Question regarding HydroDyn Strip Theory Implementation

Provide feedback, request enhancements, and get help with wind-turbine computer-aided engineering tools.

Moderators: Bonnie.Jonkman, Jason.Jonkman

Daniel.Walia
Posts: 19
Joined: Thu Dec 15, 2016 7:56 am
Organization: University of Rostock
Location: Germany

Question regarding HydroDyn Strip Theory Implementation

Postby Daniel.Walia » Thu Dec 15, 2016 3:46 pm

Dear all,

we are currently trying to run a coupled model of a RNA mounted on a TLP using FAST, HydroDyn and MoorDyn.
At this point we try to run a model without Potential Flow input Data from WAMIT so we are focussing now
on strip theory. (Later a more sofisticated model will include Potential Flow output from AQWA which will be
formated into WAMIT format)

I got some uncertainties regarding the input files for HydroDyn on the data necessary for using Strip Theory in HydroDyn.

1) There is a section called "Axial Coefficients". Hereby AxCd, AxCa and AxCp can be defined for the calculation of the lumped forces at the member ends.
Furthermore there are 3 sections for for "Hydrodynamic Coefficients" Model1/2/3 Hereby the distributed forces over each member will be calculated.

When we calculate the distributed loads on the members by i.e. Model 3 (Member-Based), what do we need the lumped forces at the member ends (joints)
in the "Axial Coefficients" section for?

In 6.3.2 in the manual it says "For example, if you want axial effects at a marine-growth boundary, you must explicitly set a joint at that
location". But still I don´t exactly understand what the lumped loads are for.

2) In the HydroDyn manual in the "Hydrodynamic Coefficients" sections it says:

"All of these models require the specification of both transverse and axial hydrodynamic
coefficients for viscous drag, added mass, and dynamic pressure (axial viscous drag is not yet
available)."

So I guess the lateral (e.g. MemberCa1) and axial (e.g. MemberAxCa1) motions shall be defined here. In the
FAST model of the OC3 Floater however the axial coefficients are set to 0. Are they ignored here? Why?
When I read about Morison equation, I only learned about lateral Drag- and Added-Mass-Coefficients. Are there publications
on axial Drag- and AM-Coefficients available?

For me this sounds contractictory to what is said in the next paragraph "... and
Ax identifies the axial coefficients to be applied for tapered members (the transverse coefficients
are identified without Ax)..." So the "Axial" coefficients apply only for tapered members?

3) I am not completely sure what C_P is. I know the Morison equation can be seperated into a drag and an added mass term.
And the first includes a drag and the latter an added mass coefficient. But I don´t remember a pressure coefficient.

4) When HydroDyn divides single members by the value assigned by "MDivSize" the distributed Strip Theory loads will be applied with a resolution depending on MDivSize.
Why does it not integrate over the member length? Is this a question of computation time?

5) The AddCLin Matrix for the OC4 Floater looks like this (in accordance to the HydroDyn Manual):

AddCLin_Matrix.png
AddCLin_Matrix.png (42.07 KiB) Viewed 2106 times


I understand that because of symmetry the green marked terms become 0. But as far as I understood the OC4 Floater is only symmetrically in one axis, therefore the orange marked terms shouldn´t be zero. Can you please explain, where is my mistake here?

Furthermore the red marked term shouldn´t be 0 as it is the product of rho*g*A. Is this value so small and can therefore be neglected? Or do all the neglected and set to 0 terms are not necessary here because there is a WAMIT input file for the OC4 model?



Thank you very much,
Daniel

Jason.Jonkman
Posts: 4627
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Jason.Jonkman » Fri Dec 16, 2016 7:01 am

Dear Daniel,

Here are my answers to your questions:

1) The axial coefficients at joints apply to ends of members that are exposed to water e.g. the bottom of the OC3-Hywind spar. See slides 11-14 of the presentation found here for more information: http://wind.nrel.gov/public/jjonkman/Pr ... onkman.pdf.

2) You must always specify all coefficients (you can't leave any blank), but you can set them to zero if appropriate. For the OC3-Hywind model, hydrodynamics are treated as a hybrid potential flow plus strip theory solution, so, only the viscous-drag terms from strip theory are used. Without marine growth and the use of model 1, this means that SimplCd is the only coefficient actually used by HydoDyn. The axial terms in models 1-3 are applied only for tapered members. We don't have great documentation on this yet, but just be aware that the axial terms in models 1-3 are unused for members that are not tapered i.e. where dR/dz = 0 (where the change in radius along the length of the member is zero).

