Hi Anthony,
your right the Airfoil profile was reversed I have corrected it.
The kink 4th foil from tip the curve is not a thro points curve in that area and doesn't match the airfoil profile tip.
I think you showed that issue in one of your screen casts.
If you have another example I can trial that would be helpful.
Cheers
Barry

hi barry,

i think we posted at the same time. see the post above this one with my reply. also, i updated the swept fan blade. you asked about this earlier, but i didn't have it completed at the time. this should help with comparisons as well. all the info to compare to should be there now. everything was manually created. there is a bit of extra info you probably don't need in there. i had also did fea on the model to check some other things with PROP_DESIGN. everything wrt PROP_DESIGN seems to look alright the best i can tell.

so this model and the one from the previous post are what i would use to compare to. i will import your geometry into the files and see how it looks.

update 1; so it looks like maybe we aren't using the same input files for the swept fan blade. your geo doesn't match up with mine. on the a400m the sweep seems to be missing on your geo. but you are getting close.

update 2; doing this comparison on the a400m blade shows there still may be an issue with PROP_DESIGN. chord distribution options 3-5 seem like they are causing some blade sweep when measured at the quarter chord. i thought i fixed this but even the non-swept stations look like they may get some sweep just due to how the geometry is constructed. i will have to look into this some more.

i just updated the link to the a400m file. prop design has been updated as well. i made another attempt at fixing the bug related to chord distribution options 3-5.

i don't know if it's possible for you to do, but if you could script the addition of dimensions for the radius, twist, and sweep angles, that would make checking your files a lot easier. you could then check them against the xyz output files. but you should now have everything you need to check the automatically created files for the a400m example and the swept computer case fan example.

i haven't checked my fix for chord distribution options 3 and 5 yet. hopefully it works ok without more bug fixes. i will look into that next. but it shouldn't affect the two cases you are running now.

update; I finished checking cds 3 and 5. found no issues. so hopefully all the bugs are fixed now. this bug was the same as the last one. all i did to fix it this time was apply the previous fix for any sweep angle. before i applied it only for sweep angles greater than zero. i hadn't realized that when you define zero it gets changed for cds 3-5.

Hi Anthony,
attached latest results of the test fan model and the A400M.
On the A400M I used the point files to generate the Span , A & B curves as these had no modification in the tip area.
The result seems to reduce the kink at the tip end also I tried the network command using the Foils plus A & B curves this again seems to give a smooth result.
I have used the cord distance between the span curve point & B curve point as my scaling factor is this correct ?
Let me know if there any problems with the attached files.

Barry

hi barry,

the swept fan blade seems to overlay to my manual geometry exactly, the best i can tell. i think on your a400m you have the old xyz points. i did a bug fix after the last geometry you sent me. i found i still hadn't fixed that bug correctly. i believe i have it fixed now. if you could use the new xyz points, then i'd be able to compare better. but visually just looking at what you sent it looks good. i suspect you have it working, but can't say for sure.

if i understand your question about scaling, yes i think that's right. for the a400m blade i have to orient the original profile. then do a separate 2d scale. i set the c-plane to the profile in question then scale it using the span point as the center and the leading edge rail point as the reference size.

i'll look into network. i have never really messed with that command before. i have just been 2 rail sweep or loft up until now.

there is a new version of prop design if you haven't gotten it yet. it has the updated bug fix. it's the same fix as before i just apply it even when no sweep is specified. it seems cds 3-5 end up with some sweep even for the straight blades. it's a result of the span moving off of the x-y plane. i hadn't noticed it before, until i was doing the comparison for you. so i'm glad you helped me find it. it affects the performance computations, so it's something i have to account for.

update; i tried the network command i got weird results with your a400m geometry, so i tried mine. on my geometry it worked a lot better. so i then tried 2 rail sweep. again with your geo on got weird results. so there may be something going on with your profiles or rails. i'm not entirely sure.

on the fan blade network worked the same with both are files. so the issue seems to be just with your a400m file. on closer inspection though, the network command severely distorts the profiles. so that doesn't seem to be a viable option. if you look at the surface compared to the profiles they are drastically different. i would stick with 2 rail sweep with multiple profiles or loft.

