Aeroengine Safety

Figure "Brush seal distinguishes from labyrinth seal": Because brush seals are seen as competing with the widely used and proven labyrinth seal designs, it is important to have a clear understanding of the sealing principles of both. This is true for new constructions as well as replacing labyrinth seals with brush seals in an engine in serial operation.

Seal effectiveness (top diagram): With the same pressure ratios, brush seals can replace labyrinths with several chambers (generally up to five; Figs. "Brush seal limits" and "Brush seal factors determining effectiveness"). The elastic deflection of the brush bristles causes their seal effectiveness to decrease more slowly over pressure cycles than is the case with labyrinths.

Pressure differences (middle diagram): The maximum possible pressure differences with brush seals depend on the strength of the brush and its pressure-dependant stiffness (Fig. "Brush seal limits"). With large pressure differences, tandem brushes are used (Ref. 7.3.1-10).

Turbulence losses (Ref. 7.3.1-14): With a comparable pressure difference, brush seals transfer far less air friction energy into the leakage flow (windage heating) than labyrinth seals. However, since the amount of leakage air is much less than in labyrinth seals, a large temperature increase in the leakage air and the brush in the gap area must be expected.

Mounting and use (bottom diagram): Brush seals can be mounted and exchanged as machine parts (Ref. 7.3.1-10). They are not an integral part of a larger engine part (e.g. turbine disk), as labyrinths commonly are. Brush seals, with the same pressure differences as labyrinth seals, take up considerably less axial space. On the other hand, brush seals are radially taller. The design of a brush should ensure as little wear of the opposite rub surface as possible. Rub-in that could be compared to that of labyrinth seals does not occur.

During overhaul, brush seals are replaced. Unlike labyrinths, they are not repaired. As with radial seals, if necessary, the contact surface must be repaired (renewed) by recoating or regrinding it.

Incitement of rotor vibration: Unlike labyrinth seals, tandem brushes do not incite shaft vibrations (Fig. "Brush seal influencing dynamic behaviour"). With tandem brushes, the circumferential ducts located on the stator are between the fins of a labyrinth, whereas the rotating ducts are not. This difference is very important. In a labyrinth, the air flow in the ducts is accelerated, whereas it is slowed in the brush seals. The speed of the air flow has a strong influence on effects such as air vibrations. The nearer the circumferential speed of the leakage air comes to that of the rotor, the higher the risk of vibrations being incited.
The decelerating effect of tandem brushes on the circumferential components of the leakage air flow prevents aerodynamic vibration incitements.

Figure "Brush seal specific terms": There are many technical terms used with brush seal technology. These are often different terms with the same meaning, or describe certain brush-specific properties.
The two bottom diagrams depict two different constructive layouts of brush seals. In the left brush, the bristles were affixed by welding. This configuration is smaller, but experience has shown that it is difficult to ensure problem-free bonding of the bristles. This problem was prevented in the right brush.

Figure "Brush seal limits" (Ref. 7.3.1-11): The bottom right diagram depicts the typical dependence of the seal effectiveness of a brush seal on the effective clearance. Effective clearance is the actually occurring leakage air flow. It is referred to as “effective clearance” because the leakage air volume depends not only on the geometric clearance cross-section, but also on other factors, such as the packet density, bristle diameter, length of the free bristles, “blow down” (also “pressure closure”), i.e. the bending of the bristles into the gap.
The seal effectiveness is characterized by the effective clearance. This is determined in experiments and takes the described additional factors into account. Three regions can be recognized on basis of the curve progression. The bending curve progression in the first zone up to about one bar (this pressure difference is below the typical range of use) is typical for brush seals. There has been no published plausible explanation for this behavior.
The second zone, from about 1 bar up to 20 in extreme cases, depending on brush configuration, corresponds to the normal operating range of brush seals. This almost linear behavior is typical. It shows, that the seal effectiveness is stabile and that there is only a small risk of the seal failing.
The third zone is characterized by the rapid climbing of the curve. This indicates that the seal effectiveness of the brush drops rapidly with increasing pressure. This behavior characterizes the pressure limit of the brush seal. This is caused by the brush bristles bending around the inner edge of the backing plate (Fig. "Operation behaviour by brush seal stiffness") when the radial clearance gap increases. Additionally, the flow resistance in the gap decreases further due to the flow-promoting contour of the bent brush.

