User Manual For

Computer Program
“2d-DMM” - Two-dimensional
Debris Mobility Model
(Version 2.0)

June 2015
 2

Terms and Conditions

2d-DMM (“the computer program” hereafter) is developed by the Geotechnical

Engineering Office (GEO) of the Civil Engineering and Development Department, the
Government of the Hong Kong Special Administrative Region. Users should read the
following terms and conditions before using the computer program.

1. The computer program is an “internal” program developed by the GEO for in-house
geotechnical designs or studies in accordance with GEO Circular No. 35. For the use of
the computer program by other departments/consultants/contractors for geotechnical
designs or studies in public works projects, guidelines laid down in Section 3.3.6 of
Chapter 4 in the Project Administration Handbook for Civil Engineering Works shall be
followed. Users assume responsibility for the selection of the program to achieve their
intended purpose, the use of and the results obtained from the computer program.

2. We shall not be responsible for installing the computer program and give no warranty that
the digital data/computer program can be processed on users’ computer system.

3. The computer program is prepared in accordance with the practice and general
information available at the time of preparation. The applicability of the computer
program may therefore change subject to the subsequent technical development. Users
assume responsibility for equipping themselves with adequate up-to-date technical
knowledge to determine the limitations of the computer program and the suitability of the
program for their intended purpose.

4. We remain the owner of the computer program at all times. Users shall not copy or
otherwise infringe upon any intellectual property rights of us, whether in whole or in part.
Unless agreed by us, the use of the computer program for projects not under the
management of us is not allowed. Prior written consent must be obtained from us for
any distribution or publication of reports or technical papers which use the computer
program in the work. In the event that consent is given, acknowledgement to us shall be
made in the reports or technical papers. Distribution or copying of the computer
program to any third party is not allowed.

5. We give no warranty that the computer program is error-free. We shall, under no

circumstances whatsoever, be responsible for or held responsible for any losses or claims
for direct or indirect damages including loss of time, money or goodwill whatsoever
which may arise out of or in consequence of relying on the use or the results of the
computer program.
 3

Contents

Page
No.

Title Page 1

Terms and Conditions 2

Contents 3

List of Figures 5

1 Introduction 6
1.1 Applications 6
1.2 Features 7
1.3 Key Assumptions 9

2 Program Structure 10
2.1 Installation of the Program 10
2.2 Program Layout 10
2.3 Opening and Saving Data Files 10
2.4 Main Menu 11

3 Data Input 14
3.1 Data Management 14
3.2 Project Title 14
3.3 Landslide Debris 14
3.4 Interface Resistance 15
3.5 Entrainment 17
3.6 Analysis Parameters 17
3.7 Runout Path Properties 18

4 Methods of Solution 19
4.1 General 19
4.2 Principles of Oldrich Hungr’s DAN Model Adopted in 19
2d-DMM
4.3 Modifications to Oldrich Hungr’s DAN Model Incorporated 21
in 2d-DMM
4.3.1 Trapezoidal Cross Section of Flow Path 21
 4

Page
No.

4.4 Model Evaluation by Back Analysis 22

4.4.1 Back Analysis of Case History for Comparison 23
with the Accepted GEO-DAN Program

5 Output 23
5.1 Elevation view 23
5.2 Textual Output 25
5.3 Data Plots 25
5.3.1 Profile Plots 26
5.3.2 Hydrograph Plots 27
5.3.3 Block Plots 27

6 Modelling Guidelines 29
6.1 Runout Path Properties 29
6.2 Interface Properties 30
6.3 Analysis Parameters 30
6.4 Nodal Velocity 30

7 References 31
 5

List of Figures

Figure Page
No. No.

1.1 Illustration of the Concept of Modelling a Flow Slide 6

(Hungr, 1995)

2.1 Home Page of 2d-DMM (Version 2.0) 10

2.2 File Menu of 2d-DMM (Version 2.0) 12

2.3 Settings Page of 2d-DMM (Version 2.0) 13

2.4 Output Page of 2d-DMM (Version 2.0) 13

3.1 Runout Path Tab on the Settings Page 19

4.1 Lagrangian Mesh in Curvilinear Co-ordinates 20

4.2 External Forces Acting on a Boundary Block 20

4.3 Rectangular Cross Section of Debris Flow Path Assumed 21

in Oldrich Hungr’s DAN Model

4.4 Trapezoidal Cross Section of Debris Flow Path Adopted in 22

the 2d-DMM

5.1 Scroll Bar on the Output Page 24

5.2 Show Point Values Function on the Output Page 25

5.3 Zoom-in Function 25

5.4 Velocity and Thickness Profiles along Chainage 26

5.5 (a) Opening a Separate Page to Enter the Location of an 28

Observation Point, and (b) Showing the Velocity,
Thickness, Cumulative Volume and Cumulative Kinetic
Energy Hydrograph Plots

5.6 Block Plots Function 29

 6

1 Introduction
1.1 Applications

2d-DMM (“2-dimensional Debris Mobility Model”) is a computer program developed

based on Oldrich Hungr’s Dynamic Analysis Numerical (DAN) model (Hungr, 1995; Hungr
& Evans, 1997) for the assessment of landslide debris mobility along a pre-defined pseudo
3-dimensional flow path. 2d-DMM incorporates consideration for a trapezoidal cross
section to define the cross sections of the landslide debris flow path (Kwan & Sun, 2006).

