Tesla’s Unique Battery Cells, Manufacturing With Robots vs Humans… (Cleantech Talk #28)

Sign up for daily news updates from CleanTechnica on email. Or follow us on Google News!

Editor’s Note: In this episode of Cleantech Talk, Matthew and I chat about Tesla’s unique 18650 batteries, Tesla manufacturing and manufacturing in general, robots, Volkswagen’s gigafactory plans, and more. Enjoy! –Zach Shahan

Welcome, everyone, to a belated set of show notes for this Cleantech Talk episode! While we strive for swift turnaround times, tardiness sometimes strikes. On the bright side, until Tesla starts releasing new vehicles on-schedule, at least we’re in good company. 🙂

Speaking of Tesla, I mentioned on the podcast that, while its batteries have an “18650” form factor, the company customizes them. Every off-the-shelf 18650-format battery contains safety features. Tesla’s custom-built 18650s don’t. Quoting from US patent 8088511:

“The inventors have found that for many applications in which a plurality of batteries are coupled together within a battery [module], such safety elements may be included elsewhere within the battery [module], thereby substantially reducing battery complexity, weight and manufacturing cost. Additionally, reducing the complexity of the cap assembly improves the ease by which the seal can be formed between the cap assembly and case.”

Because Tesla groups the batteries into modules of roughly 4–5 kWh apiece (with about 400 batteries per module), it builds its safety systems at the module level. That allows Tesla and Panasonic to simplify each battery (saves money) and even shrink it slightly (saves space) because they don’t need redundant safety features for every single cell.

Readers wanting to check out the USPTO (patent / trademark office) website to do some searching can click here. Of course, it can take years for patents to be issued, so if you want to keep tabs on what your favourite companies or inventors are up to, you can search through patent applications instead. Have fun!


As for Mercedes (and Toyota, for that matter) looking to de-emphasize robots, the Guardian article is pretty clear:

Skilled humans can change a production line in a weekend, where robots take weeks to reprogram and realign.

As if to emphasize the unpredictability of software, the day after we recorded this episode, two of my colleagues lost more than 10 person-hours debugging a highly automated spreadsheet one of them had already spent 10+ hours developing, to streamline the analysis of prospective wind farms.

The culprit? A should-have-been trivial upgrade of some software (the equivalent of going from version 3.1.1 to 3.1.2) which had the effect of adding some blank lines to its data export file format.

The highly-automated spreadsheet probably saves an hour per project, but as a result of the software upgrade, the “payback” time went from about 10 projects (10 person-hours) to 20 projects (20 person-hours). God help us if any other vendors decide to “improve” the formatting of their data export files!!

As for the calculation of how expensive software glitches — or any work stoppages, really — can be in the automotive sector, if you make 500,000 cars per year, that’s the equivalent of 10,000 cars per week. Assuming two shifts, five days per week, that’s 1,000 cars per 8-hour shift, or 125 cars per hour. With an average $5,000 gross profit per vehicle (probably low) then if your production goes down for two hours, your profits just dropped by 5000*125*2 = $1.3 million!

Combine the cost of robots (which could climb into the millions to begin with) with the risks of production stoppages due to hard-to-hunt-down software glitches, and it’s understandable that some car companies are shying away from robots.

Tesla has made a big deal about its manufacturing robots (which, I’m reliably informed, are absolutely top-of-the-line robots). In its defense, it’s only producing two vehicles, meaning its robots don’t have to be programmed to deal with the breadth of product that a Mercedes has to (perhaps a dozen models, each with sedan / hatchback / wagon / convertible and possibly other options). As Tesla’s line-up broadens, we’ll be able to see if they follow the legacy automaker’s arc.

Slowing down to speed up

It might seem counter-intuitive to suggest that Tesla might want to temporarily slow down now (with the S and X), to be better able to speed up later (with the 3).

My line of reasoning is that it’s harder to make improvements on your production line if you’re running at your full production rate: everyone will be so busy making stuff, that it will be difficult to improve how they make it. (For instance, Toyota engineers apparently make their production-line changes on the weekends, so as not to interfere with the sacrosanct weekly production plan.)

Now that Tesla has a $1.4 billion bigger cash cushion, it should have the cash to be able to slow down S and X production (or slow down the ramp-up rate), so as to more aggressively weed out remaining or lingering quality issues. The faster they put the X’s teething challenges in the rear-view mirror, the more completely they can focus on the 3!

If nothing else, this could also save a lot of money: it’s far more expensive to re-work a car that’s just come off the production line. This is because car assembly is a mass-production process, while re-work is inherently artisanal. Every re-worked car might have a different problem, which means you’re taking components apart and putting them back together in ones and twos, instead of the hundreds or thousands that you’d get on a production line.

The biggest advantage to this approach would be that the Model S and Model X production lines could then serve as lower-volume test beds or pilot projects, for the kinds of innovations Tesla may be looking to put into place for the Model 3. It’s always less stressful to try improvements out on lower-volume products, than to try to jump straight into high-volume deployments!

