CESAR1 Solvent degradation and thermal reclaiming results from TCM testing

Matthew Campbell^a, Sundus Akhter^a, Anette Knarvik^a,b, Muhammad Zeeshan^a, Ahmad Wakaa^a

^aTechnology Centre Mongstad, 5954 Mongstad, Norway

^bEquinor ASA, PO Box 8500, 4035 Stavanger, Norway

Abstract

Gå til overskriften

The Technology Centre Mongstad (TCM DA) in Norway has investigated degradation and amine losses for the non- proprietary solvent CESAR1 which is a mixture of water, amino-2-methylpropanol (AMP) and piperazine (PZ). Results have been explored during the ALIGN CCUS testing campaign which utilized the combined cycle gas turbine (CCGT) based heat and power plant (CHP) flue gas with an inlet CO₂ concentration around 3.7 vol%. It has been demonstrated that there is a significant impact on amine losses through degradation when the inlet NO₂ concentration entering the CO₂ absorber is increased. The increase in NO2 concentration in the flue gas resulted from Selective Catalytic Reduction (SCR) operation with no ammonia injection. Degradation results have also been shared for the residue fluid catalytic cracker (RFCC) flue gas from the Equinor refinery with an inlet CO₂ concentration around 13.5 vol%. Due to the impurities in the RFCC flue gas higher amine losses through degradation are observed compared to CHP flue gas testing. Also, amine losses through degradation for CESAR1 solvent were compared against historical TCM results for monoethanolamine (MEA). The results indicate significantly lower amine losses for CESAR1 as compared to MEA for both CHP and RFCC flue gases. Thermal reclaiming has also been performed on the aged CESAR1 solvent and effective operation was achieved with acceptably low amine losses during semi-continuous reclaiming operation. Future testing at TCM in the laboratory and full-scale plant are planned to have a better understanding of the major causes for amine solvent degradation.

Keywords: CESAR1, MEA, Degradation, Thermal Reclamation

1.  Introduction

Gå til overskriften

In 2019 and 2020 Technology Centre Mongstad (TCM) experimentally explored the behavior of the CESAR1 amine-based solvent for post combustion CO₂ capture. The main purpose of the testing campaigns was to generate results needed to evaluate CESAR1 performance in comparison to the current industry non-proprietary amine based solvent benchmark, which is monoethanolamine (MEA). The main areas of investigation were:

• CO₂ capture rate
• Specific reboiler duty (SRD)
• Emissions
• HSE and Operational challenges
• Solvent degradation products and rate
• Thermal reclaiming operation

This paper focuses on summarizing degradation results for a natural gas-fired combined heat and power (CHP) plant (flue gas composition ~ 3.7 vol% CO₂ and ~ 14% O₂) and for the refinery residue fluid catalytic cracker (RFCC) plant (flue gas composition ~ 13.5 vol% CO₂ and ~ 3.2% O₂). Also, a summary of CESAR1 thermal reclaiming results will be presented. This article will be divided into the following sections:

• CESAR1 degradation reaction mechanisms
• CESAR1 degradation results and rates
• Degradation comparison for CESAR1 versus TCM historical MEA
• Overview of CESAR1 thermal reclaiming
• Way forward to gain a better understanding on solvent degradation.
  

2.  CESAR1 degradation mechanisms

Gå til overskriften

Although the degradation products from a mixture of 2-amino-2-methyl-1-propanol (AMP) and piperazine (PZ) have not been explored thoroughly, several studies have been carried out in literature to propose and evaluate the degradation mechanism of PZ and AMP separately. In this section, the mechanisms for the degradation of different degradation products will be discussed.

The main areas of degradation for any amine in post-combustion CO2 capture process are the absorber sump, cross heat exchanger, reboiler, and reclaimer [1]. Thermal degradation which occurs mainly due to the high temperature of the process, and oxidative degradation because of the presence of dissolved oxygen and free radicals [1]. The major PZ and AMP degradation products are listed below in Table 1.

2.1 Overview of PZ Thermal Degradation

Gå til overskriften

The PZ molecule has the tendency to absorb two molecules of CO2 per mole of piperazine due to the presence of two amino functional groups present in a cyclic structure. The degradation pathway of PZ solvent is not completely established for each degradation product, but the mechanism for the major degradation products of PZ have been suggested [1, 3-5]. Reactions such as bimolecular nucleophilic substitution reaction (SN2), elimination, urea generation and hydrogen-abstraction are involved in the degradation pathway [1]. Iron may also affect the formation of total formate and ammonia as both increase with increase in iron concentration, suggesting that the dissolved iron might be responsible (catalyse) for the decomposition of the intermediary oxidation products to formate and ammonia [2]. Most abundant thermal degradation products are FPZ, EDA, AEP, formate and ammonia [3]. The mechanism for the thermal degradation of PZ is shown in Fig. 1.

In the first step, PZ acts as a nucleophile at the α-carbon of another molecule of a protonated PZ (P⁺) in SN2 reaction to give 1-(2-aminoethyl)-ethylpiperazine (AEAEPZ) as shown in Fig. 1. AEAEPZ has the tendency to get protonated at multiple positions due to the presence of nucleophilic primary, secondary and tertiary amino functionalities (Fig. 1 to Fig. 3). In the case of protonation of the inner-secondary amine group between C4 and C5, PZ can attack at the α-carbons (C4 and C5) to yield 1- polyethylpiperazine (PEP), protonated ethylene diamine (EDA⁺) and two molecules of 1-(2-aminoethyl)piperazine (AEP) respectively. If the tertiary nitrogen in the AEP molecule is protonated, then the α-carbon becomes prone to an SN2 reaction with another molecule of AEP nucleophile resulting in 1,4-diaminoethylpiperazine (DAEP) and AEP⁺. On the other hand, AEAEPZ can react rapidly in the presence of CO₂ in the solution to form internal urea. The internal urea, when protonated, can give a substitution reaction with another molecule of PZ which attacks on the C6 of this internal urea under SN2 reaction yielding other degradation products such as ammonium, CO₂ and quaternary amine. Similarly, EDA can also react with CO₂ to form imidazolidinone. AEP can also get protonated at multiple positions. If the protonation occurs at the terminal amine, the C7 can be approached by another molecule of AEP to yield polyAEP and ammonium while the protonation at the tertiary nitrogen gives DAEP and PZ⁺, as shown in Fig. 1.

PEP in its protonated form (at one of the tertiary amines in the cyclic structure) can undergo Hofmann’s elimination reaction to give 1-ethenylpiperazine and PZ. The elimination reaction is facilitated by higher temperatures and a mix of nucleophiles present in the solution. PEP can undergo Anti-Markovnikov hydration of 1-ethenylpiperazine in the presence of water to form 2-Hydroxyethylpiperazine (HEP) [1, 3-4]

If the tertiary amine of AEAEPZ gets protonated, it can react with a PZ molecule at positions C7 and C1 to yield hexamine and a molecule of AEAEPZ and PZ⁺ respectively (Fig.2). Similarly, protonation of the amino group in the cyclic structure of the AEAEPZ molecule followed by an attack of PZ nucleophile at C2 can give rise to the long chain hexamine shown in Fig. 3.

2.2 Overview of PZ Oxidative Degradation

Gå til overskriften

The first step in the oxidative degradation pathway is the formation of peroxide radicals. The process starts with the abstraction of a proton from one of the methylene carbon in PZ molecule thus forming a free radical which can form peroxyl radical in the presence of O₂. Degradation of the peroxyl radical may take place through; 1)intermolecular abstraction of proton (Fig. 4) or 2)intramolecular abstraction of proton (Fig. 5).

In the event of intermolecular abstraction of protons in an alkaline medium, peroxide (2) is generated together with •OH and •H. In the next step 2-hydroxyl-PZ (3) is formed, further oxidation results in the generation of (4). The [(2- aminoethyl) amino]acetaldehyde (5) undergoes further oxidization to give oxalic acid (7) and EDA (8). EDA can further degrade to give glycine (9) and glycolic acid (10). In a complex series of reactions oxalic acid and EDA can react to form 2-oxopiperazine (OPZ). The aldehyde (5) can also oxidize to give carboxylic acid (6) which undergoes ring closure reaction to give OPZ, as shown in (Fig. 4) [6-8].

Intramolecular proton abstraction may take place resulting in the peroxide radical which in the presence of •OH and through a homolytic cleavage of C2 ̶ C3 bond yields imine (11). Compound (11) degrades further to give unstable [(2-aminoethyl)amino}acetaldehyde] (12) and formaldehyde (13). Further degradation of (12) can form EDA (8) and formaldehyde. Formaldehyde undergoes oxidation to give formic acid (14) which in the presence of PZ can yield FPZ, see Fig. 5 [6- 8]

In other products, acetate and PZ can also react together to form acetyl piperazine (APZ), whereas, reaction of piperazine in the presence of CO2 gives piperazine carbamate, which undergoes nitrosation in the presence of nitrite to give 1-nitrosopiperazine (MNPZ), see Fig. 6 [9,10]

2.3 Overview of AMP Thermal Degradation

Gå til overskriften

The main thermal degradation product of AMP is DMOZD. It is formed by the reaction of AMP with CO₂ to form AMP-carbamate which is unstable and undergoes cyclisation and dehydration to yield the oxazolidinone DMOZD, see Fig. 7. Since AMP is branched and sterically hindered, it is less prone to further secondary degradation reactions compared with MEA. Thus, it is limiting the secondary degradation products formation. According to Gouedard et al. [11], some other thermal degradation products include 2,2-trimethylethanolamine, 4,4-trimethyloxazolidin-2-one, 4,4- dimethyl-1- hydroxytertiobutylimidazolidin-2-one, and l,3-bis(2-hydroxy-l,ldimethylethyl)urea [11,12].

2.4 Overview of AMP Oxidative Degradation

Gå til overskriften

The mechanism for the oxidative degradation of AMP via peroxyl radical mechanism, is shown in Fig. 8. Wang et al. [7] proposed an intramolecular hydrogen abstraction from either NH or CH bond that resulted in imine and enamine formation. Both the imine and enamine can be degraded further to give (3) and (4) with the loss of OH radical. Hydrolysis of the enamine results in the formation of ammonia and formaldehyde. The hydrolysis of compound (4) can give secondary degradation products such as acetone, acetic acid ammonia, nitrate and nitrite etc. Another major product from oxidative degradation of AMP is 2,4-lutidine. The formation of lutidine goes through a series of reactions involving imine and formaldehyde, see Fig. 8 [7, 8]

3.  CESAR1 degradation results and rates

Gå til overskriften

The analysis of degradation rate is based on 2 testing campaigns performed, in 2019 ALIGN CCUS on CHP flue gas and in 2020 TCM Owners campaign on RFCC flue gas. More details for flue gas compositions can be found in [13]. Throughout the testing campaigns, samples were taken from lean amine and chemical analysis of the samples were performed to analyze the solvent composition. The main CESAR1 degradation products which were measured and quantified during testing are listed below. All measurements of degradation products were performed by Sintef.

• 2,4-lutidine
• DMOZD – 4,4-dimethyl-2-oxazolidinone
• MNPZ – N-methylpiperazine
• OPZ – 2-oxopiperazine
• Organic acids and heat stable salts

Fig. 9 demonstrates the change in degradation product concentration during the ALIGN (CHP flue gas) testing campaign. As can be seen there are 2 main zones where detailed degradation calculations and assessments have been performed.

Zone 1: Operation from September 12, 2019, to October 12, 2019, during this testing period operation the NO₂ absorbed by the CESAR1 was on average 0.5 ppmv. This concentration of NO₂ is considered low and was the result of constant ammonia feed to the SCR upstream of the amine plant.

Zone 2: Operation from October 12, 2019, to November 1, 2019, during this testing period operation was quite stable and the NO₂ absorbed by the CESAR1 solvent was on average 2.35 ppmv. This concentration of NO₂ is considered high and was the result of no ammonia feed to the SCR upstream of the amine plant. This was not a planned test, however the variation of NO₂ concentration will allow a comparison of CESAR1 degradation rates with high and low NO₂ concentrations.

