Best practices for large sheet metal flashings

[IB55]

Hello,

I am modeling sheet metal roofing which has flashings, gutters, falls, buttstraps. I created detailed part with master sketches and used it to create multibody parts for each element. Multibody parts are getting too big and complicated with lots of features & sketches. I started to question myself if I have picked the correct modeling strategy.

Can anyone who is experienced in this type of work give me a advice what is the best method?

[IgorMir]

I personally prefer Derive component route. It is known as Muscular Modeling Technique. 
Cheers,

Igor.

[Gabriel_Watson]

Not an expert in roofing or sheet metal, but as an Inventor veteran I would say you never want to do complex structures inside a single part.

There are typically two ways to design: top-down, and bottom-up.

I usually prefer bottom-up because I work with Vault as PDM. I am not sure if you work alone or in a small team, but certainly in larger teams involving PDM you have to avoid workflows that require you to work on the top assemblies all the time.

Bottom-up you would probably build parametric components (which you can always link to an central, external Excel sheet if needed), or leverage the use of iParts (not my favorite with PDM). Top-down you have a better chance to reuse your master sketches, and as suggested above you could break apart your existing solids into different components via Derive. However, I would still think that the cleanest way to design is to not rely on a master sketch unless you are working with the Pipe or Frame Generator.

At the end of the day you also have to question how these parts will be sent to a manufacturer, and as sheet metal you would probably want to export DXFs from flat patterns, and multibody elements may not generate a flat pattern properly.

[SharkDesign]

Sounds way too complicated for multibody. 

Definitely should be a standard assembly.

If you need everything to change size if you alter something, this is the best way to do it:

https://youtu.be/8qu7jwK9M6k

[i.beridzeSPC6F]

Thank you for the video and your suggestion.

I use parameters a lot and I do link them. But it is practical only if the top assembly is not complicated or if you have around dozen parameters to control. This works well particularly if you know what may change in design. In my case I have several hundred unique parts in the top level assembly (not counting the fixings). Throughout the design and approval process anything can change till parts are fabricated and installed. Unfortunately this is the nature of construction business. I am looking for a more robust way of design to allow me quick changes. If I have a comment on the drawings today that shape of the roof is changing and in some places roof will have different pitch and in some locations gutter will be wider (just an example), I will have only few days to implement this changes and reissue drawings. And who knows what I will have to change next 🙂 

[SharkDesign]

The idea isn't too control every parameter, you control the key parameters.
So you might have one called roof width and that is the parameter the
iLogic is controlling, but when you draw your tiles they are (roofLength -
100)/tile number or whatever.

Or lintel = frontWindow + 300 or something

Not a great example, but the idea is you push a small number of parameters
and the other parameters are part level reference those.

[i.beridzeSPC6F]

Hi

I forget to mention that everting (even purlins) except of fixings are bespoke on this and most of the projects I work with. Bottom up design in such case is very to accomplish. I used to do top down but the problem with this method is adaptivity and cross reference issues. Top level assembly is very large and it needs to be divided in several sub assemblies. Adaptivity across different assemblies did not work well.

We do use Vault but I am not sure how it can help me with bottom-up method ?

With regards to the multibody parts, I create them with intend to convert to sub assembly. Once multibody part is finished with all parameters and features, I use Make Components to create respective subassembly with individual parts. This parts then can be worked on and modified from the multibody or independently (say to add holes add flanges or to create a flat pattern). My dilemma is to find a good balance between complexity of the multibody part and number of extra features added to generated standard parts. 

[SharkDesign]

Vault won't make any difference to how you model it. 

My main thoughts here are that maybe inventor is not the right software for doing this? But I am not that familiar with the construction industry.

Multibody solids are probably not right for this job. Adaptivity is usually not right for most applications. It often breaks.

[Gabriel_Watson]

Vault can help you store libraries of smaller items so you do not need to repeat the file generation from a master file again and again. It also helps bottom-up design in the sense that each subcomponent can finally be specialized to the maximum, and left to individual designers instead of one mastermind for the entire project.

