Model-based design in action
Arend-Jan Beltman, manager of the mechatronics department at the Centre for Concepts in Mechatronics, says solving design challenges in its specialty area of mechatronics without a model-based design approach would be unthinkable, given the complexities. This development approach commands more up-front work in terms of understanding the design problem and building the model, but it is a huge time saver in the end, because the model helps the team understand what is going on. "You start modeling the expected problem with the tools, and through the model, you gain insight, and you can experiment to clarify your insights and iterate the design."
Take, for example, a project the CCM completed for Agfa, which acquired the :Dotrix Modular printer brand and needed to make some design adjustments to improve the printer's repeatability. The team had determined that tension on the substrate, or printing material, caused it to stretch, triggering a misalignment in the firing mechanism of the ink jet head, which reduced precision. Operators couldn't tune the force control properties, so the team turned to The MathWorks' Simulink and Simulink Coder to design, develop, and deploy a control system that would address this issue.
The nonlinear dynamics of the substrate path were modeled in Simulink, and CCM engineers then developed controllers using their own design tools. Simulink came back into play for simulating and refining the control strategy, and Simulink Coder was tapped to generate C code automatically for the controller.
Beltman says that Simulink's automatic code generation capabilities were a big factor in accelerating development time for this project. "In the conventional way, the software engineer would need to translate the idea of the control engineer into software code, but in this case, the control engineer also creates the software code using the Simulink Coder. It saves resources, it saves communications, and it lets the software department focus on other things like user interface design."
Dr. Rudiger Neumann, head of research mechatronic systems at Festo AG, says the model-based design approach also goes a long way in streamlining the trial-and-error process that so often derails engineering projects.
Festo, which builds robots out of pneumatic drives, would not be able to optimize its controller designs efficiently for complicated systems by building and testing them on physical hardware, according to Neumann. "You can't do design in a systematic way," he says, because a movement would be developed and tested, and then hardware would have to be constructed to test the controller model. If a problem were encountered, the process would start anew, with the movement modified and the hardware redesigned to continue optimizing the controller design. "Nowadays, by testing on the model and then building the hardware, you can use the model for controller design before you even design the hardware."