Lessons Learnt from Accidents in Urban Tunnels

Prof. Andr P. Assis, PhD

(UnB / ITA) EPFL Master on Advanced Tunnelling Lausanne, Switzerland May 2011

Introduction

General Trends in the Tunnelling Industry

High risk type construction methods Trend towards design + build contracts One-sided contract conditions Tight construction schedules Low financial budgets Fierce competition in construction industries

Decade 1990

Major Tunnel Losses since 1994

PROJECT 1994 Munich Metro, Germany 1994 Heathrow Express Link, GB 1994 Metro Taipei, Taiwan 1995 Metro Los Angeles, USA 1995 Metro Taipei, Taiwan 1999 Hull Yorkshire Tunnel, UK CAUSE Collapse Collapse Collapse Collapse Collapse Collapse US$ LOSS 4 mi

US$ 141 mio US$ 12 mio US$ 9 mio US$ 29 mio US$ 55 mio US$ 9 mio US$ 24 mio

1999 TAV Bologna - Florence, Italy Collapse 2000 Metro Taegu, Korea Collapse

Major Tunnel Losses since 1994

PROJECT CAUSE 2000 TAV Bologna - Florence, Italy Collapse 2002 Taiwan High Speed Railway 2003 Shanghai Metro, PRC 2004 Singapore Metro, Spore 2005 Kaohsiung Metro, Taiwan 2005 Barcelona Metro, Spain Collapse Collapse Collapse Collapse Collapse LOSS US$ 12 mio US$ 30 mio US$ 80 mio t.b.a. t.b.a. t.b.a. t.b.a. t.b.a. t.b.a.

2005 Lausanne Metro, Switzerland Collapse 2005 Lane Cove Tunnel, Sydney Collapse 2007 Sao Paulo Metro, Brazil and so on Collapse

Statistics on Causes of Accidents

Accidents During Construction: Last Decade Scenario

Significant increase in the number of claims Insurance income <<< Claims outcome Insurance paid >>> Initial cost of the work Difficulties to insure underground works

Options of the Insurance Market

Stop insure underground works Increase insurance prices and tight conditions Professional approach to the problem involving all related parts Focus on other markets Insurance may become not feasible

Proposal of a code of practice for risk management

Aims and Results of the Code of Practice for Risk Management

Establish minimum standards for evaluation of risks and procedures of risk management Clear definition of responsibilities of all involved parts Reduce probability of losses Reduce number and size of claims Re-establish the trust of insurance companies Transfer the concept of good practice to other market sectors

No construction project is risk free. Risk can be managed, minimised, shared, transferred or accepted. It cannot be ignored.
Sir Michael Latham, 1994

18.05.2011

Lessons Learnt from the Pinheiros Station Accident in Sao Paulo, Brazil
Introduction IPT Investigation Work and Report Main IPT Report Findings Conclusions and Recommendations

Existing Technical Reports on the Pinheiros Station Accident

Barton, N. (March, 2008) IPT (June, 2008) CVA (August, 2008)

Introduction: SP Metro Line 4

Introduction: SP Metro Line 4

Pinheiros Station

Introduction: Pinheiros Station

Pinheiros Station Design (primary support)

Pinheiros Station Design (final support)

Pinheiros Station Construction Scheme

Introduction: Pinheiros Station Accident

Occurred on 12/01/2007 During the bench excavation, very close of arriving to the shaft First failure signs ~14h30 Daylight collapse at 14h54 Enormous material damages and 7 fatalities IPT commissioned the technical investigation

IPT Investigation Work and Report

IPT Commission (team of in-house specialists) Board of Consultants (4 Brazilians and 2 foreigners) Independent Auditing Firm (Rina International) Desk Studies Following-up of the collapse debris excavation Interviews with involved staff from all parties

IPT Report (main report 384 p. + 46 appendices ~3000 p. + video)

Chapters 1-3: Introduction, objectives & scope Chapter 4: Urban tunnelling Chapter 5: Trends in contractual practices Chapter 6: Pre-bidding knowledge Chapter 7: Contractual aspects of Line 4 Chapter 8: Design and construction Chapter 9: Collapse Chapter 10: Mechanism and causes Chapter 11: Conclusions and Lessons

