expert meeting on the “design of timber connections in fire

COST ACTION FP 1404 – WG 2

11-12 February, 2016Lisbon, Portugal

Chair of the Action: Joachim Schmid [email protected]

Vice Chair of the Action: Massimo Fragiacomo [email protected]

Grant Holder: Åsa Rössel as a. r [email protected]

Local organisers: Pedro Palma [email protected]

Helena Cruz [email protected]

Action website: http://www.costfp1404.com

Expert Meeting

Design of Timber Connections in Fire

Editors: Pedro Palma and Robert Jockwer

mailto:[email protected]

http://www.costfp1404.com/








Contents

Part I – Minutes

1 Participants.................................................................................................................................. 1

2 Opening the meeting............................................................................................................... 1

3 Presentations on 2016.02.11...................................................................................................2

4 Presentations on 2016.02.12................................................................................................... 7

5 Final discussion and closing................................................................................................9

Part II – Presentations

Roy Crielaard, Arup, UKDesigner's perspective on the fire design of timber connections – Example projects and common issues........................................................................................................................ 11

Dhionis Dhima, CSTB, FRFire behaviour of dowelled and bolted timber connections, and three-dimensional nailing plates.................................................................................................................................... 23

Jørgen Munch-Andersen, Træinformation, DKNext generation EC5 rules for connections and possible shortcomings for fire conditions.......................................................................................................................................... 39

Keerthi Ranasinghe, Exova BM TRADA, UKTimber connections in fire: standardisation considerations for 2020................................43

Pedro Palma, ETH Zurich, CHFire design of timber connections – assessment of current design rules and improvement proposals................................................................................................................. 53

Matteo Izzi, University of Trieste, CNR IVALSA Trees and Timber Institute, ITNumerical modelling of joints in timber structures – A state-of-the-art..........................67

Rubén Regueira, University of Santiago de Compostela, ESNumerical simulations of dovetail joints in fire......................................................................73

Daniel Brandon, SP Technical Research Institute of Sweden, SEDowel-type timber connections: improving their fire performance.................................109

Patrick Racher, CUST Univ. Blaise Pascal, FRFire safety of timber connections.............................................................................................. 127

Robert Jockwer, ETH Zurich, CHStudy on the reliability of connections at normal temperature and in fire.....................133

i

Part I – Minutes

Minutes of the Expert Meeting on the “Design of Timber Connections in Fire”

The Expert Meeting on the “Design of Timber Connections in Fire”, organisedby Working Group 2 of COST Action FP 1404, took place at the National Laboratoryfor Civil Engineering (LNEC), in Lisbon, Portugal, on 11-12 February 2016.

These minutes were prepared by Pedro Palma and Robert Jockwer.

1 ParticipantsThe meeting was attended by 11 participants from 10 countries, representing

research institutes, universities, and structural engineering companies (Table 1.1).

Table 1.1: List of participants

Name Institution/company Country

Daniel Brandon (DB) SP Technical Research Institute of Sweden SE

Dhionis Dhima (DD) Centre Scientifique et Technique du Bâtiment (CSTB) FR

Helena Cruz (HC) National Laboratory for Civil Engineering (LNEC) PT

Jørgen Munch-Andersen (JMA) Træinformation,Convener of CEN/TC250/SC5/WG5 Connections

DK

Keerthi Ranasinghe (KR) Exova BM TRADA UK

Matteo Izzi (MI) University of Trieste,CNR IVALSA Trees and Timber Institute

IT

Patrick Racher (PR) CUST Univ. Blaise Pascal FR

Pedro Palma (PP) ETH Zurich CH

Robert Jockwer (RJ) ETH Zurich CH

Roy Crielaard (RC) Arup UK

Rubén Regueira (RR) University of Santiago de Compostela ES

2 Opening the meetingThe organisers, PP and HC, open the meeting and welcome the participants. PP

gives general information on how to submit travel reimbursement requests and eligibleexpenses and proposes that the meeting runs as an open forum with presentations(participants can raise questions anytime), which is agreed by all participants. PP statesthe objectives of the meeting:

• assess the current state of knowledge on the structural fire design of timberconnections;

• identify research gaps and ways to address them;• propose improved design methods;• discuss the structure of the connections part in the next EN 1995-1-2.

1

3 Presentations on 2016.02.11

3.1 Roy Crielaard (RC), Arup, UK“Designer's perspective on the fire design of timber connections – Example projects and common issues”

Current trends in structural timber

RC presents a few example projects of current trends in structural timber. Open,exposed mass timber frame is particularly suited for office buildings, as it allows forlarge open spaces, and is usually combined with concrete or CLT cores and CLT slabs.Cladding can be used for the CLT but commonly not for the timber frames.Connections with slotted-in steel plates are preferred in practice, as well as bearingconnections (horizontal members supported in compression perpendicular to thegrain).

Issues and challenges faced by designers

RC presents issues/questions he collected from his colleagues:• Exposed timber structural elements will influence fire spread and

development. How much timber can be exposed?• The slip modulus of connections in the fire situation, according to section

4.3.4 of EN 1995-1-3 (Kfi = Ku ηf, with ηf = 0.67 for dowels and bolts andηf = 0.2 for nails and screws) does not work for tall buildings.

