Nonlinear finite elements/Partial differential equations

Partial differential equations (PDEs) are the most common method by which we model physical problems in engineering. Finite element methods are one of many ways of solving PDEs. This handout reviews the basics of PDEs and discusses some of the classes of PDEs in brief. The contents are based on Partial Differential Equations in Mechanics volumes 1 and 2 by A.P.S. Selvadurai and Nonlinear Finite Elements of Continua and Structures by T. Belytschko, W.K. Liu, and B. Moran.

Definition of a PDE

A PDE is a relationship between an unknown function of several variables and its partial derivatives.

Let $u(x_{1},x_{2},x_{3},t)$ be an unknown function. The independent variables are $x_{1}$, $x_{2}$, $x_{3}$, and $t$. We usually write

$u=u(x_{1},x_{2},x_{3},t)$

and say that $u$ is the dependent variable.

Partial derivatives are denoted by expressions such as

$u_{,1}={\frac {\partial u}{\partial x_{1}}}~;~~u_{,2}={\frac {\partial u}{\partial x_{2}}}~;~~u_{,11}={\frac {\partial ^{2}u}{\partial x_{1}\partial x_{1}}}\equiv {\frac {\partial ^{2}u}{\partial x_{1}^{2}}}~;~~u_{,12}={\frac {\partial ^{2}u}{\partial x_{1}\partial x_{2}}}~.$

Some examples of partial differential equations are

${\begin{aligned}u_{,t}=u_{,1}+u_{,2}&\Leftrightarrow {\frac {\partial u}{\partial t}}={\frac {\partial u}{\partial x_{1}}}+{\frac {\partial u}{\partial x_{2}}}\\\nabla ^{2}u=0\Leftrightarrow u_{,11}+u_{,22}+u_{,33}=0&\Leftrightarrow {\frac {\partial ^{2}u}{\partial x_{1}^{2}}}+{\frac {\partial ^{2}u}{\partial x_{2}^{2}}}+{\frac {\partial ^{2}u}{\partial x_{3}^{2}}}=0\\u_{,1111}=u_{,22}+u&\Leftrightarrow {\frac {\partial ^{4}u}{\partial x_{1}^{4}}}={\frac {\partial ^{2}u}{\partial x_{2}^{2}}}+u~.\end{aligned}}$

An example of a system of partial differential equations is

${\boldsymbol {\nabla }}({\boldsymbol {\nabla }}\bullet \mathbf {u} )+\nabla ^{2}\mathbf {u} +\mathbf {f} =\mathbf {0} \Leftrightarrow u_{k,ki}+u_{i,jj}+f_{i}=0$

In expanded form this system of equations is

${\begin{aligned}{\frac {\partial ^{2}u_{1}}{\partial x_{1}^{2}}}+{\frac {\partial ^{2}u_{2}}{\partial x_{2}\partial x_{1}}}+{\frac {\partial ^{2}u_{3}}{\partial x_{3}\partial x_{1}}}+{\frac {\partial ^{2}u_{1}}{\partial x_{1}^{2}}}+{\frac {\partial ^{2}u_{1}}{\partial x_{2}^{2}}}+{\frac {\partial ^{2}u_{1}}{\partial x_{3}^{2}}}+f_{1}&=0\\{\frac {\partial ^{2}u_{1}}{\partial x_{1}\partial x_{2}}}+{\frac {\partial ^{2}u_{2}}{\partial x_{2}^{2}}}+{\frac {\partial ^{2}u_{3}}{\partial x_{3}\partial x_{2}}}+{\frac {\partial ^{2}u_{2}}{\partial x_{1}^{2}}}+{\frac {\partial ^{2}u_{2}}{\partial x_{2}^{2}}}+{\frac {\partial ^{2}u_{2}}{\partial x_{3}^{2}}}+f_{2}&=0\\{\frac {\partial ^{2}u_{1}}{\partial x_{1}\partial x_{3}}}+{\frac {\partial ^{2}u_{2}}{\partial x_{2}\partial x_{3}}}+{\frac {\partial ^{2}u_{3}}{\partial x_{3}^{2}}}+{\frac {\partial ^{2}u_{3}}{\partial x_{1}^{2}}}+{\frac {\partial ^{2}u_{3}}{\partial x_{2}^{2}}}+{\frac {\partial ^{2}u_{3}}{\partial x_{3}^{2}}}+f_{3}&=0\end{aligned}}$

It is often more convenient to write PDEs in vector notation or index notation.

