Oil Change Metric
For most operators when to change oil in a wind turbine gearbox is harder than reading tea leaves. Oil analysis results suffer from manual sampling, testing, and inherent noise. This leaves most operators without a true condition based approach to lube system maintenance. Our advanced analytics delivers an oil health metric based on 30,000 years of gearbox oil history. We clearly recommend where oil change, flush, filter change, and similar lube system actions must take place. On average this extends the oil life in a fleet by more than 1 year. Saving thousands of dollars over the life of each turbine at the cost of tens of dollars per year.
Let’s go through some of the raw oil analysis data to show why it is near impossible to draw an actionable condition based conclusion using your current methods.
First, let’s look at the averages of the raw attributes at different oil ages. Looking to see if within 5500 oil samples, the attribute values increase over time, or if there is a definitive trend to follow. The red circles are the year that the attribute peaks:
There are no definitive trends in the raw attributes.
Next, let’s look at how these attributes change from the first oil sample to the oil sample 4 years later. The following is a study of 30 oil changes. Values that increase by less than 5% are in green:
One can see that the majority of the time the oil attribute values actually got better! It is usually not conclusive which oils should have been changed by looking at the raw data.
To solidify this concept, let’s play a game. In the following 6 samples, try to pick which one is in year 4. It’s hard:
One might think that the 3rd sample has the highest Fe, Si, and total acid number, and is therefore the most “used up”. The actual case is as follows:
The 4th sample down was a year 4 sample, and a few months later the oil was changed. Moreover, about 6 months later, the gearbox failed, and had to be replaced.
This is standard case where the oil analysis was giving information, but the operator did not have the tools to understand it.
Our analysis gives out a metric that is a function of all the attributes available. Rather than just a trend of individual noisy attributes; we utilize a multi-variate function to identify risk.
Our output would have warned that prior to the oil change, on 7/30/2008, there were significant risks. We deliver a simple “Severity” metric for the lube system. If the metric is above 0.1, there is cause for alarm. For the turbine in question, the metric was more than double that, at 0.23. In this case we would have recommended an oil change, flush, borescope, and potentially a re-alignment if necessary. This turbine would have been on high alert, and we would have likely avoided the gearbox failure that happened 6 months later due to our investigations. Our output on the above 30 cases would have been as follows:
Our oil change or health metric creates a clear condition based approach for the lube system. This is a capability that is not available any where else. We charge as low as $19/MW/yr for this tool.