The formalized specification of new functions is one thing that the automotive industry could adopt from aeronautics, according to Professor Dr. h.c. Manfred Broy of the Technical University Munich. During a synergy forum in Hamburg, Germany, named Aircraft Meets Automotive: System Architectures, the presenters and attendees tried to pin down the potential overlap between automotive and aircraft engineering.
At first sight, the trends and challenges in both branches seem to correspond in many respects.
“Systems will be more complex,” said Broy, looking not only at the automotive and aircraft industries but also at other areas such as medical systems and industrial automation. “Over the next 5 to 10 years, there will be a lot of development work and development cost. There will be more interoperability, more software-based functions, and there will be more suppliers which serve several fields. Hardware will become a commodity. Software is and will be more and more at the core of engineering."
Systems engineering will continue to gain importance because it is the only way to master the many dependencies between devices and functions. Two key facts illustrate the level of complexity. For instance, 600 or more controllers are installed in a modern civil aircraft, and between 2000 and 2500 functional features have to be controlled in a car. With regard to the current level of complexity, Broy spells out another similarity between automobile and aircraft: “Every potential reason why things can go wrong on a system level can be found in both, the car and the airplane.”
Defining a well-structured system architecture is crucial in this situation. And, according to Broy, this is where the automotive industry can learn from aeronautics. “Only 10 to 20% of all functions change with the transition from one car generation to another," he said. "However, up to 80% of the implementation changes. This is owed to unstable technical architectures which change according to progress.”
“The automotive industry is much more focused on technical architecture instead of beginning with conceptual architecture,” said Broy of his experience from intense cooperation with manufacturers.
His suggestion is to start with a formalized specification of the functional requirements and a concept of the logical system architecture. By applying model-based techniques for this purpose, the system architecture is defined totally independent of the technical realization (i.e., software, hardware, and deployment).
Beginning with conceptual architecture is a form of functional "de-combination." This approach answers the question of how a system can be structured in functionalities and how these functionalities are logically networked. This is an important advantage in the search for bugs as the search can follow the defined traces and layers of the conceptual architecture’s functional hierarchy.
“As many as 40% of all system bug problems come from the level of function hierarchy," he said. "Hence, what we really need is front-loading. We need to start with requirements, schemes of interaction, and functional dependencies.”
Putting his advice for systems engineering into a nutshell, he added: “Before building the system, the architecture should be modeled and it should make sure that it fulfills the specifications. That way we will find bugs much earlier and much faster. Also several suppliers can work on one project.”
Henning Butz, Head of Information Management and Electronic Networks Development at Airbus Industries in Hamburg, confirmed the validity of this cultural change in engineering with a comparison between past and present: “Developing a complex air-conditioning system for a civil aircraft in the classical way used to mean four or five years of trouble. During that time span, we had to identify and classify the bugs, one by one, and fix them. Typically, the fixing itself gave rise to new problems.”
As a consequence to this slow and time-consuming process, Airbus introduced formalized specification roughly a decade ago. “It is a powerful tool,” said Butz. “To analyze the practical benefit, we hired a software engineer to translate our old informal air-conditioning requirements into formalized specifications. Even though this expert did not even come from the aerospace industry, the job was done after four months. Another two months later, all bugs were fixed.”
The bottom line from the synergy forum seems to be that there are probably as many similarities as there are differences between the aircraft and automotive industry. However, in the field of skills and methodology for systems engineering, the overlap is particularly strong.
Considering the cost angle, this raises a question, said Broy. “Which differences between industries are necessary, which have to be overcome?” Looking back at the last decades, the expert concludes that in system engineering a revolution has taken place since he worked on the first system solutions for military aircraft in the late '60s.
“It all happened within the career life span of one engineer generation," he said. "I expect another revolution to happen. We are entering a completely new type of engineering culture. It is the step from assembly engineering to systems engineering.” Broy openly admitted: “There is a long way to go together. Scientifically, we have shown how to go in this direction, but we will have to find out whether it is indeed practical. Academia and industry must work more closely together.”
The AUTOSAR initiative is a step in the right direction, according to Broy and other experts, “but we need to go further. Yes, it will be very difficult and change is happening much faster than before.”
In the meantime, AUTOSAR is beginning to gain momentum. Daimler, for instance, will start to use the first few AUTOSAR components in controllers for the BR222, the next CLK, in 2010. From 2011 onward, the OEM will iteratively roll out AUTOSAR (based on release 3.0) in three steps. Beginning with the use in selected ECUs, Daimler will introduce the technology across all future lines to ensure the interoperability of the different bus systems including CAN, LIN, and FlexRay as they are all part of Daimler’s standard architecture electric/electronics (STAREE).