A seal is a component used in a turbomachine to reduce internal leakage of the working fluid and to increase the machine's efficiency. The stability of a turbomachine partially depends upon the rotodynamic coefficients of the seal. The integral control volume based rotodynamic coefficient prediction programs are no more accurate than the accuracy of the leakage mass flow rate estimation. Thus an accurate prediction of the mass flow rate through seals is extremely important, especially for rotodynamic analysis of turbomachinery. For labyrinth seals, which are widely used, the energy dissipation is achieved by a series of constrictions and cavities. When the fluid flows through the constriction (under each tooth), a part of the pressure head is converted into kinetic energy, which is dissipated through small scale turbulence-viscosity interaction in the cavity that follows. Therefore, a leakage flow rate prediction equation can be developed by comparing the seal to a series of orifices and cavities. Using this analogy, the mass flow rate is modeled as a function of the flow coefficient under each tooth and the carry over coefficient, which accounts for the turbulent dissipation of kinetic energy in a cavity. This work, based upon FLUENT CFD simulations, initially studies the influence of flow parameters, in addition to geometry, on the carry over coefficient of a cavity, developing a better model for the same. It is found that the Reynolds number and clearance to pitch ratios have a major influence and tooth width has a secondary influence on the carry over coefficient and models for the same were developed for a generic rectangular tooth on stator labyrinth seal. The discharge coefficient of the labyrinth seal tooth (with the preceding cavity) was found to be a function of the discharge coefficient of a single tooth (with no preceding cavity) and the carry over coefficient. The discharge coefficient of a single tooth is established as a function of the Reynolds number and width to clearance ratio of the tooth and a model for the same is developed. It is also verified that this model describes the discharge coefficient of the first tooth in the labyrinth seal. By comparing the coefficients of discharge of compressible flow to that of incompressible flow at the same Reynolds number, the expansion factor was found to depend only upon the pressure ratio and ratio of specific heats. A model for the same was developed. Thus using the developed models, it is possible to compute the leakage mass flow rate as well as the axial distribution of cavity pressure across the seal for known inlet and exit pressures. The model is validated against prior experimental data.

The labyrinth seal is a non-contact annular type sealing device used to reduce the internal leakage of the working fluid which is caused by the pressure difference between each stage in a turbomachine. Reducing the leakage mass flow rate of the working fluid through the labyrinth seal is desirable because it improves the efficiency of the turbomachine. The carry-over coefficient, based on the divergence angle of the jet, changed with flow parameters with fixed seal geometry while earlier models expressed the carry-over coefficient solely as a function of seal geometry. For both compressible and incompressible flows, the Reynolds number based on clearance was the only flow parameter which could influence the carry-over coefficient. In the case of incompressible flow based on the simulations for various seal geometries and operating conditions, for a given Reynolds number, the carry-over coefficient strongly depended on radial clearance to tooth width ratio. Moreover, in general, the lower the Reynolds number, the larger is the divergence angle of the jet and this results in a smaller carry-over coefficient at lower Reynolds numbers. However, during transition from laminar to turbulent, the carry-over coefficient reduced initially and once the Reynolds number attained a critical value, the carry-over coefficient increased again. In the case of compressible flow, the carry-over coefficient had been slightly increased if radial clearance to tooth width ratio and radial clearance to tooth pitch ratio were increased. Further, the carry-over coefficient did not considerably change if only radial clearance to tooth width ratio was decreased. The discharge coefficient for compressible and incompressible flows depended only on the Reynolds number based on clearance. The discharge coefficient of the tooth in a single cavity labyrinth seal was equivalent to that in a multiple tooth labyrinth seal indicating that flow downstream had negligible effect on the discharge coefficient. In particular, for compressible fluid under certain flow and seal geometric conditions, the discharge coefficient did not increase with an increase in the Reynolds number. It was correlated to the pressure ratio, Pr. Moreover, it was also related to the fact that the flow of the fluid through the constriction became compressible and the flow eventually became choked. At low pressure ratios (less than 0.7), Saikishan’s incompressible model deviated from CFD simulation results. Hence, the effects of compressibility became significant and both the carry-over coefficient compressibility factor and the discharge coefficient compressibility factor needed to be considered and included into the leakage model. The carry-over coefficient compressibility factor, phi, had two linear relationships with positive and negative slopes regarding the pressure ratios. This result was not associated with the seal geometry because the seal geometry ratios for each instance were located within the nearly same ranges. Further, the phi-Pr relationship was independent of the number of teeth regardless of single and multiple cavity labyrinth seals. The discharge coefficient compressibility factor, psi, was a linear relationship with pressure ratios across the tooth as Saikishan predicted. However, in certain flow and seal geometric conditions, Saikishan’s model needed to be modified for the deviation appearing when the pressure ratios were decreased. Hence, a modified psi-Pr relationship including Saikishan’s model was presented in order to compensate for the deviation between the simulations and his model.