A pressure booster station runs one pump on a VFD and two fixed pumps. Pressure swings at high demand, the lead pump runs alone beyond its range, and at night the system cycles on and off around a tiny leak. The drives work. The staging and sleep logic do not.
Why one variable pump is not enough
A single VFD pump holds pressure across its flow range by changing speed. Below a minimum speed the pump cannot build pressure reliably, and above its rated flow it runs out of capacity. In a system with wide demand, one variable pump handles the base, and fixed pumps stage in to cover peaks. The variable pump trims pressure while the fixed pumps run at rated speed.
The common setup keeps one pump variable and stages the others direct-on-line through contactors. At low demand only the VFD pump runs. As flow rises and the VFD reaches high speed, a fixed pump starts, and the VFD backs down to trim. Reverse happens as demand falls.
Staging thresholds
Stage a fixed pump in when the VFD pump runs near maximum speed and pressure still sits below setpoint for a sustained time. Use a delay so a brief peak does not start a pump. Stage out when the VFD slows toward minimum while the fixed pumps run, indicating the extra capacity is no longer needed.
If a fixed pump starts the moment pressure dips, the system hunts: pumps start and stop as the VFD overshoots. The threshold, delay, and the speed at which staging occurs must account for the recovery time rather than reacting to every transient.
Alternation and wear balance
Rotate which pump leads so one machine does not accumulate all the hours. Alternate by run time or each start. Balanced wear also means every pump is exercised; a standby pump that never runs can seize or hide a fault until it is needed. Log run hours and starts; the alternation logic should prefer the least-used pump rather than a fixed sequence.
Sleep at low demand
At night, demand can fall below the minimum flow the variable pump produces without overshooting pressure. Running the pump in that condition causes rapid cycling and wastes energy. Sleep logic stops the pump when it runs at minimum speed and pressure holds, indicating no meaningful draw. The system rests on the pressure tank until pressure falls to a wake level, then the pump restarts.
Size the pressure tank for the sleep cycle. A small tank means pressure falls quickly on a small leak and the pump wakes often. The draw-down volume between wake and sleep pressures should cover the acceptable interval. Raising the tank pressure band too wide just moves the variation to the users; keep the band within the pressure tolerance.
Dealing with leaks
A continuous small leak prevents sleep because pressure never holds. The station then runs the lead pump at minimum speed for hours or cycles repeatedly. Fixing leaks is the real solution; the controller cannot distinguish a leak from genuine low demand. Some systems use a low-flow detection that alarms rather than running indefinitely, which flags the leak instead of masking it.
Check valves and water hammer
Fixed pumps starting into a common header need check valves so they do not back-drive an idle pump. Staging a pump against an open check valve too fast creates a pressure spike. Soft transition, controlled valve timing, and small air vessels absorb the step changes. Water hammer at pump start and stop is often a staging and check-valve problem rather than a drive fault.
Sensor placement
The pressure transmitter should sit on the common discharge after the pumps, away from the turbulence of a single pump outlet. A sensor too close to a starting pump reads a transient and false-stages the others. Keep the sensing point representative of header pressure and isolate it from vibration.
Pump affinity and efficiency bands
Centrifugal pumps follow the affinity laws: flow scales with speed, pressure with speed squared, power with speed cubed. A pump slowed too far moves away from its best efficiency point and may not overcome static head at all. That sets the real minimum speed, which is often higher than the drive’s minimum frequency. Staging a second pump is better than running one pump below its useful range, where it heats and wears without delivering flow.
Match the pump curve to the system curve. A pump chosen for peak duty can be oversized for normal demand, and the VFD corrects that only within the curve. If the system has high static head, slowing the pump reaches a point where pressure collapses suddenly; stage before that point rather than chasing it with the drive.
Multiple variable pumps
On larger stations all pumps run on VFDs and share load, staging whole pumps as demand rises. This gives smooth control and redundancy, but the pumps must be synchronized to common pressure and their speed-sharing logic must avoid one pump taking the load while others idle. It costs more in drives and control. For smaller booster sets, one variable pump with fixed stages is simpler and adequate; full variable staging earns its cost where demand is large and continuous.
Fault and redundancy handling
If the lead VFD faults, the controller should hand lead duty to another pump or start a fixed pump under limited control rather than leaving the station down. Define behavior for a failed transmitter, a pump overload, and a loss of suction. A pressure station that depends on one sensor or one pump with no fallback fails the moment either does. Test these fault paths, not just the normal staging sequence.
Tuning the pressure loop
The pressure PID is slower than a position loop and must not react to every spike from a valve opening elsewhere. Tune with realistic draw changes, allow the pressure tank to absorb small fluctuations, and avoid aggressive integral action that winds up during staging. Watch the response when a fixed pump steps in; the VFD must back down smoothly rather than fighting the added flow.
Bottom line
Use the VFD pump to trim pressure and stage fixed pumps on delayed speed and pressure thresholds. Alternate lead pumps to balance wear and exercise standby units. Sleep on low demand with a properly sized pressure tank, and treat persistent no-sleep conditions as a leak alarm. Smooth staging depends on thresholds, check valves, and sensor placement; tuning the drives alone will not stop pressure swings and night cycling.
