I was thinking about Ice pants and related ice vest stuff on the bus the other day. One of the potential problems with the design is that it needs a pump; pumps are usually noisy and unreliable, in part because they contain a lot of moving parts. A pump that avoided these problems would be desirable for an everyday-wear ice vest, which after all only needs a small total flow rate. (To remove 200 watts of heat by warming water from 0° to 20°, you need about 2.4 milliliters per second: 200 W / (20 kcal/ℓ).)
Magnetohydrodynamic motors are pumps that solve these problems by passing a current through the liquid to be pumped at right angles to an applied magnetic field. The magnetic field produced by the current creates a pressure gradient in the fluid, which propels it silently through the passage. Jacque Fresco famously envisioned silent ships propelled by this means, although this might require unreasonably large fuel cells if you aren’t going to generate the electricity with a noisy heat engine; perhaps some navy has tried this approach for silent nuclear submarine propulsion. In, I think, 2018, the YouTube channel Cody’sLab demonstrated the use of such a pump for pumping mercury between reaction vessels for the chloralkali process.
However, electrolysis poses a potential difficulty for magnetohydrodynamic motors in water and other ionic conductors: the electrodes immersed in the liquid will tend to produce bubbles of oxygen and hydrogen. This problem can be reduced to some extent by periodically reversing the direction of the current (and of the applied magnetic field, so as to not reverse the flow direction) but not, I think, eliminated, in a classical MHD motor. In most circumstances, this is not a problem for MHD motors, though measures must be taken to prevent anodic dissolution, such as using carbon electrodes.
Perhaps the lore of electroplating has useful information here, because bubble buildup is a problem for electroplating — plating doesn’t happen in regions of the cathode covered by bubbles. But anode bubbles are no problem for electroplating.
The buildup of explosive gases in the tubing of an ice vest could not only impede liquid flow but potentially even pose a hazard of rupture and fire.
The coolant in the ice vest needs to have a melting point below that of pure water so that it can pass unimpeded through the ice pack without risk of freezing, and it needs to prevent bacterial and fungal growth inside the tubing, since those can also produce gases and potentially corrode the tubing. One possible solution to this problem is to use water sufficiently salty that almost no life forms can survive in it, and this has the advantage of greatly increasing the water’s conductivity. This probably requires keeping the water from contact with glass or metal to prevent heavy corrosion, though most common plastics would be fine.
A possible electrolysis-free way to do a magnetohydrodynamic motor to pump an electrolyte is to use eddy (electrical) currents rather than linear (electrical) currents through the electrolyte, thus eliminating the need for electrode contact — in essence, a squirt coilgun rather than a squirt railgun. This is the same approach used in squirrel-cage AC motors, coilguns, and some magnetic-levitation systems. I think eddy currents should provoke no ionic concentration gradients at all.
One way to do this would be to wrap many coils around an electrolyte-filled tube and energize the coils in the same current direction in sequence, moving the magnetic field along the length of the tube. This will induce eddy currents inside the electrolyte around the axis of the tube opposing the direction of the coil current, thus producing an opposing magnetic field, growing with a time constant related to the inductance of the single “turn” around the axis and the resistance of the electrolyte; when the coil magnetic field moves to the next coil, the thus-magnetized electrolyte will be attracted to it, with its magnetic field potentially diminishing with the same time constant. Energizing the coil behind it with the opposing direction of current will also help.
The movement speed of the applied magnetic field needs to bear a certain relation to the movement speed of the liquid and the RL time constant of the induced eddy current. If the field moves too fast (e.g., 100 000 m/s), the eddy currents induced in the liquid will not have time to build up to a level where they can produce an appreciable magnetic force; indeed, the magnetic field will be confined to the skin of the electrolyte by the skin effect. (Am I misunderstanding this? Perhaps large eddy currents occur almost instantly and the magnetic field penetrates more deeply as they begin to die away?) On the other hand, if the field moves too slowly (e.g., 1 μm/s), the eddy currents will have decayed to very low levels before the applied magnetic field moves to the next coil. Somewhere in between is a sweet spot.
