It's Monday morning. Your weekend has been consumed by your team working around the clock to get your compressor back up and running. While the planned outage was still another six months away, the hope was to run without incident for five years. Unfortunately, that was not the case. Because your focus was to get the plant back up and running, the broken impeller parts were just left in the corner of your office.
Finally, after much hard work, the plant is back on line. With the compressor running and the plant production on spec, the plant manager is pleased, as is the rest of the staff. The next weekend runs smoothly and everyone has a chance to take a deep breath. After this much needed reprieve, meetings are held in order to assess the failure, including the cost as well as the underlying cause. The resulting shutdown necessitated the exercise of a force majeure clause. As you are the maintenance manager, everyone is looking to you for answers.
You assemble a team to handle the pending investigation. The parts are shipped to a metallurgical lab to perform a failure analysis. Everyone works exhaustively, looking at time history plots. Because the machine has a good history, there is not too much concern. In fact, the machine nearly made the planned outage without any failures. So, when the outside metallurgical lab presents the results of its failure analysis revealing the cause to be "corrosion fatigue," everyone is satisfied. Seemingly, the problem is solved. Unfortunately, and shortly thereafter, the next machine fails. Another iteration of all this occurs, and the metallurgical lab comes back with the same result: corrosion fatigue.
You are somewhat perplexed by the fact that an impeller that ran for several years failed due to corrosion fatigue, and then another impeller ran for only weeks and also failed due to corrosion fatigue. The lab had no information of the length of service for each of the impellers; it simply performed a failure analysis. At this point, you are convinced that you don't have the complete picture.
In fact, you don't. The metallurgical analysis was not a failure analysis; it was a metallurgical analysis. The spare impeller that was put into service, while only running previously for weeks as opposed to years, had several years of exposure from a previous run, so corrosion fatigue was assumed. While it may have had some corrosion, everyone agreed the second failure occurred too quickly to be attributed to corrosion. So what does a failure analysis of this major piece of critical equipment involve? It is a multiphysics approach, consisting of process, controls, operations, design, maintenance, mechanical and metallurgical analysis. The cause of this failure was cavity acoustics.
What is cavity acoustics? In layman's terms, it is a condition where internal noise is amplified in the compressor case, which causes the impeller to vibrate and fail. From a technical standpoint, internal pulsation develops from an acoustic natural frequency, which is excited by a forcing function. A dynamic pressure pulsation is developed, and a forcing function couples with the mechanical natural frequency of the impeller, causing it to vibrate at a resonant frequency. When this happens, the impeller fails quickly. A change in feedstock leading to a molecular weight change caused this rare coupling in the system. It took a process analysis and an in-depth understanding of its effect to determine what happened to the compressor.
Here's an approach that one may consider:
- Historical assessment
- Process analysis
- Mechanical analysis
- Controls review
- Maintenance assessment
- Metallurgical and materials analysis
- Root-cause analysis
The problem presented required a multidisciplinary approach. It involved in-depth knowledge of process dynamics and complex cavity acoustics. The solution to this problem was a minor process change. The cost was likely millions in lost revenue for the facility. These types of problems should be reviewed by a professional engineer competent in failure analysis of these systems.
For such problems, KnightHawk Engineering Inc. is your one-stop shop.
For more information, visit www.knighthawk.com or call (281) 282-9200.