2d-DMM follows the same principle and procedure of DAN and simulates the motion
of landslide debris as a continuum after detachment using a finite difference solution of the
governing dynamic equations along its flow path with shallow-water assumptions in a
Lagrangian framework. The solution is obtained in time steps for a block assembly of
elements, representing the landslide debris as a continuum. The concept of modelling a flow
slide is illustrated in Figure 1.1. A homogeneous “apparent fluid” replaces the slide mass,
and its properties must be chosen in order to produce a similar external shape and mean
velocity of the slide mass. The velocity profile shown in Figure 1.1 is conceptual only and
would depend on the selected rheology for the debris. Two rheological models (i.e. the
friction and Voellmy models) are available in the program to simulate different landslide
movement behaviours.

Figure 1.1 Illustration of the Concept of Modelling a Flow Slide (Hungr, 1995)
 7

The key outputs of an analysis are the debris travel distance, velocity and debris
thickness at different times for a given landslide event, provided that the cross section of the
flow path is defined. These results can be used to identify the influence zone of a landslide
hazard, and to assist quantification of landslide risk and design of debris barriers.

2d-DMM has two versions that are commonly used, namely the C# Version (Version
2.0) and the Spreadsheet Versions (Versions 1.1 and 1.2). The C# Version of 2d-DMM is
developed using the programming language C#, which is an improved version of C++. The
Spreadsheet Version of 2d-DMM perform analysis using Microsoft EXCEL software.

The algorithm of 2d-DMM (Version 2.0) is updated from the previously accepted
program 2d-DMM (Version 1.2), which was developed by modifying the first accepted
program 2d-DMM (Version 1.1).

This user manual covers 2d-DMM (Version 2.0) only.

1.2 Features

The main features of 2d-DMM (Version 2.0) program are summarized below:

(a) The landslide mass is represented by a number of blocks

connected to each other, free to deform and retaining a fixed
volume of material in their descent down a curvilinear path.

(b) The coordinates of the points that specify the flow path are
entered directly into the program. In order to ensure
smoothness of the flow profile, the points are connected
using cubic spline functions, which are generated
automatically by the program. A straight line at the lower
end of the flow path is generated automatically as a
horizontal platform.

(c) The cross section of the flow path can be assumed to be a

rectangle or a trapezoid. The latter is defined by its base
width together with inclination angles at two sides. The
arrays of x coordinates and the corresponding base widths, x
coordinates and the corresponding left side angles, as well
as x coordinates and the corresponding right side angles are
entered directly into the program. In order to ensure
smoothness of the width profile, the values of the base width
and the two side angles along the flow path are connected
using cubic spline functions, which are generated
automatically by the program.

(d) Stepwise dynamic calculation is performed in a moving

Lagrangian framework to determine the motion of the
assembly of block elements, representing the landslide
 8

debris. The time interval of each time step and the overall
duration of the event are specified by users.

(e) Two rheological models for the basal flow resistance,

known as the friction and Voellmy models, are available for
the simulation of different types of landslide debris. The
program allows the models to be specified for two different
segments of the runout path. Each segment may have its
own rheological model and parameters.

(f) The change of landslide volume along the flow path due to
entrainment is simulated by entering the “entrainment rate”,
which is defined as the volume entrained per unit second
and similar to that given in Hungr & Evans (1997). The
program allows users to specify two separate entrainment
zones along the flow path, each with its own entrainment
rate.

Major modifications made for developing different versions of the computer program
are presented below.

Major modifications to 2d-DMM (Version 1.1) for development of 2d-DMM (Version

1.2):

(a) The program allows users to specify the location of the

landslide source;

(b) The program allows users to specify the initial velocity of

the landslide source; and

(c) Hydrographs of debris velocity and thickness are provided.

Major modifications to 2d-DMM (Version 1.2) for development of 2d-DMM (Version

2.0):

(a) Deposition calculation, the sliding-consolidation model, the

plastic rheological model are excluded in 2d-DMM (Version
2.0);

(b) An interactive user interface is developed to facilitate input

of data, visualization of key results and post-processing of
output data for design and analysis;

(c) In contrast to 2d-DMM (Version 1.2) where the number of

boundary blocks is fixed at 11, users are allowed to specify
the number of boundary blocks that represent the landslide
debris in 2d-DMM (Version 2.0). An increase in the
number of boundary blocks would increase the accuracy of
 9

the computed solution;

(d) Users are allowed to input the channel geometry, including

the topography, width and side angles, using x coordinates
and the corresponding topography, x coordinates and the
corresponding width, x coordinates and the corresponding
left side angles, as well as x coordinates and the
corresponding right side angles, directly instead of using 6th
order polynomial curves as in 2d-DMM (Version 1.2); and

(e) Users are able to visualize the calculation results, e.g. debris
velocity and thickness, in each time step easily on screen or
on print-outs using the newly developed plot and export
functions.

1.3 Key Assumptions

The formulation of 2d-DMM is developed based on the assumption that debris flows
are composed of continuum materials and the materials involved are homogenous. In some
circumstances, debris dynamics may be governed by the mechanics of discrete elements, for
example, rock blocks at the edge of debris mass. In general, the debris mobility analysis
undertaken by 2d-DMM cannot capture every single detail of the debris flow dynamics which
occurs during the debris transportation process.