It may be more efficient to use a small tubes to get a more concentrated magnetic field or large tubes to get less viscous losses, to use concentric tubes separating liquid layers to force the eddy currents in the outer parts of the liquid to enclose a large area of magnetic flux, to use an annular tube with just an air space in the middle for the same reason, and perhaps even to replace that air space with something like ferrite — although perhaps that would result in increasing the force on the ferrite rather than the liquid.
As an alternative to magnetohydrodynamic motors, you could drive a solid impeller.
Chemistry labs nowadays commonly use magnetic stirrers, often built into hotplates. These apply a magnetic field rotating at a few Hz to the reaction vessel, typically an Erlenmeyer flask; a magnet moving freely at the bottom of the vessel is free to rotate to align itself with the rotating magnetic field. (The magnet must be encapsulated in something; I suspect teflon.) This usually produces a substantial amount of noise, but much less than a conventional motor driving a gearbox which turns an impeller on a bearing-mounted shaft that passes through a seal. And, although it has a moving part, that moving part has tolerances measured in centimeters, so wear is not much of a problem.
You could apply the same approach to the ice vest, applying a rotating magnetic field to rotate a solid impeller entirely contained inside the liquid chamber; the low speeds involved (< 60 rpm) suggest the use of a permanent-magnet-based impeller rather than something like a squirrel cage. To prevent the clattering noises common with chemistry-lab magnetic stirrers, the impeller could be circular, like a centrifugal blower, rather than oblong; a Tesla-turbine design might work. By giving it the same overall density as the liquid, balancing the extra density of the permanent magnets by including air bubbles, you could avoid stresses from impacts and changes in the direction of gravity.
The magnets themselves might be sufficient to prevent its axis of rotation from deviating too far from the axis of the pumping chamber without requiring a shaft, and to prevent it from translating axially until it hit a wall of the chamber; thus it could normally operate without any solid-to-solid contact and thus without any wear. Whether the magnetic fields were radial (as in a squirrel-cage motor or ordinary BLDC motor) or axial (as in a pancake motor), active position control may be necessary to prevent instability that would cause it to drift closer to some of the electromagnets until it hit a wall. Perhaps fluid-bearing effects could be adequate to prevent this.
In some sense the ideal would be to have the highest MMF density and thus flux density at the center of the pumping chamber rather than near its walls, so that the magnet or magnets in the impeller would tend to drift toward the center rather than toward the walls. Given that the coils applying the magnetic field are necessarily in the walls, I’m not sure if this is feasible, but it might be a good start to wrap the coils around the center of the chamber (e.g., in the xz plane and the yz plane, if the z-axis is the rotation axis and the origin is the center of the chamber — not parallel to those planes, actually in them) rather than around cores outside its walls.
The use of a solid impeller would obviate the necessity for using an electrolyte, allowing the use of less corrosive liquid coolants (see Coolants) such as propylene glycol or a water–(propylene glycol) mixture. This in turn could perhaps enable the use of a bare metal impeller without excessive corrosion, though I would think that plastics such as PET would still be a better choice.
If the pumping chamber were spherical rather than cylindrical, the impeller could rotate freely in it when the pump was being rotated in space, thus avoiding collisions with the walls, even if the impeller itself were still cylindrical (though it could be spherical as well, like those spherical compasses people mounted on their car dashboards in the 1980s). The applied magnetic fields would eventually bring it back into alignment for proper pumping action.
Another reason to want to use a tiny, low-power pump is to reduce the amount of heat added to the water from viscous friction. If you're using a 10-watt pump that's 50% efficient, it's adding 5 watts of energy to the water, and so the water will heat up at 5 watts. While this is small compared to the total cooling load of a one-person ice vest, it is not insignificant --- someone relaxing at 70 watts might find that the reduction in "battery life" from 5 hours 17 minutes to 4 hours 56 minutes (on 4 kg of ice) was a significant loss.
An even bigger issue, though, is the weight of the battery used to power the pump. 10 watts over 4 hours is 144 kJ, which is about 250 g of lithium-ion batteries. Reducing the pump power usage down below the one-watt level, if it's possible, would lighten the battery weight of the suit dramatically.
As described in Intermittent fluid flow for heat transport, a pulsing flow is more effective than constant flow for this sort of thing, because it distributes the coolth more evenly. One way to achieve this is to use a pump with several times the needed flow rate, but only run it a small fraction of the time. This results in higher viscous losses (for a given tube diameter) but may be worth it.