The results clearly demonstrate an increased slope of nitrosamine formation in Zone 2, which as expected coincides with a significant increase in NO₂ entering the absorber.

For the 4 quantified non-volatile degradation products (DMOZD, 2,4-lutidine, MNPZ, OPZ) the total accumulation is determined for Zone 1 and Zone 2 and presented in Table 2. It should be noted that 2,4-lutidine is not included because the formation rate was observed to be zero. The rates of formation of MNPZ and OPZ have increased in Zone 2 as compared to Zone 1, due to the increased NO₂ concentration. To calculate the total amine losses resulting from degradation, it is required to know the individual formation for each degradation product and to use the stoichiometric relationship with amine moles consumed. A common manner of expressing amine losses through degradation is by the following expression: Amine loss ratio = kg of amine loss/ton CO₂ captured.

Table 2. Degradation products total accumulation in Zone 1 and Zone 2.

For this calculation it is required to quantify the total amine losses through degradation and total CO₂ captured in Zone 1 (low NO₂) and Zone 2 (high NO₂), the results are shown in Table 3. The results demonstrate that the amine loss ratio through degradation has doubled in Zone 2 as compared to Zone 1, a result of the increased NO2 concentration.

Table 3. Amine loss ratio through degradation (quantified degradation products).

The rates of degradation above consider the quantified degradation products measured. However, it should be noted that there can be other unquantified degradation products which would contribute to additional amine loss through degradation. A way to assess if there are significant unquantified degradation products, is to observe the alkalinity or nitrogen balance for solvent samples taken during the test campaign. An assessment was done to observe how the alkalinity difference is closing for the CESAR1 solvent at different times in Zone 1 and Zone 2. This was achieved by comparing the measured solvent alkalinity versus a calculated solvent alkalinity. The measured alkalinity is based on titration results where all alkaline components (known or unknown) in the solvent will contribute to total solvent alkalinity whereas the calculated solvent alkalinity considers the solvent amines and quantified degradation products. Fig. 10 demonstrate the alkalinity difference during the testing period.

The alkalinity difference increases in Zone 2 as compared to Zone 1, where at the end of Zone 2 the alkalinity difference is around 5%. This signals that as time increases, especially with high NO₂ that there are unknown or unquantified degradation products which are increasing in the CESAR1 solvent. Therefore, it can be expected that the amine loss ratios through degradation presented in Table 3 are underestimated. A revised calculation of amine loss ratio through degradation including the above alkalinity difference is shown in Table 4. The amine loss ratio in Zone 2 has increased (~ 1.85 times) from 0.091 kg amine/ton CO₂ (excluding unknown degradation products) to 0.169 kg amine/ton CO₂ (including estimate of unknown degradation products). There is no significant change or adjustment in Zone 1 amine loss ratio since the alkalinity difference was close to zero. Analysis performed to understand the degradation products contributing to the alkalinity difference are described in [14] and Section 6 of this article.

A similar degradation assessment was performed for RFCC flue gas testing during the CESAR1 Owners Campaign 2 where the amine loss ratio through degradation is shown in Table 5. The results demonstrate a ~ 5 times higher amine loss ratio through degradation when the alkalinity difference calculation (0.120 kg/tonCO₂) is used instead of only quantified degradation products (0.025 kg/tonCO₂). It is believed that the impurities in the RFCC flue gas are contributing to additional unknown degradation products in solution, as compared to CHP flue gas testing.

Table 5. Owners 2 Campaign (RFCC) Amine loss ratio through degradation.

It should be noted that the amine loss ratio’s in (RFCC) are lower than the values in Table 4 (CHP). However, since the amine loss ratio is expressed as amine loss/ton of CO₂ capture, the total amount of CO₂ captured has a large influence. For RFCC the total CO₂ captured is significantly higher than CHP due to the concentration difference in the flue gas. To illustrate this behaviour an example is presented below where key the assumptions and results are presented in Table 6. The results demonstrate the amine loss rate through degradation is ~ 2.6 times higher for RFCC flue gas (6.4 kg amine loss/hr) as compared to CHP (2.5 kg amine loss/hr) for the example below.

Table 6. Example comparing absolute amount of amine losses through degradation for CHP and RFCC flue gases.

4.  Degradation comparison for CESAR1 and MEA

Gå til overskriften

This section will provide some comparisons between CESAR1 and MEA amine-based solvents. The MEA results on amine loss rate were extracted from a TCM publication summarizing CHP flue gas testing results in [13]. Firstly, in Table 7 the total amine loss ratio is compared, where results demonstrate significantly higher amine losses for MEA as compared for MEA, for both CHP and RFCC flue gas.

Table 7. Amine loss ratio for CESAR1 & MEA.

1. Amine loss ratio calculated based on high NO₂ region and alkalinity difference approach (see section 3)
2. Amine loss ratio calculated based on alkalinity difference approach (see section 3)
3. Total amine loss is presented for MEA. For CHP flue gas amine loss due to emission is very low (< 1%) and will not impact the comparison.
4. Total amine loss is presented for MEA. For RFCC average amine loss to emission is 8% of total and will not impact the comparison.

Also, an indicator of degradation is iron (Fe) concentration in an amine solvent. As can be seen from Fig 11a below the Fe concentration is significantly lower for CESAR1 as compared to MEA, indicating lower amine degradation caused by oxidative and thermal effects.

An additional comparison of solvent degradation for CESAR1 and MEA is shown in Fig 11b. Two trends are shown, total nitrosamine (TONO) and heat stable salts (HSS) versus time. The results demonstrate that degradation to heat stable salts is significantly higher for MEA as compared to CESAR1, whereas the opposite behaviour is observed for total nitrosamines. For CESAR1, the piperazine component is a secondary amine and is very prone to nitrosamine formation in the presence of NO₂. MEA on the other hand is a primary amine and forms nitrosamine at a significantly lesser rate.

5.  CESAR1 thermal reclaiming

Gå til overskriften

The 2020 CESAR1 TCM Owners campaigns were performed with the already used CESAR1 solvent from the ALIGN CCUS test campaign in 2019 [14]. After roughly 1600 hours of CO₂ capture plant operation, reclaiming of the solvent was deemed necessary at the start of the first owners test campaign in April 2020. The main goal was to remove the solvent degradation products and metals that had accumulated during operation and refresh the solvent for the first owners test campaign. Later, after an additional 2200 hours of CO₂ capture plant operation was conducted which was followed by another thermal reclamation campaign in October 2020.

Both reclaiming operation periods were conducted in a semi-continuous operation mode similarly as previously done with MEA [15]. Fig. 12 shows a process flow diagram of the TCM plant, including the reclaiming system. The reclaimer was initially filled with a mix of demineralized water and an aqueous solution of 33 wt% NaOH. The mix was heated to around 130°C in the circulation loop before the lean amine feed was started and continued at a rate of approximately 2 m³/h. Approximately 8 ml/min of NaOH was added continuously during the reclaiming. Metals, heat stable salts (HSS) and degradation products accumulated in the reclaimer while amine and water were evaporated and led back to the stripper. This operation continued for 4 days before the amine feed was stopped. A total of 4.5 times the main plant total inventory was circulated through the reclaimer during the operation period. The temperature in the reclaimer was then raised to 144 °C in between addition of water for 2 days while the remaining amine and water evaporated and the amount of waste in the reclaimer was reduced. It was suspected that the high temperature could induce further degradation in the stream going back to the stripper, although only a limited amount of solvent is exposed to this temperature. The maximum temperature for the second reclaiming was thus limited to 137 °C. The reclaimer vessel was then emptied and cleaned, and the waste sent for disposal in three intermediate bulk containers (IBCs).

Table 8 below shows the removal rates from the reclaiming operations. The solvent quality is significantly improved during this operation. The average rates of removal were highly encouraging; 94% for metals, 89% for HSS and 83% for other known degradation products and the loss of amine through waste was lower than 5%. Long boil-off time at the end of reclaiming has been beneficial in terms of low amine loss. However, caution must be taken since exposure of high temperatures during boil-off can risk further degradation. The viscosity of waste was relatively low due to the low amine concentration, which makes draining of the reclaimer waste less challenging. Precautions are needed when handling thermal reclaimer waste as the content will contain nitrosamines and other potentially harmful degradation products.

Table 8. Loss of amine and removal of degradation products, heat stable salts and metals during reclaiming of CESAR1 solvent.

6.  Way forward to gain a better knowledge on degradation

Gå til overskriften

TCM has gained valuable knowledge and experience with the use of the non-proprietary solvents 30 wt% monoethanolamine (MEA) and CESAR1 for CO₂ capture. TCM has a laboratory scale solvent degradation rig (SDR) to allow further investigation on solvent degradation. This SDR can mimic the process conditions and configurations designed for CO₂ capture at the TCM plant. For example, similar process conditions can be in place for the flue gas, absorber and stripper as carried out during CESAR1 test campaigns [16, 17]. TCM plans to use the SDR to gain further valuable insights into the rate of solvent degradation, identification of unknown degradation products and degradation reaction mechanisms/pathways. A SDR test campaign is planned with aged CESAR1 solvent, and the results will be compared with the CESAR1 test campaigns to close the current knowledge gaps. The SDR campaign results can provide a good understanding to solvent health and environmental risks for the CESAR1 solvent systems for CO₂ capture. The results can also give sufficient knowledge of the solvent degradation expected and unexpected in a real CO₂ capture plant.

The basic goal of the SDR campaign will be to demonstrate the bench-scale studies of solvent process degradation to identify unknown components, which were previously difficult to analyze. Online instruments such as FTIR and Raman spectrometer will be utilized with SDR for monitoring solvent hygiene. In addition, degraded components will be qualitatively and quantitatively characterized by using an advanced liquid chromatography-mass spectroscopy (LCMS) instrument. During this study, new analysis methods will be developed for the degraded components of the LCMS instrument, this work will be carried out at TCM lab.

7.  Conclusions

Gå til overskriften

In this article the solvent degradation for CESAR1 has been explored. Firstly, the expected degradation mechanism’s for the main amine components (piperazine and AMP) have been presented, along with the key degradation products which have been measured and quantified during TCM testing. Results of degradation have been compiled for tests for two flues gases (1) CHP flue gas (~ 3.7 vol% CO2 and ~ 14% O2) and (2) RFCC flue gas (~ 13.5 vol% CO₂ and ~ 3.2% O2). For CHP flue gas, the results indicate a strong dependency on degradation rate as the NO2 concentration entering the CO₂ absorber is elevated. This increase in NO2 concentration leads to a greater amount of nitrosopiperazine and other degradation products in the solvent. As the CESAR1 solvent becomes more degraded, approximately 5% of unknown degradation products have been idenitfied. For RFCC flue gas, higher amine loss rates through degradation have been observed and it is believed this can be attributed to the increase of impurities in the RFCC flue gas as compared to the CHP flue gas. Future work at TCM is planned to identify the unknown degradation products and to determine what additional process conditions are key contributors to solvent degradation. A comparison of overall degradation rates for CESAR1 versus MEA was also presented, comparing rates of degradation under similar flue gas conditions. The results indicate significant reduction in degradation rate for CESAR1 compared to previous MEA tests performed at TCM. Lastly, thermal reclaiming experiments have been performed on the degraded CESAR1 solvent with a goal of refreshing the solvent for future tests. Two semi-continuous thermal reclaiming operations were conducted, the average rates of removal were highly encouraging; 94% for metals, 89% for HSS and 83% for other known degradation products and the loss of amine through waste was lower than 5%.

Acknowledgments

Gå til overskriften

The authors gratefully acknowledge the staff of TCM DA, Gassnova, Equinor, Shell and TotalEnergies for their contribution and work at the TCM DA facility. The authors also gratefully acknowledge Gassnova, Equinor, Shell, and TotalEnergies as the owners of TCM DA for their financial support and contributions.

