KARŞIT AKIŞLI RANQUE – HILSCH VORTEKS TÜPÜNDE OKSİJEN KULLANILARAK DESTEK VEKTÖR MAKİNELERİ VE LİNEER REGRESYON YÖNTEMİ İLE PERFORMANS ANALİZİ


Doğan A., Korkmaz M., Kırmacı V.

Ist-International Congress on Modern Sciences Tashkent Chemical-Technological Institute, Toskent, Özbekistan, 10 - 11 Mayıs 2022, ss.531-540

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: Toskent
  • Basıldığı Ülke: Özbekistan
  • Sayfa Sayıları: ss.531-540
  • Hacettepe Üniversitesi Adresli: Evet

Özet

Ranque-Hilsch Vortex Tube (RHVT) is a system that consists of a simple pipe with no moving parts other than the control valve, which can create the heating and cooling process using pressurized fluid. Although air is generally used as the pressurized fluid in these systems, oxygen was used as the pressurized fluid in this study. In the experimental study, counter-flow RHVT was preferred. The counterflow RHVT was designed with an inside diameter of 7 mm and a body length of 100 mm. Two, three, four, five and six orifice nozzles made of aluminum and brass materials were used in RHVT. During the experiments, the outlet control valve on the hot fluid side was left in the fully open position. In addition, while data were taken during the experiments, the inlet pressure was taken by using pressured oxygen starting from 150 kPa and up to 700 kPa at 50 kPa intervals. Performance optimization was made by calculating the difference (ΔT) between the temperature of the hot flow leaving the RHVT (Tsck) and the temperature of the cold flow leaving (Tsgk). In order to optimize the performance of RHVT, Linear Regression (LR) and Support Vector Machines (SVM) methods were used separately from machine learning methods. In the study, machine learning was applied by using 80% of all data as training data and 20% of test data in Linear Regression and Support Vector Machines methods. Estimates were made with the test data using the trained model. The coefficient of determination R2, which is the measure of the accuracy values of the test prediction results obtained by using the Linear Regression and Support Vector Machines methods was calculated. The analysis results obtained at the end of the study were interpreted.