learning from failures in timber structures...learning from failures in timber structures dr. eva...

Learning from failures

in timber structures

Dr. Eva Frühwald Hansson, Div. of Structural Engineering, LTH

[email protected]

Timber Engineering PhD course 2017

Outline

• Failure – error ?

• Why learning from failures ?

• Research projects

– Nordic Study (until 2007)

– Swedish Collapses during winter 2009/2010

• Explicit cases

• What can we learn from failures?

Definition of failure (1)

• Failure is a deviation from the status quo.

• Failure is not meeting target expectations.

• Failure is any secondary defect.

Definition of failure (2)

Ultimate limit state

• Risk for human life, negative effect on safety and performance of structure

– Direct collapse of structures

– Local cracking

– Crushing

– degradation

Serviceability limit state

• No consequence for safety of structure

– Vibrating floors

– Excessive deformations

– Moisture movements

– Mould and fungi

Why should we learn from previous

failures / collapses ?

Hypothesis:

All failures are caused by human errors.

• Errors of knowledge (inadequate training in relation to tasks)

• Errors of performance (non-professional performance, carelessness)

• Errors of intent (consciously taking short-cuts and risks to save time/money)

Should be possible to prevent failures

[Kaminetzky]

Learning from failures

Education and training are the only effective

ways to minimize failures.

The training of an engineer, architect or

contractor should provide an understanding not

only of the best solutions that may be adopted,

but also of practices that should be avoided.

(Kaminetzky, 1991)

Previous studies: common failure causes• Concrete

– material quality

– work execution

– structural design and detailing (joints, openings, supports,…)

• Steel– insufficient temporary bracing during construction

– errors in design / construction mainly of connections and details

– deficient welding

– excessive flexibility and non redundant design

– Vibration induced failures

– stability type failures

– fatigue and brittle failure

– corrosion damage

• Timber– inadequate behavior of joints

– effects of moisture exposure (imposed strains, shrinkage)

– poor durability performance

– inadequate bracing of structural system

– inadequate performance of material and products

– inadequate appreciation of load

Level of safety in timber structures

• Is the level of safety adequate for timber structures

compared to other materials?

– Implementation of Eurocode 5 (national application rules)

– Spectacular collapses of timber structures in Europe

Swedish-Finnish project 2005-2007

• Research questions investigating collapsed buildings

– Reasons for failure?

– Which type of components are most prone to failure?

– Which failure modes are most frequent?

– What can be done to avoid or reduce failures?

Survey of failure cases(Swedish-Finnish survey)

• 4 partners

• 127 cases – literature

– own investigations

• Only cases implying risk for human lives are included (ULS)!

report1. Introduction

2. Experiences from previous

failure investigations

3. Present survey of failure

cases - methodology

4. Results and interpretation

of the information collected

5. How can we learn from

previous failures

6. Summary and conclusions

Annexe– Overview failure cases

and classification

– 127 cases, 1-2 pages / case

example

Classification of error types causing failure

1. Wood material performance

2. Manufacturing errors in factory

3. Poor manufacturing principles

4. On-site alterations

5. Poor principles during erection

6. Poor design / lack of design with respect to mechanical loading

7. Poor design / lack of design with respect to environmental actions

8. Overload in relation to building regulations

9. Other / unknown reasons

Materials & products

Construction work

Design/planning

Codes

Multiple failure types: estimate of the weight of each failure type

Failure cause (127 cases)Wood material performance 1%

Manufacturing errors in factory 5%

Manufacturing principles 4%

Design

(mechanical loading)

42%

Design (environmental

actions) 11%

On-site alterations

12%

Poor principles

during erection 16%

Overload / codes 4%

Other / unknown 5%

Failure cause (127 cases)

Design / planning

53%

Construction work

27%

Overload / codes 4%

Unknown / other 5%Materials and products

11%

Type of buildings

in percentage of cases

Public 51

Industrial 23

Agricultural 7

Apartment 8

Other / unknown 11

• NOTE: Failure surveys in general can not be seen as representative for the general population of structures (cover up of mistakes is common, random sampling is impossible)