Figure "Brush seal factors determining effectiveness" (Ref. 7.3.1-11): The depicted diagrams were derived from various brush seals in a testing rig. Therefore, the diagrams apply only to the tested configurations. However, the curve progression makes possible general statements regarding seal behavior.

Influence of the fence height (free brush): The fence height has a strong influence on the seal effectiveness of the brush seal (top right diagram). The resistance to the pressure differences drops considerably as fence height increases. The maximum viable fence height is determined by the pressure differences at the seal (Fig. "Brush seal limits"). Under a certain fence height (about f=1,4), there seems to be a relatively large unsensitive zone. Therefore, the leakage air is not linearly related to the fence height. The reason for this behavior is other parameters that affect the seal effectiveness (Fig. "Brush seal limits").

Influence of bristle thickness: The pressure limit is also highly dependent on the stiffness of the brush bristles. It is considerably increased with thicker bristle diameters corresponding to the section modulus (top left diagram). The flat curve progression of the thick bristles shows almost no influence of the gap height to the seal effectiveness. However, the effective clearance is large even with small pressure gradients. The brush with thin bristles has a very small effective height at first. However, this increases rapidly if the pressure gradients increase. Evidently the bristles bend into the gap. The initial poor seal effectiveness of thick bristles is due to larger gaps between the bristles.

Influence of bristle (brush) density: The density of a brush is the number of bristles in a circumferential unit. Higher brush density is noticeable due to a considerable increase in seal effectiveness. For example, in once case, doubling the density resulted in roughly 33% more seal effectiveness. The larger number of stressed bristles and the greater friction inside the brush caused it to be more stiff. The increased stiffness of the denser brush doubled the pressure limit. Evidently the pressure gradients in the brush were not linear.

Influence of the bristle height: As the bottom left diagram shows, the leakage air flow (corresponding to the effective clearance) increases along with the bristle height. Evidently leakage increases exponentially above a certain bristle height, which indicates an additional influence. It is possible that the brush packet is blown apart. The bristle height cannot be reduced freely in order to reduce the leakage flow. Shorter brushes increase the radial stiffness, which translates into increased heating-up during rubbing. The designer must find an optimal compromise.

Influence of the effective clearance: The bottom right diagram indicates that there is a relationship between the effective clearance at different pressure levels and the clearance during rest. With a small initial clearance, the effective clearance/leakage rate increases slightly along with the pressure difference. However, a large initial gap with a high leakage rate causes slight increases in seal effectiveness as the pressure differences increase (curve drops off). This phenomenon is called “blow down”. It is caused by steeper engagement of the bristles, which are bent radially inward by the leakage air flow (Fig. "Preventing bending of brush seal bristles").

Figure "Brush seal tribo-system" (Ref. 7.3.1-1): The wear behavior of the bristles is important for the operation of the seal. The increases in the leakage rate over the operating time/cycles is dependent upon the wear behavior of the brush tribo-system (Ref. 7.3.1-9). Additionally, the failure of a brush due to overheating after rubbing is determined by the wear behavior. Therefore, a considerable part of brush development is devoted to understanding the qualitative and quantitative wear behavior of the brush.
The overlapping influences show just how complex a brush seal tribo-system is. This also explains the elaborate determination of configuration parameters, which requires special testing rigs. Semi-empirical wear models are used in attempts to make the wear-specific configuration of brush seals safer (Ref. 7.3.1-9). However, the operating behavior can deviate considerably from previous assumptions. One example is the influence of contaminants in the leakage air. Contaminants (dust, oil, wear products) are specific to the engine area in which the affected seal is located. These cause very different rubbing wear along the circumference of the brush (e.g. stiffening effect in bristles that are stuck together). The saying “the engine will tell us” is especially true for brush seals.

