Sunteți pe pagina 1din 86

Models for Raft

Foundation
Analysis
Raft Foundations:
common types
Raft Foundations Analysis:
Approximate “Static” Method
RUDIMENTARY-FLEXIBLE
RIGID APPROACH
APPROACH
(AS IN FOOTINGS)

- Static equilibrium alone is satisfied


- Deformation is entirely ignored/ settlement is treated as a
separate problem
- Behaviour of foundation material is entirely ignored (No SSI)
- Thus, the geotechnical aspect is practically non-existent
Raft Foundation Analysis:
Flexible Approaches (with SSI)
BASED ON ANALYTICAL BASED ON FE DISCRETIZATION OF
SUBGRADE MODELS SUBGRADE

Mechanical subgrade models The entire soil medium is modelled


using FE discretization techniques:
Continuum subgrade models ◦ 2D FE models for beams
◦ 3D FE models for plates
Synergy
Flexible Methods Based on Subgrade
Models: Simplest Model - Winkler
CONTINUUM IDEALIZATION MECHANICAL IDEALIZATION

Relation between surface


deformation and contact pressure
introduced - Winkler (1867):

• Thus, the model is also referred to as


Origin: the upward pressure on the
the dense-fluid model
surface of a body floating on a dense
• and have the same dimension
fluid:
• SSI is accounted for only approximately
Winkler Model: Obvious
Shortcomings
IDEALLY RIGID FOUNDATION ON IDEALLY RIGID FOUNDATION ON
CONTINUUM WINKLER’S SPRING BED

• A clear disagreement in the contact pressure


• Discontinuity in deformation in Winkler’s model
Winkler Model:
Shortcomings
IDEALLY FLEXIBLE FOUNDATION IDEALLY FLEXIBLE FOUNDATION
ON CONTINUUM ON WINKLER’S SPRING BED

• A clear disagreement in deformation


pattern
• Discontinuity in deformation in Winkler’s
model (dishing effect not well reflected)
Winkler Model:
Shortcomings
1. No interaction among Winkler’s springs!
2. Soil is represented by a single quantity, ks!
3. ks is not a fundamental soil parameter!
 Proof: the surface deflection of the elastic half space, for example, due to a uniform pressure p on
a given region is given by

 B = foundation width; I = influence factor dependent on point location and foundation shape and
size. This relation may be re-written to obtain an expression for ks consistent with Winkler’s
relation:

 Obviously, ks depends on the foundation size and shape and is not a soil parameter, whereas E and
ν are; i.e. a softer spring bed is assigned to a larger foundation regardless of the soil type.
Improvements on Winkler’s
Model Necessary!
Three approaches are followed:
 Approach 1: Developing empirical relations for ks
 This is in terms of the soil and foundation characteristics with the intention
to further use Winkler’s model due to its simplicity

 Approach 2: Developing realistic mechanical models:


 Improving Winkler’s model to account for the missing shear interaction by
interconnecting the individual springs using additional elements with
additional parameters

 Approach 3: Developing continuum models


 Gets into the mechanics of the subgrade material idealized as an elastic
continuum accounting for the boundary conditions at the ground surface.
Approach 1: Estimating ks
A number of theoretical/empirical relations are proposed
All are dependent on both soil characteristics and foundation
size/shape but different in form
Which one to use? Rather confusing!
Approach 1: Further Proposals
Some authors propose the determination of ks from
both bearing capacity and settlement
considerations:
◦ Estimate the safe bearing capacity
◦ Proportion your foundation
◦ Compute the average contact pressure
◦ Compute settlement and check if it is within the
permissible limit
◦ Compute ks as the ratio of the contact pressure to
the settlement
◦ Use this uniformly in the foundation model (FE
discretized)
Approach 1: Raft Foundation
Models on Winkler’s Spring Bed
PLATE ON WINKLER SPRINGS BEAM-STIFFENED PLATE ON
WINKLER SPRINGS

• FE-discretized plate on a vaguely represented soil!


