position using two L shape steel assemblies that were
designed to replicate a fixed support (Figure 9 and Figure 10).

5
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 9 Steel assemblies - Column base fixing

Figure 10 Steel assemblies - Column base fixings installed

Measuring equipment placement

A total of fifteen measuring gauges were installed on each tested setup. In the following part the
initial proposal will be detailed. The placement of the measuring equipment was slightly
changed for each test based on the behavior observed during the previous experiments. Some
strain gauges that recorded very small strains were moved closer to the dowel position.

• Strain gauges (total of 9 gauges)

o Side of column fork – 4 gauges (Figure 11)

o Top side of column fork – 1 gauge (Figure 12)

o Lateral sides of the column base – 2 gauges (Figure 13)

o Beam end – 2 gauges (Figure 14)

• Displacement gauge (total of 6 gauges) (Figure 15)

o Measuring column absolute displacement – 3 gauges

6
Beam to column connections – EXPERIMENT RESULTS v1.0

o Measuring beam absolute displacement – 1 gauge

o Measuring relative displacement between column and beam – 2 gauges

Figure 11 Strain gauges – Side of column fork

Figure 12 Strain gauges – Top side of column fork

Figure 13 Strain gauges – Lateral sides at the base of the column fork

7
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 14 Strain gauges – Beam end

Figure 15 Displacement gauges – Column and Beam

8
Beam to column connections – EXPERIMENT RESULTS v1.0

1.2 Number of test and variations in the assemblies

A total of five tests were conducted during this part of the research. The tests were performed
using either monotonic unidirectional or cyclic loading. Initially it was planned to have only one
monotonic test and the rest to be performed using a cyclic loading, but because of some
logistical issues and the behavior observed during the first test this has been changed and it will
be described in more detail later on.

The tests that were planed are the following:

First monotonic test


This test is necessary to calibrate the following cyclic tests. It will provide us an initial value for
the failure and maximum displacements to be expected. These values will be used to determine
the increments value for the cyclic tests and number of load cycles. The elements dimensions
can be seen in Figure 16:

Figure 16 Elements for monotonic test

In the SAFECAST report from 2012, on Design Guidelines for Connections of Precast Structures
under Seismic Actions, it is stated in the introduction that: “Any type of connection shall be
experimented with an initial type testing in order to quantify its strength and possibly the other
properties that affect its seismic behaviour. From this experimentation a design model may be
deducted, by means of which a verification by calculation can be applied on the different
connections of the same type. For a specific application one can refer to the available results of
previous experimentations like those provided in the following clauses or in other reliable
documents such as official regulations (Eurocodes, CEN product standards and Technical

9
Beam to column connections – EXPERIMENT RESULTS v1.0

specifications, …).”. Meaning that in order to use any type of precast connection, it is first
necessary to extensively study its behaviour through experimental and numerical testing.

The testing protocol proposed by SAFECAST states that two types of tests should be performed,
according to the type of loading used, monotonic loading and cyclic loading. The monotonic
(push-over) loading test is performed in order to deduce the force displacement diagrams
(Figure 17). In addition to a first quantification of the constitutive parameters, the push-over
test is performed also as preliminary, in order to define the loading steps of the subsequent
cyclic test.

Figure 17 Force-displacement diagrams - a ductile, b brittle, c over-resisting.

The experimental cyclic response is obtained by applying the load history described in Figure
18, where groups of three cycles of the same amplitude are performed step by step with
subsequent increments Δd up to the ultimate or test limit.

Figure 18 Loading history of the cyclic test

The main failure modes for dowel connections under horizontal actions are also described in
the SAFECAST guideline – longitudinal (a,b,c) and transversal (c,d,e) direction (Figure 19) of the
beam:

a. breaking of the dowel connection due to combined shear, tension and flexure on
steel bar and bearing stresses on concrete;
b. spalling of the concrete edge of the beam due to tensile stresses;
c. spalling of the concrete edge of the column due to tensile stresses;
d. flexural failure of the bearing section due to the action of bending moment;
e. pull-out of the tensioned dowel due to the action of bending moment;
f. sliding shear failure under the action of V ;

10
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 19 Forces in a dowel connection

The most detrimental failure mechanism is the one related to situation (b). This mechanism
corresponds to the situation in which the distance from the edge of the elements to the centre of
the dowel is smaller than six diameters. In order to reduce the negative effect of these small
distances, additional reinforcements should be provided near the dowel to prevent spalling.
Another important recommendation regards the dowels strength; they should be over-
proportioned, through capacity design, with respect to the critical sections at the base of the
columns.

