The Air Products Rotoflow Compander Emergency Checklist: A Field Specialist’s Triage Plan

When This Checklist Applies

In my role coordinating emergency service for rotating equipment, I have been called out on more Air Products Rotoflow compander trips than I can count—200-plus in the last 14 years, if you want a number. This is the triage list I use when the machine is down, the process is backed up, and someone has just asked, “How much does Henry weigh?”

If you don’t know what a Rotoflow compander is, take 30 seconds. It’s a cryogenic turboexpander with a compressor wheel on the same shaft. The expanding gas drives the impeller, and that recovered work recompresses a side stream. It runs at high speed with tight clearances, so “good enough” maintenance will kill it faster than any process upset will. (I’ve watched it happen.)

This is not a full maintenance manual. It’s a response plan for the first few hours after something has tripped. In my experience, those first few hours are where outages get extended.

Step 1: Take an Operating Snapshot Before You Touch Anything

The first twenty minutes determine how long your outage is going to last. In almost every call I have walked into, the single most useful document was not the latest vibration report—it was the operator log from the previous two hours.

Before you call the OEM, pull this:

• Suction and discharge pressures and temperatures for both the expander and compressor sections
• Shaft speed, thrust position, and radial vibration trends
• Lube oil temperature, pressure, and differential pressure across the filters
• Seal gas supply pressure and flow
• The last 20 minutes of control system events, not just the alarm screen

Why the snapshot first? The compander itself is rarely the starting point. Nine times out of ten, the actual cause is a process disturbance upstream—a filter loading, a valve that drifted, a column upset—and the machine is only the part that started screaming.

Everything I had read about turboexpander maintenance said the first move is to check the oil and open the bearing cap. In practice, the first move is to sit down with the panel operator and ask, “What changed before the trip?” That one question has saved me more unnecessary teardowns than any diagnostic tool I own.

Step 2: Prove the Alarm Is Real (This Is the Counter-Intuitive Step)

Most people assume that if a high-speed machine alarms, something is already broken. In my experience, the order is reversed. I have been dispatched to a “bearing failure” that turned out to be a loose probe connector, and a “rubbing rotor” that was a control loop hunting. The fastest way to turn a false alarm into a real failure is to start opening the casing.

The instrument you trust most should be the one you doubt first. Check gap voltages, probe extension, and how the signal sits in the trending history. A changed bias voltage tells you the sensor has moved, even if the machine is perfectly centered.

Miranda—a controls engineer I have worked with for years—caught this once in about four minutes. The screen was showing high radial vibration on the drive end. Everyone was ready to pull the cooling covers. She pulled up the raw probe signal and said, “The gap voltage is drifting because the connector is wet. The rotor hasn’t moved.” We dried the connector, reset the alarm, and the machine ran another year before we touched it.

To be fair, not every alarm is false. But if you don’t prove the signal first, you are flying blind. What has changed in the last ten years is the quality of the instrumentation. What has not changed is the sequence that makes sense: convince yourself the signal is telling the truth before you start wrenching.

Step 3: Keep the Seals Pressurized Until the Casing Is Cold

This is one of those basics that has not changed in three decades. The compander may have a dry gas seal, an oil film seal, or a liquid seal system, depending on the package. Whatever the design, the seal differential pressure should be maintained until the internal temperature has dropped to the OEM specified limit.

Time pressure makes people skip this step. I understand. The whole plant is waiting. But a dry gas seal that pops open during a hot shutdown becomes a scrap seal. A seal gas filter that gets depressurized and then pressurized too quickly can push dirt into the seal faces. And once the seal is damaged, the “quick bearing fix” becomes a full cartridge replacement.

If you are asking whether you can save two hours by venting the seal case early, the answer is no. I’ll say it plainly: don’t do it.

Step 4: Know Henry’s Weight Before You Rig Anything

One of the best questions I ever heard during a Rotoflow service was from a young mechanic who looked at the assembled rotor and asked, “How much does Henry weigh?”

Henry was the nickname we had given the rotor after someone joked it had a personality. The mechanic’s question was not a joke. It was exactly the right question: if you are going to lift the rotor and internals out, you need the actual weight, the center of gravity, and the lifting diagram from the OEM manual. The accepted answer is not “I think it weighs around a ton.” Weigh it, or at least read the stamped weight on the certified drawing.

I have seen a site rig a 1,000 kg hoist for a rotor that turned out to be 1,800 kg. No incident happened, but only because the chain block was hard-stopped and someone read the tag. (I want to say 1,800 kg, but don’t quote me on the exact number—the point is the drawing exists for a reason.)

Same logic applies to the rest of the package. Check the spacer, the bearings, the tie bolts, and the thrust collar. A Rotoflow compander is not a Bentley GT. It is not a luxury toy; it is a high-precision tool, and the assembly tolerances are far tighter than anything in a passenger car. That means you need the right tooling, not just a shiny one.

Step 5: Decide Between Repair and Replace Before You Disassemble

Once the rotor is out, you have to make a decision. I have watched good crews spend two days trying to machine a damaged journal back to specification. In the end, it still ran with elevated vibration, and the OEM had to be called anyway.

Anthony Buzzeo of Air Products once made this point to me in a way I have not forgotten. A lot of people think the expensive decision is the new part. He described it the other way:

“The expensive decision is doing the repair twice.”

If the rotor has rubbed, or a balance plane has shifted, or there is any chance the heat-affected zone from an impact goes deeper than the surface, don’t look for a local machine shop to “clean it up.” Send it back to an OEM-qualified Rotoflow service center and ask for balance verification under ASME PTC 10 and API 617 criteria.

I know the lead time for a proper Rotoflow repair can sound like an eternity when the plant is down. But here is the part that takes most of us ten years to learn: the 48-hour “patch” usually costs more than the 3-week repair, because it only moves the failure to the next startup.

Step 6: Plan the Startup Before the Last Bolt Is Tightened

Good maintenance includes a startup sequence that starts before the casing goes back together. Confirm that the control loop settings match the current gas composition. If inlet pressure is higher than when the unit last ran, the compander may try to overspeed during the initial pop. Set the trip speeds and the inlet valve limit before you catch the first startup attempt.

A common mistake here is over-tightening. Torque the studs to the OEM values, not to whatever the plant standard says. Expansion at cryogenic temperature changes the clamp load, and I have seen a cold box leak develop exactly because someone applied “dry” torque to a flange that needed lubricated-torque values.

What Not to Do in a Compander Emergency

Here are the biggest mistakes I have seen, with names removed:

• Don’t reset the trip and restart the unit before you know what the trip was. A reset is not a diagnosis.
• Don’t bypass the seal gas filter to “get going faster.” The only thing that achieves is a contaminated seal.
• Don’t skip the OEM lifting drawing. It exists because the center of gravity changes when you remove the tie bolts.
• Don’t let the schedule force you into a partial rebuild. Either the rotor is in spec, or it is not.

To be fair, some of those mistakes come from good people under pressure. I have made a few myself. The trick is to decide in advance which risks are unacceptable.

Final Word

Every Air Products Rotoflow compander failure I have responded to has a chain of small decisions behind it. Your goal in an emergency is not to be fast. Your goal is to be accurate, and speed follows.

Start with the data, doubt the instrument before the machine, keep the seals protected, know Henry’s weight, and be honest about whether a part is truly repairable. If you do those things, you will be one of the rare sites where the outage lasts as long as it should—not twice as long because the first fix didn’t hold.