How the Thaw Time Calculations Work

A. One shared transient model

Every method — refrigerator, countertop, running water, and the warm-water bath — is solved by the same 1-D transient heat-transfer engine. Only the external boundary condition changes between methods; the milk-and-container conduction model is identical throughout. The milk is divided along a single spatial axis into a fixed number of layers (24 for bags, 24 for bottles), so temperature can vary through the milk's depth rather than being treated as uniform.

Container geometry is approximated to that one axis:

• Storage bags → a flat slab, heated symmetrically through both broad faces (a half-slab with an insulated center plane).
• Bottles → a radial cylinder, heated through the lateral surface only. End-cap (top and bottom) heating is neglected to keep the model one-dimensional.

B. Energy, enthalpy, and phase change

Each milk layer tracks an enthalpy (energy per kilogram, relative to the freezing point) as its authoritative state, not a temperature. Temperature is then derived from enthalpy. This makes the phase change energy-conserving by construction: a layer cannot rise past the freezing point until it has absorbed the full latent heat, regardless of timestep.

Breast milk freezes below 0 °C because of dissolved solutes, so the model uses a representative freezing/melting point of −0.54 °C (Basdeki et al. 2021). Breast milk is about 87% water (Kim & Yi 2020), so rather than applying water's full 334 J/g to the entire milk mass, the latent heat is taken as an effective ~291 J/g (0.87 × 334). This is an engineering approximation derived from the water content, not a value any single study measured directly.

C. Transient conduction through the milk

Within the milk, heat moves by conduction. The governing equation is integrated in enthalpy form with a backward-Euler (implicit) scheme that is unconditionally stable, so the timestep is chosen for accuracy rather than stability:

Slab (bags):      ρ ∂H/∂t = k ∂²T/∂x²
Cylinder (bottles): ρ ∂H/∂t = (k/r) ∂/∂r ( r ∂T/∂r )

Here H is enthalpy (J/kg), T is temperature (°C), ρ is milk density, and k is milk thermal conductivity. At each step the temperature solve is a predictor for the diffusion fluxes; an enthalpy correction then sets each layer's energy from the actual net heat it received, so latent heat is conserved exactly per control volume.

D. Surface heat transfer

The outermost milk layer is coupled to the surrounding air or water through the full physical path — external convection film, container wall, and the half-cell milk conduction resistance from the surface layer's center to the wall — all in series:

q = (T_ext − T_surf) / R_tot
R_tot = 1/h + δ_wall/k_wall + (Δx/2)/k_milk

h is the external convection coefficient (much larger for water than still air), δ_wall and k_wall are the container wall thickness and conductivity, and Δx/2 is the half-cell milk distance. The container wall's thermal mass is lumped onto the surface layer so the wall itself warms and cools with the milk.

E. The warm-water bath: sizing vs. timing

The static bath is handled in two linked pieces. First, a first-law energy-balance equilibrium sizes the bath: it sets the heat the warm water loses as it cools equal to the heat the milk gains (sensible + latent) plus the heat the container wall gains:

m_w · c_w · (T_w − T_eq) = m_b · E(T_b → T_eq) + C_wall · (T_eq − T_b)
T_eq = T_target + 0.5 °C    (target supplied)
T_target = T_eq − 0.5 °C    (target unknown)

E(T_b → T_eq) is the energy per kilogram to warm (and melt) the milk from its start to the equilibrium temperature. This solver finds whichever of the water volume, water temperature, milk volume, start temperature, or target you leave blank. The +0.5 °C is modeling headroom — a conversion between the practical target you enter and the true thermodynamic equilibrium — not a safety limit.

Second, once the bath is sized, the finite-bath transient run tracks how the bath cools as the milk warms. At every step the heat delivered through all bath-contact faces (both broad faces for a bag, the single outer surface for a bottle) is subtracted from the bath:

F. Refrigerator, countertop & running water

For these methods the surrounding air or water is treated as a constant-temperature reservoir — it stays at the temperature you enter and does not cool as the milk warms. The same transient engine runs; only the boundary is an infinite reservoir instead of a finite bath. Running water uses a far higher convection coefficient than still air, so it thaws much faster. Because the model cannot measure your actual airflow or water flow, the coefficient is varied across a fast and slow bound and the result is reported as a range.

G. When is thawing "done"?

Fully thawed is the first time the phase change is complete throughout the milk — every layer has absorbed its full latent heat. Time to target is the first time the mass-weighted bulk-average milk temperature reaches your target. For a frozen start, the target time is gated on complete thaw: a warm outer region raising the average above target while any inner layer is still frozen does not count.

H. Numerical convergence

Spatial node counts (24 for slab and cylinder) were fixed by a one-time convergence study; refining further changes representative times by under ~3%. For each calculation the timestep is halved until a representative reported time stops changing, and two consecutive halvings must agree within ~2% before a result is accepted. This guards against false convergence plateaus. It is a numerical-convergence check only — it does not mean the estimate matches real-world thaw times to 2%, since the real-world inputs (airflow, contact, container dimensions) are themselves uncertain.

I. Model constants

J. What the model cannot account for

• Internal natural convection or stirring — milk is modeled as conducting, not mixing internally.
• Real bag thickness, bottle dimensions, or wall material — these use representative values.
• Bottle end-cap (top/bottom) heating, neglected by the 1-D radial model.
• Actual room airflow, fridge loading, or where the container sits.
• Your exact water flow rate and how much of the container the stream touches.
• Variation in milk composition, or how much was already partially thawed.
• Heat the static bath loses to the bowl, countertop, or surrounding air (only milk + wall + bath are tracked).

These real-world factors are why your actual thaw time can differ from the estimate — and why air-thaw, running-water, and static-bath results are shown as a range rather than a single number.