A list of other key assumptions, which are not exhaustive, is presented below:

(a) The rheological model adopted in 2d-DMM determines the

friction or the resistance experienced by debris flows.
Other internal processes, for example, energy dissipation
due to internal mixing, are not explicitly considered by the
model.

(b) The program is not able to predict lateral spreading of

landslide debris. Users need to pre-determine the
geometry of the cross section of debris flows for analysis.

(c) 2d-DMM (Version 2.0) adopts the Saint-Venant equation to

simplify the calculation of motion of landslide debris. The
assumptions of shallow flow and smooth flow path, relative
to the flow thickness, are implicit in modelling debris
mobility using 2d-DMM. Therefore, 2d-DMM (Version
2.0) cannot be used to carry out simulation which involves
rapid rate of change of momentum of landslide debris due to
abrupt slope changes. In particular, the analysis of
impedance of baffle arrays and short rigid barriers on
landslide debris cannot be performed using 2d-DMM
(Version 2.0) explicitly.
 10

2 Program Structure
2.1 Installation of the Program

2d-DMM (Version 2.0) is written in an .exe file and runs on Microsoft Windows
with .NET framework. Windows 7 is equipped with .NET framework by default. The
program is started by double-clicking the 2d-DMM.exe file. Users may create a shortcut to
2d-DMM (Version 2.0) using Windows commands. On starting, 2d-DMM (Version 2.0)
presents the Home page (Figure 2.1) and users shall start a new analysis by clicking File→
Start a new project, or open a saved analysis by clicking File→Open an existing project.

Figure 2.1 Home Page of 2d-DMM (Version 2.0)

2.2 Program Layout

2d-DMM (Version 2.0) has three main functions: data input, analysis, and data output.
The data input component allows users to define the problem geometry, material properties,
and analysis options. Once the problem is defined, users can run an analysis. Data is
collected during the analysis and can be displayed in various ways after the simulation run is
complete.

2.3 Opening and Saving Data Files

Problems created in 2d-DMM (Version 2.0) can be saved under the file extension .sds.
An ASCII character file is saved containing all the input data, including problem geometry,
material properties, and material locations.

To save a .sds file, choose Settings page→Save settings menu selection in the main
menu. To open a .sds file, choose File→Open an existing project menu selection.
 11

2.4 Main Menu

The program is controlled through the Home page (Figure 2.1). The File, the Settings
page and the Output page allows users to create/open a project, input data, run an analysis and
view the output data during and after analysis respectively. The following is a list of the
functions available in the Home page.

(a) File (Figure 2.2):

(i) Start a new project: create a new project folder with a

blank input .sds file;

(ii) Open an existing project: open a previously saved

project. The project name is identical to the name of
the project folder; and

(iii) View output files in file explorer: browse the files in the
project folder using the default file explorer.

(b) Settings page (Figure 2.3): open and display the Settings
page. The Settings page is active once a new project is
created.

(i) Save settings: save the input data to a .sds file. All
data needs to be entered to the Settings page before
clicking this menu item. Otherwise, warning
message(s) of the missing input data will display.

(ii) Load settings file: load the input data from a previously
created .sds file.

(c) Output page (Figure 2.4): open and display the Output page.
The Output page is only available when a new project is
created.

(i) Start: run a simulation.

(ii) Break: stop a simulation and clear the output files.

This menu item is active only during a simulation run.

(iii) Profile plots: shows the velocity and thickness of a total

of 10 user selected boundary blocks versus chainage
(please refer to Section 5.3.1 for further details).

(iv) Hydrograph plots: shows the time history of velocity,

thickness and cumulative kinetic energy of the landslide
debris that passes through a user defined observation
point (please refer to Section 5.3.2 for further details).
 12

(v) Block plots: shows the time history of velocity, resisting

force, self weight, block length and thickness of each
boundary block of landslide debris (please refer to
Section 5.3.3 for further details).

(d) Reset location of windows: restore the windows to the

default location.

(e) About: displays the terms and conditions of the computer

program.

(f) Resources: contains the links to the user manual and the
validation report (i.e. TN 1/2015) of 2d-DMM (Version
2.0).

(g) Close: quit the program.

Figure 2.2 File Menu of 2d-DMM (Version 2.0)

 13

Figure 2.3 Settings Page of 2d-DMM (Version 2.0)

Figure 2.4 Output Page of 2d-DMM (Version 2.0)

 14

3 Data Input
3.1 Data Management

All input data are entered on the Settings page as shown in Figure 2.3. The input data
are grouped into five categories.

(a) Project title,

(b) Landslide debris,

(c) Interface resistance,

(d) Entrainment, and

(e) Analysis parameters.

Validity checking is provided in the program. As the data is entered, individual items
are checked against their types (e.g. characters, numbers, positive numbers, etc) and their
maximum and minimum limits as appropriate. If data of a wrong type or data outside the
limits is entered on the Settings page, an error message will be displayed when the Save
settings menu item is clicked.

3.2 Project Title

User can enter the project title under this category.

3.3 Landslide Debris

In this category, users shall provide input data pertaining to the physical properties and
dimensions of the landslide debris, which comprises the density, volume, initial horizontal
length, earth pressure coefficients (active (ka), at-rest (k0) and passive (kp)), initial location of
the debris end, as well as initial velocity, as elaborated below:

(a) Density (in kg/m3): density of the landslide debris. It must

be greater than 0.

