Energy

Retro-Commissioning an Ammonia Plant: The Rules of Thumb Worth Stealing

By the operator · October 30, 2026

Most of the energy savings in an old ammonia plant aren’t in new equipment. They’re in setpoints somebody picked years ago and nobody has questioned since.

I run an underground freezer. It’s a two-stage ammonia plant feeding a calcium chloride brine loop, with rooms held anywhere from about -15 to 8 °F. This year the plant went through a retro-commissioning study (RCx) through our utility’s efficiency program. An outside refrigeration engineering team came in, looked at how the plant was actually being run, changed some things on the spot, and wrote up the rest.

The most useful thing I got out of it wasn’t the report. It was a short list of rules of thumb for what each kind of change is worth. If you run a plant like this, you can use them to decide what to look at first.

The rules of thumb

These are the study’s estimates, not numbers we measured. Treat them as a way to rank where to look, not as a promise.

Add it all up and the study estimated the package at roughly 10% of the site’s electricity. That’s a rule-of-thumb estimate, not a measured result, but for a plant that runs around the clock it’s a real number, and most of it is settings, not hardware.

The pattern worth noticing: the two biggest levers are pressure. Every degree of lift the compressors don’t have to make is energy you don’t buy. Head pressure and suction pressure are the first two things I’d look at in any plant.

What each one asks of you

None of these are free in the sense of “turn the knob and walk away.” Each one has a reason it was set where it was.

Lower head pressure. The cooling towers or condensers have to be able to hold the lower setpoint, and that’s hardest in summer. Summer head-pressure alarms are already part of life at our plant, so any reset has to be checked against the worst week of July, not an average day.

Raise suction. The coldest room sets the floor. If one room needs to hold around -15 °F, you can’t raise the suction for the whole system past what that room can live with. Check the coldest room and the coils that struggle most before you move it.

Stage the condensers. Running every fan and tower flat out all year is simple, and it’s also paying for capacity you don’t need on a cool day. Staging them to match load is one of the bigger percentages on the list.

Compressor sequencing. Which machine runs first, which loads next, and how fast they load and unload. On a mixed fleet this matters more than you’d think.

Defrost settings. Defrost puts heat back into a room you just paid to cool. Defrosting more often or longer than the coils need is waste you won’t see unless you look for it.

Doors and infiltration. About 1% sounds small, but it’s also the one you can see. Every open door and every leak in the ceiling is warm, wet air the plant has to deal with.

A sequencing lesson from the same spring

Around the same time as the study, we had a controls problem that shows why sequencing is on the list. When rooms reached temperature and the load fell off, the suction pressure dropped fast. The compressors couldn’t unload quickly enough to follow it, and the plant ended up shutting down.

The problem was the timer. The controls wouldn’t let us set the load and unload timers shorter than one minute. We asked to have them changed to seconds. Once that was done, along with a fix to a sequencing glitch the controls tech found along the way, the plant ran smoother and responded quicker.

The lesson is the same one I keep learning: check the units and the resolution of every timer in your controls. A setting that can only be adjusted in minutes can’t keep up with a pressure swing that happens in seconds.

How the study actually ran

If your utility offers a refrigeration RCx program, here’s roughly what to expect.

1. Application and a pre-visit call. They want to understand the plant before they show up.
2. Homework. They asked for equipment schedules and P&IDs. We didn’t have P&IDs. We had two drawings and an equipment list, and they filled in the rest on site. Have whatever you’ve got ready, and don’t be embarrassed if it’s thin. Ours was.
3. Site visit. A tour and an opening meeting, then time on the controls and in the engine room. Some changes were made right then.
4. Closing meeting. A written summary of every change made on site and the next steps.
5. Final report, about three weeks later. The full list of measures, split into two groups (more on that below).
6. Check-in calls and verification. Regular follow-ups, then measurement and verification before the savings are signed off. Months later we were still closing out that paperwork, so plan for it to take a while.

The study’s estimated savings came in well above the estimate it was approved on. That’s not a guarantee for your plant. It does suggest that a plant nobody has looked at with fresh eyes in a while probably has more in it than you’d guess.

Two buckets: settings first, capital second

The study split everything into two groups.

• No-cost and low-cost measures, the ones that pay back in about two years or less: condensing pressure, suction pressure, evaporator fan control, compressor sequencing and defrost settings, plus a look at doors and infiltration.
• Capital projects, with longer paybacks: variable frequency drives on fans, compressors and pumps, piping and valve changes, and controls upgrades.

The advice in the final report was simple: keep the changes already made the way they are, and go after the remaining low-cost, high-savings project next. For us that turned out to be lighting, with occupancy sensors or schedule timers. Not glamorous, but it’s the kind of thing that’s easy to overlook in a building where the lights are on all the time because nobody thinks about them.

What I’d tell another operator

• Start with pressure. Head and suction are the biggest levers on the list. Know where yours are set and why.
• Find the room that sets the floor. Before raising suction, know which room and which coils would complain first.
• Check the summer worst case. Any head-pressure reset has to survive the hottest week, not the average one.
• Audit your timers. Load, unload and defrost timers: check the units and how fine you can set them.
• Log before and after. If you change a setpoint, write down what it was, what it is now, and when. Your daily log is the only way to know if it helped. Our daily, weekly and monthly checklist is a good starting point.
• Get your paperwork in order before the visit. Equipment lists and drawings save the engineers time on site, and that time is better spent on your controls.

Related: The Defrost Game, on staggering defrosts so the compressors don’t choke, 1:20 Isn’t 120, another time-unit lesson from our controls, Demand Response From an Underground Freezer, and Snowstorm Behind the Racks, on the infiltration you can see.

← All posts