Development of a Rubber Disc Piston Seal for the Mahadaga Handpump Peter Govey, Christopher Claassen, Joseph Longenecker Messiah College 2006 Abstract We developed a new concept for a piston seal that could replace traditional leather seals in reciprocating hand pumps. The new seal is made from a stack of rubber discs, cut from truck inner tube, backed by disc cut from flat PVC sheet. The materials and construction of the seal are very simple and while it may not last as long, we hoped to have a more accessible seal for Village Level Operation and Maintenance (VLOM) hand pumps. We achieved satisfactory performance at heads less than 20 feet, but above this the pressure tended to pull the rubber discs off center, breaking the seal. We recommend that this seal design be considered for simple suction pumps used for irrigation from shallow water tables. Background Leather is the most common piston seal material in pumps around the world and has been used successfully for hundreds of years. It has excellent wear resistance when properly tanned and it conforms well to the pump cylinder. The original piston designed for the Mahadaga pump used a leather cup sandwiched between two PVC discs. Holes drilled through the discs were covered with a rubber flap, creating the check valve. Figure 17 shows an exploded view of this piston. The wire loop is a spacer so that the piston can be tightened without distorting the rubber flap. Friction is the only thing that keeps the discs from rotating with respect to each other and closing up the holes. bottom PVC disc leather cup seal 1/4" lock nut top PVC disc flattened 4" pipe rubber valve flap 1/16" thick; 1 3/4" diam. wire loop 12 guage metal washer 3/4" diameter 1/4" screw welded to rod piston rod 1/2" steel Exploded view of original Mahadaga Pump piston, dimensions in inches In 2003, Dokimoi Ergatai Pump Project member Matt Kiehl invented an innovative new piston design that used only a single rubber flap backed by the PVC disc. The first design did not create a strong seal, but we believed that it could be improved by using more rubber discs. We tested many combinations of rubber thickness, disc diameter, number of discs, and PVC backing disc diameter in the process of arriving at our final design. This design has four rubber discs cut from truck inner tube rubber approximately 1/16 inch thick and a PVC backing disc 1.75 inches in diameter. The rubber discs are slightly larger in diameter than the pump cylinder. 1
New rubber disc piston seal prototypes This piston seal also functions as a check valve. On the upstroke the rubber is drawn down around the PVC disc, squeezing it against the pump cylinder walls and sealing tightly. On the down stroke the rubber folds up away from the PVC disc, allowing water to flow around it (Figure 19). This piston is very easy to build and uses materials widely available. Even if the wear life of the rubber discs is low, they are easy and cheap to replace. Design work on this seal is not finished. It is very sensitive to the total thickness of rubber above the backing disc, a nuance that may not be easily communicated to our clients in Mahadaga. Rubber folding around piston disc on upstroke and folding away from disc on down stroke, visible through clear pipe. Testing The four main variables in our piston design are the rubber disc thickness, rubber disc diameter, PVC disc diameter, and number of rubber discs. We wanted to test the effect of each variable, so we selected representative values for each and combined them into 30 distinct piston designs. The two types of rubber were 1mm thick bicycle tube and 2 3mm thick tractor tube. The PVC discs were 38mm and 45mm, cut using the 1.75ʺ and 2ʺ hole saws. The rubber disc diameters were 59mm (slightly larger than the pipe ID) and 63mm. The number of rubber discs was one or two less than the number that caused the piston to bind. 2
Rower Pump Tests To get a rough idea of how well our piston seals performed, we built a rower pump and measured its volumetric efficiency with each seal. This pump had a chamber made from the schedule 40 type grade of PVC pipe available in Burkina Faso. For the tests with only suction head, we used a T fitting with a simple head seal over the top of the cylinder to prevent water from splashing out. On the down stroke this piston design shoots high velocity jets of water around it. To increase the head we could test at, we built a pressure simulator attachment for the top of the cylinder. The attachment had a disc which the rod could pass through while allowing very little water. The water was forced out of a side port with a valve on it. We adjusted the valve to get the right resistance so that a comfortable pumping cadence produced the desired pressure. Volumetric efficiency is the ratio of the volume of water pumped to the volume swept out by the piston (piston area x stroke length). Evaluating pistons by this method is imperfect because it does not take into account the friction force needed to get the seal. Higher friction will often correlate with a tighter seal, but may result in a greater expenditure of energy (lower mechanical efficiency). It is possible to get volumetric efficiencies greater than 100% because the momentum of the water causes it to continue flowing when the piston stops. We added rubber discs to the piston until it bound in the pipe or had excessive friction. One less disc than this usually had a good volumetric efficiency without noticeably high friction. We calculated our volumetric output from the depth of water collected in a five gallon bucket, and the pump operator obtained a reasonably consistent stroke length by eyeing a ruler held parallel to the pump rod. Rower pump used for testing piston seals, with pressure head attachment Rower Testing Results We quickly saw that all of the pistons using the 38mm PVC backing disc performed poorly, so we decided that the disc should be as large as possible and still allow adequate clearance with the pipe wall for water to flow on the down stroke. Given the limitations of our hole saw sizes, this leaves the 45mm disc, cut using a 2 inch hole saw. This disc was used for all of the following designs. We also found that the pistons using the thin bicycle tube rubber had low volumetric efficiencies as well as a tendency to bind. We think that this may be partly due to distortion of the thinner rubber as it was drawn down over the PVC disc, preventing it from contacting the cylinder wall evenly. It did appear that larger (63mm) diameter rubber discs work better than the 59mm discs when the rubber is thin. More thorough testing is needed to make any positive conclusions about this. The most effective design (piston seal 16) used four 59mm rubber discs cut from tractor inner tube rubber. Piston seal 15 used only three of these rubber discs and piston seal 20 used four discs of diameter 63mm. From the graph of volumetric efficiency vs. head, it is apparent that using the larger diameter has a slightly detrimental effect on volumetric efficiency. This may occur because the rubber has to fold up further to let water by, dampening the upward momentum of the water faster and creating more resistance on the down stroke. As is expected, removing 3
