5H86User
08-07-2009, 05:05 PM
We recently had a catastrophic failure of a connecting rod in a Carrier 5H86 reciprocating refrigeration compressor. The compressor is an 8-cylinder positive displacement machine. When the compressor operates, 2-cylinders are always active, while unloaders control the other 6-cylinders. The unloaders permit the 6-cylinders to become active as necessary, thus allowing the compressor’s cooling capacity to adjust according to need. The compressor is part of a refrigeration system that consists of a primary and secondary direct expansion coils; air cooled condenser, refrigerant receiver, associated piping, and specialties.
A root cause investigation concluded that the failure of the connecting rod was due to slugging.
The conclusion of our investigation was based upon:
· The system has been in service approximately 20 years and poor reliability has been a continual issue.
· The connecting rod that failed is associated with a cylinder that frequently loads & unloads.
· The connecting rod experienced 3 breaks in the area where it attaches to the crankshaft. Examination of the connecting rod by a metallurgist concluded:
o Two of the breaks were consistent with brittle overload fractures. The other exhibited beach marks. The beach marks suggested a fatigue crack propagated through 1/2 of the cross-section before the section failed due to overloading.
o The casting quality was good; no flaws were found.
o The connecting rod was bent approximately 5 degrees.
· The interface between the crankshaft and connecting rods associated with two cylinders exhibited severe wear. Specifically:
o For one of these interfaces the journal bearing was essentially destroyed and the interfacing surfaces exhibited severe wear. This interface is associated with the failed connecting rod. Right next to this interface is the interface for another connecting rod. This adjacent interface exhibited normal wear suggesting that lack of lubrication was not an issue.
o For the other interface, both the journal bearing and the interfacing surfaces exhibit severe wear. This interface is associated with a cylinder that is always active (loaded). Right next to this interface is the interface for another connecting rod. This adjacent interface exhibited normal wear suggesting that lack of lubrication was not an issue.
· Concurrent with the failure, there was a step in the number of cylinders loaded from 2 cylinders active to 8 cylinders active. It is believed that the corresponding change in flow velocity swept up oil that had accumulated in low spots, entrained this oil in the flow stream, and transported it to the compressor causing the slugging.
· The configuration of the compressor suction piping facilitates oil hide out that contributes to liquid oil slugging. Specifically:
o The compressor is located approximately 20 ft above the evaporator coils. A portion of the vertical riser piping is oversized. The mfg recommends this riser piping be 2.625 inch diameter, while a small portion is 3.125 inch diameter.
o There are traps in the horizontal piping, but the size of these traps is larger than recommended.
o There are two horizontal runs, one is properly sloped the other is level (no slope).
· There are two sets of evaporator coils. The primary evaporator coils are always on-line while the secondary evaporator coils come on-line as needed. The secondary evaporator coils are controlled off of suction pressure.
· The number of compressor cylinders loaded is also controlled off of suction pressure. Having suction pressure control both the secondary evaporator coils and the number of compressor cylinders loaded creates control instability. Data shows that even when the cooling loads are steady the secondary evaporator coils cycle approximately 20 times in an hour and the number of cylinders loaded cycle approximately the same.
· There is evidence suggesting that liquid ***** slugging occurs at compressor start-up. Evidence also suggests that the valves between the high-pressure side f the refrigeration piping and the low-pressure side (i.e., LLSVs & HGBVs) experience seat leakage. This allows liquid ***** to accumulate in low spots within the suction piping. In the past preventative maintenance on these valves consisted of replacing the internals. Because the body was not replaced the condition of the body seat is unknown. Recently this practice was revised to require replacement of the entire valve.
Questions:
Based on the above analysis, we are:
1. Changing the piping to install the reducing elbow in the riser (as originally called out by the vendor) instead of a reducer downstream of the elbow, to increase the velocity at the elbow and help reduce oil hide out and slugging. This will include resloping the level section.
2. Changing the setpoints for the compressor staging – there is currently a 1 psi difference between the compressor staging and the secondary coil coming online. The plan is to spread them out to prevent the fighting.
3. Installing a change to provide automatic pump down of the HVAC compressor during idle periods, to prevent ***** slugging.
4. Going to install a mod in the future that will install temperature control for the secondary coil, separating it from the suction pressure control.
