A nuclear power plant and the average soup factory would not appear to have much in common, but both are among the industries that for decades have used condition monitoring systems (CMS) for keeping a close eye on the health of their rotating equipment. "I've been into a soup plant and a nuclear power plant in one day. They both are using the same equipment. They both have vibration analysis," says David Clark from California company TurningPoint, a CMS specialist. "Now there's a big difference between a nuclear reactor and chicken noodle soup, but they both have rotating equipment and they are trying to predict failure."
The prevalence of CMS in other industries makes it all the more odd that use of vibration monitoring equipment and automated oil particulate analysis has not been standard practice when operating wind turbines, some of the biggest rotating machines on the planet. In America, only some manufacturers offer CMS as standard and only on some of their models. Most machines going up in the United States do without it.
Opinions in the US wind industry are divided on whether the extra cost of predictive and preventative maintenance compared with regular scheduled maintenance can ever pay off. There is doubt about whether the systems really provide information that can be used by operators, or whether they even work as advertised. Others argue that early discovery of problems can forestall replacement of costly components and save a bundle on operational downtime. To add to the uncertainty, some wind turbine vendors have been mixed on whether to include the systems as standard equipment, optional equipment, or not offer it at all (box).
CMS in wind turbines involves monitoring components for vibrations that could spell trouble ahead and watching out for debris in oil. Primarily, it is the gearbox that is monitored, but also other components, particularly in the drivetrain. Vibration monitoring is arguably the more sophisticated of the two. It involves the placement of vibration sensors on certain critical moving parts. The resulting data streams are captured on a constant automated basis and stored or transmitted to an operator who can look for changes that could indicate an impending hardware problem.
Automated oil particulate analysis, as its name suggests, builds upon the decades old practice of manually analysing gearbox or hydraulic oil for signs of metal particulates in the fluid, the presence of which likely indicates an impending hardware problem. But instead of waiting for this process to be done manually by a technician every three or six months, the system is automatic and can relay changes remotely to an operator or diagnostic service.
Where to start
CMS can be deployed in two ways. In many cases, it is bought as an after sales add-on to turbines of various ages operating in the field. The incremental cost -- between $5000 and $20,000 per turbine -- can substantially increase normally allotted operations and maintenance costs for a wind plant. Secondly, it can be included and paid for before a turbine leaves the factory. Most new large-scale turbines today include CMS systems as an option, but they rarely come as standard and are not always available for all models.
Proponents of the technology stress two major selling points. First, that more sophisticated and automated analyses of the internal components of a wind turbine can lead to better preventative maintenance and more cost effective scheduling of expected repairs. "If you're doing nothing but repair maintenance and fixing stuff as it breaks, you're way behind and you're never going to catch up," says Erik Smith with SKF Group, which provides full condition monitoring systems for a variety of industries.
Without naming names, he cites data from the acquisition of a wind turbine service company by a large engineering service company in the US. Its conclusions were that over the 20-year average lifetime of a typical wind turbine deploying CMS, there would be five repairs of significance. Without CMS, there will be only three repairs -- but those repairs will be much bigger, such as a full replacement or rebuilding of a gearbox, which can cost well in excess of $300,000. The use of CMS can help an operator decide when to replace some gears in a gearbox at a cost of $30,000 instead of waiting for a catastrophic failure to occur.
Scheduling of repairs is also a central sales pitch for CMS equipment. If an operator is able to identify faults and impending hardware breakdown early on, the situation can be managed more cost effectively -- particularly at a time when windsmiths and cranes are in high demand and increasingly costly. "I've seen some outrageous in and out fees," says Bo Thisted, with Upwind Solutions, a relatively new wind turbine operations and maintenance company in the US. Just getting a crane on site can cost upwards of $80,000 in the US -- partly a result of the developer scramble to complete as many projects as possible before a potential expiration of the federal production tax credit (PTC).
Smith and others say considerable value can be extracted from scheduling repairs in sequence, once a part has arrived, or a crane-call is cheaper. Meantime, a turbine showing subtle signs of wear can be "feathered back" in higher winds to reduce loads. "Are you going to get optimal performance out of that machine, no, but you're not going to have catastrophic failure," says Smith. "Or say you can baby it for a year because you know you're going to have a developer out in the area next year with a crane and you can manage the issue until then. You're the boss, you're managing your asset."