3) The Froude-Kriloff term is derived by integrating the dynamic pressure of the undisturbed incident waves around the body. In Morison's equation, the long-wavelength approximation implies that this term can be calculated using the fluid acceleration at the body centerline with a coefficient of unity. In HydroDyn, we allow the coefficient to be different than unity (e.g. to correct for members that are not exactly circular or for other calibration purposes). See slide 12 of the presentation linked above for more information.

4) The hydrodynamic loads are calculated locally at nodes along a member so that they can be properly transferred to the structural model for analysis. While FAST currently treats a floating platform as a rigid body (so integrating the hydrodynamic load along the member is OK), FAST can model bottom-fixed substructures as flexible (via the SubDyn module), so, it is important to calculate the hydrodynamics locally.

5) For the OC4-DeepCwind model, hydrodynamics are treated as a hybrid potential flow plus strip theory solution. The terms AddCLin(4,4) and AddCLin(5,5) are nonzero because of the ballasted (fluid-filled) members. See section 6.8.3 of the draft HydroDyn User's Guide and Theory Manual for more information: https://wind.nrel.gov/nwtc/docs/HydroDyn_Manual.pdf.

I hope that helps.

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Daniel.Walia
Posts: 19
Joined: Thu Dec 15, 2016 7:56 am
Organization: University of Rostock
Location: Germany

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Daniel.Walia » Mon Dec 19, 2016 3:42 am

Dear Jason,

Jason.Jonkman wrote:1) The axial coefficients at joints apply to ends of members that are exposed to water e.g. the bottom of the OC3-Hywind spar. See slides 11-14 of the presentation found here for more information: http://wind.nrel.gov/public/jjonkman/Pr ... onkman.pdf.


Now I got it. I wasn´t aware that the surface normal to the member surface where the distributed loads are applied is ment here. Slide 13 from your link was very helpfully.

Jason.Jonkman wrote:
2) You must always specify all coefficients (you can't leave any blank), but you can set them to zero if appropriate. For the OC3-Hywind model, hydrodynamics are treated as a hybrid potential flow plus strip theory solution, so, only the viscous-drag terms from strip theory are used. Without marine growth and the use of model 1, this means that SimplCd is the only coefficient actually used by HydoDyn. The axial terms in models 1-3 are applied only for tapered members. We don't have great documentation on this yet, but just be aware that the axial terms in models 1-3 are unused for members that are not tapered i.e. where dR/dz = 0 (where the change in radius along the length of the member is zero).

3) The Froude-Kriloff term is derived by integrating the dynamic pressure of the undisturbed incident waves around the body. In Morison's equation, the long-wavelength approximation implies that this term can be calculated using the fluid acceleration at the body centerline with a coefficient of unity. In HydroDyn, we allow the coefficient to be different than unity (e.g. to correct for members that are not exactly circular or for other calibration purposes). See slide 12 of the presentation linked above for more information.



Thank you, this helps alot!

Jason.Jonkman wrote:
4) The hydrodynamic loads are calculated locally at nodes along a member so that they can be properly transferred to the structural model for analysis. While FAST currently treats a floating platform as a rigid body (so integrating the hydrodynamic load along the member is OK), FAST can model bottom-fixed substructures as flexible (via the SubDyn module), so, it is important to calculate the hydrodynamics locally.



Okay that makes sense to me, that single nodes are necessary for the load mapping procedure in SubDyn.

Jason.Jonkman wrote:
5) For the OC4-DeepCwind model, hydrodynamics are treated as a hybrid potential flow plus strip theory solution. The terms AddCLin(4,4) and AddCLin(5,5) are nonzero because of the ballasted (fluid-filled) members. See section 6.8.3 of the draft HydroDyn User's Guide and Theory Manual for more information: https://wind.nrel.gov/nwtc/docs/HydroDyn_Manual.pdf.

I hope that helps.


I got one question with regard to MoorDyn. I am not sure if I should open a new thread on this. I will try it here anyow.

How do I set up pretension of the mooring lines? I didn´t find any hint on this for MoorDyn while I found out that for FEAMooring it is possible to specify pretension.

