You are on page 1of 12

NON-LINEAR NUMERICAL MODELLING OF A POST-TENSIONED

TIMBER FRAME BUILDING WITH DISSIPATIVE STEEL ANGLE


DEVICES
Michele SIMONETTI1, Tobias SMITH2, Felice C. PONZO3, Antonio DI CESARE4, Stefano
PAMPANIN5 and Domenico NIGRO6
ABSTRACT
This paper describes the numerical modelling of post-tensioned timber (Pres-Lam) frame systems
under non-linear dynamic and static loading. From the conception of the post-tensioned jointed ductile
concept it has been clear that the nature of its controlled rocking mechanism leant itself well to the use
of a lumped plasticity approach. This approach combines the use of elastic elements with springs
representing plastic rotations in the system. Two experimental testing campaigns have been used in
order to confirm the effectiveness of this modelling technique in predicting both the global (frame)
and local (beam-column) response of these systems.
The first of these tests was a full scale beam-column joint tested quasi-statically and the second
was a 3-dimensional, 3-storey, 2/3rd scale multi-storey post-tensioned timber frame building tested
dynamically. During the testing programmes the specimens were tested both with and without the
addition of dissipative steel angles which were designed to yield at a certain level of drift. These steel
angles release energy through hysteresis during lateral frame movement thus increasing damping.
Both structures were modelled using a lumped plasticity approach with springs that were calibrated
against the moment rotation design procedure used for post-tensioned timber connections.
This work has proved the adequacy of the use of two numerical modelling programs,
RUAUMOKO and SAP2000, in providing accurate representation of structural response when
calibrated against current design procedures. All testing was performed in the structural laboratory of
the University of Basilicata in Potenza, Italy. This experimental campaign is part of a series of
experimental tests in collaboration with the University of Canterbury, Christchurch, New Zealand.

INTRODUCTION
Increasingly in the design of structures, engineers rely on the use of Non-linear Time History Analysis
(NTHA) in order to verify building performance under design seismic loading. During the use of these
models several trade-offs are made between complexity, accuracy, programming time, and processing
time. In design it is crucial that simple models exist which provide sufficiently accurate building
response without requiring a large amount of time in processing or programming.

PhD Candidate, University of Basilicata, Potenza, Italy, michelesimonetti04@virgilio.it


PhD Graduand, University of Canterbury, Christchurch, New Zealand, tobias.smith@canterbury.ac.nz
3
Prof, University of Basilicata, Potenza, Italy, felice.ponzo@unibas.it
4
Dr Eng, University of Basilicata, Potenza, Italy, antodice@yahoo.it
5
Prof, University of Canterbury, Christchurch, New Zealand, stefano.pampanin@canterbury.ac.nz
6
Mr, University of Basilicata, Potenza, Italy, domenico.nigro@unibas.it
2

The post-tensioned timber concept (under the name Pres-Lam) was developed at the University
of Canterbury and has been extensively tested in the structural laboratories of the university
(Buchanan et al. 2008; Palermo et al. 2005a; Smith et al. 2007). The system, originally conceived for
use in concrete structures (Priestley et al. 1999), combines post-tensioning and dissipative reinforcing
in order to provide moment resistant connections. During the course of the post-tensioned timber
project several authors have used lumped plasticity modelling to replicate experimental testing
however the predictive dynamic abilities of this approach, with regards to post-tensioned timber
frames, have not been studied.
This paper presents the numerical modelling of two test specimens: a full-scale beam-column
joint, and a 2/3rd scale, three-storey post-tensioned timber frame building both tested in the structural
laboratory of the University of Basilicata (UNIBAS) in Potenza, Italy. Numerical models of both
structures were created prior to testing in order to predict lateral response. Models were calibrated
against current post-tensioned timber design procedures and adequately predicted non-linear static
(beam-column) and dynamic (frame) response.