(b) Volume (in m3): volume of the landslide debris. It must be

greater than 0.

(c) Initial horizontal length (in m): horizontal length of the

landslide debris at its initial location. It must be greater
than 0.

(d) Earth pressure coefficients (active (ka), at-rest (k0) and

passive (kp)): the maximum, minimum and at-rest earth
pressure coefficients. kp must be not less than k0, and k0
 15

must be not less than ka. ka must be non-negative.

(e) Pore pressure ratio for earth pressure: the pore pressure
ratio (ru) that affects the earth pressure coefficient. Users
are reminded that all stress calculations in 2d-DMM
(Version 2.0) are carried out in terms of total stress. The
use of pore pressure ratio is to convert effective earth
pressure coefficients (k) to that in total stress as follows:

Total stress earth pressure coefficient, K = k (1 – ru) + ru

(f) Initial location of debris end (in m): the chainage of the
debris tail at the source location. It must be non-negative.

(g) Initial velocity (in m/s): the initial velocity of the landslide
debris. This velocity will be applied to all blocks in the
first time step of the time marching numerical process
(i.e. velocities of all the blocks are the same and equal to the
initial velocity at t = 0). For rainfall-induced landslide, the
ground mass would detach from the source area at a
stationary state, and the initial debris velocity should be set
to zero. However, for some circumstances, specifying an
appropriate initial debris velocity for analysis may be
needed as proposed by Borstad & McClung (2009). It
must be non-negative.

The initial landslide mass should be defined, using the input parameters given in
Sections 3.3(b), (c) and (f), so as to represent the reported bulked volume (for back analysis
purpose) or the design volume of a landslide event (for design purpose). A parabolic shaped
source debris is generated by the program based on the input parameters given in Sections
3.3(b), (c) and (f), using the equation of the parabola given below.

 y  y0 2  4ax  x0  .................................................(3.1)
where x0, y0 and a are determined based on the input parameters of volume, initial horizontal
length, as well as initial location of debris end. The determined values of x0, y0 and a are
then used to form the initial parabolic shape of the source debris.

3.4 Interface Resistance

The flow path is divided into two segments, each having its own rheological model
and parameters. Two rheological models, the friction model and Voellmy model, are
available in 2d-DMM to simulate different landslide movement behaviours. Users need to
enter the chainage at which the interface changes from Segment 1 to Segment 2. Users also
need to enter the rheological parameters of these two interfaces respectively. Rheological
parameters include the friction angle ( in degree), pore pressure ratio for basal friction (ru),
turbulence coefficient (in m/s2). Users can choose to include the resistance at the interface
 16

between the channel sides and landslide debris in the simulation.

Input parameters related to the interface resistance are elaborated below:

(a) Friction angle ( in degree): the friction angle at the

interface between the channel base/sides and landslide
debris. It can be the apparent friction angle (a) if ru = 0 or
the effective stress friction angle ( ) if ru > 0. The values
of the friction angle for both two segments are required.
The friction angle is adopted for both the friction and the
Voellmy models. It must be non-negative and smaller
than 90°.

(b) Turbulent coefficient (in m/s2) (0 for the friction model):

the turbulent coefficient at the interface between the channel
base/sides and landslide debris. The values of the
turbulent coefficient for both two segments are required.
The Voellmy model is adopted in a simulation when > 0
while the friction model is invoked when = 0. It must be
non-negative.

(c) Pore pressure ratio for basal friction (ru): all stress
calculations in 2d-DMM are carried out in terms of total
stress. The apparent friction angle,  a , is related to the
effective stress friction angle ( ) and the pore pressure ratio
for basal friction (ru) as follows:

tan a  1  ru  tan  '

where u
v

To avoid causing confusion, users are suggested to adopt

and input  a in Section 3.4(a) and to adopt ru = 0 in the
program. It must be non-negative.

(d) Start location of segment 2: the chainage beyond which the

friction angle, turbulent coefficient and pore pressure ratio
for basal friction for Segment 2 are used, in the unit of m.
It must be non-negative.

(e) The check box “Include side resistance?”: the flow

resistance due to the channel sides is considered in the
simulation if this box is checked. Given that a landslide
debris experiences flow resistance from both the channel
base and the side walls, choosing this option is
recommended.
 17

3.5 Entrainment

The debris volume would start to increase with time once the most frontal block of the
debris reaches a designated entrainment zone, through adding volume to the blocks of the
debris within the entrainment zone. The program allows users to specify two separate
entrainment zones along the flow path, each with its own entrainment rate. The volume of a
block would not increase further when the thickness of the block reaches a user-specified
threshold value. Input parameters related to the entrainment are elaborated below:

(a) Start location of entrainment zone (in m): the start location
of the entrainment zone for segment 1 must be non-negative
and smaller than the end location of the entrainment zone
for segment 1. The start location of the entrainment zone
for segment 2 must be non-negative, larger than the end
location of the entrainment zone for segment 1 and smaller
than the end location of the entrainment zone for segment 2.

