June 2019

Most wind turbines tend to be in remote locations, making physical access to the gears, bearings, and control systems difficult and not always practical. Condition monitoring programs for wind turbines generally include time-scheduled oil analysis. Unfortunately, servicing of wind turbines is often limited to specific times each year. When the first large wind turbines were installed more than 20 years ago, oil samples were only taken every six months or so. Trend analysis for important oil quality parameters was difficult and erratic.

Consequently, oil analysis and condition monitoring for both onshore and offshore wind turbines are being done increasingly using online oil and machine monitoring sensors, similar to those being used in automotive engines and marine systems. The oil sampling intervals are thereby reduced from every six months or so to either daily or continuously. This allows a more precise trending of important oil and machine parameters.

Oil Properties

The most important oil properties that need to be monitored for wind turbines are:
• Viscosity.
• Acid number.
• Oxidation.
• Water content.
• Additive contents, particularly phosphorous.
• Elemental contents, particularly wear metals.

All these can now be done using electronic sensors from a number of suppliers. Bosch has developed a multifunction oil-condition sensor, which determines oil level and oil condition. The oil level information enables the system to be monitored for leaks. The oil-condition sensor constantly measures the oil’s viscosity, permittivity, conductivity, and temperature.

The measured viscosity and permittivity (or dielectric constant) are the primary values supporting the oil-condition evaluation. A novel microacoustic device determines the viscosity. This device uses the piezoelectric effect to electrically excite high-frequency mechanic (or acoustic) vibrations at a sensitive surface.

When this sensitive surface comes into contact with the oil, the electrical device parameters, such as oscillation frequency and damping, are changed according to the oil’s mechanical properties, especially viscosity. In contrast to conventional viscometers, the microacoustic sensor does not contain any moving parts.

Oil degradation and wear metals can be detected using a dielectric sensor, and several companies, including Mercedes-Benz and Delphi Corp., now use these devices for online oil monitoring. Wear metals and particulate contaminants also can be detected using online optical particle counters. When particulate contents start to trend upward, more detailed oil sampling and analysis are required.

Spectroscopy vs. Sensors

In a conventional oil analysis laboratory, Fourier Transform Infrared (FTIR) spectroscopy can be used to monitor oxidation, water content, and additive elements. An online sensor developed by the Technical University of Munich has been found to be able to provide test results for acid number, oxidation, water content, phosphorous, silicon, zinc, and copper that agree closely with those obtained by conventional FTIR.

The sensor consists of a collimated infrared light source, a cuvette, a Linear Variable Filter (LVF) and a linear pyroelectric detector array with application-specific integrated circuit (ASIC). The LVF is a spectral device that uses a Fabry-Pérot structure to provide virtually constant resolution over the required wavelength region.

Compared to conventional FTIR spectrometers, the LVF spectrometer has a reduced resolution and a limited spectral range. It operates in a range of about 1,800 cm-1 to 900 cm-1 at a wavelength dependent resolution of about 36 cm-1 at 1,800 cm-1. (Standard FTIR spectrometers normally operate in a wavelength range of 4,000 cm-1 to 500 cm-1 at a spectral resolution of 4 cm-1 or better.) Even so, the new sensor has been shown to provide acceptable correlation with laboratory test results, using data from several hundred oil samples. Extracting the oil quality parameters from the measured infrared spectra requires sophisticated data processing, using a multivariate regression model.

Internet Accessible

Data from all these sensors, all of which can be fitted to the bypass filtration line of a wind-turbine gearbox, can be fed into Internet-based data analysis systems, allowing electrical system maintenance management to track the condition of an entire array of wind turbines in real time and receive email alerts when emerging problems are identified.

Steel-to-steel contact is why many bearings fail, why proper lubrication is critical, and why The Timken Company, manufacturer of engineered bearings and power transmission components, works hard to develop advanced surface coatings that combat the problem in the toughest conditions.

To extend the life of roller bearings operating in wind turbines, Timken developed ES302, a special coating that adds durability where adhesive wear (the rubbing together of two metal surfaces) is a significant problem. Follow along as Timken bearing experts Vikram Bedekar and Doug Lucas explore the science behind this proprietary surface treatment that has been proven to prevent common damage modes in wind bearings.