DESIGN CONCEPT
Post-tensioned timber uses post-tensioning technology (frequently applied to concrete structures) in
order to connect structural timber elements. While the post-tensioning provides desirable recentering
properties, the dissipative reinforcing devices allow adequate energy dissipation from the system as
well as increased moment resistance. During lateral frame movement, controlled rocking occurs at the
beam-column and column-foundation interface which provides flag-shaped hysteretic behaviour. Posttensioned timber technology enables the design of buildings having large bay lengths (8-12m) and
reduced structural sections.
Key to the post-tensioned frame connection is the ratio between the moment resistance
provided by the post-tensioning Mpt and the total moment resistance provided by the frame Mt (Figure
1). Clearly, during design this choice affects both the damping and moment capacity of the system and
therefore changing this value will have a direct impact on both capacity and demand under seismic
loading. During both experimental campaigns the size of the structural members, building layout and
mass was not altered, however different values of post-tensioning and steel moment capacity
contributions (thus variations in the value ) have been investigated.

Figure 1. Moment response with varying levels of the parameter

The design procedure


The numerical models used were calibrated against the current post-tensioned timber design
procedure. Although a codified procedure exists for the calculation of the moment-rotation response of
post-tensioned ductile concrete structures (fib 2004), several additions and considerations must be
made in order to account for the increased flexibility of timber. The moment capacity of a posttensioned timber joint is a function of the rotation at that joint. This rotation must be calculated from
the total building drift (T) and is made up of five separate rotation contributions as shown in Figure 2.
The first three rotation contributions comprise of the elastic rotations of the beam (b), column (c) and
joint panel (j) (Buchanan and Fairweather 1993; Newcombe et al. 2010a). The beam and column
rotations can be simply calculated in relation to the moment at the connection using common member
deflection equations. Due to the low shear modulus of timber, the large axial forces in the beams
induced by the post-tensioning result in large elastic shear deformations in the joint panel which must
2

M. Simonetti T. Smith F. C. Ponzo A. Di Cesare S Pampanin D Nigro

also be accounted for. The final two rotation contributions make up the total rotation of the connection
(con). These two contributions are defined as the interface rotation (int) and the gap rotation (gap) and
are calculated and act separately. Before decompression occurs an initial stiffness relating to the
compression perpendicular to the grain on the column face is present. This initial stiffness is not
captured by the design procedure used to calculate the post decompression behaviour and therefore
must be evaluated separately. This interface rotation thus acts before the decompression point of the
beam (i.e. before the gap opens) and the gap rotation occurs after decompression.
Following decompression the Modified Monolithic Beam Analogy (MMBA) is used (Palermo
2004). This method draws an analogy between the deformations and stresses in a hybrid joint and
those occurring in a standard concrete connection. This procedure can be simply applied to the design
of a post-tensioned timber connection provided a few simple considerations are made. The procedure
involves the imposing of the gap rotation (gap) and the initial estimation of a neutral axis value (c).
Using the design procedure, the forces in the post-tensioning tendon, compression in the timber, and
the force in any dissipative element are calculated. Force equilibrium is then checked and if not
satisfied a new value of c is selected. Once force equilibrium is verified the moment contributions are
added in series about a common point.

Figure 2. Deformation in a post-tensioned timber frame

NUMERICAL MODELLING
Modelling of post-tensioned timber beam/column-foundation connections (local modelling)
From the conception of the post-tensioned jointed ductile concept it has been clear that the nature of
the controlled rocking mechanism lent itself well to the use of a lumped plasticity approach in
modelling (Palermo et al. 2005b). This method of modelling has been used in the design predictions
of the beam-column and column-foundation joint with and without the dissipating steel angles.
Recently two methods of modelling the member interface (i.e. the lumped plasticity) have been
studied and applied to post-tensioned timber; the first a series of parallel rotational springs, the second
a more complex multi-spring model (Figure 3). Although when not considering secondary effects such
as beam elongation or timber crushing the simpler rotational spring model has been shown to be
sufficiently accurate (Newcombe et al. 2010b), the rotational spring model does not predict directly
the value of post-tension force (Tpt) or the neutral axis depth (c) nor does it directly calculate the force
(Ts or Cs) or displacement (s) of dissipative elements.

