Is that reservoir too big?

By Jack L. Johnson

This question is prompted as a result of a power unit we built to supply power to the motion control lab at IDAS Engineering. Some of the requirements involved in its design were portability, small physical footprint, engine driven, 3- micron high pressure absolute filtration and a small reservoir. Moreover, we wanted a transparent reservoir so that the degree of turbulence could be evaluated under all circumstances. The power unit is seen in a side view in Figure 1

Limitations of space dictated the small reservoir size. The pump is capable of delivering over 30 gpm. However, engine speed and pump displacement are set so that steady state pump flow is 17 gpm, with surges of as much as 40 gpm in the return line. A schematic diagram is shown in Figure 2.

The reservoir started with a 5-gal tank tapped for fittings for the pump inlet and the return lines. We had the reservoir top carefully cut off, and the sides were extended with oil-resistant, transparent plastic. A transparent plastic cover completed reservoir fabrication and assembly.

As shown in Figure 1, the reservoir is atop the power unit. The plastic extensions increased total volume by a factor of about three. Conventional wisdom for reservoir sizing is generally accepted as being three times the per-minute flow through the reservoir. Using this criterion, our reservoir would have to be 51 gal, almost as big as the entire power unit. This was unacceptable at design time, so we decided to adopt the design of Figure 1. Surge flow would have dictated a 120-gal reservoir, because flow surges can be over 40 gpm!

Note that two accumulators are shown in figure 2. They have a total volume of 1.5 gal. The accumulators are vital to supplying the sudden flow demands of the motion control servo system. However, they present a special problem at shut-down time. The problem is that when the manual dump valve is shifted to the open position, the accumulators discharge through the valve and back to the tank with a sudden and large surge. The faster the valve shifts, the greater the return line flow surge.

There two other accumulators in our motion control servo application. Although not shown, they are part of the servo system load and are the same sizes as the servo system. So at peak conditions, there is a total of 3 gal of accumulator charge directed through the dump valve and back to the reservoir.

Figure 3 is a quartering side view of the reservoir as the manual dump valve approaches full open. The arm and hand at the right have the valve handle near the horizontal, full open, position.

Turbulence galore

Figure 4 is a reservoir top view of the same flow condition. The turbulence in the tiny reservoir can only be described as being explosive, and intolerable in any real system, ours included! Furthermore, Figure 5 shows a top view of the oil surface after the discharge surge has dissipated, and return flow is only 17 gpm from the unloaded pump. The explosiveness is gone. However, excessive turbulence still occurs and causes some aeration of the oil. Eventually, this would result in cavitation damage and deterioration of the pump if allowed to continue. A fix was needed.

A home-grown solution

The fittings on the reservoir would not accommodate any commercially available diffuser, and our facility does not include a complete metal fabrication capabilities. We were forced to come up with a "home grown" solution. The result is shown in Figure 6. A diffuser was fashioned from a length of ordinary window screen, long enough to produce about seven layers of wrapped screen. A very thin piece of sheet aluminum was crimped over the two long ends of the screen, and the screen was rolled into a jelly-roll shape. The assembly was then forced into the reservoir, as shown in Figure 8. The sheet metal clip, shown on the table in Figure 6, and visible just below the fluid level in Figure 8, holds the diffuser in place, especially during flow surges.

In Figure 1, the return connection can be seen as the tee in the left end of the steel (black) part of the tank. The diffuser is forced into the lower part of the reservoir, Figure 7, so that the return flow enters the diffuser axially and leaves radially inside the reservoir. It is important that there be somewhat of a seal between the ends of the diffuser and the tank sides. However, the seal does not have to be perfect. The simple metal-to-metal contact between the crimped aluminum and tank sides is sufficient.

The result of deploying the diffuser is seen in Figure 8. The reduction in turbulence and explosiveness is dramatic, because only a slight ripple can be seen on the fluid surface. The condition in Figure 8 is the same that of Figures 4 and 5. That is, the flow surge momentarily peaks at more than 30 gpm. The only indication of the surge in the photograph is the flow meter in the lower right hand corner of Figure 8, which is above full scale - about 35- to 40-gpm peak flow surge. It is startling only in its blandness.

In watching the flow meter variation in action, there are only two indications that there is a flow surge at all: the fluid level in the reservoir changes due to the accumulator charging and discharging, and the return line hose jerks as the dump valve shifts. There is absolutely no turbulence in the tank. Our problem was solved.

The diffuser system has been in use in our servo and proportional motion control lab for over four years and has been not been the source of any problems. We did find, in the beginning, some strands of screen wire in the high pressure filter element, but that's what the filter is for.

The simplicity and crudeness of the diffuser construction belie its effectiveness. But clearly, we are not trying to say that this is a commercially viable design. However, some variation of it could be easily designed and fabricated. The effectiveness is probably a result of the three dimensional jelly roll impedance presented to the flow stream, the loosely wrapped screen, and the very large area offered by the thousands of mesh openings in the screen. If the wrap were too tight, the resistance would probably be so high as to jeopardize the diffuser or cause excessive backpressure. We have not, however, researched those contentions.

In the end, we have to question our early decision to add the plastic side extensions to the reservoir. Sealing the metal-to-plastic junction has been a challenge. With the diffuser, we never fill the reservoir above about the 4-gal level. The transparency is very revealing, but transparency could probably have been achieved with fewer sealing problems.

At IDAS Engineering Inc. we do question the conventional wisdom that reservoirs must have a volume equal to three times the per minute flow rate. In our case, we turned the ratio upside down. The reservoir is closer to one-third the per-minute flow (17-gpm flow rate and 5-gal reservoir). If one considers the peak flow, 40 gpm, the tank volume is one-eighth the return flow!

Recall that the accumulators presented a special problem because of the flow surge at shutdown time. In another application, a release valve and circuit would be provided that lets the accumulators discharge automatically, and over several seconds, instead of a second or less. On the other hand, as implemented, the accumulators offered a special opportunity to test the effectiveness of our home-made diffuser.

Free video for H&P readers

IDAS Engineering has produced a 20-minute training video of the system described here in action. It is available to readers of Hydraulics & Pneumatics, at no charge. It should be of interest to anyone wishing to reduce the size of reservoirs are used in servo and proportional valve applications.

To obtain a copy, email idaseng@aol.com with your complete mailing address, a brief description of your application, and one will be mailed to you. Or call 262-642-7021 to request a copy. The video is recorded on CD, not DVD, so it plays on your Windows Media Player.