Ten Days to Feed: Dispositioning Walk-Down Findings
A 62-loop compressor seal-gas skid is 10 days from first feed. The mechanical completion certificate requires every safety-critical loop to be closed. Walk-down findings: 6 loops have no calibration certificate - 3 are low-seal-gas pressure switches that trip the compressor, 2 are bearing-temperature RTDs used for DCS indication only, and 1 is a local pressure gauge. Four transmitters have nameplate ranges different from the instrument index; the DCS was configured from the index, and their calibration certificates match the nameplate ranges. One of the four, PT-1102, is the seal-gas pressure transmitter: nameplate 0-150 psig, index 0-100 psig, DCS 0-100 psig, low trip at 80 psig. A junction box in the washdown area is missing 2 of 6 cover screws, its gasket is deformed, and the cable entries are unsealed. A replacement transmitter has a 6-week lead time.
1. Classify each finding as blocking or non-blocking for feed, with the reasoning behind each call.
Read the worked answer
Blocking (must be closed before feed):
- The 3 low-seal-gas pressure switches with no certificate. These initiate a compressor trip on a safety function. An uncalibrated trip is an unverified trip - you cannot demonstrate the setpoint or the repeatability. Non-negotiable.
- PT-1102 and any other range-mismatched transmitter that feeds a trip or alarm. See the next task; a range mismatch on a trip input is a latent failure, not a documentation issue.
- The junction box. An unsealed enclosure in a washdown area is a moisture path that corrodes terminals and eventually causes a failure or a false trip; an open enclosure in a classified area also fails the installation code. It is a 30-minute, low-cost fix - doing it now is cheaper than any analysis to defer it.
Non-blocking, subject to a documented open item:
- The 2 bearing-temperature RTDs (DCS indication only). No trip function, so indication can be verified during run-in with a simulator or a known ambient reference, provided the certificate is closed within the agreed window and recorded as an open punch item.
- The local pressure gauge with no certificate. Verify against a calibrated master and record it; a local gauge is indication only and not a protective device.
Rule applied: criticality of the function first, then cost/effort. Never trade a safety function for schedule; do trade documentation on non-safety indication, with a named owner and a date.
2. PT-1102 feeds the low seal-gas trip at 80 psig. Quantify the consequence of the mismatch and specify how it must be resolved.
Read the worked answer
Quantify: the transmitter is ranged 0-150 psig; the DCS span and the trip setpoint are built on 0-100 psig.
- At an actual 120 psig, the transmitter sits at 120/150 = 80% of its range = 16.8 mA.
- The DCS, scaled 0-100 psig, reports 80 psig and the low trip fires.
- So the compressor trips at an actual 120 psig instead of 80 psig. The trip exists precisely to protect the seals from low gas pressure; at 120 psig it is 40 psig late. The protection is effectively defeated while looking perfectly healthy on the DCS.
Resolution:
- Preferred: re-range PT-1102 to the index range (0-100 psig) and recalibrate. The existing certificate is at the nameplate range and is not valid for the index range, so a new certificate is required either way. Leave the DCS alone - it matches the index and the FAT is complete.
- If the process requires the higher range: then the instrument index, DCS scaling, trip setpoint in the cause-and-effect matrix, loop drawings, and the calibration certificate must all change together, through an MOC with process engineering approval and (if applicable) the SIS owner.
- Never change only one side. Changing the DCS alone or the transmitter alone is exactly how this class of latent error is created.
- Verify end to end: inject the trip value and confirm the full path - transmitter, DCS, ESD, and the actual compressor trip - then record the as-left range and setpoint.
3. Sequence the work over the 10 days. What would you refuse to accept, and what is the fallback if a certificate cannot be obtained?
Read the worked answer
Sequencing by (risk x effort) and by critical path:
- Day 1-2: junction box (fast, cheap, no process dependency) and any certificate work that needs no trip bypass.
- Day 3-5: all trip-related work in one window - the 3 pressure switches, PT-1102 re-range and recalibration, and the two switches stamped 85 psig versus datasheet revision 3 at 80 psig. These share the same requirement: a trip bypass with control-room involvement. Batch them into a single approved bypass rather than three separate ones, with one named operator on the phone and a defined time limit.
- Day 6-9: bearing RTDs verified with a simulator with the machine down (parallel, no shutdown dependency); local gauge against a master; paperwork and punch-list closeout.
- Day 10: buffer for re-verification and the end-to-end trip test.
What I would refuse: feed with an uncalibrated or unverified trip device, or with PT-1102 mismatched, on the argument that start-up will sort it out. A trip that fires 40 psig late is a hazard, and closing it later means taking a running machine down.
Fallback if a certificate truly cannot be obtained: a formal exception approved by the operations/start-up manager, listing the specific device, the compensating measures (for example an independent temporary trip device or a timed manual round with a logged reading interval), an expiry date, and a named owner. Not a verbal waiver.