IPT Main Report: Pre-Bidding

10 years of studies till bidding Amount of geological and geotechnical investigation and level of engineering design had been continuously upgraded very reasonable and adequate Geological-geomechanical model Hasui (1993) IPT (1997) Figueiredo Ferraz (2001)

GeologicalGeomechanical Model
N

Pinheiros Station

Caucaia Shear Zone

Structural Geology 4 families

Pre-Bidding From impressiom packer in Pinheiros station area and reginal surveys (319 poles)

From scanlines on the final surface (26 and 522 poles)

Geological interpretation considering strutctural information

142610 142630 142650 142670 142690 142710 142730 142750 142770 142790

SM-8701
178200

SM-8706 SM-8707 SM-8702

Filonite (SR-07)
SP-8709
178180

SM-8700 SM-8720 SM-8703

SM-8704 SM-6530 SM-8705

SR-06 SR-07 SR-08

178160

SM-6532 SM-8708

SR-04 SR-8584 SR-03

178140

SP-8711

SM-8719

SM-6803

SM-6802

SR-02 SR-05 DE-SM-ML4-23 SR-01 SP-8710

AS UNID

SM-8714

S NA AS

Pinheiros Station: Geomechanical sections obtained from 3D interpolation using structural geology information
142645 142665 142685 142705 142725 142745 142765 142785 178245 178245

178225

178225

178205

178205

178185

178185

178165

178165

178145

178145

178125 142645

178125 142665 142685 142705 142725 142745 142765 142785

Longitudinal sections (central, north side and south side)

WSW

ENE

720

720

710

710

700

700

690

690

680
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150

680

WSW

ENE

720

720

710

710

700

700

Blue - fill Yellow alluvium Green tertiary sediments

690

690

680 0 10 20 30 40 50 60 70 80 90 100 110 120 130

680 140

WSW

ENE

Brown weathered soil Dark gray RMR IV Medium gray RMR III Light gray RMR II (rock mass classification from designer)

720

720

710

710

700

700

690

690

680 0 10 20 30 40 50 60 70 80 90 100 110 120 130

680 140

Cross-sections obtained from 3D interpolation

NNW

eixo entrevias

SSE

NNW

eixo entrevias

SSE

720

720

720

720

710

710

710

710

700

700 700

700

690

690 690

690

680

10

20

30

40

50

60

70

80

90

100

110

680 680 120 0

10

20

30

40

50

60

70

80

90

100

110

680 120

NNW

eixo entrevias

SSE

NNW

eixo entrevias

SSE

720

720

720

720

710

710

710

710

700

700 700

700

690

690 690

690

680

10

20

30

40

50

60

70

80

90

100

110

680 680 120 0

10

20

30

40

50

60

70

80

90

100

110

680 120

GeologicalGeomechanical Model
Sub-vertical alternation of granitic and biotitic gneiss, with variable thickness (sub-parallel to the tunnel longitudinal section) Four families of discontinuities Rockmass is heterogeneous and anisotropic due to discontinuities and uneven weathering bedrock surface as egg box type Post-bidding investigation confirmed the GG model developed during the pre-bidding design

IPT Main Report: Design Shortcomings

Geomechanical model adopted did not consider the anisotropy due to discontinuities Assumptions and calculations 2D analysis shaft effect neglected Oversimplified constitutive law for the soils above tunnel Ground assumed fully drained Design analyses indicated critical stability conditions during the bench excavation phase

IPT Main Report: Design Shortcomings

Monitoring and Instrumentation Instruments 4 sections (5 convergence pins) 4 sections (3 extensometers) Some open-pipe piezometers Threshold values for the instrumentation Shaft: all threshold values defined Tunnel: only the expected value calculated, but no definition of the warning and emergency values (qualitative criteria) No evidence of backanalyses

IPT Main Report: Construction Aspects

Quality control based on self-certification Poor control of methods and materials Forepoling filling Quantity of sprayed concrete fibres Early-age strength of sprayed concrete Deficient quality management Internal auditing system Geomechanical mapping Instrumentation data interpretation

