Experimental Study on the Wear-in Behaviour of Brush Seals

In addition to increasing the thermodynamic parameters, such as the turbine inlet temperature and the compressor pressure ratio, the efficiency of turbomachinery can be significantly increased by reducing losses in the secondary air system. One possibility to reduce these losses is the use of brush seals as a replacement for conventional labyrinth seals. Their most valuable advantage over other sealing concepts is the very small gap between the bristle package and the rotor and thus a reduced leakage mass flow. This small gap can be achieved due to the great radial flexibility of the bristles without running the risk of severe detrimental deterioration in case of rubbing. Thanks to the flexible structure of the brush seal the contact forces during a rubbing event are reduced, however the frictional heat input can still be considerable.

Although some investigations on the wear behaviour of brush seals can be found in the literature, not all aspects are yet understood during the wear-in process of the seal in operation. However, it is important to understand all aspects in order to ensure a long service life of the seal. The present paper gives an overview of the procedure, the results and the knowledge gained through the long-term rub tests carried out with a tungsten carbide coated rotor with the aim of determining the time required to wear the seal in ("time-on-condition"). The total test duration of the cyclic tests was approx. 100 hours. The long-term measurements have shown that on the basis of the applied test procedure the initial wear-in phase is completed after approx. 40 cumulated operating hours and only very little wear occurs afterwards. Furthermore, it could be shown that the major part of the wear occurred within a short period of time within these 40 hours and that the properties of the test cycle have a great influence on the wear-in time.  The coupling of the rubbing intensity with time is achieved by measuring the transient rotor temperatures during rubbing. Contrary to the general assumption, during the measurements an increase in the leakage mass flow could not be observed. The leakage through the seal tended to decrease over the course of the measurements.

The tests carried out have also shown that the influence of the concentricity properties of the rotor can have a major influence on the results. In future test campaigns the influence of the presumed deviations of the rubbing temperatures caused by the concentricity deviations over the circumference ("hot spots") will be examined more closely.