l

ria
CHAPTER 8

PLATES ON ELASTIC FOUNDATION

ate
57. Bending Symmetrical with Respect to a Center. A laterally
loaded plate may rest on an elastic foundation, as in the case of a con-
crete road, an airport runway, or a mat. We begin the discussion of such
problems with the simplest assumption that the intensity of the reaction
of the subgrade is proportional to the deflections w of the plate. This
dM
intensity is then given by the expression kw. The constant /b, expressed
in pounds per square inch per inch of deflection, is called the modulus of
the foundation. The numerical value of the modulus depends largely
on the properties of the subgrade; in the case of a pavement slab or a
mat of greater extension this value may be estimated by means of the
diagram in Table 62. *
TABLE 62. VALUES OF THE MODULUS OF SUBGRADE
hte

Modulus "k" in Ib/sq in./in.

General soil rating as subgrade, subbase or base

Very poor subgrade Poor Excellent Good Best
subgrade Fair to good subgrade subgrade Good subbase base base

G-Gravel P-Poorly groded

rig

S-Sand L - Low to med. compressibility

M-Mo','very fine sand,silt H-High compressibility
C-Cloy
F-Fines, material
less than 0.1mm
0-Organic
W-WeII graded
py

Let us begin with the case of a circular plate in which the load is dis-
Co

tributed symmetrically with respect to the center. In using Eq. (58),

1
Based on Casagrande's soil classification. The table should not be regarded as a
substitute for plate bearing tests. For further information see Trans. Am. Soc. Civ.
Engrs., vol. 113, p. 901, 1948. See also K. Terzaghi, Geotechnique, vol. 5, p. 297, 1955
(Harvard Soil Mechanics Series, no. 51).

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 we add the load —kw, due to the reaction of the subgrade, to the given
lateral load q. Thus we arrive at the following differential equation for
the bent plate:

l
ria
(178)

In the particular case of a plate loaded at the center with a load P,*
q is equal to zero over the entire surface of the plate except at the center.
By introducing the notation

ate
7_ 1

Eq. (178) becomes

Since Tc is measured in pounds per cubic inch and D in pound-inches, the

dM
quantity I has the dimension of length. To simplify our further dis-
cussion it is advantageous to introduce dimensionless quantities by using
the following notations:

Then Eq. (b) becomes

hte

Using the symbol A for

we then write
rig

This is a linear differential equation of the fourth order, the general solu-
tion of which can be represented in the following form:
z = A1X1(X) + A2X2(X) + A3X3(X) + AiXt(X) (/)
py

where Ai, . . . , A4 are constants of integration and the functions

Xi, . . . , X4 are four independent solutions of Eq. (e).
We shall now try to find a solution of Eq. (e) in the form of a power
* This problem was discussed by H. Hertz, Wiedemann's Ann. Phys. u. Chem., vol.
22, p. 449, 1884; see also his " Gesammelte Werke," vol. 1, p. 288, 1895, and A. Foppl,
Co

" Vorlesungen liber technische Mechanik," vol. 5, p. 103, 1922. It is worth noting that
Hertz's investigation deals with the problem of a floating plate rather than with that
of a plate on an elastic foundation. Thus, in this case the assumption regarding the
constancy of k is fulfilled, k being the unit weight of the liquid.

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 series. Let anxn be a term of this series. Then, by differentiation, we
find

l
ria
To satisfy Eq. (e) it is necessary that each term anxn in the series have a
corresponding term an-^xn~i such that

Following this condition, all terms cancel when the series is substituted in

ate
Eq. (e); hence the series, if it is a convergent one, represents a particular
solution of the equation. From Eq. (g) it follows that

Observing also that

dM
we can conclude that there are two series satisfying Eq. (e), viz.,
hte

It may be seen from the notations (c) that for small values of the dis-
tance r, that is, for points that are close to the point of application of the
load P, the quantity x is small, and series 0') are rapidly convergent.
rig

It may be seen also that the consecutive derivatives of series (j) remain
finite at the point of application of the load (x = 0). This indicates that
these series alone are not sufficient to represent the stress conditions at
the point of application of the load where, as we know from previously
discussed cases, the bending moments become infinitely large.
py