Figure 3. The multi-spring (left) and rotational spring (right) interface model

As mentioned above the joint panel rotations of a post-tensioned timber frame are not negligible
and in order to model this a rotational spring is added in the joint panel region. Previously all
modelling of post-tensioned timber beam-column joints has been performed using the finite element
program RUAUMOKO (Carr 2006) using both multi-spring and rotational spring approaches
(Newcombe et al. 2010b). During this research, the more widely used SAP2000 structural calculation
program (Computers and Structures 2004) was used and compared with testing results.
Spring calibration
All analysis results presented have been calibrated against the moment rotation response provided by
the design procedure described above with the parameters of the rotational springs being set to match
the predicted rotational response. Post-tensioning was represented using tri-linear elastic elements for
both models with bounded Ramberg-Osgood (Kaldjian 1967) and Buoc-Wen (Wen 1980) rotational
spring models used to represent the steel elements in the RUAUMOKO and SAP2000 model,
respectively.
Calibration of the multi-spring element was performed by altering the stiffness of the spring
elements until the three parameters of moment (Mt), post-tension force (Tpt) and neutral axis depth (c)
accurately replicated analytical values for a set level of initial post-tensioning force. Following
calibration it was simply necessary to change the initial compression value across the gap (due to posttensioning elements) in order to predict the performance of the different initial testing states. As the
multi-spring model can accurately predict the gap opening, and consequent displacements, axial
springs were used to represent the performance of the yielding steel angle reinforcement. These were
calibrated against testing results obtained for a single angle element subjected to cyclic axial loading
and a Bounded Ramberg-Osgood hysteric rule was used. SAP2000 does not contain a multi-spring
model therefore only a rotational spring model was used at the beam-column joint interface. The input
of the interface elements into the two models is essentially identical however the length of the
rotational spring differed slightly. In the RUAUMOKO analysis the standard practice of setting the
spring length to 1 mm was employed however this proved unstable when applied to the SAP model
and a 10 mm spring length was used.
Modelling of a post-tensioned timber frame (global modelling)
Full post-tensioned timber beam-column joints and frames are modelled through the combination of
the rotational/multi-spring elements described above with elastic beam elements. Post-tensioned
timber frames are represented with a combination of rotational springs being assembled to represent
the post-tensioning, reinforcement and joint panel contributions (Figure 4).
Figure 4 shows a perfectly pinned based column-foundation connection for the frame. When
dissipative reinforcement is applied to the column base it provides moment capacity. This moment
capacity acts in addition to the moment capacity provided by gravity (through the column axial load).
Due to this moment capacity, a truly pinned assumption is likely to over-predict displacement and
under-predict base shears. Figure 5 shows possible options, of increasing complexity, which can be
used in order to better represent the response of the column-foundation connection. The use of a fixed
based solution is considered adequate in predicting response until column-base gap opening, however,
following gap opening it is likely to under-predict displacement and over-predict base shear. This is
b) in
due to the fixed based model being infinitely stiff and not being able to account for the sudden drop
stiffness when gapping and steel yielding occurs.
The third option shown in Figure 5 allows for the drop in stiffness due to gap opening through
the use of a rotational spring calibrated against the post-tensioned timber design procedure described
above. During this calibration the static value (i.e. without lateral earthquake loading) of column axial
load is used, however, during seismic loading this value will not remain constant, a factor which is not
considered in the rotational spring column-foundation model. The use of a calibrated multi-spring as
shown in the far right diagram in Figure 5 provides the most accurate representation of column
response. This option accounts for the sudden loss of stiffness due to gap opening but also allows for
the change in capacity due to changing axial load. As mentioned above SAP2000 does not have a
proprietary multi-spring model however two axial gapping springs (gap elements as defined in
SAP2000) can be used to create a similar effect. This two-spring approach is, however, considered to
crude for use at the beam-column joint interface.

