• Background Image

    Blog

    Journal

October 15, 2019

Preliminary Research On The Optimisation of Flying Wings In Forward Flight

By Read Liston

Longitudinal Stability and Trim

The differentiation of stability and trim is an important factor in the design of an aircraft, specifically a tailless one. In terms of static longitudinal stability, the definition remains the same for a flying wing and a traditional aircraft; a centre of gravity location ahead of the neutral point of the aircraft. For a flying wing, this neutral point is analogous to the aerodynamic centre of the wing/airfoil and is generally located along the 20% to 25% line of the chord along the span of the wing. A statically stable aircraft, however, is insufficient for steady flight, the aircraft must also be trimmed. A trimmed aircraft has all the moments about the centre of gravity balanced. With no horizontal stabiliser, the wing design of a tailless aircraft is significantly more involved, requiring both lifting and trimming capabilities. In the statically stable case, a typical main wing will produce a pitching down moment due to the fore centre of gravity location. From [1]; there are two ways in which to attain trim:

  1. Use a combination of wing sweep and wash out. 
  2. Use an airfoil that produces a positive pitching moment.

A traditional cambered airfoil produces a large negative pitching moment, exacerbating the pitching down moment. As such, an airfoil with a positive pitching moment can be used to instead balance the main wing moment. This solution is generally only applicable for a wing with no sweep. A much more feasible method is the combination of wing sweep and washout. Typical washout, or wing twist, is implemented to create a negative angle of incidence on the outboard section of the wing. Its combination with a sweptback wing creates a sufficient pitch-up moment arm to balance that of the main wing, trimming the aircraft. It follows that the larger the wing sweep angle, the less washout required (for a constant CG position). Both excessive sweepback and washout, however, have adverse effects on lift and flight performance. In practice, the most optimal solution is to combine wing sweep and washout with an airfoil that produces a minimally negative to zero pitching moment. 

A symmetric airfoil would produce a zero pitching moment, however, the minimum drag for a symmetric airfoil occurs at zero lift, therefore during level flight, an angle of incidence is required and excessive drag will be created. To obtain the desired pitching moment a reflexed airfoil is the most appropriate choice as they have minimal negative pitching moments along with a positive coefficient of lift at zero angle of attack.


Figure 1: Demonstration of the physical approach behind a reflex airfoil [1]

Cg Location

The center of gravity (CG) is one of the most important factors in the design of a flying wing. Acting as the pivot of the aircraft, its location is often influenced by many factors, particularly payload. In the case of Wren, to avoid sudden shifts in CG, its optimal location is coincident with the center of the cargo bay. This follows, that upon releasing the UGV, the flight characteristics remain unchanged. Specific internal configurations will be required in order to obtain this. 

Additionally, while a forward CG location improves longitudinal stability, it is not without its drawbacks. When the center of gravity of the aircraft is displaced from the center of lift, wing twist is necessary to provide the moment balance, as previously discussed. This twist, resulting in a down lift at the wingtips, adversely affects the maximum lift of the aircraft. The figure below demonstrates the percentage loss in a maximum lift for various CG locations.

Figure 2: Effect of CG location on the maximum lift [1]

More significantly, however, is the impact on the induced drag whose value is quadratically proportional to the distance of the CG to the elliptical center of lift. Figure 3 below, combines the losses of lift as well as the increase in induced drag, to demonstrate the optimal CG locations for a given taper ratio. Early Horten sailplanes are plotted as a reference.

Figure 3: Optimal CG locations [1]

Tests by [1] showed that the critical cg location ahead of the centre of lift before controls became unflyable was around 6-12% of the mean chord length.

Aspect Ratio

A large aspect ratio is a trade-off between reduced drag and reduced maneuverability, while the opposite is true for a smaller ratio. For a given wing area, as the aspect ratio increases, the wingspan follows suit. The adverse effect of tip vortices then occupies a smaller percentage of the entire wing. To optimise forward flight a larger aspect ratio seems to be the obvious choice however depending on the flight requirements and mission profile of the particular competition, an appropriate aspect ratio can be selected in conjunction with the average operating speed.

