CLA Episode 1: Engineering and Program Management for Quality

Transcript:

Brandon: Today we’re talking with Frank Webb, who leads engineering and program management at Edison, and Frank - The topic today is talking around quality, so I’ll be interested in your perspective on what does quality mean to you and how are your functions supporting the end result of Edison providing a high-quality product for our customers.

Frank: yeah, so when you get when we talk about quality, obviously, we're talking about product quality and even a good manufacturing organization that never ships a bad product still cares about internal quality because that drives cost – right, rework and scrap and things like that. When we're talking about it in the context of a manual assembly operation, which is usually a big flavor of what we're doing, we're talking about making a process and interlocks in the phases of the production process that can essentially reduce the opportunity for an operator to make a human error and increases the chance that we're going to catch that human error early on in the process before we add a bunch of value. So, that comes down to a lot of sequencing the high-risk things. If you can move them earlier in the process, obviously, then you'd like to do that. Sometimes, it'll be offline sub-assemblies, so we'll have a kind of a feeder sort of a line system where high-risk, maybe electronic modules or highly sensitive components, are assembled separately and then fed into a final assembly. Also, it comes down to how you're delivering direction to the operator and what you're asking the operator to do. An example, actually, that I've seen a couple of times is if you've ever heard of the red bead experiment, it's a thought experiment where you give a team of folks; usually, it's a consultant that does this, and it's a flashy thing, but they give you a little scooper to pull red and white beads out of a bucket and it you scoop it out and then the number of red beads in your bucket or in your scoop is your failure rate. So you'll give the team a whole task to how do you reduce it. and is it the way you shake it? do they do they weigh differently? The moral of the story is the red beads, and the white beads are physically identical, it's just a color. You're not giving that operator a chance or any tools to control that, so if you're you think about it that way, that's obviously a highly abstract, kind of a straw man argument, but you have to think of the way that you're presenting the task to the operator from those grounds. What does the operator really have control of that can affect quality, and how do we take things that they can't control that could affect quality and control them some other way? That’s usually where you end up with high automation, where the operator may install something in the wrong orientation, so we're going to give them a jig that's pressed with a machine, and it does, so the operator can't make that mistake. Well, if you're looking at this from the context of an assembly operation that is lower in volume and doesn't justify that kind of capital assembly, you start to think of other ways. Visual guidance, maybe - vision systems that come at a relatively low cost these days can help detect a misaligned part. Or sometimes, it's the way the tools are laid out. Is the operator reaching from here to there? Are they walking? Are they getting distracted in between the operations? An intelligent assembly of the process can really reduce the operator's chances of making a human error.

Brandon: Yeah, and that's such an interesting point, especially for people who might come from a higher volume background, right? Which is typical for the automotive industry and some other adjacent Industries. I think when you think of quality when you think of control, you think automation. You think remove the possibility of something being incorrect. It sounds like, yes, that philosophy drives us, and we're working in that direction. Hey, how can you remove ways in which this can be assembled incorrectly? But there's a limit there, right? Because part of this capital-light philosophy that makes sense for low-volume production is we can't invest millions and millions of dollars on highly automated systems that make it impossible for something to go wrong. We need to be able to bolster that with creative and intelligent manufacturing processes for how this actually comes together. Can you expand on that?

