Car designers do it with wind tunnels. Architects do it models. But how do you test the design of a complex IoT system?
It isn’t easy with potentially thousands of sensors sending millions of readings continuously. How do you check to make sure that your IoT design will work properly in real life? Check to make sure that different error conditions are handled properly, and corrective action taken on time?
Simulated IoT devices can eliminate the guesswork from designing complex systems.
Dr. Michael Grieves at the University of Michigan first proposed the idea of a “digital twin”. A digital twin is a virtual representation of a physical device to make sure that the design performs as intended. He defined Digital Twin Prototype (DTP) as an asset’s information. Such as a 3D model of the device, its Bill of Materials and Processes to fully describe the asset and how it should work. These digital proxies generate sensor readings and communicate just as their physical counterparts and help IoT designers with:
- Visibility in the operations of the machines and larger interconnected systems such as a manufacturing plant or an airport.
- Predictive modeling to forecast the future state of the machine with physics and math based techniques.
- What if analysis to simulate various conditions that are impractical to create in real life and ask the what-if questions of the model.
- Understanding of the behavior of an individual machine or thousands of devices under different situations.
- Network design to see how to connect sensors with backend business applications such as supply chain operations, manufacturing and logistics.
Why is creating a realistic digital twin so hard? The sensor readings that it generates must reflect what’s going to happen in the real world and can’t just be a replay of a single set of readings. Consider someone’s heart rate for instance. Measuring it for five minutes and then replaying that recording isn’t adequate as it doesn’t consider physical activity, weight gain or changing temperature. A digital twin must accurately represent changing conditions.
Warehouse operation case study
Take for instance a warehouse with dozens for forklifts transporting hundreds of containers. Each forklift has multiple sensor reporting things such as location, engine temperature, tire pressure and brake wear. Some of the functions such as tire pressure and brake wear (for instance) impact each other. The usage of the forklifts fluctuates based on time of day and when large shipments reach the warehouse. Sensor readings are sent to the cloud for routing the forklifts and scheduling preventative maintenance so that operations are optimized.
For the IoT designer of this warehouse application, it’s extremely valuable to simulate the forklifts, communication bottlenecks and possible breakdowns to design the optimal solution.
Oracle’s virtual twin
Oracle helps tackle this challenge with a cloud-based, virtual representation of the physical devices needed in the design. The solution has two elements (as you would expect from twins):
- Digital twin which can include a description of the devices, a 3D rendering, and details on all the sensors in the device. It continuously generates sensor readings that simulate real life operations.
- Predictive twin which models the future state and behavior of the device. This is based on historical data from other device which can simulate breakdowns and other situations that need attention.
A designer can then use a digital twin to spawn dozens of ‘virtual forklifts’ and have them send sensor readings to a cloud based IoT monitoring system. Different error conditions can be simulated to see if alerts are generated properly and the supplies needed to repair forklifts are correctly ordered. This enable the IoT design to be stress tested and refined well before it’s deployed.
Kaoru Ishikawa pioneered quality management design processes. He proposed that proactively identifying potential failure points is essential for well-designed solutions. The design process should take into account the ‘whole system’ as well as contextual data. His Ishikawa diagram approach identifies problems, by analyzing the 5Ms: Man, Machine, Method, Material, and Management. The digital twin applies this 5M model in IoT by considering all five factors in designing the solution.
A wind tunnel for testing IoT devices may never materialize. But IoT designers can come close with some advanced planning and simulation with tools such as Oracle’s digital twin.