First cyclic test


It will be performed using the same assembly as for the monotonic test (Figure 16). At this stage
we will already have two comparable results that will also give us an idea about the differences
obtained between the monotonic and cyclic test (which is expected to yield smaller strength
values)

Second cyclic test


For this we will modify the cross sections of the beam and column fork (Figure 20).

In common practice the Romanian business unit, beams have a larger “T” section in the
connection area, continuing with an “I” section (Figure 21). The shape of the end section of the
beam varies depending on several factors, such as shear force at the end of the beam and
pretensioning force. This modification reduces the concrete coverage for the dowels and we will
see how this affects the final behaviour.

11
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 20 Elements for second cyclic test

Figure 21 Beam dimensions for second cyclic test

Third cyclic test


We want to propose a modification for the last connection. As a result of Safecast recommendations, the
connection is provided with an extra dowel. This dowel protrudes trough the web of the beam (Figure
22).

Figure 22 Proposal for third cyclic test

12
Beam to column connections – EXPERIMENT RESULTS v1.0

Fourth cyclic test


This test was proposed by our coleagues from ASA Hungary. The dimensions of the elements are
the same as for the monotonic and the first cyclic tests (Figure 16). The only difference is that
instead of rigid steel dowels the elements are connected using flexible strands, the same as the
ones used for pretensioning concrete elements.

1.3 Testing assembly preparation


Step 1
• Installing the steel assemblies (Figure 23)

Figure 23 Positioning the steel assemblies

Step 2
• Positioning the concrete elements and fixing the column in place by tightening the steel
base fixings (Figure 24)

13
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 24 Positioning the concrete elements

Step 3
• After checking that everything is in its proper position; the dowels inside beam sleeves
were grouted to create the connection between the column and the beam (Figure 25).

Figure 25 Grouting the sleeves

Step 4
• Installing the install the strain gauges displacement transducers. For the first test
(monotonic) they were installed in the positions described in the initial proposal.

14
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 26 Installing strain gauges

Figure 27 Displacement transducers – Beam end and Column base

Figure 28 Displacement transducers – Column end and Relative displacement between column and beam

Step 5
• The hydraulic jack together with the force cell, were placed in position for the
monotonic test (Figure 29). In this case the beam was pulled away from the column,
which is the most unfavourable direction of loading and would give the minimum
strength of the beam to column connection based on the possible failure mechanisms
that were previously mentioned. Commented [TG1]: why?

15
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 29 Hydraulic jack and force cell

Step 6
• After all the gauges and sensors were installed, they were connected to two data
acquisition devices which offered us a total of sixteen channels to record for the whole
test (Figure 30).

Figure 30 Data acquisition devices

2 Test results and initial observations


2.1 Initial monotonic test

The first test was force controlled with a 10kN increment at each step until failure. After each
step the cracks and the connection were checked.

Unfortunately in the first test the strain gauges didn’t record any data due to a wrong setup of
the data acquisition devices. The data available is strictly based on force-displacement (Figure
31) measurements and crack widths.

16
Beam to column connections – EXPERIMENT RESULTS v1.0

Commented [TG2]: Translate into English the legend of the


graph

Figure 31 Force-Displacement - 1st test

Because of the configuration of the testing setup it is nearly impossible to eliminate completely
the friction between the structural concrete elements and the supports. For minimising as much
as possible this effect greased stainless steel plates were used in all the contact points between
the elements and the supports. Friction force was shown to be at around 25-30kN. The friction
force after this threshold continued to occur but with almost no influence on the test results.

Figure 32 1st test

Failure was reached at approximately 208kN with a maximum displacement of 182mm. It

17
Beam to column connections – EXPERIMENT RESULTS v1.0

should be noted that the displacement is much larger than expected because the base fixing
allowed for some rotation and the column was not completely fixed. This rotation will be
calculated and extracted later on for obtaining the absolute displacement.

Above the 120kN point the steel dowel (27mm threaded bar) started deforming. Yielding of the
bar vas reached at the force of 208.7kN which coincides with the failure point of the connection
(Figure 33). There was also some small crushing of the concrete around the dowel, and most
probably this would have increased if we wouldn’t have stopped the test because of safety
reasons and also to avoid any damage to the equipment.

Figure 33 Failure -dowel yielding

The maximum crack width at failure was observed at the base of the column in the area of the
cross section change (Figure 34). It had a value of 0.45mm, which shows that there was no
significant damage to the column and the failure of the assembly was strictly due to the
connection failure.

Figure 34 Position of maximum crack width

18
Beam to column connections – EXPERIMENT RESULTS v1.0

2.2 First cyclic test

For the first cyclic test the beam and column elements dimension were identical to the initial
monotonically test. The positions of the sensors were almost identical as before with just a
small change for one displacement transducer. The displacement transducer in Figure 35 was
positioned in the centre of the column base with the intent to monitor the unwanted base
rotation more accurately.

