PROBLEM SOLVING WITH 3DEC 3-1

3 PROBLEM SOLVING WITH 3DEC

This section provides guidance in the use of 3DEC in problem solving for rock mechanics engineer-
ing.* In Section 3.1, an outline of the steps recommended for performing a geomechanics analysis
is given, followed in Sections 3.2 through 3.10 by an examination of specific aspects that must be
considered in any model creation and solution. These include:
• model generation (Section 3.2);
• choice of rigid or deformable block analysis (Section 3.3);
• boundary and initial conditions (Sections 3.4 and 3.5);
• loading and sequential modeling (Section 3.6);
• choice of block and joint constitutive models and material properties
(Sections 3.7 and 3.8);
• ways to improve modeling efficiency (Section 3.9); and
• interpretation of results (Section 3.10).
Finally, the philosophy of modeling in the field of geomechanics is examined in Section 3.11; the
novice modeler in this field may wish to consult this section first. The methodology of modeling in
geomechanics can be significantly different from that in other engineering fields (such as structural
engineering). It is important to keep this in mind when performing any geomechanics analysis.

* Problem solving for coupled mechanical-thermal analysis is described in Section 1 in Optional

Features, and problem solving for dynamic analysis is discussed in Section 2 in Optional Features.

3DEC Version 5.2

3-2 User’s Guide

3.1 General Approach

The modeling of geoengineering processes involves special considerations and a design philosophy
different from that followed for design with fabricated materials. Analyses and designs for structures
and excavations in or on rocks and soils must be achieved with relatively little site-specific data and
an awareness that deformability and strength properties may vary considerably. It is impossible to
obtain complete field data at a rock or soil site. For example, information on stresses, properties
and discontinuities can only be partially known, at best.
Since the input data necessary for design predictions are limited, a numerical model in geomechanics
should be used primarily to understand the dominant mechanisms affecting the behavior of the
system. Once the behavior of the system is understood, it is then appropriate to develop simple
calculations for a design process.
This approach is oriented toward geotechnical engineering, in which there is invariably a lack of
good data; but in other applications, it may be possible to use 3DEC directly in design if sufficient
data, as well as an understanding of material behavior, are available. The results produced in a
3DEC analysis will be accurate when the program is supplied with appropriate data. Modelers
should recognize that there is a continuous spectrum of situations, as illustrated in Figure 3.1:

Typical Complicated geology; Simple geology;

inaccessible; $$$ spent on site
situation no testing budget investigation

Data NONE COMPLETE

Investigation of Bracket field behavior Predictive

Approach mechanisms by parameter studies (direct use in design)

Figure 3.1 Spectrum of modeling situations

3DEC may be used either in a fully predictive mode (right-hand side of Figure 3.1) or as a “numerical
laboratory” to test ideas (left-hand side). It is the field situation and budget, rather than the program,
that determine the types of use. If enough data of a high quality are available, 3DEC can give good
predictions.
Since most 3DEC applications will be for situations in which little data are available, this section
discusses the recommended approach for treating a numerical model as if it were a laboratory test.
The model should never be considered as a “black box” that accepts data input at one end and
produces a prediction of behavior at the other. The numerical “sample” must be prepared carefully,
and several samples tested, to gain an understanding of the problem. Table 3.1 lists the steps
recommended to perform a successful numerical experiment; each step is discussed separately.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3-3

Table 3.1 Recommended steps for numerical analysis in geomechanics

Step 1 Define the objectives for the model analysis

Step 2 Create a conceptual picture of the physical system
Step 3 Construct and run simple idealized models
Step 4 Assemble problem-specific data
Step 5 Prepare a series of detailed model runs
Step 6 Perform the model calculations
Step 7 Present results for interpretation

3.1.1 Step 1: Define the Objectives for the Model Analysis

The level of detail to be included in a model often depends on the purpose of the analysis. For
example, if the objective is to decide between two conflicting mechanisms that are proposed to
explain the behavior of a system, then a crude model may be constructed, provided that it allows
the mechanisms to occur. It is tempting to include complexity in a model just because it exists in
reality. However, complicating features should be omitted if they are likely to have little influence
on the response of the model, or if they are irrelevant to the model’s purpose. Start with a global
view and add refinement as (and if) necessary.

3.1.2 Step 2: Create a Conceptual Picture of the Physical System

It is important to have a conceptual picture of the problem to provide an initial estimate of the ex-
pected behavior under the imposed conditions. Several questions should be asked when preparing
this picture. For example: Is it anticipated that the system could become unstable? Is the pre-
dominant mechanical response linear or nonlinear? Are there well-defined discontinuities that may
affect the behavior, or does the material behave essentially as a continuum? Is there an influence
from groundwater interaction? Is the system bounded by physical structures, or do its boundaries
extend to infinity? Is there any geometric symmetry in the physical structure of the system?
These considerations will dictate the gross characteristics of the numerical model, such as the
design of the model geometry, the types of material models, the boundary conditions, and the
initial equilibrium state for the analysis. They will determine whether a three-dimensional model is
required, or whether a two-dimensional model can be used to take advantage of geometric conditions
in the physical system.

3DEC Version 5.2

3-4 User’s Guide

3.1.3 Step 3: Construct and Run Simple Idealized Models

When idealizing a physical system for numerical analysis, it is more efficient to construct and run
simple test models first, before building the detailed model. Simple models should be created at
the earliest possible stage in a project, to generate data and add to understanding. The results can
provide further insight into the conceptual picture of the system; Step 2 may need to be repeated
after simple models are run.
Simple models can reveal shortcomings that can be remedied before any significant effort is invested
in the analysis. For example, do the selected material models sufficiently represent the expected
behavior? Are the boundary conditions influencing the model response? The results from the
simple models can also help guide the plan for data collection by identifying which parameters
have the most influence on the analysis.

3.1.4 Step 4: Assemble Problem-Specific Data

The types of data required for a model analysis include:

• details of the geometry (e.g., profile of underground openings, surface topography, dam
profile, rock/soil structure);
• locations of geologic structure (e.g., faults, bedding planes, joint sets);
• material behavior (e.g., elastic/plastic properties, post-failure behavior);
• initial conditions (e.g., in-situ state of stress, pore pressures, saturation); and
• external loading (e.g., explosive loading, pressurized cavern).
Since, typically, there are large uncertainties associated with specific conditions (in particular, state
of stress, deformability and strength properties), a reasonable range of parameters must be selected
for the investigation. The results from the simple model runs (in Step 3) can often prove helpful in
determining this range, and in providing insight for the design of laboratory and field experiments
to collect the needed data.

3.1.5 Step 5: Prepare a Series of Detailed Model Runs

Most often, the numerical analysis will involve a series of computer simulations that include the
different mechanisms under investigation, and span the range of parameters derived from the as-
sembled database. When preparing a set of model runs for calculation, several aspects, such as the
following, should be considered:
1. How much time is required to perform each model calculation? It can be difficult to obtain
sufficient information to arrive at a useful conclusion if model runtimes are excessive.
Consideration should be given to performing parameter variations on multiple computers
to shorten the total computation time.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3-5

2. The state of the model should be saved at several intermediate stages so that the entire run
does not have to be repeated for each parameter variation. For example, if the analysis
involves several loading/unloading stages, the user should be able to return to any stage,
change a parameter and continue the analysis from that stage. Consideration should be
given to the amount of disk space required for save files.
3. Are there a sufficient number of monitoring locations in the model to provide for a clear
interpretation of model results, and for comparison with physical data? It is helpful
to locate several points in the model at which a record of the change of a parameter
(such as displacement, velocity or stress) can be monitored during the calculation. Also,
the maximum unbalanced force in the model should always be monitored to check the
equilibrium or failure state at each stage of an analysis.

3.1.6 Step 6: Perform the Model Calculations

It is best to first make one or two model runs, split into separate sections, before launching a series of
complete runs. The runs should be checked at each stage to ensure that the response is as expected.
Once there is assurance that the model is performing correctly, several data files can be linked
together to run a complete calculation sequence. At any time during a sequence of runs, it should
be possible to interrupt the calculation, view the results and then continue or modify the model as
appropriate.

3.1.7 Step 7: Present Results for Interpretation

The final stage of problem solving is the presentation of the results for a clear interpretation of
the analysis. This is best accomplished by displaying the results graphically, either directly on
the computer screen or as output to a hardcopy plotting device. The graphical output should be
presented in a format that can be directly compared to field measurements and observations. Plots
should clearly identify regions of interest from the analysis, such as locations of calculated stress
concentrations or areas of stable movement versus unstable movement in the model. The numeric
values of any variable in the model should also be readily available for more detailed interpretation
by the modeler.
We recommend that these seven steps be followed to solve geoengineering problems efficiently.
The following sections describe the application of 3DEC to meet the specific aspects of each of the
steps in this modeling approach.

3DEC Version 5.2

3-6 User’s Guide

3.2 Model Generation

One of the main tasks when performing a 3DEC analysis is building a numerical representation
of the real-world situation. All 3DEC models will have an outer boundary to which boundary
conditions will be applied. In addition, most models will have features such as joints, faults,
cracks, excavations or material property interfaces. Quite often, the features to be modeled are not
continuous through the entire model. The complex shapes which result from the intersections of
noncontinuous planes in space are difficult to create and visualize. The purpose of this section is
to give some guidance for using a systematic approach to model building.
The first rule is keep the model simple. Usually, a numerical model is being developed because the
real-life situation is too complex to understand. There is a tendency to attempt to include every
possible feature in a numerical model. This results in a 3DEC model that is also too complex
to understand. The goal of the modeler should be to understand the mechanisms, properties and
parameters that determine the model’s behavior. The time required to calculate a solution also
increases with increasing complexity. Therefore, it is to the modeler’s advantage to start with
a model that only includes the minimum of features. The complexity of the model can then be
increased by adding features one at a time and noting the effect. By using this approach, the
modeler has the best chance of gaining understanding of the parameters that are critical to the
real-life situation.
3DEC is different from conventional numerical programs in the way the model geometry is created.
A 3DEC model can be created in two ways: (1) by splitting a polyhedron into separate polyhedra;
and (2) by creating separate polyhedra and joining them together. For most geomechanics analyses,
a single block is created first, with a size that encompasses the physical region being analyzed.
Then, this block is cut into smaller blocks whose boundaries represent both geologic features and
engineered structures in the model. This cutting process is termed collectively as joint generation;
however, “joints” represent both physically real geologic structures and boundaries of man-made
structures or materials that will be removed or changed during the subsequent stages of the 3DEC
analysis. In this latter case, the joints are fictitious entities and their presence should not influence
model results. The representation of fictitious joints is discussed in Section 3.2.3.

3.2.1 Fitting the 3DEC Model to a Problem Region

The 3DEC model geometry must represent the physical problem to an extent sufficient to capture
the dominant mechanisms related to the geologic structure in the region of interest. The following
aspects must be considered:
1. How much detail should be used to represent the geologic structure (e.g., faults,
joints, bedding planes, etc.)?
2. How will the location of the model boundaries influence model results?
3. If deformable blocks are used, what density of zoning is required for accurate
solution in the region of interest?
All three aspects determine the size of the 3DEC model that is practical for analysis.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3-7

As mentioned above, there are two different starting points in building 3DEC models. The first
method is to describe a simple starting shape and slice it up to create the desired geometrical
features. The second involves defining complex polyhedral shapes and putting them together to
form the continuous mass. Both of these approaches make use of the POLYHEDRON command.
There are seven forms of the POLYHEDRON command available in 3DEC:
POLYHEDRON face
POLYHEDRON brick
POLYHEDRON cube
POLYHEDRON drum
POLYHEDRON prism
POLYHEDRON tunnel
POLYHEDRON tet
By using the POLYHEDRON face command, polyhedra of virtually any shape can be defined. Each
face is defined by a list of vertex coordinates. The list must be entered in clockwise order, looking
at the face from outside the polyhedra. All points on a face must be coplanar, and the resulting
polyhedra created by the face commands must be convex. Continuation lines are allowed, but the
coordinates for each vertex may not be split between lines. All faces required to close the polyhedra
must be specified. A simple example of using the POLYHEDRON face command to generate a cube
(1 unit on each side) is:

Example 3.1 A cube generated with the POLYHEDRON face command

new
poly &
face 0,0,0 1,0,0 1,1,0 0,1,0 &
face 0,0,0 0,0,1 1,0,1 1,0,0 &
face 0,0,0 0,1,0 0,1,1 0,0,1 &
face 1,1,1 1,1,0 1,0,0 1,0,1 &
face 1,1,1 1,0,1 0,0,1 0,1,1 &
face 1,1,1 0,1,1 0,1,0 1,1,0
plot create plot Blocks
plot block
plot reset
plot set dip 120 dd 30
ret

The model created with the example is shown in Figure 3.2. In this case, a simple regular right-
angled solid is produced. The command can also be used to create complex shapes. Because of the
large amount of input required, the POLYHEDRON face command is often best used in conjunction

3DEC Version 5.2

3-8 User’s Guide

with the external preprocessor program PGEN – see Section 4 in Theory and Background for
examples.

E

pGre00

0 u0
i
0G
pGrelG,lrG0p9Ge9p:0

  90 

Figure 3.2 Cubic model created with the POLYHEDRON face command

The POLYHEDRON brick command provides a simpler alternative to the POLYHEDRON face com-
mand when the problem region is a regular six-sided “brick-shaped” region. The parameters for the
brick keyword are the x-, y-, and z-limits of the solid (i.e., the region extends from coordinates xl
to xu in the x-direction, from coordinates yl to yu in the y-direction, and from coordinates zl to zu
in the z-direction). For example, to create the same model generated in Figure 3.2, the command
is:

Example 3.2 A cube generated with the POLYHEDRON brick command

new
poly brick 0,1 0,1 0,1
plot create plot Blocks
plot block
plot reset
plot set dip 120 dd 30
ret

POLYHEDRON cube is a tool for generating irregularly shaped boundaries. These boundaries may
represent geologic contacts or the borders of excavations. The resulting shapes are easy to zone,

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3-9

and can be zoned as mixed discretization zones for plasticity. However, the shapes can be complex
in only two dimensions. See Section 3.2.3.2 for an example of the complex use of this tool.
POLYHEDRON drum creates drum-shaped blocks with parallel ends. This is similar to POLYHEDRON
prism but uses a simpler command syntax.
The POLYHEDRON prism command is an extension of the POLYHEDRON brick command to create
prism-shaped polyhedra. The two parallel faces of the prism are defined by an arbitrary number of
vertices. The opposing vertices on each face are then automatically connected to form the prism.
The first face (face a) is defined by vertices entered in either a clockwise or counterclockwise order.
The opposite face (face b) must have its vertices entered in the same order as the corresponding
vertices for face a. Faces a and b must be planar and convex. The prism shown in Figure 3.3 is
created by the commands listed in Example 3.3:

Example 3.3 An octahedral-shaped prism generated with the POLYHEDRON prism command
new
poly prism &
a (0.0,0.0,0.0 ) (-0.5,0.0,0.87) (-0.5,0.0,1.87) (0.0,0.0,2.74) &
(1.0,0.0,2.74) ( 1.5,0.0,1.87) ( 1.5,0.0,0.87) (1.0,0.0, 0.0) &
b (0.0,4.0,0.0 ) (-0.5,4.0,0.87) (-0.5,4.0,1.87) (0.0,4.0,2.74) &
(1.0,4.0,2.74) ( 1.5,4.0,1.87) ( 1.5,4.0,0.87) (1.0,4.0, 0.0)
plot create plot Blocks
plot block
plot reset
plot set dip 120 dd 20
ret

3DEC Version 5.2

3 - 10 User’s Guide

E

pGre00

0 u0
i
0G
pGrelG,lrG0pPGeP,G0

  P0 

Figure 3.3 An octahedral-shaped prism generated with the POLYHEDRON

prism command

The POLYHEDRON tunnel command is specifically designed to generate a circular-shaped tunnel

model. This command works by constructing the model with individual blocks. This is in contrast
to the TUNNEL command (discussed in Section 3.2.3), which cuts an arbitrarily shaped tunnel out of
an existing block. The blocks created by the POLYHEDRON tunnel command are all six-sided with
low aspect ratios, and are intended for use with the GENERATE quad command. (This command
is recommended for deformable-block plasticity analysis – see Section 3.3.) The user needs only
to specify the orientation, dimensions and number of blocks to be used for the tunnel. Additional
jointing can be added with the JSET command, if desired. For example, to create a tunnel model
with the following dimensions:

radius 2.0 m
length 20.0 m
outside boundary 3.0 r
dip horizontal
heading south
blocks in each octant 1
annular blocks 2
blocks along the axis 3

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 11

Use the POLYHEDRON tunnel command as given in Example 3.4.

Example 3.4 A tunnel model generated with the POLYHEDRON tunnel command
new
poly tunnel rad=2 leng=-10,10 ratr=3.0 dip=0 dd=0 nr=2 nt=1 nx=3
plot create plot Blocks
plot block
plot reset
delete range x -2,2 y -10,10 z -2,2
ret

The model with the tunnel deleted is shown in Figure 3.4:

E

pGre00

0 u0
i
0G
pGrelG,lrG0pPGeP,G0

  P0 
r

Figure 3.4 Tunnel model created with the POLYHEDRON tunnel command

The POLYHEDRON tet command creates a tetrahedral block by specifying the four vertex coordinates.
Vertices can be listed in any order.

3DEC Version 5.2

3 - 12 User’s Guide

3.2.2 Joint Generation

The JSET command is used to make additional cuts in the solids created with the POLYHEDRON
command, to define joints, faults, and holes or excavations. The JSET command can be used to
make single cuts or multiple parallel cuts. Statistical parameters may be used to vary orientation,
spacing and persistence to match logged jointing data.
The JSET command is first demonstrated in this section for making single cuts. Some planning
should be done to optimize the sequence in which joints are created. Joints that define the geometry
of excavations are usually cut first (see Section 3.2.3), followed by the minor joints or joint sets.
Through-going faults are usually defined last.
The primary keywords for the JSET command are dip, dd (dip direction) and origin. Unless other
keywords are used, the JSET command will create a single plane cutting through the model in the
orientation specified. The origin point may be any point on the plane. Figure 3.5 shows how the
orientations for dip and dip direction relate to the coordinate axes in 3DEC. The dip range is from
0 to 90◦ ; the dip direction range is from 0 to 360◦ .

Up (z)

North (y)

= Dip direction
= Dip
lin e
ik e
S tr

East (x)

Joint plane

Figure 3.5 Terms describing the attitude of an inclined plane:

dip angle, α, is positive measured downward from the horizontal
(xy) plane; dip direction, β, is positive measured clockwise from
north (y)

Control of the continuity of cuts made using the JSET command is accomplished with the HIDE and
SEEK commands. The JSET command will only cut blocks that are currently visible.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 13

Example 3.5 illustrates the creation of a noncontinuous joint:

Example 3.5 Creation of a noncontinuous vertical joint

new
poly brick 0,1 0,1 0,1 ; create a block
plot create plot Blocks ; create plot
plot block ; plot block model
plot reset ; reset viewpoint
jset dip 0 dd 0 origin 0,0,.5 ; make a horizontal cut
hide range plane dip 0 dd 0 origin 0,0,.5 below ; hide the bottom block
jset dip 90 dd 90 origin .5,0,0 ; vertical cut through top block only
seek ; make all blocks visible
ret

Figure 3.6 shows the full model and the joint structure plot for this example. Note that the vertical
joint does not penetrate the bottom block. 3DEC automatically assigns a joint ID number = 2 to
the horizontal joint and a joint ID number = 3 to the vertical joint. If desired, the joint ID number
can be controlled with the JSET command. For example,
jset dip 0 dd 0 or 0,.5,0 id = 1000

will create a horizontal joint with an ID number of 1000.

The ID numbers for joint faces and contacts are given sequentially as they are created. Therefore,
if two faults that intersect are defined, the edge to edge contacts at the line of intersection will have
the joint IDs of the second fault defined. This order becomes important if different properties are
to be assigned to the different faults. Property numbers of the face-to-face contacts that make up
most of the area of joints can be assigned using the CHANGE command, in which the particular
joint to be changed is identified using either its joint ID number or orientation. The difficulty
comes in assigning property numbers to the edge-to-edge contacts that are created at joint and fault
intersections. It is easy to assign property numbers to edge-to-edge contacts by use of the joint ID
number; it is difficult to assign property numbers to edge-to-edge contacts by orientation, because
they have a different orientation than either of the two intersecting planes that created them. There
is a PLOT option that allows plotting of joint material properties (the joint material item in the
options menu). This plot is useful for checking that the edge-to-edge contacts are assigned the
correct properties.

3DEC Version 5.2

3 - 14 User’s Guide

E

pGre00

0 u0
i
0G
pGrelG,lrG0pPGeP,G0

  P0 
r

(a) full-solid view

E

pGre00

0 u0
i
0G
pGrelG,lrG0pPGeP,r0


  P0 0
r

(b) joint-structure view

Figure 3.6 Model created with the JSET and HIDE commands

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 15

Concave blocks can be made by using the JOIN command. The blocks that have been joined are
still convex, but the join logic locks the interface between them. For example, add the following
commands at the end of Example 3.5:
hide (0.5,1.0) (0,1) (0.5,1.0)
join on

Figure 3.7 shows the concave block that is created. Note that only visible blocks can be joined.

E

pGre00

0 u0
i
0G
pGrelG,lrG0pPGeP,r0

  P0 
r

Figure 3.7 Concave block created with the JOIN command

Joined blocks are plotted in the same color on the graphics screen. Also, contacts between joined
blocks are identified as master-slave (m-s) contacts. Type LIST contact to check the contact type.
Note that “slaved” blocks will be automatically joined if they are connected to the same “master”
block. For example, if block A and block B are joined, and block A and block C are joined, then
block B will be joined automatically to block C.
The JSET command can also be used to generate a set of joints automatically, based on physically
measured parameters (i.e., joint dip, dip direction, spacing and persistence). By hiding selected
blocks, a set of noncontinuous joints can be generated. In Example 3.6, a jointed rock slope
containing both shallow and deeply dipping joint sets is created. Two noncontinuous fractures are
also created to define a rock wedge in the slope (see Figure 3.8).

3DEC Version 5.2

3 - 16 User’s Guide

Example 3.6 Rock slope containing continuous and noncontinuous joints

new
poly brick 0 80 -30 80 0 50
plot create plot Blocks
plot block
plot reset
plot set dip 110 dd 30
; shallow-dipping fracture planes (continuous)
jset dip 2.45 dd 235 ori 30 0 12.5
jset dip 2.45 dd 315 ori 35 0 30
; high angle foliation planes (continuous)
jset dip 76 dd 270 spac 16 num 3 ori 30,0,12.5
; intersecting discontinuities (non-continuous)
hide range x 0,80 y 0,50 z 0,10
hide range x 55,80 y 0,50 z 0,50
jset dip 70 dd 200 ori 0 35 0
jset dip 60 dd 330 ori 50 15 50
seek
hide range x 0,30 y -30,80 z 13,50
ret

E

pGre00

0 u0
i
0G
pGrelG,lrG0pPGeP,r0

  P0 

Figure 3.8 Rock slope containing continuous and noncontinuous joints

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 17

Bear in mind that joints are displayed as straight-line segments in the 3DEC model; many segments
may be required to fit an irregular joint structure. The modeler must choose a level at which the
3DEC joint geometry will match the physical jointing pattern. The effect of geometric irregularity
on the response of a joint can also be taken into account via the joint material model (e.g., by varying
properties along the joint).
One final point concerning joint generation: When using continuum programs, it is usually ap-
propriate to take advantage of symmetry conditions with excavation shapes, in order to reduce the
size of the model. Symmetry conditions cannot be imposed as easily with discontinuum programs,
because the presence of discontinuous features precludes symmetry except for special cases. For
example, it is not possible to impose a vertical plane of symmetry through the model shown in
Figure 3.8, because the joints in the model are not aligned with the vertical axis.

3.2.3 Creating Internal Boundary Shapes

When fitting the 3DEC model to the problem region, polyhedral boundaries must also be defined
to coincide with boundary shapes of the physical problem. These may be internal boundaries
representing excavations or holes, or external boundaries representing, for example, man-made
structures such as earth dams, or natural features such as mountain slopes. If the physical problem
has a complicated boundary, it is important to assess whether simplification will have any effect
on the questions that need to be answered (i.e., whether a simpler geometry will be sufficient to
reproduce the important mechanisms).
All physical boundaries to be represented in the model simulation (including regions that will be
added, or excavations created at a later stage in the simulation) must be defined before the solution
process begins. Shapes of structures that will be added later in a sequential analysis must be
defined and then “removed” (via the EXCAVATE command). Excavated blocks are added with the
FILL command. Note that only deformable blocks can be excavated and filled.
The creation of boundary shapes is performed with the following commands:
JSET
TUNNEL
POLYHEDRON cube
Each command cuts the polyhedra into one or more segments that are fitted together in the desired
shape. The JSET commands create planar joint segments, as discussed in Section 3.2.2. The TUNNEL
command cuts a shape into the blocks. The POLYHEDRON cube command creates a volume of cubed
blocks, and cuts a boundary.

