Lightweight Design worldwide

, Volume 11, Issue 2, pp 42–47 | Cite as

Production of fiber composite structures by means of cooperating robots

  • Dominik Delisle
  • Markus Schreiber
  • Christian Krombholz
  • Jan Stüve
Production Fiber Deposition

Scientists at the Center for Lightweight Production Technology at the German Aerospace Center are, for the first time worldwide, using several cooperating robots for the production of fiber composite components. The multi-head technology automatically deposits the fibers and reduces production times and costs. As a result, the scientists have implemented a time-efficient task distribution for the robot units.

Automated Fiber Deposition

The use of fiber composite components as structural components in the fields of aviation, aerospace, wind energy and automotives has experienced steady growth in recent decades. In aviation in particular, growing structural-mechanical, economic and ecological requirements for new generations of civil and military aircraft have led to an increased use of fiber composite components.

Due to the high material and production costs, however, the use for primary structures in aviation is worthwhile only as of a certain flight duration, with the result that fiber composite components, such as for wings and fuselage sections, are used in particular for new models of long-haul aircraft.

To produce these structures, the technologies of automated tape laying (ATL) and automated fiber placement (AFP) are predominantly used to reduce production times and material waste in comparison to that of manual production by means of automated production, while at the same time increasing reproducibility and component quality. The current manufacturing processes are purely serial. By means of portal, gantry or robot kinematics, the fiber material is deposited on a shaping mold in individual material lines (called “courses”) by an end effector, Figure 1. Closed layers of different sizes, thicknesses and fiber orientations result from numerous courses, which are positioned adjacent to and on top of one another, together forming a component.
Figure 1

The deposition of individual courses with AFP technology, which in their entirety form a closed fiber layer (© DLR)

The current production of components through automated fiber placement technology is characterized not only by fiber deposition, but also by planned and unplanned boundary processes [8]. The proportion of fiber deposition over the entire production time depends on many factors and is usually in the order of 13 to 40 %. Significant shares of the standstill times of a production plant currently result from visual inspections and reworks, planned and unplanned maintenance, as well as material loading processes of the placement units [3, 9].

The proportion of fiber deposition in the total production time is about 13 to 40 %.

Since minimizing the material costs for carbon fiber reinforced plastics is currently limited, the primary focus of component cost reduction is on the manufacturing costs and, therefore, on the optimization of the production processes [10]. In particular, the increase in the production rate and economy of the manufacturing processes leads to a fulfillment of the requirements of current and future aircraft programs [1, 5].

The German Aerospace Center (DLR) has therefore developed a new manufacturing approach through which the current serial production process can be significantly accelerated by parallelizing the work steps, while productivity can be increased by decoupling the maintenance processes.

Through the use of several cost-effective robot units which operate autonomously and automatically on a rail system, tasks can be parallelized, new production technologies can be integrated into component production and/or standstill times can be reduced due to redundant units. The research platform Grofi, developed and implemented at the DLR in Stade, currently has six robot units that represent different technologies for automated fiber deposition [4]. With the Grofi system, a concept has been implemented in which each robot unit of the system is equipped with its own control device and includes all systems and materials that are required for production. This design results in a high degree of flexibility, since each unit is already capable of performing the demanding manufacturing tasks itself. There is a superordinate central control unit that coordinates the robots by distributing information to the units, and which, at the same time, represents a central collection point for data for monitoring and controlling production. The communication between the manufacturing units and the central controller has to operate at a high cycle rate and free of losses and malfunctions to prevent any collisions and to make the data available for long-term evaluation. Depending on the production order and the available placement units, the intelligent overall plant control system assigns manufacturing tasks both offline and online.

Multi-head Approach

To meet the aforementioned challenges of increasing the efficiency of fiber composite production processes, the Lufo (Luftfahrtforschungsprogramm) VI project Ewima (efficient wing cover manufacturing) involves the bundling of experience from previous Lufo IV projects to concatenate the experience and results of previous LuFo IV projects to develop innovative, highly efficient manufacturing technologies. One focus of the project is the development and verification of the multi-head approach by means of several simultaneously operating production units for the production of large fiber composite structures, Figure 2.
Figure 2

While production robots work in the left-hand area of the Grofi system, placement units can be set up and maintained at the same time to the right (© DLR)

The verification of the multi-head capability of the AFP research platform Grofi could be demonstrated within the project. For this purpose, a carbon fiber reinforced structure was produced, for the first time worldwide, using two simultaneously acting robot units with overlapping, dynamic working areas in the AFP process, Figure 3. A generic wing shell model with a wingspan of 8 m was used as a production demonstrator. The complexity of the demonstrator presents all the challenges of a real wing.
Figure 3

