A Predictive Cooling Model Applied to the NASA Energy Efficient Engine in a Collaborative Turbine Pre-Design Process

Abstract

Accurately estimating turbine cooling requirements at a preliminary design stage is both a challenge and a necessity for an accurate modeling of the overall propulsive system. Cooling requirements largely affect both compressor and combustor operating conditions with an overall impact on thermal efficiency and specific fuel consumption. A more accurate initial estimation reduces lead time and development costs of preliminary design tasks. Despite the necessity, it is a challenging task as cooling requirements are dependent on component material properties as well as on blade airfoil profile and internal geometry. Thus far, the 1D meanline preliminary design tool PrEDiCT combined with empirical data has provided reliable initial estimations for turbine cooling requirements. For next generation aero-engines, both operating conditions and turbine technologies may change significantly. As the turbine inlet temperature rises and new high-temperature material and cooling technologies entry into service, this empirical data becomes inadequate. To address this challenge, a cooling estimation method based on the model from Holland and Thake is built into PrEDiCT. Despite its extensive adoption in the literature, there is nonetheless a lack of information regarding the empirical values attributed to some model parameters and how these can be influenced through turbine design decisions. In the present work, these parameters are either calculated or calibrated for two cooled high-pressure turbines (HPTs) developed by P&W and GE within the NASA Energy Efficient Engine program. In the absence of a complete performance dataset in the literature, both HPTs are modeled in a collaborative process between turbine and engine performance. A data analysis provides a better understanding of empirical values attributed to some model parameters as well as reference data for future studies. In this line of thought and based on the provided data, one of the cooled turbine blades is modeled using the higher fidelity blade cooling design tool PICCOOLO. The resulting output is compared and discussed with that from the cooling model built into PrEDiCT. This model requires limited user-input, but is nonetheless a semi-empirical approach. In an iterative process, coupling this model in PrEDiCT with PICCOOLO per blade row may be a solution to eliminate user-input and reduce error. This next step in the development of a design process between 0D performance, 1D turbine meanline and 2D cooling design is briefly discussed. Additional parametric analysis provide an insight on how operating conditions and turbine design may affect cooling requirements. Apart from turbine inlet temperature, turbine inlet pressure has a noticeable effect on the estimated cooling requirements. This is a relevant insight when scaling the cooling system of aero-engines with a wide operating range, where high altitude operating points may exhibit increased thermal loads. Furthermore, preliminary design parameters such as reaction, loading or Zweifel number affect both blade row characteristics and flow conditions. The resulting effect on the cooling requirements estimated in PrEDiCT is reported and discussed. This provides a better understanding of how turbine preliminary design can be oriented for reducing cooling requirements. In addition, a sensitivity analysis on cooling model parameters is introduced to point out new technologies with the highest potential for reduction in cooling mass flow. Finally, limitations of this cooling model are briefly discussed. A scaling factor is required to estimate additional bleed air for platform cooling, disc cooling or seal purge. Furthermore, this approach is based on averaged quantities both for material temperature and flow conditions. As a result, aspects such as inflow temperature profiles or temperature distribution over the turbine blade cannot be modelled.