You are on page 1of 7

1

Non-Existence of Solution for Horizontally Rigid Half-space

A. J. Anyaegbunam A.M.ASCE1; N. N. Osadebe2; and O. J. Eze-Uzomaka3


1
Lecturer, Dept. of Civil Engrg., Univ. of Nigeria, Nsukka, A.M. ASCE, email: ajacivil@gmail.com
2
Prof. of Struct. Engrg, Dept. of Civil Engrg., Univ. of Nigeria, Nsukka
3
Prof. of Civil Engrg, Dept. of Civil Engrg., Univ. of Nigeria, Nsukka

Abstract: The Westergaard expressions for stresses and displacements in a halfspace of Westergaard material are
shown to be invalid because the vertical shear stress does not vanish at the plane boundary of the halfspace. The same
inadequacy is discovered in the solution for a horizontally rigid cross-anisotropic halfspace deduced from Michell's
expressions. A surprising conclusion is arrived at, namely: there is currently no exact elastic solution for the stresses and
displacements in a horizontally rigid cross-anisotropic halfspace. The Westergaard theory should cease to be regarded as
an exact elastic solution for a problem of theory of elasticity.

Author Keywords: Horizontally rigid, exact elastic solution, isotropic halfspace, cross-anisotropic halfspace,
stresses boundary condition.

Introduction
Boussinesq (1885) published his well known formulae for stresses and displacements due to a vertical point load on the
surface of a linearly elastic, isotropic (Eh = Ev), homogeneous semi-infinite continuum (halfspace). Because of the size of
the earth, it could be regarded as nearly semi-infinite and Boussinesq’s solution was immediately applied to problems in
geotechnical engineering in the 1920’s and 30’s. In that period, Boussinesq’s solution was integrated to obtain the
solutions for circular and rectangular loaded areas. A series of test cases revealed that the Boussinesq formula over
predicted immediate elastic settlements. Michell (1900), in view of the recognized anisotropy of many materials,
developed limited expressions for stresses and displacements in a homogeneous cross-anisotropic halfspace due to a
surface point load. Unfortunately, for the civil engineering profession, Michell’s seminal paper was hidden away in a
mathematics journal where it remained hidden until it was extracted and presented to the engineering world by Barden
(1963). Barden’s aim was to introduce an improved procedure for settlement computations that can cater for over-
consolidated clays that are known to be cross-anisotropic.
Westergaard (1938) developed the expressions for stresses and displacements due to a vertical point load on an
isotropic and linearly elastic halfspace in which the horizontal normal strains εx and εy and horizontal shear strain γxy are
zero, thus making the halfspace indirectly horizontally rigid. The Westergaard material is an isotropic material with finite
horizontal and vertical normal moduli and poison’s ratios (Eh = Ev = E and νhv = νvh = ν) but with infinite horizontal
shear modulus (Ghh = ∞). The Westergaard medium is hypothetical and does not exist in nature for it assumes isotropy
with respect to normal moduli E but anisotropy with respect to shear moduli. Westergaard invented his
stress/displacement theory for varved clays, which are clays with numerous horizontal seams of silt and fine sand, and it
has been accepted ever since its inception in 1938 till date with its validity remaining unquestioned (Barden 1963; Harr
1966; Harrison and Gerrard 1972; Hooper 1975; Gazetas 1981; Gazetas 1982; Bowles 1997; Wang and Cheung 2001;
Arora 2008; Tekinsoy et al. 2009). The major reason why Westergaard Solution had been accepted in soil mechanics and
foundation engineering was because it gave lower values for immediate elastic settlements than those calculated from the
Boussinesq formula, which was known to overpredicted settlements.
Barden (1963), by using Michell’s results, developed the expression for the vertical stress only in a horizontally rigid
cross-anisotropic halfspace and concluded that it resembled Westergaard’s. More will be said on this later.
The purpose of this note is triple fold, namely to show that
1. The Westergaard solution is invalid by not satisfying the shear stress boundary condition at the ground surface;
2. The solution for a point load on a horizontally rigid (extreme type) cross-anisotropic halfspace is also in error for the
same reason.
3. The problem of stresses and displacements in an extreme type of cross-anisotropic, elastic, horizontally rigid (Eh = ∞)
halfspace has no exact elastic solution.
The last task is done in a non-rigorous fashion within the known classes of solutions for a cross-anisotropic halfspace.

