Breaking the Bottleneck | Issue 41
[2/19/2024] 4D Printing, Battery Plant Scrap Rates, Robot Headwinds, Quilter, & More!
Breaking the Bottleneck is a weekly manufacturing technology newsletter with perspectives, interviews, news, funding announcements, manufacturing market maps, and a startup database!
🗺️Check out these market maps for new discrete/process manufacturing tech!
💥 Seen any interesting startups recently? I’d love to chat with them!
🏭 If you were forwarded this and found it interesting, please sign up!
Content I Enjoyed Last Week 🏭🗞️🔬 📚
News
America Wanted a Homegrown Solar Industry. China Is Building a Lot of It. [WSJ]
For years, the U.S. imposed increasing barriers to Chinese solar panel imports to protect domestic suppliers. However, following the IRA, which introduced substantial production subsidies, major Chinese solar companies are now establishing or expanding their panel factories in the U.S., with locations ranging from Ohio to Texas. These companies account for nearly a quarter of the roughly 80 gigawatts of new solar panel capacity announced since the legislation, potentially benefiting from up to $1.4 billion annually in government subsidies. These new U.S.-based factories are significant in scale, with at least four backed by Chinese manufacturers expected to come online this year. They could supply over half of the 33 gigawatts of panels installed in the U.S. last year. Prominent Chinese companies, such as Longi Green Energy Technology and Trina Solar, are leading this expansion. Trina Solar, in particular, has announced a $200 million investment in a Dallas-area factory capable of producing five gigawatts of solar panels annually. The move of Chinese solar manufacturers to the U.S. presents both opportunities and challenges. While it boosts local economies and supports U.S. clean-energy goals, it also raises concerns about the U.S. becoming dependent on China for clean energy.
Battery Plant Scrap Rates Can Hit 90% At Ramp Up [AutoWeek]
There are significant challenges and costs associated with launching large-scale, automated electric vehicle (EV) battery plants, specifically high scrap rates during the pilot production phase. Dr. Tal Sholklapper, co-founder of Voltaiq, a startup that uses analytics to improve battery production yields, highlights that initial production yields can be extremely low, with only 10% to 20% of production being usable. This results not only in wastage but also in the need to manage hazardous waste, which can be a significant cost for companies. The journey to profitability in this sector can take three to four years due to these inefficiencies. Reducing scrap rates by even one percentage point can mean tens of millions in added profit annually. For instance, the Tesla/Panasonic Gigafactory in Nevada had a scrap rate of 84% in the third quarter of 2017, soon after production started. Tesla redirected its best engineering talent to improve these rates, and by 2019, yields at the Gigafactory had improved to 75%.
4D Printing: Researchers Unlock Shape Memory Resins [Engineering]
4D printing is a term used to describe shape memory objects. These objects can change form after production, typically triggered by heat or energy. While many shape memory objects exist, they are usually made using conventional methods due to the high cost of creating specialized resins. However, researchers have developed a more cost-effective shape-memory resin. This advancement involves mixing liquid crystals into a vat of resin and toluene at 70°C. After thorough mixing, the resin is left overnight to allow the toluene to evaporate, resulting in a mixture of pure resin and liquid crystals. These researchers successfully used this resin in a DLP resin 3D printer to create various lattices and smart structures. These 3D-printed structures demonstrated predictable distortion when heated and returned to their original shapes upon cooling resulting in behavior consistent over hundreds of cycles. The potential applications of these shape-changing structures include using them as temperature and strain sensors, where the structure bends to engage an electrical contact upon heating.
BatMan’s Laser-patterning Process Enhances Battery Capabilities [NREL]
The U.S. Department of Energy's National Renewable Energy Laboratory (NREL) has made a significant breakthrough in battery manufacturing through the BatMan project, focusing on enhancing electric vehicle (EV) batteries. The project, short for Battery Advanced Technology Manufacturing And Novelty, introduced a novel laser patterning process to improve the microstructure of battery electrode materials. The team's innovation lies in using lasers to pattern electrode surfaces, enhancing porosity, conductivity, and wettability. The project leverages a high-throughput laser patterning process, integrated with roll-to-roll manufacturing techniques. This method allows for rapid and precise modification of electrode structures, improving battery capabilities at minimal additional manufacturing cost. It also utilizes computational simulations, advanced characterization, and laboratory-scale prototyping to refine the laser ablation technique.
Heavy Machinery Meets AI [HBR]
The manufacturing industry is undergoing a transformational shift with the increasing integration of advanced software into products. This approach, known as fusion strategy, differs significantly from traditional IoT strategy and is categorized into four types: fusion products, fusion services, fusion systems, and fusion solutions. These categories encompass a range of approaches from leveraging data from single products to integrating multiple systems for customer outcomes. Companies like Tesla and Rolls-Royce are examples of early adopters, using fusion strategies to create innovative products and services. John Deere has also adopted such an approach with See & Spray, a revolutionary fully self-driving tractor and weed killer. The See & Spray features a 120-foot carbon-fiber boom equipped with 36 cameras that can scan 2,100 square feet per second. It utilizes 10 onboard vision-processing units that handle nearly four gigabytes of data per second. This system, powered by AI and deep learning, distinguishes crops from weeds and sprays the latter, moving through fields at 12 miles per hour. Furthermore, Deere collects extensive data from its modern farm equipment, including the See & Spray. This data gathered from about 500,000 machines over 325 million acres, is analyzed in Deere’s JDLink system. This cloud-enabled system generates immediate and future enhancements for both equipment and farms to drive optimal management of seeds, fertilizers, and weeds.
