COMPANY CONFIDENTIAL. DO NOT DISCLOSE.
Ever a pioneer, Tesla also leads the charge in digital transformation
State-of-the-art manufacturing systems are creating data streams with the potential to make zero-emission electric vehicles more affordable
Challenge: Though looming climate crisis has made it clear the world must transition away from energy consumables, renewable energy technologies remain prohibitively expensive for the vast majority of consumers. In the automotive industry, the high price tag associated with zero-emission electric vehicles is directly linked to the traditionally high cost of developing and manufacturing bleeding-edge innovations.
Solution: Visionaries at Tesla are taking a two-pronged approach to cost reduction: utilizing high-volume digital factories to maximize economies of scale while tightly optimizing processes through advanced analytics. JMP® statistical discovery software provides engineering teams with a single platform for data access, analysis and sharing that removes barriers and complexities from engineering workflows. JMP, says Process Quality Engineer Samuel Troptard, “is a really complete solution.”
Results: By adopting one software for the entire process quality workflow, Tesla has increased engineering efficiency in several key areas. Engineers can now qualify and release new process tools into production faster and with fewer out-of-specification occurrences; the automation of routine tasks makes analyses more reproducible and instills a universal statistical best practice; and the dynamic visualization of data helps domain experts and data scientists to work more collaboratively. All have contributed to cost and time reduction – something Troptard says makes a strong case for JMP and, he adds, aligns well with Tesla’s spirit of accelerated invention.
_____________________________________________________________________________________
With the traditional transportation industry contributing nearly a third of worldwide greenhouse gas emissions, scientists warn that a complete shift to sustainable transport is urgent if we are to forestall climate crisis. At the consumer level, US-based maker of electric vehicles Tesla has pioneered not only the technologies needed to make that shift a reality; it has also put zero-emission cars on the road at a record pace, with Tesla’s Model 3 becoming the first electric car to pass one million in global sales.
To achieve its mission of expediting the transition to sustainable transport, however, Tesla is confronting the reality that the high cost of its products puts them out of reach for most consumers. The company is therefore focused on innovating increasingly affordable electric vehicle technology. “By reducing the cost of [components],” the company announced in a recent press release, “Tesla can make products available to more and more people, allowing us to make the biggest possible impact on transitioning the world to sustainable energy.”
Advanced digital factory systems are a key tool of innovation
High-volume, renewable energy-powered digital factories – and the economies of scale they create – now form a cornerstone of Tesla’s strategy to accelerate sustainable transportation. In announcing the formation of Tesla Advanced Automation Germany in 2016, the company wrote that: “As the machine that builds the machine, our factories are so important that we believe they will ultimately deserve an order of magnitude more attention in engineering than what they produce. At very high production volumes, the factory becomes more of a product than the product itself.”
As Tesla Automation has grown in the pursuant years, this philosophy has become even more deeply embedded in the organization’s culture where analytics goes hand-in-hand with some of the most advanced digital manufacturing systems in the world. As the dedicated manufacturer of automated machinery for Tesla’s microprocessors, memory chips, sensors, steering controllers and lithium-ion batteries, among other things, Tesla Automation is heavily focused on innovating new fit-for-purpose robotics systems.
“In the construction of these kinds of specialty machines, there is always a new challenge,” says Samuel Troptard, Process Quality Engineer at Tesla’s facility in Prüm, Germany. The quality team’s remit, he explains, is therefore to consider the relationship between the design of the technology, the process by which it is manufactured and ultimately, the quality and reliability of its life after production. The site’s advanced digital systems generate robust data streams that Troptard says are essential in the statistical approaches that constitute the team’s workflow. “Statistics isn’t always the complete solution, but it does have a huge impact [on the way our factories operate].”
JMP, an end-to-end statistical workflow tool, cuts data access, preparation and analysis time by two thirds
Cost is an area where the impact of analytics is most deeply felt. Though cost is always a key factor in development, Troptard says, statistical approaches to the start of production enable Tesla’s quality engineers to directly quantify the cost of each problem or out-of-specification arising. It therefore becomes easier to prioritize the order in which problems should be solved.
Of similar importance in both development and production is speed. Statistical approaches can dramatically accelerate innovation cycles, but without the right tools to manage the analytics workflow, Troptard explains, unnecessary complexity tends to offset pace advantages. That’s why, he says, Tesla adopted JMP® statistical discovery software, an end-to-end solution for process and equipment engineering that reduces both cost and development time.
When engineers first brought the tool to Tesla’s management, they demonstrated how JMP could help to accelerate processes at the company’s site in Prüm. “The most [compelling] reason was to save time which, I think, is the best argument for JMP,” Troptard says. “JMP accelerates many things. For example, whereas previously we had to construct a data sheet with formulas and Excel macros before we could even calculate a small standard capability factor, with JMP you can get there in just two clicks.”