The influences on the operating behavior can be summarized in main categories:

Operating conditions: The following excerpt from a 1998 conference paper is characteristic (Ref. 7.3.1-10 ): “One of the places it (the improvement of engine performance) must start is at the design phase with greater understanding of the engine operating environment which leads to more robust brush seals. This is a challenge for all current and future hardware.” Uneven in- and outflow at the brush seal can cause bristles to deflect or vibrate and make them wear heavily against one another (Fig. "Brush seal damaging turbulence"). If oxidation also occurs, the material removal from the fresh metal surfaces created by the wear will accelerate further (Fig. "Brush seal bristle oxidation").

Tribo-system: The wear of the brush contact surfaces depends on more than the behavior of the bristles during rubbing. Oxidation of the fresh wear surfaces can promote material removal, or even protect against it due to the lubricating effect of the oxides. Naturally, in a brush tribo-system, the rub surface plays an important role as a wear partner. Its surface structure and material properties are important factors.

Construction: The designer can influence the wear behavior of a brush by adjusting the dimensions of the seal parts, such as the cover plate and brush (also see Ref. 7.3.1-17 and Fig. "Brush seals damage relevant features"). The options for the bristle diameter, brush angle, and bristle height allow specific dimensioning for a broad range of operating conditions. However, at this point in time, there are no comparatively verified configuration data of the type available for other machine parts, such as roller bearings. Evidently, there is still a good deal of fundamental work that needs to be done. Additionally, the configuration of the gap and the prescribed geometric tolerances and clearance changes during operation are important. Experience has shown that the problem of installing brushes without damaging them is usually recognized too late. This can have an effect on the wear and wear patterns. Similar problems are caused by design-specific air turbulences around the brush. Typical causes are rotating screw heads or discrete injections (e.g. in collectors for turbine rotor blade cooling air).

Manufacture of the brush and rub surface: Evenness of the brush properties around the entire circumference, maintenance of close tolerances, and the type of brush fastener are typical manufacture-dependant characteristics that influence brush wear. Damage to the bristles during manufacture (e.g. strength loss due to corrosion or overly high processing temperatures) can worsen their operating behavior (wear, leakage, life span).

Figure "Compressor rotor perforated by rubbing": The stiffness of a brush is an important factor for its tendency to have problems during operation. The stiffness is dependent upon the constructive dimensioning of the brush. Less obvious influences include the operating environment, such as pressure gradients or leakage air flow. Brush stiffness is increased by so-called “blow down” or “Pressure closure” (Ref. 7.3.1-15, Fig. "Preventing bending of brush seal bristles"). This stiffening is caused by the leakage air flow, some of which is guided radially inward along the bristles. Therefore, the front plate length is an important factor (Ref. 7.3.1-8).
Noticeable stiffening is also caused by the pressure difference at the brush, which presses the bristles into one another and against the backing plate (Figs. "Brush seal stiffness", Operation behaviour by brush seal stiffness" and "Preventing bending of brush seal bristles"). This creates such high friction forces that they prevent deflection of the brush during clearance losses (rubbing). Also, the deflected brush only returns to its original shape after the pressure has dropped, i.e. the engine has been shut down. This reduces the contact time with the glide surface and minimizes wear at the bristle tips (Fig. "Brush seal leakage hysteresis"). This behavior is the prerequisite for an acceptable life span.
Stiffness changes due to changes in the pressure differences can be reduced by a relatively simple change to the brush lay-on length on the “backing plate”, which entails perforating the potential lay-on area (bottom diagram, Fig. "Preventing bending of brush seal bristles"). These brushes have a correspondingly lower hysterisis than conventional brush seals (Fig. "Brush seal leakage behavior") and are therefore referred to as “Low Hysteresis Seals” (LHS; Ref. 7.3.1-3).
Manufacturing can have unexpected effects. These include jamming of the bristles or fusing of the bristle tips during machining of the inner diameter by means of electrical discharge machining. Auxiliary material used during manufacture, for fastening the bristles, for example, can cause the hairs to stick together and considerably increase the stiffness of the brush.

