Hi Prop,
In the file with the .005 fillat, I just worked it and did the trimming... Remember Michael describing the viewport being meshed by your video card, and turning everything into triangles instead of curves.. That part would mean at some type of zoom level, you wont be seeing whats really there.. "Maybe that instersection was actually right!!""

But this is just only going to get worse, as Michael had described.

The G1 fillet was easy. I just removed the little radius on the trailing edge and then ran the fillet tool..

Please note that everywhere you look, it will be "removed" for actual work.. It's just there as part of a proper mathematical representation... But like mentioned here in a wiki for airfoils, you'll note that they mention changing some numbers to use 0 as the little radius, when when you need to do computational work with the airfoil:

http://en.wikipedia.org/wiki/NACA_airfoil

Equation for a symmetrical 4-digit NACA airfoil

The formula for the shape of a NACA 00xx foil, with "xx" being replaced by the percentage of thickness to chord, is:



where:

c is the chord length,
x is the position along the chord from 0 to c,
y is the half thickness at a given value of x (centerline to surface), and
t is the maximum thickness as a fraction of the chord (so 100 t gives the last two digits in the NACA 4-digit denomination).

Note that in this equation, at (x/c) = 1 (the trailing edge of the airfoil), the thickness is not quite zero. If a zero-thickness trailing edge is required, for example for computational work, one of the coefficients should be modified such that they sum to zero. Modifying the last coefficient (i.e. to -0.1036) will result in the smallest change to the overall shape of the airfoil. The leading edge approximates a cylinder with a radius of:



Now the coordinates (xU,yU) of the upper airfoil surface, and (xL,yL) of the lower airfoil surface are:

Hi Anthony, it likely is possible to do it, but it's probably going to require some fairly difficult steps like maybe working part of the time at one scale and doing other operations at a different scale.

Just fundamentally the model that you're trying to build is not a particularly good fit for CAD technology in general - a mixture of extreme different scaled features within the same design will lead to many problems.

Once again, my best advice is to not try and create a model that has that kind of problematic structure in it - maybe put a line in there or make the end just come to a simple sharp point.

It is still not very clear to me why you think that little tiny radius is necessary - what kind of CNC are you expecting to build with that will be able to cut such tiny features?

- Michael

If you actually get it built, and then actually get it "Meshed" for FEA, hopfully you have access to an SGI Octane III at a minimum.

When the tool cuts the metal there will be a radius. It won't come out as a straight line. That is what I was trying to say. Also, from a stress and fatigue point of view you can not have a straight line or sharp features of any kind or the part will crack and fail. There will and should always be some sort of fillet at that boundary. The downstream analysis for the CAD model would be FEA, CFD, aeroelastic etc. You want the correct geometry for a lot of reasons. My program outputs the information to build the le and te based on the naca 65a009 definition. The radius will vary with each design but the part bothering you will never change. The te will always be very very small compared to the rest of the model. Making it a point or line is not only not a real geometry but if it was made it will fail instantly. If you look at turbine or compressor blades they have small fillets at the base. Those will be way smaller than in this model.

Yeah it takes a lot of elements, a lot of memory, etc. About a 10,000 workstation at a minimum and ansys costs over 40,000. Its not cheap but done everyday. I used to do it for years.

Well, it's kindof a conundrum there prop.. Yes, stuff like this is done everyday. But the guys doing it would just design it the way it needed to be...

Can you show me the cnc machine that is being used everyday that works at a .000006 tolerance and the tooling that it uses to make a fillet of that radius in a corner area???

Hi Anthony, if that's really what you need then MoI is probably not the right tool for your particular job - you may need a more specialized kind of program to deal with unusual (unusual in the sense of what MoI is oriented towards working with) criteria like that.

MoI is just not specifically designed to handle this particular task that you need to do.

