|
Abstract
|
The issue of sustainability has become a strategic imperative for researchers attempting to address en-ergy and environmental concerns using biorefinery approach. Exergy-based methods have shown sig-nificant promises in terms of their ability to reliably locate the hotspots of resource degradation in
biorefineries. The key step in analyzing biorefineries exegetically is to calculate biomass chemical exergy
which is a very computationally-intensive task. Interestingly, proximate and ultimate analysis methods
show potential to reflect the chemical exergy content of biomass. Hence, the present study was devoted
to introducing a novel hybrid intelligent approach to determine the chemical exergy content of biomass
based on both the composition analysis methods. In the developed hybrid models, input score variables
in each inner loop of partial least square (PLS) approach were correlated with its output score variables
using hybrid adaptive neuro-fuzzy inference system and particle swarm optimization algorithm (ANFIS-PSO). Both the developed modeling systems showed acceptable accuracy in determining the chemical
exergy values of biomass materials. The model derived from ultimate analysis was slightly more accurate
than that from proximate analysis (mean absolute percentage error of 0.207 vs. 0.506, respectively).
Nevertheless, simple and inexpensive character of proximate analysis can facilitate real-world applica-tions of the respective model. Overall, the developed model can pave the way for developing sustainable
biorefineries by computing the chemical exergy of biomass more accurately than complex thermody-namic models
|