Who are you?
Laurent Truscello: I'm Laurent Truscello, Head of Products and Innovation at Carl Berger-Levrault. All business sectors have started to instrument and track their assets in software such as the one we offer. Two years ago, we joined a larger group, in particular to support our international development.
Are we still talking about maintenance today, or are we going way beyond CMMS?
Laurent Truscello: It's a little old school to talk about CMMS - computer-aided maintenance management - or EAM. Asset Management (asset management in the broadest sense). There are a number of different terminologies used to describe traceability, knowledge of this equipment, and how to optimize it within the process. As I said in my introduction, companies have been equipping themselves for more than thirty years: the largest, then the smallest in size, have seen themselves equipped to control their assets; and then, as a result, maximize their production.
To answer the question, are we still talking about CMMS? Today, we're going to talk about CMMS 4.0 to keep up with the Industry 4.0 trend. Perhaps we're talking more generally about a digital equipment platform. We're going to start integrating a whole range of other notions. What's going to change is that the management body is still necessary, and it's wanted, graphically, in these technologies - the web, mobility. So some things have evolved, while still being called "maintenance management". In other words, we bring information to technicians in the field, and in exchange, we obviously share this information via mobility or other technologies.
But where I'd say the factor evolves over the years is from two angles. The first is management, which we've already talked about. The second is everything to do with the digital representation of equipment. In other words, just as the parts we manufacture are represented in 3D, the equipment in the production chain is also modeled by the manufacturers. And the idea is not only to understand and trace the activity that takes place on this equipment - its knowledge and the activity that takes place on it - but also to understand where it takes place on this equipment, to better understand equipment that is increasingly complex, where maintenance work needs to be carried out. You need this valve, this conveyor cart, this motor, this robot - because there's a lot of this kind of equipment in the sectors we've just been talking about. As equipment becomes more and more complex, we also need to have a representation to help us understand it better, and to understand where failures may occur, so we can respond to them more effectively.
You mentioned process control in the cosmetics industry. You immediately replied "knowledge of the equipment". Can't a process be mastered without equipment, without detailed knowledge of what it takes to produce it?
Laurent Truscello: Generally speaking, there are several aspects to production. There's the raw material, the input to the process, which will be transformed. You have the means. And there are two main types of means at this point: human means, the people who are there to control and supply; and then the equipment, which will itself fulfil a manufacturing, packaging, control or other function. And so, unless we're really in a sewing manufacturing business, for example - and even then, there would be equipment - today, it's clear that it's an important part of the process, and one that's communicating more and more. It's no longer simply a static element that produces something, it's an element that's going to talk, that's going to exchange much more with its human or machine environment.
So here again, your raw material is data?
Laurent Truscello: Exactly. And this data is fed by the graphic aspect I was talking about, the management aspect, and of course, new data that arrives en masse: either through SCADA and supervision tools that are already in place, or thanks to the democratization of connected objects. We supplement the information thanks to these objects, which are already increasingly integrated within the machines themselves, or we add others when we're in particular contexts; because even if the machines are increasingly equipped with measurement elements, obviously, these machines are integrated into a more complex system. In the pharmaceutical or cosmetics industries, it's a set of end-to-end equipment that is used to produce or manufacture a finished product. And so, in this context, we may need to complete the instrumentation. So it's this set of additional and existing data that will provide this very real vision of the equipment for analysis purposes.
And where we're starting to move into what we might call 4.0, is really this ability to gather data - we could already start doing that - but it's [this ability] to analyze it in the light of the history and understanding of current equipment, to propose operating models by detecting weak signals, in order to propose maintenance prognosis. In other words, it's no longer just a question of technical action, but sometimes also of adjustment to avoid increasing wear and tear, or overheating, which in turn will lead to the unavailability of all or part of the process.
Could there be a misunderstanding of the data, or even a competition over who decides, who interprets? You were talking earlier about an instrumented valve, or a flowmeter, which produce their own data. So you add an extra layer, you use this data. How does it work? Who has the truth in the end, who decides?
Laurent Truscello: In any case, the decision-maker is still someone who will make a human decision. In any case, what we have is a maintenance prognosis that will lead to someone being alerted to a piece of information, or to an action that needs to be carried out. And this can be an act of control, sometimes, an act of verification - because you can also be supervised. We're going to check, especially at the beginning, that what we're proposing is coherent. And we're really into this notion of alert.
What does exist, however, and what can already be automated, are measurement self-correction loops. In other words, when we come to collect information, not to mention the complexity of assembling this information - just the raw data - if a detector, a sensor starts to malfunction, today we have our own filter algorithms that will detect that the sensor itself is drifting, and that it is the one that is giving erroneous information.
And so you've put your finger on an essential element: trust in data is an important element. So we need systems that are capable of checking themselves to some extent. And then there's blind trust in the system. The aim is to analyze weak signals, to understand sometimes complex systems in order to help in certain situations. We're not at the stage of total instrumentation of a line controlled by maintenance assisted by artificial intelligence. Today, we're working on isolated projects for critical machines or sets of critical systems to improve highly targeted productivity, with very precise results expected.
