Distribution Pressure Control

What the station is trying to do

A distribution system needs pressure in a band: high enough at the worst point for fire flow and for the customer on the hill, low enough that it does not burst old mains or blow past pressure reducing valves. Demand changes continuously, from the night minimum to the morning peak to a hydrant flow, and the pump station has to deliver whatever the demand is while the pressure stays put. Variable speed pumping is how that is done without a storage tank floating on the system.

The basic loop

A pressure transmitter on the discharge header, downstream of the check valves, measures the controlled variable. A PID controller compares it with the setpoint and its output becomes the speed reference to the lead pump drive. Demand rises, pressure sags, the controller raises speed, the pump delivers more, pressure recovers. The loop is straightforward; most of the difficulty is in everything around it.

The setpoint is worth thinking about. Pressure at the pump discharge is easy to measure but it is not what the system needs; the system needs pressure at the critical point, which is the highest or most distant customer. Friction loss between the station and that point grows with flow, so a fixed discharge setpoint over-pressures the system at night and under-pressures it at the peak. Two approaches fix this: measure at the critical point over telemetry and control on that, or raise the discharge setpoint with flow using a curve that approximates the friction loss.

Speed limits

Minimum speed

Below some speed, usually around 40 to 60 percent depending on the pump curve and the static head, the pump cannot generate enough head to open the check valve and it churns water with no flow. That heats the water and the pump. The controller output is clamped at the minimum speed, and the sleep logic takes over below it.

Maximum speed

Normally 100 percent of the motor rated speed. Running above rated speed on a drive is possible but the power required rises with the cube of speed and the motor and pump are not rated for it without engineering review.

Ramp rates

The drive accelerates and decelerates at set rates. Too fast and every speed change is a pressure transient that the loop then chases. A few tens of seconds from minimum to maximum is common for distribution pumping.

Staging pumps

One pump covers a range of demand. Beyond it a second pump has to run, and the logic that adds and removes pumps is where most pressure stations misbehave.

1. 1

   Stage up.

   When the lead pump has been at maximum speed for a set time and pressure is still below setpoint by a set margin, start the lag pump. Bring the lag pump up to a matching speed and then let the controller drive both together, or run the lead at full speed and control on the lag; each has advocates and both work when the transitions are managed.

2. 2

   Stage down.

   When the running pumps have been at or near minimum speed for a set time with pressure at setpoint, stop one. Raise the speed of the remaining pump as the other stops so that pressure does not dip.

3. 3

   Alternate.

   Rotate which pump is lead on each cycle or each day so that wear is shared and every pump is proven regularly.

4. 4

   Protect the transitions.

   Delays on both stage decisions, a deadband between the stage up and stage down conditions, and a minimum run time on each pump. Without these the station adds and removes a pump every few minutes at the demand that sits on the boundary.

Sleep and small demand

At night the demand may be less than the flow of one pump at minimum speed. The pump then runs at minimum, pressure rises above setpoint, and the loop has nowhere to go. The standard answer is a sleep mode: when the pump has been at minimum speed with pressure above setpoint for a delay, stop it, and restart when pressure falls to a wake setpoint. Something has to hold pressure while the pump is asleep. A hydropneumatic tank does it with stored air; a small jockey pump does it with a pump sized for the leakage. Without one or the other the main pump short cycles all night.

Why pressure loops hunt

A hunting pressure loop swings above and below setpoint continuously, and the pump speed swings with it. It wears the drive, disturbs the system, and trips low and high pressure alarms. The causes are few and they are almost always in this list.

• Tuning too aggressive. Pressure responds to speed almost instantly, so a loop tuned like a slow temperature loop overshoots. Reduce gain, lengthen the integral time, and use no derivative unless there is a specific reason.
• Transmitter too close to the pump. Immediately downstream of the pump the pressure is turbulent and follows every speed change. Move the transmitter downstream of the check valves and a length of straight pipe, or filter the signal.
• Ramp rate faster than the loop. Every controller output change becomes a pressure step, which the controller then corrects. Slow the ramp.
• Staging without delays. The second pump starts on a dip, raises pressure, stops on the rise, and the cycle repeats. Add delays and deadbands.
• Integral windup at the speed limits. When the output is clamped at minimum or maximum the integral term keeps accumulating, and when the limit releases the controller overshoots. Use a controller with anti-windup, which every modern PLC PID instruction has, and confirm it is enabled.
• A pressure reducing valve or an altitude valve downstream fighting the station. Two controllers holding pressure at two points with overlapping ranges will oscillate against each other. Separate their setpoints.

Constant speed stations

Not every station has drives. A constant speed station holds pressure in a band with pressure switches or a transmitter and setpoints, starting a pump when pressure falls to the low setpoint and stopping it at the high setpoint, with a hydropneumatic tank sized to keep the cycles per hour within the motor rating. The control is simpler and the pressure varies more. The staging, alternation, and signal failure rules are the same.