The German Aerospace Center (DLR) has developed a flexible automation platform for its own work cell to build a series of CFRP aerospace structures that are manufactured using the same process-dry non-crimped fabric, pasted laminate and resin infusion- Rear ballast bulkhead and fuselage panel. Instead of teaching collaborative robots, they define their own collision-free paths to place slices into the tool and place it into the tool based on CAD and process definition inputs and AI algorithms. This AI-driven automation is one of the foundations of the future intelligent Composites 4.0 factory.
Composite 4.0 is a small galaxy in the Industry 4.0 universe. Industry 4.0 is the digital transformation of how products and services are designed, produced, delivered, operated, maintained and decommissioned. The goal of composite manufacturing is to use automation, sensors and data, 5G communications, software, and other evolving digital technologies to make products and processes more efficient, smarter, and more adaptable.
Composite manufacturers are making this digital transformation along a spectrum. The initial steps include on-line inspection and optimization of the process to reduce waste and cost, while improving part quality and yield. More advanced solutions are dedicated to intelligent and autonomous production, which is not only agile, but also capable of responding and even predicting the ever-changing market and customer needs.
"Composite 4.0 is not an end, but a tool," explains Dr. Michael Emonts, managing director of the AZL Aachen Lightweight Integrated Production Center at RWTH Aachen University, Germany. His iComposite 4.0 project demonstrates an adaptive process chain that has the potential to reduce vehicle chassis costs by 50-64%
Christian Koppenberg, general manager of composite parts manufacturer Dynexa (Laudenbach, Germany), believes: "It is only digitalization and digital transformation that are different. Digital transformation has actually changed the processes behind the enterprise and opened up new opportunities and business models. "
"Composite 4.0 does not only use robots," Dr. Michael Kupke, head of the German Aerospace Center (DLR) Lightweight Production Technology Center (ZLP in Augsburg), asserts that the center has developed a work cell equipped with artificial intelligence [AI]. The collaborative robot can switch from producing composite ballast bulkheads to fuselage panels without reprogramming or retraining. "This technology ensures that you don't have to teach robots because there is no business reason. Composite 4.0 can not only increase efficiency and cut costs. How companies think and deal with changes in production methods will determine which companies can survive and which cannot survive."
Adaptive preforming: RTM
"The idea of the composite material 4.0 project is to combine dry, long glass fiber (25-30 mm) jets, and then through automatic fiber placement (AFP) with unidirectional carbon fiber mesh reinforcement, from cost-effective chips and drag blocks Create prefabs in the process,” Emonts explained. "The selected prototype machine, under the floor of the car, was previously made with more expensive textiles, which also caused more than 60% of the waste."
Composite 4.0 conversion requires integrated fiber spraying, fiber deposition and subsequent resin transfer molding (RTM) processes so that they react to each other and adapt based on the quality measurement part between steps (Figure 1). “We use a machine vision system from Apodius GmbH (Aachen, Germany) with an optical laser sensor and camera module to describe the surface topology spray pre-forming,” Emonts said. Apodius has improved the software to analyze fiber percentages in all directions. The iComposite 4.0 line compares this with digital design and decides whether it meets mechanical requirements. If so, use standard UD grids for reinforcement. If not, it decides where to place the extra UD fiber layer.
The AZL Aachen led project combines 3D fiber spraying, automatic UD tape laying and RTM into an adaptive process chain. Automation allows processes to react based on each other
For the impact on the quality of the preform, laser technology is used to evaluate the quality of the preform.
However, these additional UD layers may cause part thickness and geometry to exceed tolerances. He explained: “Therefore, we combine the preforming line with the adaptive RTM process. If necessary, the thickness of the parts can be adjusted by increasing the pressure on certain parts of the press.” This is also automated, and the purpose is to replace the production line operator. Intervention, but it does require the use of measurement data and standard FEA software to simulate part performance.
