Common Errors In Multi-Compartment Tissue Models

Why Small Equation Errors Change the Model

A multi-compartment tissue model can have the right compartments and still describe the wrong process. A reversed sign can make drug flow uphill in concentration when it should flow downhill. A concentration flux inserted directly into an amount equation can change the implied rate by a factor involving volume. Multiplying by membrane area twice can inflate exchange. And replacing a finite-rate flux with an equilibrium constraint removes the very delay the model was intended to represent.

The audit below uses two compartments: extracellular space $e$ and intracellular space $i$. The same checks apply to an added compartment such as an endosome by auditing each membrane interface separately. Assume fixed compartment volumes, no sources or losses during the exchange considered, and a passive, finite-rate membrane flux. These assumptions isolate exchange; additional processes can be added as separate terms.

Definition: Let $A_e,A_i$ be drug amounts and $C_e=A_e/V_e,\ C_i=A_i/V_i$ be concentrations in compartments with fixed volumes $V_e,V_i$. Define $K_{i:e}=C_i/C_e$ at equilibrium. Let $PS$ be the permeability-area product, with units of volume per time. Define positive flux $J_{e\to i}$ as amount transferred per time from $e$ to $i$. For this model, $$ J_{e\to i}=PS\left(C_e-\frac{C_i}{K_{i:e}}\right),\qquad \frac{dA_e}{dt}=-J_{e\to i},\qquad \frac{dA_i}{dt}=+J_{e\to i}. $$

With this convention, the flux is positive when $C_e>C_i/K_{i:e}$, zero at equilibrium, and negative when it points from intracellular to extracellular space. A negative value is not automatically an error: it means the actual transfer is opposite the chosen positive direction. The paired amount equations use equal and opposite fluxes, so exchange alone conserves $A_e+A_i$.

Audit the Direction Before the Arithmetic

A reliable sign check has two parts. First, write down what a positive flux means in words. Second, choose concentrations for which the direction is unambiguous and check that the donor loses amount while the recipient gains it. Do not infer the sign convention from an equation after the equations have already been coded.

Worked Example: Finding a Reversed Flux Sign

Suppose $V_e=2$ L, $V_i=1$ L, $PS=0.6$ L/h, $K_{i:e}=2$, $C_e=8$ mg/L, and $C_i=10$ mg/L. Determine the flux direction and the concentration rates of change.

The effective intracellular concentration on the extracellular scale is $C_i/K_{i:e}=10/2=5$ mg/L. Therefore, $$ J_{e\to i}=0.6\left(8-\frac{10}{2}\right)=0.6(3)=1.8\ \text{mg/h}. $$

The positive result means $1.8$ mg/h moves from extracellular to intracellular space. The corresponding amount rates are $dA_e/dt=-1.8$ mg/h and $dA_i/dt=+1.8$ mg/h. Since volumes are fixed, the concentration rates are $$ \frac{dC_e}{dt}=\frac{-1.8}{2}=-0.9\ \text{mg/L/h},\qquad \frac{dC_i}{dt}=\frac{1.8}{1}=1.8\ \text{mg/L/h}. $$

A sign-reversed implementation would make extracellular amount rise and intracellular amount fall in this case. That contradicts the stated flux direction. The amount-rate sum here is $-1.8+1.8=0$ mg/h, as required for exchange without loss.

Keep Amount Rates and Concentration Rates Distinct

The membrane flux $J_{e\to i}$ is an amount rate, not a concentration rate. If the state variables are amounts, use $dA_e/dt=-J_{e\to i}$ and $dA_i/dt=+J_{e\to i}$. If the state variables are concentrations, divide each amount rate by the volume of its own compartment. A shared flux therefore produces equal and opposite amount changes, but generally not equal and opposite concentration changes.

Worked Example: Converting One Flux to Concentration Rates

Let $V_e=3$ L, $V_i=1.5$ L, $PS=0.4$ L/h, $K_{i:e}=1.5$, $C_e=12$ mg/L, and $C_i=9$ mg/L. Calculate the flux and both forms of the compartment equations.

