Self-priming pump that primes slowly, loses flow and short-cycles
A bungalow lifts water from an underground sump to a 1,000-litre HDPE tank on the roof with a 0.5 HP self-priming monoblock pump. The suction line is 25 mm GI: 5.5 m vertical from the lowest sump water level to the pump axis, plus a 4 m horizontal run, with a foot valve and three elbows. On its way it rises over a 600 mm plinth beam, creating a high loop above the pump. Delivery is 20 m of 25 mm GI rising 9 m to the tank, with a gate valve and no non-return valve. The pump takes three to four minutes to prime, runs, then its flow falls away after about ten minutes and the roof tank fills very slowly. When a garden tap on the delivery is opened, the pump cuts in and out about every twenty seconds.
1. Work out the suction-side condition. Is the pump being asked to do something it physically cannot do, and what does the plinth-beam loop add?
Read the worked answer
Static suction lift. The figure that matters is from the lowest sump level to the pump axis: 5.5 m. At Jaipur's altitude (about 400 m) atmospheric pressure is roughly 10.3 − 0.5 ≈ 9.8 m of water, so the absolute ceiling is under 10 m - but that is a laboratory figure. A real reciprocating-free centrifugal impeller cannot pull within several metres of it: dissolved air comes out of solution and the water cavitates. Practical limits are around 7–8 m for a good pump and 6–7 m once friction is counted.
Add the friction. The 4 m horizontal run plus three elbows plus the foot valve strainer adds perhaps 1–2 m of equivalent loss at this flow. Effective suction side ≈ 5.5 + 1.5 ≈ 7 m. That is at or beyond the practical limit, which explains the three-to-four-minute priming and the flow that decays as the pump gets hot and air accumulates.
The high loop is worse than the friction. The crest of the loop is above the pump, so any air in the suction line collects there. Water velocity in a 25 mm line is far too low to sweep a bubble downhill against buoyancy, so the pocket grows until it breaks the water column and the pump loses prime. The loop also means the line cannot self-vent forward. Remove the loop or fit an air release at its crest; do not simply add a second pump.
Sizing note. A suction line should normally be at least one size larger than the delivery, so 25 mm suction on a 25 mm delivery is already marginal.
2. Match each symptom to its physical cause: slow priming, flow decaying after ten minutes, tank filling very slowly, and twenty-second cycling.
Read the worked answer
- Slow priming: air must be evacuated through a long, high suction line with a crest that traps it; a slightly leaking foot valve means the column has to be rebuilt from scratch every start. Priming will always be slow until the lift and the loop are reduced.
- Flow decaying after ten minutes: the trapped air pocket at the crest progressively throttles the line, or the pump begins to cavitate as the water heats and gives off vapour. A worn foot valve letting air in at the joint, or a loose union washer on the suction side, produces the same slow decline.
- Tank fills very slowly: the delivery is 25 mm over 20 m with a 9 m lift - the pump is operating far from its best-efficiency point because the total head (9 m static plus suction problems plus friction) is high for 0.5 HP; a partly throttled gate valve adds to it. Actual output should be checked against the pump curve with a bucket test rather than assumed.
- Cycling every twenty seconds: there is no pressure vessel and no non-return valve. On a small draw-off the pump reaches cut-out almost instantly, the tap takes water, pressure falls to cut-in, and the pump restarts - a handful of litres per cycle. This hammers the motor, the pressure switch and the pipework, and shortens pump life more than continuous running does. The fix is a hydro-pneumatic (diaphragm) tank plus a non-return valve on the delivery, with cut-in/cut-out set about 0.7–1 bar apart so each cycle delivers at least a minute of running.
3. Recommend the corrected arrangement, item by item, and state how you would prove it works before handing over.
Read the worked answer
Suction side
- Re-route the suction so it rises continuously from the sump to the pump with no high loop; if a crest is unavoidable, fit a small air release at the highest point.
- Increase to 32 mm suction pipe, or shorten the run; replace 90° elbows with 45° bends.
- Fit a good-quality foot valve with a strainer, at least 300 mm clear of the sump floor, and check every joint and union on the suction side for air tightness - smear joints with soap solution while running and watch for a change in priming.
- Better still, if the sump allows, mount the pump lower or use a submersible, so lift is reduced to 2–3 m.
Delivery side
- Fit a spring-loaded non-return valve immediately after the pump, then the gate valve.
- Add a diaphragm pressure tank sized so that the switch differential gives at least one minute between starts; set cut-in below the static head needed to reach the roof tank.
- Check the duty point against the pump curve and consider a 0.75 HP unit if total head really is around 15–18 m.
Proving it
- Bucket-and-stopwatch test the delivery flow at the roof tank.
- Time a full cycle at a small draw-off (should be over a minute).
- Record suction vacuum if a gauge is available; a vacuum much above about 0.6 bar indicates the lift is still too high.
- Run for an hour and confirm the pump does not lose prime or overheat, then hand over with the settings recorded.