IPT Main Report: Construction Aspects

Risk Management (contingency and emergency actions) Three main design violations during construction Inversion of the excavation direction of the bench towards the shaft Increase of the bench height (4 to ~5 m) Change of the bench excavation sequence (also the rate: 1.8 m/d in January 2007 and 0.9 m/day in December 2006)

-5 Recalque (mm)

7.0+86 - P1 7.0+86 - P2
-10

7.0+86 - P3 7.0+96 - P1 7.0+96 - P2 7.0+96 - P3 7.1+06 - P1 7.1+06 - P2 7.1+06 - P3

-15

Few Days before Collapse

13/12/06 23/12/06 02/01/07 12/01/07
23/12/06 02/01/07 12/01/07

-20

7.1+15 - P1 7.1+15 - P2 7.1+15 - P3

-25 23/11/06 03/12/06

Data

5,0

0,0

-5,0

-10,0 Convergncia (mm)

-15,0

-20,0

-25,0

-30,0 7.0+86 7.0+96 -35,0 7.1+6 7.1+15 -40,0 15/08/06 25/08/06 04/09/06 14/09/06 24/09/06 04/10/06 14/10/06 24/10/06 03/11/06 13/11/06 23/11/06 03/12/06 13/12/06 22/01/07

Data

Meeting on 11/01/2007
Installation of bolts in the tunnel bench walls decided No enough bolts in stock, despite it was forecasted in the design as contingency action (15% installed but all borehole drilled) No clear definition on the need to stop the works (contradictory version among participants) Three blasting on the 12/01/2007 (two around 8 h, one in each platform tunnel, and a third one around 12 h)

The Colapse
Fall of small concrete blocks Fracture propagation from the shaft till 1/3 of the tunnel length, position 11 h Fall of 6 to 8 lattice girders in the left-hand side wall Colapse daylight on surface at 14h54 Colapse of the north wall of the shaft at 15h30 (last event)

Main Report Findings: Collapse Evidences

eixo

1 0 -1 -2 -3 -4 -5 R e c alq u e (m m ) -6 -7 -8 -9 -10 -11 -12 -13 -14 23 /1 1/0 6 03 /1 2/0 6 13 /1 2/0 6 23 /1 2/0 6 02 /0 1/0 7 12 /0 1/0 7 22 /0 1/0 7

02/01/2007

13/12

23/12

TN.E-1 TN.E-2 TN.E-3 TN.F-1 TN.F-2 TN.F-3 TN.G-1 TN.G-2 TN.G-3 MS1 TN.H1 TN.H-3

Collapse: Instrumentation Data

Instrument Extensometer Convergence Pins (Settlement) Convergence

Position Axis

Data

Calculated (mm) -0,7 -0,7 -0,7 -0,9 -0,5 -0,2

Observed on 11/01/07 (mm) -11 -12 -7 -20 -7 -21

Observed / Calculated 17 19 10 22 13 95

Lateral Wall Axis Upper Lower P2-P3

Main Report Findings: Collapse Evidences

Instrumentation evidences
30 PINO 2 25 20 15 10 5 0 ESTACAS 7,0+86 / 7,0+97 / 7,1+06 / 7,1+15 25 20 15 10 5 0 ESTACAS 7,0+86 / 7,0+97 / 7,1+06 / 7,1+15 30 PINO 3 NOV29 DEZ15 DEZ27 JAN02 JAN08 JAN09 JAN10 JAN11 JAN12 JAN12T

VISTA GERAL DOS ESCOMBROS

Main Report Findings: Collapse Evidences

Debris position evidences

Section 7,0+87

Section 7,1+04

Section 7,1+13

Main Findings: Collapse Mechanism

Medidas de recalque nos tassmetros referenciadas a 11/01/07
142735 142745 142755 142765 142775 142785

