Best of Jim: Designing for Assembly vs Designing for Serviceability

This is an old post from another blog of mine, from April of 2014. But as my current client project involves adapting something new to a 20-year-old piece of kit that, while very very functional, was NOT designed with serviceability in mind, as the very agile folks crawling in and on and under it to replace all the hydraulic hoses have learned, so it’s been fresh on my mind the past couple of weeks.

I spent most of today in the driveway replacing the air conditioning evaporator core in my truck. While I was doing so, I figured out that that unit had great design for assembly (DFA), but only mediocre design for serviceability. But what do I mean by that?

(We’ll pause for the inevitable aside here)

“You fixed your own air conditioner? But… doesn’t that require a bunch of special tools?”

Yes, and I happen to own them – I’ve been servicing auto air conditioners for somewhere very near 30 years now. Paid for a lot of parts when I first restored my ‘65 Mustang in high school by doing AC work.

I *did* have to buy a couple tools today as I discovered I didn’t have them. But they’re hardly specialty tools: a 1-¼” combination wrench as I the largest I had was 1-1/8″ (which I also needed), and a set of picks so I could replace the sealing O-rings on a couple of fittings properly. I thought I had some, but I couldn’t find them, so now I do have some. And more tools is always more better.

Fixing an air conditioner is fairly straightforward if you have the right tools and know how to use them. But that’s another discussion.

(Okay, back to our main narrative.)

The HVAC module in the ’99 F350 has portions of itself both under the instrument panel and under the hood. The AC evaporator, blower motor/fan, and blower motor resistor pack (and I’ve now replaced all of these items – hey, the truck is 15 years old and has 331,000 miles on it [November 2017 update: it’s 18-plus years old and has 368k miles on it now. The AC still blows meat-locker cold from this servicing]. Stuff’s gonna wear out. Still far less money than dropping $60k on a new equivalent truck) reside underhood, which is, in fact, a lot easier to access for service than if it was under the instrument panel.

If I have to roll the IP in that truck, I’m writing a check. There are jobs that I’m perfectly capable of doing, I just don’t want to.

It appeared that there is a hatch or cover you can remove to access the evaporator core, but I wasn’t entirely sure how it was attached beyond the half-dozen screws I could see. So I did a quick bit of research online and discovered there are in fact 13 screws holding that cover. AND you have to remove the cold-side intercooler boost tube for access in addition to the brackets I could see would be in the way.

I added a set of silicone boost tube boots to my parts order. I can drive it without AC (I’ve been doing so for a couple months now, anyway). I can’t drive it without that being functional. They’d never even been loosened up, and I wanted backups just in case.

(As it happens, I’ll be sending those back, unopened, as I didn’t need them. Huzzah for the longevity of Nomex-reinforced silicone and T-bolt band clamps!)

And after I removed the boost tube, the two brackets that were also in the way, the receiver/dryer (replaced that, too, since I was opening up the AC anyway), the receiver/dryer bracket…

(Okay, aside time: new receiver/dryer came in this nifty plastic clamshell housing with its own mounts, clearly intended to replace the 3-piece metal bracket on the original. Probably something changed for later model years – my truck is the first year of this model. As it turns out, they also changed the bolt pattern and the new setup doesn’t bolt up where the old setup was. Pop the spring clips to split the clamshell, put new receiver/dryer into old bracket, assembly is the reverse of disassembly…)

…and then I could start in on the access cover. Removed the screws I could see using a variety of means including an 18″ long ¼” drive extension. And now I discovered that while this thing was very easy to put together (design for assembly) out in the open, once it was in the vehicle, servicing it was going to be moderately entertaining. There are 3 screws I simply cannot see at all, 2 on the bottom and one on the lower corner of the rear flange. On the plus side, I can get my hand back in there, so with my nifty M8 ratcheting box wrench, I could in fact remove them.

There’s almost enough room to get the cover out cleanly. The hang-up is an assembly aid: a locating pin molded into the case that fits through a mating hole on the cover.

