Prediction, Monitoring and Evaluation of Performance of Geotechnical Structures. Prévision, contrôle et l’évaluation de la performance des structures géotechniques Arsenio Negro Jr. Bureau de Projetos e Consultoria Ltd., Sao Paulo, Brazil Kjell Karlsrud Norwegian Geotechnical Institute, Oslo, Norway Sri Srithar & Max Ervin Golder Associates Pty Ltd, Melbourne, Australia Eduard Vorster Aurecon South Africa (Pty) Ltd, Pretoria, South Africa
Transcript
Prediction, Monitoring and Evaluation of Performance of Geotechnical Structures.
Prévision, contrôle et l’évaluation de la performance des structures géotechniques
Arsenio Negro Jr.
Bureau de Projetos e Consultoria Ltd., Sao Paulo, Brazil
Kjell Karlsrud
Norwegian Geotechnical Institute, Oslo, Norway
Sri Srithar & Max Ervin
Golder Associates Pty Ltd, Melbourne, Australia
Eduard Vorster
Aurecon South Africa (Pty) Ltd, Pretoria, South Africa
Contents
1. Introduction
2. Foundations
3. Earth Fills
4. Supported Excavation
5. Tunnels
6. Geotechnical Instrumentation
7. Final Remarks
1. Introduction
Design not for the most unfavourable conditions but(more economically) for the most probable conditions(Interactive Design).
Basic requirements for Interactive Design:
1º) need for contingency plans for any foreseengeotechnical condition;
2º) need to anticipate response of the geotechnicalstructure;
3º) keen ability to interpret the observed performanceor to detect subtle deviations.
The first two requirements are undisputed.
The last is normally taken for granted and overlooked(in many cases resulting in failures).
Mere comparisons between measurements andpredicted quantities may not always suffice toanticipate problems.
(These should be supplemented by full understandingof response of the geotechnical structure and byadequate assessment of the predictions made).
Response of Geotechnical Structures = f (stress path+ failure mode)
Four geotechnical structures are contemplated:
-Foundations (deep)
-Earth Fills
-Supported Excavations
-Tunnels
For each structure the report reviews:
-typical responses
-measuring particular performance
-evaluation of prediction
-performance evaluation
-specificities of the Interactive Design application oneach of structure
Final chapter on Field Instrumentation is included, for reviewing and discussing new trends and recent developments.
2. Foundations
State-of-practice in deep foundation design: stillmostly conservative and based on assessing anultimate bearing pressure and then applying a factorof safety.
Settlement of the foundation is assessed but asettlement based design is not routinely adopted as adesign method.
Evaluation of prediction of deep foundations:essentially model tests or full scale field testing.
Full scale field tests is important, if:
- there are uncertainties in the design
- the design is pushing the boundaries of acceptedpractice
A test bored pile with 2.5 m diameter founded at about86 m depth in dense clayey/silty sand was performedfor My Thuan Bridge in Vietnam.
My Thuan Bridge in Vietnam
Typical Osterberg Cell® (O-Cell®) monitoring system setup in a pile (courtesy LOADTEST).
Results of an Osterberg Cell test at toe of pile (adapted fromFellenius, 1999).
The pile was load tested with Osterberg cells. Pile toe results shown below.
Note initial stiffer response due to residual load lockedin during pile construction (unload-reload cycle byboring and pile self weight and unmeasuredassociated displacements).
Hyperbolic fit considering residual load:
-Residual toe load of 10MN
-Initial modulus of 100MPa
-Ultimate load > 30MN
Key performance parameters = total anddifferential settlements.
Poulos et al (2001) tolerable settlement criteria innext Table (to be used as a guide only for low riskstructures; high risk structures should be assessedindividually).
Type of Structure Type of Damage Criterion Limiting Value
Framed buildings and reinforced load bearing
walls
Structural damage Angular distortion 1/150 to 1/250
Cracking in walls and partitions
Angular distortion
1/5001/1000 to 1/1400 for end bays
Visual appearance Tilt 1/300
Connection to services Total settlement 50 to75 mm sands50 to 135 mm for clays
Tall buildings Operation of elevators Tilt 1/1200 to 1/2000
Unreinforced load bearing walls
Cracking by relative sag Deflection ratio* 1/2500 for wall length/height =11/1250 for wall length/height =5
Cracking by relative hog Deflection ratio* 1/5000 for wall length/height =11/2500 for wall length/height =5
Bridges
Ride quality Total settlement 100 mm
Function Horizontal movement 38 mm (15 in)
Structural damage Angular distortion
1/250 for multi span1/200 for single span
Tolerable Settlements (after Poulos et al. 2001).
*deflection ratio = maximum relative deflection in a panel/panel length
Normally it is difficult to change a foundation designwhen a structure is partly built.
Interactive design is not generally feasible forfoundations (reason for the general conservatismadopted in the foundation design?).
3. Earth Fills
Evaluation of prediction of an earth-fill:instrumented trial fill.
Trial embankments:
-for high variability soil conditions prevail
-if no previous experience with certain soils
Structure on earth fill Criterion Limiting Value
Road including bridge approach road
Long term total settlement after road construction
50 mm
Differential settlement 20 mm over 5 m
Bridge abutmentLateral movement after footing installation
25 mm (not applicable, if footing is designed for lateral soil movement)
Building on shallow foundations Angular distortion 1/150 to 1/250 (will depend on the type of building)
Building - Service connection through earth-fill
Differential settlement
50 mm (will depend on type of connection)
Buried service pipe Angular distortion 1/200 (will depend on the type of pipe)
Clay liner Angular distortion 1/5 to 1/200 (will depend on the type of clay)
Typical settlement criteria adopted for earth-fill in soft ground.