Wind project developers are listening to these arguments. Two of the main players in the US market, Horizon Wind, the American subsidiary of Portugal's EDP, and Invenergy, have hundreds of turbines coming online in the coming months and years. Both do not regularly use CMS but are in the process of initiating programs to evaluate the systems and believe it may be a prudent investment with a quantifiable return.
Invenergy's pilot program to test and evaluate the CMS options in the field is driven partly by wind projects rolling off their manufacturer warranties as they get older. Invenergy has eight wind plants in operation, three others coming online shortly and expects to be operating 1500 MW by the end of 2008. Major turbine orders with GE have already been locked in through 2010 that will bring the company's total operating wind plant to 3500 MW. Minus a first 15 turbine project using Vestas machines, all Invenergy's wind plant use GE turbines. None of these units, even those scheduled for 2010 delivery, include CMS, says Frank Pizzileo, Invenergy's head of asset management. The pilot program will determine whether to add third-party CMS to turbines in the company fleet. "Our fleet is growing at a tremendous rate," says Pizzileo. Middle of next year, the company's first wind plant will roll off warranty and they will continue to roll off from that point forward. "Now we're starting to think forward. Do we really have enough telemetry to assess the condition of our equipment prior to warranty rolling off and once off warranty to maintain units efficiently? What we don't want is to be running in the dark," says Pizzileo.
Horizon is on the same page. "Horizon has an interest and is evaluating what it wants to implement going forward," says the company's operations director, Mike Kelly. "We believe that predictive maintenance going forward and preserving the value of the asset is going to be enhanced by condition monitoring."
Kelly echoes fears that run deep in the industry that gearboxes will continue to be a problem area. "It seems to be that we may be facing more gearbox maintenance in the future than the industry had anticipated and so there may be even more value in the vibration monitoring and oil particulate analysis," he says.
That being said, Kelly says the case for CMS on smaller units has not been very compelling and that the return on investment for adding CMS to new units is not a certainty. The reason European owners have been quicker to embrace CMS is because of insurance requirements. Americans are figuring it out on a need basis, he explains. "The Americans haven't been forced into it by insurance companies and instead have been looking at it from a value basis; what's my return on investment for purchasing this -- and that hasn't been especially clear."
Kelly says the answer is beginning to be refined and the company hopes for its own results going forward. Horizon will test vibration monitoring units and automated oil particulate analysis units and likely coordinate with turbine vendors to facilitate the test program since some systems could come with future deliveries of turbines.
"That's where this begins to get a little political because you are in theory making a modification to your turbine and you've got a warranty provider that has a warranty responsibility and you need to make sure you are not compromising your turbine," says Kelly. "Not that this can't be done but it requires certain protocols and some manufacturers are more sensitive than others."
Lack of trust
Some operators add CMS technology prior to warranties expiring to find out the condition of the turbine -- and potential liabilities -- they might be left responsible for. The knowledge they gain can be used to make any last minute warranty filings with turbine operators. This is where a lack of trust in the industry begins to shine through, including assertions that turbine suppliers are not keen to embrace CMS technology because detailed operational data is valuable, sometimes proprietary, and could even lead to more warranty claims.
TurningPoint's Clark is among a number of industry insiders that have voiced concerns. The California company specialises in vibration monitoring and provides its systems to Clipper Windpower's turbines. Clark provides an example of once having conducted three vibration monitoring tests on one turbine for three different clients. One was the turbine manufacturer that wanted to know the gearbox it chose as standardised for its turbines was going to last. The second was the gearbox manufacturer wanting to make sure the gearbox was properly installed. And third was the end user who did not trust the first two and wanted an impartial third party opinion of the hardware.
"I think that speaks clearly to what we're hinting at, which is that there isn't a lot of trust there," says Clark, who recalls being asked the following question by a large manufacturer that was considering using TurningPoint's CMS equipment on their turbines: "Why would I want to use this information if it may point to there being a problem with my turbine before it comes off warranty to a customer."
Clark's response was that it does not matter who benefits. It could be the manufacturer avoiding a catastrophic failure early during the warranty stage or an owner after the warranty has expired. "Condition monitoring is indiscriminate, it's not biased, no one is cutting a check to dissuade you from seeing something. It's blind justice," he says.
