While the team at 3D Hub is working on the mechanical design, and manufacture of the Galin Pump prototype, here, we will discuss the development of the vane control algorithm.

The first thing to point out is that Galin pump is different to all existing instances of pumps. This is because spinning the shaft of this pump is not sufficient to create pumping action. To reiterate the point consider that most pumps have a single drive shaft, rotating this shaft causes the pumping action to occur. All existing pumps are separable from the mechanism that drives (rotates) the shaft. Not so in Galin pump, where the driving mechanism is integrated with the pump. There is no drive shaft to spin; the input is electricity, and the output is flow rate. This may sound radical, even like an over engineered design. Let me convince you otherwise.

To understand why this setup actually offers an advantage, let us take a step back, and consider that pumps are rarely (some exceptions are bicycle and water piston pumps) driven by hand. Most pumps require a means of rotating their shaft in order to operate. Consequently, most pumps require an automated method of producing rotary shaft power to operate, i.e. a pump driver. I stated this in general terms, but commonly the pump driver is either an electrical machine, or internal combustion engine.

Now, in reality, the pump is always coupled with a driver, and hence it does not really make sense to consider the efficiency of a pump as separable from the driver. What this means in practice is that, while there may be pumps in existence that offer volumetric and mechanical/hydraulic efficiencies of close to 100% - if you cannot efficiently couple that with a driver the overall results will always be poor. Stated another way: it is always important to consider the context that the pump will operate in - and select accordingly. It is system efficiency that is the metric to look at. Here’s another place where this same argument is made.

In Galin pump the driver and pump mechanism are integrated. The is no headache to be had choosing the pump, driver mechanism, and coupling. As Galin pump offers integrated sensing of differential pressure (between input and output ports) as well as real-time flow measurements, additional relief valves, and sensors may become redundant. Finally, because the pump has integrated computing, adjusting the operating point of the pump can be done in real-time. Galin pump is the answer to improving system efficiencies that industry has been seeking.

Having described the expected behaviour of Galin pump, let us look into the details of how it actually achieves this. (FYI - towards the end of this page, we discuss and present calculations for the limits on motion of the pump as determined by the performance of the electrical machines used to drive it.)

First, to reiterate, rotating the shaft of Galin pump at a constant speed will not result in pumping action. The are in fact two shafts (one on each side of the chamber) and the rotation of these shafts needs to follow a specific pattern to achieve pumping of fluid from the input to the output ports. Figure 1 is a schematic drawing of the Galin pump chamber with the input and output ports labelled, one pair of vanes is attached to the shaft (A) that is directed out of the page, and one pair of vanes is attached to the shaft (B) directed into the page. Our goal is to move the vanes in such a way as to open up a volume between the vanes opposite the input port without also, at the same time, creating a channel to the output port. Figure 2 shows a schematic that achieves this; shaft A has rotated 100 degrees, and during the same time interval shaft B has rotated 10 degrees, the volumes between the vanes now contain fluid that has come in via the input ports. Figure 3 shows the position of the vanes after shaft A rotates 10 degrees, and shaft B rotates 100 degrees, the volume between the vanes has now been pushed out of the output ports, and at the same time, new volumes have opened up to intake fluid from the input ports. This sequence of motion, if repeated indefinitely, will cause input and output flows equal to the two volumes between the vanes (minus whatever leakage occurs between the volumes) at each time step. It takes 4 time steps to complete a full shaft revolution, during which time, we have four input and four output strokes. Consequently, the flow of fluid through the pump is equal to: 4 * 2 * (maximum volume between the vanes) / (4 * (time step)).

Figure 4 (position/speed/torque(acceleration)) shows the trajectory of motion of shafts A and B in order to achieve the described sequence. This is an idealised trajectory - it will only ever be approximated by the electrical machines. This is a good place to point out the fact that under this trajectory of motion, the output fluid flow will be pulsing - which may be a problem. One immediate mitigating factor (compared to a positive displacement single piston pump) is that we have, as just described four (compared to one) output pulses per shaft revolution. However, another, even more important factor is that because we are controlling the position/speed/torque of the vanes with electrical machines, and we sense differential pressure, we area able to in real-time, use the motors to adjust the trajectory of the vanes in order to minimise the amplitude of the pulsations. So, for example, instead of applying constant acceleration, and then equal but opposite deceleration, which leads to a triangular change in speed - shown in figure 4, we can adjust the acceleration to smooth out the speed profile and thereby dampen the output pulsing amplitudes.

The important feature that the motor control should achieve is to get the vanes to move to the correct place inside the chamber during the required time step. Decreasing the time step results in flow increase, and increasing the time step results in decreased flow. This is all very easy and simple to understand. The limit on the maximum speed (and hence flow) that is achievable will depend on the pressure of the fluid, and torque limits of the electrical machines. What we should point out is that, in the event that the motors are commanded to achieve this position pattern via constant acceleration, followed by constant deceleration (as opposed to a sequence of incremental position changes) we are essentially putting the motors into regeneration mode in order to slow down the vanes. During this regeneration phase, current is pushed back into the power supply, hence why a small battery, and/or super-capacitors are required so as to absorb the electrical energy before giving it back again. ts