Frank: Yeah, so when I think about that I think it really comes down to fixtures and process design, and also it really comes down to feedback from the operator. So, that's the part where, if you have a really integrated sort of a culture in the organization, and you really involve the operators, the folks that are going to be doing the work and assembling the product, in the product launch process, and you time that product launch process intelligently, so that you’re expecting a stop-and-rethink process at numerous phases uh along the production process. It oftentimes, at first glance, if you look at that program timeline, it feels like we're starting and stopping, which is generally never something you want to do in a production operation if you can avoid it. But we're not really starting and stopping - we're shifting focus. We're doing a build, and then we're learning from it, and then we're doing a slightly larger build, and we're learning from it, and we're doing a slightly larger build than that before we get to a steady state operation. That gives us time for both the product to mature (sometimes those initial units make it out in the field somewhere, and engineering wants to make a change, the product isn't performing, or there's an optimization we can make or cost optimization) but also, we're optimizing the process, and that comes down to the tools, the fixtures, the equipment, the instructions we're providing, the sequence of those operations, but also you're tuning up your training program for the operators. You're learning where the operators are giving you feedback - that, hey, I keep messing this part up, or this requires a lot of focus, it seems fiddly – right? That's one of the worst things you want to hear is fiddly in a manual assembly operation, but if you have that context set in the right way up front and you plan the program around that, you really, it's like slow down to speed up, right? you're going to do an intentional operation, you're going to intentionally learn from it, you're going to do another one, and you're going to intentionally learn from that. So, usually, in our program launches, there are three or four alpha or beta phase builds that build on each other before we get to something that's actually as robust as we want it to be.

Brandon: So it sounds like this thought process is critical to executing those upfront high-change engineering builds successfully, but then maybe even more importantly, it's critical to provide the fuel and the learnings that allow us to operate at a rate as we scale up and really be humming and deliver I mean not that we're getting to these high volume limits but when we get to the rate of what we're doing whether it's hundreds or low thousands or whatever then sets us up for success by learning from all those early builds.

Frank: Exactly

Brandon: Maybe the last topic I'd be curious to get your thoughts on – is this idea of product-centric quality, right? So, especially, I mean, obviously, Edison as an organization, we cover a wide range of products, from small components up to full vehicles and upfits and such, but for this, these types of build these capital light, these lower volume builds, can you talk to the importance of understanding that each product is unique and accounting for that intentionally when putting together a quality plan?

Frank: Yeah, each product is absolutely unique. We do things that range from a junction box up to a full vehicle upfit or sometimes powertrain swaps and things like that. So, every product is different, but the thought process that goes into it has to be looked at from the product’s perspective and also from the people's perspective. So, if you look at it from the perspective of the operator and the product, we think, how could these two work better together? And so, when we give DFM feedback, that's where a lot of it comes from. There's obviously the low-hanging fruit stuff like, oh, commonize your fastener drive types - that's something that any manufacturing engineer will be familiar with, and when they see multiple derived types, they will get a little bit stressed out. But, if you start to think about that a little bit differently, you're going to think about maybe two products that can't fit in each other's places, but visually they're not very distinct. That can drive inefficiency, even if it won't assemble the wrong way, the operator having a visual - one's green, one's blue - or maybe we can reorganize something to have a different feature on top so it's inherently, intuitively obvious to the operator which one I'm grabbing, in what order I grab on the tab, and then I grab them without the tab. Even if the tab isn't necessarily a functional tab, something that's visually distinct to the operator is part of the product, and if the product can withstand something like that, we try to think a little bit creatively. Sometimes, it's just the color of the insulation on something that that can help make the learning curve for the operator getting really into the groove with an assembly step shorter, right, that something is visually intuitive to the operator, and then that's part of the product. The product needs to be intuitive to assemble, within reason obviously, right we don't want to go and completely reassemble everything to be Legos, but there is identifying those hot spots during the initial builds, and we need to think about this product and a small change can really uh not just drive efficiency and reduce errors and cost quality and all that, but can make it so that the operator doesn't feel like they're having to put a ton of effort into the learning curve. The learning curve comes naturally, and then you get that kind of the flow state of – alright, now we can take a step back, we're not stressing out about which product, which part goes in what order, now we can think about how do these actually fit together, how am I being presented with the parts and the tools, and how is it being articulated to me in work instructions, and how can we really tune that up.

Brandon: Yeah, thanks. That makes a lot of sense, and there's a ton more information. We can get layers and layers deeper, which I'm sure over the course of this capital-light assembly series, we will, but given the short nature that we're trying to Target here, I think that we can wrap it up for here. So, Frank, I really appreciate it. I think this is a good overview of what quality looks like and how we think about quality for low-volume Capital light assembly, and I appreciate you uh joining the conversation.

Frank: Absolutely