Three trends in manufacturing innovation for 2024

Different manufacturing customers are in different stages of their digital transformation journeys and therefore have different preferences on how they want to approach the transformation process. To kick-start their digital transformations, manufacturing customers are looking to extract insights from their data and deliver business outcomes, which is now simpler than ever due to technologies such as data lakes, the internet of things (IoT), machine learning (ML), and artificial intelligence (AI). Moreover, technologies such as generative AI, digital twins, and digital engineering have recently taken the manufacturing industry by storm.

In this blog, we highlight three trends in manufacturing innovation in 2024: generative AI, digital twins, and digital engineering and how using these technologies in the cloud are helping drive innovation for manufacturing customers.

1/Generative AI in manufacturing

Precedence Research estimates that by 2032, $68.4 billion will be spent on AI in the manufacturing industry. This is an increase in investment of roughly 33.5 percent over the next 9 years. This increased investment is influenced by the emerging adoption of generative AI technologies across businesses. Generative AI allows manufacturing companies to operate more efficiently, leading to quicker product development and assembly, longer equipment lifetimes, reduced machine downtime, and improved effectiveness of operational equipment. Three common generative AI applications within manufacturing include the summarization of machine and corporate manuals, natural language product search, and machine error resolution.

Manual summarization

Using techniques such as retrieval augmented generation (RAG), companies are able to provide large language models (LLMs) access to their internal documentation and machine manuals, helping workers more quickly find information about corporate policy, machine configurations, and steps to solve malfunctions—all resulting in increased productivity. Additionally, RAG provides the ability to use internal information to create documents that previously were created manually. Employees are able to ask questions across a range of contexts. For instance, the query “the air pressure gauge of the welding machine is broken—how can I fix this?” returns step-by-step instructions derived from the welding machine’s manual. Enter the request “generate a bill of materials (BOM) for the following 20 products,” and have a BOM generated with the price for each product pulled from an internal database. Say “I have an upcoming vacation—what is our vacation policy for PTO?” and receive an answer from corporate documents. This is often the first use case that companies undertake, because it can be set up in less than 1 week and has an immediate impact on employees. An example architecture for building this solution can be found under the Text Use Case section at the Generative AI Application Builder on AWS solution on the AWS Solutions Library, vetted solutions and guidance for business and technical use cases.

Natural language product search

Natural language product search gives a company’s employees and customers the ability to search information about the company’s products, resulting in a better customer experience and the ability to find the right product on the first try. Questions such as “I have a 2010 Dodge Charger—what parts would increase its horsepower with the lowest installation effort and are currently in stock?”, “I have X refrigerator—what compartment models would fit inside it?”, and “I need a circuit breaker for electrical panelboard X. What do we currently have in our inventory stock?”, can now be quickly answered. This increased access to information reduces the percentage of products returned and tends to increase product ratings because customers are receiving the best products for their individual needs. With information stored in different locations, you’re able to use tools and libraries such as LangChain to query information from different sources. A detailed solution can be found in the AWS Machine Learning Blog post, “Reinventing the data experience: Use generative AI and modern data architecture to unlock insights.”

Error remediation

Manufacturing companies that use sensors to monitor their machines have recently started using generative AI to more quickly remediate errors and reduce downtime. When sensors predict or detect errors with a machine, messages containing the error can be sent to LLMs that have access to machine manuals, which can generate the steps needed to fix or prevent the error and send this information to an onsite engineer for remediation. Reduced downtime helps all manufacturing processes continue as expected and means that customers aren’t affected by delays in product development. The error remediation solution would make use of an architecture similar to that used by the manual summarization solution but would also incorporate AWS IoT Core, a service for simply and securely connecting devices to the cloud and related services.

2/Streamlining product development with digital engineering

Modern product design requires sophisticated data storage, compute, and collaboration. Many manufacturing companies are looking to use digital engineering to help reduce costs for research and design, accelerate time to market, optimize production efficiency, meet sustainability objectives, and create new revenue streams. The four core areas in which AWS customers are using digital engineering are computer-aided engineering (CAE), product data and lifecycle management (PDLM), computer-aided design (CAD), and electronic design automation (EDA).

Computer-aided engineering and validation

CAE is the use of computer software to simulate performance for the purpose of improving product designs or assisting in the resolution of engineering problems for a wide range of industries. CAE processes include simulation, validation, and optimization of products, processes, and manufacturing tools. Customers in the manufacturing industry are faced with increasingly complex design and verification challenges. Specific applications used in manufacturing and design include computational fluid dynamics (CFD), the finite element method (FEM), electromagnetic and thermal simulation, design optimization/generative design, and impact/multi-physics/multi-body simulations, among others.

