theleanmag — #18 — february 2024 LEAN AND DIGITIZATION: PUTTING PEOPLE AT THE CENTER by Jeff Liker, Jeremy Frank, Joe Li, and Yahya Kahn Walk into a Toyota plant and the first graphs. The performance of each line is vividly overwhelming image is paper on walls and displayed on huge computer screens updated easels everywhere. Color coded charts, automatically in real time and the numbers are graphs, figures with places for data seem to fill remarkable—making what was planned for the every nook and cranny. Andon lights are firing shift on time, without vehicles pulled off line off constantly playing music and team leaders for quality problems, 96% or more of the time. come running to respond. The most common Even old robots that many companies would or biggest problems of each shift are worked replace are running over 99% of the time. on by teams at the front line using disciplined What is missing, or at least very limited, are the problem solving. computer displays that have become popular in the digitization of the factories. Toyota believes strongly in the power of visual For automated processes, skilled maintenance management to highlight gaps between team members perform detailed preventative standard and actual and people’s brains to maintenance for one full shift everyday, also close those gaps. In the Toyota plant in the guided by visual reminders on charts and 32 theleanmag — #18 — february 2024 United Kingdom they operate with the motto: humans and our endless capacity for creative “Use brains not money.” thinking and innovative problem solving. As we will see, Toyota is indeed cautiously stepping into the world of the Internet of The lean movement views the Toyota Things. Production System as a benchmark for best practice in manufacturing and even today large companies typically have departments The Nature of Problems and Problem of continuous improvement experts roaming Solving: Predictive versus Adaptive the plants, leading improvement workshops, We see too often lean advocates and IT and preaching the virtues of lean. In the people talking past each other about the meantime, we have another movement power and usefulness of advanced digital sweeping manufacturing in the 21st century— technologies. The IT people claim their Industry 4.0. This is a movement to digitize technology offers “solutions” that: the factory promising a revolution that will make yesterday’s manual methods obsolete, · Predict where problems are likely to occur. and companies that do not get on board also · Immediately flag problems that have just obsolete. occurred. · Closed-loop control systems automatically Why does Toyota persist in using legacy manual methods with paper? Why isn’t this make adjustments to equipment to fix the manufacturing leader leading the way with problem. · In cases where people are involved digitization? And can these movements work together—lean and digitization? provide detailed guidance, perhaps through augmented reality, to walk even the novice through a recipe to fix the We think the answer is yes they can, but it problems. requires stepping away from the orthodoxy of · Provide real-time data on performance lean and the bold, if sometimes arrogant, promises of “solutions” provided by the digital relative to targets so gaps can be quickly factory. It requires understanding the reality of and efficiently closed. the types of problems that millions face in typical days on the shop floor and how those Meanwhile, lean advocates caution that the problems can best be solved. It requires real world is far more complex than that and being realistic about the unlikely visions of people, with our immense problem-solving dark factories with most people gone, leaving capability, still need to be in the center: only engineers who develop the automated systems. Most important, it requires valuing 33 theleanmag — #18 — february 2024 theleanmag—15—May 2023 · In the highly variable and uncertain world of manufacturing novel problems occur These seemingly conflicting world views are regularly, many of which have no known actually focusing on two different problem recipe. solving scenarios—predictive versus adaptive. Predictive problem solving are cases where · The automation that is supposed to fix the machine itself breaks down frequently the problem can be easily diagnosed and the and needs to be fixed. solutions that will work predicted in advance of actually taking action. Often these solutions · The more complex the technology the are based on engineering knowledge more likely it will break down in expressed in mathematical models. unpredictable ways. Essentially through prior knowledge the · People with our unparalleled brain people or system are trying to predict the capacity and finely tuned senses can future. predict that a problem will occur and find Adaptive problem solving are cases where the creative ways to solve the problems problem is vague, the solution is not clear, outperforming even the most advanced AI and it takes creative thinking and trials systems. organized as experiments to learn our way to · Often simple, inexpensive solutions can the solution. For these types of problems be very effective and save a lot of money. intimate knowledge of the conditions of · People are visual creatures and to be failure, and past experience with similar most effective need simple displays of problems, can be a great asset. People on actual versus standard highlighting what the front-line who live with the equipment is green, yellow and red at a glance. and processes have this knowledge. Predic ve Problem Solving Adap ve Problem Solving Problem defini on Clear Requires Learning Solu on Well understood Requires Learning Problem Solvers Experts or Authority Stakeholders Process Efficient Decision Making Act Experimentally Timeline Results ASAP Expecta ons Fix the Problem A tude Confidence and Skill Time Needed to Follow Process Achieve Targets and Develop People Curiosity Figure 1: Comparing Predictive and Adaptive Problem Solving 34 theleanmag — #18 — february 2024 · If used effectively by well-trained people it Table 1, adapted from O’Malley and Cebula, contrasts these two problem solving can lead to large increases in equipment scenarios. If we labeled the “Predictive performance very quickly. Problem Solving” column as the Digital · In many, maybe most cases, the insights advocates view of the world and the from the analysis