Page **2** of **3**

### Re: Help about BEM and NREL's Phase VI

Posted: **Fri Sep 02, 2016 5:21 am**

by **Wesley.Miguel**

Sir, thanks for the file that you provided. I checked it and observed that the data concerning structural properties covers the radius until 5m only.

I'm using the r= 5.029m.

any idea about this one sir? should I extrapolate for instance to get the data @ 5.029m?

### Re: Help about BEM and NREL's Phase VI

Posted: **Fri Sep 02, 2016 6:07 am**

by **Jason.Jonkman**

Dear Wesley,

You could extrapolate, or perhaps use the data from r=5 at the tip.

Please keep in mind that these are "estimates" and their accuracy is not guaranteed. You may need to do "tuning" anway to get the desired response.

Also keep in mind that the UAE Phase VI rotor was very rigid for conventional blade standards. It may not matter how accurate the properties are because the aero-elastic effects of this rotor are mostly negligible.

Best regards,

### Re: Help about BEM and NREL's Phase VI

Posted: **Thu Dec 01, 2016 10:48 pm**

by **Wesley.Miguel**

Sir,

Good day!

Sir, if I'll use NREL Phase VI for aerodynamic validation, could I use cn, ct as the parameters of comparison for my BEM code? I've seen a tabulated data about this (i.e. Jonkman, 2003), though I have some reservation upon reading that

these data (cl, cd, wind speed, aoa, cn, ct) reflects the operational environment of the wind turbine. Should I use the experimental data concerning power, torque (as functions of wind speed) instead?

Kind regards,

Wesley

### Re: Help about BEM and NREL's Phase VI

Posted: **Fri Dec 02, 2016 6:10 am**

by **Jason.Jonkman**

Dear Wesley,

I guess it all depends on what you intend to validate. All channels recording during the NREL Phase VI Unsteady Aerodynamics Experiment (UAE) will well tested/calibrated.

Best regards,

### Re: Help about BEM and NREL's Phase VI

Posted: **Mon Dec 05, 2016 9:16 am**

by **Wesley.Miguel**

I see, thanks sir.

My data got closer to the computed power, however variation exists in cn, ct.

Sir, concerning the formula for cn, ct= are they in terms of inflow angle or angle of attack?

btw, sir, regarding structural modelling,any idea how's proportional modelling approached?

Kind regards,

Wesley

### Re: Help about BEM and NREL's Phase VI

Posted: **Mon Dec 05, 2016 1:21 pm**

by **Jason.Jonkman**

Dear Wesley,

I'm sorry, but I don't understand your questions. Please clarify.

Best regards,

### Re: Help about BEM and NREL's Phase VI

Posted: **Fri Jan 13, 2017 12:13 am**

by **Wesley.Miguel**

Sir,

I'm doing BEM using Excel for initial testing.

I notice for the blade element at the tip (i.e. r=R), the axial and tangential induction factor becomes 1 and -1 after the first run. This becomes a problem when the second run starts as the Relative Velocity at that element is zero. Any idea about this one sir?

Kind regards,

Wesley

### Re: Help about BEM and NREL's Phase VI

Posted: **Fri Jan 13, 2017 6:22 am**

by **Jason.Jonkman**

Dear Wesley,

I'm assuming this Excel file you are referring to has your own implementation of BEM?

When looking at the limit of the BEM solution with the Prandtl tip-loss model as r --> R at the tip, I would expect that a --> 1 and a' --> 0, which is what we have implemented in AeroDyn v15. (AeroDyn v14 and earlier never allowed nodes to lie right at the tip.)

In our AeroDyn v15 implementation, we simply set a = 1 and a' = 0 for any nodes lying right at the tip when Prandtl tip loss is enabled. You can do the same in your Excel implementation of BEM, or you can probably get around any problems right at the tip by ensuring that nodes are not placed within some tolerance of the tip location.

