Design of Risk Mitigation Measures

for Natural Terrain Landslide Hazards

November 2020

Natural vs Man-made Slope : Different Strategies

Mitigation
Landslide Known
Man-made Potential Upgrading Landslide Works Natural
Risk Landslide
Slopes Instability Works Prevention (or hybrid Hillsides
Reduction Hazards
approach)

Reducing Overall Landslide Risk in Hong Kong

2
Frontal Characteristics of Channelised Debris Flows
Bouldery front

Yu Tung Road Shek Pik

Shek Mun Kap Yi O Village Nam Chung Tsuen

Note:
- debris flows
tends to develop
a bouldery front
- large tree trunks
may be present 3

Common Structural Countermeasures for

Mitigation of Natural Terrain Hazards

Rigid debris-resisting barrier Flexible debris-resisting barrier

4
 Examples of Natural Terrain
Mitigation Measures Used in Hong Kong

Examples of Natural Terrain

Mitigation Measures Used in Hong Kong
Design of Debris-resisting Barriers
• Design should account for:
 debris runout characteristics (e.g. runout
distance, velocity, thickness)
 debris impact load & boulder impact load

 multiple phases of landslide impact including

static load
 geotechnical stability & structural response

 debris run-up height

 retention capacity

• Detailing
7

A. Design of Rigid Barriers

This covers both rigid

and flexible barriers

8
 Current Design Approach for Rigid Barriers
• Pseudo-dynamic force equilibrium analysis [Force Approach]
• Assume a multiple surge impact model
• Soil debris impact load – use hydrodynamic equation
• Impact large boulders – use simplified Hertz equation
• Superposition of soil debris and boulder loads for geotechnical stability and structural
design
• Cushion material may be prescribed by designers in front of barrier without theoretical
calculations to reduce the impulse associated with the dynamic impact
• Deflector may be provided at crest of rigid barrier

L ≥ HMax

Barrier
wall
stem

Prescriptive rockfill
gabion cushion
Deflector

Load model

Current Design Approach for Rigid Barriers

Hydro-static approach:

p max = k ⋅ ρ ⋅ g ⋅ h where pmax = maximum debris flow impact pressure

k = empirical factor = 1.0
ρ = density of debris flow
g = gravity
h = debris flow height

Hydro-dynamic approach:

where α = dynamic pressure coefficient

v = velocity of debris at impact

Current practice in Hong Kong:

1. GEO Report 270 recommended that α be taken to be = 2.5, which accounts for the
impact load induced by soil debris and any small boulders with size up to 0.5 m.

2. For boulder size >0.5 m, the boulder impact load needs to be allowed for
separately.
10
Calibration of Hydro-dynamic Model
in Flume Test

Storage container
Automatic door (30o)

Transportation zone
(20o)
Instrumentation port

Transparent sidewall

Runout zone
(Horizontal)

11

Proposed Recommendation for α value

(not yet implemented)

β
Proposal under consideration:
• Dynamic pressure coefficient (α) of 1.5 is recommended Plan View
• A partial load factor remains as 1.0
• The above is applicable to both geotechnical stability check
and structural design
• h = height of debris front; w = width of debris front

Note:
1. Boulder (of all sizes) impact is to be considered separately.
Current approach for assessing boulder impact
• Impact load of boulder (>0.5m) is highly transient and of high magnitude.
Jiangjia Ravine (field) Centrifuge Tests
Debris impact with hard Debris impact with hard
inclusion inclusion

• Current practice assumes boulder load as a pseudo-static force which is

derived from simplified Hertz equation with a reduction factor, Kc, of 0.1
where Eb = Elastic modulus of boulder
EB = Elastic modulus of barrier
= Poisson ratio of boulder F
= Poisson ratio of barrier
R = Radius of boulder
v = Velocity of boulder
m = Mass of boulder

• Implications: barriers typically require heavy foundations and tie-back in

order to achieve geotechnical stability for sliding and overturning.

Boulder Impact on Rigid Barrier

(without cushioning material)
GEO Report No. 270 (2012)
Hertz Equation further simplified:
F = 4000 Kc vb1.2 rb2 (in kN) (from Hertz Equation F = Kc n αb1.5 )
Using typical values of Poisson ratios and Young’s moduli assumed for boulder and rigid barrier:

Kc is load reduction factor (taken as 0.1)

vb is velocity of boulder

rb is radius of boulder

n is 4 rb0.5 / 3π (kb + kB)

αb = (5 mb vb2 / 4 n)0.4
mb = 4 / 3 π ρ rb 3
Eb = 50 x 109 N/m2 (i.e. 50 GPa)
EB = 25 x 109 N/m2 (i.e. 25 GPa)
μb = 0.2
μB = 0.2
ρ = 2650 kg/m3 14
New Design Approach for boulder impact on rigid barrier
- Displacement Approach
• GEO has developed a “Displacement Approach” for assessing
geotechnical stability of rigid barriers subjected to boulder impact.
• This approach considers the energy transfer during the impact process
to calculate the resultant translational and rotational movements of
the rigid barrier, taking into account the inertial resistance of barrier in
resisting the boulder impact.
∆

∆C.G. (gain in PE)

Work done by friction

Translational Rotational
KE2
∆cg = Mg 15

Partial Safety factors

The partial safety factors based on Geoguide 1 (GEO, 1993)
should be adopted in establishing the stabilisation and
destabilisation forces at the ultimate limit state for the
design scenarios.

A partial safety factor =1.0 may be applied to the calculated

debris impact load and boulder impact load.