(b) End location of the entrainment zone (in m): the end
location of the entrainment zone for segments 1 must be
non-negative, larger than the start location of the
entrainment zone for segment 1 and smaller than the start
location of the entrainment zone for segment 2. The end
location of the entrainment zone for segment 2 must be
non-negative and larger than the start location of the
entrainment zone for segment 2.

(c) Entrainment rate (in m3/s): entrainment rate is the rate of

change in debris volume. Due to these formulation
requirements, it is a trial and error process to obtain a design
entrainment volume. It must be non-negative.

(d) Threshold entrainment depth (in m): the threshold debris

thickness above which the volume of a block would not
increase further. It must be non-negative.

3.6 Analysis Parameters

The parameters for the stepwise time marching finite difference solution are required,
which comprises the time step, maximum flow duration, number of time steps for updating as
well as the number of blocks, as elaborated below:

(a) Time step (in s): the prototype time interval for one
calculation step. It must be greater than 0 and smaller than
the maximum flow duration.

(b) Maximum flow duration (in s): the prototype time at

which the calculation stops. It must be greater than the
time step.
 18

(c) Number of time steps for updating: the number of

calculation steps between screen updating and data output.
For example, screen updating and data output are carried out
in a time interval of 0.1 s by setting the maximum flow
duration of 1 s and the number of time steps for updating of
10. It must be greater than 1.

(d) Number of blocks: the number of boundary blocks that the

landslide debris is split into for analysis. There must be at
least three blocks. The recommended minimum number of
boundary blocks is 11. The recommended maximum
number of boundary blocks is 50. In order to prevent
instability due to numerical divergence and to increase
precision, the more the boundary blocks are used, the
smaller the time step should be for analysis.

3.7 Runout Path Properties

The geometry of the runout path is specified using points. The input points are
entered in the tables of the Runout path tab on the Settings page (Figure 3.1). Four types of
points can be entered, namely topography, channel width, left side angle and right side angle.
The topography and the width of the channel are entered in meters. The side angles are
entered in radian and should not be greater than 1.57. The total number of each type of
points should not exceed 500. The input points are entered in ascending order of the x
coordinates, as shown in Figure 3.1. The input points are linked together using a series of
interconnected cubic spline functions to produce a continuous and differentiable flow path for
the calculation of debris motion. The cubic spline functions are prepared prior to each
simulation run.

All points in the tables of the Runout path tab can be copied from Microsoft Excel and
pasted into the tables by pressing Ctrl+v. Likewise, the points can be copied to Microsoft
Excel from the tables by pressing Ctrl+c. The Clear button helps users to remove all points
in a table.
 19

Figure 3.1 Runout Path Tab on the Settings Page

4 Methods of Solution
4.1 General

The 2d-DMM (Version 2.0) is developed in-house by the GEO. This program adopts
the principles of the DAN model proposed by Hungr (1995) and extended by Hungr & Evans
(1997) and Kwan & Sun (2006).

4.2 Principles of Oldrich Hungr’s DAN Model Adopted in 2d-DMM

2d-DMM, the same as DAN, is a simple and versatile continuum numerical model
which may be used to simulate the motion of a deformable landslide debris along a pre-defined
curvilinear flow path using a finite difference solution of the governing dynamic equations in a
Lagrangian framework. The solution is obtained by stepwise calculations in small time steps
for a block assembly of elements, representing the landslide debris as a continuum (Figure 4.1).

Once the landslide mass simulated by a number of blocks along a 2-dimensional curve
representing the runout path is set up, the net driving force of the blocks can be obtained from
the tangential component of the self-weight of the block, the basal resistance and difference in
inter-block normal forces, as shown in Figure 4.2. This out-of-balance driving force is then
 20

used to determine the resultant change in velocity through Newton’s second law. Time
marching numerical integration along the flow path gives the velocity and travel distance of
the landslide debris at different times and predict the debris thickness, provided that the cross
section of the flow path (which is assumed to be trapezoidal) is defined.

Notes: (1) Boundary blocks are numbered i = 1 to n + 1.

(2) Mass blocks are numbered j = 1 to n.

Figure 4.1 Lagrangian Mesh in Curvilinear Co-ordinates

Figure 4.2 External Forces Acting on a Boundary Block

In 2d-DMM, two basal resistance rheological models, known as the friction model and
Voellmy model, are available to simulate different landslide movement mechanisms.
Friction and Voellmy models are commonly used for modelling landslides in Hong Kong.

The original DAN model proposed by Hungr (1995) was extended by Hungr & Evans
(1997) to make it possible to simulate volume change of landslide debris along its propagation.
This procedure has also been incorporated in 2d-DMM (Version 2.0).

It is essential that users understand the approximations and assumptions involved in

the model and consider its limitations. Details of the theoretical background of the DAN
model are presented in Hungr (1995) and Hungr & Evans (1997). Users are recommended
to study this before using the program.
 21

4.3 Modifications to Oldrich Hungr’s DAN Model Incorporated into 2d-DMM

4.3.1 Trapezoidal Cross Section of Flow Path

A major improvement to Oldrich Hungr’s DAN model that has been incorporated into
2d-DMM is to model the cross section of a flow path in a trapezoidal shape. The DAN
model approximates the cross section of flow path as rectangular-shaped, as shown in
Figure 4.3. Such approximation affects the analysis in the following ways.

