DAARWIN: Reducing Geotechnical Uncertainty by

Connecting Numerical Models with Monitoring Data

Cristian de Santos, PhD.
CEO & Co-Founder
SAALG GEOMECHANICS

Shaping the future of a sustainable, efficient and safe

construction industry.

Geotechnical Society of Singapore SEMINAR, 20th July 2023

SAALG & MYSELF
W H O W E A R E

 Cristian de Santos, PhD

 CEO & Co-founder of SAALG Geomechanics: 2016 - Present

 PhD Geotechnical Engineering (UPC / Caltech): 2007-2015

“Backanalysis Methodology Based on Multiple Optimization Techniques

for Geotechnical Problems”

(Prof. Antonio Gens and Prof. Alberto Ledesma)

 BS & MS Geological Engineering (UPC): 2001-2007

“Deformaciones del revestimiento y asientos producidos por la

construcción de un túnel mediante tuneladora”

(Prof. Eduardo Alonso and Eng. Nicola Della Valle)

SAALG & MYSELF
W H O W E A R E

 SAALG GEOMECHANICS

“A Barcelona-Based Startup specialized in geotechnical

software to reduce geotechnical uncertainty”
SAALG & MYSELF
W H O W E A R E

 SAALG GEOMECHANICS

“A Barcelona-Based Startup specialized in geotechnical

software to reduce geotechnical uncertainty”

PARTNER OF
SAALG & MYSELF
W H O W E A R E

 DAARWIN

HISTORICAL INFORMATION PREDICTIVE MODELS MONITORING DATA CONSTRUCTION PROGRESS

SAALG & MYSELF
W H O W E A R E

 WHO TRUST ON US
 ASSET OWNERS

 CONTRACTORS

 ENGINEERING FIRMS
TA B L E O F C O N T E N T S
 Introduction
 Problem: Geotechnical Uncertainty
 Traditional Solution
 Backanalysis
 Genetic Algorithms (optimization technique)

 Case Study 1: The Girona High-Speed Train Station

 Project – Methodology - Results
 Daarwin implementation

 Case Study 2: Excavation Box in UK (in collaboration with Mott MacDonald)

 Case Study 3: Commercial Building in Singapore (in collaboration with Kajima)

 Conclusions & Recommendations

INTRODUCTION
P R O B L E M : G E O T E C H N I C A L U N C E R T A I N Y

“Not knowing how the ground behaves

has a high environmental & financial cost
to the construction industry and to
society”

OVER-DIMENSIONED UNFORSEEN EVENTS DURING

DESIGNS CONSTRUCTION & MANTENANCE

DELAYS MALFUNCTION FAILURE

INTRODUCTION
T R A D I T I O N A L S O L U T I O N

 INCREASE ground investigation (during pre-design)

 USE sophisticated numerical tools (during design)
 MEASURE the real soil response (during construction & exploitation)

Disconnected

Ground Investigation Numerical Tools Measurements

Project Lifecycle

Poor Information Unrealistic Parameters Data not Analyzed

(only compare with threshold vales)

“The industry is wasting the potential of

numerical tools and missing the high value of
analyzing monitoring data”
INTRODUCTION
B A C K A N A L Y S I S

Computed with a model, i.e. F.E. Model

Measurements
Depend on “p” parameters: x = M (p)

In Situ Measurements Calculations

Sliding
Micrometer

Inclinometer Inclinometer

The problem is reduced to minimize J

INTRODUCTION
G E N E T I C A L G O R I T H M S ( O P T I M I Z A T I O N T E C H N I Q U E )

K0

Eurref [kN/m2]
INTRODUCTION
G E N E T I C A L G O R I T H M S ( O P T I M I Z A T I O N T E C H N I Q U E )

Initial Generation

K0

Eurref [kN/m2]
INTRODUCTION
G E N E T I C A L G O R I T H M S ( O P T I M I Z A T I O N T E C H N I Q U E )

Second Generation

K0

Eurref [kN/m2]
INTRODUCTION
G E N E T I C A L G O R I T H M S ( O P T I M I Z A T I O N T E C H N I Q U E )

Third Generation

K0

Eurref [kN/m2]
INTRODUCTION
G E N E T I C A L G O R I T H M S ( O P T I M I Z A T I O N T E C H N I Q U E )

Fourth Generation

K0

Eurref [kN/m2]
INTRODUCTION
G E N E T I C A L G O R I T H M S ( O P T I M I Z A T I O N T E C H N I Q U E )

Final Generation

K0

Eurref [kN/m2]
TA B L E O F C O N T E N T S
 Introduction
 Problem: Geotechnical Uncertainty
 Traditional Solution
 Backanalysis
 Genetic Algorithms (optimization technique)

 Case Study 1: The Girona High-Speed Train Station

 Project – Methodology - Results
 Daarwin implementation

 Case Study 2: Excavation Box in UK (in collaboration with Mott MacDonald)