16
 Partial Safety Factors

17

Design impact scenarios

• GEO Report No. 270, para. 7(i)

18
 Multiple phases of debris impact for
rigid and flexible barriers (GEO TGN 44)
Flowing and Deposition Sequence of Debris Loading on Barrier

h
Barrier F = α ρ v2 h

pd = α ρd v2 α = dynamic pressure
coefficient
 Debris front reaches the barrier and deposition begins (first debris surge)

ρ = density of debris flow

h
Moving debris pd = α ρd v2
(in kg/m3)
d
Debris
deposited ps= Κ ρs g d
K=1 v = debris velocity at
 Debris climbs above deposited debris (subsequent debris surges) impact (in m/s)
h
pd = α ρd v2
d
h = debris thickness (in m)
Debris deposited

ps= Κ ρs g d

 Debris piles up behind the barrier (last debris surge may fill up the barrier)
19

Design impact scenarios

20
Design impact scenarios

21

Design impact scenarios

22
 Design impact scenarios

Recommend to check
against debris overflow
and impact at the top of
barriers as a good practice

Accounting for self weight of impacting debris

in geotechnical stability check

As debris is a mass to be
retained, its self-weight mg

should be included in the

calculation of the
restoring moment for the
geotechnical stability
check for the barrier (see
GEO TGN 47).
 New Methodology for Debris Velocity Reduction
under Successive Debris Impacts (GEO TGN 44)
[Applicable to both rigid and flexible barriers]
hn 10o
1. Calculate deposition
nth phase volume of each
••
• phase
10o
h2
2nd phase
h1 1st phase

Hydrographs at
wall location
2. Read off the time
nth phase
from cumulative
2nd phase volume hydrograph
1st phase for each phase of
impact
12 Suggested
10
8
design 3. Discretise the debris
Velocity

6 envelope
velocity hydrograph
(m/s)

4
2
0

25

Design for multiple phase impact

- Assumptions on debris velocity for
subsequent impact phases
V1
14 V2
Debris velocity (m/s)

12
Vn
10
8
6
4
2
0

• Take maximum velocity at each time step

• Limit the attenuation to a Vn value of not less

than 0.7 Vmax (i.e. V1)
26
 Assumptions on Geometry of
Deposition Zones for Different Phases
hn 10o
nth phase
••
•
10o
h2
2nd phase
h1 1st phase

• h : designers may take the larger of (i) the maximum debris

thickness passing through barrier wall location from debris
mobility analysis, (ii) a nominal thickness of 0.5 m, (iii) the
estimated thickness of boulder accumulated front

• Inclination θ
 take 10o (= assumed run-up angle as recommended by
Sun & Law (2011) based on numerical analyses and field
observations)

27

Updates on Design Guidance

for Rigid Barriers
• Some updated design guidance for rigid barriers was
promulgated by GEO in TGN 47

• New guidance was given on the following aspects:

(a) Beneficial effect of self-weight of impacting debris

(b) Assumption of simultaneous impacts by several boulders

is too conservative and need not be made

(c) Use realistic topographic profile in debris mobility

assessment
28
Use realistic topographic profile of the
landslide runout path

Debris slows down on

level ground

• GEO TGN 47 recommends that the design profile of the

debris retention zone should be used in the debris mobility
analysis for barrier design purposes in order to simulate the
slowing down effect as shown above.

Other design considerations for

rigid debris-resisting barriers

• Check debris run-up

• Check retention capacity

• Proper detailing

30
 Debris run-up height
• Bernoulli’s principle:

Total head = elevation + pressure head + velocity head

• GEO Report No. 270:

Debris run-up height

• Analytical solution based on mass and
momentum conservation:
 Debris run-up height

Design against debris run-up

Design line

34
 Detailing against debris overspilling
• For example, provide deflector at barrier crest

35

Test on Deflector - Water flow with and without deflector

Test Details
Volume of Water = 1.5 m3
Flow Depth (frontal) = 80 mm
Flow velocity (frontal) = 7.5 m/s

Control Test: No Deflector (noticeable water overtopped barrier)

Deflector at
L = 0.5 x flow depth (i.e. 40 mm)

Test D1: Deflector provided (no spillage of water was observed)

A deflector (with a protruding length of 0.5 x flow depth) can effectively prevent
36
spillage of water flow.
 Numerical simulation using CFD-DEM
model (HKUST)
• Solid phase is modelled by Discrete Element Method which applies Newton's laws of
motion to every particle
• Flow of continuum fluid is described by the local averaged Navier–Stokes
equations solved by CFD approach.

Flow details
• Flow volume: 200 m3
• Flow depth: 0.5 m
• Hard inclusion: 0.4 m dia. boulders
• Barrier geometry : 5 m high x 10 m
wide
• Deflector provided at a length of 0.25m

Key observations:
• No signs of spillage of debris (both fluid part and hard inclusions) were observed.
• The flow predominately deflected backward when interacting with the deflector.
• Hard inclusions did not spill out from the model barrier and quickly settled in the fluid
mix upon run-up.

Updated Recommendation on Crest Deflector

Current Practice on Geometrical Requirement of a Deflector
• horizontal projected length shall be, at least, 1 x debris flow depth.

Proposed Recommendation
• The required horizontal projected length of the deflector shall be, at least, half of the
frontal debris flow depth.
• In selecting a suitable shape and form for a deflector (e.g. reinforced concrete or
galvanized steel), due regard should be given to various factors including construction
cost and constructability (including temporary works required), maintenance
requirement, and landscaping requirement of the barriers, etc.