Morwell River case history:
Large earth-fill project for river diversion, Victoria,Australia.
Details:
-70 m wide, 3.5 km long river diversion channel,mostly through loosely backfilled brown coal mine pit,Latrobe Valley
-13 million m3 of earthworks
-maximum height of embankment: 60 m
-30 m high built over 50 m of uncontrolled looselydumped fill (mixture of materials including clays withvery high variability)
Perspective view of river diversion embankment.
Dumped fill:
-soft to firm clays to 30 m depth and stiff clays below
-loose to medium dense sand and gravel layers
-profiles of CPT tip resistance in the next Figure toillustrate the variability.
0 2 4 6Con e B earing (M Pa )
-10
0
10
20
Dep
th(R
L,m
)
CPT tip resistance profiles in the dumped fill from 6 CPTs.
High variability of foundation and associated risks:trial embankment of about 1 million m3 of earth fill(400 m long, 200 m wide and up to 15 m high).
Location of trial embankment: within the footprint ofthe main diversion embankment.
Cross section of the main and trial embankmentsshown in the next Figure.
Final Embankment ProfileTrial Embankment Profile
Dumped Fill (Mid Field) Coal Dyke East Field (Current Mining Pit)
Cross section of Morwell River diversion embankment.
Typical settlements measured in settlement plates and magneticextensometers in trial embankment.
Typical excess pore pressures measured in vibrating wirepiezometers in trial embankment.
Comparison of CPT tip resistance profile before and after trialembankment.
Results of trial embankment:
- settlement of up to 1.5 m
-40% of above due to compression of naturalmaterials below dumped fill (layers of high moisturebrown coal)
-rate of settlement faster than initially anticipated(90% in 1 year)
- degree of settlements ≠ degree of observed porepressure dissipation in the dumped fill
-dissipation in upper 25 m in 1 year = 70 to 80% and25% below
-increase in su of 25 kPa in the top 25 m and 5 kPabelow.
These were a key findings demanding design review ofmain embankment.
Project required construction of culverts forconveyors, with fills up to 45 m above the base of theculverts and with varying foundation conditions.
Culverts were designed for maximum settlements upto 2 m and associated differential settlements.
Culverts were built parallel to each other at 50 mspacing, but at different levels.
Construction was completed successfully after designreview using data from trial fill.
4. Supported Excavations
Stability
Geotechnical stability of supported excavations isgoverned by:
- classical bottom heave mechanisms in soft to medium stiff clays;
- hydraulic uplift stability in frictional soils.
In stiff clays, or sands above the water table, stabilityis not an issue (provided support is properlydesigned).
Geometry definitions related to bottom heave stability analyses in clays.
In cohesionless or layered soils below WL, the wallshould be extended below the base of excavation toprevent hydraulic uplift or heave failure.
Failure mechanisms divided into two categories:
1 - “piping heave” failure, typical for homogeneoussoils (next Figure)
Seepage into excavation in uniform sand deposit.
2 - in layered soils the failure by uniform lifting of asoil plug from a high-permeable layer below the walltoe - next Figure.
CLAY
CLAY
SAND
ZL
uS
τs
Hydraulic plug uplift failure from confined sand layer.
Loads
In soft clays: loads = f (FSBH)
(shown by parametric FE analyses)
Normalized sum of strut loads, Ktotal, is defined asthe sum of maximum strut load at any one level atany stage of excavation divided by the loadcorresponding to vertical overburden pressure:
Ktotal = ∑Pmax/0.5γH2
0.0
0.5
1.0
1.5
2.0
0.6 1.0 1.4 1.8 2.2 2.6FACTOR OF SAFETY AGAINST BASAL HEAVE
NOR
M. T
OTA
L S
TRU
T FO
RCE;
Kto
t =
F
/ 0.5
H
2
A1, A2, A3B1, B2, B3C1, C2D1
Factor of safety against basal heave
Nor
m. t
otal
stru
t for
ce, K
tota
l
0.0
0.5
1.0
1.5
2.0
0.6 1.0 1.4 1.8 2.2 2.6FACTOR OF SAFETY AGAINST BASAL HEAVE
NOR
M. T
OTA
L S
TRU
T FO
RCE;
Kto
t =
F
/ 0.5
H
2
A1, A2, A3B1, B2, B3C1, C2D1
Factor of safety against basal heave
Nor
m. t
otal
stru
t for
ce, K
tota
l
0.0
0.5
1.0
1.5
2.0
0.6 1.0 1.4 1.8 2.2 2.6FACTOR OF SAFETY AGAINST BASAL HEAVE
NOR
M. T
OTA
L S
TRU
T FO
RCE;
Kto
t =
F
/ 0.5
H
2
A1, A2, A3B1, B2, B3C1, C2D1
Factor of safety against basal heave
Nor
m. t
otal
stru
t for
ce, K
tota
l
Normalised sum of maximum strut loads, Ktotal, in relation to bottomheave safety factor for excavation in soft normally consolidated clay.
