How implementation of AI across an automotive factory floor helped increase productivity while providing a six-figure monthly cost-saving due to reductions in scrap and energy consumption.
Content Summary
Executive Summary
Manufacturers Are Turning to Digitalization and Automation
IIoT as Backbone for the Plant of the Future
Sustainable Autonomous Manufacturing Using AI as a Service
Executive Summary
The pandemic’s impact on global supply chains accelerated the need for manufacturers to reshore their efforts as they worked to reduce international dependency and diversify their supplier base.
To achieve this, many manufacturers began investing in factory digitalization, automation and IoT deployments, innovating their processes and using technologies to improve efficiency, resilience and supply chain visibility. Today they find themselves operating at the edge of their capacities, balancing multiple opposing forces, including heightened stakeholder demand and the need to increase the pace of innovation.
Digital connectivity, IoT and AI-as-a-Service automation technologies are key enablers offering manufacturers the tools they need to not only meet these challenges head-on but also capitalize on the opportunities they present. Together, IoT and AI technologies can help manufacturers optimize plant utilization, mitigate unplanned downtime, reduce wasted resources and time, and improve their environmental impact.
In this article, we look at how implementation of AI across an automotive factory floor helped increase productivity while providing a six-figure monthly cost-saving due to reductions in scrap and energy consumption.
Manufacturers Are Turning to Digitalization and Automation
Companies are using IoT technologies and AI to increase output and reduce costs, even as they grapple with staffing shortages
As the pace of industrial development continues to accelerate, so do business demands and expectations.
Meanwhile, businesses are grappling with mitigating and reversing their impact on the climate, their communities and the planet.
As the engine of industrial progress, manufacturing is at the intersection of these opposing forces. On one hand, there is an exponential acceleration to deliver more, faster. While on the other hand, manufacturers are under pressure to reduce waste and energy consumption, and identify, improve and eliminate harmful practices.
With the pandemic’s role in “the Great Resignation” many sectors, including manufacturing, are finding it harder to retain skilled staff and hire new employees to fill vacant roles.
Today’s manufacturers are increasingly turning to digitalization and automation through the use of Internet of Things (IoT) technologies and artificial intelligence (AI). Both offer many benefits including increased output, reduced human dependency and reduced costs, as well as improvements in energy and resource efficiency.
Despite significant deployments of connectivity, sensor and IoT technologies across both the information technology (IT) and operational technology (OT) domains, many manufacturers are discovering there is still a long way to go to reap the benefits of digital manufacturing – let alone be able to address the increasing pressure from consumers, investors and clients to become more sustainable.
For manufacturers, the urgency to become more sustainable, drive increased efficiency and output, and overcome staffing challenges comes at a time when the global community is struggling with compromised supply chains, dwindling workforces and skyrocketing energy prices.
By using AI and IoT, manufacturers can embrace technology to overcome these challenges and capitalize on the opportunities before them.
The Case for Increased Manufacturing Automation
A European automotive component manufacturer was under increasing pressure from its clients to increase capacity, reduce turnaround times, and retool for frequently changing specifications and supply chains.
It needed to achieve this while mitigating against energy cost increases, improving the company’s environmental footprint and looking after the well-being of its employees while increasing profits.
The plant had already invested in several types of data-capture IoT sensors, including vibration, sound and electrical supply, and the implementation of machine vision technologies for component inspection and analyzing worker behavior on the factory floor.
These sensors captured vast quantities of data, which were manually analyzed and compiled into reports and dashboards to facilitate operational insights and process improvement.
While time-consuming to generate, these insights already reduced costs by minimizing unplanned downtime and increasing output. However, the manual analysis was both timeconsuming and difficult to cost-effectively scale across other equipment, stations and production lines.
To achieve greater operational efficiency and waste reduction across the entire plant, the business looked toward using AI to extract additional value from its IoT investment.
Focusing on reducing both equipment downtime and energy consumption, the manufacturer implemented an AI proof-ofconcept to explore how IoT data could be leveraged further to enhance utilization and eliminate wasted time, materials and energy.
The factory operates hundreds of weld types across the entire factory floor. Weld faults are directly attributed to scrap, energy waste and downtime impact that multiplies across the downstream processes, making welding a critical sub-process with potentially huge impacts. As such, the weld process was an ideal candidate for the AI proof-of-concept.
The proof-of-concept helped uncover trends across all the weld station IoT sensor data and allowed the factory to create dynamic control plans to optimize the plant’s unique process parameters across each of the hundreds of weld types – something that was previously unobtainable using manual analytics processes.
The impact of the proof-of-concept was substantial.
Demonstrating a sustainable reduction of more than half of the welding defects, the implementation saved hundreds of thousands of dollars every month, from the reduction of waste materials alone.
Furthermore, the significant reduction in weld defects reduced downtime and re-work across production lines, increasing production output and reducing energy consumption.
