Skip to main content

Digitalization in EV battery cell manufacturing – The state of play

By 18. October 2022No Comments

We’ve wrote about battery production on this blog before, but I want to dive deeper into this industry for multiple reasons:

  1. A lot of so-called Giga-factories are being built right now
  2. Cell production hasn’t been “figured out” yet
  3. It is a very important piece of the green mobility future

In this post, I want to give a summary of our current understanding of the industry and where it is headed – especially regarding the digitalization aspect of manufacturing operations.

The basics

To kick this off, some basic information for any reader who is not familiar with this industry:

These are battery cells (they also come in rectangular shapes):

These are packaged together to form a battery pack (which includes battery management, housing and cooling usually):

These battery packs will then be assembled into the final product – a car, truck, storage unit etc. No matter in which end-product the cells end up in, the early manufacturing stages for cell production are usually always the same (Tesla is working on a dried electrode process, if they can get that working it would be a possible game change – see also here: #2 Tesla’s Secret Sauce // Maxwell Dry Battery Electrode by “The Limiting Factor”)

The big players

Now, who are the companies that produce these cells? The distribution of globally production volume based on a McKinsey report from 2020 is as follows:

So while activity is picking up heavily in the US and the EU to catch up with the China, the current market is dominated by:

I added the flags to show just how far SEA and especially China hast left the “Western World” behind. It seems like the Car manufacturers of this world have woken up though –  at least there is a multitude of production facilities in various stages, both in the EU and in the USA:

With all these factories springing into action the next few years, what about the level of digitalization in this industry?

The state of digitalization

BCG published an article ( – being released in 2018 it’s a bit dated) that estimated the savings potential of “smart factory concepts” to be about 20% for battery production. So it seems to make sense to invest into digitalizing the manufacturing process at the right places. In order to lay some groundwork, a team of researches from the Technical University Munich summarized the parameters that a manufacturer would need to capture just for the electrode manufacturing part ( – leaving out the slitting/cutting, cell assembly and filling step).

They come out at 120 different parameters that influence the resulting cell!

To name a few of them from the Coating step:

  1. Width
  2. Porosity
  3. Coating density
  4. Morphology of the electrode
  5. Moisture content

As one can see, the physical nature of these parameters can be very different. But in order to make sense of them, you need to capture them (at least the most important ones)! To do that, there is a multitude of vendors out there, offering their hardware and software solutions.
To name a few:

  1. ISRA Vision – In-line visual inspection of surfaces (hard- and software)
  2. MeSys – Various surface/depth/weight measurement solutions
  3. Inficon – Gas/humidity/pressure analytics
  4. Manz – whole production lines, but also weighing and electrical measurement (among others)
  5. Hamamatsu – X-ray equipment and more
  6. Mitsubishi Electric – “Line scan bars”-Hardware for high quality image acquisition
  7. Balluff – Many sensor-types like web edge detectors, positioning sensors etc.

Once all relevant sensors are in place, the next questions becomes one of automatically generating the most amount of value out of the data they generate.

Based on our current understanding, this is the phase in which most of the industries out there are currently in the middle of. A lot of data is being generated, but between the multitude of sensors & specific spot solutions that solve a singular problem and the MES/ERP-System, there is gap – a gap of capabilities & solutions in the middle layer. 

A recent visualization of a research paper shows that gap by displaying no active layer between Machine and MES and storing data first before any AI comes into play (the IPA does not extend to the process itself):

But right there between Machine and MES is a prime spot for AI to generate value and enable a more automated production process – but we are only on the way there.

Two points to underscore that statement: 

  1. The German Ministry BMBF poured 150m EUR into cell production research just this year
  2. The DBE process that Tesla is working on has issues related to engineering but also process stability which could be alleviated by AI monitoring

To get to the vision of a fully digitalized and smart factory, one step is to feed multiple data sources & types into a single system that looks for dependencies & root causes in real-time.

That is what Synsor is working on.

Our system is able to intake visual and non-visual/machine data through various interfaces (like REST, OPC-UA, MQTT, CSV, ftp-servers etc.), train our patent-pending AI on the allowed state of the product and process and then detect anomalies and potential dependencies within the incoming data streams.
As a result: we can warn in real-time about deviations from the norm and prevent defects, shutdowns, reworks or breakages. 

If you sell production machines or produce yourself and are interested – let us know!

Leave a Reply