38
 Retention Capacity for Rigid Barrier
Retention capacity of rigid barrier = area of the
shaded triangle x width

May assume tanγ = 1/2 to 3/4 × tan θ (i.e. gradient of the deposition area)
39

Retention Capacity for Rigid Barrier

• Total active volume of debris reaching/to be retained by the
barrier should be based on Natural Terrain Hazard Study
(NTHS).
• Designers should make reference to the calculated
cumulative debris volume entering the debris retention zone
with consideration of an appropriate debris bulking factor
(may assume 5% to 10%) in order to assess the design
retention volume.
• The retention capacity should be checked with due regard to
the dimensions of the barrier and the topographic conditions
(i.e. gradient of the deposition area).

40
Detailing of Rigid Barriers

GEO TGN 35
supplements
guidance given in
GEO Report No.
104

41

Landscape Treatments

42
 Illustrative example of
design of rigid barrier
• A rigid barrier is proposed to intercept a channelised
debris flow
• Debris density = 2,000 kg/m3
• = unit weight/g = 20 kN/m3 x (1/g)
• = 2 kN/m2 x (s2/m2)
• Width of debris run-out path = 7.5 m
• Width of barrier = 15 m
• Many small boulders (< 0.5 m) and one large boulder
(1.5 m) are found along the predicted debris run-out
path
• A DMM assessment has been carried out using realistic
topographic profile
43

Illustrative example
(using current design approach)
• Estimate the debris velocity (v) and debris
thickness (h) for the first two phases of impact
• Determine the debris impact force (F) for the
first two phases
• Determine the impact force (F) due to the
large boulder
• Determine the run-up height (hf) for the first
phase of impact

44
 Debris Mobility Model

Observation point of the

design barrier

Landslide source

45

46
At chainage 180 m

47

At chainage 180 m

48
 Answers

49

Cumulative volume after 1st debris phase = 0.5 x 4.8 m x 0.85 m x 15 m = 31 m3

Cumulative volume after 2nd debris phase = 0.5 x 9.6 m x 1.7 m x 15 m = 122 m3

0.85 m

0.85 m
10 10°
0.85 / tan10° =° 4.8 m

50
2nd debris phase

1st debris phase

8.5 10.2

51

v = 9.1 m/s for 1st debris phase

v = 8.6 m/s for 2nd debris phase

8.5 10.2

The max. debris thickness passing through the barrier

location under free-field condition (TGN44)

h = 0.85 m for all debris phases

52
 Impact forces (current approach)
First debris phase of dynamic debris impact:
F = (2.5 x 2 x 9.12) x 0.85 x 7.5 = 2,639 kN
Impact pressure height & width

Second debris phase of dynamic debris impact:

F = (2.5 x 2 x 8.62) x 0.85 x 7.5 = 2,357 kN

Boulder impact:
F = 0.1 x 4,000 x 9.11.2 x (1.5/2)2 = 3,184 kN
53

Debris run-up height

• Froude number (Fr) = v/(gh)0.5
• Fr = 9.1/(9.81 x 0.85)0.5 = 3.2
• From the group:
• hf/h = 5.2
• hf = 4.4 m
• [h = 0.85 m]

54
 B. Design of Flexible Barriers

55

Rockfall Barrier

Rockfall Barrier/boulder fence

- Designed to catch falling rock blocks or boulders
- Design guidelines widely accepted
- Design capacity based on full-scale test
- Design based on consideration of kinetic energy of impacting rock

56
Successful cases retaining landslide debris

Case Study 2 - Veltheim,

Switzerland in 2008

Case Study 3 - St-Léonard,

Switzerland in 2008
Case Study 1 - Illgraben,
Switzerland in 2006

 Impact velocities
5 m/s - 11 m/s

Case Study 4 - Pyrenees,

Spain in 2010
57

Post-Supported Flexible Rockfall Barriers

Brake
element Top rope

Side rope

Net-mesh
post

Bottom rope

58
 Flexible Debris-resisting Barriers (may
have additional intermediate ropes)
Brake element
(energy dissipation
device) Top rope

Side rope
Intermediate rope

Net-mesh

post

Bottom rope

48

Typical components

Principal net Wire ropes Anchor

Secondary mesh Post Brake elements foundation

60
 Rockfall flexible barriers
Maccaferri RMC 300A (3000kJ)
• Energy rating certified by full-
scale impact tests

Geobrugg RXE-8000 (8000kJ)

Trumer TSC-3000-ZD (3000kJ)
Geobrugg RXE-5000 (5000kJ)

61

Debris-resisting flexible barriers

• Pressure rating certified by
limited full-scale impact
tests & numerical analyses
only for selected products
• Usually require site-specific
design using Analytical
Approach
Geobrugg VX Barrier (60-160kPa) Maccaferri DF Barrier (80-200kPa)

Maccaferri
(2011)
 Different Structural Forms of Flexible Barriers
Side-anchored barriers
Post-supported barriers Side-anchored barriers (with intermediate posts)

Intermediate posts

Design of Empirical Flexible Barriers for

Mitigation of Open Hillslope Landslides (OHL)

• see GEO TGN 37

• Based on probabilistic concept with the use of
the available local information on the observed
characteristics of past open hillslope landslides
• Assumed 25% reduction in energy rating of
rockfall barrier to account for the areal nature of
debris loading, which is different to that of a
point load impact by a boulder

64
 Design practice in Hong Kong
Design of debris-
resisting flexible
barrier

Empirical design Analytical design

CD & TD
OH catchment catchments
(qualifying criteria OH catchment
in TGN37 satisfied) (qualifying criteria
NOT satisfied) Kwan & Cheung
(2012)