require people skilled in “Adaptive Problem Solving” column as the adaptive problem solving to find the lean view of the world we would have a solutions that will work. pretty good fit. The Toyota Production System for Who is right? Is it possible both are? There Equipment Health are many problems that can be solved Equipment maintenance appears to be right predictively by those with expert knowledge, in the wheelhouse of the Industry 4.0 or from computer algorithms, though far advocates. Industry 4.0 forecasts a world fewer than the digital thinkers suggest in their where repetitive manual processes are at pitches for digitalization, and also many some point fully replaced by automated problems that require pure adaptation and equipment. In a sense this is an engineering learning by stakeholders experimenting, paradise where the more predictable and though fewer than the lean advocates consistent programmable automation takes suggest. over for less consistent human operators who have personal needs. The catch is that most Let’s consider a case example of a class of of maintenance is done by those technologies that use sensors to measure unpredictable humans who need deep vibration, temperature, current, load, and time knowledge, high levels of technical and and have the following characteristics: problem-solving skill, and creativity to identify the cause of a breakdown and figure out how · There is well established engineering to do the repair. science on how different pieces of equipment function optimally that can use these indicators to predict, detect and The Toyota Production System is centered on sometimes even diagnose problems. people and usually applied to repetitive manual work, like that on an auto assembly · It is not so much AI as well understood line. Just-in-Time is about working toward a engineering equations. high value-added flow of product to the · When coupled with sensors and the customer, without waste in the process. In Internet of Things, the analysis provides the plant this requires getting the right information that is not always intuitive to materials at the right time in the right amount most operating personnel in factories and to the team members who do the work. the insights are often surprising. 35 theleanmag — #18 — february 2024 theleanmag—15—May 2023 “[digitization] requires valuing humans and our endless capacity for creative thinking and innovative problem solving. members pulled the “andon” cord when out of standard condition were noticed which could lead to line stops. To have a consistent flow of the product and its constituent parts you need stable processes which requires Standardization. There are standards for product design, standards for process design, and standardized work for people performing manual tasks. The reason People are in the center is they not only perform the work, but are expected to continuously improve the system, including reacting to out of standard conditions with creative problem solving. The ownership and responsibility is at the front-line and work groups comprised of team members, team leaders, and group leaders are responsible for achieving aggressive performance targets and perform some of the simpler preventative maintenance.1 Since there is always variation in various parts So how does this apply to equipment of the system, including the human, things maintenance? The goal is to have healthy will invariably be out of standard so corrective equipment that is ready and able to produce action is needed. Jidoka, the second pillar, exactly the parts needed when they are addresses out of standard conditions. It needed with perfect quality. began with Sakichi Toyoda who founded the original loom works that led to Toyota Motor Just-in-time (JIT) maintenance means the Company. He ultimately invented a fully right maintenance at the right time, neither automated loom. Along the way he added too early or too late. Preventative physical sensors that stopped the loom when maintenance is critical to Toyota’s high level a single thread broke and signaled there was of operational availability of equipment, but a problem by a flag popping up. Later this led one could argue that, without the right data, it to identifying potential failure modes and is often performed too early. That is, a broad designing into equipment sensors to detect set of maintenance procedures are done out of standard conditions and stop the whether the equipment needs it or not to run equipment when certain thresholds were at a high level Just-In-Case it might otherwise reached. For manual processes team lead to equipment breakdown. 36 theleanmag — #18 — february 2024 Another approach to maintenance is to wait Many of the equations that they use to predict for it to fail, like the original sensor on the equipment failure can be traced back loom when a thread broke. This reactive decades to fundamental engineering science. When there is an analytical equation that approach requires quick response to first effectively predicts it is not necessary to have contain the problem and get the equipment AI systems scouring big data to find patterns. running and then solve the problem at the You simply plug the right numbers into the root cause so it is unlikely to happen again. right mathematical model. Toyota is one of Arguably this is too late. their clients. Toyota is increasingly moving toward the world of predictive maintenance. They wish to know in advance for a particular machine that Toyota Plant Example it was going to fail in a specific way and take Toyota has had great success with KCF corrective action before the problem occurs. technology for both predictive and adaptive There are advanced methods, through the problems. “Internet of Things” and analytics for example, As a predictive example, sensors were that are increasing what we can predict. installed on the drive motors of conveyors Unfortunately, any digital system today can and analysis of vibration patterns was able to only measure certain variables, there are predict impending failure. When the limits to the analytical models, and because vibrations exceed thresholds, the software most factory systems are complex, unique triggers a text and email. Team leaders and and interconnected, there are a limited set of members follow up immediately and problems that can be managed purely on a generally determine they need to schedule predictive