WESTERGAARD’S SOLUTION

The Westergaard procedure will be briefly reproduced here because his original article is inaccessible. Westergaard
assumed that the horizontal displacements and strains are zero, i.e. u = v = εx = εy = γxy = 0, so that from Hooke’s law
νσ z
σx =σy = (1)
(1 − ν )

ASCE, Journal of Geotechnical and Geoenvironmental Engineering, 2011, 137(4), April, 431- 434
http://ascelibrary.org/gto/resource/1/jggefk/v137/i4 , doi: 10.1061/(ASCE)GT.1943-5606.0000444.
2

(1 − ν ) Eε z
σ= (2)
(1 + ν )(1 − 2ν )
τ zx = Gγ zx , τ zy = γ zy (3)

z R

r σz

Fig.1. Point Load on Surface of a half-space

Using the strain – displacement relations


ε z = ∂w / ∂z , γ zx = ∂w / ∂x , γ zy = ∂w / ∂y (4)
the vertical equilibrium equation can be expressed as

∂2w ∂2w ∂2w


+ + =0 (5)
∂x 2 ∂y 2 ∂Z 2
where Ζ = ηz and
1/ 2
⎛ 1 − 2ν ⎞
η =⎜ ⎟
⎝ 2 − 2ν ⎠

Westergaard obtained the solution of Eq.(5) for a point load on the half-space (Fig. 1) to be

P 1
w = η (1 + ν ) (6a)
πE (r + η 2 z 2 )1 / 2
2

So that at the surface of the half-space z = 0

P
w( r ,0) = η (1 + ν ) (6b)
πEr
and also

Pηz
σz = (7)
2π ( r + η 2 z 2 ) 3 / 2
2

Pη x
τ zx = (8)
2π (r + η 2 z 2 ) 3 / 2
2

Pηy
τ zy = (9)
2π (r + η 2 z 2 ) 3 / 2
2

By combining τzx and τzy it is obtained that

ASCE, Journal of Geotechnical and Geoenvironmental Engineering, 2011, 137(4), April, 431- 434
http://ascelibrary.org/gto/resource/1/jggefk/v137/i4 , doi: 10.1061/(ASCE)GT.1943-5606.0000444.
3

Pη r
τ zr = (10)
2π ( r + η 2 z 2 ) 3 / 2
2

where τrz is the vertical/horizontal shear stress in the r-z plane.


At the surface of the halfspace z = 0, Eq. (10) gives


τ rz = τ rz (r ,0) = (11)
2πr 2
Hence, the shear stress τzr does not vanish at the stress-free surface of the halfspace at z = 0, r > 0.
The problem solved by Westergaard is shown in Fig. 2

P
τrz (r,0)

τ = τrz (r,0)

z R τ τ

r τ

σz

(a) Loading (b) Stresses on a Surface Element

Fig. 2 The Problem solved by Westergaard

Consequently, the Westergaard solution is invalid for it does not satisfy one of the necessary stress boundary conditions
of the original problem which is that the surface z = 0 should be free of shear stresses. Thus, the Westergaard formulas
are not the exact elastic solution for a point Load on the surface of a halfspace of Wetergaard medium. The error in the
Westergaard analysis has remained unknown until today because of the emphasis laid on vertical stress σz and vertical
displacement w in geotechnical engineering such that they are the only ones whose formulae are reproduced in
publications.
It is noteworthy that Harrison and Gerrard (1972), Hooper (1975) and Gazetas (1982), have misinterpreted the
Westregaard medium to be an extreme type of cross-anisotropic material with infinite horizontal normal modulus (Eh = ∞
or n = ∞) and νvh = 0. The Westergaard medium has finite normal moduli and Poisson’s ratios in both the vertical and
horizontal directions, and its horizontal rigidity derives from the artificial imposition of zero normal horizontal strains.
The extreme type of cross-anisotropic medium has infinite horizontal normal modulus.