References

Gå til overskriften

• Freeman, S.A., (PhD thesis), Thermal Degradation and Oxidation of Aqueous Piperazine. University of Texas at Austin 2011.
• S.A. Mazari et al., Degradation study of piperazine, its blends and structural analogs for CO2 capture: A review. Int. J. Greenh. Gas Control 2014, 31, 214–228.
• Nielsen, P.T., et al., Piperazine degradation in pilot plants. EnergyProc. 2013, 37, 1912–1923
• Freeman, S.A., et al. Thermal degradation of aqueous piperazine for CO2 capture. 1. Effect of process conditions and comparison of thermal stability of CO2 capture amines. Ind. Eng. Chem. Res. 2012, 51 (22), 7719–7725.
• Freeman, S. A., et al. Thermal degradation of aqueous piperazine for CO2 capture: 2. Product types and generation rates. Ind. Eng. Chem. Res. 2012,51 (22), 7726–7735.
• Namjoshi, O., et al., Thermal degradation of piperazine blends with diamines. Energy Proc. 2013, 37, 1904–1911.
• Mark J. G, et al., Kinetics of N-Nitrosopiperazine Formation from Nitrite and Piperazine in CO2 Capture. Environ. Sci. Technol.,2013, 47 (7), 3528-3534.
• Wang, T., et al., Oxidative degradation of aqueous 2-amino-2-methyl-1-propanol solvent for post combustion CO2 capture. Ind. Eng. Chem. Res. 2012, 51 (18), 6529–6536.
• Wang, T., et al., Oxidative degradation of aqueous PZ solution and AMP/PZ blends for post combustion carbon dioxide capture. Int. J. Greenh. Gas Control 2014, 24, 98–105.
• Fine, N.A., et al., Thermal decomposition of N-nitrosopiperazine. Energy Proc. 2013, 37, 1678–1686.
• Gouedard, C., et al., Amine degradation in CO2 capture. I. A review. Int. J. Greenh. Gas Control 2012 10, 244.
• Naser, S. M., et al., Thermal Degradation Rate of 2-Amino-2-methyl-1-propanol to Cyclic 4,4-Dimethyl-1,3-oxazolidin-2-one; Mechanistic Aspects and Kinetics Investigation. Ind. Eng. Chem. Res., 2017, 56, 34, 9437–9445.
• Morken., et al., CO2 capture with monoethanolamine: solvent management and environmental impacts during long term operation at the Technology Centre Mongstad. International Journal of Greenhouse Gas Control, 2019, 82, 175–183.
• Benquet, et al., First Process Results and Operational Experience with CESAR1 Solvent at TCM with High Capture Rates (ALIGN-CCUS Project). 15^th International Conference on Greenhouse Gas Control Technologies, GHGT, 2021.
• Morken, et al., Degradation and Emission Results of Amine Plant Operations from MEA Testing at the CO2 Technology Centre Monstad. Energy Procedia. 2017, 114, 1245-1262.
• Aslak, E. et.al., A new test rig for studies of degradation of CO2 absorption solvents at process conditions; comparison of test rig results and pilot plant data for degradation of MEA. 2013, 37, 717-726.
• Gjernes, E. et.at., Health and environmental impact of amine based post combustion CO2 capture, Energy Procedia. 2013, 37, 2013, 735-742.

Multivariate data analysis of online-sensors and spectroscopic data for the prediction of solvent composition parameters for MEA (2022)

Lars E. Williams^a, Audun Drageset^b*, Bjørn Grunga^a
aUniversity of Bergen, Department of Chemistry, Allégaten 41, 5020 Bergen, Norway ^bTechnology Centre Mongstad (TCM), 5954 Mongstad, Norway

Abstract

Gå til overskriften

Cost-effective operation of amine-based post-combustion CO₂ capture facilities is important for successfully implementing the technology on a broad industrial scale to reach current climate objectives. Technology Centre Mongstad has benchmarked performance of such technologies in a generic amine plant since 2012. This work utilized historic plant process and laboratory data collected during a test campaign with 2-aminoethan-1-ol (MEA) in 2015. The aim of this work was to employ multivariate analysis to develop models to predict laboratory results for CO₂ content (Total Inorganic Carbon) and amine functionalities (total alkalinity) in the amine solvent. Predictive models were made based on process variables alone, spectroscopic data alone and data fusion models. The process model could explain 99% of the variance for total inorganic carbon in the Lean solvent stream. The Rich solvent is more chemically complex and requires the use of spectroscopic data to explain 95-99% of the variance. In this work we demonstrated how multivariate data analysis can be employed to predict solvent parameters that can be reported in real time for improved control of the capture process.

1. Introduction

Gå til overskriften

Decarbonizing heavy industries is a key for achieving the carbon mitigation goals outlined in the 6th report from the Intergovernmental Panel on Climate Change (IPCC-6) [1]. Amine-based carbon capture is among the most mature technologies for decarbonizing existing industrial point sources for CO₂ emissions. Technology Centre Mongstad (TCM) has operated and demonstrated generic and proprietary amine solvents for post-combustion carbon capture (PCCC) since 2012 [2]. TCM is located on the west coast of Norway in the vicinity of Equinor’s oil refinery at Mongstad. With access to two distinctly different industrial flue gases: combined-cycle gas turbine (CCGT)-based combined-heat-and-power plant (CHP) and RFCC (Residual fluid catalytic cracker) and the ability to manipulate these flue gases (through dilution and CO₂ recycling), TCM can assess CO₂ capture technologies under conditions that are representative of multiple industries emissions [3]. Among the main objectives of TCM’s test campaigns are risk reduction (economic and environmental) for commercial application and full-scale deployment of Carbon Capture and Storage (CCS). Key among these test campaigns are the open test campaigns with non-proprietary solvents like aqueous 2-aminoethan-1-ol (commonly known as Monoethanolamine or MEA) and the aqueous blend of 2-amino-2-methylpropan-1-ol (AMP) and Piperazine (PZ). Data and learnings from these campaigns can be disseminated in line with TCM’s purpose to ensure safe technology implementation to combat climate change.

MEA is a first-generation amine-based CO₂ capture solvent. Amine based absorption is a reversible reaction between an aqueous amine and an industrial flue gas containing CO₂. The basic amine functionality reacts with CO₂ to form a carbamate, removing the CO₂ from the gas and trapping it in the liquid phase. This reaction is carried out in the absorber containing a packed gas-liquid contactor to ensure high mass transfer between the two phases at lower temperatures (30 – 60 °C, depending on solvent and plant configuration). The reaction can then be reversed by applying heat via a steam boiler in the stripper section (see Figure 1). CO₂ is released as a product gas and the liquid amine is regenerated and returned to the absorber. A capture plant operating with MEA can capture over 90% of the CO₂ (advanced solvents have demonstrated over 98% capture) and generates CO₂ product with high purity (99.9%) [3a]. The major challenge is the operational cost [4]. The capture plant should be operated under optimal conditions to minimize energy consumption mainly tied to removing CO₂ from the solvent. This is often reported as Specific reboiler duty (SRD). To achieve this, operators are reliant on accurate gas composition data at the inlet and outlet of the absorber as well as the solvent composition. Gas composition is monitored online (via gas chromatography or infrared spectroscopy) and operators can quickly act on any changes in for example the CO₂ concentration from the industrial source. In contrast, the solvent composition is usually measured through extractive samples and laboratory analysis, and results are only available after multiple hours and in some cases days.

This work utilized plant data from TCM’s MEA test campaign conducted in 2015 (July to October) funded by Gassnova, Equinor (former Statoil), Shell and Sasol (TCM’s owners in that period). The campaign’s primary objective was to conduct an updated baseline and plant performance with MEA and to verify plant mass balance over a set of operational conditions, as well as other technology knowledge gaps [5]. The plant was operated with a 30 wt% aqueous MEA solution and the CHP flue gas (3.6 vol% CO₂) for most of the test period. Throughout the campaign laboratory samples were collected and analysed to (a) ensure tests were conducted with correct amine concentration, (b) record resulting lean loading (mole CO₂ per mole of amine) during process optimization and (c) monitor solvent degradation and plant corrosion.

The TCM amine plant has a large array of analytical instruments in the rich (CO₂ rich solvent after absorption) and lean (CO₂ lean solvent after stripping solvent) streams. Among these are temperature, pH, conductivity, density, and pressure, see Figure 1. The data is used for general purpose applications in characterising the physical parameters during a test campaign. Such analytical instruments can potentially be used to predict solvent parameters currently only available via laboratory analysis like (1) Total Inorganic Carbon (TIC); a measure of CO₂ in the solvent, (2) Total alkalinity (total NH functionality in the solvent as determined via an acid base titration) and (3) amine concentration. These are used to calculate the CO₂ loading of the solvent (mole CO₂ per mole of amine). During the campaign Total alkalinity was used as an analogue for amine concentration as the NH functionality is predominantly from MEA. It is expected that predictive models would benefit from the addition of chemical information acquired through spectroscopy, as spectroscopic methods like FTIR can give information about chemical bonds and functionality present in the solvent. Such chemical information is necessary if the degradation [6] of the solvent and its impact on the plant are to be monitored online [7].

In this paper we present how common online measurement principles like pH and conductivity can be used to predict solvent parameters which were previously only obtained through laboratory analysis. In addition, the limitations of this concept as well as how the addition of spectroscopy can improve the model accuracy is discussed. The implementation of such models can improve plant efficiency and lower the frequency of sampling and analysis resulting in a reduction of both exposure risks to operators and as well as costs for operating a laboratory.

1.1 Latent variable modelling

Gå til overskriften

Latent variables are linear combinations of the measured variables. They represent excellent tools for data visualization and quantitative modelling. Two main types of latent variable analysis have been used in this work. Principal Component Analysis (PCA) [8] and Partial Least Squares Regression (PLS) [9].

Any data set not based on an orthogonal design will have correlations among the variables. This makes it possible to go from a large number of measured variables to a much smaller number of latent variables while preserving the information content of the data. This concept is particularly useful for data exploration but can also be used in classification and regression analysis. In PCA, latent variables are referred to as principal components. These are constructed so that they capture as much of the variance in the data as possible. This is known as the maximum variance criterion. Each measured variable contributes to each latent variable, but the amount of contribution is different for different variables. Each measured object has a score on each latent variable, just like every object has a value for the measured variables. The collection of scores on a latent variable is referred to as a score vector, and a bivariate scatter plot of the first two score vectors after PCA is the two-dimensional plot that explains as much variation as possible. For this reason, PCA is extremely popular for data exploration, and it is used in a variety of scientific fields, although under different names.

All data contains noise. This fact means that the usefulness of principal components extends beyond data exploration. They can be used in classification and discrimination analysis, and regression. This can be done by calculating enough principal components to capture the systematic variation in the data. Using these principal components in further analysis and ignoring the residual variation left unmodelled ensures that noise in the data does not pollute the quality of the subsequent models. In this aspect, PCA can be seen as a denoising technique.

In this work, principal components have been used for data fusion [10], which is the term used for uniting different sets of data into one data set. The present work deals with traditional process data and infrared spectra used to characterize the state of an amine solution used to absorb CO₂. It is tempting to combine these two measurement types to better describe the system’s state. Doing so, one immediately finds that the number of relevant spectral variables is significantly larger than the number of process variables. Simply fusing the two sets of measurements by adding one set of variables to the other will lead to a data matrix completely dominated by the spectral data. Deleting spectral variables so that the two sets become equal in size is another strategy, but there is a risk of throwing away useful information.

Furthermore, making a proper variable selection is not trivial. In this work, PCA is done on the spectral data. The number of principal components calculated and retained is enough to capture the systematic variation of the data. This number is independent of the number of variables in the data set but is decided by the number of independent sources of variation in the data. In this way one can reduce thousands of measured variables to a handful of principal components without any information loss. As the spectral variables are very correlated the reduction is substantial. This means the process data is joined with the significant score vectors from a PCA, not the measured spectral variables.