0

10

20

30

40

50

60

70

80

90

100

sp

an

[m

]

Span

16% < 10 m

84% > 10 m

25 m

Age at failure

0

5

10

15

20

25

0 1 2 3 4 5 6-10 11-15 16-20 21-25 26-30 31-35 36-40

age /years

% o

f f

ailu

res

0 1 2 3 4 5 6-10 11-15 16-20 21-25 26-30 31-35 36-40

years

0

15

10

5

20

25

% o

f fa

ilure

s

Type of structural elements that failed


beam 47

truss 34

bracing 29

joint 23

arch 8

column 4

frame 2

dowel-type 57

punched metal plate 10

glued 7

other 27

Correlated with typical structural elements?!

Failure modes


• instability 30

• bending failure 15

• tension failure perp. to grain 11

• shear failure 9

• drying cracks 9

• excessive deflection 7

• tension failure 5

• corrosion of fasteners / decay 4

• withdrawal of fasteners 3

• compression (buckling) 2

• other / unknown 21

Timber, steel and concrete buildings:

failure causes

Failure cause[in % of cases]

Timber [own survey]

Steel [Oehme, Vogt]

Concrete [Brand, Glatz ]

Design / planning 53 35 40

Construction work 27 25 40

Maintenance / reuse 35

material 11

other 9 5 20

difficult to compare – definition of categories, number of cases etc.

Question: Are engineers better at designing

steel and concrete structures !?

How can we learn from previous failures?

human errors

Errors of intentErrors of knowledge Errors of performance

improved

training and

education

more efficient

Quality

Assurance (QA)

more efficient

Quality Assurance

(QA) ?

53 % design / planning

27 % construction work

Training & education

• Should focus on technical aspects which are typical causes for failure

• Training of engineers and control in the design / planning phase most important (most errors!)

• Training & education measurements– Lectures on good and bad examples for students / practicing

engineers

– Database on good / bad examples

– Certain technical aspects

– …

Learning from each others mistakes

Training & education: examples for issues to be emphasized

• Bracing to avoid instability both during construction and in the finished structure

• Situations with risk for perpendicular to grain tensile failure (joints, double-tapered beams, curved beams,…)

• Consideration of moisture effects

• Design of joints

• Estimation of loading conditions

• Estimation of real behavior of the structure

Training & education:

building site professionals

• Increasing the competence of building site professionals

– Professional training

– Assigned training / certified personnel to perform

certain tasks

– Continuous courses and seminars

• External quality control by impartial and certified

personnel

Collapses in Sweden

winter 2009/2010/2011

• Snowy winter

• Many collapses during

january/february 2010

Number of collapses

per municipality

Reported collapses

Winter 2009/2010

Nu

mb

er

of co

llap

ses

Collapses in Sweden during winter 2009/2010/2011

• Swedish government assigned Boverket (National Board of

Housing, Building and Planning) to investigate the reasons for

collapses

• Investigation was carried out by SP (Technical Research Institute of

Sweden)

• Parallel projects with common focus with other financers

– SLU (Swedish Agricultural University)

– Skanska Teknik

• The three projects were coordinated and resulted in a common

report, which was published in may 2011 (reported to the

government): Roof collapses winter 2009/2010 and 2010/2011 –

reasons and proposals for actions

Report: Roof collapses winter 2009/2010 and 2010/2011

– reasons and proposals for actions; Carl-Johan Johansson, Camilla Lidgren, Christer Nilsson, Roberto Crocetti

Contents

1. Introduction

2. Description of snow and weather conditions

3. Collapses in SP database (statistics, collapse reasons)

4. Experience from earlier snowy winters / other publications

5. Boverkets snowloads and form factors

6. Weaknesses and critical factors in different roof constructions (dependent on structural material)

7. Calculation errors and calculation programs

8. Building permit process

9. Construction and procurement

10. Special consideration of agricultural buildings

11. Interviews with suppliers

12. Conclusions and proposals for action

Collapses in Sweden

winter 2009/2010/2011

• About 3500 damages reported to insurance companies

• 180 collapses in SP database (167 + 13)