Deformations
For soft to medium stiff clays:
Deformations = f (depth of excav. + depth firm stratum)
FSBH = f (depth of excav. + depth firm stratum)
Deformations = f (FSBH)
Non linear anisotropic soft soil model in FE analyses(Karlsrud and Andresen, 2008) showed:
% wall deflection = 0.2% H for FSBH ≥ 1.8
% wall deflection = 2 % H for FSBH ≈ 1.0
0.0
0.5
1.0
1.5
2.0
0.6 1.0 1.4 1.8 2.2 2.6
FACTO R OF SAFETY AGAINST BASAL HEAVE
A1, A2, A3B1, B2, B3C1, C2D1
Best fit
0.0
0.5
1.0
1.5
2.0
0.6 1.0 1.4 1.8 2.2 2.6
FACTO R OF SAFETY AGAINST BASAL HEAVE
A1, A2, A3B1, B2, B3C1, C2D1
Best fit
0.0
0.5
1.0
1.5
2.0
0.6 1.0 1.4 1.8 2.2 2.6
FACTO R OF SAFETY AGAINST BASAL HEAVE
A1, A2, A3B1, B2, B3C1, C2D1
Best fitBest fit h
max
/H (%
)
Factor of safety against basal heave
0.0
0.5
1.0
1.5
2.0
0.6 1.0 1.4 1.8 2.2 2.6
FACTO R OF SAFETY AGAINST BASAL HEAVE
A1, A2, A3B1, B2, B3C1, C2D1
Best fit
0.0
0.5
1.0
1.5
2.0
0.6 1.0 1.4 1.8 2.2 2.6
FACTO R OF SAFETY AGAINST BASAL HEAVE
A1, A2, A3B1, B2, B3C1, C2D1
Best fit
0.0
0.5
1.0
1.5
2.0
0.6 1.0 1.4 1.8 2.2 2.6
FACTO R OF SAFETY AGAINST BASAL HEAVE
A1, A2, A3B1, B2, B3C1, C2D1
Best fitBest fit h
max
/H (%
)
Factor of safety against basal heave
Normalised maximum wall displacement against basal heave safetyfactor from parametric FE analyses (n.c. clay).
591 case records reviewed by Moorman (2004) with:
K= EI/γws4 : system stiffness
E = modulus of wall
I = section modulus of wall
s = vertical spacing of struts/anchors
γw = unit weight of water
For stiff clays, data show:
-wall deflections smaller than expected from earlier works (0.05 to 0.3 % H).
-wall deflections ≠ f (system stiffness)
Relationship between normalized maximum lateral displacement,uhmax/H, system stiffness and safety factor correlations. From Moorman(2004).
Lateral extension of superficial subsidence = f (depthto firm bottom)
Settlement profile from expected lateraldisplacement: next Figure
Relationship between wall movement and ground settlements asproposed by Karlsrud (1997a) for soft/loose soils.
a) width of the excavation (extent of vertical stressrelief under excavation)
b) external loads from existing buildings
c) the stress-strain relation, drained and undrained
d) 3D-effects (end walls)
Other factors affecting relation between settlementand wall movements:
Measuring Excavation Performance
Karlsrud (1997) suggested:
-purpose of monitoring divided in 7 categories (nextTable)
-weigh the usefulness of measurements as afunction of the purpose (5 = high priority; 1 = notrequired)
Purpose CategoryHor.displ.at wall
Hor disp.behind wall
Ground surface
settlement
Vertical distribution of
ground movements
Settlement of
surrounding structures
Tilt and strain in surrounding
structures
A-Verify basis for design 5 3 4 1 3 1
B- Warning against failure 5 3 2 1 2 1
C- Observationaldesign approach 5 3 4 1 4 2
D- Influence on surroundings 4 3 5 2 5 4
E- Verify quality of construction 4 2 4 1 3 2
F- Improve design rules 5 4 4 2 4 3
G-Enhance knowledge 5 5 5 5 4 3
Table (a). Weighted value of different types of deformation measurements (value 5 high-amust, value 1 low-not required).
Purpose CategoryLoads in struts
oranchors
Temperature in struts
Strain in wall
Porepressurewithin the excavation
Porepressure
outside the excavation
Earth and pore pressures
against the wall
A-Verify basis for design 3 3 3 1-5 3-5 1
B- Warning against failure 5 4 3 1-5 1-3 1
C- Observationaldesign approach 5 3 4 1-5 1-3 1
D- Influence on surroundings 1 1 1 1-2 3-5 1
E- Verify quality of construction 4 2 1 1-4 1-4 1
F- Improve design rules 5 3 3 1-4 1-4 4
G-Enhance knowledge 5 3 5 2-4 3-5 5
Table (b) Weighted value of other measurements (value 5 high-a must, value 1 low-not required).
Acceptable deformations
Burland and Wroth (1975):
-defined set of displacement criteria (next Figure)
-related damage potencial to sagging and hoggingdisplacement pattern
a) Sagging building
b) Hogging building
Next Table relates settlement parameters defined andacceptance performance criteria (from parts ofTaipei Metro in the 1980’s).
Building type Setllementδmax (mm)
Absolute rotationθmax (rad)
Angulardistortio
nβmax
(rad)
Hoggingratio
Δ/L (rad)
Sagging ratio
Δ/L (rad)
Multi-storey framed building on raft foundation
45 2x10-3 2x10-3 0.8x10-3 1.2x10-3
Concrete framed building on footings
40 2x10-3 2x10-3 0.6x10-3 0.8x10-3
Brick building on footings 25 2x10-3 0.4x10-3 0.2x10-3 0.4x10-3
Example of settlement criteria used for excavations in connectionwith Taipei Metro.
