You were recently hired on as a senior maintenance technician at a plant with a maintenance department that is clearly undersized for the job it must do. For example, you’re the only senior maintenance technician.
One of the plant’s big money-makers is a complex machine. This machine is under PLC control. It uses a variety of inputs, including two flowmeters, six temperature switches covering different temperature ranges, some pressure switches, and some proximity switches.
The maintenance manager wants you to figure out why this machine has such high scrap rates. He told you, “We calibrated everything and went through the PLC logic, and we cannot figure it out. The operators end up making a lot of manual adjustments to get the scrap rates as low as they are, but they need to be a lot lower.”
Where might you start — and what might you consider — to get to the bottom of this?
Answer to Quiz
Normally when troubleshooting a PLC-controlled system, you start by examining the final control elements. Then you look at the inputs (sensors, switches, etc.), and finally, you check the PLC (hardware, connections, and programming—typically in that order). In this case, you get a big clue that the problem is in the input side of the system. The maintenance manager is almost certainly wrong in his belief the maintenance department has “calibrated everything.”
- Flowmeter calibration requires specialized equipment, and it’s unlikely that a plant with insufficient investment in staff has invested in such equipment. They may have done loop checks or similar, but not a calibration. Review the maintenance history on these to make sure. It’s fairly typical to outsource the calibration of these meters ,and if that cannot be scheduled to occur during a scheduled shutdown, the plant will have to buy a new unit for each model of flowmeter it has. Each new one will have to be installed as a substitute while the original is out for calibration (after being calibrated, it would then serve as a spare). This lack of calibration is likely your smoking gun.
- Calibrating temperature switches typically requires a sand bath or other special apparatus that can provide evenly distributed heat in the measurement area at temperatures well beyond the boiling point of water. If you don’t see any such a device in the maintenance shop, you’ve found one issue.
- Look at the calibration records for the pressure switches. Also look at the calibration procedures and equipment. For example, what is used to measure the applied pressure and is it calibrated to a tighter tolerance than the pressure switches? What about the method of calibration? Find the last tech who calibrated one of these and ask him to do a calibration while you watch. He should increase pressure at a rate that does not shock the sensor. Also watch this tech reinstall a given switch. Does he, for example, put thread tape on all the threads or just those threads that will remain engaged in the mounting boss? If this tape is on all threads, that is an error. The purpose of the tape is to provide lubrication for the final tightening of the threaded parts. You do not want it in contact with the process fluid, and you do not want it covering any part of the sensing element. Misapplication of this tape could well be your smoking gun.
Once you’ve been able to verify the sensing elements are correctly calibrated and correctly installed, visually trace their signal wiring. Look for damage to it (e.g., it got kinked because somebody stepped on it) ,and look for its routing (a common mistake is to run it too close to power wiring and thus induce a false signal in it).
Now turn your attention to the final control elements. An easy way to identify these on the system drawings is these are the devices fed by the PLC output module(s). They will be things like motors and valves. As before, check the proper calibration (as in the case of valves) and maintenance (as in the case of motors). For example, if the machine has a feeder pipe with a globe valve that adjusts its position per the PLC signal to it, then that valve should have been stroked at five positions using a loop simulator device.
PLC programming doesn’t “go bad”, unless someone messes with it. If this is where the operators are manually making adjustments, you have found yet another problem area. PLC modules do go bad (start by checking for the power light) and so do connections (typically, where the spade or ring lug is crimped to the wire).
In summary:
- Ensure the inputs are accurate (measurement function).
- Ensure the outputs are accurate (control function).
- Restore the PLC programming to its original state or where it was prior to the manual adjustments (there should be a backup). Check PLC hardware and connections.
If this doesn’t fix the problem, you have at least eliminated the PLC system as being implicated. Other causes could be operator error, a change in raw materials (someone in purchasing “saved money” by going with a lowball supplier), a change in ambient conditions, or something else that is outside the purvey of your undersized maintenance department.
