design of experiments/parametric variation ('shape study')

[rgcoe]

Using the nominal design variables (r_1 = 20 m, r_2 = 30 m, d_1 = ?, d_2 = 42 m) and ranges given in SNL-WaterPower/WecOptTool#1, conduct a design of experiments study to better understand the sensitivity of the system to the different design variables.

[zmorrell-sand]

It looks like DAKOTA will do a very good job at performing this task, along with what is mentioned in issue SNL-WaterPower/WecOptTool_old#14 . I will try to get interfaces set up for both issues by the end of the week. I am thinking that a Box-Wilson Central Composite Design would be a good approach for both, as it is supported natively by DAKOTA, and it has the nice property of placing test points equidistant around the starting point.

[zmorrell-sand]

@ryancoe
I am having some issues with the RM3_debug.m file.

The first issue, which I found a workaround to using scp, is that Git LFS isn't installed on the Linux machines. This made it rather difficult to get the RM3_BEM.mat file to download properly. The cloned file only included the SHA256 signature, but not the actual file. I'm still not entirely sure what the issue was, but I got around it by downloading it onto the windows machine then using scp to copy the file to the correct location on the Linux machine,

The second, and currently more pressing issue that I have come across is that on line 11 of the RM3_debug.m file, you call a function called bretschneider, which is not in the WecOptTool source code as near as I can tell. I did a little bit of digging online, and I think that you are calling it from a tool called Wafo. Is Wafo a dependency for this tool, or is there somewhere else that I need to be looking for the file to add it to the path?

Thank You,
Zachary Morrell

[rgcoe]

Zach, On the Git LFS: You can install a local version of Git with LFS on the Linux machine and point to this if you like. On Wafo: Yes, this is a dependency, good catch! I think I state this in the root/README.md, but can’t remember. Either way, I think it be good if we had a “setup” script that alters the Matlab path to include the WecOptTool directories and checks if Wafo is in the path. Ryan

…

-- Ryan Coe, PhD Water Power Technologies (Org. 8822) | Sandia National Laboratories Email: Ryan.Coe@sandia.gov<mailto:Ryan.Coe@sandia.gov> | Office: (505)845-9064<tel:(505)845-9064> | Mobile: (513)910-8809<tel:(513)910-8809> On May 29, 2019, at 1:30 PM, zmorrell-sand <notifications@github.com<mailto:notifications@github.com>> wrote: @ryancoe<https://github.com/ryancoe> I am having some issues with the RM3_debug.m file. The first issue, which I found a workaround to using scp, is that Git LFS isn't installed on the Linux machines. This made it rather difficult to get the RM3_BEM.mat file to download properly. The cloned file only included the SHA256 signature, but not the actual file. I'm still not entirely sure what the issue was, but I got around it by downloading it onto the windows machine then using scp to copy the file to the correct location on the Linux machine, The second, and currently more pressing issue that I have come across is that on line 11 of the RM3_debug.m file, you call a function called bretschneider, which is not in the WecOptTool source code as near as I can tell. I did a little bit of digging online, and I think that you are calling it from a tool called Wafo. Is Wafo a dependency for this tool, or is there somewhere else that I need to be looking for the file to add it to the path? Thank You, Zachary Morrell — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<https://github.com/SNL-WaterPower/WecOptTool/issues/11?email_source=notifications&email_token=ABKUW77XSUDA4UNT6TO4NUDPX24UNA5CNFSM4HNP4DA2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWQCF4A#issuecomment-497033968>, or mute the thread<https://github.com/notifications/unsubscribe-auth/ABKUW74PX2YNTT5IKFLMTUDPX24UNANCNFSM4HNP4DAQ>.

[rgcoe]

@zmorrell-sand - Just wanted to note our conversation here:

1. Add the following responses:
   • power
   • volume
   • surface area
   • power/volume
   • power/(volume)^1/3
   • power/(volume)^2/3
   • power/(surface area)
   • power/(surface area)^1/2
2. Rerun the study on an HPC
3. Present the results with the plots you've created and those shown below

[rgcoe]

Let's pick this back up again with the goal of demonstrating the usage of a parametric variation study to understand the objective function sensitivity to various design variables (e.g., by creating an example script).