Coupled analysis
Energy approach Force approach (e.g. Kwan et al.,
(DN 1/2012) (DN 1/2012) 2018; Cheung et al.,
TGN 37 2018)

65

Which flexible barrier can be designed

using Empirical Approach?
NTHS: 200 m
 No evidence of existing signs of distress, CDF
continuing hazardous movement or incipient
instability;
 No newly emerged hazardous situation, new signs
of distress, etc.;
 Not susceptible to deep-seated landslide hazards. OHL TDF

Past landslide <20

m3 Past landslide of 100
m3
OHL
200 m

Squatter
Past landslide = 60 m3

200 m

66
 Framework for Use of
Prescriptive Flexible Barriers for Mitigation of OHL

Can only use if all the prescribed qualifying criteria are satisfied
– for details, see GEO TGN 37

67

Analytical design approach

• Should be adopted for mitigation of the following:
– OHL hazards where the qualifying criteria for
prescriptive barriers cannot be met
– Debris flow hazards

• Three design approaches:

– Energy approach (Kwan & Cheung, 2012; Sun & Law, 2015)
– Force approach (Kwan & Cheung, 2012; Sze et al., 2018)
– Coupled analysis (e.g. Kwan et al., 2018; Cheung et al., 2018)

68
Analytical design of Flexible Barriers

• The flexible barrier will undergo significant

deformation upon debris impact and dissipate the
impact energy.
• As a rule of thumb, the barrier should be able to
sustain its structural integrity at a maximum
deformation of not less than 10% of the barrier
height.

69

GEO DN 1/2012
Design approaches for flexible barrier
Debris-resisting capacity of flexible barriers
can be determined using either:
Energy Approach or
Force Approach

Other checking requirements:

Design retention
Overall slope stability
Serviceability (i.e. movement) Potential
Slip
Surface

70
 B1. Energy Approach
• Key design consideration:
Energy Dissipation Capacity (EDC) of barrier
> Energy loading (E) of debris flow

Calculation of “E”
What is the energy brought
by debris flow to a flexible
barrier?
* There is internal energy
dissipation as the debris
interacts with the barrier upon
impact
E < ½ mtotalvmax2
71

GEO’s work

Particulate Sun & Law carried out PFC simulations and proposed
Flow analytical solutions to calculate energy loading of
Computation debris impact
(PFC) analysis

Pile-up mechanism

Run-up mechanism

72
 Flexible barrier
• Energy Loading (Pile-up mechanism)

Major Assumptions:
- Coulomb Friction at the base (µ = tan φ)
- Debris accumulated behaves as a rigid mass
73

Flexible barrier
• Energy Loading (Run-up mechanism)

Major Assumptions:
- Inclination of the ramp (γ)
- Coulomb Friction at the base (µ = tan φ)
74
 Determine Energy Loading “E”
Calculate the flow velocity
and depth hydrograph at the
design location of flexible
barrier
(TGNs 29, 34, 38)

Calculate the upper bound

impact energy based on pile-
where
up and run-up mechanisms
(GEO Report 309)

Calculate the upper bound

energy requirements by
assuming no energy loss

Determine the design energy

requirement of the flexible
barrier i.e. “E”
75

Step 1 – Debris Mobility Assessment

Design
location of
flexible barrier Velocity and thickness
hydrograph

76
 Step 2 – Pile-up and run-up mechanisms
(GEO Report 309)
• The energy required to be absorbed by the flexible
barrier is calculated based on the limiting deposition
processes of landslide debris

Note: The above simplified mechanisms cannot be adopted where the slope angle ≥
interface friction angle at the base of the debris plug.
In this case, designer may assume that the energy loading is given by the kinetic energy
of the entire debris mass. 77

Step 2 – Energy loading assuming pile-up

Check whether the friction of deposits

could take up all the impact load

Calculate the upper bound energy

requirement based on the assumed
pile-up mechanism

 The above is not applicable when

the slope inclination behind the
barrier > basal friction angle
Step 2 – Energy loading assuming run-up

)2 2
Calculate the upper bound energy
requirement based on the assumed
run-up mechanism

 The formula for Ep and Er are

based on simplified analytical
model.

 Actual energy required to be

absorbed by the flexible barrier for
retention of the landslide debris is
likely to be less.

79

Step 3 – Calculate kinetic energy of the entire

landslide debris mass (Ek)

There are two ways to calculate the

total kinetic energy of landslide debris

(a) All kinetic energy of landslide debris (b) All kinetic energy of landslide debris that
when the debris front reaches the design passes through the design location of
location of flexible barrier flexible barrier

80
 Step 3 – Upper bound energy requirement
• Instead of taking the upper bound as E = ½ mtotalvmax2
Ek1 and Ek2 are defined:

v8 v7
v6 v5
v4 v3 v2 v1

Ek1 Ek2

81

Step 4 –Calculate the Energy Loading (E)

The minimum of the total energy

(Ek) and the energy calculated
according to the limiting
mechanisms (i.e. maximum of Ep
and Er)

Why are we taking max (Ep, Er)? This is because the debris flow can
behave either in the form of pile-up (Ep) or run-up (Er) which depends on
its rheology and composition (i.e. whether it behaves more as a viscous
flow or as a dry granular flow). So, it is on safe side to assume the
maximum value of Ep or Er.

The main reason of using min (Ek, max (Ep, Er)) is because Ek is a free-
field kinetic energy which does not consider the deposition mechanism
behind the rigid barrier. Therefore, it is more realistic and economical to
assume min (Ek, max (Ep, Er)).
82
 Worked Example

The width of the runout trail is assumed to be 6 m. The channel cross section is rectangular in
shape. The landslide volume is assumed to be 200 m3. The flexible barrier is located at CH 152 m.

