Law enforcement organizations have this down to a science. Arrive at any crime scene, and you'll find yourself immediately in the midst of a flurry of activity. After the Under Pressure Food Mart is burgled, the area is roped off,witnesses are gathered together and segregated from other onlookers, fingerprints are being lifted, and suspects may already be in custody.More cops are there to guard the area from accidental or purposeful intrusion.
The amount of resources expended on a major (or even many minor) crime scene can be truly mind-boggling.You'll find the team leader, who directs general responsibilities. The photographer documents visual evidence, a sketch artist takes descriptions and draws the crime scene, and a number of officers guard the area. Investigators interview people at the scene, while more patrolmen canvass the local residents for more data. Specially-trained evidence gathering personnel process the evidence and ensure the documentation is foolproof. Investigators immediately start researching the backgrounds on suspects, looking for clues in past history.
Why is this immediate effort so massive? When actually analyzed, the number of man-hours invested, equipment expended and depreciated, and the inter-departmental coordination required add up to a hefty wad of cash that the taxpayers must pony up. Of course, this must have been determined to be appropriate, or local law-enforcement efforts would be shut down. Is this initial level of investigation really necessary? In fact, why not wait a few days for everyone to calm down, let the emotions die off? After all, we are hurting the business owner by restricting access to the shop, bothering his customers, even appropriating pieces of his store or inventory. Let him get back on his feet.What makes this worth the effort?
What makes this acceptable is the fact that there is really no other method available that can reliably produce the required results. If the photographer was not there, there would be no record of the actual environment at the scene. Evidence that is not quickly and accurately recorded will be lost or modified, with no hope of retrieval. We could wait to begin researching background information, but this will just prolong the successful completion of the investigation beyond reasonable time-limits. Sweeping up and throwing away the broken glass gets the business up and running, but for how long? Without this process in place, the crime is almost guaranteed to happen again.The stricken store may install bars on the windows, but the criminal still atlarge will just find another way in, or move on to the next store down the street.
The process of determining the cause of an equipment malfunction can often seem as daunting as a major crime scene investigation. It often appears to require expert knowledge about how the equipment was operated, how it was installed, the original design specs, changes in the environment, how it was actually being used, etc. Luckily, with just the right combination of repair expertise, root cause analysis, and corrective action implementation, the process does not necessarily
have to be harder to get more productive and lasting results. The right systems have usually already been purchased and put in place at most production facilities to get the data required for an accurate and detailed failure analysis. Unfortunately, the employment of these resources is not always optimum. A smarter approach to the gathering of evidence, the correct interpretation of what that evidence is telling you and the judicious application of corrective actions will put those expensive monitoring systems to work for you.
The evidence gathering process
Most companies already have many systems in place that can help the troubleshooter narrow down his focus, but often times the data is no longer available. The act of repairing the gear has already modified, moved, or destroyed key pieces of evidence. Although the failure appeared to be minor at first, these data points can be crucial to finding an actual root cause of equipment damage.Where do you get the evidence you need to determine the actual root cause of the failure?
A good place to start is with the equipment operators. How often have you heard (AFTER the gear is down),"Oh, yeah, it's been doing that for a while," or "It's always been that way." This can be one of the most frustrating times in the life of the maintenance manager, listening to an operator describe in detail the telltail signs that his gear is about to fail.However, at this point in the failure analysis, this is just INFORMATION TO BE GATHERED. The fact that the operator did not inform anyone about the previous abnormalities is yet another data point. Again, this is only data that can be used later for root cause analysis and corrective actions. Do not draw any conclusions at this time.
Some companies have trained their operators to immediately document the conditions encountered at the time of a failure. The data is often written on a standard form or in the operator's log using an approved format. In either event, the report should include some basic information:
Equipment monitoring records and recordings contain a wealth of information.Vibration monitoring recordings, thermal images, and oil analysis results can all be used to determine the timeline of events leading up to the failure.You may not know what to do with the data yet, but have it available and ready for further scrutiny.
