David Clark, Condition Monitoring Specialist, Bachmann Electronic, www.bachmann.info
Here’s why condition-monitoring programs fail – and the “repairs” to ensure their success
Condition monitoring drives predictive maintenance. It’s an accepted fact which is why it is so wide spread in many other industries besides wind. Because the wind industry seems a slow learner, let’s look at the pitfalls and mistakes made in condition-based monitoring (CBM) programs and learn what it takes to run one successfully.
1 The analyst
It starts with people. The focus here is on the vibration analysis more than the other condition-monitoring technologies. This one comes from condition monitoring not being part of the maintenance culture in wind. Consider paper mills. Condition monitoring there has been part of the culture for decades. Extreme cost penalties, $40,000 to $70,000 per hour of down time, does wonders for motivating a reliability-centered-maintenance culture. The real problem is that few people are qualified to analyze data who are vibration certified, have experience in wind, experience with a given turbine model, and other specific application knowledge.
How do you know if you have a good analyst? Start with someone certified in vibration analysis. Experience with the unique application of wind is also imperative as well as experience with many installations. Even then, that is no guarantee. Most certifications are so heavily focused on signal processing and handheld devices that this is no assurance of a good analyst. Why? Because signal processing done in most commercially available software and handheld devices are not used in wind. Some of the best analysts I have encountered are not certified. A few of the worst actually were.
Some who claim to have experience with so-many installations misdiagnose the most fundamental vibration readings or system set-up. If you were to show those readings to someone in another industrial market, they would know exactly what is wrong with the equipment in question. So there is no guarantee that installations make the analyst. Bottom line: vibration analysis readings are not subjective.
So, if qualified analysts are rare, a certification does not guarantee a good analyst, and neither does the number of installs, what does? Here are some guidelines for selecting an analyst:
• Ask how many installs they have
• Ask if they are certified. If yes, ask to
what level and by whom?
• Ask how much experience they have
and with which turbine models?
How would you know that you made a poor selection? When they:
• Tell you they need months of data
• Tell you they need a baseline or
readings from inception
• Have false alarms even after months
• When you get the feeling they are
learning and testing on your towers
The solution: There is no easy answer here. Good analysts in wind are rare. If unsure, outsource the job to a trusted analyst. See the guidelines for selecting an analyst.
2 No one is looking at the data
Imagine buying a race car and not budgeting for a driver. Follow me on this one because I have seen it dozens of times in all industries. Here’s what happens: The issues (fires or alarms) detected with condition monitoring are addressed (or put out) for the time being, and the drama subsides. Attention eventually drifts away but the issue slowly returns. The person who was looking at the data has been tasked with the project du jour and slowly abandons the CBM program. Consistently looking at and analyzing the data is key for long-term program success.
Another version of this scenario is that the person tapped to look at the data becomes involved in other projects after the fires are put out. Condition monitoring drives maintenance. If no one is at the wheel, good things do not happen.
I know of many installed systems that are running, generating useful, actionable data, and yet absolutely no one is looking it. The name of the game is consistency, especially with the slow manner in which wind turbine drivetrains fail.
The solution: Consistent analysis today, tomorrow, and for the life of the project. Allocate staff.
3 No one acting on the data
The best part of condition monitoring is picking out what is wrong. It’s the payoff. It’s essential to have an open line of communication between condition monitoring and maintenance or site personal. The best analysts and systems installed are of little value if the predictive maintenance information does not reach the people who perform the repairs. Those people could be with the turbine manufacturer, your O&M service provider, or your internal services group.
The solution: Provide an effective communication tool between the maintenance and condition monitoring staffs, and other appropriate units. All parties must have a clear understanding of the impacts of their actions.
4 Too many management changes
I know of a condition-based monitoring manager who had seven supervisors in six years. Each one asked him why the company had to have a CBM program and staff when they had no maintenance issues. What the newbie supervisors didn’t understand was that CBM was responsible for the asset stability and reliability. The first supervisor created the program and the second one eliminated the program. Six months later they were back where they started prior to CBM with reliability issues because no one was predicting maintenance.
One smart analyst in a similar situation detailed a case study in which a single save was worth $465,000. He subsequently handed this report to each new supervisor with the proud statement: “This is what I do.” On that success, the program was not cut again despite five more supervisors in five years.
Another variation of this scenario arises when other important people within the organization do not understand how impactful condition monitoring is to other aspects of business.
Here is a short list of the functions in an organization (other than O&M) positively impacted by proper condition monitoring:
• Production and PPAs
The solution: Create a cost-justification case study and a presentation for other departments within your organization so they understand the impact of CBM to their concerns.
5 No culture of predictive maintenance
This goes hand-in-hand with the mistake above. Several other industrial markets have condition monitoring ingrained into their culture. Getting it there only took a few decades. Creating a culture of predictive maintenance in a new industry is not as daunting as one would think. Several vertical markets, such as paper mills, mines, steel mills, refineries, traditional power plants, and many other industrial markets have installed and use condition monitoring.
There is good news and bad news here. The good news is that resources are available for establishing and understanding the basics of condition monitoring and its best practices. In addition, many positive examples of CBM systems show how managers avoided costly trouble. The bad news: Not much translates into wind.
How do you know if you are working in a culture of predictive maintenance? Simple. Ask yourself this: Which of the following categories describes the practices at my company? It:
• Waits until something breaks and
then repairs it.
• Repairs at set intervals, and repairs
• Predicts maintenance and provides
maintenance on company terms.
If you answered the third, you have the right culture. If you didn’t answer the third option, you don’t. It is interesting that the motivation in condition monitoring and predictive maintenance depends largely on the age, warranty status, and reliability of the fleet. This is understandable, of course, until these failures start costing you money. It’s human nature: out of sight, out of mind.
The solution: Understand and educate yourself and your company on the effects of condition monitoring done right. Read a few books on the topic of condition monitoring or reliability-centered maintenance. Or just ask yourself: What would it mean if I could predict impending failures and their resulting impact?
6 Mistaking SCADA data for predictive maintenance
There is a huge difference between SCADA data and CBM data. The first one changes every few minutes and has many inputs. The second one has few inputs and doesn’t change for weeks or months. One is focused on operational conditions and one is focused on the actual condition of the component. Now, some will say that temperature is a condition monitoring tool. Perhaps it can be in some cases, but not consistently, reliably, or with much predictability.
Here’s an example of why SCADA data (temperature) generally speaking is not a good predictor. On more than one model of wind turbine, the high-speed shaft (HSS) bearing spins in the housing where the temperature probe is located. When the bearing spins in the housing, it generates enough heat to melt the tip of the probe making the temperature probe readings unreliable. With the heat undetected, the bearing may run to failure.
Another example is that some components do run at different temperatures from day-one of operation. Does this mean that temperature is not a good indicator for predictive maintenance? It depends on two factors: Does the temperature sensor provide accurate data? Does it provide lead time to failures? If the answer is no, then use something else.
Conversely, various condition monitoring methods and technologies provide significant lead times and some specific details as to failure modes other than “it’s hot.” Truthfully, by the time a bearing is hot, the damage has been done.
The solution: Understand that SCADA and CBM have different functions and produce different failure data at different failure rates.
7 Never starting a CBM program
The worst CBM program was the one that was never started.
The solution: Start one. O&M costs go down, reliability goes up. WPE