It may be something that I can adjust in the future, but it's a difficult task to try and expose a whole bunch of adjustable tolerance values in the UI for every kind of action that you want to do - that would probably help with what you're trying to do in this particular case but would also result in a much greater avalanche of technical info being presented to the user all the time, making for a much more complex and more difficult to use program.

It's a major goal for MoI to be easy to use and to not try to require a math degree to operate it...

- Michael

burrman

the ter is .000067 meters roughly. i attached some info that may help. the moi model here is in the actual dimensions. probably through all the scaling some confusion arose. the issue bugging michael is that the largest dimension / smallest dimension (radius / ter) is approx. 47619. for a compressor or turbine blade that would be a lot smaller. But even the chord / ter is still 4761.9. Somewhere in there lies the rub I believe. Burrman got it to work so I wanna see if I can too. I like MoI better than rhino because it is simpler to use. I think you have a great program here. I wouldn't change it.

Hi Anthony,

> the ter is .0067 meters roughly.

But that's not the case in the A400M Propeller Example.3dm file that you posted previously.

In that file the little rounded tip here:



Seems to be 0.0000666 meters as you can see... That's 2 orders of magnitude smaller than what you wrote here - is one of these incorrect?

Then of course to actually cut an arc shape on your CNC you will need a precision of some fraction of that, maybe a precision of around 1/1000 of a millimeter or so to cut that accurately...


EDIT: I see you updated with the correct figures above, I didn't see that when I first posted this one.


- Michael

Yeah its really small and if the chord gets smaller like the first points that started this thread it would get even smaller. That's why I switched to this example. The large blend that burrman did at the base is what i was going for. But it would still have to blend into the ter. I used to work for McCauley propeller and Hamilton Standard. We made these types of blades all the time. Even guys with hand grinders could do it. The CNC was just an example. I was trying to say that machines don't make lines or points. There will be some radius based on the drill bit or what not. At P&W we drilled with lasers. Believe me there are a lot of ways to do all sorts of things, that's all I was trying to say. Unfortunately the ratio (radius/ter) that I believe is bugging you I can't change. It comes from the fact that the blade has to have a certain chord. The airfoil spec does the rest. Making the ter larger is an option, but not ideal. I was trying to say it still needs to be a radius for a real perspective and from a stress perspective. So if it becomes too small to manufacture as burrman is saying it would have to be enlarged. I was saying my code just uses the naca spec. Its not like I drew it in there for fun or to bug you guys. I'm ok with making it bigger if need be at this point. But what I sent is the actual definition.

What I can do is change the radius / chord ratio. That will increase the ter and still follow the exact geo. What would be the min ter that MoI can deal with given the same radius of 3.170912 meters?

Hi Anthony, yeah the part that's particularly difficult here is the mix of different proportions within the same surface, since that little radius pushes out with the sweep to a much larger length than that radius in the sweep direction.

That means you've got a surface fragment that is like a long very thin sliver - things like that will cause some particular difficulties, particularly with stuff like join, since with a slivery surface like that both of its long edges may be within tolerance of the other piece you're trying to join it to and it's kind of easy for it to be basically treated as if it was a degenerate collapsed line without any surface area at all.

So anyway, this mix of different proportions within the same surface is particularly difficult - if you had something like a tiny part that was all made up of surfaces that were also of the same proportions that would probably be a lot more feasible.

- Michael

> What would be the min ter that MoI can deal with given
> the same radius of 3.170912 meters?

I'm not sure of the exact value that you'd need, but anything that increases the size of the surfaces that are generated so that they're not so extremely thin and slivery for each surface piece would probably help.

- Michael

I played around in a spreadsheet a bit and changing the r/c wouldn't change the ter that much. It also will drop the efficiency which is bad. It looks like ter is going to be really small no matter what I do. So in the worst case manufacturing scenario they would just have to make it bigger. I've never heard of any issues not being able to meet ter in manufacturing though. It's possible they didn't bring it to our attention, I'm not sure.

Here is just a solid with no fillet applied to the connection of hub and blade... I could get about .001 m before the fillet starts to fold over itself... What were you looking to put in there?

