While a software tool supplier may want to deliver a complete set of development tools, automakers and tier one electronics suppliers usually wind up mixing and matching the capabilities of several suppliers. As a result, cooperative competition links suppliers of simulation, verification, code development, production code generation, hardware-in-loop, and test and calibration equipment. As the entire automotive industry struggles to rapidly realign with the changing consumer requirements, cooperative efforts are more important than ever.
Changing Consumer Requirements
With $4.00 and higher gasoline prices, US car buyers have dropped the SUV and other high fuel consumption vehicles like a bad habit and finally reversed the trend of more trucks than cars. Carmakers with high mileage vehicles have had increased sales but those companies that relied heavily on SUVs and trucks, such as GM, Ford and Chrysler have been hit hard. Total sales for the Detroit Three for 2008 vehicles have plummeted.
In spite of the current financial situation, carmakers have to develop vehicles that people are going to buy. Today, one of the first things on the “must-have” list is fuel economy. With accelerated or entirely new programs to satisfy current customer requirements, design and development tools have become more important than ever.
“So far, I haven’t seen a whole lot of hesitation on buying tools because of the need for hybrids and fuel economy, which has been growing,” said Kevin Kott, President, dSpace Inc.
In fact, the rapidly changing environment has accelerated the acceptance and spread of software tools, reinforcing the reasons for adapting these tools. “One of the reasons is because design cycle times are changing, they are getting shorter,” said Paul Latiolais, Marketing Manager, Synopsys, Inc. “The other aspect of this, related to it, is that consumer demands are changing more rapidly than they use to.” With even shorter design windows, design tools can shave time from the design and development schedule. There are additional reasons for implementing the latest design tools.
“The introduction of a hybrid adds a whole new level of complexity in the electronics area that is really, really driving a high demand for tools,” said Mark Jensen, product line Mmanager, Process Tools (and team leader, Customer Training), Vector CANtech. “A lot of our new business is coming out of these newly formed or at least newly expanded hybrid groups.” The need to change quickly and added complexity have brought automakers to tool suppliers with very specific requests.
“One of the big things they are looking for is the ability to integrate multiple development tools that have not previously been linked,” said Jensen. Modeling new control algorithms in one tool and then modeling network communication or control strategy in another tool is a common practice. Now engineers are looking for ways that these tools can communicate with each other in an automated way and take what had previously been segregated development activities and bring them together in a single combined development strategy. Since the tools are generally from different companies, it means that these tools have to work along common interface boundaries. That does not mean that tool suppliers have stopped trying to provide as much of the total solution as possible. Figure 1 shows how Vector’s eASEE project and process management tool addresses data management in the V model for software development.
“What we have seen just over the last year is multiple customers come to us and ask us for the end to end solution-- the interoperability of everything from developing your requirements to developing the systems engineering, the development and release of the components but all the way up the validation side of the V,” said John Cain, vice-president, sales and marketing, Vector CANtech. The key is interoperability.
“There seems to be a lot of distaste right now for tools that are standalone,” said Jensen. “If the tool doesn’t interoperate with other tools or doesn’t at least have very good data exchange capabilities with other tools, it tends to be frowned upon.” Automakers and tier one suppliers have high expectations for new tool and new tool functions to support integration of the tools that are already in place. One of the cooperative efforts to ensure interoperability is shown in Figure 2. Vector and The MathWorks worked together so that Vector’s CANape and MathWorks’ Simulink can exchange simulation data
In the current economical environment, cost is more important than ever and the right development tools can reduce costs as well as development time. “In the data management, we’ve seen reductions in costs in their process systems of everything ranging from like 20 to 90 percent,” said Cain. “From end to end, just a guesstimate, might be a 40 to 50 percent reduction in cost.” The cost savings appears to correlate well with the time savings in the overall process.
