Riassunto analitico
The objective of this activity, carried out in collaboration with the company CNH Industrial, is to create a Digital Twin of the front axle suspension of a medium-range tractor, in this case a New Holland T7 LWB CVT STAGE V, which means it is the long wheelbase version, with a CVT transmission and it is the last model created which respects the last regulations in terms of emissions.
We started our activity creating a multi body model in Simcenter 3D, the software for the dynamic modelling, of the whole tractor, obtaining a high-fidelity model. We modelled with precision the tires, which are very important to define the correct dynamics of a machine, especially of a tractor. The front axle suspension of this tractor is of hydraulic type, with a Panhard system. This hydraulic system has been implemented exploiting the software Amesim.
To control and handle the models built in these two different software we used MATLAB, in particular Simulink, which served as our control system indeed. We then created interface blocks in Simulink for the multi body model from Simcenter 3D and for the hydraulic model from Amesim, creating a so called co-simulation. In this way we can control the simulation from the MATLAB environment, making also Simcenter 3D and Amesim working together. In this system the hydraulic model receives real inputs, which control the opening and closing of the valves, and gives as output a force, which is fed to the dynamic model, as force acting in the suspension cylinder. The dynamic model then returns signals of velocity and displacement of the cylinder, which are sent back as inputs for the hydraulic model, obtaining a closed loop system. The digital twin is now completed, we need to check and validate it.
The maneuver in which we tested our Digital Twin is a levelling one, where the tractor is still and the suspension cylinder goes from its nominal medium position, to the fully compressed and the fully extended one, then returning to the medium position. We compared the results we got from the dynamic model in terms of displacement of the suspension cylinder with a reference coming from a real measurement.
We noticed that our model follows very well the real reference, the two displacement curves are very similar, but only when the accumulators valves of the hydraulic system are closed. When these valves are opened and the accumulators are taken into consideration, the behavior is not good, our model cannot follow the reference and the simulation presents an error and stops. This is due to the fact that the data we feed the accumulators’ valves with are not coincident with the signals that drive the other valves. A time offset is present between these signals, making impossible the simulation considering the accumulators.
We also evaluated the creation of a Reduced Order Model (ROM), with the Ansys Twin Builder tool, to reduce the simulation time of our model. This analysis is just at its very beginning, but for that we could see, the ROM at this point of development is not very good for describing our system, which is not linear and can be subjected to many different excitations. The ROM is very fast in terms of computational time, but not accurate with different inputs, especially in representing the upper and lower end of the suspension cylinder stroke, which are often overcome.
|