83

Step 1 - Debris mobility assessment

Voellmy rheology is adopted in this worked example. The apparent friction angle and turbulent coefficient
are adopted as 8o and 1000 m/s2 respectively.

Using debris mobility model to determine the velocity hydrograph and the debris thickness hydrograph at
the location of the flexible barrier (i.e. CH 152 m) respectively as shown in the following figures.

84
 Step 2a - Calculation of energy loading assuming pile-up
mechanism (Ep)

In this worked example, the maximum velocity and the maximum debris thickness are 4 m/s and 0.25 m
respectively. The calculated discharge rate is 6 m3/s (= 4 m/s (velocity) × 0.25 m (thickness) × 6 m (width)).

The duration of impact in the pile-up mechanism (Ts) cannot be longer than the ratio of total active volume of
landslide debris (V) to the discharge rate of landslide debris (Qo). Therefore, Ts should be checked in
accordance with Equation 6.11a of TN 1/2012 as follows:

In this worked example, Ts = 3.7 s which is less than 33.4 s (= V/Qo = 200/6). Therefore, the pile-up
mechanism is considered theoretically permissible.

85

Step 2a - Calculation of energy loading assuming pile-up

mechanism (Ep)

After checking the impact duration, Equation 6.13b of TN 1/2012 is used to calculate the energy
loading based on the pile-up mechanism (Ep). The Equation is presented below:

ρ = bulk density of the landslide debris deposit = 2200 kg/m3 (see Section B.3.2 of DN1/2012)

Qo = discharge rate = 6 m3/s (calculated by debris mobility analysis)

With the input parameters above, Ep = 197 kJ is obtained.

86
 Step 2b - Calculation of energy loading assuming run-up
mechanism (Er)

Equation 6.19 of TN 1/2012 is used to calculate the energy loading based on the run-up mechanism
(Er). The Equation is presented below:

where

h0 = debris depth = 0.25 m (calculated by debris mobility analysis)

γ = inclination of ramp formed by debris accumulated behind flexible barrier = 10o

(see Section B.3.2 of DN 1/2012)

With the input parameters above, Er = 13 kJ is obtained.

87

Step 3 - Calculation of total kinetic energy of landslide debris (Ek)

The total kinetic energy of the landslide debris (Ek) should be taken as the maximum of

(i) the kinetic energy of whole active landslide debris when the debris front reaches the barrier
location (Ek1); and
(ii) the theoretical kinetic energy of landslide debris that travelled beyond the barrier location (Ek2).

Ek1 can be obtained from the output of the debris mobility assessment. For example, 2d-DMM calculates
kinetic energy of landslide debris at each time step. Users first determine the time at which the debris
front reaches the barrier location from either velocity hydrograph or debris thickness hydrograph. The
kinetic energy of whole active landslide debris at that time can be extracted from 2d-DMM output as
illustrated in the following figure:

88
 Step 3 - Calculation of total kinetic energy of landslide debris (Ek)

Ek2 can be obtained by coupling the velocity hydrograph and debris thickness hydrograph, denoted by
v(t) and h(t) in following equation:

where
v(t) = debris velocity at barrier location at time t (m/s)
A(h(t)) = cross sectional area which is a function of debris depth h(t); where rectangular cross
section is assumed A = h(t) * w
h(t) = debris thickness at barrier location at time t (m)
w = debris width at barrier location, assumed to be constant
ρ = debris density (in kg/m3)
∆t = observation time interval (s)

In this worked example, Ek1 and Ek2 are 913 kJ and 1,279 kJ respectively. Ek is therefore taken as
1,279 kJ.

89

Step 4 - Determining energy loading for design of flexible debris-

resisting barriers (E)

The energy loading (E) is determined as follows (Equation B.4 of

DN 1/2012):

E = min [Ek, max (Ep, Er)]

Thus in our present example, the following is applicable:

E = min [1279, max (197, 13)] = 197 kJ

90
 Determine Energy Dissipation Capacity
“EDC”
• Key design consideration:
Energy Dissipation Capacity (EDC) of barrier
> Energy loading (E) of debris flow

EDC can be established based on

structural analysis (Kwan & Cheung, 2012)
field testing (TGN 37)

e.g. EDC of a certified proprietary rockfall barrier rated at 2000

kJ = 2000 * 0.75 = 1500 kJ.
91

B2. Force Approach for Flexible Barriers

Barrier Design Loading Debris-resisting Overall
Configuration Scenarios Capacity Stability

• Structural • Sequential • Modelling to • Ensure slope

form vs site applications account for stability up
setting of dynamic & non-linear load- to required
• Structural static loadings deformation safety
capacity & behavior of standard
retention barrier
capacity

92
 Loading on Flexible Barrier
Dynamic impact pressure
• α = 2.0 [current approach] Static earth pressure
• This is deemed to cover a boulder size up to
2 m in diameter
Pd = αρdv2 velocity Ps = Kdρsg
2.0 1.0 deposited 2,000 kg/m3
2,000 kg/m3 depth

• Overflow drag

• Boulders

93

Multiple phases of debris impact

Flowing and Deposition Sequence of Loading on Barrier
Debris
h
Barrier

Variation in debris pd = α ρd v2
velocity is
considered in (i) (Debris front reaches the barrier and deposition begins
(first debris surge)
multiple phases of
debris impacts as h
recommended in Moving debris
pd = α ρd v2
d
TGN 44 Debris
deposited ps= Κ ρs g d
(ii) (Debris climbs above deposited debris
(subsequent debris surge)
h
pd = α ρd v2
d