3DEC Version 5.2

3 - 18 User’s Guide

3.2.3.1 Tunnel Command

The TUNNEL command creates a tunnel whose boundary is formed by planar segments that connect
two faces (designated as face A and face B). The shape of the tunnel face is prescribed by an arbitrary
number of vertices with coordinates (x1,y1,z1), (x2,y2,z2), (x3,y3,z3), etc. The same number of
vertices must exist on both faces; the faces may be positioned either inside or outside of the model.
Example 3.7 presents a simple example for the creation of a horseshoe-shaped tunnel. The command
REMOVE is used to delete the tunnel region (defined as region 1). The DELETE command may also
be used; with the REMOVE command, the deleted region can still be viewed in plot mode, if desired.
The resulting tunnel is shown in Figure 3.9.

Example 3.7 Tunnel created with the TUNNEL command

new
poly brick -1.5,1.5 -1.5,1.5 -1.5,1.5
plot create plot Blocks
plot block colorby material
plot reset
plot set dip 110 dd 20
tunnel radial region 1 &
a (-0.30,-1.5,0.00) (-0.3,-1.5,0.40) (-0.25,-1.5,0.47) &
(-0.15,-1.5,0.52) ( 0.0,-1.5,0.55) ( 0.15,-1.5,0.52) &
( 0.25,-1.5,0.47) ( 0.3,-1.5,0.40) ( 0.30,-1.5,0.00) &
b (-0.30, 1.5,0.00) (-0.3, 1.5,0.40) (-0.25, 1.5,0.47) &
(-0.15, 1.5,0.52) ( 0.0, 1.5,0.55) ( 0.15, 1.5,0.52) &
( 0.25, 1.5,0.47) ( 0.3, 1.5,0.40) ( 0.30, 1.5,0.00)
remove range region 1
ret

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 19

E

pGre00

0 u0
i
0G
pGrelG,lrG0pPGeP,p0

  P0 
r

Figure 3.9 Tunnel created with TUNNEL command

Usually, the model will be brought to an equilibrium state before the tunnel is excavated. The
user must be careful that the fictitious joints along the tunnel boundary do not influence model
response during the initial equilibrium calculation. If the TUNNEL command is used, the blocks are
automatically joined at the fictitious joints. If JSET commands are used to create fictitious joints,
then it is recommended that the JOIN command be used to join the blocks separated by the fictitious
joint. Use the LIST contact command to identify whether the contact is a master-slave (m-s) contact
between joined blocks.
3.2.3.2 POLYHEDRON cube

POLYHEDRON cube is a tool for generating irregularly shaped boundaries. These boundaries may
represent geologic contacts or the borders of excavations. This is intended as an alternative to
the PGEN program. The advantage of POLYHEDRON cube over the PGEN program is that the
resulting shapes are easier to zone, and can be zoned as mixed discretization zones for plasticity.
The disadvantage is that the shapes can be complex in only two dimensions. Using PGEN, the
shapes can be complex in three dimensions.
The POLYHEDRON cube process is relatively simple. The user specifies information about the size
and orientation of the shape to be created. 3DEC then generates an area of six-sided polyhedra
(cubes) that occupies this area. The orientation of the region can be along any line in space, and
can be rotated about that line. After the cubes are created, 3DEC uses the coordinates specified in
a data file (normally “overlay.txt”) to cut the geometry out of the cubes. Each cube is cut only once
by the boundary, so the resolution of the shape is defined by the size of the cubes.
The orientation of the cube area is defined by the dip, ddirection, top and rotate keywords that define
a control plane (see Figure 3.10). All blocks are created behind this control plane. The size and

3DEC Version 5.2

3 - 20 User’s Guide

shape of the cubes are defined by the number and spacing keywords. An outer box can be created as
a zoning transition with the box keyword. The IDs of the jointing are controlled by the id keyword.
The region numbers can be set by the inside and outside keywords.
The file that controls the cutting of the boundary shape is a simple text file of x,y,z coordinates. A
DXF file may also be used, but the coordinates in the DXF file must define a contiguous polygon.
The PGEN program may be used to edit the DXF file to connect discontinuous segments.

top point

rotate nz

ny

y x z

nx

spacing

control plane
defined by dip and
dip direction

Figure 3.10 Elements of the POLYHEDRON cube command

This simple example demonstrates the construction of a 500-cube area that is cut by the polygon
stored in “overlay.txt.”

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 21

Example 3.8 Data file that generates a model using POLYHEDRON cube command
new
;
; example of use of poly cube command
;
config LH
poly brick 0 100 0 100 0 40
plot create plot Blocks
plot block colorby reg
plot reset
poly cube dip 90 dd 180 num 10 10 5 spac 4 4 8 top 50 50 0
seek
hide range reg 1
ret

The cutting geometry is controlled by the data points in the file “overlay.txt”:
31 31 0
31 41 0
37 51 0
41 61 0
51 69 0
63 61 0
63 51 0
47 43 0
; end of segment

3DEC Version 5.2

3 - 22 User’s Guide

E

pGre00

0 u0
i
0G
pGrelG,lrr0rr5,S5re0

  50

G
p

Figure 3.11 Resultant geometry from example

3.2.4 Selecting the Coordinate System

As shown in Figure 3.5, the reference axes for 3DEC are a left-handed set (x,y,z), oriented by
default as x (east), y (north) and z (vertical). In generating a 3DEC model of a given region, the
following conditions may require reorientation for input into the model:
(1) geometry of the problem structures (e.g., mine layout, tunnel locations, etc.);
(2) geometry of the geologic features (e.g., faults and joint sets); and
(3) orientation of the in-situ stress field.
In the ideal situation, the problem geometries would align with the 3DEC reference axes. In general
though, this is not the case, and one or two of the problem geometries will require transformation
to the 3DEC model reference frame.
Typically, it is best to orient the 3DEC model axes to align with the geometry of the problem
structures (e.g., the mine grid or centerline of a tunnel). For graphic presentation, it is best to
position the origin of the model axes at the center of the structure for which the analysis is intended.
In this case, the geologic features may require reorientation from global to local problem axes (see
Section 3.2.5, below), and the field principal stresses may require transformation for application to
the model boundary. The recommended procedure to execute this transformation is discussed in
Section 3.5.8.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 23

For reasons of numerical precision, it is best to truncate coordinates to some convenient number.
For example, if the units for a model are from x = 15,423 to 15,443, it is best to truncate the
coordinates by 15,400 so that the model coordinates range from 23 to 43.

3.2.5 Orientation of Geologic Features to the Model Axes

If the problem axes do not align with the model axes (i.e., positive z-axis (vertical), x-axis (east)
and y-axis (north)), then the orientation of geologic features will have to be transformed from the
problem axes to the model axes. This can be accomplished by using a stereonet.
The reorientation of structural features to the model reference axes is demonstrated for the case
of a single fault that crosses a mine raise. The raise is oriented with an axis dip of 84◦ and dip
direction of 125◦ , and the fault is oriented 11◦ / 148◦ . The 3DEC model axes are located with the
z-axis directed down the raise, the y-axis lying in the vertical plane containing the raise dip vector,
and the x-axis lying in the horizontal plane. The origin of the model axes is located at the center of
the raise (see Figure 3.12). The dip and dip direction of the fault must be redefined relative to the
x,y,z-model axes as oriented in the figure.

UP

ventilation
raise Raise Orientation
> = 125º
= = 84º
>
E
=
N35º
x
Model Axes

y
z

Figure 3.12 Orientation of 3DEC model axes (x,y,z) relative to north-east-

vertical reference axes

3DEC Version 5.2

3 - 24 User’s Guide

The fault pole is plotted on the lower hemisphere stereonet shown in Figure 3.14. The relation
between the fault and the model axes is found by plotting the positive x-, y- and z-axes of the model
on the stereonet. The dihedral angle between the fault pole and each axis is then read from the
stereonet. In this example, the dihedral angles are 86◦ for x-axis to fault pole, 73◦ for z-axis to fault
pole, and 17◦ for y-axis to fault pole.
On a second stereonet (Figure 3.13), the model axes are oriented to align with the axes of the
stereonet: z (vertical), x (east) and y (north). The intersection of the three dihedral angles on this
plot gives the pole of the fault (plotted on the upper hemisphere). The orientation of the fault relative
to the model axes is thus 74◦ / 2◦ .

3.2.6 Choice of Model Scale

Analysis of rock mass response involves several different scales. It is impossible (and undesirable)
to include all features, details or rock mass response mechanisms in one model. It is also well-
recognized that the location of the far-field boundary in the model can have a significant influence
on results obtained for underground excavations. However, in three-dimensional analysis, it is
not always feasible to place boundaries sufficiently far from the excavations to avoid adversely
affecting the results. These two observations taken together suggest that a reasonable modeling
approach involves starting with a large global model and proceeding through reasonable models to
the smallest size required, with increasing complexity and detail added at each stage. 3DEC has an
automatic method of recording stresses at specified locations so that they can be applied as boundary
tractions on smaller problems. This unique feature ensures stress compatibility between larger and
smaller models. See Section 3.4.4.2 for additional information on applying this technique.

3.2.7 Incorporation of Discontinuities

Selection of joint geometry for input to a model is a crucial step in distinct element analysis.
Typically, only a very small percentage of joints can actually be included in the model, so that
models of reasonable size and execution speed for practical analysis can be created. Thus, the
modeler must filter joint geometry data, and select only those joints that are most critical to the
mechanical response by identifying those that are most susceptible to slip for the prescribed loading
conditions. This may involve, for example, determining whether sufficient kinematic freedom is
provided (e.g., through the use of block theory) or comparing in-situ observations and records (e.g.,
microseismic records to identify key joints).
Once a consistent set of joints is selected, the geometric parameters (e.g., strike, dip, location) for
these features are input into the model, and the practicality of the analysis in terms of required
memory and runtime are assessed. If the model size is too large, the number of joints must be
reduced, and it is necessary to further filter the input to bring the model to a practical size. This
dilemma of balancing model size and critical joint structures is addressed by Hart (1993).
In some cases, exactly how important certain discontinuities are, or whether discontinuities should
be incorporated into the model at all, may not be clear. The guiding philosophy in these cases
is to first run the models either with no discontinuities, or with the discontinuities in a welded
state. Comparisons should then be made with observations. If adequate calibration of the model

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 25

behavior can be obtained without discontinuities, then discontinuities need not be modeled. By
rerunning the model and allowing certain discontinuities to slip, the modeler can determine exactly
how the discontinuities affect the system behavior, and whether that behavior agrees more closely
with observations.

3DEC Version 5.2

3 - 26 User’s Guide

x-axis

86º
73º

pole to fault
17º E
z-axis

fault plane

Figure 3.13 Stereonet plot of fault relative to model axes

z
17º

73º
fault 86º
pole on
upper hemisphere

fault plane

74º/2º

y (up)

fault
pole on
lower hemisphere

Figure 3.14 Stereonet plot of pole to fault and model reference axes relative
to problem north-east axes

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 27

3.3 Selection of Deformable versus Rigid Blocks

An important consideration when doing a discontinuum analysis is whether to use rigid or de-
formable blocks to represent the behavior of intact material. The considerations for rigid versus
deformable blocks are discussed in this section. If a deformable block analysis is required, there are
several different models available to simulate block deformability; these are discussed in Section 3.7.
As mentioned in Section 1.1 in Theory and Background, early distinct element codes assumed
that blocks were rigid. However, the importance of including block deformability has become
recognized, particularly for stability analyses of underground openings and studies of seismic
response of buried structures. One of the most obvious reasons to include block deformability in a
distinct element analysis is the requirement to represent the “Poisson’s ratio effect” of a confined
rock mass.

3.3.1 Poisson’s Effect

Rock mechanics problems are usually very sensitive to the Poisson’s ratio chosen for a rock mass.
This is because joints and intact rock are pressure-sensitive; their failure criteria are functions of
the confining stress (e.g., the Mohr-Coulomb criterion). Capturing the true Poisson behavior of a
jointed rock mass is critical for meaningful numerical modeling.
The effective Poisson’s ratio of a rock mass comprises two parts: a component due to the jointing;
and a component due to the elastic properties of the intact rock. Except at shallow depths or
low confining stress levels, the compressibility of the intact rock makes a large contribution to
the compressibility of a rock mass as a whole. Thus, the Poisson’s ratio of the intact rock has a
significant effect on the Poisson’s ratio of a jointed rock mass.
Strictly speaking, a single Poisson’s ratio, ν, is defined only for isotropic elastic materials. However,
there are only a few jointing patterns that lead to isotropic elastic properties for a rock mass.
Therefore, it is convenient to define a “Poisson effect” that can be used for discussion of anisotropic
materials.
The Poisson effect will be defined as the ratio of horizontal-to-vertical stress when a load is applied
in the vertical direction and no strain is allowed in the horizontal direction; plane-strain conditions
are assumed. As an example, the Poisson effect for an isotropic elastic material is

σxx ν
= (3.1)
σzz 1−ν

Consider the Poisson effect produced by the vertical jointing pattern shown in Figure 3.15. If this
jointing were modeled with rigid blocks, applying a vertical stress would produce no horizontal
stress at all. This is clearly unrealistic, because the horizontal stress produced by the Poisson’s ratio
of the intact rock is ignored.

3DEC Version 5.2

3 - 28 User’s Guide

yy

yy
Figure 3.15 Model for Poisson’s effect in rock with vertical and horizontal
jointing

The joints and intact rock act in series. In other words, the stresses acting on the joints and on the
rock are identical. The total strain of the jointed rock mass is the sum of the strain due to the jointing
and the strain due to the compressibility of the rock. The elastic properties of the rock mass as a
whole can be derived by adding the compliances of the jointing and the intact rock:

    
xx rock jointing σxx
= C +C (3.2)
yy σyy

If the intact rock were modeled as an isotropic elastic material, its compliance matrix would be

 
1+ν 1−ν −ν
C rock = (3.3)
E −ν 1−ν

The compliance matrix due to the jointing is

 1 
Skn 0
C jointing = (3.4)
1
0 Skn

where S is the joint spacing, and kn is the normal stiffness of the joints.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 29

If xx = 0 in Eq. (3.2), then

(total)
σxx C
= − 12
(total)
(3.5)
σyy C 11

where C (total) = C (rock) + C (j ointing) .

Thus, the Poisson effect for the rock mass as a whole is

σxx ν (1 + ν)
= (3.6)
σyy E/(Skn ) + (1 + ν)(1 − ν)

Eq. (3.6) is graphed as a function of the ratio E/(Skn ) in Figure 3.16. Also graphed are the results
of several two-dimensional UDEC simulations run to verify the formula. The ratio E/(Skn ) is a
measure of the stiffness of the intact rock in relation to the stiffness of the joints. For low values
of E/(Skn ), the Poisson effect for the rock mass is dominated by the elastic properties of the intact
rock. For high values of E/(Skn ), the Poisson effect is dominated by the jointing.
Now consider the Poisson effect produced by joints dipping at various angles. The Poisson effect
is a function of the orientation and elastic properties of the joints. Consider the special case shown
in Figure 3.17. A rock mass contains two sets of equally spaced joints dipping at an angle, θ , from
the horizontal. The elastic properties of the joints consist of a normal stiffness, kn , and a shear
stiffness, ks . The blocks of intact rock are assumed to be completely rigid.

UDEC Simulations
0.4
Analytic Solution

0.3
yy

0.2
xx

0.1

0
0 0.5 1 1.5 2 2.5 3
E
SK n
Figure 3.16 Poisson’s effect for vertically jointed rock
(ν = 0.3 for intact rock)

3DEC Version 5.2

3 - 30 User’s Guide

yy

S
O

yy

Figure 3.17 Model for Poisson’s effect in rock with joints dipping at angle θ
from the horizontal and with spacing S

The Poisson effect for this jointing pattern is

σxx cos2 θ [(kn / ks ) − 1]

= (3.7)
σyy sin2 θ + cos2 θ (kn / ks )

This formula is illustrated graphically for several values of θ in Figure 3.18. Also shown are the
results of numerical simulations using UDEC. The UDEC simulations agree closely with Eq. (3.7).

0.8

0.6
yy
xx

0.4
Analytic Solution UDEC Simulation
o
O = 20
o
0.2 O = 45
o
O = 60

0
2 4 6 8 10
Kn
Ks

Figure 3.18 Poisson’s effect for jointed rock at various joint angles (blocks
are rigid)

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 31

Eq. (3.7) demonstrates the importance of using realistic values for joint shear stiffness in numerical
models. The ratio of shear stiffness to normal stiffness dramatically affects the Poisson response
of a rock mass. If shear stiffness is equal to normal stiffness, the Poisson effect is zero. For more
reasonable values of kn /ks , from 2.0 to 10.0, the Poisson effect is quite high, up to 0.9.
Next, the contribution of the elastic properties of the intact rock will be examined for the case of
θ = 45◦ . Following the analysis for the vertical jointing case, the intact rock will be treated as an
isotropic elastic material. The elastic properties of the rock mass as a whole will be derived by
adding the compliances of the jointing and the intact rock.
The compliance matrix due to the two equally spaced sets of joints dipping at 45◦ is
 
1 ks + kn ks − kn
C (j ointing)
=
2S kn ks ks − kn ks + kn

Thus, the Poisson effect for the rock mass as a whole is

σxx [ν(1 + ν)] / E + (kn − ks ) / (2S kn ks )

= (3.8)
σyy [(1 + ν)(1 − ν)] / E + (kn + ks ) / (2S kn ks )

Eq. (3.8) is graphed for several values of the ratio E/(Skn ) for the case of ν = 0.2 (see Figure 3.19).
Also plotted are the results of UDEC simulations. For low values of E/(Skn ), the Poisson effect of
a rock mass is dominated by the elastic properties of the intact rock. For high values of E/(Skn ),
the Poisson effect is dominated by the jointing.

0.8 E SK n
= )
Rigid
Block
s (
0.6
yy

0.5
E SK n =
xx

0.4

Rock with No Joints ( E SK n

= 0 )
0.2 UDEC Simulation

0
2 4 6 8 10
Kn
Ks

Figure 3.19 Poisson’s effect for rock with two equally spaced joint sets
with θ = 45◦ (blocks are deformable with ν = 0.2)

3DEC Version 5.2

3 - 32 User’s Guide

3.3.2 Zoning for Deformable Blocks

If a deformable block analysis is required, models with different densities of block zoning should
be evaluated once the block cutting and boundary location have been established. The GENERATE
edge and GENERATE quad commands are used to specify the zoning density. The highest density
of zoning should be in regions of high stress or strain gradients (e.g., in the vicinity of excavations).
For greatest accuracy, the aspect ratio of zone dimensions (i.e., tetrahedron base length to height
ratio) should also be as near unity as possible; anything above 5:1 is potentially inaccurate. Neither
is it advisable to have large jumps in zone size between adjacent polyhedra. The ratio between
zone volumes in adjacent polyhedra should not exceed roughly 4:1 for reasonable accuracy. Use
the LIST max command to find the maximum and minimum zone volumes.
The GENERATE edge command will automatically create tetrahedral zones within an arbitrarily
shaped convex polyhedron. If block-cutting results in blocks that are long and thin, it is recom-
mended that these blocks be further cut and joined before generating zones. By doing this, zones
with an aspect ratio closer to unity can be generated.
The GENERATE quad command will only generate zones within six-sided polyhedra. This command
creates mixed-discretization (m-d) zones (two overlays of five tetrahedral zones) that provide better
accuracy for problems involving failure and collapse of the intact blocks. (See Section 1.2.2.5 in
Theory and Background for details.) It is recommended that the GENERATE quad command be
used for analyses involving plastic failure of the intact material. The GENERATE edge command
can provide reasonable accuracy for certain failure modes (e.g., confined compression loading);
however, this type of zoning does not produce an accurate prediction for collapse loads in bearing
capacity problems. Comparisons of results using GENERATE quad versus GENERATE edge for
plasticity analysis are given in Sections 5 and 6 in the Verifications volume.
The GENERATE edge command produces zoning that is more computationally efficient, and is
recommended for blocks in regions where extensive intact material failure is not anticipated. In order
to improve the calculational efficiency for models involving intact material failure, the GENERATE
quad command can be used to generate m-d zones around an excavation, and the GENERATE edge
command can then be used to generate zones in blocks at greater distance from the excavation (or
for surrounding blocks that are not six-sided).
The GENERATE hotetra command produces high-order tetrahedral zones that can be used in blocks
that cannot be zoned with m-d zoning. These zones have additional gridpoint nodes and are
more accurate than the standard tetrahedral zoning. Note that high-order tetrahedral zones are not
compatible with the far-field dynamic boundary or the zone models.
The analysis of large models can also be aided by using the GENERATE center command, which
is a variation of the GENERATE edge command. This command allows the sizes of the tetrahedral
zones to be increased gradually, outward from a central point.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 33

3.4 Boundary Conditions

The boundary conditions in a numerical model consist of the values of field variables (e.g., stress,
displacement) that are prescribed at the boundary of the model. Boundaries are of two categories:
real and artificial. Real boundaries exist in the physical object being modeled (e.g., a tunnel surface
or the ground surface). Artificial boundaries do not exist in reality, but they must be introduced in
order to enclose the chosen number of elements (e.g., blocks). The conditions that can be imposed on
each type are similar; these conditions are discussed first. Then (in Section 3.4.4), some suggestions
are made concerning the location and choice of artificial boundaries, and the effect they have on
the solution.
Mechanical boundaries are of two main types: prescribed displacement and prescribed stress. A
free surface is a special case of the prescribed-stress boundaries. The two types of mechanical
boundaries are described in Sections 3.4.1 and 3.4.2. Viscous boundaries, which are used for
dynamic analysis, are described in Section 2 in Optional Features.

3.4.1 Stress Boundary

By default, the boundaries of a 3DEC model are free of stress and any constraint. Forces or stresses
may be applied to any boundary, or part of a boundary, with the BOUNDARY command. Note that
forces and stresses can be applied to either rigid or deformable blocks. Individual components
of the stress tensor (σxx , σyy , σzz , σxy , σxz and σyz ) are specified with the stress keyword. For
example, the command
boundary (0,10) (-1,1) (0,10) stress 0,0,-1e6 0,0,0

would apply σxx = 0, σyy = 0, σzz = −106 and zero shear stresses to a model boundary lying
within the coordinate window 0 < x < 10, −1 < y < −1, 0 < z < 10. The user should always
make sure that the window encompasses all of the boundary vertices designated for the assigned
boundary condition. This can be done using the command
list boundary state

Each exterior boundary vertex will be listed with its assigned boundary code. (See the LIST boundary
command in Section 1.3 in the Command Reference.) The boundary can move during a model
calculation, so the user must make sure that the coordinate window is large enough to include the
appropriate boundary vertices at the time the BOUNDARY command is executed.
Alternatively, boundary conditions can be specified along a boundary defined by the orientation of
the boundary face. For example, the same boundary condition as above can be applied with the
command
boundary dip 90 dd 180 or 0,1,0 above stress 0,0,-1e6 0,0,0

This will apply the boundary condition along the boundary face located within the range defined
by a plane with a dip angle of 90◦ , a dip direction of 270◦ , and above the position x = 0, y = 1, z =
0.