"Currently, simulations of the mechanical properties of parts are performed offline," Emonts said. "We generated a database of process and component changes, created algorithms to react to each change, and verified these changes through FEA. Therefore, based on the changes measured by the wire, the algorithm guides it to perform appropriate mitigation. In order to make This line can be adapted in-situ, and the next step will be to add machine learning. At the same time, AZL is working on multiple composite 4.0 projects, including the production of self-optimizing hybrid thermoplastic composites, and the integration with tape-based custom blanks Stiffened injection molded parts.
Zero defect carbon fiber reinforced plastic wing skin
The ZAero project is another important composite material 4.0 project, which was launched in 2016. The project aims to increase the productivity of large carbon fiber reinforced plastic (CFRP) structures, such as wing skins. By using prepreg AFP or Danobat (Elgoibar, Spain) automatic dry material placement (ADMP, see "Proving the feasibility of dry fabrics for the infusion of large aircraft structures"), defects can be reduced through automatic online inspections. Process monitoring during resin infusion or prepreg curing will predict the curing state and shorten cycle time. The collected process and defect data is used with FEA to predict part performance. Then enter it into a decision support tool to resolve the defects that have been discovered. A part flow simulation for CFRP wing skins was developed and imported into the tool to help optimize rework strategies (Figure 2). Today, many of these parts are reworked during the manufacturing process, but only after NDI. Earlier rework and improved process control are indeed the goal of the ZAero project, and it is also the driving force for achieving the goal of increasing productivity by 15% and reducing production costs and waste by 50%.
The ZAero project demonstrated the online defect control of CFRP stiffened panels. Shown counterclockwise from the top left: laser sensor integrated with AFP, defect classification, flow chart, showing the combination of FEA modeling with part performance and decision support tools and part process simulation of CFRP wing skin (top right), with Optimize rework strategy.
Through the final review in September 2019, the prepreg AFP sensor developed by the project leader Profactor (Steyr, Austria) not only realizes automatic online inspection, but also can be used for on-site calibration of parts. "The sensor can detect standard defects such as gaps, overlaps, FOD, fluff and twisted tows, as well as the morning and evening cutting of each tow," said Dr. Christian Eitzinger, head of machine vision at Profactor. The missing tow can be automatically corrected by accurately placing the other tow in the omitted position. However, the machine must be stopped to remove fluff or twisted tow. "Using the database established by the 3D Experience for CATIA of Dassault Systèmes (Paris, France), we can calculate the impact on part performance based on the size, shape and type of the defect. It only takes a few seconds to process all the defects in the layer. Then, the machine operates The staff decides which defects can be kept and which repairs must be made."
For the monitoring and control of the infusion process, Airbus (Toulouse, France), through its subsidiary InFactory Solutions (Tauckkirchen, Germany), has developed three sensors to measure temperature, curing state and resin flow frontier. "We have integrated them with CATIA 3D Experience and have shown that data can be reliably acquired and added to the digital thread of each part," Eitzinger said.
The last three parts of the demonstrator are upper fender sections with three stringers. For this part, Profactor's decision support tool was demonstrated live at the partner FIDAMC (Madrid, Spain). The tool has been connected to part flow simulation based on Siemens PLM (Plano, Texas, USA) Tecnomatix Plant Simulation software , And run on Profactor server in Austria. In addition to establishing a defect database, ZAero also conducted machine learning experiments. The generated computer model designed by hand is combined with the deep neural network to detect and classify the defects. Even in the presence of artificially created defect data, different areas (gap, overlap, etc.) can be detected in the real ADMP monitoring data. The correct classification rate of drag, burr ball) reaches 95% for deep network training (similar to how to calibrate an ultrasonic testing system on a series of intentional defects).
"We will definitely pursue some kind of next stage result," Eitzinger said. At the same time, Profactor is commercializing modular sensors to solve fiber orientation and defects in the automatic web-laying process. InFactory Solutions also offers its AFP and resin injection sensors, and fiber placement partners Danobat and MTorres (Torres de Elorz in Torres de Elorz, Navarre, Spain) are now selling equipment with integrated online inspection.