Michell’s Solution
Barden’s solution to the extreme-condition problem of a horizontally rigid cross-anitropic halfspace was deduced from
Michell’s (1900) solution and hence the latter need to be considered. Michell (1900) first derived the expressions for
vertical normal and shear stresses everywhere and surface vertical displacements in an elastic cross-anisotropic halfspace
due to a surface point load.
Michell’s solution for the case when the characteristic equation has real and distinct roots is:
Pz ⎛ 1 1 ⎞
σ z (r , z ) = ⎜⎜ 3 − 3 ⎟⎟ (12a)
2π (k 2 − k1 ) ⎝ R1 R2 ⎠
Pr ⎛ 1 1 ⎞
τ zr (r , z ) = ⎜⎜ 3 − 3 ⎟⎟ (12b)
2π (k 2 − k1 ) ⎝ R1 R2 ⎠
P
w(r ,0) = ξ o
r

ASCE, Journal of Geotechnical and Geoenvironmental Engineering, 2011, 137(4), April, 431- 434
http://ascelibrary.org/gto/resource/1/jggefk/v137/i4 , doi: 10.1061/(ASCE)GT.1943-5606.0000444.
4

Where k1 and k2 are the real roots of the characteristic equation

( AC − F 2 − 2 FL) 2 C
k4 − k + =0 (13a)
AL A
1/ 2
⎡ AC − F 2 − 2 FL − D3 ⎤
k1 = ⎢ ⎥ (13b)
⎢⎣ 2 AL ⎥⎦
1/ 2
⎡ AC − F 2 − 2 FL + D3 ⎤
k2 = ⎢ ⎥ (13c)
⎣⎢ 2 AL ⎦⎥

D1 = [ AC − L]2 − [F + L]2 , D2 = [ AC + L]2 − [F + L]2 (14a)

D3 = D1D2 = [AC − F2][ AC − (F + 2L)2 ] (14b)

R1 = [r2 + (z/k1)2 ]1/2 , R2 = [r2 + (z/k2)2 ]1/2 (15)


1/ 2
1 ⎛ AD2 ⎞
ξo = ⎜ ⎟ (16)
2π ( AC − F 2 ) ⎝ L ⎠

Where A, C, F, L and N are elastic constants defined in the Appendix .

Solution for Horizontally Rigid Halfspace


The nature of the solution for a cross-anisotropic halfspace (Gerrard and Harrison 1970, Gerrard and Wardle 1973)
depends on whether
AC > F+2L.
• Class 1: The characteristic equation has two real distinct roots, i.e, if D1 > 0 or
• Class 2 The characteristic equation has two equal, i.e, if D1 = 0 or AC = F+2L.
• Class 3: The characteristic equation has two complex roots, i.e, if D1 < 0 or AC < F+2L.
An additional constraint placed on the elastic constants so that the strain energy is non-negative is that (Hearmon 1961;
Gerard 1972)
1−ν1−2nν22 > 0 (17a)
(1 − ν 1 )
or n< (17b)
2ν 22
If n is allowed to tend to infinity (the extreme type of cross-anisotropic material) then by virtue of Eq. (17) ν2 will
tend to zero, in agreement with Harrison and Gerrard (1972), Hooper (1975), Gazetas (1982) and the elastic constants
specified in the Appendix will assume the values
nE v
A= , C = Ev (18a)
(1 − ν 12 )
ν E nEv
F= 3 v , N= (18b)
(1 − ν 1 ) 2(1 + ν 1 )
from which it is observed that A and N dominate the other elastic constants as n tends to infinity.
Thus, the extreme cross-anisotropic medium will fall under class 1, and will have two real and distinct roots, so that
Michell’s solution will apply to it. From Eq. (13) and Eq. (14) it could be deduced that
k1 = 0 (19a)

1/2
k2 = C/L = Ev/Gvh = 1/m2 where m = [Ghv /Ev] (19b)
Finally, stresses and vertical surface displacement are given by

ASCE, Journal of Geotechnical and Geoenvironmental Engineering, 2011, 137(4), April, 431- 434
http://ascelibrary.org/gto/resource/1/jggefk/v137/i4 , doi: 10.1061/(ASCE)GT.1943-5606.0000444.
5

