Return to search

Multi-scale multi-physics model and hybrid computational framework for predicting dynamics of hydraulic rod seals

Rod seals are one of the most critical components of hydraulic systems. However, the fundamental physics of seal behavior is still poorly understood and the seal designers have virtually no analytical tools with which to predict the behavior of potential seal designs. In pursuit of a comprehensive physics based seal analysis/ design tool, in this work, a multi-scale multi-physics (MSMP) seal model is developed. The model solves the transient problem involving macro-scale viscoelastic deformation mechanics, macro-scale contact, micro-scale two phase fluid mechanics in the sealing zone, micro-scale asperity contact mechanics and micro-scale deformation mechanics of the sealing edge in a strongly coupled manner. The model takes into account surface roughness, mixed lubrication, cavitation and two phase flow, transient squeeze film effects and the dynamic operation as well as the effect of macro/micro/nano scale viscoelasticity. A hybrid finite element-finite volume-statistical computational framework is developed to solve the highly coupled multi-physics interactions of the MSMP model simultaneously. Surface characterization experiments are performed to extract the parameters like RMS roughness, asperity density, autocorrelation length and asperity radius needed by MSMP. To remove the high frequency noise without removing the high frequency real surface features, a wavelet transform based adaptive surface extraction method is implemented. Dynamic mechanical analysis (DMA) is performed to extract the macro-scale viscoelastic parameters of the seal. Through atomic force microscopy (AFM) experiments, the local micro/nano scale elastic moduli were found to be varying within two orders of magnitude higher than the bulk of the polymer. Significant differences in local stiffness, adhesion and the relaxation time scales of individual surface asperities were also observed. With the MSMP model, dynamic seal performance was analyzed. The results confirmed the mixed lubrication and the effect of surface roughness. Thicker fluid films during instroke and cavitation during the outstroke were found to be important for non-leakage. Seal behavior was a function of the complex dual dependence on the time varying sealed pressure and hydrodynamic effects. Viscoelasticity is seen to critically affect the leakage and friction characteristics. It produces thicker fluid films and produces a significant increase in Poiseuille component of flow during instroke. Ignoring viscoelasticity leads to under-prediction of the time required to reach the zero leakage state. Several high pressure - high frequency sealing applications were analyzed. In such applications, a new phenomenon of "secondary contact" was observed. Viscoelastic creep was seen to critically affect the contact pressure and hence the friction characteristics. In high frequency applications, viscoelasticity induced significant differences in Poiseuille flow and friction force from cycle to cycle. Cycle frequency was seen to play an important role in governing visco-elastohydrodynamics and the leakage of such seals. The seals need to be designed by considering the relationship between relaxation time scales of the polymer and the cycle frequencies. Study also revealed the presence of characteristics like "critical temperature" and "critical frequency". Using the multi-physics modeling capability of MSMP framework, several novel seal designs using smart materials like piezo-ceramic embedded polymers are proposed and analyzed. The MSMP computational framework developed here has a great potential to be used as a stand-alone seal design and analysis software in academic and industrial research.

Identiferoai:union.ndltd.org:GATECH/oai:smartech.gatech.edu:1853/37272
Date25 October 2010
CreatorsThatte, Azam
PublisherGeorgia Institute of Technology
Source SetsGeorgia Tech Electronic Thesis and Dissertation Archive
Detected LanguageEnglish
TypeDissertation

Page generated in 0.002 seconds