Propagation of uncertainty in high-performance liquid chromatography with UV–VIS detection

Detection, Quantification, and Propagation of Uncertainty in High‐Throughput Experimentation by Monte Carlo Methods

“…Pipetting a volume of 10 lL, the RSD was 6.5 % (buffer), respectively 9.8 % (protein); pipetting higher volumes of around 200 lL, the RSD was only 1.5 % (buffer), respectively 1.3 % (protein). The increase of RSDs for buffer and protein pipetting and the pipetted means are very similar; therefore, the pipetting data was summarized and analogously to the previous section a functional relationship between the data and their RSD could be stated to: (12) with a coefficient of determination of 95.83 %. Pipetting smaller volumes is generally more deviation-prone as drops can stick to the pipetting tip or adhesive forces hinder pouring out the complete volume in the tip.…”

“…Monte Carlo (MC) sampling, a concept of numerical in silico experimentation based on the law of large numbers, came up in the 1970s parallel to rising computational power and is frequently applied to problems of uncertainty propagation, e.g., in physics and chemistry [10]. Applications of MC sampling-based uncertainty propagation in the field of bioseparation technologies are described in [11][12][13][14]. However, to the authors' knowledge, bioseparation experimentation on a robotic station has never been subject to uncertainty analyses.…”

Uncertainty budget for final assay of a pharmaceutical product based on RP–HPLC

“…All short-time random variations in A are included in the repeatability of the method [4], which then includes small differences in injection volumes, separation yields, sensitivity of UV-detector, etc.…”

Uncertainty analysis in mass transfer parameter estimations for chromatographic separation of glucose and fructose

“…The objective of this work is to analyze how the uncertainty in the values of some variables that are involved in the characterization of a chromatographic column, propagates in the estimation of kinetic parameters of the model front velocity. For this purpose, an alternative procedure to the the use of the Monte Carlo method [7,5] is applied, based on the use of response surface [9] for obtaining the probability distribution function (pdf) of the estimated parameters. This proposal has the advantage, over the Monte Carlo method, that require shorter compute time and to offer direct interpretation of how acting factors uncertainty has influence over the variation of the estimated parameters.…”