Machinery history and repair records are invaluable. These records can be on paper or in electronic format. They can be used to discover long-term trends in equipment operational status and down-time analysis. Has this happened before? What caused it that time? How did we fix it last time? Did that fix work?
At this point, the usefulness of these records is established by past maintenance practices. Entries in these records that say (more or less), "Process pump #3 down due to pump failure" is much less useful than,"Process pump #3 secured (run hours 2910). Smoke noted issuing from mechanical seal upon initial start-up.Discovered clogged flush line. Line cleared, flow verified, seal replaced and retested." The second entry contains a wealth of information that can be used for a much better analysis of the reason it failed versus just a single failure datapoint. This entry would probably take the maintenance supervisor an extra three minutes to complete.
When should entries be made in the machinery history log? Best practice is to make a minimum of two entries: one immediately at the initial failure, and one following repair and retest. If further indications were found, special troubleshooting methods were employed, or the troubleshooting was very complex, more entries can be made as required. Bottom line: for electronic recording systems, there cannot be too much data. Paper systems may require a more judicious use of space to prevent an unmanageable clutter, but can still contain a good amount of information.
Another important information resource is the broken piece of equipment itself. It is critical that the troubleshooter look at the failed part to determine not only what broke, but how it broke. The failure mode and failure agents must be determined to find and eliminate the actual cause of the failure.
Sequencing the analysis
The sequence of the data-gathering steps is actually quite important. The operator should immediately write down his indication. The troubleshooter should talk to the operator early on to get his thoughts while it is still fresh in his mind. But when can equipment repair begin? After all, working in parallel to find the cause, while simultaneously preparing for the repair, just seems like good sense. However, this is where an enormous amount of information is often lost, destroyed, or altered. The following example illustrates how working ahead of the analysis can lead to frustrating re-work.
A plant was having its entire main condensate system overhauled. New piping was being installed, and the condensate pumps were to be rebuilt. Work began on the system by removing the pumps and hauling them to the pump shop for refurbishment. Piping in the system was cut out and replaced to correct below-spec minimum wall thicknesses.
The pumps were spec'ed out, rebuilt, and hydrostatically tested in the shop. No issues were found.
Two months after their removal, the pumps were re-installed in the system. The system was filled, vented, and tested one pump at a time. After running for 20 hours, the lower pump bearing failed, as indicated by excessive vibration.
The pump was removed from the system and inspected. The lower pump bearing was found to have failed. The bearing was replaced and the pump re-installed. Twelve hours after start-up for run-in, the bearing again failed.
This time, the ace pump rebuilder was called in. Obviously, someone was not installing the bearing correctly. He had been doing this for years, and would make sure the job was done correctly this time. He personally supervised the rebuilding and retesting of the pump. It was run on a test fixture for 80 hours,with all vibration measurements well within spec. Everything looked fine from his perspective.He saw nothing that he recognized as a problem from his experience.
The pump was again re-installed and retested. The bearing failed for the third time after 20 hours of operation. Each bearing replacement cost over $23,000 just in parts and labor. So far, this equated to nearly $70,000, not including the slip in delivery date, extra time and effort expended by the expert pump supervisor, and extensive pre-installation vibration testing on the third go-around. Yet, the pump was in worse shape than before the overhaul.
Finding the culprit
From this example,with the data you have been given, the cause of the bearing failure will not be obvious. Even the expert is left scratching his head. How do you go about finding the cause of this type of failure?
The sequence of evidence gathering listed above was followed for all three bearing failures. Obviously, there must be something else going on that even the "pump guru" was not aware of or hadn't thought of. What do you do?
This facility fell into one of the traps that many companies stumble into.Repairs were commenced before the failure analysis was complete. Companies want to get ahead and disassemble the pump, but this can lead to the disruption (or destruction) of evidence needed to determine the cause. But, wait a minute.We determined earlier that one of the most important pieces of the puzzle is the failed component. How can we analyze the bearing if we don't first disassemble the pump? We seem to need to know the possible causes before we even start the disassembly!
