Skip to main content

Analysis of Enhanced Diffusion in Taylor Dispersion via a Model Problem

  • Chapter
Hamiltonian Partial Differential Equations and Applications

Part of the book series: Fields Institute Communications ((FIC,volume 75))

Abstract

We consider a simple model of the evolution of the concentration of a tracer, subject to a background shear flow by a fluid with viscosity ν ≪ 1 in an infinite channel. Taylor observed in the 1950s that, in such a setting, the tracer diffuses at a rate proportional to 1∕ν, rather than the expected rate proportional to ν. We provide a mathematical explanation for this enhanced diffusion using a combination of Fourier analysis and center manifold theory. More precisely, we show that, while the high modes of the concentration decay exponentially, the low modes decay algebraically, but at an enhanced rate. Moreover, the behavior of the low modes is governed by finite-dimensional dynamics on an appropriate center manifold, which corresponds exactly to diffusion by a fluid with viscosity proportional to 1∕ν.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 39.99
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD 54.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info
Hardcover Book
USD 54.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

References

  1. Aris, R.: On the dispersion of a solute in a fluid flowing through a tube. Proc. Roy. Soc. Lond. Ser. A 235(1200), 67–77 (1956)

    Article  Google Scholar 

  2. Beck, M., Eugene Wayne, C.: Metastability and rapid convergence to quasi-stationary bar states for the two-dimensional Navier-Stokes equations. Proc. Roy. Soc. Edinb. Sect. A Math. 143, 905–927, 10 (2013)

    Google Scholar 

  3. Chatwin, P.C., C.M. Allen, C.M.: Mathematical models of dispersion in rivers and estuaries. Ann. Rev. Fluid Mech. 17, 119–149 (1985)

    Google Scholar 

  4. Chen, X.-Y., Hale, J. K., Tan, B.: Invariant foliations for C 1 semigroups in Banach spaces. J. Differ. Equ. 139(2), 283–318 (1997)

    Article  MATH  MathSciNet  Google Scholar 

  5. Eugene Wayne, C.: Invariant manifolds for parabolic partial differential equations on unbounded domains. Arch. Ration. Mech. Anal. 138(3), 279–306 (1997)

    Article  MATH  Google Scholar 

  6. Gallay, T., Eugene Wayne, C.: Invariant manifolds and the long-time asymptotics of the Navier-Stokes and vorticity equations on \(\mathbb{R}^{2}\). Arch. Ration. Mech. Anal. 163(3), 209–258 (2002)

    Article  MATH  MathSciNet  Google Scholar 

  7. Latini, M., Bernoff, A.J.: Transient anomalous diffusion in Poiseuille flow. J. Fluid Mech. 441(8), 399–411, (2001)

    MATH  Google Scholar 

  8. Mercer, G.N., Roberts, A.J.: A centre manifold description of contaminant dispersion in channels with varying flow properties. SIAM J. Appl. Math. 50(6), 1547–1565 (1990)

    Article  MATH  MathSciNet  Google Scholar 

  9. Taylor, G.: Dispersion of soluble matter in solvent flowing slowly through a tube. Proc. Roy. Soc. Lond. Ser. A 219(1137), 186–203 (1953)

    Article  Google Scholar 

  10. Taylor, G.: Dispersion of matter in turbulent flow through a tube. Proc. Roy. Soc. Lond. Ser. A 223(1155), 446–468 (1954)

    Article  Google Scholar 

Download references

Acknowledgements

The work of OC and CEW was supported in part by the NSF through grant DMS-1311553. The work of MB was supported in part by a Sloan Fellowship and NSF grant DMS-1411460. MB and CEW thank Tasso Kaper and Edgar Knobloch for pointing out a possible connection between their prior work in [2] and the phenomenon of Taylor dispersion, and we all gratefully acknowledge the many insightful and extremely helpful comments of the anonymous referee.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to C. Eugene Wayne .

Editor information

Editors and Affiliations

Additional information

Dedicated to Walter Craig, with admiration and affection, on his 60th birthday.

Appendix: Convergence to the Center Manifold

Appendix: Convergence to the Center Manifold

The purpose of this appendix is to show that the center manifold constructed in Sect. 2 attracts all solutions. We know that we can construct the invariant manifold for the Eq. (21) (for a k ′and b k ′)globally since we have explicit formulas which hold for all values of a k andη. In this note we show that any trajectory will converge toward the center manifold on a time scale of \(\mathcal{O}(1/\nu )\).