For this reason the particular solution X 3 of Eq. (e) will be taken in
the following form:
X 3 = X 1 log x + F3(z) (k)
in which F3(x) is a function of x which can again be represented by a
Co

power series. By differentiation we find

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 and substituting X 3 for z in Eq. (e), we obtain

l
ria
Since Xi satisfies Eq. (e) and is represented by the first of the series (J)1
we obtain the following equation for determining Fz(x):

ate
Taking F 3 (x) in the form of the series

and substituting this series in Eq. (I), we determine the coefficients 64,
dM
&8, & 12, . . . so that the resulting equation will be satisfied. Observing
that
AA(&4z4) = 42 • 22 • 64
we find, by equating to zero the sum of the terms that do not contain X1
that
hte

Equating to zero the sum of the terms containing x4, we find

rig

In general, we find

Thus the third particular solution of Eq. (e) is

py

The fourth particular integral X4 of Eq. (e) is obtained in a similar

manner by taking
Co

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 By substituting the particular solutions (j), (n), and (o) in expression
(/) we obtain the general solution of Eq. (e) in the following form:

l
ria
ate
It remains now to determine in each particular case the constants of inte-
gration Ai, . . . , A4 so as to satisfy the boundary conditions.
Let us consider the case in which the edge of a circular plate of radius a
dM
is entirely free. Making use of expression (52) for the radial moments
and expression (55) for the radial shear force Qn we write the boundary
conditions as
hte

In addition to these two conditions we have two more conditions that

hold at the center of the plate; viz., the deflection at the center of the
plate must be finite, and the sum of the shearing forces distributed over
the lateral surface of an infinitesimal circular cylinder cut out of the plate
at its center must balance the concentrated force P. From the first of
rig

these two conditions it follows that the constant A3 in the general solu-
tion (p) vanishes. The second condition gives

or, by using notation (a),

py

where e is the radius of the infinitesimal cylinder. Substituting Iz for w

Co

in this equation and using for z expression (p), we find that for an infinitely
small value of x equal to e/l the equation reduces to

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 from which

l
Having the values of the constants A3 anji A4, the remaining two con-

ria
stants Ai and A2 can be found from Eqs. (q). For given dimensions of
the plate and given moduli of the plate and of the foundation these equa-
tions furnish two linear equations in Ax and A2.
Let us take, as an example, a plate of radius a = 5 in. and of such rigidity that

ate
We apply at the center a load P such that

dM
Using this value of A4 and substituting Iz for w, we find, by using expression (p) and
taking x = a/I = 1, that Eqs. (q) give
0.50(Mi -f 0.250A2 = 4.062A4 = 4.062 • 102 • 10~6
0.687A1 - 8.483A2 = 11.09A4 = 11.09 • 102 • 10~5
These equations give
Ai = 86 • 10~4 A2 = - 6 4 • 10~6
hte

Substituting these values in expression (p) and retaining only the terms that contain
x to a power not larger than the fourth, we obtain the following expression for the
deflection:

The deflection at the center (x = 0) is then

rig

and the deflection at the boundary (x = 1) is

py

The difference of these deflections is comparatively small, and the pressure distribution
over the foundation differs only slightly from a uniform distribution.
If we take the radius of the plate two times larger (a = 10 in.) and retain the
previous values for the rigidities D and k, x becomes equal to 2 at the boundary, and
Eqs. {q) reduce to
0.826A1 + 1.980A2 = 1.208A4
Co

2.665A1 - 5.745A2 = 16.37A4

These equations give

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 The deflection is obtained from expression (p) as

l
ria
The deflections at the center and at the boundary of the plate are, respectively,
wm&x = 2.10~2 in. and wm\n = 0.88 • 10~2 in.
It is thus seen that, if the radius of the plate is twice as large as the quantity /, the
distribution of pressure over the foundation is already far from uniform. The applica-

ate
tion of the strain energy method to the problem of bending of a plate on elastic sub-
grade will be shown in Art. 80.
58. Application of Bessel Functions to the Problem of the Circular Plate. The
general solution (/) of Eq. (e) in the preceding article can also be represented in terms
of Bessel functions. To this end we introduce into Eq. (e) a new variable £ = x V i ;
thus we arrive at the equation
A'A'z - 2 = 0 (a)
dM
in which the symbol A' stands for

