Reynolds transport theorem

{{Short description|3D generalization of the Leibniz integral rule}}

{{Calculus|expanded=differential}}In differential calculus, the Reynolds transport theorem (also known as the Leibniz–Reynolds transport theorem), or simply the Reynolds theorem, named after Osborne Reynolds (1842–1912), is a three-dimensional generalization of the Leibniz integral rule. It is used to recast time derivatives of integrated quantities and is useful in formulating the basic equations of continuum mechanics.

Consider integrating {{math|1=f = f(x,t)}} over the time-dependent region {{math|Ω(t)}} that has boundary {{math|∂Ω(t)}}, then taking the derivative with respect to time:

\frac{d}{dt}\int_{\Omega(t)} \mathbf{f}\,dV.

If we wish to move the derivative into the integral, there are two issues: the time dependence of {{math|f}}, and the introduction of and removal of space from {{math|Ω}} due to its dynamic boundary. Reynolds transport theorem provides the necessary framework.

General form

Reynolds transport theorem can be expressed as follows:{{cite book |authorlink=L. Gary Leal |first=L. G. |last=Leal |year=2007 |title=Advanced transport phenomena: fluid mechanics and convective transport processes |publisher=Cambridge University Press |isbn=978-0-521-84910-4 |page=23 }}{{cite book |authorlink=Osborne Reynolds |first=O. |last=Reynolds |year=1903 |title=Papers on Mechanical and Physical Subjects |volume=3, The Sub-Mechanics of the Universe |publisher=Cambridge University Press |location=Cambridge |pages=12–13 }}{{cite book |authorlink=Jerrold E. Marsden |first=J. E. |last=Marsden |authorlink2=Anthony Tromba |first2=A. |last2=Tromba |year=2003 |title=Vector Calculus |edition=5th |publisher=W. H. Freeman |location=New York |isbn=978-0-7167-4992-9 }}

\frac{d}{dt}\int_{\Omega(t)} \mathbf{f}\,dV = \int_{\Omega(t)} \frac{\partial \mathbf{f}}{\partial t}\,dV + \int_{\partial \Omega(t)} \left(\mathbf{v}_b\cdot\mathbf{n}\right)\mathbf{f}\,dA

in which {{math|n(x,t)}} is the outward-pointing unit normal vector, {{math|x}} is a point in the region and is the variable of integration, {{math|dV}} and {{math|dA}} are volume and surface elements at {{math|x}}, and {{math|vb(x,t)}} is the velocity of the area element (not the flow velocity). The function {{math|f}} may be tensor-, vector- or scalar-valued.{{cite book |first=H. |last=Yamaguchi |title=Engineering Fluid Mechanics |location=Dordrecht |publisher=Springer |year=2008 |page=23 |isbn=978-1-4020-6741-9 }} Note that the integral on the left hand side is a function solely of time, and so the total derivative has been used.

Form for a material element

In continuum mechanics, this theorem is often used for material elements. These are parcels of fluids or solids which no material enters or leaves. If {{math|Ω(t)}} is a material element then there is a velocity function {{math|1=v = v(x,t)}}, and the boundary elements obey

\mathbf{v}_b\cdot\mathbf{n}=\mathbf{v}\cdot\mathbf{n}.

This condition may be substituted to obtain:{{cite book |authorlink=Ted Belytschko |first=T. |last=Belytschko |first2=W. K. |last2=Liu |first3=B. |last3=Moran |year=2000 |title=Nonlinear Finite Elements for Continua and Structures |publisher=John Wiley and Sons |location=New York |isbn=0-471-98773-5 }}

\frac{d}{dt}\left(\int_{\Omega(t)} \mathbf{f}\,dV\right) = \int_{\Omega(t)} \frac{\partial \mathbf{f}}{\partial t}\,dV + \int_{\partial \Omega(t)} (\mathbf{v}\cdot\mathbf{n})\mathbf{f}\,dA.

{{math proof| title = Proof for a material element | proof =

Let {{math|Ω0}} be reference configuration of the region {{math|Ω(t)}}. Let

the motion and the deformation gradient be given by

\mathbf{x} = \boldsymbol{\varphi}(\mathbf{X}, t),

\boldsymbol{F}(\mathbf{X},t) = \boldsymbol{\nabla}\boldsymbol{\varphi}.

Let {{math|1=J(X,t) = det F(X,t)}}. Define

\hat{\mathbf{f}}(\mathbf{X}, t) = \mathbf{f}(\boldsymbol{\varphi}(\mathbf{X}, t), t).

Then the integrals in the current and the reference configurations are related by

\int_{\Omega(t)} \mathbf{f}(\mathbf{x},t)\,dV

= \int_{\Omega_0} \mathbf{f}(\boldsymbol{\varphi}(\mathbf{X},t),t) \, J(\mathbf{X},t) \,dV_0

= \int_{\Omega_0} \hat{\mathbf{f}}(\mathbf{X},t) \, J(\mathbf{X},t) \, dV_0.

