Anyone here work with wind turbine blades and understand fatigue?

[Lt72884]

I am looking for some guidance in understanding a few things. I figured why not ask at the rocket forums because this place has a very varied group of people haha
I understand how fatigue works and what it is, but im getting some strange outputs from a numerical methods based opensource program (openFAST) when finding "time until failure in seconds"

The software uses the ultimate design load and then calulates time to failure. So what i did, since i do not have access to ANSYS or any fea software, i adjusted the ultimate design load in the software until the time to failure was 20 years. This L_ult value happened to be 806kN at a wind speed of 12.5m/s

Here is whats strange. When i use a wind speed of 4.5m/s, and use the 806kN as the ultimate load, the time to failure is 726,000 years. So, once again, i adjust the ultimate load until the time to failure is 20 years. in the slower wind speed case, the time to failure of 20 years had a ultimate load of 280kN This seems way way off. The wind speed should not affect the ultimate design load.

so this begs the question, of whats going on?
Ultimate design load is what the entire component can take before failure. IE, a popsicle stick can handle 20lbs force (bending) before it breaks.
Im not seeing how the ultimate design load can be drastically reduced to 280kN when the ONLY input to the system that changed was the wind speed..

thanks for any knowledge and advice given for this

 

[bjphoenix]

There are too many variables involved, and we don't know the algorithm that is being used to estimate fatigue life. The algorithm seems to be suspicious since it gives 726,000 years for one case. Normally as fatigue life is extended you reach a point where the item will not fail in fatigue for any length of time.

Fatigue life is normally a function of the applied load vs. the ultimate capacity. If both scenarios have the same fatigue life, then both scenarios would have the same relationship of applied load vs. ultimate capacity. If the first scenario has a given wind speed and the second scenario has a wind speed that only applies 13% as much load to the system, you would expect that this would only work if the item had 13% of the ultimate strength.

Even if you perform FEA analysis, that only gives you forces, or perhaps stresses, in the system. You then need a separate analysis to predict fatigue life from the stress range. The shape of the object and the type of material play into this. Imperfections in the system are important, things like grooves, notches, transitions, bolt holes, welds.

I'll admit that all of the above relates to steel structures, and somewhat to aluminum. I believe fatigue behavior is slightly different for composite structures.

 

[Blast it Tom!]

Yeah. I largely agree with @bjphoenix. Other than to say, with composites, you don't have an endurance limit, it'll break after enough stress cycles.
Wind speed as in input may be fed to the algorithm as affecting rpm. The slower rpm would affect the "wall clock" time to failure. What you're really after is no. of cycles to failure. But I can see how that can quickly get complicated. The load on the blade is a function of wind speed, for both propulsion, a bending load which increases as speed increases, and direct centrifugal stress at the blade root, also increasing with the square of the speed. Both of these have additional alternating components due to gravity as the thing spins. And that leaves out possible gyroscopic forces as the fan turns into the wind. Even the points of maximum stress likely change due to varying the blade pitch.
So lower design loads, whatever such a single number like that could represent, could really stretch out the life.
But like I always say to my younger cohorts, don't trust a black box. Pop the hood and see what that calculation is actually doing. You do NOT want to see a wind turbine coming apart at speed, believe me.
Or, as another member's signature line says, "Every simulation is guilty until proven innocent."

ETA: Obviously it's a stress range and stress cycles that causes fatigue cracking. I forgot a big one -startups and shutdowns. So that fatigue scenario is quite complicated, with perhaps lower stress ranges when operating, but at a higher number of cycles/time, and startups and shutdowns, with bigger stress ranges and lower frequency. So there must be a lot of figuring going into that one number.

 

[Rschub]

This seems way way off. The wind speed should not affect the ultimate design load.

This is a fundamental error, the wind speed is the applied load.

 

[Blast it Tom!]