Figure "Operation behaviour by brush seal stiffness": The brush stiffness during rest without seal function can be determined by the modulus of elasticity of the bristle material and the geometry of the bristles (top left diagram). The brush has a corresponding resistance to radial deflection of the rotor (Ref. 7.3.1-5). This makes calculation of the rubbing forces possible. The bottom right diagram shows the pressure of the bristles on the glide surface (also see Fig. "Brush seal distinguishes from labyrinth seal"). Understandably, relatively high lay-on pressures can be created, resulting in fusion and plastification of the bristle tips during rubbing.
Increases of brush stiffness due to friction between the bristles and against the backing plate must also be taken into account. The stiffness increases due to this effect are typically the same magnitude as would be necessary for a pressure difference of about 4 bar (Ref. 7.3.1-3).
The bottom left diagram shows further influences on brush stiffness (Ref. 7.3.1-8). The pressure difference and the corresponding leakage flow deflect the bristles outward towards the rotor (blow down) and bend them around the edge of the backing plate (blow out). Noticeable blow out is a clear sign for overstressing of the brush due to an overly large pressure difference (Fig. "Brush seal limits").

Figure "Brush seal deterioration of sealing effect": Naturally, the seal effectiveness of a new brush only has limited use for predicting the operating behavior over long run times. Therefore, this property can not serve as the only measure of seal quality.
Brush seals have different long-term performance than labyrinth seals. A labyrinth seal is worn down after the first incident of clearance loss. This decreases the seal effectiveness considerably and early (top left diagram). However, leakage does not increase much after this, even after long operating times (e.g. due to erosion). On the other hand, the elasticity of brushes allows them to deflect when they contact the rotor. Wear occurs, but it is relatively minor. The rubbing process repeats during every start-up/shut-down cycle. This is due to thermally and mechanically induced radial relative motions of the rotor and static brush components. The decrease in seal effectiveness takes place over the course of many load cycles. If a sufficient pressure drop occurs (top left diagram), then the brush sinks (also see Fig. "Brush seal leakage behavior") due to the elastic resilience of the brush, causing rubbing (usually start-up/shut-down cycles, Ref. 7.3.1-10). This hysteresis behavior is typical for brush seals (Fig. "Brush seal leakage behavior"). In References 7.3.1-10 and 7.3.1-3, this behavior can be seen in compressor exit seals up to about 10³ cycles. After this time, the gap has become so large that the seal performance is similar to that of labyrinth seals.
The designer can control this behavior by selecting a suitable original clearance. In order to minimize the loss of seal effectiveness in the first period of operation, Low Hysteresis Seals (LHS) are recommended (Fig. "Preventing bending of brush seal bristles"). The friction on the backing plate of these seals is reduced through a constructive measure. However, these seals have a larger leakage amount during start-up (before deflection; bottom diagram, Ref. 7.3.1-3). Additionally, they are sensitive to bristle flutter (vibrations, Ref. 7.3.1-13).
For a conventional brush, the top curve corresponds to the behavior after deflection, the bottom curve is behavior without deflection. It is interesting that the leakage decreases over time in the curve without deflection. This behavior can be explained by the blow-down effect. The increase of the leakage amount over time in the deflected brush is most likely due to the bristles jamming and rubbing against one another. The leakage rate is considerably increased after a single deflection. The top right diagram shows how, after a complete loss of pressure (engine shut-down) in case B, process A reoccurs. This is explained by the temporary jamming of the bristles while deflected due to friction forces between the bristles themselves and between the bristles and the backing plate.