Stronger Bearings Wanted

Wind turbines place huge demands on mainshaft and gearbox bearings as many wind-farm operators already know, having experienced the disruption of a major mechanical overhaul that can cost $100,000 or more anytime a bearing is unexpectedly taken out of service.

By the late 2000s, with bearing failures mounting, the U.S. wind-energy industry began to widely recognize that turbines designed to produce electricity for 20 years were requiring large corrective work within five to 10 years. Early wind-turbine designs, for instance, tended to use two-row mainshaft spherical roller bearings that, due to the dynamic loading forces commonly experienced in turbines, were required to support a radial reaction as well as wind-thrust loading on only the downwind row of rollers.

This uneven pressure led to premature mainshaft bearing wear while also transferring damaging axial deflection loads to gearbox bearings. Furthermore, wind-energy producers found it challenging to keep bearings well-lubricated under these complex loading conditions. In wind turbines, lubricant films are particularly difficult to maintain, and bearing damage can occur quickly. As years passed, the urgency for wear-resistant bearing solutions grew stronger as turbine operators looked for answers.

Advancing Wind Energy

Timken has been committed to the wind industry for nearly 20 years with dedicated research and development, engineering, and field service teams actively engaged in solving the industry’s biggest challenges. By 2009, the Timken R&D team was nearing the production stage on a special plasma coating process for roller bearings that it had been pioneering since the late 1990s.

At that time, physical vapor deposition — or PVD — was a technology unfamiliar to most. PVD is a vacuum deposition method used to produce thin films on objects such as bearing rollers, gears, and engine components. Previously, Timken had demonstrated promising results for PVD-coated bearings in various oil and gas, mining, military and pump drive applications and saw an opportunity to help wind producers increase their productivity by bringing the same surface engineering technology to turbine mainshaft and gearbox bearings.

About PVD

PVD uses physical processes such as heating, sputtering, or transfer by laser or electron beams to produce a vapor of material, which is then deposited on the object that requires coating. This highly complex process is a rigorous one that must be controlled within world-class production facilities and with stringent design and quality specifications. For wind bearings, Timken chose a plasma-based PVD sputtering method whereby atoms are ejected from a source material (the target) by high-energy particles and transferred to a substrate.

First, bearing rollers are placed in a chamber in a vacuum environment and introduced to very low quantities of a special mixture of reactive gases (Ar/C2H2). These gases react with the metallic targets (such as WC) in the presence of strong electric and magnetic fields. The atoms from the target are then dislodged by high-energy collisions (that is, plasma ions bombard the source material, causing some of it to ionize). These ionized atoms then condense on the surface of the object in microthin layers.

PVD coatings are more uniform, more durable (up to six times) than conventional electroplated or spray coatings. PVD is also considered to be a more environmentally friendly process as it does not produce any hazardous waste materials to dispose of.

Once confined to special laboratories, PVD methods have been scaled up in recent years to meet the needs of heavy industry, with ongoing development allowing the production of new coatings with properties that were previously unachievable.

Essentially, PVD methods allow coatings to be deposited on surfaces atom by atom. This precise, nanometer control of the coating structure allows Timken to create coatings that possess the exact performance characteristics an application requires. This level of accuracy and consistency is simply not achievable using a traditional bath dip or spray-paint process. The result is a fabrication of atomic layers having the exact microstructure and properties desired.

Solving for Steel-on-Steel

By employing PVD coating technology to make roller surface less “steel-like” than the bearing rings, Timken theorized it could effectively combat the steel-on-steel adhesive wear issues plaguing turbine mainshaft and gearbox bearings.

The coating Timken developed — dubbed ES302 — is a super-thin engineered combination of nanometer-sized ceramic particles embedded in an amorphous hydrocarbon matrix measured in fractions of a human hair. The coating’s nanocomposite, metal-carbide matrix design provides extremely hard and low-friction, near-surface properties that inhibit microwelding and adhesive wear at the roller and race interfaces.

Essentially, ES302 is both hard and wear-resistant like a drill bit, yet slippery and nonstick like a frying pan. When applied to bearing rollers, the coated surface acts as dissimilar material to the steel rings to reduce adhesive wear. Such coatings have proven effective in low-lubrication conditions or where debris is present, offering a wide range of advantages in heavy-duty bearing applications. Benefits can include reduced friction and torque, reduced component scuffing and smearing damage, decreased false-brinelling wear, and greater oil-out protection.

