Skip to content
Xiaoming XuSCIENCE · SYSTEMS · PEOPLE
Research Observatory
MODEL & EXPERIMENT · PARTICLE SCIENCE

Particles in transition.

Shear changes structure. Structure changes dissolution.

Explore the link from deflocculation to accessible surface area and in-vitro dissolution. Compare the mechanistic model with the published observations.

Triamcinolone-inspiredIn-vitro mechanismsStudy: Smith et al., 2021
BEFORE THE DISSOLUTION EXPERIMENT

Why let particles
form loose flocs?

Flocculation is the reversible association of small solid particles into loose clusters. The primary crystals remain distinct, with liquid between them.

Controlled flocculation can improve physical stability on the shelf: an open sediment is less prone to packing into a hard cake and can be easier to redisperse.

01

Loose association
Weak particle interactions form an open network.

02

Lower caking risk
Liquid-filled spaces help resist tight, irreversible packing.

03

Redispersibility
Mixing can redistribute flocs through the liquid.

Flocs may settle faster than individual particles. Formulation design balances settling, caking risk and redispersibility—and that same structure can influence dissolution.

Scientific context: Smith et al., 2021 · Introduction & section 3.5
INSIDE A SUSPENSION VIALSCHEMATIC
CrystalsLoose flocsOpen sedimentRedispersed

Individual crystals.

Small solid particles are distributed through the liquid.

Conceptual animation with accelerated stages. It does not calculate settling time, caking or shelf life. Dashed links indicate reversible association.

SHEAR → ACCESSIBLE SURFACE → DISSOLUTION

A mechanism you can explore.

Change a pre-shear pulse, then follow structure and dissolution after dilution. Compare with a sample given the same waiting time and vessel conditions, without the pulse.

Exploratory model · illustrative parameters · not fitted to the published curves.
01 / SHEAR HISTORY8.0×relative pre-shear intensity
02 / INITIAL AREA4.22×accessible area vs no pulse
03 / INITIAL RATE4.22×same mass-transfer coefficient
No pre-shearSelected pre-shear

Illustrative particles. Layout conveys the dispersion index, not tracked trajectories; crystal size follows simulated shrinkage. Deflocculation exposes existing surface—it does not fracture crystals.

0.00 min

Accessible surface area

6.44cm²27.19cm²

Breakup changes accessibility; crystal shrinkage reduces area.

Instantaneous dissolution rate

1.46%/min6.18%/min

Noyes–Whitney flux at the current area and concentration.

Cumulative drug dissolved

0.0%0.0%

Same dose, volume, solubility and vessel shear in both samples.

The dissolution rate follows dMd/dt = k Aeff(Cs − C), where k = D/h. Here k stays fixed to isolate surface accessibility. Agitation can also change mass transfer in real experiments.

WHAT THE MODEL REVEALS

Surface access
changes the response.

BREAKUP & RE-FLOCCULATION

Shear history can persist.

The pulse changes the initial dispersion index. Breakup and re-flocculation then compete under the same background shear in both samples.

ACCESSIBILITY & GEOMETRY

Wetting matters too.

If liquid already reaches all primary surfaces inside a floc, breakup need not increase effective area. Set internal access to 100% to explore this limiting case.

DRIVING FORCE

Dissolution slows as solution fills.

Solid mass and crystal size decrease. Concentration approaches solubility, reducing Cs − C. In a saturated vessel, increasing surface area cannot overcome the equilibrium limit.

Equations, assumptions and parameter provenance

1. Mean-field structural kinetics

df/dt = kb g² (1 − f) − (ka0 + ka1 g) (Ms/M₀) f

f is a mean-field dispersion/exposure index, not a measured particle-count fraction. g = G/Gref is relative shear intensity. The quadratic breakup and concentration-dependent first-order re-flocculation proxy are declared assumptions. They are inspired by competing aggregation/breakup models, not equations fitted to this suspension or copied from the paper. During preconditioning, solid mass is held fixed; dissolution starts at dilution. The reference sample undergoes the same wait at g = 0.

2. Accessible area and shrinking primary crystals

x = (Ms/M₀)1/3; Afree = 6M₀ x²/(ρd₀)
χ = 1/F + p(1 − 1/F); Aeff = Afree[χ + (1 − χ)f]

F is the compact-equivalent floc-to-primary diameter ratio. For F³ identical primaries, the compact-envelope area divided by their separate total area is 1/F. Parameter p adds internal surface access, from 0 (shielded limit) to 1 (all primary surfaces accessible). This is an idealized shielding model; porous real flocs need not behave as compact spheres. Uniform spherical primary-crystal shrinkage is a mean-field approximation, not a full size-resolved population balance.

3. Closed-vessel Noyes–Whitney dissolution

dMd/dt = k Aeff max(Cs − Md/V, 0); Ms + Md = M₀

k is an effective mass-transfer coefficient (D/h). Volume, solubility, density and k are constant; the vessel is well mixed. No removal, replacement, absorption or in-vivo extrapolation is included. Breakup does not create drug mass or smaller primary crystals. The numerical solver evolves x and f, checks physical bounds and reports rates from the same state as the charts.

Where the defaults come from

The 2 µm primary size, approximately 15 µg/mL water solubility, 150 mL test volume and 1.32 mg dose (33 µL × 40 mg/mL) are informed by the 2021 study. Density, floc factor, internal access, dispersion index, kinetic coefficients and k are illustrative assumptions. The simulator is not calibrated to its dissolution, D10 or obscuration curves. Its x-derived diameter is a model primary diameter, not measured D10 or a recreated PSD.

Related scientific basis: Serra & Casamitjana: shear-dependent aggregation and breakup; Gao et al.: diffusion-based, non-sink dissolution modeling. Their calibrated material parameters are not transferred to triamcinolone here.

The deterministic solver uses fourth-order Runge–Kutta with a default 0.05-second step and a separate 0.05-minute output grid. Unit, mass-balance, limiting-case, analytical structural-relaxation and time-step-refinement checks accompany the maintained implementation. These numerical checks verify the model calculation; experimental calibration remains a separate step.