>> From: Joe & Jan Wurts <[EMAIL PROTECTED]> >> Somewhere around the beginning of the year, I set out to do some >>systems design, analysis, and optimization work on the... >Joe, >could you please elaborate a little more on your optimization approach? Oleg, great questions. These questions indicate that you at least understand much of the challenges, if not also how to solve these problems. In short, my analysis tools were: Excel, a vortex lattice code, and an airfoil design and analysis tool that is highly regarded in the aero industry. I used some custom Excel stuff that is formulated to help out in digesting the results from the vortex lattice and airfoil tools. Not very difficult to put together, just the typical eqns, along with some macros and VB. Not anything that is terribly user friendly, but enough for me to do the job. As you probably know, setting up the constraints, and the objective evaluation functions is really the difficult task. All else is just math. As for the piece parts, the wing planform design is optimized considering spanwise lift distribution to get good handling qualities, along with good overall total lift and drag efficiency. This is somewhat intuitive in nature, one has to look at the local lift coefficients along the span in order to arrive at the "handling qualities" part of it. The rest of the stuff is more objective. In addition to the handling qualities, one needs to get a good e (Oswalds efficiency) out of the planform, along with a high useful total lift coefficient (total wing Cl/local Cl). The latter is somewhat in opposition to the handling qualities, but manageable. Also, one should be taking into account the varying local reynolds number along the span in formulating the total aero characteristics of the wing (lift and drag). Also included in the wing optimization process was the weight of the wing. As a first order approximation, the wing weight is a function of the area (skin weight at minimum gage), volume (foam cores for vacuum bagging), along with a second order influence due to total thickness (thin high aspect ratio wings need more spar structure). The spar structure is really not as much of a driver for HLG as one would think, but should be included. As for the tails, a simple trade on tail boom length vs. tail size was made, with a simplistic boom length vs weight eqn, along with tail area vs tail weight eqn. Functional goals included optimizing the minimum sink, mid-range cruise speed L/D, and very low Cl profile drag. Defining a useful objective function here is not a trivial task, and will be left up to the reader. Variables that were traded for the optimization include: planform airfoil(s) wing area tail boom length TE angle as applicable TE length as applicable Fixed values include: Radio gear weight (2ch differing from 4 ch) wing span tail volumes nose pod weight Qualitative ratings include: spanwise local lift distribution cl - cd bucket width with TE deflections Constraints include: airfoil thickness (at servo and and TE) The process is highly iterative in nature. I started off with using an Encore type HLG to get me the scaled Re's along the wing. After getting this, I started whacking at the airfoil development. After developing a few candidate airfoils for evaluation, the planform was brought into consideration for a round of optimization. With the "optimized" airfoil, I did a planform optimization using the developed airfoils. In the case of the poly ship, I ended up doing another round of airfoil optimization, as the optimal aspect ratio increased, driving down the wing area which pushed for another bout of airfoil optimization. This design/analysis/optimization loop is fairly straightforward for the poly ship, but grew some hair for the 4 ch ship. The additional variable of TE deflections added considerably to the design cycle. Also, it made for more qualitative evaluations. Airfoil 1 might produce a better peak efficiency, but airfoil 2 might produce a wider "bucket". Which is better? And how wide should that bucket be to be considered optimal? There is a reason why nobody is selling a program that has an optimal airplane design button. There are just too many qualitative judgments that go into a realistically constrained design. After the first few orbits around the design loop, I tossed in the tail boom optimization as well, and went for another orbit. The result, the toys that I flew at Poway. The poly ship did not get the tail boom iterated, and the 4 ch ship was a bit shorter compared to the optimal, but the min sink sensitivity was really flat between the chosen length and the optimal length. Note: optimal really should have quotes around it. It was optimal by my evaluations, objective functions, and constraints, but might not be optimal by anothers evaluation. Now, you ask my opinions on some of the "skinny" type platforms that are showing up out