Figure 4.3 Rectangular Cross Section of Debris Flow Path Assumed in Oldrich
Hungr’s DAN Model

(a) The widths of all debris blocks remain the same at any
particular location, irrespective of the debris thickness
flowing across that particular location. Such error may be
negligible for steep-side cross sections resembling
rectangles, but will affect the motion of landslide debris in
cross sections with sloping sides where the surface width
should vary with depth.

(b) A pre-defined width profile along the flow path is required

for analysis. Therefore, users need to guess the width of
debris tail when assessing debris mobility of a predicted
event.

(c) The flow resistance is assumed in the motion equation to act

on the channel base only, while in reality it acts on the
wetted perimeter of the channel.

In order to overcome the above shortcomings resulted from the assumption of a

rectangular flow channel, 2d-DMM has incorporated a functionality module so that users can
select to approximate the flow channel in a trapezoidal shape, as shown in Figure 4.4.
Modifications to the calculation formulae of Oldrich Hungr’s DAN model have been made.
 22

Figure 4.4 Trapezoidal Cross Section of Debris Flow Path Adopted in 2d-DMM

4.4 Model Evaluation by Back Analysis

The validity of a computer model can be verified by back analysing a number of

well-documented case histories and laboratory experiments.

(a) Back Analysis of Hong Kong Natural Terrain Landslides

using Oldrich Hungr’s DAN Model

Oldrich Hungr’s DAN model has been used to back analyse

a number of landslides in Hong Kong, including some
twenty natural terrain landslides (Hungr, 1998; Ayotte &
Hungr, 1998) with debris volumes ranging from about
50 m3 to about 26,000 m3.

The back analyses of the mobility of the natural terrain

landslides by Hungr (1998) and Ayotte & Hungr (1998)
have demonstrated that the continuum model can reasonably
predict the downslope motions of landslide debris.
Subsequently, the results of these back analyses have been
used to develop an empirical correlation to facilitate the
assessment of mobility. Lo (2001) presents this empirical
correlation and suggests this approach as one possible
method for estimating debris mobility. In addition, Lo
(2001) recommends “continuum models can be used to
estimate debris mobility, ... but only those models, such as
DAN, ... which have been calibrated against field
measurements, should be used.”

(b) Back Analysis of Hong Kong Natural Terrain Landslides

using GEO-DAN

A number of back analyses of natural terrain landslides in

Hong Kong have been carried out using the GEO-DAN
program for comparison with those previously produced by
Oldrich Hungr’s DAN model. The results conclude that
 23

the GEO-DAN program performs as well as the DAN

program previously developed by Oldrich Hungr, in
simulating the motion of landslide after failure.

(c) Back Analysis of Laboratory Experiment using GEO-DAN

Apart from natural terrain landslides in Hong Kong, a back

analysis has been carried out using GEO-DAN to closely
simulate an experiment on the flow of sand down a steeply
inclined sand-lined chute described by Hungr (1995). The
back analysis has been described in details in the user
manual of the “GEO-DAN” program. The results
conclude that GEO-DAN correctly simulates the movement
of the sand throughout its path as well as the shape of the
flow slide profile and that the GEO-DAN program performs
as well as the DAN program previously developed by
Oldrich Hungr, in simulating the motion of the sand in the
experiment.

4.4.1 Back Analysis of Case History for Comparison with the Accepted GEO-DAN
Program

The back analysis of Liu Pok Landslide (King, 1997) which occurred in 1993 has been
carried out using the 2d-DMM (Version 1.0). The same analysis has been previously carried
out by the accepted GEO-DAN program and the results included in its user manual. A
comparison of the results from the analysis has demonstrated that 2d-DMM (Versions 1.1)
and 2d-DMM (Version 1.0), when no new functionality module is selected, are the same as
the accepted GEO-DAN program. Also, a comparison of the results from the analysis has
demonstrated that 2d-DMM (Versions 1.2) and 2d-DMM (Versions 1.1), when no new
functionality module is selected, are the same.

A comparison of results of back-analyses of some historical and hypothetical

landslides between 2d-DMM (Versions 2.0) and the existing pre-accepted programs,
2d-DMM (Version 1.2) and DAN-W (Release 10) were carried out. It is demonstrated that
2d-DMM (Version 2.0) produces results consistent with both the pre-accepted programs (Law
& Ko, 2015).

5 Output

Output from 2d-DMM (Version 2.0) are grouped into three categories, namely
elevation view, textual output and data plots.

5.1 Elevation View

When users press the Start button at the top left corner of the Output page to start an
analysis, the motion of the landslide is displayed real-time at a output time interval equals to
 24

Time step (i.e. Section 3.6(a)) times Number of time steps for updating (i.e. Section 3.6(c)).
For example, screen update and data output are carried out in a time interval of 0.1 s by
setting the maximum flow duration of 1 s and the number of time steps for updating of 10.

Users can review the motion of the landslide debris throughout the flow duration using
the scroll bar as shown in Figure 5.1. The scroll bar function is active after a simulation run
is completed. The progress of the simulation run is given by a progress bar at the lower right
corner of the Output page.

A menu which contains additional functions, such as copying, printing and scaling the
side view of the flow profile, can be brought up by clicking the right mouse button on the
flow profile. In addition, by choosing the Show point values option on the menu, x and y
coordinates of a boundary block over which the mouse cursor is placed are displayed, as
illustrated in Figure 5.2.

