Best approach for a configurator

[mat_hijs]

Within our company we have been looking for the best workflow to go from design, to fabrication, to mounting on site. We make curtain walls, which mostly consist of mullions, transoms, glazing, panels, … Our biggest issue lies with the mullions and transoms because these are the most complex, the most varying and we actually fabricate these internally.

A mullion (or transom, but I’ll focus on the mullion) is basically a purchased profile, which is cut in the correct length (and sometimes angle) with a bunch of holes, cutouts, …

There are a bunch of things that have to be assembled onto this profile during our fabrication like gaskets, screws, plates, … I have made most of these things as content center parts because most are purchased and can vary in length.

At the moment I make one “master” assembly for a mullion and try to put in every known variable, which can then be configured by entering the parameters on an ilogic form and running some illogic rules (in the correct order, as some things are calculated during or after a specific rule ran).

Because we can have a couple hundred different mullions in a big project this means we now have to copy this master assembly (and the underlying profile part, which always has to be 1 to 1) lots of times. This is done manually at the moment, but probably could be done with illogic fairly easily.

If I have a variation that wasn’t included in my master assembly (or part) I can just add a feature manually after copying the master, or decide to change the master assembly from that point on. There is no link between existing copies and my original master assembly.

After the models are done I constrain them onto a grid I created (using a macro that will find the pre-created work features existing in both the grid and the master assembly) to rebuild our complete curtain wall. And then create a detailed fabrication drawing for every copy of my master assembly (again using a macro to put dimensions between pre-created work features existing in the master assembly).

This works, period. Are there things to improve about this? Most definitely!

Now, my colleague has a bit of a different view on this whole thing. He basically wants to enter all the data into excel, and start building assemblies based on that data. Notice I said building, because he wants to skip making a master assembly, and go straight to generating parts and placing them in a new empty assembly. I believe that in principle this would be possible, however I also believe that the code would be immensely complex and long for a non-programmer (IF we get it to work at all, that is).

In some sense I can see some benefits in this workflow, because if you’d have to change something to all the copies you already made, you could edit the excel, edit the code and rebuild the assemblies. And of course, not having to manually copy everything could save a little bit of time too (even though I’m very wary of this because it could overwrite existing files without you even noticing).

However, here are some issues that I suspect we will run into:

First, we are not programmers, so I think that this would be very hard to make this, and the chance to have issues in the code which will cause errors or issues further on seems very big to me.

Second, because we are not programmers I’m not convinced that writing or editing this code would be easier or faster than creating or editing a master file like I have done previously.

Third, I believe that when you’re just entering numbers into excel without any visual check the chance of making errors increases, and maybe this could even take longer than entering this data straight into Inventor.

Fourth, I think that if we want to create a somewhat detailed drawing we’ll either have to create dimensions manually or have some work features existing in our assembly to create dimensions programmatically. If these all also have to be created with code, this will further increase the complexity and the length of the code.

Fifth, to rebuild our complete curtain wall in an efficient way I also think we would need some work features to constrain everything.

One very important factor in this whole thing is that we NEVER know every single variation we will encounter before we start. We NEVER have ALL data gathered before we start.

To give an example to this, I’m currently working on a 13 floor building. The first 7 floors are already fabricated, 5 of which are already mounted on site, but I’m still waiting on information to make the remaining floors.

Also it could very well be that from 300 different mullions, only one has some specific hole or cutout, in some way I’m flexible at the moment because I can just change that one specific mullion without having an impact on all the others. I think that if you want to build everything from an excel, you would need to have some parameter or something in every single mullion which would make it so there’s no hole, except in the one exception where the parameter would make it so that there is a hole.

Now, I’m wondering what the best way forward would be. My initial idea, my colleagues idea, a mix of the two, something completely different, …?

If anyone could share their experiences, thoughts, ideas, … that could be very helpful!

Sorry for writing a book on this post, I hope my explanation is clear enough to understand my general points, if not, feel free to ask more questions.

[mat_hijs]

@thomas.fitzgerald Could you possibly share your thoughts on this?

