The first aim of the study was to model the influence of water amount and air pressure on various batter properties using Response Surface Methodology. Batter quality was assessed through density, water content, colour, spreadability and fluorescence spectra. Quadratic models using two variables well represented spreading time, water content and final temperature, but they failed to fairly represent initial density, overrun and L*a*b* values. In addition, simplified models using a single variable also well represented the data: final density was modelled by a linear equation involving pressure, whereas initial density, water content and final temperature were modelled by a linear equation involving water amount. Spreading time was modelled using a quadratic equation using water amount. Experimental results were compared with expertise rules used by operators to control the industrial process. Indeed, operators often used water amount and air pressure as controlling variables. It was found that experimental results were in agreement with expertise rules. The second aim was to investigate the link between smart lab-measurement methods such as fluorescence spectroscopy and simple macroscopical properties used by operators such as water content, density, spreading time and colour. By applying hierarchical clustering analysis to NADH and tryptophan merged spectra, batter samples manufactured at various water amounts and pressure levels were clearly separated at a high level of discrimination. Neither water content nor spreading time were satisfactorily predicted from NADH or tryptophan spectra using PLS.
©2011 Walter de Gruyter GmbH & Co. KG, Berlin/Boston