• Model Incompatibility!
Approach 1: Pseudo-coupled Springs

Still others propose the use


of pseudo-coupled Winkler’s
springs to bring about the
missing dishing effect:
◦ Subdivide mat into sub-
regions (Liao, 1991; Horvath
1993);
◦ Use double modulus at the
edges (ACI, 1993) and
intermediate values in
between the central zone
and the edges
Approach 2: Development of
advanced mechanical models
A number of mechanical models were proposed to
introduce interaction among Winkler’s springs
Two-parameter models:
◦ Filonenko-Borodich model (1950)
◦ Hetényi model (1950)
◦ Pasternak model (1954)
◦ etc

Three-parameter models
◦ Kerr Model (1965, 1985)

Higher order models


◦ Kerr and Rhines (1967) (superfluous and irrelevant)
Two-Parameter Mechanical Models:
Filonenko-Borodich (FB) Model (1950)

Shear interaction is effected by a stretched membrane


under an all-round tension, T, joining Winkler’s springs
Two parameters are involved: kP and T
Two-Parameter Mechanical Models:
Filonenko-Borodich (FB) Model (1950)
A stretched membrane under
tension, T, connects Winkler’s springs.
At any location x, a shear force, V,
that varies with x is developed,
whereas the horizontal tension force
remains constant.
For moment equilibrium:

M O 0
 Vdx  Tdw
Or V  T dw dx
Two-Parameter Mechanical Models:
Filonenko-Borodich (FB) Model (1950)
For vertical force equilibrium:
F y 0
 V  V  dV   pdx  k P wdx  0
 k P w  dV dx  p x 

Substituting for V in this equation:


p x   k P w x   Tw x 
For a 3D problem (a plate): Note that Winkler’s simple

p  x, y   k P w x, y   T 2 w x, y 
linear relationship is
transformed into a 2nd –
order p.d.e. – a result of the
Where, introduced interaction
 2   2 x 2   2 y 2 among the springs.
Two-Parameter Mechanical Models:
Pasternak Model (1954)
Shear interaction is effected by a pure-shear element
joining Winkler’s springs
Two parameters are involved: kP and GP
Two-Parameter Mechanical Models:
Pasternak Model (1954)
A pure shear element with a
parameter, GP, connects Winkler’s
springs.
The force on each face is a pure shear
force.
Vertical force equilibrium requires as in
the FB model:
dV
p  kP w 
dx
d  .1
 p  kP w 
dx
p x   k P w x   GP w x 

Forpax3D
, y  problem:
 k P w x, y   GP  2 w x, y 
Two-Parameter Mechanical Models:
Hetényi Model (1950)
In Hetényi model, shear interaction is supposed to be effected
by a flexural element
Two parameters are thus involved: kP and EI
Two-Parameter Mechanical Models:
Hetényi Model (1950)
Vertical equilibrium requires as
before
dV
p  kP w 
dx

Noting that V=dM/dx and M =


-EId2w/dx2, the relation for a
constant EI becomes
p  x   k P w x   EIw iv  x 
This model is inappropriate for
the intended purpose, since the
For 3D problems: missing link is shear interaction,
not flexure, which is irrelevant in
p x, y   k P w x, y   D 2 2 w x, y  soils
Three-Parameter Mechanical Models:
Modified Pasternak/Kerr Model (1965)
The development of this model was motivated by the continuum model
of Reissner, whose mathematical relation has yet another term involving
the 2nd derivative of p (see first slides on Approach 3)
It was devised by adding a second bed of springs on top of Pasternak’s
model
It thus involves three parameters: ku, kl and GK
Three-Parameter Mechanical Models:
Modified Pasternak/Kerr Model (1965)
For equilibrium of the upper spring
F y  0  ku dx w  wl   p dx
p
 wl  w 
ku
The lower part is identical to Pasternak’s
model except w is replaced
2
by wl. Thus,
d wl
p x   kl wl  x   GK
dx 2
Inserting the first relation into the
second and rearranging:
GK kk G k
p x   p x   u l w x   K u w x 
ku  kl ku  kl ku  kl
For a 3D problem:
GK kk G k
p  x, y    2 p x, y   u l w x   K u  2 w x, y 
k u  kl ku  kl ku  k l
Three-Parameter Mechanical Models:
Modified Filonenko-Borodich Model (1964)
This model is identical to the modified Pasternak model with the mere
replacement of GK by T as observed earlier
Three parameters are involved: ku, kl and T
Mathematical relation:

T kk Tku
p  x, y    2 p x, y   u l w x    2 w x, y 
ku  kl ku  kl ku  kl
Higher Order Mechanical Models:
Modified Hetényi Model(1960)
Hetényi (1960) also proposed a modified three-
parameter version of his two-parameter model using
his flexural element as a connector.
Once again, it is important to point out that this model
is inappropriate, as it involves the 4th derivatives of p
and w, and also, flexure is irrelevant to soils.
Higher Order Mechanical Models:
Kerr and Rhines (1967)
Kerr and Rhines (1967) thought that any level of
accuracy could be achieved by subsequent
modification of these models through the addition of
more spring beds and shear elements

…..etc

Kerr and Rhines’ approach is, however, more of an academic exercise


than is a realistic and practical one.
Approach 3: Continuum Models -
Reissner’s Model (1958)
Reissner (1958) idealized the soil medium as an elastic layer of
thickness, H, overlying a perfectly rigid medium as shown.
He made the simplifying assumption of zero in-plane stress
components in advance; i.e.
 x   y   xy  0
Approach 3: Development of Continuum
Models - Reissner’s Model (1958)
Thus, the stress tensor/matrix takes the form:
 0 0  xz 
 
 0 0  yz 
   
 xz yz z 

The stress (equilibrium) equations yield:


 x , x   xy , y   xz , z  0   xz , z  0
 y , y   xy , x   yz , z  0   yz , z  0
 z , z   xz , x   yz , y  0   z  Qz  c; Q   xz , x   yz , y

Note that neglecting the in-plane stresses has also the consequence of
the vertical shear stresses being constant (see first two equations). This
is obviously unrealistic.
Approach 3: Development of Continuum
Models - Reissner’s Model (1958)
Using the stress boundary condition at the surface, the last
equilibrium Equation becomes:
 z  Qz  p
Inserting this in the combined stress-strain and strain-
displacement relationship for the vertical direction,
integrating w.r.t. z and applying the displacement b.c’s at the
bottom (zero) and at the surface (w0), one obtains

where
Approach 3: Development of Continuum
Models - Reissner’s Model (1958)
The vertical shear stresses are given by

We substitute for w(x,y) in these equations, integrate w.r.t. z, the


remaining b.c’s applied and the resulting equations solved for the
shear stresses:

These expressions and that of together with the expression for Q


are inserted in the 3rd equilibrium equation and rearranged to
obtain
Continuum Models:
Reissner’s Model (1964)
Note that this result is what motivated Kerr(1964) to modify
Pasternak’s two-parameter model and come up with his
three-parameter mechanical model presented earlier that
has a similar mathematical form and order. It is given by
GK kk G k
p  x, y    2 p x, y   u l w x   K u  2 w x, y 
k u  kl ku  kl ku  k l

Compare it with Reissner’s model just found:


Continuum Models:
Vlasov’s Model (1966)
Vlasov and Leontev (1966) represented the soil medium the
same way as Reissner did
They made, however, the following simplifying assumptions:
◦ Zero horizontal displacement components; i.e.
◦ The vertical deformation shape, , are proposed from optimization
using variational principles
Continuum Models:
Vlasov’s Model (1966)
They obtained the following mathematical model:

Where

Note that this is equivalent in form and order to the two-


parameter Pasternak and/or F-B models:
Mechanical and Continuum Models:
Observations
On Mechanical modelling:
◦ Whereas the initial attempts to bring about the missing interaction
among Winkler’s springs using few additional elements are desirable,
the subsequent attempts to come up with endless number of
elements and forms of the mechanical models (e.g. Kerr and Rhines)
make little sense.
On Continuum modelling:
◦ All the continuum models developed are highly simplified and, one
way or another, violate the basic rules of mechanics from the outset.
◦ The simplifying assumptions seem to be somewhat arbitrary and
lacking rationality.
These observations motivated the search for better models
On Continuum Models:
Questions raised
1. Is it after all necessary to neglect any
deformation, strain or stress components of the
continuum in advance to devise a subgrade
model?
2. Can’t we make more reasonable assumptions
based on known facts, if necessary?
3. Wouldn’t it even be possible to arrive at a
generalized model without making any
simplifying assumptions so that specific models
can be derived as needed?
Developing Improved Continuum Models

