Nondimensionalizing Reaction-Diffusion Systems

Why Nondimensionalize a Reaction-Diffusion Model?

A reaction-diffusion equation combines local chemical change with transport through space. Its parameters may have different units and magnitudes, making it difficult to see directly whether a predicted spatial pattern is controlled mainly by reaction or by diffusion. Nondimensionalization rescales the variables so that the equation is expressed using dimensionless quantities. In doing so, it exposes ratios of competing processes and clarifies which parameters govern the behavior over a specified spatial scale.

The key ratio for a first-order reaction is the Damköhler number. It compares the time required for diffusion across a chosen length with the time required for reaction. It is not an intrinsic property of a molecule alone: its value depends on the reaction rate, the diffusion coefficient, and the length scale relevant to the question.

Definition: For first-order reaction with rate constant $k$, diffusion coefficient $D$, and characteristic length $L$, the Damköhler number is $$ \mathrm{Da}=\frac{kL^2}{D}. $$ Equivalently, $\mathrm{Da}$ is the diffusion time $L^2/D$ divided by the reaction time $1/k$. A large value indicates that reaction is fast relative to diffusion across $L$; a small value indicates that diffusion is fast relative to reaction across $L$.

Scaling the First-Order Reaction-Diffusion Equation

Consider a single concentration $c(\mathbf{x},t)$ in a homogeneous, stationary medium. Assume that diffusion is isotropic with constant coefficient $D$, reaction is a spatially uniform first-order loss with rate constant $k$, and a uniform source supplies material at rate $q$. The model is

\frac{\partial c}{\partial t} = D\nabla^2 c-kc+q.

Here $c$ has units of concentration, $D$ has units of length squared per time, $k$ has units of inverse time, and $q$ has units of concentration per time. We choose a characteristic length $L$, a concentration scale $C_0$, and use the diffusion time $L^2/D$ as the time scale. Define dimensionless position $\boldsymbol{\xi}$, time $\tau$, and concentration $u$ by

\mathbf{x}=L\boldsymbol{\xi},\qquad t=\frac{L^2}{D}\tau,\qquad c=C_0u.

These definitions give $\partial c/\partial t=(C_0D/L^2)\partial u/\partial\tau$ and $\nabla^2 c=(C_0/L^2)\nabla_{\boldsymbol{\xi}}^2u$. Substituting into the dimensional equation and dividing by $C_0D/L^2$ yields

\frac{\partial u}{\partial \tau} = \nabla_{\boldsymbol{\xi}}^2u - \frac{kL^2}{D}u + \frac{qL^2}{DC_0} = \nabla_{\boldsymbol{\xi}}^2u-\mathrm{Da}\,u+Q, \qquad Q=\frac{qL^2}{DC_0}.

The diffusion coefficient is now one because the chosen time scale is the time for diffusion over $L$. The dimensionless source strength $Q$ compares source input over that diffusion time with the concentration scale $C_0$. The loss term is weighted by $\mathrm{Da}$, which compares the two processes of interest.

The same scaling must be applied to initial and boundary conditions. For example, an initial condition $c(\mathbf{x},0)=c_0(\mathbf{x})$ becomes $u(\boldsymbol{\xi},0)=c_0(L\boldsymbol{\xi})/C_0$. A fixed boundary concentration $c=C_b$ becomes $u=C_b/C_0$. A zero-flux boundary remains zero flux, since the dimensional normal derivative is a constant scale factor times the dimensionless normal derivative. Keeping these conditions consistent is necessary: the dimensionless equation alone does not specify a spatial solution.

Why the Damköhler Number Compares Rates

For diffusion across a distance $L$, the characteristic time is $t_D=L^2/D$. This follows from the diffusion term: a concentration varying over $L$ has a second spatial derivative of order $c/L^2$, so its diffusion-driven rate of change is of order $Dc/L^2$. Dividing the concentration by this rate gives $L^2/D$. For first-order loss, the rate is $kc$, so the characteristic time to change concentration by an amount of order $c$ is $t_R=1/k$. Their ratio is

\frac{t_D}{t_R} = \frac{L^2/D}{1/k} = \frac{kL^2}{D} = \mathrm{Da}.

Thus, $\mathrm{Da}\ll 1$ means the characteristic diffusion time is shorter than the reaction time: transport can redistribute material across the chosen distance before much first-order loss occurs. When $\mathrm{Da}\gg 1$, reaction can substantially consume material before diffusion transports it across that distance. When $\mathrm{Da}$ is of order one, neither process is safely negligible on that scale. These comparisons describe relative timescales, not guarantees about a particular concentration profile. Boundary forcing, initial conditions, sources, and geometry also matter.