To anyone used to bottom-up design, your process is much more labor-intensive, no matter how bespoke. It takes longer to set up the bottom-up structure, but once you have parametric design going it becomes faster, and you can also leverage Copy Design from Vault to create copies and customize small changes later.

[cadman777]

Sounds like you need to use a structural steel software.

Tekla: very expensive and steep learning curve.

AS (Autodesk's): expensive, and ask in that forum about the learning curve.

SDS2: very expensive, never tried it but hear it's got a very steep learning curve.

Parabuild: not so expensive, much less learning curve (works on top of AutoCAD and BricsCAD). I demo'd it and really like it compared to Tekla. Tekla does WAY more, but for my little world, it's WAY overkill and unaffordable.

BricsCAD: more of a BIM software (like Revit)

There are others, but those are the main ones I can recall off the top of my head.

I personally would get Parabuild for miscellaneous steel and small structures.

The reason you get structural software is for the 'automatic' drawings and various material lists you can generate.
Inventor is not suited to that out-of-the-box, and requires programming to accomplish all that.

The best thing about structural software is the connections.
Inventor has NOTHING like that.

Very frustrating doing structural w/Inventor.

[IgorMir · Accepted solution]

Hi Gabriel,
How long does it take to set up the bottom up design (please see the Rack Size. jpeg)? The end product is shown in 02-ga. jpeg. By changing parameters in master file - the whole model updates as expected. And drawings too. No crashes, no major rework due to the "sick constrains" is needed. 
As a matter of fact - I always change some dimensions in the Master file after I am done with modeling. Just to see, that all parts are linked together properly. Once I am satisfied with the result -  I change the dimensions back to what it should be. It is fun to watch - how it all gets updated.
Cheers,

Igor.

---

@Gabriel_Watson wrote:

To anyone used to bottom-up design, your process is much more labor-intensive, no matter how bespoke. It takes longer to set up the bottom-up structure, but once you have parametric design going it becomes faster, and you can also leverage Copy Design from Vault to create copies and customize small changes later.

---

[IB55]

I have to agree with IgorMir.
Bottom-up design for the assembly which will have 200-300 unique parts and around 3000 occurrences will take much, much longer than Top-down method. More importantly to make quick changes in to the design without issues will be almost impossible. And this changes can be anything: to satisfy requirements from architect, structural engineer, fabricator, main contractor, other interfacing trades.

Bringing Vault subject in this topic is absolutely irrelevant. There are no library items used in this design and not a single part will be reused in any future projects. Therefore is nothing to publish for re-use. It would be different if the work is in manufacturing industry where serial production of similar or identical products is the key. In construction is very opposite: I have not yet met any architects who wants to replicate existing design.

Thank you to all for your input.
I was more hoping to get workflow ideas from somebody who is working in construction with this type of products (roofs, fascias, soffits, flashing, architectural metalwork etc. ).

[cadman777 · Accepted solution]

@IB55,

Yes, that's the kind of work I have done.

That's why I suggested you use either an architectural or structural software.

They have built-in libraries and are very good with BOMs and automatic drawings.

Inventor sucks at all of that.

I can't suggest an architectural software b/c I haven't used any.
But I have done research on them.

There are many kinds, from framing, to trusses, to roofing, to whatever.

Everybody wants an 'automatic' estimator so they can fire all the people who used to do that.

But it comes at a steep price.

Anyway, if you insist on using Inventor for this kind of project, then your best bet is to quickly author a simple profile for each trim piece and then use it from CC. I've done that on one-off projects, and it's not very difficult, b/c once you have your vendor's template setup, you can reuse it to author a new profile.

Then just use a MasterSketch that's wireframe, and if you have railings and other things that need wireframes, make sub-mastersketches that connect to the MasterSketch by Deriving the relevant geometry into the sub-mastersketches. I've been doing it this way for years and have encountered all the kinds of things you talk about. Changes at the last minute require a simple change of a few dimensions in the MasterSketch, and then everything updates. Then all you have to do is scrub your drawings. No big deal.

[IB55]

Thank you Chris,

You just firmed up what I had in mind and gave me good idea about railings.