Prediction and Evaluation of Excavation Behaviour
General perception :
-important projects are designed on poor soilinvestigation;
-numerical capabilities are more incorporated indesign practice than determination of relevant soilparameters.
a) ensuring that the safety level of the excavation isacceptable;
b) ensuring that displacements are within acceptablelimits.
Dealing with observed performance
Type and magnitude of alert and alarm levels forsupported excavation to be set depend on the twodesign criteria:
If measurements ≠ prediction: investigate discrepancyin any project stage.
Importance of designers to be available and involvedin all construction stage.
Interactive Excavation Design
Interactive design not well suited for deep excavationprojects (unless for a long excavation in uniformconditions).
For long cut-and-cover excavations (>500m) considerto vary anchors spacing or other elements in aninteractive process.
Preferable to start off with a conservative design, andthen move into a less conservative direction.
5. Tunnels
Evaluation of Prediction
Negro (2009) reviewed the practice, in Brazil, fordesigning urban tunnels in soil. The next twoFigures present how practitioners estimatesettlements routinely and how they assess liningloads in plane static systems.
3,3
16,7
50
20
10
0 25 50 75 100
A
B
C
D
E
30 answers
Frequency
A. empirical; B. semi-empirical; C. numerical; D. numerically derived; E. no indication of preference.
%
Preferred methods for settlements estimates in Brazil.
7,4
7,4
11,1
48,2
25,9
0 25 50 75 100
A
B
C
D
E
27 answers
Frequency
A. closed form solutions; B. analytical ring and spring solutions; C. numerical solutions with lining as bar elements and soil as discrete springs; D. finite element analysis; E. finite difference analysis.
%
Preferred methods for lining loads assessments in Brazil, using 2Dstatic systems.
Negro and Queiroz (2000) reviewed results of 65published comparisons made from 1977 to 1998.
This review was extended, by adding comparisonsmade after that period, totalling more than onehundred cases reviewed.
Next Tables present a complementary andcomprehensive list of comparisons made in the lastdecade, clearly not attempting to be exhaustive andretaining the same structure and criteria usedbefore.
Review of some numerical predictions of shallow tunnels usingnumerical modelling.
Surface Settlement Subsurf. Settl.
Horiz. Displ.
Lining LoadsAuthor/Year Origin Tunnel Ground Construction
Method Pred. Type Numerical Simulation Anal. C.M.S.
Mg D is t D tr Mg D tr Mg D tr Mg D tr
R. L.
661 Addenbrooke et al. 1997
UK Jubilee Line Extension
Stiff Clay Shield C1 y Imp. Conv. 2D-FE NLEPy4 < ? G - - - - - - 1
67Benmebarek et
al. 1998 FranceLyon Metro
Line D Silty Soils Slurry Shield C1 y Imp. Conv. 2D -FD EPf < - R ? G > G - - 2
68 Almeida e Sousa 1998 Portugal Sao Paulo Metro Tropical Lateritic
Porous Clay NATM C1 x 3-D 3D-FE NLEPy5 ? ? G ? G < R - - 2
69 Almeida e Sousa 1998 Portugal Sao Paulo Metro Tropical Lateritic
Porous Clay NATM C1 z Stress Red. 2D-FE NLEPy5 ? ? G > R ? R - - 2
70 Conceicao et al. 1998
Portugal Mato Forte Marlly Limestone Mined B1 y Stress Red. 2D-FE EPf ? < G - - < G - - 2
71Martins et al.
1998 Portugal Porto - Tunnel 4Granitic Residual Soil NATM
C1? x 3-D 3D-FE EPf ? ? G - - - - - - 1
72 Bakker et al. 1999 Netherlands 2nd Heinenoord
TunnelHolocene's Sands and Clays Slurry Shield C1 y Imp. Conv. 2D-FE EPf - - - - - - - > R 1
73 Bakker et al. 1999 Netherlands 2nd Heinenoord
TunnelHolocene's Sands and Clays Slurry Shield C1 y 3-D 3D-FE EPf - - - - - - - > R 1
74 Benmebarek et al. 1999
France Lyon Metro Line D
Alluvial Clays and Sands
Slurry Shield C1 y Imp. Conv. 2D -FD EPf ? < R - - ? G - - 2
76 Dias et al. 1999 France Line 2 Cairo Metro Alluvial Sand Slurry Shield C1 y 3-D 3D -FD EPf > - - - - - - - - 1
77 Dias et al. 1999 France Line 2 Cairo Metro Alluvial Sand Slurry Shield C1 y Imp. Conv. 2D -FD EPf < - - ? G - - - - 1
78 Tang et al. 1999
U.K. Heathrow Express Trial
London Clay NATM C1 x 3-D 3D-FE EPf < < G > G - - - - 1
79Gioda &
Locatelli 1999
Italy"Monteolimpio
2" Italy-Switzerland
Alluvial Sand Deposit NATM C1 y Stress Red. 2D-FE E < ? R - - > R - - 2
80 Dias et al. 2001 France Lyon Metro Line D
Silty Soils Slurry Shield C1 x 3-D 3D-FD? EPf ? ? R - - - - - - 1
81 Faria et al. 2001
Brazil Brasilia Metro Porous Clay NATM C1 z 3-D 3D-FE EPy ? ? R ? R - - - - 1
82 Wu et al. 2001 GermanyHigh-Speed
Line Cologne -Frankfurt
Weathered Sedimentary Rock NATM B1 x Stress Red. 2D-FE EPf ? ? G - - - - - - 1
83 Bakker et al. 2009
Netherlands 2nd Heinenoord Tunnel
Holocene's Sands and Clays
Slurry Shield B1 z 3-D 3D-FE EPf ? ? G - - - - - - 1
Review of some numerical predictions of shallow tunnels usingnumerical modelling (continued).