The soil medium is physically represented the same way as


in Vlasov’s and Reissner’s models
But, no stress, deformation or strain component is
neglected!
The stress tensor has no zero element!

  x  xy  xz 
 
 xy  y  yz 
   
 xz yz z 
1st Generation Improved
Continuum Models
Improved models were derived based on the following
premises:
◦ Instead of neglecting the normal stresses, and , their ratio with the
vertical normal stress, , can be reasonably estimated based on
existing knowledge of lateral earth pressure theories and stress
distribution.
◦ The depth-wise variation of the vertical shear stresses, and , can
be reasonably estimated based on observation of existing results
for various surface loading cases

Accordingly, three different model variants were


obtained (Worku, 2009):
◦ Interestingly, all are of the same order and form as Reissner’s
mathematical model, though with different coefficients
1st Generation Improved
Continuum Models: Variant 1
Assumptions:
1. The normal stresses are related linearly; i.e.
 x  k x z ;  y  k y z

2. The vertical shear stresses are constant with depth (This


assumption is made to merely show that Reissner’s model is a
special case of this model variant)
The following mathematical
 GH 2
 2 E modelGH
is obtained:
p     p  w0   2 w0
 12 E  H 3
  1   k x  k y 
kx  k y  0
Reissner’s model is obtained when
1st Generation Improved
Continuum Models: Variant 2
Assumptions:
1. The normal stresses are related linearly; i.e.
 x  k x z ;  y  k y z

2. The vertical shear stresses vary bi-linearly (This is based on


observation of plots for various surface loading cases)
The following

mathematical
2

model is obtained:
GH E  GH  2
p  2.35   2 p  w0  2.22  w0
 12 E  H  3 
  1   k x  k y 
1st Generation Improved
Continuum Models: Variant 3
Assumptions:
1. The normal stresses are related according to a decaying exponential
function (based on observation of plots); i.e.

 x  rx e z z ;  y  ry e z z

2. The vertical shear stresses vary bi-linearly (This is based on


observation of plots for various surface loading cases)
A similar mathematical model with different expressions of the
coefficients was obtained:

**All three variants show the same form and order of differential
equations!
Developing a Generalized
Continuum Model
The remarkable similarity of the mathematical equations of the
three variants and Reissner’s model encouraged a search for a
generalized model.
To this end, the horizontal normal and the vertical shear stress
components were expressed in a more generalized form as follows:

*Note that there is no simplifying assumptions made here.


**The above representations are mathematically permissible.
Developing a Generalized
Continuum Model
After satisfying all the pertinent conditions, the following
generalized model was obtained (Worku, 2010 (ASCE, GSP
199))

Where the coefficients are given by the following definite


integrals:
Developing a Generalized
Continuum Model
This generalized mathematical model is still of the same
order as Reissner’s model and the 1st generation models
proposed earlier. This is one of the major findings!
Any number of variant models can be obtained by making
different combinations of assumptions regarding
◦ The lateral to vertical normal stress ratios and
◦ the depth-wise variation of the vertical shear stress components

It can be shown that the previously suggested 1st generation


models and all existing continuum models (Reissner, Vlasov
& Leontev, etc) can be regarded as special cases of this
generalized (umbrella) model.
This is the second major finding!
Developing a Generalized
Continuum Model
The mechanical equivalent of this generalized
continuum model is the Kerr or the modified FB
model
Hence,
◦ No need to look for more complicated mechanical models
than the three-parameter Kerr model!!
◦ The plethora of mechanical models proposed by Kerr and
Rhines (1967) presented above with ever increasing order
and complexity are irrelevant.
◦ The maximum form and order of the mathematical model
is: p  c  2 p  c w  c  2 w
1 2 0 3 0
Developing a Generalized
Continuum Model
Furthermore, It has been shown that
◦ The generalized model can be reduced to a Pasternak-type model, if
only the horizontal deformations u and v are neglected resulting in