M. Simonetti T. Smith F. C. Ponzo A. Di Cesare S Pampanin D Nigro

Figure 4. SAP2000 and RUAUMOKO frame numerical models

Figure 5. Base connection modelling options considered for frame numerical model

EXPERIMENTAL TEST SET UP


Beam-column joint quasi-static testing
The full-scale beam-column joint (Smith et al. 2014), shown in Figure 6, was made from glulam grade
GL32h (EN 1995-1-1 2004) and both column and beam were 483 x 240 mm. The high strength steel
tendon (Fy = 1530 N/mm2; ft = 1760 N/mm2) placed at the centre of the beam had a diameter of 15.2
mm and was replaced by a 26.5 mm high strength steel (fy = 1050 N/mm2 E = 170 kN/mm2) bar when
initial post-tensioning forces were above 150 kN. Due to the nature of timber as an orthotropic
material it is undesirable to have excessive loading in the perpendicular to grain direction. As
displayed in Figure 6b, the column section (where compression was perpendicular to the grain) was
reinforced using 8 mm, 120 mm length screws. A steel tube of 100 mm diameter was glued into the
beam with a protruding length of 40 mm. This transferred a vertical load to the column.
On completion of a series of post-tensioned only testing, dissipation was provided through the
use of steel angles which were reduced in a certain area to a design width in order to obtain desired
hysteretic performance. This method of dissipation is based on the DIS CAM concept developed by
Dolce et al. (2006). A section of mild steel equal angle 100x100x10 mm was milled down to provide a
region of concentrated yielding and thus hysteretic energy dissipation. The variables used to control
the characteristics of angle performance were the thickness of the worked area (tA), the height of the
worked area (LA) and the length of the angle (bA). The width of the angle used in testing was bA = 80
mm and the strength of steel was fyk = 355 MPa. To characterize the stand alone performance of the
steel angle testing was performed along with finite element analysis (Ponzo et al. 2011).
The test specimen was loaded quasi-statically with a load protocol which applied increasing
levels of inter-storey drift (as top of column displacement) with two cycles being applied at each drift
level. A maximum drift level of 2.5% was reached in each test.

Figure 6. a) Beam-column joint and b) details showing dissipating elements and shear transfer system

Three storey dynamic frame testing


The three storey test frame (Ponzo et al. 2012) shown in Figure 7 was made from glulam grade
GL32h. A scale factor of 2/3 was applied to a prototype structure resulting in an interstorey height of 2
m and a building footprint of 4 m x 3 m. The building was designed to represent an office structure
(live loading Q = 3 kPa) with the final floor being a rooftop garden. All design was performed in
accordance with the current version of the Italian design codes (NTC 2008).

Figure 7. Experimental model constructed in UNIBAS lab

Beam-column joints were detailed in the same manner as the beam-column testing with the use
of internal tube shear keys and horizontal screw reinforcing. The flooring of the building was made
from solid glulam panels. As with the beam-column, testing was performed with and without the
addition of dissipative steel reinforcing angles at the beam-column connections. Instrumentation of the
structure consisted of a combination of potentiometers, load cells and accelerometers.
Seismic inputs
The testing input was a set of 7 spectra compatible earthquakes selected from the European strongmotion database. The code spectrum used to select the set of seven earthquakes was defined in
accordance with the current Eurocode for seismic design (EN 1998-1:2003 2003) giving a PGA for the
design spectrum of ag = 0.44 g (Soil class B medium soil). A smaller set of three ground motions
was selected which provided the best representation of the design spectra as shown on the left of Table

M. Simonetti T. Smith F. C. Ponzo A. Di Cesare S Pampanin D Nigro

1. This smaller set of seismic intensities was progressively increased in acceleration until the design
performance criterion was achieved.
Table 1. Characteristics of selected earthquakes and comparison with the code spectrum
ID Code

Location

Mw

30

PGA (g)

196

Sa (m/sec )

535*1.5

001228x

Izmit, Turkey

7.6

0.357

000196x

Montenegro

6.9

0.454

000535y

Erzican,
Turkey

6.6

0.769

1228*1.5

20

Average
Code Spectrum
10

0
0

Tsc (sec)

EXPERIMENTAL TESTING RESULTS


Beam-column joint quasi-static testing
The moment/force inter-storey drift/displacement responses of a selection of beam-column specimen
testing are shown in Figure 8. The post-tensioned only testing plots, shown in Figure 8a, display the
significant increase in moment resistance that occurred with the increase in initial tendon load. It can
be seen that an increase in initial tendon load from 50 kN to 250 kN led to a four times increase in
connection capacity at a drift of 2.5%. This graph also displays that without the addition of special
dissipation devices the fact that all materials in the system remain elastic meant that nominal energy
release occurred.
Displacement (mm)

