INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING, VOL.

21, 585--600 (1985)

AN ISOPARAMETRIC JOINT/INTERFACE ELEMENT

FOR FINITE ELEMENT ANALYSIS

G . BEER
Ikpurlmcni of Cioil Engineering, Uniucmify of Queenslund, SILuciu. A U S T R A L I A

SUMMARY
A generally applicable and simple joint/interface element for three- and two-dimensional finite element
analysis is presented. The proposed element can model joints/interfaces between solid finite elements and shell
finite elements. The derivation of the joint element stiffness is presented and algorithms for the treatment of
nonlinear joint behaviour discussed. The performance of the element is tested on typical problems involving
shell-to-shell and shell-to-solid interfaces.

lNTRODUCTION
In the finitc clcment analysis in solid mechanics, situations arise where discontinuous behaviour
occurs between finite elements. Examples are interfaces between dissimilar materials and joints or
fractures in the material. Applications include soil-structure interaction problems and problems
involving a rock mass which is intersected by joints and planes of weakness.
Various methods have been proposed in the past. The discontinuous bchaviour at the interface
has been modelled using constraint equationsLg2or by connecting the elements with each other by
discrete spring^.^.^ Other worker^^.^ have treated the joint/interface as a quasi-continuum of small
thickness, i.e. by using continuum finite elements which contain planes of weakncss.
Special joint finite elements have been developed by G ~ o d m a nGhaboussi
,~ et ~ 1 Wilson'
, ~ and
others.lOT1' Most of the applications are in two dimensions (plane stress, plane strain), although
some three-dimensional analyses have been reported.",' '
The purpose of this paper is to present a generally applicable joint element which is based on
similar assumptions as in References 8 and 10, but differs from it in the following:

1 . An isoparametric formulation is used.

2. The formulation is quite general and the joint element can be used with three-dimensional and
two-dimensional shell and solid finite elements.
3. The joint element has zero thickness and this has obvious advantages when modelling rock
joints and fractures.

In the present paper, attention is given to joints which are initially in contact. However, the joint
element can be applied to more general contact problems with a suitable modification of the
analysis procedure.
In the following, a stiffness matrix is derived for a generic isoparametric joint element. Then the
solution algorithms uscd for the analysis of problems involving nonlinear joint bchaviour arc

0029-5981/85/04 0585-16$01.60 Received Murch 84

((:) I985 by John Wiley & Sons, Ltd.
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
G. BEER
586
discussed. Finally, test examples are presented which show the performance of the element in three-
dimensional situations.

THEORY
The theory is developed for the most complex case first, i.e. for the shell-to-shell contact in three
dimensions. This is then taken as a generic element with the formulation for other applications
being mutations or simplifications of this case. Figure 1 shows the interface configurations which
can be analysed.

Three-dimensional analysis

1. Shell-shell contact element. Figure 2 shows two shell elements in contact. In this particular
case, a thick shell elementi2is used. The displacements at any point on the bottom surface of the top
shell element are given by

where N , are isoparametric shape functions and v l i and vZi are unit vectors defined in Reference 13
and Figure 2.

20

f
30
Figure 1. Possible jointbinterface configurations in two- and three-dimensional stress analysis
10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
587
ISOPARAMETRIC JOINTiINTERFACE ELEMENT

Similarly, we can write for the top surface of the bottom element:
Figure 2. Joint between two shell elements

= “,,Nz””nll.OP
Equation (1) can be written in matrix form as

NTOP

where I is a 3 x 3 unit matrix.

where

and

or
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
G . BEER
588
W'

Figure 3. Definition of vectors in three-dimensional joint

A local co-ordinate system is established at a point on the contact (Figure 3) by first finding the
vector normal to the contact surface. This is obtained as a cross-product of two vectors:

v3=

The derivatives in (8) are coefficients of the Jacobian matrix of the co-ordinate transformation:

where {.xi,yi,z i ) pop are the midsurface node co-ordinates of the top shell element.
The unit vector in the direction normal to the contact is

n = -v3
A
where A is the length of V, and represents the unit mapped area of the contact plane.
The two tangent vectors are formed" by

and
s2 = s1 x n

When the direction of n is exactly in the x direction then s1 is obtained by

s, =[]x n
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ISOPARAMETRIC JOINT/INTERFACE ELEMENT 589

The displacements in the directions normal and tangential to the contact surface are obtained by

with

The relative displacements at the interface are

-Slip in directions 1 and 2
--Convergence/separation

where

and

ae = i "%OT
OP1

The following relationships between tractions acting on the contact and relative displacements
are defined:

where tsl, t,, are shear tractions, t, the contact pressure and k,71,ks2, k, are shear and normal
stiffnesses. Equation (20) can be written in matrix form:
t=D6 (21)
where D is the elasticity matrix, i.e.