Pmz
σ z (r , z ) = (20)
2π (r + m 2 z 2 ) 3 / 2
2

Pmx
τ rz (r , z ) = (21)
2π (r + m 2 z 2 ) 3 / 2
2

P
w(r ,0) = (22)
2πE v mr
1/2
where m = limit as n → ∞ of [Ghv /Ev] .
If Barden’s defnition of Ghv = L is used, then m = 1 and the formula given for vertical stress on p.203 of Barden (1963)
is reproduced.
From Eq. (21)
Pm
τ rz (r ,0) = (23)
2πr 2
Hence, again, the shear stress τzr does not vanish at the stress-free surface of the halfspace at z = 0, r > 0 as required by
the stress boundary condition. Thus, the solution for the extreme type of cross-anisotropic material is invalid.
As had been shown, there are three classes of cross-anisotropic media with the extreme type cross-anisotropic
medium belonging to the class 1 medium for which no solution could be found for the problem of a surface loaded
halfspace. It is futile to seek the solution of this problem for class 2 and 3 media by virtue of their exclusion from the
possible solution space. Hence, the problem of stresses and displacements in an extreme type of cross-anisotropic, elastic,
horizontally rigid (Eh = ∞) halfspace has no exact elastic solution.
This lack of a solution for the horizontally rigid cross-anisotropic medium is an inexplicable mystery for the cross-
anisotropic medium with finite horizontal modulus has a solution that satisfies the necessary stress-boundary condition at
the surface. Consequently, the horizontally rigid cross-anisotropic halfspace can be regarded as a degenerate halfspace
because what applies to its class is inapplicable to it.
Examination of Eqs. (6b), (7), (20) and (22) reveals that the vertical stress and surface displacement in the
Westergaard medium do not agree exactly with those in the extreme type cross-anisotropic medium, except m = η in the
case of vertical stress, and m = 1/[2η(1+ν)] in the case of surface displacement. Interestingly, Madhav and Subha Rao
(1964) had queried Barden on this lack of exact agreement between the two solutions. This paper has answered that
question by explaining that the Westergaard medium and the extreme type cross-anisotropic medium are not the same
though they resemble one another in having zero normal horizontal strains. The Westergaard medium has finite normal
moduli and Poisson’s ratios in both the vertical and horizontal directions, and its horizontal rigidity derives from the
artificial imposition of zero normal horizontal strains. The extreme type of cross-anisotropic medium has infinite
horizontal normal modulus.

Conclusion
The Westergaard medium has been distinguished from the horizontally rigid (extreme type) cross-anisotropic half-space.
Investigation of the Westergaard solution revealed it to be invalid, that is, it is not what it is supposed to be: the exact
analytical elastic solution for a Point load on a ‘Westergaard’ halfspace because the shear stress τzr does not vanish at the
stress-free surface of the halfspace. A similar defect is observed in the solution for the horizontally rigid cross-anisotropic
halfspace. The latter is a degenerate halfspace that has no solution when it is loaded at its plane boundary. The non-
validity of the existing solutions has remained undiscovered till today because of the emphasis laid on vertical normal
stress and vertical displacement in foundation engineering. It is recommended that the use of the Westergaard formulas
be discontinued for they are neither analytical nor empirical.

Appendix: Stress Relations For Cross-Anisotropic Material

σ = Dε (24)

where σ = [σx σy σz τxy τxz τyz ] T

ε = [εx εy εz γxy γxz γyz ]T

where the matrix D is given in Michell’s symbolism as

ASCE, Journal of Geotechnical and Geoenvironmental Engineering, 2011, 137(4), April, 431- 434
http://ascelibrary.org/gto/resource/1/jggefk/v137/i4 , doi: 10.1061/(ASCE)GT.1943-5606.0000444.
6

A A-2N F 0 0 0
A-2N A F 0 0 0
F F C 0 0 0
D = 0 0 0 N 0 0 (25)
0 0 0 0 L 0
0 0 0 0 0 L

A = f ( 1 − nν22), A− 2N = f (ν1 + nν22), F = f ν2(1 + ν1), C = f (1 − ν12)/n

N = Ghh = nEV/2(1 + ν1), f = nEV/ [(1 + ν1) (1− ν1 − 2nν22)]

n = Eh/Ev = modulus ratio, ν1 = νhh = Poisson’s ratio in the horizontal plane

ν2 = νvh = Poisson’s ratio measuring the effect of vertical strain on horizontal strain

ν3 = νhv = nν2 = Poisson’s ratio measuring the effect of horizontal strain on vertical strain
Numerous attempts have been made to express Ghv in terms of other elastic constants.
According to Barden (1963)
nEv
L = Ghv = (26)
(1 + n + 2nν 2 )

This has, however, been stated to be wrong by subsequent authors. Dooley (1964) had criticized Eq. (26) as being
incorrect for assuming that the shear modulus for a pair of axes inclined at 45o to the x-z axes equals that for the x-z axes.
Gazetas (1981) based on Carrier (1946) has surmised that a good approximation for Ghv is
( AC − F 2 )
Ghv = (27)
( A + C + 2F )

Harrison and Gerrard (1972) had arrived at the following result for a horizontally rigid cross-anisotropic material (n
=∞)
(1 − ν 1 − ν 3 )
m= (28)
2(1 − ν 3 )

With the conditions of positive strain energy demanding that 0 ≤ m ≤ 0.75.