That this derivation is for a material element is implicit in the time constancy of the reference configuration: it is constant in material coordinates. The time derivative of an integral over a volume is defined as

\frac{d}{dt}\left( \int_{\Omega(t)} \mathbf{f}(\mathbf{x},t)\,dV\right) = \lim_{\Delta t \to 0} \frac{1}{\Delta t} \left(\int_{\Omega(t + \Delta t)} \mathbf{f}(\mathbf{x},t+\Delta t)\,dV - \int_{\Omega(t)} \mathbf{f}(\mathbf{x},t)\,dV\right).

Converting into integrals over the reference configuration, we get

\frac{d}{dt} \left( \int_{\Omega(t)} \mathbf{f}(\mathbf{x},t) \, dV\right) =

\lim_{\Delta t \to 0} \frac{1}{\Delta t} \left(\int_{\Omega_0} \hat{\mathbf{f}}(\mathbf{X},t+\Delta t)\, J(\mathbf{X},t+\Delta t)\,dV_0 - \int_{\Omega_0} \hat{\mathbf{f}}(\mathbf{X},t)\, J(\mathbf{X},t)\, dV_0\right).

Since {{math|Ω0}} is independent of time, we have

\begin{align}

\frac{d}{dt}\left( \int_{\Omega(t)} \mathbf{f}(\mathbf{x},t)\,dV\right)

&= \int_{\Omega_0} \left(\lim_{\Delta t \to 0} \frac{ \hat{\mathbf{f}}(\mathbf{X},t+\Delta t)\, J(\mathbf{X},t+\Delta t) - \hat{\mathbf{f}}(\mathbf{X},t)\, J(\mathbf{X},t)}{\Delta t} \right)\,dV_0 \\

&= \int_{\Omega_0} \frac{\partial}{\partial t}\left(\hat{\mathbf{f}}(\mathbf{X},t)\, J(\mathbf{X},t)\right)\,dV_0 \\

&= \int_{\Omega_0} \left( \frac{\partial}{\partial t}\big(\hat{\mathbf{f}}(\mathbf{X},t)\big)\, J(\mathbf{X},t)+ \hat{\mathbf{f}}(\mathbf{X},t)\,\frac{\partial}{\partial t}\big(J(\mathbf{X},t)\big)\right) \,dV_0.

\end{align}

The time derivative of {{mvar|J}} is given by:{{cite book |authorlink=Morton Gurtin |last=Gurtin |first=M. E. |year=1981 |title=An Introduction to Continuum Mechanics |publisher=Academic Press |location=New York |page=77 |isbn=0-12-309750-9 }}

\begin{align}

\frac{\partial J(\mathbf{X},t)}{\partial t} &= \frac{\partial}{\partial t}(\det\boldsymbol{F}) \\

&= (\det\boldsymbol{F}) \operatorname{tr}\left(\boldsymbol{F}^{-1} \frac{\partial\boldsymbol F}{\partial t}\right)\\

&= (\det\boldsymbol{F}) \operatorname{tr}\left(\frac{\partial \boldsymbol{X}}{\partial \boldsymbol{\varphi}} \frac{\partial}{\partial t}\left(\frac{\partial\boldsymbol{\varphi}}{\partial\boldsymbol{X}}\right)\right)\\

&= (\det\boldsymbol{F}) \operatorname{tr}\left(\frac{\partial \boldsymbol{X}}{\partial \boldsymbol{\varphi}} \frac{\partial}{\partial \boldsymbol{X}}\left(\frac{\partial\boldsymbol{\varphi}}{\partial t}\right)\right) \\

&= (\det\boldsymbol{F}) \operatorname{tr}\left( \frac{\partial}{\partial \boldsymbol{x}}\left(\frac{\partial\boldsymbol{\varphi}}{\partial t}\right)\right) \\

&= (\det\boldsymbol{F})(\boldsymbol{\nabla} \cdot \mathbf{v}) \\

&= J(\mathbf{X},t)\,\boldsymbol{\nabla} \cdot \mathbf{v}\big(\boldsymbol{\varphi}(\mathbf{X},t),t\big) \\

&= J(\mathbf{X},t)\,\boldsymbol{\nabla} \cdot \mathbf{v}(\mathbf{x},t). \end{align}

Therefore,

\begin{align}

\frac{d}{dt}\left( \int_{\Omega(t)} \mathbf{f}(\mathbf{x},t)\,dV\right)

&= \int_{\Omega_0} \left( \frac{\partial}{\partial t}\left(\hat{\mathbf{f}}(\mathbf{X},t)\right)\,J(\mathbf{X},t)+ \hat{\mathbf{f}}(\mathbf{X},t)\,J(\mathbf{X},t)\,\boldsymbol{\nabla} \cdot \mathbf{v}(\mathbf{x},t)\right) \,dV_0 \\

&= \int_{\Omega_0} \left(\frac{\partial}{\partial t}\left(\hat{\mathbf{f}}(\mathbf{X},t)\right)+ \hat{\mathbf{f}}(\mathbf{X},t)\,\boldsymbol{\nabla} \cdot \mathbf{v}(\mathbf{x},t)\right)\,J(\mathbf{X},t) \,dV_0 \\

&= \int_{\Omega(t)} \left(\dot{\mathbf{f}}(\mathbf{x},t)+ \mathbf{f}(\mathbf{x},t)\,\boldsymbol{\nabla} \cdot \mathbf{v}(\mathbf{x},t)\right)\,dV.