• 37 cases investigated thoroughly

Type of building and structural material

Structural material42 % steel

47 % timber

11 % glulam

Type of building number %

Sportinghalls, icerinks, eventhalls 14 8

Riding halls 6 4

Schools 2 1

Shops 5 3

Industrial buildings 4 2

Storage buildings 22 13

Agricultural buildings 65 38

Other 52 31

Sum 170 100

18%

Public buildings

SpanN

um

ber

of colla

pses

Span [m]

Roof inclination

arch

Num

ber

of colla

pses

Roof inclination [degrees]

Age of Structure (year of completion)

-1919 1930ies 1950ies 1970ies 1990ies 2010-

Num

ber

of colla

pses

1920ies 1940ies 1960ies 1980ies 2000-2009

Snowy winter 1976-1977

Observations regarding snow loads• Not more snow than in the code (1 exeption)

• No occasional thawing; wind from north/east during the wholesnowing period led to large snowdriftinghighly nonsymmetrical snowloads

• Nonsymmetrical snow loads even for roofs with low inclination(not at all in Swedish standard; non-symmetry larger than in Eurocode)

• Measured snow load on roof equals snow load on ground(in the standard, a form factor of 0.8 is used?!)

• Large differences in snow depth for large roofs

earlier

now

Reasons for collapse (37 cases)

Structural material Errors in

/ lack of

design

Errors in

material /

compo-

nent

Lack of

main-

tenance

Errors on

building

site

Other

(e.g.

snow

load)

Steel

(beam, frame, arch)

8 1 - 3 5

Glulam

(beam, frame)

3 2 - 4 3

Timber

(framework/lattice)

5 1 2 4 3

Sum 16 4 2 11 11

% 43 11 5 30 30

Nordic study (2007) % 53 11 * 27 9

No statistics for whole sample (170 cases) available

Reasons for failure in collapsed timber

and glulam structures

• Timber structures

– Missing stabilizing elements

– Rot (lack of maintenance)

– Purlin as gerber system – very sensitive to varying loading in different spans

• Glulam structures

– Risk for lateral buckling neglected

– Cracks in tension rod (steel quality)

– Missing anchor plate in tension rod connection

– Tension failure in tension rod

– Wrong detailing of tension rod connection

Critical aspects in glulam structures

Example 1: error on building site

Three-hinged frame, tension rod

Anchor plate dimensions

115x 55x 25 mm3

Missing anchor plate leading to

the nut being drawn through

the glulam beam

Collapse after 18 years

Critical aspects in glulam structures

Example 2: error of design

Eccentricity between tension rod

connection and steel column

high tension stresses

perpendicular to grain

High shear stresses

120 240 240

90

Dubbel dragstångAnkarbricka

Limträ balk

Upplagsplåt

(stålpelare)

15

6

Modelled by Henrik Danielsson,

Structural Mechanics, Lund University

Critical aspects in glulam structures (general)as presented in the report

GOOD / BETTER BAD / WORSE

Notched beams

Holes

Too short support length for

inclined beams

Luftspalt

(min 15mm)


Cut lamellae for tapered

beams

Tension perpendicular to

grain in special beams

Loose ridge

Screw

Reinforce-

ment

”HELA” LAMELLER SNEDSKURNA LAMELLERUrtag

Lös

nock

C

Critical aspects in glulam structures (general) as presented in the report


Design of joints

Primary / secondary

beam

Loading perpendicular to

the grain

Största delen av fästdon

ovanpå primärbalkens

neutrallager

Critical aspects in glulam structures (general) as presented in the report

Agricultural buildings

• Special rules for agricultural buildings

– No building permit required

– 15% higher capacity of nailed connections

• 63 collapsed agricultural buildings in SP database

• Building material

– 70% Timber framework

– 10% Timber traditional 2-storey building

– 5% Glulam

– 15% Steel

Agricultural buildings - Age (year of completion)

-19 30-39 50-59 70-79 90-99 10-

20-29 40-49 60-69 80-89 00-09

Num

ber

of colla

pses

Agricultural buildings – Design

• Difficult to say whether there was any design at all or who did the design

• ”in some cases the design was right which does not help much ifother parts are wrongly designed or are missing…”