Debris deposited

ps= Κ ρs g d

(iii) (Debris piles up behind the barrier (last

debris surge filling up the barrier) 94
 Variation in debris velocity considered in
multiple phases of debris impact (TGN 44)
V1
14 V2

Debris velocity (m/s)

12
Vn
10
8
6
4
2
0

• Take maximum velocity at each time step

• Limit the attenuation to a Vn value of not less
than 0.7 Vmax (i.e. V1)

95

Structural Capacity of Flexible Barrier

• Appropriate structural analysis that takes account

of load-deformation characteristics of all major
structural components should be carried out.
• The barrier structure including the foundations
should be designed to relevant structural codes
and standards.
• The structural design should be carried out by a
competent structural engineer with design
experience in steel structures.
• The structural design should be independently
checked and certified by a RSE or RPE(Structural).

96
 Numerical Computer Programs
for Structural Analysis
• Models that take into account the non-linear material behavior of flexible
barriers under sequential loadings:
• e.g. NIDA-MNN (Ref. Prof S L Chan)

• e.g. LS-DYNA

97

Retention Capacity for Flexible Barrier

For the purposes of retention volume calculation, Geobrugg

(2012) recommended the following:

(a) the volume of bulged-out portion of the flexible barrier

should be ignored
(b) the deposition profile behind the flexible barrier should
be assumed to be horizontal (i.e. storage angle γ = 0)

98
 B3. Coupled analysis
of debris-flexible barrier interaction
Advanced debris mobility modelling Explicit structural modelling of flexible barrier

Debris flow cases in Hong Kong Rockfall test (Volkwein, 2004)

Coupled analysis
Finite element program LS-DYNA
- couples debris mobility modelling together with
explicit structural modelling of flexible barriers
- adopts Arbitrary Lagrangian-Eulerian (ALE)
mesh-based continuum model
- provides insight into dynamic debris-barrier
interaction through the coupled analysis
99

Soil-Structure Interaction Analysis

• Force approach is essentially a decoupled analysis
Debris mobility modelling Structural response of barrier

+
• Coupled analysis explicitly consider the debris-barrier
interaction

100
 Coupled Numerical Analyses using LS-DYNA
Key Insights:
 Parametric study: energy transfer from debris to flexible barrier
is generally less than 35% of total debris kinetic energy
 Kinetic energy of debris appears to be dissipated significantly
through internal shearing and re-direction of flow momentum
upon debris impact
Impact Scenario
Debris volume = 85 m3 LS-DYNA results Centrifuge test results
Impact velocity = 8.6 m/s
Slope gradient = 15o

End of impact

101

Detailing of Flexible Barrier

• GEO TGN 48 stipulates a number of
recommendations on the detailing of flexible debris-
resisting barriers. It covers:
– barrier alignment
– anchor locations and detailing
– detailing of barriers traversing a stream course from a
drainage perspective

102
 Multiple Barriers as Mitigation Measures
Single barrier Multiple barriers
• at outlet of drainage line • a series of smaller barriers along a
• close to toe of hillside drainage line
• to retain a certain amount of debris
• terminal barrier provided at
drainage line outlet
• more robust and more suited for
large scale landslide hazards

103

New Design Framework for Multiple Barriers

Simplified analytical solution

104
 Single Barrier vs Multiple Barriers
Aspect Single barrier at drainage outlet Multiple barriers along drainage line and
terminal barrier at drainage outlet
Robustness of  Hazard mitigation relies on a single barrier to  More defence lines; seems better
hazard resist a large volume of landslide debris,  Impedance provided by the intermediate barriers retards
mitigation consequence could be serious if the barrier mobility of debris and reduces entrainment
is overwhelmed.

Aesthetic and  Higher barrier, possibly closer to existing  Barriers are smaller, but larger combined physical
environmental developments, would be needed, hence footprint
impact higher visual impact  Landscaping measures needed

Land  May be sizable in order to create a  Land required by each of the intermediate barriers and
consumption sufficiently large space for construction of terminal barrier may not be large.
debris retention zone  Stand a higher chance of encroaching into Country Parks
or other conservation areas.

Future  Routine maintenance needed  Routine maintenance for intermediate barriers is not
maintenance required
 Alert system to detect landslide impact on barriers is
required
 Condition survey every 5 to 10 years or immediately after
landslide event
 Routine maintenance required for terminal barriers

Construction  Comparatively easier  May be strenuous

 Construction at several locations needed

Cost  Cost may be lower  More barriers are needed and construction of
intermediate barriers may be fraught with the constraint
of accessibility
105

Examples of Other Mitigation Measures:

Drainage Tunnels at Po Shan
for Deep-seated Landslide Hazard

Sub-vertical
drains

Drainage
tunnels
3-D images of the innovative scheme Sub-vertical drains
• An innovative regional
groundwater regulation system
• The pioneering use of a drainage
tunnel cum subvertical drain pipes
system to improve hillside stability
against large-scale landslide
106
 Nature-based Solutions (Bioengineering Measures)
to repair natural terrain landslide scars
Upper Keung Shan
Brushlayer

Fat Wah
Temple

Shotcreted
Scar

Live Fascines/Pole Drains Live Cribwall Hedgelayer

107

Thank you

108
 Back-up slides

109

Displacement Approach - Sliding

Sliding movement of barrier subject due to boulder impact
can be predicted by the following:

- Conservation of Momentum

- KE2 can be obtained from the

following:

KE

KE0 = Kinetic energy of impactor

λ = Mass ratio of barrier and
impactor
COR = Coefficient of restitution

KE2 is sensitive to mass

ratio, λ.
Larger λ, lower KE2
110
 Displacement Approach -
Overturning
Overturning of barrier subject to boulder impact can be
predicted by the following:

KE
∆cg = Mg2

111

(a) Verification of Displacement approach (Overturning)

conducted at University of Melbourne
• Model barrier: 0.8 m x 0.2 m thick L-shaped
concrete block
• Impactor: 5 kg steel iron ball
(drop height: 0.5 m to 1.5 m)
• Measured horizontal displacement (a measure of
rise of center of gravity or overturning) was
compared to that predicted by Displacement
Approach.
Laser
sensor

112
 (b) Large-scale Test (at Kadoorie Centre) to further
verify Displacement Approach

High-speed camera
(from side)
θrotat
Laser e High-speed camera
Scanner

Laser
Scanner
20°

∆slide
Side view

Compare sliding and rotation movements of barrier predicted by

Displacement Approach and those observed in the field flume test.

(c) Further validation of Displacement Approach by

advanced numerical analysis (computer program LS Dyna)

Boulder Diameter = 1 m;
Impact Velocity = 10 m/s
Concrete Grade C30
T32 -200 (i.e. 0.7% reinforcement content)

Numerical Prediction based on

output Displacement Approach’s
equations
Sliding 0.2 1.9
displacement
(mm)
Rotation (deg) 0.0213 0.0683 Other energy loss observed in
numerical analyses, e.g. concrete
fracturing at impact location, and wall
Prediction by Displacement Approach is conservative! base due to bending
114
 Barrier/boulder
mass ratio > 30

0.8
sliding is <20 mm m
10 m
Mass of barrier (230 tonne,
10 m long, 5m high)
Impact velocity: 10 m/s

5m

0.8
m

Barrier/boulder
mass ratio > 30
Sensitivity check of barrier/boulder mass
ratio of 30 for barriers with other
geometries shows sliding and rotation
Rotation is <1 deg.
are not significant.

Non-linearity of barrier’s displacement vs boulder mass

Proposed Recommendation – Checking geotechnical

stability of barrier due to boulder impact
 Proposed new approach - “Enhanced Flexural Stiffness Method
(EFSM)” to assess effect of boulder impact on structural response

Key principle:
• Conservation
of momentum

F = boulder impact force

COR = Coefficient of Restitution (typically
0.2 to 0.3)
λ = ratio between participating mass of
barrier and mass of boulder
v = boulder impact velocity
m = mass of boulder
k = flexural stiffness of barrier
Physical meaning of “boulder impact load” - an equivalent static force that would give rise to the
∆ = deflection at tip a deflection same as that due to dynamic impact for a given impact scenario.

Participating Mass (m*) in EFSM

• Experiments shows boulder impact at tip Shape function for tip load of a
of wall will lead to oscillation of wall in cantilever beam
Simple Harmonic Motion.
• There is interchange of KE and strain
energy of wall stem under SHM
oscillation.
• Strain energy:
• KE
•

m = mass of boulder Put z = l, then

l = height of barrier
 Verifications of EFSM
Displacement Boulder Impact
at tip of beam Force (N)
due to bending (equivalent
(mm) static force)
Experimental 16.9 1014
Result
EFSM 16.4 984
FSM (Hungr 40 2426
et al, 1984)
Small scale impact tests

Calibrated numerical Models

0.8 m

10 m

5.2 m

Boulder dia. = 0.5 to 1.5 m

Impact velocity = 10 m/s

119

Test Results
Observed Peak Deflection
Observed Observed at Barrier Crest
Test Peak Deflection Prediction using Strain level
No. at Barrier Crest EFSM (mm) of concrete
(mm) and Steel
A1 4.0 5.5
A2 5.7 8.0
A3 8.9 10.2
A4 11.9 11.9
A5 5.9 6.1 Within
A6 9.5 9.4 Elastic
A7 13.0 13.0 range
B1 5.2 5.2
B2 8.2 8.1
B3 11.5 11.6
B4 14.4 13.3 Deflection time-history
of barrier (Test No. A6)
• Predicted deflections (which is a measure of flexural response) based
on EFSM match very well with the experimental results.
(Note: In the above tests, the barrier is caused to bend predominately, but not to
slide, rotate, fracture locally. In reality, all these would co-exist, and the extent of
bending would be smaller in real-life impact than that predicted by EFSM.) 120
 Boulder Impact with cushion
materials

Tip Deflection of Reduction in Tip Deflection

Tip Deflection of Barrier
Test Barrier (which is a measure of
No. without cushion with cushion bending) due to cushioning
(mm) effect (%)
(mm)

C1 11.9 1.8 - 3.9 67 - 85%

C2 13.0 1.3 - 3.2 75 - 90 %

C3 11.9 2.5 - 3.5 71 - 79 %

C4 13.0 1.3 - 2.1 84 - 90 %

Key Observations:
• In the tests, the barrier’s flexural response was significantly reduced
by 67 to 90%, due to the presence of a layer of 0.5 m thick rockfill
cushion layer.
121

2 Degree of Freedom (DOF) Mass-Spring-Damper Model for Cushion Material

- Analysis of contact response of
cushion and global flexural response
of barrier in an integrated manner Analytically derived (Sun et al., 2015,
Perera et al., 2016) and depending on
the peak contact force, mass ratio,
impact energy, indentation rate, etc.,
validated with large-scale impact
tests on rockfill gabion

cushion

RC wall

Governing equations (time step: 0.01 ms << impact duration)