▪ PR agrees and mentions that, according to his observations during firetests, it should be Kfi = Kser ηf. Especially in large structures and buildings,where deformations during and/or after fire events may be an issue.

• Are connections, or should they be, expected to carry any load after therequired fire resistance period and how to assess the residual load-carryingcapacity of a connection after a given exposure.

• Fire resistance of connections between CLT floor and wall-elements (e.g.angle brackets).

• Fire behaviour of timber connections loaded in shear, bending, shear andtensions (as most fire resistance tests have been performed on connectionsloaded in tension).

▪ DD says that fire tests on steel-to-timber connections loaded perp. tograin and loaded in bending have been performed in France. PP says thattests on shear beam-to-column connections have been performed inSwitzerland.

• Impact of constructions tolerances, such as gaps between members and slotwidths, in the fire behaviour of connections.

▪ PP says that in the tests he mentioned before, the influence of the gapbetween the beam and the column was assessed and it seems that 10 mmmight be an upper limit, above which the fire resistance is severelyreduced.

2

• Performance of various protection materials (timber, gypsum, intumescentpaint, etc.). The need and effectiveness of intumescent paints applied onexposed bolts (nuts and washers) and dowels is questioned. An alternativecommonly used in practice is to protect the fasteners with wooden plugs.

• Impact of elevated temperatures (θ ≈ 50°C) in connections (e.g. how isembedment strength affected?).

▪ JMA states that this temperature might already reduce the strengthconsiderably. PP says that it also depends on the heating rate: in fire theheating rate is very high and, given the low thermal conductivity of wood,only a limited outer layer is influenced by temperature.

• What are the key parameters influencing fire resistance? Fastener type,fastener diameter, side member thickness, applied load?

▪ PP says that member thickness is more important than fastener diameter.PR states that the dowel diameter has only a small influence in thetemperature profiles. Also, failure modes might change (e.g. from blockshear at normal temperature to embedment in fire). KR mentions thatrope effect has to be disregarded in fire.

3.2 Dhionis Dhima (DD), CSTB, FR“Fire behaviour of dowelled and bolted timber connections, and three-dimensional nailing plates”

Experimental results

DD presents the fire resistance tests performed in France since 1999, on timber-to-timber and steel-to-timber dowelled and bolted connections (and the correspondingtests at normal temperature). DD states that the failure mode is longitudinal splittingalong the rows of fasteners and PR agrees. PP mentions that the large deformations aremost likely due embedment failures and splitting might occur only at the very end ofthe tests, as shown by the photos.

Proposed design method

DD presents simulation results that show the load distribution between fasteners.The “1 bolt for every 4 dowels” rule was based on the tests, but the simulations showthat the load carried by the bolt decreases significantly with time, whereas if onlydowels are used the load distribution between fasteners remains mostly constant. DDshows that simulations can follow the time-displacement curves measured during thefire tests, but only up to the steep part of the curve, right before failure.

Based on numerical simulations and experiments on steel-to-timber dowelledconnections, DD presents a design method based in a single empirical formula for allconnections and loading directions:

t fi=c 1+ c 2⋅d + c 3⋅t 1+c 4⋅ln (η)+c 5⋅d⋅ln(η )+ c 6⋅t 1⋅ln(η ) (3.1)

3

The results are then adjusted using a coefficient k, which is different for differentconnection typologies and loading directions (1.2 < k < 2.1):

t fi,d=k⋅t fi (3.2)

The method gives safe estimations of the fire resistance, even in situations inwhich EC5 is known to give unsafe results.

Beam-to-beam shear connections with three-dimensional nailing plates

DD shows results of fire resistance tests on beam-to-beam shear connectionswith three-dimensional nailing plates and slotted-in steel plates. Based on the testresults, DD states that the nailing plates reach 15 min of fire resistance for a load ratioof 30% and up to 30 min for a load ratio of 10%. KR mentions that the declared load-carrying capacity of joist hangers is not reliable (issue with EOTA Technical ReportR16), so the load applied during the fire tests should be based on tests performed atnormal temperature and not in the manufacturer's declared value. The exact failuremechanism of three-dimensional plates strongly depends on the test configuration. Inreality the load ratio might be higher than 10-30%. DD states that the slotted-in plateswith a bottom flange exhibit a better fire performance, because after the fasteners failthe member will rest on it. PP mentions that for longer exposures it might have theopposite effect, as it conducts more heat into the connection area.

RC asks about the observed failure modes and DD answers that in fire no blockshear failure has been observed. JMA states that plasticity at elevated temperature mightbe the reason.

RC asks about the differences between the connections reaching fire resistanceslower than 60 min and higher than 70 min. DD answers that the side member thicknessis the reason.

3.3 Jørgen Munch-Andersen (JMA), Træinformation, DK“Next generation EC5 rules for connections and possible shortcomings forfire conditions”

JMA presents the new proposed structure for the section on timber connectionsof EN 1995-1-1.

