Online analysis of MEA solvent in a pilot plant was demonstrated by application of the developed model using a mobile setup for extractive sampling to an FTIR analyser.
Our results show that liquid FTIR by ATR combined with multivariate methods such as PLS is well suited for analysis of amine solvents, both in the lab and for online process monitoring. Published Ltd. E-mail address: aslak. Published by Elsevier Ltd. Conventionally, these liquid analyses have been performed manually by the use of relatively time-demanding methods.
Manual off-line liquid analysis does not provide input data for real-time process control and optimization. In addition, the methods used are generally labour-intensive and impractical for studies of transient conditions during dynamic changes in a process. Both from a scientific- and an operational aspect, real-time data of CO2 loading and amine concentration during operation of experimental equipment and CO2 capture plants would provide much useful information for process characterization and control.
Fourier Transform Infrared FTIR spectroscopy has substantial potential for scientific purposes and as a quantitative quality control tool for the industry. FTIR spectrometry has the advantage, compared with other methods that several compounds can be detected simultaneously, and the monitoring can be performed continuously.
FTIR ATR liquid analysis of CO2 absorption in amine solutions has demonstrated that both carbamate and carbonate formation can be monitored using this technique [1]. The advantage of choosing FTIR as a quantitative technique lies in its ability to readily carry out multi-component analysis in association with multivariate analysis methods such as Partial Least Squares PLS regression [2].
The objective of this work has been to develop a robust calibration model for liquid analysis of mono-ethanolamine MEA solutions with different amine concentration and CO2 loading.
One half of the stock solution was loaded with CO2 by carefully bubbling CO2 through a sinter into the solution until the desired amount of 0. The solutions with other CO2 loadings were then obtained by mixing the unloaded stock solution A and high loaded solution stock solution B gravimetrically in different ratios.
The MEA method is based on potentiometric titration with 0. The measured spectral region was cm During online measurements, 2 scans per spectrum were applied in order to obtain one spectrum per second from the analyser. The spectral regions from cm-1 and cm-1 were chosen for the PLS-1 calibration model. In order to compensate for baseline drift and differences between instruments, the model includes pre-processing of the spectra: in addition to mean centering of the spectra, the average intensity in the region to cm-1 near absorptionless region in all calibration spectra is offset-corrected to intensity zero.
Normalization is applied by setting the integral in the region cm-1 equal to 1 this region has a near iso-spectric point for the calibration spectra. Cross validation was applied as diagnostic type for the multivariate models. Figure 1 shows an overlay of all spectra from the calibration set. Spectral resolution: 32 cm-1 and 32 scans pr. Method parameters To maximize the amount of brine samples analyzed on a TOC analyzer, the sample volume injected into the combustion tube for oxidation was reduced.
During a typical TOC oxidation step, the carrier gas is turned off after the sample is injected into the combustion chamber to allow for expansion of the liquid sample into a gas. Reproducibility and accuracy decreased dramatically after the first failure, as observed in the higher standard deviations and percent RSDs. In addition to the decreased reproducibility, occasional sample replicate outliers yielded poor peak integration.
As the precision decreased, the number of poor sample integrations increased, which eventually caused the accuracy to fail. A total of At this rate, typical duplicate analyses would last for samples using 0. The analysis would last four days, averaging 40 brine samples 28 percent NaCl per day before the The warning signs of salt overload are poor precision and peak shape of sample integrations. The same sample load criteria can be applied to samples with lower amounts of salt such as seawater percent NaCl.
Metals in salts, such as sodium, bond with the quartz of the combustion tube through the process of devitrification. Higher temperatures will increase the devitrification rate of the quartz combustion tube causing it to crystallize. Increased devitrification may cause the combustion tube to weaken and possibly break. Conclusion The TOC analyzer used in this study found that increased deposition of salts onto the catalyst eventually causes incorrect response for organics at about 23 g of NaCl.
Therefore, to allow a safety factor for the analysis of salty samples, the analyzer could tolerate a maximum of 15 to 20 g of NaCl before the combustion tube and catalyst must be cleaned. Both combustion tube and catalyst should be cleaned with 2 N HCl solution and reagent grade water.
The combustion tube, o-rings and injection port should be inspected for aging and salt buildup when cleaned. Fresh quartz wool should be used after each cleaning. Sample and gas lines are color coded for quick identification. Internal components are carefully laid out with the customer in mind.
A static injection port is utilized instead of a slider injection port. No moving parts during sample injection allows for consistent, long-term, leak-free operation. As a result, down time and cost of operation is kept to a minimum. Since the NDIR detector has an especially wide linear range, the need for multiple calibration curves or time-wasting reruns is minimized.
0コメント