Reaction network determination using calibration free analytical data
Advancing the chemical engineering fundamentals
Chemical Reaction Engineering (T2-2P)
Keywords: Keywords: Reaction network, analytical data, kinetic modelling
Reaction network determination using calibration free analytical data
K. Novakovic, M.J. Willis and A.R. Wright
University of Newcastle, School of Chemical Engineering and Advanced Materials,
Newcastle upon Tyne, NE1 7RU, UK
katarina.novakovic@ncl.ac.uk; mark.willis@ncl.ac.uk; a.r.wright@ncl.ac.uk;
The determination of a reliable reaction network that can be used in simulation studies is the initial requirement for reaction engineering applications such as scale up, moving from a batch process to a continuous one, thermal safety studies etc. The usual method used to determine a reaction network is to postulate a number of different structures which are then fitted to experimental data. The reaction network whose ordinary differential equation model provides the highest prediction accuracy with respect to the experimental data is taken to be the correct structure.
The experimental data used are generally the species time-concentration data obtained from a series of batch or fed-batch experiments, using one of a number of analytical devices, including GC (Gas Chromatography), GC-MS (Gas Chromatography-Mass Spectrometry), HPLC (High Performance Liquid Chromatography) etc. The usual method used to obtain species’ concentrations with these devices is to use an internal standard; a chemical that is inert to other components of the reaction system investigated and is present in each sample in known concentration. This requires determination of a constant called the response factor which relates the area of the peak produced by a known concentration of the internal standard and the area of the peak produced by an unknown concentration of the species whose concentration is being measured. The response factor differs with each species and needs to be determined prior to sample analysis therefore requiring a standard for each species, i.e. that each individual species has been synthesised.
A common problem is that some of the products and intermediates species standards are not available. This may be because the species cannot be isolated or synthesised or that obtaining a satisfactory purity for the standard is difficult and time consuming. Within process development, time to market is a critical process driver. The ability to differentiate between potential network structures while not being reliant on species standard characterisation would have the potential to remove a significant bottleneck in process development time. The aim of this work is to mathematically demonstrate that a network structure can be obtained without the need for response factor determination prior to sample analysis. As the ratio of the species area to the reference standard area observed through, for example GC-MS analysis, is directly proportional to the species concentration this data may be used directly to elucidate network structure.
The approach is demonstrated and critically assessed through simulated case studies addressing reaction systems where (1) there are differing numbers of species (2) some of the species are not detected by the analytical device (3) the reactant initial conditions relate to a known ratio of the species area to the reference standard area and (4) species of known initial concentration are not detected. Finally, the practicality of the method is demonstrated using experimental data obtained from the L-proline catalysed aldol reaction of p-nitrobenzaldehyde with acetone (Novakovic et al. 2006).
References
K. Novakovic, C. Grosjean, T. Schütz, M.J. Willis, A.R. Wright and A. Whiting (2006). “Enhancing process development using high throughput technologies. A case study using an L-proline catalysed aldol reaction”, 17th International Congress of Chemical and Process Engineering, CHISA 2006.
Presented Tuesday 18, 13:30 to 15:00, in session Chemical Reaction Engineering (T2-2P).