• 8.1 Introduction: principles, definitions, references (e.g.. EN 14592, P-clauses);• 8.2 Axial capacity of a single fastener;• 8.3 Lateral capacity of a single fastener;• 8.4 Combined axial and lateral capacity of a single fastener;• 8.5 Load distribution in a group of fasteners;• 8.6 Timber strength• 8.7-8.9: Shear connectors, glued-in Rods, punched metal plate fasteners…

JMA says that spacings and edge/end distances to avoid brittle failures arecurrently only diameter-dependent, but they could be also dependent on the thicknessof the timber members (required area or volume of timber per fasteners) or on the

4

load-carrying capacity of the fastener (modern fasteners). The participants agree thatchanges in minimum spacings, distances and thicknesses for the design at normaltemperature will require readjusting some design rules for fire, namely the simplifiedrules. PP states that fastener spacing perpendicular to the grain (currently 4·d for boltsand 3·d for dowels) might have to be increased for longer exposures (above 30 min), assimulations and fire tests on connections loaded perpendicularly to the grain show.

Summary of the challenges for the standardisation of connections in general:• strength parameters are not consistent, some are measured, some are

calculated; CE-marking is not consistent, reference testing should berequired;

• transformation of strength parameters to other materials (OSB, LVL, etc.)should be possible;

• spacing requirements have to take modern high capacity fasteners intoaccount.

Some challenges for the fire situation:• definition of penetration depth and zero strength layer for the fastening of

protective panels during fire.

3.4 Keerthi Ranasinghe (KR), Exova BM TRADA, UK“Timber connections in fire: standardisation considerations for 2020”

KR presents a review of the chapters for connections in EN 1995-1-1 (chapter 8)and EN 1995-1-2 (chapter 6). from an end-user's perspective. Current chapters arecomplex to follow and hopefully it will be possible to make them clearer and simpler inthe future. The challenge is to present the current technical content in a betterstructure. RC suggests that, as in the concrete part, tabulated values could be given firstand calculations methods afterwards. DD supports the need for simple and logicalmethods for the engineer (advanced methods, such as finite element simulations, arestill not practical for the design of timber connections). PR agrees thatEN 1995-1-2:2004 might have inconsistencies and lack clarity, but when it waspublished there was no basis to build on, and an urgent need to publish it.

KR asks why, in the simplified rules for unprotected connections, are theincreased dimensions afi calculated with the notional charring rate βn and not the one-dimensional charring rate β0. PR answers that it could be to make the determination ofafi similar to dchar,n in the reduced cross-section method (Equation 4.1 of EN 1995-1-2),only with a kflux instead of a zero-strength layer.

KR mentions that glued-in timber plugs (used to protect fasteners) are alsosupposed to be calculated using afi, and therefore βn and kflux, which seems strange giventhat there are no exposed metal parts anymore. PR mentioned that it could be forsimplification.

KR suggests to keep only one design method in the standard and move additionalor alternative methods to an annex.

5

3.5 Pedro Palma (PP), ETH Zurich, CH“Fire design of timber connections – assessment of current design rules and improvement proposals”

PP presents and overview of published fire resistance tests on timberconnections, before and after the publication of EN 1995-1-2:2004.

Simplified rules

PP shows that for timber-to-timber connections the current simplified rules canlead to an unsafe design, namely for high degrees of utilisation in the fire situation (ratiobetween the actions in the fire situation and the load-carrying capacity at normaltemperature Efi/R20°C), and proposes limiting the degree of utilisation to 0.3 for thesimplified rules to be applicable. PP proposes changing the kflux for each connectiontypology, so that the simplified rules always give safe estimates of the fire resistance.

PP states that for steel-to-timber connections with steel plates with unprotectededges in general, the values in Table 6.2 of EN 1995-1-2:2004 appear to have beentaken from the 1994 edition of the Holz Brandschutz Handbuch, which has significantlyincreased these values in the most recent, 2009, edition. For connections withunprotected edges on one or two sides, PP says that the table should not be used forbolted connections, as it gives unsafe results, and that for dowelled connections there isno experimental data to support the values for R60.

Reduced load method

PP states that the thickness of the side member t1 is the most importantparameter regarding fire resistance and proposes changing the current design method,making the k parameters dependent not only on the type of fasteners but also on thethickness of the side member t1:

R fi=e−k⋅t fi⋅R 20°C , with k=k ( t 1)=c 1+ c 2⋅t 1 (3.3)

The proposed coefficients for the k parameters are based on fitting a one-parameter exponential model to test results grouped by thickness. PP states that thediameter only plays a minor role, compared to the thickness of the side member and itdoes not improve the estimates of the proposed model.

DD states that the proposed minimum side member thickness t1 = 28mm fornailed connections is too low. PP answers that it was the minimum thickness used inthe tests performed by J. Norén, in Sweden, and that the maximum period of validity tfi

for the parameters k could also be made dependent on the side member thickness.

PR clarifies a discussion about the taking into account the effective number offasteners to determine the load-carrying capacity in the fire situation: up to a fireresistance of 30 min, there is no need to reduce the number of fasteners, but above thatthere might be.

6

4 Presentations on 2016.02.12

4.1 Matteo Izzi (MI), University of Trieste, CNR IVALSA Trees and Timber Institute, IT“Numerical modelling of joints in timber structures – A state-of-the-art”

MI presents different modelling strategies for timber connections in fire.Orthotropic models with plasticity in compression and brittle failures in tension havebeen developed for thermal-structural analyses, but are computationally verydemanding. MI has successfully adapted Carmen Sandhaas' model (Abaqus' UMATsubroutine) to thermal-stress analysis, but the simulations are very unstable, especiallywhen reaching ultimate failure. MI states that there is no need for additional newmodels but the existing ones have to be extended and the complexity of the modelsstrongly depends on the objectives. MI and DB discuss about the numericalimplementation of failure in wood and will continue to work on this topic.