Figure "Brush seal leakage behavior": These diagrams illustrate the dependency of the leakage rate and the adjusting pressure ratios in a brush seal when the clearance gap changes. The top diagrams (Ref. 7.3.1-5) show the tendency of this dependency in stationary trials, i.e. changing the pressure ratios around a non-rotating seal.
The bottom diagram (also see Ref. 7.3.1-16) shows the typical hysteresis of the leakage flow progression in a conventional seal (the brush is affixed by a weld, Fig. "Brush seal specific terms").
During changes in the operating state of an engine, the following factors influence the formation of the hysteresis of the leakage air flow during changes in the pressure gradients.
When an engine is started, the pressure level increases, as does the pressure gradient at the seal. This causes the leakage air flow to increase. At the same time, the clearance gap changes due to centrifugal forces and heat strain (see Chapter 7.1). The increased pressure gradient densifies the brush and provides a higher resistance to the leakage air flow. This can even result in a temporary drop in the leakage air flow (Fig. "Brush seal deterioration of sealing effect"). Additionally, when leakage air flows through the brush radially, it presses it inward (blow down, pressure closure, Fig. "Preventing bending of brush seal bristles"), making the leakage gap smaller and improving the seal effectiveness. This causes the curve to level off at higher pressure ratios.
Clearance losses during unsteady states of operation, such as start-up, shut-down, and output changes deflect the brush radially outward. This also happens when the rotor is temporarily deflected due to imbalances or vibrations. Usually, clearance losses cause the leakage gap to close and decrease the leakage flow. When the gap reopens, however, e.g. when the engine is shut down, the size of the friction forces of the brush against the backing plate and between the bristles determine how long it takes for the brush to spring back and reduce the clearance gap again. Until the brush springs back, the leakage air flow should be relatively large, and may possibly increase periodically.
The hysteresis is partly dependent on the friction forces in and on the brush (see Fig. "Brush seal deterioration of sealing effect"). It can evidently be reduced by constructive designs such as recesses in the backing plate (Fig. "Brush seals for high pressure ratio"). In attached brushes, such as in Fig. "Brush seal specific terms", the hysteresis can be suitably adjusted by the ring-shaped air spaces.
The reset forces are influenced by many specific configuration factors, such as bristle thickness, material combination of the rubbing partners, bristle height, and fence height (Fig. "Brush seal specific terms").
Due to the many influences from design, configuration, and operation, very different hysteresis progressions can be expected. Therefore, the diagram should only be seen as depicting tendencies. However, the difficulty of obtaining sufficiently certain realistic leakage measurements from test configurations is easy to see. These include statements concerning long-term performance of brush seals, especially when factors such as wear and oxidation are considered.
The saying, “the engine will tell us” is especially true for brush seals.

Figure "Brush seal leakage hysteresis": A characteristic of brush seals during operation is the hysteresis of the leakage rate (radial clearance) due to changes in the pressure ratios (bottom diagram). The solid curve shows the leakage rate, i.e. seal effectiveness. The broken line is the radial clearance between the brush and rotor. This behavior can be explained by the radially-acting force components during start-up and shut-down. The lower curve branch corresponds to engine start-up (increased pressure and pressure ratios at the seal; top left detail). The top curve branch corresponds to engine shut-down (top right detail).

The important, radially-acting force components are shown in the top diagrams:
F_B is the radial component of the friction force created by the contact of the bristles with one another and against the backing plate. At pressure ratios typical around brush seals, this force is considerably greater than the other, inward-acting components.

F_E is the elastic spring-back force after the bristles have been deflected.

F_G is the dynamically induced lifting force on the bristles due to an air cushion forming between the glide surface and the bristle tips.

F_BD is the force from the blow down effect caused by the leakage air flow that flows radially inward through the brush.

F_A is the radially-acting component of the rubbing force between the brush and rotor.

