Centrifugal compressor anti-surge: mapping, sizing and tuning

Centrifugal compressor surge is flow reversal within a few cycles of the unstable operating point. It loads the thrust bearings, rubs seals and impellers, and can wreck the machine before anyone sees an alarm. The anti-surge valve is the last active barrier. If it is too small or too slow, the machine is exposed. If it is grossly oversized, the loop hunts, the process gets kicked at every recycle event, and the valve and piping pick up noise and vibration. Emerson's sizing guidance notes that anti-surge valves are typically grossly oversized for common service conditions. This article covers how to map the surge line, size the recycle valve for the duty it actually performs, and tune the loop so the valve opens before the compressor reaches the limit.

Standards and requirements context

  • API 617 covers axial and centrifugal compressors and expander-compressors for petroleum, chemical and gas industry services. It defines the vendor's performance curves and the stated surge point, which is the starting data for any surge line.
  • API 670 covers machinery protection systems. Vibration, axial position and temperature trips work independently of the anti-surge controller, and they have to be coordinated with it.
  • IEC 61511 applies if the anti-surge function, or a separate trip that depends on it, is assigned a safety integrity level. In that case the valve, solenoids, transmitters and proof-test intervals form a safety instrumented function and each needs its own verification.

Most installations run anti-surge as a basic process control function, with machinery protection and process shutdown as separate layers. Settle where that boundary sits during design, not at commissioning.

Surge line mapping

Source data and its limits

The vendor's predicted surge limit is a starting point, not the final answer. Where you can, use a field surge test or a shop test with the same gas family. Test points taken near the limit will show whether the predicted line is conservative or optimistic.

Coordinates that make the line portable

A surge line drawn in raw flow and pressure coordinates moves with suction pressure, temperature, molecular weight and speed. So controllers use reduced or invariant coordinates, where the limit collapses to a single curve across operating conditions. Whatever coordinate system your vendor uses, check three things:

  1. The flow measurement sits in the same location and on the same basis as the one used to build the line.
  2. Gas composition changes are handled by a molecular weight or compressibility input, or the line is built conservatively for the worst case.
  3. Suction conditions are measured at the compressor, not at a remote point with its own dynamics.

Surge control line and margin

The surge control line (SCL) sits on the stable side of the surge limit line (SLL), and the controller acts when the operating point reaches the SCL. A wider margin buys the valve more time, but it forces more recycle and wastes more energy. So the margin should come from system dynamics, not habit. Treat the number here as a screening value. The actual margin has to come from a dynamic assessment of your own system.

A second, faster protection line, often called a recycle trip line or safety-on line, sits between the SCL and the SLL. When the operating point crosses it, the valve steps open by a fixed amount regardless of what the PI loop is doing. That covers fast disturbances, like a downstream valve closing or a driver trip, which a PI loop simply can't keep up with.

Recycle valve sizing

Size for the duty, not the nameplate flow

Honeywell describes the recycle loop as the valve plus the pipework upstream and downstream of it, circulating gas from discharge back to suction (Honeywell, 2022). The valve has to pass enough flow to return the compressor to a stable point under the worst credible conditions. It is not sized for normal operating flow. That range is a screening check only. Rigorous sizing starts from the surge flow at the relevant conditions.

A sound sizing sequence:

  1. Identify the operating cases that produce the highest surge flow. These are usually the highest speed or head case and the lightest or heaviest gas, depending on the machine.
  2. For each case, determine the recycle flow needed to hold the operating point at or beyond the SCL. Include a system-resistance estimate that shows where the operating point lands once the valve opens.
  3. Calculate the valve pressure drop from the actual discharge pressure, cooler and piping losses, and the suction pressure at that recycle flow. In a closed loop with a cooler, suction pressure rises as recycle builds, so don't use the starting value.
  4. Compute the required flow coefficient with the gas sizing equations in your control valve standard, using real gas properties and checking for choked flow.
  5. Select the valve and trim so the required capacity falls inside the controllable part of the valve characteristic at the normal operating point, and the valve still has margin left in the extreme case.
  6. Check start-up, shutdown and settle-out duties.

Why oversizing hurts

An oversized valve spends nearly all its life at a very small opening. Positioner resolution, stiction and the trim characteristic then take over, and small controller moves produce large flow changes. The result is limit cycling, process upsets from a valve that is "barely open" but already passing significant flow, and more noise.

Velocity and noise

Check velocity downstream, in the pipe and at the valve outlet, not only at the trim. If the calculation flags a problem, specify noise-attenuating trim. Then confirm the piping supports and downstream tie-in can handle the dynamic loads.