Automated production of the demonstrator component with two AFP units (© DLR)

During the manufacturing process, relevant process data is recorded by the placement units, and collected and managed by the overall control system. Subsequent to the multi-head deposition of the project, the collected process data was evaluated and used for a comparison of the production times of multi-head and single-head operation. Since the production process with one placement unit was merely simulated in the present case instead of being conducted in reality, validation of the simulation environment of the Grofi research platform was performed in advance. For this purpose, a representative partial sequence of the component comprising 305 courses and containing both reinforcing layers as well as layers extending over the entire component was considered.

A comparison of the recorded production times of the multi-head operation with the results of the pre-simulation showed a deviation of less than 2 % for the observed sequence. On this basis, the comparison of the production times of the single-head and multi-head operation with two placement units was performed. The results are shown in Figure 4. The diagram shows that in the present case, a time saving of almost 30 % can be expected through the use of two placement units. This estimation reflects the current state of development. A more detailed analysis of the production times of the multi-head deposition shows that further development of the plant control system holds additional potential in terms of saving time. As an example, synchronization mechanisms are used for the current multi-head operation that ensure a very high level of safety with regard to collision avoidance. Further development and testing of the online collision monitoring of the Grofi research platform will enable these mechanisms to be adapted, and additional production time to be saved. Another relevant aspect is the current data and communication structure of the NC programs which are to be processed. This is designed for research operations, and can be optimized for industrial use.
Figure 4

Evaluation of the production times of the multi-head deposition system (© DLR)

Sequence Planning

A meaningful subdivision of the entire process into smaller work packages is first necessary for the production of a fiber composite component with cooperating placement units. For the multi-head approach, a course was chosen as the smallest possible unit.

Based on NC data concerning the entire component and which is generated with commercial CAD/CAM tools, a course-based division into individual partial programs is effected. Production times and topology-related dependencies of the individual courses are determined with the help of a simulation and planning environment developed for Grofi. This ensures that the layer structure specified by the design is maintained during the cross-ply manufacturing and the use of multiple units.

The information concerning the production times and priority relationships of the individual courses serves as input for scheduling, which divides the work packages among the available placement units and determines a processing sequence. A manufacturing execution system (MES) is connected to the Grofi research platform and uses the schedules to execute the programs of the placement units.

Mathematical Hurdles

It is essential for time-efficient manufacturing processes with cooperating machines to have work schedules available in which as many tasks as possible can be performed in parallel, and in which the units do not obstruct each other. The properties of conventional mathematical models are not adequate for mapping the manufacturing process of machines that work next to one another on rails. A new model and appropriate solution heuristics for this problem class were therefore developed at the DLR. The greatest difficulty in modeling is that the machines have to move in space and along the rails when working, while the schedules have to prevent collisions between the robots. In addition, the developed model takes into account that the schedule can still be processed stably even if the actions of the robots deviate from the previously simulated movements, such as for the purpose of correcting movements.

The solution heuristics have to be designed in such a way that they can split schedules for several thousand orders among several machines. In the mathematical context, this task is one of the problem classes that are difficult to solve.

The developed software is able to split the manufacturing steps of a manufacturing process designed in the traditional sense for one machine over several production units and to translate this plan into code that can be executed to control the system. To test the plans, a virtual map of the system was developed in which multi-head operation can be simulated and visualized in advance, Figure 5.
Figure 5

Virtual twin of the system for simulating and visualizing multi-head operation (© DLR)

The planning of the processes currently takes place offline, this is before production. In the future, online concepts are to be developed which will make it possible to modify the processes during production in order to respond to maintenance or errors. Since this enables the Grofi concept to have set-up work performed in the maintenance area while production is in progress and to have redundant units maintained for these times, it is intended to fully exploit the potential of this approach in order to maximize production efficiency.


Compared to current production facilities that have only one fiber placement head, the use of a flexible and automated production facility for manufacturing large components made of fiber reinforced plastics offers advantages with regard to production times per component. These time advantages also affect production costs related to components. Figure 6 shows the component costs of a 20-m-long component, taking into account a defined production rate of 60 components per month and the resulting number of required production lines. The considered configurations differ in terms of the number of placement units involved during fiber deposition (1st number) as well as in terms of the available redundant units (2nd number). Material loading times and an otherwise error-free process with automated quality assurance are also required in order to eliminate process times for current visual inspections. Further information concerning the DLR developments for automated online quality assurance is described in [2] and [6].
Figure 6

Production costs of a wing shell depending on the plant topology: placement units involved in the fiber deposition (1st number) and available redundant units (2nd number) (© DLR)

In future, the findings of the project Ewima will be expanded, and investigations with regard to technology combinations such as AFP and ATL, digitalization, in-situ verification and multi-material deposition, will be performed.

