loading...
Temperatures over the trip: ambient, cargo-space air, outer layer and core of the load; green dashes = allowed range
Shelf life used (%) over the trip for the whole load, the outer layer and the core; red dashes = 100 %
Histogram of the shelf life used at delivery over the synthetic Monte-Carlo shipments; grey line = 100 %
Arrhenius ageing rate relative to storage at the reference temperature (log scale), with the current outer, core and ambient temperatures
Starting.
running...
The science
Heat transfer into an insulated load. The cargo space is modelled as three lumped temperatures: the air (Ta), an outer layer of the load (Ts, 30 % of the mass: outer cartons, pallet edges) and the core (Tc, 70 %). Heat enters through the walls with the conductance UA and moves inward through two thermal resistances:
Ca dTa/dt = UAeff (Tamb − Ta) + hA (Ts − Ta) − Qcool
Cs dTs/dt = hA (Ta − Ts) + kc (Tc − Ts) + qresp,s
Cc dTc/dt = kc (Ts − Tc) + qresp,c
Cs + Cc = m cp is the heat capacity of the product (tonnes of food hold megajoules per kelvin), so a stopped refrigeration unit warms the air within minutes, the outer cartons within hours and the core within a day. That lag is why a short excursion on the data logger (air probe) is often harmless, while a long one is not. The time constant of the whole load is τ ≈ m cp/UA. While a door is open the conductance is multiplied by (1 + kdoor), and on a dock with the unit off by (1 + 0.25 kdoor). The unit is a proportional thermostat, Qcool = Qmax d(Tamb) clamp[(Ta − Tset)/band, 0, 1], with its capacity derated by 1.2 % per kelvin above 25 C ambient; the electricity is Qcool/COP with COP = 0.4 Tset/(Tamb − Tset) (about 40 % of the Carnot value, clamped to 0.9..3.5). Phase-change packs are handled with the apparent-heat-capacity method: near the melting point the outer layer gets the extra capacity L m g(T − Tpcm), g a smooth bump of width ±1 K and unit area, with L = 200 kJ/kg, so a pack plateau lasts about L m / (heat leak). Fruit and vegetables also give off respiration heat q = q0 m ebT (strawberries: 0.05 W/kg at 0 C, b = 0.09 per K).
Spoilage kinetics (Arrhenius). The rate of the dominant quality-loss reaction follows k(T) = A exp(−Ea/RT), so relative to storage at the reference temperature the ageing rate is
r(T) = exp[ (Ea/R) (1/Tref − 1/T) ]
With Ea = 90 kJ/mol every 10 K near 5 C multiplies the rate by about 3.7. The fraction of the shelf life that is used up is S = ∫ r(T(t)) dt / SL, where SL is the shelf life (or, for a vaccine, the stability budget) at Tref; the page integrates it separately for the outer layer and the core and weights them 30:70. When S reaches 100 % the verdict is FAIL. A load is also failed when freeze-sensitive goods have an outer layer below −0.5 C for 30 minutes, or (vaccine) when it spends more than 120 minutes above its limit. The mean kinetic temperature (Haynes 1971), MKT = (Ea/R) / (−ln[ mean exp(−Ea/RT) ]), is the constant temperature that would age the product as much as the real, fluctuating one; it is always above the arithmetic mean, to second order by (Ea/R) σ2/(2T2) − σ2/T.
Synthetic shipments. The Monte-Carlo panel simulates the chosen number of shipments in your browser, each with its own seeded weather (mean temperature ±3 K, hour-scale noise), delay (log-normal, up to +80 %), number and length of door openings, load temperature at loading and, with the chosen probability, a refrigeration fault (median 2.5 h) at a random time. The histogram of the used shelf life and the share delivered within specification show what a rule such as "keep the air probe within the band" leaves to chance. The downloadable CSV is a synthetic data-logger record (air and core probes with Gaussian sensor noise); nothing in it is measured data.
Honest simplifications. (1) Three lumped nodes: there is no spatial airflow, no pallet-to-pallet difference and no stratification, so the front-to-back and top-to-bottom gradients of a real trailer are missing; the colours of the cartons only show the outer and the core node. (2) The thermal parameters (hA = 0.12 W/K per kg, kc = 0.05 W/K per kg, air plus structure capacity 3 kJ/K + 40 J/K per kg) are plausible round numbers, not measurements of a particular vehicle. (3) Ea, the reference temperatures and the shelf lives are illustrative values for each product class; the vaccine stability budget (30 days at 5 C) is a teaching number, and real decisions about a vaccine or any food follow the manufacturer's data, the product label and the applicable good-distribution rules. (4) A single reaction describes ageing; real products have several mechanisms, lag phases and growth limits. (5) The thermostat is proportional, not on/off with a defrost cycle. (6) Brands and carriers are not modelled.
Checks run in the page. (a) A passive warm-up over 12 h agrees with the matrix-exponential solution of the same linear three-node system to better than 0.02 K; (b) with the thermostat, the air temperature after 200 h equals the closed form (UA Tamb + K Tset)/(UA + K); (c) with phase-change packs the enthalpy change (including the latent heat from the closed-form integral of g) equals the time-integrated heat flow; (d) a vaccine held at 25 C for a day uses exactly the shelf-life fraction given by the Arrhenius formula; (e) the mean kinetic temperature of a sinusoidal swing matches the second-order estimate, and exceeds the arithmetic mean.