Tesla Model Y efficiency exceeds early-production Model 3
teslarati.comReally interesting topic. For those interested I highly recommend the Tesla Model Y teardown videos from Sandy Munro of Munro & Associates. He is a really great engineer who used to be one of the Heads of Manufacturing at Ford among lots of other positions. His company does pretty much everything, from tearing down vehicles to, redesigning products for manufacturing, to building whole factories. He worked on the EV1 as well.
In his videos he takes apart a Model Y step by step and they analyses what it would cost to make with very detailed models:
https://www.youtube.com/playlist?list=PLkiDlGyJnprdFftxAZ85a...
In terms of efficiency, one of the highlights of the Model Y is the heatpump. Having a Heatpump itself is not new, other EV had it, but not Teslas. Tesla internally designed and built a new Heatpump that improves efficiency quite a bit, and it is credited with making Model Y get almost Model 3 range even while being bigger. Elon even said he had one of those heatpumps in his bedroom because its one of the best things any of his companies have ever done. Sandy Munro confirms that he has never seen that in all the heatpumps they have seen.
There are lots of good information on pretty much everything including electronics as well, its well worth watching.
>And just like the Model 3s, Model Ys that were produced after June 2020 exhibited a significant improvement in efficiency
Won't the MPGe values be higher for these vehicles because they are being driven exclusively during summer months so far, when batteries are less efficient in the winter months?
As measured in MPGe. I had to look it up.
Miles per gallon gasoline equivalent https://en.wikipedia.org/wiki/Miles_per_gallon_gasoline_equi...
It's probably a fine metric.
Would a total cost per mile (projected over lifetime) would be more comprehensive?
even miles/km per kwh input should work fine enough
If the efficiency gains vary by production year, that implies it's hardware improvements, as opposed to software tweaks (which would be rolled out fleet-wide regardless of year). I wonder what goes into that experience curve effect?
Tesla doesn’t change hardware by build year. They make hardware changes every day/week/month.
Per your reply, from TFA:
"Tesla has a habit of rolling out improvements to its vehicles as soon as they are available. The latest Teslas are therefore expected to have the best tech that the company has to offer at the time of their production."
Realizing that Tesla's slipstreaming improvements, completely ignoring model years, was the 3rd or 4th "Oh, this is different." aha moment I've had. And why I think their ridiculous valuation might be semi-rational.
Another aha, per Conways' Law, was learning that Tesla didn't honor the traditional organization structure. Permitting them to create the octovalve and omni heater. (Sorry, I keep forgetting the names given for those parts.)
Right now I can't imagine any other automobile manufacturer shaking their ant farms the same way, allowing organizational blank slate resets. Such is the benefit of no legacy.
>They make hardware changes every day/week/month.
How are they sourcing parts and integrating them into the manufacturing process "every day"? This seems like an SV understanding of manufacturing; Tesla may be more agile than the average auto OEM, but there is still a complex procurement prcoess that requires lead time, as well as spare parts for fleet repair and maintenance.
Do you have an example of a piece of hardware they are swapping every day, week or month? Maybe this is why they have a hard time servicing their cars.
> How are they sourcing parts and integrating them into the manufacturing process "every day
They know by VIN exactly what parts go in what car.
That said, the coding equivalent is that they make interfaces - and swap out interfaces. You can make a LOT of changes at the implementation level, that do not change an interface.
> Do you have an example of a piece of hardware they are swapping every day, week or month?
Sure, they are swapping the entire body caste machine from ~20 parts down to 1..
https://www.foundrymag.com/molds-cores/media-gallery/2193204...
went live this week I believe, while they are actively cranking out thousands of model Ys per day.
Another example - the octovalve - a very special decide that lets heat and cold get shared 8 ways as needed between motors / battery / and cabin is a huge part of the model Y. According to a teardown
https://www.thestreet.com/tesla/news/munro-tesla-model-y-oct...
the valve has had 13 changes in the 6 months the car has been in production. 13! In a single part that is the size of a small backpack.
> Maybe this is why they have a hard time servicing their cars.
This has not been my experience at all. I think QC is a little low on brand new cars during rush season, but service has been top notch. Make sure you aren't relying on stories from 2016.
Really? How do they support all those hardware configurations? Is there a policy framework they use to manage software development do it works well with all those configurations?
Keep the interface the same, but change the implementation.
Same way you can easily refactor code.
For an example see https://www.thestreet.com/tesla/news/munro-tesla-model-y-oct...
Comparative data across the Tesla fleet, by model and production years, is so interesting!
I love the Tesla approach to continuous improvement, and the Nikola app seems really cool!
This is David. I made Nikola. Glad you like the app! Happy to answer any questions you or anyone else has on the data / research here.