See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/319727271

Solution Methodologies

Chapter · January 2005

DOI: 10.1007/0-8176-4420-2_4

CITATIONS READS

0 773

2 authors:

Omri Rand Vladimir Rovenski

Technion - Israel Institute of Technology University of Haifa
116 PUBLICATIONS 971 CITATIONS 236 PUBLICATIONS 800 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Vladimir Rovenski on 09 October 2017.

The user has requested enhancement of the downloaded file.

4
Solution Methodologies

This chapter discusses various solution methodologies for the two-dimensional BVPs intro-
duced in Chapter 3 that may be adopted for diversified applications. In particular, these meth-
ods will be useful for the anisotropic beam analyses presented in subsequent chapters. Nev-
ertheless, the same methods are beneficial for many other applications such as the analysis of
plates, see S.3.5, or the analysis of anisotropic foundations discussed, for example, by (Mu-
ravskii, 2001).
In order to keep the discussion generic enough and capable of handling many and different
domain shapes, the solutions are oriented to global polynomial analysis. Since this class of
polynomial solutions does not always supply exact solutions (i.e. some combination of geome-
tries and loading modes require other specific types of analytical solutions), some approxima-
tions become inevitable.
The proposed approach keeps its analytical nature in many senses. This is due to the fact
that we shall use an analytical elimination process that ensures exact fulfillment of the field
equations, while the entire approximation, if required, migrates to the boundary conditions. In
addition, all approximation procedures that will be derived in this chapter are realized by a
least-squares process, where all operations are carried out in a rational manner that loses no
accuracy, and yields an “exact-approximate” solution. In other words, the resulting solution
may be considered as exact since it can be analytically proved to be the “closest” (or the “most
accurate”) polynomial solution according to the least-squares criteria.
The above technique has several advantages. First, the exactness of fulfilling the field equa-
tions ensures integrability of the deformation components (since as shown in Chapter 3, the
field equations emerge from compatibility requirements). Secondly, it enables a simple track-
ing of the error over the boundary line. Hence, the error interpretation becomes more intuitive
(e.g., for a plane-strain problem, the error may be interpreted as “additional loading over the
contour”). The influence of such an error may be estimated more easily than the consequence
of an error in the fulfillment of the field equations. It also keeps the solution accuracy in areas
(over the domain) that are far from the boundary.
To clarify the derivation, the general solutions of the field equations are separated into two
parts: the particular solutions, and the general (or “prescribed”) solutions of the homogeneous
 126 4. Solution Methodologies

field equations. Although such a separation into particular and homogeneous solutions is not
always essential, it greatly contributes to an insight into the solution ingredients.
The last part of this chapter is devoted to the complex potentials method which exhibits some
analytical superiority. The method is derived and then implemented via Fourier series analysis.

While only a few, representative examples appear in the text, the solution schemes in this
chapter are documented as programs as well, see S.P.4. The reader may rerun all examples and
many others as required. In addition, these programs are extensively used in further chapters
of this book.

4.1 Unified Formulation of Two-Dimensional BVPs

Prior to the solution methodologies presentation, we shall review here the general formulation
of BVPs in a simply connected x, y-plane domain Ω, and their solutions’ separation into par-
ticular and homogeneous ones. The BVPs will be expressed using harmonic (Λ) or biharmonic
(Φ) functions (in some contexts, the above functions are also accompanied by the adjective
“generalized”). In the equations documented below we assume (x, y) ∈ Ω for the field equa-
tions, and (x, y) ∈ ∂Ω for the boundary conditions. In addition, all given boundary functions
d
are assumed to be single-valued. The ds -case of boundary conditions leaves the values of the
solution functions Λ, Φ and the derivatives Φ, x , Φ, y undetermined by a constant, and for the
sake of simplicity one may put {Λ, Φ, Φ, x , Φ, y }(0, 0) = {0, 0, 0, 0}.

The Dirichlet/Neumann BVPs are, see (3.121), (3.128), (3.195), (3.216),

(2)
∇3 Λ = F0Λ over Ω, (4.1a)
d
Dirichlet: Λ = G3 or Λ = F3 on ∂Ω, (4.1b)
ds
Neumann: Dn1 Λ = F3Λ on ∂Ω. (4.1c)

The above BVPs should be solved under the conditions (3.126), (3.129),

F3 = 0 or F3Λ = F0Λ . (4.2)
∂Ω ∂Ω Ω

The biharmonic BVP for plane-strain is

(4)
∇1 Φ = F0 over Ω, (4.3a)
d
{Φ,x , Φ,y } = {−F1 , F2 } or {Φ,x , Φ,y } = {−G1 , G2 } on ∂Ω (4.3b)
ds
which should be solved under the single-value conditions (3.72),

{F1 , F2 , F2 y − F1 x} = {0, 0, 0}, (4.4)

∂Ω

or, by integrating (by parts) the third equation of (4.4) and using Fi = d
ds Gi ,

{G2 cos(n̄, x) + G1 cos(n̄, y)} = 0. (4.5)

∂Ω

(4) (4)
For plane-stress analysis, ∇1 should be replaced by ∇3 , see (3.58), (3.202), (3.206), (alter-
natively, bi j may be replaced by ai j in the resulting expressions).

View publication stats