Adding shuttles to axial piston pumps can substantially increase overall efficiency.
Peter Achten, Robin Mommers and Jeroen Potma • INNAS BV, Breda, The Netherlands
Given today’s push to limit global warming and embrace a carbon-neutral future, there is a heightened focus on improving the efficiency of industrial and mobile equipment and their underlying fluid power systems. One area of keen interest involves minimizing the losses in hydraulic pumps and motors.
Part of the losses of hydrostatic devices are commutation losses, sometimes called compression losses, that result from the unavoidable and continuous process of compressing and expanding oil in the displacement chambers of a pump or motor. Commutation losses are strongly pressure related, but also depend on other factors like the type of oil, the pump construction, and the design and sizing of any silencing grooves. Depending on the operating conditions, commutation losses can amount to 50% of the total losses of a pump.
Researchers are investigating shuttles as a means to reduce or maybe even eliminate these losses. Shuttles are small connections between the individual and neighboring cylinders or displacement chambers. Each shuttle consists of a small cylinder which has two seats, one at each end, and contains a small ball. The shuttle ball can move back and forth inside its cylinder and, while the ball moves, oil flows in and out of the shuttle. Shuttles were first introduced for hydraulic transformers and, in 2021, were presented as a solution in pumps and motors.
Engineers at INNAS recently examined the effects of shuttles on efficiency in a fixed-displacement slipper type pump from Bosch Rexroth, which was tested with and without shuttles. Results show that adding shuttles to the pump substantially increased both average and peak efficiency over a wide range of speeds and pressures.
Shuttle basics
Hydrostatic pumps operate between two pressure levels: a low pressure level at the supply side and a high pressure level at the delivery side of the pump. For example, a nine-piston axial piston pump switches 18 times per revolution from one pressure level to the other: from low to high pressure in the bottom dead center (BDC) and from high to low pressure in the top dead center (TDC). At 3,000 rpm, this amounts to 900 switches per second.
This switching process, called commutation, occurs not only in hydrostatic pumps but also in motors and hydraulic transformers — in both fixed and variable displacement versions. A few pumps use valves to facilitate and enable the switching process, but the vast majority of hydrostatic machines uses a distributor, similar to the function of a commutator in direct-current electric motors with brushes. In axial piston devices, the valve plate or port plate acts as the distributor. Similar components are found in radial piston pumps and motors, vane pumps, gerotor and geroler motors, and even in gear pumps.
In an ideal machine, for example an ideal axial piston pump, piston movement compresses and expands the oil. For instance at BDC, the piston and its cylinder move from the low pressure kidney (the kidney-shape port machined through the valve plate) to the high pressure kidney. In between these two kidneys is the sealing land. When the barrel port is closed off by the sealing land, it has (for a short time) no connection to either the low or the high pressure side of the pump. In that case, the ongoing piston movement will result in compression of the oil inside its cylinder, since the barrel port of the cylinder is closed by the sealing land.
In this ideal pump, the sealing land around BDC needs to be dimensioned such that compression stops precisely at the moment when the pressure level in the barrel cylinder reaches the pressure level of the high pressure kidney. If the sealing land is too short, compression will not be enough and oil in the cylinder will open prematurely to the high pressure kidney. If the sealing land is too long, high pressure spikes will occur. Also in the commutation zone of TDC, the sealing land needs to be dimensioned exactly right (that is, not too long or too short) to avoid short circuiting or cavitation.
The problem is that the valve plate’s geometry and dimensions are fixed after production, whereas the ‘ideal’ timing of commutation cannot be defined as a constant. The amount of compression and expansion strongly depends on parameters like the pump pressure, the bulk modulus of the oil, the amount of leakage (and hence the oil temperature) and even the rotational speed of the pump. In current pumps and motors, manufacturers accept that commutation can’t be ideal. Instead, as a compromise, most commutation conflicts are softened by means of silencing grooves or pressure relief grooves.
The design of the valve plate and its silencing grooves has been a subject of numerous studies because it determines the noise, pressure pulsations, torque variations, and compression and expansion losses (commutation). Most of these concepts are meant to reduce noise in pumps, not power losses.
In 2021, INNAS presented a more general solution for applying shuttles in hydrostatic pumps and motors. Shuttles can, in theory, completely eliminate commutation losses and strongly increase overall efficiency. The solution can be applied in any hydrostatic machine which uses a distributor or valve plate and is applicable for both fixed and variable displacement machines.
Incorporating shuttles
To demonstrate the effects of shuttles on pump efficiency, INNAS specialists first tested a standard 28 cc Bosch Rexroth A4FO28/32R, fixed-displacement axial piston pump with no modifications, across a wide range of operating conditions. Efficiency plots at various speeds and pressures are shown in the nearby graphics.
After that, the pump was adapted to incorporate shuttles:
- Connections (shuttle pathways) were made between each pair of neighboring barrel cylinders;
- A new valve plate was designed and manufactured, without silencing grooves but with larger sealing lands;
- The spherical bearing surfaces of the new port plate and the barrel were lapped together to match both surface profiles.
Aside from the new valve plate and adaptation of the barrel, all other components were unchanged. This way, any effects measured can only be attributed to the shuttles and the new valve plate.
As the original A4FO28 is not designed for shuttles, fitting them into the existing barrels was a challenge. The shuttles had to be positioned to use the pressure difference across the barrel ports, and the shuttle volume needed to be large enough for any pump operation between 0 and 400 bar. Finally, the shuttles must be positioned such that centrifugal forces don’t influence the movement and positioning of the shuttle balls.
Each barrel cylinder connects to two shuttles: one leading the direction of rotation and one following. The shuttle balls are 4.5 mm diameter ceramic balls with a mass of 0.37 gm. The shuttle chamber has a 4.9 mm diameter. The shuttle balls make a stroke of 6 mm inside the shuttle cylinders and, in the end positions, the balls push into a conical seat and act as a check valve.
