“…In Figure 1 are presented the singularity (s ι ) and ill conditioning (c ι ) measures (12) of the estimation matrix O (9a) as function of the sensor location pair (l 1 , l 2 ), for a completely observable structure σ (5) with estimation order pair κ = (κ 1 , κ 2 ) = (6, 6) and κ ι = κ 1 + κ 2 = 12 = n = N (i.e., complete model with observable structure), showing that: (i) the largest singularity and illconditioning values are obtained with the sensor stage location pair (l 1 , l 2 ) ≈ (1 to 2, 12), (ii) the smallest singularity and illconditioning values are obtained with the sensor location (l 1 , l 2 ) ≈ (1, 2) [ or (11, 12)] with two adjacent sensors in the bottom (or top) of the column, followed by the sensor location (l 1 , l 2 ) ≈ (5, 6) with two adjacent sensors above or below the feed tray (5). These consideration lead to the following conclusions: (i) the two sensors should not be in the same section, and (ii) the best location pair for complete profile estimation is given by These location results are in agreement with location criteria employed in: (i) two-point temperature PI control of distillation columns (Tolliver, 1980;Castellanos-Sahagun et al, 2005), and (ii) previous distillation column studies with EKF (Baratti et al, 1995;Oisiovici and Cruz, 2000).…”