"I hate MBD". Why? 4

[pmarc]

"I hate MBD". This statement was made in another discussion about UOS all over profile application. As I have been pretty deeply involved in MBD for some time (drawing creation, proper CMM consumption of PMI characteristics, etc.), and personally see huge advantages of this methodology over traditional 2D approach to product definition, I would be really interested to see/hear/understand why people hate MBD so much. Of course, if possible, I would also hope to receive a positive feedback about it.

Could anyone share opinion?

Thank you.

 

[drawoh]

pmarc,

About forty years ago, I worked as an inspector in an auto parts plant. I inspected oil strainers by sitting them on a fixture and checking to see how they cleared the base. I never saw drawings. This is a successful application of MBD. The oil strainers were weird and non-orthogonal. The company made hundreds of these things per hour. They needed me to not be very skilled or expensive. I never saw drawings. I needed to inspect stuff quickly.

We have a business partner who is a big aviation company with its own CNC machining facilities. They design large, complete parts to be CNC[ ]machined from billet, in-house. I do not know how extensively they do MBD, but they do it on these drawings. Probably, it makes sense for them.

Our manager has seen this and announced that we are to do MBD. It is "the new way of doing things". We design machined components, mostly for limited production. We have successfully avoided doing detailed dimensioning. At least one of our fabricators takes our models and completes the fabrication drawings. Our inspector may have to do this too.

It does not take me long to detailed fabrication drawings. I am often curious about the dimensions, so I apply reference dimensions to the MBD drawings.

MBD saves money when fabrication and inspection can be automated. It may be the only way to document weird, non-orthogonal curves and other shapes. In my environment, it saves a small amount of my time, and it creates more work for everyone else.

--
JHG

 

[3DDave]

The part of MBD I see as not likely to work is when there is no access to the engineering model. As long as everyone in the chain has the tools required, then it seems not too bad. It also seems like it's a problem when there is a problem and the system is not homogeneous. I can say surface ID XYX relative to ID RST is out of tolerance ID JKL, but those aren't going to be in any STEP file that gets transferred to an incompatible system, so I have no way to describe beyond, the bigger surface on the left .. not that one .. the other one. Maybe everyone has to create a WebEx account.

As long as all the pieces are lined up it should work great. I don't know of any CMM software that accepts native CAD models and can add the inspected variations in a form the engineers can directly interact with in the original CAD tool. If I'm sending models out I expect the same courtesy and get the same models back with the inspections data linked to them.

What I dislike about MBD is that it decreases the interaction of the user with the model in crafting an explanation of how they think it should act. The MBD software often has this notion of 'annotation planes' which is all the work of a drawing with none of the clarity.

On the drawing I can clearly state that some requirement on some sheet in some view and zone is such and such and a guy on the other end can be certain he/she is looking at the same thing no matter what software they have. In CAD MBD? Send a STEP model -and- a drawing.

Not that sending models is a bad thing as a manufacturing tool. But if the model does not encapsulate the allowable variation in a mathematically/programatically defined format, there's likely to be trouble.

tl;dr if there is an end-to-end scheme in place with 100% fidelity, it seems good. I haven't seen that yet, but maybe it's out there.

 

[Oaklandishh]

Model Based Definition?

 

[JNieman]

As a designer I love the concept. As a manufacturing planner I love the potential for implementing near-paperless production without sacrificing quality and efficiency.

As someone who also takes customers data when we make others' parts... I sometime facepalm over how it's implemented.

When you send a model accompanied by a PDF with screenshots of 37 views, each with view-specific annotation normal to the view-plane... I question their understanding of the process. There are many places where management, whether business management or quality management, put too many wrenches into the mix because too many people want to half-implement the process and end up breaking it by not committing to it. So they end up with MBD as-defined on paper.

I'll sometimes get a STP model with screenshots from CATIA, but none of the annotation came with the STP model. What's the point, then? Sure, I can program machinery to the STP model but now I'm back to manually programming the CMM from prints, basically. Except now they are crappy haphazard screenshots with annotation that sometimes isn't even normal to the view plane. One step forward, two steps back.

However, as I mentioned before, I love the concept. I think that, when done right, it makes for an accurate, efficient process with reduced waste and reduced risk during manufacturing.