But since warranties are increasingly expensive and running for far fewer years than previously (Windpower Monthly, January 2008), this may be less of a concern for turbine vendors. "Almost useless," is how Ed Duggan of Oak Creek Energy systems describes today's warranties. His company just installed eight Vestas 3 MW turbines in Tehachapi, California, but opted out of both CMS equipment or a long warranty through Vestas.
Cost has ultimately been the largest hurdle to the widespread use of CMS, says TurningPoint's Clark. Beyond his company and SKF, other big names in the American CMS market include Swantech, DLI and the Schaeffler Group, parent of FAG Bearings, a major supplier to the wind industry. There are many more in Europe. Most CMS units, whether vibration monitoring or oil particulate analysis, fall into the $5000-$20,000 per turbine range -- a cost that adds up for a major wind farm.
America's Electric Power Research Institute (EPRI) is one of the few independent organisations to have taken a look at the cost benefits of CMS in the wind industry. It came to a favourable conclusion, but noted cost concerns. Assuming that a CMS provided a warning early enough to provoke a remedy before major failure, that value could provide a cost savings of between $75,000 and $225,000 per event for megawatt scale turbines, according to EPRI. But it adds: "The high initial cost of installing such systems on all wind turbines at a project or in a fleet of turbines is likely to create a significant barrier for wind developers, owners and operators to make the required investment."
Proponents like Clark stress that avoiding just one major failure event pays for the system and there is good reason for why the technology is commonplace in many industries, particularly in manufacturing and process industries. Chris Walford with wind consultancy Global Energy Concepts (GEC) agrees that condition monitoring technology has proven its worth for decades, but does not believe a wind power plant's needs can be compared with a standard process or manufacturing operation.
"The value proposition for putting this on turbines is not the same as for process plants where everything has to run at the same time," says Walford. He describes assembly line factories where steel and other inputs go in one end and completed products come out the other. Sometimes this chain runs a full kilometre or more and all the equipment must be working in perfect unison throughout the entire process line.
"So if you lose one bearing you can shut down something that can cost you ten thousand dollars a minute to run. So it's a no-brainer. They do whatever they can to monitor that equipment," he says. "But for wind turbines, you've got maybe a hundred out there. You lose one, it's a problem, but the whole plant doesn't stop. It's incremental as opposed to a serial issue."
Another issue is whether actually CMS works or not. Not surprisingly, companies selling the equipment are quick to say it works, but impartial experts are not fully convinced. "Our position is, yeah it works, but not for all failure modes and not one hundred percent of the time," says Walford. The systems are extremely sensitive to the stage of the failures they are designed to detect. Results with stationary test equipment in a laboratory may be relatively easy to prove, but it is a different story altogether on an operational turbine, with all its moving parts, constantly changing loads, in a noisy outdoor environment.
This was precisely the experience of Brad Adams, of Whitewater Energy Corp, which operates hundreds of wind turbines -- predominantly older units -- in the San Gorgonio Pass near Palm Springs, California. The company bought an expensive portable bearing health indicator that measured vibration and shocks at different sensor locations on gearboxes. It proved a "time-consuming cumbersome process" that few of the company technicians knew how to operate.
"It sits idle most of the time, although we have used it," says Adams. "The biggest problem was achieving baseline information on what is normal and what is not. There are a lot of stray vibrations from other moving parts that muddy the analysis, compounding our problem."
Use of the technology has been more commonplace in Europe but there too the results are not conclusive. "I know there are places in Europe that say they've been doing this for years. The truth is, I have not seen any definitive studies by any of these people that say they've been doing this for years that shows really clear success on a number of instances," says Walford. "Maybe a spot here or there where they say they caught a bearing but they probably could have caught that flaw driving by in a truck and listening to it. I really haven't seen anything that demonstrates that it works at any level where people say I must have that."
Particularly in Germany, equipment that can detect faults before they lead to major component failure is a prerequisite for wind turbine insurance policy. German insurance company Allianz led this charge, says Walford. "I think it would be really good to have an honest dialogue between people on this side of the pond and that side of the pond to see what they are doing and how successful it has been and if the owners are really as pleased with this as the insurance companies are. I don't know the answer to that."
