Prediction and Optimization of Heavy Clay Products Quality - Advanced Materials for Agriculture, Food and Environmental Safety - page 88

inl uenced by many factors due to the heterogeneous nature of used raw
materials. Mineral transformations during i ring are inl uenced by various
factors: chemical and mineralogical composition, grain-size distribution,
maximum heating temperature, heating rate, duration of i ring, and kiln
atmosphere. h e knowledge of the chemical composition of heavy clays
is essential for dei ning suitable compositions required for brick produc-
tion. Important parameters of i red products that help in evaluation of
the heavy clay deposits are compressive strength ( CS ), water absorption
( WA ), i ring shrinkage ( FS ), weight loss during i ring ( WLF ) and volume
mass of cubes ( VMC ). h ese parameters can be also indicative for i nal
products' durability and quality. Dei ning the mutual relationship among
characteristics of heavy clay, processing conditions, and properties of the
i nal products is a topic that has been extensively studied in previous years
[1-7]. Various reports have been published, but only a few of them are
related to the application of the mathematical modeling for better under-
standing of the overall system behavior and the quality of i nal products.
Recently, mathematical modeling has been increasingly used for the study
of the given systems and composition to i nal product properties relations
[1-4, 8, 9]. Developed empirical models give a reasonable i t to experi-
mental data and successfully predict technological parameters. In the
case of brick production, nonlinear models are found to be more suitable
due to variability in chemical composition of raw material and nonlinear
behavior during i ring of shaped products. h e heterogeneous nature of
the heavy clays implied a complex material-process-product relationship.
h erefore, second order polynomial (SOP) and artii cial neural network
(ANN) models have gained momentum for modeling and control of brick
processing [1-4].
Artii cial neural network (ANN) models are recognized as a good mod-
eling tool since they provide the solution to the problems from a set of
experimental data, and are capable of handling complex systems with
nonlinearities and interactions between decision variables [4]. Sensitivity
analysis was used for ANN outputs testing, by changing one input variable
at a time, for +2.241% or -2.241% of its nominal value, while all the other
variables were i xed to their baseline values. Input variables' ef ects were
calculated with reference to the same central point in the input space, thus
increasing the comparability of the results obtained [1, 10].
h e specii c objective in this study was to investigate the ef ect of chemi-
cal composition and i ring temperature on compressive strength, water
absorption, i ring shrinkage, and weight loss during i ring and volume
mass of cubes, for the brick production process. h e performance of ANN
was compared with the performance of second order polynomial models