Thermal mass and thermal lag simulator
Your cold room alarmed at 3 a.m. and the air was above 8 °C for twenty minutes. Whether the product ever got there depends on its thermal mass — work out the time constant for your load and see how much of that excursion it actually followed.
10 g of aqueous liquid with about 25 cm² of surface.
On this model the product itself went past 8.0 °C, for about 15 min, peaking at 9.15 °C. Thermal mass is not protecting you here — the event was long relative to the 22 min time constant. Treat this as a reason to investigate properly, with measured product temperature, rather than as a quantified exposure.
The same event against every load
| Load | τ | Product peak | Step followed | Over limit |
|---|---|---|---|---|
| 2 mL vial | 14 min | 10.34 °C | 76% | 20 min |
| 10 mL vial | 22 min | 9.15 °C | 59% | 15 min |
| 500 mL infusion bag | 56 min | 7.12 °C | 30% | never |
| 5 L bottle | 2.8 h | 5.77 °C | 11% | never |
| 10 L bulk container | 3.5 h | 5.64 °C | 9% | never |
Same room, same 20 min at 12.0 °C, same air movement. The only thing that changes is what was sitting in it. This is why a single air sensor cannot answer the question for a mixed store.
What this model assumes
- The load is one uniform temperature throughout. Lumped capacitance holds when the resistance to heat flow inside the body is small next to the resistance at its surface. It is a good description of a small watery load in air and a poor one of a large solid block, where the outside warms long before the middle.
- Nothing is packaged. Every preset here is a bare surface exchanging heat with the air. A carton adds its own thermal resistance, which usually dominates everything modelled here — a first version of this tool had a pallet preset that came out responding faster than a 10 L bottle, which is backwards. The effective coefficient for a particular carton, stacked a particular way, is not something this tool can derive, so it does not pretend to.
- The air steps instantly and holds steady. Real excursions ramp, fluctuate and recover unevenly. A square step is the worst realistic case for a given peak and duration, which makes it a safe shape for asking “could the product have got there?”
- The heat-transfer coefficient is a category, not a measurement. It is the number the answer is most sensitive to and the least knowable. The three settings above span roughly a factor of six, which is about the honest spread for air. If it matters, measure your load’s response instead of choosing a category.
A model is not a measurement. Nothing on this page is evidence about a real batch, and it should not appear in a deviation report as though it were. Its proper use is upstream of that: sizing an alarm delay, deciding whether a product-simulating probe is worth fitting, and knowing which of your stores genuinely need one.
Runs entirely in your browser · nothing you enter is sent to a server
How the calculation works
- 01The lumped-capacitance model, the standard first-order treatment of a body exchanging heat with its surroundings: the time constant is tau = mc/(hA), and the temperature follows T(t) = T_air + (T_0 - T_air)e^(-t/tau).
- 02Each step is integrated with the closed-form exponential rather than by a forward difference, so a long step cannot overshoot and understate the very lag the tool exists to show.
- 03One time constant is the time to cover 63.2% of a step change. The simulator reports tau directly, because that single number tells you whether a given excursion could have mattered before any other arithmetic.
- 04Heat-transfer coefficients are offered as three broad categories for air, spanning roughly a factor of six, because that is the honest range and it is the input the answer is most sensitive to.
Limitations
- This is a model, not a measurement. It is not evidence about a real batch and should not appear in a deviation report as though it were. The only thing that shows what a product did is a sensor in or against the product.
- Lumped capacitance assumes the load is one uniform temperature throughout. That describes a small watery load in air well and a large solid block badly, where the outside warms long before the middle does.
- Every load here is a bare surface. Packaging adds its own thermal resistance, which usually dominates everything modelled — so there is deliberately no preset for a cartoned or palletised load, because the effective coefficient for a particular carton stacked a particular way cannot be derived here.
- The air is modelled as stepping instantly and holding steady. Real excursions ramp and fluctuate; a square step is the worst realistic case for a given peak and duration, which makes it safe for asking whether the product could have got there, and wrong for quantifying exposure.
Frequently asked questions
- What is thermal lag?
- Thermal lag is the delay between a change in air temperature and the same change in the product sitting in that air. Heat has to cross the product's surface and spread through its mass, and both take time. The consequence is that a monitoring sensor measuring air responds far faster than the goods it is there to protect, so a brief air excursion can be a genuine alarm without ever becoming a product excursion.
- What is the thermal time constant?
- The time constant, usually written tau, is mass times specific heat capacity divided by the surface heat-transfer coefficient times surface area. It is the time the product takes to cover 63.2% of any step change in the air around it. After three time constants it has covered about 95%. Comparing tau with how long your excursion lasted is the quickest way to tell whether thermal mass was protecting you: an event much shorter than tau moved the product very little.
- Does thermal mass mean I can ignore a temperature excursion?
- No. An air excursion is a real event that needs recording and investigating whatever the product did, and the decision about product quality belongs to the people who own the stability data, not to a calculation. What thermal mass legitimately changes is how you design the system around it: how long an alarm delay you can defend before a genuine event is being ignored, whether a product-simulating probe is worth fitting, and which of your stores actually need one.
- Why does a small vial respond faster than a large container?
- Because the ratio of surface area to mass falls as things get bigger. A 2 mL vial has a large surface for its small contents and equilibrates in tens of minutes; ten litres of the same liquid has far more mass per unit of surface and takes hours. This is why a single air sensor cannot answer the excursion question for a mixed store: the same twenty minutes at 12 °C means something quite different to a vial than to a bulk container on the next shelf.
- Can I use a thermal lag calculation in a deviation report?
- Not as evidence of what the product experienced. A model with an assumed heat-transfer coefficient and an idealised geometry cannot substitute for a measurement, and presenting it as one invites exactly the challenge you do not want. The defensible version is to measure: place a sensor in a product-simulating load, or in the product itself where that is possible, so that the next excursion produces data rather than an argument.
- What is a product-simulating probe?
- A sensor placed inside something with a thermal mass similar to the product — often a vial or bottle of glycol or water — so that what it records tracks the goods rather than the air. It converts the question this simulator answers by calculation into something the monitoring system answers by measurement. Working out the time constant of your load is a good way to choose what the simulator should be filled with and how large it needs to be.
Need the reasoning, not just the number?
Ask ValiTrac AI and see the standards evidence and the engine calculation behind the answer.
Ask ValiTrac AIRead the theory
Handling temperature excursions in storage and distribution
An excursion is a deviation to be investigated against the product's stability data, not a number to be averaged away. Mean kinetic temperature is structurally incapable of seeing a short one: six hours at 12 °C in a 40-day cold-chain profile moves MKT by 0.08 °C.
Sensor response time and why it matters for monitoring
A sensor's response time sets how fast it can see a change. Bare-bead sensors respond in seconds; sensors in glycol or metal blocks in many minutes. Choose deliberately.
Alarm limits vs acceptance limits: setting them correctly
Acceptance limits come from the product; alarm limits sit inside them to give time to act; alert (warning) limits sit inside those. Uncertainty and delay decide the gaps.
Mean kinetic temperature (MKT): what it is and when to use it
MKT is not an average. It is the single temperature that would do the same cumulative chemical damage as the varying profile actually recorded, and because degradation is exponential it always sits above the arithmetic mean whenever the temperature varied at all.
Temperature monitoring during transport
Every temperature-sensitive shipment carries a calibrated logger placed with the product; the record is reviewed on receipt and excursions handled before release.