Clean Energy Smart Manufacturing Innovation Institute

Jim Davis, Vice Provost – Information Technology, UCLA and Co-founder Smart Manufacturing Leadership Coalition

Jim Davis, Vice Provost – Information Technology, UCLA and Co-founder Smart Manufacturing Leadership Coalition
Jim Davis, Vice Provost – Information Technology, UCLA and Co-founder Smart Manufacturing Leadership Coalition

Smart Manufacturing and other data-oriented advanced manufacturing initiatives – IIoT (Industrial Internet of Things), Industry 4.0, Manufacturing 4.0 and Digital Manufacturing – have generally converged on business opportunity with performance, productivity, and growth in customized precision products in an increasingly dynamic demand trade space. There is considerable associated discussion about business disruption and the disruptive roles of data and Information Technology (IT). In fact, it is hard to find any discussion that does not have something about data and disruption – the disruptive enterprise, the platform effect, big data, digital transformation, machine learning, AI, cloud, edge, block chain, and cyber-security, to name a few. Noticeably missing is what enables the data to be disruptive.

From a business standpoint, there is strong economic impetus toward precision products that are purpose-built, safe and more safely manufactured, environmentally neutral, and manufactured with far less energy and material throughout the value chain. The Clean Energy Smart Manufacturing Innovation Institute (CESMII) has shown that the economics of improved precision, performance, and productivity with energy savings alone are sufficient to stimulate growth and new markets for small, medium, and large manufacturer value chains.

Achieving full business potential involves the business and technology disruption of restructuring highly compartmentalized manufacturing line operations and supply chain enterprises so that data can be used securely where and when it is needed. Greater product precision is achieved locally, often in smaller lots, but with extended upstream and downstream interoperability and global orchestration. There is additional disruptive opportunity with virtual enterprise businesses that can more rapidly form and maximize the use of the physical facilities relative to market opportunity. Data are at the center. None of these business changes are aimed at driving further cost reductions or automation to reduce staff. Disruption is about new and changing markets, revenue generation, productivity, and market growth by using data and IT to reorient and redistribute manufacturing intelligence and innovation.

No manufacturer, though, is going to invest in a smart operation all at one time let alone disrupt an existing operation. When it comes to stitching data together in our current environments and with current products, the industry is managing to maximum complexity, cost, and time, not simplicity. The technology is out of reach for most. Cyber security and business trust methodologies are mostly designed to block interaction, and the difficulties in reaching and executing agreements around data present sizable barriers. Technologies, along with critical business, organizational, and workforce changes, need to be implementable in small trusted steps by small, medium, and large companies alike. There needs to be immediate and predictable benefit at a small fraction of the cost and time it takes today. This is a journey of significant change in accelerated small steps.

So what are we really talking about with Smart Manufacturing and data disruption? Consider a simple line operation comprised of three primary manufacturing assets: a CAD/ CAM front end driving a 3D metal parts printer followed by a milling machine*. These are already Operational Technology (OT) rich with measurement, data, and controller systems. The business objective is to develop, qualify, and produce small volumes of specialized metal parts substantially faster, at lower cost, with zero defects and with less energy and material usage than with a traditional casting process. There are multiple material suppliers upstream and multiple OEMs downstream forming intertwined value and supply chains.

What is ‘smart’ is to enhance the OT (the IT to function) to radically push precision and performance capability. An acoustic sensor combined with machine learning can be used as a sensor system that ‘listens’ for quality and defects while the part is being milled. A high speed camera system with image analysis can be used as a machine vision system to analyze the part both on the printer and the milling machine, and a nanostructure surface sensor system can be used to do a precise analysis of the surface microstructure in situ. Machine listening, viewing, and measuring the integrity of the part can then be fused into a modeling system that does real-time, insitu qualification, optimizes the design and updates the CAD/ CAM model that drives the 3D printer. During production, it is smart to validate the input materials and the design data, monitor for variations and anomalies, and control and optimize the printer and milling machine taking into account the condition and maintenance of the two together.

The IT requirements are significant in that huge amounts of millisecond acoustic, image and microstructure data from different vendor products need to be networked, interconnected, streamed, ingested, contextualized, stored, managed, and provisioned for different software analytics and models. The models, often from different vendors, and the data need to be orchestrated and executed with time-assurance in a secure cyberphysical system-of-systems framework. The overall ‘cyber-physical system’ can self-interrogate to assure and/ or diagnose OT and IT operations that can affect each other locally as well as upstream and downstream. A smart worker looks for broader anomalies, ensures data for learning or tuning of models, analyzes the operational data for insights and improvements, and looks for new sensor and modeling technologies. Hardware, software, data, and model configurations, and the OT and IT infrastructure are reusable.

* This hybrid 3D printer line operation is a public demonstration of a 
Smart Manufacturing Cyber Physical System built by the Clean Energy Smart 
Manufacturing Innovation Institute (CESMII) Gulf Coast Regional Manufacturing 
Center at Texas A & M University and the CESMII SM Platform team and its commercial

Putting this all together, big data, data transformation, cloud, and machine learning are all well and good but they really won’t have much impact unless some hard questions are answered about enabling integrated OT and IT practices and the interconnected trust and cyber-security environments in which they will be deployed:

Good Data – There is so much talk about big data but the primary issue is Good Data. Machine learning, AI, actuation, and decision-making depend primarily on good data and data that have been contextualized for objectives.

Agnostic Interconnectivity – Unless there is a better, faster, and lower cost way to interconnect networks, software products, infrastructures and companies, the vision of smart manufacturing is unreachable.

DevOps – The old way of designing to long-horizon specifications is not viable. There is a need to work with new insights from the data, new vendor products and configurations of these products to increase capability and benefit in small steps without having to rebuild infrastructure. Staging and moving a new data/software application or even a security update into production during production is needed to accelerate benefit.

Reusability – Easier, less costly, and faster reuse of operational data and software/hardware as previously developed configurations is needed to move out of building oneoff systems. This also clears a path to reuse by search, not synthesis, to significantly lower complexity and barriers to access.

Smart Worker – Machines and operations need to do what they do best and the workforce needs to do what it does best. Workforce and advanced sensors, controls, platforms, and models need to interoperate so the capabilities of humans and machines are both extended.

Enterprise Security, Trust and Situational Resilience  – The ability to validate, secure, trust, know state, and have the necessary time assurances of all of the OT and IT functions at any point in time throughout the enterprise is critical. Localized diagnostics are important but an enterprise approach needs that broader kind of resilience with situational response and the ability to predict and self-interrogate.

The era of data opportunity and data disruption in manufacturing has kicked off and Big Data, AI and machine learning, and networked, edge and cloud applications and products are proliferating. These are insufficient, though. Enabling technology, infrastructure, workforce, and business practices throughout the industry need to be developed and matched to an accelerating stepwise journey. Viability of needed enterprise approaches extend beyond any one company so public-private partnerships are important. CIOs, CTOs, and operational technologists together need to maintain a line of sight on BOTH the new data methodologies and how they are enabled.