The Effect of Self-Optimisation Targets on the Methylation of Alcohols Using Dimethyl Carbonate in Supercritical CO2

Abstract

This paper describes the next stage in our development of self-optimising reactors. We demonstrate that the same reaction can be optimised for a series of different criteria including yield, space-time yield, E factor and a weighted yield function (the product of space-time yield and yield). In different experiments, we achieved 97.6% yield, space-time yield of 42.9 kg/L/h and E factors of 1.4 and 3.3 (including CO 2 ) and the weighted yield, which gave a promising balance between yield, E factor and space-time yield.

Remote-controlled experiments with cloud chemistry

“…More recently, the automation has been taken to a higher level with the development of self-optimizing reactors that have an analytical device (chromatograph or spectrometer) downstream of the reactor that determines the composition of the product stream and quantifies the amount of each different product formed [9][10][11][12] . These data, which are obtained under different reaction conditions, are then fed into an algorithm that calculates a new set of reaction conditions predicted to give a higher yield of the desired product [13][14][15] . The procedure is applied iteratively to maximize the yield without the need for any intervention by the operator.…”

Enhanced process development using automated continuous reactors by self-optimisation algorithms and statistical empirical modelling

“…Self-optimising reactors have been designed using a variety of analytical techniques including IR [14][15][16] and NMR spectroscopy 17 , mass spectrometry 16,18 , gas [19][20][21][22] and liquid 23,24 chromatography. In this paper, a feedback-controlled flow reactor, equipped with an at-line HPLC system, is used to provide fast separation and quantification of the desired compounds.…”

Real-Time Feedback Control Using Online Attenuated Total Reflection Fourier Transform Infrared (ATR FT-IR) Spectroscopy for Continuous Flow Optimization and Process Knowledge

“…13 We showed that the apparatus could be programmed to optimize the reaction for any one of a range of different criteria (i.e., yield, space-time yield, E factor or a weighted yield function). 14 Although we reported that self-optimization is fast compared to manual optimization of one parameter at a time, the process is still quite lengthy (.2 days for 4 parameters) largely because of the time required for the GLC analysis. Here, using the solvent-free etherification of 1-pentanol as an example, we report how the whole process of optimization can be greatly accelerated by using Fourier transform infrared (FT-IR) spectroscopy analysis.…”