Displacement (mm)

80

-80

60

Moment (kNm)

30
25
20
15
10
5
0
-5
-10
-15
-20
-25
-30

Moment/Force to Column Top


v Drift/Displacement

80

40
20
0

50kN Initial
100kN Initial
150 kN Initial
200kN Initial
250kN Initial

-20
-40
-60
-80

-100
-3

-2

-1

-40

80
60
40
20
0
-20
-40
-60

PT ONLY 100kN Initial


PT 100kN Initial Double Angle

-80

-3

-2

-1

40

80

-80

Moment (kNm)

30
25
20
15
10
5
0
-5
-10
-15
-20
-25
-30

40
20
0
-20
-40
-60

PT ONLY 150 kN Initial


PT 150 kN Initial Double Angle

-80
-100
-3

-2

-1

Drift (%)

30
25
20
15
10
5
0
-5
-10
-15
-20
-25
-30

-40

40

80

100

Moment/Force to Column Top


v Drift/Displacement

80
60

Moment (kNm)

Moment/Force to Column Top


v Drift/Displacement

60

30
25
20
15
10
5
0
-5
-10
-15
-20
-25
-30

Drift (%)
Displacement (mm)

100
80

80

-100

Force to Column Top (kN)

-40

40

Moment/Force to Column Top


v Drift/Displacement

Drift (%)
Displacement (mm)
-80

100

Force to Column Top (kN)

40

Force to Column Top (kN)

Moment (kNm)

-40

Force to Column Top (kN)

-80
100

40
20
0
-20
-40
-60

PT ONLY 200 kN Initial


PT 200 kN Initial Double Angle

-80
-100
-3

-2

-1

Drift (%)

Figure 8. Post-tensioned (PT) beam-column connection testing results: a) PT Only Testing, b) 100 kN Initial
with and without Dissipation, c) 150 kN Initial with and without Dissipation and d) 250 kN Initial with and
without Dissipation

Three storey dynamic frame testing


A series of shaking table tests with increasing PGA levels were performed both with and without
dissipative devices. Figure 9 shows the total input force as measured by the dynamic actuator versus
first floor drift at 75% and 100% of PGA intensity for model without and with dissipative steel angles,
respectively, considering three seismic inputs. The figure clearly shows the development of the typical
flag-shaped behaviour when dissipative steel angles were added as was seen for the beam-column

quasi-static testing described above. Figure 10 shows photos of the maximum response of the structure
during testing with the dissipative reinforcing subjected to ground motion 000535y at a PGA intensity
level of 100%.
150

001228x

000196x

000535y

Ram force (kN)

100
50
0
-50
-100

Shaking foundation ram force versus first floor drift

-150

a)

-4

-2

-4
4

-2
0
2
First floor drift (%)

-4
4

-2

150

001228x

000196x

000535y

Ram force (kN)

100
50
0
-50
-100

Shaking foundation ram force versus first floor drift

-150

b)

-4

-2

-4
4

-2
0
2
First floor drift (%)

-4
4

-2

Figure 9. Shaking foundation ram force versus first floor drift for test model a) without dissipation (75%PGA)
and b) with dissipation (100%PGA)

Figure 10. Maximum positive and negative drift response of the structure with dissipation (000535y at 100%).

NUMERICAL MODELLING COMPARISONS


Non-linear static modelling of beam-column joint
Comparisons between the results of the numerical analysis and testing for the non-linear static beam
column are shown in Figure 11. As can be seen the performance comparison of the PT ONLY test
response is good for all cases and models studied. The initial stiffness of the beam-column joint, an
important parameter in time-history analysis, is accurately predicted for all cases. Comparisons of the
models with the inclusion of the steel dissipative angles are displayed in terms of the penultimate test
and numerical analysis cycle. This cycle has been chosen as the ultimate cycle of testing had reduced
capacity due to the onset of angle failure. As shown, all of the modelling methods provided adequate
prediction of the beam-column joint loading behaviour. It can be noted however that both rotational

