Modelación cinética multimodelo del compostaje aeróbico de sólidos a partir de literatura: ajuste de parámetros y guía para diseño experimental.
Loading...
Date
2025
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Concepción
Abstract
La modelación cinética del compostaje aeróbico es fundamental para el diseño de reactores a escala industrial; no obstante, su aplicación práctica se ve limitada por la alta variabilidad de los parámetros reportados en la literatura y el uso generalizado de modelos simplificados sin validación estadística robusta. Esta memoria de título aborda esta problemática mediante la evaluación crítica de cuatro modelos fenomenológicos (Primer Orden, Bifásico, Gompertz Modificado y Arrhenius) aplicados a un "Conjunto de Datos de Referencia", construido a partir de la curaduría y procesamiento de series temporales de degradación de diversos residuos orgánicos.
Utilizando algoritmos de regresión no lineal y el Criterio de Información de Akaike corregido (AICc), se demostró la inexistencia de un modelo cinético universal. Mientras que los residuos de cocina y sustratos altamente biodegradables se ajustan adecuadamente al modelo de Primer Orden (R2>0.95), los sustratos complejos como lodos sanitarios y estiércoles requieren modelos multiparamétricos (Gompertz o Bifásico) para describir fenómenos de latencia (lag phase) y degradación heterogénea. Adicionalmente, el análisis de incertidumbre mediante técnica Bootstrap reveló rangos de confianza amplios en los parámetros cinéticos publicados, evidenciando un riesgo de subdimensionamiento en el diseño ingenieril actual. Como respuesta, se propone un protocolo experimental estandarizado y de bajo costo, diseñado para asegurar la identificabilidad paramétrica en futuros estudios, validando así el tránsito desde el ajuste empírico hacia la predicción fenomenológica confiable.
Kinetic modeling of aerobic composting is essential for industrial reactor design; however, its practical application is hindered by the high variability of parameters reported in literature and the widespread use of simplified models lacking robust statistical validation. This thesis addresses this challenge by critically evaluating four phenomenological models (First-Order, Biphasic, Modified Gompertz, and Arrhenius) applied to a curated "Reference Dataset" (Golden Dataset), established through the processing of degradation time-series from diverse organic wastes. Using non-linear regression algorithms and the corrected Akaike Information Criterion (AICc), the non-existence of a universal kinetic model was demonstrated. While kitchen waste and highly biodegradable substrates fit adequately to the First-Order model (R2>0.95), complex substrates such as sewage sludge and manure require multi-parameter models (Gompertz or Biphasic) to accurately describe lag phases and heterogeneous degradation. Additionally, uncertainty analysis using the Bootstrap technique revealed wide confidence intervals in published kinetic parameters, highlighting a risk of under-dimensioning in current engineering designs. In response, a standardized, cost-effective experimental protocol is proposed, designed to ensure parametric identifiability in future studies, thereby validating the transition from empirical fitting to reliable phenomenological prediction.
Kinetic modeling of aerobic composting is essential for industrial reactor design; however, its practical application is hindered by the high variability of parameters reported in literature and the widespread use of simplified models lacking robust statistical validation. This thesis addresses this challenge by critically evaluating four phenomenological models (First-Order, Biphasic, Modified Gompertz, and Arrhenius) applied to a curated "Reference Dataset" (Golden Dataset), established through the processing of degradation time-series from diverse organic wastes. Using non-linear regression algorithms and the corrected Akaike Information Criterion (AICc), the non-existence of a universal kinetic model was demonstrated. While kitchen waste and highly biodegradable substrates fit adequately to the First-Order model (R2>0.95), complex substrates such as sewage sludge and manure require multi-parameter models (Gompertz or Biphasic) to accurately describe lag phases and heterogeneous degradation. Additionally, uncertainty analysis using the Bootstrap technique revealed wide confidence intervals in published kinetic parameters, highlighting a risk of under-dimensioning in current engineering designs. In response, a standardized, cost-effective experimental protocol is proposed, designed to ensure parametric identifiability in future studies, thereby validating the transition from empirical fitting to reliable phenomenological prediction.
Description
Tesis presentada para optar al título de Ingeniero/a Civil Químico/a.
Keywords
Cinética química, Diseño de experimentos, Desechos orgánicos