This is a great question. It runs to the core of why many troubleshooting and repair scenarios end with a rework of the same failure.
Having an advantage
It's human nature to seek an advantage when dealing with a problem. So, let's walk through the above example, using the TapRooT®'s Equifactor® Equipment Troubleshooting module to help narrow down the cause of the failure—even before disassembling the equipment. (Equifactor is a system that has incorporated the troubleshooting expertise of Heinz Bloch into easy-to-use tables that allow the troubleshooter to narrow down the causes of equipment and component failures during the early stages of the troubleshooting effort.)
As shown in Fig. 1, the first step is to diagram exactly what happened. This is done using TapRooT's SnapCharT® function.
By using this system, a timeline is set up with all the known data incorporated into an easyto- understand format. It may be tempting to skip this part ("I know what happened!"), but this is a crucial step in understanding exactly what happened when.
Now, since this is an equipment-related failure, the Equifactor module is brought to bear. Using its logical tables, most causes can quickly be ruled out, and causes previously not thought of are brought to light.You can eliminate many of these causes right away (the pump had been verified in balance, the shaft was not bent, etc). The possible remaining causes are now known, and valuable data can be brought to the jobsite to find the actual cause. You now know the right questions to ask during the equipment teardown:
1. Is there a misalignment between the pump and motor?
2. Is there casing distortion due to excessive pipe strain?
At this point, you can continue the investigation just like any other. Since you know what to ask, you know what to look for.You can go to the job site and gather the extra data that you need. In this case, before the pump is unbolted from the foundation, you notice the riggers are connecting chain falls to the discharge piping and the pump.When questioned, the riggers tell you that it took chain falls to get the piping aligned during installation, and there will be quite a bit of tension as the flange bolts are loosened.
The root cause
"We found the root cause!" "Those mechanics obviously don't know what they're doing and are flexing the pipe (and the pump casing) too much." "Tell those mechanics to line it up right next time!" Do those remarks sound reasonable? Of course not.Unfortunately, they are the type of responses that are heard over and over again throughout industry.
"Tell those guys to be more careful." This has the same effect as telling your son (after he's run over the family mailbox) to drive more carefully in the future.You'll get a half-hearted "OK," and still nothing changes.
While, the root cause analysis is not over,we finally have the information we need to start the analysis. In the TapRooT system, the data gained from this investigation is now fed back into the SnapCharT, and problem factors are highlighted, as shown in Fig. 2.
The highlighted problems are not the root causes, but they are the major indicators that will now be used by the rest of the TapRooT system to find the actual root causes.After completing the investigation and running all these indicators completely through the system, several root causes may be found. For example:
This is another point in the incident investigation process that often fails. Corrective actions must now be assigned that are meaningful, achievable, and the results measurable. For example, it does no good to tell the workers to be more careful. Each of the root causes must be addressed on its own merit, with corrective actions assigned, carried through, and audited.
Who has time for this type of analysis? In reality, all best-in-class companies have found the time. The time spent properly following up on equipment failures is rarely wasted time. In fact, the savings are compounded two-fold. In this particular case, the time spent conducting a proper equipment failure analysis would have saved the shipyard the three weeks and over $150,000 in delays after the first bearing failure. In addition, if the corrective actions are not implemented, this same issue is almost guaranteed to happen again, causing repeat equipment failures and delays further down the road.
Unfortunately, this scenario is not an isolated case. Every plant has at least one of these stories to tell. Not every plant can say it has come up with a proven system that has averted further repeat problems. As reflected in Table I, studies have shown that industry is not meeting the best practice mix of maintenance resource strategies:
Industry seems to be spending large sums of money on predictive maintenance systems, allowing users to know WHEN the gear is about to fail, but none of these systems can tell you WHY. It is up to the trained investigator, with the right tools, to be able to avoid the costly repeat failures that continue to plague the manufacturing field. MT