M. Simonetti T. Smith F. C. Ponzo A. Di Cesare S Pampanin D Nigro

models do not adequately reproduce the unloading behaviour of the joint. This is due to the way in
which the angle elements are modelled. In the multi-spring model only positive displacement is
applied which creates a larger hysteretic area than the rotational spring model where loading is also
reversed (+ve and ve rotations).
The failure to replicate unloading impacts on the hysteretic damping prediction of the model,
and the value of equivalent viscous damping (related to the ability of joints to dissipate input energy
through yielding). For the test cycle shown in Figure 11 the values of equivalent viscous damping are
shown in Table 2. As shown, both rotational spring models tend to under estimate the value of
damping while the multi-spring model overestimates damping.
The models presented all provide reasonably accurate predictions of the non-linear static
response of the beam-column joint. The rotational spring models have been shown to be almost
identical between the two programmes used with the only difference being the selection of hysteric
rule used based on the available hysteric rules of each selected calculation program.
25

15
10
5

15
10
5

15
10
5

Force (kN)

PT 100 kN
2 Angles

30
20
10

0
20
40
60
80
Displacement (mm)
PT 150 kN
2 Angles

30
20
10

Force (kN)

20
40
60
80
Displacement (mm)

20
40
60
80
Displacement (mm)

Test Results

20
40
60
80
Displacement (mm)

30
20
10

0
0

250 kN Initial

20

0
0

Force (kN)

25

150 kN Initial

20

Force (kN)

50 kN Initial

20

Force (kN)

Force (kN)

25

PT 250 kN
2 Angles

0
0

20
40
60
80
Displacement (mm)

SAP2000 Rotational

Ruaumoko Rotational

20
40
60
80
Displacement (mm)
Ruaumoko Multi-Spring

Figure 11. Comparison between Testing Results and Modelling Options

Table 2. Values of equivalent viscous damping (in %) for testing and numerical predictions
Test
100 kN 2 Angle
150 kN 2 Angle
250 kN 2 Angle

14.7
12.5
7.8

SAP2000
Rot.
13.2
9.9
5.7

RUA.
Rot.
11.6
9.8
6.6

RUA.
Multi
17.2
14.4
10.5

Dynamic model identification of three-storey frame


Dynamic model identification testing was carried out in order to find the natural frequencies of
vibration considering two different excitation sources: hammer impact excitations and sine-sweep
ground motion. In Table 2 the first three periods of the frame, considering the hammer impact test
response on the third floor of the structure with dissipative devices, are compared with the 2-D
SAP2000 and RUAUMOKO numerical predictions. The numerical blind predictions (Ti,num) for both
models matched well with experimental results Ti,exp

Ti,num SAP (sec)

Ti,num RUA (sec)

0.53

0.53

0.50

0.13

0.12

0.12

0.06

0.05

0.05

T3 = 0.06s

Ti,exp (sec)

T2 = 0.13s

Mode

T1 = 0.53s

Table 3. Comparison between dynamic experimental behaviour and SAP2000 blind predictions

10

15

Frequency (Hz)

20

25

Dynamic comparisons
As mentioned above the fixity of the base of the column has a significant impact on the total response
of the non-linear dynamic numerical model. In this section the experimental outcomes are compared
with SAP2000 and RUAUMOKO numerical results considering the more accurate, but most complex,
multi-spring base model.
The experimental outcomes of the dynamic testing are compared in Figure 3 with the SAP2000
and RUAUMOKO NTHA results considering the 3rd floor displacements for three test cases
(000196xa, 000535ya and 001228ya) with an intensity level of 75% of the design PGA with and
without dissipative steel angles. As shown both models provided very similar and sufficiently accurate
results when compared to testing. It can be seen however that during certain sections of structural
response the models were out of phase with the test structure. In all cases however this was during the
tail of the response and did not impact on the prediction of maximum values.
0.15

0.15

SAP2000

196

EXPER.

0.05

0.05

-0.05

-0.05

-0.15

-0.15
0

10

15

20

0.15

RUAU.