Dynexa's digital transformation journey
Dynexa is a composite parts manufacturer specializing in CFRP tubes and shafts. "We have been trying to digitize everything." Managing Director Koppenberg said. "We have got rid of manual and simulation processes and mainly integrated everything into our ERP [Enterprise Resource Planning] system. But how do we do this in the manufacturing industry? We understand that we are in the work procedure agreement or process All the content put in is code, which is the basis of digital conversion. But where is it stored? On the local server, in the cloud or in the computer? We will ask five people and get seven of the measures that should be taken. Answers."
Fortunately, the German government has developed a plan for universities to provide free Industry 4.0 consulting for small and medium enterprises (SMEs). Dynexa started to cooperate with Darmstadt "Mittelstand (SME) 4.0" Competence Center. Koppenberg recalled: “They said they don’t need to worry about the digital architecture, but pay more attention to what needs to be measured and how to do it.” “We chose a process that involves a lot of manual measurement. We know that we have encountered quality, time and The cost issue."
Dynexa uses a wet silk winding process. The key step is resin pickup, in which the dry fiber is run onto a compaction roller, which rolls out of the resin bath. The doctor blade is close to the compaction roller, and the doctor blade determines the amount of resin to be combined with the dry filament before winding Koppenberg said: "If we collect too much resin, we may exceed the specified pipe diameter, but if there is too much resin, we may fall below the minimum allowed diameter. "
He pointed out: “If you don’t take measurements, you will only know the final diameter after curing after all the value is invested.” “So the operator must stop the machine, measure the part, write it down and restart it. Based on years of experience , We know what the thickness of the laminate should be at each stage of winding. Therefore, the operator can compare the measurement results and adjust the squeegee as necessary to correct the resin pickup, but this is very manual and requires the skills of the operator And experience."
In cooperation with the Darmstadt SME 4.0 Competence Center, Dynexa has developed a digital camera system that monitors the thickness of the tube/shaft and automatically adjusts the resin reclaim (bottom) (top) during the filament winding process, thus eliminating manual measurement. Improved efficiency and cost. Source | Dynexa
To digitize this, Dynexa talked with countless laser and camera manufacturers. "They will say,'We have a solution,' but no one can make it work," Koebenberg pointed out. However, the University of Darmstadt team enabled the use of cameras by determining the corrections required due to certain physical factors, such as light reflected from wet surfaces. He added: "Now, we have connected the winding machine to the measuring device, which operates in a very standardized way."
The team developed a database of correction tables and decision algorithms that enable the filament winding machine to know the goals of each stage of the specific tube to be wound. Koppenberg explained: "If the input of the measuring device shows that the resin pickup is not where it should be, the filament winding machine can adjust the scraper to restore it to the specification state without stopping the winding for measurement."
Now, every winding machine has a digital measuring system and an Ethernet card. "The most expensive part is the cable to install and run to the server," Koppenberger quipped, "but now we can communicate with each machine and collect all the data." There is another benefit. "In the past, operators programmed on the machine, but once they are connected to the server, we can program on any desktop or portable computer. This further reduces downtime and eliminates another production bottleneck. "
The first step of digitization allowed Dynexa to improve process control, quality and efficiency, making it more cost-effective. It also stimulated further transformation.
Changing the paradigm of composite materials
Airborne Aircraft Corporation (The Hague, Netherlands) launched an on-demand manufacturing platform for its own suppliers in September 2019 for automated manufacturing of composite materials. Suppliers can use this tool to enter designs into a web-based platform. Then, the system instantly creates machine code and determines the production duration and cost. The product can then be customized and produced in an automated production cell after ordering. The platform uses an airborne automatic laminating unit (ALC) to process thermosetting prepregs. It will be extended to other processes, such as a high-volume thermoplastic composite (TPC) production line developed for SABIC (Riyadh, Saudi Arabia) specialty products business unit.
