The scaling test for the cooling performance of turbine vane has been widely used, as it has a relatively low cost and reduced difficulty compared with tests in the real engine conditions. However, the similarity between scaling tests and real conditions become increasingly difficult to be maintained as the cooling configurations become more and more complex. Although there are some similarity criteria have been proposed, the performance of those criteria are still uncertain.

This two-part paper explores the suitable criterion for the temperature and pressure scaling tests to measure the overall cooling effectiveness, taking the internally-cooled Mark Ⅱvane as an example. In this part, Artificial Neural Network (ANN) models were established and trained by hundreds of experimentally validated Conjugate Heat Transfer CFD results. These ANN models could efficiently predict the overall cooling effectiveness distribution and other parameters according to physical boundary conditions(such as inlet total pressure and temperature), which were then utilized to verify the criteria’s accuracy in temperature and pressure scaling. The multi-layer feed-forward network configuration and back-propagation was used for ANNs’ training. All the ANNs has 5 inputs and 12 neurons in the hidden layer. The correlation coefficient of ANNs is over 0.99, suggesting the prediction of ANNs is accurate.

After comparing the scaling error, the results show that matched-Biot number criterion has the best performance at both constant coolant temperature and temperature ratio conditions. The maximum scaling error is 6.4% and 1.2% for those two conditions, respectively. Apart from that, scaling at constant coolant temperature conditions (changing temperature ratio) can obtain an relatively low mainstream pressure compared with constant temperature ratio conditions for the same target temperature, however, the scaling error would increase significantly when the  goes below 0.7.

Furthermore, the performance of those scaling criteria based on empirical correlations without the ANN models were also investigated, which is more feasible in engineering use. The accuracy of matched-Biot number criterion has an obvious decrease in this condition, but it is still the best selection at constant coolant temperature conditions. While the mass flow ratio criterion becomes the most accurate one at constant temperature ratio conditions. In terms of distribution similarity, both the matched-Biot number and mass flow ratio criterion based on empirical correlations can maintain the similarity of overall cooling effectiveness on pressure side. However, mass flow ratio criterion has a better performance on suction side, which can accurately scale the location of transition point and leading edge.

In the second part of the paper, a series of experiments conducted on Mark Ⅱ vane model at different lower temperature and pressure conditions are reported to validate the results given above.


Investigation of Similarity Criteria for an Internally-Cooled Turbine Vane Based on Artificial Neural Networks:  Part I \u2014 Comparison of Scaling Criteria

Division / Sponsor:
International Gas Turbine Institute (IGTI)

Paper Proceedings: https://asmedigitalcollection.asme.org/GT/GT2020/volume/84058

Turbo Expo - Virtual Conference

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.