F_R is the radially outward-acting force on the bristles due to the movement of the rotor against the brush.

If the pressure increases, F_B,, F_E and F_BD act against deflection and increase the force necessary for the rotor to deflect the brush. This deflecting force is primarily comprised of F_A,(resulting from the rubbing force) and F_R,(the radial force from the movement of the rotor against the brush). This results in contact (minimal clearance and leakage air flow) between the rotor and the brush. The metallic contact is interrupted when sufficiently large dynamic air forces F_G (similar to those in an air bearing) have formed.
If pressure decreases, the friction force of the bristles against one another and against the backing plate resists spring-back and blow down. Because this force (F_B) is very large, the brush cannot spring back with the rotor movement. Complete spring-back occurs only when the pressure has greatly decreased (engine shut-down; see Fig. "Brush seal leakage behavior").
Therefore, it can be assumed that a maximally deflected brush will not come back into contact with the rotor until the engine is shut down. This minimizes wear and can be attributed to the start-up/shut-down cycles. During operation, a relatively large clearance gap is created, which is similar to those in run-in labyrinth seals.
If the brush is installed with some excess, then the bristles are already laying against the rotor and are elastically deflected during standstill. In this case, the seal effectiveness is very high during the first few cycles (bottom curve in the diagram), but the material contact during engine power-up will cause wear after a short time, which leads to a permanent clearance gap. Therefore, brush excess does not ensure a lasting improvement in seal effectiveness of a brush seal. On the other hand, excess increases the likelihood that unintentional rotation of the rotor against the brush angle will seriously damage the bristles by bending them.

Figure "Brush seal influencing dynamic behaviour" (Ref. 7.3.1-18): Aside from good seal effectiveness, brush seals also have a vibration-damping effect on rotors.
Labyrinth seals can cause rotor-dynamic shaft instabilities (see Chapter 7.2.2). The vibrations are incited by the gas flow that rotates inside the seal. These vibrations can be weakened or prevented by a turbulence retarder. This is installed above the seal and lowers the inlet tangential velocity of the flow to the seal.
The design of a brush seal causes it to have, aside from seal effectiveness, a turbulence retarding effect. This is due to the standing brush packet, which retards the tangential components of the leakage flow. In the cited text, comparative tests of a four-stage tandem brush and a nine-fin labyrinth were conducted in a testing rig. These seals were selected with relatively large axial length, in order to determine their influence on the rotor dynamics. Evidently, seal effectiveness was of secondary importance. Rotor vibrations were incited from the outside.
It was shown that the labyrinth seal in the testing configuration always had negative stiffness, i.e. promoted deflection due to decreasing counteracting force. The brush seal had positive stiffness, i.e. it had an increasing force that acted against deflection. The cross-coupled stiffness coefficient (in circumferential direction) is lower in the brush seal than in the labyrinth seal, and independent of the inlet preswirl. The labyrinth has a destabilizing effect on the shaft due to the cross-coupled stiffness coefficient, which is not the case with brushes.
On the other hand, the direct damping is greater in the labyrinth than in the brush seal.
The relationship of the circumferential speed of the flow in the seal to the circumferential speed of the rotor is a suitable measure of the dynamic stability of a seal (also see Chapter 7.2.2). This ratio is very small in brush seals, and virtually independent of all test parameters. This is a characteristic of a vibration-stable seal. On the other hand, at high tangential inflow speeds, the whirl frequency ratio in the labyrinth seals indicates instability.

