KYLA · technical working model · public versionpublic build 2026-09-08 10:42

What governs the cooling power?

Calculate the whole chain from the blood to the cooling medium, for any cooling method. Pick a scenario and a product, and you are done. The sliders are still there under the plus signs if you want to dig, but nobody needs to touch them.

Read first. The model is correctly set up but fed with literature values and assumptions, not measurements. It shows directions and ranks interventions. Before a number goes public it must be measured. KYLA's internal construction is not published in this version; it is represented by lumped, measured quantities that give identical results.

1 · Who is using it

Sets skin temperature, blood-flow ceiling, grip moisture and how hard the body defends itself against cold.

2 · Which method

About this product: construction, sources and what is measured, derived or guessed

3 · How cautious do you want to be

Same model, but fed with pessimistic, middle or optimistic assumptions within the range we have support for.

What if… the six questions that actually matter

These are the sliders where the answer changes in a way that matters for the product. Everything else is either a property of the product you have chosen, or a physiological constant. If you move something here you are still building the same product, you are just asking a question about it.

How it is used

Grip length, seconds

KYLA is built for short grips between sets. Shorter grips give a higher average output, because the skin does not have time to get cold, but more grips are needed for the same total.

Rest between grips, minutes

The surface has time to recover during the rest because the reservoir is locked at the melting point. Longer rest does, however, mean more leakage to the room.

The product's two big design choices

Melting temperature, °C

0 = ice. Our product = 14. Drag it down and watch the output rise a little, peak around 13-14 and then fall when the skin gets so cold that the vessels constrict.

Contact area per hand, cm²

KYLA ~55. CoolMitt glove ~100. Apex tube bundle ~37. Towel ~130. Vest ~600.
For glabrous skin only the first 90 cm² count as AVA-coupled; a glove that covers the whole hand does not get more blood flow, just more ordinary skin.

The user and the room

Hand grip and moisture

Ambient temperature, °C

Room 22. Gym 25. Hot day 30. Car in the sun 40.

About the model what sits behind the numbers

The model calculates the whole chain from the blood to the cooling medium: Q = (skin temperature − surface temperature) × contact area / (contact resistance + tissue resistance + product resistance), limited by what the blood flow can carry away and by the product's own capacity.

The tissue resistance is not constant. It rises when the skin gets cold and the vessels constrict, which is the feedback that makes a product that is too cold worse in the end, not better.

KYLA's internal construction is not published. It is represented by lumped, measured quantities: stored energy as a function of temperature, the internal base resistance, how fast the melt front develops, and the contact area. The results are identical to the full model. Competitor products are built from their published specifications and the physics of their architecture; none of them has been tested by us.

Physiological assumptions the model's most sensitive and least measured part

These are set by the scenario and by the literature. They do not need to be touched to use the tool, but this is where the uncertainty lives. If you want to know how much our numbers are really worth, drag the AVA threshold and see how much everything moves.

Set by the scenario

Skin temperature in the palm

Central drive of the vessels (flow floor)

How much flow remains when the AVAs have closed locally. The literature says that the local closure is powerful and happens even in heat-stressed people, so the floor is small: 5% at rest, 10% training, 18% overheated. The main effect of the central drive is instead that the skin is warmer to begin with, which in itself keeps the AVAs above the threshold.

Blood-borne ceiling, palm (55 cm²)

What the AVA shunts in ONE palm can carry away with fully open vessels. Converted to power density per cm² and scaled with the AVA-coupled area (max 90 cm²).

System factor for two hands

Thermally, two hands always extract 2x: each side draws its own heat from its own hand, and empties at its own pace. To the core we count 1.65x, because two hands together approach the perfusion limit and a larger share of the heat is then drawn from the hand's own tissue instead of from the core blood.
That thermal is 2.0 while core is 1.65 is therefore not an inconsistency, but the signature of a partly binding limit. Everything in the range 1.5 to 2.0 is physiologically defensible; 1.65 is a deliberately cautious middle choice pending calorimetry.

Skin blood flow, whole body

Used for non-glabrous skin (neck, torso, face), where the ceiling follows the body's general skin circulation instead of AVA shunts. Rest ~275, training ~2500, maximal heat stress ~7500 ml/min spread over ~1.8 m² of skin.

Literature values

Vessels halve at skin temperature

This is the SKIN temperature, not the product's. The skin is always considerably warmer than the cooling surface, because the heat flow requires a gradient the whole way. During a KYLA grip the surface is 14 C but the skin ~26 C.
21.5 C is measured, not guessed. Bergersen, Eriksen and Walløe (1997) cooled one hand in a water bath and measured the blood velocity in the finger artery with ultrasound Doppler. From 35 down to 27 C nothing happened; at ~21.5 C (spread 20-23 C between subjects) the AVA pulsations ceased abruptly. Grahn and Heller independently give ~20 C as the lower limit for the cold source and therefore run their own product at 18-22 C.

Tissue resistance, constricted vessels

With open AVA shunts the resistance is ~2.3e-3, because 37-degree blood reaches almost all the way up to the surface. When the AVAs close it rises only moderately, to ~4e-3: the cooled patch still has nutritive capillary flow and is surrounded by warm, perfused tissue that conducts heat sideways.
Closing the AVAs therefore does not lower the watts very much. What collapses is the core coupling, because the heat is then drawn from the hand instead of from the core blood. That is why an ice cube can measure high watts and still not cool the body.
If you set this high the feedback becomes self-reinforcing and the model locks at the threshold. That is mathematically correct but physiologically unlikely for a gripped cylinder, which only cools a patch of the palm.