Principal components can also be used in regression analysis. This is referred to as Principal Component Regression (PCR). In multiple linear regression (MLR), one would model a response (e.g., total alkalinity of a solution) as a function of a set of measured variables, such as the process variables used in this work. There are many problems with this approach. In MLR it is assumed that the error in the independent variables is much less than the error in the response. This is not necessarily the case, and if this is violated the MLR model is not the optimal model. A further complication is that the regressors are assumed to be independent variables – they should be independent of each other. If they are not, neither the model predictions nor the interpretation of the effects of the regressors (the regression coefficients) may be trusted. Moving from a set of highly collinear measured variables to a set of orthogonal latent variables and using these as regressors elegantly avoids this problem. A further benefit of this approach is that the principal components are much less influenced by noise than the original data set. Where MLR uses all the data available, a latent variable regression model using principal components would only use the systematic part of the data as regressors.

While PCR in most cases is favourable to MLR, it is still not the preferred latent variable regression technique. The problem with the principal components is that they are constructed to capture all the main sources of systematic behaviour. Not all these systematic sources of variation will be relevant to the regression problem. A simple example would be the prediction of the concentration of a compound in a sample using spectroscopy. While changes in the analyte concentration surely impacts the spectra, so will any changes in the concentrations of other compounds present in the sample. This is systematic behaviour that will be picked up by the principal components, but it is not relevant for the prediction of our analyte. PLS is the solution to this problem.

In PLS the latent variables are calculated differently from the ones in PCA. Instead of focusing on the variance of the independent variables, PLS uses the covariance between the independent variables and the response as the latent variables. This means that the latent variables are directly related to the response modelled. Where PCR asks the question “what are the major sources of variation in my data?”, PLS asks “what part of my data varies in the same way as the response?”. This makes PLS a more powerful regression technique than PCR. They are both latent variable regression techniques, but the latent variables used in the regression are quite different.

2. Experimental section

Gå til overskriften

The TCM amine plant is a generic plant designed and built by Aker Solutions and Kværner with a capacity to treat up to 60 000 Sm³/h post combustion flue gas. The plant was operated with the Combined Heat and Power flue gas with a CO₂ concentration of 3.6 % and aqueous MEA (30 wt%). The absorber tower was operated with an 18- and 24- meter packing section and a lean amine flow of 43 000 – 70 000 kg/h. The stripper section was operated at 120.0 –121.5 °C. Detailed plant parameters for different test phases are described in the literature [5]. Solvent parameters were monitored via in-line liquid conductivity (Endress Hauser, resistance measurement conductivity meter), density (Coriolis mass flow, Proline Promass 80F) and pH meter (Endress Hauser, potentiometric pH measurement) installed after circulation pumps on rich and lean solvent flow (see Table 1). The process data are measured at different time intervals. In this work, values averaged over a period of 15 minutes were used for all variables.

The process data needs to be cleaned up prior to further analysis as the measurements are carried out continuously, regardless of the system state. Measurements taken during outlying conditions were removed from the data sets. Examples of outlying conditions are system shutdown periods, the recovery phase after such shutdowns and periods of MEA reclamation.

Extractive liquid samples were collected via a fast loop system equipped with a process sampler (DOPAK Inc.) by operators on a regular basis and delivered to TCM onsite laboratory for storage and further analysis.
Rich and lean liquid samples were analysed with a Bruker Alpha FTIR Spectrometer with a diamond ATR cell (Bruker Corporation). Spectra were recorded between ~4000 and ~425 cm-1. A plot of the raw spectra (Figure 2) shows that there is no relevant information above 3640 cm-1 or below 800 cm-1. The same applies to the region between 2750 cm- 1 and 1730 cm-1. These regions were removed from the spectra prior to further analysis.

The number of samples used in the models varies for the different responses and type of data. The analytical laboratory did not measure all responses for all samples, and IR spectra were not recorded for all samples. Table 2 shows the number of samples used in each model.

Total Alkalinity is analysed via automated acid base titration with HCl (1.0 M). The reported uncertainty is 2%. Total Inorganic Carbon (TIC) is analysed with a TOC/TN Elementar (Elementar Analysensysteme GmbH). The reported uncertainty is 4%.

3. Results and discussion

Gå til overskriften

3.1 Prediction of properties using only process data

Gå til overskriften

3.1.1 Lean TIC predicted from process data

Gå til overskriften

As the process data is a mix of variables expressed using different measurement units, the data was standardized and mean centred prior to modelling. An initial PLS regression of the process data with the lean TIC as a response yielded an 8-component model explaining more than 98 % of the variance in the TIC. The number of components was determined using cross validation [11]. This is more than satisfactory as the uncertainty in the laboratory measurements is approximately 4 %.

Not all recorded variables contribute to the regression models. Since the presence of irrelevant predictors may be of detriment to the model, care was taken to remove any predictor variables with a small regression coefficient and a large uncertainty. This also leads to a simpler model – one with fewer PLS components. The final Lean TIC model contained six PLS components and captured 99 % of the variance of the response. Figure 3 shows the regression coefficients of the variables in the final model.

3.1.2 Rich Total alkalinity predicted from the process data

Gå til overskriften

The enriched amine solution resulting from the CO2 capture is more complex chemically. Various chemical reactions take place in the mixture, and it is expected that a model based on the process data alone will struggle to explain the behaviour of the solution.

Figure 4 shows the poor performance of an optimized (irrelevant variables removed) PLS model of the Rich Total Alkalinity value. The two-component model only explains 52.81% of the response. This demonstrates that while the process variables may be enough to satisfactorily model the simpler lean system, more information is needed to obtain acceptable models for the more complex rich flow.

3.2 Predictions from IR spectra

Gå til overskriften

As shown in 4.1.2, process data alone is not sufficient for prediction of many of the properties. Infrared spectra represent an alternative information source. An infrared spectrum contains information on the functional groups and chemical bonds present in a sample. Figure 6 shows two spectra: One from a lean sample and one from a rich sample. It is evident that the two samples have strong similarities, but also that there are differences.

The models presented in this section use whole spectral profiles (from the regions described in the Experimental section) as predictor variables. This necessitates different pre-treatment compared to the process data. Standardization is not an appropriate tool for profile data, as that would inflate small, noisy variables and reduce the contribution from the larger, more interesting variables. The effects that usually cause problems when using spectral profiles as predictors are additive baseline effects and multiplicative effects due to light scattering. In this work, the baseline effects were handled by Savitzky-Golay second order differentiation [12] with a window size of 25 and a cubic polynomial. The multiplicative effects were handled by Extended Multiplicative Scatter Correction [13].

3.2.1 Rich Total Alkalinity modelled from IR spectra

Gå til overskriften

Figure 7 shows the performance of a three component PLS model for predicting the total alkalinity in the rich samples. For this model, the Total Alkalinity values were root transformed. This improves the model quality. The model captures 95.56 % of the variance of the data and compared to the model shown in Figure 5 the IR model improves the performance to a remarkable extent.

3.2.2. Rich TIC modelled from IR spectra

Gå til overskriften

Although not shown in this work, attempts were made to model the TIC of the rich solvent using only the online process data. The resulting model was only able to explain 59.46% of the response variance. It is therefore interesting to see if the performance improves when replacing the process data with the IR spectra. The best model of the IR data explained 81.66% of the response. While this is an improvement, Figure 8 shows that there is still room for improvement. The next logical thing to try is then a combination of the process and IR data.

3.3 Combining process data and IR spectra: Data fusion model of the rich TIC

Gå til overskriften

An immediate challenge presents itself when trying to combine the online process data and the IR spectra. The number of spectral variables is more than 250 times larger than the process variables. This means that a simple fusion where the spectra are simply added to the process variables will lead to a data matrix completely dominated by the spectral information. More advanced fusion methods are therefore needed.

In this work, the spectral data was decomposed using Principal Component Analysis. Enough components were extracted to explain 99 % of the variance of the spectra. The number of components, which typically is quite small, was found using cross validation. The corresponding significant score vectors were subsequently appended to the process data, resulting in new variables.

Three principal components were enough to reach more than 99 % variance explained for the spectra used in the modelling of the rich TIC. The resulting model was able to explain 87 % of the variance of the response. The performance is illustrated in Figure 9. While far from perfect, it is still the best model variable, and is good enough to predict the general variations in the response.

While the model still has problems with picking up the finer changes in the response, it is clear that the combination of process and IR data has improved the performance.

4. Conclusion

Gå til overskriften

This work has demonstrated the possibility of predicting relevant parameters describing the state of a MEA solvent used in CO₂ capture. For some responses, solid models are achieved only using online process measurements. More complex situations benefit from the usage of IR spectra. The best models are achieved by combining both sets of data using data fusion techniques.

This work has been carried out on historical data from a campaign several years old. This represents challenges, as the detection and exclusion of outlying conditions becomes more difficult. A model is never better than the data from which it was created, so cleaning the data from outlying conditions is important.

This work demonstrates how multivariate models from process and spectral data can be used to predict the state of the solvent mixture and the efficiency of the capturing process. It is paramount that the sample set from which the model is created is representative for all the states the system may occupy in the future. Samples from both fresh and various degrees of degraded solvent states must be used. This should not represent a problem for a plant operating under relatively constant conditions.

Continuous monitoring of plant conditions is vital for cost-effective plant operation. This includes the monitoring of solvent conditions both in the lean and the rich stream. Laboratory analyses contribute to the operational expenses (OPEX) of the capture plant and a significant reduction in the scope of laboratory analysis can be a major cost saving for a full-scale plant. In addition, models can be incorporated into the control system so operators can take immediate action to keep the plant running optimally. As this work demonstrates, models can accurately predict standard laboratory parameters and can improve response time to changes in the plant. This work illustrate how technology developers can utilize process measurements and spectroscopy to improve the control of the plant. The method is not solvent or plant specific and can be used in screening historic datasets to guide further technology development.

5. Acknowledgment

Gå til overskriften

The authors gratefully acknowledge the staff of TCM DA, Gassnova, Equinor, Shell and TotalEnergies for their contribution and work at the TCM DA facility. The authors also gratefully acknowledge Gassnova, Equinor, Shell, and TotalEnergies as the owners of TCM DA for their financial support and contributions.