Excitement for ES302

During ES302 development, Timken experts traveled the world to talk to wind-farm operators, OEMs, and others about the problems they were experiencing. This direct customer contact was key to uncovering real-world challenges, building collaborative relationships, and gathering extensive feedback on wear-resistant bearings. In 2011, Timken began sending prototype parts out for evaluation and received promising reports in return. In one instance, a bearing that had previously lasted just 200 hours was documented to have crossed the 2,000-hour threshold with no disruptions, prompting an enthusiastic email from the customer.

At the same time, Timken conducted laboratory testing that indicated an ES302-coated bearing could experience up to six times greater life when operated in standard conditions, and up to three times greater life under debris-contaminated conditions compared to an uncoated bearing.

The company soon established a healthy aftermarket strategy of replacing standard mainshaft and gearbox bearings with wear-resistant bearings featuring ES302 coating on the rollers. Furthermore, the new coating was showing promise in agricultural applications such as large farming implements, tractors and combines. Timken aerospace customers also found uses for ES302, along with paper mills and cement plants where high temperatures, water ingress, and abrasive particles pose a constant threat to bearings.

Ultimately, the investment by Timken to identify fundamental problems in customers’ applications and to work on the cutting edge of engineering and scientific knowledge to solve them is what spurred the success of the ES302 project. As materials and processes shift over time, it is a collaborative mindset that keeps Timken focused on the problems customers care most about.

Award-Winning Innovation

Since commercializing its proprietary coating, Timken has garnered widespread industry acclaim including the prestigious Engineering Materials Achievement Award from ASM International in 2017. The pre-eminent materials information society recognized Timken for ES302’s “significant contributions to the field of materials science and evidence of industrial impact.” Previous winners of the annual award, established in 1969, include General Electric Co., Northrop Grumman, IBM, Boeing and General Motors Corp., to name a few.

For Timken development team members, the honor brought home a broader sense of how much their customers and technical peers appreciated the work that went into developing ES302. Many Timken engineers and analysts made valuable contributions to the project, including Gary Doll, Timken professor of surface engineering and director of the Timken Engineered Surfaces Laboratory at the University of Akron; Ryan Evans, director of R&D for bearings; Jim Gnagy, product development specialist; and Tim Springer, program manager.

Field-Proven Performance

Recently Timken was presented an opportunity to reinforce its initial lab results when one 230/600-series ES302-coated mainshaft bearing was removed from a 1.5-MW turbine in New Mexico (due to nonbearing-related issues) and returned for analysis following seven years of reliable service. Because the turbine was planned to be repowered, the operator didn’t relubricate the bearings for the last two years of operation.

A close examination showed that the bearing exhibited little to no adhesive wear, and it demonstrated the expected surface finish and remained consistent with the original design profile, leading Timken to conclude the bearing was in “very good” condition for this stage of operation and the lack of relubrication for two years. Furthermore, there was no evidence of the bearing entering the next stages of damage. Based on its own application engineering practices, Timken can confidently estimate the bearing would have continued to provide reliable, trouble-free operation into the 15- to 20-year time frame.

The Power Behind the Power

Currently there are more than 1,000 ES302-coated bearings installed in wind turbines around the world. According to the American Wind Energy Association (AWEA), U.S. wind power alone now supports a record 114,000 jobs, more than 500 domestic factories and more than $1 billion a year in revenue for states and communities hosting wind farms. With the ability to power more than 30 million homes, the industry is bringing economic opportunities to towns and cities across America.

Timken is proud to play its part in helping to grow renewable energy technologies by producing wear-resistant bearings that customers do not have to think about again. ES302 and other surface engineering developments from Timken help bearings do their jobs dependably wherever rotating parts require them.

Today, Timken researchers continue to look for ways to reduce costs and make coatings more accessible to a wider variety of customer applications. By partnering with energy producers, OEMs, and centers of excellence worldwide, the company is delivering on its mission to engineer the next generation of coating technologies that improve bearing life and eliminate downtime.

How Bearing Damage Occurs

In a standard mainshaft spherical roller bearing, radial and thrust forces result in higher loads on the downwind row of the bearing, as well as a full 360-degree loaded arc of rollers.

