Re: Food Prices

On Thu, 12 Jun 2025 17:49:59 -0400, Frank Krygowski
<frkrygow@sbcglobal.net> wrote:

On 6/12/2025 3:31 PM, Jeff Liebermann wrote:

 
Amazing.  I didn't know that turbine tip speed destroyed generators.
Usually, the gearbox wears out before the generator.
 
"Wind turbine gearboxes: When is 20 years not 20 years?"
<https://tamarindo.global/insight/analysis/wind-turbine-gearboxes-when-is-20-years-not-20-years/>
"So almost all gearboxes in a wind farm are likely to fail within 20
years."

>
I'm pretty skeptical about that site's claims.

You're probably correct.  I picked an article that Tom could hopefully
understand.

"It is well-known that many wind turbine gearboxes have a design life of
20 years.

To split hairs, there's no such thing as a "design life".  There is
basic rating life (L10), recommended bearing service life, modified
rating life, and some calculated rating life based on application and
environmental considerations. 

"Calculation of service life"
<https://koyo.jtekt.co.jp/en/support/bearing-knowledge/5-2000.html>

"Wind Turbine Main Bearing Rating Lives as Determined by IEC 61400-1
and ISO 281"
<https://docs.nrel.gov/docs/fy23osti/86299.pdf>
"Field data show 22 - 25% failure rate at year 20"
Source for the above is:
"Main Bearing Replacement and Damage - A Field Data Study on 15
Gigawatts of Wind Energy Capacity"
<https://docs.nrel.gov/docs/fy23osti/86228.pdf>
See Pg 10 "Results".  Read the analysis.  A 95% confidence survival
estimate of bearing life is usually done with the Kaplan-Meier
calculator:
<https://en.wikipedia.org/wiki/Kaplan%E2%80%93Meier_estimator>
although there are other methods:
<https://en.wikipedia.org/wiki/Survival_analysis>

"However, it is also well-known that a great number of gearboxes don’t
last for 20 years and fail prematurely. Why the discrepancy?

Because "design life" (whatever that might be) and premature failure
are two different calculations.  Design life is probably the MTBF of
all the bearing components when they are simply multiplied together.
The premature failure calculation is probably where 95% of the bearing
assemblies survive to 20 years.

"The answer lies in the way that gear and bearing lives are defined. We
can’t predict the exact date on which a component will fail, but we can
estimate the probability that it will last for a given duration.

Yep.  That's where the 95% survival rate comes from.

"A very simple calculation shows that, if we combine the expected life
for every bearing in a drivetrain to calculate a ‘system level’ life,
the probability of one or more bearings failing within 20 years is up to
93%. So almost all gearboxes in a wind farm are likely to fail within 20
years. It seems shocking, but isn’t far from reality."

Well, one mistake.  The 93% is the survival rate, not the failure
rate.  The big problem is that the author does not mention the bathtub
curve.
<https://upkeep.com/learning/bathtub-curve/>
That's where there are more failures at the beginning (initial
manufacturing and installation errors) and at the end of life.  In
between, there are few failures.  The trick is to replace the part
before end of life.

They seem to be implying that each gear, bearing, shaft, etc. are
individually designed for 20 year life, but that because there are
always some premature failures, 93% will have _something_ fail sooner
than 20 years.

That was the mistake I previously mentioned, where the 93% is the
survival rate and not the failure rate.  Does a 7% failure rate sound
more reasonable?  Go to:
<https://upkeep.com/learning/bathtub-curve/>
See graph on page 12.  The author calls it "hazard rate" or
"conditional failure rate". 
"In the current context, h(t) estimates the probability of a main
bearing being replaced in the next month, given that it has survived
to time, t (for t in years)."
Notice from the graph that for most of the 20 years, the "hazard rate"
is 0.015%.

ISTM that doesn't mean the wind turbine comes down. It means the failed
bearing must be replaced, just like a car's failed wheel bearing, failed
throwout bearing, etc. You don't junk a car at the first bearing failure.

Agreed.  That's the likely conclusion based on our life's experiences.
However, the problem here is how to correctly calculate when
replacement is necessary. 

--
Jeff Liebermann                 jeffl@cruzio.com
PO Box 272      http://www.LearnByDestroying.com
Ben Lomond CA 95005-0272
Skype: JeffLiebermann      AE6KS    831-336-2558