I had a quick look, that openFAST is quite the package! Have we provided any help? It seems to me, after some consideration, that in order to be able to take one load value and output an expected design life, they must have what's called a "duty cycle", one more severe than is likely to be encountered in service. This would include operating hours per year - and apparently you can set the speed - as well as expected number of starts from zero to speed, which is often a bigger stress range than experienced during operating. Then as well, there may be an allowance for some upset conditions, feathered prop during a certain number of high wind events/year, etc.

Anyway, hope we helped!
Edited to emphasize that the user enters the speed and the stresses and number of cycles from that are then incorporated into the fatigue life calculation.

 

[Hobie1dog]

interesting topic. I keep remembering the video of the wind turbines blades flying apart.

 

[Lt72884]

This is a fundamental error, the wind speed is the applied load.

correct, wind speed is the thing that makes the blades turn its strange that it effects it in an opposite way.
I just find it odd that at such slow wind speeds, the blade can only handle 280kN for 20 years, but at very fast wind speeds, the ultimate load is 806kN for 20 years.

all parameters and initial conditions for the blade are the same. All i did was change the wind speed to test it and noticed the loading values were very strange

 

[Lt72884]

I had a quick look, that openFAST is quite the package! Have we provided any help? It seems to me, after some consideration, that in order to be able to take one load value and output an expected design life, they must have what's called a "duty cycle", one more severe than is likely to be encountered in service. This would include operating hours per year - and apparently you can set the speed - as well as expected number of starts from zero to speed, which is often a bigger stress range than experienced during operating. Then as well, there may be an allowance for some upset conditions, feathered prop during a certain number of high wind events/year, etc.

Anyway, hope we helped!
Edited to emphasize that the user enters the speed and the stresses and number of cycles from that are then incorporated into the fatigue life calculation.

It is a great software and helps alot with analysis. I have made sure all initial conditions are the exact same between the two tests. The only difference between the 2 trial runs is the wind speed. Nothing else in the parameters has changed. pitch angle, azimuth, startup and length of trial, plus about 15 other parameters.

A slower wind speed should mean the blade should last longer due to less loading, with the parameters i have set. I have verified that the max loading on the blade is less with a smaller windspeed, than the larger windspeed, but when the software calculated the time of life, it states that the lower wind speed can handle only 280kN for 20 years, but at a speed 3 times higher, it can handle 806kN for 20 years.
the output should read "at 4m/s the blade will last considerably longer than 20 years due to the ultimate loading being 806kN for a 12m/s wind speed"

 

[techrat]

I understand how fatigue works and what it is

As someone still a member of the workforce, I understand fatigue is, all too well. (i need coffee)....

 

[Lt72884]

interesting topic. I keep remembering the video of the wind turbines blades flying apart.

i have seen a few videos of that and its pretty crazy

 

[Lt72884]

exactly.

As someone still a member of the workforce, I understand fatigue is, all too well. (i need coffee)....

Exactly haha.

 

[Blast it Tom!]

It is a great software and helps alot with analysis. I have made sure all initial conditions are the exact same between the two tests. The only difference between the 2 trial runs is the wind speed. Nothing else in the parameters has changed. pitch angle, azimuth, startup and length of trial, plus about 15 other parameters.

A slower wind speed should mean the blade should last longer due to less loading, with the parameters i have set. I have verified that the max loading on the blade is less with a smaller windspeed, than the larger windspeed, but when the software calculated the time of life, it states that the lower wind speed can handle only 280kN for 20 years, but at a speed 3 times higher, it can handle 806kN for 20 years.
the output should read "at 4m/s the blade will last considerably longer than 20 years due to the ultimate loading being 806kN for a 12m/s wind speed"

Right! I somehow didn't get that out of you initial post, sorry!

 

[robopup]

It's generally unreasonable to expect actionable answers to solve a specific problem you have with some complex solver without providing your input file (to someone who knows what they're doing). My day job is at an FEA company and if a customer has a problem with a simulation, I usually won't lift a finger until they send me their input deck unless they have a really good reason not to. Without the input, you're almost always just throwing darts at the board.

Also, have you looked into what physics this software is solving? Relevant details of the material model used? Maybe you don't have an issue with your input file; you're somehow outside the assumptions of the model.

