Online monitoring of solvent and acid gas concentration in a CO2 absorption process using MEA

Online monitoring of solvent and acid gas concentration in a CO2 absorption process using MEA

Van Eckeveld, A.

Boersma, B.J. (mentor)

Mechanical, Maritime and Materials Engineering

Process and Energy

Energy Technology

2013-10-23

Enhanced greenhouse gas emisions in the past century contributed to the global average temperature rise. The emission of greenhouse gases, therefore, should be limited in the coming decades. One way to do so is by capturing the greenhouse gasses (like CO2) from industrial flue gases (post-combustion capture). This can be achieved in a chemical absorption process using monoethanolamine as a solvent. In this report, the development of a method has been developed for the real-time liquid analysis of the solvent and absorbed acid gas concentrations in a post-combustion capture process using monoethanolamine as a solvent, capturing CO2. Online monitoring of the dynamic behaviour of these process properties is of major importance from a process control and a scientific perspective. Currently this is only achieved using fourier transfrom infrared spectroscopy, combined with a multivariate calibration method called chemometrics. This is, however, an expensive method and there are limitations with respect to robustness and actual process installation. The developed method is based on cheap and easy measurable properties of the solvent. Inverse Least-Squares models were built at two temperature levels, based on a set of 29 lab-prepared samples with different MEA and CO2 concentrations. Density, conductivity, refractive index and sonic speed measurements were used as input data in model development since this combination of quantities showed the best performance with respect to predictive accuracy and apparatus costs of all assessed analytical techniques. The developed model has been validated during continuous operation of a mini CO2 capture plant. The concentrations of MEA and CO2 were predicted with a mean error of 0.53 and 0.31 [wt%] for MEA and CO2 respectively. Process dynamics, like step-changes in the CO2 flue gas concentration, were covered accurately as well. The model showed good robustness to changes in solvent temperature. Robustness of the model against solvent degradation was assessed using samples from the TNO pilot plant at the Maasvlakte and during a three week experiment at a microplant. First some degraded samples were added to the model calibration set for improved robustness purposes. Both the polluted model and the unpolluted model were able to monitor the MEA and CO2 concentrations of the pilot plant samples and in the microplant accurately. The addition of degraded samples to the calibration set did improve the predictive accuracy of the model if the degradation products in the monitored process were of the same nature as the degradation products that were added to the calibration set. The combination of the model with process data has shortly been assessed, but further study in this direction is required. Combining density, conductivity, refractive index and sonic speed measurements with a multivariate chemometric method enables the real-time and accurate monitoring of the acid gas and monoethanolamine concentrations in CO2 absorption processes, even under degrading process conditions.

CCS
process monitoring

http://resolver.tudelft.nl/uuid:a1eeb0e3-92ad-4c1e-80cd-30ab117b4a79

2016-10-23

Student theses

master thesis

(c) 2013 Van Eckeveld, A.