The stiffness matrix is obtained by the standard procedure of minimizing the total potential
energy.13 The work done by the tractions at the interface is

(
= iaeT BTDBd A)ae
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
590 G . BEER

The work done by the forces at the nodes of the contact element is
W = aeTFe
Following the usual finite element procedure^,'^ the following equilibrium equation is obtained:
Fe = K”ae (26)
where the stiffness matrix for the contact element is

The integration is carried out numerically within the mapped unit square using 2 x 2 Gauss
quadrature.

2. Shell-solid contact Assuming that the solid element is at the bottom then
= [N I, N,I. .........N”I1
NBOT (28)
where n is the number of nodes on the solid element surface in contact. The displacements of the
bottom surface are

The derivation of the stiffness matrix follows the same procedure as before.

3. Solid-solid contact If there is a solid element both on top and bottom then in addition to (28) we
have
NTOP = [N,IN,I.. ........N J I (30)
The coordinates of the contact are now computed by

Two-dimensional analyses
For two-dimensional analyses (i.e. plane strain/plane stress) the expressions derived previously
can be used, except that certain degrees-of-freedom do not appear.
For the shell element nodes the displacement vector is reduced to

(32)

and for the solid element nodes to

(33)

Thus for two-dimensional analysis the coefficients corresponding to the deleted degrees-of-
freedom do not appear.
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ISOPARAMETRIC JOINT/INTERFACE ELEMENT 59 1

Figure 4. Definition of vectors in two-dimensional joint

The local co-ordinate system for a two-dimensional analysis is shown in Figure 4. The vector
normal to the contact is found by

the local directions are

n = -v3 (35)
A
and

s ={ !}x n

Therefore
8 = [s, n] (37)
and
(38)

For an elastic analysis the stiffness matrices of all finite elements and joint elements are
assembled in the usual way to give
Ka=F (39)
For a joint in contact, k, and kn are theoretically infinite. For the analysis procedure to work, large
finite values have to be used. These values could represent the stiffness of any joint infill.
Alternatively, the values for k, and k,scan be chosen in such a way that the elastic slip and closing is
negligible compared with the displacements of the elements adjacent to the joint. It has been
shown' that when double precision is used for the variables in the solution phase, the stiffness
values can be quite high before any numerical ill-conditioning is experienced.

ANALYSIS PROCEDURE FOR NONLINEAR JOINTS

The nonlinear behaviour of joints is characterized by slip and separation taking place at the joint
plane.
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
592 G . BEER

For a joint with no tensile strength, separation of joint planes will occur when the traction
normal to the joint plane becomes tensile ( = positive). The yield function for this case is
F, = t, (40)
Alternatively, the joint can be given a tensile strength (i.e. F , = t, - T ) .
If the shear strength of the joint is exceeded, irreversible slip occurs. The Mohr-Coulomb yield
function is
+
F , = t, K (41)
where
t s = J(t,: + t . 3 (42)
and
u = O if F n 2 0 (43)
K = t;tan+ - c F , > 0

The joint properties are: the angle of friction 4, cohesion c and, if required, tensile strength T.
A time-dependent plasticity (viscoplastic) model is used here.I4 It is assumed that at the end of
each time step the total relative joint displacements consist of an elastic part and-if the point has
yielded-a visco-plastic part:
6 = a='+ p (44)
The visco-plastic displacement components are computed from

For a simple linear time dependence:

4(F)=O F<O
&(F)= F F >0
In equations (45), ys and yn are material parameters which indicate the time dependence of the
visco-plastic shear and normal deformation. The magnitude of the time step AT cannot be chosen
freely and has to satisfy certain stability criteria in order to obtain a stable solution. It has been
shown14 that for certain choices of AT the visco-plastic algorithm becomes an 'initial stress'
procedure. In the present model this time step is

The visco-plastic algorithm can be extended to allow for joint dilatancy and peak and residual
strength which are important parameters in rock mechanics.
The nonlinear analysis proceeds in time steps. At time zero the loading (applied forces,
excavation forces, etc.)is applied. If visco-plastic strains have developed, then a residual load vector

s
AFe = BTD8"'d V,

is computed for each yielded joint element and applied at the next time step.
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ISOPARAMETRIC JOINTANTERFACE ELEMENT 593

In many cases it is advisable to apply the loading in increments to ensure that the right loading
path is followed. Convergence rate can be accelerated by updating the joint stiffness matrix at
certain intervals.
The procedure for the nonlinear analysis is summarized in Appendix I, in a FORTRAN 77 style
diagram.

TEST EXAMPLES
Three test examples are presented to verify the joint element and demonstrate its applicability.