3DEC Version 5.2

3 - 34 User’s Guide

Compressive stresses have a negative sign, in accordance with the general sign convention for
internal stresses in 3DEC. Also, 3DEC actually applies stress components as forces, or tractions,
which result from a stress tensor acting on the given boundary plane. The tractions are divided into
two components, permanent and transient. Permanent tractions are constant loads, and transient
tractions are time-varying loads applied for dynamic analysis (see Section 2 in Optional Features)
by using the history keyword on the same command line as the stress keyword. Various forms
of time-varying histories can be applied, including linear-varying, sine and cosine wave, and user-
supplied functions; these are described in Section 1.3 in the Command Reference (see BOUNDARY
history).
Individual forces can be applied to the model boundary of rigid or deformable blocks by using the
xload, yload and zload keywords, which specify x-, y- and z-components of an applied force vector.
3.4.1.1 Applied Stress Gradient

The BOUNDARY command may take additional keywords xgrad, ygrad and zgrad, which allow the
applied stresses or forces to vary linearly over the specified range. Six parameters follow each of
these keywords and describe the variation of the stress components in the x-, y- or z-direction:
xgrad sxxx syyx szzx sxyx sxzx syzx
ygrad sxxy syyy szzy sxyy sxzy syzy
zgrad sxxz syyz szzz sxyz sxzz syzz
The stresses vary linearly with distance from the global coordinate origin of x = 0, y = 0, z = 0:

◦
σxx = σxx + (sxxx)x + (sxxy)y + (sxxz)z
◦
σyy = σyy + (syyx)x + (syyy)y + (syyz)z
◦
σzz = σzz + (szzx)x + (szzy)y + (szzz)z
◦
σxy = σxy + (sxyx)x + (sxyy)y + (sxyz)z (3.9)
◦
σxz = σxz + (sxzx)x + (sxzy)y + (sxzz)z
◦
σyz = σyz + (syzx)x + (syzy)y + (syzz)z

◦ , σ ◦ , σ ◦ , σ ◦ , σ ◦ and σ ◦ are the stress components at the origin.

where σxx yy zz xy xz yz

The operation of this feature is best explained by an example:

boundary -.1,.1 0,10 -100,0 stress 0,0,-10e6 0,0,0 zgrad 0,0,1e5 0,0,0

3DEC Version 5.2

 PROBLEM SOLVING WITH 3DEC 3 - 35

The stresses at the origin (x = 0, y = 0, z = 0) are:

◦
σxx = 0
◦
σyy = 0
◦
σzz = −10 × 106
◦
σxy = 0
◦
σxz = 0
◦
σyz = 0

The equation for the z-variation in stress component σzz is

σzz = −10 × 106 + (105 )z

The value for σzz at z = −100 is then −20 × 106 . At points in between, the z-variation is linearly
scaled to the relative z-distance from the origin.
Typically, applied stress gradients are used to reproduce the effects of increasing stress with depth
caused by gravity. It is important to make sure that the applied gradient is compatible with the
gradient specified with the INSITU command and the value of gravitational acceleration (GRAVITY
command). Section 3.5 provides more details on this matter.
3.4.1.2 Changing Boundary Stresses

As discussed above, transient loading can be performed with the history keyword for dynamic
analysis. For static analysis, it may also be necessary to alter the values of applied stresses during
the course of a 3DEC simulation. For example, the load on a footing may change. To effect a
sudden change in an existing applied stress or load, a new BOUNDARY command is given with the
range that encompasses the same boundary vertices as in the original command, but with a change
in stress value or variation.
In this case, the new value will be added to the existing value.* If the stress is to be removed,
the current value should be given with an opposite sign. If a transient load is changed (i.e., a load
assigned with the history keyword), any new load with the same history type is added to the existing
load; however, a new transient load with a different history type replaces the old transient load.

* The user should be aware that this approach is different than the one used in the Itasca code FLAC,
in which the stresses are updated to the new values when a new boundary condition is specified.

3DEC Version 5.2

3 - 36 User’s Guide

3.4.1.3 Checking the Boundary Condition

The boundary stresses and loads may be verified with the command LIST bound. The LIST bound
command lists the boundary vertex addresses, along with current values and conditions assigned to
each vertex. Once a BOUNDARY command is issued, a boundary vertex list is created for the model
face to which the boundary condition is assigned. Optional keywords can be used with the LIST
bound command to check the different conditions along the boundary. For example,
list bound force

lists the permanent forces (fx,fy,fz) and incremental forces (fxi,fyi,fzi) added during the current
loading stage. If transient loads are applied (with the BOUNDARY. . . hist command), the total
forces refer to the permanent plus transient loads at the current cycle number.
The command
list bound state

identifies the type of boundary condition assigned to a boundary vertex.

3.4.1.4 Cautions and Advice

In this section, some miscellaneous difficulties with stress boundaries are described. With 3DEC, it
is possible to apply stresses to the boundary of a body that has no displacement constraints (unlike
many finite element programs, which require some constraints). The body will react in exactly the
same way as a real body would (i.e., if the boundary stresses are not in equilibrium, then the whole
body will start moving).
A similar, but more subtle, effect arises when material is excavated from a body that is supported
by a stress boundary condition: the body is initially in equilibrium under gravity, but the removal
of material reduces the weight. The whole body then starts moving upward, as demonstrated in
Example 3.9 and illustrated in Figure 3.20.

Example 3.9 Uplift when material is removed

new
poly brick 0,10 0,10 0,10
plot create plot Blocks
plot block axes
plot reset
plot set dip 110 dd 330
;
jset dip 0 dd 180 origin 0,0,5
gen quad ndiv 4 4 4
;
change cons 1
prop mat=1 dens 1000 bulk 8e9 g 5e9
prop jmat=1 jkn 1e10 jks 1e10

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 37

gravity 0,0,-10
;
bound stress 0,0,-1e5 0,0,0 range z 0.0
bound xvel = 0.0 range x 0.0
bound xvel = 0.0 range x 10.0
bound yvel = 0.0 range y 0.0
bound yvel = 0.0 range y 10.0
insitu stress 0,0,-1e5 0,0,0 zgrad 0,0,1e4 0,0,0
;
hist zdisp 5,5,2.5
step 300
;
excavate range z 5,10
step 100
;
plot clear
plot block colorby mat fill off axes scale = 10
plot add vel line color red
ret

E

pGre007

0 u0
i
0.GG
pGrelG,lrG0p:Ge:,,0

7  :0 
r
 
o 

 :0GiGGS-4S
:0e,i-p33

Figure 3.20 Uplift when material is removed

The difficulty encountered in running this data file can be eliminated by fixing the bottom boundary,
rather than supporting it with stresses. Section 3.4.4 contains information relating to the location
of such artificial boundaries.

3DEC Version 5.2

3 - 38 User’s Guide

Finally, the stress boundary affects all degrees-of-freedom. Velocity boundary conditions must,
therefore, be prescribed after stress boundary conditions affecting the same boundary corners. If
the stress boundary is applied after the velocity boundary, the effect of the prescribed velocity will
be lost. Example 3.10 demonstrates this problem:

Example 3.10 Mixing stress and velocity boundary conditions

new
poly brick 0,10 0,10 0,10
plot create plot Blocks
plot block
plot reset
;
gen quad ndiv 4 4 4
;
prop mat=1 dens 1000 bulk 8e9 g 5e9
bound zvel = 0.0 range z 0.0
bound stress -1e5,0,0 0,0,0 range x 0.0
bound stress -1e5,0,0 0,0,0 range x 10.0
bound stress 0,-1e5,0 0,0,0 range y 0.0
bound stress 0,-1e5,0 0,0,0 range y 10.0
bound stress 0,0,-2e5 0,0,0 range z 10.0
;
hist zdisp 0,0,0
step 100
;
plot clear
plot axes scale = 10 block fill off vel line color blue
ret

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 39

E

pGre004

0 u0
i
0rGG
pGrelG,lrG0pPGeP,,0

4  P05 
 
o 

 P0GiGGeS.SGp
P0eeirS:

Figure 3.21 Mixing stress and velocity boundary conditions

The fixed z-velocity boundary condition along the bottom boundary of the model is removed along
the bottom edges of the block when the stress boundaries are applied. These points move downward,
as indicated by the velocity vector plot in Figure 3.21, when the model is loaded.

3.4.2 Displacement Boundary

Displacements cannot be controlled directly in 3DEC; in fact, they play no part in the calculation
process, as explained in Section 1.2.2 in Theory and Background. In order to apply a given
displacement to a boundary of a deformable block model, it is necessary to fix the boundary and
prescribe the boundary’s velocity for a given number of steps (using the BOUNDARY command). If
the desired displacement is D, a velocity, V , is applied for a time increment, T (e.g., D = V T ),
where T = tN , t is the timestep and N is the number of steps (or cycles). In practice, V should
be kept small and N large, in order to minimize shocks to the system being modeled.
The BOUNDARY command is used to fix the velocity of gridpoints of deformable blocks in the x-,
y- or z-direction (BOUNDARY xvel, yvel or zvel) or in the normal direction (BOUNDARY nvel) along
boundaries not aligned with the x-, y- and z-axes. The velocity of rigid or deformable blocks can
be fixed with the FIX command (at the current velocity). Use the APPLY command to specify a
velocity other than the current value for rigid blocks. The velocity can also be altered with a FISH
function.
Time-varying velocity histories can be applied via the BOUNDARY . . . hist command for deformable
blocks or the APPLY . . . hist command for rigid blocks. This history keyword must appear on the

3DEC Version 5.2

3 - 40 User’s Guide

same line as BOUNDARY xvel, BOUNDARY yvel or BOUNDARY zvel to prescribe a velocity history.
Histories can also be applied as FISH functions. As discussed in Section 3.4.1.4, velocity boundaries
should always be assigned after stress boundaries. Fixed velocity conditions can be removed for
deformable blocks with the BOUNDARY xfree, BOUNDARY yfree or BOUNDARY zfree command, and
for rigid blocks with the FREE command.

3.4.3 Real Boundaries – Choosing the Right Type

It is sometimes difficult to know the type of boundary condition to apply to a particular surface on
the body being modeled. For example, in modeling a laboratory triaxial test, should the load applied
by the platen be regarded as a stress boundary, or should the platen be treated as a rigid displacement
boundary? Of course, the whole testing machine, including the platen, could be modeled, but that
might be very time-consuming. Remember that 3DEC takes a long time to converge if there is a
large contrast in stiffnesses. In general, if the object applying the load is very stiff compared with
the sample (say, more than 20 times stiffer), then it may be treated as a rigid boundary. If it is
soft compared with the sample (say, 20 times softer), then it may be modeled as a stress-controlled
boundary. Clearly, a fluid pressure acting on the surface of a body is in the latter category. Footings
on jointed rock can often be represented as rigid boundaries that move with constant velocity, for
the purposes of finding the collapse load of the rock. This approach has another advantage: it is
much easier to control the test and obtain a good load/displacement graph. It is well-known that
stiff testing machines are more stable than soft testing machines.

3.4.4 Artificial Boundaries

Artificial boundaries fall into two categories: planes of symmetry and planes of truncation. Sym-
metry planes take advantage of symmetry conditions in a physical system; truncation planes are
needed when modeling an infinite or very large system.
3.4.4.1 Symmetry Planes

Sometimes it is possible to take advantage of the fact that the geometry and loading in a system
are symmetrical about one or more planes. For example, if everything is symmetrical about a
vertical (yz) plane, then the horizontal displacements on that plane will be zero. Therefore, we can
make that plane a boundary, and fix all gridpoints in the horizontal direction, using the command
BOUNDARY xvel=0. If velocities on the plane of symmetry are not already zero, they will be set to
zero with this command. In the case considered, the y-component and z-component of velocity on
the vertical plane of symmetry are not affected; they should not be fixed. Similar considerations
apply to a horizontal plane of symmetry. The command BOUNDARY nvel=0 can be used to set planes
of symmetry that lie at angles to the coordinate axes.
As discussed in Section 3.2.3, the presence of discontinuities makes the application of symmetry
planes more difficult. When using symmetry planes in 3DEC, the modeler should always be careful
to consider the effect of joint orientation.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 41

3.4.4.2 Boundary Truncation – Location of the Far-Field Boundary

Analysis of rock-mass response involves several different scales. It is impossible (and undesirable)
to include all features, details or rock-mass response mechanisms in one model. It is also well-
recognized that location of the far-field boundary in the model can have a significant influence
on results obtained for underground excavations. However, in three-dimensional analysis, it is
not always feasible to place boundaries sufficiently far from the excavations to avoid adversely
affecting the results. These two observations, taken together, suggest that a reasonable modeling
approach involves starting with a large global model and proceeding through reasonable models
to the smallest size required, with increasing complexity and detail added at each stage. 3DEC
has an automatic method of recording stresses at specified locations so that they can be applied as
boundary tractions on smaller problems. This unique feature ensures stress compatibility between
larger and smaller models.
An example of the application of this technique is a mine in which the local topographic relief is
significant relative to the size of the mine, and in which a model incorporating both the surrounding
topography and the mine geometry is prohibitively large. A coarse regional model incorporating
a relatively large sample of the surrounding topography is first created, as shown in Figure 3.22.
Following this, a smaller central portion of the regional model is generated, incorporating details of
the mining geometry, which is more finely zoned. Boundary tractions applied to the detailed model
are transferred from the regional model at points corresponding to the locations of the detailed
model boundaries by the following procedure:
1. Run Model A.
2. Enter the command LIST brick xl xu yl yu zl zu force file filename, where (xl,
xu), (yl, yu) and (zl, zu) correspond to the boundaries of Model B.
3. The loads printed to the filename for the LIST brick command can be applied
to Model B by using the CALL filename command.
An important modeling decision to be made when using this procedure is how extensive a volume
should be incorporated into each of the regional and detailed models.
In the regional model, the topographic relief should be small relative to the external dimensions
of the model. A criterion recommended for selecting the depth of the model is that stresses near
the bottom of the model should be relatively uniform (i.e., it should not be possible to determine
whether a point is beneath a mountain peak or a valley); otherwise, the finite depth of the model
will influence the stress field near the surface. Analyses performed using two-dimensional models
indicate that to satisfy this criterion, the thickness of the model beneath the valleys should be
approximately three times the topographic amplitude.

3DEC Version 5.2

3 - 42 User’s Guide

Model B

Model A

Figure 3.22 Models used to transfer stress boundary conditions

Boundary locations for the detailed model are chosen using the conventional criterion that induced
stresses at the boundary location caused by internal changes (such as mining) should not be sig-
nificant. An estimate of this distance can be obtained using the elastic solution for stress around a
t , at a distance
sphere in a hydrostatic stress field. The induced tangential component of stress, σind
R from a cavity of radius a in a hydrostatic stress field po is

t
σind a3
= (3.10)
po 2 R3

Because the decay of induced stress away from the excavation is more rapid than for a circular hole,
the two-dimensional equivalent boundaries around three-dimensional excavations do not have to
be particularly far. For example, at a distance of approximately 2a, the induced stress is only 5%
of the hydrostatic level, and only 1% at a distance of 3.7a. Solutions for other shaped openings can
also be used, depending on the actual excavation shape.
An important aspect of the point-wise boundary traction transfer to the smaller model is that a more
complex boundary stress distribution can be generated than by conventional linear variations. The
dimensions of boundaries in the detailed model can, therefore, be small relative to the surrounding
topographic relief. This would be impossible to achieve using standard linearly varying boundary
stress distributions.
An alternative to the stress transfer technique can be used when the entire model is not too big to
fit in memory, but runs too slowly to allow investigation of such things as strength variations or

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 43

extraction sequences. In this case, the COUPLE command may be used. The COUPLE command
allows the decoupling of an inner and outer region. After reaching internal equilibrium, the large
mass shown in Figure 3.22 (Model A) is defined as a single region (for example, region 1 – see the
MARK command).
The outer area can be removed from calculation cycles with the COUPLE 1 off command. From this
point, 3DEC does not include the blocks that lie in region 1 in the cycle calculations. The model
forces at the interface between region 1 and all other regions are held constant. Changes in the
inner model that affect the interface will not cause a change in the forces at the boundaries unless
an outer region is turned back on (COUPLE 1 on). The status of any region can be determined with
the LIST region command.

3DEC Version 5.2

3 - 44 User’s Guide

3.5 Initial Conditions

In all civil or mining engineering projects, there is an in-situ state of stress in the ground, before any
excavation or construction is started. By setting initial conditions in the 3DEC model, an attempt is
made to reproduce this in-situ state, because it can influence the subsequent behavior of the model.
Ideally, information about the initial state comes from field measurements. But, when these are not
available, the model can be run for a range of possible conditions. Although the range is potentially
infinite, there are a number of constraining factors (e.g., the system must be in equilibrium, and the
chosen yield and slip criteria must not be violated anywhere).
In a uniform layer of soil or rock with a free surface, the vertical stresses are usually equal to gρz,
where g is the gravitational acceleration, ρ is the mass density of the material and z is the depth below
surface. However, the in-situ horizontal stresses are more difficult to estimate. There is a common
(but erroneous) belief that there is some “natural” ratio between horizontal and vertical stress, given
by ν/(1 − ν), where ν is the Poisson’s ratio. This formula is derived from the assumption that
gravity is suddenly applied to an elastic mass of material in which lateral movement is prevented.
This condition hardly ever applies in practice, due to repeated tectonic movements, material failure,
overburden removal and locked-in stresses due to faulting and localization (see Section 3.11.3). Of
course, if we had enough knowledge of the history of a particular volume of material, we might
simulate the whole process numerically, in order to arrive at the initial conditions for our planned
engineering works. This approach is not usually feasible. Typically, we compromise: a set of
stresses is installed in the model, and then 3DEC is run until an equilibrium state is obtained. It is
important to realize that there are an infinite number of equilibrium states for any given system.
In the following sections, we examine progressively more complicated situations, and the ways
in which the initial conditions may be specified. The user is encouraged to experiment with the
various data files that are presented.

3.5.1 Uniform Stresses in an Unjointed Medium: No Gravity

For an excavation deep underground, the gravitational variation of stress from top to bottom of the
excavation may be neglected because the variation is small in comparison with the magnitude of
stress acting on the volume of rock to be modeled. The GRAVITY command may be omitted, causing
the gravitational acceleration to default to zero. The initial stresses are installed with the INSITU
command – e.g.,
insitu stress -5e6 -5e6 -1e7 0.0 0.0 0.0

The components σ11 (or σxx ), σ22 (or σyy ) and σ33 (or σzz ) are set to compressive stresses of 5×106 ,
5 × 106 and 5 × 107 , respectively, throughout the model. Range parameters may be added if the
stresses are to be restricted to a subregion of the model. The INSITU command sets all stresses to
the given values, but there is no guarantee that the stresses will be in equilibrium. There are at least
three possible problems. First, the stresses may violate the yield criterion of a nonlinear constitutive
model assigned to deformable blocks. In this case, plastic flow of zones in the blocks will occur
immediately after the STEP command is given, and the stresses will readjust; this possibility should
be checked by doing one trial step and examining the response (e.g., PLOT plas). Second, the stress

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 45

state may result in slip or separation along joints within the model. The command PLOT vel should
indicate locations where joint movement is occurring. Third, the prescribed stresses at the model
boundary may not equal the given initial stresses. In this case, the boundary gridpoints will start
to move as soon as a STEP command is given; again, output should be checked (e.g., PLOT vel) for
this possibility.
The commands in Example 3.11 produce a single block with initial stresses that are in equilibrium
with prescribed boundary stresses:

Example 3.11 Initial and boundary stresses in equilibrium

new
poly brick 0,10 0,10 0,10
plot create plot Blocks
plot block
plot reset
;
gen edge 2.0
;
prop mat=1 dens 1000 bulk 8e9 g 5e9
bound stress -5e6, 0, 0 0,0,0 range x 0.0
bound stress -5e6, 0, 0 0,0,0 range x 10.0
bound stress 0,-5e6, 0 0,0,0 range y 0.0
bound stress 0,-5e6, 0 0,0,0 range y 10.0
bound stress 0, 0,-1e7 0,0,0 range z 0.0
bound stress 0, 0,-1e7 0,0,0 range z 10.0
insitu stress -5e6 -5e6 -1e7 0,0,0
;warning will appear due to lack of joints in block model

step 1
plot clear
plot block fill off zonetensor stress colorby minimum
ret

3.5.2 Stresses with Gradients in an Unjointed Medium: Uniform Material

Variation in stress with depth cannot be ignored near the ground surface – the GRAVITY command
is used to inform 3DEC that gravitational acceleration operates on the model. It is important to
understand that the GRAVITY command does not directly cause stresses to appear in the model;
it simply causes body forces to act on all gridpoints of deformable blocks (or centroids of rigid
blocks). These body forces correspond to the weight of material surrounding each gridpoint. If
no initial stresses are present, the forces will cause the material to move (during stepping) in the
direction of the forces until equal and opposite forces are generated by zone stresses. Given the
appropriate boundary conditions (e.g., fixed bottom, roller side boundaries), the model will, in fact,
generate its own gravitational stresses compatible with the applied gravity. However, this process is

3DEC Version 5.2

3 - 46 User’s Guide

inefficient, since many hundreds of steps may be necessary for equilibrium. It is better to initialize
the internal stresses such that they satisfy both equilibrium and the gravitational gradient. The
INSITU command must include the xgrad, ygrad and zgrad parameters so that the stress gradient
matches the gravitational gradient, gρ. The internal stresses must also match boundary stresses at
stress boundaries.
Even though the boundary and in-situ stresses are specified to produce a force balance, some cycling
of the model is normally required. This is because the boundary forces are only applied at the end
of a cycle; a small force imbalance is produced by the in-situ stresses. Usually, this imbalance is
reduced within a few hundred cycles.
Consider, for example, a 20 m × 20 m × 20 m box of homogeneous unjointed material at a depth
of 200 m underground, with fixed base and stress boundaries on the other five sides. Example 3.12
produces an equilibrium system for this problem condition:

Example 3.12 Initial stress state with gravitational gradient

new
poly brick 0,20 0,20 0,20
plot create plot Blocks
plot block
plot reset
;
gen edge 4.0
;
prop mat=1 dens 2500 bulk 5e9 g 3e9 phi 35
change cons 2
;
gravity 0 0 -10
;
bound stress -2.75e6,-2.75e6,-5.5e6 0,0,0 &
zgrad 1.25e4, 1.25e4, 2.5e4 0,0,0 range x 0.0
bound stress -2.75e6,-2.75e6,-5.5e6 0,0,0 &
zgrad 1.25e4, 1.25e4, 2.5e4 0,0,0 range x 20.0
bound stress -2.75e6,-2.75e6,-5.5e6 0,0,0 &
zgrad 1.25e4, 1.25e4, 2.5e4 0,0,0 range y 0.0
bound stress -2.75e6,-2.75e6,-5.5e6 0,0,0 &
zgrad 1.25e4, 1.25e4, 2.5e4 0,0,0 range y 20.0
bound stress 0, 0,-5.0e6 0,0,0 range z 20.0
bound zvel = 0.0 range z 0.0
insitu stress -2.75e6,-2.75e6,-5.5e6 0,0,0 &
zgrad 1.25e4, 1.25e4, 2.5e4 0,0,0
;warning will appear due to lack of joints in block model

step 500
;

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 47

plot clear
plot block fill off zonetensor stress colorby minimum reverse
ret

In this example, horizontal stresses and gradients are equal to half the vertical stresses and gradients,
but they may be set at any value that does not violate the yield criterion (Mohr-Coulomb, in this
case). After preparing a data file such as the one above, the model should be cycled to check that an
equilibrium state is reached. If material failure does occur (e.g., reduce phi = 10◦ ), this will show as
an unbalanced force magnitude that is roughly the same order of magnitude as the applied loading.