To help solve these issues and support their use cases, many customers scale compute in the AWS Cloud with the help of AWS services such as AWS Batch, which offers batch processing, ML model training, and analysis at any scale, along with AWS ParallelCluster, an open-source cluster management tool that makes it simple to deploy and manage high performance computing (HPC) clusters on AWS. For example, Amgen, a leading manufacturer and distributor of pharmaceuticals, saw a significant surge in consumer and data volume through these solutions. Justin Porth, principal data engineer (senior manager of information systems) at Amgen, and Venki Anantharam, director of information systems at Amgen, stated that “the advances made for Amgen’s HPC capabilities as a part of this initiative are fully aligned with the IS City Plan and architectural standards. They not only benefit the digital integration and predictive technologies (DIPT) and final product technologies (FPT) organizations, but the design, methodology, and experience will help accelerate [next-generation sequencing].” To learn more about the AWS solutions for CAE, check out the Scale-Out Computing on AWS solution on the AWS Solutions Library.

Product data and lifecycle management

PDLM helps manufacturing customers reduce the cost, time, and risk associated with product development. Product lifecycle management (PLM) solutions on AWS can help businesses meet availability and compliance requirements while also achieving increased PLM performance and eliminating data silos. Manufacturing customers often face three major challenges in this area: performance and scalability, data friction, and global collaboration. Customers who use AWS solutions to solve these challenges can expect to see the following benefits: improved time to market with less unanticipated iteration; unlocked engineering time for innovation and validation; and improved quality with reduced rework. Madhavi Isanaka, CIO of Rivian, stated that “partnering with AWS lets Rivian focus on sustainable product data and lifecycle management and delivery—not on IT. And with AWS, we are running our key development applications faster than on premises, including 56 percent faster on Elements, 35 percent faster on Siemens, and 20 percent faster on Ansys.” To start using AWS services to help streamline PDLM, check out the PLM solutions on the AWS Solutions Library.

Computer-aided design (CAD)

Computer-aided design is the use of computers to aid in the creation, modification, analysis, or optimization of a design. Manufacturing customers are often faced with increasingly complex designs and challenges associated with collaboration, exacerbated by the fact that remote work has become a permanent solution for R&D workers and research and development teams. Simultaneously, industrial customers face pressure to get products to market faster and to decrease costs while increasing design reusability.

To help their research and design teams reduce the cost, time, and risk associated with the introduction of new products, many customers use existing AWS managed services of both application and desktop streaming types, such as Amazon AppStream 2.0, secure, reliable, and scalable access to applications from any location, and Amazon WorkSpaces, all-inclusive, fully persistent virtual desktops for every worker. John Rousseau, vice president of technical operations at Onshape, said that “Onshape has used AWS to provide our CAD service since 2013. In all that time, we have never questioned our choice of AWS. The global reliability of the AWS service allows my team to focus on improving the Onshape customer experience. The security infrastructure and tools in AWS substantially help us secure our customers’ data.” For more information, check out solutions for engineering design applications and desktops on the AWS Solutions Library.

Electronic design automation (EDA)

EDA is a category of software tools for designing electronic systems such as integrated circuits and printed circuit boards. Increasingly complex chips and system-on-chip (SoC) products are driving the need for greater processing power, memory, and storage, making HPC a general necessity. EDA deployed on the AWS Cloud is helping address these challenges across the entire semiconductor supply chain. Philippe Moyer, vice president of design enablement at Arm, said that “using AWS, our EDA workload characterization turnaround time was reduced from a few months to a few weeks.” To learn more about EDA solutions on AWS, check out solutions for hi-tech electronics and semiconductor on the AWS Solutions Library.

3/Optimizing performance with digital twins

Another way in which manufacturing customers have been innovating is through the use of digital twins to model manufacturing environments. Investment in digital twins continues to increase. According to a report by MarketsandMarkets, the market size of digital twins is projected to grow at a CAGR (compound annual growth rate) of 61.3 percent from 2023 to 2028 and is expected eventually to be worth $110.1 billion. This reflects the increasing adoption of digital twins across various industries, including manufacturing, as organizations recognize the potential for improved operational efficiency, reduced downtime, and enhanced maintenance practices.

A digital twin digitally replicates a real-world asset in a virtual environment, including its functionality, features, and behavior. This is accomplished by using data collected from smart sensors in near real time to simulate the behavior of the object and monitor operations. Digital twins can replicate individual pieces of machinery, production lines, or end-to-end manufacturing environments. Users can interact with and update these environments in near real time. Digitally modeling an object in near real time opens up opportunities to extract business value. These opportunities include predictive capabilities, fine-tuning performance, remote monitoring and tuning, and increasing efficiency.

