Biggest DFM Mistakes – What is DFM? (3 of 3)

The Biggest DFM Mistakes: Designing Things You Can't Actually Make (S2 E18, Part 3 of 3)
The Integrate Intelligently Podcast, Season 2, Episode 18 — with Jeff Brickler and Scott Brickler (CADTALK). Design for Manufacturing, Part 3 of 3.

Jeff: Welcome back to The Integrate Intelligently Podcast, Season 2, Episode 18. I'm your host, Jeff Brickler, and today we continue our series on design for manufacturing with founder and CEO of CADTALK Software, Scott Brickler.

Before we jump in: if you haven't subscribed to the podcast, left a comment, or reached out to Scott or me on LinkedIn, I'd encourage you to. We can only grow this channel and help improve manufacturing with your help spreading the word. Subscribe on YouTube or your favorite podcast platform, and share it with anyone else in your organization.

Scott, welcome back. How's it going?

Scott: It's been a hot summer so far, and we're enjoying it nonetheless. Getting ready to go on a cruise with the family next month in August. Looking forward to this conversation about DFM mistakes — I'm sure there are a lot of them.

Jeff: If you haven't listened to the first two episodes, go back and listen. We covered design for manufacturing fundamentals — DFM 101, the intro — and then part two was implementation: getting your engineers on board and what a design for manufacturing approach might look like.

Today we go deeper into mistakes. DFM mistakes, and how to deal with them, manage them, and fix them.

As a reminder, if you haven't been following this series: DFM is the concept that during the design stage, design engineers take into account how a part is manufactured. Scott has talked in the last few episodes about optimization, and design for manufacturing is one avenue of optimization when designing a product.

So tell me about mistakes. Maybe start with a story about one you experienced in manufacturing, and how mistakes affect manufacturing companies.

Designing Things You Can't Make
Scott: It's been a while since I was actually in a manufacturing company, but I've seen cases where engineers design things that aren't manufacturable.

I mentioned this a couple of episodes ago — it is possible in a CAD system to design something that would be prohibitively expensive or impossible to actually make. Now, it's getting more capable with 3D printing, where you can make some fairly wild things. But when you think about material removal processes, mills and lasers where you're cutting material away, there's a real possibility of creating something you simply cannot make. You can design it, but you can't build it.

That would typically be the biggest mistake. You don't see a lot of it, but you do still see it.

Think about how an engineer designs something. There are roughly two approaches. One is that they simulate how you'd make it — extruding a block and then cutting it away — so they're roughly simulating what a machine would do. The other is what I'd call a blob approach, where you start with a blob and move it around and mold it into what you want.

In my experience, the second approach tends to produce things that aren't really makeable.

The other mistakes are things that are difficult to assemble, or weren't thought through. I have to hang this wire harness on the back of the machine, and I can't reach the screws on the back to get nuts on them. I can't get there from here. So what should be a one-person job becomes a two-person job with walkie-talkies just to put the thing together. It's ridiculous.

In virtual space I can put that together easily. But on the ground with a 2,000 pound machine, I can't spin it around and look at the bottom. Like the car example a couple of episodes ago — things that are easy to assemble on the computer and very difficult in production.

I usually see that around assembly rather than the machining itself. It's possible to put together, it's just going to be stupidly hard.

And once again, they're optimizing on multiple things. They get tunnel vision — how do I take the most cost out of this, how do I make it so we only bend it out of one piece of metal — and they create a situation where the thing is difficult to work with.

Holes That Don't Line Up
Jeff: I tie everything back to working on cars as kids. But also the simple stuff — the bolt holes don't line up. You're on the shop floor, they've cut it or drilled the holes or milled it, and it comes time to assemble and you're saying these things don't line up. Show me on the model how these line up. And the engineer says, well, it lines up for me. Or to your point, there's a nut that goes on the back of this — where am I going to get that, how am I going to get the nut on there?

Scott: Good example — that transmission job we talked about as kids. Think about lining up holes on two heavy things in space. You can do it easily in CAD: line this hole up with that hole, and it just does it. But in actuality you're holding an engine and a transmission, and they're both heavy. How do you get those holes to line up?

A design for manufacturing principle would be a guide pin — a rounded pin and a receiving hole, so the two heavy things get guided into alignment. It doesn't fasten anything, it's just a guide pin to align everything. Then you put the bolts in.

That's a DFM thing somebody added because they realized they couldn't align two heavy objects easily without a guide.

You see it in construction too. If you're putting in a big girder for a building, they don't try to line up holes and then thread bolts through. They design it with the bolts pointing up and stationary, then lower the girder down over the bolts so they align into the holes, and then put the nuts on. Because that's easier than threading bolts through two heavy things.

Little things like that, that you only think about when you consider what it takes to actually build it. And there are thousands of them.

There's another one — design for gravity. Use gravity to your advantage. If you're fighting gravity, it's going to be more difficult. Are you putting bolts in from the bottom and pushing up against gravity, or putting them in from the top and letting gravity align them? These are cool little tricks that incorporate the real world into your design philosophy.

Material Choices and What You Can Actually Bend
Jeff: It isn't just alignment, it's also material choice. You mentioned bend radii in a previous episode — what material you use matters. We can't bend that radius, we don't have the capability. So you end up spending more money on tooling, or outsourcing it.

Scott: Think about a straight right angle. You could design something with a straight right angle and no radius. If it had to be that way, you'd probably have to mill it — take a block of metal and mill the angle out of it. Or you could buy angle off the shelf. But to create that exact effect you'd have to mill it.