1. For cases 1 to 65, see Negro and Queiroz (2000).
2. Type of prediction according to Lambe (1973) classification. The question mark indicates a certain degree of uncertainty; x: actual prediction; y: back analyses; z: prediction with previously calibrated model; w: any of the former x, y or z.
3. Abbreviations: Pred. : prediction; Anal. : Type of Analysis; C.M.S. : constitutive model for soil; Subsurf. Settl. : subsurface settlement; Horiz. Displ. : horizontal displacements; Mg. : maximum magnitude; Dist. : maximum distortion; Dtr. : overall distribution; R.L. : Rank Level for the comparisons; E: Linear Elastic model; NLE Nonlinear Elastic model; NLEPy Nonlinear Elastic Plastic model with distinct yield and failure surfaces; EPf: Elastic Plastic model with yield and failure surfaces coinciding; EPy: Elastic Plastic model with distinct yield and failure surfaces; Core Removal: Progressive Core Removal; Stress Red. : Ground Stress Reduction; Imposed Conv. : Imposed Tunnel Convergence; Stiff. Red. : Ground Stiffness Reduction; Imp. Stress : Imposed Stress; ≡: calculated value approximately equal to measured value; >: calculated value greater than measured value; <: calculated value smaller than measured value; G: good; R: regular; P: poor; FE : finite element; FD : finite difference; 2 and 3D : two and three dimensional.
4. Anisotropic ICFEP. 5. Lades's model. 6. With strain softenning.
2008 China Thunder Bay Silty and Sandy Soils TBM C1 x 3-D 3D-FD EPf ? ? G ? G > G - - 2
110 Shahin et al. 2008 Japan 2D Model Aluminium Rods Self Weight B1? x Imp. Conv. 2D -FE EPy ? ? G - - - - ?9 G 3
Surface Settlement Subsurf. Settl.
Horiz. Displ.
Lining LoadsAuthor/Year Origin Tunnel Ground Construction
Method Pred. Type Numerical Simulation Anal. C.M.S.
Mg D is t D tr Mg D tr Mg D tr Mg D tr
R. L.
Prediction When it Made Results
A Before event -
B During event not known
B1 During event known
C After event not known
C1 After event known
Negro and Queiroz (2000) appendage:
(x): actual prediction
(y): back-analysis
(z): prediction with previously calibrated model
(w): not clearly identified case
Classification of prediction, according to Lambe (1973):appendage by Negro and Queiroz (2000).
Comparison Level
Vertical Displacements
Horizontal Displacements
Lining Loads
1 X or X or X
2 X and X
3 X or X and X
4 X and X and X
Rank Levels for comparisons between prediction and performance oftunnels in soil (Negro, 1998).
Usual lack of details in case history description or inthe analysis can alone hinder any carefulassessment of the prediction.
Notwithstanding this, an attempt was made toidentify the efficiency of the modelling tools used,related to soil type, construction method, type ofprediction, of numerical simulation, of constitutivemodel used.
No clear correlation was found between thosefactors and the certainty of the prediction. Thisdifficulty, as earlier, allowed only a broad appraisalto be attempted in what follows.
0 20 40 60 80 100
C1
B1
A
B
C
x y z w
Frequency
%
Types of prediction according to Lambe (1973), with appendage byNegro and Queiroz (2000).
Similarly to the previous review no clear relationwas found in the current one, relating quality ofprediction and features of the numerical modelling.
Over the last decade, an increased use of 3D FEanalysis was noted and some yielded the bestranked results reviewed (case 88, by Melo andPereira 2002, on a slurry shield and case 104 byMarques et al. 2006, on NATM).
There is a clear need of less biased evaluations oftype A prediction, preferably of rank level 4.
Evaluation of performance
Evaluation of performance is usually done by straightcomparison of field measurements with predictedquantities. The use of limiting displacements forassessment of performance of shallow tunnels hasshortcomings.
This is particularly true for stiff to hard groundmasses: a near ultimate state condition may bereached with ground displacements of few centimeters(raising undue skepticism on the validity of a givenlimiting displacement).
This was the case of some documented tunnel failuresin which reduced magnitude limiting displacementswere exceeded just prior to collapse.
Next Table illustrates some of these cases, all of thembuilt by the so – called NATM.
Maximum surface and crown settlements measured prior to the collapseof some NATM tunnels.
Project Itaquera (Brazil) S. Amaro (Brazil)
Heathrow (U.K.) Pinheiros (Brazil)
Year 1989 1993 1994 2007
Location Sao Paulo, Brazil Sao Paulo, Brazil Heathrow, U.K. Sao Paulo, Brazil
Surface Settlement before collapse (cm) 0.8 3.4(1) 5.5 6.3(1)
Crown settlement before collapse (cm) 2.7 3.5 6.0 3.4
Reference Sozio et. al., 1998 This report HSE, 2000 This report
Note: (1) Includes settlements due to drainage.
Cording et al (1971) proposed for performanceevaluation of underground rock caverns the ratiomeasured to calculated elastic displacement atopening contour:
ratio ≤ 2: opening stable
5 ≤ ratio ≤ 10: opening unstable
Kuwajima and Rocha (2005) presented a type Aprediction for displacements around the PinheirosMetro Station, Sao Paulo, Brazil, a shallow rockcavern, with a thin rock cover, below residual andsedimentary soils, that eventually collapsed during itsconstruction (Assis et al, 2008 and Barton, 2008).