Hi Anthony,
I used the a400m updated file you posted for Karsten is there a later version in the new Prop Designer ?
Re the scaling factor I used the distance from Span to B curve points for my calculations and if I understand you use Span to A (leading edge) curve points I will check what difference that might make.
Regards the network option see attached photo and just to be sure I only use the A & B curves plus the airfiol profiles the span curve is not used.
Barry

hi barry,

sorry to cause you trouble. i updated the link to the a400m files a few days ago. if you download it again you should be good. i tried not using the span before i posted before but rhino wouldn't do it. perhaps moi is better in this regard. i will play with the network command some more and see if i can get what you are showing.

technically you probably could use either rail to scale the profiles. however, i trimmed the trailing edge on my model. so that voids the trailing edge rail. i would have to make a new rail based off the shorter profile, if i wanted to scale off the te rail. so it's not really worth the extra trouble, when i can just use the le rail instead.

update; yeah rhino says it needs 3 or more open curves to do a network surface. so perhaps moi is better there

hey barry,

i was trying to make a different blade for the a400m, so that the kink would no longer be there. i wanted to make a straight blade with airfoil sweep. as i started to work on that, i noticed something seems broke with prop design opt. i need to look into what's run. once i get that fixed i think the blade you are working with will change again. it was made using the broken version of opt. so it's not sized to the inputs specified. i'll get back with you when i get this fixed.

Hello Anthony,

while my experiments I saw that for some examples with an outdated Version of your Software where the number of Points for Span and the rails differ. 19 for A, 18 for the span and 17 for B.

Is this correct?

Have a nice day

wow that's bizarre. no definitely not correct. i just posted the update. i also updated the a400m example. now the straight blade has the same amount of total sweep. but does it without any blade sweep. this gets rid of the kink. but you loose about 2% propeller efficiency. however, i don't feel like you can really build the bent blades in a way that would perform as predicted. so the straight blade with airfoil sweep is probably the better way to go. i have also been thinking about any kind of sweep and whether or not projecting onto cylinders makes sense. for a purely straight blade with no sweep, so chord distributions 1 and 2, the idea of projecting onto cylinders makes sense. but once you add sweep, i'm not sure it applies anymore. so i just wanted to make sure you had the option to apply the projection or not. you'd have to build and test to see how they work.

for the bug you found with rail points, can you provide the file or example you ran. i mainly need to know the number of points to define the blade, the mesh density multiplier, and the number of tip steps. that way i can see what you are seeing. but definitely sounds like something is messed up there.

besides updating the a400m example, i found a bug in opt that was changing one of the tolerance values. i changed one of the optimization criterias slightly as well. so it's now working the way i expected it to. both input files for the a400m changed. the bug was causing the shaft power target to be voided. now both blades are drawing the specified shp.

Delete

Hi Anthony,
in the photo you will see the difference I have in your new projected curves and my calculated ones.
It appears that your projected curves have been rotated back and no long lie on the cylinder is this correct?
It's no problem for me to allow for this.
Barry

hi barry,

there are four profile positions in total. for the file in question, if you turn on the 'imported geometry layer' and the 'projected airfoils' layer you will see all four positions. the 'comparison geometry' layer i just copied over what it seemed like you were outputting. to make it easier to compare to your work.

i want to update this model again, to match the current version of prop design. however, i'm thinking it would be better to use the new a400m straight blade geometry. it doesn't have that bend at the tip but performances about the same. in reality, i don't think the bent blade would actually achieve the predicted performance anyways, due to the bend. there really isn't a way to manufacture the bend that would match prop design. unfortunately, this is a case where prop design modelling the blade as a line is an issue when you try to go to reality.

i can still work with the bent blade if you want. let me know what you would prefer.