I also don't believe the efficiencies apply to many sectors due to the lack of efficiency-of-scale and because some parts are not best inspected by computerized inspection equipment. But I'm not going to berate a system for not being a 100% solution for the world.

 

[MintJulep]

Model Based Definition.

Plenty of people equate that to "Send them the model" (or a STEP or other extract from the model)

Mostly that defines the nominal geometry.

To fully define a part you need geometry, tolerances and an inspection plan.

I'd suspect that the "I hate MBD" crowd is doing it wrong.

 

[powerhound]

I think Mint is onto something here. What's not to love about MBD? If it's because one does not have the technology to interrogate the model then that's one thing but it isn't MBDs fault.

John Acosta, GDTP Senior Level
Manufacturing Engineering Tech

 

[pmarc]

Thanks a lot for all responses. Please keep them coming!

As I am short of time today, I will try to refer to at least some of the statements made thus far in next day or two.

 

[3DDave]

PH - It can't be MBD's fault - it has none. The only fault is with those who claim it is applicable or usable in situations where it is not.

The only MBD I've seen is sending models out with few or no tolerance controls and later writing changes against certain areas when the parts don't fit. It works only because the suppliers manufacture with smaller variations than the design requires. It's possible the parts could be cheaper were there greater variation allowed in the supplier process.

I've mentioned before that tolerancing is most valuable when the expected variation is similar to the required limits. If the process result varies much less than the requirement there is little point making the measurement and therefore a lot of people think there is little point in documenting the requirement.

 

[BUGGAR]

Sort of related: I'm designing a fabrication using waterjet cut parts. How do I design for the waterjet cut parts? Right now I include individual dxf parts files for each part as requested. But practically, I need a key plan for talking with the suppliers, assigning part names, materials, etc. And how to specify cutting speeds/finishes? Some shops use layers, some colors. I currently dimension the fabrication assembly drawing but do not dimension the parts to be waterjet cut. The shop I work with now, the guys just feed the file in and remove the part when done with no real inspection (unless there is a bust - then it's always "my bad file"). I can't find any drafting standards on waterjet or similar.

 

[JNieman]

There are no drafting standards for waterjetting because every software will do it differently.

It's not your responsibility to give them a perfect CAD file so that all they have to do is "push button, deliver part" That's almost offensively absurd.

I say this as someone who can program, set up, and maintain Flow waterjets.

We use discretion on the finish if the customer doesn't call for specific roughness call outs. When in doubt, a call with their buyer helps, if we don't just err on the side of safety and run it at high quality.

It would be absurd of me to expect our customers to know what colors FlowMaster software uses to define their 5 or 6 preset cutting speeds, and to /assume/ we would /only/ use those 5 and never custom settings, and to deliver a CAD file using those colors and ONLY those colors.

What if they should tab the part? What about nesting? Where should the lead-in/lead-out be?

Why do you pay them if not to do their own job? I hope you get a discount if they're making you supply these things already done for them.

How do you design for waterjet parts? You make a print that defines what you want the final product to look like, at a minimum. If you're being nice, you can supply a DXF of the nominal geometry. If you're being really nice, you've cleaned up the file, deleted overlapping geometry, double checked for gaps, put everything on one layer, set everything 'bylayer' so the colors are easy to change.

 

[pmarc]

Again, thanks a lot for all replies.

I would like to start from MintJulep's post, because I too think that it touches a really important aspect of the problem. A lot of people tend to have negative opinions/feelings about certain concepts (not only MBD, Model Based Definition), while in reality this is just a result of incorrect understanding of what the concepts are really about.

With that said, MBD does not mean that detailed definition of a product is not needed any more. MBD is just as much about full and unambiguous dimensioning and tolerancing as any other traditional method of product definition. The main big difference is that in traditional approach (let's call CAD generated 2D drawings this way), unlike in MBD, geometric characteristics specified on the drawing are not embedded in CAD model, thus can't be "automatically" used/consumed downstream as a source of allowable variation by manufacturing, inspection or quality planning.

I agree that MBD (or actually the broader concept called MBE, Model Based Enterprise) requires that everyone in a food-chain has access to the engineering model. But that is exactly what this is about. That is why it is called Model Based Definition or Enterprise. The CAD file, and everything that is in it, is a master that drives the entire process.