Worked Example: A Tissue-Scale First-Order Sink

Suppose a hypothetical soluble mediator diffuses through tissue with $D=120\ \mu\mathrm{m}^2/\mathrm{min}$. It undergoes first-order loss with $k=0.030\ \mathrm{min}^{-1}$. For a tissue length scale $L=300\ \mu\mathrm{m}$, calculate the Damköhler number and compare the characteristic times.

Check the model quantities. The assumed first-order rate constant has units of inverse time, and $D$ has units of length squared per time. Therefore $kL^2/D$ is dimensionless.

Calculate the ratio.

\mathrm{Da} = \frac{(0.030\ \mathrm{min}^{-1})(300\ \mu\mathrm{m})^2} {120\ \mu\mathrm{m}^2/\mathrm{min}} = \frac{2700\ \mu\mathrm{m}^2/\mathrm{min}} {120\ \mu\mathrm{m}^2/\mathrm{min}} = 22.5.

The diffusion time is $t_D=L^2/D=90{,}000/120=750\ \mathrm{min}$, while the reaction time is $t_R=1/k=33.3\ \mathrm{min}$. Their ratio is $750/33.3\approx22.5$, agreeing with the dimensionless calculation. Under the stated assumptions, reaction is fast relative to diffusion across $300\ \mu\mathrm{m}$. This comparison alone does not specify the full tissue concentration profile.

Length Scale and Reaction Law Matter

The length $L$ must represent the distance over which transport and reaction are being compared: for example, a tissue thickness or the distance between a source and a region of interest. Because $\mathrm{Da}$ contains $L^2$, changing that choice can change the conclusion substantially. A value calculated over a whole organ should not automatically be used to describe transport across a thin local region.

For a general reaction term $R(c)$, with units of concentration per time, the dimensional equation is $\partial c/\partial t=D\nabla^2c+R(c)$. Under the same scales, it becomes

\frac{\partial u}{\partial\tau} = \nabla_{\boldsymbol{\xi}}^2u + \frac{L^2}{DC_0}R(C_0u).

If the reaction has a characteristic rate scale $R_0$, its dimensionless strength is $R_0L^2/(DC_0)$. For nonlinear kinetics, the dimensionless reaction term may also depend on additional ratios, such as a half-saturation concentration divided by $C_0$. In that setting, one should not imply that a single constant Damköhler number fully describes reaction at every concentration.

Worked Example: Changing the Length Scale

Consider a hypothetical first-order loss with $k=0.020\ \mathrm{min}^{-1}$ and diffusion coefficient $D=100\ \mu\mathrm{m}^2/\mathrm{min}$. Compare the Damköhler number for two possible transport distances, $50\ \mu\mathrm{m}$ and $100\ \mu\mathrm{m}$.

For $L=50\ \mu\mathrm{m}$,

\mathrm{Da}_{50} = \frac{(0.020)(50^2)}{100} = \frac{50}{100} = 0.50.

The corresponding times are $t_D=50^2/100=25\ \mathrm{min}$ and $t_R=1/0.020=50\ \mathrm{min}$. Diffusion is faster than reaction across this distance.

For $L=100\ \mu\mathrm{m}$,

\mathrm{Da}_{100} = \frac{(0.020)(100^2)}{100} = \frac{200}{100} = 2.00.

Now $t_D=100^2/100=100\ \mathrm{min}$, while $t_R$ remains $50\ \mathrm{min}$. Reaction is faster than diffusion across this longer distance. Doubling $L$ multiplies $\mathrm{Da}$ by four, since $\mathrm{Da}$ is proportional to $L^2$. The two results are not contradictory: they answer questions about different spatial scales.

Worked Example: Saturable Reaction and a Concentration Scale

A hypothetical mediator is removed by a saturable reaction, $R(c)=-V_{\max}c/(K_m+c)$, with $V_{\max}=0.40\ \mathrm{nM}/\mathrm{min}$ and $K_m=10\ \mathrm{nM}$. Let $D=80\ \mu\mathrm{m}^2/\mathrm{min}$, $L=200\ \mu\mathrm{m}$, and $C_0=20\ \mathrm{nM}$. Nondimensionalize the reaction term and compare its rate at $c=C_0$ with diffusion.

Set $c=C_0u$. The dimensionless reaction term is

\frac{L^2}{DC_0}R(C_0u) = -\frac{V_{\max}L^2}{DC_0} \frac{u}{K_m/C_0+u} = -\mathrm{Da}_{\max}\frac{u}{\kappa+u}, \qquad \kappa=\frac{K_m}{C_0}.