3.5.3 Stresses with Gradients in a Nonuniform Material

It is more difficult to give the initial stresses when materials of different densities are present.
Consider a layered system with a free surface, enclosed in a box with roller side boundaries and a
fixed base. Suppose that the material has the following density distribution:

1600 kg/m3 from 0 to 10 m depth

2000 kg/m3 from 10 to 15 m

2200 kg/m3 from 15 to 25 m

An equilibrium state is produced by the data file in Example 3.13:

Example 3.13 Initial stress gradient in a nonuniform material

new
poly brick 0,20 0,20 0,25
plot create plot Blocks
plot block
plot reset
;
jset dip 0.0 origin 0,0,10
jset dip 0.0 origin 0,0,15
gen edge 2.0
;
change mat 1 range z 0,10
change mat 2 range z 10,15
change mat 3 range z 15,25
prop mat=1 dens 1600 bulk 5e9 g 3e9
prop mat=2 dens 2000 bulk 5e9 g 3e9
prop mat=3 dens 2200 bulk 5e9 g 3e9
change jmat 1
prop jmat 1 jkn 1e10 jks 1e10 jcoh 1e10

3DEC Version 5.2

3 - 48 User’s Guide

;
gravity 0 0 -10
;
insitu stress 0.0,0.0,-4.8e5 0,0,0 &
zgrad 0.0,0.0, 2.2e4 0,0,0 range z 0,10
insitu stress 0.0,0.0,-4.6e5 0,0,0 &
zgrad 0.0,0.0, 2.0e4 0,0,0 range z 10,15
insitu stress 0.0,0.0,-4.0e5 0,0,0 &
zgrad 0.0,0.0, 1.6e4 0,0,0 range z 15,25
bound xvel = 0.0 range x 0.0
bound xvel = 0.0 range x 20.0
bound yvel = 0.0 range y 0.0
bound yvel = 0.0 range y 20.0
bound zvel = 0.0 range z 0.0
;
hist unbal
step 500
;
plot create plot Z-stress
plot set dip 90 dd 180
plot cut add plane origin (10,10,12.5) name Plane normal (0,1,0)
plot add bcont szz minimum -6e5 maximum 0 interval 1.25e5 wireframe off

An individual block is created for each material density; fictitious joints separate each block. The
internal stress profile is calculated manually for each block from the known overburden above it.
Note that the example is simplified: in a real case, the elastic moduli would vary, and there would
be horizontal stresses. If high horizontal stresses exist in a layer, these may also be installed with
the INSITU command.
This example is not in equilibrium at one calculation step; approximately 500 steps are required.
The presence of the fictitious joints also prevents the model from being in equilibrium when the
initial stresses match the boundary stresses. A jointed model will often require more steps to
equilibrate than an unjointed model.
For models in which the surface of the model corresponds to the actual surface of the earth, the INSITU
topo command may be used. With this command, vertical stresses are automatically calculated for
each zone by integrating upwards and summing the weight contribution of each overlying zone.
Horizontal stresses are then given as ratios of the vertical stress.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 49

3.5.4 Compaction within a Model with Nonuniform Zoning

Puzzling results are sometimes observed when a model with nonuniform zoning is allowed to come
to equilibrium under gravity. A model that is composed of deformable blocks of different sizes will
usually have nonuniform zoning. When a Mohr-Coulomb (or other nonlinear constitutive) model is
assigned to the blocks, the final stress state and displacement pattern are not uniform, even though
the boundaries are straight and the free surface is flat. The data file in Example 3.14 illustrates the
effect (see Figure 3.23 for the generated plot showing vertical stress contours):

Example 3.14 Nonuniform stress initialized in a model with nonuniform zoning

new
poly brick 0,10 0,2 0,10
plot create plot Blocks
plot block
plot reset
;
jset dip 90.0 dd 90 origin 3,0,0
hide range plane dip 90.0 dd 90 origin 3,0,0 above
;gen edge .3
gen quad ndiv 1,5,10
seek
;gen edge 1.0
gen quad ndiv 1,2,10
;
change cons 2
prop mat=1 dens 2000 bulk 2e8 g 1e8 phi 30
prop jmat 1 jkn 1e8 jks 1e8 jcoh 1e10 jten 1e10
;
gravity 0 0 -10
;
bound xvel = 0.0 range x -0.1 0.1
bound xvel = 0.0 range x 9.9 10.1
bound yvel = 0.0 range y -0.1 0.1
bound yvel = 0.0 range y 1.9 2.1
bound zvel = 0.0 range z -0.1 0.1
;
hist unbal
sav ex3_14.3dsav
step 500
plot clear
plot bcont szz max 0 min -200000 int 20000
sav ex3_14a.3dsav
; pause key
; optional method 1
restore ex3_14.3dsav

3DEC Version 5.2

3 - 50 User’s Guide

insitu stress -1.5e5 -1.5e5,-2.0e5, 0,0,0 &

zgrad 1.5e4 1.5e4, 2.0e4, 0,0,0
step 400
sav ex3_14-1.3dsav
; pause key
;
; optional method 2
restore ex3_14.3dsav
prop mat=1 bcoh 1e10 bten 1e10
step 750
prop mat=1 bcoh 0.0 bten 0.0
step 250
sav ex3_14-2.3dsav
ret

E

pGre00

0 u0
i
0SGG
pGrelG,lrG0p:Ge:,50
S
 
GiGGGGnGG
lpiGGGGnG.
l.iGGGGnG.
leiGGGGnG.
l-iGGGGnG.
lriGGGGnGS
lripGGGnGS
lri.GGGnGS
lrieGGGnGS
lri-GGGnGS
lpiGGGGnGS

Figure 3.23 Nonuniform stresses

Since we have roller boundaries on the four sides, we might expect the material to move down
equally on all sides. However, the zones are not the same size in the blocks. For static analysis,
3DEC tries to keep the timestep equal for all zones, so it increases the inertial mass for the gridpoints
of the smaller zones to compensate for their size. These gridpoints then accelerate more slowly
than those for the larger zones. This would have no effect on the final state of a linear material, but
it causes nonuniformity in a material that is path-dependent. For a Mohr-Coulomb material without
cohesion, the situation is similar to dropping sand from some height into a container and expecting

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 51

the final state to be uniform. In reality, a large amount of plastic flow would occur because the
confining stress does not build up immediately. Even with a uniformly zoned model, this approach
is not a good one because the horizontal stresses depend on the dynamics of the process.
The best solution is to use the INSITU stress command to set initial stresses to conform to the
desired Ko value (ratio of horizontal to vertical stress). For example, the STEP 1000 command in
the previous data file could be replaced by the following lines:
insitu stress (-1.5e5,-2.0e5,-1.5e5,0,0,0) &
ygrad (1.5e4,2.0e4,1.5e4,0,0,0)
step 400

A stable state is achieved with Ko = 0.75; fewer steps are needed to reach equilibrium, and the
stress state is uniform (see Figure 3.24). Note that there is a slight nonuniformity, but this is related
to the contouring routine and the coarseness of the zoning.
Alternatively, the model can be run with an elastic behavior for the initial equilibrium calculation,
and then changed to the nonlinear behavior model for the final state. Replace the STEP 1000
command with the following lines:
prop mat=1 btens=1e10 bcoh=1e10
step 750
prop mat=1 btens=0 bcoh=0
step 250

The result is the same as that shown in Figure 3.24. The material is prevented from yielding during
the compaction process, but the original properties are restored when equilibrium is achieved.

3DEC Version 5.2

3 - 52 User’s Guide

E

pGre00

0 u0
i
0rGGG
pGrelG,lrG0p8Ge8,30
S
 
GiGGGGnGG
lpiGGGGnG.
l.iGGGGnG.
leiGGGGnG.
l3iGGGGnG.
lriGGGGnGS
lripGGGnGS
lri.GGGnGS
lrieGGGnGS
lri3GGGnGS
lpiGGGGnGS

Figure 3.24 Uniform stresses

3.5.5 Initial Stresses following a Model Change

There may be situations in which one material model for deformable blocks is used in the process
of reaching a desired stress distribution, but another model is used for the subsequent simulation.
Models can be changed for entire blocks (via the CHANGE command). If one model is replaced by
another non-null model, the stresses in the affected zones are preserved:

Example 3.15 Initial stresses following a model change

new
poly brick 0,5 0,5 0,5
plot create plot Blocks
plot block colorby model
plot reset
;
gen edge 1.0
;
prop mat=1 dens 2000 bulk 3e8 g 2e8
;
gravity 0 0 -10
;
bound xvel = 0.0 zvel = 0.0 range z 0.0

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 53

;
hist unbal
step 250
;
;pause key
change cons 2
prop mat=1 dens 2000 bulk 3e8 g 2e8 phi 34
ret

At this point in the run, the stresses generated by the initial elastic model still exist and act as initial
stresses for the region containing the new Mohr-Coulomb model.
Two points should be remembered. First, if a null block is created (via the EXCAVATE command)
in any part of the model (even if it is subsequently replaced by another non-null block), all stresses
are removed from the null block. Second, if one material model is replaced by another and the
stresses should physically be zero in the new model, then an INSITU command must be used to reset
the stresses to zero in this region. This situation would occur if rock is mined out and replaced by
backfill; the backfill should start its life without stress.

3.5.6 Stresses in a Jointed Medium

A spatial heterogeneity in an initial stress state can develop in a jointed and fractured medium. This
results from the stress path followed during the geologic history of the medium, and the physical
processes (related to fracturing and slip and separation along discontinuities) that may have occurred
at different stages in the history. Spatial heterogeneity of the stress state can be an important factor in
the design of underground excavations, particularly if the resulting stress concentrations adversely
influence the excavation stability.
It is very difficult to determine whether the stress state installed in a jointed model is representative
of the in-situ state of stress. As will be discussed in Section 3.11.2, statistical analyses may provide
a means to develop confidence in the model representation. One such study using UDEC is reported
by Brady et al. (1986).
There are certain modeling aspects that should be considered when bringing a jointed model to an
equilibrated state. First, the INSITU command should be invoked after all joints are generated in
the model. Then the normal and shear stresses along joints will be initialized, corresponding to the
initial stress values resolved along the plane of each joint.
As mentioned previously, a jointed model will not be in equilibrium initially, even when internal
stresses are set to match boundary stresses: some calculation steps are required, and the unbalanced
force should be monitored. In addition, histories of velocities or displacements should be recorded
at various locations in the model. These are good indicators of the calculation step at which motion
is negligible. The user should always ensure that motion in the model has essentially stopped for
the equilibrium stress state before beginning the next stage of an analysis.

3DEC Version 5.2

3 - 54 User’s Guide

It is possible that, for the specified initial stress state and joint strength properties, some joints will
slip or separate when the model is brought to an equilibrated state. Joint slip that is confined within
the model is acceptable; “locked-in” stresses at the joint ends will result. However, the user should
avoid conditions for which joint failure extends to the model boundary. This indicates that the
model conditions are not well-posed. It may be necessary to reevaluate the assigned stress state,
joint properties and joint orientations and locations. If conditions are such that joint failure still
extends to a boundary, then a fixed boundary condition should be considered. This implies that the
joint is truncated at the boundary.
The data file in Example 3.16 demonstrates the case of a joint dipping at 60◦ , confined between
two joints dipping at 20◦ . The 60◦ joint slips for the prescribed initial stress, while the 20◦ joints
do not. The friction angle for all joints is 30◦ .

Example 3.16 Slip of a confined joint

new
poly brick -10,10 -10,10 -20,0
plot create plot Blocks
plot block
plot reset
;
jset dip 60 dd 90 origin 0,0,-10
jset dip 20 dd 90 origin 0,0,-8
jset dip 20 dd 90 origin 0,0,-12
join on range z -20,-14
join on range z -7,0
gen edge 2.0
;
prop mat=1 dens 2000 bulk 8e9 g 5e9
prop jmat=1 jkn 5e11 jks 2.5e11 jfric 30
insitu stress -2.5e6,-2.5e6,-1e7 0,0,0
;
bound stress -2.5e6 0 0 0 0 0 range x -10.0
bound stress -2.5e6 0 0 0 0 0 range x 10.0
bound stress 0 -2.5e6 0 0 0 0 range y -10.0
bound stress 0 -2.5e6 0 0 0 0 range y 10.0
bound stress 0 0 -1e7 0 0 0 range z 0.0
bound zvel 0.0 range z -20.0
;
hist unbal
hist zdis 0,0,-10
step 1000
;
plot clear
plot cut add plane origin (0,0,-10) normal (0,1,0)
plot add jointvector shear shearoffset 0.1 plane on

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 55

plot add blockcontour sxz fill on wireframe off

plot add joint colorby material clear addlabel "1" black line width 3
plot set orientation (90,0,0) center (0,0,-10) mag 2.5
ret

Figure 3.25 shows a block plot on which the region of joint slip is indicated by joint shear vectors.
Contours of σxz are also plotted and show the areas of locked-in stresses near the ends of the 60◦
joint.

E

pGre0b0E

0y u0b
i
0rGGG
pGrelG,lrG0p4Ge4.G0

i i 
i 

 40GiGGG,33ppr
40r.:3i:,
i
 

40 0
0
eip-3p5nGS
eiGGGG5nGS
SiGGGG5nGS
.iGGGG5nGS
,iGGGG5nGS
piGGGG5nGS
riGGGG5nGS
GiGGGG5nGG
lriGGGG5nGS
lpiGGGG5nGS
l,iGGGG5nGS
l.iGGGG5nGS
l.i--3.5nGS



40 0
0
E  740 0
r
E  740 0
r

Figure 3.25 Slip of a confined joint; plot shows shear stress contours

3.5.7 Determination of the In-situ Stress State

Knowledge of the virgin stress field is required in order to establish appropriate boundary and initial
conditions for models of underground excavations. However, often it is not possible to perform
in-situ stress measurements sufficiently far from underground excavations or topographic features
to determine virgin stresses. This is particularly true for mines using massive mining methods.
Boundary stresses may be approximated from regional stress compilations if available (e.g., Müller
et al. 1992; Lindner and Halpern 1978), but three-dimensional modeling can be used to quantify
the various forms of induced stress, such as those generated by topography, excavations or material
property variations. Subtracting the induced stress components from the total or measured stress
enables the virgin stress field to be computed. The latter method has the advantage of being able to
utilize stress measurements, which are known to include induced stresses from various sources (or

3DEC Version 5.2

3 - 56 User’s Guide

to check whether the measurements are free of induced stress). It also enables a computation of the
local stress field to be made, which is not necessarily represented by the regional stress field. The
following procedure has been used in three-dimensional stress analyses to estimate virgin stresses
from in-situ measurements that were influenced by various forms of induced stress.
The total stress field, σtot , at any point is the sum of virgin stress plus any induced stress components,
σind . Virgin stress, in turn, is composed of gravitational stress, σgrav , plus an as-yet-undetermined
additional horizontal component that will be referred to as a tectonic component, σtec . There
are various geological reasons for why this additional horizontal component of stress should be
incorporated into the total stress tensor. Eq. (3.11) relates these components:

σtot = σgrav + σtec + σind (3.11)

The induced stress, in turn, is composed of gravitational and tectonic components:

σind = σ̄grav + σ̄tec (3.12)

Substituting Eq. (3.12) into Eq. (3.11) produces

σtot = σgrav + σ̄grav + σtec + σ̄tec (3.13)

which, upon re-grouping, becomes

σtec + σ̄tec = σtot − (σgrav + σ̄grav ) (3.14)

Terms on the right-hand side of Eq. (3.14) are either known (stress measurements are representative
of the total stress field), or can be computed using a model with only gravitational loading.
To start the computation process, it should be assumed that the problem geometry (i.e., topography,
nearby excavations) is the primary factor generating the induced stress field, and that material
property variations produce only second-order effects. (Experience has shown this to be a reasonable
assumption.)
First, a model is constructed, taking into account the topography and excavation geometry. This
is run with gravitational loading only. The resulting stress field at the stress measurement points
accounts for gravity, plus the gravitational component of induced stress caused by the problem
geometry. The resultant calculated vertical stresses are compared to the corresponding measured
vertical stresses, and the measured stresses (all components) are adjusted to bring the measured
vertical stress into agreement with the calculated vertical stress. This latter step essentially scales
the measured stress to the model. If the computed and measured vertical components of stress are
found to differ by a large amount, then either the model is incorrect (e.g., incorrect densities), there
are other unknown sources of induced stress (e.g., locked-in stresses from geological processes), or
there are significant errors in the measurements. In these situations, it is wise to investigate further,

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 57

to determine the reason for the stress anomaly, because confidence in the stress field is a critical
design requirement.
The unknown tectonic components can be solved by applying unit normal or shear stress boundary
conditions to the model, and computing the resultant stress level at the stress measurement point.
The correct “far-field” tectonic boundary stress is computed by scaling the unit stress results to
match the magnitudes of components obtained using Eq. (3.14). The total stress field is specified
by the combination of horizontal tectonic stresses applied at the boundary of the model, as well as
gravitational stresses.

3.5.8 Transferring Field Stresses to Model Stresses

A utility command, LIST principal, is provided to transform field stresses into a set of stress com-
ponents referenced to the local problem axes defined for the 3DEC model. The orientation of the
local (model) axes is defined by the dip and dip direction of the local y-axis. The local z-axis lies in
the vertical plane containing the y-axis dip vector, and the x-axis lies in the horizontal plane. The
command calculates local stress components on the basis of the following input data:
LIST princ min int max min dipdd dip,dd max dipdd dip,dd
The user must ensure that the directions of σ1 , σ2 and σ3 are orthogonal.
The following example illustrates the transformation of field stresses to boundary stresses for the
3DEC model.
The field stresses for this problem are:

σ1 = 30 MPa directed 24◦ / 231◦ (dip / dip direction)

σ2 = 15 MPa directed 66◦ / 36◦ (dip / dip direction)
σ3 = 12 MPa directed 84◦ / 128◦ (dip / dip direction)

3DEC Version 5.2

3 - 58 User’s Guide

Stress Data for 3DEC (tension positive)

Field Stress:

Principal Stress 1 Magnitude −30.0 Dip 24.0 Bearing 231.0

Principal Stress 2 Magnitude −15.0 Dip 66.0 Bearing 36.0

Principal Stress 3 Magnitude −12.0 Dip 84.0 Bearing 128.0

3DEC Command:
list principal -30 -15 -12 min az 231 24 int az 36 66

Stresses Relative to Model Axes:

SXX −21.19 SYY −18.22 SZZ −17.59

SXY −7.53 SYZ −3.37 SXZ −4.63

3.5.9 Topographical Stresses

The command INSITU topograph automatically calculates gravity loading in models that have an
irregular top surface. The stresses are calculated based on the density of the overlying materials
and specify Ko values.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 59

3.6 Loading and Sequential Modeling

By applying different model loading conditions at different stages of an analysis, it is possible to

simulate changes in physical loading, such as sequences of excavation and construction. Changes
in loading may be specified in a number of ways (e.g., by applying new stress or displacement
boundaries, by changing the material model in blocks to either a null material or to a different
material model, or by changing material properties).
It is important to recognize that sequential modeling follows the stages of an engineering work.
In most analyses, each work stage corresponds to a different static solution following a loading
change (i.e., physical time is not a parameter). 3DEC can perform calculations for heat transfer
and dynamic mechanical analysis as well (see Sections 1 and 2 in Optional Features). In these
cases, a static solution for an equilibrium stress state may be followed, for example, by a dynamic
calculation for an applied explosive excitation or a transient calculation for flow through joints.
Time-dependent behavior, on the other hand, cannot be simulated directly. Some engineering
judgment must be used to estimate the effects of time. For example, a model parameter may be
changed after a predetermined amount of displacement or strain has occurred. This displacement
may be estimated to have occurred over a given period of time.
A loading change must cause unbalanced forces to develop in order to effect a change in model
response. Therefore, changing the elastic properties will have no effect; changing strength properties
will, if the change causes the current stress state to exceed the failure limit.
The recommended approach to sequential modeling is demonstrated by the following example. This
problem involves the stability analysis of an underground opening in jointed rock, and includes the
evaluation of different types of support measures. The stages to be analyzed are:
(1) equilibration at the in-situ stress state;
(2) excavation of the tunnel; and
(3) application of the tunnel support.
The objective is to investigate the stability of the excavation under in-situ conditions, and assess
the effect of the support measures. Three types of support are evaluated: local reinforcement rock
bolts, cable bolts and a continuous concrete liner. Note that this model is greatly simplified for
rapid execution, but it still illustrates the recommended steps for loading and sequential modeling.
The tunnel is located in rock containing three major faults: one dipping at 65◦ with a dip direction
of 40◦ ; the second dipping at 70◦ with a dip direction of 270◦ ; and the third dipping at 60◦ with a
dip direction of 130◦ . The tunnel is horseshoe-shaped and is centered along the y-axis of the model.
The tunnel is created with the TUNNEL command. A second TUNNEL command is also used to define
the location of the concrete liner. Note that this must be done before any cycling is performed. The
model is created with the following series of commands, beginning with Example 3.17:

3DEC Version 5.2

3 - 60 User’s Guide

Example 3.17 Stability analysis of an underground excavation – initial model

new
poly brick -10 10 -10 10 -10 10
plot create Plot Blocks
plot block
plot reset
; --- tunz: FISH function to define tunnel geometry parameters --------
; ... tunnel along Y axis, from YYA to YYB
; ... semi-circular roof, centered at (TXC,TZC)
;
def tunz
;
yya = -10.0
yyb = 10.0
;
; --- outer surface ---
txb1 = -4.0
tzb1 = -4.0
txb2 = 4.0
tzb2 = -4.0
;
txc = 0.0
tzc = 0.0
tr = 4.0
tx1 = txc + tr * cos(180*degrad)
tz1 = tzc + tr * sin(180*degrad)
tx2 = txc + tr * cos(135*degrad)
tz2 = tzc + tr * sin(135*degrad)
tx3 = txc + tr * cos(90*degrad)
tz3 = tzc + tr * sin(90*degrad)
tx4 = txc + tr * cos(45*degrad)
tz4 = tzc + tr * sin(45*degrad)
tx5 = txc + tr * cos(0*degrad)
tz5 = tzc + tr * sin(0*degrad)
;
; --- inner surface ---
; thickness th
th = 0.5
txb1i = -4.0 + th
tzb1i = -4.0 + th
txb2i = 4.0 - th
tzb2i = -4.0 + th
;
txc = 0.0

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 61

tzc = 0.0
tri = tr - th
tx1i = txc + tri * cos(180*degrad)
tz1i = tzc + tri * sin(180*degrad)
tx2i = txc + tri * cos(135*degrad)
tz2i = tzc + tri * sin(135*degrad)
tx3i = txc + tri * cos(90*degrad)
tz3i = tzc + tri * sin(90*degrad)
tx4i = txc + tri * cos(45*degrad)
tz4i = tzc + tri * sin(45*degrad)
tx5i = txc + tri * cos(0*degrad)
tz5i = tzc + tri * sin(0*degrad)
;
end
;
; ---------------------------------------------------------------------
; (execute function)
@tunz
;
; create outer surface
tunnel radial &
a @txb1,@yya,@tzb1 @tx1,@yya,@tz1 &
@tx2 ,@yya,@tz2 @tx3,@yya,@tz3 &
@tx4 ,@yya,@tz4 @tx5,@yya,@tz5 @txb2,@yya,@tzb2 &
b @txb1,@yyb,@tzb1 @tx1,@yyb,@tz1 &
@tx2 ,@yyb,@tz2 @tx3,@yyb,@tz3 &
@tx4 ,@yyb,@tz4 @tx5,@yyb,@tz5 @txb2,@yyb,@tzb2 &
reg 5
hide range reg 0
;
; create inner surface
tunnel radial &
a @txb1i,@yya,@tzb1i @tx1i,@yya,@tz1i &
@tx2i ,@yya ,@tz2i @tx3i,@yya,@tz3i &
@tx4i ,@yya ,@tz4i @tx5i,@yya,@tz5i @txb2i,@yya,@tzb2i &
b @txb1i,@yyb,@tzb1i @tx1i,@yyb,@tz1i &
@tx2i ,@yyb ,@tz2i @tx3i,@yyb,@tz3i &
@tx4i ,@yyb ,@tz4i @tx5i,@yyb,@tz5i @txb2i,@yyb,@tzb2i &
reg 7
seek
;
; --- NOTE: region inside inner surface is REG 7
; region between surface (to be liner) is REG 5
;
save tun_a.3dsav
;

3DEC Version 5.2

3 - 62 User’s Guide

; --- joints --- 3 joints to form a wedge in the roof

;
jset dd 270 dip 70 origin 0 0,5.7 id 10
jset dd 40 dip 65 origin 0 0,5.7 id 10
jset dd 130 dip 60 origin 0 0,5.7 id 10
;
save tun_b.3dsav
;
; --- mesh generation ---
; rock blocks
hide range reg 5 7
gen edge 5
;
; liner
find range reg 5
gen edge 2
;
find range reg 7
gen edge 2
;
save tun_z.3dsav
plot clear
plot block colorby reg wireframe off
plot add joint wireframe on
plot set dip 110 dd 30
ret

Figure 3.26 shows the resulting model configuration. The tunnel geometry parameters are defined
in the FISH function tunz. The inner region of the tunnel is assigned region number 7, and the
region corresponding to the liner is assigned region number 5. Note that the tunnel is created first,
and then the physical joint set is generated. Blocks are joined automatically with the TUNNEL
commands. Zone generation is performed separately for the rock blocks, the liner blocks and the
interior region of the tunnel.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 63

E

pGre00

0 u0
i
0G
pGrelG,lrG0p4Ge4.r0

  40

G
S
-

Figure 3.26 3DEC model of tunnel region

Material properties are assigned to the rock blocks (mat 1), the concrete liner blocks (mat 5), the
rock joints (jmat 1), the concrete-concrete joints (jmat 5) and the concrete-rock interface (jmat 6).
The in-situ stress state and boundary conditions are applied, assuming the tunnel is at a depth of
200 m and the ratio of horizontal to vertical stress is 0.5. Note that for a practical simulation, the
boundaries are too close to the tunnel excavation and should be moved to a greater distance to
minimize their influence on the model results (see Section 3.4.4.2).
The commands to assign material properties and achieve the initial stress state are listed in Exam-
ple 3.18:

Example 3.18 Stability analysis of an underground excavation – initial equilibrium stress

state
rest tun_z.3dsav
;
; --- properties ---
;
; --- MAT 1 : rock ---
; density = 2700 kg/m3 = 0.0027e6 kg/m3
; E=50 GPa, Poisson’s ratio=0.2
prop mat 1 dens 0.0027 k 27778 g 20833
;
; --- MAT=5 : concrete liner ---
; density = 2400 kg/m3 = 0.0024e6 kg/m3

3DEC Version 5.2

3 - 64 User’s Guide

; E=30 GPa, Poisson’s ratio=0.2

prop mat 5 dens 0.0025 k 16667 g 12500
;
; --- JMAT=1 : rock joints ---
prop jmat 1 jkn 10000 jks 2000 jfric 25
;
; --- JMAT=5 : concrete-concrete joints (elastic) ---
prop jmat 5 jkn 30000 jks 12000 jcoh 1e6 jtens 1e6
;
; --- JMAT=6 : concrete-rock interface ---
prop jmat 6 jkn 10000 jks 2000 jfric 0.001
;
; --- assign material numbers ---
; initially all materials are rock
change mat 1
change jmat 1
;
; --- insitu stress state ---
; assume tunnel at 200 m depth
; vertical stress: szz=(0.0027*g)*(z-200)
; at z=0: szz=-5.4
; z-gradient of szz: 0.027
; (positive: less compression going up)
; horizontal sxx=syy=0.5*szz
;
insitu stress -2.7 -2.7 -5.4 0 0 0 &
zgrad 0.0135 0.0135 0.027 0 0 0
;
; gravity
grav 0 0 -10
;
; --- boundary conditions for insitu stress state ---
; top of model (z=10): szz=-0.027*190=-5.13
bound stress 0 0 -5.13 0 0 0 range z 10.0
; bottom
bound zvel 0 range z -10.0
; sides
bound xvel 0 range x -10.0
bound xvel 0 range x 10.0
bound yvel 0 range y -10.0
bound yvel 0 range y 10.0
;
; --- histories to monitor convergence ---
hist nc=1 unbal
; top of model
hist xdis 0 0 10 ydis 0 0 10 zdis 0 0 10

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 65

;
save tun_c0.3dsav
plot create plot Hist
plot hist 2 3 4 yaxis label ’Displacement’
;
cycle 1000
save tun_c.3dsav
ret

The maximum unbalanced force in the model and displacements at the top boundary are monitored
to help make sure that an initial equilibrium stress state is reached within 1000 cycles. Figure 3.27
shows the x-, y- and z-displacement histories for the gridpoint (x = 0, y = 0, z = 10) at the top of
the model:

E
 
pGre005

0 u0
i
0rGGG 
pGrelG,lrG0p4Ge4.p0

 
p0!lY 
0 0pp:S,p
,0"lY 
0 0pp:S,p
.0#lY 
0 0pp:S,p
i0 


 
e6-c















         

e6-o

Figure 3.27 Displacement histories at top of model

If the tunnel is excavated without support, a rock wedge detaches and falls from the roof. This
is shown by running Example 3.19; the tunnel is excavated with the DELETE command, and the
z-displacement at a location in the roof is monitored while the model is cycled. Figure 3.29 plots
the z-displacement history and indicates that the position is moving downward. Figure 3.28 shows
a close-up view of the detached wedge, with surrounding blocks hidden for better viewing.

