Design improvements on labyrinth seal teeth and a honeycomb land are examined by three-dimensional CFD numerical modeling of the flow field. The only objective is reduction of the total leakage through the new seal. CFD assumptions and analysis was validated by comparison with leakage data from labyrinth seal experiments conducted by Stocker [1]. The baseline chosen for comparison of sealing effectiveness is a conventional low clearance straight-through labyrinth seal with four teeth and a honeycomb land of symmetrical hexagonal cells. The proposed new seal has a staggered honeycomb land and straight teeth with an inclined notch. CFD predicts ∼17% reduction in seal leakage at a radial clearance of 0.005 inch (0.122mm) due to higher wall friction and flow turbulence.

HP turbine cooling systems utilize a TOBI (tangential on-board injector) nozzle and seal-plate to deliver cooling air to the airfoils with minimal loss in pressure and a benefit in relative total temperature. It was necessary to have a low TOBI exit pressure to control excess leakage across adjacent labyrinth seals with a trade-off of maintaining adequate pressure to supply blade cooling. Consequently, pumping vanes were needed on the seal-plate to restore pressure, which adds parasitic work to the turbine and a corresponding loss of engine efficiency. Engine testing revealed recirculation zones attached to the radial pumping vanes. Subsequent numerical simulation corroborated the recirculation and resulting circumferential asymmetry in flow distribution to the cooled airfoils. As a consequence, excess cooling air is supplied to the HP rotor with some airfoils receiving the minimum amount of cooling air. This study seeks to improve the flow distribution and pumping effectiveness by means of contouring the pumping vanes, increasing pumping vane solidity and/or controlling diffusion by re-contouring the seal-plate surface. Design features that increase total through flow when overall pressure ratio is fixed are also explored.

The flow discharge through thick orifices with approaching flow normal and inclined to the orifice axis was numerically predicted. The objective is to validate Fluent code predictions of discharge coefficient against Rohde et al. experimental data and arrive at an optimum mesh in terms of effort, simulation time and accuracy. CFD simulations were performed on several Rohde et al. experimental models with orifice thickness to length ratio varying from 0.5 to 4.0, with sharp and rounded inlet edges and orifice axis inclined 45 degrees to the approaching flow. The approach Mach number varies from 0.07 to 0.55 and orifice velocity head ratio ranges from 1.3 to 200. Simulations are performed using FLUENT V6.3 with k-ω SST model and 2nd order discretization scheme. Coarse mesh results at selected test points were compared with fine mesh results and a best meshing practice was determined. CFD computation at selected points was also performed with Realizable K-ε, v2f and RSM turbulence model. CFD predictions of discharge coefficient show good agreement with Rohde et al. experimental data.

Gas turbine engine secondary flow systems are sensitive to variation in part dimensions, clearances, flow coefficients, swirl ratios, head loss factors, tolerances, boundary conditions, etc. This paper reveals a process and software application which embodies the process wherein both offer a measurable contribution to secondary airflow system reliability. The probabilistic methodology is empirically validated by (1) applying it to an engine component that failed in a validation test, and (2) demonstrating that a multiple order sensitivity analysis performed during detailed design was unable to detect a failure mode while a probabilistic analysis revealed a small yet significant risk of catastrophic failure. Therefore, a secondary flow analyst does not have a justifiable reason to be highly confident of a design qualified by a first, second, or higher order sensitivity analysis. The last example empirically demonstrates compatibility of optimization techniques with probabilistic methods (as part of the process) to quantify the likelihood of failure and reveal an optimized design space of key characteristics where risk is eliminated and the effects of variation are controlled. Trade study analysis is more valuable if it includes a quantitative evaluation of the effects of variation on alternate designs and the response to failure modes. A key feature of the software application is a relational database with the capability to configure and effectively manage flow networks in many forms including, a status model, failure modes of the status model, multiple alternative designs as well as failure modes specific to an alternative design.

Gai turbine engine secondary flow systems are sensitive to variation in part dimensions, clearances, flow coefficients, swirl ratios, head loss factors, tolerances, boundary conditions, etc. This paper reveals a process and software application, which embodies the process, wherein both offer a measurable contribution to secondary airflow system reliability. The probabilistic methodology is empirically validated by (I) applying it to an engine component that failed in a validation test and (2) demonstrating that a multiple order sensitivity analysis performed during detailed design was unable to detect a failure mode while a probabilistic analysis revealed a small yet signiflcant risk of failure. Therefore, a secondary flow analyst does not have a justiflable reason to be highly confldent of a design qualified by a flrst, second, or higher order sensitivity analysis. The last example empirically demonstrates the compatibility of optimization techniques with probabilistic methods (as part of the process) to quantify the likelihood of failure and reveal an optimized design space of key characteristics, where risk is eliminated and the effects of variation are controlled. Trade study analysis is more valuable if it includes a quantitative evaluation of the effects of variation on alternate designs and the response to failure modes. A key feature of'the software application is a relational database with the capabiliry to configure and effectively manage flow networks in many forms including a status model, failure modes of the status model, multiple alternative designs, as well as failure modes speciflc to an alternative design.