Airborne launched its composite online platform in 2019, where customers can input designs and receive the cost of laminates/parts produced by Airborne's automated manufacturing unit and estimate the delivery time, including Falcon's high-volume thermoplastic developed in future expansions Composite material production line SABIC (bottom).
"The platform is a key element of how we view the digital future of composite manufacturing," said Marcus Kremers, chief technology officer of the airborne company. "Five years ago, we changed from a parts manufacturing business model to helping customers achieve automation and digitalization. We are developing a series of solutions so that customers can easily use composite materials for construction." The product portfolio includes ALC, automatic honeycomb irrigation SABIC calls it the Digital Composites Manufacturing Line (DCML), and Airborne calls it Falcon. The latter is an example of an airborne customized solution. Kramers said: "We embed the manufacturing knowledge of composite materials and parts into these automated systems so that customers do not have to be experts."
There are three business models for airborne: buying automation equipment, renting or renting to airborne to operate through manufacturing as a service (MaaS). The onboard automation unit has sensors and an online inspection system that can generate alarms based on a database of customer-defined defects and tolerances. "The Falcon product line has a low tolerance for visual quality defects," Kremers pointed out, "but our aerospace automation is more driven by structural tolerances. We are also constantly improving the self-learning and adaptive capabilities of technology. For example, our next software version for ALC tape laying will have the ability to identify defects and modify production procedures on the fly."
In the long term, the vision is to expand the on-demand platform to collect composite parts production capabilities distributed across multiple companies and regions. Kremers cited Protolabs (Maple Plains, Minnesota, USA), which can provide prototypes for injection molding, sheet metal, CNC machining or 3D printing on demand in just one day. Similarly, Plyable's online application (Oxford, UK) provides molds for manufacturing composite materials, ranging from polyurethane panels to composite materials, including carbon fiber materials and 3D printing tools. "This is another way of organizing the value chain," Kremers said. "We are making machines and software that make composite parts possible."
AZL Aachen also achieves this goal with its ultra-fast consolidation machine, which is used to produce multilayer TPC laminates in less than five seconds. It was commercialized in 2019, using Conbility (Aachen, Germany) robot automatic laser-assisted AFP applicator and 25 mm wide UD tape, and the principle of split flow (this is the latest technology in the high-speed printing industry). Produce various thicknesses of simply nailed or fully consolidated TPC laminates with local reinforcement. "Our vision is to provide scalable machines to support online platforms," Emonts said. "The production line can have multiple workstations, and each workstation has multiple AFP applicators. The customer will enter the requirements and get options about the loose-leaf book, cost and delivery. After the final determination, the applicators communicate with each other to organize the production, and Not an operator. This is completely intelligent production of custom composite materials."
The main focus of ZLP is the automated production of CFRP structures. Florian Krebs, head of the ZLP flexible automation team, pointed out: “It is difficult to prove that only one part or program is automated.” “But if you move from a task-specific machine to an automation platform that can be reconfigured with almost no additional setup, you Now there is a business solution. The more flexible the platform, the faster the return on investment."
The work cell shown in the opening image is part of the ZLP project PROTEC NSR, which aims to manufacture a series of parts following the same process route: pick-and-place stacking of dry non-crimp fabrics and resin infusion. “This process is designed for the rear ballast bulkhead of the Airbus A350, but you can also manufacture fuselage panels or wing covers on this production line because the steps are similar,” Krebs said.
"To achieve a flexible automation platform, certain technical foundations are needed, including robot algorithms, sensors and how to understand the data they generate," Kupke said. "For example, the design of the PROTEC NSR production line is towards the greatest modularity-all modules are interconnected to demonstrate a self-configurable, correctable and optimized system, scalable size and complexity."
Figure 5. Digital structure of flexible intelligent automation
For the PROTEC NSR project, ZLP has developed a flexible automation platform that can produce CFRP rear pressure diaphragms (top right) or fuselage panels (bottom right), and can quickly switch between these by simply changing the CAD file. The instructions defined by the engineer in the tool chain (dark blue box) control how to interpret the CAD data to automatically generate new process steps.
