We’re surrounded by digital manufacturing softwares, and we use extremely complex products every day. But we don’t think about the effort it takes to design and manufacture these products.
Take cars, for example. We use cars throughout our daily life, and we don’t really think about how much of a technical marvel they actually are. Large companies with many departments and suppliers manufacture cars, and they do so by using numerous software programs and databases. The very term ‘product lifecycle management’ includes the software needed to design complex products such as cars, as well as to plan their manufacturing, while being able to simulate the products in every phase of their lifecycles.
The most pressing problem with creating a fully functional car is with the amount of complexity involved. Complexity steadily increases with product variety, market demands and new technology, such as electric cars, or how the Internet of Things interacts with one’s car.
One also must consider the demands one places on one’s personal connected devices and which devices already have an internet connection, which could add to the car’s complexity if they connect to the car to play music, provide navigation instructions or gather information on the car’s performance. Fulfilling these demands is a huge task if one wants to successfully build a car. If one tries to fulfill those demands using traditional organisations and today’s digital manufacturing technology, one will fail. Those organisations and software will be overwhelmed because they can’t handle the challenges and intricacies of this complexity headed one’s way.
How are companies supposed to deal with all of the complexity that comes with today’s digital manufacturing software? If we look at all of the departments in companies completing work for the products one buys, we see that even though organisations may be on the same tasks, few know how to use deeper data integration and digitalisation to complete that work much more efficiently. Ideally, everything should be commonly engineered and thoroughly tested before actual production begins. This engineering and testing can be completed today, but it’s mostly done by people in different departments who use their own tools, often with weak interfaces. Some organisations are taking a partial digital manufacturing approach, but crucial domains, especially electrics, automation and mechatronics validation, are still separate from everything else in these organisations. None of this will help as one moves deeper into the digital manufacturing age.
Digital manufacturing software & the smart factory
To understand why this process is still so complicated, and the value a new level of digital manufacturing software could bring to smart factories, we need to look at the beginnings of digital manufacturing technology.
The use of digital manufacturing software became more widespread around the year 2000. As the digital factory was first coming to be, companies were looking for ways to safeguard their factory planning and production planning, and they wanted to predict how profitable their products would be. These digital factories, which began using digital technology to create a digital representation of entire physical value chains, began taking off because the software and hardware to handle such a digital factory were, for the first time, mature enough. They could support the increased demand coming from a new level of product and production complexity.
These digital factory software systems are widely used today and generally follow the approach of dividing the planning tasks as “products,” “processes,” “resources” and “plants,” and these structures are stored and managed in PLM systems. On top of this, data models in sophisticated software tools are developed for detailed process, cell, line and factory planning, as well as engineering and simulation.
In recent years, simulation systems’ functionalities have also enhanced to move closer to virtual commissioning. This technology allows a convergence of digital factory models with the real factory; actual factory control software drives the digital models and safeguards the engineering results. The systems’ development is slowly moving in the right direction, but the progression of digital manufacturing software isn’t happening fast enough. These systems aren’t enough to help companies meet all of the demands they currently face or will face in the future, and companies will need to have agile systems in place to respond not just to market fluctuations, but also to customer demands.
Digital manufacturing software changes for a smart factory
What would it mean for smart manufacturing’s future if its engineering and validation could happen by being fully integrated, including all involved planning and engineering domains? What kind of holistic planning data could offer this during production lifecycle?
There’s still quite a bit of work left to do with digital factories before most companies reach that point. If you look at digital factory implementations today, you’ll recognise three major deficits that lead to inefficiencies and waste. These deficits hinder the digital factory to unleash its full potential.
First smart factory hurdle: No common databases
In digital manufacturing, the main tasks are planning, engineering and simulating. Although several simulation technologies for different use cases are well-established – for example, for kinematics simulation and robotics, or for discrete event simulation – there are still no common databases together with the engineering tools.
For example, the plant structure containing the data that’s the basis for most planning tasks, is seldom directly synchronised with changes that happen during simulation. This hasn’t yet happened because planning and simulation tools were developed independently and with one-way interfaces.
For a long time, a common platform for all product, process, resource and plant data for all authoring tools was out of reach. There were planning databases, engineering and simulation tools and still in parallel a huge amount of spreadsheet-based data. All contained data for a project is interconnected.
One result is the input for others to continue, but not only in one direction. Changes always happen, and you need feedback loops and round-trips. This isn’t possible with current the established systems and workflows this is not possible. A new type of data backbone is necessary that can handle and manage the needed for smart factory technology. This would increase the engineering efficiency dramatically and enable companies to deal with the complexity of the upcoming projects.
Second smart factory hurdle: No electrical or automation data
Most digital factories currently contain product, process, resources and plant data – but they don’t include the electrical and automation data to run the plants that would actually bring a digital factory to life. The electrical and automation data is missing so far because they’re usually defined in later project stages, and this definition usually comes from separate departments or suppliers in a waterfall-structured organisation. Electrical and automation data also have a different content and structure than more mechanically-driven digital factory models.
Software companies mostly lack all needed competencies to integrate this data in a mechatronics approach that all engineering domains could consistently use. Major portions of crucial, necessary data are still not part of the digital factory; they’re handled in offline systems, which lack strong interfaces and contain no or little change in the management.
This also means that today’s virtual commissioning tools are decoupled from the digital factory models and need to be fed through data interface, even though they need electrical and automation data. This decoupling becomes a source of errors and inefficient change management, which makes companies less efficient at responding to rapid market fluctuations. So, companies should increase their efforts working on mechatronics. This means that interdisciplinary teams need to be built to align all aspects of the respective production systems. The effort to create modular mechatronics building blocks pays off. Those modules should include the best practice and are tested. This reduces aliment efforts in the projects and increases engineering quality. Based on such modules, the simulation models can also be quickly configured for virtual commissioning.
Third smart factory hurdle: No actual production updates
Today’s digital factory tools are mainly used until the commissioning phase of a production system, and they aren’t updated according to the actual production. During production, an organisation’s own planning and optimisation tools will often come into play. The digital factory models that took lots of effort to build become useless during production: they aren’t up-to-date with the changes that happened directly on the shopfloor.
If these models need to be used again for the next planning or integration project, for a study or for an optimisation task, they would need to be manually updated, because no technologies are implemented to automatically align the digital models with the real production. If a smart factory is implemented holistically, these models could be used to synchronise the digital factory and the real factory by using the control programs and network information on the shop floor. Those exactly represent the current state of a production, and this information should have its counterpart in the digital models. The result would be the connected digital twin.