Figure 35 Displacement transducer for monitoring the base rotation

For the cyclic test the load displacement increments were chosen based on the initial monotonic
test and the Safecast testing protocol recommendations. Based on this the maximum
displacement from the first test was divided into three increment groups( 60mm, 120mm and
180mm).

In compression the whole strength of the assembly is larger than in tension because of the
contact between the concrete elements (beam end and the column fork). Based on this initial
observation it was expected that the failure will not occur in compression and the load cycles
were chosen in such a way that the final step to represent tension which is the most
unfavourable loading direction.

19
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 36 Force-Displacement - 1st cyclic test

The following graphs show the maximum loads at each step of the cyclic test and the force
reduction for intermediate equal displacements taken from each cycle:

Figure 37 Max. load Force - 1st cyclic test

20
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 38 Force variation for 6cm displacements - 1st cyclic test

Figure 39 Force variation for 12cm displacements - 1st cyclic test

As expected the failure has occurred in tension at 267kN with a maximum displacement of
aproximately 267mm (Figure 36). Failure was a combination of steel dowel yelding and
concrete spalling in the column fork, around the dowel (Figure 40).

21
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 40 Failure mechanism - 1st cyclic test

In this case the maximum crack width was slightly larger, 0.60mm, and in the same position as
for the first test (base of the column in the area of cross section change). This indicates that
there is no significant damage to the column and the failure of the assembly is strictly related to
the failure of the connection.

2.3 Second cyclic test

In this test the beam web was larger, increase from18 cm to 30 cm. This also reduces the
thickness of the fork sides in the column (Figure 20). In this situation the steel dowels and the
sleeves inside the beam are also shifted, outwards, to the sides of the elements, resulting in a
substantial reduction of concrete cover around these connection elements.

Due to inaccurate positioning of the sleeves in the beam during the manufacturing process it
was not possible to properly assemble the two concrete elements together. The bending length
of the steel dowel between the two elements was 3cm longer than planned (Figure 41).

22
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 41 Dowel bending length

Failure was reached in tension at 228kN with a maximum displacement of 200kN. The failure
mechanism was again a combination of the steel dowel yielding and concrete spalling. It is
possible that the failure load is smaller than for the previous test because of the larger bending
length of the dowel, this directly influences the capacity of the dowel. But also the smaller
concrete cover for the dowel should be considered. The extent to which each of these factors
affects the final value still remains to be investigated.

Figure 42 Force-Displacement - 2nd cyclic test

23
Beam to column connections – EXPERIMENT RESULTS v1.0

The following graphs show the maximum loads at each step of the cyclic test and the force
reduction for intermediate equal displacements taken from each cycle:

Figure 43 Max. load Force - 2nd cyclic test

Figure 44 Force variation for 6cm displacements - 2nd cyclic test

24
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 45 Force variation for 12cm displacements - 2nd cyclic test

At the final step there was a large relative displacement between the column and the beam. The
gap between them has increased from 3.5cm to 9.5cm (Figure 46).

Figure 46 Gap increase between elements

Maximum crack width at the base of the column in the area of the cross section change was
0.50mm.

25
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 47 Failure mechanism - 2nd cyclic test

The following graphs show a comparison of the force reduction for intermediate equal
displacements taken from each cycle:

Figure 48 Force variation for 6cm displacements - Comparison

26
Beam to column connections – EXPERIMENT RESULTS v1.0

Figure 49 Force variation for 12cm displacements – Comparison

Figure 50 shows the variation of the shear resistance with different edge-to-dowel distances. It can be
concluded that the edge to-dowel-distance in the direction of the forces governs the shear capacity of the overall
connections when the concrete cover of the dowel perpendicular to the applied force is equal or greater than4xD.
Thus for practical applications it is recommended that the concrete cover is taken as a minimum of 4xD in all
cases and a minimum of 6xD in the direction of the applied shear.

Figure 50 SAFECAST - Shear resistance of dowels as function of concrete cover

27
Beam to column connections – EXPERIMENT RESULTS v1.0

3 Initial conclusions

1. It is clear that the connection fails before the plastic hinge develops at the base of the
column. In this situation the hypothesis that the connection remains in the elastic range
during horizontal actions is not valid and the hypothesis of rigid diaphragm floors is also
disputable.

2. Behaviour factor q permits us to design structures for forces smaller than those
corresponding to a linear elastic response, but in this situation what behaviour factor
should we choose considering that there is no plastic hinge formation and no dissipation
mechanism?

3. Adequate formulas are needed for calculating these critical connections.

4. Using these types of structures with the described connection solution should be limited
to areas with a low to medium seismicity. The question that arises is what is the
maximum peak ground acceleration for which these structures can be designed?

28

You might also like