Following that success, the manufacturer now operates AI across other areas of the factory floor. The AI-as-a-Service algorithms now identify hidden trends and correlations between seemingly unconnected data sets gathered from presses, robotics and injection molds, which alert workers to both predictive and prescriptive maintenance scheduling.
With this enhanced operation awareness and confidence in production assets, the factory can not only reliably and consistently operate at the edge of its capability, but also do so while minimizing risk and environmental impact.
IIoT as Backbone for the Plant of the Future
A PAN-OPERATIONAL, SELF-OPTIMIZING ECOSYSTEM
A complete IIoT solution platform will retrieve, store, stream, visualize and compute preprocessed industrial data—bringing it into a unified, accessible environment for stakeholders across the entire manufacturing enterprise value chain. In so doing, it will allow teams, companywide, to act self-consistently on the insights of advanced analytics. They will do so at the interface where informational and end-to-end operational technologies have merged.
“Manufacturing Lighthouses” have already embarked on the journey to this state of future production and realized Data Maturity’s starring role. These excerpts are from the World Economic Forum (WEF) white paper of January 2022 entitled “The Data-Driven Journey Towards Manufacturing Excellence”:
“Companies that master the use of data… have the greatest impact on productivity and customer experiences, as well as society and the environment….” “Operations … become largely selfmanaging, with historical and real-time data being used as input parameters for self-training algorithms, which get “smarter” as their experience grows.”
Its conclusions are based on a Boston Consulting Group (BCG) survey conducted with more than 1,300 manufacturing executives, the Data-Maturity self-assessments of eighteen leading manufacturers from diverse industries, and actual business use cases of data-derived value-creation instances for production operations.
The WEF report confirms what many manufacturers may already intuit:
- Future production is trending towards autonomous manufacturing.
- Data is the foundation of its intelligence.
- Digital Maturity is a prerequisite for mobilizing smart industries.
Because they contain the most potential scope for immediate, high-impact improvement to the Cost of Goods Sold (COGS) without additional CAPEX— production, maintenance, and quality are three crucial (and interconnected) domains perfectly poised to automate data-driven decision making. As stated in the WEF white paper, manufacturers “…must compete on several levels at once to improve their value proposition: ever-increasing quality, reduced prices and digitally enhanced productivity. This triple goal requires continuous improvement in manufacturing efficiency. In addition, manufacturers realize that they need to take the climate imperative into account and accelerate the transition to net-zero.”
In light of this value proposition, let’s assume that an IIoT data platform, as described above, has been established at a manufacturing plant to enhance process optimization and machine health. This IIoT platform would need to be primed to leverage live and historical industrial edge data to be AI-ready. To this end, the data platform would support a wide range of data sources, connecting to machine sensors and hundreds of PLC and SCADA types. A purposebuilt edge device will continuously collect, store, and analyze plant data across ISA 95 levels 1, 2, and 3.
Meanwhile, ideally, a data ingest stack will capture and pre-process factory data up to millisecond resolution and organize each source into a human-interpretable hierarchy of data streams.
With these foundations in place, data scientists can build a unified view of production and unleash the deep learning insights of advanced analytics—to optimize production processes, apply prescriptive maintenance, or both.
ALIGNING TEAMS AROUND NEXT-LEVEL PRODUCTION
As for optimizing a production line, an AIenabled system can create a dependable feedback loop to push the limits of normal operation and discover deeper efficiencies in any advanced process. Data-driven discovery (from the edge of the production site’s best operating regime) accounts for critical tolerances. It can then safely seek new optimal regions—where improved sustainability measures such as plant efficiency will mutually reinforce production KPI benchmarking such as throughput and quality. This feedback loop has the added benefit of ensuring production and quality teams align around a centrally agreed means of optimizing the line for continual improvement. AI prescriptions directly feed back to the control plan and create a completely autonomous plant.
TARGETING DEEPER SUSTAINABILITY EFFICIENCIES
By capturing data and sustaining an autonomous feedback loop, we can start perceiving production problems differently. Datadriven discovery allows us to reimage the way we look at a plant’s control plan without exposing the production regime to any risk. For example, after optimizing for a given objective, we might still find ourselves in an operating region that contains inherent bias. Recognizing this bias, we know there is room to evolve the process further.
The insights of AI modeling can then be applied to discover alternative operating regions for further optimization. Sustainability objectives can be worked into such a frameshift. Why? Sustainability isn’t far removed from any conventional production objective—running more efficiently, reducing scrap, or increasing the operating life of a line can all be pegged to sustainability goals.
The trick is to attain these sustainability efficiency goals securely. AI-as-a-Service manages this security by embedding a feedback loop into the IIoT platform for a fully connected plant, which (as mentioned) also factors in critical tolerances. Gradually moving the production line’s manifold to find better (in this case, more globally efficient) states happens as the AI incrementally prescribes novel operating regions tied to environmental metrics.
These new regions are a symbiosis of where the plant safely and efficiently worked before, together with variations that reinvent the manifold of previous operations—both self-consistently with the manufacturing process and in a way that targets increasingly sustainable production.