3DEC Version 5.2

3 - 66 User’s Guide

Example 3.19 Stability analysis of an underground excavation – unsupported tunnel

rest tun_c.3dsav
plot create plot Blocks
plot block
plot reset
; delete interior blocks
remove range reg 5 7
;
; history point at tunnel roof
reset disp time hist
hide
seek range xr .44 .46
hist zdis 1.8 0 3.72
hist label 1 ’Vertical Displacement’
seek
;

plot create plot Hist

pl hist 1 yaxis label ’Displacement’
cycle 5000
;
save tun_x.3dsav

hide range plane dip 60 dd 130 or 0 0 5.7 above

plot current plot Blocks
plot clear
plot block colorby mat velocity line color black
plot set dip 100 dd 0 mag 2.5
ret

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 67


E

pGre00

0 u0
i

0eGGG
pGrelG,lrG0p4Ge4.S0



r0$ 0% 

i0 



 
eH-cH

















         

eH-r

Figure 3.28 z-displacement history at tunnel roof

E

pGre007

0 u0
i
0eGGG
pGrelG,lrG0p4Ge4.S0

7  40 
r
 
o 

 40GiSr3-p-
40riee:S

Figure 3.29 Close-up view of wedge in roof (surrounding blocks hidden)

3DEC Version 5.2

3 - 68 User’s Guide

The effect of rock bolt support is evaluated first for local reinforcement elements (STRUCTURE
axial), and then for fully bonded cable elements (STRUCTURE cable). See Section 3 in Theory and
Background for a detailed description of these two types of structural support.
Example 3.20 lists the commands to excavate the tunnel and install the local reinforcement elements,
and Example 3.21 lists those for cable element support. Note that we use the REMOVE command
to excavate the tunnel this time. This has the same effect as the DELETE command, but now we
can view the excavated region with the PLOT block exc command. The reinforcement elements
and cable elements are positioned in the same locations in the side walls and roof of the tunnel.
Figure 3.30 shows the location of the cable elements around the tunnel excavation.

E

pGre00

0 u0
i
0,GGG
pGrelG,lrG0p4Ge4.:0

  40 
r

  840 0%
-

Figure 3.30 Cable bolts positioned around tunnel excavation

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 69

Example 3.20 Stability analysis of an underground excavation – local reinforcement support

rest tun_c.3dsav
;
; delete interior blocks
remove range region 7
;
; delete liner blocks
remove range reg 5
; --- install axial elements ---
struct axial -8 -5 -2 -3.9 -5 -2 prop 7
struct axial -8 0 -2 -3.9 0 -2 prop 7
struct axial -8 5 -2 -3.9 5 -2 prop 7
struct axial -6.8 -5 6.8 -2.8 -5 2.8 prop 7
struct axial -6.8 0 6.8 -2.8 0 2.8 prop 7
struct axial -6.8 5 6.8 -2.8 5 2.8 prop 7
;
struct axial 8 -5 -2 3.9 -5 -2 prop 7
struct axial 8 0 -2 3.9 0 -2 prop 7
struct axial 8 5 -2 3.9 5 -2 prop 7
struct axial 6.8 -5 6.8 2.8 -5 2.8 prop 7
struct axial 6.8 0 6.8 2.8 0 2.8 prop 7
struct axial 6.8 5 6.8 2.8 5 2.8 prop 7
;
struct axial 0 -5 4 0 -5 8 prop 7
struct axial 0 0 4 0 0 8 prop 7
struct axial 0 5 4 0 5 8 prop 7
;
struct prop 7 rkax 250 rlen 0.10 rult 0.55
;
reset disp time hist
; history point at tunnel roof
hide
seek range xr .44 .46
hist zdis 1.8 0 3.72
hist label 1 ’Vertical Displacement’
seek
;
plot create plot Hist
plot hist 1 yaxis label ’Displacement’
cy 2000
;
save tun_lr.3dsav
ret

3DEC Version 5.2

3 - 70 User’s Guide

Example 3.21 Stability analysis of an underground excavation – fully grouted cable support
rest tun_c.3dsav
;
; delete interior blocks
plot create plot Blocks
plot block
remove range region 7
;
; delete liner blocks
remove range reg 5
;
; --- install cable elements ---
struct cable -8 -5 -2 -4.05 -5 -2 prop 8 seg 4
struct cable -8 0 -2 -4.05 0 -2 prop 8 seg 4
struct cable -8 5 -2 -4.05 5 -2 prop 8 seg 4
struct cable -6.8 -5 6.8 -2.85 -5 2.85 prop 8 seg 4
struct cable -6.8 0 6.8 -2.85 0 2.85 prop 8 seg 4
struct cable -6.8 5 6.8 -2.85 5 2.85 prop 8 seg 4
;
struct cable 8 -5 -2 4.05 -5 -2 prop 8 seg 4
struct cable 8 0 -2 4.05 0 -2 prop 8 seg 4
struct cable 8 5 -2 4.05 5 -2 prop 8 seg 4
struct cable 6.8 -5 6.8 2.85 -5 2.85 prop 8 seg 4
struct cable 6.8 0 6.8 2.85 0 2.85 prop 8 seg 4
struct cable 6.8 5 6.8 2.85 5 2.85 prop 8 seg 4
;
struct cable 0 -5 4.1 0 -5 8 prop 8 seg 4
struct cable 0 0 4.1 0 0 8 prop 8 seg 4
struct cable 0 5 4.1 0 5 8 prop 8 seg 4
;
; start with high SBOND
struct prop 8 area 5e-4 e 100000 yield 0.55 kbond 15e4 sbond 1e6
;
reset disp time hist
; history point at tunnel roof
hide
seek range xr .44 .46
hist zdis 1.8 0 3.72
hist label 1 ’Vertical Displacement’
seek
;
cycle 500
;
; set real SBOND

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 71

struct prop 8 sbond 0.8

;
plot create plot VertDisp
plot hist 1 yaxis label ’Displacement’
cy 1500
;
save tun_cab.3dsav

plot current plot Blocks

plot clear
plot block excavate on colorby material cable
plot set dip 100 dd 10
ret

The roof is stabilized for both types of reinforcement. The z-displacement history now indicates
that the wedge movement stops at roughly 25 mm displacement for both the reinforcement elements
and the cable elements (see Figures 3.31 and 3.32).

E

pGre00

0 u0
i 
0,GGG
pGrelG,lrG0p4Ge4.30



r0$ 0% 

i0





 
eH-cr


















         

eH-r

Figure 3.31 z-displacement history at tunnel roof – reinforcement element

support

3DEC Version 5.2

3 - 72 User’s Guide


E

pGre00

0 u0
i 
0,GGG
pGrelG,lrG0p4Ge4.:0



r0$ 0% 
 
i0




 
eH-cr

















         

eH-r

Figure 3.32 z-displacement history at tunnel roof – cable support

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 73

The axial forces that develop in the support are greatest in the roof elements. This is shown for
both the reinforcement elements and the cable elements by the axial force plots in Figures 3.33 and
3.34:

E

pGre008

0 u0
i
0,GGG
pGrelG,lrG0p4Ge4.30

8  40 
r
o  !too! o 

 40Gi,S-,e3
40ri-G.33

Figure 3.33 Axial forces in reinforcement elements

E

pGre005

0 u0
i
0,GGG
pGrelG,lrG0p4Ge4.:0

5  40 
r
 oo!
 40GiG3:S.r:
40-ieSpS-

Figure 3.34 Axial forces in cable elements

3DEC Version 5.2

3 - 74 User’s Guide

The model of a tunnel excavation and support sequence should simulate the change in stresses
around the tunnel as the excavation advances, before the tunnel support is installed. This can
be done in a 3DEC model by excavating the tunnel in sections and installing support after each
excavation section. This is the recommended approach to simulate support loading changes due to
tunnel advancement.
Alternatively, in this simplified model we simulate the effect of tunnel advancement by reducing
the tractions at the tunnel periphery in increments, and installing the liner before the tractions are
completely removed. This demonstrates an approach for simulating a gradual excavation of a tunnel
section. Example 3.22 shows the data file for this approach:

Example 3.22 Stability analysis of an underground excavation – reduce tunnel tractions by

50% and install liner
rest tun_c.3dsav
plot create plot Blocks
plot block
plot set dip 100 dd 20
; delete interior blocks
delete range region 7
;
; excavate liner blocks (not deleted)
excavate range reg 5
;
; history point at tunnel roof
hide
seek range xr .44 .46
hist xdis 1.8 0 3.72 ydis 1.8 0 3.72 zdis 1.8 0 3.72
hist label 7 ’Vertical Displacement’
seek
;
; simulate the removal of approximately 50% of insitu stress
; applying at liner-rock interface a stress state
; szz=-2.7 sxx=syy=-1.35
;
bound str -1.35 -1.35 -2.7 0 0 0 range x -4.1,-3.9 y -11 11 z -4.1 0.1
bound str -1.35 -1.35 -2.7 0 0 0 range x 3.9,4.1 y -11 11 z -4.1 0.1
bound str -1.35 -1.35 -2.7 0 0 0 range x -4.1,4.1 y -11 11 z -4.1 -3.9
; note: need to include all faces on tunnel surface
; (inner radius must be a bit smaller than 4.0)
bound str -1.35 -1.35 -2.7 0 0 0 &
range zr -0.1,4.1 cyl end1 0,-11,0 end2 0,11,0 rad 3.5,4.1
;
; must fix again end-surfaces that were freed by BOU STRESS
bou yvel 0 range yr -10.1,-9.9
bou yvel 0 range yr 9.9,10.1

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 75

;
; check that sum of applied forces on tunnel surface is zero
list boun force range x -5,5 y -11,11 z -5,5
; check that sum of applied x and z forces on tunnel surface is near zero
;
cycle 2000
;
save tun_l1.3dsav
;
; --- insert liner ---
; remove loads from tunnel surface
;
bound xfree yfree zfree range x -4.1,-3.9 y -11,11 z -4.1,0.1
bound xfree yfree zfree range x 3.9,4.1 y -11,11 z -4.1,0.1
bound xfree yfree zfree range x -4.1,4.1 y -11,11 z -4.1,-3.9
bound xfree yfree zfree range zr -0.1,4.1 &
cyl end1 0,-11,0 end2 0,11,0 rad 3.5,4.1
;
; must fix again end-surfaces
bou yvel 0 range yr -10.1 -9.9
bou yvel 0 range yr 9.9 10.1
;
; insert liner
fill mat 5 jmat 5 range reg 5
;
; join liner blocks
join on range reg 5
;
; assign rock-liner interface material
change jmat 6 range rint 0 5
;
plot create plot Hist
plot hist 7 yaxis label ’Displacement’
cy 2000
;
save tun_l2.3dsav
hide
seek range reg 5
plot current plot Blocks
plot clear
plot reset
plot block
plot set dip 100 dd 20
ret

3DEC Version 5.2

3 - 76 User’s Guide

The BOUNDARY command is used to apply 50% of the in-situ stress state to the liner-rock interface,
and the BOUNDARY range covers all faces on the tunnel surface. The applied stresses at the tunnel
surface should produce traction forces on the surface that sum to zero; this can be checked with
the LIST bound force command. The model is cycled to an equilibrium state with tunnel tractions
reduced by 50%. Then, the tractions are removed completely and the liner is installed (with the
FILL region command). Figure 3.35 shows the liner blocks created for this model. The model is
cycled to a new equilibrium state. The load that develops in the liner is due to the reduction of the
tractions from 50% to zero.
Note that the selection of a 50% reduction in tunnel tractions in this example is arbitrary and only
for demonstration purposes. If it is necessary to simulate a gradual excavation, it may be necessary
to reduce the tractions in smaller increments to minimize the effects of transient stress waves on
the response of the model.

E

pGre00

0 u0
i
0SGGG
pGrelG,lrG0p:Ge:S.0

  :0 
r
S

Figure 3.35 Thick concrete liner support – liner blocks

The displacement of the roof is monitored in Figure 3.36. Roughly 1 mm of vertical displacement
occurs when the tractions are reduced by 50%, and an additional 1 mm displacement after the tunnel
tractions are completely removed and the liner is installed.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 77

E

pGre00

0 u0
i
0SGGG 
pGrelrGlpS0p8r-8,p0


50$ 0% 
 
i0



 
eH-cr













       

eH-r

Figure 3.36 z-displacement history at tunnel roof – tunnel liner added after
tractions reduced by 50%

If a more representative model of the liner behavior (including an elastic-plastic response) is re-
quired, then mixed-discretization zoning (see Section 3.3.2) should be used to define the liner with
a minimum of five m-d zones across the liner thickness. The POLYHEDRON prism command can be
used to create liner blocks for the m-d zones. Example 3.23 presents a data file to create the liner
with m-d zoning:

Example 3.23 Stability analysis of an underground excavation – liner with m-d zoning
rest tun_c0.3dsav
; delete interior blocks
delete range reg 5 7
; ---------------------------------------------------------------------
; --- insert support with POLY prism commands ---
;
poly prism a @txb1 @yya @tzb1 @tx1 @yya @tz1 &
@tx1i @yya @tz1i @txb1i @yya @tzb1i &
b @txb1 @yyb @tzb1 @tx1 @yyb @tz1 &
@tx1i @yyb @tz1i @txb1i @yyb @tzb1i &
reg 8
;
poly prism a @tx1 @yya @tz1 @tx2 @yya @tz2 &
@tx2i @yya @tz2i @tx1i @yya @tz1i &

3DEC Version 5.2

3 - 78 User’s Guide

b @tx1 @yyb @tz1 @tx2 @yyb @tz2 &

@tx2i @yyb @tz2i @tx1i @yyb @tz1i &
reg 8
;
poly prism a @tx2 @yya @tz2 @tx3 @yya @tz3 &
@tx3i @yya @tz3i @tx2i @yya @tz2i &
b @tx2 @yyb @tz2 @tx3 @yyb @tz3 &
@tx3i @yyb @tz3i @tx2i @yyb @tz2i &
reg 8
;
poly prism a @tx3 @yya @tz3 @tx4 @yya @tz4 &
@tx4i @yya @tz4i @tx3i @yya @tz3i &
b @tx3 @yyb @tz3 @tx4 @yyb @tz4 &
@tx4i @yyb @tz4i @tx3i @yyb @tz3i &
reg 8
;
poly prism a @tx4 @yya @tz4 @tx5 @yya @tz5 &
@tx5i @yya @tz5i @tx4i @yya @tz4i &
b @tx4 @yyb @tz4 @tx5 @yyb @tz5 &
@tx5i @yyb @tz5i @tx4i @yyb @tz4i &
reg 8
;
poly prism a @tx5 @yya @tz5 @txb2 @yya @tzb2 &
@txb2i @yya @tzb2i @tx5i @yya @tz5i &
b @tx5 @yyb @tz5 @txb2 @yyb @tzb2 &
@txb2i @yyb @tzb2i @tx5i @yyb @tz5i &
reg 8
;
poly prism a @txb2 @yya @tzb2 @txb1 @yya @tzb1 &
@txb1i @yya @tzb1i @txb2i @yya @tzb2i &
b @txb2 @yyb @tzb2 @txb1 @yyb @tzb1 &
@txb1i @yyb @tzb1i @txb2i @yyb @tzb2i &
reg 8
;
hide
seek range reg 8
plot create plot Blocks
plot block colorby reg axes scale 10
plot set dip 100 dd 20

gen quad ndiv 5 10 5 rmul 1.0

plot create plot Wireframe
plot block fill off
plot set dip 100 dd 20
;
change mat 5 range reg 8

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 79

change jmat 5 range rint 8 8

change jmat 6 range rint 0 8
;
; --- MAT=5 : concrete liner ---
; density = 2400 kg/m3 = 0.0024e6 kg/m3
; E=30 GPa, Poisson’s ratio=0.2
prop mat 5 dens 0.0025 k 16667 g 12500
;
; --- JMAT=5 : concrete-concrete joints (elastic) ---
prop jmat 5 jkn 30000 jks 12000 jcoh 1e6 jtens 1e6
;
; --- JMAT=6 : concrete-rock interface ---
prop jmat 6 jkn 10000 jks 2000 jfric 35
;
; ---------------------------------------------------------------------
;
; history point at tunnel roof
reset disp time hist
hide
seek range xr .44 .46
hist zdis 1.8 0 3.72
hist label 1 ’Vertical Displacement’
hide
seek range reg 8
;
plot create plot Hist
pl hist 1 yaxis label ’Displacement’
cycle 2000
;
save tun_lin3.3dsav
ret

Note that the prism-shaped blocks must be created before cycling is initiated. In this example, we
delete the blocks in region 5, and insert prism-shaped blocks for the liner (defined now as region
8). The m-d zoning is created with the GENERATE quad command, and only elastic behavior
is assigned to the liner material. If we wish to evaluate the elastic-plastic response, the bilinear
material model (ZONE model subi with the ubiquitous joint behavior suppressed) can be assigned
to the liner material. The liner supports the entire load in this example. (We could also reduce the
tractions as before, in Example 3.22.) Figure 3.37 illustrates the liner blocks for this case, and
Figure 3.38 shows the m-d zoning within the liner.

3DEC Version 5.2

3 - 80 User’s Guide

E

pGre00

0 u0
i
0pGGG
pGrelG,lrG0p5Ge5S:0

  50

Figure 3.37 Thick concrete liner support – prism-shaped liner blocks

E

pGre00

0 u0
i
0pGGG
pGrelG,lrG0p5Ge5S:0

  50 

Figure 3.38 Thick concrete liner support – mixed-discretization zoning in

liner blocks

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 81

Figure 3.39 shows the plot of the z-displacement in the roof for this case. Approximately 1.6 mm
displacement occurs when the liner supports the tunnel. The stresses in the liner are plotted in
Figure 3.40.

E

pGre00

0 u0
i
0pGGG
pGrelG,lrG0p5Ge5S:0 


r0$ 0% 

i0 


 
eH-cr











         

eH-r

Figure 3.39 z-displacement history at tunnel roof – support by prism-shaped

liner blocks

E

pGre00E

0 u0
i
0pGGG
pGrelG,lrG0p7G-7GG0


70 0
0
E  709 
" mo# moP"mo  ro
 

70 0
0
-ie,:-5lG.
lpiSGGG5lGr
l-iSGGG5lGr
lripSGG5nGG
lri-SGG5nGG
lpipSGG5nGG
lpi-SGG5nGG
l,ipSGG5nGG
l,i-SGG5nGG
l.ipSGG5nGG
l.i-SGG5nGG
l.i:.Gp5nGG

Figure 3.40 Principal stress distribution in top section of liner

3DEC Version 5.2

3 - 82 User’s Guide

3.7 Choice of Constitutive Model

This section provides an overview of the block and joint constitutive models in 3DEC, as well as
recommendations concerning when to use a model. Section 1 in Constitutive Models presents back-
ground information on the block constitutive model formulations. Also see user-defined models and
extended material models in Optional Features. The joint models are described in Section 1.2.2.3
in Theory and Background and Section 2 in Constitutive Models.

3.7.1 Deformable-Block Material Models

There are four built-in block material models in 3DEC:

(1) null (EXCAVATE command);
(2) elastic, isotropic (CHANGE cons = 1);
(3) elastic, anisotropic (CHANGE cons = 3); and
(4) Mohr-Coulomb plasticity (CHANGE cons = 2).

Note that the null model is assigned with the EXCAVATE command. The other four models are
assigned with the CHANGE cons command. Model properties are then specified for the non-null
models with the PROPERTY mat command for material property numbers, and the property numbers
are assigned to the blocks with the CHANGE mat command.
Each block model is designed to represent a specific type of constitutive behavior commonly
associated with geologic materials. The null model is used to represent material that is removed
from the model. The elastic, isotropic model is valid for homogeneous, isotropic, continuous
materials which exhibit linear stress-strain behavior. The elastic, anisotropic model is appropriate
for elastic materials that exhibit a well-defined elastic anisotropy. The Mohr-Coulomb plasticity
model is used for materials that yield when subjected to shear loading. But the yield stress depends
on the major and minor principal stresses only; the intermediate principal stress has no effect on
yield.
The material models in 3DEC are primarily intended for applications related to geotechnical en-
gineering (e.g., underground construction, mining, slope stability, foundations, earth and rock-fill
dams). When selecting a constitutive model for a particular engineering analysis, the following two
considerations should be kept in mind:
1. What are the known characteristics of the material being modeled?
2. What is the intended application of the model analysis?

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 83

Table 3.2 presents a summary of the 3DEC block models, along with examples of representative
materials and possible applications of the models. The elastic block model is generally applicable
for cases in which slip along discontinuities is the predominant mechanism for failure. The Mohr-
Coulomb model should be used when stress levels are such that failure of intact material is expected.
Mohr-Coulomb parameters for cohesion and friction angle are usually available more often than
other properties for geoengineering materials.

Table 3.2 3DEC block constitutive models

Model Representative Material Example Application

null void holes, excavations, regions in which

material will be added at later stage
elastic homogeneous, isotropic continuum; manufactured materials (e.g., steel)
linear stress-strain behavior loaded below strength limit; factor-of-
safety calculation
Drucker-Prager limited application; soft clays with low common model for comparison to
plasticity friction implicit finite-element programs
Mohr-Coulomb loose and cemented granular materials; general soil or rock mechanics (e.g.,
plasticity soils, rock, concrete slope stability and underground
excavation)

Other material models (e.g., Hoek-Brown, strain softening) are available with the ZONE model
command. Details are given in the Constitutive Models manual.