10

15

20

10

15

20

10

15

20

0.15

535
0.05

0.05

-0.05

-0.05

-0.15

-0.15
0

10

15

0.15

20
0.15

1228

0.05

0.05

-0.05

-0.05

-0.15

-0.15
0

10

15

20

Figure 12. 3rd floor displacement comparisons between SAP2000, RUAUMOKO and testing results without (on
the left) and with (on the right) dissipative

Three key indicators have been selected for comparison between the testing results and the
numerical models and are displayed in Figure 13: the maximum 1st level drift, the maximum 3rd level
acceleration and the maximum base shear for the configuration with and without steel angles at
varying PGA intensity levels. As shown in Figure 13, the numerical predictions provided an accurate
representation of the experimental performance for both numerical models showing differences of less
than 20 25% when compared with experimental outcomes. For this test frame the SAP2000 model
provided a more accurate prediction of maximum test response.

10

M. Simonetti T. Smith F. C. Ponzo A. Di Cesare S Pampanin D Nigro

0.8
0.4
0

60
40
20
0

20

40

60

80 100

20

40

60

Without dissipation

0.8
0.4
0
20

40

60

80 100

1
Average Testing

0.5

Average SAP
Average RUA

20

40

60

80 100

2.5

80
60
40
20
0

1.5

80 100

100

1.2

0
0

Base shear (Vb, kN)

1.6
3rd Level acc. (g)

80

1st Level drift (rad)

3rd Level acc. (g)

With dissipation
1.2

2.5
1st Level drift (rad)

100
Base shear (Vb, kN)

1.6

11

2
1.5
1
Average Testing

0.5

Average SAP
Average RUA

0
0

20 40 60 80 100
Percentage of PGA

20

40

60

80 100

Figure 13. Comparison between SAP2000, RUAUMOKO and experimental results.

CONCLUSIONS
This paper has presented an overview of the non-linear static and dynamic modelling for the
prediction of post-tensioned timber frame response. Recent quasi-static and dynamic post-tensioned
timber testing performed at the University of Basilicata in Potenza Italy has then been used to assess
the adequacy of these models in predicting response.
The models consisted of a series of elastic elements and non-linear springs which have been
calibrated against the current post-tensioned timber design procedure that has also been briefly
described. Two analysis programmes, RUAUMOKO and the more widely spread SAP2000 were used.
The ability of the non-linear model to predict the static local (beam-column joint) behaviour has
been verified through comparison against quasi-static beam-column testing. All models considered
accurately represented initial and non-linear stiffness as well as providing sufficiently accurate
representation of hysteretic energy dissipation.
The ability of the non-linear model to predict dynamic global (frame) response has been verified
through comparison against 2/3rd scale dynamic frame testing. Firstly the dynamic characteristics (first
three natural periods) of the test frame were compared favourably against results provided by the
numerical models. Secondly the key parameters of third floor displacement, maximum base shear,
maximum drift and maximum third floor acceleration were compared. All values were accurately
(within 20%) predicted with the SAP2000 programme providing improved accuracy for this test
frame. During testing a complex multi-spring model was used at the base of the column which may
not be feasible for use outside of research applications.
In a world were practicing engineers are relying increasingly on the use of non-linear static
(push-overs) and dynamic (NLTH) analysis it is important that the models used are robust and able to
predict system response. Current methods for the non-linear analysis of post-tensioned timber
buildings have been compared against a selection of test results. This comparison has shown that the
modelling techniques were able to predict with sufficient accuracy both local and global response
providing further confidence in their use in design.

ACKNOWLEDGEMENT
Authors would like to acknowledge the financial support of the Structural Timber Innovation
Company (STIC, New Zealand) and FederLegnoArredo (FLA, Italy).