Carmakers, especially in specific departments, are well aware of the capability that development tools bring in terms of cost and time savings. Reporting the results from using these tools has become an integral part of SAE papers and industry forums. At the annual 2008 SAE Congress, GM reported on virtual manufacturing efforts and distributed computing (SAE 2008-01-0288, Designing Automotive Subsystems Using Virtual Manufacturing and Distributed Systems) as well as variable valve timing (SAE 2008-01-0901, Modeling and Simulating a VVT System for Robust Design).
Both of these projects used Synopsys’ Saber simulator, a multidomain or multi-physics (electrical, thermal, hydraulic, mechanical and more) modeling and analysis tool. In one case, the authors concluded that the GM powertrain and its suppliers who participated in the Signal Delivery Subsystem (SDSS) modeling and simulation process reported reductions in both development time and costs. In the other, it was noted that compared to hardware prototypes, “simulation, using a virtual prototype is a vastly more practical and useful alternative due to the time and cost considerations involved when making multiple permutations.”
In another SAE paper (2008- 01-0085, Model-Based Design for Hybrid Electric Vehicle Systems), several MathWorks’ engineers showed how model-based design using Simulink and other tools can be implemented in the design of a hybrid electric vehicle. Figure 3 shows average and detailed models that were developed and tested. The authors concluded that “continuous validation and verification of requirements throughout the design process reduced errors and development time.”
Hybrids are just one of the approaches to increase fuel economy and reduce reliance on foreign oil. Diesel engines are the solution for higher fuel economy in Europe and certainly will be part of the mix in the United States with clean diesel fuel available now and ongoing efforts to limit particulates. Flex fuel systems with E-85 Ethanol / gasoline mix are another solution from several carmakers. “All of these are changes and all of these have to rapidly be put into the cars and if you rely on traditional systems like physical prototyping then you can be in trouble,” said Synopsys’ Latiolais. “I don’t think there is any way to do it now with just physical prototyping. Those days are gone.”
Weight – There’s More
Reducing the amount of copper in the vehicle especially in the wiring harness reduces initial cost and improves the fuel economy. In addition to increased steel costs, copper costs have also risen dramatically. The cash price of copper has already risen 31 percent this year. According to some reports, the average weight of a vehicle wiring harness was about 80 pounds a couple of years ago and was forecast to double in five years given the current trend. Greater utilization of the networking capability could reverse that trend.
“People are using a lot more bus structures in the vehicle than would be necessary if they actually were utilizing all of the bandwidth of available structures,” said Serge Leef, general manager of the System-Level Engineering Division, Mentor Graphics. “There are actually internal rules inside some large American companies that you cannot put any more traffic than 30 percent on a CAN bus,” he said. The low utilization rate increases the amount of copper and adds to the vehicle’s total weight. Eliminating unnecessary weight contributes to better fuel efficiency in any size vehicle, no matter what type of powertrain is used.
In Europe, some companies use tools and embedded software, from Mentor Graphics and others to drive the utilization of onboard buses to much higher levels. “Our software enables utilization of upwards of 90 percent,” said Leef.
Mentor Graphics’ Volcano Network Architect is a system design and architecture tool that allows the design of a top down network by declaring all the ECUs, all the messages and all the rules associated with messages. Using the tool to describe the system, Volcano automatically generates the layers of software that manage the communications on the actual ECUs.
The other part of the Mentor Graphics solution is Volcano Target Package (VTP), embedded software that gets loaded into every ECU. It provides an interaction layer that insolates the applications developer from the actual mechanics of the communications and allows better utilization. A bus utilization of 90 percent versus the typical 30 percent found on most of today’s vehicles would reduce the weight of copper and increase fuel economy.
Architecting the Complete System
While powertrain changes including hybrids and clean diesel technology that meets US standards are some of the more obvious areas to expedite, they are not the only ones. Other vehicle differentiating technologies, such as infotainment and body electronics, as well as business as usual hardware improvements need to change fast and be correct the first time, too.
“We have seen a lot more focus on simulating systems and simulating multiple ECUs on networks,” said Vector’s Cain. “This obviously cuts down the time to market because it reduces the amount of prototyping work that goes on in the development side of the V.”