3.7.2 Joint Material Models

There are four built-in models available to represent the material behavior of discontinuities:
(1) Coulomb slip with weakening (CHANGE jcons = 1);
(2) perfectly plastic Coulomb slip (CHANGE jcons = 2);
(3) continuously yielding (CHANGE jcons = 3); and
(4) elastic (CHANGE jcons = 7).
The joint models are assigned to one or more contacts by using the CHANGE jcons command.
Joint model properties are then specified with the PROPERTY jmat command for material property
numbers, and the property numbers are assigned to the contacts with the CHANGE jmat command.
The joint constitutive models are designed to be representative of the physical response of rock
joints. The Coulomb slip model (the default) provides a linear representation of joint stiffness and

3DEC Version 5.2

3 - 84 User’s Guide

yield limit, and is based upon elastic stiffness, frictional, cohesive and tensile strength properties, and
dilation characteristics common to rock joints. The model simulates displacement-weakening of
the joint by loss of cohesive and tensile strength at the onset of shear or tensile failure. (A variation
of this model (CHANGE jcons = 2), in which the cohesion and tensile strength are maintained
following failure, is also available.) The continuously yielding joint model is a more complex
model that simulates continuous weakening behavior as a function of accumulated plastic-shear
displacement. See Section 2 in Constitutive Models for details.
Table 3.3 summarizes the 3DEC joint models, and presents examples of representative materials
and possible applications. The Coulomb slip model is most applicable for general engineering
studies. Coulomb friction and cohesion properties are usually available more often than other joint
properties.

Table 3.3 3DEC joint constitutive models

Model Representative Material Example Application

Coulomb joints, faults, bedding planes in rock general rock mechanics (e.g., under-
ground excavation)
continuously rock joints displaying progressive cyclic loading and load reversal with
yielding damage and hysteretic behavior predominant hysteretic loop; dynamic
analysis

3.7.3 Selection of an Appropriate Model

A problem analysis should always start with simple block and joint material models. In most cases,
an elastic block model (cons = 1) and a joint area contact Coulomb slip model (jcons = 1) should
be used first. The elastic block model only requires three material parameters: mass density, bulk
modulus and shear modulus (see Section 3.8.1.2). The Coulomb slip model requires six parameters:
normal and shear stiffness, friction angle, cohesion, tensile strength and dilation angle. Estimates
and references for these properties are given in Section 3.8.2. These material models provide a
simple perspective of stress-deformation behavior in the 3DEC model; the results of these analyses
can help the user assess whether a more complex (or simpler) material model is needed to describe
the block or joint behavior. For example, if the stresses and deformations in the blocks are low
compared to the joint movements, then a simpler, rigid block model may be sufficient.
It is often helpful to run simple tests of the selected material model before using it to solve the full-
scale boundary-value problem. This can provide insight into the expected response of the model
compared to the known response of the physical material.
The following example illustrates the use of a simple test model. The problem application is the
analysis of joint slip around an underground excavation. A simple model is created to evaluate the
adequacy of the Coulomb slip model to represent the response of a joint subjected to shear loading.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 85

The test is a simulation of a direct shear test, which consists of a single horizontal joint that is first
subjected to a normal confining stress, and then to a unidirectional shear displacement. Figure 3.41
shows the model; the joint is defined by one contact that is composed of 10 sub-contacts.

E

pGre00m

0 u0
i
0rSrGG
pGrelG,lrG0p:G5:rp0

m  :0x 
 
o 

 :0GiGGSrp.3r
:0piGG3Sr

Figure 3.41 Direct shear test model

First, a normal stress of 20 MPa, which is representative of the confining stress acting on the joint, is
applied. A horizontal velocity is then applied to the top block to produce a shear displacement that
is also representative of the displacements expected in the problem application. For demonstration
purposes, we only apply a small shear displacement of less than 1 mm to this model.
The average normal and shear stresses, and normal and shear displacements along the joint, are
measured with a FISH function (av str). With this information we can determine the peak and
residual shear strengths and dilation that are produced with the different models. The data file for
this test using the Coulomb slip model is

Example 3.24 Direct shear test with Coulomb slip model

new
; Coulomb slip joint model
; direct shear test
;
poly brick -0.15,0.15 -0.10,0.10 -0.10,0
poly brick -0.10,0.10 -0.10,0.10 0,0.10
plot create plot Blocks

3DEC Version 5.2

3 - 86 User’s Guide

plot block
plot reset
gen edge 0.2
plot reset
;pause key
;
prop mat=1 dens=0.0026 k=4000 g=3000
;
; Coulomb slip model
change jcons=1
prop jmat=1 jkn=100000 jks=100000 jfric=30.0 jdil 15 zdil 6e-4
;prop jmat=1 jcoh 10 ; use for weakening behavior
;
hide range z 0 1
boun xvel 0 yvel 0 zvel 0 range z -1 0.1
seek
;
; normal load
bound str 0 0 -50 0 0 0 range z 0.09 0.11
;
step 100
;
; function to calculate average joint stresses
; and average joint displacements
;
def av_str
whilestepping
sstav = 0.0
nstav = 0.0
njdisp = 0.0
sjdisp = 0.0
ncon = 0
jarea = 0.04
ic = contact_head
loop while ic # 0
icsub = c_cx(ic)
loop while icsub # 0
ncon = ncon + 1
sstav = sstav + cx_xsforce(icsub)
nstav = nstav + cx_nforce(icsub)
njdisp = njdisp + cx_ndis(icsub)
sjdisp = sjdisp + cx_xsdis(icsub)
icsub = cx_next(icsub)
endloop
if ncon # 0
sstav = sstav / jarea

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 87

nstav = nstav / jarea

njdisp = njdisp / ncon
sjdisp = - sjdisp / ncon
endif
ic = c_next{ic)
endloop
end
;
reset jdisp
; shear load
hide range z -.1 0.0
bound xvel=0.005 range z -.1 1.1
bound yvel 0 range z -1 1
seek
;
;
hist unbal nc 5
hist @sstav @nstav @njdisp @sjdisp
;
hist sdis -1 -1 0 ndis -1 -1 0
hist sdis -1 1 0 ndis -1 1 0
hist sdis 0 0 0 ndis 0 0 0
hist sstr -1 -1 0 nstr -1 -1 0
hist sstr -1 1 0 nstr -1 1 0
hist sstr 0 0 0 nstr 0 0 0
hist sfor -1 -1 0 nfor -1 -1 0
hist label 2 ’Shear Stress’
hist label 5 ’Shear Displacement’
hist label 4 ’Normal Displacement’

hist type 16
plot create plot Hist
plot hist 2 vs 5 yaxis label ’Shear Stress’ &
xaxis label ’Shear Displacement’
cyc 15000
;pause key
plot clear
plot hist 4 vs 5 yaxis label ’Normal Displacement’ &
xaxis label ’Shear Displacement’
save cs_1.3dsav
return

The average shear stress versus shear displacement along the joint is plotted in Figure 3.42, and
the average normal displacement versus shear displacement is plotted in Figure 3.43. These plots
indicate that joint slip occurs for the prescribed model properties and conditions. The loading
slope in Figure 3.42 is linear until a peak shear strength of approximately 2.9 MPa is reached. As

3DEC Version 5.2

3 - 88 User’s Guide

indicated in Figure 3.43, the joint begins to dilate when the joint fails in shear, at roughly 0.3 mm
shear displacement. Dilation occurs until the limiting shear displacement (zdilation = 0.6 mm) is
reached for zero dilation. The maximum average dilation is approximately 0.077 mm.

E
 
pGre00

0 u0
i
0rSrGG 
pGrelG,lrG0p:G5:rp0

 
p0&0
i0S0&0% 
 




 e
 e-^-

















         
 e 
e-^xH

Figure 3.42 Average shear stress versus shear displacement

– Coulomb slip model

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 89

E

pGre00h

0 u0
i
0rSrGG 
pGrelG,lrG0p:G5:rp0


.0 0% 
 
i0S0&0% 


$e 
e-^x5













         
 e 
e-^xN

Figure 3.43 Average normal displacement versus shear displacement

– Coulomb slip model

As these results indicate, the Coulomb slip model (jcons = 1) only defines a limiting shear strength
value for the joint. The dilation that occurs after the joint begins to slip is approximated as a linear
function of the dilation angle, with a dilation limit that is a function of the shear displacement.
(These functions are described in Section 1.2.2.3 in Theory and Background.)
Other modifications to the joint behavior are also available for the Coulomb slip model. For
example, a displacement-weakening behavior can be approximated by including a joint cohesion
of 10 MPa (PROPERTY jmat 1 coh = 10). At the onset of failure, the cohesion is set to zero. The
results shown in Figure 3.44 illustrate the peak and residual strengths that develop when the effect
of cohesion is included. Note that the drop in strength occurs abruptly. The maximum dilation (as
shown in Figure 3.36) is lower than the previous case without cohesion for the same limiting shear
displacement, because more shear displacement occurs before the joint fails initially. (Compare
Figure 3.45 to Figure 3.43.)

3DEC Version 5.2

3 - 90 User’s Guide

E
 
pGre00

0 u0
i
0rSrGG 
pGrelG,lrG0p:G5:p.0

 
p0&0
i0S0&0% 
 



 e
 e-^-











         
 e 
e-^xH

Figure 3.44 Average shear stress versus shear displacement

– Coulomb slip model with peak and residual strength

Note that if no weakening behavior is associated with a joint that has a cohesive strength, the
command CHANGE jcons = 2 should be given in place of CHANGE jcons = 1. In this case, there will
be no change in the cohesion when the joint fails.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 91

E
 
pGre00h

0 u0
i
0rSrGG 
pGrelG,lrG0p:G5:p.0

 
.0 0% 

i0S0&0% 



$e 
e-^x5

















         
 e 
e-^xN

Figure 3.45 Average normal displacement versus shear displacement

– Coulomb slip model with peak and residual strength

The displacement-weakening behavior is produced automatically with the continuously yielding

joint model (jcons = 3). This model simulates the progressive damage of the joint under shear.
For details and an example direct shear test with the continuously yielding model, see Section 2 in
Constitutive Models.
The material properties for the Coulomb slip model in these examples were selected to produce
roughly the same response for joint shear strength and dilation for joints subjected to unidirectional
shearing. For an actual application, properties should be selected (and adjusted as necessary)
to simulate the response of the joints under the expected loading conditions. In most cases, the
Coulomb model parameters are relatively easy to estimate (see Section 3.8.2), and the simple
modifications that are available with the Coulomb model may be sufficient to approximate the joint
behavior.
For other joint models, such as the continuously yielding model, the determination of properties
is more involved. In order to use the continuously yielding model, it is necessary to run a series
of joint shear tests to best-fit the model properties to physical test results. It is recommended that
simple shear tests always be performed, regardless of the joint model selected, to ensure that the
joint behaves as expected under the anticipated problem conditions.
If it is necessary to simulate a complicated joint response, then a more complex joint model may be
required. However, before going to a more complex model, it is usually helpful to apply a simple
model first, to establish a basis for evaluating the influence of the more complicated joint behavior.

3DEC Version 5.2

3 - 92 User’s Guide

3.8 Material Properties

3DEC requires material properties for both the intact blocks and the discontinuities. This section
provides an overview of typical properties used to represent the behavior of jointed rock, and
presents guidelines for selecting the appropriate properties for a given model. There are also
special considerations, such as the definition of post-failure properties and the extrapolation of
laboratory-measured properties to the field scale. These topics are also discussed.
The selection of properties is often the most difficult element in the generation of a model because
of the high uncertainty in the property database. It should be kept in mind when performing an
analysis, especially in geomechanics, that the problem will always involve a data-limited system;
the field data will never be known completely. However, with the appropriate selection of properties
based on the available database, important insight to the physical problem can still be achieved.
This approach to modeling is discussed further in Section 3.11.

3.8.1 Block Properties

Properties assigned to blocks are generally derived from laboratory testing programs. The following
four sections describe intrinsic (laboratory-scale) properties and list common values for various
rocks.
3.8.1.1 Mass Density

The mass density is required for every non-void material in a 3DEC model. This property has units
of mass divided by volume, and does not include the gravitational acceleration. In many cases,
the unit weight of a material is prescribed. If the unit weight is given with units of force divided
by volume, then this value must be divided by the gravitational acceleration before entering it as
3DEC input for density.
3.8.1.2 Intrinsic Deformability Properties

All material models for deformable blocks in 3DEC assume an isotropic material behavior in the
elastic range described by two elastic constants (bulk modulus, K, and shear modulus, G). The
elastic constants, K and G, are used in 3DEC instead of Young’s modulus, E, and Poisson’s ratio,
ν, because it is believed that bulk and shear moduli correspond to more fundamental aspects of
material behavior than do Young’s modulus and Poisson’s ratio. (See note 8 in Section 3.9 for
justification for using (K,G) rather than (E,ν).)
The equations to convert from (E,ν) to (K,G) are:

E
K=
3(1 − 2ν)
(3.15)
E
G=
2(1 + ν)

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 93

Eq. (3.15) should not be used blindly when ν is near 0.5, since the computed value of K will be
unrealistically high and convergence to the solution will be very slow. It is better to fix the value
of K at its known physical value (estimated from an isotropic compaction test or from the p-wave
speed), and then compute G from K and ν.
Some typical values for elastic constants are summarized in Table 3.4 for selected rocks:

Table 3.4 Selected elastic constants (laboratory-scale) for rocks

(adapted from Goodman 1980)
E (GPa) ν K (GPa) G (GPa)
Berea sandstone 19.3 0.38 26.8 7.0
Hackensack siltstone 26.3 0.22 15.6 10.8
Bedford limestone 28.5 0.29 22.6 11.1
Micaceous shale 11.1 0.29 8.8 4.3
Cherokee marble 55.8 0.25 37.2 22.3
Nevada Test Site granite 73.8 0.22 43.9 30.2

3.8.1.3 Intrinsic Strength Properties

The basic criterion for block material failure in 3DEC is the Mohr-Coulomb relation, which is a
linear failure surface corresponding to shear failure:


fs = σ1 − σ3 Nφ + 2c Nφ (3.16)

where: Nφ = (1 + sin φ)/(1 − sin φ);

σ1 = major principal stress (compressive stress is negative);
σ3 = minor principal stress;
φ = friction angle; and
c = cohesion.
Shear yield is detected if fs < 0. The two strength constants, φ and c, are conventionally derived
from laboratory triaxial tests.

3DEC Version 5.2

3 - 94 User’s Guide

The Mohr-Coulomb criterion loses its physical validity when the normal stress becomes tensile.
But, for simplicity, the surface is extended into the tensile region to the point at which σ3 equals the
uniaxial tensile strength, σ t . The minor principal stress can never exceed the tensile strength – i.e.,

ft = σ3 − σ t (3.17)

Tensile yield is detected if ft > 0. Tensile strength for rock and concrete is usually derived from
a Brazilian (or indirect tensile) test. Note that the tensile strength cannot exceed the value of σ3
corresponding to the apex limit for the Mohr-Coulomb relation. This maximum value is given by

c
t
σmax = (3.18)
tan φ

Typical values of cohesion, friction angle and tensile strength for a representative set of rock
specimens are listed in Table 3.5:

Table 3.5 Selected strength properties (laboratory-scale) for rocks

(adapted from Goodman 1980)
friction cohesion tensile
angle (MPa) strength
(degrees) (MPa)
Berea sandstone 27.8 27.2 1.17
Repetto siltstone 32.1 34.7 —
Muddy shale 14.4 38.4 —
Sioux quartzite 48.0 70.6 —
Indiana limestone 42.0 6.72 1.58
Stone Mountain granite 51.0 55.1 —
Nevada Test Site basalt 31.0 66.2 13.1

The ubiquitous-joint component of the bilinear model also requires strength properties for the planes
of weakness. Joint properties are discussed in Section 3.8.2. The properties for joint cohesion and
friction angle also apply for the ubiquitous-joint model.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 95

3.8.1.4 Post-Failure Properties

In many instances, particularly in mining engineering, the response of a material after the onset
of failure is an important factor in the engineering design. Consequently, the post-failure behavior
must be simulated in the material model. In 3DEC, this is accomplished with properties that define
four types of post-failure response:
(1) shear dilatancy;
(2) shear hardening/softening;
(3) volumetric hardening/softening; and
(4) tensile softening.
These properties are only activated after the onset of failure, as defined by the Mohr-Coulomb
relation. Shear dilatancy is assigned for the Mohr-Coulomb and bilinear strain-hardening/softening
ubiquitous joint model. Hardening/softening parameters are assigned for the bilinear model.

Shear Dilatancy – Shear dilatancy, or dilatancy, is the change in volume that occurs with shear
distortion of a material. Dilatancy is characterized by a dilation angle, ψ, which is related to the
ratio of plastic volume change to plastic shear strain. This angle can be specified in the block
plasticity models in 3DEC. The dilation angle is typically determined from triaxial tests or shear
box tests. For example, the idealized relation for dilatancy, based on the Mohr-Coulomb failure
surface, is depicted for a triaxial test in Figure 3.46. The dilation angle is found from the plot
of volumetric strain versus axial strain. Note that the initial slope for this plot corresponds to the
elastic regime, while the slope used to measure the dilation angle corresponds to the plastic regime.

|I1 - I3|

I1
2 c cos B - (I1 - I3) sin B
E
1

A1
elastic plastic

I2 = I3
Av
I3

atan (1-2K) 2 sin O

atan
1 - sin O
A1

Figure 3.46 Idealized relation for dilation angle, ψ, from triaxial test results
(Vermeer and de Borst 1984)

3DEC Version 5.2

3 - 96 User’s Guide

For soils, rocks and concrete, the dilation angle is generally significantly smaller than the friction
angle of the material. Vermeer and de Borst (1984) report the following typical values for ψ:
dense sand 15◦
loose sand < 10◦
normally consolidated clay 0◦
granulated and intact marble 12◦ − 20◦
concrete 12◦
Vermeer and de Borst observe that values for the dilation angle are approximately between 0◦ and
20◦ , whether the material is soil, rock or concrete. The default value for the dilation angle is zero
for all of the constitutive models in 3DEC.
The dilation angle can also be prescribed for the joints in the ubiquitous-joint component of the
bilinear model. This property is typically determined from direct shear tests, and common values
can be found in the references discussed in Section 3.8.2.
Shear Hardening/Softening – The initiation of material hardening or softening is a gradual process
once plastic yield begins. At failure, deformation becomes more and more inelastic as a result of
micro-cracking in concrete and rock, and particle sliding in soil. This also leads to degradation
of strength in these materials and the initiation of shear bands. These phenomena, related to
localization, are discussed further in Section 3.11.
In 3DEC, shear hardening and softening are simulated by making Mohr-Coulomb properties (cohe-
sion and friction, along with dilation) functions of plastic strain (see Section 1.6.4 in Constitutive
Models). These functions are accessed from the bilinear model, and can be specified by using the
TABLE command.
Hardening and softening parameters must be calibrated for each specific analysis with values that
are generally back-calculated from results of laboratory triaxial tests. This is usually an iterative
process. Investigators have developed expressions for hardening and softening. For example,
Vermeer and de Borst (1984) propose the frictional hardening relation

√
ep ef
sin φm = 2 sin φ for ep ≤ ef
ep + ef
(3.19)
sin φm = sin φ for ep > ef

where: φ = ultimate friction angle;

φm = mobilized friction angle;
ep = plastic strain; and
ef = hardening constant.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 97

Numerical testing conditions can influence the model response for shear hardening/softening be-
havior. The rate of loading can introduce inertial effects; this can be controlled by monitoring
the unbalanced force and reducing the loading rate accordingly. A FISH function can be used to
control the loading rate automatically. The results are also mesh-dependent. Thus, it is important
to evaluate the model behavior for differing zone sizes and mesh orientations whenever performing
an analysis involving shear hardening or softening.
Tensile Softening – At the initiation of tensile failure, the tensile strength of a material will generally
drop to zero. The Mohr-Coulomb model (cons = 2) does not strain-soften in tensile failure. The
tensile stress remains constant after the tensile stress limit has been reached. The bilinear zone
model (subi) will also have a constant tensile strength unless a tensile softening table is defined.
The rate at which the tensile strength drops, or tensile softening occurs, can also be controlled by
the plastic tensile strain in 3DEC. This function is accessed from the bilinear model (subi), and
can be specified by using the TABLE command.
A simple tension test (Example 3.25) illustrates perfectly plastic tensile failure, as built into the
Mohr-Coulomb model, and brittle tensile failure in the bilinear model (subi). The model is a tension
test on a cubic block composed of Mohr-Coulomb material. The ends of the sample are pulled apart
at a constant velocity. The test is performed with both the cons 2 and the subi block models.

Example 3.25 Tension test on tensile-softening material

new
set small
poly brick 0 1 0 1 0 1
plot create plot Blocks
plot block
plot reset
gen quad ndiv 1 1 1 rmul 1
; Mohr-Coulomb (cons = 2) model
change cons 2
prop mat=1 dens=2500 k=1.19e10 g=1.1e10 bcoh 2.72e5 phi 44 bten 2e5
; bilinear model
;config cpp
;zone model subi
;prop mat 1 dens 2500
;zone bulk=1.19e10 shear=1.1e10 coh 2.72e5 fric 44 ten 2e5
;zone coh 1e20 jten 1e20 fric 44
;zone ttab 1
;table 1 0 2e5 9e-6 0
;
bound zvel -1e-5 range z 0.0
bound zvel 1e-5 range z 1.0
def ax_str
str = 0.0
ib = block_head

3DEC Version 5.2

3 - 98 User’s Guide

ig = b_gp(ib)
nx = 0
ny = 0
xbpos = 0.0
xbdis = 0.0
xtpos = 0.0
xtdis = 0.0
zbpos = 0.0
zbdis = 0.0
ztpos = 0.0
ztdis = 0.0
loop while ig # 0
if gp_z(ig) > ztop then
igb = gp_bou(ig)
str = str + bou_zforce{igb)
ztpos = ztpos + gp_z(ig)
ztdis = ztdis + gp_zdis(ig)
end_if
if gp_z(ig) < zbot then
nz = nz + 1
zbpos = zbpos + gp_z(ig)
zbdis = zbdis + gp_zdis(ig)
end_if
if gp_x(ig) < xbot then
nx = nx + 1
xbpos = xbpos + gp_x(ig)
xbdis = xbdis + gp_xdis(ig)
end_if
if gp_x(ig) > xtop then
xtpos = xtpos + gp_x(ig)
xtdis = xtdis + gp_xdis(ig)
end_if
ig = gp_next(ig)
end_loop
ax_str = str / area
xbdis = xbdis / nx
xbpos = xbpos / nx
xtdis = xtdis / nx
xtpos = xtpos / nx
zbdis = zbdis / nz
zbpos = zbpos / nz
ztdis = ztdis / nz
ztpos = ztpos / nz
ex_str = (xtdis - xbdis) / (xtpos - xbpos)
ez_str = (ztdis - zbdis) / (ztpos - zbpos)
end

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 99

set @area = 1.0 @ztop = 0.9 @zbot = 0.1 @xbot = 0.1 @xtop = 0.9
hist @ax_str @ex_str @ez_str
damp local
plot create plot Hist
plot his 1 vs 3 xaxis label ’Vertical Strain’ &
yaxis label ’Vertical Stress’
step 20000
save mc.sav

plot clear
plot his 2 vs 3 xaxis label ’Vertical Strain’ &
yaxis label ’Horizontal Strain’
ret

The plot of σzz -stress versus zz-strain (Figure 3.47) shows that the average stress remains constant
for the Mohr-Coulomb model in cons 2. The stress will remain constant in the bilinear model
without tensile softening. The brittleness of the tensile softening can be controlled by the plastic
tensile-strain function. If Example 3.25 is repeated with the bilinear model (subi) and a tensile
softening table, an instantaneous softening response can be produced, as shown in Figure 3.48.