For every shaft revolution, the full-loaded arc of rollers increases the number of stress cycles occurring on a point on the inner raceway of the bearing.

Given that a mainshaft typically rotates at slow speeds in the 25 to 35 rpm range, this may not generate a significant lubricant film, even with higher-viscosity lubricants.

Taken together, the result is greater loading, more stress cycles, and thinner lubricant film thickness on the downwind row of the bearing, which increases the risk of micropitting and other damage modes.

One ice detector solution for OEMs and operators is to borrow off-the-shelf technology from the global telecommunications industry.

Smart phones are manufactured in extremely high volumes around the world. This has forced the price of optical lasers and LEDs down to just pennies apiece. Plastic optical fibers are likewise made in vast quantities, with prices just pennies a foot.

With today’s tele-components being reliable, common, and cheap, they should make excellent building blocks for wind-turbine ice sensors — and they do. (Figure 1)

Modern ice detectors are entirely optical. In the modern ice sensor, a tele-component light source and receiver monitor the opacity and refractive index of whatever substance is in contact with the probe. Other than ice itself, little else would have intimate contact with an ice sensor high atop a wind-turbine nacelle. Liquid water simply runs off by gravity, but ice sticks to the optical surfaces for sensing.

Since they have no moving parts, modern ice sensors and housings can be scaled very small — about the size of a thumb — and potted solid against the elements. This is an improvement over hermetic sensor housings, because solid sensors are robust against moisture and vibration as well.

Modern ice detectors work as a combined optical spectrometer and optical switch. A change in opacity registers as rime ice; a change in refractive index registers as clear ice.

The wavelength of the sensor’s excitation signal is invisible to the human eye, so as not to cause a distraction. At the same time, its clear sensor probe is useful to display a blue ice-indicating LED, helpful to turbine technicians working aloft (Figure 2).

Modern ice sensors eliminate many of the problems with legacy ice sensors, being generally lower in cost, lighter in weight, and smaller in size. Simple relay contacts eliminate the need for interpretation and send a “go/no-go” “ice/no-ice” indicating status to the host control system. This simplifies the interface while reducing software complexity and cost.

Optical ice sensors (Figure 3) are lightweight and easy to carry aloft and install. They run on any convenient DC power source 8-30 volts and need only 2 watts to look for ice. Unlike legacy sensors, modern ice sensors detect and report the absolute presence of ice on a wind turbine, do not artificially melt the sensed ice away, and do not require additional power supplies and power budgets.

Bearing heat signatures

Following the path of legacy ice sensors, modern ice sensors were also recently developed for aircraft. Smaller, lighter, more sensitive, and cost effective, they are easily adapted to wind turbines and autonomous drones for wind farms.

Other, wider applications for modern ice sensors include virtually any place with a need to sense the formation of hazardous ice and take corrective action in a timely manner, such as heat pumps, radio and TV broadcast towers, HVAC cooling towers, hydrogen car engines, traffic signals, vehicular bridges and overpasses, remote power lines, water-level sensors in oil and gas installations, and more.

Ice sensor prices are eroding because overall demand and production volumes are increasing.

They are increasing because modern ice sensors enjoy wider applications in high-volume drones and refrigerator defrost controls. The ultimate manufacturing volume for optical ice sensors is projected to approach that of musical greeting cards (Figure 4) in multi-million-unit volumes, with prices expected in the sub-$10 range.

Tested in the the world’s largest icing wind tunnel at NASA Glenn (Figure 5), modern optical ice sensors comply with SAE Aerospace Standard AS 5498 ¶ 5.2.1.1.1 for FAA-mandated ice protection of autonomous BVLOS (beyond visual line of sight) drones.

FAA compliance is a major breakthrough for wind-farm drones. Modern ice sensors invite drones to use the darkness of night to autonomously monitor the IR heat signatures of turbine bearings and other equipment for signs of impending failure.

If the sensors flag anomalous high temperatures, observe nighttime intruders, the movement of wildlife, flocks of birds, or colonies of bats, the sound of an ultrasonic horn will safely shoo them away, autonomously, without damage to the blades or requiring the services of an FAA-licensed drone pilot. The pilot is built right into the operating system, along with a virtual GPS-defined “fence,” delineating airspace beyond which the drone is disabled, and cannot fly.