He explained his modules, including CAD models, process definitions, process models that allow simulation and execution of processes, manufacturing execution modules, sensors for data acquisition, software for data annotation, and databases for storage.
"On the left side of the module, you make a plan. Then the execution module will execute the plan." Kupke said. "In the process steps, we obtain data from all involved machines and processes, such as cutting machines, robots, buildings (temperature, pressure, humidity), cameras during pick-and-place, etc. We analyze the data in real time during processing , And automatically use metadata to annotate the collected data to feed it to the database, which forms the basis of the digital twin in the process. The most important thing about digital twins is to have a central repository, which is a source of truth. The CAD model and process definition of each part are part of its unique authenticity."
After installing these modules, the production line can run autonomously at the push of a button. The robot infers the workpiece to be cut next from the CAD model, production plan and camera, and looks for the workpiece on the table (for example, from the other 100 workpieces). "They decide how to configure the gripper to pick it up and place it in the tool, and know where to place it." The robot determines each start/end path of all process steps based on the production plan and knows when it is completed. "Usually, these paths are taught by people," Kupke pointed out. "But in our system, each path is automatically defined, there is no collision, and it is real-time. If the CAD model or process definition is changed, the robot will adjust without any additional teaching work. But , What if you completely change parts? With this type of automation, you can make changes very quickly. This is the way to achieve flexible production. Our role in ZLP is to develop technical bricks and link them together to pave Level this road."
Opportunity and ontology
The COVID-19 pandemic highlights the value of flexible production. It also creates an increasingly unpredictable business environment. "In the past two to three years, everything has become more turbulent," Dynexa's Bruckhoff said. "Our customers really want quick answers in order to respond to their customers. By providing a new online ecosystem, we make the entire supply chain more competitive."
This is recognized by the aviation industry. "We need a digital production line and the basis of the entire site to achieve horizontal and vertical integration," said Marc Fette, Chairman of the Technical Department of the VDI Aviation Technology Department-the German Engineers Association and the Composite Technology Center (CTC) Airbus R&T subsidiary. CTC projects in Composites 4.0 include material and asset tracking, collaborative robots, and advanced process chains. But Fette emphasized the need for ontology—ontology is a term and general protocol for digital communication and data exchange. (See the online sidebar "Composite 4.0 Architecture and Ontology".)
He explained: "You need a given factory's overall network of all machines and production systems, but this must also be extended to the entire value creation chain, including engineering, procurement, logistics and materials, and process certification and other disciplines, on the one hand. On the other hand. On the one hand, all stakeholders, such as suppliers, must be considered and involved in this change process. We have seen many pilot projects, but when you look at it in detail, there is still a lack of strategies for adopting a holistic approach for each company or production chain. "
He continued: "We have a huge global supplier network, and they have the same requirements to operate as a digital supply chain. Most of our suppliers are small and medium-sized enterprises serving aircraft manufacturers (Airbus and Boeing). If Without discussions on common standards, these challenges can be passed on to suppliers. They may not be able to afford two different sets of standards for all machines, including documentation, evaluation data, network security, etc."
Fette acknowledged that these ideas are important and said that aircraft OEMs are developing plans to address these challenges. "But there are many obstacles like this, and they are really complex. They involve not only technology, but also social, economic, ergonomics and legal issues. This is a process of ideological change. We have just begun. But to succeed, we It must be understood that these new systems depend on people, and these people must be involved, not only original equipment manufacturers, but also the entire global network."
ZLP's Krebs pointed out: "The market in all walks of life is becoming more and more fragmented, and everyone is facing transformation and transformation." ZLP's Kupke added: "Many people don't see this as an opportunity. "But those who do see this opportunity foresee the realization of personalized access to Composites enabled by Composites 4.0, and as a result, the market has become broader, including applications that we have just begun to conceive.