Centers for Lightweight Production Technology

With the Centers for Lightweight Production Technology (ZLP) in the German cities of Stade and Augsburg, the DLR operates two research facilities dedicated to the transfer of laboratory scale production technologies to real scale industrial applications. The installed research platforms are highly flexible in terms of the technologies used, and allow the production of aerospace-grade large components. The ZLP in Stade operates platforms researching multi-robotic fiber deposition, in-autoclave and out-of-autoclave infusion and curing processes, as well as high-volume component manufacturing in automated textile preforming and resin transfer molding processes. The ZLP in Stade is organizationally connected to the Institute for Composite Structures and Adaptronics of the DLR in Braunschweig.



The findings presented here result from the project “ZLP Stade — Large Components Appropriate for Fiber Composites and Online Quality Assurance in Autoclaves,” which was funded by the Economic Development Program of NBank, as well as the joint project “Development of Technologies for Efficient CFRP Wing Shell Production,” funded by the German Federal Ministry for Economic Affairs and Energy (BMWi) on the basis of a resolution of the German Bundestag. We would like to thank the members of the project-accompanying committees for their cooperation and financial support.


  1. [1]
    Ahrens, M.: Fibre Placement Offers Cheaper Composites. In: Reinforced Plastics 42 (1998), No. 6, p. 25CrossRefGoogle Scholar
  2. [2]
    Apmann, H.; Mayer, M.; Fortkamp, K.; Haschenburger, A.; Krombholz, C.; Meister, S.; Ucan, H.; Zapp, P.: Verfahren der inline-Qualitätssicherung und der zerstörungsfreien Prüfung innerhalb der Fertigungslinie von Faser-Metall-Laminaten. Deutscher Luft- und Raumfahrtkongress, 2017Google Scholar
  3. [3]
    Halbritter A.; Harper, R.: Big Parts Demand Big Changes to the Fiber Placement Status Quo. SME Composites Manufacturing, 2012.Google Scholar
  4. [4]
    Krombholz, C.: GroFi: Large-scale Fiber Placement Research Facility. In: Journal of large-scale research facilities 2 (2016)Google Scholar
  5. [5]
    Dirk H., Lukaszewicz, J. A.; Ward, C.; Potter, K. D.: The Engineering Aspects of Automated Prepreg Layup: History, Present and future. In: Composites Part B: Engineering, 43 (2012), No. 3, pp. 997–1009CrossRefGoogle Scholar
  6. [6]
    Meister, S.; Krombholz, C.: Konzeptionierung und Implementierung einer bildbasierten online-Qualitätssicherung für den automatisierten Faserlegeprozess. DLR internal report, 2017Google Scholar
  7. [7]
    Menzen, A.; Krombholz, C.: Wirtschaftlichkeitsbetrachtung von Faserlege- und Anlagentechnologien für Großbauteile der Luftfahrt. DLR internal report, 2016Google Scholar
  8. [8]
    Potter, K. D.: Understanding the Origins of Defects and Variability in Composites Manufacture. Conference on Composite Materials (ICCM) 17, Edinburgh, 2009Google Scholar
  9. [9]
    Rudberg, T.; Nielson, J.; Henscheid, M.; Cemenska, J.: Improving AFP Cell Performance. In: SAE International Journal of Aerospace 7 (2014), No. 2, pp. 317–321CrossRefGoogle Scholar
  10. [10]
    Schmidt, C.; Schultz, C.; Weber, P.; Denkena, B.: Evaluation of Eddy Current Testing for Quality Assurance and Process Monitoring of Automated Fiber Placement. In: Composites Part B: Engineering, 56 (2014), pp. 109–116CrossRefGoogle Scholar

Copyright information

© Springer Fachmedien Wiesbaden 2018

Authors and Affiliations

  • Dominik Delisle
    • 1
  • Markus Schreiber
    • 1
  • Christian Krombholz
    • 1
  • Jan Stüve
    • 2
  1. 1.DLRStadeGermany
  2. 2.Germany

Personalised recommendations