4.2 Rubén Regueira (RR), University of Santiago de Compostela, ES“Numerical simulations of dovetail joints in fire”

RR presents simulations of dovetail joints exposed to fire. Tests show hightemperatures (up to 100 °C) inside the notch but the simulations initially did not. Anadditional heat source has to be simulated in the notch side surfaces to reach thetemperatures observed in the tests. It appears that the gaps in the connections areimportant and have to be evaluated more in detail. PR and DD state they performedsimilar tests and did not observe any charring inside the notch. RR clarifies that at100 °C charring would not be visible, but strength would be reduced. RR says thatfailure criteria for normal temperature, namely Tsai-Wu, is not suited for elevatedtemperatures, as the temperature-reduced strengths are too low and failure is reachedprematurely.

PP says that two fire resistance tests were performed on dovetail connections inSwitzerland; the results showed that the stronger connection at normal temperature(smaller notch) exhibited a lower fire resistance, as the charring reached the notchedsurfaces at an earlier stage. DD added that the same was observed in tests performed inFrance. PR explains the progressive failure of dovetails connection observed in tests,which could differ in normal temperature and in fire.

4.3 Daniel Brandon (DB), SP Technical Research Institute of Sweden, SE“Dowel-type timber connections: improving their fire performance”

DB presents his research on dowelled connections with non-metallic parts – FRPdowels and DVW flitch plates. Test results show that FRP dowels conduct much lessheat than steel dowels and the charring depth along the dowels is similar to that of thesurrounding timber. Numerical simulations were based on finite differences method,using a model of a single fastener (elasto-plastic beam on supported by springs withnon-linear behaviour). Temperatures are obtained from heat transfer simulations andthe properties of the mechanical model are reduced accordingly. The model is able to

7

capture the entire experimental displacement-time curve, including the final steep partbefore failure. DB states that reducing the modulus of elasticity after failure causesnumerical problems in FE models, but in his model a secant modulus of elasticity wasused instead, which is numerical much more robust.

DB discusses further aspects of connections in fire, namely the encapsulation ofthe connection is often problematic, e.g. wood plugs to protect the fasteners cause areduction of effective embedment strength. An alternative are the expanded-tubeconnections studied by A. Leijten, in the Netherlands. PR states that widespreadadoption of these connections is limited because self-tapping screws are much easier toapply and also increase, strength and ductility. PP points out that reinforcementsconduct more heat into the connection area and have been shown to reduce the fireresistance. RC asks about the costs of expanded tube fasteners and mentions that FRPdowels have the advantage to use the same construction process as the conventionalmetallic dowels.

4.4 Patrick Racher (PR), CUST Univ. Blaise Pascal, FR“Fire safety of timber connections”

PR presents FE simulations performed in France, in which the non-linearbehaviour was limited to a restricted area around the fasteners, overcoming numericalinstabilities. PR sates that the heating in the connection area is not dependent on thefastener diameter and shows depth-temperature curves for connections with fastenerswith different diameters.

PR states that there are two key issues regarding timber connections in fire: stiffness(very important for tall buildings, as mentioned earlier by RC) and load-carrying capacity.

According to PR, the slip modulus for connections in the fire situation should beKfi = Kser ηf, instead of Kfi = Ku ηf proposed in section 4.3.4 of EN 1995-1-2.

PR discusses two approaches for the load-carrying capacity in fire: reducing theload-carrying capacity of the connection (as in the current reduced load method) orreducing the resistance of a single fastener and then calculating the load-carryingcapacity of the connections (as in EN 1993-1-2). The first approach is well establishedbut has the problem that, in the available experimental data, the load-carrying capacityat normal temperature is not always adequately determined (different parameters aretaken, or not, into account: friction, rope effect, effective number of fasteners) and theempirical formulas might not be accurate. The second approach is similar to what isused in steel structures and a reduced embedment strength (dependent on the requiredfire resistance) would be used in Johansen's failure modes. PR states that adaptedJohansen's model gives good results for fire resistances up to 30 min, but above thatthreshold the thermal behaviour of the steel plate has an increasing influence in the firebehaviour and the fire resistance is overestimated.

PR mentions that there are no fire tests on connections with screws, even thoughthere are becoming ever more popular. PR also mentions that there is no test standardfor fire resistance tests of timber connections.

8

4.5 Robert Jockwer (RJ), ETH Zurich, CH“Study on the reliability of connections at normal temperature and in fire”

RJ presents a study on the reliability of timber connections at normal temperatureusing the framework for uncertainty quantification UQLab. Results show that thereliability index reduces with the thickness of side members, as shear splitting andtension perpendicular to the grain failure modes become more relevant. Also,increasing the strength of the dowel can be problematic, as it can change the failuremode towards brittle failures, reducing the reliability index. In fire, as charring reducesthe thickness of the side members, brittle failures may become dominant, thereforereducing the reliability index of the connections. The model should serve as a tool toevaluate the failure behaviour of connections in cold and fire situation. Optimalconfigurations with a preferred failure behaviour can be identified for the cold situationand required member thickness and distances can be specified. If the expected fireresistance and the corresponding failure mode of these configurations in the firesituation can be estimated, safety factors can be adjusted accordingly. RJ asks for testdata to calibrate the model, e.g. the brittle failure modes discussed by PR and DD.

