Loading...

Investigation of Key Parameters in the Spray Drying Process using a Combination of Numerical Simulation and Neural Networks

Ghadim Mahanipoor, Mohammad Hossein | 2025

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58733 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Moosavi, Ali; Nouri Boroojerdi, Ali
  7. Abstract:
  8. Spray drying has evolved from traditional natural moisture evaporation methods into an advanced industrial technology that enables rapid and controlled drying of liquid materials. This process converts the feed liquid into a uniform powder, significantly improving product quality and shelf life. Spray dryers are widely used in the food, pharmaceutical, and chemical industries due to their high efficiency, flexibility, and broad applicability. However, spray drying faces challenges such as particle deposition on the chamber walls, reduced product yield, and potential contamination or even ignition of the deposits. These deposits not only pose safety risks but also require frequent cleaning, increasing operational costs. Additionally, high energy consumption for producing hot air, while maintaining product quality and production rate, makes energy optimization and yield improvement major challenges. Conducting full-scale experiments on spray dryers is time-consuming and costly due to the difficulty of measuring key parameters such as flow, temperature, and humidity inside the chamber. Numerical simulation has been used as a cost-effective method to analyze droplet behavior, heat and mass transfer, and predict process outputs, though accurate simulations require significant computational time. Therefore, in this study, machine learning algorithms were employed to develop predictive models based on numerical simulation data, capable of accurately forecasting the performance parameters of spray dryers. Two neural networks were developed, one with input features including operating conditions and droplet characteristics, and the other also including geometric dimensions, to predict outputs such as droplet temperature, size, moisture content, and process conditions. Prediction accuracy was compared using two approaches, neural networks and random forest, enabling selection of the most suitable model. Finally, spray dryer performance was optimized in terms of energy consumption and particle deposition using genetic algorithm, particle swarm optimization, and a hybrid particle swarm-genetic algorithm method
  9. Keywords:
  10. Spray Dryer ; Particle Deposition ; Numerical Simulation ; Neural Network ; Optimization

 Digital Object List

 Bookmark

No TOC