Marine propulsion system provides power for the sailing and operations of a ship, and can be treated as the heart of a ship. It has long been the main focus of marine engineers and researchers. After decades of development, torsional vibration still remains as one of the main problems in a marine propulsion system. In fact, torsional vibration is not only a problem of the propulsion system. Due to the fact that it can be easily transferred from the engine to other systems, via. the drive train, torsional vibration also harms the other systems in the ship. How to control and reduce the vibration of the propulsion system is a topic that the whole ship engineering community concerns for years, and is exactly the motivation of the present thesis.
The fast-developed digital twin concept is one of the most popular topics in recent years with the discussions of new Industry 4.0 and Industrial Internet of Things (IoT). The digital twin technique has also greatly developed to be able to handle problems in diverse industrial applications.
The fundamental idea of a digital twin is to build a virtual replica of a physical asset/ system with respect to dimensions, material properties, working environment, and even real-time interactions with external conditions. In other words, the digital twin not only duplicate the appearance of its physical twin, but also shares all realtime measurements / conditions, and reflects exactly what the real physical twin experiences. The ability of digital twin to reflect the most real status of a system / asset is nothing a conventional simulation in modern CAE (computer assisted engineering) can compare. In conventional simulations, no real-time input/output can be achieved and the predictions are always lagged behind the real situation.
In view of all the appealing advantages of the digital twin concept, we have decided to make the try and combine our problem (vibration of marine propulsion system) with digital twin. The aim would be to explore the possibilities in this field.
A test rig in the machinery lab of NTNU was used as the main object of this thesis. This test rig is a simplified marine propulsion system. It consists of a motor, a gearbox, a shaft, and finally a generator which represents a propeller in real propulsion systems. Based on all the detailed parameters of this test rig, I built a physical based digital twin model in Ansys Twin Builder, and studies the speed control and vibration of the system.
It is worthy to mention that Ansys Twin Builder is a rather new package launched by ANSYS earlier this year (January of 2019). It is a brand new package for digital twin building applications, and quite ambitious for providing digital twin solutions. A challenge therefore also lays in properly using this new tool without any earlier experiences. This exploration has taken a majority time during this semester and eventually a proper test rig digital twin model was built up in Ansys Twin Builder.
Firstly, a system-level model of the test rig was built, including the motor, gearbox, shaft and the generator. Then I have designed a speed control system for the motor, as well as a torque control system for the generator. Based on this model, we are able to apply simulations. In the simulations, I observe that the motor speed is able to achieve the rated speed in about 4-6 seconds and keep a stable speed afterwards, so is for the generator. This clearly indicates that the built digital twin is reliable under the speed and torque control system. A modal analysis of the system was conducted, and the natural frequency of the system was obtained. The obtained nature frequency was compared with the system frequency got from the speed response of both the motor and generator, which verified the model.
Based on this verified model, I did some sensitivity analysis to study how different parameters impact the speed response. The investigated parameters include shaft stiffness, moment of inertia of the generator rotor, gain in PID controller (part of the speed control system), etc. The change of natural frequency of the whole system is clearly observed as either the shaft stiffness or the moment of inertia of the generator rotor changes. By applying different values of shaft stiffness and the moment of inertia of the generator rotor, strong vibrations were observed when the induced nature frequency of the system gets close to the rotational frequency of the generator, due to resonances. Therefore, in order to reduce torsional vibration of the whole system, the shaft stiffness should be set to values that the consequent system nature frequency is far from the generator’s rotational frequency.