By reducing basic analysis time, he says, the tool also frees up engineers to pursue more complex, value-adding problems. “You can spend less time talking about the problem and more time talking about the solution,” he adds, noting that JMP has helped the team to remove barriers and complexities, increasing efficiency without requiring management to increase head count.
“I would say I’m at least three times faster with JMP,” Troptard attests. “And there are things I can do in JMP that were not possible before. For example, there are things that would be really difficult to do in Excel and JMP is 100 times faster because it’s already implemented as a special function.”
Migrating all data processes into a single tool boosts reproducibility and standardizes around best practices
Whether it’s pre-configured platforms that can be used out of the box for methods like design of experiments (DOE) or the dynamic exploratory platform Graph Builder, Troptard says JMP manages data workflows from end to end – access to discovery to decision. JMP provides Tesla with the ability to troubleshoot existing processes and equipment, develop new methods to support the next bleeding-edge technology, and hone processes quickly and economically. “The solution of JMP is really complete,” Troptard says. “When I begin a project with JMP, I can finish the job completely all within JMP because of its different functions… JMP is a really complete solution – it’s my everyday tool.”
By uniting so many tools into one single software, Troptard says that JMP eliminates the need for engineers to license multiple platforms that solve the same problems. “JMP is the right balance between universal software like Excel and dedicated software quality software for process analysis,” he says.
“You may be able to do all kinds of things with Excel, but it’s not fast. And you have to know a lot of tricks to reach your goal. On the other hand, [a process analysis software] may be really specific – but it’s limited in the wide range of things you might need a tool for. Moreover, with JMP, I’ve found I am able to perform analysis much quicker and with less configuration. Variability charts, for example, are really fast. In two clicks, you can read your answer.”
Customization through scripted dashboards has helped to further automate data processes at all stages of the analytics workflow, instilling best practices and speeding repeat analyses. For example, Troptard shows how he scripts table transformations, making the columns into rows and vice versa.
“We can check visually to see if all positions are measured and make sure that the data is plausible. Then I can decide whether we continue the tests on the machine,” he explains. “If we run more products, we may want to recheck the hardware or software. After we define any issues in the software, or if the data is not passing, then we’ll have some discussions. JMP graphics show visually any arisings that are out of specification so that we can react quickly.”
Standardized scripts are yet another way of realizing economies of scale, and at Tesla, scripts for repeat analyses and process monitoring can be quickly rolled out to an entire team of engineers. Those more adept at complex analysis can therefore more easily support domain experts working closer to products and processes. Furthermore, Troptard adds, shared scripts instill best practices and JMP’s in-built tools allow for improved reproducibility.
JMP is a complete digital replacement for traditional prototypes
As JMP has become more deeply embedded in daily workflows, the tool now provides Troptard’s team with process insight on par with the physical prototypes that were previously standard. “Unlike typical quality engineers, we’re not looking at a constant process,” Troptard says. “Each machine is quite unique and therefore our role is to first improve the design and then to develop and verify the process for production.”
Whereas traditionally prototypes provided insight into the critical core of a process and helped to predefine parameters for testing, it was not always easy to simulate the innerworkings of such complex machinery. And prototyping was often expensive. “So our idea, in essence, was to replace the [traditional] prototype with JMP,” Troptard says. “With JMP, we can save power, reduce analysis time, reduce the number of trials and manipulations of the machine, and reach our goal faster. Then we can apply [learnings] directly to the machine and run additional trials to verify. That is also something we can do in JMP. It’s a complete prototype replacement.”
The tool also facilitates data sharing and discussion, whether in the improvement or development phase. And for the end user of the machine, Troptard adds, “it’s also an advantage to be able to share results quickly and carefully and say ‘here is the capability,’ for example, so you can make a decision quickly about the result.”
A culture of analytics enablement boosts employee retention and helps attract high-level engineering talent
Perhaps even more impactful than speed and optimization, Tesla’s embrace of digital transformation and its pioneering approach to standardizing analytics is helping to build the Tesla brand in the eyes of the engineering community at large. Highly automated standard systems have streamlined data processes to the point where engineering teams are freed up from rote data tasks to think creatively and innovate at an unprecedented pace. It’s an exciting opportunity that helps draw in the best and brightest new hires – as well as retain existing talent.
“Tesla is a beloved brand, which certainly makes the job more attractive,” Troptard says. “It’s also good for the company because we are able to attract very high-level engineering talent. We have the opportunity to work with great people – and it is this diversity that makes working for Tesla so good.”
The level of statistical savvy among new hires is high, he says, but “what’s really new is the way we analyze the data.” Domain experts not only work closely with data scientists, but they are also empowered to apply analytics to their own data. “What I find really important is how we make analytics part of the practice of our job… I see analytics as a lever to make improvements.”
“The time saving factor is really a good point for JMP, but more than that – it’s also in line with the Tesla spirit: working at an accelerated pace to keep the advanced lens we already have.” Perhaps the world where every driver has an electric vehicle is not so far off after all.