Even the perception that CMS is widely embraced in Europe appears non-conclusive. "The majority of the wind turbines in general are not equipped with condition monitoring systems," says Karl Steingroever with the Machinery Components and Safety Department of Germanischer Lloyd (GL), a wind turbine certification outfit of international repute. He believes that less than 10% of all wind turbines turning in Germany are equipped with CMS, though the percentage is far higher than that if just newer units are considered.
Steingroever agrees that a lack of data to support CMS is hampering its adoption. "Up to now there are not many publications available, where it is shown, that by a CMS the beginning of a failure was detected and by this a long standstill of a wind turbine was avoided. If such cases would be published more frequently, wind turbine operators would become more interested in CMS."
That being said, GL is positive on the use of CMS. The company provides offline condition monitoring checks during periodic monitoring of wind turbines. For offshore wind turbines with GL-certification, a certified CMS is already mandatory. "GL's basic opinion on CMS is that CMS is becoming more and more important, especially for larger wind turbines and for offshore installations," says Steingroever.
Adding data to data
Horizon's Kelly questions the need for the voluminous data streams that come out of CMS units and wonders what operators can use it for. In many instances, an operator needs to pay for a costly third party service to evaluate data that for some systems can be "too complex and mysterious," he says. "Does the slightest hint trigger a visual inspection, or does the particulate counting need to be taking into consideration...these are all questions we are trying to answer because we're going to have a fleet of a thousand of these things and what do you do with a mass of data coming out."
He is not alone in his doubts. Why add more and increasingly complicated data streams when existing ones are not being fully utilised, ask other experts. GEC's Walford believes a greatly overlooked issue related to monitoring is the incomplete use of current Supervisory Control And Data Acquisition (SCADA) systems that come standard with all modern turbines. They are vastly underused, he says, and with the right analysis can be used to diagnose problems in a similar way that condition-monitoring technologies can be used. "[Operators] don't use a tenth of the info they're getting fed," says Walford.
Sandy Butterfield, chief wind turbine engineer at the National Renewable Energy Laboratory (NREL) in the US, agrees. He says it would be hard to detect impending failures with the existing SCADA systems and there is a lot of room for improvement, including perhaps better integration of SCADA with condition monitoring technology. "That's really where the big plum is," says Butterfield. He has heard many in the industry say that information from SCADA systems is woefully underutilised.
"I think that absolutely needs to be exploited. We need to do a better job of capitalising on the data streams that exist already," says Butterfield. "Without doing that, it's kind of hard to convince [the owners] to spend more money on more data, which is going to require more techs to look at the data."
Maximising existing data streams and windsmith talent is where Thomas Jonsson of Energy Maintenance Service (EMS) believes big gains can be made. He knows what it takes to maintain gearboxes, having previously been the head of the North American gearbox business for Finland's Moventas, a long time supplier to the wind industry. "I've always said the best condition monitoring system is having a technician climbing up and down the tower and using all his five senses," says Jonsson, who also agrees that today's SCADA systems are underutilised.
He says operators could look at all the alarm modes and shut downs currently logged by SCADA systems. Those faults could then be rated or have other factors put behind them to allow them to be multiplied out to determine how certain components might be aging. "If you would look at the root cause for faults, maybe you would learn more," he says.
Jonsson believes there is promise in CMS technologies, particularly as a good tool for research and development of turbines when it can be placed on the turbines from the beginning, providing a good baseline to track changes from. It is known that GE uses CMS on its "exploratory" gearboxes, testing new models or new vendor products thoroughly for months using conditioning monitoring equipment.
If anything approaching a consensus is emerging it is this: the value of adding CMS as an after sales product to operational turbines is unclear. On today's modern, and increasingly large and expensive turbines, however, the additional cost is relatively smaller, the economic risk associated with component failure is higher, and it is likely worthwhile for the extra marginal cost. "If you have a big machine, you can't afford not to try it," says Walford.
Looking ahead, Walford says the equipment will soon be on all turbines. "In five years you will see it on all big turbines. It will be a standard package," he says. But getting the data from day one will be paramount. "The sooner the owners get on board and start establishing teams and procedures and baseline data, the better. If they wait a couple years, they will wish they had done it two years earlier."
It may not be Walford, however, who has the last word. GE's decision not to use CMS reflects a desire among at least some wind project developers for cheaper, more basic turbines that are simply repaired or replaced when they break down. In the end, market economics will be the deciding factor -- and they will vary from country to country, often dependent on the cost of labour.