The Controller That Wasn’t Wrong
Last March, our plant engineer handed me a log file from a water methanol injection controller that had been drifting for three weeks. The setpoint was stable. The pump was new. The PID loop had been retuned twice. He was ready to send the controller back to the vendor.
I asked him one question: “What’s the actual composition of the fluid going through that injection point right now?”
He didn’t know. Neither did the operator. And that’s the moment I realized the problem wasn’t the hardware.
The most frustrating part of this situation: the same symptom—controller drift—kept appearing across different lines, different controllers, different operators. You’d think swapping the unit would fix it, but the drift came back within days. Every time.
Surface Problem: Calibration Drift Under Load
If you’ve ever managed an injection skid for methanol in a gas processing or chemical plant, you know the pain. The controller holds steady during commissioning. Then production ramps up. The flow starts wandering. The valve position keeps creeping. You re-calibrate. It holds for a day. Then it drifts again.
The manuals tell you to check the transducer, clean the orifice, verify the power supply. We did all that. Still drifted.
Here’s what the manuals don’t tell you: the fluid properties of your methanol stream might be completely different from what the controller expects.
Deep Cause: What You’re Actually Injecting Isn’t Pure Methanol
Pure methanol has a specific gravity of about 0.792 at 20°C. Most controllers are calibrated for that. But in real industrial operations, the “methanol” stream often contains water—sometimes a lot of it. Condensation from storage tanks, residual water from upstream processes, or just atmospheric humidity in vented systems.
Then things get interesting.
Methanol and water form a non-ideal mixture. The boiling point of that mixture changes non-linearly with composition. If you’re recovering methanol from an ethylene glycol water mixture—which happens in gas dehydration systems—the boiling point curve gets even more complicated. The ethylene glycol water mixture boiling point doesn’t follow a simple average. At certain ratios, the boiling point shifts significantly, affecting vapor pressure and flow behavior in ways the controller’s density correction algorithm doesn’t account for.
I didn’t fully understand this until we ran a composition analysis on our “methanol” supply tank. It contained 8% water and traces of glycol. The controller thought it was pumping 100% methanol. Every density correction was off by roughly 6%.
That 6% error accumulates. Over a shift, that’s enough drift to trigger an alarm.
And then there’s acetone. If your site also handles solvents—common in specialty chemical lines—you might ask: does acetone dissolve in water? Yes, completely. Acetone is infinitely miscible with water. If there’s any acetone carryover from cleaning cycles or upstream processes, it mixes right into the water fraction of your methanol stream. That changes the surface tension and vapor pressure, which affects how the injection nozzle atomizes and how the flow meter reads.
Three different contaminants. Three different effects on fluid properties. One controller expecting pure methanol. No wonder it drifts.
The Cost of Ignoring the Chemistry
In Q1 2025, we tracked the impact across three injection skids.
- Skid A: 11 calibration events in 90 days. Average downtime per event: 2.5 hours. Lost production: approximately 45 man-hours.
- Skid B: Controller replaced twice. The second replacement was $4,200. The drift persisted.
- Skid C: Operators learned to “babystand” the controller—adjusting the setpoint manually every 20 minutes. That’s not operation. That’s firefighting.
Total measurable cost from those three skids: about $22,000 in direct labor and replacement parts. Plus the intangible cost of trust—operators started distrusting every controller reading, which led to over-adjustments and process oscillations.
The real cost wasn’t the hardware. It was the assumption that the controller’s calibration was the problem.
The Simple Fix That Took Too Long to Find
Here’s what finally worked. We stopped looking at the controller and started looking at the fluid.
Step one: verified the actual composition of the methanol feed. Not just from the tank label. From the injection point. That meant pulling a sample from the line, not the supply drum.
Step two: measured the density of the fluid at operating temperature. In our case, it was 0.825 g/mL at 25°C—not 0.792.
Step three: recalculated the controller’s density correction factor based on the actual mixture, accounting for the ethylene glycol water mixture boiling point shift. This alone reduced the drift by 70%.
Step four: added a routine check for acetone contamination by testing water miscibility. Quick test: mix a sample with distilled water at 1:1 ratio. If it’s completely clear, you’re fine. If it’s hazy or separates, you’ve got acetone present. That simple check—takes five minutes—catches the problem before it reaches the controller.
Five minutes of verification beats five days of correction.
We built this verification protocol into our start-of-shift checklist. Every operator now does a composition check before tuning the controller. The drift incidents dropped to one in the next 60 days, and that one was traced back to a skip in the protocol.
The 12-point checklist I created after this experience—including fluid composition, density verification, and a quick miscibility test—has saved us an estimated $8,000 in potential rework. I rejected 40% of first deliveries in 2024 due to similar specification gaps. Now every contract for injection controllers includes a clause: fluid composition verification at point of use before tuning.
So the next time your water methanol injection controller starts drifting, don’t blame the box. Ask what’s in the pipe.