• Timber framework: stabilizing elements are often missing (missingin design or error on building site)

• Purlins often designed as continuous beams, but built as singlespan beams – purlins have too low capacity and cannot reallystabilize the building against buckling and lateral torsional buckling

Agricultural buildings – loads

• Snowload often highly non-symmetric

• Non-symmetric snowload due to snow removal, leading to collapse

in one case

• Many buildings close to each other – snow from roofs with high

inclination slides down on roofs with lower inclination low

inclination roof should be designed to withstand 50% of the other

roofs snowload

Agricultural buildings – reasons for collapse (10 cases)

Structure

TF = timber

framework

T2 = timber

2-storey

Span

[m]

Inclination

[degr.]

Errors

in / lack

of

design

Errors in

material /

component

Lack of

main-

tenance

Errors

on

building

site

Snow

differently

distributed

on roof

compared

to code

Higher

snowload

on roof

compared

to code

when

completed

TF 34/2 10 X

TF 31/2 10 X X X

TF 20/2 14 X X

T2 14 45 X

TF 14 23 X

TF 32/2 15 X X X

Glulam 18 15 X X

TF 15 10 X

TF 16 16 X

Steel 14 14 X X

% 60 10 10 20 60 10

Errors in design / lack of design

Snow distribution on the roof

Danish experience of collapses of agricultural buildings 2010

• Ventilation hoods on the roof lead to snow accumulation on the leeward side

• Insurance company checked 60 buildings, only 9 were OK (freefrom mistakes)

• Typical errors

– Wrong bracing

– missing bracing (sometimes despite notation on the girder whereto place the bracing)

– bracing not tightened

– bracing cut off

– Wrong execution of nailed connections

Interviews with suppliers (steel, steel plates, glulam)

• What type is the collapsed structure – is that type still used or was the design changed?

• Control of design of collapsed structures with old and new codes –difference in some cases due to higher snow loads / non-symmetricsnow loads and use of Eurocode

• Typical errors according to suppliers

– removed tension rods

– corrosion due to lack of maintenance

– new building generating snow pockets

– Notches/holes in glulam beams made on building site (not in the design) information leaflet on holes and notches follows the delivery

– Collapse of steel sheathing above the primary beams due to highbending moment and high support reaction (now designed withlarger safety margin, instructions for snow removal on homepage)

Interviews with suppliers (steel, sheathing, glulam)

• Typical problems according to suppliers

– communication between the different designers (errors in design of stability and supports, information on snow pockets etc)

– Communication between designers and suppliers

– Lack of information from client to supplier difficult to make good design

• Sensitive structures/elements according to suppliers

– Arches

– Gerber systems

– Large deformations

– Fracture at support

Conclusions and proposals for action:

Collapses• Slender roof structures (steel, timber, glulam) have collapsed

• > 60% of collapsed buildings were built from 1980 and on

• Low inclination (in 50% of the cases < 15 degrees)

• Reasons for failure (from thorough study of 37 cases)

– No design /wrong design (including neglected snow pockets): 43%

– Errors on building site: 30%

– Material or component: 11%

– Lack of maintenance: 5%

– Other (including overload of snow): 30%


Agricultural buildings

• Mostly timber roofs

• Reasons for collapse mostly carelessness and

ignorance

• Often built without competence (do it yourself)

• Often no design at all

• Proposals for action:

– Inform farmers about the responsibility to be commissioner of

a building project by broschures and seminars

– Request from insurance companies that they require control

and supervision


Snow load and form factors• Value of snow load was increased for 2/3 of places in 2006

(decreased in only few cases)

• Snow load in current standard is plausible

• Form factors

– For roofs with low inclination (<15 degrees), the standard

prescribes symmetrical load, however, in reality the snowload

was highly nonsymmetrical introduction of Eurocode will

improve this, but is that enough?

– Snow load on flat roof = 80% of snow load on ground – WHY?