Write \(b_{k} = B_{k} + h_{k}(a_{k-2},\ldots,a_{0},\eta )\). We’ll prove that for any choice of initial conditions B k goes to zero like \(\sim e^{-\nu t}\).

First note that

$$\displaystyle\begin{array}{rcl} b_{k}'& =& B_{k}' +\sum _{ \ell=0,even}^{k-2}(\partial _{ a_{\ell}}h_{k})a_{\ell}' + (\partial _{\eta }h_{k})\eta ' {}\\ & =& B_{k}' -\sum _{\ell=2,even}^{k-2}(\partial _{ a_{\ell}}h_{k})\eta \frac{\ell} {2}a_{\ell} -\sum _{\ell=2,even}^{k-2}(\partial _{ a_{\ell}}h_{k})\eta b_{\ell-2} - (\partial _{\eta }h_{k})\eta ^{2} {}\\ & =& B_{k}' -\eta \sum _{\ell=2,even}^{k-2}(\partial _{ a_{\ell}}h_{k})B_{\ell-2} +\sum _{ \ell=2,even}^{k-2}(\partial _{ a_{\ell}}h_{k})(-\eta \frac{\ell} {2}a_{\ell} -\eta h_{\ell-2}) - (\partial _{\eta }h_{k})\eta ^{2}. {}\\ \end{array}$$

Thus, using the equation for b k ′ in (21) we have

$$\displaystyle\begin{array}{rcl} & & B_{k}' -\eta \sum _{\ell=2,even}^{k-2}(\partial _{ a_{\ell}}h_{k})B_{\ell-2} + (\nu +\eta \frac{k} {2})B_{k} + \frac{2\eta } {\nu } B_{k-2} {}\\ & & \quad = -(\nu +\eta \frac{k} {2})h_{k} + \frac{\eta } {\nu ^{2}}a_{k-2} {}\\ & & \qquad \ -\sum _{\ell=2,even}^{k-2}(\partial _{ a_{\ell}}h_{k})(-\eta \frac{\ell} {2}a_{\ell} -\eta h_{\ell-2}) - (\partial _{\eta }h_{k})\eta ^{2} -\frac{2\eta } {\nu } h_{k-2}. {}\\ \end{array}$$

The key observation is that the terms on the right hand side precisely represent the invariance equation that defines h k (and hence they all cancel), leaving the equation

$$\displaystyle\begin{array}{rcl} B_{k}'\quad = -(\nu +\eta \frac{k} {2})B_{k} -\frac{2\eta } {\nu } B_{k-2} +\eta \sum _{ \ell=2,even}^{k-2}(\partial _{ a_{\ell}}h_{k})B_{\ell-2}.& &{}\end{array}$$
(51)

We now see that the system of equations for B k is homogeneous, linear, upper-triangular (but non-autonomous), and hence can be analyzed inductively. We’ll show that

$$\displaystyle\begin{array}{rcl} \vert B_{k}(t)\vert < \frac{C(N)} {\nu ^{k/2}} e^{-\nu t}& & {}\\ \end{array}$$

for \(t > \frac{1} {\nu }\). We’ll only prove this for the even-indexed subsystem; the proof for the odd-subsystem is analogous. Notice that the base case, k = 0, holds since (51) for k = 0 reads B 0′ = −ν B 0, which implies \(B_{0} \sim e^{-\nu t}\). Now let’s proceed with the induction argument: assume for j = 0, 2, … k, that

$$\displaystyle\begin{array}{rcl} \vert B_{j}(t)\vert < \frac{C(N)} {\nu ^{j/2}} e^{-\nu t}.& & {}\\ \end{array}$$

Next, we write the equation for B k+2 from (51) with a key difference in the way the last term is written:

$$\displaystyle\begin{array}{rcl} B_{k+2}' = -\left (\nu B_{k+2} +\eta \frac{k + 2} {2} B_{k+2}\right ) -\frac{2\eta } {\nu } B_{k} + \frac{\eta } {\nu }\sum _{\ell=1}^{\frac{k} {2} }C_{k+2-2\ell}^{k+2} \frac{\eta ^{\ell}} {\nu ^{2\ell}}B_{k-2\ell}.& &{}\end{array}$$
(52)