So, if we're being concrete, right now you've got pharmacy and cosmetics customers. What do they expect from you? What have you developed for them? What do they use you for?
Laurent Truscello: Already, they are using - and very quickly seizing - what is known as CMMS 2.0 or 3.0; in other words, they already have traceability, data and a history of their equipment. This is what we're going to see across the board, in any case for all players of a certain size, although there are still some who have Excel files, and who are starting to centralize them, or to share information when they have several sites. So, let's say, the fairly classic things, electronic signatures and information traceability.
Because, as we were saying, you asked me the question: is the equipment part of the process? Well, equipment is part of the process, because when certification audits are carried out - in the pharmaceutical industry in particular - maintenance-related equipment and processes are audited in the same way as production processes. We have to guarantee who worked at what time, on which product, in relation to which batch. So this notion is obviously important. Today, there's a gradual trend towards connected object technologies, but it's a gradual trend. That's it, and we're hearing a lot about it, but between hearing about it...
What does it mean in practical terms?
Laurent Truscello: In concrete terms, this means instrumenting pieces of line, or pieces of process - because we've judged this particular piece of equipment to be critical - and monitoring them to identify weak signals. The aim of these strategies is twofold. It's to increase equipment performance while reducing preventive maintenance. Because, if you like, it's quite easy to prevent a complex system from breaking down. All you have to do is check it every day, every hour, every minute, and take constant action. So you maximize your rounds, maximize your lubrication, maximize your maintenance. So it's all about the notion of balance. And so, when you're looking to optimize costs and quality, you're going to go for this type of technology.
But as I was saying, today we have a number of pilot projects. Here, we're going to be working on very specific and precise things, to optimize what I've just told you. The pharmaceutical and cosmetics industries are cutting-edge industries, so they already have management and supervision tools. So often, we'll make use of what already exists, and improve it to take a new step forward and move from curative maintenance to preventive or programmed maintenance, assisted - I'll call it that - by artificial intelligence; because we want to optimize, we want to be very precise, and at that point we need to make use of certain technologies.
That's not really the case today, but I'm thinking a lot about energy savings in the future. In the building industry, for example, and in other sectors, we can sense that there is a will to do so, because the energy stakes are high. But I'm convinced that we'll have to address this issue, or this desire to address energy consumption on equipment.
Are there any special features specific to the pharmaceutical and cosmetics sector?
Laurent Truscello: Compared to other business sectors, it's this very, very strong desire for traceability. It's really, really important to guarantee this historical record. And this also means that, if tomorrow elements are computer-driven, well, this notion of guaranteeing what was done, how it was done, is perhaps a stronger, more predominant element than in other sectors of activity - on the equipment itself... No, because cleanrooms can be found in other sectors of activity. I'd say that this whole "equipment" aspect can be found in other sectors, but it's not specific to them. What I would really note is the notion of traceability; and obviously, I didn't note it, but safety. So, this notion of quality, control and safety is obviously extremely important.
Traceability, safety, do we go as far as liability?
Laurent Truscello: This raises a number of questions. Indeed, if the algorithm is the one that controls the equipment. Today, we haven't gone that far. The algorithm proposes something to the technical teams and helps them to better understand. But we're not in the business of algorithms that automatically act on equipment actuators. This can already create a barrier and avoid this type of response immediately.
How quickly will this evolve? It's hard to say, especially if we start to move towards complete, interlinked systems. But for the time being, at least, we've chosen to be in the assistance business, really - that's the clear word - and not in the action business. This means that it's not the algorithm that acts on the system. The algorithm informs, feeds reflection. It can also suggest things, that's its purpose. But it doesn't act.
So what next?
Laurent Truscello: I think there's perhaps one new element that needs to be taken into consideration. Today, we're still in the business of acquiring data, processing it and making proposals. Perhaps what will evolve is with the first mobile applications. We've started to bring this information closer to the field, and we've started to give tools to the technicians, the agents in the field. It's clear that with these new approaches, there are new technologies that will also come to the aid of agents in the field, in real time, if only for their safety. So it could be augmented reality that assists with certain complex missions. It can be real-time communication, so that when he's in front of an installation, he can see if it's supplied, we can see if there's any remaining load in certain networks, we can tell him "watch out, the valve". In fact, sometimes, on complex systems, we have valves in close proximity, so "watch out, it's not valve A, it's valve B that needs to be turned". And we'll be able to give him real assistance tools for complex missions.
And so I think that bringing these notions of digital twins, of data in the field, and not just having curves or that sort of thing, but in overlay of reality or in a situation in front of the equipment, is going to be a challenge of the future because the equipment itself is becoming increasingly sophisticated. How do we bring this technology and this information to the agents in the field? This, for example, is one of the subjects on which Carl Berger-Levrault is working hard, in connection with what we might call augmented maintenance. How do we find technologies to bring this information closer to the agents?
Interview by Nicolas Gosse
www.carl-software.fr