Detecting improvement opportunities
NOTES FROM THE GEMBA – A culture in which problems are tackled as soon as they appear and the production and product development teams work closely together is helping this French healthcare technology company to thrive.
Words: Catherine Chabiron, lean coach and member of Institut Lean France.
Photo by: Quentin Reytinas
This is a world of high technology, where the work content can take up to 45 minutes for a given operator to complete before she/he can move on to the next product. The work requires extremely small gestures, a sophisticated understanding of what could go wrong if the operating standard is not adhered to, and a lot of training. Welcome to Trixell, a company that designs, develops and produces X-ray flat panel digital detectors designed for a wide range of medical applications in radiology. Its customers include Thales, Siemens Healthineers, and Philips Healthcare, which are the three shareholders of the Trixell joint-venture.
Much like flat TV screens, this technology is fairly new, which is why these three corporations decided to form a joint venture as three main actors in the healthcare field. They want to mitigate the impact of the initial investment. Only 15 years ago, hospitals had to work with heavy detectors weighing on average 25 kilograms, which used a tube technology and could only be installed in a fixed spot – with patients being moved in and out. Portable flat panel detectors today are much thinner (15 millimeters) and lighter (2.8 kilograms) and they can easily be carried around the hospital. Trixell produced 3,000 detectors in 2007 – in just three variants – whereas their 2018 annual output was above 13,000 detectors, with 50 different products.
I am meeting Jean-Loïc Mourrain, Industrial Director at Trixell, whose function covers both production and process engineering. Product design takes place at the same site, although managed by a different department. “Does this bother you?” I ask Jean-Loïc. I am thinking of the many R&D departments I have come across that are physically located far away from production (and its daily problems) and wonder whether Trixell suffers from the same issue. “Well,” he says, “Not any more. You’ll see why as we progress on the gemba. It’s important to say that innovation is not only product related; it can also stem from the process. At Trixell, we have a very sophisticated process on the scintillator (a key element of our detector) and both production and process engineering are on my team. This makes collaboration easier.”
My plan is to check out Trixell’s Obeyas and see how they manage their continuous improvement efforts. And I am lucky, as the daily meeting with the current shift of operators is about to start in the production Obeya. The team is unusually large, close to 20 people. This is more than I would expect for an efficient stand-up meeting and to sustain kaizen activities, but Olivier, in charge of production, will later confirm the production team is organized by workbench, each with their own Quality Totem on which they work quality issues. The four production teams are, however, sharing the same daily meeting.
JIDOKA, OR “STOP-AND-FIX”
The meeting starts with the review of the daily production and a look at the week’s target, set to the customer’s takt time.
At Trixell (and, indeed, in the entire Thales Radiology branch), Stop-and-Fix is the name given to Jidoka. If an operator spots a quality issue or believes that there might be a quality issue, he/she will trigger an alert (Thales and Trixell also call it “Stop at first doubt”). The production will be stopped. Needless to say, this is key when it comes to such type of technology: pursuing the assembly of costly components when a risk has been spotted would undoubtedly result in components being unusable elsewhere if the product is scrapped, not to mention the precious time spent on an assembly that may later have to be taken apart. Stopping at first doubt is also critical to the very high level of reliability Trixell is after. Once the alert is raised, solutions will be discussed on the spot (containment measures, picking up new components, stopping the production altogether), at the team’s Quality Totem. Once contained, and the customer protected, the problem is tracked on the stop-and-fix board, while design changes, process changes or other solutions are being refined.
Stop-and-Fix is fundamental, but so is right-first-time. Today, six parts have been set aside on a daily production of 54 products (each made up of 70 to 90 parts, depending on the version). This is a major breakthrough: four years ago, a far larger number of parts would have been set aside on a given day. The team tracks the internal repair rate, which was cut by half over the past two years. The teams are still working on it, but remember, when you compare Trixell’s activity to the automotive world where lean started, with one car produced per minute, the takt time in Trixell ranges between 25 and 30 minutes. The work content is adjusted accordingly, and the number of technical gestures required from a single operator far exceeds what you would see on a car assembly workstation.
Support functions also attend the stand-up meeting and the group leader now turns to the maintenance engineer to get an update on faulty equipment and to collect any new information or request from operators. There is a dedicated board for Equipment, which includes self-assessment by operators of the workstations, when they take over from the previous shift. The topic of the day is a screwer still out of service and a cooling malfunction that takes the temperature of the product lower than it is supposed to. An equipment status board is not so common in Obeyas, and it is great to see time spent on what can make or break the day for an operator!
The group leader ends by sharing a piece of information: Siemens Healthineers, both shareholder and customer, has thoroughly thanked Trixell for their reactivity to a recent after-sales issue, which was promptly handled and solved. As people go back to work, I see smiles on their faces.
COLLABORATION ON QUALITY ISSUES
Olivier, the production manager, takes me aside at the end of the meeting. “This room is now going to be used by the production management team – including product quality and maintenance – on a similar stand-up meeting,” he says. And indeed, I see them gathering in front of the stop-and-fix and other boards next to it that were not used by the previous group.
Olivier gives me an explanation before they get started: “We have that meeting twice a day. The point is to discuss how to solve our Stop-and-Fix issues. We have started to plot them on a large ishikawa, and, in the morning, all the key players join in to discuss those – product engineering, maintenance, method, supply chain, production and technical manufacturing support. In the afternoon, the meeting is limited to production management, as a follow-up.”
