Model-Based Design can reduce costs, enable teamwork, and a more efficient development process, leading to a better end-product with a shorter development cycle.
It is no secret that automotive companies are working harder and harder to reduce their development costs. Just about everyone stands to benefit from any advances made in this arena, from the consumers to the automakers to the suppliers everyone would like to see these costs come down. With the rising costs of fuel, labour, and healthcare, costs that the automakers find difficult to control, they are aggressively taking steps to reduce the costs they can control, namely development costs. And we are not only talking about money – time is also a huge factor in this game. Let’s take a look at some of the factors contribute to those costs, and how we can work to offset them. To choose a specific example, we will look at developing an automotive transmission system.
The most obvious cost is the things you build that you cannot sell: prototypes. They are painfully expensive in the material cost and the time it takes to develop them. A single vehicle prototype for even a low priced car can cost upwards of 500x the final price of the vehicle. Having the ability to reduce the number of prototypes built would reduce development costs significantly.
All companies make mistakes. The difference between the successful ones and the unsuccessful ones is not necessarily the number of mistakes made, but rather when they are detected. Mistakes detected in the development process, for example in the writing of the specification, are far less costly to fix than mistakes detected in the last stages of development, for example as the ECU and the hardware are finally united. And let’s not even think about the cost of mistakes that are not detected until after the product is in the field. In a perfect world, our development process would prevent us from making these mistakes. Recognising that this is impossible, having the ability to detect these mistakes as early as possible in the development process would drastically reduce our development costs.
Someone has to design all of those fancy features and systems, and that not only involves a number of people, but multiple teams and sometimes multiple companies. The difference between the companies who thrive and those who simply survive is not always the number of people, but rather how effectively those people are used. The trick is to maximise the contribution of each member of the team towards the final product. Companies which succeed in making the product of the team effort greater than the sum of its parts are the ones left standing head and shoulders above the others.
So how do we achieve success with all of these things stacked against us? Is there a path that can reduce costs while improving the end result? Taking advantage of computer simulation is the first step towards this. Likely, this solution is already in use at your company, but the key is making sure that it is used as effectively as possible, and that is the main focus of Model-Based Design.
So what is Model-Based Design? It is the way to maximise the value of the contributions of the team of engineers at your company. It is designed to streamline the traditional development process so that all engineering effort is focussed on developing the final solution.
The traditional development process goes through four phases:
For an automotive transmission, engineers first write a specification to try and capture the requirements of the system, usually focussed on the electrical, mechanical, and hydraulic system as well as the controller and the performance the system must achieve. For an automotive transmission, this could include requirements of how quickly it can accelerate the vehicle to a specific speed. In the next phase, another set of engineers must attempt to interpret this specification and then come up with a design that will meet the requirements spelled out in the specification. Products of this phase of the development process often include hardware prototypes, those expensive items mentioned earlier. After that, another set of engineers will take this design and implement it. This traditionally involves hand-coding the control code, which is a further interpretation of the original specification. In the last phase, all of the individual components are finally united and tested against one another to ensure that the final product meets the requirements defined in the specification. The transmission will be tested to ensure that it meets performance, efficiency, and even durability requirements.
In the traditional development process outlined above, we essentially have at least three different versions of the same transmission. We have a specification, prototypes, and a final design. Computer simulation is used in each phase, but often in separate teams with each team re-interpreting the specification and re-developing tests to check their design. Designs are done at the component level (gearing) and perhaps at the subsystem level (transmission or controller), but because the teams are separated and working independently, the result can only be as good as the sum of the individual contributions, and will never be optimised as a complete system. More likely, the system will only be as good as the weakest element in the chain.
With Model-Based Design, instead of working on isolated teams, the teams work on creating computer models that can be integrated with one another at the system level, and that are directly connected to the specification. As the name “Model-Based Design” implies, the model is the foundation for the development process, linking all phases and teams involved in the development process.
In the specification phase, engineers need to have a closer link between the computer model and the specification to quickly be able to see how they correspond with one another. Ideally, the two would be bidirectionally linked, so that engineers can quickly see how specific requirements were implemented, and to see which requirements the specific pieces of the model cover. An example of this link is shown with The MathWorks Simulink Verification and Validation tool, which allows Simulink models to be directly linked with various forms of electronic specifications (DOORS, Word, and HTML etc.). This link to the specification allows engineers to continually check to see if the implementation properly covers a requirement, and to see if the requirements make sense.
In the design phase, engineers need to have the ability to try as many designs as possible and to quickly analyse them in order to find the best one. Creating computer models is much cheaper than making hardware prototypes, and provides the ability to easily change them and compare them. Using the Model-Based Design approach, the model created in specification process is enhanced to include more details and to test different designs. Ideally, the designs could be automatically compared to the specification in order to avoid tedious manual comparison. An example of this automatic comparison is shown in The MathWorks Simulink Verification tool, which automatically compares the results of the simulation with the specification and informs the user if requirements are not met. Automating the process of comparing the design to the specification allows test suites to be run around the clock, and reduces the amount of human error in evaluating a design.
In the implementation phase, engineers need to have the ability to implement the design without reinterpreting the work done in the design phase. Using the Model- Based Design approach, the model we have been developing is automatically converted into production code to be downloaded onto the automotive transmission ECU. This saves lots of time, and drastically reduces the human error that can be introduced by manually coding the design developed in the previous phase.
In the testing phase, engineers need the ability to test each component as well as the entire system. Once the ECU is programmed, ideally engineers would be able to first test that against the computer model of the transmission that was developed in order to ensure that the implementation still meets the specification, and to measure the changes that come with implementing it in hardware. Using Model-Based Design, the computer model of the transmission can be downloaded onto a real-time computer and connected to the ECU. This allows engineers to test the system without a prototype, enabling them to find mistakes earlier and without the expense of the prototype or the test track or testrig required to test a hardware transmission. The tests developed in the design phase can also be reused, again allowing us to avoid reinventing the wheel.
Reducing costs is critical to success, as well as having an efficient design process that produces a good end product. Streamlining the design process is a key step towards achieving this goal, and the best way to do that is to implement Model-Based Design. Having a set of tools that enable different teams of engineers to work together and developing a model that can be used throughout the entire development process is the best way to ensure that you maximise the contribution of all engineers on the development team, while reducing redundant work, human error, and costly prototypes. While nothing can guarantee success, this formula will certainly maximise the chances that your product will be produced on time with the lowest cost and the highest quality possible.
Keywords: Model-Based Design, The MathWorks Simulink Verification too