 Case Study 3: Commercial Building in Singapore (in collaboration with Kajima)

 Conclusions & Recommendations

T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

 Located 80km North to Barcelona City (SPAIN)

 Population of 97.586 people

T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

Cathedral

Girona High-Speed Railway Station

Bridge Historical Building
(3m far from the station) (next to the station)

The station is surrounded by many other

infrastructures and buildings in the downtown of
Girona City
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

Large underground excavation under water conditions in the downtown of Girona City
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

Fill Material

Alluvial

Shallow Silty Clay (PLA)

Shallow Sandy Gravel (PGA)

Deep Silty Clay (PLB)

Deep Sandy Gravel (PGB)

T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

Constitutive γ dry γ sat Kh Kv E50Ref EOedRef EurRef C Φ Psi

Material K0nc Ko m ν Rf (kN/m2) R
Model (kN/m3) (kN/m3) (m/dia) (m/dia) (kN/m2) (kN/m2) (kN/m2) (kN/m2) (°) (°)
Fill M-C 18 20 - 0,5 0,514 0,514 2,00E+04 - - - 0,3 10 40 0 - 0,6
Aluvial M-C 18 20 - 0,5 0,514 0,514 5,00E+04 - - - 0,3 0 40 0 - 0,6
Pl_A HS 19 21 0,5 0,5 0,026 0,026 2,50E+04 2,50E+04 8,75E+04 1 0,2 15 28 0 0,9 0,6
Pl_B HS 19 21 0,5 0,5 0,026 0,026 5E+04 5E+04 1,75E+05 1 0,2 15 28 0 0,9 0,6
Pg_A HS 19 21 0,5 0,5 0,321 0,321 2,50E+04 2,50E+04 8,75E+04 1 0,2 10 35 0 0,9 0,6
Pg_B HS 19 21 0,5 0,5 0,321 0,321 5E+04 5E+04 1,75E+05 1 0,2 10 35 0 0,9 0,6

Eurref Cohesive
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

Plaxis 2D Plaxis 3D
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

Design Monitoring Data

NUMERICAL MODEL
INCLINOMETER
(using design soil parameters)
(sensor that measures horizontal
displacements depth-wise)
NO MATCH

Displacements predicted Real displacements obtained after

before starting excavation finishing an excavation phase
(NON-CALIBRATED NUMERICAL MODEL) (DATA FROM SENSORS)
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

PGA
PLA
6 Different backanalysis, 1 for each excavation phase
PGB

PLB

Identification of the stiffness moduli (E50ref, Eoedref and

Monitoring Data
Eurref ) of 4 different layers (PGA, PLA, PGB, PLB)

Search Space (38.891.389.681 combinations):

Min_E50ref=5MPa; Max_ E50ref=500MPa

Min_Eurref=25MPa; Max_ Eurref=2500MPa

T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

6 Different backanalysis, 1 for each excavation phase

Exc_1 Exc_1 – Exc_2 Exc_1 – Exc_3 Exc_1 – Exc_4 Exc_1 – Exc_5 Exc_1 – Exc_6
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

The backanalize stiffness (Eurref ) is closer to the values

obtained from the Cross-Hole tests

Original Eurref Cohesive

Backanalaze Eurref Cohesive

T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

Traditional Method Monitoring Data Backanalysis

NO MATCH MATCH

Displacements predicted Real displacements obtained after Daarwin predictions obtained

before starting excavation finishing an excavation phase after the first excavation phase
(NON-CALIBRATED NUMERICAL MODEL) (DATA FROM SENSORS) (CALIBRATED NUMERICAL MODEL)
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S
T H E G I R O N A H I G H -S P E E D T R A I N STAT I O N
P R O J E C T - M E T H O D O L O G Y - R E S U L T S

≈600m

Exc_7

50 Excavation phases of 12m long 1 Excavation phase of 600m long

Cost of Excavation phase Exc_7 Cost of Excavation phase Exc_7

• Excavation: 1.011.378,31 € • Excavation: 194.148,99 € (saving: 817.229,32 €)
• Structure: 4.285.347,90 € • Structure: 3.371.358,52 € (saving: 913.989,39 €)
• Joints: 324.354,72 € • Joints: 5.021,72 € (saving: 319.333,00 €)

5.621.080,93 € 3.570.529,22 €

MAXIMIZING THE EXCAVATION PRODUCTIVITY OF PHASE EXC_7

2.050.551,71 € (savings)
º
TA B L E O F C O N T E N T S
 Introduction
 Problem: Geotechnical Uncertainty
 Traditional Solution
 Backanalysis
 Genetic Algorithms (optimization technique)

 Case Study 1: The Girona High-Speed Train Station

 Project – Methodology - Results
 Daarwin implementation

 Case Study 2: Excavation Box in UK (in collaboration with Mott MacDonald)

 Case Study 3: Commercial Building in Singapore (in collaboration with Kajima)

