The time may not yet be right for electric powered cars but the use of electricity in cars is spreading, with an electric motor hidden away in just about every nook and cranny. The days when you had to do anything so menial as wind a window down by hand are long gone, while powered wing-mirrors and seat adjustment are de rigeur on top-of-the-range models. And where there are motors, there are bound to be drives.
The problem is that these motors have to handle a lot of power. The starter motor in particular needs a large current, all of which has to pass through the drive’s power unit. The essential component in that power unit is a chip measuring just 1cm2. Unsurprisingly, this gets extremely hot. Therefore getting the thermal design right is an important issue for the designers of such systems.
Normally, electronics are attached to a metal plate which acts as a heat sink, with the heat dispersing into the air. But for higher-powered automotive applications this is not sufficient. If these systems are to be melt-proof, extra cooling is required.
Automotive manufacturers tend to think that the answer lies with water, as Claus Petersen, chief executive of Danfoss Silicon Power, explained. ‘Whenever we speak to automotive manufacturers about cooling, they will only talk water. There are two problems with this. First, electricity and water don’t mix so we have to keep the water sealed away from the electronics, but that’s fairly simple. The second is more complex: water cooling is not particularly efficient, and needs some engineering.’
The method normally used for water-cooling involves counter-current cooling, a technique familiar both from condensers in chemistry labs and the huge heat exchangers in chemical plants. The heat sink plate of the power unit is mounted in a water bath which is sealed by gaskets to ensure that no water can get into the electronics housing. On one side of the water bath an inlet allows coolant into the system. The outlet is on the other side. The coolant – from the engine’s own water/antifreeze supply and powered by the normal water pump – flows across the plate taking excess heat with it. On its own, this has severe drawbacks, mainly due tovarying temperature in different parts of the unit (temperature gradients); the coolant at the inlet is much colder than that at the outlet so components mounted on the outlet side tend to burn out much faster than those near the inlet.
To increase cooling efficiency, engineers can machine the immersed side of the baseplates into complex shapes known as pin-fins. This makes the coolant’s flow pattern more turbulent, increases the time it is in contact with the plate and spreads the cooling effect more evenly. However, it’s costly to grind out or cut shapes into the thick copper sheeting needed for the heat sink. Machined base-plates are five times more expensive than flat ones, said Petersen, and even this technique can’t get rid of temperature gradients altogether.
As with many innovations, it took a sideways look at the problem for Danfoss engineers to find the solution. Temperature gradients are caused by the coolants being in contact with the hot plate for a long time. To make cooling more efficient, it’s better for it to be on the plate for a very short time. So rather than having the coolant flow across the whole surface, a system has been developed whereby coolant is sprayed on to the plate in jets, then led away quickly before it can warm up enough to form a temperature gradient. The liquid only needs to travel a short distance which minimises pressure drop and cuts the load the system exerts on the engine’s coolant pump.
Also, because the technique itself creates turbulence, there’s no need for a shaped baseplate – a completely flat, and therefore cheap, one can be used. And the fluid handling system too can be made from cheap moulded plastics. Known as Shower Power, the system is now in late-stage testing with several large car companies and shows promise for other high-power applications.
Of course, even simple ideas tend to be complicated when they’re implemented and Shower Power is no exception. The development team had to work out whether it was more efficient to have many shower nozzles or few, how far to let the coolant circulate on the baseplate surface and the geometry of the flow-path of the coolant.
The number of nozzles needed came as a surprise, said Petersen: reducing the number increased the cooling efficiency dramatically as it also increased the velocity of the coolant flow. For a standard-sized power module of 180mm x 100mm, the optimum number of nozzles is 20-25. Another design decision saw the team link the inlet and outlet with a narrow channel which changes direction in a series of hairpin turns. Every time the liquid is forced to change direction it is mixed – a process using energy drawn from the hot baseplate surface.
The meandering channels, which are 2-3mm across, are arranged like interlocking fingers so that one finger supplies coolant while its neighbour drains it away. Each inlet, outlet and channel makes up a square ‘cell’, each arranged symmetrically so each has the same pressure drop and, therefore, the same cooling effect. ‘We are achieving true parallel cooling,’ said Petersen.
The module is made up of three units: the power module itself with its integral heat sink baseboard made from a flat sheet of a metal such as copper or a material with superior heat conducting properties such as an aluminium-silicon-carbon; the flow management unit moulded from plastic with its inlet nozzles, meandering channels and cells; and a plastic or metal base unit containing the inlet and outlet pipes. This unit, which is basically a slope-floored trough into which the other components are sealed, can be built into the housing for the rest of the drive’s components.
The team used design techniques such as computational fluid dynamics (CFD) to optimise the number of bends in the channels and their cross-section (so as to minimise the risk of their being blocked by debris in the coolant). CFD also allows for custom design; in a unit where one component runs hotter than the others, the channels beneath the hot spot can be made narrower, increasing the coolant flow velocity and thus cooling efficiency.
The improvement over pin-fin countercurrent technology is marked, said Petersen.First and most noticeably, thermography reveals that there is no temperature gradient: where the pin-fin design shows that the inlet end is cooler than the outlet, the Shower Power module shows each cell as having an almost homogeneoustemperature pattern across its area. And further analysis points up these improvements: for both types of cooling the heat transfer coefficient increases with flow-rate but this increase is far steeper with the Shower Power module. Petersen said that the Shower Power is 25-30 per cent more efficient than the pin-fin unit.
The size of the modules is limited as there is an increasing chance of athermal runaway as the number of cells increases but this doesn’t limit power capacity. It’s a simple matter to link standard-sized modules together in ‘multiple packs’, each pack connected to its own Shower Power module. ‘The interlocking finger principle is repeated on a larger scale so that all the unit cells underneath all the modules are fed with the same coolant temperature,’ said Petersen.
Different-shaped modules are also possible; for example you could make a curved base for a module to be installed inside a tubular housing.
According to Petersen, while the units can be used wherever high energy densities are handled, automotive is the largest potential market. Danfoss is now in talks with several large companies, and discussions and testing are underway. So it may not be long before car makers’ new designs benefit from a cold shower.