References

Gå til overskriften

[1] Intergovernmental Panel on Climate Change (IPCC), 2021: Summary for Policymakers. In: Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [MassonDelmotte, V., P. Zhai, A. Pirani, S.L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen, L. Goldfarb, M.I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J.B.R. Matthews, T.K. Maycock, T. Waterfield, O. Yelekçi, R. Yu, and B. Zhou (eds.)]. Cambridge University Press.
[2] a) E. Gjernes, S. Pedersen, D. Jain, K. I. Åsen, O. A. Hvidsten, G. de Koeijer, L. Faramarzi, T. de Cazenove, Documenting modes of operation with cost saving potential at the Technology Centre Mongstad, 14th Greenhouse Gas Control Technologies Conference Melbourne 21-26 October 2018 (GHGT-14), b) A. Morken, S. Pedersen, E. R. Kleppe, A. Wisthaler, K. Vernstad, Ø. Ullestad, N. E. Flø, L. Faramarzi, E. S. Hamborg, Energy Procedia, 114, 2017, 1245-1262, c) C. Benquet, A. Knarvik, E. Gjernes, O. A. Hvidsten, E. R. Kleppe, S. Akhter, First Process Results and Operational Experience with CESAR1 Solvent at TCM with High Capture Rates (ALIGN-CCUS Project), Proceedings of the 15th Greenhouse Gas Control Technologies Conference 15-18 March 2021.
[3] a) K. Johnsen, E. R. Kleppe, L. Faramarzi, C. Benquet, E. Gjernes, T. de Cazenove, A. K. Morken, N. Flø, M. I. Shah, M. Aronsson, Ø. Ullestad, CO2 product quality: assessment of the range and level of impurities in the CO2 product stream from MEA testing at the Technology Centre Mongstad (TCM), Proceedings of the 14th Greenhouse Gas Control Technologies Conference Melbourne 21-26 October 2018 (GHGT-14), b)G. Lombardo, M. I. Shah, B. Fostås, O. A. Hvodsten, L. Faramarzi, T. de Cazenove, H. Lepaumier, P. Rogiers, Results from testing of a Brownian diffusion filter for reducing the aerosol concentration in a residual fluidized catalytic cracker flue gas at the Technology Centre Mongstad, 14th Greenhouse Gas Control Technologies Conference Melbourne 21-26 October 2018 (GHGT-14).
[4] a) E. Gjenres, S. Pedersen, D. Jain, K. I. Åsen, O. A. Hvidsten, G, de Koijer, L. Faramerzi, T, de Cazenove, Documenting modes of operation with cost saving potential at the Technology Centre Mongstad, 14th Greenhouse Gas Control Technologies Conference Melbourne 21-26 October 2018 (GHGT-14). b) L. Faramerzi, D. Thimsen, S. Hume, A. Maxon, G. Watson, E. Gjernes, B. F. Fostås, G. Lombardo, T. Cents, A.K. Morken, M. I. Shah, T, de Cazenove, E. S. Hamborg, Results from MEA testing at the CO2 Technology Centre Mongstad: Verification of baseline results in 2015, Energy Procedia, 2017, 114, 1128-1145.
[5] E. Gjernes, S. Pedersen, T. Cents, G. Cents, G. Watson, B. F. Fostås, M. I. Shah, G. Lombardo, C. Desvingnes, N. E. Flø, A. K. Morken, T. de Cazenove, L. Faramerzi, E. S. Hamborg, Energy Procedia, 114, 2017, 1146-1157.
[6] N. E. Flø, L. Faramerzi, T. de Cazenove, O. A. Hvidsten, A. K. Morken, E. S. Hamborg, K. Vernstad, G. Watson, S. Pedersen, T. Cents, B. F. Fostås, M. I. Shah, G. Lombardo, E. Gjernes, Results from MEA Degradation and Reclaiming Processes at the CO2 Technology Centre Mongstad, Energy Procedia, 2017, 114, 1307-1324.
[7] S. Hjelmaas, E. Storheim, N. E. Flø, E. S. Thorjussen, A. K. Morken, L. Faramerzi, T. de Cazenove, E. S. Hamborg, Results from MEA amine plant corrosion processes at the CO2 Technology Centre Mongstad, Energy Procedia, 2017, 114, 1166-1178.
[8] H. Hotelling, Analysis of a complex of statistical variables into principal components, Journal of Educational. Psychology., 1933, 24, 417-441
[9] S. Wold, M. Sjöström, L. Eriksson, PLS-regression: a basic tool of chemometrics,m Chemometrics and Intelligent Laboratory Systems, 2001, 58, 109-130
[10] E. Borràs, J. Ferré, R. Boqué, M. Mestres, L. Aceña, O. Busto, Data fusion methodologies for food and beverage authentication and quality assessment – a review, Analytica Chimica Acta, 2015, 891, 1-14
[11] P. Filzmoser, B. Liebmann, K. Varmuza, Repeated double cross validation, Journal of Chemometrics, 2009, 23, 160-171
[12] A. Savitzky, M.J.E. Golay, Smoothing and Differentiation of Data by Simplified Least Squares Procedures, Analytical Chemistry, 36, 1627- 1639
[13] H. Martemns, E. Stark, Extended multiplicative signal correction and spectral interference subtraction: new preprocessing methods foir near infrared spectroscopy, Journal of Pharmaceutical and Biomedical Analysis, 1991, 9, 625-635

CO₂ capture with monoethanolamine: Solvent management and environmental impacts during long term operation at the Technology Centre Mongstad (TCM) (2019)

Anne Kolstad Morken^a,b,⁎, Steinar Pedersena,^b, Stein Olav Nessea,^b, Nina Enaasen Fløa,^b, Kim Johnsena,^b, Jane Karin Feste^a,b, Thomas de Cazenove^a, Leila Faramarzi^a,b, Kai Vernstad^c

^aTechnology Centre Mongstad (TCM), 5954 Mongstad, Norway ^bEquinor ASA, P.O. Box 8500, 4035 Stavanger, Norway ^cSintef Materialer og Kjemi, Avd Bioteknologi, Sem Sælands vei 2, 7034 Trondheim, Norway

Abstract 

Gå til overskriften

The owners of the Technology Centre Mongstad (TCM DA) started a monoethanolamine (MEA) test campaign in June 2017. The main objective was to produce knowledge and information that can be used to reduce the cost as well as technical, environmental and financial risks of commercial scale deployment of post-combustion capture (PCC). The campaign covered experimental activities in the amine plant from the 12th of June 2017 until the 30^th of July 2018. A wide range of operating conditions were applied, thus giving a unique opportunity to study the impacts on the solvent quality, degradation behavior, corrosion tendency and emissions to the atmosphere. The current work describes how solvent quality and low emissions to atmosphere can be maintained during long- term operation by appropriate solvent management.

This article is behind a paywall. For futher information: https://www.sciencedirect.com/science/article/abs/pii/S1750583618307576?via%3Dihub

Degradation and Emission Results of Amine Plant Operations from MEA Testing at the CO₂ Technology Centre Mongstad (2017)

Anne Kolstad Morken^a,b,*, Steinar Pedersen^b, Eirik Romslo Kleppe^a, Armin Wisthaler^d, Kai Vernstad^e, Øyvind Ullestad^a,b, Nina Enaasen Flø^a, Leila Faramarzi^a,b, Espen Steinseth Hamborg^a,b

^aCO₂ Technology Centre Mongstad (TCM DA), 5954 Mongstad, Norway ^bStatoil ASA, PO Box 8500, 4035 Stavanger, Norway ^cGassnova SF, Dokkvegen 10, 3920 Porsgrunn, Norway ^dUniversity of Oslo, Department of Chemistry, P.O. Box 1033 Blindern, 0315 Oslo, Norway ^eSintef Materialer og Kjemi, Avd Bioteknologi, Sem Sælands vei 2, 7034 Trondheim, Norway *Corresponding author

Abstract

Gå til overskriften

In 2015, the CO₂ Technology Centre Mongstad (TCM DA), operated a test campaign using aqueous monoethanolamine (MEA) solvent at 30 wt%. The main objective was to demonstrate and document the  performance of the TCM DA Amine Plant located in Mongstad, Norway. This paper will present several aspects concerning degradation of the solvent and atmospheric emissions from amine based CO₂ removal processes. The work aims to; (1) quantify the amounts and compositions of the degraded solvent (2) report results from atmospheric emissions measurements of amines and amine based degradation products; and (3) present Ambient Air measurement done during a 2 month campaign.

1. Introduction

Gå til overskriften

The CO₂ Technology Centre Mongstad (TCM DA) is located next to the Statoil refinery in Mongstad, Norway. TCM DA is a joint venture set up by Gassnova representing the Norwegian state, Statoil, Shell, and Sasol. The facility run by TCM DA entered the operational phase in August 2012 and it is one of the largest post-combustion CO₂ capture test centres in the world.  A unique aspect of the facility is that either a  flue gas slipstream from a  natural gas turbine based combined heat and power (CHP) plant or an equivalent volumetric flow from a residual fluidized catalytic cracker (RFCC) unit can be used  for CO₂ capture. The CHP  flue gas contains about 3.5% CO₂ and the RFCC flue gas contains about 13-14% CO₂. One of the main test plants at TCM DA is a highly flexible and well-instrumented amine plant. The amine plant was designed and constructed by Aker Solutions and Kværner to accommodate a variety of technologies, with capabilities of treating flue gas streams of up to 60,000 standard cubic meters per hour. The plant is being offered to vendors of solvent based CO₂ capture technologies to, among others, test; (1) the performance of their solvent technology, and (2) technologies aimed to reduce the  atmospheric  emissions and environmental impact of amines and amine based degradation products from such solvent based CO₂ capture processes. The objective of TCM DA is to test, verify, and demonstrate CO₂  capture technologies suitable for deployment at full-scale. Up to now the vendors Aker Solutions, Alstom, Cansolv Technologies Inc. and Carbon Clean Solutions Ltd. have successfully used the TCM DA facilities to verify their CO₂ capture technologies.

From July to October 2015 TCM DA, in collaboration with partners, operated a test campaign using the non- proprietary aqueous monoethanolamine (MEA) solvent at 30 wt%.

2. The amine plant and operating conditions

Gå til overskriften

The MEA campaign was started 6th of July 2015 with flue gas introduction to the amine plant. The campaign  lasted to 17th of October 2015. Operational hours are counted as hours  with both flue gas and solvent circulation.  The entire campaign gave a total of 1960 hours of operation (figure 1).

A process flow diagram including sample points for the amine plant is given in figure 2. A more detailed description of the TCM DA amine plant and the TCM sample handling system can be found elsewhere [1,2,3].  Liquid and gas sampling, target component groups and analytical measurement techniques are described in sections 2.3 to 2.5 below.

Several operating conditions are important with respect to the solvent degradation and emission rates of amines and degradation products. Detailed information about the operating conditions and all the test activities and performance results from the MEA campaign, can be found in Gjernes et al [12].

The flue gas composition downstream the Direct Contact Cooler (DCC) from the CHP and the RFCC are providing a range of test conditions and the solvent will be exposed to a corresponding range in CO₂ and O₂ concentrations, as well as NO_x, SO_x and particles. Solvent amines react with the flue gas components and give rise to the degradation products as illustrated in figure 3. Degradation reactions of MEA and specific degradation  products that where monitored during this campaign is given in section 3 below.

When the solvent is exposed to higher temperatures in combination with the flue gas components,  the  degradation reactions are accelerated. Also the accumulation in the solvent of transition metal elements due to corrosion may contribute to degradation [11]. Process units with high temperature exposure are the stripper and reboiler system and the hot part of the solvent circulation loop. For more process details see Table 1. The inventory and the residence time of solvent in the hot areas are decisive for degradation, for more details regarding the inventory see Flø et al [13].

2.1 Liquid samples

Gå til overskriften

The solvent amine, ammonia, and some degradation products were analyzed by TCM DA and Statoil Crude Oil and Products laboratories (CP Lab). Alkyl amines, aldehydes, ketone, generic nitrosamines, solvent specific nitrosamines and nitramines were analyzed by SINTEF laboratories. Total Nitrogen (Kjeldahl) was analyzed by LabNett Stjørdal, table 2 gives an overview of the different techniques used.

Organic acids and anions were measured by Ion Chromatograph (IC) and Total Heat Stable Salts (HSS) by ion exchange and following titration.

2.2 Emission samples

Gå til overskriften

TCM DA applies different measurement techniques to monitor and quantify the amounts and concentrations of emitted compounds. There are three different flue gas streams, flue gas inlet to the absorber (downstream DCC), absorber outlet and CO2-stripper outlet. Online instruments are connected via heated sampling lines to sampling probes. The amine and other emissions were monitored and confirmed by isokinetic sampling and the following online analyzers in Table 3. A full description of emission monitoring at TCM is given in Morken et al [1]. For a more detailed description of the general online equipment see Lombardo and Gjernes [6,12].

2.3 Ambient Air measurements, instrumentation and locations

Gå til overskriften

The ultra-sensitive proton-transfer-reaction quadrupole ion guide time-of-flight mass spectrometer (PTR-QiToF- MS) from IONICON was used for detecting trace gases at low pptv levels in ambient air in the vicinity of Technology Centre Mongstad. These novel ambient air measurements were performed in August and September  2015 by University of Oslo. Measurements were carried out in three different geographic locations, Sundsbø (60º46’10.1’’N, 5º09’08.6’’E), Sande (60º50’56.6’’N, 5º00’21.0’’E) and Mongstad West (60º48’45.7’’N, 5º00’43.4’’E). These sites were chosen from earlier measurement done by Norwegian Institute for Air Research (NILU) and dispersion models done by NILU [5]. For more technical details and results regarding this surveillance see Mikoviny et al [10].