for karsten's question. i can't duplicate what you are seeing. i get 64 points for all the rails, using the default settings of 13, 50, 1. let me know how you were seeing those odd results. just to make sure you weren't talking about the significant digits, i checked the source code write formatting and they are all the same. G25.17 is the default format for the output numbers.

hey karsten,

i had one thought on the issue you mentioned. i'm looking at xyz output. i don't know if you are looking at some of the files in the two test cases. i think i saved some of them with spaces i added. the added spaces helped me check the dimensions used in the examples. not sure if that is what you are seeing or not. let me know.

Hi Anthony,
attached latest curves created with node editor for the A400M.
They match the manual ones you created. If you require the other angled options let me know.

Cheers
Barry

hi barry,

looks like you nailed the position and everything. the only problem i see is that when i do a 2 rail sweep using your airfoils they have more definition than the ones i made. so the surface comes out on the heavy side. not sure how you are ending up with more control points than me. even mine come out a little weird too. mine has something to do with the projection. once i do the projection the curves are no longer defined the same. but there's nothing i can do about it. on yours, they come out with even more points than mine. but it seems like you worked out all the bugs.

just as a reminder on features, can you also do the same thing without them being projected to cylinders? so the airfoils would be in the same place as you have now. that would be a useful option.

also, many posts up there was a list of various profiles that would be useful to have. lastly, if at all possible to automate adding in the dimensions using max's dimension script until michael gets his own dimensions going. in the test file i have the dimensions so you can see where they would go. mine are laid out in a bit of a mess. rhino doesn't have a way to move them that i know of. so if you don't arrange them right you have to do them over. so i just left them all messy.

but even what you got now would still be a huge time saver. it takes a really long time to get to the point you have automated. so you have saved a lot of work for people.

thanks a lot for doing this.

Hello ANthony,

I've made a screencast of what is possible. Some things are not implemented at the moment - erasing the high density of profiles, patch the end of the blade, ...
For the patch at the end I have an idea, that i've tesed manually.
But I hope you will get an impression of what is possible.



Have a nice day
Karsten

p.s.: Sorry for my english - The slang is called denglish:-)

Hi Karsten,
Interesting video and I’ve got a couple of questions if you don’t mind.
The things I would like to add to my nod is how you fillet and blend the Airfoil.
At the moment my method require doing this manually then joining and naming. I would like the blend to be automatic and the fillet to be adjustable.
Your new blend node does it use the point on the ‘A’ curve as it’s peak position?
Cheers have a good weekend.
Barry

hi karsten,

thanks for the video, it was great. that's really quite amazing. i think this will be extremely useful to a lot of people. creating new rails is a great idea. that's something that was too much of a pain for me to do manually. it's interesting how much you can automate. you're english is very good. you reminded me of a professor i had in college, dr. ying. everyone said he spoke yinglish. no one could understand anything he said. so i was worried at first. but your english is fine.

seems like you got all the major issues dealt with. i look forward to playing with this in the future.

hi barry,

if i understood you right, in your last post you were talking about working on doing the root fillet. the file you wanted a link to awhile ago has some retention geometry in it. you could use a shape similar to that to attach the blade to, then try to fillet it. i have found this geometry to be about the best to go with. it fits the most blades in it. i have been manually making it, as it varies blade to blade. i'm not entirely sure if you could automate it. i'm not sure of exactly what factors go into sizing it. i found rhino messes up the geometry a lot in subtle ways. the best way to get rhino to accurately make it was to extrude a whole cylinder and then do a boolean split with those funky red curved surfaces. anything else i tried would be malformed. which is typical rhino. that's why i am looking forward to switching to moi. but this should give you an idea of what sort of retention to connect the blade to. unfortunately, simple pie shaped pieces often don't allow you to fit all the blades. so you need that weird cut to get a lot of blades to fit onto a hub. this is the same link as before, just posting it again for convenience. let me know if you have any questions.