Kuwajima and Rocha (op. cit.) performed a 3Dsequential finite element analysis, assuming for therock a linear elastic plastic behaviour with a nonassociated plastic flow to the Mohr-Coulomb failurecriterion.
Analysis results:
-no plastic zones formed in the rock mass, whichbehaved essentially as a linear elastic ground.
-maximum settlements of the rock cover of the orderof 9 mm.
Following Cording et al (op.cit) suggestions, formeasured settlements smaller than 18 mm, the cavitywould have been essentially stable. Instabilities wereto be expected for deep settlements greater than 45to 90mm.
The station collapsed on the 12 of January 2007 justafter the field instrumentation measured anaccumulated settlement of 34 mm in the rock cover(previous Table), a ‘gray zone’ value, larger than thelower limit offered but smaller than the upper limitproposed by the criterion.
Pinheiros Metro Station Collapse (12th January 2007)
It appears that dimensionless quantities derivedfrom displacements as well as from other variables,can operate better for generalizations and calibrationwith the practice.
Moreover, redundancy of evaluation is required dueto the very nature of the assessment, involvingcomplex ground conditions as well as complexboundary conditions.
Accordingly, a review is presented herein ofperformance indicators for tunnels in soil, some ofthem related to serviceability, some to ultimate state,some already in use and some other not quite so.
a) Limiting crown settlement to tunnel diameter ratio (Sc /D).
b) Limiting surface to crown settlement ratio at tunnel axis (Ss/Sc ).
Normalized settlement ratios observed in some tunnels case histories(modified from Ward and Pender, 1981).
c) Limiting surface and subsurface distortions().
d) Longitudinal distortion index (LDI).
LDI(x) = u(x)/ x
Distributions of LDI along the cover of a stable (a) and an unstable (b)tunnel.
As collapse is approached, the displacementvectors orientation change at these pointsshowing increasing angle to the vertical against thedirection of excavation.
Note that in non homogeneous ground, the changingorientation of displacement vector is related to thetunnel approaching zones of contrasting stiffness asmajor faults or dikes crossing a rock tunnel.
This effect was noted and explored by Schubert andBudil (1995) as an element of performanceanticipation for deep tunnels.
Next Figure taken from Grossauer, Schubert andSellner (2005) illustrates this effect.
Settlement and displacement vector orientation of points at a deeptunnel contour as a fault is approached (modified from Grossauer etal, 2005).
e) Volume of surface settlement (%Vs).
Construction Quality Range of %VS
High < or = 0.5%
Normal 0.5 and 1%
Poor 1 and 3%
Pre-failure condition 3 and 40%
Construction quality and volume of surface settlements.
f) Volume of soil lost (loss of ground, %Vl)
%Vl = 100 . Sc . 2(R + y) / 2 R2
Cording & Hansmire (1975)
Technology Ground type Range of loss of ground (% Vl)
Open face tunnelling Stiff clays 1 to 2
NATM Stiff clays 0.5 to 1.5
EPB and slurry shields Sands >0.5
EPB and slurry shields Soft clays 1 to 2
EPB and slurry shields Mixed face 2 to 4
Typical losses of ground (from Mair and Taylor, 1997).
Soil Type Limiting Distortion Range R/R (%)
Stiff to hard clay (OF < 2.5 – 3) 0.15 – 0.40
Soft clays or silts (OF > 2.5 – 3) 0.25 – 0.75
Dense or cohesive sands, most residual soils 0.05 – 0.25
(1) Add 0.1 – 0.3% for tunnels in compressed air, depending on air pressure.
(2) Add appropriate distortion for external effects such as passing neighbouring tunnels
(3) Values assume reasonable care in construction, and standard excavation and lining methods.
h) Reference lining load (%Overburden).
Provided good ground control conditions are present,ground stress release at lining activation varies from20 to 70% (Negro and Eisenstein, 1997) and finalaverage ground stresses onto the lining correspondedto 25 to 75% of the ground in situ stresses at tunnelaxis. More frequently than not they represent 50%of in situ stresses.
i) Maximum lining load.
Check buckling of thin linings in weak soils under highground stresses.
j) Dimensionless crown displacement (Uc ).
Uc = Sc . Eco / Dcro
Inspection of model test results revealed that Uc ≥1.8 corresponds to near collapse condition (fulldevelopment of high shear strain concentrations).
This would be a near ultimate limiting crowndisplacement. For good ground control conditions.(shear band formation is not noted) where tunnelserviceability is not jeopardized by excessivedeformations (Sc/D smaller than 3 to 4%) and lossesof ground are acceptable (up to 3%), the limitingserviceability crown displacement Uc ≤ 1.0 (FS ofabout 1.5).
These criteria were tested in few dozens of welldocumented case histories of actual tunnelconstructions and was proved satisfactory.
Limitations of the criteria:
-requirement to know the profile of the initial tangentmodulus of elasticity, at the section where the crownmeasurements are being taken.
-stress path dependency of this modulus,complicating the assessment.
Next Table reproduces dimensionless crowndisplacements at the four tunnel cases given inearlier Table prior to their collapses.
The results seem to be in agreement with thecorrelation given above for ultimate state.