◦ The model can also be reduced to a Winkler-type model, if the


vertical shear stresses alone are neglected yielding

◦ This latter result shows in a different way that Winkler’s model


indeed misses the vertical shear interaction among soil elements,
but interaction among the normal stresses can exist
Model Equivalence
GENERALIZED CONTINUUM EQUIVALENT MECHANICAL
MODEL MODEL
Synthesis of the Mechanical and
Continuum Models
The equivalence of the two equations enables us to obtain
the open parameters of the mechanical models by equating
corresponding coefficients. Thus, (Worku, 2010, 2013):

When the shear interaction in the mechanical model is


neglected, there remain two beds of springs connected in
series that can be replaced by a single bed with its effective
coefficient obtained as
Note that this result was directly
obtained from the continuum
Generating Specific Types of
Models
We are now in a position to generate
variants of models by specifying two
stress functions of depth:
1. Normal stress ratios: g x  z     x  z  z ; g y  z    y  z  z 
2. Variation with depth of the shear
  z  and  yz  z 
stresses xz
This can be made based on observations
of plots of existing results in stress
distribution theory (e.g. Poulos and
Davis, 1980)
Generating Realistic Individual
Models
The general trend of the depth-
wise variations is as shown:
These functions can be
represented by exponential and
bilinear functions as shown g x  z  and g y  z 
schematically
 x  z   g x  z  z ;  y  z   g y  z  z
I xz ( z ) and I yz ( z )
g x  z   rx e z
; g y  z   ry e z

Note: Coefficients are determined


from curve fitting
Generating Realistic Models
The resulting model is

After synthesis, the mechanical model parameters are found as

K B
Note that H is now replaced by
K
is used as a calibration factor for further refinement of the model
Calibration of the Models
The proposed model can be refined through calibration
Calibration is possible with the use of
◦ FE-based software,
◦ Existing theoretical results, or
◦ Lab model testing.

The first two are more feasible and handy


Two calibration efforts have been made using these tools:
◦ Calibration using Plaxis for beams
◦ Calibration using theoretical results for plates (needs more
work)
Calibration of the Models for
Beams
A calibration effort for beams was
made using Plaxis
The study conducted revealed that
the factor is insensitive to one of the
most important factors - the soil-
beam relative rigidity (see plots) -
and may be taken as a constantK  3
value of around
(Worku & Degu, 2010)
A more rigorous study is currently
underway encompassing a wider
range of relative rigidity. Results
indicate values between 2 and 3.
Calibration of the Models: for
Plates
Accordingly, the following model parameters are
proposed for use in beams (Worku & Degu, 2010)

Similarly, a calibration exercise was done for plates of


circular and rectangular shapes based on available
theoretical results. The proposed values of the
calibration factor are (Worku, 2013):
Calibration of the Models: for
Plates
With implementation of the model in mind, the work
was extended to deriving calibrated parameters for
Pasternak’s mechanical model (Worku, 2014)
These parameters are derived by seeking equivalence of
the deformations of a beam on Pasternak model with
that on the calibrated Kerr model. The model
parameters obtained are:

Note that the same values of calibration factors


proposed for Kerr’s model are applicable.
Summary of Findings
Therefore, based on representation of the soil medium
as an elastic continuum of finite thickness, calibrated
parameters are found for all three levels of mechanical
models:

◦ Winkler (single-parameter) model:


Summary of Findings
Pasternak/FB (two-parameter) Model
Summary of Findings
Modified Pasternak or Kerr
(three-parameter) Model
Analysis of Beams on Elastic
Foundations
The governing differential equation is
obtained from equilibrium
consideration of the isolated
differential element of the beam:

 V  x   Bp x   Bq x 

With V(x)=M’(x) and M=-EIw’’(x), one


obtains

 EIw  x    Bp x   Bq x 


p(x) is brought in from the applicable


subgrade model
1. Beams on Winkler’s Subgrade
When the subgrade is idealized using Winkler’s simple
model of , the DE for constant EI becomes
EIw iv  x   Bk s w x   Bq x 

The homogenous/complementary solution has the


general form of
w x   Aex

After substitution and rearranging, one obtains the


characteristic polynomial:
 4  44  0

Where is called rigidity factor, with the dimension of L-1,


relating the rigidities of the soil and beam and defined as:
  4 k s B 4 EI  b
1. Beams on Winkler’s …
The characteristic polynomial can be reformed using the
substitution to
y4  4  0