Example 1. Simply-supported plate with horizontal joint

This example was chosen to verify that the element can model inelastic slip. The mesh in Figure 5
represents one-quarter of a simply-supported plate with a horizontal joint and consists of 18 thick
shell elements and 9 joint elements. The plate is subjected to a uniformly distributed load of unity.
Various analyses were performed with different angles of friction. Figure 6 shows contours of slip
for d, = 0", C = 0 (largest slip is contour No. 12, smallest = 2). Figure 7 shows the displaced shape.
The central deflection obtained for 4 = 0" and C = 0 was 0.39032 and this compares well with the
theoretical value of 0.39180. Two time steps with one update of the joint stiffness matrix were
required for full convergence.

Exumple 2. Plate resting on injinite elusfic half-space

In this example, both slip and separation occurs. The mesh in Figure 8 again represents only
one-quarter of the problem with symmetry conditions applied. The mesh consists of 9 shell
elements, 9 joint elements, 50 parabolic 20-node solid elements and 45 infinite domain 16-node
elements which model the infinite space surrounding the mesh.
The plate was subjected to a concentrated force at the middle, and two analyses were made. In

t'
E = 1000
v = 0
q = l
-- A X

a = &

3d=0,2

Figure 5. Example 1: mesh and assumptions

 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
594 G . BEER

Figure 6. Example 1: contours of slip (4= O", c' = 0 )

Figure 7. Example 1: displaccd shape (4= 0". C = 0)

the first analysis it was assumed that no slip can occur on that part of the interface which has not
separated (no slip interface) and in the second, that the interface between the plate and the ground
cannot resist any shear (smooth interface). The results of the analysis are plotted in Figure 9.
It can be seen that in both cases the corner of the plate lifts off the ground, but this is-as
expected--more pronounced if slip is allowed to occur at the interface.
10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
/
595

€:igure 9. Examplc 2: contact prcssurc along A-A for elastic and elasto plastic joint (compression ncgative)
I-
JOINT
ISOPARAMETRIC JOINT/INTERFACE ELEMENT

Figure 8. Example 2 mesh and assumptions

v = o
AF
I

tP. 103
L
t-
t’

iI

,
i
i
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
596 G . BEER

SCALE 1 I 00.
DISPL. SCALE I 1 mm 0 10.00 mm

Figure 10. Example 2: displaccd shapc for no-slip intcrhce

For the no-slip case, the results are similar to the ones obtained by Cheung and Nag’s using a
considerably simpler approach of modelling the interface. The values cannot be compared directly
here, because in Reference 13 contact pressures are only given along the edge of the plate, whereas
Figure 9 shows the pressures slightly inside the edge along the Gaussian integration joints. The
displaced shape for the no-slip case is shown in Figure 10. For both analyses, convergence was
achieved after three time steps with updated stiffness matrix.

Exainple 3. Sequentid excavution ? f a square cavity

The third example is the analysis of the sequential excavation of a square cavity in an infinite
prestressed medium. The mesh for one-quarter of the problem with conditions of symmetry applied
is shown in Figure 11. It consists of 72 parabolic solid elements to model the rock mass, 36 shell
elements to model the rock plate, 72 joint elements and 72 ‘infinite domain’ elements. The initial
stress is assumed to be 35 MPa in vertical direction and the cavity is excavated in three stages. At
each stage the elements corresponding to excavation sequence number in Figure 11 are removed
from the mesh and their nodal forces applied in the opposite direction.
Figure 12 shows plots of displaced shapes after each excavation stage (the displacements are
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ISOPARAMETRIC JOINT/INTERFACE ELEMENT 597

‘t infinite domain
elements

Im

E -80000 MPa
v = 0,25

+JO -0
7m Y
I. e: = -35 MPa
Y4

I X

Figure 11. Examplc 3: mcsh and assumptions

cxaggerated). The plots show that slip and separation is taking place on the joint planes. Figure 13
shows contours of slip which has taken place at excavation stage 2, with the contour No. 9 being
the largest amount of slip and 1 the smallest.

SUMMARY AND DISCUSSION

A joint/interface element has been presented which performs well in three-dimensional stress
analysis involving nonlinear slip and separation. The formulation of the element is quite general
and mutations of the generic element can be used for shell-to-shell, shell-to-solid and solid-to-solid
interfaces in either two or three dimensions. The element is more sophisticated than joint elements
developed previously, and because of the assumption that it has zero thickness is particularly
suited for modelling rock joints and fractures. One of the main aims in developing this element has
becn to apply it to the modelling of excavations in rock, where the rock mass is traversed by
bedding planes and fractures. Currently, back-analyses of the behaviour of shale rock exposures
where both opening and slip occur is being undertaken and the findings will be reported in a future
publication.
The joint element presented here, however, is not restricted to problems in geomechanics. One
10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ISOPARAMETRIC JOINT/INTERFACE ELEMENT 599

Figure 13. Examplc 3: contours of rnagnitudc of slip (excavation stagc 2)

can envisage applications in mechanical engineering, concrete technology and many more. Also,
the formulation could be extended to include problems in dynamics such as earthquake
engineering.