Fictiv Launches Materials.AI [Fictiv]
Robot Sales Plummeted 30% in 2023 [Engineering.com]
In 2023, North American robot sales experienced a significant decline of 30% following two years of record orders, as reported by the Association for Advancing Automation (A3). The year saw 31,159 robots sold, a notable decrease from 44,196 in 2022 and 39,708 in 2021. The distribution of these sales was almost even between the automotive sector (15,723) and non-automotive industries (15,436). This represented a 34% drop in sales to automotive OEMs and suppliers and a 25% decrease in other industries compared to 2022. Despite the overall decline, certain non-automotive industries still showed strong demand for robots in 2023. The metal industry led this demand, followed by sectors like semiconductor & electronics/photonics, food & consumer goods, life sciences, pharmaceutical and biomedical, and plastics & rubber.
Research:
Xometry's 2024 Automotive Manufacturing Industry Survey [Xometry]
Xometry and Thomasnet conducted a benchmark study forecasting 2024 automotive manufacturing trends. Below are some highlights:
Amid the backdrop of the 2023 United Auto Workers strike, two-thirds (67%) of those surveyed believe that their company will increase worker wages within the next 12 months.
More than three-quarters (79%) are likely to reshore products or raw materials suppliers within the next year, but the cost is the biggest barrier to sourcing automotive supplies domestically.
An overwhelming majority (84%) indicate that it’s difficult to bridge the gap between their current status of electric vehicle innovation and where the Biden administration wants them to be in 2024
Artificial intelligence (AI) is an untapped technology in the automotive sector, with close to half (48%) of respondents not currently implementing AI in automotive production
Generative Expressive Robot Behaviors Using Large Language Models
A new study by researchers at the University of Toronto, Google DeepMind, and Hoku Labs proposes an innovative solution for creating expressive behaviors in mobile robots. The technique, named GenEM (Generative Expressive Motion), leverages large language models (LLMs) to understand environmental context and enable robots to mimic human-like expressive behaviors. GenEM uses the vast social context available in LLMs to dynamically generate expressive behavior without the need for extensive training or rule creation. It allows robots to perform actions like nodding or making eye contact, adapting to human feedback and different robot types. GenEM operates by processing natural language instructions through a sequence of LLM agents. Each agent has a distinct role in interpreting the social context and translating desired behaviors into API calls for the robot. The process begins with an instruction in natural language, followed by the LLM using chain-of-thought reasoning to describe a human response in a given situation. Another LLM agent then translates this human motion into a procedure for the robot based on its capabilities. Finally, an agent maps this procedure to executable code for the robot.
Automotive R&D Transformation: Optimizing GenAI’s Value [McKinsey]
The transition to electric vehicle (EV) technology, the rise of software-defined vehicles, and the emergence of generative AI (gen AI) are reshaping the automotive industry. A workshop with European automotive and manufacturing executives focused on the impact of gen AI revealed that 75% of companies are experimenting with gen AI, with investments reaching up to €5 million for over 40% of respondents. However, most gen AI pilot programs are limited to specific stages of the R&D process, indicating a lack of systematic integration. The potential of gen AI in R&D is substantial, with expectations of improving processes by 10-20%. Use cases include requirements engineering, software testing, product design, and optimization. Gen AI has also demonstrated efficiency gains in documentation and compliance tasks, as well as in the design segment. Despite the opportunities, the adoption of gen AI faces significant barriers, chiefly organizational and cultural changes. A value-centered approach, framing gen AI as an enabler, and addressing legal and ethical considerations are crucial for successful integration.
Chart of the Week:
Podcasts:
The Shifting Manufacturing Landscape [Fabricator Podcast]
Advanced Precision Machining w/ Sarah & Chris Doyle [Capacity from Fulcrum]
Manufacturing Deals🏭💵
Daedalus - A german company deploying software to control and optimize the “shop floor” producing components for the medical device, aerospace, defense, and semiconductor industries
$21 million [Series A] - Led by NGP Capital and joined by Khosla Ventures and Addition
Aizon - A company building a GMP compliant AI platform for pharmaceutical manufacturers
$20 million [Series C] - Led by New Vale Capital and joined by Atlantic Bridge, Crosslink Capital, and Uncork Capital
Quilter - A company using reinforcement learning to automate circuit board design
$10 million [Series A] - Led by Benchmark
Note: Microsoft and OpenAI are reportedly in talks to invest as much as $500 million in humanoid robotics startup Figure AI raising Figure’s value to $1.9 billion.
Planned Downtime 🏭🧑🔧
Deadpool vs Wolverine
The Art of Watchmaking at Nomos