Once again thank you and your team for your time and engagement in providing those tools and support.

Best regards,
Daniel

Jason.Jonkman
Posts: 4627
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Jason.Jonkman » Mon Dec 19, 2016 6:04 am

Dear Daniel,

The pretension in MoorDyn is solved for as part of the initial static solution based on the geometry and material properties of the mooring system.

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Daniel.Walia
Posts: 19
Joined: Thu Dec 15, 2016 7:56 am
Organization: University of Rostock
Location: Germany

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Daniel.Walia » Mon Dec 19, 2016 7:46 am

Dear Jason,

so Moordyn calculates the pre-tension by the delta between UnstrLen and the distance between the line end- and starting point. Thank you.

One final question on MemberCp1 and MemberCp2 in the Model3 of the HydroDyn inputfile. In the OC4-DeepCwind model they are set to 0 for all members, because the OC4-DeepCwind model includes a potential flow solution. Thus, if our model doesn´t include a potential flow solution setting MemberCp1 and MemberCp2 is adequate?

Best regards,
Daniel

Jason.Jonkman
Posts: 4627
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Jason.Jonkman » Mon Dec 19, 2016 2:02 pm

Dear Daniel,

Your understanding of MoorDyn is correct.

If your model doesn't include a potential-flow solution--i.e., your model is a strip-theory-only solution--then you should set the dynamic pressure coefficients, added-mass coefficients, and viscous drag coefficients appropriately.

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Daniel.Walia
Posts: 19
Joined: Thu Dec 15, 2016 7:56 am
Organization: University of Rostock
Location: Germany

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Daniel.Walia » Mon Jan 02, 2017 4:48 am

Dear Jason, I wish you a happy new year.

Thank you for answers. The FAST simulation runs now without errors.

Anyhow I got further questions regarding the strip theory model. When importing the geometry into paraview the visualization shows some inconsistencies at member joints.
I am sure that I have defined the nodes at member joints correctly. With respect to the OC4-DeepCwind model the second member of a joint is defined by its start/end node which is at the intersection of the axis of the second member and the surface of the first member. Anyhow the visualization in paraview (attachment screenshot.png and screenshot2.png) shows some discontinuities at member joints.
Is this only a result of the visualization and doesn´t effect calculation results of hydrodynamic analysis or is this an error in the model?
Or should the member joint be defined at the intersection of the axes of both members?

The second question concerns horizontal members. In the paraview output (attachment screenshot3.png) horizontal pipe members look like sheets. They have the same properties as vertical members. So I am wondering here if this is a question of visualization or an modeling error because I think also lateral wave loads on the horizontal members have to be taken into account.

Best regards,
Daniel
Attachments
screenshot3.png
screenshot3.png (8.73 KiB) Viewed 2059 times
screenshot2.png
screenshot2.png (22.83 KiB) Viewed 2059 times
screenshot.png
screenshot.png (34.57 KiB) Viewed 2059 times

Jason.Jonkman
Posts: 4627
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Jason.Jonkman » Mon Jan 02, 2017 6:40 am

Dear Daniel,

I think what you are seeing is only a limitation of the FAST visualization capability, and not an improper model set up. The FAST ReadMe file (https://wind.nrel.gov/nwtc/docs/README_FAST8.pdf) refers to the following limitation associated with the visualization of HydroDyn's strip-theory members:

Note that HydroDyn currently uses the identity matrix for the orientations of this mesh, so elements that are not completely vertical will not be visualized correctly (horizontal elements look like planes instead of cylinders).

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Daniel.Walia
Posts: 19
Joined: Thu Dec 15, 2016 7:56 am
Organization: University of Rostock
Location: Germany

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Daniel.Walia » Mon Jan 02, 2017 8:10 am

Dear Jason,

thank you very much for your response.

