J-20 5th Gen Fighter Thread VIII

[phrozenflame]

Well i understand. They are using NURBS which basically STEP file. and they said it's alternative not substitute. Nor it said requires smaller surface or "resolution"

Your second paper doesn't even talk about modeling resolution at all nor even in the microwave range it talks about compact RCS range using extremely short wavelength for measurement. Are you really reading them ?
----------

Well TBH This is the kind of answer i wish you could provide... This is from POFACET's manual.


View attachment 116304


Look at the Rule of Thumb section.
-Object must be large which my model is.
-Facet's edge which basically the facet's dimension.. it has to be bigger than the wavelength.. not smaller apparently.

But then ANSYS is SBR+ solver while POFACETS is PO. Will the rule of thumb works ?

That's how a clear and concise explanation is.

Also i apologize for being rude and i actually forget to thank you for your care.

I just want to say me and many others really appreciate your work and really appreciate the effort done by Aircraft 101.

 

[latenlazy]

F-117, B-2, and even F-22 weren't done by what counts as supercomputers by modern standards.
Provided @Silentflanker's models are close enough to real things in geometry and hotspots (edges, intakes, protruding elements etc) treatments, they can be used at the very least as the best available relative estimates.

At the very least, here we have a consistent and transparent methodology.

1) That assumes he took the right approach to the modeling (this is why I emphasized the importance of approximating curvature if he wasn’t going to increase polygon counts).
2) the F-117 B-2 and F-22 all used a whole lot of experimental data as part of their models and those models were almost certainly recursively updated.

So no, I don’t think it’s a given that you can get a decent approximation with a hobby exercise. Methodological transparency is indeed good *but* the point of transparency is to enable the ability to judge and assess methodology because ultimately what you care about is accuracy. Transparency on a bad methodology is not a good result.

 

[HighGround]

1) That assumes he took the right approach to the modeling (this is why I emphasized the importance of approximating curvature if he wasn’t going to increase polygon counts).
2) the F-117 B-2 and F-22 all used a whole lot of experimental data as part of their models and those models were almost certainly recursively updated.

So no, I don’t think it’s a given that you can get a decent approximation with a hobby exercise. Methodological transparency is indeed good *but* the point of transparency is to enable the ability to judge and assess methodology because ultimately what you care about is accuracy. Transparency on a bad methodology is not a good result.

To be honest, all of your criticism went way over my head. It's a little hard for people to understand who don't have a background in physics. I've a strong background in math, any reading you'd suggest to understand this particular topic better?

 

[latenlazy]

To be honest, all of your criticism went way over my head. It's a little hard for people to understand who don't have a background in physics. I've a strong background in math, any reading you'd suggest to understand this particular topic better?

Take a sphere and facet it into a dodecahedron. Would you expect both of these two shapes to reflect EM waves the same way. Take an RF beam with a wavelength of 3 cm. Do you expect that beam to interact the same with a dodecahedron with 7 cm edges as one with 2 cm edges. His model is constructed from polygons. Where there are edges and flat surfaces and corners in many parts of his model, a plane is smooth with much fewer flat surfaces and corners. Where in real life an RF beam might interact with a curved surface that has no corners, one edge, and no flat surfaces his model would have multiple edges, flat surfaces, and corners. He never dealt with these problem in his model. Can you get a rough idea of where the plane is reflecting RF energy at any specular angle? Sure you can but that’s not the same as the question how much, how tightly, etc. All those shape factors I just talk about introduce variance. If your variance overpowers the magnitude of scale by which your key performance parameter is defined (aka the significance) then you cannot answer the question you’re interested in answering. Even for comparative purposes, if the difference between two objects is at the millimeter level but your models can only be precise at the centimeter level, your error factors can drown out the difference, and you can’t say with any certainty which two objects is actually a better performer than the other. To answer more precise questions you need more precise models.

 

[HighGround]

Take a sphere and facet it into a dodecahedron. Would you expect both of these two shapes to reflect EM waves the same way. Take an RF beam with a wavelength of 3 cm. Do you expect that beam to interact the same with a dodecahedron with 7 cm edges as one with 2 cm edges. His model is constructed from polygons. Where there are edges and flat surfaces and corners in many parts of his model, a plane is smooth with much fewer flat surfaces and corners. Where in real life an RF beam might interact with a curved surface that has no corners, one edge, and no flat surfaces his model would have multiple edges, flat surfaces, and corners. He never dealt with these problem in his model. Can you get a rough idea of where the plane is reflecting RF energy at any specular angle? Sure you can but that’s not the same question of how much, how tightly, etc. All those shape factors I just talk about introduce variance. If your variance overpowers the magnitude of scale by which your key performance parameter is defined (aka the significance) then you cannot answer the question you’re interested in answering. Even for comparative purposes, if the difference between two objects is at the millimeter level but your models can only be precise at the centimeter level, your error factors can drown out the difference, and you can’t say with any certainty which two objects is actually a better performer than the other. To answer more precise questions you need more precise models.

I see, yes, that does clear up a lot thank you.

 

[taxiya]

F-117, B-2, and even F-22 weren't done by what counts as supercomputers by modern standards.
Provided @Silentflanker's models are close enough to real things in geometry and hotspots (edges, intakes, protruding elements etc) treatments, they can be used at the very least as the best available relative estimates.

The problem is the bold texts. You see, scientific research start by data gathering and sampling from the real thing. Less than that is just imagining, no matter how it seems to be close.

Any research will have errors. The final error (or accuracy) is the multiplication of errors introduced in the chain of study. The very first error is how accurate the model represents the real thing. This error can not be determined by thinking, but by measurement.

At the very least, here we have a consistent and transparent methodology.

A flawed methodology.

 

[by78]

High-resolution images from the Air Force Day even in Changchun.

 

[by78]

 

[by78]

 

[by78]