Figure 4: Display of the affect aspect ratio plays on the coefficient of lift [2]

Taper Ratio and Stall

Due to the delicate balance of the wing, local stall can disturb the flight significantly. For example, stall at a wing root when ahead of the CG will cause a nose dive while local stall further outboard, behind the CG will cause a pitch up. The ideal location for flow separation can, therefore, be inferred to be in line with the CG A crude approximation to achieve this given an idealised lift distribution, a minimum taper ratio of 0.6 should be selected with an increase to 0.7 perhaps being more desirable, as proved by [1]. Moreover, it was discovered that for a sweptback wing with washout, the optimal taper ratio to minimise parasitic drag lies between 0.7 and 1. Conversely, a strong taper ratio will produce generally all positive results apart from the increased induced drag above and center of lift implications displayed in figure 3. 

While it seems that the Horten design neglect the previous fact, by having strong tapers, authors Nickel and Wohlfahrt, stress that a flying wing with a taper ratio less than (greater taper) 0.7 has either:

  • No optimal Lift Distribution.
  • Unfavourable stall Characteristics.

Sweep

As mentioned, sweep is necessary to balance the locations of the centre of gravity and aerodynamic centre, below is a brief summary of the effects different amount of sweep has on the aircraft. 

For larger sweep:

  • Creates a larger moment arm for control surfaces located toward the wingtips.
  • Creates a larger moment arm for wingtip washout, increasing its effectiveness.
  • Creates less lift when compared to a straight wing by approximately the cos of the sweep angle. 

The opposite is true for smaller sweep angles. 

Washout

The amount of washout required can be either a small amount, needing elevon deflection at low speed, a large amount, needing elevon deflection at high speed or somewhere in between. The selection should depend on the operating lift coefficient. For Wren, this would be cruising conditions given the influence of the quad-motors at low speeds speed, take-off, and landing.

The washout amount can be determined using the moment balance equation:

The value of C_mw can be expressed in the form below, and experimentally found from the graph.

Figure 5: Washout function in terms of aspect ratio and sweep angle [1]

Lift Distribution

Recent studies from [3] express the benefits of a bell-shaped lift distribution in its ability to create superior control and proverse yaw. Additionally, it is known that the Horten brothers implemented this lift distribution similarly in their design. The effect of proverse yaw was not attained by the Horten’s however, due to the inboard nature of their elevons explained by [3]. The bell-shaped lift distribution developed by Prandtl and implemented by Horten and NASA [3], however, produces more drag for a given wing when compared to an elliptical lift distribution as mathematically explained in detail by [1]. Horten and NASA increased the aspect ratio (Span) to reduce this loss in drag to attempt to justify the benefits of proverse yaw. A firm expression from [1] suggests that, particularly for models (small scale aircraft), other techniques are more effective in reducing adverse yaw and that a bell-shaped lift distribution is inappropriate. 

It would worth building models from the theory of both [1] and [3] to compare the results first hand. 

Figure 6: Horten Bell-shaped lift distribution [1]

Horten Design Philosophy [2]

  • Followed the above, used large washout at the wingtips 
  • Wingtip twist created a bell-shaped lift distribution which was theorised to reduce adverse yaw.
  • They used a larger aspect ratio to make up for the loss in lift near the tips. 
Figure 7: Photograph of the Horten IX/ HO 229 

Northrop Design Philosophy [2]

  • Early designs pre B-2, Chose to avoid excess twist and used pusher propellers relying on vertical stabilisers.
  • Pusher props along the trailing edge of the wing created a stabilising effect
B2
  • Due to the fly-by-wire relaxed stability excess wingtip twist was not needed and instead, the flight computer handled the numerous control surfaces to stabilise the aircraft. 
  • The B-2 uses wingtip mounted control surfaces that crack open to create drag and thus a yawing moment. This could potentially be implemented into WREN however the response is known to be very non-linear, resulting in difficulty in control. When the B-2 is cruising the yaw control surfaces are cracked open to the point where they are only just catching the air. The flight computer then can fine-tune the aircraft’s dynamic response.
Figure 8: Northrop Grumman B2 with yaw control surfaces cracked open

References 

[1] K. NIckel and M. W. Wohlfahrt ‘Tailless Aircraft in Theory and Practise’  AIAA educational series 

[2] D. Raymer ‘Aircraft Design A conceptual Approach’ AIAA educational series. 

[3] A.H Bowers ‘On The Wings of the Minimum Induced Drag’ 2016.