Users can zoom-in and view the landslide debris by dragging the mouse on the plot as
illustrated in Figure 5.3. When the box Keep zoom value on the lower right corner is
checked, the current zoom-in setting would be maintained for viewing at another time interval
adjusted using the scroll bar. The scale will be restored to default by not checking the Keep
zoom value and moving the scroll bar.

The scales of the x and y axes are set automatically according to the geometry of the
flow path. The box Same scale for both axes on the lower right corner can be checked to
equate the scale of the x axis to the y axis, if users would like to have a better sense on the
gradient of the flow path.

The scroll bar for viewing the motion

ofScroll bardebris
landslide for viewing the the
throughout motion
of the
flow landslide debris throughout
duration.
the flow duration

Figure 5.1 Scroll Bar on the Output Page

 25

Figure 5.2 Show Point Values Function on the Output Page

Zoom by dragging Closer inspection of

landslide debris

Figure 5.3 Zoom-in Function

5.2 Textual Output

Every time when the graphical display is updated, the status of the landslide mass at
that time step will be saved as a separate text file in the project folder, which can be located
and opened by clicking Project→View output files in file explorer in the main menu. These
include the positions (in x-y co-ordinates), velocity, thickness, width, slope gradient, gravity
driving force, lateral earth force for each block of a landslide mass. Instead of digging deep
into the raw textual data for post-processing, 2d-DMM (Version 2.0) provides a number of
graphical plots functions to facilitate the viewing and analysis of the textual data, as
elaborated in Section 5.3.

5.3 Data Plots

The program has included the profile plots, hydrograph plots and block plots functions
 26

to graphically display output data, as elaborated below.

5.3.1 Profile Plots

The velocity and thickness profiles along chainage can be plotted using the profile plot
function by clicking the Profile plots, as shown in Figure 5.4, on the Output page. A total of
10 boundary blocks can be selected by users using the pull-down menu. The plots are
updated by clicking the Go button. Similar to the elevation view, the menu of additional
functions can be brought up by right clicking the mouse on the plot. The text data behind
each line on the plots are stored as text files, e.g. line1-Velocity.txt, line2-Velocity.txt,
line1-Depth.txt, etc., in the project folder. Users can make use of the text data for further
analysis and plotting.

Figure 5.4 Velocity and Thickness Profiles along Chainage

 27

5.3.2 Hydrograph Plots

Users can place an observation point anywhere along the flow path to monitor the
velocity, thickness, cumulative volume and cumulative kinetic energy of the landslide debris
at that point against time. To set an observation point, users shall click the Hydrograph plots
on the Output page and input the chainage of the observation point on the screen. The
velocity, thickness, cumulative volume and cumulative kinetic energy hydrograph plots are
shown by clicking the Go button. A solid vertical line is shown on the elevation view to
indicate the location of the observation point, as shown in Figure 5.5. Similar to the
elevation view on the Output page, the menu of additional functions is brought up by right
clicking the mouse on the plots. The text data behind the hydrograph plots are stored as
VelocityCSV.csv, FlowDepthCSV.csv, VolumeCumCSV.csv and EnergyCumCSV.csv
respectively in the project folder. Users can make use of the text data for further analysis
and plotting.

The cumulative volume at the observation point is calculated using the following
equation.
T
Vop   Ac vi dt ....................................................... (5.1)
0
where Vop = the cumulative volume at the observation point (in m3)
= maximum flow duration (in s)
= cross sectional area at the observation point (in m2)
vi = velocity of boundary block i that reaches the observation point (in m/s)
t = time (in s)

The cumulative kinetic energy at the observation point is calculated using the
following equation.

 T

2000 
Eop 
3
Ac vi dt .................................................. (5.2)
0

where Eop = the cumulative kinetic energy at the observation point (in kJ)
= density of landslide debris (in kg/m3)
= maximum flow duration (in s)
= cross sectional area at the observation point (in m2)
vi = velocity of boundary block i that reaches the observation point (in m/s)
t = time (in s)

5.3.3 Block Plots

The velocity, flow resistance, gravity driving force, lateral earth force, tangential
length of the boundary blocks, as well as the thickness of all boundary blocks at each output
time interval throughout the flow duration can be viewed by clicking the Block plots on the
Output page and setting the prototype time using the scroll bar at the bottom of the plot, as
shown in Figure 5.6. Similar to the flow profile on the Output page, the menu of additional
functions is brought up by right clicking the mouse on the plots. Users can click the Show
data in Excel button to view the text data behind the block plots for further analysis and
plotting.
 28

The solid line showing the

location of the observation
point.

Figure 5.5 (a) Opening a Separate Page to Enter the Location of an Observation Point,
and (b) Showing the Velocity, Thickness, Cumulative Volume and
Cumulative Kinetic Energy Hydrograph Plots
 29

Figure 5.6 Block Plots Function

6 Modelling Guidelines

This section presents some general modelling guidelines for predicting debris mobility
of natural terrain landslides using 2d-DMM.

6.1 Runout Path Properties

Channel geometry in 2d-DMM (Version 2.0) is specified using x-y co-ordinates of a

flow path. The recommendation regarding specification of flow path using points given in
Hungr (2010) is relevant:

“The input profile should be made reasonably smooth to avoid

instability. Do not use too many points and avoid details such
as minor steps in the profile. Round out abrupt slope changes.
Users should test the influence of such simplification (usually it
has relatively small effect on the results, but excessive
roughness could unrealistically reduce the runout). Ideally, a
slope profile should have about 15-25 input points.”