Best regards,
Daniel

Daniel.Walia
Posts: 19
Joined: Thu Dec 15, 2016 7:56 am
Organization: University of Rostock
Location: Germany

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Daniel.Walia » Wed Dec 05, 2018 9:27 am

Dear Jason,

I am sorry for necro-bumping this thread. But after going again a little bit deeper into HydroDyn I learned that there are still some uncertainties on my side which I´d like to clearify.
Regarding the axial coefficients:

- Axial coefficients have to be defined for the member ends as shown in your presentation in the first link provided above on slide 11. When having a look on our model as shown in the screenshot attached, I have to define axial coefficients at three different points. At the bottom where the vertical column ends, as well as where there the two points, the hull surface increases for the buoyancy bodies.
Now I am not sure which CA and CD to choose here because I don´ t understand about the way of calculation here. When having a look into the DNV RP-C205: https://rules.dnvgl.com/docs/pdf/dnv/codes/docs/2010-10/rp-c205.pdf I´d assume a value für C_A of 2/PI as recommended on page 118 of C205 and shown in the screenshot attached. Hereby I´d assume that we are here talking about a vertical motion of the surface through the fluid (or vice versa).
Is that correct?

If yes, this would mean that I also have to determine the respective C_D value which I wasn´ t able to find yet. Also I think find C_A and C_D for the circular rings for the other member joints as shown in the screenshot might be tricky.

Best regards,
Daniel
Attachments
dnvgl.png
dnvgl.png (36.45 KiB) Viewed 1486 times
Screenshot_2018-12-05_16-59-09.png
Screenshot_2018-12-05_16-59-09.png (65.11 KiB) Viewed 1486 times

Jason.Jonkman
Posts: 4627
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Jason.Jonkman » Wed Dec 05, 2018 10:22 am

Dear Daniel,

Your understanding is correct. For a flat circular disk of radius "a", HydroDyn would require you to define the Ca on each end of the disk, but HydroDyn would set the reference volume to equal that of half a sphere, 2/3*pi*a^3, so setting Ca = 2/pi on both sides would be equivalent to defining a single Ca = 2/pi with a reference volume equaling the full sphere (4/3*pi*a^3).

For the buoyancy bodies connected to the vertical column, the reference volume HydroDyn would use for each would be equal to 2/3*pi*(a_body^3 - a_column^3).

I agree that estimating Cd is never easy.

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Daniel.Walia
Posts: 19
Joined: Thu Dec 15, 2016 7:56 am
Organization: University of Rostock
Location: Germany

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Daniel.Walia » Thu Dec 06, 2018 8:09 am

Dear Jason,

thank you for the explanation. Just to make sure that I understood this correctly:

For a circular disc which is subjected to flow an added mass coefficient of Ca = 2/Pi is suitable. This can be applied to different joints of the structure. The only important factor is the shape of the surface. For the design shown in the screenshot above Ca=2/Pi is applicable for both types of member ends, for the lower one as well as for the upper two which are connected to a slender column. The fact that not the entire cross section of the latter is exposed to the flow is taken into account by HydroDyn´s implementation (what you have just explained above). Thus, these special geometric characteristics which result from the transition between vertical column and B-Body have not to be taken into account when chosing Ca for this member end.

Best regards,
Daniel

Jason.Jonkman
Posts: 4627
Joined: Thu Nov 03, 2005 4:38 pm
Location: Boulder, CO
Contact:

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Jason.Jonkman » Thu Dec 06, 2018 10:49 am

Dear Daniel,

Yes, I agree that you can reasonably use a value of Ca = 2/pi for all member ends in the absence of more information. Of course, it would be better if you had more information that would take the actual geometry into account e.g. the pressure distribution from a potential-flow solution, experimental measurements, or high-fidelity simulations could be used to derive more accurate coefficients for the actual substructure geometry.

Best regards,
Jason Jonkman, Ph.D.
Senior Engineer | National Wind Technology Center (NWTC)

National Renewable Energy Laboratory (NREL)
15013 Denver West Parkway | Golden, CO 80401
+1 (303) 384 – 7026 | Fax: +1 (303) 384 – 6901
nwtc.nrel.gov

Daniel.Walia
Posts: 19
Joined: Thu Dec 15, 2016 7:56 am
Organization: University of Rostock
Location: Germany

Re: Question regarding HydroDyn Strip Theory Implementation

Postby Daniel.Walia » Fri Dec 07, 2018 12:22 am

Dear Jason,

thank you very much for these informations. For a more sophisticated model we will definetely implement the pressure distribution from a potential-flow solution. At this point just wanted to make sure that I understood how HydroDyn deals with the added mass and the corresponding Ca, for a "lower-fidelity" simulation, correctly.

Thank you and best regards,
Daniel


Return to “Computer-Aided Engineering Software Tools”

Who is online

Users browsing this forum: No registered users and 1 guest