Milton Javier Munoz Neira ¹, Manuel Roa ¹, Rodrigo Correa ¹

Affiliations
1 Universidad Industrial de Santander - UIS Colombia, Santander, CO

Abstract

Background/Objectives: This article includes some of the most relevant experimental results related to the thermal properties of coffee hulls taken from samples of a Colombian coffee variety Castilla. Methods: Coffee hull samples were analyzed using IR, DSC, TGA and DTGA. Analytical thermal experiments described its thermal behavior when it was heated at 5°C/min from room temperature to about 900°C. Furthermore, we include some drying curves and analyze the observed delay effect of the hull presence on the coffee drying process. Findings: Some of the main thermal transitions where observed in the range of 50-300°C. The IR test showed that its main composition was cellulose. Such delay was related to the hull composition. The diffusion coefficients of water in coffee grains with hull, without its hull, and of hull alone, are also reported. Applications: This work provides important information to understand in a better form some thermal characteristics and inference in drying process of coffee grains hull of variety Castilla, original from Colombia northeast region.

Full Text

   

Juan Carlos López ¹, Edgar García ¹, Rodrigo Correa ¹

Affiliations
1 Universidad Industrial de Santander Colombia, CO

Abstract

Background/Objectives: This article describes a way for estimating the internal heat generation function during microwave heating processes. Methods: We solve the corresponding inverse problem to estimate the internal heat generation function. We consider an illustrative example dealing with the heating of solid spherical Silicon Carbide samples. We used two numerical strategies: The Spiral Optimization Algorithm and the traditional Levenberg-Marquardt method. Findings: Even if both approaches differ in nature, our results indicated an excellent agreement between both numerical strategies. Applications: With this method, it is possible to estimate with high precision this important parameter in microwave heating processes.

Full Text

   

Ernesto Aguilera ¹, Ivan Amaya ², Rodrigo Correa ¹

Affiliations
1 Escuela de Ingenierías Eléctrica, Electrónica y de Telecomunicaciones, Universidad Industrial de Santander, CO
2 Escuela de Ingeniería y Ciencias, Tecnologico de Monterrey, MX

Abstract

Background/Objectives: In this work, we present an efficient design methodology that uses boosting algorithms to improve the accuracy of any given learning algorithm by combining the output of individual weak learners. Methods: First, a finite-difference time-domain model of a loaded rectangular wave guide yields the desired input-output response of a microwave heating system. Then, it is used to train neural networks used as weak learners in the boosting algorithm. Findings: The method is easy to implement and have a tendency not to over fit the training data. Data show that performance of the boosting algorithm increases with the number of neural networks. An example that uses 34 neural networks, with three hidden layers, fits 96 of 100 temperature profiles of the heating system with a previously defined ischolar_main mean square error below 1°C. Applications: Two simple examples of inverse modelling problems of the heating system were solved efficiently using the output of the boosting algorithm.

Full Text

   

Milton Munoz ¹, Ivan Amaya ², Rodrigo Correa ¹

Affiliations
1 Universidad Industrial de Santander, UIS, Santander, CO
2 Tecnologico de Monterrey, Monterrey, Nuevo Leon, MX

Abstract

Background/Objectives: We present an alternative for estimating drying curves and diffusion coefficients of coffee beans (Castilla variety), based on global optimization strategies. Methods: Four optimization algorithms were tested for adjusting drying curves. Based on the parameters that were found, we determined the diffusion coefficient. Algorithms were tuned up on 11 non-linear systems, prior to using them for adjusting the curves. Their performance were assessed through error dispersion analysis, as well as through the number of evaluations of the objective function and run time. Findings: On non-linear systems, Particle Swarm Optimization (PSO) and Drone Squadron Optimization (DSO) exhibited the best performance in terms of error. When used for estimating drying curves, PSO, DSO and Genetic Algorithms (GA) achieved determination coefficients beyond 0.99. Even so, GA had the lowest run time. Applications: Our experiments offer an alternative with excellent precision for estimating parameters of the drying function and of its diffusion coefficients for different coffee beans.

Full Text