E

pGre00v

0 u0
i
0pGGGG 
pGrelG,lrG0p7G-7,p0


r0'0c()C 
i0,0*'0c()C


 
r
 roH%













         
 
r
 roH-%

Figure 3.47 σzz -stress versus zz-strain for tension test with cons 2 model

3DEC Version 5.2

3 - 100 User’s Guide

E

pGre00)

0 u0
i
0pGGGG 
pGrelG,lrG0p7G-7.G0


r0'0c()C 
i0,0*'0c()C



 
r
 roH%













         
 
r
 roH-%

Figure 3.48 σzz -stress versus zz-strain for tension test with subi model and
tensile-softening table

The average xx-strain and yy-strain across the model decrease until tensile failure is initiated. With
the cons 2 model, this strain is not controlled in the post-failure region; the model will continue to
contract as indicated by the plot of xx-strain versus zz-strain in Figure 3.49.
With the bilinear model, the strain is affected after the onset of tensile failure; the model expands
in the x- and y-directions as tensile softening occurs, as indicated in Figure 3.50.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 101

E

pGre00v

0 u0
i
0pGGGG 
pGrelG,lrG0p7G-7,p0

 
p0'0c()C
i0,0*'0c()C




&
r
 rzo-














         
 
r
 rzo-%

Figure 3.49 xx-strain versus zz-strain for tension test with cons 2 model

E

pGre00)

0 u0
i

0pGGGG
pGrelG,lrG0p7G-7.G0



p0'0c()C
i0,0*'0c()C 


&
r
 rzo-%


















         
 
r
 rzo-%

Figure 3.50 xx-strain versus zz-strain for tension test with subi model and
tensile-softening table

3DEC Version 5.2

3 - 102 User’s Guide

Note that local damping (DAMPING local) is used to minimize oscillations that can arise when
the abrupt tensile failure occurs. Alternatively, if adaptive global damping is used by giving the
DAMPING auto command, oscillations are observed in the stress/strain plots. With adaptive global
damping, the damping parameter is continually decreased as the model is stretched. When tensile
failure occurs, the global damping parameter is low, and oscillations that may affect the final solution
state are produced. With local damping, the amount of damping varies from gridpoint to gridpoint,
and is proportional to the unbalanced force. This damping minimizes the oscillations that are
produced when the abrupt tensile failure occurs. (See Section 1.2.3.2 in Theory and Background
for further discussion on damping.)
The brittleness of the tensile softening can be controlled by the plastic tensile strain function, by
using the bilinear model instead of the Mohr-Coulomb model. As with the shear-softening, the
tensile-softening must be calibrated for each specific problem and mesh size, since the results will
be mesh-dependent.
3.8.1.5 Extrapolation to Field-Scale Properties

The material properties used in the 3DEC model should correspond as closely as possible to the
actual values of the physical problem. Laboratory-measured properties generally should not be used
directly in a 3DEC model for a full-scale problem. The presence of discontinuities in the model
will account for a good portion of the scaling effect on properties. However, some adjustment of
block properties will still probably be required to represent the influence of heterogeneities and
micro-fractures, fissures and other small discontinuities on the rock mass response.
Several empirical approaches have been proposed to derive field-scale properties. Some of the more
commonly accepted methods are discussed.
Deformability of a rock mass is generally defined by a modulus of deformation, Em . If the rock
mass contains a set of relatively parallel, continuous joints with uniform spacing, the value for Em
can be estimated by treating the rock mass as an equivalent transversely isotropic continuum. The
relations in Section 3.8.2 can then be used to estimate Em in the direction normal to the joint set.
Deformation moduli can also be estimated for cases involving more than one set of discontinuities.
The references listed in Section 3.8.2 provide solutions for multiple joint sets.
In practice, the rock mass structure is often much too irregular, or sufficient data are not available,
to use the preceding approach. It is common to determine Em from a force-displacement curve
obtained from an in-situ compression test. Such tests include plate bearing tests, flatjack tests and
dilatometer tests.
Bieniawski (1978) developed an empirical relation for Em based on field test results at sites through-
out the world. The relation is based on rock mass rating (RMR). For rocks with a rating higher than
55, the test data can be approximately fit to

Em = 2(RMR) − 100 (3.20)

The units of Em are GPa.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 103

For values of Em between 1 and 10 GPa, Serafim and Pereira (1983) found a better fit, given by

RMR−10
Em = 10 40 (3.21)

References by Goodman (1980) and Brady and Brown (1985) provide additional discussion on
these methods.
The most commonly accepted approach to estimate rock mass strength is that proposed by Hoek
and Brown (1980). They developed the empirical rock mass strength criterion,

σ1s = σ3 + (mσc σ3 + sσc2 )1/2 (3.22)

where: σ1s = major principal stress at peak strength;

σ3 = minor principal stress;
m and s = constants that depend on the properties of the rock, and the
extent to which it has been broken before being subjected to
failure stresses; and
σc = uniaxial compressive strength of intact rock material.
The unconfined compressive strength for a rock mass is given by

qm = σc s 1/2 (3.23)

and the uniaxial tensile strength of a rock mass is

1
σt = σc [m − (m2 + 4s)1/2 ] (3.24)
2

Table 3.6 (from Hoek and Brown 1988) presents typical values for m and s for undisturbed and
disturbed rock masses.
It is possible to estimate Mohr-Coulomb friction angle and cohesion from the Hoek-Brown criterion
(see, for example, Hoek 1990).
For a given value of σ3 , a tangent to the function (Eq. (3.22)) will represent an equivalent Mohr-
Coulomb yield criterion in the form

σ1 = Nφ σ3 + σcM (3.25)

where Nφ = 1+sin φ
1−sin φ = tan2 ( φ2 + 45◦ )

3DEC Version 5.2

3 - 104 User’s Guide

By substitution, σcM is

σcM = σ1 − σ3 Nφ = σ3 + σ3 σc m + σc2 s − σ3 Nφ = σ3 (1 − Nφ ) + σ3 σc m + σc2 s

σcM is the apparent uniaxial compressive strength of the rock mass for that value of σ3 .
The tangent to the Eq. (3.22) is defined by

∂σ1 σc m
Nφ (σ3 ) = = 1 +  (3.26)
∂σ3 2 σ3 · σc m + sσc2

The cohesion (c) and friction angle (φ) can then be obtained from Nφ and σcM :


φ = 2 tan−1 Nφ − 90◦ (3.27)

σM
c = c (3.28)
2 Nφ

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 105

Table 3.6 Typical values for Hoek-Brown rock-mass strength parameters

(adapted from Hoek and Brown (1988))
Disturbed rock mass m and s values Undisturbed rock mass m and s values

FINE-GRAINED POLYMINERALLIC IGNEOUS

IGNEOUS & METAMORPHIC CRYSTALLINE

ROCKS — amphibolite, gabbro gneiss, granite,
CRYSTALLINE ROCKS — andesite, dolerite,
mudstone, siltstone, shale and slate (normal to

CRYSTALS AND POORLY DEVELOPED

ARENACEOUS ROCKS WITH STRONG
LITHIFIED ARGILLACEOUS ROCKS —
DEVELOPED CRYSTAL CLEAVAGE —

COARSE-GRAINED POLYMINERALLIC
CRYSTAL CLEAVAGE — sandstone and
CARBONATE ROCKS WITH WELL-
EMPIRICAL FAILURE CRITERION

dolomite, limestone and marble

σ'1 = σ'3 ÷ √(mσcσ'3 ÷ sσ2c)
σ'1 = major principal effective stress

norite, quartz-diorite
σ'3 = minor principaI effective stress

diabase and thyolite

σc = uniaxial compressive strength
of intact rock, and

cleavage)
m and s are empirical constants.

quartzite
INTACT ROCK SAMPLES
Laboratory specimens free m 7.00 10.00 15.00 17.00 25.00
from discontinuities s 1.00 1.00 1.00 1.00 1.00
CSIR rating: RMR = 100 m 7.00 10.00 15.00 17.00 25.00
NGI rating: Q = 500 s 1.00 1.00 1.00 1.00 1.00
VERY GOOD QUALITY ROCK MASS
Tightly interlocking undisturbed rock m 2.40 3.43 5.14 5.82 8.56
with unweathered joints at 1 to 3 m s 0.082 0.082 0.082 0.082 0.082
CSIR rating RMR = 85 m 4.10 5.85 8.78 9.95 14.63
NGI rating: Q = 100 s 0.189 0.189 0.189 0.189 0.189
GOOD QUALITY ROCK MASS
Fresh to slightly weathered rock, slightly m 0.575 0.821 1.231 1.395 2.052
disturbed with joints at 1 to 3 m s 0.00293 0.00293 0.00293 0.00293 0.00293
CSIR rating: RMR = 65 m 2.006 2.865 4.298 4.871 7.163
NGI rating: Q = 10 s 0.0205 0.0205 0.0205 0.0205 0.0205
FAIR QUALITY ROCK MASS
Several sets of moderately weathered m 0.128 0.183 0.275 0.311 0.458
joints spaced at 0.3 to 1 m s 0.00009 0.00009 0.00009 0.00009 0.00009
CSIR rating: RMR = 44 m 0.947 1.353 2.03 2.301 3.383
NGI rating: Q = 1 s 0.00198 0.00198 0.00198 0.00198 0.00198

POOR QUALITY ROCK MASS

Numerous weathered joints at 30-500 mm, m 0.029 0.041 0.061 0.069 0.102
some gouge; clean compacted waste rock s 0.000003 0.000003 0.000003 0.000003 0.000003
CSIR rating: RMR = 23 m 0.447 0.639 0.959 1.087 1.598
NGI rating: Q = 0.1 s 0.00019 0.00019 0.00019 0.00019 0.00019

VERY POOR QUALITY ROCK MASS

Numerous heavily weathered joints spaced m 0.007 0.01 0.015 0.017 0.025
<50 mm with gouge; waste rock with fines s 0.0000001 0.0000001 0.0000001 0.0000001 0.0000001
CSIR rating: RMR = 3 m 0.219 0.313 0.469 0.532 0.782
NGI rating: Q = 0.01 s 0.00002 0.00002 0.00002 0.00002 0.00002

3DEC Version 5.2

3 - 106 User’s Guide

3.8.2 Joint Properties

Joint properties are conventionally derived from laboratory testing (e.g., triaxial and direct shear
tests). These tests can produce physical properties for joint friction angle, cohesion, dilation angle
and tensile strength, as well as joint normal and shear stiffnesses. The joint cohesion and friction
angle correspond to the parameters in the Coulomb strength criterion.
Values for normal and shear stiffnesses for rock joints typically can range from roughly 10 to 100
MPa/m for joints with soft clay in-filling, to over 100 GPa/m for tight joints in granite and basalt.
Published data on stiffness properties for rock joints are limited; summaries of data can be found
in Kulhawy (1975), Rosso (1976) and Bandis et al. (1983).
Approximate stiffness values can be back-calculated from information on the deformability and
joint structure in the jointed rock mass and the deformability of the intact rock. If the jointed rock
mass is assumed to have the same deformational response as an equivalent elastic continuum, then
relations between jointed rock properties and equivalent continuum properties can be derived.
For uniaxial loading of rock containing a single set of uniformly spaced joints oriented normal to
the direction of loading, the following relation applies:

1 1 1
= +
Em Er kn s

or

Em Er
kn = (3.29)
s (Er − Em )

where:Em = rock mass Young’s modulus;

Er = intact rock Young’s modulus;
kn = joint normal stiffness; and
s = joint spacing.

A similar expression can be derived for joint shear stiffness:

Gm Gr
ks = (3.30)
s (Gr − Gm )

where:Gm = rock mass shear modulus;

Gr = intact rock shear modulus; and
ks = joint shear stiffness.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 107

The equivalent continuum assumption, when extended to three orthogonal joint sets, produces the
following relations:

 −1
1 1
Ei = + (i = 1, 2, 3) (3.31)
Er si kni
 −1
1 1 1
Gij = + + (i , j = 1, 2, 3) (3.32)
Gr si ksi sj ksj

Several expressions have been derived for two- and three-dimensional characterizations and multiple
joint sets. References for these derivations can be found in Singh (1973), Gerrard (1982(a) and (b))
and Fossum (1985).
There is a limit to the maximum joint stiffnesses that are reasonable to use in a 3DEC model. If
the physical normal and shear stiffnesses are less than ten times the equivalent stiffness of adjacent
zones (see Eq. (3.33) in Section 3.9), then there is no problem with using physical values. If the
ratio is more than ten, the solution time will be significantly longer than for the case in which the
ratio is limited to ten, without much change in the behavior of the system. Serious consideration
should be given to reducing supplied values of normal and shear stiffnesses to improve solution
efficiency. There may also be problems with block interpenetration if the normal stiffness, kn , is
very low. A rough estimate should be made of the joint normal displacement that would result from
the application of typical stresses in the system (u = σ/kn ). This displacement should be small
compared to a typical zone size. If it is greater than, say, 10% of an adjacent zone size, then either
there is an error in one of the numbers or the stiffness should be increased.
Published strength properties are more readily available for joints than for stiffness properties.
Summaries can be found, for example, in Jaeger and Cook (1969), Kulhawy (1975) and Barton
(1976). Friction angles can vary from less than 10◦ for smooth joints in weak rock (such as tuff),
to over 50◦ for rough joints in hard rock (such as granite). Joint cohesion can range from zero
cohesion to values approaching the compressive strength of the surrounding rock.
It is important to recognize that joint properties measured in the laboratory typically are not rep-
resentative of those for real joints in the field. Scale dependence of joint properties is a major
question in rock mechanics. Often, the only way to guide the choice of appropriate parameters is
by comparison to similar joint properties derived from field tests; however, field test observations
are extremely limited. Some results are reported by Kulhawy (1975).

3DEC Version 5.2

3 - 108 User’s Guide

3.9 Tips and Advice

When problem solving with 3DEC, it is often important to try to optimize the model for the most
efficient analysis. This section provides several suggestions on ways to improve a model run. Some
common pitfalls that should be avoided when preparing a 3DEC calculation are also listed.
1. Setting Model Tolerances
There is one main tolerance (ATOL) that is used globally to control the oper-
ations of 3DEC. This tolerance or derivatives of it are used to control:
minimum distance between gridpoints;
minimum area (atol*atol) of sub-contacts;
tolerance to link gridpoints of joined blocks;
tolerance to select vertices close to the command plane;
planarity of block cutting;
contact detection;
minimum block edge length; and
joint plane plot
The default value of ATOL is calculated based on the size of the average model
dimensions in x, y and z. The default value calculated by 3DEC is not always
appropriate for all models. The validity of the automatically assigned ATOL
decreases as the range in block sizes increases.
Some of the side effects of having an ATOL that is too small are:
creation of very small blocks during block cutting
thin blocks that collapse during cycling
zero or negative zone volume during cycling
very small timesteps and long solve times
Some of the side effects of having an ATOL that is too large are:
rejected cuts during block cutting;
nonplanar block faces;
contacts that form across thin blocks;

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 109

gridpoints that are slaved when they should not be (block join logic);
and
inability to calculate the normals to some block faces.
It is recommended that users define an ATOL prior to cutting any blocks with
JSET commands. We do not recommend changing ATOL after cutting blocks.
The value of ATOL should not be greater that 20% of the smallest block edge
or zone edge in the model. The value of ATOL can be made smaller that this
value to resolve issues such as rejected cuts to blocks when using the JSET
command.
2. Designing the Model
It is tempting to try to build as much detail of the geologic structure as possible
into a 3DEC model. There are three main arguments against this approach:
(1) It is futile to ever expect to have sufficient data to model a jointed rock mass
in every detail. (2) The computer hardware requirements for a detailed model
quickly exceed what is typically available for engineering projects. (3) Most
importantly, a controlled engineering understanding of model results becomes
less effective as more detail is added.
Two considerations should be kept in mind when creating the 3DEC model.
The first is whether a discontinuum analysis is actually required. This depends
in large part on the ratio of the scale of the physical system under investigation
to the average spacing of the joint structure. For example, if an excavation
is made in a rock mass containing a single joint set with an average spacing
of 1 m or less, and the minimum dimension of the excavation is 10 m, then
a continuum analysis with a ubiquitous-joint material model may be a more
reasonable approach. At this scale, the continuum analysis can produce a re-
sponse that is broadly equivalent to that obtained when the joints are explicitly
modeled. The analysis with 3DEC provides a more detailed analysis of the
failure mechanism, but may require more computation time than the contin-
uum analysis. In instances where the ratio of the physical system scale to
the joint spacing exceeds approximately 10:1, a continuum analysis may be
preferred. Analyses should be conducted with both continuum and discontin-
uum analyses when there is a question as to whether the continuum analysis
is sufficient to represent the discontinuum response. See Board et al. (1996)
for an illustration of the application of discontinuum and continuum analyses
(using UDEC and FLAC) in a comparative study of toppling behavior of a
jointed rock slope.
The second consideration is the extent to which the detailed geologic structure
should be included in the model. A detailed representation that includes the
most critical joint structure (see Section 3.2.7) is usually only required within
a limited region surrounding the area of interest (e.g., within a few tunnel radii
surrounding a tunnel excavation). Generally, the greater jointing detail only
needs to extend from the area of interest to a distance sufficient to encompass

3DEC Version 5.2

3 - 110 User’s Guide

the region in which failure is anticipated. That is, the detailed geologic struc-
ture should extend beyond the distance to which joint slip and separation are
calculated.
3. Check Model Runtime
The solution time for a 3DEC run is a function of both the number of rigid
blocks or gridpoints in deformable blocks, and the number of contacts in a
model. If there are very few contacts in the model, then the time is pro-
portional to N 3/2 , where N is the number of rigid blocks or gridpoints in
deformable blocks. This formula holds for elastic problems. The runtime will
vary somewhat, but not substantially, for plasticity problems.
The solution time will increase as more contacts are created in the model. It
is important to check the speed of calculation on your computer for a specific
model. An easy way to do this is to run the benchmark test described in
Section 9. Then use this speed to estimate the speed of calculation for the
specific model, based on interpolation from the number of gridpoints and
contacts.
4. Effects on Runtime
3DEC will take a longer time to converge if:
(a) there are large contrasts in stiffness in block materials or in joint
materials, or in block versus joint materials; or
(b) there are large contrasts in block or zone sizes.
The code becomes less efficient as these contrasts become greater. The effect
of a contrast in stiffness should be investigated before performing a detailed
analysis. For example, for mechanical-only calculations, joint normal and
shear stiffnesses should be kept smaller than ten times the equivalent stiffness
of the stiffest neighboring zone in blocks adjoining the joint – i.e.,

  
K + 4/3G
kn and ks ≤ 10.0 max (3.33)
zmin

where K and G are the bulk and shear moduli, respectively, of the block
material, and zmin is the smallest dimension of the zone adjoining the joint
in the normal direction. If the joint stiffnesses are greater than 10 times the
equivalent stiffness, the solution time of the model will be significantly longer
than for the case in which the ratio is limited to ten, without a significant
change in the behavior of the system.
On the other hand, there may be problems if the normal stiffness, kn , is very
low. A rough estimate of the joint normal displacement that would result from

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 111

the application of typical stresses in the system (u = σ/kn ) should be made.

This displacement should be small compared to a typical zone size. If it is
greater than roughly 10% of an adjacent zone size, then either there is an error
in one of the numbers or the stiffness should be increased.
5. Considerations for Density of Zoning
3DEC uses constant-strain elements in deformable blocks. If the gradient
of stress/strain is high, many zones will be needed to represent the varying
distribution. Run the same problem with different zoning densities to check
the effect. Constant-strain zones are used in 3DEC because a better accuracy
is achieved when plastic flow is modeled with many low-order elements rather
than with a few high-order elements.
Try to keep the zoning as uniform as possible, particularly in the region of
interest. Avoid long, thin zones with aspect ratios greater than 5:1.
6. Check Model Response
3DEC shows how a system behaves. Make frequent simple tests to check
whether you are doing what you think you are doing. For example, if a
loading condition and geometry are symmetrical, make sure that the response
is symmetrical. After making a change in the model, execute a few calculation
steps (for example, 5 or 10) to verify that the initial response is of the correct
sign and in the correct location. Do back-of-the-envelope estimates of the
expected order of magnitude of stress or displacements, and compare to the
3DEC output.
If you apply a violent shock to the model, you will get a violent response. If
you do physically unreasonable things to the model, you must expect strange
results. If you get unexpected results at a given stage of an analysis, review
the steps you followed up to this stage.
Critically examine the output before proceeding with the model simulation.
If, for example, everything appears reasonable except for large velocities in
one corner block, do not go on until you understand the reason for this. In this
case, you may not have fixed a boundary corner properly.
7. Use Bulk and Shear Moduli
It is better to use bulk modulus, K, and shear modulus, G, than Young’s
modulus, E, and Poisson’s ratio, ν, for elastic properties in 3DEC.
The pair (K, G) makes sense for all elastic materials that do not violate thermo-
dynamic principles. The pair (E, ν) does not make sense for certain admissible
materials. At one extreme we have materials that resist volumetric change but
not shear; at the other extreme, materials that resist shear but not volumetric
change. The first type of material corresponds to finite K and zero G; the sec-
ond to zero K and finite G. However, the pair (E, ν) is not able to characterize

3DEC Version 5.2

3 - 112 User’s Guide

either the first or second type of material. If we exclude the two limiting cases
(conventionally, ν = 0.5 and ν = −1), the equations

3K(1 − 2ν) = E
(3.34)
2G(1 + ν) = E

relate the two sets of constants. These equations hold, however close we ap-
proach (but do not reach) the limiting cases. We do not need to relate them
to physical tests that may or may not be feasible; the equations are simply the
consequence of two possible ways of defining coefficients of proportionality.
Suppose we have a material in which the resistance to distortion progressively
reduces, but in which the resistance to volume change remains constant. ν
approaches 0.5 in this case. The equation 3K(1 − 2ν) = E must still be satis-
fied. There are two possibilities (argued on algebraic, not physical, grounds):
either E remains finite (and nonzero) and K tends to an arbitrarily large value;
or K remains finite and E tends to zero. We rule out the first possibility be-
cause there is a limiting compressibility to all known materials (e.g., 2 GPa for
water, which has a Poisson’s ratio of 0.5). This leaves the second, in which E
is varying drastically, even though we supposed that the material’s principal
mode of elastic resistance was unchanging. We deduce that the parameters
(E, ν) are inadequate to express the material behavior.
8. Choice of Damping
In most instances, it is recommended that DAMPING local be used for static
analyses. This is generally appropriate for static analysis, for the reasons
given in Section 1.2.3.2 in Theory and Background. Also, as demonstrated
in Example 3.25, local damping is more suitable to minimize oscillations that
may arise when abrupt failure occurs in the model.
In some cases, particularly when calculating an initial equilibrium state, it may
be more computationally efficient to use DAMPING auto. As discussed in Sec-
tion 1.2.3.2 in Theory and Background, local damping is most efficient when
velocity components at gridpoints pass through zero periodically, because the
mass-adjustment process depends on velocity sign changes. Adaptive global
damping, on the other hand, applies a constant damping factor that is not
affected by velocity sign-changes. If velocities act predominantly in one di-
rection (e.g., due to gravity loading), then a system with local damping may
take longer to converge than one with adaptive global damping. When in doubt,
it is usually best to run the model with both DAMPING local and DAMPING auto,
and compare the calculation steps required to reach convergence.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 113

9. Avoiding Rounding Errors

Most of the calculations in 3DEC are in double precision (e.g., coordinate up-
dates and displacements). Usually, the significant figures that are available in
double-precision, 64-bit computer arithmetic are sufficient to keep rounding
errors from affecting the solution. However, there are circumstances in which
numerical rounding errors can affect the results of an analysis. For exam-
ple, rounding errors may become noticeable in problems that run for several
hundred thousand cycles with a low applied velocity.
To minimize rounding error problems, avoid large coordinates when creating
a model. For example, it is tempting to adopt the same coordinate values as
those used in mine plan and section drawings when creating a model for a
mining application. While this makes it easier to refer plots and results back
to the mine drawings, it may inadvertently magnify the rounding errors in the
computation.
3DEC always updates coordinates progressively as deformations develop. If
the initial coordinates are very large numerically (e.g., a coordinate range of
10,000 to 10,100), significant accuracy may be lost when small displacement
increments are added to the coordinates. Also, searches for block contacts,
which involve differences between coordinate values, may become unreliable.
By changing the coordinates to range from 0 to 100, the problem can be
avoided.
10. Determining Collapse Loads
In order to determine a collapse load, it often is better to use “strain-controlled”
boundary conditions instead of “stress-controlled” boundary conditions (i.e.,
apply a constant velocity and measure the boundary reaction forces, rather than
apply forces and measure displacements). A system that collapses becomes
difficult to control as the applied load approaches the collapse load. This is
true of a real system as well as a model system.
11. Determining Factor of Safety
3DEC does not calculate a “factor of safety” directly. If you need a safety
factor, it can be determined for any selected parameter by calculating the ratio
of the selected parameter value under given conditions, to the parameter value
that results in failure. For example:

applied load to cause failure

FL =
design load

tan (actual failure angle)

Fφ =
tan (friction angle at failure)

3DEC Version 5.2

3 - 114 User’s Guide

Note that the larger value is always divided by the smaller value (assuming
that the system does not fail under the actual conditions). The definition of
failure must be established by the user.

3DEC Version 5.2

 PROBLEM SOLVING WITH 3DEC 3 - 115

3.10 Interpretation

Because 3DEC models a nonlinear system as it evolves in time, the interpretation of results may
be more difficult than with a conventional finite element program that produces a “solution” at
the end of its calculation phase. There are several indicators that can be used to assess the state
of the numerical model for a static analysis* (e.g., whether the system is stable, unstable or is in
steady-state plastic flow). The various indicators are described below.