This polynomial has the following four distinct roots:


y1  1  i   y 3 ; y 2  1  i   y 4

And hence
 1   1  i    3 ;  2     1  i     4

With these roots, the solution becomes


w x   A1e1x  A2 e 2 x  A3e 3 x  A4 e 4 x
1. Beams on Winkler’s …
Applying Euler’s expansion to the complex exponential
functions the general solution takes the following form:
w x   e x  C1 cos x  C 2 sin x   e  x  C3 cos x  C 4 sin x 

With further expansion of the real-valued exponential


functions, one obtains the following alternative form
w x    B1 cosh x  B2 sinh x  B3 cos x  B4 sin x 

The four open coefficients are to be determined from the


prevailing boundary conditions of the beam
Either of the two forms can be used as found convenient
1.1 Infinite Beams on Winkler’s
Subgrade - Concentrated Force
We consider now a beam of infinite length subjected to a
concentrated force, P, as shown (classical case)
◦ Note that the origin is directly under the force

The first form of the solution is convenient


The boundary conditions are


 k s w x  dx  P 2
0
1.1 Infinite Beams on Winkler’s Subgrade - Concentrated
Force

The first form of the solution is more convenient for this


case.
◦ The 1st b.c. requires
◦ The 2nd b.c. requires
◦ The 3rd b.c. requires
P P
C 

2 4 EI 
4
2k s B

Inserting this back in w(x), one obtains the solution


The slope, moment and shear are then found by
successive differentiation.
1.1 Infinite Beams on Winkler’s
Subgrade - Concentrated Force
Thus, the deflection, slope, bending moment and shear
force are obtained as
P P2
w x   A x  ;   x   w x    B  x 
2k s B ks B
P
M  x    EIw x   C  x 
4
P
V  x    EIw x    D x 
2
Where
A x   e  x  cos x  sin x ; B x   e  x sin x
C  x   e x  cos x  sin x ; D x   e x cos x

The dimensionless functions to D can be easily plotted


against the dimensionless length,
1.1 Infinite Beams on Winkler’s
Subgrade - Concentrated Force:
 Plots of to D
Variation of normalized quantities
0.4

0.2
Normalized responses

0
-6 -4 -2 0 2 4 6
-0.2

-0.4

-0.6

-0.8

-1

-1.2

Normalized length

Deflection Slope Moment Shear


1.1 Infinite Beams on Winkler’s
Subgrade - Concentrated Force
The following are some observations on the trend of the
curves
1. The beam is lifted over part of the length of the beam
implying that the Winkler springs carry both tension
and compression;
2. For the beam may be considered as long (infinite)
beam as all quantities are less than 0.5% of their
respective maximum values;
3. The definition of long or short beam is a function of
the relative soil-beam rigidity, and not of the beam or
the soil in isolation.
4. The maximum deflection, bending moment and shear
1.2 Infinite Beams on Winkler’s
Subgrade - Uniform Load
We now consider a uniform load acting on a finite portion of
the infinite beam as shown

The solution follows through integration of those obtained


above for a concentrated force. Two cases may be identified.
Case1: Point within loaded area
 r e   cos   sin   d  
p  0 
wJ 
2k  s e   cos   sin   d 
 0 
1.2 Infinite Beams on Winkler’s
Subgrade - Uniform Load
The above integral reduces to
p
wJ   2  D  r   D  s  
2k s

Similarly follow the slope, moment and shear equations:


p pB
J   A r   A s  ; MJ   B   r   B   s  ;
2k s 4 2

pB
VJ   C  r   C  s  
4
Noteworthy is that for large r and s (i.e. an infinite beam uniformly
loaded by p over its entire length), all quantities diminish to zero except
the deflection, which become . Hence, the load is directly transmitted to
the subgrade without causing internal stresses in the beam.
1.2 Infinite Beams on Winkler’s
Subgrade - Uniform Load
Case 2: Point outside loaded area (left/right) of the load
In this case, one obtains
p p
wJ    D   r   D   s  ; J   A r   A s  ;
2k s 2k s
pB pB
M J   2  B   r   B   s  ; VJ   C  r   C  s  
4 4
1.2 Infinite Beams on Winkler’s
Subgrade - Concentrated Moment
The problem of a concentrated moment acting on an
infinite beam as shown can be regarded as a limiting
case of a combination of two concentrated forces acting
at a small distance, e, apart to yield a concentrated
couple.