Best regards,

### Re: Help about BEM and NREL's Phase VI

Posted: **Fri Jan 13, 2017 7:14 am**

by **Wesley.Miguel**

Jason.Jonkman wrote:Dear Wesley,

I'm assuming this Excel file you are referring to has your own implementation of BEM?

When looking at the limit of the BEM solution with the Prandtl tip-loss model as r --> R at the tip, I would expect that a --> 1 and a' --> 0, which is what we have implemented in AeroDyn v15. (AeroDyn v14 and earlier never allowed nodes to lie right at the tip.)

In our AeroDyn v15 implementation, we simply set a = 1 and a' = 0 for any nodes lying right at the tip when Prandtl tip loss is enabled. You can do the same in your Excel implementation of BEM, or you can probably get around any problems right at the tip by ensuring that nodes are not placed within some tolerance of the tip location.

Best regards,

Sir, thanks for this!

Yes sir, the Excel file has my own implementation (i.e. following the usual procedure of BEM as explained in several texts).

The value of a, a' you provided sir are irrespective of the empirical correction used concerning turbulent wake state (e.g. Glauert, Buhl)?

As for the matlab implementation, the path to follow sir is using newton raphson?

Kind regards,

Wesley

### Re: Help about BEM and NREL's Phase VI

Posted: **Fri Jan 13, 2017 8:01 am**

by **Jason.Jonkman**

Dear Wesley,

Regarding the limiting values at the tip, "yes," that is correct.

There are different methods to solve the BEM equations. Newton iterations are one method. In AeroDyn v15, we use Brent's method.

Best regards,

### Re: Help about BEM and NREL's Phase VI

Posted: **Wed Jan 18, 2017 7:13 am**

by **Wesley.Miguel**

Jason.Jonkman wrote:Dear Wesley,

Regarding the limiting values at the tip, "yes," that is correct.

There are different methods to solve the BEM equations. Newton iterations are one method. In AeroDyn v15, we use Brent's method.

Best regards,

I see,

thanks sir!

Kind regards,

Wesley

### Re: Help about BEM and NREL's Phase VI

Posted: **Tue Feb 28, 2017 6:09 am**

by **Wesley.Miguel**

Sir,

Good day!

Sir, I'm using trapezoidal rule in the computation of Tangential Force/Torque/Power.

The formula requires indicating segments (i.e. distance between two radial positions, r(n+1)-r(n)).

My computation started with r=1.257m (since this is the section where s809 is utilized).

Sir, if I will get the Tangential force for the first blade element (n=1), any idea what would the segment be?

Will the segment of the first blade element covers be r=1.257 and r=1.343?

Kind regards,

Wesley

### Re: Help about BEM and NREL's Phase VI

Posted: **Tue Feb 28, 2017 6:22 am**

by **Jason.Jonkman**

Dear Wesley,

I'm sorry, but I don't understand your question. Are you using AeroDyn v15, an older version of AeroDyn, or something else? What does your nodal distribution look like?

Best regards,

### Re: Help about BEM and NREL's Phase VI

Posted: **Sun Dec 15, 2019 1:09 pm**

by **KumaraRaja.Eedara**

Dear Jason,

I have implemented axial induction factor calculation in MATLAB as given in the paper "Ning, Andrew, et al. "Development and validation of a new blade element momentum skewed-wake model within AeroDyn." 33rd Wind Energy Symposium. 2015".

I get good match for AxInd and TanInd between MATLAB code and FAST at the middle section of the blade, but at the tip , FAST reports, AxInd =1 and TanInd =-1; (I chose Wakemod=1, included tip, hub loss, drag is taken in the induction factor calculation)

But my matlab implementation gives values different from 1 and -1. From reading this post, I realized that AxInd and TanInd are hardcoded to be 1 and -1 respectively near tip. Is it incorrect to proceed with the values given by the algorithm given in the paper? I understand the problem with the implementation in paper right at the tip. Could you please elaborate on the implications of this on the further calculations.

