Performance analysis and modeling of a closed-loop heat pump dryer for bay leaves using artificial neural network


AKTAŞ M., Sevik S., ÖZDEMİR M. B., Gonen E.

APPLIED THERMAL ENGINEERING, cilt.87, ss.714-723, 2015 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 87
  • Basım Tarihi: 2015
  • Doi Numarası: 10.1016/j.applthermaleng.2015.05.049
  • Dergi Adı: APPLIED THERMAL ENGINEERING
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.714-723
  • Anahtar Kelimeler: Bay leaf, Drying, Heat pump dryer, Artificial neural network (ANN), Moisture output, Energy consumption, FREEZE-DRYING BEHAVIORS, PREDICTION, ENERGY, SYSTEM, PARAMETERS, MUSHROOM, TOMATO, DESIGN
  • Gazi Üniversitesi Adresli: Evet

Özet

This study discusses the performance analysis and modeling of bay leaves drying in a closed-loop heat pump dryer. To control relative humidity and temperatures, a controller has been designed, developed and tested to perform at low temperature drying applications (40 degrees C, 45 degrees C and 50 degrees C). New techniques are applied simultaneously such as control of relative humidity, drying air temperature and air velocity for the closed-loop heat pump dryer. A closed-loop heat pump dryer using water as a secondary fluid has been used to determine drying characteristics of bay leaves. Moreover, a Programmable Logic Control (PLC) has been used to control drying air temperature, air velocity, relative humidity and the mass obtained for the dried product. The performance of the facility has been carried out but several experimental tests under different psychrometric conditions from the test results. COPhp and COPws values were 2.8-3.7 and 2.4-3.2, respectively, while Energy Utilization Ratio (EUR) values were found to vary 0.22-0.75. From experimental data the system was analyzed and modeled by using Artificial Neural Network (ANN) and drying kinetics of bay leaf. The ANN model was used to predict the moisture content (MC, g water/g dry matter) and the total energy consumption (TEC, kWh) of the system. The back-propagation learning algorithm with Levenberg-Marquardt (LM) and Fermi transfer function were used in the network. The coefficient of determination (R-2), the root means square error (RMSE) and the mean absolute percentage error (MAPE) were calculated as 0.996, 0.0002053, 0.4161673 and 0.997, 0.0005013, 0.4280322, respectively. Accordingly, it can be concluded that predicted MC and TEC results are in good agreement with experimental results. (C) 2015 Elsevier Ltd. All rights reserved.