Now Eq. (a) is equivalent to equation

and also to
hte

Hence Eq. (a) is satisfied by the solutions of the Bessel differential equation

as well as by the solutions of the equation

rig

which is transformable into Eq. (d) by substituting %i for £. Thus the combined solu-
tion of Eqs. (d) and (e) can be written as
z = BxI0(X Vi) + BJ0(Xi Vi) + B*K0(x Vi) + BAK0(xi Vi) (/)
py

I0 and K0 being Bessel functions of the first and second kind, respectively, and of
imaginary argument, whereas Bi, Bi, . . . are arbitrary constants. The argument
x being real, all functions contained in Eq. (/) appear in a complex form. To single
out the real part of the solution, it is convenient to introduce four other functions, first
used by Lord Kelvin and defined by the relations1
Co

1
See, for instance, G. N. Watson, "Theory of Bessel Functions," p. 81, Cambridge,
1948.

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 Setting, furthermore,

l
where the new constants Ci, C2, . . . are real, we obtain the following expression for

ria
the deflections of the plate:
w = Ci ber x + C2 bei x -f C3 kei x + C4 ker x (h)
All functions herein contained are tabulated functions,1 real for real values of the
argument.
For small values of the argument we have

ate
ber x = 1 - z4/64 -f . . .
bei x = x*/4 - z«/2,304 -f . . .
(l)
ker x = - log x + log 2 - T + **V16 + • • •
kei x - -(z 2 /4) log Z - TT/4 + (1 + log 2 - 7 )s74 + • • •
in which y = 0.5772157 - i s Euler's constant and log 2 - y = 0.11593 • • •
For large values of the argument the following asymptotic expressions hold:
dM
hte

in which <r « x/\/2.

The general solution (h) can be used for the analysis of any symmetrical bending of
a circular plate, with or without a hole, resting on an elastic foundation. The four
constants C, corresponding in the most general case to four boundary conditions, must
be determined in each particular case.2
rig

1
See "Tables of Bessel Functions /0(2) and Ji(z) for Complex Arguments,"
Columbia University Press, New York, 1943, and "Tables of Bessel Functions
Yo(z) and Yx{z) for Complex Arguments," Columbia University Press, New York,
1950. We have
py

* Many particular solutions of this problem are given by F. Schleicher in his

Co

book " Kreisplatten auf elastischer Unterlage," Berlin, 1926, which also contains
tables of functions Z1 (x) = ber x, Z2(x) = —bei x} Zz(x) = — (2/ir) kei x, and
ZA(X) •» — (2/TT) ker x as well as the first derivatives of those functions. An abbrevi-
ated table of the functions Z and their first derivatives is given in Art. 118, where
they are denoted by the symbol ^.

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 We shall confine ourselves to the case of an infinitely extended plate carrying a
single load P at the point x = 0. Now, from the four functions forming solution Qi)1
the first two functions increase indefinitely with increasing argument in accordance
with Eqs. (J); and the function ker x becomes infinitely large at the origin, as we can

l
conclude from Eqs. (J). Accordingly, setting C1 = C2 • CA = 0, solution (h) is

ria
reduced to
w = Cz kei x (k)

In order to determine the constant Cz, we calculate, by means of Eqs. (i), the shearing
force [see Eqs. (193)]

ate
As x decreases, the value of Qr tends to CzD/I3X — CzD/l2r. On the other hand, upon
distributing the load P uniformly over the circumference with radius r, we have
Qr = —P/2irr. Equating both expressions obtained for Qr, we have
dM
Substitution of Cz into Eq. (k) yields, finally, the complete solution of Hertz's problem
in the form

(179)
hte

wu
and the corresponding reaction of the subgrade is given by p = kw = •—• The
variation of these quantities along a meridional section through the deflection
surface of the plate is shown in Fig. 131, together with similar curves based on a theory
which will be discussed in Art. 61.
At the origin we have kei x = —T/4 and the deflection under the load becomes
rig

(180)

For the reaction of the subgrade at the same point we obtain

(181)
py

If we take an infinitely large plate with the conditions of rigidity and loading assumed
on page 264, the deflection under the load becomes
Co

as compared with the value of 0.02 in. obtained for a finite circular plate with the
radius a = 21.
The distribution of the bending moments due to the concentrated load is shown in

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 l
(Elost. solid)

ria
(Hertz)

ate
dM (Elosisolid)
hte
rig
py

(c)
FIG. 131
Fig. 131c. It is seen that the radial moments become negative at some distance from
the load, their numerically largest value being about —0.02P. The positive moments
Co

are infinitely large at the origin, but at a small1 distance from the point of application
of the load they can be easily calculated by taking the function kei x in the form (i).
Upon applying formulas (52) and (53) to expression (179), we arrive at the results
1
As compared with the characteristic length I = s/D/k.

