A chemical manufacturing plant with three, three-cylinder reciprocating compressors was accustomed to performing intensive maintenance on the compressors.
For about 20 years, the maintenance on these three machines was not performed on any schedule; it was simply performed whenever a machine broke down. Periodically, a machine would break a piston rod, be taken out of service, and a standby machine would be put on line.
Figure 1. Compressor Showing “W” Cylinder Formation
It had become accepted that the problems with the compressors stemmed from carbon buildup that was simply unavoidable. However, with a little investigation and some basic oil analysis activity, the underlying problem was identified and a solution that not only reduced maintenance cost, but also improved plant safety was implemented.
The process gas entering the compressors was both flammable and toxic and came from the chemical process plant to the compressors through a drying tower before going back to the plant for further processing. The double-acting compressor cylinders were in a “W” arrangement with both the outer cylinders of the W being first-stage cylinders and the center one being a second-stage cylinder.
The pistons were double-acting with lubricated piston rings. The pistons and gland packings were originally lubricated with a mineral-based compressor oil, but had been switched to lubrication with a synthetic polyalphaolefin (PAO)-type oil. The piston and gland packings were lubricated from a separate lubricator skid comprised of a small oil tank and an oil pump.
The switch to a PAO oil had been recommended by the local representative of the oil company in an attempt to reduce the carbon buildup. However, the change had no effect at all.
Figure 2. Carbonized Compressor Valve
A visit (by the author) to the plant and the compressor house revealed a large amount of used or broken machinery lying around. Only one of the three compressors was actually operating, one was on standby and the other had been stripped down by maintenance personnel under the direction of the manufacturer’s local service representative.
Apparently, the plant had operated the machinery in this fashion for around 20 years and the compressor house was virtually the permanent place of employment for the manufacturer’s service person!
The service person and a plant operator indicated that the machines were regularly taken out of service due to carbon buildup inside the cylinders and valves. Sometimes, the end clearances inside the cylinders became so reduced that the pistons would touch the cylinder head and the piston rod would snap underneath the piston attachment.
Usually, the contact took place on the downward stroke with the underside of the piston striking the bottom of the head at dead center (Figure 3).
Somewhat surprisingly, when this happened the failed piston rod would push the piston back up the cylinder on the next stroke and the compressor would keep running. The only clue to a piston rod breaking was pressure dropping off in the plant as the gas flow diminished.
Even though this problem appeared to be rather serious, the plant operator and the service person were not concerned in the least, as this happened fairly regularly. They just considered it a sign that the machine was due for maintenance.
Inspection of a dismantled machine’s parts revealed the heavy carbon deposits that had been a characteristic of these machines for years. All internal surfaces in the cylinders were heavily fouled with oily, black, hard carbon deposits. The gas passages of the inlet and outlet valves were heavily choked with carbon.
The gas packings were a lubricated metallic Babbitt-type and were also heavily contaminated with the carbonized debris. The piston rods were heavily worn by the carbon that was finding its way into the gas packings and subsequently wearing the rod surface away.
A broken piston rod was examined and a fracture immediately below its piston attachment was discovered. A change in material characteristic about three millimeters deep all round the periphery of the rod was also readily apparent.
The rod was taken to a metallurgy lab and cleaned up. At that point, it became clear that the apparent discoloration seen at the jobsite was in fact a very thick layer of metallic flame-sprayed buildup.
At first this seemed bizarre, but upon talking to the plant supervisor it became clear that the replacement piston rods were not new. Because the rods didn’t last long, they were sent to a local workshop to be refurbished - metal sprayed back to the original diameter - and were then sent back to the stores to be reinstalled in the compressors.
A discussion with the local repair shop personnel, who were preparing and flame-spraying the rods, revealed that the repair shop didn’t know what the piston rods were or what they did. To them, they were a worn piece of round bar with a thread at each end.
They just prepared the rods with a coarse machined finish (perfect conditions for stress propagation), sprayed them and then finished them off by turning them to a given diameter.
The reclaimed rods, which were considered fully refurbished items, were then put back into stores. It was notable that neither the plant nor the reclamation workshop understood what was required in refurbishing a piston rod or what the parameters might be to scrap the rod.
The manufacturer’s representative gave no advice on the condition or refurbishing of the rods; he just used the parts that were furnished by the client. In fact, the broken rod had a flame-sprayed coating that was three millimeters thick on a rod that was only 47 millimeters in diameter. So the effective diameter was reduced to about 41 millimeters (Figure 4).
The shop performing the refurbishment didn’t think there was any limit to how thick the metal spray coating could be as long as the final diameter met the specifications.
Figure 4. Section Through Piston Rod
The plant manager and the plant chemical engineer believed the machinery fouling was due to “unreacted” carbon, a normal result of the process, and they didn’t consider it much of a problem because it had occurred for years.
Their unreacted carbon theory had been accepted and went unchallenged despite the fact that there was another plant next door that had no problems with its (nonlubricated) reciprocating compressors.
However, further investigation rendered this explanation weak. It was noted that the gas entering the compressors traveled through an upstream drying tower that distributed a 98 percent sulfuric acid mist.
Therefore, an experiment was conducted to determine if the sulfuric acid could be affecting the oil. A small jar of the PAO oil was taken to the on-site laboratory and one of the plant’s chemists added a few drops of 98 percent sulfuric acid to the oil.
Immediately, the oil went black, and after the addition of a few more drops, free carbon was produced from the acid-oil mixture. This evidence prompted personnel to retain a tribology lab to perform some tests on the stability of the oil when mixed with concentrated sulfuric acid. A brief transcript of the findings is shown in Table 1.
During one of the tests, enough sulfur dioxide gas was liberated from the resulting reaction to cause the laboratory to be temporarily evacuated! Large quantities of carbon were produced from the acid-oil mixtures. At this point, the following findings were presented to plant management:
During the following months, the staff involved were re-educated about the importance of proper refurbishment procedures, and a full procedure for piston rod refurbishing was written and put into place. In addition, the manufacturer’s service department was dismissed and a third party was brought in to perform any needed compressor work.
A project was implemented to replace the three old lubricated machines with two modern nonlubricated machines similar to those installed in the plant next door. This approach was thought preferable to converting the existing machines, although nonlubricated conversion was technically feasible.
The plant continues to run profitably and the compressors now run year-round with just an annual routine service. Any problems with unreacted carbon clogging the machines have been eliminated, and the 20-year old stories about the process gas affecting the machines are steadily being forgotten.
About the Author
Stephen H. Shakeshaft is a mechanical engineer based in the United Kingdom. He is the principal consultant and director of Stephen H. Shakeshaft Consulting Ltd. More information about Stephen H. Shakeshaft Consulting Ltd. can be found at www.shs-consulting.co.uk.