5 Final discussion and closing

5.1 Final discussion

It is mentioned that if fastener spacings and edge/end distances or minimummember thicknesses change in the design at normal temperature, the fire design has tobe adjusted accordingly. JMA says that no immediate changes planned. RC says thatwarning statements could be included in EN 1995-1-1: “(…) this minimumspacing/thickness might not be sufficient for the fire design (…)”.

5.2 Closing

PP thanks HC for hosting the meeting and all the participants for joining,presenting their work, and actively participating in the discussions, especially PR andDD.

9

Part II – Presentations

Numerical modelling of joints in timber structures: A state-of-the-art

COST Action FP1404 – Design of Timber Connections in Fire, LNEC, Lisbon, Portugal.

Matteo Izzi

Ph.D. CandidateUniversity of TriestePiazzale Europa 1, 34127 Trieste, Italy

Research AssistantCNR IVALSA Trees and Timber InstituteVia Biasi 75, 38010 San Michele all’Adige, Italy

February 12, 2016

Motivations


The technological advancements of finite elements (FE) software packages have lead toan increasing use of numerical modelling

Numerical simulations have become an important tool to:Verify and extend the experimental resultsLimit the experimental testing to a minimumProvide time-saving and cost-effective solutionsSimulate situations that have not been tested

FE modelling has been successfully applied is several fields of timber engineering:Seismic design (ambient conditions)Heat conduction of timber and bio-based building productsStructural fire engineeringFire safety engineeringCreep/long-term effects

67

Introduction


The creation of a numerical model requires the careful consideration of several aspects,i.e.:

GeometryBoundary conditions (mechanical supports, external loads)Interactions (fire, hard contact between fasteners and timber)

Little assumptions and simplification

Generalresults

Numericalmodel

Experimentaltest setup

Simple model

”Bermuda Triangle” for numerical modelling

State-of-the-Art: Geometry and boundary conditions


The model schematization may vary depending on the problem under analysis:

2D modelling space with shell elements, e.g. for:Heat transfer analyses within the cross sectionThermal-structural analyses of timber floors/beams

3D modelling space with shell elements, e.g. for:Structural/thermal-structural analyses of timber floors/beams

3D modelling space with solid elements, e.g. for:Thermal-structural analyses of timber studs/floors/beams/connections

The use of 3D solid modelling is encouraged, although the computational effort requiredto carry out the simulations is higher

68

State-of-the-Art: Material models for timber


Mechanical behaviour of real timber:Orthotropic materialDifferent failure mechanisms, depending on applied loads and internal stresses

Mechanical behaviour of simulated timber, several approaches:Elastic orthotropic material with isotropic plasticity (unique yield strength for bothtension and compression)“equivalent” elastic isotropic material with Continuum Damage Mechanics plasticity(different yield strength in tension and compression)Advanced numerical modelling using a UMAT (User MATerial) subroutines

Basic information to characterize the material models are:Density*, conductivity* and specific heat*

Mechanical behaviour*

Thermal expansion coefficient (* defined as temperature-dependent parameters)

The mechanical and physical properties are not always easy to obtain, and their valuemay depend on the approach used to measure them (stationary/transient heat flux)

State-of-the-Art: Other limitations


Aside to the above mentioned considerations, there are other limitations related to thetechnology and the software packages used to carry out the simulations:

The computational effort is very high, leading to very time-consuming analyses

The amount of data saved by each simulations is very big, and it’s interpretation isnot always “user friendly”

Not all the software packages allow to perform coupled thermal-structural analyses(simultaneous application of fire and external loads). Therefore the simulation iscarried out in two separate steps:

1st step: heat transfer analysis simulating the heat conduction process withinthe cross section

2nd step: thermal-structural analysis, in which the temperature distributionresulting from the 1st step is used as an input for the mechanical simulation

69

Advance numerical modelling: UMAT Sandhaas


Features:Elasto-plastic orthotropic materialDeveloped for 3D modelling using solid elementsEight different failure mechanismsElasto-plastic in compression, brittle in tension

Add-ons:Strength and stiffness reduction due to temperature (EN 1995-1-2) for coupledthermal-structural analyses

Limitations:Origin-oriented approachTime-consuming analysesConvergence issues

Sandhaas C (2012) Mechanical Behaviour Of Timber Joints With Slotted-In Steel Plates. Ph.D. Thesis. Delft Universityof Technology, Delft, The Netherlands.

Sandhaas C, Van de Kuilen J-WG, Blaß HJ (2012) Constitutive Model For Wood Based On Continuum DamageMechanic. World Conference on Timber Engineering (WCTE), Auckland, New Zealand.