Dynamic blade inspection

Modern fifth generation ice sensors are capable of detecting the formation of ice films only molecules thick.

Upon sensing an initial film of ice molecules, the modern ice sensor of a BVLOS wind-farm drone (Figure 6) would help protect the drone against ice by sending an interrupt to the host operating system and prompt it to briefly reverse the fan blades. The top surfaces of a drone’s fan blades are more susceptible to ice molecules than the bottom surfaces because air above the blades is colder due to lower pressure (lift), while the air below the blades is warmer due to higher pressure.

Warmer air briefly swept over the top of the fan blades overcomes newly-formed ice molecules’ latent heat of fusion, changes their state back into liquid water, and retards formation of ice on the drone. (The latent heat of fusion of water is 80 calories/gram.)
This proven fan-reversing ice removal technique is common on HVAC cooling towers for large data centers, apartment buildings, campus buildings, hospitals, etc., which also use modern optical ice sensors for ice protection.

Daytime and nighttime detection

Protected against ice during daylight hours on a fixed schedule under SCADA control, autonomous wind-farm drones would take off and inspect a specific circuit of turbines.
Rather than stop the turbines and lose valuable production time, each dedicated wind-farm drone would take advantage of its inherent agility and inspect turbine blades in a spiral on the fly, without stopping the turbine. Record high resolution blade images and other key features, fly autonomously to the next turbine like a honey bee collecting nectar in a field of flowers. It then would return to its garage for a robotic battery swap-out or charge.

Protected against ice during nighttime hours, autonomous wind-farm drones would also serve as night watchmen, using the cover of darkness to monitor the heat signature of turbine bearings and other equipment for signs of impending failure.

In the unlikely event of fire or other anomaly, SCADA would immediately dispatch its BVLOS drone for video monitoring and situational awareness before technicians and fire apparatus could arrive.

In addition to end-of-warranty inspections and preventive maintenance, inherent advantages of ice-protected autonomous drones include safety, objectivity, uniformity, accuracy, low cost, and real time dissemination of critical data.

Ice on wind turbines may be hazardous to people and things, but the use of abundant, low-cost optical components to make ice sensors provides simple, inexpensive solutions for turbines and dedicated on-site drones.

SIDEBAR

A dozen other ways to detect ice

Daytime visual: Visual observation of ice on turbine members during daytime hours, with field glasses or naked eye.
Nighttime visual: Strategic illumination of turbine members with either white or red light. A red light shining upward through an acrylic panel is inconspicuous at night, but when ice forms on it, the red light diffuses and provides a visual form of ice alert.

Mechanical obstruction: A scraper rotating on an exposed surface. As ice accretes on the surface, the torque required to rotate the scraper increases. At some preset torque, a signal is generated, setting off an ice alert.

Differential pressure: A hollow probe senses air pressure through small orifices on its upwind face. As ice blocks the orifices, a pressure differential signal is created setting off an ice alert.

Latent heat: A periodic current pulse is sent through a resistance element to heat an exposed probe. If ice has accreted on the probe, the temperature increase will temporarily dwell at 0°C and set off an ice alert.

Vibration: Ice on a vibrating reed slows its resonant frequency, which is detected and used to energize a heater, which melts the ice and pulses pneumatic boots on an aircraft wing. The frequency of pulses sets off an ice alert.

Microwave: A microwave transducer consisting of a resonant surface waveguide is embedded into an exposed surface on which ice accretes. The ice acts as part of the waveguide; changes its phase, and sets off an ice alert.

Electromagnetic beam interruption: An EM source is placed on one side of a flattened tube and directed at a sensor on the other side of the tube. As ice accretes on the tube, the signal is blocked and sets off an ice alert.

Ultrasonic: Sound waves are reflected off an ice/air interface and sensed by a transducer. When ice is present, the reflected waves are perturbed, sensed, and used to set off an ice alert.

Capacitance: A total impedance ice detector uses the capacitance of a surface-mounted electrical circuit to determine the presence and thickness of ice and sets off an ice alert.

Optically occluding: An optical source directs light at an optical receiver. An accreting surface close to the beam accretes the ice, senses ice blocking the path of the beam, and sets off an ice alert.

Optically refractive: An optical source sets off an ice alert by using un-collimated light to monitor the opacity and optical index-of-refraction of whatever substance is on the probe, either air or ice.