So do you care about that being a sharp edge, or is it just easier to design it that way because that's how you started? If you design in the bend radius, you know you can form it, and forming takes far less time and cost than cutting all that material away.

It's a rudimentary example, but by not putting a chamfer or a bend radius on it, you're deciding at the design stage whether it's going to be formed or not.

And you might say my tolerances are too tight — forming with a press brake only holds a certain tolerance, and this needs to be tight because it's going into a slot gauge or a channel. Then you have to do it the way you have to do it, because it has to work. Those are the constraints you're dealing with while designing.

Lead Times and Single-Source Risk
Jeff: I also think about lead times. Sometimes the lead time on a purchased part matters. It isn't the only deciding factor, but you might need to select among several parts based on lead time and what will work. You need more information — otherwise you design a part, and now we need this special component and I can only get it from one supplier.

We see this in semiconductors. There's a lot of talk about this now, with all the rhetoric around semiconductors and AI — the Dutch company that makes the extreme ultraviolet lithography machines that allow manufacturers to make the really fine chips. Only that manufacturer makes them, and only certain manufacturers have the capability to use them.

So designing for manufacturing can create situations where you get bottlenecked to one supplier. And if anything goes wrong with that supplier, you're hosed.

Scott: Because you've optimized in a way where the number of people who can supply that optimization is lower. You end up with the example from a couple of episodes ago — trucks that need a single chip to come off the assembly line. You have a single-source supply chain issue, and when that source goes out, your supply chain is disrupted.

I predict you'll see design for manufacturing incorporate more supply chain constraints than ever, because of COVID and everything else that's happened. Being resilient to supply chain disruption will be a factor of design for manufacturing in the future.

Jeff: Some of the platforms we work with are already coming out with supply chain modules — capabilities to pivot and make better supply chain decisions. And that's going to be needed all the way back in engineering. Even if you have better supply chain information in one system, if you aren't designing with it in mind, you won't know about it. You'll keep doing the same thing: I designed this, this is how it's made, go figure out how to make it.

Scott: And there are all kinds of compromises you can make to deal with supply chain issues.

I've used the stainless finish example before. Number four stainless is polished on both sides, number three on one side. If you specified a part that only needs number three, you could technically make it out of number four. But you couldn't make a part specified as number four out of number three, because both sides aren't polished.

From an optimization standpoint, if I have number four, I can make the number three parts out of it. It costs a little more, but at least I have it.

That kind of game-time decision making on the shop floor will become more of a thing — material substitutions based on supply chain, running optimization algorithms to figure out in real time what's available and how quickly you need it. All of that done automatically, especially in a world of AI.

Jeff: And making faster decisions throughout, from the design phase all the way to shipping out the door, with fewer mistakes.

How to Avoid the Mistakes
Jeff: So to avoid these mistakes, what's the framework or thought process a design engineer needs?

Scott: You always need a reality check while designing. Can I actually build this? Can you visualize how you'd do it if you were that person — if you were assembling it or making it? You may not be a manufacturing engineer, but as an engineer you have a general sense of how things are made.

Then have discussions about the capabilities in your shop and at your vendors during the design phase. Do a spot check. There are also design reviews, where you ask colleagues to look at your design and consider how easy it will be to make.

Those are good principles for anything you're building. We do the same with code, and with any engineering discipline. And in the future you'll be able to ask AI to help with that too.

Jeff: Some systems are already starting to have that — some of the platforms we work with use AI to help make those decisions.

Scott: It's going to be a huge revolution of AI-assisted everything. AI-assisted design is definitely in the future.

Jeff: And AI-assisted DFM, I'd imagine.

What Do Engineers Actually Learn About Manufacturing?
Jeff: A question, since I'm not an engineer. You focused on industrial engineering, which puts you in the middle, closer to manufacturing. When a design engineer finishes school and starts working at a manufacturing company, how much do they understand about manufacturing itself? How much training is involved?

You worked at a company with a shop floor you could walk out to. But I'd imagine at some companies the engineer is more removed from the manufacturing process. How much manufacturing knowledge does an engineer get in school, and how do they get it otherwise?

Scott: I went to school for industrial and manufacturing engineering, so we got quite a bit of it.

A typical mechanical or electrical engineer — more mechanical than electrical — has some knowledge of it. And some of it is just experience. If they work somewhere with machining, they pick it up quickly, because designs get kicked back a lot. There's the iterative approach: this isn't going to work, we can't make this.

Jeff: When they're starting out, they probably have knowledge similar to what a doctor has about nutrition. They have the basics, but they can't do much with it unless they really study it.

Scott: And you have to know the nuances of your company and its capabilities. They generally know how certain things are made and the general principles, but they learn a lot from experience.

Wrapping Up the Series
Jeff: Another good conversation — not just about CADTALK, but about design for manufacturing generally.

A quick plug: CADTALK has a design for manufacturing module that helps get data out of the ERP and into the designer's hands at the time of design, to help make better decisions. I'd encourage everybody to download our DFM PDF — there's a link in the show notes.

Subscribe to the podcast, give us a thumbs up, comment, share it internally, and connect with Scott and me on LinkedIn. Fair warning again: if you look for me on LinkedIn there's more than one Jeff Brickler, so make sure you get the right one. All that information is in the show notes.

A preview of our next episode: we'll give some practical examples — maybe using CADTALK, maybe just using what you have today — of how to implement a design for manufacturing strategy with your team. No cost, low cost, or just practical steps.

Scott: Looking forward to it.

Jeff: That's all for today. Thanks for the conversation, and thanks everyone.