there. First, I'll make a note on the poly ship that I defined. I intentionally finished with about 15 - 20% more wing area than the theoretical optimal solution, as the sensitivity was really flat, and I wanted to have a bit more "spare" Cl for manuever capabilities. I lost about 1% in the hang time in order to pay for this, something that I thought was a reasonable trade. It turns out that one can get almost the same overall result over a fairly large area range via trading airfoil for chord length (wing area). So, just what is best? Dunno... just gave it my best shot. The challenge with the real high AR type solution on a poly, is that the airfoil has to compromise more at the very low Cl (high speed) region. As these toys were aimed at the Poway type conditions, I chose not to go that route. Also, I have a pretty high speed throw, so minimizing the drag for the throw condition is pretty important to me, but might not be quite so valued for another. For a high AR 4 ch wing, I'm quite worried about the aeroelastic and flutter issues. It is quite difficult to get sufficient torsional stiffness to survive at 80 mph without flutter. I have not seen a good solution here yet. Fortunately, my numbers did not lead me to have to solve this one, as I got pointed to a lower AR solution. Now, to the very long tail booms. I'm a fan of compromising somewhere in the middle for the tail boom. My optimizations show that going too long is just as bad as too short. But, I've tossed in a weight penalty for the longer booms that is derived via the required total stiffness to preserve a similar aeroelastic structure. Some of the long tail boom configurations out there are unsuitable for me in the wind due to aeroelastic issues. They work very well for many, but I think that I might be getting to too high of a throw speed in the wind. Also, the handling qualities get a little "different" for my taste. Due to the large damping from the long tail boom, one needs very high control deflections compared to a shorter tail boom to get the same pitch or yaw rate. Might be a personal thing, but it does not feel right to me. Now, for some things I noticed during the process. For airfoils, in the low Re regime that HLGs play at, it is hard to get the thickness and camber far enough forward. I discarded some airfoils due to a lack of construction capability. In fact, the airfoil used on the 4 ch bird was initially discarded as I did not think that it could be built successfully, but Phil somehow made it happen. I had designed up a series of different airfoils that went from root to tip in order to fit the servo, get enough TE thickness for control surfaces, and suitable spar thickness, but we did not end up using them. The 4 ch airfoil had the max thickness and camber even farther forward than the poly airfoil. I jokingly call it the "flat top" airfoil, as there are only small amounts of curvature in the aft 60% of the upper surface. A little bit of creative shaping near the hinge line can pay off as well, but steals even more from the TE thickness. Both airfoils had a far blunter LE than the 6063 does. I did a little check on my now old 6063 Encore, and it turns out the the LE on it was blunter than it should be. Maybe that is one reason that I really liked it, it was already headed towards the right direction... The wing planform also had a pretty low sensitivity to the "safety" value the rolloff of local cl vs. spanwise location). The higher the safety value, the higher the tip Re, which pretty much paid for the reduced span efficiency and then some. Where it lost was in the total lift capability before the wing root got out of the drag bucket. There are lotsa trades that can be done here with using airfoils varying vs. span in order to tailor the airfoil cl - cd bucket to match the local requirements, but one tends to lose on the edgess of the flight envelope. My multi-airfoil 4 ch wing design is in this vein. The bottom line, after all this typing: The design/analysis/optimization routine is highly dependent on your definition of your explicit and your unknown implicit constraints. It is the unknown implicit ones that hamper your finding the next breakthrough in design. It was a lot of fun and work to go through this process, and it enabled me to understand and develop sufficient tool sets and knowledge to attack the far more complex tasks of designing an F3B and F3J ship. Stay tuned for further details on these. Regards, Joe Wurts PS All of the above opinions are mine, and are quite likely not applicable to many. IMHO,The design process above resulted in high performance thoroughbreds, and might not result in the optimal fun fly HLG. That said, they are less extreme than the 6063 Encore (easier to fly, more forgiving, and more performance). RCSE-List facilities provided by Model Airplane News. Send "subscribe" and "unsubscribe" requests to [EMAIL PROTECTED]