The conclusions in Ref. 7.3.1-18 for the inspected brush seal are as follows:
“1.) The rotor-dynamic coefficients (direct and cross-coupled damping and stiffness coefficients) are independent of seal spacing or inlet tangential velocity.
2.) Direct stiffness is frequency dependent and always positive. The mean value of the direct stiffness coefficient increased slightly with increase in pressure ratio and showed no dependence on the other test parameters.
3.) Cross-coupled stiffness is very low and generally negative, therefore, stabilizing. The cross-coupled stiffness coefficient is generally independent of all test parameters, particularly the inlet preswirl, in contrast to conventional labyrinth seals.
4.) Direct damping slightly increases with increase in rotor speed. The intercept of the fluid forces imposed on the seal …vs frequency at w=0 is not zero. This could be attributed to Coulomb friction caused by bristle contacting the shaft.
5.) The whirl frequency ratio indicates that the brush seal is extremely stable. This ratio did not vary with any of the test parameters.“

Figure "Brush seal application suitable designs" (Ref. 7.3.1-1): The space-saving design, good seal effectiveness, and damping effect on rotor vibrations make the use of brush seals in many parts of the engine.
The brush can be installed similar to a roller bearing or integrated into the larger part (top left diagram). The fastening of the brush seal can be adjusted to the operating environment. These configurations allow segmented brush seals, as are used in axially split housings, for example (bottom right diagram).
Brush seals in tandem arrangement (top right diagram) are always used in cases where the pressure ratio to be controlled would overload a single brush (see Fig. "Brush seal limits"). Tandem brushes can be several staggered single brushes or an integrated design.
The bottom left diagram shows brushes used as intermediate stage seals. The brush also serves to stiffen the inner shroud. It is also thinkable that the pronounced inner friction of the brush serves to dampen vibrations of the stator assembly.

Figure "Brush seals in modern engines": This modern turbofan engine was one of the first to be outfitted with brush seals. They were used in the area of the front bearing chamber (top diagram) and at the compressor exit (bottom diagram). The higher pressure ratios in these areas required a tandem design in order to prevent the brush being overloaded (blow out, see Fig. "Brush seal limits"). It is interesting that a “safeguarding” labyrinth seal was also used at the compressor exit. The rub surface was created by a disk-like forming of the shaft around the flange and not integrated into the shaft cone. This type of integrated bracket for brush contact surfaces is now found in other engines in the same area.

Figure "Brush seals replacing labyrinth seals" (Ref. 7.3.1-10): In the model range of a large modern turbofan engine, an early version (top diagram) had labyrinth seals at the compressor exit and the cooling air intake on the front side of the first high-pressure turbine disk (outside and inside; diagram bottom left column). At the compressor exit, a four-stage and a three-stage labyrinth were used. A later, higher performance version had brush seals at these four places. The diagrams in the right column depict the two compressor seals and the inner turbine seal. The comparatively small space taken up by the brush seal is clearly recognizable. There are evidently several constructive variations of the turbine seals, so that it is unclear to what degree the currently used inner seal depicted here corresponds to which engine design variation. These have overhaul intervals of up to 20,000 hours. Originally, all seals were in a two-brush tandem design. This was replaced by a three-brush tandem configuration. This replacement indicates that the two-brush configuration was not effective. The pressure ratio was most likely too large for two brushes and overloaded them (see Fig. "Brush seal limits"). An unallowably high loss of cooling air can cause high temperature levels in the turbine rotor blades and be very expensive.
The engines now have run times between 600 and 11,000 hours. The outer brush seal in the high-pressure turbine is near the disk annulus and seals the hot gas flow out. This seal is subject to high rubbing speeds and temperatures. It is not surprising that these seals were the most heavily damaged. They were rendered useless after 10,000 hours of operation due to heavy and uneven bristle wear. The inner brush seals of the high-pressure turbine, however, were relatively free from wear and could be reused. On the whole, the reports indicate that the airline which uses these engines feels that considerably more robust brush seals are necessary in order to ensure that the necessary safe operating times are reached.

