“At least it wasn’t raining hard,” I muttered. The French operator next to me couldn’t appreciate my sarcasm. On the burner deck, we were cold, shivering, soaked to the skin. The burners had decided to die at about midnight on the fourth night of the start-up northeast of Paris. In the beginning, we had four starter coils, now we were down to one, which we protected like a child. Painfully, we moved from burner to burner, wiring and unwiring our starter until all the burners were lit. What was the cause of our lamp-lighting exercise? The NEMA 12 coils needed to be in a NEMA 4X box, although this might not have been enough given the wind that howled 50 ft. in the air.
One of the first things you learn in industrial construction is how often protecting exposed electrical components seems to be a low priority. I’ve found National Electrical Code (NEC) standards to be generally ineffectual in many applications. The problem starts with the design of conduit, which consists of square threading instead of pitched threading like pipe.
Years ago I asked an electrical engineer why more-water-tight pitched threading isn’t used. I was told that square threading reduced the risk of spurs of metal. The idea is that square threading won’t cut wire as easily as pitched threading. Perhaps this is true, but as someone who has managed to cut himself on conduit, the reasoning seems thin.
The fittings themselves aren’t watertight, which I discovered while filling in for an electrical engineer during the design of a water-treatment plant involving a submersible pump. I suppose I always suspected it. The electrical engineer reviewing my design informed me that this is so water would drain from the fitting. Anyone who has watched an instrument short out because condensed water flowed from the conduit into the housing knows the fallacy of this argument. Ever since, I’ve designed conduit with drain holes and so that the flexible conduit loops upward; electricians hate me for it but I’ve delayed start-ups until I had it done my way.
And, then, there’s dust. To my knowledge there’s no way to keep dust out of an electrical panel that must vent heat to the outside. Dust build-up was a continual problem in a plant in California where a thermal scan became a line item on the spring cleaning list.
There are other, more insidious troublemakers than water and dust in an instrument layout design that affect reliability. One of these is atmospheric corrosion. Many problems can be minimized by keeping contaminants out of boxes.
The capital cost of a purge system — say, about $1,200, installed — is cheap compared to the damage caused by a single failed connection. During one start-up at a TiO2 plant, we were down for an extra day because the terminations in a marshalling panel were completely gone after only five years. We had to scour the company for new termination strips and pay plenty of overtime on a project where everyone was already working 18-hour days. And, this didn’t include the half day we spent monkeying around pretending we could fix the problem. Assuming $10,000 an hour for lost production, which is a good rule of thumb in the chemical industry if you can’t be more specific, the plant took at least a $240,000 hit — and that doesn’t consider customer satisfaction.
However, predicting savings based on this loss can be complicated. By comparing the installation cost with purge against what it turned out to be without it, and adding in the operating cost ($600/year for about 30 SCFH for a 9–ft3 box), the payback would be about 2.5 years, assuming 10% interest and a 15-year service life for the purged box. Obviously, if the risk to production could be accurately added in, the payback period would be measured in days instead of years.
At one company, the plant manager had the HVAC filters removed because he didn’t have replacements. Too bad, the filters were keeping the dust out of the PLC. It must be human nature — slowly developing problems tend to be ignored. After accumulating for several months, the dust must have shorted out something and the PLC was knocked out of commission. Let’s see, a $15 filter compared to two weeks of downtime — that’s got to be a payback period of about, what, five minutes?
Given the huge expenditure of time and brainpower that goes into your control system, doesn’t it make sense to protect it as best you can?