The driving difference is $12-9/1.5=6$ mg/L, giving $$ J_{e\to i}=0.4(6)=2.4\ \text{mg/h}. $$ Thus $dA_e/dt=-2.4$ mg/h and $dA_i/dt=+2.4$ mg/h. The concentration equations must account for different volumes: $$ \frac{dC_e}{dt}=\frac{-2.4}{3}=-0.8\ \text{mg/L/h},\qquad \frac{dC_i}{dt}=\frac{2.4}{1.5}=1.6\ \text{mg/L/h}. $$

Putting $-2.4$ and $+2.4$ directly into concentration-rate equations would be wrong: those values are in mg/h, not mg/L/h, and would ignore the volumes. Likewise, conservation is checked by summing amount rates, not by adding the two concentration rates.

This distinction is also useful when changing model formulations. With constant $V_e$, $dA_e/dt=V_e\,dC_e/dt$; if volume varies, that simple conversion does not hold without accounting for the changing volume. Here, fixed volumes are part of the stated model assumptions.

Count Membrane Area Exactly Once

Permeability per unit area and a permeability-area product are different quantities. If $P$ is permeability per unit area and $S$ is membrane area, then $PS$ is the product used in the flux equation. If a parameter has already been supplied or fitted as $PS$, it already includes area and must not be multiplied by $S$ again. For an interface between two compartments, use the area for that interface once; do not add another area factor merely because the flux appears with opposite signs in the two compartment equations.

Worked Example: Detecting a Duplicated Membrane Area

Suppose $P=0.03$ L/(h·m²) and the relevant membrane area is $S=20$ m². The concentrations are $C_e=10$ mg/L and $C_i=8$ mg/L, with $K_{i:e}=2$. Calculate the flux.

First form the permeability-area product: $$ PS=(0.03\ \text{L/(h·m²)})(20\ \text{m²})=0.6\ \text{L/h}. $$ The driving concentration difference is $10-8/2=6$ mg/L, so $$ J_{e\to i}=(0.6\ \text{L/h})(6\ \text{mg/L})=3.6\ \text{mg/h}. $$

If the model already stores $PS=0.6$ L/h but then multiplies it by $S=20$ again, it has counted area twice. Numerically that would produce a spurious $72$ mg/h if the area were treated as the same numerical factor, and the extra multiplication would also be inconsistent with the stated units of $PS$. The correction is to use either $P$ and $S$ together or a precomputed $PS$, but not both representations at once.

Finite-Rate Exchange Is Not Instantaneous Equilibrium

A finite $PS$ gives a differential equation: the concentrations move toward equilibrium over time. An instantaneous-equilibrium formulation instead imposes the algebraic condition $C_i=K_{i:e}C_e$ at every modeled time. The two formulations can have similar behavior when exchange is very fast relative to other processes, but they are not the same model. Setting an exchange flux to zero because the compartments are assumed to equilibrate instantly is not a valid way to retain a finite-rate exchange equation.

Derivation: For an isolated two-compartment system, let $D=C_i-K_{i:e}C_e$ measure departure from equilibrium. The amount equations imply $$ \frac{dD}{dt}=\frac{J_{e\to i}}{V_i}+\frac{K_{i:e}J_{e\to i}}{V_e}. $$ Since $C_e-C_i/K_{i:e}=-D/K_{i:e}$, it follows that $$ \frac{dD}{dt}=-PS\left(\frac{1}{K_{i:e}V_i}+\frac{1}{V_e}\right)D. $$ Thus $D$ decays exponentially with time constant $$ \tau=\frac{1}{PS\left(\frac{1}{K_{i:e}V_i}+\frac{1}{V_e}\right)}. $$ This finite time constant describes relaxation toward equilibrium. Instantaneous equilibration instead imposes $D=0$ without representing that relaxation.

Worked Example: Checking for Accidental Instantaneous Equilibration

Consider $V_e=2$ L, $V_i=1$ L, $K_{i:e}=3$, $PS=0.5$ L/h, and a conserved total amount of $100$ mg. At equilibrium, $C_i=3C_e$. Find the equilibrium concentrations and the finite-rate exchange time constant.

The total amount at equilibrium is $V_eC_e+V_iC_i$. Substituting $C_i=3C_e$ gives $$ 100=(2)(C_e)+(1)(3C_e)=5C_e, \qquad C_e=20\ \text{mg/L}. $$ Therefore $C_i=60$ mg/L. The amounts are $40$ mg extracellular and $60$ mg intracellular, which sum to the given $100$ mg.