And once the cover is off, there’s almost enough room to get the evaporator core out cleanly – the culprit again is the locating pin, which is over an inch long, 3/8″ across, and pointed so that it will indeed mess up the fins on the aluminum evaporator core if you aren’t careful. Fortunately, I learned this while removing the leaky old one (nothing made me happier than seeing the obvious witnesses to the leaking refrigerant & oil charge on that old evaporator. My diagnostics were accurate!!!! Also, I’d just spent $150 on the part & a couple hours of labor already) and was able to protect the shiny new one.

So… design for assembly: you can easily put it together in a logical sequence. That module was assembled out in the open at a supplier, trucked to the assembly plant, installed in the cab before the cab was decked to the frame. In all of those operations, there was full and easy access to everything that needs to be screwed, connected, or otherwise attached.

But once the thing is all together, and something breaks or wears out, some poor schlemiel (today, that was me!) has to get to it to fix it, and once all the other stuff has been installed around it, just getting to it is often harder than the actual removal and installation. At least it was accessible, instead of being buried under the IP!

Today, the actual act of swapping the evaporator core was about 20 minutes, not counting evacuating the air from the system and recharging it with refrigerant. But, even knowing what needed to be done in what order and simply doing it, I spent about 4 hours doing the R&I, plus evacuation & charging. I haven’t looked it up, but I’m pretty sure I didn’t beat the flat rate book [EDIT TO ADD: No, I most assuredly did not beat the book, as the book gives 1.5 hours for this job. Could I have done it in 1.5 hours? Now, yes. While sorting out stuff like learning that the new parts don’t quite mate up as-received, no]. I was very methodical – I didn’t drop any fasteners (magnetic fastener bowl FTW), and most importantly, I didn’t strip any of the fasteners threaded into plastic out – rather than trying to beat the book as a pro mechanic might do.

Designing for serviceability could have taken a couple forms here. The simplest would have been to use tabs and slots on the bottom of the cover, eliminating the hidden fasteners (this would also aid assembly: fewer threaded fasteners is always better than more threaded fasteners when doing mass production). The cover and case could be designed in such a way that you could remove the evaporator core without having to remove the boost tube (this is assembly-neutral).

The bottom line is that designing for assembly is now a common practice, and it’s proven. In the 24 years [27-1/2 now] I’ve been an engineer, though, designing for serviceability has most often been an afterthought, if it’s been thought of at all.

(Often, it isn’t. Early in my career at Ford, I was at a design aid meeting around the mock-up buck for what became the 1992 Taurus. An engineer from another department was showing where his evaporative emissions canister purge solenoid would be placed on the firewall. Once the engine/transmission cradle was decked, though, you couldn’t get to it at all without removing the engine – a 5 hour job. I suggested a location only about 4″ away that was easy to get to after decking, making a repair about a 10 minute task. “But that extra hose & wire will cost us another $0.05/vehicle!” he said. I asked him what the warranty repair rate was. Taking that, the annual production volume, and the then-current rate for warranty labor at a dealership, I quickly did some math and showed him that spending a nickel a vehicle would net-save the company $1.5 million in the first year of production alone. “Huh. I never thought about that.”)

Designing for serviceability is generally assembly-neutral to a net benefit to assembly, but makes life so much easier for consumers and service personnel alike. This isn’t just about cars & trucks – any machinery that may require warranty repairs can greatly reduce the cost of that warranty (and, also important for customers, downtime of the machine as repairs are faster), and once it’s out of warranty, the cost of repair that the customer has to bear. Commercial (and some individual) customers often calculate lifetime costs before making a purchase decision – they may spend more up front for something that’s going to have reduced service costs over its lifetime.

Something to think about if you design stuff, buy & use stuff, or ever have to work on stuff.


This is just a placeholder. Eventually, I hope to put here some tales of engineering, and some of the projects I’ve done — within the bounds of confidentiality agreements, of course. I’ll be sharing tips and techniques that may be of use, and highlighting some of the ways our services may help you solve a tricky engineering problem.

Thanks for stopping by, and we look forward to serving you!