Current thinking is that Ghv is an independent elastic constant that needs be determined experimentally.

References

Arora, K. R. (2008). Soil mechanics and foundation engineering, Standard Publishers Distributors, Delhi, India, p.243
Barden, M. (1963). “Stresses and displacements in a cross-anisotropic soil.” Géotechnique, 13 (3), 198-210.
Boussinesq, J. (1885). Application des potentials a l’etude de l’equilibre et du mouvement des solides elastiques,
Gauthier-Villars, Paris (in French).
Bowles, J. E. (1997). Foundation analysis and design, McGraw-Hill, New York, 301-302.
Carrier, G. F. (1946). “Propagation of waves in orthotropic media.” Q. Appl. Math., 4, 160-165.
Dooley, J.C. (1964). “Discussion of L. Barden's stress and displacements in a cross-anisotropic soil.”
Géotechnique, 14(3), 278-279.
Gazetas, G. (1981). “Strip foundations on a cross-anisotropic soil layer subjected to dynamic loading.”Géotechnique,
31(2), 161-179.
Gazetas, G. (1982). “ Stresses and displacements in cross-anisotropic soils.”J. Geotech. Eng. Div., 108(4), 532-553.
Gerrard C. M. (1972). “Discussion of L. Barden's stress and displacements in a cross-anisotropic soil, Géotechnique,
22(2), 372-376.
Gerrard, C. M. and Harrison, W. J. (1970). “Circular loads applied to a cross-anisotropic half-space.” Tech. Paper No. 8,
Appl. Geomechanics Div., Commonwealth Scientific and Industrial Research Org., Canberra, Australia.
Gerrard, C. M. and Wardle, L. J. (1973).” Solutions for point loads and generalized circular loads applied to a cross

ASCE, Journal of Geotechnical and Geoenvironmental Engineering, 2011, 137(4), April, 431- 434
7

anisotropic half-space.” Tech. Paper No. 13, Appl. Geomechanics Div., Commonwealth Scientific and Industrial
Research Org., Sydney, Australia.
Harr, M.E. (1966). Foundations of theoretical soil mechanics, McGraw-Hill, New York, 104-108.
Harrison, W. J. and Gerrard, C. M. (1972). “Elastic theory applied to reinforced earth.” J.Soil Mech. Found. Div,
98(12), 1325-1345.
Hearmon, R. F. S. (1961). An introduction to applied anisotropic elasticity, Oxford Univ. Press, Oxford.
Hooper J.A. (1975). “Elastic settlement of a circular raft in adhesive contact with a transversely isotropic medium.”
Géotechnique, 25(4), 691-711.
Madhav, M. R., and Subha Rao, K. S. (1964). “Correspondence on L. Barden’s paper.” Géotechnique, 14(3), p. 59.
Michell, J. H. (1900). “The stress distribution in an aelotropic solid with infinite boundary plane.” Proc. London Math.
Soc., 32, 247-258
Tekinsoy, M. A., Ta Kiran, T., kayadelen, C. and Baran, T. (2009). “An approximation to the stress distribution analysis
for anisotropic clayey soil.” Sci. Res. Essays, 4(2), 78-87.
Wang, Y. H., and Cheung, Y. K. (2001). “Plate on cross-anisotropic foundation analyzed by the finite element method,
Comput. Getech., 28(1), 37 - 54
Westergaard, H.M. (1938). “A problem of elasticity suggested by a problem in soil mechanics: Soft material reinforced
by numerous strong horizontal sheets.” Contributions to the Mechanics of Solids, Stephen Timoshenko 60th birthday
anniversary volume, Macmillan, New York.

ASCE, Journal of Geotechnical and Geoenvironmental Engineering, 2011, 137(4), April, 431- 434

You might also like