 Conclusions & Recommendations

E XC AVAT I O N B OX I N U K
I N C O L L A B O R A T I O N W I T H M O T T M A C D O N A L D

Box 1
E XC AVAT I O N B OX I N U K
I N C O L L A B O R A T I O N W I T H M O T T M A C D O N A L D
E XC AVAT I O N B OX I N U K
I N C O L L A B O R A T I O N W I T H M O T T M A C D O N A L D

Mercia Mudstone Grade II/I parameters (Characteristic range)

 UCS: 2 ~ 3 Mpa
 GSI: 35 ~ 55
E XC AVAT I O N B OX I N U K
I N C O L L A B O R A T I O N W I T H M O T T M A C D O N A L D

Observation of excavation from Observation of excavation from +81

+87.3 to +81.3 mAOD to +77 mAOD

 Mass strength revealed during bulk excavation, higher than indicated

by GI boreholes (bh’s);

 Mudstone less fractured in-situ than indicated by bh fracture indices

(drilling induced disturbance)
E XC AVAT I O N B OX I N U K
I N C O L L A B O R A T I O N W I T H M O T T M A C D O N A L D

Backanalysis
E XC AVAT I O N B OX I N U K
I N C O L L A B O R A T I O N W I T H M O T T M A C D O N A L D

“The benefits of DAARWIN were quantify

with 1 month program reduction, £300K
cost savings, and 440 tCO2e saving”
TA B L E O F C O N T E N T S
 Introduction
 Problem: Geotechnical Uncertainty
 Traditional Solution
 Backanalysis
 Genetic Algorithms (optimization technique)

 Case Study 1: The Girona High-Speed Train Station

 Project – Methodology - Results
 Daarwin implementation

 Case Study 2: Excavation Box in UK (in collaboration with Mott MacDonald)

 Case Study 3: Commercial Building in Singapore (in collaboration with Kajima)

 Conclusions & Recommendations

COMMERCIAL BUILING IN SINGAPORE
I N C O L L A B O R A T I O N W I T H K A J I M A

Section 01

Section 02
COMMERCIAL BUILING IN SINGAPORE
I N C O L L A B O R A T I O N W I T H K A J I M A

Inclinometer

Section 01

Section 02
COMMERCIAL BUILING IN SINGAPORE
I N C O L L A B O R A T I O N W I T H K A J I M A

Excavation below L1 Excavation below B1

Excavation below B2 Excavation to FEL

COMMERCIAL BUILING IN SINGAPORE
I N C O L L A B O R A T I O N W I T H K A J I M A

All Data Backanalysis used Data

COMMERCIAL BUILING IN SINGAPORE
I N C O L L A B O R A T I O N W I T H K A J I M A

 Using data from Excavation Phase B1 only

 Final population of 4398 good individuals out of a population of 4925 individuals after 50 generations
COMMERCIAL BUILING IN SINGAPORE
I N C O L L A B O R A T I O N W I T H K A J I M A

 Using data from Excavation Phase B1 and B2

 Final population of 4939 good individuals out of a population of 5430 individuals after 50 generations
COMMERCIAL BUILING IN SINGAPORE
I N C O L L A B O R A T I O N W I T H K A J I M A

 Using data from Excavation Phase B1, B2, and FEL

 Final population of 3682 good individuals out of a population of 5147 individuals after 50 generations
TA B L E O F C O N T E N T S
 Introduction
 Problem: Geotechnical Uncertainty
 Traditional Solution
 Backanalysis
 Genetic Algorithms (optimization technique)

 Case Study 1: The Girona High-Speed Train Station

 Project – Methodology - Results
 Daarwin implementation

 Case Study 2: Excavation Box in UK (in collaboration with Mott MacDonald)

 Case Study 3: Commercial Building in Singapore (in collaboration with Kajima)

 Conclusions & Recommendations

C O N C LU S I O N S & R E C O M M E N DAT I O N S
B A C K A N A L Y S I S I N G E O T E C H N I C A L P R O B L E M S

 Backanalysis is a powerful methodology to calibrate sophisticated numerical models in early

construction stages.

 Validate the hypothesis use in design

 Detect instabilities in advance

 Optimize the design and the construction procedure

 Check the assumption that the numerical model is representing the physics of the problem

 Make sure that the monitoring data is showing consistent values before using them

 Have a clear vision of how is the construction site every day that you have measurements
(correlate calculation phases with construction phases)

 To effectively implement changes derived from a backanalysis, better to plan them in advance
(during design). Al stakeholders must be involved.
 THANK YOU FOR YOUR ATTENTION
Geotechnical Society of Singapore SEMINAR, 20th July 2023

Shaping the future of a sustainable, efficient and safe

construction industry.

Cristian de Santos, PhD.

CEO & Co-Founder
EMAIL: [email protected]
PHONE: +34 625 08 08 70

PARTNER OF