Project Itaquera (Brazil) S. Amaro (Brazil) Heathrow
(U.K.) Pinheiros (Brazil)
Ground cover Hard sandy clay Stiff silty clay Hard grey
London clayResidual soil, gneiss
saprolite
Estimated Eco (MPa) 300 60 200 1000
Crown settlement before collapse (mm) 27 35 60 34
Uc before collapse 2.07 1.87 4.00 5.86
Dimensionless crown displacements at some tunnels prior to collapse.
k) Non-conforming horizontal longitudinal displacement of tunnel face (uf)
Require 3D linear elastic FE analysis to get maximumface extrusion uf. Else consider approximate solution:
uf ≈ 0.5 D cro / Eso
where:
cro = in situ radial stress at crown
Eso = in situ tangent modulus of soil at springline
Non conforming condition if ratio of measured toestimated extrusion ≥ 2.
l)Measured to calculated crown settlementratio at tunnel face.
Require 3D linear elastic FE analysis. Else considerapproximate solution:
Ucf = 0.375-0.147Ko
The estimated crown settlement results:
Scf = D cro (0.375-0.147Ko)/Eco
The accumulated frequency distribution of measuredto calculated crown settlement ratios at the face inmore than 50 tunnel projects are presented in thenext Figure.
Frequency distribution of measured to calculated crown settlementratio at tunnel face of some tunnels.
Kolmogorov-Smirnov adherence test:
-a log normal accumulated frequency distribution fitsbetter the data than a normal Gaussian distribution.
-50% of the cases, the settlement ratio is close to 1.0.
-90% of the cases the ratio varies between 0.3 and2.0.
-10% of the cases in which this ratio exceeded 2.0were identified as having shown poor ground controland excessive loss of ground through the face.
A non conforming condition is at play for ratios greaterthan 2.0. This figure could be taken as a safeultimate limiting value.
m) Measured to calculated transverse springlinedisplacement ratio at tunnel face.
Require 3D linear elastic FE analysis. Or else considerestimate by:
Ssf = (0.210 – 0.033/Ko) D sro / Eso
If ratio > 1 non-conforming value
If ratio > 2 (?) ultimate limiting value
n) Measured to calculated longitudinal distortion ratio.
Require 3D linear elastic FE analysis. Else considerestimate by:
dmax 0.6 cro /Eco
If ratio > 1 non-conforming (serviceability)
o) Limiting dimensionless crown settlementincrement after prefabricated lininginstallation.
Item Symbol Definition or ConceptLimiting Values
Serviceability Ultimate State
a Sc/D Limiting crown settlement to tunnel diameter ratio.0.03 to 0.04
(or minimum required clearance)
0.03 to 0.15
b Ss/ScLimiting surface to crown settlement ratio at tunnel axis (or
settlement increments ratio). - 1.0
c Limiting subsurface distortions. - 1/10 for soft and 1/30 for stiffer soils
dLDI Longitudinal distortion index. - Change in sign
Displacement vector orientation. - Increasing angle to vertical
e %Vs Volume of surface settlement. 0.5 to 1% 3 to 40%
f %Vl Volume of soil lost (loss of ground). 0.5 to 4% (usual.)4 to 6% (excep.) 8 to 10%
g D/D% Lining distortions. 0.05 to 0.75% (dep. on soil) -
h %Overburden Reference lining load. 25 to 75% -
i p Maximum lining load. - 3EI / R3+Es (1+)
j Uc Dimensionless crown displacement. 1.0 1.8
k uf ratio measured to calculated maximum face extrusion (uf) ratio. > 2 (non-conforming) -
l Scf ratio Measured to calculated crown settlement ratio at tunnel face (Scf). 1.0 (non-conforming) 2.0 (safe)
m Ssf ratio Measured to calculated transverse springline displacement ratio at tunnel face (Ssf).
1.0 (non-conforming) 2.0
n dmax ratio Measured to calculated longitudinal distortion ratio (dmax). > 1.0 (non-conforming) -
o UcultLimiting dimensionless crown settlement increment after
Summary of Performance Indicators for shallow tunnels in soil.
Interactive Tunnel Design.
Tunnels in soil are ideal structures for its use.
Soil tunnelling is essentially an industrial process (costof the product is dependent on the production rate).
Condition for its application requires possibility formodification of design during construction (notuncommon in traditional tunnel construction contracts,less common in TBM contracts).
Next Table explores the potentials of applyinginteractive design in ductile and brittle scenarios fortunnels in soil.
Careful when dealing with a ductile groundenvironment using a brittle support or lining system:they are incompatible and should not be used for thebest use of the interactive design.
Avoid the use of brittle support in a brittle groundmass, for the sake of safety.
These two conditions led to the decline of unreinforcedconcrete lining and to the use of minimum steelreinforcement which is generally specified in thedesign of the tunnel lining.
Feature Ductile Brittle
Failure development
Gradual development of ground settlements with large ground losses, intense lining cracking and large distortions.
Abrupt failure, after limited groundsettlement, small loss of ground, minorlining cracking and smaller distortions.
Governing limit state Serviceability. Ultimate.
Predicability based on experience Reasonable, rich, case histories. Limited, less case histories.
Numerical predicability Reasonable predictions. Difficult and complex due to strainsoftening.
Instrumentation Simple instrumentation is valuable. Simple instrumentation may not detectpre-failure displacements.
Contingencies Fairly ample time for action. Too short time for proper action.
Impact on interactive design
Good, induced damages can be controlled, requires use of ductile lining for conformance, good conditions for optimizations and savings.
Bad, requires conservative design andconstruction, requires use of ductile liningfor safety, poor opportunities foroptimizations and savings.
Ductile and brittle scenarios for tunnels in soil and the impact oninteractive design (modified from Nicholson, 1996).