basis. Therefore, we still need replacement of the motor by maintenance. people to find the root cause and develop a As the plant expanded it use, this type of creative solution to solve the problem. Often predictive maintenance action played out solving the problem is a matter of trying hundreds of times over thousands of pieces things, experimenting, that is, adaptive of critical plant machinery, and built problem solving. confidence in the system with the team members. Intelligent Sensors Case Examples Going beyond predictive maintenance, the KCF Technologies is one technology supplier team members also realized many cases where the predictive data enabled them to that believes in the power of digital make adaptive improvements to system information to enhance, not replace creative configurations and standard operating problem solving by people. Their core procedures. One example of this was in the technology is advanced sensors, with the paint shop. The KCF data consistently computer power of advanced analytics. 37 theleanmag — #18 — february 2024 showed imbalance on the paint shop exhaust to figure out the best way to extend the fans, related to build-up of paint on the fan press’s operational life by an additional two blades from the process. years. This data enabled the team members to The team deployed a series of “IoT HUBS”, assess and update the frequency and which are specifically designed to capture methods for proactively cleaning the fan and process data on complex, intermittent blades, thus eliminating the imbalance machinery like stamping presses. The sensors problem that would have otherwise led to were installed on a press line to test its premature failure of the fan support bearings efficacy and data indicated severe and the motor and structure. The solution was deterioration in several key components of not obvious immediately and took a more in- the press, though the solutions were not depth problem solving process with some obvious. Through a process of going to see, trial and error. identifying several key causes of failure, and There are even examples of a combination of testing various ideas the team came up with predictive and adaptive. An example of this an alternative for the hydraulic brake pump system—which was so out of date that was at a Toyota facility that stamps out steel replacement parts were no longer available— parts. This was an older plant with aging that resolved a number of the issues. The equipment. Toward the end of 2022, they had technology also was also able to anticipate a debilitating equipment failure on a sub- the need to replace the main motor bearings, system of the stamping press. This resulted which predictably failed 3 months later. in significant hours of downtime, substantial stress on the team members, and But they found even this level of problem considerable capital expenditures. The team solving was not enough. The slide adjust was caught off guard by unplanned failures assembly required a more intensive problem and consequently began a collaboration with solving approach that built upon the KCF in April 2023 to manage the health of foundational current and vibration data their assets more proactively. supplied by the sensors. The team noticed a change in the waveform that looked like it could be a rotor bar fault in the making and Simply adjusting their already robust shortly after observed a spike in vibration on maintenance practices would not be sufficient the gearbox. As is often the case, there was to address the particular problems that come not much downtime to inspect the issue, and with maintaining older equipment. Toyota’s the slide adjust was hard to access. A cross- team determined they needed a combination functional team identified possible checks of advanced technology and technical and ranked them from easiest to hardest, expertise. They formed an interdisciplinary creating a feasible and optimized proactive team of manufacturing engineers, corporate maintenance plan, supported by remote and site maintenance staff, and KCF’s analysts 38 theleanmag — #18 — february 2024 continuous online monitoring until the solving. Their experience also suggests that planned downtime and scheduled inspection over time, as the low-hanging fruit is picked, several months later. an increasing number of the problems require adaptive problem solving. Conclusion We recommend that manufacturing The lean movement started with its model, companies that have not invested heavily in the Toyota Production System. At the core of the problem-solving capabilities of their front- lean are people at the gemba, where the line staff should do so in parallel with work is done, who respond to abnormalities introducing advanced digital technologies. highlighted by the system. This may mean going slower in introducing the technologies, including piloting individual Industry 4.0 is a vision for automated factories modules before widespread deployment. run largely by computers that plan and Some companies are partnering lean decide what to do and when. Presumably practitioners with internal IT experts with problems are solved by the computer great results, much as we saw in the systems, often with the help of advanced AI collaboration between Toyota plant experts and analytical systems. and KCF technology experts. We suspect that over time automation and There is a different type of problem solving, digital technologies will lead to fewer people adaptive, in which the problem cannot be in the factories doing manual work. But there easily defined and solutions are not obvious. will be an even greater need for thinking Adaptive problem solving requires people at people at the gemba solving problems as the gemba who with their senses can find the they are identified by software and people. In root cause and solve the problem, often this way the technology supports the people through a series of experiments. and process. n We have given examples with one type of technology, offered by KCF Technologies, of sensors and systems based on engineering equations that can predict when equipment will fail and often provide solutions of the predictive variety. “Often” is not always. KCF looked at data from customers and estimated References: that roughly 50 percent of problems were 1.Jeffrey K. Liker, The Toyota Way, second predictive and 50 required adaptive problem edition, McGraw Hill, 2020. 39