Here $\kappa=10/20=0.500$, and the maximum-rate scale gives

\mathrm{Da}_{\max} = \frac{(0.40\ \mathrm{nM}/\mathrm{min})(200\ \mu\mathrm{m})^2} {(80\ \mu\mathrm{m}^2/\mathrm{min})(20\ \mathrm{nM})} = \frac{16{,}000}{1{,}600} = 10.0.

At $u=1$, the dimensionless loss term has magnitude $10(1)/(0.500+1)=6.67$. The same comparison in physical times uses the local fractional loss rate $V_{\max}/(K_m+C_0)=0.40/30=0.0133\ \mathrm{min}^{-1}$. Its reciprocal is $75.0\ \mathrm{min}$, while $t_D=L^2/D=40{,}000/80=500\ \mathrm{min}$. Thus the local reaction-to-diffusion ratio at $C_0$ is $500/75.0=6.67$. The maximum-scale value $10.0$ and the value at $C_0$ differ because the reaction is saturable; the additional ratio $\kappa$ is needed to describe its concentration dependence.

Assumptions, Diagnostics, and Alternatives

The simple expression $\mathrm{Da}=kL^2/D$ applies to a uniform first-order rate constant and a single chosen length scale. The equation used above also assumes a continuum concentration field, constant diffusion, and no advection. If diffusion varies by location, reaction rates vary across tissue, or multiple species are coupled, the relevant rate comparisons may differ by region or reaction. A single reported Damköhler number can then conceal important variation; state which parameters and scales it uses.

A useful diagnostic is to recover the dimensionless equation directly from the dimensional one. Every term in the original equation must have units of concentration per time, and every term after scaling must be dimensionless. In particular, $kL^2/D$ has no units, whereas $kL/D$ does not. Also check the scaled initial and boundary conditions: a common source of error is to scale the differential equation but leave a boundary concentration in dimensional units.

Diffusion-time scaling is convenient when the main question is whether reaction competes with transport, because it sets the diffusion coefficient to one and leaves $\mathrm{Da}$ multiplying reaction. An alternative is to use the reaction time $1/k$ as the time scale. The resulting dimensionless equation has diffusion coefficient $D/(kL^2)=1/\mathrm{Da}$. These are equivalent choices: one puts the ratio on the reaction term, the other on the diffusion term. Neither changes the underlying model or the physical interpretation.

Nondimensionalization does not replace solving or validating a model. A small $\mathrm{Da}$ does not prove that concentration is spatially uniform, and a large $\mathrm{Da}$ does not, by itself, determine the size or shape of a gradient. Spatially varying inputs, boundary conditions, geometry, and initial conditions can all generate structure. The number identifies a relative timescale, not a complete prediction.

Common pitfall: It is incorrect to say that a reaction is “fast” or “slow” from $k$ alone, without specifying the transport distance and diffusion coefficient. The correction is to state $L$, $D$, and the rate law, then calculate the appropriate dimensionless ratio. For first-order loss, that ratio is $kL^2/D$; for nonlinear kinetics, state the concentration scale and any additional dimensionless parameters.

Check Your Understanding

Answer each question, then compare your reasoning with the solutions.

A hypothetical first-order sink has $k=0.010\ \mathrm{min}^{-1}$, $D=50\ \mu\mathrm{m}^2/\mathrm{min}$, and $L=100\ \mu\mathrm{m}$. Calculate $\mathrm{Da}$ and interpret it.
For fixed $k$ and $D$, what happens to $\mathrm{Da}$ if the chosen length scale is tripled?
Why does a saturable reaction generally require more information than one Damköhler number to describe its dimensionless behavior?
When $t_D=20\ \mathrm{min}$ and $t_R=80\ \mathrm{min}$, calculate $\mathrm{Da}$. Which process is faster over the stated length scale?

Solutions. (1) $\mathrm{Da}=(0.010)(100^2)/50=100/50=2.00$, so reaction is faster than diffusion across the specified $100\ \mu\mathrm{m}$ scale. (2) Since $\mathrm{Da}=kL^2/D$, tripling $L$ multiplies $\mathrm{Da}$ by $3^2=9$. (3) Saturable reaction rates depend on concentration, so scaling introduces a concentration ratio such as $K_m/C_0$ as well as a reaction-strength ratio. (4) $\mathrm{Da}=t_D/t_R=20/80=0.250$; diffusion is faster than reaction over that scale.