Given what has been provided, do the members of this forum see any other vulnerabilities or have any suggestions to help eliminate these failures? We have looked at this for weeks, but at this point we may be too close to the issue to see other solutions, which is why we are posting this here.
Thanks in advance for the help.
5H86User
A root cause investigation concluded that the failure of the connecting rod was due to slugging.
The conclusion of our investigation was based upon:
· The system has been in service approximately 20 years and poor reliability has been a continual issue.
· The connecting rod that failed is associated with a cylinder that frequently loads & unloads.
· The connecting rod experienced 3 breaks in the area where it attaches to the crankshaft. Examination of the connecting rod by a metallurgist concluded:
o Two of the breaks were consistent with brittle overload fractures. The other exhibited beach marks. The beach marks suggested a fatigue crack propagated through 1/2 of the cross-section before the section failed due to overloading.
o The casting quality was good; no flaws were found.
o The connecting rod was bent approximately 5 degrees.
· The interface between the crankshaft and connecting rods associated with two cylinders exhibited severe wear. Specifically:
o For one of these interfaces the journal bearing was essentially destroyed and the interfacing surfaces exhibited severe wear. This interface is associated with the failed connecting rod. Right next to this interface is the interface for another connecting rod. This adjacent interface exhibited normal wear suggesting that lack of lubrication was not an issue.
o For the other interface, both the journal bearing and the interfacing surfaces exhibit severe wear. This interface is associated with a cylinder that is always active (loaded). Right next to this interface is the interface for another connecting rod. This adjacent interface exhibited normal wear suggesting that lack of lubrication was not an issue.
· Concurrent with the failure, there was a step in the number of cylinders loaded from 2 cylinders active to 8 cylinders active. It is believed that the corresponding change in flow velocity swept up oil that had accumulated in low spots, entrained this oil in the flow stream, and transported it to the compressor causing the slugging.
· The configuration of the compressor suction piping facilitates oil hide out that contributes to liquid oil slugging. Specifically:
o The compressor is located approximately 20 ft above the evaporator coils. A portion of the vertical riser piping is oversized. The mfg recommends this riser piping be 2.625 inch diameter, while a small portion is 3.125 inch diameter.
o There are traps in the horizontal piping, but the size of these traps is larger than recommended.
o There are two horizontal runs, one is properly sloped the other is level (no slope).
· There are two sets of evaporator coils. The primary evaporator coils are always on-line while the secondary evaporator coils come on-line as needed. The secondary evaporator coils are controlled off of suction pressure.
· The number of compressor cylinders loaded is also controlled off of suction pressure. Having suction pressure control both the secondary evaporator coils and the number of compressor cylinders loaded creates control instability. Data shows that even when the cooling loads are steady the secondary evaporator coils cycle approximately 20 times in an hour and the number of cylinders loaded cycle approximately the same.
· There is evidence suggesting that liquid ***** slugging occurs at compressor start-up. Evidence also suggests that the valves between the high-pressure side f the refrigeration piping and the low-pressure side (i.e., LLSVs & HGBVs) experience seat leakage. This allows liquid ***** to accumulate in low spots within the suction piping. In the past preventative maintenance on these valves consisted of replacing the internals. Because the body was not replaced the condition of the body seat is unknown. Recently this practice was revised to require replacement of the entire valve.
Questions:
Based on the above analysis, we are:
1. Changing the piping to install the reducing elbow in the riser (as originally called out by the vendor) instead of a reducer downstream of the elbow, to increase the velocity at the elbow and help reduce oil hide out and slugging. This will include resloping the level section.
2. Changing the setpoints for the compressor staging – there is currently a 1 psi difference between the compressor staging and the secondary coil coming online. The plan is to spread them out to prevent the fighting.
3. Installing a change to provide automatic pump down of the HVAC compressor during idle periods, to prevent ***** slugging.
4. Going to install a mod in the future that will install temperature control for the secondary coil, separating it from the suction pressure control.
Given what has been provided, do the members of this forum see any other vulnerabilities or have any suggestions to help eliminate these failures? We have looked at this for weeks, but at this point we may be too close to the issue to see other solutions, which is why we are posting this here.
Thanks in advance for the help.
5H86User