You can listen to Cleantech Talk via the embedded player above, on iTunes, or on SoundCloud. And you can check out our Cleantech Talk archives for previous shows if you haven’t listened to them yet.

Have a tip for CleanTechnica? Want to advertise? Want to suggest a guest for our CleanTech Talk podcast? Contact us here.

CleanTechnica Holiday Wish Book

Holiday Wish Book Cover

Click to download.

Our Latest EVObsession Video

I don't like paywalls. You don't like paywalls. Who likes paywalls? Here at CleanTechnica, we implemented a limited paywall for a while, but it always felt wrong — and it was always tough to decide what we should put behind there. In theory, your most exclusive and best content goes behind a paywall. But then fewer people read it!! So, we've decided to completely nix paywalls here at CleanTechnica. But...
Like other media companies, we need reader support! If you support us, please chip in a bit monthly to help our team write, edit, and publish 15 cleantech stories a day!
Thank you!

CleanTechnica uses affiliate links. See our policy here.

Matthew Klippenstein

Matthew Klippenstein is a renewable energy consultant in Vancouver, Canada. He has chronicled the Canadian electric car market for GreenCarReports.com since 2013, and has provided commentary (in English and French) for print, television, radio, web and podcast media. An early guest on "The Energy Transition Show", his work has also been discussed on "The Energy Gang".

Matthew Klippenstein has 37 posts and counting. See all posts by Matthew Klippenstein

14 thoughts on “Tesla’s Unique Battery Cells, Manufacturing With Robots vs Humans… (Cleantech Talk #28)

  • I currently get paid close to 6-8 thousand dollars a month from freelancing online. For everybody looking to complete simple at home jobs for several hours every day from comfort of your home and make valuable paycheck for doing it… This is a work for you… SELF40.COM


  • Other problems with rework is that it inherently a lower quality and higher cost product. Those costs are not material at lower volumes and can be absorbed by assembly laborers and they can learn the root cause of the rework; none of that is the case at high volume. Rework rarely gets validated so the process going into the product is unproven.

  • That robot argument can be turned on its head. If your line runs for two hours with no (or minimal) staff, then you are millions ahead. Ditto if it runs at double speed (or faster).
    If you haven’t seen how fast robots are now, I recommend this video:

    • What’s the output loss when a new human has to be trained to do a task on an assembly line? How about when a critical worker is a bit under the weather?

      Programming times will shorten. A fully automated factory can run 24/365. Not even Christmas day off. No slowdown on the day following Thanksgiving due to too much turkey and eggnog….

      • Above, Zachary compares robot software with desktop computer software. Not the same. A robot which is functioning at peak has no need for an operating system upgrade. (Indeed, nor do most desktops, but where would that leave Apple and Microsoft). It is only code change that causes bugs. Systems in steady state run for years without a glitch.

        • Also, there is a huge difference between reloading a working program, and developing a new one. If you are switching back and forth between three models, nobody in their right mind is going to rewrite it each time.

      • I think this man v machine thing in large part explains Swanson’s Law.
        Below a given volume, man wins as described in the article. Above a given volume, machines win. Double the volume from there and your marginal cost is little more than the cost of materials, and even the cost of materials fall with larger tonnages.
        This has philosophical implications. What works, what is affordable, what is efficient, is whatever we chose to make work as a species.
        We’ve seen it with so many products now, TVs, phones, panels. There can be zero doubt that EVs, batteries and grid storage will become cheap, and they will be made by robots.

      • Yeah… there should be a law prohibiting/restricting these types of unremitting inefficiencies. LOL

    • Thanks Carl. And that video is 2 years old now!

  • I just put “Cleantech Talk” on my phone. Fortunately it is listed as a podcast.

  • As a machinist that worked with industrial robotics for decades, I know for a fact that it does not take weeks to setup the machine for a change of parts. CNC (Computer Numerical Control) equipment is designed to be flexible. Most of the time it will only take a few hours.

  • FWIW: I think there’s a lesson buried in the article – keep the number of car models and options to a minimum and production is greatly simplified. I still remember my 1968 Volvo: Your choices: color, auto vs manual transmission. Maybe the radio was an option. At that time A/C was an after-buy add on. I think Model T only coming in Black was a bit extreme 🙂

    I think a moderately loaded “standard” car allowing a lower price might be greatly appreciated by the public. Differentiate in simple areas like outside color, upholstery color/material.

    I really wonder if anyone has done a tradeoff on whether the whole zillion models with a zillion options actually pays for itself. Or whether it’s one of those “everybody else is doing it” things.

    • With computers doing the communicating, and made to order cars, they can probably get away with a lot, and still be efficient. That said, Tesla does eliminate unpopular options, and Elon is very focussed on “the machine that builds the machine”, and designing the car to be easy to build. They are buying some experience, like that Audi guy, and I assume all new gear for the model 3 lines. It is going to be interesting to watch.

Comments are closed.