Surge Drum Level Loop Amplifying an Upstream Cycle
A feed surge drum (6 ft ID, 8 ft between level taps) receives 90–150 gpm of reactor feed in a roughly 5-minute cycle from an upstream unit. Its outlet goes to a fired heater and then to a reactor; average throughput is 120 gpm. The level controller is a PID with proportional band 10% and integral 0.5 min driving the outlet valve. Outlet flow swings between 60 and 190 gpm, the charge pump trips on low flow (below 80 gpm) about twice a week, and the heater outlet temperature swings ±25 °F. The pump has been checked and is healthy, and the level transmitter and flow meter have been calibrated. Operations wants to install a larger pump and a bigger heater.
1. Quantify the drum's holdup and calculate the level swing if the outlet were held constant.
Read the worked answer
Volume
A = π(3 ft)² = 28.27 ft²
V = 28.27 × 8 = 226 ft³ = 226 × 7.48 = 1,692 gal
Holdup at the 120 gpm average = 1,692/120 = 14.1 min of inventory between taps.
Level swing with a constant outlet
The inlet runs at 150 gpm for half the 5-minute cycle, accumulating:
ΔV = (150 − 120) gpm × 2.5 min = 75 gal (and −75 gal on the other half)
As a fraction of the span: 75/1,692 = 4.4% → 0.044 × 8 ft = 0.36 ft, about 4.3 in.
So a completely uncontrolled outlet would let the level wander only ±4 inches. The drum is an excellent surge volume; the 8 ft span is over 20 times the disturbance it must absorb. The drum is not the problem and does not need to be bigger.
2. Explain why the outlet swings 60–190 gpm when the inlet only moves 90–150 gpm, and what the tuning is doing.
Read the worked answer
The loop is passing the disturbance straight through and adding to it.
With a 10% proportional band, a level error of 4.4% of span produces 44% of full controller output. The controller is effectively tuned to hold the level to a fraction of an inch, which is neither necessary nor possible given the flow noise and the valve's response. Integral action at 0.5 min repeats this every cycle, and the combination of a fast integral and the transport lag through the drum and valve typically produces a decaying oscillation of period comparable to the disturbance - so the 5-minute upstream cycle is re-transmitted downstream with added amplitude and phase lag.
The numbers confirm it: outlet swing (±65 gpm) is larger than inlet swing (±30 gpm). The controller is the amplifier. The heater then sees this flow swing and produces the ±25 °F outlet temperature swing, which is a symptom, not a separate fault.
The pump trips because it is being asked to follow a 60-to-190 gpm square wave. A larger pump and a bigger heater would cost capital and leave the oscillation exactly where it is - probably worse, since a larger pump swings more flow per unit of valve travel.
3. Recommend a control strategy and predict the resulting performance.
Read the worked answer
Retune the level loop as an averaging (surge) controller.
- Proportional band 150–200%, not 10%.
- Integral time 5–10 min (or P-only with a slow manual reset).
- No derivative - level is already noisy and derivative will amplify it.
With a 200% PB, the same 4.4% level error produces only about 2% of valve output change, so the drum absorbs the upstream cycle and delivers a nearly steady flow to the heater. The level will drift a few percent of span around setpoint - which the drum can easily afford, since the worst-case excursion with no control at all is 4 inches.
Watch the valve range. With the outlet nearly constant, the valve sits near a fixed position; make sure that position is not at the bottom of its range where a small integral action causes a large flow change (a characteristic or split-range problem). If it is, consider a level-to-flow cascade: the level controller trims the setpoint of an outlet flow controller. The flow controller gives smooth valve action, and the level controller only moves the setpoint slowly.
Also:
- Put a 5–10 s delay on the pump low-flow trip so a single valve stroke cannot shut the unit down. The trip is protecting the pump from a dry run, not from a 2-second dip.
- Do not buy a larger pump or heater until the loop has been retuned and observed for a week. Expect the heater outlet swing to fall to a few degrees, and the pump trips to stop.