Order of a PDE

The order of a PDE is determined by the highest derivative in the equation. For example,

${\begin{aligned}{\frac {\partial u}{\partial t}}-{\frac {\partial u}{\partial x}}&=0~~~~{\text{is a first-order PDE.}}\\{\frac {\partial ^{2}u}{\partial x_{1}^{2}}}+{\frac {\partial ^{2}u}{\partial x_{2}^{2}}}+{\frac {\partial ^{2}u}{\partial x_{3}^{2}}}&=0~~~~{\text{is a second-order PDE.}}\\{\frac {\partial ^{4}u}{\partial x_{1}^{4}}}+{\frac {\partial ^{2}u}{\partial x_{2}^{2}}}-u&=0~~~~{\text{is a fourth-order PDE.}}\\\left({\frac {\partial u}{\partial x_{1}}}\right)^{3}+{\frac {\partial u}{\partial x_{2}}}+u^{4}&=0~~~~{\text{is a first-order PDE.}}\end{aligned}}$

Linear and nonlinear PDEs

A linear PDE is one that is of first degree in all of its field variables and partial derivatives. For example,

${\begin{aligned}{\frac {\partial u}{\partial x_{1}}}+{\frac {\partial u}{\partial x_{2}}}&=0~~~{\text{is linear}}~.\\{\frac {\partial u}{\partial x_{1}}}+\left({\frac {\partial u}{\partial x_{2}}}\right)^{2}&=0~~~{\text{is nonlinear}}~.\\{\frac {\partial u}{\partial x_{1}}}+{\frac {\partial u}{\partial x_{2}}}+u^{2}&=0~~~{\text{is nonlinear}}~.\\{\frac {\partial ^{2}u}{\partial x_{1}^{2}}}+{\frac {\partial ^{2}u}{\partial x_{2}^{2}}}&=x_{1}~~~{\text{is linear}}~.\\{\frac {\partial ^{2}u}{\partial x_{1}^{2}}}+u{\frac {\partial ^{2}u}{\partial x_{2}^{2}}}&=0~~~{\text{is quasilinear}}~.\end{aligned}}$

The above equations can also be written in operator notation as

${\begin{aligned}D(u)=0&~~{\text{where}}~~D:={\frac {\partial }{\partial x_{1}}}+{\frac {\partial }{\partial x_{2}}}~.\\D(u)=0&~~{\text{where}}~~D:={\frac {\partial }{\partial x_{1}}}+\left({\frac {\partial }{\partial x_{2}}}\right)^{2}~.\\D(u)=0&~~{\text{where}}~~D:={\frac {\partial }{\partial x_{1}}}+{\frac {\partial }{\partial x_{2}}}+u^{2}~.\\D(u)=x_{1}&~~{\text{where}}~~D:={\frac {\partial ^{2}}{\partial x_{1}^{2}}}+{\frac {\partial ^{2}}{\partial x_{2}^{2}}}~.\\D(u)=0&~~{\text{where}}~~D:={\frac {\partial ^{2}}{\partial x_{1}^{2}}}+u{\frac {\partial ^{2}}{\partial x_{2}^{2}}}~.\end{aligned}}$

Homogeneous PDEs

Let $L$ be a linear operator. Then a linear partial differential equation can be written in the form

$L(u)=f(x_{1},x_{2},x_{3},t)~.$

If $f(x_{1},x_{2},x_{3},t)=0$, the PDE is called homogeneous. For example,

${\begin{aligned}{\frac {\partial u}{\partial t}}+{\frac {\partial u}{\partial x_{1}}}+{\frac {\partial u}{\partial x_{2}}}+{\frac {\partial u}{\partial x_{3}}}&=0~~~{\text{is homogeneous}}~.\\{\frac {\partial u}{\partial t}}+{\frac {\partial u}{\partial x_{1}}}+{\frac {\partial u}{\partial x_{2}}}+{\frac {\partial u}{\partial x_{3}}}&=x_{1}+x_{2}~~~{\text{is nonhomogeneous}}~.\\\end{aligned}}$

Elliptic, Hyperbolic, and Parabolic PDEs

We usually come across three-types of second-order PDEs in mechanics. These are classified as elliptic, hyperbolic, and parabolic.

