Fruit Slice Dehydrator (3D)

A Blender-built cabinet dehydrator: moisture diffuses out of 60 apple rings; a balance, a cut-away slab and live graphs show the physics.

Switch to 2D

Presets

Playback

Camera

Layers

Quality

Experiment

How long until the mean moisture reaches a_w = 0.60? Measured by a fresh simulation, compared with the analytic root.

Model checks

The science

Moisture moves by diffusion

Let X be the moisture content on a dry basis (kg water per kg dry solids). Inside a slice of thickness 2L the water follows Fick's second law, dX/dt = D d²X/dx², with the symmetry plane at x = 0 and both faces at x = ±L. Water leaves the faces into the air by convection: -D dX/dx = k_m (X_s - X_eq). The ratio of internal to external resistance is the Biot number Bi = k_m L / D; time enters as the Fourier number Fo = D t / L².

Closed-form series

For constant conditions the moisture ratio MR = (X̄ - X_eq)/(X_0 - X_eq) is MR = Σ 2Bi² e^(-β_n² Fo) / [β_n² (β_n² + Bi² + Bi)] where β_n tan β_n = Bi. For large Bi (fast air) this becomes 8/π² Σ e^(-(2n+1)² π² Fo/4)/(2n+1)²; for small Bi the slice stays nearly uniform and the decay is surface-limited, MR ≈ e^(-Bi Fo). At late times only the first root survives: MR → C_1 e^(-β_1² D t / L²), so the slope of ln MR against time is β_1² D / L².

Temperature, air speed and the isotherm

Diffusivity follows Arrhenius: D = D_60 exp[-(E_a/R)(1/T - 1/333.15 K)]. The mass-transfer coefficient scales as k_m ∝ v^0.8 (turbulent boundary layer, illustrative constant). The air sets the equilibrium moisture through a GAB sorption isotherm X_eq = X_m C K a_w / [(1 - K a_w)(1 - K a_w + C K a_w)] with illustrative apple-like constants (X_m = 0.12, C = 5, K = 0.9). Water activity a_w is the humidity the slice would hold at equilibrium; the readout inverts the isotherm for the mean moisture. A value near 0.60 is often quoted as a target for shelf-stable dried fruit, but this page is an educational model, not food-safety guidance.

Why the curve falls

At first the surface is wet and the rate is set by the air; as the surface dries, the slice becomes a diffusion problem and the rate falls as moisture has to travel farther. Thin slices (small L) win quadratically because the time scale is L²/D. A fast fan only helps until Bi is large; after that only temperature and thickness matter.

Checks run in the page

  • Solver vs series: a finite-volume Crank-Nicolson code (60 cells) against the closed-form mean and profile at three Fourier numbers and two Biot numbers.
  • Water balance: water lost from storage equals the time-integral of the surface flux to rounding error.
  • Monte Carlo: 20 000 random walkers with absorbing faces reproduce the Bi → ∞ series within 3.5 standard errors (and the solver in the same limit matches).
  • Late-time slope: the measured decay rate equals β_1² D / L².
  • Experiment: the simulated time to a_w = 0.60 equals the root of the analytic series.

Honest simplifications

One-dimensional diffusion through the thickness (the rim and the core hole are ignored); constant D (real D falls strongly as the tissue dries); the slice stays at air temperature (evaporative cooling is ignored); no shrinkage in the maths (the shrinking rings are a visual estimate from an ideal volume-additivity rule); no case-hardening, enzymatic browning or sugar crystallisation; the GAB constants ignore temperature; all 60 slices are identical, so tray-to-tray differences in a real cabinet are absent; k_m and D values are illustrative orders of magnitude, not measurements. Real drying times vary widely with variety, pre-treatment and cabinet design.

What to try

  • Run Thin chips and Thick slices: time to a_w 0.60 scales close to the square of thickness.
  • Compare Strong fan and Weak fan: at 0.2 m/s the Biot number is below 2 and the profile stays flat; at 4 m/s the surface dries at once and a steep gradient forms.
  • Choose Humid day: the slice stalls at the a_w of the air and never reaches 0.60.
  • Press Jump to dry, then change the air live: the dashed analytic curves hide because they assume constant conditions.
Loading model...
Loading Blender model...
Drying curve: mean moisture vs time (analytic series dashed)
Falling-rate curve: water loss per slice vs moisture
Moisture profile across the half-thickness now
Water activity of the mean moisture vs time
Made with 🟠 Blender · Three.js