• Proposal for action: investigate form factors for snow load


Weaknesses of different constructions

• Steel: Slender structures need to be properly stabilized (needs gooddesign and execution)

• Glulam: local stresses (notch, slotted-in plate)

• Nailplate timber roofs: lack of bracing of compressed elements

• Steel arch covered with textile: lack of bracing against lateral buckling

• Roof sheathing: gerbersystems sensitive to nonuniform loading; shouldbe designed in highest safety class (usually in medium); sheathing is often too thin proposal: inform on correct design of sheathing

• Risk for progressive collapse is often neglected

• Proposal for action: formulate a broschure on weaknesses of different roof types (primary/secondary elements, detailing, risk for progressive collapse)


Design programs

• Many different programs

• Suppliers have own programs

• Programs are based on different models

• Many programs do not consider important load combinations


How many contractors?

• No difference whether one or several contractors

• Lacks in design and execution

• lack of compentence for some contractors

• Who has overall responsibility when many

contractors?


Building permit process

• Depending on when the building was erected, different meetings/control plans were needed

• Usually, the meetings were held, control plans were agreed upon – but not specified what is checked in the control plan, no documentation

• Need for a competent person linking all the different contractors, which only have responsibility for their part

• Different checks are missing, e.g. control of effect of snow pockets

• Conclusion: Collapses could not have prevented even if all rules had been followed

• Proposals for action

– Control plan should include more technical information

– One responsible designer / expert needed to put together all pieces

– Reconstruction/extension of buildings: design should include both old and new parts

– Different parts from different suppliers: make sure that the different parts interact correctly

Examples1 – Siemens Arena, Denmark

2 – Jyväskylä Exhibition Hall, Finland

3 – Buckling of trusses

Siemens Arena, Denmark

• Inaugurated February 2002

• Failed January 3, 2003

• 8000 seats

• Building costs: 6 millions Euros

• Calculated cost to rebuild structure:

25% of new price

Martin Hansson, 2004

Siemens Arena – the fish shape

72 m

(double: center to center 10 m)

5 m

about 6.5 m

No diagonals!

R=175 m

R=380 m

Failure

No load at failure

2 out of 12 trusses failed

Failure initiation

The failure joint

Slotted-in steel plates

Failure section

Reduced

section

- Tension failure

The failed cross section

glued-in spacer

block 45x160

160 160 160

533

430


4 slotted-in 8 mm

steel plates in 10

mm wide slots

160 160 160

533

430


Dowels 12 mm

160 160 160

533

430

Experts explanation

Early stage estimation not corrected during detailing!

• Area reduction not considered

- depth in bottom chord from 533 to 430 mm

- slots for steel plates

- holes for dowels

160 160 160

533

430

Experts explanation

Early stage estimation not corrected during detailing!

• Moment caused by very stiff connection –assumed as hinged

• Angle between stresses at surface and grain direction

• Used tensile strength about 50% higher than code value

Total effect:

The actual strength in joints was about 25% of the “designed strength”

The actual stresses were 40% higher than the characteristic 5% strength

All trusses will fail sooner or later…

Why those two? the trusses had large knots and low quality glueline glulam quality determined which of the trusses failed

Why just then? Duration of load

Experts explanation

Restauration- cables

Design related causes strength design

environmental actions

Construction related causes Poor principles during construction

Alterations on-site of intended design or products

Deficiency in design rules / codes prediction of capacity

Extreme loading

Causes related to wood material and wood products Inadequate quality of wood material

Poor manufacturing principles for wood products

Manufacturing errors in factory

Causes related to utilisation of the structureMisuse or lack of maintenance of the structure

Other causes

X

(X)

(X)

Jyväskylä Exhibition hall, Finland

• Glulam truss, about 55 meter span

• Inaugurated January 17, 2003

• Failed February 1, 2003 (the day after an

exhibition)

• Number of people in building: 10

• Nobody hurt

Martin Hansson, 2004

Structure

55 m

(center to center 9 m)

4.8 m

Double trusses

The failure

Failure load

• 52 kg/m2 snow load (25 % of design value)

• Failure load = 51% of expected

characteristic 5% capacity

Failure initiation

Failure investigation

Slotted-in steel plate

Dowels (hidden, no inspection!)