The equations of elasticity (without inertial terms) are elliptic PDEs. Hyperbolic PDEs describe wave propagation phenomena. The heat conduction equation is an example of a parabolic PDE.

Each type of PDE has certain characteristics that help determine if a particular finite element approach is appropriate to the problem being described by the PDE. Interestingly, just knowing the type of PDE can give us insight into how smooth the solution is, how fast information propagates, and the effect of initial and boundary conditions.

• In hyperbolic PDEs, the smoothness of the solution depends on the smoothness of the initial and boundary conditions. For instance, if there is a jump in the data at the start or at the boundaries, then the jump will propagate as a discontinuity in the solution. If, in addition, the PDE is nonlinear, then shocks may develop even though the initial conditions and the boundary conditions are smooth. In a system modeled with a hyperbolic PDE, information travels at a finite speed referred to as the wavespeed. Information is not transmitted until the wave arrives.
• In contrast, the solutions of elliptic PDEs are always smooth, even if the initial and boundary conditions are rough (though there may be singularities at sharp corners). In addition, boundary data at any point affect the solution at all points in the domain.
• Parabolic PDEs are usually time dependent and represent the diffusion-like processes. Solutions are smooth in space but may possess singularities. However, information travels at infinite speed in a parabolic system.

Suppose we have a second-order PDE of the form

$a(x_{1},x_{2}){\frac {\partial ^{2}u}{\partial x_{1}^{2}}}+b(x_{1},x_{2}){\frac {\partial ^{2}u}{\partial x_{1}\partial x_{2}}}+c(x_{1},x_{2}){\frac {\partial ^{2}u}{\partial x_{2}^{2}}}+d(x_{1},x_{2}){\frac {\partial u}{\partial x_{1}}}+e(x_{1},x_{2}){\frac {\partial u}{\partial x_{2}}}+f(x_{1},x_{2})u=g(x_{1},x_{2})$

Then, the PDE is called elliptic if

${b^{2}-4ac<0}$

An example is

${\frac {\partial ^{2}u}{\partial x_{1}^{2}}}+{\frac {\partial ^{2}u}{\partial x_{1}\partial x_{2}}}+{\frac {\partial ^{2}u}{\partial x_{2}^{2}}}=x_{1}{\frac {\partial u}{\partial x_{1}}}$

The PDE is called hyperbolic if

${b^{2}-4ac>0}$

An example is

${\frac {\partial ^{2}u}{\partial x_{1}^{2}}}+3{\frac {\partial ^{2}u}{\partial x_{1}\partial x_{2}}}+{\frac {\partial ^{2}u}{\partial x_{2}^{2}}}=x_{1}{\frac {\partial u}{\partial x_{1}}}$

The PDE is called parabolic if

${b^{2}-4ac=0}$

An example is

${\frac {\partial ^{2}u}{\partial x_{1}^{2}}}+2{\frac {\partial ^{2}u}{\partial x_{1}\partial x_{2}}}+{\frac {\partial ^{2}u}{\partial x_{2}^{2}}}=x_{1}{\frac {\partial u}{\partial x_{1}}}$

Solutions to Common PDEs

Partial differential equation appear in several areas of physics and engineering. A firm grasp of how to solve ordinary differential equations is required to solve PDEs. In particular, solutions to the Sturm-Liouville problems should be familiar to anyone attempting to solve PDEs.

• A tutorial on how to solve the Laplace equation
• A tutorial on how to solve the Poisson equation
• A tutorial on how to apply the method of separation of variables

Application of PDEs in Physics and Engineering

There are many applications of partial differential equations in physics and engineering. Here are some examples:

• The diffusion equation
  • The heat equation
• The wave equation
• Maxwell's equations

Resources

• w:Numerical partial differential equations
• MIT 18.03 Differential Equations, Spring 2006

The Heat conduction equation of 2-D is elliptic in space and parabolic in time.