Population-Balance Models for Particle Dissolution

From Size Classes to a Dynamic Population

A bin-based dissolution calculation can describe how much solid is present in several particle-size classes at a particular time. A population-balance model adds the time-dependent movement between those classes: as particles dissolve, particles leave larger classes and enter smaller ones. This makes the model useful for following both the evolving size distribution and the transfer of drug from solid particles into solution.

The bookkeeping must distinguish particle number from drug mass. A transfer of one particle between two classes changes the number in each class, but it releases only the difference between the particle’s solid drug masses assigned to those classes. Counting each transferred particle as if its entire drug content dissolved would overstate the dissolution rate.

The \(q_i\) values are class-level representatives: they can be calculated from class sizes and the drug content per particle. The model does not track every particle’s exact size within its class. A transfer from class \(i\) to \(i-1\) is therefore a discrete approximation: it reduces the assigned solid mass by \(q_i-q_{i-1}\) per particle transferred.

Class-Transfer Equations

For a class in the interior of the size range, the population increases through arrivals from the next larger class and decreases through departures to the next smaller class. The same balance applies at the boundaries, provided that the smallest class has an outflow \(J_1\) and the largest class has no external inflow.

A common closure is a first-order transition rate, \(J_i=k_iN_i\), where \(k_i\) has units of inverse time. This is a modeling choice, not a requirement of population balances: the transfer rate could instead depend on other state variables or be calculated from a size-coordinate flux. Whatever the closure, \(J_i\) must have units of particles per time and should be nonnegative for dissolution-driven shrinkage.

The total particle count is \(N_{\mathrm{tot}}=\sum_{i=1}^{m}N_i\). Adding the class equations cancels internal transfers:

Thus, internal movement does not create or destroy particles. The total count falls only when particles leave the smallest class. Particle number is not conserved through complete dissolution, whereas drug mass can still be conserved by accounting for the dissolved material.

Drug-Mass Balance and Its Derivation

The total solid drug mass is \(S(t)=\sum_{i=1}^{m}q_iN_i(t)\). Substituting the class equations gives a direct check on the solid-mass change. In the rearrangement below, each transfer rate contributes the mass difference between the class it leaves and the class it enters.

To see why the second line follows, re-index the inflow sum: \(\sum_{i=1}^{m}q_iJ_{i+1}=\sum_{i=2}^{m+1}q_{i-1}J_i\). Its \(i=m+1\) term is zero because \(J_{m+1}=0\); subtracting \(\sum_{i=1}^{m}q_iJ_i\) leaves \(-\sum_{i=1}^{m}J_i(q_i-q_{i-1})\). The negative of the solid-mass change is consequently the dissolution rate:

For a well-mixed dissolved-drug amount \(A(t)\), let \(U(t)\) denote drug leaving solution through uptake or another specified sink. Then \(dA/dt=R_{\mathrm{diss}}-U\). If \(B(t)\) is the cumulative amount removed through that sink, \(dB/dt=U\). Adding the three balances gives \(d(S+A+B)/dt=0\). This ledger assumes no other source or loss, such as degradation, and makes the accounting boundary explicit.

Continuous Size-Coordinate Form

A continuous population balance provides a useful interpretation of the class model. Let \(x\) be a particle-size coordinate that increases with particle size, and let \(n(x,t)\) be the number of particles per unit size. Suppose particles shrink with \(\dot{x}=-v(x,t)\), where \(v\geq0\). With no particle birth, aggregation, or external influx, conservation of particles along the size coordinate gives:

Define \(q(x)\) as drug mass per particle at size \(x\), with \(q(0)=0\), and let solid mass be \(S(t)=\int_0^\infty q(x)n(x,t)\,dx\). If there is no incoming particle flux at the upper boundary and the flux and mass are well behaved, integration by parts gives:

The boundary term is zero under the stated conditions: there is no influx from arbitrarily large sizes, and \(q(0)=0\) at complete dissolution. If \(q'(x)\geq0\), then the continuous dissolution rate is \(R_{\mathrm{diss}}=\int_0^\infty q'(x)v(x,t)n(x,t)\,dx\geq0\). The integral expresses the mass released by particles as they pass through size space. The discrete formula approximates this continuous loss by assigning a finite mass decrement to each class transition.

Checks, Comparisons, and Failure Modes

A population balance is a dynamic extension of bin-based dissolution calculations: it updates class populations through time rather than treating each bin only as a separate contribution to a snapshot. A continuous size-coordinate equation avoids fixed class boundaries but requires a size-dependent flux and numerical treatment of the size domain. Particle-level simulation can represent individual trajectories, but it may be more computationally demanding when the population is large. These approaches differ in representation; each still needs consistent particle and drug-mass accounting.

Useful diagnostics follow directly from the balances. Check that every internal transfer appears once as an outflow and once as an inflow, that the smallest-class outflow is treated as complete dissolution, and that the drug released per transfer is the correct mass difference. At every simulated time, inspect whether \(N_i\geq0\), \(R_{\mathrm{diss}}\geq0\), and the total mass ledger closes to numerical tolerance. Large mass residuals can reveal a missing transfer, mismatched units, an incorrectly signed flux, or a dissolved-drug update that does not use the same dissolution rate as the solid balance.

The class representation has limits. A single \(q_i\) cannot describe variation among particles within a class, and the transfer rule approximates continuous shrinkage with jumps. Results can depend on class boundaries and representative masses, especially when classes are wide. Refining the classes can test discretization sensitivity, but does not by itself establish that the assumed transfer-rate closure is appropriate. The model also does not determine a dissolution law merely by writing the balance: the fluxes \(J_i\) or the shrinkage velocity \(v\) must be specified and supported separately.

Common Mistake

Wrong reasoning: “Ten particles leave the smallest class per hour, and each contains 0.20 mg, so all 2.0 mg/h is the dissolution rate”—or, for a larger-class transition, “every particle transferred releases its entire class mass.” The second statement ignores the solid drug mass the particle still carries after entering the smaller class.

Correction: Count only the difference between the assigned masses on either side of each transition. A class-2 particle entering class 1 releases \(q_2-q_1\), not \(q_2\). Include the full \(q_1\) only when the particle leaves class 1 and is treated as completely dissolved. Then verify that the resulting \(R_{\mathrm{diss}}\) equals \(-dS/dt\).