ACKNOWLEDGEMENT

The work reported here has been carried out as part of a research project entitled ‘Development of
efficient 3-D stress analysis techniques’ which has been sponsored by the Mining Research Section
of Mount Isa Mines Ltd. Australia.

APPENDIX I: NUMERICAL PROCEDURE FOR NONLINEAR ANALYSIS

FOR LOAD INCREMENT i
a’ = Displucemrnt at time O
to = Tractions at time 0
AF’ = Applied loading
 10970207, 1985, 4, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/nme.1620210402 by Dalian University Of, Wiley Online Library on [28/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
600 G . BEER

K O= Stiffjless mutrix at time 0

DO n= 1, N o of Time Steps
SOLVE K ” - l A a ” = A F ” -
a” = an- + Aa”
DO e = I , N o of Joint Elements
DO 9 = I , No of’ Gauss Points
(Att’)e,g= DB(Aa”),
(t”)e,g = (tn- ‘ L , g + (Atn)e,g
COMPUTE 6”
(t“),,, = (t”),,, - DSVP
IF (UPDATE STIFFNESS) THEN
IF (F, > 0) k, N 0, k, N 0
IF (F, > 0) k, N 0
COMPUTE CONTRIBUTION TO STIFFNESS MATRIX (27)
END IF
COMPUTE CONTRIBUTION TO LOAD VECTOR (48)
END DO
END DO
ASSEMBLE AF,l
IF (UPDATE STIFFNESS) ASSEMBLE K“
ELSE K” N K ” -
END DO

REFERENCES
1. A. Francavilla and 0 . C. Zienkiewicz, ‘A note on numerical computation of elastic contact problcms’, lnt. .j. numo.
methods eng., 9, 9 13-924 ( I 975).
2. T. D. Sachdeva and C. V. Ramakrishnan, ‘A finite element solution for the two-dimensional elastic contact problem’,
I n t . .j. numer. methods eny., 17, 1259-1271 (1981).
3. L. R. Hcrmnnn, ‘Finitc clcment analysis of contact problems’, Proc. A.S.C.E., 104(EM5) (I 978).
4. R. Frank et ul., ‘Numerical analysis of contacts in geomcchanics’, Proc. 4th Int Conf o n Numerical Methods in
Gcomcchanics. A. A. Balkcma, Rortcrdam (I982), pp. 37.-42.
5. 0. C. Zienkiewicz et nl.. ‘Analysis o f non linear problems in rock mechanics with particular reference to jointed rock
systems, 2nd ISRM Conf., Belgrade, pp. 8- 14 (1970).
6. G. N. Pande and K. G. Sharma, ‘On joint/interFace elements and associated problems of numerical ill-conditioning’,
lnt. .j. numer. methods geomech, 2, 293-300 (1979).
7. R. E. Goodman et al., ‘A model for the mechanics of jointed rocks’, Journal A.S.C.h’., 94(SM3) (1968).
8. 7‘. Gbaboussi et al., ‘Finite elements for rock joints and interfaces’, Proc. A.S.C.E., 99(SMlO) (1974).
9. E. L. Wilson, ‘Finite elements for foundations, joints and fluids’, ch. 10 in Finite Elements in Gromechanics (Ed. G .
Gudchus), Wilcy, 1977.
10. D. E. Van Delen and R. D. Ewig, ‘BMINES-- a finite element code for rock mechanics applications’, Proc. 22nd U.S.
Symp. on Rock Mechanics (Ed. H. H. Einstein).
1 I . C . S. Dcsai, Constilutiuc, Equtrtions,fbr Soil Media in Numericul Me1hod.s in Geomechunics, (Ed. J. B. Martins), Riedel,
Holland, 1981.
12. S. Ahniad, B. M. lronsand 0. C. Zicnkiewicz, ‘Analysis ofthick and thin shcll structurcs by curved finitcclements’, I n l .
j. numer. melhods eny., 2, 4 19- 45 I ( I 970).
13. ‘3.C . Zicnkiewicz, The Finite Element Method, 3rd edn, McGraw-Hill, London, 1977.
14. 0. C. Zicnkicwicz and 1. C. Cormeau, ‘Visco-plasticity, plasticity and creep in elastic solids~--~a unified numerical
solution approach’, In/. j. numer. methods eny., 8, 821-845 (1974).
15. Y. K . Cheung and D. K. Nag, ‘Plates and beams on elastic foundations--linear and non-Linear behaviour, Geotech.,
18, 250-260 ( I 968).
16. G. Beer, ‘Infinitc domain elements’, Int.,j. numer. methods eng., 17, 43-52 (1981).