3. MEA solvent and degradation theory

Gå til overskriften

3.1 Oxidative and Thermal degradation

Gå til overskriften

The degradation mechanisms for MEA have been extensively studied in the literature [4,5,8,11,14]. The main degradation reaction pathways with most important degradation products are indicated and proposed in figure 7 below. Oxidative degradation is induced by O₂ and produces oxidized fragments of the solvent. Organic acids, ammonia and aldehydes are the main products from this degradation route. Ammonia and aldehydes are observed in the emission samples. The organic acids react with MEA and various degradation products are formed in subsequent reactions. These products are identified in the solvent samples.

The carbamate degradation route requires CO₂ and fairly high temperatures. The thermal degradation of MEA occurs predominantly in the reboiler and stripper packing due to exposure to high temperature. While the initial products of thermal degradation have been identified, the kinetics of the thermal degradation pathways has not been clearly defined. Davis and Rochelle [14] indicate that thermal degradation is minor when reboiler temperature  is  held below 110°C but it accelerates above 130 °C. Carbamate polymerization due to high temperature is the main cause of thermal degradation of MEA. This degradation is also compounded when the CO₂ loading of the solution is increased. MEA concentrations can be kept at 30 wt % to minimize thermal degradation and prevent corrosion in industrial applications.

3.2 HSS components

Gå til overskriften

Heat Stable Salts (HSS) are salts in the amine solution that is not affected by heat. The heat stable salt does not regenerate in the regenerator and remains in the circulating amine system. Total HSS are measured by a titration procedure which prepares the sample with a strong cation exchange resin. Individual HSS anions are measured by  Ion Chromatography (IC). The different anions measured by IC are summarized in table 4.

The identified anions are summed to provide a total HSS. In general,  Total HSS by titration should be the same  or larger than the sum of anions by IC, figure 6 (h). Total HSS are reported as the wt% of the equivalent amount of amine. This means if HSS concentration were 1 mole/kg (eq/kg) of solution, it will be 6.1 wt% as MEA (1).                                                                    

3.3 Degradation components in solvent, from emission and in Ambient Air

Gå til overskriften

The degradation components measured during the MEA campaign were based on information found from literature [4]. All components from solvent and emission samples in Table 5 were analyzed by Sintef. The analyzing measuring technique was primarily LC-MS-QQQ. The mixture of the different degradation components  are  hereafter called D-mix. Analysis of Ambient Air components were done by University of Oslo [10].

4.  Results and discussions

Gå til overskriften

The first observable sign of degradation was color change of the solvent. The color of the solvent changed rapidly after the first contact with the flue gas. Samples taken before introduction of flue gas show a colorless solvent. Only hours after start up, the color started to change from colorless to yellow, and more and more orange and dark brown as seen in figure 4. After reclaiming 12^th of October, the color is more like the color that appeared in the start of the campaign when the solvent was fresh.

4.1 Heat stable salts in the solvent

Gå til overskriften

Figure 5 and 6 shows how the levels of organic acids and anions developed during the entire campaign. Figure 5 shows overall heat stable salts development where 5a) are Total Heat Stable salts reported as wt% MEA, and 5b) results from individually IC results from each component. Figure 6 (a-g) shows more detailed development of all the individual components. The main anion formed is formate and the level of this component reach 3000 mg/L before reclaiming. Glyoxylic acid is assumed to be one of the formed organic acids during the degradation process [7]. It was not possible to analyze for this component as there were no available method at the time. An unknown component of significant response on the IC chromatogram was found. The area of the unknown component in the chromatogram was significant, and the component was calibrated with a mix of the other components. The result from this unknown component is rather uncertain, see figure 6 g). All other IC results have a  repeatability  uncertainty of ± 20%.

Propionate (C₂H₅COO^–) and nitrite (NO₂-) were not detected above 10 mg/L which is the limit of detection on the Ion Chromatograph.

4.2 Degradation products in the solvent

Gå til overskriften

A simplified scheme for MEA degradation is proposed in figure 7. Oxidation reactions lead to formation of the organic acids and the emission products ammonia and aldehydes. This is indicated in the left blue square of the figure. Reactions between MEA and the organic acids, CO₂ and additional free MEA lead to formation of the degradation products identified in the lean solvent samples. This is indicated in the large red square of the figure. A nitrogen mass balance based on solvent analysis are presented and compared to literature data in section 4.5 below.

The rate of formation of the degradation products is a function of temperature (faster kinetics), CO₂ loading (more carbamate present), and MEA concentration. The identified degradation products in the solvent samples and the accumulation of these as function of operational hours are shown in figure 8.TCM performed a MIST test after 1314 hours of operation and also did a CO₂ recycling test with higher CO₂ content in the CHP flue gas [12]. The results shown after 1314 hours are not consistent with the other samples and cannot be explained. Results from the reclaiming part of the 2015 MEA campaign is given in [13].

It is seen that the dominant degradation products in the solvent are N-(2-hydroxyethyl)glycine (HeGly) and 4-(2- hydroxyethyl)piperazine-2-one (HEPO). This corresponds to the oxidation pathway via glyoxylate and subsequent reaction with MEA given in figure 7. The identification of the nitroso-compound nitroso-Hegly (No-HeGly) in the solvent further confirms this degradation route.

4.3 Nitroso- and Nitramines in solvent

Gå til overskriften

Two solvent specific nitrosamines, N-nitrosodiethanolamine (NDELA) and N-nitroso-2-hydroxyethyl-glycine (Nitroso-HeGly), were detected in the solvent as the degradation process progressed. The total concentrations of nitrosamines (TONO) were measured to be 2351 µmol/L after 1850 hours of operation, see figure 9. Since MEA is a primary amine it is not expected to form a stable nitrosamine. The identified compounds are thus formed from secondary amines occurring as impurities in the solvent or being formed during the degradation reactions. As is shown in Figure 9 a), there are still some unidentified nitrosamines in the degraded solvent sample. These nitrosamines are formed from high molecular weight amines and have low volatility.  Figure 9 b) shows a decrease   in the level of total nitrosamines after reclaiming of the solvent.

Nitrosamines are formed after reaction with NO_x in the flue gas [8]. During the MIST test, RFCC flue gas was used, and as this flue gas contains more NO_x than flue gas from the Combined Heat and Power Plant, this could explain the higher amount of nitrosamines in this MEA2 campaign compared with the first MEA1 campaign from TCM [1].

The solvent specific nitramine (MEA-NO2) was detected at a concentration of approximately 4 mg/L after 1850 hours of operation. Methylnitramine (MA-NO2) and Dimethylnitramine (DMA-NO2) were also analyzed, but the responses on the LC MS QQQ were below the limit of detection (< 0.1 mg/L).

4.4 Nitrogen mass balance of the solvent

Gå til overskriften

A nitrogen balance of the solvent was done after 1850 hours of operation, just before reclaiming, see table 6.

Total Nitrogen in lean amine was measured to be 8.3 wt%, which give a total of 222964 mole N. The sum of the different degradation products found gives a total of 205567 moles. This gives 7.8 mole% of nitrogen that is not found by analysis, these components are hereafter called unidentified components. Some of the unidentified components are assumed to be long chain molecules. Dissolved ammonium and ammonia in the solvent were not measured; this means that they will presumably have some contribution to the amount of the unidentified components. Table 6 shows an overview of all the components that were analyzed, and the contributions of each component to the total amount of nitrogen.

4.5 Solvent loss

Gå til overskriften

Excluding plant leakage, MEA loss can occur in the following ways:

• MEA emitted via Absorber (after water wash section)
• MEA emitted via stripper upper product after the condenser
• MEA degraded product via NH3 formation, which is detected after the wash section and from the CO₂ product stream
• Liquid sampling, which was taken for analysis
• Unexpected loss due to leakage through joints and pumps
• Wash water (absorber, stripper)
• Reclaimer waste

Lab samples and reclaimer waste are a part of the total inventory calculation. MEA was charged into the amine makeup tank from trucks. From the amine make up tank, MEA can either be charged into the storage tank or directly to the process loop. A total of 30088 Kg of pure MEA was filled into the makeup tank, while a total of 23208 Kg of MEA was discharged from the plant after the end of campaign. This gave a total loss of 7622 Kg pure MEA. Total CO₂ capture in the campaign was 4941 ton, and this give a loss of 1.5 kg MEA/ton CO₂ captured.

A nitrogen mass balance of the total solvent system was also done. The accumulated NH3 emission from the absorber and stripper corresponds to approximately 67% of the total MEA loss, while the nitrogen  detected  identified degradation compounds (D-mix) constitutes approximately 16% of the MEA loss. Table 7 gives a short summary of the degraded product produced per mole amine lost. These results are similar to the results reported by IEAGHG [11]. Total Nitrogen analysis was performed, and it is reasonable to assume that long-chain degradation compounds constitute some amount of the unidentified loss.

The nitrogen mass balance for the entire campaign gives a loss of MEA that corresponds to 1.6 kg MEA/ton CO₂ captured. There is a small gap between the two different methods of calculation, and average value  is used. From  this MEA 2 campaign it is concluded that the loss of solvent was 1.6 ± 0.1 kg/ton CO₂ captured.

5.  Emissions of amines and amine based degradation products

Gå til overskriften

5.1 Analysis of emission from depleted flue gas

Gå til overskriften

Emission to Air from TCM DA amine plant has two sources, the amine absorber and the CO₂-stack. At TCM the CO₂ product stream is sent into the atmosphere, which will not be the case for a full-scale CO₂ capture plant. As the contribution from this stream is small considered to the absorber (1-3%), data from this stream is not given in this paper.

TCM DA applies different measurement techniques to monitor and quantify the amounts and concentrations of emitted compounds. A description of the TCM DA overall system for emission control and monitoring is given elsewhere [1]. The emission was followed up by FTIR, PTR-TOF-MS, PTR-QMS, isokinetic sampling and by 3rd party (FORCE Technology) [9].

MEA emissions are highly related to aerosols in the flue gas [6]. Even at low mass concentrations of aerosols, increased MEA emissions have been measured and reported. In September 2015 TCM investigated the relation between flue gas particle content, mainly related to sulphuric acid mist particles and dust, and corresponding MEA amine emissions. This “MIST test” was based on aerosol number concentration and size distribution, to evaluate the maximum aerosol number concentration acceptable for operation with a solvent based on MEA [6]. TCM received a temporary emission permit given for this campaign from the Norwegian environmental agency (NEA). The temporary permit gave allowance to increase MEA emission from 6 ppmV to 500 ppmV for maximum 4 days of testing.

The Mist test was a planned temporary campaign lasting for only two  weeks. The rest of the MEA campaign  were performed without issues regarding mist, impurities and aerosols, as flue gas from the combined heat and  power plant does not contain particles and impurities. Detailed information about all the test activities and performance from the MEA campaign can be found in Gjernes et al [12].

Figures 10 – 13 provide the daily average ammonia, MEA, acetaldehyde and formaldehyde emissions and operational hours throughout the campaign. Some daily averages of ammonia emissions indicate higher emissions than allowed in the TCM DA emission permit. Any such emission peaks were communicated to the NEA. These incidents were administratively handled by NEA, and the campaign continued as planned. These higher levels were due to amine plant start-up activities, where molecular ammonia (or other amine compounds), i.e. ammonia (or other amine compounds) are unreacted with CO₂, are by convection transferred by the flue gas through the absorber and eventually emitted to atmosphere. The emissions follow a  Gaussian like trend, i.e. an emission peak  is observed  until the emission levels settles at a lower steady state level. Test activities with increased CO₂ content in the  flue  gas combined with high temperatures in the solvent, water washes and flue gas, gave high ammonia emissions.

A start-up procedure conducted in the following order will reduce such start-up emission peaks;

• MEA solvent circulation starts at ambient temperatures
• Flue gas is introduced and the CO₂ loading process of the entire MEA solvent inventory occurs  at ambient temperatures, until CO₂ in the MEA solvent are in equilibrium with CO₂  in the  incoming flue gas

• Heat is applied to the stripper section in order start the continuous CO₂ removal process

By following the aforementioned start-up order, the amount of emitted molecular ammonia  and  amine compounds are decreased as the presence of these compounds in the gas phase inside the absorber is reduced, and hence less gaseous ammonia and amine compounds are transferred through the absorber by convection. 19th of May 2016, TCM received a new permanent emission permit from NEA allowing 100 ppmV ammonia emissions as  a  daily average.