One bus architecture improvement that needs to be implemented is the use of the FlexRay protocol for more bandwidth and/or deterministic/ fault-tolerant operation. Hardware-in-the-loop (HIL) simulation and evaluation can speed up this integration. Figure 4 shows dSpace’s approach for FlexRay HIL analysis.
Besides adding a new bus, reducing the number of ECUs communicating between each other is another highly desirable automotive systemlevel change. With 40 or 50 or even 60 ECUs, the communications between them is at or possibly even beyond a reasonable limit. “There has been a fairly big effort to try to minimize the number of ECUs and to do that, there is a whole bunch of architectural problems,” said dSpace’s Kott.
dSpace System Desk allows the software architect to start with the overall architecture of the system and develop the interconnections between the ECU hardware and software modules. This allows mapping of software modules to ECUs based on required functionality and performing simulations to optimize the number of ECUs.
No Yielding Ground on Quality and Reliability
Perhaps the biggest differentiator for automakers is quality. While the goal of development tools is to improve quality, one company, Mentor Graphics, recently announced that its virtual prototyping tool, SystemVision, supports Design for Six Sigma (DFSS) methodologies to achieve cost-effective design innovation by a modeldriven development process. Combined with model-driven development, DFSS methodologies can result in significant improvements in both productivity and quality with virtual prototyping, automated data collection, and statistical analyses used in the modeldriven development process. That model-driven process is increasingly collaborative.
Synopsys virtual platforms form the basis for extensive collaboration in the design process. Figure 5 shows the different design tools that they offer for different portions of the design cycle and examples of the collaborating companies. This level of collaboration can identify quality problems early in the design process. “We allow the semiconductor vendor to get feedback from the rest of the design chain very early,” said Frank Schirrmeister, director, product marketing, System-Level Solutions, Synopsys, Inc.
How much time can be saved in automotive design using this approach is difficult to qualify exactly but there are ways to estimate the savings. “We have seen in the wireless world cycle time reductions up to 9 to 12 months,” said Schirrmeister. In automotive, he expects more complicated chips like multimedia ICs to have similar results so simulation and analysis can be performed as much as a year before the actual silicon is available using the virtual platform.
Standards Support Collaboration
Interoperability can occur without specific collaboration between toolmakers. Standards activities from ISO or SAE or automotive-specific private consortiums such as AUTOSAR or the Association for Standardization of Automation and Measuring Systems (ASAM) allow tool vendors to provide implicit integration by supporting these data and interface standards. “So even without knowing a lot about the other tools that are out there, if you know you are conforming to standards and implementing them correctly, there is a very good chance you’re going to have some level of interoperability right out of the box,” said Vector’s Jensen.
But collaboration is pervasive. Perhaps one of the more interesting demonstrations of industrywide cooperation is a user group meeting, such as MathWorks Automotive Conference where over 20 software tool and test and development companies participated. The pervasive use of MathWorks MATLAB and Simulink tools as well as MathWorks collaborative efforts, such as their Connections program, support cooperation with third party solutions. Some of the papers presented at the conference demonstrated the interoperability that can be achieved. In one paper, ASAM-MBFS: A Standardized Block Library as Enabler of Efficient Model- Based Collaboration, co-authored by engineers from Continental and Audi, the authors concluded that the ASAM- Model-Based Function Specification (MBFS) provides a standard for efficient exchange of functional models representing ECU algorithms or executable specifications. “Models can be transferred between different modeling tools and between different companies and organizations,” they observed.
The role of tools and automakers’ attitude toward software development is certainly changing. “The auto industry is really starting to shift where a lot of the intellectual property is in the models, the features that they have got to create,” said Jon Friedman, automotive industry marketing manager, The MathWorks. Noting the importance of getting to market first with innovations such as Ford’s Sync or GM’s OnStar, he concludes, “It’s not hard to copy, but if you get that head start on people, you get a real advantage.”