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 (182)

l
ria
A comparison of the foregoing expressions with Eqs. (90) and (91) shows that the
stress condition in a plate in the vicinity of the load in Hertz's case is identical with
that of a simply supported circular plate with a radius a = 2le~y = 1.123J, except for
p
a moment M' = M\ = — — (1 — v), which is superimposed on the moments of the
OTT

ate
circular plate.
Let us consider now the case in which the load P is distributed over the area of a
circle with a radius c, small in comparison with I. The bending moments at the center
of a circular plate carrying such a load are

dM
This results from Eq. (83), if we neglect there the term c2/a2 against unity. By
substituting a = 2le~f into Eq. (m) and adding the moment —P/8TT(1 — v), we obtain
at the center of the loaded circle of the infinitely large plate the moments

(183)
hte

Stresses resulting from Eq. (183) must be corrected by means of the thick-plate theory
in the case of a highly concentrated load. Such a corrected stress formula is given on
page 275.
In the case of a load uniformly distributed over the area of a small rectangle, we may
proceed as described in Art. 37. The equivalent of a square area, in particular, is a
circle with the radius c = 0.57w, u being the length of the side of the square (see page
162). Substituting this into Eq. (183) we obtain
rig

The effect of any group of concentrated loads on the deflections of the infinitely
large plate can be calculated by summing up the deflections produced by each load
py

separately.

59. Rectangular and Continuous Plates on Elastic Foundation. An

example of a plate resting on elastic subgrade and supported at the same
time along a rectangular boundary is shown in Fig. 132, which represents
Co

a beam of a rectangular tubular cross section pressed into an elastic

foundation by the loads P. The bottom plate of the beam, loaded by
the elastic reactions of the foundation, is supported by the vertical sides
of the tube and by the transverse diaphragms indicated in the figure by

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 dashed lines. It is assumed again that the intensity of the reaction p at
any point of the bottom plate is proportional to the deflection w at that
point, so that p = kw, k being the modulus of the foundation.

l
In accordance with this assumption, the differential equation for the

ria
deflection, written in rectangular coordinates, becomes

where q, as before, is the intensity of the lateral

load.

ate
Let us begin with the case shown in Fig. 132.
If W0 denotes the deflection of the edges of the
bottom plate, and w the deflection of this plate
with respect to the plane of its boundary, the
intensity of the reaction of the foundation at
any point is k(wQ — w), and Eq. (a) becomes
dM
Taking the coordinate axes as shown in the
FIG. 132 figure and assuming that the edges of the plate
parallel to the y axis are simply supported and the other two edges are
clamped, the boundary conditions are
hte

The deflection w can be taken in the form of a series:

rig

The first series on the right-hand side is a particular solution of Eq. (6)
py

representing the deflection of a simply supported strip resting on an

elastic foundation. The second series is the solution of the homogene-
ous equation
Co

Hence the functions Ym have to satisfy the ordinary differential equation

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 Using notations

l
ria
and taking the solution of Eq. (g) in the form erv, we obtain for r the
following four roots:

The corresponding four independent particular solutions of Eq. (g) are

ate
e0mV C O g ymy e-PmV C 0 S ymy ePmV S[n ymy g-0mV g ^ ymy (j)

which can be taken also in the following form:

cosh pmy cos ymy sinh pmy cos ymy „.
dM cosh pmy sin ymy sinh pmy sin ymy
From symmetry it can be concluded that Yn in our case is an even func-
tion of y. Hence, by using integrals (k), we obtain
Yn = Am cosh $my cos ymy + Bn sinh j3TOt/ sin ymy
and the deflection of the plate is
hte

This expression satisfies the boundary conditions (c). To satisfy con-

ditions (d) we must choose the constants Am and Bm so as to satisfy the
equations
rig
py
Co

Substituting these values of Am and Bm in expression (Z), we obtain the

required deflection of the plate.
The problem of the plate with all four edges simply supported can be
solved by using Eq. (a). Taking the coordinate axes as shown in Fig. 59

