If time really is literally money – thinking in the region of € 10,000 or more per day – how do you ensure a new control system of a machine can be put into action as quick as possible? And, after installation production can start up as quickly (and also as safely) as possible? By testing the control system in advance. Virtual that is. This can be done through an HIL test. What HIL stands for? For Hardware In the Loop.
Specifically, we are talking about a ‘machine’ at Marin. A marine research institute where ship behavior is tested on models in an enormous basin. Above that basin, a measuring carriage, packed with equipment, follows the movements of the model. In this case, the measuring carriage is the machine. VSE Industrial Automation designed the new control system and works together with Controllab Products on the testing of the controls of this measuring carriage. They do this using Hardware In the Loop, or HIL testing.
We went to Enschede to be a live witness of the initial testing. We wanted to get close up to see how virtual testing is done. Jan van Vuuren and Leendert Uithoven from VSE are here together with Christian Kleijn and Paul Weustink from Controllab to tell us all about it. Firstly, about what an HIL simulation is. “Hardware-In-the-Loop simulation is a technique that involves machine control systems being tested by linking them to a simulated machine. It is a very effective method of testing control systems under standard conditions, but also under conditions that you normally only encounter in emergencies. You can stretch the limits of whatever your testing as much as possible because if something does go wrong, the problem is only within the simulation. You can never test a control system to its limits in a real situation if there is a chance that the model will crash and that the equipment damaged. With an HIL simulation you can define the actual limits and set up the control system such that the trial run stops before any damage is incurred. The control system therefore becomes safer.”
The control system cannot see the difference between whether it was a ‘real’ or a ‘fake’ test. In the simulated tests it reacts in exactly the same way as in real tests and the control system can therefore be perfected for the ‘real’ scenario. This ensures a time saving of four weeks for the measuring carriage and the deployment of it – meaning less downtime, and we haven't even mentioned the fact that it can be tested much extensively.
However, it is not only the operation of the control system that is tested, the setup of the measuring carriage is also tailored to the ideal situation. Uithoven: “There are operators working on the measuring carriage. It is really important for them that the control system is not adjusted too strictly – that wouldn’t be comfortable for working. However, if the control system is configured too softly, then the measuring carriage again doesn’t follow the model adequately. We can therefore now find the ideal setting beforehand. Simulated tests can be repeated endlessly in a short space of time with guaranteed identical test conditions. The real tests must be 100% identical to the simulated test. Controllab made sure of that.”
Weustink explains what is involved in simulating the ‘machine’: “We provide hardware with real-time machine simulators and services such as HIL testing based on FMEA – Failure Mode and Effect Analysis. This way we can provide Marin with certainty in advance about whether the control systems meets their requirements. In the computer we simulate the dynamic behavior of the ship model, the measuring carriage, the sensors and the actuators. The model calculates five hundred times per second and communicates with the control system one hundred times per second in real time. How something moves in a simulation, must of course match how something moves in reality. To be able to determine this and program it properly, requires a great deal of physics knowledge. How, for example, a model responds to currents and what the characteristics of the model are, must all be included. And on top of that you also need to include human factors in your testing. Operators might sometimes do something completely different to what you are expecting. And is the operating sequence that seems so logical, logical for everyone? These are all things that need to be properly documented. Where is the test valid and where not? Which parameters are used? Which specifications are used?”
Of course you learn most in testing if something goes wrong. So the data you get during testing if an error occurs is very valuable data. Uithoven: “You are then able to look at where and why something went wrong. And what you want to do with that data. Does something need to be changed? Are there rules that need to be adjusted? Was it an unexpected error or an expected one? And simulating the operation also provides an insight into whether the measuring carriage can really do what is expected. Or maybe even more.”
There are several monitors and a control box on a desk. Uithoven and Weustink are sitting behind the desk – ready to perform the tests. One of the screens shows a 3D simulation of the reality, a basin, a measuring carriage, a model and operators. The other screens show graphs, a control interface and data. After selecting a scenario and pressing a few buttons, the first test can take place. As spectacular as a ‘real’ test can be, as boring is a simulated test. Mind you, that is for the unsuspecting spectator. For the testers it is, of course, highly interesting. The collected data is evaluated and documented and another scenario is tested soon after. And this is how it is possible to achieve a lot in a short space of time. In a much shorter period of time than in reality. In a basin you would have to shut down the basin each time and also take the model and the measuring carriage back to the starting position before you can run a new test. Not to mention the operators that are required and the fact that in reality you cannot test anywhere near everything you would want.
Kleijn: “Simulating situations and mimicking destructive situations is also widely used in the aircraft industry. Just think of the air crash investigation tv show. If you can find out what caused a particular error, you can stop it from happening again. The advantage of this to the aircraft industry is of course obvious, but other industries are also using this technology more and more frequently. You can also greatly improve the performance of your products. A great deal of simulations are, for example, used in the development of autonomous cars.”
Van Vuuren does not dare to say where the technology will end. The acceleration of technology is continuously taking an unexpected direction. What the future holds remains exciting. Controllab and VSE have found each other in this new test method. Kleijn: “We have shared a lot of knowledge along the way and have learned a lot from each other. We both see great potential for these simulations. Having simulations as part of the design process helps to save a lot of time throughout your whole project and can also help make your results foolproof. Quite pleasing, actually.”