Manufacturing companies are investing in digital twins to drive innovation and enhance their operational capabilities. For instance, Siemens has committed significant amounts to developing digital twin technology, recognizing its potential to revolutionize manufacturing processes. By creating a digital replica of a physical asset, Siemens can simulate and optimize that asset’s performance, leading to improved quality and efficiency. For example, General Electric (GE) uses digital twins for wind turbines, an instance in which the technology delivers predictive maintenance capabilities and near real-time performance monitoring. Information such as wind speed, electricity output, temperature, component stress, and more is continuously collected and can be monitored from virtually anywhere. This helps the relevant teams consider a variety of scenarios, including power output based on wind conditions, and empowers engineers in the field to operate turbines more efficiently. It also facilitates tracking the productivity of turbines and fleets, as well as performing predictive maintenance. Notable investments such as those by Siemens and GE highlight the growing importance of digital twins as a strategic tool for driving operational excellence in the manufacturing industry.

Optimization

Continuous data provides near real-time insights pertaining to tools, inventory, and data that is relevant to performance and efficiency. With the most up-to-date data on hand, users can more quickly make decisions that help optimize performance. This also means that issues can be detected and acted upon earlier, reducing the downtime of equipment and production lines. Moreover, having access to up-to-date data helps businesses better assess risk and make better informed operational decisions. For example, as noted above, GE uses digital twins for wind turbines, and this facilitates the collection and analysis of data from physical sensors on AWS, which in turn means that turbine operations can be analyzed in near real time to increase efficiency.

Digital twins can also be used to conduct what-if scenarios, and to model changes in output, efficiency, or KPIs relative to your business outcomes. GE uses digital twins to consider a variety of what-if scenarios, such as how wind will affect electrical output, to help engineers operate turbines more efficiently.

For example, in the automotive industry of today, the development of new cars takes place mostly in virtual environments, and many companies use the digital twins of their prototypes to refine designs. A twin usually encompasses an entire car—its software, mechanics, electrics, and physical behavior. This modeling helps simulate and validate each step of the car’s development to identify problems and possible failures before producing real parts. For instance, physical behavior can be mimicked using a car’s 3D data to optimize material behavior, airflow, or heat buildup.

In applying this method, companies save both time and money by reducing the number of physical prototypes. In addition, it empowers individuals from different disciplines to work simultaneously on the same project, simplifying the configuration of different product versions. Finally, by selecting the necessary equipment and designing production cells digitally, it becomes possible to simulate how all the parts come together in a production cell.

Predictive maintenance

Digital twins have revolutionized predictive maintenance in the manufacturing industry by offering near real-time insights into equipment performance, facilitating proactive maintenance that prevents costly downtime. By creating virtual replicas of physical assets, companies can monitor and analyze their operations, identifying potential issues before they escalate and scheduling maintenance at optimal times to minimize disruptions. This predictive maintenance approach not only extends the life span of machinery but also reduces overall maintenance costs, enhancing operational efficiency and productivity.

Remote monitoring

Digital twins facilitate remote monitoring, allowing manufacturers to oversee their assets from virtually anywhere in the world. Through near real-time data collection and analysis, companies can assess equipment performance, identify inefficiencies, and make data-driven decisions to optimize production processes. Additionally, remote monitoring gives teams the tools to respond swiftly to issues, reducing the need for onsite inspections and enhancing the overall safety of manufacturing facilities. As a result, there’s been a significant increase in the adoption of digital twins for predictive maintenance and remote monitoring, with companies recognizing the substantial cost savings and operational benefits of the technology.

Chemicals manufacturer INVISTA uses AWS IoT TwinMaker, a service that makes it simpler to create digital twins, to remotely monitor its manufacturing environment, among other use cases. Jerry Grunewald, vice president of operations transformation at INVISTA, said that “INVISTA is using AWS IoT TwinMaker to help our field personnel efficiently address operational notifications and alerts from plant floors across multiple, distributed locations. With AWS IoT TwinMaker, we can quickly and easily build a digital twin of our manufacturing operations to give field workers a consolidated view of all asset and operational data. By doing so, INVISTA operations is making significant progress toward our vision of the connected worker as an outcome of our transformation effort. For example, a field technician could pinpoint the source of equipment anomalies and identify appropriate corrective action.” AWS IoT TwinMaker empowers you to create virtual representations of any physical environment with existing data sources, ultimately helping customers take advantage of the benefits of digital twins.

Conclusion

Manufacturing innovation is accelerating due to advances in technologies like generative AI, digital twins, and digital engineering. Companies are seeing benefits such as faster and more efficient product development, increased operational efficiency, and reduced costs and risk. While investment and adoption are continuing to grow in these areas, we will continue to see manufacturing companies innovate further with these technologies to transform their operations. Overall, the manufacturing industry is being revolutionized by cloud-based innovations that are driving productivity, sustainability, and competitive advantage.

If you want to learn more about how AWS helps manufacturing customers innovate, explore solutions for manufacturing and industrial on the AWS Solutions Library. You can also reach out to your AWS account team if you are interested in implementing any of these trends.