An AI-guided plant leverages data to eliminate human bias and discovers operating paradigms beyond the reach of expert human capacity. The functionality of the AI-ready platform is embedded with automated pipelines through which advanced deep learning prescriptions transmit to process engineers and plant managers.
These prescriptions manifest on an HMI as easily actionable parameter adjustments.
With human-interpretable cues, operators are empowered to affect process optimization or machine health interventions ahead of production loss. Only an AI-ready IIoT platform will safeguard the evolution towards sustainable, autonomous production.
Sustainable Autonomous Manufacturing Using AI as a Service
Minimizing expense and risk while facilitating the integration of AI capabilities across the factory floor
Like most investments, AI is a technology that needs to find its justification.
The hype around AI can lead many businesses to speculatively invest, however, without a sound business case or proven return on investment, AI can be a distraction rather than a game-changer.
Industry 4.0 helped manufacturers transform their operations allowing them to continue to embrace the need to integrate IoT sensors and devices to capture the performance and health of production assets. However, many are finding it difficult to extract relevant, timely insights from this data.
With AI as a Service, manufacturers can selectively identify already digitized production assets for specific technology evaluations to build and develop positive ROI business cases.
This approach minimizes capital expenditure and operational risk while facilitating targeted and appropriate incremental integration of AI capabilities across the factory floor.
Humanistic AI in Action
AI often gets a bad rap, with detractors citing a future where artificial intelligence systems and robots will take our jobs.
The history of industrialization, technology advancement and the search for artificial intelligence systems is littered with examples of job obsoletion.
AI should be considered just another tool that can help humans achieve something greater. This augmentation of human intelligence with artificial intelligence is known as humanistic AI and is the process of deriving a result greater than the best each can achieve in isolation.
While highly trained engineers and data scientists can find correlations and patterns across three to six dimensions of data. AI systems can work with significantly more data sets, allowing them to identify trends across more complex systems, significantly faster than even a team of experienced engineers.
However, these AI systems can also depend heavily on human understanding. To gain accurate insights, these systems must be trained on known, labeled data. As such, during implementation, real-world experience from trained engineers is vital in the success of the AI system moving forward.
But this is just the beginning of the human- AI partnership.
The benefits of predictive analytics include the ability to schedule maintenance on production machinery in a controlled fashion rather than having unexpected interruptions to the production line. This in itself is valuable for operators and the business as a whole. Beyond this, prescriptive maintenance AI systems preemptively alert workers of any maintenance requirements and can suggest the required course of action.
Most AI systems stop at this point — at the point of suggestion. It is then up to the experienced operators and engineers to take those suggestions, verify them and implement them.
Together these humanistic AI systems can save time and money while improving safety and operational efficiency, freeing up valuable human capability and engineering resources to be redeployed into other areas.
Human Operational Resilience
The COVID pandemic caused many manufacturers to take unprecedented actions. Two years after the initial outbreak some of these actions are only still beginning to be felt.
From the moment factories and businesses shut down for quarantine, a lack of staff forced businesses into a new paradigm of uncertainty.
This global phenomenon, along with other developments such as workforce distribution and decentralization, is partly to blame for the difficulty in retaining and attracting staff.
Building more intelligent, autonomous business practices and systems through the use of AI can help prepare for these lessunforeseen future disruptive events.
Balancing Profit, People and the Planet
During the last couple of years, the combined effect of the climate crisis and the pandemic have escalated awareness and action toward sustainability.
While historically the reduction of scrap might be seen as a financially motivated endeavor from a sustainability point of view, waste reduction has a much greater impact. Looking at it through the sustainability lens helps highlight the positive impact it can have on the environment, communities and biodiversity.
Similarly, when implemented with integrity, the mitigation of energy costs should be seen as an opportunity to not only manage price risk but also have a positive sustainability impact.
The United Nations Sustainable Development Goals (UN SDGs) demonstrate the interconnected criticality of people, planet and profit, and is one of several global sustainability frameworks rapidly growing in awareness and adoption.
The implementation of AI in autonomous manufacturing should be seen as an opportunity to balance the perceived conflict among profit, people and planet.
UN SDG 12, Ensure Sustainable Consumption and Production Patterns, is focused on ensuring sustainable consumption and production patterns and is broken down into 11 specific targets. Many of these targets put an emphasis on waste reduction and the elimination of pollutants.
Specifically, two targets aim to achieve sustainable management and efficient use of natural resources, and substantially reduce waste generation through prevention, reduction, recycling and reuse, respectively.
These are targets which AI-enabled autonomous manufacturing techniques can have a genuine and meaningful impact.
Due to the interwoven nature of the SDGs, an impact on SDG 12 will influence impacts aligned with other goals. Reducing waste through smarter consumption of raw materials will likely also reduce energy consumption and environmental pollutants, thereby having secondary impacts that could align with both human health and environmental focused goals.