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 (page 105) and using the Navier solution, the deflection of the plate is

l
ria
In similar manner let the series

ate
represent the distribution of the given load, and the series

represent the reaction of the subgrade. Substituting the series (n) in

dM
the left-hand side and the series (o) and (p) in the right-hand side of
Eq. (a), we obtain

As an example, let us consider the bending of the plate by a force P

concentrated at some point (£,7?). In such a case
hte

by Eq. (6) on page 111. By substitution of expressions (q) and (r) into
Eq. (n) we finally obtain
rig

Having the deflection of the plate produced by a concentrated force,

the deflection produced by any kind of lateral loading is obtained by the
py

method of superposition. Take, as an example, the case of a uniformly

distributed load of the intensity q. Substituting q d£ drj for P in expres-
sion (s) and integrating between the limits 0 and a and between 0 and 6,
we obtain
Co

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 When k is equal to zero, this deflection reduces to that given in Navier
solution (131) for the deflection of a uniformly loaded plate.1
Let us consider now the case represented in Fig. 133. A large plate

l
which rests on an elastic foundation is loaded at equidistant points along

ria
the x axis by forces P.* We shall take the coordinate axes as shown in

ate
FIG. 133
the figure and use Eq. (/), since there is no distributed lateral load. Let
us consider a solution of this equation in the form of the series
dM
in which the first term

represents the deflection of an infinitely long strip of unit width parallel

hte

to the y axis loaded at y = 0 by a load P/a [see Eq. (283), page 471].
The other terms of the series satisfy the requirement of symmetry that
the tangent to the deflection surface in the x direction shall have a zero
slope at the loaded points and at the points midway between the loads.
We take for functions Ym those of the particular integrals (j) which
vanish for infinite values of y. Hence,
rig

To satisfy the symmetry condition (dw/dy)y==o = 0 we must take in this

expression
py

1
The case of a rectangular plate with prescribed deflections and moments on two
opposite edges and various boundary conditions on two others was discussed by H. J.
Fletcher and C. J. Thorne, J. Appl. Mechanics, vol. 19, p. 361, 1952. Many graphs
are given in that paper.
Co

* This problem has been discussed by H. M. Westergaard; see Ingenifiren, vol. 32,
p. 513, 1923. Practical applications of the solution of this problem in concrete road
design are discussed by H. M. Westergaard in the journal Public Roads, vol. 7, p. 25,
1926; vol. 10, p. 65, 1929; and vol. 14, p. 185, 1933.

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 Hence, by introducing the new constants A'm = Am/ym, we represent the
deflections (u) in the following form:

l
ria
In order to express the constants A'm in terms of the magnitude of loads
P, we consider the shearing force Qy acting along the normal section of
the plate through the x axis. From symmetry we conclude that this
force vanishes at all points except the points of application of the loads

ate
P, at which points the shearing forces must give resultants equal to
— P/2. It was shown in the discussion of a similar distribution of
shearing forces in Art. 54 (see page 248) that the shear forces can be
represented by the series
dM
The shearing force, as calculated from expression (v), is
hte

Comparing these two expressions for the shearing force, we find

or, by using notations (i),

rig

Substituting this in expression (v), we finally obtain

py

The maximum deflection is evidently under the loads P and is obtained

by substituting x = a/2, y = 0 in expression (w), which gives
Co

(184)

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 The deflection in the particular case of one isolated load P acting on
an infinitely large plate can also be obtained by setting a = <*> in formula
(184). In such a case the first term in the formula vanishes, and by using

l
notations (i) we obtain

ria
ate
Using the substitution

we find
(185)
dM
in accordance with the result (180). With this magnitude of the deflec-
tion, the maximum pressure on the elastic foundation is

(186)

The maximum tensile stress is at the bottom of the plate under the point
of application of the load. The theory developed above gives an infinite
hte

value for the bending moment at this point, and recourse should be had
to the theory of thick plates (see Art. 26). In the above-mentioned
investigation by Westergaard the following formula for calculating maxi-
mum tensile stress at the bottom of the plate is established by using the
thick-plate theory:
rig

Here h denotes the thickness of the plate, and

py

where c is the radius of the circular area over which the load P is assumed
to be uniformly distributed. For c = 0 the case of the concentrated
force is obtained.
In the case of a square loaded area u by u, we have to replace c by
Co

0.57u (see page 162).