Figure "Brush seal at turbine shroud" (Ref. 7.3.1-4): This example shows the broad range of potential uses for brush seals. It also shows the high expense and time necessary for developing brush seals that have satisfactory operating behavior in extreme operating environments and conditions.
For the low-pressure turbine of the depicted helicopter engine, a fin seal was designed, manufactured, and tested in the engine. The trial run was conducted over 21 operating hours with cyclical and constant operating conditions. The shroud temperature was 620°C and its circumferential speed was 335 m/sec (about 20,000 RPM).
The brush seal was intended to replace a honeycomb seal in the serially produced engine. The seal had to be radially slit at an angle for proper installation. The life span requirements demanded that the seal was replaced at specific intervals (overhauls). The brush, which was made of Haynes 25 bristles, created a smooth contact surface on the blade shrouds without damaging them directly. Evidently, the brush material smeared (deposited) on the contact surface. The normal angle of the shrouds caused the corresponding contact surface geometry.
The brush wore down very quickly at first, but then the wear slowed considerably. Evidently, the brush had rapidly been worn down to the point that there was no contact between it and the shroud annulus throughout most of the test time. Brush wear is noticeably accelerated by cyclical operation. In this case, pronounced hysteresis of the brush seems to have an effect (Figs. "Brush seal leakage behavior" and "Brush seal leakage hysteresis").
These tests were published in 1993. Even now, about 8 years later, it is still not known if this type of seal use can be developed to find serial application.

Figure "Brush seal possible applications": As the example in Fig. "Brush seal at turbine shroud" shows, there are many possible different uses of brush seals in engines. In order to avoid the additional problems caused by high operating temperatures (oxidation of the thin bristles, loss of resilience due to creeping-induced plastic deformations), brushes are most suited to applications in the cooler and moderately warm engine areas. This makes the compressor region especially interesting. The left diagram shows a hypothetical use of a brush as a tip seal on shroudless stator vanes.
The right diagram depicts a brush seal used as a tip seal for compressor rotor blades. The self-supporting set-on shroud ring could, for example, be made from high-strength, high-stiffness, fiber-reinforced titanium.

Figure "Brush seal possible designs": These examples of use show that, aside from seal effectiveness, there are other properties of brush seals that can be used to an advantage.

Fastening of freely expanding parts: The top left diagram depicts the hot part region of a helicopter engine (Ref. 7.3.1-6). The clearance above the blades of the high-pressure turbine disk (arrow) is vital for the efficiency and performance of the high-pressure turbine and, therefore, that of a significant part of the engine. This gap must be minimized over the entire period of operation. To this end, a seal ring (turbine ring) is mounted on the housing, under which the blades run past. Ceramic is a suitable material for this ring due to its properties: high heat strength (saves cooling air), low heat strain (maintains clearance), and high resistance to erosion and wear. However, the brittleness of the ceramic and the differences in heat strain between it and the housing make it difficult to fasten. The top right diagram shows a constructive solution in which a four brush tandem configuration elastically supports the ceramic turbine ring on the housing. The low thermal conductivity of the brush also prevents large amounts of heat from being transferred into the housing.
Similar uses have been reported in combustion chambers and turbine stator vanes. In these cases, the brush seal takes over the seal function and/or support of metallic parts (Ref. 7.3.1-8). Serial implementation has not been reported. Brush seals can also be combined with glide rings, in which case the brush seal is intended to make the glide ring more easily adjustable (Ref. 7.3.1-12).

Labyrinth seals with brushes as an abradable surface (middle left diagrams): Ref. 7.3.1-7 describes tests in which labyrinth fins on the rotor run into a wide brush packet. Its behavior during dynamic tests (cyclical changing of the overlap) was good, compared with conventional labyrinth seals. However, as would be expected, the brush wear during rubbing was problematic.

Sealing the bearing chambers of slowly running shafts. These seals are designed for tough lubricants such viscous oil or grease. The middle right diagrams show different seal variations, in which the brush seal is mounted on the shaft and rotates with it. In this case, the rub surface is on the static part.

Intermediate stage seals of turbine rotors: The bottom diagram shows an example of an industrial gas turbine (Ref. 7.3.1-8), which uses a multiple brush tandem configuration due to the high pressure ratios.