Also, I don't see hard coded values for AxInd and TanInd in the source code (given below). Where is this implemented in the code?

Code: Select all

` ! compute axial induction factor`

if (phi > 0.0_ReKi) then ! momentum/empirical

! update axial induction factor

if (k <= 2.0_ReKi/3.0_ReKi) then ! momentum state

if ( EqualRealNos( k, -1.0_ReKi) ) then

k = k - 0.1_ReKi ! Need to bump k to avoid singularities

end if

a = k/(1.0_ReKi+k)

else ! Glauert(Buhl) correction

g1 = 2.0_ReKi*F*k - (10.0_ReKi/9-F)

g2 = 2.0_ReKi*F*k - (4.0_ReKi/3-F)*F

g3 = 2.0_ReKi*F*k - (25.0_ReKi/9-2*F)

if (abs(g3) < 1e-6_ReKi) then ! avoid singularity

a = 1.0_ReKi - 1.0_ReKi/2.0/sqrt(g2)

else

a = (g1 - sqrt(g2)) / g3

end if

end if

else ! propeller brake region (a and ap not directly used but update anyway) !bjj: huh? when k is slightly larger than 1, a is definitely getting used (and causing issues)...

if (k > 1.0_ReKi .and. .not. EqualRealNos(k, 1.0_ReKi) ) then

!if (sigma_pcn > Fsphi) then

a = k/(k-1.0_ReKi) !sigma_pcn / (sigma_pcn - Fsphi ) !

! axial induction is blowing up, so I'm putting a band-aid here. BJJ 25-Feb-2016

a = min(a, 10.0_ReKi )

else

a = 0.0_ReKi ! dummy value

end if

end if

! compute tangential induction factor

if ( cphi==0.0_ReKi ) then ! We don't want NaN here

kp = HUGE(kp)

else

kp = sigma_p*ct/4.0_ReKi/F/sphi/cphi

end if

! Per conversation with Rick, we should only trigger this if phi = 0 , so we will return predefined values as if phi=0.0

if (EqualRealNos(kp, 1.0_ReKi)) then

fzero = 0.0_ReKi

a = 0.0_ReKi

ap = 0.0_ReKi

return

end if

ap = kp/(1.0_ReKi-kp)

! tangential induction is blowing up, so we're putting a band-aid here. GJH, JMJ, BJJ 1-Sep-2015

if ( abs(ap) > 10.0_ReKi ) then

ap = sign( 10.0_ReKi, ap )

end if

!bjj: 3-jun-2015: TODO: was able to trigger divide-by-zero here using ccBlade_UAE.dvr without tiploss or hubloss

if (.not. wakerotation) then

ap = 0.0_ReKi

kp = 0.0_ReKi

end if

Regards,

Kumara

### Re: Help about BEM and NREL's Phase VI

Posted: **Mon Dec 16, 2019 9:36 am**

by **Jason.Jonkman**

Dear Kumara,

The BEM solution of AeroDyn generally follows Ning et al's paper from AIAA SciTech 2015, but some of the details have been changed to ensure AeroDyn is robust for all cases, e.g., aero-elastic interaction under turbulent wind inflow.

Which version of FAST / AeroDyn are you using? In the current version of OpenFAST / AeroDyn v15, I would expect the axial induction to be 1.0 and the tangential induction to be 0.0 at the root and tip, when hub and tip losses are enabled. This is calculated in routine BEMTUncoupled/BEMTU_InductionWithResidual(). I would recommend upgrading if you are not at least using OpenFAST 0.1 or newer because AeroDyn v15 was upgraded and made more robust between FAST v8.16 and OpenFAST v0.1.

We need to develop documentation with the most up-to-date implementation of BEM, based on updates to Ning et al's paper, but we do not currently have the resources to do so. Until then, the source code is probably your best resource as to the details of the implementation.

Best regards,