Searching for correspondence between mechanismic/deductive and empirical/inductive models at industrial level by means of dimensional analysis
Fragiskos A. Batzias, Athanasia Bountri, Dimitrios K. Sidiras · AIP conference proceedings · 2012
In this work, a methodology is developed under the form of an algorithmic procedure, including 16 activity stages and 6 decision nodes, to reveal any correspondence between mechanismic/deductive and empirical/inductive models used for simulated industrial processes. The main parts of this procedure are: (i) determination of the set of dimensionless groups, which is the most representative for describing the industrial process under consideration, (ii) regression analysis of the corresponding empirical/inductive model to estimate its goodness of fitting to experimental data, (iii) determination of the most relevant mechanismic model and estimation of its parameter values, (iv) incorporation of these values into a model similar to the original inductive one, (v) regression of the new hybrid model on the same dataset for checking goodness of fitting, (vi) parameter identification by comparing the original with the new model. An implementation is presented referring to adsorption of methylene blue on lignocellulosic biomass wasted from wood processing industrial units.