[Gabriel_Watson]

Through the use of libraries and copy design, we make industrial washers bespoke to client specification. Conveyors and everything repeatable becomes iFactory powered and we can adjust size on tables or, like I said, simply copy design and modify a few things here and there.
I understand that if you have projects that never repeat in likeliness you can model those faster through top-down design, but over time any company should compose a library of components and assemblies. This is how Inventor Factory also works, even leveraging components from 2D AutoCAD.

You can show me all day how to model an individual assembly is faster in top-down, but having to re-generate most components using Derive should take longer over multiple projects.

[cadman777]

Excellent, glad to hear it.

Below is an example of a project (not architectural) that I did where nearly every part is a vendor part I authored in the CC. Took about 4 hours to get them all authored, but that accounted for only a small % of the time from start to finish. All the parts are used on docks and water-side access equipment (by EMS - aluminum extrusions supplier). There's pipe and tube (easy to convert existing CC), L, U, I, misc sections, FB, RB, fittings, etc. Lots of different shapes and sizes! I included in the profile only as much detail as needed for dimensional purposes. So for example, the doc deck planks have raised ribs across them. I didn't detail that b/c it would eat up computer resources. i just modeled the part to the top surface of the ridges to keep it simple. 

There's a very lengthy discussion on this forum about structural work. It began around 2010. The guy worked w/his company to create structural parts and assemblies that they could 'pick-n-place' to get structural work done. I've not achieved that level of process yet, but have been trying for years. That's why I'm trying to learn VB.Net, to do that kind of thing. I'm betting you could achieve some of that with architectural millwork, esp. if you have AutoCAD 2d drawings to work from.

Speaking of millworks...the other day I was looking at a way to create 3d milling profiles from 2d bitmaps (pixel files) using Rhino3d. That is an entirely different area of work. I was looking at what's mostly for hobbyists, but it's pretty slick what can be done. Only thing is, the more automated you want it, the more money have to spend on software. I consider Rhino3D/Grasshopper combo to 'beat all'. It's used extensively by architects. Haven't had much time to see if it can be adapted to my kind of work though...

[IgorMir]

Where did you see me mentioning Top down approach? I have never said it. But apparently - it is lost in definition. Let me clarify. I am two hands up for Muscular modeling technique. And in my line of work I use Derive components always.

I hope - it is clear now.

Cheers,

Igor.

---

@Gabriel_Watson wrote:

You can show me all day how to model an individual assembly is faster in top-down, but having to re-generate most components using Derive should take longer over multiple projects.

---

[BDCollett]

---

@IgorMir wrote:

Where did you see me mentioning Top down approach? I have never said it. But apparently - it is lost in definition. Let me clarify. I am two hands up for Muscular modeling technique. And in my line of work I use Derive components always.

I hope - it is clear now.

Cheers,

Igor.

---

@Gabriel_Watson wrote:

You can show me all day how to model an individual assembly is faster in top-down, but having to re-generate most components using Derive should take longer over multiple projects.

---

---

The images and description you showed looked like normal top-down modelling. Master sketch - derived parts.

I have never heard of "Muscular modeling technique". In what way does it vary?

I would agree with you either way, the method you linked seems logical for this, regardless of having a team working on it. 

Plan it out accordingly, split it up into a few master models if needed but this is still the way to go.

[IgorMir]

Hi Ben,

That definition has been mentioned in one of the Sean Dotson's tutorials in 2005. I was trying to allocate an electronic version of it, but couldn't find it. All I got is a hard copy of it.
Anyway - here is a link to one of the post where the technique was discussed. 
https://forums.autodesk.com/t5/inventor-forum/tutorial-muscular-modeling/m-p/1377287
Cheers,

Igor.

---

@BDCollett wrote:

---

---

I have never heard of "Muscular modeling technique". In what way does it vary?

I would agree with you either way, the method you linked seems logical for this, regardless of having a team working on it. 

Plan it out accordingly, split it up into a few master models if needed but this is still the way to go.

---

[cadman777]

I quit using 'muscular modeling' many years ago b/c there was too much computer overhead associated w/that method.

Way back then I decided to use 'skeletal modeling' with MasterSketches instead, which allows me to make relatively large models of discrete parts.