[matt_jlt]

Its sounds interesting what you are doing. But it's hard to give advice as i don't know exactly what you are doing and what the assemblies look like.. I deal mostly with mining / heavy industry and a few random things thrown in there so not too familiar with curtain walls either. 

I have even had instances of clients asking for custom code and it just ended up they weren't using the software in to it's full capacity / correctly and all it required was a change in their workflow and there was no custom code needed along with increased productivity / speed and reliability with their models.

You are right about it will take more code and effort especially if you are unfamiliar with coding. I have done a fair few 'automated' type models for various clients and no one situation is similar. It depends entirely on your workflow, how your libraries are setup etc. although it does seem to be a trend that people seem to like having an external driving spreadsheet.

Advantages of the master sketch approach is it's less code / already working for you and you don't need to invest time and money in another solution. If the intention is to save time, you need to weigh up how much time you will actually save vs the amount of time it will take to develop a new method.

On the other hand, having it not constrained but all placed by code etc. is very reliable, but if your code isn't up to scratch, or falls over / doesnt have these 'variations' built into it. It could mean more effort if things change.

And yes a mix of the two is possible aswell. It sounds like there are lots of variables that is almost impossible for me to provide any real help. You would need someone to come up with a clear step by step process and outline before you got started.

[thomas.behrendY5FX4]

Sounds like you want rather a fix product that is already on the market like Speedmax. 

They use the Inventor API to remote control a template. You have to design all the part files, position the parts in a configuration once and when you are done, you can combine the parts with a few clicks in a self developed mask and get the 3D model.

In your case, maybe design a template with named iproperties you can set with the configurator, create a new assembly with the parts and set the iproperties. Don´t make a direct add-in, make a standalone exe that connects to Inventor, that´s easier to develop.

Details on how to do it are beyond my actual understanding how you work. I´m a dev in a company for high security access restrictions ranging from simple fireproof steel doors to giant nuclear powerplant / fushion reactor plant doors.

[dkovac5Q3G4]

Did you maybe discover the best way how to do the work? Do you have some videos or something of your workflow ? I am also in the same industry and we are using Inventor...but mainly iparts ,iassembly with excel and similar but I think there is a better way with ilogic just there is hard to find time to learnt it and itc.

[mat_hijs]

I'm still doing what I used to do at the moment, taking my code and models to the next level step by step for every new project I do. The main reason for this is that I don't really have the time to build something like this, I just have to do it while still doing all my other work. I haven't used a lot of iparts and no iassemblies at all. I'm interested in exploring them though. The drawback I have with iassemblies is that (if I'm not mistaken) it only makes new configurations from existing components, which won't work for me as I want to create (some of) the components automatically based on the configuration.

[A.Acheson]

Is there a chance to attach even a basic concept of your worklow? No major detail on parts required just some simple extrusions to show the file linkages etc. From what I have gauged from many workplaces is that there should always be time allowed for process improvement especially if you have a lean design team. Otherwise your team faces burn out from tedious tasks and just the sheer lack of people to do the workload. Time spend in R&D is usually time well spend later on (works best if your team has allready has ideas of the improvement needed) 

From my expierience iAssemblies will only work if the products is a design that is consistently changing. There cannot be custom iparts or any user interaction other than to switch members. Otherwise the change becomes a new member. It is usefully if your just changing a size of a frame but it can be inflexible depending in the changes. 

For custom work a regular assembly with iparts(standard or custom) works well ma ually or with ilogic driving the members. For custom iparts unfortunately you need to select a path for the custom member each time which is a pain. You can drive this through ilogic and this might speed up the part placement. 

[mat_hijs]

I attached a Pack and Go of a simple mullion. To configure it you should go through the form tab per tab in order to get the best results. Like I said, this is a simple variation where everything is either vertical or horizontal. I also have variations where things could be at any angle, which makes everything that much more complicated.

I totally agree about the R&D thing, but my company simply can't miss me doing my actual job at the moment, so that means no reserved time for developing things like this. Which means doing it on the go for the specific needs of the current project. Not as elegant or clean as it probably could be, but it gets the job done.