The case of equidistant loads P applied along the edge of a semi-infinite
plate, as shown in Fig. 134, can also be treated in a similar way. The
final formula for the maximum tensile stress at the bottom of the plate

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 under the load when the distance a is large is

l
where b is calculated as in the previous case, and c is the radius of the

ria
semicircular area over which the load P is assumed to be uniformly
distributed. Formulas (x) and (y) have proved very useful in the design
of concrete roads, in which case the
circle of radius c represents the area
of contact of the wheel tire with the

ate
road surface.1
60. Plate Carrying Rows of Equidistant
Columns. As a last example, let us con-
sider an infinite plate or mat resting on
elastic subgrade and carrying equidistant
and equal loads P, each load being distrib-
dM
uted uniformly over the area w by v of a
rectangle, as shown in Fig. 135. The

FIG. 134 FIG. 135

bending of such a "reversed flat slab" may be treated by means of the previously dis-
hte

cussed Westergaard's solution, using simple series.2 Much simpler, however, and,
except for the case of a highly concentrated load, also adequate is the solution in
double series, making use of Navier's method.
Conditions of symmetry compel us to represent the lateral load due to the columns
in form of a cosine series:
rig

The intensity of the given load is equal to P/uv within the shadowed rectangles in
Fig. 135 and is zero elsewhere. Thus, proceeding in the usual manner, i.e., multiplying
Eq. (a) by cos —— cos —— dx dy and integrating between the limits — a/2, -{-a/2
py

a b
1
The problem of stress distribution near the load applied at a corner of a large plate
has not yet been solved with the same reliability as the problems discussed above.
Several empiric and semiempiric stress formulas regarding that case may be found in
"Concrete Pavement Design," p. 79, Portland Cement Association, Chicago, 1951.
Co

Noteworthy experimental results concerning this problem were obtained by M. Dantu,

Ann. ponts et chausstes, vol. 122, p. 337, 1952. See also L. D. Black, Trans. Eng.
Inst. Canada, vol. 2, p. 129, 1958, and D. E. Nevel, ibid., p. 132.
2
See W. Muller, Ingr.-Arch., vol. 20, p. 278, 1952, and Osterr. Ingr.-Arch., vol. 6,
p. 404, 1952.

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 for x) —6/2, + 6 / 2 for y, we have

l
where emn = 1 for m ^ 0, n ^ 0

ria
e,nn = £ for w = 0, n ^ 0 or m ^ 0, n = 0
€mn = -J- for w = n = 0
In the particular case of m — 0 or n = 0 the coefficient itself is readily obtained as a
limit value of the expression (6).
Now, in accordance with Eq. (a) we take for deflections the series

ate
and the relation between the coefficients amn and Amn is easily established by the same
reasoning as before (see page 272). Thus, using the notation
dM
we obtain

Substituting this in the series (c) and observing Eq. (6) we have the final result 1
hte

The bending moments of the plate are now obtained by the usual differentiation, and
the distribution of the pressure between the plate and the subgrade is found by multi-
plication of expression (/) by the modulus k.
The particular case k = 0 corresponds to a uniformly distributed reaction, of the
subgrade, i.e., to the case of a "reversed fiat slab" uniformly loaded with q — P/ab.
rig

It is seen from Eq. (/) that the introduction of the modulus tends to reduce the deflec-
tions and also the bending moments of the plate.
The case of a rectangular plate of finite dimensions resting on an elastic foundation
and submitted to the action of a concentrated load has been discussed by H. Happel. 2
The Ritz method (see page 344) has been used to determine the deflections of this
plate, and it was shown in the particular example of a centrally loaded square plate
py

that the series representing the deflection converges rapidly and that the deflection
can be calculated with sufficient accuracy by taking only the first few terms of the
series.3
1
Due to V. Lewe, Bauingenieur, vol. 3, p. 453, 1923.
2
Math. Z., vol. 6, p. 203, 1920. See also F. Halbritter, Bautechnik, vol. 26, p. 181,
Co

1949.
3
The problem of a square plate on an elastic foundation has also been investigated
experimentally; see the paper by J. Vint and W. N. Elgood, Phil. Mag., ser. 7, vol. 19,
p. 1, 1935; and that by G. Murphy, Iowa State Coll. Eng. Expt. Sta. Bull. 135, 1937.