The adapted barrel was also designed so that the shuttles could be altered, permitting a change in the size, mass or material of the shuttle balls. It was a compromise solution in which one end of the shuttles has a 90° angled connection to the cylinder, which results in a larger flow restriction and, thus, in higher pressure differences than preferable.
New valve plate
A new valve plate design is also needed. The new plate does not rely on silencing grooves. Instead, the sealing lands around the top and bottom dead centers (TDC and BDC) are increased. The length of the sealing lands is determined by the maximum pump pressure and should be at least large enough to allow for a full compression (after BDC) and full expansion (after TDC). The sealing lands may even be larger, as long as the shuttles are large enough to compensate for any differences or mismatches.
In operation, the barrel ports move across the port plate. In the new plate design, the sealing lands have a length much larger than the arc length of the barrel port. The difference allows for full compression and expansion of the oil in the barrel cylinder, even when the pump operates at maximum pressure. The distance between the end of the low pressure kidney and BDC, and between the end of the high pressure kidney and TDC, is half the arc length of the barrel port. Consequently, the barrel ports close exactly at the TDC and BDC positions of the pistons. Immediately after reaching one of the two dead centers, piston movement starts the compression (BDC) or expansion (TDC) of the oil contents of the cylinder.
If the pump operates at a lower pressure than maximum, the compression or expansion will stop after the pressure in the barrel cylinder has reached the pressure level of the next kidney. At that point, the shuttle ball of the leading shuttle will start to move, thus avoiding any overpressure or cavitation.
Contrary to the positive overlap of the new valve plate, the old valve plate has a negative overlap with respect to the silencing grooves. Consequently, there is some degree of short-circuiting. Further, the barrel ports are immediately connected to the next kidney, which results in an extremely fast compression or expansion with high pressure-change rates. Silencing grooves limit these rates to some extent, but nevertheless the rates are very high, resulting not only in dissipative losses across the silencing grooves, but also in strong pressure and flow pulsations, noise, and possibly cavitation and wear.
Test results
The pump was tested in the INNAS lab following Bosch Rexroth pump specifications. All tests were performed with Shell Tellus Oil S2 MX46 at a supply temperature of 50°C (±0.5°C). In order to measure at a maximum allowable rotational speed of 3,750 rpm, the supply pressure needed to be 1 bar above atmospheric pressure.
The A4FO28 was tested in 104 different operating points, ranging between 50 and 400 bar and between 10 and 3,750 rpm. All operating points were measured at steady state conditions for a period of 10 seconds with 0.05 seconds interval between the individual measurement points. These data are averaged in post-processing. A run-in procedure was performed before each test, for the original and the adapted pump.
At low rotational speeds, pump leakage can become larger than the flow that the pump can deliver. In those cases, the pump is not capable of maintaining the desired pressure, and an additional supply pump is used to maintain the desired pressure level. In such cases, it was no longer possible to measure and define the overall efficiency.
Results show that shuttles and the complementary new valve plate have significantly improved pump performance. Peak efficiency increased from 91.9% to 96.4%. In the range between 500 and 3,000 rpm, and 100 and 400 bar, average overall efficiency has increased from 90.8% to 94.3%.
The contour plots shown nearby represent the measured overall efficiency of the A4FO28, without and with shuttles. Only at 50 bar and speeds ≥ 3,500 rpm was there no improvement of the overall efficiency. In all other cases the overall efficiency increased, especially at pump pressures above 200 bar. This was expected, as shuttles reduce the commutation losses which are strongly pressure dependent.
The most remarkable results are at operating speeds below 1,000 rpm. In this area, overall efficiency increased by more than 30%. For instance at 100 rpm and 400 bar, the shuttles have increased the overall efficiency from 49.6% to 75.3%, an increase of 33.5%.
The experiments have proven the shuttles substantially increase overall efficiency. Total losses are reduced by 41% on average, and up to 60% at high pump pressure and low rotational speeds.
The improvement is largely due to a reduction in hydromechanical losses. Part of the improvement is a result of lower commutation losses, which was the reason for developing the shuttle solution. But measurements show that, on top of the reduced commutation losses, the shuttles also cut friction losses in the pump itself, especially at low rotational speeds. The shuttles also reduce the volumetric power losses by 56% and the drain leakage by about 70% on average. The reason for the reduction of the volumetric loss is uncertain. One “guess” is that leakage between the slippers and swash plate is reduced. During the suction stroke, the pistons and slippers are pulled away from the swash plate. The retainer plate keeps the slippers close to the swash plate, thereby limiting the gap to about 20 µm. When the piston passes the BDC, the piston and slipper return to the swash plate. This takes time, but since the commutation is almost instantaneous, oil at a high pressure leaks via the gap between the slipper and swash plate to the case, while the slipper is returned to the swash plate and the gap height is reduced again to a few micrometers. More research is needed to fully explain the results. But the measured data shows unequivocally the positive effects of shuttles on pump efficiency.
It should be noted that this is the first test result of shuttles in a conventional pump. We expect that the shuttles can be improved even further when shuttles are integrated in the design of new pumps, instead of adapting conventional pumps.
According to the design principle, the shuttles also work in hydrostatic motors, and in two-quadrant pump/motors. In theory, the shuttles might also work in variable displacement machines. In these pumps and motors, the detrimental effects of the dead volume is much more important than in fixed displacement machines. It is therefore expected that the shuttles could result in even higher efficiency improvements than in fixed displacement machines. Further development work and testing are needed in this direction.
INNAS
innas.com
Filed Under: Components Oil Coolers, Pumps & Motors, Trending