Tissue heat capacity

Determines how quickly the skin gets cold. The time constant is R x C, that is ~37 s for a KYLA grip. After 90 s the skin has gone 91% of the way to equilibrium. That is why short grips give more than the equilibrium figure: the gradient is larger the whole time and the AVAs do not have time to close. ~2 mm of tissue with rho*c 3.6e6 gives 7200 J/m2K.

Definitions

Charge temperature (energy is counted from here)

Bulk temperature of the cooling medium when it comes out of the fridge, 8 C by default. The stored energy is the measured enthalpy from here up to the cut-off. A unit that has warmed through to 14 C has already absorbed about half of its reserve, so charge cold and keep it in the container.

Surface temperature at the first grip

Nobody takes the product straight from the fridge into the hand. On the way to the gym the aluminium shell warms ahead of the PCM core, so the grip surface is around 12 C when the session starts while the core still holds its charge. This only sets where the surface trajectory begins; the energy budget is unaffected. At 12 C the first grip delivers full core coupling. Set it equal to the charge temperature to model a grip straight out of the fridge.

Usable until the surface reaches

This is where we stop counting the product as usable. It keeps drawing heat even after that, but ever more weakly. 18 C is our working definition, 20 C is generous. This is what "active minutes" and the session length are counted against, so move it and see how much it matters.

Summary

Thermal W

0 W

To the core W

0 W

Whole system

0 W

Skin temp, °C

- C

Active time, min

- min

Core temp, °C

- C

How long a session it lasts

With the grip pattern below, and counted until the product is no longer usable.

Session length

- min

Number of cooled sets

Active time in the hand

- min

Sustained output, details output per grip and fade over the session

Sustained output over a real session

The boxes above show equilibrium, that is what the product gives if you hold it for as long as you like. That is not how it is used. Here a real pattern is run instead: short grips with rest in between, where each grip both drains the reservoir and benefits from the skin not yet having had time to get cold.

Grip length and rest are set under What if… in the left column.

At the plateau, W

- W

The product's stable operating point

10 min average, W

- W

20 min average, W

- W

First grip versus last grip shows whether the product holds its output or fades. A product that starts high and drops is a different kind of product from one that stays flat, even if the average happens to be the same.

Passive hold time how long it stays cold without being used

Passive hold time

How long the product stays cold without being used, in the selected storage.

Storage between uses

Where the resistance sits skin, blood flow or product

Skin to surface Tissue / blood flow Product

Comparison in the selected scenario

All methods calculated with the same user and environment, so the comparison is fair.

All figures are for the WHOLE system, that is both hands for the products that cool two hands (whether as two pieces or one double-sided unit) and one hand for those that only cool one. The Units · hands column shows both how many pieces the product consists of and how many hands it cools. These are not the same thing: Therabody and NICE ROCC are ONE unit that cools both palms, while our cylinder and Apex Narwhals are two separate pieces.
Both power columns come from the same simulation, so the difference between them is purely degradation over the session and not two different definitions.
The table is sorted on SUSTAINED W, that is the average over the first ten active minutes. A product that runs out after five minutes therefore gets half the average, because the remaining minutes count as zero. It is the only column that penalises a strong but short-lived product fairly.
Thermal W is what a calorimeter would measure. To the core W is what actually lowers body temperature, after vasoconstriction and after non-glabrous skin insulates the core. The difference between the two columns is the whole point of palm cooling.
Two units give 2.0x thermally but 1.65x to the core, because two hands together approach the perfusion limit. The factor can be changed under Physiology.

Detailed terms for technical review

Term	Value	Comment
Cooling medium temperature		PCM or liquid inside
Surface against the skin		what the palm actually meets under load. For Apex it should land close to their published 12.8 C.
Skin surface temperature		controls the vasoconstriction
Temperature difference		deep skin minus cooling surface
R contact		1 / contact conductance
R tissue		rises when the vessels constrict
R product		architecture dependent
Vessel opening		share of the palm's MAXIMUM AVA flow that is still open. 100% = the shunts wide open, blood rushing just under the skin. 10% = they have closed and only ordinary capillary flow remains. Controls both the blood ceiling and the core coupling.
Blood-borne ceiling		what the circulation can carry
Core coupling		the body part's anatomy
Plate drift under load		Peltier: how far above its advertised setpoint the plate ends up under a warm palm. Larger = weaker element or heat sink; the setpoint holds only unloaded
Hot side temperature		Peltier: must be cooled away
Calculated COP		falls when the hot side gets hot
Heat flux per cm²		compare with Heller and Grahn's measured 0.17 W/cm² at a 16 C surface, and 0.23 with negative pressure
Stored energy		usable, per unit
Passive leakage		what the surroundings steal

KYLA technical working model · Q = (T_skin − T_surface) × A / (R_contact + R_tissue + R_product). R_tissue is not constant: it rises when the skin gets cold and the vessels constrict, which is the feedback that eventually makes colder products worse. Power to the core = min(thermal power, blood-borne ceiling) × the body part's core coupling. Literature values and assumptions, not measured data.