one of the rubber discs causes the volumetric efficiency to drop off more quickly, but it also cuts the friction. We tested some designs that include various combinations of the two rubber disc diameters, and these fell somewhere in between the performance of either of them exclusively. When testing the seals at higher pressures we had some trouble with the rubber discs becoming pulled off center, reducing the efficiency of the seal. 120% Volumetric Efficiency v. Head for Rubber Piston Seal Designs 100% Volumetric Efficiency 80% 60% 40% 20% piston seal 16 piston seal 15 piston seal 20 piston seal 31 piston seal 32 0% 0 10 20 30 40 50 Head (feet) David Roncin and Ruth Kitchin built and tested another variation that attempted to avoid the main problem with the rubber piston seal: It is very sensitive to the total thickness of rubber. Their seal (#31) used four thick rubber discs 53mm in diameter and a single thin rubber disc 57mm in diameter (equal to the pipe ID). This seal had decent volumetric efficiency but noticeable friction. David and Ruth also built a simple labyrinth seal using three PVC discs sandwiched together. The outer discs were 55mm in diameter and the inner disc was 53mm. This seal had very low friction but much worse volumetric efficiency. Manufacturing processes likely contributed to the inefficiency, sine the discs had to me made by filing them to the correct diameter, thus causing inconsistency and gaps with the pump cylinder wall. The PVC disc labyrinth seal prototype, piston seal 32 4
Stationary Piston Seal Testing Device After eliminating many piston seal designs, we did additional testing in a more controlled environment by constructing a stationary piston seal test device. This device allowed the piston to be fixed in the pump chamber and pressurized with a centrifugal pump to simulate pumping conditions. The leak rate of water past the seal could then be measured Stationary piston seal test device The graph indicates that leak rate increases exponentially with increase in pressure, as expected. We have concluded, however, that this pressure test device is not very useful. More accurate results can be obtained from our durability test apparatus which provides a steady stroke to control the tests. The static piston is probably not a good simulation of actual pumping conditions, although it may give an idea of at what pressure the leak rate begins its dramatic increase. 140.00 Leak Rate of Piston Design 16: Four truck rubber 58mm diameter discs 120.00 Leak Rate (ml/s) 100.00 80.00 60.00 40.00 Piston Design 16 20.00 0.00 0 5 10 15 20 25 30 35 Pressure (psi) Leak rate measurements using the stationary piston test apparatus. 5
Performance Testing Done with the Durability Testing Apparatus. We obtained the following performance numbers during the durability testing apparatus pump. Tests are at the DTA stroke rate of 30/min with a 40 cm stroke. With a leather piston seal at 51 feet of head, volumetric efficiency was 109% and the flow rate was 1933 L/hour. With rubber piston seal 16 at 46 feet of head, volumetric efficiency was 88% and the flow rate was 1547 L/hour. We were unable to raise the pressure to the same head as used for the leather seal, probably due to the much greater leakage past the rubber seal. Subsequent testing of the rubber seal at these heads revealed major problems. Within 5 minutes of starting the pump, flow was reduced to a trickle. The rubber was found to have been pulled off center as seen in the picture below. Several repetitions resulted in the same problem, also in only a few minutes of run time. We tried tightening the nuts that hold the rubber in place, but this did not help. It seems that the pressure in the cylinder is too great for the clamping force on the rubber to resist. If the nuts are tightened too much, the rubber is deformed and will not lay flat on the PVC backing disc. This was a similar result to the earlier rower pump tests at high heads, but much more severe, probably because we let it go longer. Rubber piston seal with distorted rubber discs, resulting in failure. Conclusions We conclude from this testing that this piston seal design is workable for low head pumps within the suction limit. The rubber should be relatively thick and its diameter should be only slightly larger than the inner diameter of the pump cylinder. The PVC backing cylinder should have about 5mm clearance with the cylinder walls. The total thickness of rubber should be found for a given PVC disc and cylinder diameter; then it will be possible to build pistons using varying thicknesses of rubber without testing each one with several numbers of discs. Reliability of construction could be improved by using templates to cut out the rubber discs and a fixture to measure the total thickness of rubber. More testing should be done with piston seal 31 to see if the rubber thickness sensitivity problem can be reduced. If a simple replacement for the leather seal is possible, we suggest investigating the labyrinth seal, similar to piston seal 32 with more stacked discs. This could possibly be combined in series with the rubber seal so that the labyrinth absorbed most of the pressure while the rubber prevented the small amount of leakage that gets around a labyrinth seal. Alternate manufacturing methods will have to be developed to effectively create the labyrinth seal, such as simple injection molding or lathe turning. A final note of caution: We do not know how the rubber will wear in a PVC cylinder. This requires durability testing of the type done on the head seal of the Mahadaga Hand Pump. 6