\end{align}

where \dot{\mathbf{f}} is the material time derivative of {{math|f}}. The material derivative is given by

\dot{\mathbf{f}}(\mathbf{x},t) = \frac{\partial \mathbf{f}(\mathbf{x},t)}{\partial t} + \big(\boldsymbol{\nabla} \mathbf{f}(\mathbf{x},t)\big)\cdot\mathbf{v}(\mathbf{x},t).

Therefore,

\frac{d}{dt}\left( \int_{\Omega(t)} \mathbf{f}(\mathbf{x},t)\,dV\right) =

\int_{\Omega(t)} \left( \frac{\partial \mathbf{f}(\mathbf{x},t)}{\partial t} + \big(\boldsymbol{\nabla} \mathbf{f}(\mathbf{x},t)\big) \cdot\mathbf{v}(\mathbf{x},t) + \mathbf{f}(\mathbf{x},t)\,\boldsymbol{\nabla} \cdot \mathbf{v}(\mathbf{x},t) \right) \,dV,

or,

\frac{d}{dt}\left( \int_{\Omega(t)} \mathbf{f}\,dV\right)

= \int_{\Omega(t)} \left( \frac{\partial \mathbf{f}}{\partial t} + \boldsymbol{\nabla} \mathbf{f}\cdot\mathbf{v} + \mathbf{f}\,\boldsymbol{\nabla} \cdot \mathbf{v}\right)\,dV.

Using the identity

\boldsymbol{\nabla} \cdot (\mathbf{v}\otimes\mathbf{w}) = \mathbf{v}(\boldsymbol{\nabla} \cdot \mathbf{w}) + \boldsymbol{\nabla}\mathbf{v}\cdot\mathbf{w},

we then have

\frac{d}{dt}\left( \int_{\Omega(t)} \mathbf{f}\,dV\right)

= \int_{\Omega(t)} \left(\frac{\partial \mathbf{f}}{\partial t} + \boldsymbol{\nabla} \cdot (\mathbf{f}\otimes\mathbf{v})\right)\,dV.

Using the divergence theorem and the identity {{math|1=(ab) · n = (b · n)a}}, we have

\begin{align}

\frac{d}{dt}\left( \int_{\Omega(t)} \mathbf{f}\,dV\right)

&= \int_{\Omega(t)}\frac{\partial \mathbf{f}}{\partial t}\,dV + \int_{\partial \Omega(t)}(\mathbf{f}\otimes\mathbf{v})\cdot\mathbf{n}\,dA \\

&= \int_{\Omega(t)}\frac{\partial \mathbf{f}}{\partial t}\,dV + \int_{\partial \Omega(t)}(\mathbf{v}\cdot\mathbf{n})\mathbf{f}\,dA.

\end{align}

Q.E.D.

}}

A special case

If we take {{math|Ω}} to be constant with respect to time, then {{math|vb {{=}} 0}} and the identity reduces to

\frac{d}{dt}\int_{\Omega} f\,dV = \int_{\Omega} \frac{\partial f}{\partial t}\,dV.

as expected. (This simplification is not possible if the flow velocity is incorrectly used in place of the velocity of an area element.)

= Interpretation and reduction to one dimension =

The theorem is the higher-dimensional extension of differentiation under the integral sign and reduces to that expression in some cases. Suppose {{mvar|f}} is independent of {{mvar|y}} and {{mvar|z}}, and that {{math|Ω(t)}} is a unit square in the {{mvar|yz}}-plane and has {{mvar|x}} limits {{math|a(t)}} and {{math|b(t)}}. Then Reynolds transport theorem reduces to

\frac{d}{dt}\int_{a(t)}^{b(t)} f(x,t)\,dx = \int_{a(t)}^{b(t)} \frac{\partial f}{\partial t}\,dx + \frac{\partial b(t)}{\partial t} f\big(b(t),t\big) - \frac{\partial a(t)}{\partial t} f\big(a(t),t\big) \,,

which, up to swapping {{mvar|x}} and {{mvar|t}}, is the standard expression for differentiation under the integral sign.

See also

{{Portal|Mathematics}}

  • {{annotated link|Leibniz integral rule}}

{{clear}}

References

{{Reflist}}