For the portion to the right of the moment one can


show that M 2 M 3
wJ  0
B  x  ; J  0
C  x  ;
ks B ks B
M0 M 0
MJ  D  x  ; VJ   A x 
2 2
Beams of Finite Length on Winkler’s
Subgrade: Method of Superposition
Consider a beam of finite length as shown subjected to
a concentrated force, P, and a distributed load, p.
The finite beam may be considered as a portion of an
infinite beam under the same loading

With the origin of the coordinate axes as shown, we


need first to determine the moments and shear forces
in the infinite beam at A and B – the ends of the finite
beam – due to the given loads. These are denoted by
MA, VA, MB and VB
Beams of Finite Length on Winkler’s
Subgrade: Method of Superposition
Next the conditioning moments, M0A and M0B and the
conditioning shear forces, V0A and V0B, at A and B of the
infinite beam are obtained to satisfy the boundary
conditions in the finite beam.
◦ For example, the forces V0A and V0B give rise to the
moments of and , respectively, at A.
◦ Similarly, the moments M0A and M0B give rise to the
moments of and respectively, at the same point A.
Beams of Finite Length on Winkler’s
Subgrade: Method of Superposition
To satisfy the zero-moment condition at A in the finite
beam, the sum of these four should balance with MA;
i.e. V C  0  V C  l   M D 0  M D l   M
0A 0B 0A 0B
A
4 4 2 2

Similar equations can be derived for all four boundary


conditions to result in the matrix equation of
 C w  R   F  or  R   C w  1  F 
Where
◦ {R} is the unknown vector of end conditioning
forces/moments
◦ {F} is the external force vector
◦ [Cw] is the coefficient matrix
Beams of Finite Length on Winkler’s
Subgrade: Method of Superposition
The coefficient matrix, [Cw] can be easily found as
 C  0 C  l  D 0 D  l  
 4 4 2 2 
 C  l  C  0 D  l  D 0  
 
 C w    4D 0 4
D  l  
2

2 
  A 0  A l  
 2 2 2 2
 D  l  D 0    
  A l  A 0 
 2 2 2 2 

It can be shown that


V0 A  M 0A 
  P0  P0;   M 0  M 0
V0 B  M 0B 
Beams of Finite Length on Winkler’s
Subgrade: Method of Superposition
Where
P0  4 E1  S A 1  D l    M A 1  A l   ;
P0  4 E 2  S A 1  D l    M A 1  A l   ;
2 E1
M 0    S A 1  C  l    2M A 1  D l   ;

2E
M 0   2  S A 1  C  l    2M A 1  D l   

And
S A  1 M A  1
  V A VB  ;   M A  M B 0
S A  2 M A  2
E1 
   2 1  D   l    1  D   l     1  A  l    1  C   l    1

E2 
Beams on Pasternak’s (or FB’s)
Subgrade
When the subgrade is idealized using Pasternak’s (or FB’s)
model of , the d.e. of the beam becomes
EIw iv  x   G P Bw x   Bk s w x   Bq x  ; or
EIw iv  x   TBw x   Bk s w x   Bq  x 

The homogenous solution has the usual form of


w x   Ae mx

After substitution, rearranging and introducing and , one


obtains the characteristic polynomial:
Tˆ 2 kˆs
m 4
m  0
EI EI
Beams on Pasternak’s (or FB’s)
Subgrade

