An introduction to the digital transformation of lean manufacturing
The top goal of most digital transformation (DX) initiatives sounds suspiciously similar to that of lean-manufacturing methodologies — to minimize an operation’s waste and cost while concurrently maximizing its quality (defect-free) throughput of customer-coveted product. Benefiting a vast array of industries outside manufacturing, today lean-manufacturing approaches are applied in most all settings that allow collected data to inform decisions or designs. Those approaches that meld DX technologies to achieve lean objectives are called lean digital or (less commonly) digital lean.
These so-called lean digital approaches are advancing and permanently transforming modern automation — buttressing organizations against the market effects of supply-chain issues and natural disasters as well as the effects of the recent pandemic and global inflation of raw-goods prices. Lean-digital adaptability has also helped operations adjust to skilled-labor shortages.
Complementing lean-management approaches such as Six Sigma, lean digital also addresses the same waste issues that traditional approaches target — including overproduction of out-of-favor product; excessive lead times caused by internal mis-dedication of resources; production of defective end products; overly complicated routes of materials or workpieces through a production facility; and misguided stocking of certain items in inventory. Many such issues are addressed with lean-digital realtime data reporting on and control over quantities, times, processes, standards, and corrections of current operational conditions.
Production, preparation, process lean approaches (3P lean approaches) emphasize non-siloed cross-team collaboration using simple 2D or 3D physical mockups of an automated setting (for example, made of notecards on a pinboard or cardboard blocks on a table) to design new processes. In contrast, lean digital 3P approaches connect either existing equipment in a facility already in operation or future facility’s digital twin to data-aggregating IIoT systems. Then digital 3P teams access these models (and data analytics they support) to conceptualize and consider improved iterations of operations with fewer bottlenecks and issues than current or proposed designs. Such digital 3P approaches allow remote workers to join in the collaboration to optimize parameters or machine sections with which they’re most familiar.
Of course, such new designs should include consideration of personnel who will ultimately drive the operation. Automated installations have always been complex, but recent decades have seen rapid proliferation of operational variables to produce or package a given set of offerings. COVID-19 only hastened this trend … and rendered efficient employee onboarding and training tools all the more important for companies aiming to embrace the continuous-improvement of DX while allowing new and longtime personnel to participate in ever-evolving business approaches.
Kanban lean approaches are predefined communication systems signaling to plant personnel when work should be diverted to replenish a hardware, subcomponent, or raw-material bin. In traditional operations, it’s the job of supply-adjacent personnel to flag those bins for refilling. In contrast, lean digital kanban approaches employ bin-monitoring sensors (often employing RFID technology) to automatically and immediately detect and signal when a bin needs refilling. Complementary software technologies such as machine learning and digital twins can simulate the typical drawdowns and refills a bin sees to further optimize its stock redeliveries.
Total productive maintenance or TPM lean approaches support equipment health by periodically triggering service calls before failures (large or small) occur. The scheduling of such checkups and tune-ups depends on machine runtime, cycle intensities, and various environmental factors. Lean digital approaches on the other hand employ controls as well as force, temperature, and vibration sensors along with machine-learning software to inform predictive-maintenance routines that are at once more effective and less demanding on maintenance personnel. Case in point: Some industrial PCs (IPCs) and ruggedized automation controllers provide realtime data logging and processing — with the latter via software including statistical process control, multiple regression analysis, AI functions (including similar-waveform recognition), and Mahalanobis-Taguchi system multivariate data diagnosis and prediction technology. Such analytics software can send realtime feedback to the factory floor to allow for the combination of IIoT data with seasoned plant-personnel expertise where applicable to yield comprehensive predictive-maintenance programs. Again, the IIoT data can be supplied in realtime to immediately drive decisions.
Heijunka lean approaches (much like just-in-time or JIT approaches) aim to smooth out production cycles by strategically scheduling and right-sizing disparate products’ production runs. Products being alternated appear on a Heijunka calendar only occasionally altered. In contrast, lean digital Heijunka leverages software analysis of previous runs’ historical data to calculate the most suitable run size and order for a given machine and resource availability status. In this way, lean digital Heijunka helps operations make the most of today’s exceptionally adaptable automated machinery — as well as provide maximum support to plant personnel executing sophisticated manual tasks.
Software and other DX design support
As just illustrated, software that unifies various aspects of automation development is key to lean digital and DX efforts. Such software typically provides one universal engineering environment or more commonly an integrated development environment (IDE) to let engineers connect machinery with IT and other facility architectures in a familiar software window. In fact, IDEs can also help minimize downtime — the primary enemy of most manufacturing, packaging, and processing.
Consider how traditionally executed machine changeovers degrade throughput. As lean digital concepts see increased adoption, changeovers (to satisfy consumer demands for greater product customization and personalization) are more frequent and easier to execute. In practice, this is where IDEs really shine — in making automated installations both modular and scalable. That’s illustrated by some suppliers’ software structures that render machine modules capable of coupling automatically and self-coordinating when being joined or rejoined with alternative production-line arrangements.
Another strength of standardized IDEs is how they can facilitate the integration of more exotic automation equipment. Consider linear-motor conveyors — a relatively new type of intelligent transport system or linear transport system aimed at replacing inflexible mechanical-heavy arrangements necessitating complicated changeover routines (if they allow changeovers at all). Programming for some such systems is relatively simple in part because suppliers allow the systems’ setup within the same software and Ethernet-based network as the controls for less exotic motion systems. That simplifies the adoption of these flexible automation options, so machine builders and engineers designing for captive plant use can employ them where appropriate without trepidation.
For engineers lacking the time to learn whole new programming platforms, many automation suppliers’ software contains automation, industry, and motion-control libraries (including point to point, NC, gearing, camming, flying saw, and collision-avoidance routines for the latter) all in one ecosystem. Many integrate with standard IDEs such as Microsoft Visual Studio as well as programming via IEC 61131-3 languages and computer-science standards. Otherwise, industrial controllers might accept established IEC 61131-3 programming methodology with the option of adding newer code better suited to IIoT applications … or accommodate programming languages having both established and newer presences in automation applications … including PLCopen, C/C++, Node-Red, Python, Blockly, Java, and G-Code. All this software standardization imparts benefits that are synergistic with multi-function hardware formats such as the gateways, HMIs, controllers, and motors that don’t rigidly tie set machine functions to specific components.
Consumer-grade computing informs digital-lean approaches
Consider the astonishing advancement of microcontrollers over the last 20 years. The Raspberry Pi and similar offerings along with smartphones and tablets are now capable of astounding computing power for a low price — and have found direct use in (and revolutionized approaches to) computing for automation. With hardware now so capable, the dominant value of today’s consumer products is its software.
With the adoption lag typical of manufacturing especially, industrial computing is finally seeing the same trend. After all, industrial-automation processors today can reside almost anywhere — in standalone housings, inside a drive, or on a motor. To illustrate: For many applications, PLCs are no longer pieces of hardware but simply one of a system’s many software modules. Paramount is what runs on a system’s processor. So on some new platforms designed to support digital lean, engineers can run on hardware and scale via software to simplify the maintenance of parts and replacements. Applying functions is as simple as downloading the necessary software, just like on a smartphone.