3.10.1 Unbalanced Force

Forces are accumulated at each centroid of rigid blocks, and at each gridpoint of deformable blocks.
At equilibrium (or steady plastic flow in deformable blocks), the algebraic sum of these forces is
almost zero (i.e., the forces acting on one side of the block centroid or gridpoint nearly balance
those acting on the other). During timestepping, the maximum unbalanced force is determined for
the whole model; this force is displayed continuously on the screen. It can also be saved as a history
and viewed as a graph. The unbalanced force is important in assessing the state of the model for
static analysis, but its magnitude must be compared with the magnitude of typical internal forces
acting in the model. In other words, it is necessary to know what constitutes a “small” force. A
representative internal gridpoint force for deformable blocks may be found by multiplying stress by
zone area perpendicular to the force, using values that are typical in the area of interest in the model.
Denoting R as the ratio of maximum unbalanced force to the representative internal force (expressed
as a percentage), the value of R will never decrease to zero. However, a value of 1% or 0.1% may be
acceptable as denoting equilibrium, depending on the degree of precision required (e.g., R = 1%
may be good enough for an intermediate stage in a sequence of operations, while R = 0.1% may
be used if a final stress or displacement distribution is required for inclusion in a report or paper).
Note that a low value of R only indicates that forces balance at all gridpoints. However, steady
plastic flow may be occurring, without acceleration. In order to distinguish between this condition
and “true” equilibrium, other indicators (such as those described below) should be consulted.

3.10.2 Block/Gridpoint Velocities

The velocities of rigid blocks and gridpoints of deformable blocks may be assessed by plotting the
whole field of velocities (using the PLOT vel command), or by selecting certain key points in the
model and tracking their velocities with histories (HISTORY xvel, HISTORY yvel and HISTORY zvel).
Both types of plots are useful. Steady-state conditions are indicated if the velocity histories show
horizontal traces in their final stages. If they have all converged to near-zero (in comparison to their
starting values), then absolute equilibrium has occurred. If a history has converged to a nonzero
value, then either the block is falling, or steady plastic flow is occurring at the block/gridpoint
corresponding to that history. If one or more velocity history plots show fluctuating velocities, then
the system is likely to be in a transient condition.

* Interpretation of the state of a model for a dynamic analysis is discussed in Section 2 in Optional
Features.

3DEC Version 5.2

3 - 116 User’s Guide

The plot of the field of velocity vectors is more difficult to interpret, since both the magnitudes and
the nature of the pattern are important. As with gridpoint forces, velocities never decrease precisely
to zero. The magnitude of velocity should be viewed in relation to the displacement that would
occur if a significant number of steps (e.g., 1000) were to be executed. For example, if current
displacements in the system are of the order of 1 cm, the maximum velocity in the velocity plot is
10−3 m/sec and the timestep is 10−5 sec, then 1000 steps would produce an additional displacement
of 10−5 m or 10−3 cm, which is 0.1% of the current displacements. In this case, it can be said
that the system is in equilibrium, even if the velocities all seem to be “flowing” in one direction.
More often, the vectors appear to be random (or almost random) in direction and (possibly) in
magnitude. This condition occurs when the changes in gridpoint force fall below the accuracy limit
of the computer, which is around six decimal digits. A random velocity field of low amplitude is
an infallible indicator of block stability and no plastic flow.
If the vectors in the velocity field are coherent (i.e., there is some systematic pattern) and their
magnitude is quite large (using the criterion described above), then either blocks are falling or
slipping, plastic flow is occurring within blocks, or the system is still adjusting elastically (e.g.,
damped elastic oscillation is taking place). To confirm that continuing plastic flow is occurring,
a plot of plasticity indicators should be consulted (see Section 3.10.3). If, however, the motion
involves elastic oscillation, then the magnitude should be observed in order to indicate whether such
movement is significant. Seemingly meaningful patterns of oscillation may be seen. However, if
amplitude is low, then the motion has no physical significance.

3.10.3 Plastic Indicators for Block Failure

For most of the nonlinear block models in 3DEC, the command PLOT block colorby state displays
those zones in which the stresses satisfy the yield criterion. Such an indication usually denotes
that plastic flow is occurring, but it is possible for a block zone simply to “sit” on the yield surface
without any significant flow taking place. It is important to look at the whole pattern of plasticity
indicators to see whether a mechanism has developed. A failure mechanism is indicated if there is
a contiguous line of active plastic zones that join two surfaces. The diagnosis is confirmed if the
velocity plot also indicates motion corresponding to the same mechanism. Note that initial plastic
flow often occurs at the beginning of a simulation, but subsequent stress redistribution unloads
the yielding elements so that their stresses no longer satisfy the yield criterion (“yielded in past”).
Only the actively yielding elements (“at yield surface”) are important to the detection of a failure
mechanism. If there is no contiguous line or band of active plastic zones between boundaries, two
patterns should be compared before and after the execution of, say, 500 steps. Is the region of
active yield increasing or decreasing? If it is decreasing, then the system is probably heading for
equilibrium; if it is increasing, then ultimate failure may be possible.
If a condition of continuing plastic flow has been diagnosed, one further question should be asked:
Does the active flow band(s) include zones adjacent to artificial boundaries? The term “artificial
boundary” refers to a boundary that does not correspond to a physical entity, but exists simply to
limit the size of the model that is used. If plastic flow occurs along such a boundary, then the
solution is not realistic because the mechanism of failure is influenced by a nonphysical entity. This
comment only applies to the final steady-state solution; intermediate stages may exhibit flow along
boundaries.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 117

3.10.4 Histories

In any problem, there are certain variables that are of particular interest (e.g., displacements may be
of concern in one problem, but stresses may be of concern in another). Liberal use should be made
of the HISTORY command to track these important variables in the regions of interest. After some
timestepping has taken place, the plots of these histories often provide the way to find out what the
system is doing.

3.10.5 Factor of Safety

A factor of safety can be determined in 3DEC for any selected parameter by calculating the ratio
of the selected parameter value under given conditions to the parameter value that results in failure.
For example:

applied load to cause failure

FL =
design load

tan (actual friction angle)

Fφ =
tan (friction angle at failure)

Note that the larger value is always divided by the smaller value (assuming that the system does
not fail under the actual conditions). The definition of failure must be established by the user. A
comparison of this approach based on strength reduction for determining factor of safety to that
based upon limit analysis solutions is given by Dawson and Roth (1999) and Dawson et al. (1999).
The strength reduction method for determining factor of safety is implemented in 3DEC through the
SOLVE fos command. This command implements an automatic search for factor of safety using the
bracketing approach, as described in Dawson et al. (1999). The SOLVE fos command may be given
at any stage in a 3DEC run, provided that all non-null zones contain Mohr Coulomb-based models.
During the solution process, properties will be changed, but the original state will be restored on
completion, or if the fos search is terminated prematurely with the <Esc> key.
When 3DEC is executing the SOLVE fos command, the bracketing values for F are printed con-
tinuously to the screen so that you can tell how the solution is progressing. If terminated by the
<Esc> key, the solution process terminates, and the best current estimate of fos is displayed. Save
files are produced for the last stable state and the last unstable state, so that velocity vectors, etc.
can be plotted. This allows a visual representation of the failure mode to be created. It is highly
recommended that you verify that the automatic bracketing process has correctly identified both
the stable and unstable models. Some modes of failure may occur without detection.
Example 3.26 is a simple model that illustrates how the command works. The run stops at F = 1.00.
Figure 3.51 shows shear strain as well as velocity, which allows the user to see that the failure
would be in the continuum instead of on the joint surfaces.

3DEC Version 5.2

3 - 118 User’s Guide

It is recommended that the SOLVE fos command be given at an equilibrium state of a model (to
improve solution time), but this is not essential.
The procedure used by 3DEC during execution of SOLVE fos is as follows. First the code finds
a “representative number of steps” (denoted by Nr ), which characterizes the response time of the
system. Nr is found by setting the cohesion to a large value, making a large change to the internal
stresses, and finding how many steps are necessary for the system to return to equilibrium. Then
for a given factor of safety, F , Nr steps are executed. If the unbalanced force ratio is less than 10−5 ,
then the system is in equilibrium. If the unbalanced force ratio is greater than 10−5 , then another
Nr steps are executed, exiting the loop if the force ratio is less than 10−5 . The mean value of force
ratio, averaged over the current span of Nr steps, is compared with the mean force ratio over the
previous Nr steps. If the difference is less than 10%, the system is deemed to be in nonequilibrium,
and the loop is exited with the new nonequilibrium F . If the above-mentioned difference is greater
than 10%, blocks of Nr steps are continued until: (1) the difference is less than 10%; or (2) 6 such
blocks have been executed; or (3) the force ratio is less than 10−5 . The justification for case (1) is
that the mean force ratio is converging to a steady value that is greater than that corresponding to
equilibrium; the system must be in continuous motion.
The default settings of the SOLVE fos command will adjust the strength parameters of friction and
cohesion for Mohr-Coulomb type failure in both the zones and the joints. If the constitutive model
does not use a Mohr-Coulomb type failure, this process will not apply (i.e., the CY joint model
will not produce a factor of safety using this command). The mhoekbrown model and ubiquitous
model have been modified to work with SOLVE fos; however, the older standard hoekbrown cannot
be used with SOLVE fos. Currently this technique may not be used with DLL zone or joint models.
There is an option to include tensile strength in the calculations for either the zones or the joints.
The user may also exclude certain material types from inclusion in the factor-of-safety calculations.
(Such an exclusion will not apply to properties that have been assigned using the ZONE or JOINT
command.)

Example 3.26 Factor-of-safety calculation for a simple joint model

new
poly brick 0 20 0 1 0 12
plot create plot Blocks
plot block
plot reset
jset dip 0 dd 180 or 0 0 2
hide range plane dip 0 dd 180 or 0 0 2 below
jset dip 45 dd 270 or 12 0 12
hide range plane dip 45 dd 270 or 12 0 12 below
del
seek
jset dip 60 dd 90 or 14 0 12 spac 3.6 num 10
jset dip 20 dd 270 or 20 0 12 spac 2.2 num 10
gen edge .7

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 119

change cons 2
prop mat 1 dens 2000.0
prop mat 1 bulk 1.0e8 shear 3.0e7 bcohesion 1e20
prop mat 1 bfriction 20.0 bdilation 20.0 btens 1e10
prop jmat 1 jkn 1e10 jks 1e9 jfric 30 jcoh 1e20
gravity 0,0 -10.0
boun zvel 0 range z 0
boun xvel 0 range z 0
boun xvel 0 range x 20
boun xvel 0 range x 0
boun yvel 0 range y 0
boun yvel 0 range y 1
damp auto
solve
plot clear
plot contour velocity
plot add velocity
plot reset
prop jmat 1 jcoh 1e4
prop mat 1 bcoh 12380.0
solve fos associated
ret

E

pGre00x

0 u0
i
0prSeG
pGrelG,lrG0p:G3:Gr0
 
l' 
(
ri3G.GVlG,
ri3GGGVlG,
ri-GGGVlG,
ri5GGGVlG,
rieGGGVlG,
riSGGGVlG,
ri.GGGVlG,
ri,GGGVlG,
ripGGGVlG,
rirGGGVlG,
riGGGGVlG,
3iGGGGVlG.
-iGGGGVlG.
5iGGGGVlG.
eiGGGGVlG.
SiGGGGVlG.
.iGGGGVlG.
,iGGGGVlG.
piGGGGVlG.
riGGGGVlG.
GiGGGGVnGG
 
l 

 :0GiGGr3G,3e
:0,Gei.e5
)
l!l!

$0M0riG,

Figure 3.51 Plot showing area of failure in slope

3DEC Version 5.2

3 - 120 User’s Guide

3.11 Modeling Methodology

3.11.1 Modeling of Data-Limited Systems

In a field such as geomechanics, where data are not always available, the methodology used in
numerical modeling should be different from that used in a field such as mechanical engineering.
Starfield and Cundall (1988) provide suggestions for an approach to modeling that is appropriate
for a data-limited system. This paper should be consulted before any serious modeling with 3DEC
is attempted. In essence, the approach recognizes that field data (such as in-situ stresses, material
properties and geological features) will never be known completely. It is futile to expect the model
to provide design data (such as expected displacements) when there is massive uncertainty in the
input data. However, a numerical model is still useful in providing a picture of the mechanisms
that may occur in particular physical systems. The model acts to educate the intuition of the design
engineer by providing a series of cause-and-effect examples. The models may be simple, with
assumed data that are consistent with known field data and engineering judgement. It is a waste of
effort to construct a very large and complicated model that may be just as difficult to understand as
the real case.
Of course, if extensive field data are available, then it may be incorporated into a comprehensive
model that can yield design information directly. More commonly, however, the data-limited
model does not produce such information directly, but provides insight into mechanisms that may
occur. The designer can then do simple calculations, based on these mechanisms, that estimate the
parameters of interest or the stability conditions.

3.11.2 Modeling of Chaotic Systems

In some calculations, especially in those involving discontinuous materials, the results can be
extremely sensitive to very small changes in initial conditions, or trivial changes in loading sequence.
At first glance, this situation may seem unsatisfactory, and may be taken as a reason to mistrust the
computer simulations. However, the sensitivity exists in the physical system being modeled. There
appear to be a least two sources for the seemingly erratic behavior:
1. There are certain geometric patterns of discontinuities that force the system to
choose (apparently at random) between two alternative outcomes; the subse-
quent evolution depends on which choice is made. For example, Figure 3.52
illustrates a small portion of a jointed rock mass. If block A is forced to move
down relative to B, it can either go to the left or to the right of B; the choice
will depend on microscopic irregularities in geometry, properties or kinetic
energy.

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 121

Figure 3.52 A small portion of a jointed rock mass

2. There are processes in the system that can be described as “softening” or, more
generally, as cases of positive feedback. In a fairly uniform stress field, small
perturbations are magnified in the subsequent evolution because when a region
has more strain, it softens more and thereby attracts more strain, and so on, in
a cycle of positive feedback.
Both phenomena give rise to behavior that is chaotic in its extreme form (Gleick 1987, and Thompson
and Stewart 1986). The study of chaotic systems reveals that the detailed evolution of such a system
is not predictable, even in principle. The observed sensitivity of the computer model to small changes
in initial conditions or numerical factors is simply a reflection of a similar sensitivity in the real world
to small irregularities. There is no point in pursuing ever more “accurate” calculations, because the
resulting model is unrepresentative of the real world, where conditions are not perfect. What should
our modeling strategy be in the face of a chaotic system? It appears that the best we can expect
from such a model is a finite spectrum of expected behavior; the statistics of a chaotic system are
well-defined. We need to construct models that contain distributions of initial irregularities (e.g.,
by using 3DEC ’s adev parameter on the JSET command). Each model should be run several times,
with different distributions of irregularities. Under these conditions, we may expect the fluctuations
in behavior to be triggered by the imposed irregularities, rather than by artifacts of the numerical
solution scheme. We can express the results in a statistical form.

3DEC Version 5.2

3 - 122 User’s Guide

3.11.3 Localization, Physical Instability and Path-Dependence

In many systems that can be modeled with 3DEC, there may be several paths that the solution
may take, depending on rather small changes in initial conditions. This phenomenon is termed
bifurcation. For example, a shear test on an elastic/plastic material may either deform uniformly,
or it may exhibit shear bands, in which the shear strain is localized rather than being uniformly
distributed. It appears that if a numerical model has enough degrees of freedom (i.e., enough
elements), then localization is to be expected. Indeed, theoretical work on the bifurcation process
(e.g., Rudnicki and Rice 1975, and Vardoulakis 1980) shows that shear bands form even if the
material does not strain-soften, provided that the dilation angle is lower than the friction angle. The
“simple” Mohr-Coulomb material should always exhibit localization if enough elements exist to
resolve one or more localized bands. A strain-softening material is more prone to produce bands.
Some computer programs appear incapable of reproducing band formation, although the phe-
nomenon is to be expected physically. However, 3DEC is able to allow bands to develop and
evolve, partly because it models the dynamic equations of motion (i.e., the kinetic energy that ac-
companies band formation is released and dissipated in a physically realistic way). Several papers
document the use of two-dimensional FLAC in modeling shear band formation (Cundall 1989, 1990
and 1991). These should be consulted for details concerning the solution process. One aspect that
is not treated well by 3DEC is the thickness of a shear band. In reality, the thickness of a band
is determined by internal features of the material, such as grain size. These features are not built
into 3DEC ’s constitutive models. Hence, the bands in 3DEC collapse down to the smallest width
that can be resolved by the grid: one grid-width if the band is parallel to the grid; or about three
grid-widths if the band cuts across the grid at an arbitrary angle. Although the overall physics of
band formation is modeled correctly by 3DEC, band thickness and band spacing are grid-dependent.
Furthermore, if the strain-softening model is used with a weakening material, the load/displacement
relation generated by 3DEC for a simulated test is strongly grid-dependent. This is because the
strain concentrated in a band depends on the width of the band (in length units), which depends on
zone size, as we have seen. Hence, smaller zones lead to more softening, since we move out more
rapidly on the strain axis of the given softening curve. To correct this grid dependence, some sort
of length scale must be built into the constitutive model. There is controversy at present concerning
the best way to do this. It is anticipated that future versions of 3DEC will include a length scale in
the constitutive models – probably involving the use of a Cosserat material, in which internal spins
and moments are taken into account. In the meantime, the processes of softening and localization
may be modeled, but it must be recognized that the grid size and angle affect the results; models
must be calibrated for each grid used.
One topic that involves chaos, physical instability and bifurcation is path-dependence. In most
nonlinear, inelastic systems, there are an infinite number of solutions that satisfy equilibrium,
compatibility and the constitutive relations. There is no “correct” solution to the physical problem
unless the path is specified. If the path is not specified, all possible solutions are correct. This
situation can cause endless debate among modelers and users, particularly if a seemingly irrelevant
parameter in the solution process (e.g., damping) is seen to affect the final result. All the solutions
are valid numerically. For example, a simulation of a mining excavation with low damping may
show a large overshoot and, hence, large final displacements, while high damping will eliminate
the overshoot and give lower final displacements. Which one is more realistic? It depends on the
path. If the excavation is done by explosion (i.e., suddenly), then the solution with overshoot may

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 123

be the appropriate one; if the excavation is done by pick and shovel (i.e., gradually), then the second
case may be more appropriate. For cases in which path-dependence is a factor, modeling should
be done in a way that mimics the way the system evolves physically.

3DEC Version 5.2

3 - 124 User’s Guide

3.12 References

Bandis, S. C., A. C. Lumsden and N. R. Barton. “Fundamentals of Rock Joint Deformation,” Int.
J. Rock Mech. Min. Sci. & Geomech. Abstr., 20(6), 249-268 (1983).
Barton, N. “The Shear Strength of Rock and Rock Joints,” Int. J. Rock Mech. Min. Sci. & Geotech.
Abstr., 13, 255-279 (1976).
Bieniawski, Z. T. “Determining Rock Mass Deformability: Experience from Case Histories,” Int.
J. Rock Mech. Min. Sci., 15, 237-247 (1978).
Board, M., et al. “Comparative Analysis of Toppling Behaviour at Chuquicamata Open-Pit Mine,
Chile,” Trans. Instn. Min. Metall., Sec. A, 105, A11-A21, 1996.
Brady, B. H. G., and E. T. Brown. Rock Mechanics for Underground Mining. London: George
Allen and Unwin. (1985).
Brady, B. H. G., J. V. Lemos and P. A. Cundall. “Stress Measurement Schemes for Jointed and
Fractured Rock,” in Rock Stress and Rock Stress Measurements, pp. 167-176. Luleå, Sweden:
Centek Publishers (1986).
Cundall, P. A. “Numerical Experiments on Localization in Frictional Materials,” Ingenieur-Archiv,
59, 148-159 (1989).
Cundall, P. A. “Numerical Modelling of Jointed and Faulted Rock,” in Mechanics of Jointed and
Faulted Rock, pp. 11-18. Rotterdam: A. A. Balkema (1990).
Cundall, P. A. “Shear Band Initiation and Evolution in Frictional Materials,” in Mechanics Com-
puting in 1990s and Beyond (Proceedings of the Conference, Columbus, Ohio, May 1991), Vol.
2: Structural and Material Mechanics, pp. 1279-1289. New York: ASME (1991).
Dawson, E. M., and W. H. Roth. “Slope Stability Analysis with FLAC,” in FLAC and Numerical
Modeling in Geomechanics (Proceedings of the International FLAC Symposium on Numerical
Modeling in Geomechanics, Minneapolis, Minnesota, September 1999), pp. 3-9. C. Detournay
and R. Hart, eds. Rotterdam: A. A. Balkema (1999).
Dawson, E. M., W. H. Roth and A. Drescher. “Slope Stability Analysis by Strength Reduction,”
Géotechnique, 49(6), 835-840 (1999).
Fossum, A. F. “Technical Note: Effective Elastic Properties for a Randomly Jointed Rock Mass,”
Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 22(6), 467-470 (1985).
Gerrard, C. M. “Elastic Models of Rock Masses Having One, Two and Three Sets of Joints,” Int.
J. Rock Mech. Min. Sci. & Geomech. Abstr., 19, 15-23 (1982b).
Gerrard, C. M. “Equivalent Elastic Moduli of a Rock Mass Consisting of Orthorhombic Layers,”
Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 19, 9-14 (1982a).
Gleick, J. Chaos: Making a New Science. New York: Penguin Books (1987).

3DEC Version 5.2

PROBLEM SOLVING WITH 3DEC 3 - 125

Goodman, R. E. Introduction to Rock Mechanics. New York: John Wiley and Sons (1980).
Hart, R. D. “An Introduction to Distinct Element Modeling for Rock Engineering,” in Compre-
hensive Rock Engineering, Vol. 2, pp. 245-261. Oxford: Pergamon Press, Ltd. (1993).
Hoek, E. “Estimating Mohr-Coulomb Friction and Cohesion Values from the Hoek-Brown Failure
Criterion,” Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 27(3), 227-229 (1990).
Hoek, E., and E. T. Brown. “The Hoek-Brown Failure Criterion – a 1988 Update,” in Rock
Engineering for Underground Excavations, pp. 31-38. Toronto: University of Toronto (1988).
Hoek, E., and E. T. Brown. Underground Excavations in Rock. London: Instn. Min. Metall.
(1980).
Jaeger, J. C., and N. G. W. Cook. Fundamentals of Rock Mechanics, 2nd Ed. London: Chapman
and Hall (1969).
Kulhawy, Fred H. “Stress Deformation Properties of Rock and Rock Discontinuities,” Engineering
Geology, 9, 327-350 (1975).
Lindner, E. N., and J. A. Halpern. “In-Situ Stress in North America: A Compilation,” Int. J. Rock
Mech. Min. Sci. & Geomech. Abstr., 15, 183-203 (1978).
Müller, B., et al. “Regional Patterns of Tectonic Stress In Europe,” J. Geophys. Res., 97(B8),
11783-11803 (1992).
Rosso, R. S. “A Comparison of Joint Stiffness Measurements in Direct Shear, Triaxial Compression,
and In-Situ,” Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 13, 167-172 (1976).
Rudnicki, J. W., and J. R. Rice. “Conditions for the Localization of the Deformation in Pressure-
Sensitive Dilatant Materials,” J. Mech. Phys. Solids, 23, 371-394 (1975).
Serafim, J. L., and J. P. Pereira. “Considerations of the Geomechanical Classification of Bieniawski,”
in Proceedings of the International Symposium on Engineering Geology and Underground Con-
struction Lisbon 1983, Vol. 1, pp. II.33-42. Lisbon: SPGILNEC (1983).
Singh, B. “Continuum Characterization of Jointed Rock Masses: Part I – The Constitutive Equa-
tions,” Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 10, 311-335 (1973).
Starfield, A. M., and P. A. Cundall. “Towards a Methodology for Rock Mechanics Modelling,” Int.
J. Rock Mech. Min. Sci. & Geomech. Abstr., 25(3), 99-106 (1988).
Thompson, J. M. T., and H. B. Stewart. Nonlinear Dynamics and Chaos. New York: John Wiley
and Sons (1986).
Vardoulakis, I. “Shear Band Inclination and Shear Modulus of Sand in Biaxial Tests,” Int. J. Numer.
Anal. Meth. in Geomechanics, 4, 103-119 (1980).
Vermeer, P. A., and R. de Borst. “Non-Associated Plasticity for Soils, Concrete and Rock,” Heron,
29(3), 3-64 (1984).

3DEC Version 5.2

3 - 126 User’s Guide

3DEC Version 5.2