For achieving the TCM objectives, it is important that variables are measured with high degree of accuracy. This will ensure that high quality data are obtained and thus a high quality of test results can be provided. This is significant not only for technology test reports but also for emissions reporting to the Norwegian Environmental Agency (NEA). A failure to estimate the inaccuracies of measurements will complicate the test planning, reporting   to NEA and operation and maintenance of the test facility. Apart from accuracies of different variables, repeatability or precision of measurements for each of the variables on different streams also needs to be estimated. One quality assurance (QA) test is to compare different monitoring techniques. This was done during the MIST  test,  and depleted flue gas out of the absorber was measured by four different independent measurements; two FTIR’s, PTR- TOF-MS and PTR-QMS. All the different measurement techniques showed very similar results. The result of this  QA is shown in figure 14 and 15. TCM is a demo-plant where many types of online emission measurement equipment are tested, providing useful information for commercial projects.

Seven emission isokinetic sampling campaigns have been carried out in order to follow up on emissions form the absorber. Results from these measurements can be found in table 8. Overall the results are similar to the results reported by Morken et al [1].

TCM has shown earlier that the absorber wash water sections are found to effectively reduce  possible  atmospheric emissions from amine based solvent system [1]. Atmospheric emissions of monoethanolamine (MEA) were very low throughout the entire campaign, and determined to be in the parts per billion (ppb) ranges.

Atmospheric emissions of MEA amine based degradation products such as nitrosamines and nitramines were below detectable levels. Atmospheric emissions of alkyl amines in the low ppb range. Results from isokinetic measurements can be seen in table 8. These results confirm the emission results  from earlier MEA campaign at  TCM [1].

Conclusions

Gå til overskriften

During the MEA 2015 campaign at TCM the degradation products being formed in the solvent and released to   the atmosphere were closely monitored. Based on an overall nitrogen mass balance it was concluded that less than 8% of total nitrogen introduced into the plant was not identified. The solvent loss calculated as pure MEA was 1.6 ±

0.1 kg/ton CO₂ captured. The major contributors to the loss were ammonia emission (67% of loss) and identified degradation products in the solvent (16% of loss). Emissions to air from the absorber stack were monitored by five different independent on-line measurement instruments and by regular manual sampling. The four on-line methods provided very similar results. The manual sampling results confirmed results from earlier MEA campaign at TCM. The MEA and alkyl amines emissions are in the parts per billion ranges and nitrosamines and nitramines were below detectable levels.

Acknowledgements

Gå til overskriften

The authors gratefully acknowledge the staff of TCM DA, Gassnova, Statoil, Shell and Sasol for their  contribution and work at the TCM DA facility. The authors also gratefully acknowledge The TCM DA operation team, lab team at TCM DA and Statoil CP laboratory, Technology Committee, University of Oslo and Sintef for  their contribution and work at the TCM DA facilities.

References

Gå til overskriften

1. Morken AK, Nenseter B, Pedersen S, Chhaganlal M, Feste J, Tyborgnes RB, Ullestad Ø, Ulvatn H, Zhu L, Mikoviny T, Wisthaler A, Cents T, Bade OM, Knudsen J, De Koeijer G, Falk-Pedersen O, Hamborg ES. Emission results of amine plant operations from MEA testing at the CO₂ Technology Centre Mongstad, Energy Procedia 63, 6023-6038.
2. Thimsen D, Maxson A, Smith V, Cents T, Falk-Pedersen O, Gorset O, Hamborg E S. Results from MEA testing at the CO2 Technology Centre Mongstad. Part I: Post-Combustion CO₂ Capture Testing Methodology. Energy Procedia 2014.
3. Hamborg ES, Smith V, Cents T, Brigman N, Falk-Pedersen O, Chhaganlal M, Feste J K, Ullestad Ø, Ulvatn H, Gorset O, Askestad I, Gram L K, Fostås B F, Shah M I, Maxson A, Thimsen D. Results from MEA testing at the CO₂ Technology Centre Mongstad. Part II: Verification of baseline results. Energy Procedia 2014
4. Da Silva EF, Hoff KA, Booth A; Emissions from CO₂ capture plants; an overview, Energy Procedia 37 (2013) 784-790
5. Lepaumier H, Picq D, Carrette P-L, New Amines for CO2 Capture. I. Mechanisms of Amine Degradation in the Presence of CO₂, Ind. Eng. Chem. Res., 2009, 48 (20), pp 9061–9067
6. Lombardo G, Fostås BF, Shah MI, Morken AK, Hvidsten OA, Mertens J, Hamborg ES. Results from aerosol measurement in amine plant treating gas turbine and residue catalytic cracker flue gases at the CO₂ Technology Centre Mongstad. Energy Procedia (GHGT-13), Forthcoming 2017.
7. Vevelstad SJ, Eide-Haugmo I, Da Silva EF, Svendsen HF, Degradation of MEA; a theoretical study, Energy Procedia 4 (2011) 1608–1615
8. Fostås B, Gangstad A, Nenseter B, Pedersen S, Sjøvoll M, Sørensen AL, Effects of NO_x in the flue gas degradation of MEA, Energy Procedia 4 (2011) 1566–1573
9. Faramarzi L, Thimsen D, Hume S, Maxson A, Watson G, Pedersen S, Gjernes E, Fostås BF, Lombardo G, Cents T, Morken AK, Shah MI, de Cazenove T, Hamborg ES. Results from MEA testing at the CO₂ Technology Centre Mongstad: Verification of baseline results in 2015. Energy Procedia (GHGT-13), Forthcoming 2017.
10. Mikoviny T, Nielsen C.J, Tana W, Wisthaler A , Zhu L, Morken A.K, Nilsen T.N, Ambient Measurements of Amines by PTR-QiTOF: Instrument Performance Assessment and Results from Field Measurements in the Vicinity of TCM, Mongstad, Forthcoming 2017.
11. IEAGHG “Evaluation of reclaimer sludge disposal from post-combustion CO₂ capture”, 2015/02, March 2014
12. Gjernes E, Pedersen S, Cents T, Watson G, Fostås BF, Shah MI, Lombardo G, Desvignes C, Flø NE, Morken AK, de Cazenove T, Faramarzi L, Hamborg ES. Results from 30 wt% MEA performance testing at the CO₂ Technology Centre Mongstad. Energy Procedia (GHGT-13), Forthcoming 2017.
13. Flø NE, Faramarzi L, de Cazenove T, Hvidtsten OA, Morken AK, Hamborg ES, Vernstad K, Watson G, Pedersen S, Cents T, Fostås BF, Shah MI, Lombardo G, Gjernes E. Results from MEA degradation and reclaiming processes at the CO₂ Technology Centre Mongstad. Energy Procedia (GHGT-13), Forthcoming 2017.
14. Davis J, Rochelle G, Thermal degradation of monoethanolamine at stripper conditions, Energy Procedia 1 (2009) 327-333.

Results from MEA Degradation and Reclaiming Processes at the CO₂ Technology Centre Mongstad (2017)

Nina Enaasen Flø^a, Leila Faramarzi^a,b, Thomas de Cazenove^a, Odd Arne Hvidsten^a,b, Anne Kolstad Morken^a,b, Espen Steinseth Hamborg^a,b,*, Kai Vernstad^c, Guillaume Watson^d, Steinar Pedersen^b, Toine Cents^e, Berit F. Fostås^b, Muhammad Ismail Shah^a,f, Gerard Lombardo^a,f, Erik Gjernes^f

^aCO₂ Technology Centre Mongstad, 5954 Mongstad, Norway ^bStatoil ASA, PO Box 8500, 4035 Stavanger, Norway ^cSINTEF, Strindveien 4, 7034 Trondheim, Norway ^dShell Global Solutions International B.V., PO Box 663, 2501CR The Hague, The Netherlands ^eSasol Technology, PO Box 5486, Johannesburg 2000, South Africa ^fGassnova SF, Dokkvegen 10, 3920 Porsgrunn, Norway *Corresponding author

Abstract

Gå til overskriften

In 2015, the CO₂ Technology Center Mongstad (TCM DA), operated a test campaign using aqueous monoethanolamine (MEA) solvent at 30 wt%. The main objective was to demonstrate and document the performance of the TCM DA Amine Plant located in Mongstad, Norway. As part of the test campaign, thermal reclaiming was performed in order to eliminate accumulated degradation products and improve the solvent performance. This paper presents results and discussions concerning formation and monitoring of amine degradation products along with experiences related to the thermal reclaiming process and its operational procedure. Evaluations of the efficiency of thermal reclaiming and the solvent improvement after reclaiming are also presented.

1. Introduction

Gå til overskriften

The CO₂ Technology Centre Mongstad (TCM DA) is located next to the Statoil refinery in Mongstad, Norway. TCM DA is a joint venture set up by Gassnova representing the Norwegian state, Statoil, Shell, and Sasol. The facility run by TCM DA entered the operational phase in August 2012 and it is one of the largest post-combustion CO₂ capture test centres in the world. A unique aspect of the facility is that either a flue gas slipstream from a natural gas turbine based combined heat and power (CHP) plant or an equivalent volumetric flow from a residual fluidized catalytic cracker (RFCC) unit can be used for CO₂ capture. The CHP flue gas contains about 3.5% CO₂ and the RFCC flue gas contains about 13-14% CO₂. One of the main test plants at TCM DA is a highly flexible and well-instrumented amine plant. The amine plant was designed and constructed by Aker Solutions and Kværner to accommodate a variety of technologies, with capabilities of treating flue gas streams of up to 60,000 standard cubic meters per hour. The plant is being offered to vendors of solvent based CO₂ capture technologies to, among others, test; (1) the performance of their solvent technology, and (2) technologies aimed to reduce the atmospheric emissions and environmental impact of amines and amine based degradation products from such solvent based CO₂ capture processes. The objective of TCM DA is to test, verify, and demonstrate CO₂ capture technologies suitable for deployment at full-scale. Up to now the vendors Aker Solutions, Alstom, Shell Cansolv Technologies Inc. and Carbon Clean Solutions Ltd. have successfully used the TCM DA facilities to verify their CO₂ capture technologies. From July to October 2015 TCM DA, in collaboration with partners, operated a test campaign using the non- proprietary aqueous monoethanolamine (MEA) solvent at 30 wt%. After testing a variety of process conditions for a total of 1843 hours, clear evidence of solvent degradation was observed. The test campaign proceeded with thermal reclaiming of the solvent in order to eliminate accumulated degradation products and demonstrate improvement of solvent performance. This work presents results concerning MEA degradation monitoring and reclaiming operation at TCM DA. Various design and operational factors that affect degradation rates are discussed, the efficiency of thermal reclaiming is estimated and experiences related to the reclaiming process and its operational procedure is shared.

1.1 Solvent degradation mechanisms

Gå til overskriften

Amine solvents degrade due to exposure to heat (thermal degradation), presence of oxygen (oxidative degradation) and reactions of the amine with flue gas contaminants such as SO_x, NO_x, halogenated compounds, hydrocarbons and other impurities. Thermal degradation occurs mainly in the stripper section and is strongly dependent on the stripper operating temperature. The main thermal degradation products in MEA are Oxazolidin-2- one (OZD), MEA urea, HEIA, HEEDA [1]. The rate of formation of these products depend on the operating temperature (faster kinetics), CO₂ loading (more carbamate present) and MEA concentration. Oxidative degradation is mainly an issue for post-combustion CO₂ capture where the solvent is exposed to oxygen present in the flue gas. This occurs mainly in the absorber, where the level of oxygen is significant. Amine oxidation is also shown to be catalyzed by transition metal ions and will typically results in formation of ammonia and different organic acids [2]. In a second step, the organic acids will form heat stable salts (HSS) which are difficult to regenerate under normal regeneration conditions (atmospheric pressure and temperature around 120°C) [1]. These salts will therefore remain and accumulate in the circulated solvent. Amine degradation may also be induced by flue gas contaminants such as sulfur, polysulfide and CO. This issue has become especially evident for oil refinery flue gases such as gas originating from RFCC units [2]. Nonvolatile contaminants causing amine degradation can also arise from other sources such as make-up water, anti-foam agents, lubricants and corrosion inhibitors [2].