April 1, 2019

Quick and Simple Wings for WREN’s First Prototype!

By Daniel Wong

After 2-3 long nights of preparation and manufacturing, Arfin and I built quick wings for our first prototype of WREN. These will be used to conduct tests in our first flight of the full model with the aim of testing avionics for out flying wing VTOL aircraft and WREN’s overall aerodynamics.

They will be joined with the aligning carbon fibre tubes to the main midsection/fuselage. The reason for the quick change wings is to give us the ability to swap out and use other sized wings tips, so we can fly using the optimal setup for any situation.

We have manufactured using a simple laser-cut rib design, consisting of spaced ribs and supporting struts across the length of the wings. This was combined with foamboard leading and trailing edges to ensure no flow separation and wrapped in solar film for a smooth aerodynamic surface. Finally, foamboard ailerons were attached to a servo and cloth tape hinges to enable the aircraft to steer.

This build was quite a learning experience, as they rapidly learn the technique of developing a sufficiently rigid internal structure, shrinking the wrap by heat and attaching ailerons. There are certainly many problems and improvements that can be made, alongside the many hiccups along the way but I hope to practice this technique further into the future alongside another big project in the works!

November 18, 2018

Vacuum Unit and Foam Moulds

By Matthew Payor

This article is about the initial setup of a portable vacuum unit for performing layups.  Here’s some action pictures.  We’ve since rebuilt the unit to address some shortcomings.

So we’re hitting build season soon.  Coming off the back of constructing the IBIS airframe earlier this year, there were a bunch of aspects to our composites manufacturing that we wanted to revise and test.  Furthermore, the UNSW Redback and Sunswift car teams moved to a much nicer workshop, leaving behind their composites layup room that we previously relied on for laying on our moulds.  So Daniel Mann, Adam Temesvary, and I over the past month or so have built a portable vacuum unit for performing layups, and with this have performed some small scale tests, looking to improve our mould design and manufacture, train some more members in the process, and have some fun along the way.

Building the Vacuum Unit

The vacuum unit was essentially cobbled together by Dan and I from parts ever-generously donated by the dumpster elves.  The pump unit traveled to UNSW from a bin at CSIRO Lindfield, the cylinder was thrown out by a workshop.  The fittings, valves, and bolts were thrown out from the old UNSW car team workshop.  The acrylic panel was formerly an interior window from a renovated lab.  The electrics hail from a random assortment of old disassembled machines.  We mixed all these ingredients together, left them in a furnace overnight, and then chipped off the scale the next morning to reveal our beautiful “new” portable vacuum unit.

Making the Control Panel

Yes, it is overdone and could’ve been much simpler.  Dan and I argued for over an hour about whether to have a fuse or breaker so we put in both.  But there’s something terribly satisfying about flicking one of those old, really strongly detented switches.  So we made it so you have to switch three of them to start the pump.  And then crank down some big ball valves.  More is better, right?

Doing some layups

From the last build, we mainly wanted to address the amount of handwork involved in finishing our moulds.  The problem being that when sanding a 1.2m^2 mould, then recoating it, then sanding it again up to 4 or 5 times, the probability of making a mistake is quite high.  Mistakes come in the form of the edge of some sandpaper nicking the foam in a complex corner, residue picked up onto the sandpaper leaving scratch marks, or taking too much material off a feature.  The idea we came up with was to replace handwork with machine work.  With offcuts of carbon fibre and foam about 300mm^2, I put together some test moulds – an art palette I drew up, and some bicycle seats found on Grabcad.  Many thanks to internet heroes Max Morozov and Eric van Helmond for providing such models on a shared public platform.

Adam and I tested leaving a freshly machined mould in place in the mill, sealing it, and then re-finishing it the next day.  We compared the results from using PVA vs resin for sealing, and also tried thinning the PVA with water and alcohol at various ratios.  In the end we came to the conclusion that a proper resin coat was required to get the kindof finish quality we wanted – PVA sealing still left some small voids that would have to be filled by applying a thicker coat of mould release wax later on.  However the resin, in setting, heats up and soaks a fair amount into the foam we use, and results in a slight shrinkage of radial dimensions due to the heat effecting the foam.  We found that using thinned PVA first (using about 2:1 with water, alcohol proved worse) would harden up the surface of the foam to recieve the resin better.  In finishing the moulds, we tried finishing with -0.2mm radial and axial stock prior to resin coating, so that when we went to finish on the resin coat we wouldn’t machine most of it away, however it proved difficult to coax the dimensions out, probably a waste of time for the kindof work we do.  We’ll be buying some gel coat in ensuing weeks to retry this.  Another change we made was to use a spray gun to apply the polyvinyl alcohol to the moulds.  This allowed us to build up lots of thin coats to create a much stronger and more even coating than when we used to hand-apply it.