REFERENCES
Buchanan, A., Deam, B., Fragiacomo, M., Pampanin, S., and Palermo, A. (2008). "Multi-Storey Prestressed
Timber Buildings in New Zealand." Journal of the International Association for Bridge and Structural
Engineering, 18(2), 166-173.
Buchanan, A. H., and Fairweather, R. H. (1993). "Seismic Design of Glulam Structures." Bulletin of the New
Zealand Society for Eathquake Engineering, 26(4), 415-436.
Carr, A. (2006). "RUAUMOKO." A. Carr, Christchurch, New Zealand, Inelastic Dynamic Analysis Software.
fib Bulletin 27. (2004). "Seismic Design of Precast Concrete Building Structures." fib
CSI Computers and Structures Inc. (2004). "SAP2000: Static and Dynamic Finite Analysis of Structures."
Berkeley, CA, USA.
Dolce, M., Moroni, C., Nigro, D., Ponzo, F. C., Santarsiero, G., Croce, M. D., Canio, G. D., Ranieri, N.,
Caponero, M., Berardis, S., Goretti, A., Spina, D., Lamonaca, B., and Marnetto, R. (2006). "TREMA
Project Experimental Evaluation of the Seismic Performance of a R/C 1/4 Scaled Model Upgraded with
the DIS-CAM System " 2nd International fib Congress, Naples, Italy.
NTC. (2008). "Norme Tecniche per le Costruzioni." Il Ministro delle Infrastrutture.
Kaldjian, M. J. (1967). "Moment-curvature of Beams as Ramberg-Osgood functions." Journal of the Structural
Division, American Society of Civil Engineers, 93(ST5), 53-65.
Newcombe, M., Cusiel, M., Pampanin, S., Palermo, A., and Buchanan, A. H. (2010a). "Simplified Design of
Post-tensioned Timber Frames." CIB W18 Workshop on Timber Structures, Nelson, New Zealand.
Newcombe, M. P., Pampanin, S., and Buchanan, A. H. (2010b). "Numerical Modelling and Analysis of a TwoStorey Post-Tensioned Timber Frame with Floor Diaphragms." 14th European Conference on
Earthquake Engineering, Ohrid, Republic of Macedonia.
Palermo, A. (2004). "The Use of Controlled Rocking in the Seismic Design of Bridges," Politecnico di Milano,
Milan.Doctor of Engineering.
Palermo, A., Pampanin, S., Buchanan, A., and Newcombe, M. (2005a). "Seismic Design of Multi-Storey
Buildings using Laminated Veneer Lumber (LVL)." 2005 New Zealand Society for Earthquake
Engineering Conference, Wairakei Resort, Taupo, New Zealand.
Palermo, A., Pampanin, S., and Carr, A. (2005b). "Efficiency of Simplified Alternative Modelling Approaches to
Predict the Seismic Responce of Precast Concrete Hybrid Systems." fib symposium "Keep Concrete
Attractive, Budapest, Hungary.
Ponzo, F., Di Cesare, A., Nigro, D., and Smith, T. (2011). "Dissipative Steel Angles Devices for the Increased
Seismic Performance of Low Level Damping Systems." XXIII Congresso CTA, Lacco Ameno, Ischia,
Italy.
Ponzo, F. C., Smith, T., Di Cesare, A., Pampanin, S., Carradine, D., and Nigro, D. (2012). "Shaking Table Test
of a Multistorey Post-tensioned Glulam Building: Design and Construction." 12th World Conference on
Timber Engineering, Auckland, New Zealand.
Priestley, N., Sritharan, S., Conley, J., and Pampanin, S. (1999). "Preliminary Results and Conclusions From the
PRESSS Five-Story Precast Concrete Test Building." PCI Journal(November-December 1999), 42-67.
Smith, T., Ludwig, F., Pampanin, S., Fragiacomo, M., Buchanan, A., Deam, B., and Palermo, A. (2007).
"Seismic Response of Hybrid-LVL Coupled Walls Under Quasi-Static and Pseudo-Dynamic Testing."
2007 New Zealand Society for Earthquake Engineering Conference, Palmerston North, New Zealand,
8.
Smith, T., Ponzo, F. C., Di Cesare, A., Pampanin, S., Carradine, D., Buchanan, A. H., and Nigro, D. (2014).
"Post-tensioned Glulam Beam-Column Joints with Advanced Damping Systems: Testing and
Numerical Analysis." Journal of Earthquake Engineering, 18(1), 147 - 167.
EN 1998-1:2003. (2003). "Design of structures for earthquake resistance - Part 1: General rules, seismic actions
and rules for buildings." European Committee for Standardization.
EN 1995-1-1:2004. (2004). "Design of Timber Structures Part 1-1: General - Common Rules and Rules for
Buildings." European Committee for Standardization.
Wen, Y. (1980). "Equivalent Linearization for Hysteretic Systems under Random Excitation." ASME Journal for
Applied Mechanics, 47(1), 150-154.

12

You might also like