I dont think that anything is really bugging Michael. You will just need to jump through a bunch of hoops to model this. Just like in Rhino. But the point may be, if you are just starting out with MoI, there will be alot of low level work for you (Only more to come) then it may prove difficult for you.

Anything that looks reasonable is fine. I'm just looking for smooth blends that looks like a machine or human would do. So that it can be analyzed later for stress and fatigue.

I'm curious how you set the model up to be able to fillet. That seems to be my biggest problem. I have a bunch of surfaces in the right position but can't get them to fillet. I noticed your model was all one piece and it was able to perform a fillet. Do you have to join them somehow? Being new at this is difficult. But I can clearly see MoI is much nicer than Rhino. I just have to figure it out. I've been at this all day, so I'm tired. I'm going to go to sleep. Thanks again for all the help.

I watched the sunburst tutorial and that's almost exactly what I'm doing, but I just can't seem to get it done.

NACA airfoil equations give a zero radius at the trailing edge (a point). Any radius put there is a manufacturing one and shouldn't affect the performance. I have never seen a propellor with a trailing edge as sharp as a scalpel and would suggest there would be no performance gain even if you could achieve it.

Yes, the model i posted is listed as a "solid" in the properties display. There are a couple different fillets that MoI does... Edge and surface.. The surface one works best when done on a solid object though can create a fillet with seperated surfaces more robustley, but those will also require a lower level of work to get the result into a solid.. It's a good idea to work with solids as that will then have well defined surfaces and structure.

To get your model into a solid form for filleting I first rebuilt the arcs to use in the sweep as I had mentioned before. After I completed the sweep I seperated everything into seperate surfaces and ran a trim on the surfaces of the sweep object and the hub object, using the "keep" option to keep the exterior surfaces I wanted to keep. Then I ran join on all the surfaces and created a solid form that I posted here.

The fillet of that edge will only go so far before it folds over onto itself.. If you put a .25 fillet on that edge, you will see the results on that little trailing edge patch that is created... Note the difference in the 2 verticle patches that run up to it... This is whats causeing the issue for the fillet corner patch area. If it was a sharp, it will be able to make a more well defined patch, but at some value, you would have to look at it and ask "What would I expect to see put there???" A proper patch structure cant really twist in and around on itself to satisfy the area...

"""" I noticed your model was all one piece and it was able to perform a fillet. Do you have to join them somehow? Being new at this is difficult.""""

It's not really going to be anything you find in the manual for doing what we are talking about here, but those 3 videos Michael has there, cover things pretty well. Even Rhino wont be having specific direction on working in the manner you are describing. Even the higher end stuff may have a need for understanding whats happening under the hood to have success!

Maybe Danny, who is a user here, can find the time to look at your file with NX, a very expensive modeler using one of the more powerful modeling kernels, and comment on the type of results it produces with simple "boolean and fillet" with no need to jump through hoops...

Bueller??? Bueller????

As an example, I created your model as 2 solids "pre-boolean" and sent it to a program that uses the parasolids modeling kernel. Powershape.. Importing the parts in it first starts me through a process of setting up "tolerances" that I want it to use on the part.. Something like a 6 step process.. Then I booleaned the 2 parts together and it started an automated wizard for setting up the tolerances for gap allowance/disallowance and stitching of the surface to ensure a water tight solid.. I then ran a fillet on the edges we are discussing here.. It ran the same size as the MoI file I presented and also created the exact same twisted patch with the larger radius!!! (Than really fits with that geometry)





The red is the underside of the corner patch rolling over onto itself and folding upwards..

At this point, there will be some complicated "Help me fix this area" tools, asking questions that I need to know the answer to, to try and create something there..The help me tools may not exist for the issue and then I will need to know what to do. It may be a week long process to finish just this one component of the model. Then, the very next issue starts all over again... Maybe a months work or more???

Anyway.

@ Bur: Have you the free Version of PowerSHAPE?