Dumb Stuff

So something I personally wanted to try was laying up with plant matter, just for the sake of interest.  I think palm leaf bowls are cool, and moulds that size were about the right size for our offcuts of foam.  Unfortunately we do not have food safe resin, so I thought the next best thing to a bowl would be a paint palette.  It took 3 attempts to figure out a good prep procedure for leaves.  You need to remove the volatiles and water from the plant matter before it makes friends with the resin, without taking it too far and turning them into dust.  The process I landed on was boiling leaves in soapy, salty water, and then repeatedly in plain water until the tea produced was weak and the leaves had thinned and lost some colour; then drying them between towels.  I used some maple leaves donated by a tree next to my apartment, the tea along the way was quite drinkable.

The technique when laying also needed to be different to when laying fibre weaves.  I made a lot of mistakes and Adam essentially guided us to success here.  Whilst the leaves are slightly permeable, the resin is much less able to move around under the vacuum pressure than with carbon fibre and fibreglass.  The successful product has some bubbles and a thick layer of resin on the mould side because we applied too much resin to the base of the mould, since normally that resin will be pulled up and through fibre weaves by the vacuum.

Mould Design

There’s a bunch of annoyances that I’ve come across during layups and finishing that we can retroactively work to mitigate through better mould design.  For example, its good to design a mould that has plenty of clearance from the geometry and a pooling area around the geometry to prevent resin from overflowing around the edges of the mould.  Sometimes these considerations for containing resin make the release more difficult.  If you inset geometry vertically to put a border around it, that vertical section then adds issues to the ease of release.  It’s a balancing act.  What is also useful is to work a lip into the first finishing of moulds prior to resin coating to keep that resin from spilling into the mill.  Its also very important to round the corners of the mould.  Fillets are a good thing to but square edges are a major strain on the vacuum bag. Making the mould thinner in the Z means the bag doesn’t have to work as hard, but the foam is more bendy.  Another balancing act.

Training

Layup Training Session

In the end this was a great opportunity to use some offcuts and leftovers to test and develop new processes, to dust off rusty hands and train up some new people.  The training lay had a few mistakes which impacted the results – a good learning experience for us all.  Due to a hurried sticking down of tape and cutting of the vacuum bag,  we didn’t give enough slack in the vac bag in one axis.  This lead to some voids and creases in the lay.  The small area in the negative seat mould had the polyvinyl torn off during handling prior to laying, and this left a noticable spot with matte surface finish.  Good fun in the end, a few days work resulting in some cute proof of concepts.  We walk out of this with some fancy kit, better informed mould design, and a far easier mould preparation process.

October 11, 2018

Cost effective method for making moulds for composites

By Anthony Sobbi

At FLU we’ve been experimenting with a cost-effective method to make moulds for our carbon fibre parts.  Here is our current process to make smooth carbon fibre or glass fibre pieces:

  1.  Mill the mould from XPS foam and use 240 grit sandpaper to smooth out the mould.
  2. Use 400 grit to finish sand
  3. Use PVA wood glue to seal the mould. This prevents the foam piece absorbing the resin. Also requires less work than using gel coat. 
  4. Use 240 grit to sand down the PVA and then use 400 grit to give it a smooth finish. 
  5. Add mould release onto the surface.

Even though gel coat will give you the best finish, PVA works wonders! It’s cheap and very easy to work with.

We found that certain stores have different densities of XPS. In Australia for example, our major hardware store sells XPS but it’s horrible to mill. We get our XPS from a foam supplier that normally use it for building installation but it’s great to mill with. The density of their XPS is around 35 kg/m³.

We are going to be experimenting with a few new methods over the next few weeks so we will post up our discoveries as we go!