High quality traditional 2D drawing requires high quality GD&T (among other things), but even that can be worthless if downstream users don't understand GD&T. In MBD/MBE high quality GD&T applied to the model can be quickly and seamlessly transferred downstream without a need of being interpreted in between by any human being. This, to me, is something really huge. It is quite impressive to see when for example a CMM software imports a CAD file with a geometric definition in it, and with the press of one button it is able to extract that definition to create all inspection paths for verification of the definition.

Of course, a real life is not always that beautiful. In many cases it is impossible to take full advantage of MBD (for different reasons). Sometimes, as it was already stated, MBD probably does not make sense at all. But it is an interesting tool/methodology that definitely opens up new interesting possibilities not available in traditional 2D approach thus far.

 

[BUGGAR]

Responding from the building and bridges construction community, we have Building Information Modeling (BIM), which does the same but for buildings. The idea is perfect in theory, if difficult in application. The major killer is that BIM is being usurped (as in Microsoft) by AutoCAD. Now that they have captured most of the Architectural/Engineering world, they are forcing us onto subscriptions and their related problems and costs.

JNieman: Good info! I'm new to specifying waterjet cut parts. It's like learning how to specify welds - you have to give the welder just enough info so that he can strike an arc immediately without wasting his time figuring our what you want.

 

[drawoh]

The part of MBD I see as not likely to work is when there is no access to the engineering model
...

As a designer I love the concept. As a manufacturing planner I love the potential for implementing near-paperless production without sacrificing quality and efficiency.

As someone who also takes customers data when we make others' parts... I sometime facepalm over how it's implemented.
...

When you want parts fabricated, you need to communicate with your vendor. A whole bunch of people out there are rule driven, and they do nothing more than follow steps of a process. This is not communication. Maybe it is harder for them to completely screw up drawings. Some fabrication works well with MBD. Some fabrication does not.

I am sending drawings out to jobbing shops, and they are not well adapted to MBD. Joe Osborn does not approve of MBD, and this is the environment I am working in.

--
JHG

 

[KENAT]

Full blown high end MBD probably works adequately in environments where the supply chain is somewhat 'integrated' i.e. using compatible SW/file types and having a common understanding of how stuff is meant to be used etc. I especially like the aspect of CMM pulling in the tolerance requirements (though still requires the CMM/User combination to properly understand implication of that GD&T which I'm not entirely confident in).

However, for the rest of us it can be problematic.

Also despite having read Y14.41 cover to cover a few years ago, it still seemed somewhat incomplete for the user, and more about telling the CAD packages etc. what they need to allow for.

The whole idea of having saved views in the model with corresponding tolerancing & notes etc. applied seemed less user friendly than a more conventional drawing where everything is laid out for you.

As Drawoh implies, even if you go the hybrid route of dumb model and partial drawing there are issues. If the QA guys and other users can't interrogate the model efficiently then a full drawing is probably more useful.

We originally introduced it (hybrid) for use with castings and molded parts etc. that had relatively complex hard to detail geometry and where the end result basically depended on the tooling. Interface (or otherwise key) features were fully dimensioned on the drawing while the other geometry was covered by the model + an all over profile note or similar.

On a recent project where they were trying to go 'fast' they went the route of sending more stuff (i.e. machined parts) out effectively model only with badly done & generally inadequate hybrid drawings. They more or less got away with it (or perhaps just muddled through) on the Engineering prototypes but as they've moved to production have been having more and more issues.

I'm bemused every time I hear of places that claim to be fully MBD and then send out the aforementioned screen shots with the models - clearly they'd be better going true hybrid rather than faking full MBD.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[moon161]

One place I worked, made chillers and such on an assembly line. SWX E drawings of the assy were sent to flat screens. Both the bom printout (Navision) and the assembly model (SWX e drawing) had to be arranged in strict touch order for the assemblers. I thought it was daft that assembly order had to be pushed all the way up to SWX and Navision where it was hard to change, or had a perverse effect on the modeling process. An assembly drawing with a few sheets on paper or a flat panel would have been the easiest thing in the world.