This article elaborates the concept of splitter vanes in controlling secondary flow. Secondary flow vortices are formed by the rotation of vorticity filaments, located in the endwall boundary layers, as the filaments move through the passage. The connection between the number of stators and the secondary kinetic energy suggests that the only way to significantly reduce the mixing loss is to increase the number of blades in the row. The designs evaluated were produced with fast turn-around computational fluid dynamics (10 minutes per solution) and automated optimization techniques. Experimental tests showed that the theory was correct, and that by increasing vane count, the secondary kinetic energy was reduced by up to 80%.
Turbine design involves many engineering disciplines. The final product is a compromise between aerodynamic performance and constraints arising from mechanical, structural or material requirements. As turbine efficiency increases, engineers must revisit the performance penalties associated with these compromises and develop new ideas to improve the design. In this article, we describe one such concept: splitter vanes.
An example of an aerodynamic challenge created by a mechanical requirement is the use of turbine stators to encase components that pass through the main gas path. These components could be part of the engine structure or pipes carrying oil or air. Engineers have used two approaches to tackle this problem. In the first approach, an additional row of non-turning faired struts is added to house the components. This increases machine length, weight and wetted surface area, all reducing performance. In the second, an existing stator row is adapted to accommodate the components. In this case the machine length remains almost constant. However, the modified stators are thick, have a low aspect ratio, and secondary flows dominate.
Secondary flow is defined as fluid with a velocity component in a direction normal to the average flow. Secondary flow is typically characterised by vortices such as those that dig away at the riverbed upstream of a bridge buttress.
“Secondary flow vortices are formed by the rotation of vorticity filaments, located in the endwall boundary layers, as the filaments move through the passage. Around each stator leading edge the inlet boundary layer rolls up into a vortex tube. A vortex “leg” enters the passage on each side of the stator. The leg next to the pressure surface at the leading edge (PS leg) sweeps across the passage, entraining more vorticity as it does so, to produce the dominant flow structure known as the “passage vortex”. The leg formed at the suction side (SS leg) remains close to the suction surface forming the counter vortex, a much smaller flow feature.
This model predicts that the secondary vorticity (wsec) at row exit is a function of inlet velocity gradient and row turning only and thus, for a fixed inlet boundary layer profile and stator turning, the secondary vorticity will be constant.
Although the secondary vorticity is of interest, the designer is principally concerned with the associated aerodynamic loss. The primary loss contribution is the dissipation of the secondary kinetic energy (SKE) as the vortices mix out. The SKE of a vortex is proportional to the square of its circulation. In turn, the circulation is proportional to the width of the passage (the reciprocal of the stator count). Thus, summing the secondary kinetic energy across every stator, we find an inverse dependence with stator count. Therefore, when low stator counts are used, as is common in current designs, SKE is a large contributor to aerodynamic loss.
The effect of increasing the number of stators is to produce a higher number of smaller passage vortices and a net reduction in secondary kinetic energy.
The connection between the number of stators and the secondary kinetic energy suggests that the only way to significantly reduce the mixing loss is to increase the number of blades in the row. However, the large thickness needed to pass the structural or pipe components means that the stator count is limited.
The solution requires challenging one of the most common features of a turbomachine – that all blades in a row are the same. Once the possibility of a “non-uniform” stator row with thick blades shielding components and thinner “splitter vanes” to reduce the secondary flow is considered, the design space is greatly expanded.
It was found that both the stators and splitter vanes must be designed simultaneously to achieve peak performance. This increases not only the design possibilities but also the complexity of any numerical simulations performed. The designs evaluated in the current work were produced with fast turn-around computational fluid dynamics (10 minutes per solution) and automated optimization techniques.
The Horseshoe Vortex Jump
During the design process a critical flow feature, only found in non-uniform blade rows, was identified. If the leg of the horseshoe vortex of the thick stator passes upstream of the splitter vane leading edge the vorticity that the designer intended for the first passage is now diverted to the second. This results in a single large passage vortex rather than two smaller ones. In this situation the primary benefit from including splitter vanes is not achieved. Through careful profile design, it was possible to avoid the horseshoe vortex jump and hence successfully reduce the secondary flow strength, improving stage performance.
Experimental tests showed that the underlying theory was correct and that by increasing vane count the secondary kinetic energy was reduced by up to 80%. This in turn lead to increases in stage efficiency of almost 1%, representing a significant fuel saving .