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 61. Bending of Plates Resting on a Semi-infinite Elastic Solid. So far, the settling
of the subgrade at some point of its surface has been assumed as proportional to the
pressure between the plate and the subgrade at the same point, and consequently as
independent of the pressure elsewhere. This is correct in the case of a floating plate,

l
considered by Hertz (see page 260), but in the case of a coherent subgrade such a

ria
hypothesis approximates but crudely the actual behavior of the subgrade; a better
approximation can sometimes be obtained on the basis of the following assumptions:
1. The foundation has the properties of a semi-infinite elastic body.
2. The plate rests on the subgrade without friction.
3. A perfect contact between the plate and foundation also exists in the case of a
negative mutual pressure.

ate
This last supposition appears arbitrary; however, a negative pressure between plate
and subgrade actually is compensated, more or less, by the weight of the plate.
The elastic properties of the elastic foundation may be characterized, if isotropy is
assumed, by a Young modulus E0 and a Poisson ratio v0. The approximate numerical
values1 of these constants, depending on the nature of the subgrade and based on
results of dynamical tests, are given in Table 63, together with the value of the
constantdM ka = (a)
^ 4 )
used in t h e following.
T A B L E 63. V A L U E S O F E L A S T I C C O N S T A N T S D E P E N D I N G ON
NATURE OF FOUNDATION

Subgrade E0, psi v0 k0, psi

Clay 11,000 0.17 5,700

hte

Loess and clay 13,000 0.42 7,900

Medium sand 14,000-18,500 0.33-0.23 7,900-9,800
Sand and gravel 40,000 0.31 22,000
Liassic plastic clay 38,000 0.44 23,500
Lime (air-slaked) 165,000-190,000 0.32-0.38 92,000-110,000
Sandstone 1,600,000 0.26 860,000
rig

We restrict the further consideration to the case of an infinitely large plate in a state
of axial symmetry. Using polar coordinates r, 0, we can write the plate equation as
DAAw(r) = q(r) - pir) (b)
where q(r) denotes the given surface loading and p(r) the reaction of the subgrade.
Let KQ{r,p,<p) be the deflection at the point (r,0) of the subgrade surface due to a
py

normal unit load applied on this surface (P,<P). The form of the ''influence function"
K0 depends merely upon the nature of the foundation. Making use of some properties
of the Bessel functions, it can be shown2 that Eq. (b) is satisfied by the expression
Co

1
Due to E. Schultze and H. Muhs, " Bodenuntersuchungen fur Ingenieurbauten,"
Berlin, 1950. See also Veroffentl. Degebo, Heft 4, p. 37, 1936.
2
The solution of the problem in this general form is due to D. L. HoIl, Proc. Fifth
Intern. Congr. Appl. Mech., Cambridge, Mass., 1938.

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 In Eq. (c) J0 denotes the Bessel function of zero order; the term depending on the
nature of the subgrade is

l
ria
in which the form of Ko is defined by

s being the distance between points (r,0) and (P,<P). Finally

ate
is the term depending on the intensity q(p) of the symmetrical loading at r = p.
In the particular case of a load P uniformly distributed along the periphery of a
circle with a radius c, we have

In the case of the load P distributed uniformly over the area of the same circle, Eq.
dM
(e) yields

where the Bessel function is of the order one. Finally, where a load is concentrated
at the origin (p = 0), we obtain from Eq. (/)
hte

As for the distribution of the reactive pressure, the respective function p(r) is
obtained from Eq. (6), the term

being previously expressed through its Fourier-Bessel transform (e). Thus, we obtain
rig

Now let us consider two particular cases with respect to the physical nature of the
subgrade. For a,floatingplate (Art. 57) the influence function Ko(s) is zero everywhere
except at s = 0, where the unit force is applied. With regard to Eq. (d) the quan-
py

tity Ko(a) then must be a constant. In order to get from Eq. (c) the expression
w(r) = p(r)/k, this in accordance with the definition of the modulus, we have to
assume K0(a) = 1/k. Using the previous notation Z4 = D/k (page 260), we obtain
from Eq. (c) the expression
Co

which actually satisfies the differential equation (178) of the floating plate.
In the case of an isotropic semi-infinite medium we have, by a result due to Bous-

