MADSPEC
Project co-financed by the Junta de Castilla y León, the Institute for Business Competitiveness, and the European Union through the European Regional Development Fund (ERDF).
Project co-financed by the Junta de Castilla y León, the Institute for Business Competitiveness, and the European Union through the European Regional Development Fund (ERDF).
OBJETIVE
MADSPEC is an industrial research project aimed at evaluating the potential of vibrational spectrometry, specifically in the near-infrared (NIR) range and FTIR-ATR (Fourier Transform Infrared with Attenuated Total Reflectance), to determine the physico-mechanical properties of industrial wood. If positive results are achieved, this project will pave the way for companies manufacturing and using wood to adopt these fast, simple, and cost-effective techniques as alternatives to complex and expensive laboratory procedures.
DEVELOPMENT
To develop predictive models, NIR and FTIR-ATR spectrometries require the construction of a spectral database containing known values for the properties to be analyzed. In this study, wood from various species (Scots pine, walnut, chestnut, cherry, and gombé) was used, and some samples were subjected to different preservation treatments, including raw maintenance, air drying, steaming, and thermal treatment.
Reference Analytical Techniques
The physico-mechanical variables studied in this project are: global modulus of elasticity in bending, axial bending strength, density, hardness, and moisture content. These variables will be analyzed through tests conducted in CESEFOR’s specialized laboratories using universal testing machines and other standardized methods.
Statistical Analysis
The samples analyzed through laboratory tests will subsequently be scanned using a self-developed portable spectrometer prototype (SATree-Boscalia), a commercial NIR spectrometer (Micronir by Viavi), and the FTIR-ATR spectrometer (Carry 630) in collaboration with the CRETUS-EcoPast group (USC). Since the variables involved are continuous, the chemometric treatments applied will consist of regression models: principal component analysis (PCA) and partial least squares regression (PLSR).