For the finite-rate model, $$ \tau=\frac{1}{0.5\left(\frac{1}{3(1)}+\frac{1}{2}\right)} =\frac{1}{0.5\left(\frac{5}{6}\right)} =2.4\ \text{h}. $$ A model that sets $C_i=3C_e$ immediately at every time point omits this 2.4-hour relaxation scale. It may be a deliberate rapid-equilibrium approximation, but it is not the same as the stated finite-rate model.

Practical Diagnostics and Modeling Choices

Several quick checks catch many implementation errors before simulation results are interpreted. For a passive exchange term, test a state with $C_e=C_i/K_{i:e}$: the exchange flux should be zero. Test a state with $C_e>C_i/K_{i:e}$: the extracellular amount should decrease and the intracellular amount should increase. For exchange alone, the two amount rates should sum to zero. Finally, confirm that the parameter in the flux is either a permeability-area product or a permeability paired with area, and verify that concentration-rate equations include the appropriate volume divisors.

These checks do not guarantee that the full model is correct. A simulation that instantly satisfies the equilibrium ratio despite a finite exchange rate may indicate that the code has imposed an algebraic constraint or omitted a dynamic state. A very large but finite $PS$ can also produce rapid changes that are difficult to resolve numerically; that calls for checking the exchange time scale and simulation behavior, not silently replacing the model with instantaneous equilibrium. When other processes such as blood inflow, metabolism, or transport are present, mass conservation applies to the complete balance including those terms, rather than to exchange in isolation.

Representation	What is modeled	Key audit
Finite-rate, amount states	Amount changes driven by membrane flux	Use equal and opposite amount rates at the interface
Finite-rate, concentration states	Concentration changes driven by the same amount flux	Divide each flux by that compartment's volume
Instantaneous equilibrium	An algebraic partition relation, without a resolved exchange delay	Use only when the equilibrium approximation is intended
Explicit $P$ and area, or precomputed $PS$	The same area-scaled exchange capacity in two parameterizations	Count the membrane area once

Choosing amount states or concentration states is a modeling representation choice; with fixed volumes and correctly transformed equations, they encode the same exchange process. Instantaneous equilibrium is a different structural assumption, not simply a different notation. The appropriate choice depends on whether finite exchange delays matter for the question being modeled. Neither a good mass-balance check nor plausible concentration profiles alone can establish that all biological assumptions are appropriate.

Common Mistake

Wrong reasoning: “The flux is positive into the cell, so I add it to both compartment equations; also, the two concentration rates should cancel because the amounts are conserved.” This reverses the donor sign and confuses amount rates with concentration rates. The correction is to define one flux direction, subtract that flux from the donor amount, add it to the recipient amount, and divide by each compartment’s volume only when writing concentration rates. The concentration rates need not cancel when the volumes differ.

Check Your Understanding

For each question, identify the appropriate correction or calculation. Solutions follow so that each check can be verified.

If $J_{e\to i}=2$ mg/h and exchange is the only process, what are $dA_e/dt$ and $dA_i/dt$?
If those same compartments have $V_e=4$ L and $V_i=2$ L, what are the corresponding concentration rates?
A model uses $PS=0.8$ L/h, where $PS$ is already precomputed, and then multiplies the flux by membrane area. What should be checked?
What does it mean if a finite-rate exchange model is instead coded to enforce $C_i=K_{i:e}C_e$ at every time?

Solutions. 1. By the positive-direction convention, $dA_e/dt=-2$ mg/h and $dA_i/dt=+2$ mg/h; their sum is zero. 2. Divide each amount rate by its own volume: $dC_e/dt=-2/4=-0.5$ mg/L/h and $dC_i/dt=2/2=1$ mg/L/h. 3. Check whether area has been counted twice. If $PS$ already includes area, use it directly in the flux; multiply by area only when starting with a per-area permeability $P$. 4. The code has imposed instantaneous equilibrium, an algebraic approximation that removes the finite exchange relaxation time. It should be retained only if that approximation is intended; otherwise restore the finite-rate flux and dynamic compartment equations.