Time delay in sizing

The Seoul National University work proposes a graphical method for sizing recycle valves that explicitly includes time delay (Oh, 2015). The point is practical. A valve that is large enough in steady state can still be too small if gas has to travel through a big volume before the effect reaches the compressor. You can't separate sizing from response time.

Dynamic response

Valve and actuator requirements

Together these give you a window: fast enough to beat surge, but not so fast that the downstream pressure and flow transient shoves the operating point somewhere else.

Hitting the stroke time is a system problem:

  • Size the actuator and supply volume for the stroke under worst-case differential pressure, not just the bench test.
  • Use boosters or quick-exhaust devices, and keep tubing short and large enough.
  • Fit a positioner suited to rapid opening, and check that its tuning isn't slowing the stroke.
  • Make fail-open the default on loss of signal or air, with the compressor trip logic aware of the valve state.
  • Place the check valve and the recycle take-off so the check valve doesn't trap volume that delays the effect of recycle.

Where the time goes

Total response is the sum of several delays. Count all of them:

Element What it adds Practical lever
Flow transmitter and scan time Detection delay Fast transmitter, short scan, minimal filtering
Controller algorithm Computation and output delay Dedicated surge controller, fast execution
Signal path and solenoid Delay before the actuator moves Direct wiring, quick-exhaust, avoid slow relays
Actuator and valve stroke Mechanical delay Booster, adequate air supply, correct sizing
Piping inventory Delay until recycle gas reaches the compressor Short recycle line, take-off close to discharge, valve close to the take-off

Controller tuning

Anti-surge control doesn't tune like a flow loop. The practical approach has three parts:

  • PI loop on the SCL. It maintains margin during slow changes. Keep the gain moderate and the integral action strong enough to recover, but avoid overshoot that opens the valve more than necessary. Many controllers use a nonlinear or gap-based response, so the loop reacts harder the closer the operating point gets to the SCL.
  • Recycle trip step. Set the step size and decay from open-loop tests. The step has to be large enough to stop the approach to surge, and the decay has to let the valve close again without chattering.
  • Valve closing. Closing should be slower and stable, with logic that prevents re-entry into the trip zone.

Validate by stroking the valve in the field and recording the response. Then approach the SCL in a controlled test, with the operating team on watch. Moore and Brun (2018) lay out the controller functions and tuning logic in tutorial form that a control engineer can follow.

Illustrative scenario (guidance only, not a sizing calculation)

A two-stage export compressor on a gas plant has a recycle valve that was sized from nameplate flow, and it sits at a very low opening in normal running. Operators see slow oscillation in suction pressure and frequent small recycle events. A review would go like this:

  1. Compare the installed valve capacity with the recycle flow required at the worst surge case. That shows whether the valve is oversized in the control range.
  2. Check the valve position at normal running. If the valve is nearly closed and the loop hunts, suspect positioner resolution and the trim characteristic.
  3. Stroke-test the valve and measure the opening time against the design requirement.
  4. Review the recycle piping. A long run between discharge and valve points to a need for more margin or a modified layout.
  5. Retune the PI loop and recycle trip step after any hardware change, then repeat the controlled approach test.

The fix might be a smaller trim, a different characteristic, or a revised margin. The data decides. A general rule won't.

Practical checklist

  • Confirm the source of the surge limit line: vendor prediction, shop test or field test, and the gas used.
  • Verify the flow measurement location and the coordinate system the controller uses.
  • Document the basis for the SCL margin, including piping inventory effects.
  • Size the recycle valve from the worst-case surge flow, using real gas properties and choked-flow checks.
  • Check valve position at normal operation, so it isn't running near the seat.
  • Confirm outlet velocity and noise against the design limit, including the 0.3 Mach design target.
  • Confirm stroke time under actual differential pressure, and fail-open action on loss of air or signal.
  • Review the check valve and recycle line layout for trapped volume.
  • Test the recycle trip function and record the response.
  • Keep a record of tuning values and test data for future comparison.
  • For any work that opens the recycle loop or valve, isolate and depressurise the section, verify zero energy, apply lockout/tagout, and observe hazardous-area rules. Use gas detection and vent to a safe location before breaking any joint.

Conclusion

Anti-surge performance rests on three things being right together: a surge line that matches the real machine and gas, a valve sized for the duty it performs, and a loop that acts before the compressor reaches the limit. Start by checking where your surge line came from and where the installed valve sits in its operating range. Then stroke-test the valve and compare its response with the delay budget. If margins have been inflated to cover slow response, fix the response first and then recover the energy.