6. Geotechnical Instrumentation
FIBRE OPTIC SENSORS
Point Sensors Long gauge sensors Distributed sensors
Fabry -Perot Interferometers
Fibre bragg grating
SOFO Interferometric Sensors Raman scattering
Brillouin scattering
Classification of FO sensing technologies used in industry (Glišić andInaudi, 2007).
Summary of FO sensing types and Typical Performance (Glišić andInaudi, 2007).
Fabry-Perot Interferometry
Fibre Bragg Grating
SOFO Interferometry
Raman Scattering Brillouin Scattering
Sensor Type Point Point Long-gauge (integral strain) Distributed Distributed
Main measureable parameters
StrainTemperature
Pressure
StrainTemperatureAccelerationWater level
DeformationStrain
TiltTemperature Strain
Temperature
Multiplexing Parallel In-line and parallel Parallel Distributed Distributed
Measurement points in one line 1 10 - 50 1 10 000 30 000
Typical accuracy
Strain = 1e� = 100 mT2 = 0.1°C
P3 = 0.25% full scale
Strain = 1D� = 1 mT2 = 0.1°C
Strain = 1eD� = 1m
Tilt = 30 mradT2 = 0.1°C Strain = 20
T2 = 0.2°C
Most promising
Summary of FO sensing types and Typical Performance (Glišić andInaudi, 2007) (continued).
Fabry-Perot Interferometry
Fibre Bragg Grating
SOFO Interferometry
Raman Scattering Brillouin Scattering
Sensor Type Point Point Long-gauge (integral strain) Distributed Distributed
Precision and Stability using the best possible
equipment and installation
?5Strain = 1 T2 = 0.1°C
D1 = 2 m (independent of sensor length,
proven over more than 10 years)
T2 = 0.1°C Strain = 20T2 = 0.2°C
Typical Resolution Strain = 1T2 = 0.1°C
Strain = 1 T2 = 0.1°C Strain = 1
Range (length of pulse) = 1 mStrain = 1 T2 = 0.2°C
Range - - 20 m gauge 8 km
30 km, 250 km with range extenders; 500 km
expected by mirroring 250 km setups, but not yet proven (Nikles et al.,
2005)
Fibre type Multimode Single mode Single mode Multimode Single mode
Note:
= Deformation.2. T = Temperature.3. P = Pressure.4. ? = No reference could be obtained which indicates typical precision.
Most promising
Construction of the brine pipeline, with FO sensor placed at 6 o’clockposition (Nikles et al, 2004). BOTDR.
Measured profile before and after leakage. Vertical scale representsBrillouin frequency shift, while the horizontal scale represents thelength of the monitored section (Nikles et al, 2004).
Schematic representation of remote monitoring using repeaters andremote generation for optical signal generation and processing; (a) Up to75 km; (b) More than 75 km. (Nikles et al, 2005). BOTDR.
Instrumentation layout and photos of the sensor system duringinstallation (Vorster, 2005). BOTDR.
Typical result indicating joint formation in a prestressed concretepipeline (Vorster, 2005). BOTDR.
New geotechnical FO Inclinometer based ondistributed Brillouin scattering sensors (including areference fibre).
A prototype: could use Smartape or Smartprofile strainsensors applied at 0, 90, 180 and 360 degrees over a100 mm HDPE pipe, adequately protected.
For a standard strain resolution of 2010-6 m/m and ata spatial resolution of say 1500mm, one could measurerotations larger than 1:1,666 in the A and B orthogonaldirections.
The prototype could be horizontally installed parallel tothe axis of a shallow tunnel in soil, prior to itsconstruction, using a robust and precise horizontaldirectional driving drilling system, at a certain depthbelow surface, in lengths of a city block (from 100 m to200m).
Conceptual FO Inclinometer installed around a tunnel in soil.
Redundancy offered by the four distributed Brillouinsensor assures assessment of spatial bending of thetube through the average curvature.
The average straining of the tube provides informationon its longitudinal traction or compression, wheninstalled parallel to the tunnel and as the tunnelheading advances.
The unstrained reference fibre used allows absolutestrain measurements.
If temperature sensors are included, the inclinometerhorizontally installed can detected risk of blow-outs, if(warmer) air pressure is lost ahead the tunnel face, orif excessive (also warmer) slurry or water flow isestablished ahead the TBM face, possibly creatingexcessive ground heave.
7. Final Remarks
•Straight comparisons between measurements andpredictions may not be sufficient anticipate problems.
•Evaluation of prediction of deep foundations and earthfills in soft ground demand field test (pile testing andtrial fill).
•Numerical capabilities are more incorporated in designpractice than determination of relevant soil parameters.
•Interactive design is most suited for long geotechnicalstructures such as tunnels (excepted in brittlescenarios) and least suited for foundations (possiblereason for conservative design of foundations?)
•Need for less biased evaluation of type A prediction of rank level 4.
•Need for proper recording case histories and numerical analyses.
•Need to use performance indicators with redundancyfor performance evaluation.
•F0 distributed sensors (optical time-domainreflectometers) are most promising for geotechnicalinstrumentation.
ACKNOWLEDGMENTS
The authors wish to thank their work institutions,namely Bureau de Projetos e Consultoria (Brazil),NGI – Norwegian Geotechnical Institute (Norway),Golder Associates (Australia) and Aurecon SouthAfrica (Pty) Ltd (South Africa), for supporting thisreport preparation and for their continued interest inbridging development between Academy andIndustry.