TN.H1
178205 178205

TN.G1

TN.F1 TN.E1 TN.G2

178195 178195

TN.H3 TN.F2 TN.G3 TN.E2 TN.F3

178185 178185

TN.E3

178175

178175

?
TN.H1

P-2 P-2

TN.G1

178165 142735 142745 142755

P-2
142765 142775 142785

178165

TN.F1 TN.E1 TN.G2 TN.H3 TN.F2 TN.G3 TN.E2

P-2

P-3
TN.F3

P-3
TN.E3

P-3 P-3

Main Report Findings: Risk Factors and Causes

Non-Validated Design
Oversimplified geomechanical model Structural tunnel model Assumptions and completeness of calculations and simulations No definition of threshold values for monitoring Deficient GG mapping Deficient analysis and interpretation of monitoring data No evidence of back-analyses and design validation

Main Report Findings: Risk Factors and Causes

Non-Validated Construction Procedure
Change of excavation direction Increase of bench height Change of blasting scheme Deficient quality control Increase of excavation rate Deficient construction management (lack of bolts) No decision to stop works Deficient plans of contingency and emergency actions

Main Report Findings: Risk Factors and Causes

Accident Collapse and its Consequences Collapse of Pinheiros Station Presence of transit and pedestrians Fault of the emergency plan of actions

Non-validated design

Non-validated construction procedure and poor management

Risk Factors and Causes: Foreseeability and Other Aspects

Different ground conditions Excessive rain Seismic activity Pipe leakage
4,0 2,0 0,0 -2,0 Recalque (mm) -4,0 -6,0 -8,0 P1 -10,0 P2 P3 -12,0 P4 P5 -14,0 avano 7085 7090 7095

7130

7125

7120

7115

7110 Estaca (m)

7105

7100

Foreseeability

-16,0 29/11/06 04/12/06 09/12/06 14/12/06 19/12/06 24/12/06 29/12/06 03/01/07

7080

Data

Clear under good practice of engineering Misty by faults in several engineering processes

Conclusions

Geological model complex but data was fully disclosure no major changes by no means claim based on Different Ground Conditions Causes are related to shortcomings in engineering processes (design and construction) systemic fault process Lessons and recommendations to engineering and contractual arrangements

Recommendations for Future Contractual Arrangements

Keep fair balance among quality, schedule and costs Mix of technical and performance specifications quality control Independent auditing and full disclosure of control parameters owners must keep control Incorporate risk management and risk sharing

Lessons Learnt

Pre-Bidding Documents
Geological and geotechnical data as much as possible Full disclosure of all GG data
Geological model GG Data Report Geotechnical Base Report Different Ground Conditions Owner

Lessons Learnt

Design Documents
Geomechanical model Structural model of the tunnel Assumptions, completeness and type of calculations and simulations Continuum media? Type of model and parameters 2D or 3D analysis? Monitoring threshold values

Design Reviewer

Lessons Learnt

Design during Construction

Complementary investigation and mapping of all GG conditions Monitoring interpretation Design back-analysis Design Validation

Lessons Learnt

Construction
Faithful to the design agreement changes in

Quality control (materials and services) Integrated risk and construction management contingency and emergency actions

Lessons Learnt

Role of Contracts
Keep fair balance among quality, schedule and costs Mix of technical and performance specifications quality control Independent auditing and full disclosure of control parameters Incorporate risk management and risk sharing

Urban tunnelling is a great and increasing demand worldwide Urban tunnelling is challenging due to urban environment and constraints Urban tunnelling is likely dominated by limit admissible damage criteria Risk management has to be incorporated in all project phases

The worst happening is not to have an accident, it is to learn nothing from it.

Kovari, K. & Ramoni. M. (2004). Urban Tunnelling in Soft Ground Using TBMs. International Congress on Mechanised Tunnelling: Challenging Case Histories, Keynote Lecture, Politecnico di Torino, Turin, Italy (www.ita-aites.org). Munich Re (2006). Code of Practice for Risk Management of Tunnel Works: Future Tunnelling Insurance from the Insurers Point of View. ITA Open Session, ITA World Tunnel Congress, Seoul, South Korea. Munich Re (2007). Insurance Cover as Part of the General Risk Management Strategy. ITA Open Session on Public Private Partnership Projects, ITA World Tunnel Congress, Prague, Czech Republic. Seidenfuss, T. (2006). Collapses in Tunnelling. Master Thesis, Stuttgart University of Applied Sciences, Stuttgart, Germany, 179 p.