The characteristic polynomial has the four separate roots


of
2
Tˆ ˆ
k  Tˆ 
m1, 2,3, 4   i s   
2 EI EI  2 EI 

The solution has the usual form:


w x   A1e m1x  A2 e m2 x  A3 e m3 x  A4 e m4 x

Three cases arise depending on whether the internal


square root is positive, zero or negative
Infinite Beams on Pasternak’s (or
 FB’s) Subgrade - Case I:
This case is the most common case in practice, as the
shear interaction can rarely be larger than double the
square root expression, which is the product of the beam
rigidity and the soil stiffness.
In this case, the internal square root is positive, and the
four roots form two pairs of conjugate complex numbers
expressed as
m1, 2,3, 4      i 

where
ˆ Tˆ ˆ Tˆ ˆ kˆs
  
2
;   
2
;  4
4 EI 4 EI 4 EI
Infinite Beams on Pasternak’s (or FB’s)
 
Subgrade - Case I:

With these roots substituted back, one obtains

   
w x   C1e x  C 2 e x cos x  C 3 e x  C 4 e x sin x

For an infinite beam, , so that the solution reduces to

w x   e x  C 2 cos x  C 4 sin x 


Infinite Beams on Pasternak’s (FB’s) Subgrade
 
Subjected to a Concentrated Force –
Case I:

We consider now a beam of infinite length subjected to a


concentrated force, P, as in Winkler’s subgrade

The first form of the solution is convenient


The remaining boundary conditions are:
, and

 k s w x  dx  P 2
0
Infinite Beams on Pasternak’s (FB’s) Subgrade
Subjected to a Concentrated Force –
 
Case I:

The 1 b.c. requires
st C4  C 2 ; and

  
w x   Ce x
 cos x  sin x 
  

Pˆ2
The 2nd b.c. requires C
2kˆ  s

This finally results in


P ˆ2 x
w x   e   cos x   sin x 
ˆ
2k s 
Infinite Beams on Pasternak’s (FB’s) Subgrade
 
Subjected to a Concentrated Force –
Case I:

Thus, the deflection, slope, bending moment and shear


force are obtained as
P ˆ2 x
w x   e   cos x   sin  x  ;
ˆ
2k s 
P 1 x
  x   w x    e sin  x
4 EI 
P 1 x
M  x    EIw x   e   cos  x   sin  x 
4 

V  x    EIw x   
P 1 x
4 

e 2 cos  x   2   2  sin x 
Infinite Beams on Pasternak’s (FB’s) Subgrade
 
Subjected to a Concentrated Moment –
Case I:

If a clockwise moment, M0, is applied at Point O, instead


of a force, the deflection, slope, bending moment and
shear force are similarly obtained as
M 0 1 x
w x   e sin  x;
4 EI 
M 1 x
  x  0 e   cos x   sin  x 
4 EI 

M  x  0
M 1 x
4 
  
e 2 cos x   2   2 sin  x 
M 0 ˆ2 x
V  x   e   cos x   sin  x 
2 
Infinite Beams on Pasternak’s (FB’s) Subgrade
 
Subjected to a Concentrated Moment –
Case I:

Note that, for T=0, both the above sets of equations for a
concentrated force and a moment reduce to the
corresponding formulas obtained for Winkler’s subgrade.
The solution for a uniformly distributed load can also be
similarly obtained through integration as in the case of
Winkler’s subgrade.
Finite beams can also be treated in a similar fashion
Beams on Kerr’s Subgrade
When the subgrade is idealized using Kerr’s model, the
mathematical relationship between the surface deflection and
the contact pressure is given by
T k k Tk u
p x   p   u l w  w
ku  kl ku  kl ku  kl

p(x) can be eliminated from this equation and that of the beam
given by
EIwiv  x   Bp x   Bq x 

The resulting d.e is


EIT vi  ku  kl  iv TB
w  x   EI   w  x   TBw x   k l Bw x   q x   q  x 
ku  ku  ku
Beams on Kerr’s Subgrade
The homogenous equation can be written as

w vi  x   a1 w iv  x   a 2 w x   a3 w x   0
Where
 ku  kl  ku B ku kl B
a1   ; a 2  ; a3  
 T  EI EIT

After substituting the usual exponential function of , the


above sixth order o.d.e. gives rise to the following sixth
order characteristic polynomial:
t 6  a1t 4  a 2 t 2  a3  0

Three cases arise in the process. Generally, each case


results in six roots. The solution will thus generally have
six terms with six integration constants involving more
work than the beam on Winkler and Pasternak models.

S-ar putea să vă placă și