Advance numerical modelling: Future perspectives (1 of 2)


Development of a new UMAT based on Tsai-Wu failure criterion

Features:Simple strength criterion for anisotropic materialsTakes into account the difference in strengths due to positive and negative stressesCan be specialized to account for:

Different material symmetriesMulti-dimensional spaceMulti-axial stresses

Possible use:More or less everywhere

Current challenges for implementation:Plastic fluxes

Tsai SW, Wu EM (1971) A General Theory of Strength for Anisotropic Materials. Journal of Composite Materials 5(1):58-80, doi: 10.1177/002199837100500106.

70

Advance numerical modelling: Future perspectives (2 of 2)


Numerical modelling of timber connections using UEL (User ELements)

Features:Elasto-plastic beam in a non-linear medium that acts only in compressionCalibrated using quantities with whom practicing engineers are familiarAdapts to any input history (either force or displacement)Develops pinching as gaps are formed

Possible use:Steel-to-timber jointsTimber-to-timber (2/3 members) jointsSlotted-in steel plates joints

Current challenges for implementation:Introducing the effect of fire

Rinaldin G, Amadio C, Fragiacomo M (2013) A component approach for the hysteretic behaviour of connections incross-laminated wooden structures. Earthquake Engineering & Structural Dynamics, 42(13): 2023-2042, doi:10.1002/eqe.2310.

Numerical modelling of joints in timber structures: A state-of-the-art


Matteo Izzi

Ph.D. CandidateUniversity of TriestePiazzale Europa 1, 34127 Trieste, Italy

Research AssistantCNR IVALSA Trees and Timber InstituteVia Biasi 75, 38010 San Michele all’Adige, Italy

E-mail: [email protected]

71

MMS Axis (Mechanics, Materials & Structures) - Patrick RACHER – COST FP1404, Lisboa 1

Patrick RacherDhionis Dhima

Fire safety of timberconnections


3D thermo-mechanical FEMJoint scale

t1(tfi)

F

θx

θyθz

F i, tfi

Local stiffness variation

Plasticity beneath fasteners

Cinematic and static criteria in fire depend on :

K0(θ)

K90(θ) Kw(θ)

K0(θ)

K90(θ) Kw(θ)Structural scale

General consideration

Rjoint,tfi

Kser,tfi

127


Slip modulus of mechanical Fasteners

Joist hangers

33,0KK serfi =

Timber to timber and steel to timber joints

67,0KK serfi =

25

20

15

10

5

0

0 10 20 30 40 50 Temps (mn)

Glissement(mm)

Test N° 7tfi=27mn

Test N° 8tfi=22mn

Test N° 4tfi=38mn

25

20

15

10

5

0

0 10 20 30 40 50 Temps (mn)

Glissement(mm)

Test N° 7tfi=27mn

Test N° 8tfi=22mn

Test N° 4tfi=38mn

Slip (mm)

Time (mn)


Tb3

Tb2Tb1

2x10T1 ,T2 ,T3

2x15

TintTb3 Tb2 Tb1

10 10T1, T2

20

50 100 150 200 250 300 350 θ (°C)

20

40

60

80

100

Profondeur(mm) Broches 20mm Boulons 20mm

Broches 12mm Boulons 12mmBrochesprotégées

50 100 150 200 250 300 350 θ (°C)

20

40

60

80

100

Profondeur(mm) Broches 20mm Boulons 20mm

Broches 12mm Boulons 12mmBrochesprotégées

FastenersDowels

DowelsProtectedDowels

Bolts

BoltsDepth

Fasteners heating

128


Fire resistance

Reduction factor (on material properties and fastener resistance)

Load ratio (capacity of the connection )

Which approach ?

0

0,2

0,4

0,6

0,8

1

0 100 200 300

k (%)fh

θ (°C)

EN 1995-1.2Model x

Embedding strength

ηd


Behaviour of steel plate

- tfi < 30 mn : Johansen ~ FEM

- tfi > 30 mn : Johansen > FEM

0

10

20

30

40

50 100 150 200 250t1 (mm)

Rj (kN)

Johansen

0

10

20

30

40

50

50 100 150 200 250t1 (mm)

Rj (kN)

Simulation

Cold30 mn60 mn90 mn

Reduction factor

129


Steel to timber fastener design

Reduction factor

Timber behaviour

- t1 = 50 mm

30 < tfi < 60 mn : Plate behaviour

- t1 = 125 or 200 mm

0,0

0,2

0,4

0,6

0,8

1,0

0 15 30 45 60 75 90

kRj

(%)

tfi (mn)

t1 = 50 mmt1 = 125 mmt1 = 200 mm

( ) 11Rjfi1, RtkR ⋅=


Timber to timberfastener design

( ) 11Rjfi1, RtkR ⋅=Mode 2

Mode 1

3

2

FEM

Reduction factor

130


Load ratio

0,0

0,1

0,2

0,3

0,4

0,5

0 15 30 45 60 75 90

ηexp

0,4

0,3

0,2

0,1

0tfi (mn)

Bois sur bois

Bois-Métalη = 0,58e-0,020tfi

η = 0,80e-0,027tfi

ηd


135mm

1

4

Fi/F (%)

t (mn)

0

2

4

6

8

10

0 10 20 30 40 50

g (mm)

t (mn)

t1=60mm t1=135mm

Bolt

ts

F

t1

Dowels

1 2 3 4

60mm 1

4

neff (tfi)

Extensions to connections

Failure criteria

131


Pagode de Fogong-Shanxi (1056)

TIMBER, The 3rd millenium material

Thank youfor your attention ...

