Model updating of coupled hydraulic-mechanical systems

This project deals with models of hydraulic drive systems which appear in construction machinery, agricultural machinery, machine tools, steel mills, offshore platforms, or traction drives. Such systems consist of hydraulic and mechanical parts that are mutually coupled; for a prediction of pressure pulsations and mechanical vibrations, accurate coupled hydraulic-mechanical models are required. This project shall demonstrate how experiments can help to improve dynamic models of coupled hydraulic-mechanical systems. Although it has been attempted to calculate the mechanical vibrations of pipelines that are filled with a pulsating fluid, the results sometimes failed to agree with measurements. On the other hand, the respective models contain parameters that are not known exactly. It is obvious to ask how parameters can be found to obtain an optimal match between theory and experiment; this is the purpose of model updating. Experimental data for model updating shall be provided by experiments in which the coupled hydraulic-mechanical system is excited by a hydraulic or a mechanical device, such as a valve or an impact hammer. Pressure pulsations and mechanical vibrations shall be measured, the transfer behavior shall be calculated in the frequency domain, and resonance frequencies shall be identified; coupled hydraulic-mechanical vibration mode shapes shall be extracted by modal analysis. Resonance frequencies and mode shapes shall also be calculated from theoretical models and compared with experimental results. Using the sensitivities of theoretical models with respect to parameter changes, optimal parameter values shall be found by minimizing the error between theoretical and experimental resonance frequencies, mode shapes, or other dynamic characteristics. If necessary, model structures shall be extended to obtain more appropriate system descriptions. Although model updating has been developed for mechanical systems and is frequently applied in structural dynamics, the concept is new for hydraulic and coupled hydraulic-mechanical systems. Two case studies shall demonstrate the process of model updating. The first case study treats the torsional vibrations of a pump drivetrain in combination with the pressure pulsations of an attached hydraulic system. The second case study investigates fluid-structure interaction in hydraulic pipeline systems; by an extension of the model structure and subsequent model updating, successful predictions are expected where models failed in the past.

In this project, three different experiments were carried out with coupled hydraulic-mechanical systems. In each case, the dynamic behaviour of the system was modelled in a frequency range up to several hundred Hertz. Many parameters of those models were not known in advance and should be identified from measurements. In the first experiment, four hydraulic cavities were connected by short pipelines and coupled to a mechanical two-mass-oscillator which was mounted on the flexible cover of a cavity. Hydraulic and mechanical excitation could be realized by a servo-valve and an impact hammer, respectively. Pressure and acceleration responses could be measured. The scope of this assembly was purely scientific, it was called a "research machine". The basic dynamic model was obvious and the mechanical part could be removed for a separate identification of the purely hydraulic system. After moderate results with a first optimization strategy and some model extensions concerned with damping, a successful two-step optimization procedure was implemented. This procedure provided physically reasonable values for all identified parameters and lead to an excellent agreement between calculated and measured results, which was robust with respect to physical changes in the assembly. The second experiment dealt with the torsional vibrations of a pump drivetrain in combination with the pressure pulsations of an attached hydraulic system. By a simple model, a pressure pulsation compensation effect was predicted, which could be confirmed by measurements. This experiment lead to an appropriate patent application. In the third experiment, an oil-filled pipeline elbow was investigated. One pipeline end was connected to an accumulator, and the entire assembly was hung up on elastic strings. After filling and pressurizing, the pipeline elbow was disconnected from supply. It was excited by impact hammer blows in various locations and directions while pressures, accelerations, and strains were measured. Parameter identification was tackled with the two-step optimization procedure developed for the first experiment, which was complicated by the fact that the assembly possessed a large number of mechanical resonances. Nevertheless, a fair agreement between calculated and measured results was achieved. Overall, the project paved a way for accurate dynamic models of hydraulic drive systems. Such models are required for the analysis of energy consumption, which will be a necessary contribution in the development of energy efficient hydraulic systems.