 

[Blast it Tom!]

It's generally unreasonable to expect actionable answers to solve a specific problem you have with some complex solver without providing your input file (to someone who knows what they're doing). My day job is at an FEA company and if a customer has a problem with a simulation, I usually won't lift a finger until they send me their input deck unless they have a really good reason not to. Without the input, you're almost always just throwing darts at the board.

Also, have you looked into what physics this software is solving? Relevant details of the material model used? Maybe you don't have an issue with your input file; you're somehow outside the assumptions of the model.

I pity you poor fellows! I'm usually the guy on the other end of the line. But you are right - make sure it's solving the problem that you think it is solving!

 

[Rschub]

correct, wind speed is the thing that makes the blades turn its strange that it effects it in an opposite way.
I just find it odd that at such slow wind speeds, the blade can only handle 280kN for 20 years, but at very fast wind speeds, the ultimate load is 806kN for 20 years.

The program is giving you the design load.

At 4.5 m/sec the blades must designed to fail at 280KN, to satisfy the stated 20 yr. fatigue criteria.

At 12.5 m/sec the blades must be designed to fail at 806KN, to satisfy the stated 20 yr. fatigue criteria.

This is a case of not understanding what the program is doing.

The results are due to how composites handle fatigue, which is beyond the scope of this forum.

 

[Blast it Tom!]

The program is giving you the design load.

At 4.5 m/sec the blades must designed to fail at 280KN, to satisfy the stated 20 yr. fatigue criteria.

At 12.5 m/sec the blades must be designed to fail at 806KN, to satisfy the stated 20 yr. fatigue criteria.

This is a case of not understanding what the program is doing.

The results are due to how composites handle fatigue, which is beyond the scope of this forum.

It sounds like you might have it! That explains what appeared to be an upside down relationship very well!

 

[Rschub]

I feel like my last post came off as a little harsh and if so, I apologize.

 

[Rob Campbell]

Fatigue comes in two forms, high cycle and low cycle. High cycle fatigue is caused by vibration which exposes the object to relative high frequencies. The second is low cycle. In a turbine engine, his is caused by cycling the load on the blades by moving the throttle from idle to full power and back again, This is called total accumulated cycles (TACs). After failure, visual inspection of the break reveals if it's high or low cycle fatigue that caused the failure.

 

[Blast it Tom!]

Fatigue comes in two forms, high cycle and low cycle. High cycle fatigue is caused by vibration which exposes the object to relative high frequencies. The second is low cycle. In a turbine engine, his is caused by cycling the load on the blades by moving the throttle from idle to full power and back again, This is called total accumulated cycles (TACs). After failure, visual inspection of the break reveals if it's high or low cycle fatigue that caused the failure.

Right. And in metals, the (rough) boundary between the two is that low cycle involves plastic strain (picture bending a paper clip back and forth); high cycle, the plastic strain is confined to a very small volume at a stress riser, notch, crack initiation or crack tip, etc.

Composites are a whole 'nuther thing completely, my hat is off to those who deal with them!

 

[Rschub]

Fatigue life calculations in composites are a relatively new field, within the last 10-15 years. A lot of it was proprietary, since the people doing it are military or Boeing. It was assumed at one time that composites had nearly unlimited fatigue life, this turned out to be optimistic when you try to optimize for strength/weight.

There are a lot of factors that are involved, manufacturing variations, moisture intrusion, UV exposure, environmental chemicals, or how many eagles per year hit the blades, Etc.

Even the big guys don't always get it right.

 

[Blast it Tom!]

Fatigue life calculations in composites are a relatively new field, within the last 10-15 years. A lot of it was proprietary, since the people doing it are military or Boeing. It was assumed at one time that composites had nearly unlimited fatigue life, this turned out to be optimistic when you try to optimize for strength/weight.

There are a lot of factors that are involved, manufacturing variations, moisture intrusion, UV exposure, environmental chemicals, or how many eagles per year hit the blades, Etc.

Even the big guys don't always get it right.

I sure appreciate that. OceanGate, anyone?