In all cases, users should adopt reasonable number of points to represent smooth slope
surfaces for realistic computation results. If this basic assumption is violated, unreliable
results are inevitable. Besides, smooth profiles of channel width and side angles are also
recommended. The complex interactions between abrupt change of channel width and side
 30

angles are not modelled in 2d-DMM (Version 2.0) which is a two-dimensional numerical
formulation. Please refer to Section 1.3(c) for details of this program assumption.

6.2 Interface Properties

GEO Technical Guidance Notes No. 29 (GEO, 2011), No. 34 (GEO, 2012) and No. 38
(GEO, 2013) contain recommendations on the appropriate rheological parameters for
simulation of channelised debris flows, open hillslope failures and failures within topographic
depression catchments for forward prediction purpose. Users of 2d-DMM (Version 2.0) are
recommended to refer to these guidelines for more details on selecting the appropriate
rheological models and parameters.

All stress calculations in 2d-DMM are carried out in terms of total stress. Please refer
to Section 3.4 of this manual for details of the rheological parameters used in this program.

6.3 Analysis Parameters

2d-DMM (Version 2.0) adopts finite difference numerical scheme in which the
resolution of computed velocity, basal resistance, debris thickness, etc, along the landslide
debris would normally be improved by using more blocks. However, increasing the number
of blocks would not only demand longer processing time (Hungr, 2010), but also involve
higher chance of numerical errors or fluctuating velocity and thickness due to boundary
blocks overriding each other. A balance needs to be struck between solution resolution,
processing time and numerical stability.

The number of time steps between updating graphical display (and producing
numerical output) should be chosen such that the time interval between every update is set
between 0.2 s and 1.0 s, which is considered appropriate and sufficient for producing smooth
curves in the velocity-time graph and for identifying the maximum frontal velocity.

6.4 Nodal Velocity

Frontal velocity can be taken as the velocity of the node at the front end of landslide
mass while the runout is the travel distance of that node when the landslide mass has come to
rest. However, it is not realistic for the computer model to achieve an exactly zero velocity
at the front boundary block for determining the total runout. In fact, since no numerical
damping is applied in the program, velocity of the boundary blocks (including the frontal
boundary block) will be fluctuating between very small positive and small negative values,
and the boundary blocks (including the frontal boundary block) may never come to rest (i.e.
the calculated velocity genuinely equals to zero). Therefore, users should set a threshold
value of frontal velocity, below which the frontal boundary block can be considered at rest, in
order to determine the total run-out. For example, a frontal boundary block can be
considered at rest if its velocity fluctuates between -0.01 m/s and +0.01 m/s.
 31

7 References

Ayotte, D. & Hungr, O. (1998). Runout Analysis of Debris Flows and Debris Avalanches in
Hong Kong. Report prepared for the Geotechnical Engineering Office, Hong Kong.

Borstad, C.P. & McClung, D.M. (2009). Sensitivity Analyses in Snow Avalche Dynamics
Modelling and Implications when Modeling Extreme Events. Canadian
Geotechnical Journal, vol. 46, pp 1024-1033.

GEO (2011). Guidelines on the Assessment of Debris Mobility for Channelised Debris
Flows (GEO Technical Guidance Note No. 29). Geotechnical Engineering Office,
Hong Kong, 6 p.

GEO (2012). Guidelines on Assessment of Debris Mobility for Open Hillslope Failures
(GEO Technical Guidance Note No. 34). Geotechnical Engineering Office, Hong
Kong, 16 p.

GEO (2013). Guidelines on the Assessment of Debris Mobility for Failures within
Topographic Depression Catchments (GEO Technical Guidance Note No. 38).
Geotechnical Engineering Office, Hong Kong, 8 p.

Hungr, O. (1995). A model for the runout analysis of rapid flow slides, debris flows and
avalanches. Canadian Geotechnical Journal, vol. 32, pp 610-623.

Hungr, O. (2010). Use ’s Man al of DAN-W. O. Hungr. Geotechnical Research Inc.,

Canada, 61 p.

Hungr, O. & Evans, S.G. (1997). A dynamic model for landslide with changing mass.
Proceedings of the International Symposium on Engineering Geology and the
Environment, vol. 1, pp 719-724.

King, J.P. (1997). Natural Terrain Landslide Study - Damage to Liu Pok School by a
Natural Terrain Landslide (Discussion Note No. DN 2/97). Geotechnical
Engineering Office, Hong Kong, 52 p.

Kwan, J.S.H. & Sun, H.W. (2006). An improved landslide mobility model. Canadian
Geotechnical Journal, vol. 43, pp 531-539.

Law, R.P.H & Ko, F.W.Y. (2015). Validation of Geotechnical Computer Program
“2d-DMM (Ve sion 2.0)” (Technical Note No. TN 1/2015). Geotechnical
Engineering Office, Hong Kong, 67 p.

Lo, D.O.K. (2001). Review of Natural Terrain Landslide Debris-resisting Barrier Design
(GEO Report No. 104). Geotechnical Engineering Office, Hong Kong, 91 p.