132

��

��

��

��

��

��

� � ��

��

��

��

Table 1: Tentative target reliability indices �� (and associated target failure rates) related to one year reference period and ultimate limit states

1 2 3 4

Relative cost of safety

measure

Minor consequences

of failure

Moderate

consequences of

failure

Large

consequences of

failure

Large (A) �=3.1 (pF�10-3) �=3.3 (pF � 5 10-4) �=3.7 (pF � 10-4)

Normal (B) �=3.7 (pF�10-4) �=4.2 (pF � 10-5) �=4.4 (pF � 5 10-6)

Small (C) �=4.2 (pF�10-5) �=4.4 (pF � 5 10-6) �=4.7 (pF � 10-6)

�� β ��

��!�� !��"�� !�� #�$�� % &'(�')

�� !"�� #� � �� # ��$

�� $

%&�� &�

'(&� �� )

��"�� *�+� ��

�� $

%&�

��,�� γ

��,��

⇒ -� �� * �� .' � ��

133

��

��

�� − ��

zk�� ·R�

γ�− γ�G� − γ�Q� ≥ 0

��

z ��

��

Ak�� · f�,0,�

γ�− γ�G� − γ�Q� ≥ 0

��

A ��

� � ��

��

��

�� − ��

� � ��

134

−

g = z · − −

x2

x1

g(x1,x2)=0

gX1 X2

12 ��j

j��R

α1�R

α2�R

u1

u2

�R

g(u1,u2)=0

β

−

P (g ≤ 0) = P (z · − − ≤ 0)

β

(P ≈ 10−3

) (P ≈ 5 · 10−4

) (P ≈ 10−4

)(P ≈ 10−4

) (P ≈ 10−5

) (P ≈ 5 · 10−6

)(P ≈ 10−5

) (P ≈ 5 · 10−6

) (P ≈ 10−6

)

135

��

��

�� − ��

α =Q

G

Definition of markers: Span Weight of roof or ceiling

short heavy large intermediate short or large light

Snow load 1) Imposed load 2)

Altitude [kN/m2] Category [kN/m2] 0 m 0.76 A 1.5

100 m 0.91 B1/D1 2 500 m 2.53 B2/C1 3

1000 m 3.73 C2 4 2000 m 31.76 C3-C5/D2 5

1) according to DIN EN 1991-1-3/NA 2) according to DIN EN 1991-1-1/NA

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

A B1/D1 B2/C1 C2 C3-C5/D2

Load

ratio

α [-

]

Category of imposed load (DIN EN 1991-1-1/NA)

Ceiling beams

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 100 500 1000 2000

Load

ratio

α [-

]

Site altitude above sea level [m]

Rafters

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 100 500 1000 2000

Load

ratio

α [-

]

Site altitude above sea level [m]

Roof trusses of halls

∗

∗�� !"

��

#$

�� − ��

��

R1,1

2= fh,1t1d

R1,2 = fh,2t2d

R1,3

2= fh,3t3d

t1

t3

t2

d

% !"

136

��

��

�� − ��

��

R2,12

2= −d

2

t1fh,1fh,2fh,1 + fh,2

+

+

√(d

2

t1fh,1fh,2fh,1 + fh,2

)2

+

(t21dfh,1 + 4My

)dfh,1fh,2

2fh,1 + fh,2

� � ��

��

��

�� − ��

��

R3,12

2=

√4Mydfh,1fh,2fh,1 + fh,2

��

137

−

Rconnection = min {2R1,i, R2,i +R2,j , R3,i +R3,j , R2,i +R3,j}

Mode 1 Mode 2 Mode 3

0

500

1000

1500

2000

2500

3000

3500

4000

4500

0 20 40 60 80 100

F v,R

k [N

]

penetration length t2

with rope effectwithout rope effect

t

R

− fh

fh = AρBdCε

ABCε

15 20 25 30 35 40 45 50 55��

15

20

25

30

35

40

45

50

55

� ��

��

��

��

∗

138

��

��

�� − �� My

My = 0.3fud2.6

�� fu � ��

��

�� 2 � � ��2 �

�� 4%�� 4%�� 4%�� 4%�� 4%

∗�� !

"# $ #%

��

&��

�� − ��

��

Rsplit = ti

√Gc,iE0,id (2h− d)

h

��

Gc �� !"�� #$E0 �%& '��((�( ! #�� ∗

��

∗�� "''(!

"% $ #%

139

��

��

�� − ��

��

Rv,split = 2tilendfv

Rt,split =1

1− μtilendft,90

lend

lend

��

��

��

�� − ��

�� ∗

Gc =1

κ1

(1 +

κ22κ1

(1−

√1 +

4κ1κ2

))��

!��

κ1 =1− κ3GII,c

κ2 =κ3GI,c

κ3 =

(σt,90

σv

)2

(σt,90

σv

)2+√

E90E0

∗�� !!�"

�# � ��

140

−

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2��

0

200

400

600

800

1000

1200

��

��

��

�

�

−

300 350 400 450 500 550 6000

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0.55

Density ρ [ kg/m3 ]

Gf,I

[N/mm

]

LarsenRiberholtGustafssonJockwer

∗

CoV

G 0.3 20%

∗

141

��

��

�� − ��

0 0.5 1 1.5 2 2.5 30

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Gf,II [ N/mm ]

Cumulativeden

sity

[-]

Method 1Method 2

∗

�� CoV

G�� 1.14 �� 31%

∗��

�� !