Hunt & Crossley model
(for viscoelastic materials)

Force Equilibrium

Contact Flexural Response

Inertia Force
force of barrier
(participating mass of
cushion and barrier)
Parametric Study on 2DOF
Effect of Impact Velocity
Boulder Boulder Impact Force for Boulder Impact Force for Reduction in
Impact flexural response for flexural response for flexural response
Velocity barrier without cushion barrier with cushion (kN)
(m/s) (kN)
1m thick
(i.e. EFSM) (i.e. 2 DOF Model)
6 523 280 46%
8 697 384 45%
10 871 490 44%
12 1045 597 43%

Effect of Boulder Size

Boulder Boulder Impact Force for Boulder Impact Reduction

Diameter (m) flexural response for barrier Force for flexural in flexural
without cushion (kN) response for barrier response
with cushion (kN)

(i.e. EFSM) (i.e. 2 DOF Model)

0.5 114 72 37%
Boulder diameter from 0.5 m to 1.5 m
1.0 871 490 44%
Boulder velocity = 10 m/s 1.5 2628 1449 45%

Use of rockfill gabion to reduce

flexural response due to boulder impact
Current Practice
Cushion materials are used in a prescriptive manner (i.e.
no calculations are carried out)

Findings
• If a minimum 500 mm thick rockfill gabion is provided,
flexural response predicted using EFSM and 2DOF mass
spring damper model can be reduced by 35%

124
 125

Modelling of debris flow

 Continuum materials: Arbitrary Lagrangian-Eulerian analysis
 Shear strength: internal friction angle
 Interface resistance: basal friction angle
 Initialization:

Debris Injection Dam Break

Debris
Injection Source Source
Release
Mechanism
Flume

Debris
Explicitly Modelling of flexible barrier
Sliding contacts
Explicit modelling of elements

Sliding nodes

Brake elements

Debris-barrier coupling

Validation against full-scale field tests

(Kwan et al., 2018; Cheung et al., 2018)
• Veltheim, Switzerland • LS-DYNA model
(28 m3, 10 m/s) Energy-dissipating device
Post
Upslope Top cable
cable
3.5 m

Net
15 m Bottom cable

40 m

8m
Results Debris mobility
5 2

Flow thickness (m)

4 1.5
Arrival time (s)

3
LS- 1
2 DYNA
Field Flow thickness 0.5
1
0 0
5 10 15 20 25 30 35
Chainage (m)
Barrier response
Force / Deformation Field LS-DYNA
Peak bottom cable force 71 kN 79 kN
Peak top cable force 39 kN 63 kN
Peak upslope cable force 42 kN 46 kN
Maximum bulging of net 1.8 m 2.3 m

Variations amongst suppliers

Brake Elements
Principal Net Energy dissipation device)

Geobrug
g

Geobrugg Trumer (2013)

(2009)

Maccaferri

Isofer
(2012)
Others such as cable and brake element arrangement, post Trumer
and anchor head connection details, etc.
 Background
• Flexible barriers are commonly used in mitigating risk
arising from rockfall (point load) in the past.
• Flexible barriers were occasionally hit by debris (area
load) and they were found to be able to stop and
retain a certain amount of landslide debris.
• Flexible barriers are attractive natural terrain landslide
risk mitigation measures because they are less
intrusive as compared with rigid barriers.

131

Other instrumented case studies

Rüdlingen, Switzerland in 2010

reported by Springman
USGS Flume Tests
Volume: 150 m3
Volume of debris: 10 m3
Debris impact velocity: 5 to 9 m/s

132
 General zoning for detailing

• 3 zones can be broadly distinguished as shown

above
• Key elements to be considered in the detailing
of rigid barriers in these 3 zones are
documented in TGN 35
133

Discharge zone
• Drainage provision and erosion protection
- extreme situation (overflow):

discharge weir at wall crest

erosion in front of the barrier

134
 Debris retention zone
• Access means to facilitate debris clearance & routine maintenance

• Other prescriptive provisions

- baffles and cushioning material

135

136
 Debris impact scenarios
Flowing and Deposition Sequence of Debris Loading on Barrier
v1 > v2 > v3
h
Barrier

pd = α ρd v2
v1
 Debris front reaches the barrier and deposition begins (first debris surge)

h vMoving
2 debris
pd = α ρd v2
d
Debris
deposited ps= Κ ρs g d

 Debris climbs above deposited debris (subsequent debris surges)

h v3 pd = α ρd v2
d

Debris deposited

ps= Κ ρs g d

 Debris piles up behind the barrier (last debris surge filling up the barrier)
TGN44
137

Examples of Energy-Rated Rockfall Barriers

Geobrugg Maccaferri Trumer

1000 kJ

3000 kJ

5000 kJ

8000 kJ NA

- Post-supported barriers in rectangular panels.

- All are designed with high ‘flexibility’, some comprising top and bottom lateral
cables without any intermediate cables.
 Debris mobility analyses
• Use of continuum models:
Open Hillside Failures Debris Flows in Channelised Debris
Topographic Flows
Depressions

Friction model Voellmy model Voellmy model

TGN 34 TGN 38 TGN 29

Vol. <= 500 m3, φa=25° φa=18° φa=11°

Vol. > 500 m3, φa=20° ζ=1000 m/s2 ζ=500 m/s2
(for catchments without
adverse site setting)
φa=8°
ζ=500 m/s2
(for catchments with
adverse site setting)
139