Extremely poor manufacturing

of dowel joints• steel plate was placed incorrectly

compared to center line of beams

• Positioning of dowels was poor:

Some not penetrating the steel plate

• Drilled holes varied in depth and

were oversized

• Partly too short embedding length

for dowels


Slotted-in steel plate

dowel

In failure joint only

7 of 33 dowels

were found !

Inadequate formula in old draft

of Eurocode 5 for block shear

failure. Corrected years before

the incident.


• Roof elements constructed as two-span, but support

reactions on trusses designed for single-span (+25%)

• Lateral stability of trusses not sufficient

progressive failure






Extreme loading





Other causes

X

(X)

Buckling of trusses






Extreme loading





Other causes

X

What can we learn from failures?

• All materials are different: steel, concrete and

timber have common but also their special

problems and failures types

An engineer good at designing steel and / or

concrete structures is not automatically excellent

at designing timber structures

What can we learn from failures?

Extra education / training should be provided to introduce engineers to the special problems in timber engineering

Implementationdatabase / books on failure cases – learn from each

others mistakes

Handbook/ material for design of timber structures including good and bad solutions, focus on typical problems

Courses in design of timber structures

external control – internal control

Literature

• Kaminetzky: Design and Construction failures - lessons from forensicinvestigations, McGraw-Hill, 1991

• Martin Hansson: presentations on failures in timber structures, 2004

• Accident Investigation Board Finland http://www.onnettomuustutkinta.fi/2601.htm

• http://www.maintenanceresources.com/ReferenceLibrary/FailureAnalysis/FailureModes.htm

• Dröge, Dröge: Schäden an Holztragwerken, Schadenfreies Bauen, Band 28, Fraunhofer IRB Verlag, 2003

• Brand, Glatz: Schäden an Tragwerken aus Stahlbeton, SchadenfreiesBauen, Band 14, Fraunhofer IRB Verlag, 2005

• Oehme, Vogt: Schäden an Tragwerken aus Stahl, Schadenfreies Bauen, Band 30, Fraunhofer IRB Verlag, 2003

• Frühwald et al, 2007: Design of safe timber structures – how can welearn from structural failures in concrete, steel and timber? Division ofStructural Engineering, Lund University, Report TVBK-3053

• Report: Roof collapses winter 2009/2010 and 2010/2011 – reasons and proposals for actions; Carl-Johan Johansson, Camilla Lidgren, Christer Nilsson, Roberto Crocetti; Report 2011:32

• Boverket: Erfarenheter från takras i Sverige vintern 2009/2010. En delredovisning av Boverkets regeringsuppdrag M2010/2276/H

More literature…

• Dietsch, Wolfrum, Winter: Evaluation of wide-span timber structures – results and

recommendations, WCTE 2008

• Winter, Kreuzinger: The Bad Reichenhall ice-arena collapse and the necessary

consequences for wide span timber structures, WCTE 2008

• Dietsch, Merk, Mestek, Winter: Structural safety and rehabilitation of connections in

wide-span timber structures – two exemplary truss systems, WCTE 2008

• Dietsch, Winter: Typische Tragwerksmängel im Ingenieurholzbau und Empfehlungen

für Planung, Ausführung und Instandhaltung, 8. Grazer Holzbau-Fachtagung, 2009

• Blass, Frese: Schadensanalyse von Hallentragwerken aus Holz, Karlsruher Berichte

zum Ingenieurholzbau 16

• Frühwald, Thelandersson, Fülöp, Toratti: Robustness Evaluation of Failed Timber

Structures, COST Action TU0601, Robustness of Structures, 1st Workshop, February

4-5, 2008, ETH Zurich, Zurich, Switzerland

• Frühwald Hansson: Analysis of structural failures in timber structures: Typical causes

for failure and failure modes, Engineering Structures 33 (2011) 2978–2982

• Studiengemeinschaft Holzleimbau e.V.: GUIDELINE FOR A FIRST EVALUATION OF

LARGE-SPAN TIMBER ROOF STRUCTURES

learning from failures in timber structures...learning from failures in timber structures dr. eva...

Documents