CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Bootstrapping myself (http://www.cfd-online.com/Forums/main/82656-bootstrapping-myself.html)

 Maury Markowitz December 1, 2010 18:04

Bootstrapping myself

CFD is overkill for what I'm trying to do, so if anyone had a simpler suggestion for what follows, please let me know!

There is a considerable amount of debate in the solar world about how to model wind forces on solar panels on the top of buildings. Most suggest using the ASCE 7-05 wind loading codes, but there are many "sub codes" to chapter 6 that give very different answers, and each one has arguments in its favor. If you use the Canadian codes you get an entirely different set of numbers, perhaps 2x the US values. In either event, the numbers are so large as to seem unphysical.

I've been trying to figure out if any of these are accurate, and tracked down a number of papers from wind tunnel studies, including one that is pretty much identical to an install we're working on. However, these don't "close the loop" to the actual forces, leaving the answer in the form of Cf's, which doesn't help much because there's all those ways to use the Cf's!

One way to solve this would be to just model it and measure the lift. That's why I'm here. But even CFD doesn't make that too easy, what you get is a bunch of numbers at different points. So...

I'm looking to bootstrap myself to the point where I feel confident at the most basic level. I want to take the 3D model, turn it into a mesh, run a simulation, and then from the result, calculate the NET FORCES (lift, drag, torques).

Can anyone suggest a resource for getting me going?

 TMG December 2, 2010 10:29

I work for a CFD company so you can argue that I'm not objective, but this is exactly what you should be using CFD for, rather than guessing using all sorts of sub-codes that give different answers. CFD is routinely used to get drag and lift on anything from airplanes to automobiles to submarines and you don't just wind up with a bunch of numbers at different points. If you use the right program you get the number you are interested in directly and easily. If you are uncomfortable jumping into CFD yourself then find a consultant that will do the job for you the first time and show you how to do it.

 gocarts December 2, 2010 12:22

Solar Panel CFD Simulation

1 Attachment(s)
Is this what you are looking for?

http://www.cfd-online.com/Forums/att...screenshot.jpg

Confession: I represent Symscape the developer of Caedium (shown in the screenshot), so you might find me a little bias...

 TMG December 2, 2010 12:44

Richard - thank you for helping to prove my point, even if its not my favorite CFD code :)
Maury - there are a number of codes around in addition to the one shown above that can solve this problem, not to mention consultants that will run it for you. In this day and age, there is no reason to guess, no reason to build the structure 2x as strong and cost 2x more than they reasonably need to be.

 f-w December 2, 2010 14:37

If you have the normal force coefficient (Cf) curve from a paper that uses panels that are almost identical to yours, why not calculate the forces? This is the whole point of non-dimensional coefficients.

Cf=Fn/[.5*density*area*velocity^2]

Make sure you're using Cf at the correct inclination angle, and then calculate Fn. The next step is to use trigonometry to break Fn into drag and lift. Toques can then be calculated if center of pressure is given (if not look for papers that mention typical locations).

This will get you an answer much quicker (and cheaper) than the time it will take to set up a simulation correctly.

 Maury Markowitz December 2, 2010 16:53

Quote:
 Originally Posted by f-w (Post 285770) If you have the normal force coefficient (Cf) curve from a paper that uses panels that are almost identical to yours, why not calculate the forces? This is the whole point of non-dimensional coefficients.
We do this now, but here's the problem:

There's a great paper from Romania addressing this issue, "Steady Wind Pressures on Solar Panels on Flat Roofed Buildings". They come up with a value called C(LR), the net C for the top and bottom. For most wind angles it's nearly zero, and only becomes an issue when the wind is from the back (as one would expect) when it gets as high as 1.0

Ok great, but... Do we calculate forces based on this number and then apply a safety factor? What safety factor? What gust factor should I use?

It's simple in theory, but getting an engineer to sign off on this, ugggg!

 Maury Markowitz December 2, 2010 16:59

Quote:
 Originally Posted by gocarts (Post 285754) Is this what you are looking for?
_Almost_ exactly what we're looking for.

But how do I go from a bunch of colored arrows to actual forces? There's a force gauge on the lower right, but what is it measuring?

 gocarts December 2, 2010 17:33

Solar Array Drag and Lift Forces

The Force monitor in the lower right is reporting the total Drag (Force in X direction - parallel to the inlet flow) and the total Lift (force in the Y direction - perpendicular to the inlet flow and upwards) acting on the solar array surfaces, where 100 iterations = 1s.

CFD solvers typically make a guess at a solution and then iteratively improve it - what the monitor shows is that there is little variation in force with increasing iteration count (pseudo time) beyond 2s, so essentially the force has converged, so you read off the final values at 5s for your forces.

All CFD codes will report forces etc., often it's the primary requirement of running a CFD simulation. However, without the 'colored arrows' how do you know that you didn't mess up the boundary conditions - say the arrows show the flow going backwards...consider them a confirmation that you set up the simulation correctly.

Further, flow visualization is also a valuable aid in understanding how the flow is responding to your design and that insight might help you make beneficial changes in shape or location.

You are likely aware that the flow over your solar array is grossly affected by how close it is to the edge of a building - the upwash over the lip of a flat roof and around your array plays a major role in determining the forces acting on the array. The flow visualization provides a deeper insight than the raw drag and lift numbers alone.

 f-w December 2, 2010 18:58

Quote:
 Originally Posted by Maury Markowitz (Post 285786) We do this now, but here's the problem: There's a great paper from Romania addressing this issue, "Steady Wind Pressures on Solar Panels on Flat Roofed Buildings". They come up with a value called C(LR), the net C for the top and bottom. For most wind angles it's nearly zero, and only becomes an issue when the wind is from the back (as one would expect) when it gets as high as 1.0 Ok great, but... Do we calculate forces based on this number and then apply a safety factor? What safety factor? What gust factor should I use? It's simple in theory, but getting an engineer to sign off on this, ugggg!
I assume before designing a product, the designers would agree upon the winds, gusts, and safety factors that a product should withstand (and hope clients read the label). Quick publication search might turn up some typical values. If this is a one-off (custom install), you might consider doing a local meteorological study to determine these numbers.

Of course, design for the worst case scenario; and, if none of the papers cover the range of conditions, further testing (numerical or experimental) is a requirement, unless someone with experience can sign it off. As gocarts point out, every design aspect matters, and if papers are not available (or if more understanding is sought), then testing, once again, is required (assuming you can setup and interpret the results correctly). Or, as TMG points out, you can outsource it (and learn from them, which will allow you to develop an in-house capability faster).

Publication or CFD, you're going to have to ask yourself the same design questions.

Good Luck!

 All times are GMT -4. The time now is 08:37.