Once again there is a crack found in the inlet tubesheet in your high-pressure high temperature heat exchanger. As head of the maintenance engineering effort, you know that plant management will ask you if it can run safely and reliability until the next scheduled shutdown. In the daily production meeting the issue comes up and you recommend that a fitness for service (FFS) analysis be performed. Non-destructive techniques are reviewed and there is too much resulting uncertainty over the reliability of the characterization of the crack work. Without the proper characterization of the crack it is unlikely that a FFS could be performed and the unit would have to be shutdown and repaired during a non-scheduled outage. In the back of your mind you remember an application where frequency response testing was done on compressor blades to assess the effects of microcracks that were difficult to characterize in depth due to location of the fur tree.
The situation described is not uncommon. Every facility has a goal to run safely and reliably. Some situations just do not fit the traditional approach due to lack of available technology for the task at hand.
Every structure, especially when placed under load, has a natural frequency and mode shape for every frequency. Examples of these loads can be normal operational, some percentage in excess of that, hydrotests, etc. In the compressor blade scenario a baseline was determined based on a good blade under normal loads. The blades with microcracks were checked with acoustic frequencies and mode shapes characterized. Most of the proposed replacement blades had the same frequency response and mode shape as measured by accelerometers as non-damaged blades. In addition non- destructive testing that could be done found nothing indicating problems. The blades could be run. Several of the blades however had a very distinct and different response so those particular blades could not be used. The blades were tested in the machine for the best and most realistic response.
For the situation involving the heat exchanger inlet tubesheet the situation is more complex. A “cluster” of accelerometers tied into a data acquisition system would need to evaluate the response around each tube section. In such a system one is looking for/at the variance from the established norm for that situation. For both of the situations described it is best to have acoustic sensors attached to the equipment as well to assist in the overall evaluation.
A general methodology that would apply to both static and rotating equipment is as follows:
Exhaust all standard non-destructive techniques (NDT) and follow standard applicable Code procedures. If a positive conclusion cannot be reached with standard NDT then the following may be considered as applicable.