maths.freeFluid Dynamics › The stress tensor and Cauchy's momentum equation

The stress tensor and Cauchy's momentum equation

Surface forces as a linear map from normals to tractions, the momentum balance, and why the stress is symmetric.

Two kinds of force act on a piece of fluid. Body forces act on every particle, like gravity \(\rho\mathbf{g}\) per unit volume. Surface forces are exerted by the neighbouring fluid across the boundary: on a small surface patch with unit normal \(\mathbf{n}\) the fluid on the side \(\mathbf{n}\) points to pushes with a force \(\mathbf{t}(\mathbf{n})\,dS\), where \(\mathbf{t}\) is the traction. A priori \(\mathbf{t}\) could depend on \(\mathbf{n}\) in any way.

Cauchy's stress theorem (Cauchy, 1823) says the dependence is linear: there is a matrix \(\sigma(\mathbf{x},t)\), the stress tensor, with \[ \mathbf{t}(\mathbf{n}) = \sigma\,\mathbf{n}, \qquad t_i = \sigma_{ij}n_j. \] The argument balances forces on a tiny tetrahedron with three faces on the coordinate planes. Surface forces scale like the area, \(h^2\), while inertia and body forces scale like the volume, \(h^3\); letting \(h\to0\) forces the tractions on the four faces to balance, which is exactly linearity. The entry \(\sigma_{ij}\) is the \(i\)-component of force per area on a face whose normal points along \(j\).

Newton's second law for a material region \(V\) reads \(\frac{d}{dt}\int_V\rho\mathbf{u}\,dV = \int_V\rho\mathbf{f}\,dV + \oint_{\partial V}\sigma\mathbf{n}\,dS\). The divergence theorem and conservation of mass turn this into Cauchy's momentum equation \[ \rho\frac{D\mathbf{u}}{Dt} = \nabla\cdot\sigma + \rho\mathbf{f}, \qquad (\nabla\cdot\sigma)_i = \partial_j\sigma_{ij}. \] Balancing angular momentum in the same way (with no body torques) shows \(\sigma_{ij} = \sigma_{ji}\). A symmetric matrix has real eigenvalues and orthogonal eigenvectors, the principal stresses and principal axes: on planes normal to a principal axis the traction is purely normal.

In a fluid at rest there can be no shear stress, because a fluid yields to any shear. Every direction must then be a principal direction, which forces all eigenvalues equal: \(\sigma = -p\,I\), with \(p\) the pressure (the minus sign because pressure pushes inward). In a moving fluid we write \(\sigma = -pI + \tau\), where \(\tau\) is the viscous or deviatoric stress. Cauchy's equation has three components but \(\sigma\) has six independent entries, so it cannot be solved until a constitutive law says how \(\tau\) depends on the motion. Setting \(\tau = 0\) gives the Euler equations; a linear law in the rate of strain gives Navier-Stokes.

Picture it: cut the fluid with an imaginary knife in every direction through one point; the force per area on each cut is \(\sigma\) applied to the cut's normal, and the three principal cuts feel pure push or pull. Think it: \(\sigma\) is a symmetric linear map, and the mean normal stress \(-\tfrac13\operatorname{tr}\sigma\) is the part every direction agrees on: that is the mechanical definition of pressure.

مثال کار شده · eigenvalues of [[2,1,0],[1,2,0],[0,0,1]]

Eigenvalues of [[2,1,0],[1,2,0],[0,0,1]]

\left[\begin{matrix}2 & 1 & 0\\1 & 2 & 0\\0 & 0 & 1\end{matrix}\right]

قدم به قدم

  1. \det(A - \lambda I) = 0

    Eigenvalues are the roots of the characteristic polynomial.

  2. \det\left[\begin{matrix}2 - \lambda & 1 & 0\\1 & 2 - \lambda & 0\\0 & 0 & 1 - \lambda\end{matrix}\right] = 0

    Subtract λ from the diagonal.

  3. - \lambda^{3} + 5 \lambda^{2} - 7 \lambda + 3 = 0

    Expand the determinant.

  4. - \left(\lambda - 3\right) \left(\lambda - 1\right)^{2} = 0

    Factor.

  5. \lambda = 3, \lambda = 1\ (\times 2)

    Eigenvalues (with multiplicity).

  6. \lambda = 1:\ (A - \lambda I)\mathbf{v} = 0 \Rightarrow \mathbf{v} = \left[\begin{matrix}-1\\1\\0\end{matrix}\right]

    Solve (A − 1I)v = 0 for a basis eigenvector.

  7. \lambda = 1:\ (A - \lambda I)\mathbf{v} = 0 \Rightarrow \mathbf{v} = \left[\begin{matrix}0\\0\\1\end{matrix}\right]

    Solve (A − 1I)v = 0 for a basis eigenvector.

  8. \lambda = 3:\ (A - \lambda I)\mathbf{v} = 0 \Rightarrow \mathbf{v} = \left[\begin{matrix}1\\1\\0\end{matrix}\right]

    Solve (A − 3I)v = 0 for a basis eigenvector.

جواب رو نشون بده
\lambda = 3,\; \lambda = 1

حالا تو يه مساله انتخاب کن، يا خودت رو بنويس يا بکشي هر گام، يه تصوير، جواب مخفي ميشه تا وقتي که تو بپرسي

خودت رو به کار بسپار

یک حساب رایگان یادداشت‌هایی را در هر درس اضافه می‌کند، یک ثبت از آنچه که تمام کرده‌اید، مسائل حل شده شما در یک مکان، و یک معلم که می‌توانید در مورد این صفحه از او بپرسید. خود ریاضیات برای همه باز است، چه وارد شده باشید و چه نباشید.

ثبت نام ورود

نمادهای استفاده‌شده در اینجا

هر نماد را برای تعریف کامل، تصویر و معنی هر حرف در آن بزنید.

چطور: The stress tensor and Cauchy's momentum equation

  1. Write the stress as a symmetric 3 by 3 matrix.
  2. For the traction on a surface, multiply the matrix by the unit normal (normalise it first).
  3. The normal stress is the traction dotted with the normal; the rest of the traction is shear.
  4. Principal stresses are the eigenvalues; the pressure is minus one third of the trace.

سوالاتي که مردم ميپرسن

Why does the stress have to be symmetric?

If it were not, the torque on a tiny cube would scale like its volume times a finite amount while its moment of inertia scales like the fifth power of its size, so it would spin up infinitely fast as the cube shrinks.

Is pressure a vector?

No. Pressure is a scalar; the force it produces on a surface is -p times the normal, so the direction comes from the surface, not from the pressure.

What do I need before starting fluid dynamics?

Multivariable calculus (divergence, curl, the divergence and Stokes theorems), linear algebra (symmetric matrices and eigenvalues) and partial differential equations (the heat and Laplace equations). The last lessons also use Sobolev spaces, which are introduced where they are needed.

What are the Navier-Stokes equations in one sentence?

Newton's second law for each particle of a viscous incompressible fluid: acceleration equals the pressure force plus viscous diffusion of momentum, with the constraint that the velocity field is divergence-free.

Why is two-dimensional flow easier than three-dimensional flow?

In 2D the vorticity is a scalar that is only carried and diffused, so its maximum never grows; in 3D vortex lines can be stretched, which amplifies vorticity, and no known bound rules out unlimited growth.

Is this course physics or mathematics?

Both, in order. The first half derives the equations from physical principles and solves classical flows; the second half treats the equations as mathematical objects and studies which of their properties can be proved.

بیشتر در Fluid Dynamics