FT-IR: a suitable process analytical technique for post combustion - - PowerPoint PPT Presentation

FT-IR: a suitable process analytical technique for post combustion capture of CO2

Eva Sanchez Fernandez, Annemiek van de Runstraat, Leon Geers, Earl Goetheer TNO Gas Treatment Delft, the Netherlands

TNO Gas Treatment Group 23/5/11 1

Content

Why PAT? Principles of FT-IR Model construction Demonstration Conclusions

TNO Gas Treatment Group 23/5/11 2

Why PAT

Compromising capturing efficiency: Thermal degradation of amine due to heating/cooling cycles Chemical degradation of amine due to formation heat stable salts Contamination of solution with other species in flue gas Evaporation of water from amine solution Temperature fluctuations  Optimal performance  Better process understanding necessary  But ... first step is monitoring concentrations Goal: Development of in-line monitoring system for concentrations of CO2, active amine, and heat stable salts

TNO Gas Treatment Group 23/5/11 3

Framework: post combustion capture of CO2

CO2 content Amine content HSS content

TNO Gas Treatment Group 23/5/11 4

Measurement technique: Fourier Transform Infrared Spectroscopy

Flexible  many different species In-line applicable  no “running to the lab” with samples Fast  order of a minute Non-contaminating  without addition of internal standards

C N H Infrared light absorption Chemical bonds stretch or bend

Model system: potassium b-alanine

TNO Gas Treatment Group 23/5/11 5

Comparison potassium salt and - carbamate

Potassium salt β-alanine Potassium salt β-alanine carbamate Carbonate Conclusion: No straight forward dependency species and FT-IR peaks

TNO Gas Treatment Group 23/5/11 6

From IR spectra to concentrations?

[Amine-] [CO3

2-]

[HCO3

• ]

[Amine-CO2

• ]

[Heat Stable Salts] [ ... ? ... ]

?

• Absorption is wavelength dependent  peaks of specific bonds

may overlap

• Water absorbs over broad band  might obscure peaks of other

species

• Chemical reactions  peak shifting or disappearance
• Temperature fluctuations affect spectra

TNO Gas Treatment Group 23/5/11 7

Partial Least Squares Model – Creation

: : : : 16.9 … 9.99 0.01 n 23.5 … 6.20 2.33 2 12.3 … 2.60 1.31 1 Yp … Y2 Y1 # : : : : 16.9 … 9.99 0.01 n 23.5 … 6.20 2.33 2 12.3 … 2.60 1.31 1 p … 2 1 #

Spectra (X) Concentrations (Y)

1 2 3 ... ... n

Covariance matrix XTYYTX Basis functions

TNO Gas Treatment Group 23/5/11 8

Coefficients T1..m contain information to calculate species concentrations More basis functions  better fit BUT more rigid model Caveat: Non-informative correlations may be present in spectra  preprocessing necessary

Partial Least Squares Model – Regression

= T1* + T2* + .. + Tm* + error

TNO Gas Treatment Group 23/5/11 9

Experiments

Calibration + validation 28 Stock solutions were created of amine in 4 different concentrations CO2 and SO2 were added with a bubbler CO2 concentration measured with phosphoric acid method IR Spectra collected with FTIR spectrometer and ATR flowcell Testing 7 samples from pilot plant capture installation

• f both lean and rich streams

TNO Gas Treatment Group 23/5/11 10

Results – Basis functions

Amine peaks (e.g.H-N-H bend) N-H stretch C-O stretch N-H, but also S-O

TNO Gas Treatment Group 23/5/11 11

Calibration and validation matrices

(37) (15)

TNO Gas Treatment Group 23/5/11 12

Results – Validation

1 2 3 4 5 6 1 2 3 4 5 6 Reconstructed Amine conc. (M) Measured Amine concentration (M) 0.5 1 1.5 2 2.5 0.5 1 1.5 2 2.5 Reconstructed CO2 conc. (M) Measured CO2 concentration (M) 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 Reconstructed SOx conc. (M) Measured SOx concentration (M)

TNO Gas Treatment Group 23/5/11 13

Results – Pilot plant

RMS Error: ~0.04M

TNO Gas Treatment Group 23/5/11 14

Results pilot tests

TNO Gas Treatment Group 23/5/11 15

Microplant pilot equipped with on-line FT-IR

TNO Gas Treatment Group 23/5/11 16

Test 1: 100 spectra recorded on-line show stable operation Time Colour intensity is measure

• f peak

intensity

TNO Gas Treatment Group 23/5/11 17

Test 2: process event identification

CO2 + SO2 SO2

Water replenishement

TNO Gas Treatment Group 23/5/11 18

Test 3: Model robustness

CO2 CO2 + SO2 CO2 + SO2 + NOx

TNO Gas Treatment Group 23/5/11 19

Test 3: Model robustness – SO2 in rich stream

CO2 CO2 + SO2 CO2 + SO2 + NOx

TNO Gas Treatment Group 23/5/11 20

Conclusions

Methodology for reconstructing solvent concentration from FT-IR concentrations developed Accuracy of predictions on-line within 10% For calibrated species in calibrated range Model can be trained to discard noise from contaminations of other species Process event identification possible using this tool  extra input of information for process control FT-IR is a suitable process analytical tool

TNO Gas Treatment Group 23/5/11 21

Acknowledgement:

Ralph Joh and Rudiger Schneider (Siemens)