There is dire need for the system certification of condition monitoring, mostly because current CMS vendors offer at least nine different approaches for basic sensor selection and placement. This confusion originally prompted the call for a certification for owners to assist in the selection of a viable CMS for wind.
After speaking at recently NREL and UVIG events, I was asked if there was a certification for CMS systems in the wind industry. The answer is yes…and no. To investigate further, we will go through the available standards for CMS in wind to find what is helpful.
The standard for vibration in wind turbines is listed as ISO 10816-21. This ISO standard states it “does not apply to diagnostics or fault detection.” It also states that the vibration “evaluation zone boundary values are not intended for use as acceptance values. These must be agreed upon between the manufacturer and user.” So how does the standard help?
To be clear, if you adhere to ISO 10816-21, there is no guarantee you will have a successful CMS program. Also note that the standard applies to a rather wide range of wind turbines, from 200 kw up to 3-MW ratings.
So, what does the standard suggest regarding monitoring? It suggests the use of piezoelectric accelerometers and separates equipment into rotating and non-rotating components.
For rotating components, the standard suggests:
Sensor placement – For rotating components (drive train is a given), it advises the use of sensors in three axis across the drive train. This is not standard for many reasons. Using accelerometers in three axes is not beneficial in determining a failure component. A single sensor can work except when determining some defects. However, as this standard states, its purpose is not intended to measure specific defects.
The main bearing – Three-axis measurements are recommended but not demodulated readings for detecting early bearing defects, nor time waveform measurements. The standard makes no suggestion for the number of lines-of-resolution in the spectrum, averaging, or overlap.
Gearbox – Three-axis measurements again but no demodulated reading for early bearing defect detection nor time waveform measurements. No suggestion for lines-of-resolution in the spectrum, averaging or overlap.
Generator – Three-axis measurements yet no demodulated reading for early bearing defect detection nor time waveform measurements. No suggestion for lines-of-resolution in the spectrum, averaging or overlap.
Number of sensors
It is not practical to use the suggested 17 sensors across the drive train. Typically, most commercial CMS systems use 7 to 8 sensors. Seventeen of them are cost prohibitive. They will also generate too much data. Consider that 7 to 8 sensors installed on 100 towers will produce an annual data volume is in excess of two-million measurements based upon once-a-day intervals.
Using three sensors on a generator bearing to find a bad bearing within is simply not practical. Installation costs go up, data volume increases, hardware costs double, and network load increases as well to get the data out. So this portion of the specification is simply not practical.
On Non-rotating equipment:
Sensor placement – The standard suggests three axes in two locations to monitor the bedplate of the tower for structural purposes. This is more for a process and controls purpose than a condition-monitoring function. Most bedplate monitoring I have been involved with is at a prototype level, certainly not at a fleet level.
Number of sensors – The standard specifies installing eight low-frequency sensors. Again, as with 17 sensors on the drive train, adding eight expensive low frequency sensors is not a good suggestion. It is doubtful that any owner would install eight sensors per tower to understand the vibration of the structure.
Measurement parameters – The standard suggest performing evaluations in 10-minute periods. This really requires specialized equipment. It is certainly not something that is fleet advisable. At 10 minutes per tower, the data volume would be significant, and the data storage and analysis substantial.
Vibration levels – The standard suggests measuring only acceleration and velocity. The accompanying chart (above) shows overall vibration levels in velocity measurement units.
Other standards
ISO 10816-21 standard does not give the end user guidance for acceptable vibration levels, diagnostics, or fault detection. So what can it offer the owner? It does advise as to structural vibration measurements and parameters as well as sensor types and placement. To be clear, not one CMS system commercially available today, marketed and installed, meets this standard.
However, portions of two other suggested standards might be helpful to owners. They are:
• VDI 3832, for rolling element bearing noise and vibration. This also has a wide range of application in wind turbine size. And,
• ISO 13373-2, techniques for bearing and gearbox defects analysis.
So between rotating and non-rotating components, the suggested 25 sensors per tower is literally three times the amount normally installed. Owners show pause at the current eight-accelerometer pricing per tower. Twenty-five is likely not realistic.
Here is what’s missing and crucial to a successful CMS program:
Quality of the sensors – There are at least 10 to 12 sensor factors that are not covered. Cables are not even discussed.
Ability of the software – There are no specifications on measurement parameters, measurement types, resolution, averaging, and a dozen other measurement factors.
Ability of the analyst – There currently are three companies that certify vibration analysts. It certainly makes sense to have an analyst who is certified but this is no guarantee that the person is a good analyst. Ironically, the certifying bodies that teach the theory and background of vibration analysis measurements, teach very little on actual vibration analysis.
We will discuss these important factors in a following article. We will also look at what makes sense for a standard for CMS installs to benefit the owners, ISPs, and manufacturers.