��

��

�� − �� fv

0 1 2 3 4 5 6 7 8 9 100

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

fv [ N/mm2 ]

Cumulativeden

sity

[-]

I-Beam-SpecimenEN 408 SpecimenWeibulLognormal

∗

�� CoV

f��"#��$ �� 4.1 ��2 13%f��%!&' �� 5.5 ��2 19%

∗(��)�� *�� +&&,�

+& � !

142

��

��

�� − �� ft,90

0 0.5 1 1.5 2 2.5 3 3.5 40

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

ft,90 [ N/mm2 ]

Cumulativedistribution[-]

V =0.57 dm3

V =10 dm3

WeibulLognormal

∗

�� CoV

f��=0.57dm3 �� 1.9 ��2 30%f��=10dm3 �� 0.76 ��2 30%

∗�� !

� " #$

��

��

�� − ��

% ��&�� '�� (

�� CoV

f��ρ �� 420 10%A �� B �� C �� ! "�#�ε ��

f� �� 437 4%E� �� 11500 23%G� �� 0.3 20%G�� 1.05 30%f� �� 5 25%f�� $�� 2 30%

�� " #$

143

��

��

�� − ��

� ��

�� CoV

G �� 2 10%Q �� 8 53%

��

��

��

�� − ��

��

E0 f� f�!"#ρ ��

E0 � ��

f� � ��

��

B C ε

A ��

B � ��

C � �

��

144

−lend

h

t1

t3

t2

dlend

dt1,2t2lend 10dh 10d

−

0 10 20 30 40 50 60 70 80 90 100��

0

1

2

3

4

5

6

7

8

9

10

��

0 10 20 30 40 50 60 70 80 90 100��

0

1

2

3

4

5

6

7

8

9

10

��

� ��

145

−

0 20 40 60 80 100 120��

0

0.5

1

1.5

2

2.5

3

��

0 10 20 30 40 50 60 70 80 90 100��

0

1

2

3

4

5

6

7

8

9

10

��

� ��

t1,3

−

0 20 40 60 80 100 120��

0

0.5

1

1.5

2

2.5

3

��

��

��

��

0 10 20 30 40 50 60 70 80 90 100��

0

1

2

3

4

5

6

7

8

9

10

��

��

��

��

t1,3

146

−

lend

ht1

t3

t2

dlend

⇒ t1 t2 lend h

−

0 5 10 15 20 25 30 35 40 45 50��

0

0.5

1

1.5

2

2.5

3

3.5

��

��

��

��

t = 0

0 5 10 15 20 25 30 35 40 45 50 55 60� ��

0

1

2

3

4

5

6

7

8

9

10

��

� ��

fu = 400 2

147

��

�� − ��

��

��

��

�� β ��

��

��

��

→ ��

��

��

�� − ��

��

��

��

��

. . .

��

�� !�� "#$$%&�

��

148

��

�� !�� "�� #��#�� !��

�� $%�&�'($&)(&&�*��+�� ,� -� �� .�� /� ��0�� *�� !�� 1�� ! �� 23 (�%�

�� (��'04)%&�,�� ,�� 5� 61��"�� '

��#'77888�9� �"5!��:7�,��:8�� (��

�� /�* �� 2 � ;�� 2�!��!� ;��8�+��

,�� 6� ,� � ��44%�� !� �� /�* �� *��<�1�� 5� *�� 3��

,�� 6� ,� � �� =�9�� 6� ��$�� 5�� #��5 �� 8�� 5#� �� " �� #$ %& �� '&� -�� .��5� /�#�� 3�� >?�%7@%>0>$�

-�� ,� ��$�� 6 #��"�"� �� 8��: �� "��5 � � �� 8�� 8�� 5#�� " �� #$ %& �� '&� -�� .��5� /�#�� 3�� >?�%7@%>0>��

-�� ,�� !�� :� .�� ,��:8�� 6 � �� 2�� "� �� !��A�� !�� "��5� �� " �� #$ %& �� ()� #�!� ��)?�%7($)��)�� BCD9E� �8�� /�#�� 3�� >?�%7($>��>��

=�� .� �� /� ,� ��44�� !5 �� 8�� #��#�� !�� " �� #$ %& �� *'� = "�� /��!�� /�#�� 3�� >?�%7�@>�4>��

=�9�� 6� ,� �� -�� ,�� ,�� 6� ,� � ��(�� 1�8 �� #��"�"��5 �� "�� 8�� 8�� 5#� �� " �� #$ %& �� '+� #�!� ��)?�%7@0)0)�� 2��"��!�� /�#�� 3�� >?�%7@0>0>��

��=2 ��44$�� *�� !5 � ��8�� ! ��!�� 8�� %�%�'(%�)(%0�

B�� .� �� E�� =�� .� �� /� ,�� >�� 6�� .�8�� 44�� 6##�� "�� >��>��>�� #�� #�� B � ��

@@ 7 @(

��

��

��

��

��

��

@( 7 @(

149

expert meeting on the “design of timber connections in fire

Documents