A statistical approach for pyrolysis process description and optimization

According to RED II targets by 2030 the renewable energy consumption must be raised up to 32%, particularly 14% of the energy supplied for road and rail transport must be renewable. As part of renewable energy sources, residual biomass plays a significant role, because of its availability and the possibility of obtaining advanced biofuels and biochemical trough several thermochemical and biochemical processes, such as pyrolysis, gasification and aerobic digestion. The main challenge of using residual biomass is its properties variability (e.g., O/C ratio, ashes content, extractives, cellulose, hemicellulose, lignin and so on) according to type, geographic localization and seasonality. For this reason, properties and yields of the products are highly variable. Furthermore, the complexity of the processes involved in biomass valorization often makes the classical chemical-physical modelling approach very challenging. In this framework, the research community as well as the industrial sector are focusing their attention on the adoption of new provisional tools, based on advanced statistical methods, machine learning techniques and operative research techniques for decision-making support, in order to improve the sustainability utilization of the biomass resources. In particular, this work represents the first attempt to this new modelling approach, focusing its attention on the thermochemical valorization of residual biomass. The principal components analysis (PCA), which projects the original data in principal component (PC) space for assessing which input parameters have the highest score on the PC reducing the input data dimension, without losing valuable information, is used for the fast pyrolysis process. A broad database with about 450 observations is developed collecting data from scientific literature. The database includes both qualitative and quantitative information regarding main biomass characteristics (e.g., type, proximate and ultimate analysis), operative parameters (e.g., temperature, pressure, type of reactor) pyrolysis products and byproducts (e.g., yields, quality). PCA is applied to a subset of the original data considering only fast pyrolysis process, non-catalytic, in a fluidized-bed reactor with a temperature range of 450-550°C.