12.4 Final Element Testing and Commissioning

Topic 3.O is testing: loop test, calibration, simulation of values, diagnostics, stroke, fail-action, and limit-switch setting. On the PE Control Systems exam this is still Domain 3 (final elements), not Domain 5 (safety systems). You will see partial-stroke language that overlaps SIS valves. Keep the distinction: here you are commissioning a BPCS valve, damper, or solenoid package and proving the mechanical and I/O path. IEC 61511 proof-test coverage, dangerous-undetected failure fractions, and test-interval math are later chapters. Do not import SIL arithmetic into a cold-loop punchlist.

Loop test, calibration, range, simulation, diagnostics

Commissioning usually runs from paper to motion:

1. Walkdown — tag, flow direction, air supply, tubing, packing, mounting, heater or tracing if specified.
2. Cold loop check — continuity, polarity, shield landing, analog 4–20 mA path from the output card to the I/P or positioner, discrete path to solenoids and from limit switches. This proves wiring and assignment. It does not prove the spring.
3. Calibration / range verification — apply known inputs and record outputs. A five-point check (0, 25, 50, 75, 100%, or 4, 8, 12, 16, 20 mA) is the usual BPCS analog pattern. Calibration adjusts; verification documents as-found/as-left. If you only force 12 mA in the DCS graphic, you have not verified the field converter.
4. Simulation of values — name what you bypassed. A DCS force of the PV tests logic with the field disconnected. A calibrator at the marshalling cabinet tests cable plus I/O. A communicator at the positioner tests the last meters. Write that on the loop sheet or you will sign for a test you did not run.
5. Diagnostics — modern positioners report travel deviation, supply pressure, actuator pressure, and stick-slip. A healthy diagnostic is a baseline, not a fail-action test and not a PSV.

Stroke tests: full versus partial (BPCS first)

A full-stroke test commands 0–100% (or open/close for discrete), times the stroke, and watches for sticking, overshoot, and seating. On a BPCS valve that can be taken off-line during an outage, full-stroke is the honest test: you prove travel, packing, air-supply adequacy, and often the seat. Stroke time is loop dead time for analog control; it is also a specification number on many on/off valves.

A partial-stroke test (PST) moves a small fraction of travel (often about 10%–20%) and returns. It proves the valve is not stuck at the current position. It does not prove full seating, full Cv, limit switches at the far end, or the fail-action spring all the way home. PST is common on SIS ESD valves that cannot be full-stroked in service. For a BPCS control valve you can isolate, do not substitute PST for a full-stroke just because the phrase appears in a vendor brochure. If the exam stem is an SIS proof test, that is Domain 5; if the stem is a control valve during commissioning, full-stroke during the outage is the expected BPCS answer.

Positioner calibration, fail-action, limit switches

Positioner calibration teaches the instrument the actual 0% and 100% stops (auto-stroke / auto-tune), then you verify 4 mA, 12 mA, and 20 mA against stem travel. Characterization (linear versus equal-percent) is a positioner setting. It is not a property of the actuator spring. If the cam or software characteristic is wrong, the loop gain you tuned in the DCS is wrong even when 4 mA and 20 mA hit the stops.

Fail-action test: remove the thing the fail position depends on, and watch the stem. For pneumatic fail-closed, bleed the air (or de-energize the dump solenoid) and confirm the valve closes. Pulling the DCS output to 4 mA while the positioner still has supply air is a commanded close, not an air-fail test. For electric actuators, kill power (or the analog command, depending on the specified fail) and confirm FC, FO, or fail-last as designed. Fail-last on a modulating electric operator is a real specified state; do not assume spring-return.

Limit switches are set at actual end of travel after the valve has been stroked, not at a drawing dimension and not at 50% unless that is the specified trip. Open feedback should pick up when the valve is open; close feedback when it is closed. Using a limit switch as the only proof of fail-action is circular if you never removed air.

Worked example: simulated 12 mA versus installed travel, bench versus installed spring

An air-to-open globe valve has an actuator bench-set 6–30 psig: on the bench, with no packing pinch and no process load, the stem starts to move at 6 psig and reaches rated travel at 30 psig. A calibrated positioner on the bench takes a 12 mA command and puts the stem at 50% travel. That is the unloaded signature.

Installed, the same valve sees 200 psig upstream. Packing was tightened after a leak. A technician simulates 12 mA at the marshalling cabinet (loop-powered positioner).

• Healthy positioner in AUTO: the stem should still go to 50%, because the positioner raises actuator pressure above the bench midpoint until travel matches command. Diagnostics should show higher than expected actuator pressure. Simulated mA plus actual stem travel plus a diagnostic match is a real installed check.
• Direct I/P onto the 6–30 spring, no positioner: people still map 4–20 mA to 3–15 psig. Midscale 12 mA then loads about 9 psig. Installed travel is about (9 − 6) / (30 − 6) = 12.5%, not 50%, even before packing and unbalanced plug forces. The DCS thinks 50%. The process sees a nearly closed valve. Bench set was never the installed characteristic.
• DCS graphic forced to 50% with the analog output still at 4 mA in the field: you simulated a controller value, not a valve. Loop logic may be fine; the final element was not in the test.

The PE point is mechanical: simulated current proves the electrical slice you connected to; stem travel proves the mechanical slice; bench spring range is an unloaded number. Process unbalanced forces and packing shift the installed signature. A positioner can hide that shift from the DCS percent display while still telling on itself in actuator pressure. Checkout that never looks at the stem is not checkout of the final element.