Model-Based Design Accelerates Development Cycles
When you need to boost performance, reduce operating costs or investigate design space, model-based design is a powerful tool. With model-based design, virtual models are at the center of the development process and help you shorten development cycles and substantially reduce overall development costs.
In the systems development cycle, the V-model is usually shown as a diagram:
On the left, the project is defined, and the team verifies whether the concept meets the requirements, such as regulations, specifications and customer demand. The different subsystems are then implemented and unit tested before they are integrated and the validation can be performed. This means that a big part of the test and validation waits until the very end of a project, potentially leading to huge redesign costs when problems are identified at the end of the process. Model-based design is essential to moving as much test and validation as possible into an earlier stage of the process.
For the past two decades, the system-modeling team at Wolfram MathCore has focused on solving tricky problems with the help of model-based design. Sometimes team members were brought into projects early in the development process, when concepts were being tested, but more often they were called at a later stage, when fatal failures had occurred during operation or new requirements emerged, and a quick fix was needed.
One case of a late redesign occurred when one of MathCore’s customers, who had developed a 180-kilogram (397-pound) drone, signed a contract for drones operating from ships. The drones were supposed to start and land autonomously from the ships. However, the system had not been initially designed for this, which raised many new concerns. Would the drones be able to cope with the potential high-impact landings? How should the landing algorithm be updated? Would the control system be able to handle the task?
One solution would’ve been to run numerous landing tests and adjust the design as needed. However, the obvious risk was that the drones would crash during testing. On top of this, waiting for the right weather conditions for the extensive tests needed would take a lot of time and money. Instead, MathCore was asked to build models for the drone, the ship and the sea, and to use these models to test the system virtually.
Once the model was developed, the team was able to plug in the drone’s landing algorithm and control system, and run hundreds of simulations in any sea conditions. The simulations quickly identified deficiencies in both the landing algorithm and control system—flaws that would almost certainly cause crashes in somewhat rougher weather conditions.
Using this virtual environment, both the landing algorithm and the control system could be tested and improved; even the drone design itself was updated as a result of this testing. The likelihood of successful landings in different conditions could be calculated. Considering external factors like wind and cruise speed as well as the stochastic behavior of the waves, guidelines for safe operation could be created. Once safe limits had been found, real-life tests could successfully be performed to validate the virtual system.
While using model-based design from the start of a project is ideal, this case demonstrates how it can also be applied successfully to reduce development time, cost and errors in later stages of the development cycle.