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 sinesq,1 KQ(s) - (1 - wD/rE* and, by Eq. (d), K(a) = 2(1 - wD/Eaa, or

l
ria
where k0 is the elastic constant defined by Eq. (a). Writing for brevity,

we finally obtain the solution (c) in the more special form2

ate
In the particular case of a load concentrated at the origin, expression (m) in con-
nection with (h) yields
dM (187)

where X is written for ah. Therefore, the deflection under the load is

(188)

against the result 0.125Pl2/D of Hertz. The distribution of the pressure is readily
obtained from the general expression (j). We have at any point
hte

(189)

and especially under the load

rig

(190)

in comparison with the value of 0.125P/Z* obtained by Hertz. If we assume equal

values of wm*x in both cases, formula (190) yields a value for pmax which is 2.37 times
as large as the value from Hertz's formula (181). In such a case the relation
py

Z — 1.241Zo must hold, and curves of the respective deflections as calculated from
Eqs. (179) and (187) are shown in Fig. 131a. Figure 1316 shows in like manner the
variation of the pressure; this time, in order to obtain equal values for pmax in both
cases, it must be assumed that I — 0.806Z0.
It can be shown, finally, that the magnitude of bending moments in the vicinity of
1
Co

SeC, for example, S. Timoshenko and J. N. Goodier, "Theory of Elasticity,"

2d ed., p. 365, New York, 1951.
2
For this result see also S. Woinowsky-Krieger, Ingr.-Arch., vol. 3, p. 250, 1932, and
vol. 17, p. 142, 1949; K. Marguerre, Z. angew. Math. Mech., vol. 17, p. 229, 1937;
A. H. A. Hogg, Phil Mag., vol. 25, p. 576, 1938.

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

 the concentrated load is the same for foundations of both kinds if expressed in terms
of the dimensionless argument x = r/l and x = r/h, respectively. We conclude from
this fact that expressions for bending moments, such as given by Eq. (183), can also be
used for a plate resting on an isotropic elastic medium if we replace I by Zo.. Proceeding

l
in this manner with the stress formula {x) of Westergaard (page 275), we arrive at the

ria
formula

in which ko is given by Eq. (a), and b denotes the same quantity as on page 275.
The problem of the bending of a finite circular plate leads to an infinite set of linear

ate
equations for the coefficients of the series, which has to represent the deflections of
such a plate. 1
The use of the method of finite differences should also be considered in handling the
problem of finite circular plates. 2
The bending of an infinite plate supported by an elastic layer, which rests in its turn
on a perfectly rigid base, 3 and the problem of a semi-infinite pavement slab 4 have also
been discussed.
Stresses due to a highly concentrated surface load should be corrected in accordance
dM
with the general theory of thick plates. However, a special theory of thick plates
supported elastically has also been established.6
x
See H. Borowicka, Ingr.-Arch., vol. 10, p. 113, 1939; A. G. Ishkova, Doklady
Akad. Nauk 8.8.S.R., vol. 56, p. 129, 1947; G. Pickett and F. J. McCormick, Proc.
First U.S. Nail. Congr. Appl. Mech., p. 331, Chicago, 1951. The effect of raising the
outer portion of the plate submitted to a central load was discussed by H. Jung,
Ingr.-Arch., vol. 20, p. 8, 1952. For bending of rectangular plates see M. I. Gor-
bounov-Posadov, Priklad. Mat. Mekhan., vol. 4, p. 68, 1940.
hte

2
A. Habel, Bauingenieur, vol. 18, p. 188, 1937; for application to rectangular plates
see G. Pickett, W. C. Janes, M. E. Raville, and F. J. McCormick, Kansas State Coll.
Eng. Expt. Sta. Bull. 65, 1951.
3
A. H. A. Hogg, Phil. Mag., vol. 35, p. 265, 1944,
4
G. Pickett and S. Badaruddin, Proc. Ninth Intern. Congr. Appl. Mech., vol. 6,
p. 396, Brussels, 1957.
5
The first discussion of the statical and dynamical behavior of such plates is due to
K. Marguerre, Ingr.-Arch., vol. 4, p. 332, 1933; see also I. Szab6, Ingr.-Arch., vol. 19,
rig

pp. 128, 342, 1951; Z. angew. Math. Mech., vol. 32, p. 145, 1952. For application of
E. Reissner's theory see P. M. Naghdi and J. C. Rowley, Proc. First Midwest Conf.
Solid Mech. (Univ. Illinois), 1953, p. 119, and D. Frederick, J. Appl. Mechanics,
vol. 23, p. 195, 1956.
py
Co

Copyright © 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com