Every issue spotted at the Stop-and-Fix ends up one way or the other on the ishikawa (4 Ms: Milieu, Machines, Materials, Manpower). The point is to collaborate in front of the board and discuss how to progress on the problem. Experimentation is their motto, as learning comes from trying out options. To fix their potential materials problems, they have tried two ways.
One of the options they had was to design a board displaying the Supplier non-compliance Top 5, with pictures of the elements at fault. The idea was to invite suppliers to come over to one of the daily meetings and explain how they planned to fix the situation. This is a sort of shaming process, however, as no one likes to be in such a Top 5. It might even work if the supplier is not far away, but in today’s world it’s just not very common.
The other option they experimented with was to go and visit the supplier premises and see together what could be done. This achieved far more and brought satisfaction to both sides. The management team has numerous stories to tell, all showing that both parties learned in the process and improved the situation thanks to the gemba walk in the supplier’s facility. Olivier remembers an instance where they were asked by a foam supplier to contribute to the investment of a clean room, but it turned out at the gemba that a thorough cleaning of the workstations was enough to contain the problem. Trixell is extremely demanding on this, and would reject any polluted component with the stop-and-fix, as particles of metal could create smears or blotches on the screens of a flat detector, possibly even malfunctions (the detector is used by many people and moves a lot around the hospital).
The same is true for another metal part. The cutting operation is sub-contracted to a jail – one of the small industrial operations that are handled by inmates. As Trixell visited the workshop, they observed the preparation of the workstations and saw how they were being cleaned with a sponge. They asked whether the workstation could be vacuum-cleaned instead, and when that change was implemented, the rate of non-compliance decreased dramatically.
In the meantime, Jean-Loïc is back. He did not want to attend the operators stand-up meeting to ensure they would feel free to speak openly. We continue the exchange in front of the ishikawa. “What has changed with the Stop-and-Fix approach is that we have Engineering Project Leaders come over here and discuss production and assembly issues with us. Recurring production issues can now be taken into account by engineers before a new product is released, to immediately correct the design and make the life of operators easier.” Jean-Loïc has even brought the newly-appointed R&D Director to the board, at their first meeting (no PowerPoint, straight to the collaboration spot between Production and R&D).
Moreover, as soon as a Stop-and-Fix alert is raised, and if design, process or quality are at stake, system experts are immediately invited to the gemba. “As they do this regularly, they are confronted with the reality of operations and, more often than not, eagerly look for root causes.” (This is far more efficient, I must admit, than exchanging emails on possible solutions while checking data on a Computer Aided Design software.)
Change takes time, and Trixell is no exception. Olivier confirms that it took three years to build mutual trust between operators, supervisors and engineers, and move past the hot-potato game that was being played. Trust is now visible in the production Obeya: the board showing the improvement proposals made by the operators themselves is really dynamic. None of the dates I check are more than two weeks old. Olivier confirms: “As soon as an idea is proposed, we try to help the operator implement it. We give them the time and tools they need. If the idea proves valuable, they will more often than not convince their colleagues of the relevance of the new approach.”
DEEP LEARNING AND VISIBLE KNOWLEDGE
A common difficulty in daily operations is that we tend to initiate a lot of actions but also forget about them once the Do part of Plan Do Check Act (PDCA) is over with. Trixell, however, spends time on the Check, too: after the countermeasures have been implemented (Do), past issues are monitored for a while (Check) in the production Obeya using a dedicated watch board, to ensure the problem is definitely under control and the identified cause is tackled. Jean-Loïc and I have a great conversation on the Act that should then take place: beyond the update of work standards, how do you actually retain the knowledge you have just acquired? Before-and-after pictures with an explanation of why the new approach should be followed? Trade-off curves? Cause trees where the observed effect is shown with its confirmed causes? They are not yet sure of this at Trixell, but they are clearly trying to find out by means of experimentation.
We end the tour in the Management Obeya, where the management team regularly meets (they call it the Compass Room). While daily operations (and associated learning) are managed in production obeyas, the Compass Room is designed for deep learning across the management team, as they anticipate and handle future challenges. There is a wall dedicated to the customers and quality as perceived by them. A space shows projects where conditions for success are expressed on a multi-criteria radar chart: an opportunity to discuss options and share learning.
A3s are also presented and discussed according to a schedule. Nathalie, who is in charge of lean in Trixell, has engaged in a lot of experiments, observation, and cause-and-effect investigation around leveled pull flows. She has learned a lot on the subject, accepting how important it is to have a detailed understanding of complex flows and interactions. Collaborating with production group leaders on lead-time, work content and variability, in a dedicated room rather than through Excel, helped refine the number of kanban cards in circulation, reduce the overdue cards and finally the overall work-in-progress. WIP, with regards to the production output, was practically cut in half in the last three years. Productivity increased by 20% over the same period, while capex went down: not because of reduced ambitions but thanks to an optimised use of the production means.
I am intrigued by a specific board in the management Obeya, called Deep Diving, and I turn to Jean-Loïc to ask him to explain what it is. I remember seeing, in his office, a description of what he had labelled the golden detector. “Are those the things you need to crack for future innovations?” I ask. “In a way, yes,” he tells me. “I write down all the challenges we have today, on which we need to spend time and learn to learn. If we can crack those, and the others that will come our way, we will pave the way for the creation of new competitive advantages in the future.”
In order to ensure efficient day-to-day procedures at hospitals in the most intractable situations, quality and robustness of products are priorities for Trixell. In this way, developing internal competences in order to fully satisfy customer needs is a sure way to sustain growth.