This last sum is obtained from reindexing and using the fact that the formula for h k in Proposition 1 implies that \((\partial _{a_{\ell}}h_{k})\) consists of a single term. Let’s proceed by noting that the equation

$$\displaystyle\begin{array}{rcl} y' = -\left (\nu y + a\eta y\right )& & {}\\ \end{array}$$

has the exact solution

$$\displaystyle\begin{array}{rcl} y = e^{-\nu (t-t_{0})}(1 + t)^{-a}(1 + t_{ 0})^{a}.& & {}\\ \end{array}$$

To derive this solution, it may help to recall that \(\eta = \frac{1} {1+t}\). Applying this to (52) with \(a = \frac{k+2} {2}\) and using Duhamel’s formula, we obtain

$$\displaystyle\begin{array}{rcl} B_{k+2}(t)& =& e^{-\nu (t-t_{0})}(1 + t)^{-\frac{k+2} {2} }(1 + t_{0})^{\frac{k+2} {2} }B_{k+2}(0) + D_{k}^{k+2}(t) \\ & & +\sum _{\ell=1}^{\frac{k+2} {2} -1}D_{k-2\ell}^{k+2}(t) {}\end{array}$$
(53)

where the Duhamel terms D k−2 k+2 satisfy

$$\displaystyle\begin{array}{rcl} D_{k-2\ell}^{k+2}(t)& & \\ & \sim & \frac{C} {\nu ^{2\ell+1}}\int _{t_{0}}^{t}e^{-\nu (t-s)}(1 + t)^{-\frac{k+2} {2} }(1 + s)^{\frac{k+2} {2} }(1 + s)^{-\ell-1} \frac{1} {\nu ^{\frac{k-2\ell} {2} }} e^{-\nu s}ds.{}\end{array}$$
(54)

Notice in the above Duhamel term, we have substituted, using the induction hypothesis \(\vert B_{k-2\ell}(t)\vert \leq \frac{C} {\nu ^{\frac{k-2\ell} {2} }} e^{-\nu t}\) (we also assume \(t > t_{0} > \frac{1} {\nu }\)). These Duhamel terms are the most slowly decaying terms in the solution formula (53). Proceeding, we simplify (54) and obtain (for all ),

$$\displaystyle\begin{array}{rcl} D_{k-2\ell}^{k+2}(t)& \sim & \frac{C} {\nu ^{\frac{k} {2} +\ell+1}}e^{-\nu t}(1 + t)^{-\left (\frac{k+2} {2} \right )}\left ((1 + t)^{\frac{k} {2} -\ell+1} - (1 + t_{0})^{\frac{k} {2} -\ell+1}\right ) {}\\ & =& \frac{C} {\nu ^{\frac{k} {2} +1}}e^{-\nu t}\left ( \frac{1} {\nu ^{\ell}(1 + t)^{\ell}} - \frac{1} {\nu ^{\ell}(1 + t_{0})^{\ell}} \frac{(1 + t_{0})^{\frac{k} {2} +1}} {(1 + t)^{\frac{k} {2} +1}} \right ). {}\\ \end{array}$$

Now since \(t > t_{0} > \frac{1} {\nu }\), we obtain

$$\displaystyle\begin{array}{rcl} \vert D_{k-2\ell}^{k+2}(t)\vert \leq \frac{C} {\nu ^{\frac{k+2} {2} }} e^{-\nu t},& & {}\\ \end{array}$$

and subsequently we obtain, for \(t > t_{0} > \frac{1} {\nu }\),

$$\displaystyle\begin{array}{rcl} \vert B_{k+2}(t)\vert \leq \frac{C} {\nu ^{\frac{k+2} {2} }} e^{-\nu t}& & {}\\ \end{array}$$

as desired. □ 

Rights and permissions

Reprints and permissions

Copyright information

© 2015 Springer Science+Business Media New York

About this chapter

Cite this chapter

Beck, M., Chaudhary, O., Eugene Wayne, C. (2015). Analysis of Enhanced Diffusion in Taylor Dispersion via a Model Problem. In: Guyenne, P., Nicholls, D., Sulem, C. (eds) Hamiltonian Partial Differential Equations and Applications. Fields Institute Communications, vol 75. Springer, New York, NY. https://doi.org/10.1007/978-1-4939-2950-4_2

Download citation

Publish with us

Policies and ethics