As mentioned briefly in Heating my apartment with a plastic tub of hot water, intermittent fluid flow can transport heat better than continuous fluid flow with the same average speed. The case there is that heating my apartment with the shared hot-water heater would require 19 mℓ/s, but drawing that continuously would just heat up the pipes in the wall, not the apartment. If, instead, I draw 300 mℓ/s for two minutes, I can put 36 liters of hot water into a bucket or radiator, which can then release the heat into the apartment over the next half hour. This eliminates more than 85% of the heat loss — although the half-liter or so of water left in the pipe for that half hour will lose nearly all of its heat, that’s only 3% of the total heat.
This is a general property of heat transport through thin, lossy tubes: if the amount of time the transport fluid spends in the lossy tube is pushed to extremes by using pulsing flow, the overall system efficiency improves, sometimes dramatically. I speculate that, in the mass-transport realm, this might be a reason vertebrate hearts use the shocking, violent pulsing motion that they do, incurring an obvious waste of energy from unnecessary viscous energy losses, rather than the peristaltic motion used by earthworm hearts and insect hearts: that way, most of the blood can spend more of its time in the capillaries and less of its time uselessly losing oxygen in large vessels, and furthermore even the venous part of the non-pulmonary capillaries gets fully-oxygenated blood, and even the venous part of the pulmonary capillaries gets fully-deoxygenated blood to oxygenate and decarbonate.
This has a direct application to ice vest design (see Ice pants.) Here you have a substantial amount of coolant (salt water, say) sitting in tubing inside the vest, absorbing heat from your body, and a substantial amount of coolant sitting in tubing inside an ice pack, releasing heat into the melting ice. If you use a continuous slow flow, all the coolant passing through the transitional tubing between these two points is going to lose an unnecessarily large amount of heat, and additionally, part of the vest will be ice cold, while another part will barely be cold at all, decreasing heat transfer; an analogous phenomenon will impede heat transfer within the ice pack, where the coolant near the exit is already too cold to release much additional heat. If, instead, you use periodic sudden pulses that replace most of the vest coolant at once, the whole vest will be a spatially uniform (but temporally oscillating) temperature, increasing heat transfer substantially.
Probably the way to do this electrically is to use a small electric motor to wind up a spring, then release a brake on the spring to drive the pump. This avoids the need for a large electric motor.
I think I hadn’t noticed this previously because of an aesthetic or philosophical inclination toward symmetrical, steady-state solutions, which are easier to analyze, rather than oscillatory or asymmetric solutions: wheels rather than legs, turbines rather than pistons, linearity rather than nonlinearity, flat rather than curved walls, continuous rather than crenellated or fractal webs. I speculate that this inclination is shared by much of modern science and engineering and represents a significant blind spot.