Several degradation processes often occur simultaneously to produce a wide range of degradation products. Accumulation of amine degradation products affects the solvent properties. They are known to increase the solvent viscosity and surface tension which again affects heat transfer coefficients, diffusion coefficients, and mass transfer rates [3]. This will again lead to loss of solvent capacity and increased energy numbers. Further, degradation products might lead to corrosion, fouling and foaming [2], which again increases operational and maintenance costs and might cause long-term technical integrity issues. Dissolved metal elements originating from corrosion are also as mentioned above expected to contribute as catalysts for oxidative degradation [1].

1.2 Solvent refreshing options

Gå til overskriften

In order to reduce the impact of unwanted impurities and minimize the operational and maintenance issues listed above, a number of techniques have been suggested. Wang et al. (2015) have published an extensive review of amine reclaiming technologies and other techniques to handle this issue, including purging (bleed and feed), neutralization, ion exchange, adsorption, electrodialysis, and different thermal reclamation techniques [4]. Dumée et al. (2012) also presents a thorough comparison of the most promising techniques [1]. A summary is provided below.

• Bleed and feed

Bleed and feed is a simple operational procedure where a portion of the degraded solvent is continuously or periodically purged and replaced with fresh solvent. However, amine replacement and disposal might make this technique rather costly, particularly for specialized and expensive solvents. Further, a certain level of degradation products needs to build up before effecting bleed and feed in order to minimize replacement and disposal costs.

• Neutralization

Neutralization converts amine HSS to sodium or potassium HSS by addition of NaOH or KOH, according to the following reaction using NaOH as an example:

NaOH + [AmineH⁺ RCOO^– ] ➔ Amine + H 0 +  [Na⁺ RCOO^– ]                                                                            (1)

Neutralization maintains the amine capacity; however, there is no reduction in salt content of the solvent. The amine becomes more and more contaminated by salts that contribute to higher solvent density and viscosity, reduced surface tension, and possibly foaming and fouling. Eventually, the solvent needs to be discarded.

• Ion exchange

Ion exchange is a technology where the amine HSS ion is replaced with a friendlier ion. For example, an anion exchange removes HSS anions, replacing them with hydroxide ions, which frees the amine and let it return together with water to the process. The HSS anions are later removed from the resin by regeneration with NaOH. The practice of removing HSS from amine systems by ion exchange has presented many technical and operational challenges, and several researchers report doubt in the practical efficiency for amine applications. High consumption of chemical and water for resin regeneration together with generation of large amounts of waste are mentioned as other disadvantages. Further, ion exchangers are not capable of removing uncharged contaminants, i.e. degradation products originating from thermal degradation. It is still regarded a relatively economical method, especially for low levels of contaminants. However, if poorly designed and/or operated it can cause significant solvent losses and sodium slippage into the main amine process.

• Adsorption

Adsorption on activated carbon is a widely used method to remove high-boiling or surface active organic compounds [5]. However, activated carbon it is not able to remove significant amount of degradation products [2].

• Electrodialysis

Electrodialysis has also been suggested as a method to purify amine solutions. It uses a stack of alternating anionic and cationic ion-exchange membranes to selectively remove charged contaminants from the solvent [1]. The main disadvantage also for this method is inability to remove uncharged amine degradation products originating from thermal degradation and hydrocarbons [2].

• Thermal reclaiming

Thermal reclaiming is usually conducted on a small slipstream extracted from the stripper reboiler on a semi- continuous basis [2, 5]. The amine solution is vaporized in the reclaimer vessel and returned as vapour to the main process, while the less volatile degradation products and other contaminants accumulate in the reclaimer vessel bottoms. Stoichiometric addition of NaOH during thermal reclaiming allows recovery of the amine from the amine heat stable salts by conversion to sodium salts, according to Reaction (1). Thermal reclaiming has long been a recognized reclamation method for MEA. Compared with secondary and tertiary amines, MEA has a low normal boiling point allowing it to vaporize without degrading significantly. For other amines with higher boiling points vacuum might be required in order to prevent thermal degradation during reclaiming. The fact that MEA reclaiming units can be operated at the stripper pressure eliminates the need for a separate condenser for the reclaiming system and reduces the overall energy demand. In this approach the reclaimer product vapour which contains MEA is directly sent to the stripper [1, 5]. A major disadvantage of thermal reclaiming is the formation of an aqueous slurry waste whose disposal poses a challenge for the CO₂ capture process. The amount of waste depends on various parameters such as the flow rate of the slip stream fed to the reclaimer, the amount of basic solution used to liberate MEA from heat stable salts, solvent conditions and overall operating conditions of the plant. According to the International Energy Agency (IEA) about 3.2 kg of amine reclaimer waste is generated per ton of CO₂ captured from coal fired flue gases using MEA [3]. However, depending on gas pre-treatment, combustion fuel, the type of amine used and the capture process itself, the reclaimer waste generation can vary in the range of 0.1-14.9 kg waste/ ton CO₂ [3]. Collecting representative samples of reclaimer waste is complicated and so far there is limited information in the public domain that fully represents amine reclaiming waste for CO₂ capture processes. Using the Flour Econamine FG^SM system as a reference, Nurrokhmah et al. (2013) have investigated methods to characterize MEA reclaiming waste along with possible waste treatment and reuse options. Thermal reclaiming is also mentioned to be energy extensive. However, alternative reclaiming technologies such as ion exchange and electrodialysis are not able to remove metals and non-ionic products and the potential efficiency of HSS removal is not as high as for thermal reclaiming [1].

2. The TCM DA amine plant

Gå til overskriften

An illustration of the TCM DA amine test unit is presented in Figure 1, and a short description is given in the following. Flue gas is cooled down and saturated with water in a direct contact cooler (DCC) before it enters the absorber. At TCM DA there are two possible sources of flue gas, i.e. exhaust gas originating from the natural gas fired combined heat and power plant and industry gas originating from the residue fluidized catalytic cracker. Both flue gas sources have their individual flue gas fans and DCCs as illustrated in Figure 1. Product CO₂ can also be recirculated back to the CHP gas absorber inlet to adjust the CO₂ content. For RFCC gas there is an option of mixing in air to adjust the CO₂ content. The conditioned flue gas is contacted counter-currently with the amine solvent in the absorber tower. CO₂ from the flue gas is absorbed yielding a solvent rich in CO₂ and a depleted flue gas with low CO₂ content. The depleted flue gas is released to the atmosphere after passing two sections of water wash. Typical absorber conditions are close to ambient pressure and temperatures of 40 – 80 °C, depending on the CO₂ content in the incoming flue gas. The CO₂ rich solvent is pre-heated in the lean/rich cross heat exchanger before it enters the stripper column where the chemical reactions are reversed to desorb CO₂ and regenerate the solvent. Heat is provided through steam in a thermosiphon reboiler to maintain regeneration conditions, i.e. 100 – 120 °C and pressure around 1 barg. The product CO₂ is released to the atmosphere, while the regenerated lean solvent is pumped back to the absorber via the lean/rich cross heat exchanger and the lean cooler.

The TCM DA amine test unit is also equipped with a thermal reclaimer which treats a slip stream of the lean solvent coming from the stripper. The thermal reclaimer uses additional heat provided by steam to separate the useful solvent from the degradation products which are accumulated in the solvent over time. The reclaimer vapour contains useful solvent which is recycled back to the main process, while the waste remains in the reclaimer and is periodically discharged. Water and NaOH can be added to the reclaimer unit on demand. The operating pressure corresponds to the stripper pressure.

The reclaiming system consists of a flash vessel and a steam heater, as illustrated in Figure 1. The dimensions of the reclaimer vessel is 2.3m x 3.0 m (IDxTT) and it is designed for an operating volume of 1 – 7 m³, which corresponds to approximately 2 – 14 % of the total solvent inventory of the plant.

2.1 MEA campaign overview

Gå til overskriften

The MEA test campaign was conducted from 06/07/2015 to 17/10 2015. During the total 1960 hours of operation a wide range of operational process conditions were executed and a total of 4941 tons of CO₂ was captured. The variation of gas and solvent flow rates and stripper bottom temperatures are presented in Figure 2, while further details on typical operating process conditions are presented in Table 1 of Gjernes et al. (2017) [7]. The test campaign was operating on 30 ± 2 wt% MEA and the ranges of the lean and rich CO₂ loadings during the campaign was 0.19 – 0.29 and 0.46 – 0.53 mol CO₂/mol MEA, respectively. The majority of the campaign was operated with CHP flue gas; however, for a shorter period of 9 days from 16/09/2015 to 24/09/2015 it was operated on a mixture of CHP and RFCC gas, as indicated in Figure 2. Thermal reclaiming was performed towards the end of the campaign, after 1838 hours of operation. Reclaiming was performed for 92 hours, and the plant was run for an additional 28 hours after the reclaiming period before the campaign was concluded 17/10/2015.

3. Solvent degradation during the test campaign

Gå til overskriften

3.1 Process conditions that influenced solvent degradation

Gå til overskriften

The MEA test campaign was conducted by executing a wide range of process conditions with frequent operational set-point changes. Such a shifting operating environment might accelerate solvent degradation. The average stripper bottom temperature was 120 °C, with a maximum of 122.5 °C. Superheated MP steam in the temperature range of 130 – 150 °C was used as heat source in the stripper reboiler. The reboiler skin temperature for which the solvent is exposed to, can therefore be assumed to be around 130°C. The solvent will undergo thermal degradation when exposed to temperatures at this level.

The majority of the campaign was operated with CHP flue gas. However, as part of specific mist testing where the aim was to induce formation of aerosols and study its effect on emissions, the plant was operated on a mixture of CHP and RFCC gas [8]. The mist testing where more specifically conducted by;

1. Increasing the concentration of CO₂ in the feed flue gas up to 12 vol% by recycling parts of the captured CO₂ to the absorber flue gas inlet.
2. Mixing portions of the RFCC flue gas with the CHP flue gas.

Up to 10 % mixing of RFCC gas in CHP gas was tested. Typical CHP and RFCC gas concentrations downstream the DCCs are presented in Table 1. As seen in the table, the CHP flue gas contains significant amounts of oxygen which causes oxidative degradation. Exposure to higher concentrations of CO₂ and RFCC gas impurities during the mist testing accelerated the rate of solvent degradation. Further, metal particulate material present in the RFCC gas might have contributed as catalysts for oxidative degradation.

3.2 The impact of process design on solvent degradation

Gå til overskriften

As mentioned above, the main factors causing solvent degradation was elevated operating temperature in the stripper section and exposure to oxygen and contaminants in the flue gas. The effect of thermal and oxidative degradation will not only depend on these factors themselves, but also on the solvent residence times in the sections of the plant where these factors are significant, i.e. the part of the plant where the solvent is exposed to higher temperatures and oxygen and gas contaminants.

The hot solvent inventory (desorber packing, desorber sump, reboiler, hot part of the lean/rich cross heat exchanger and the hot lean and rich solvent piping) calculated for CHP baseline operating conditions are presented in Table 2. For details about the CHP baseline operating conditions it is referred to Faramarzi et al. (2017) [9]. The total of 13.4 m³ hot solvent inventory is quite significant and corresponds to about 35% of the total solvent inventory. The corresponding solvent residence time is about 20 minutes for CHP baseline operating conditions. The main contributor to the hot solvent inventory is clearly the rather long hot lean solvent pipe, which contributes to 60% of the total hot solvent inventory. The reboiler itself has a rather low solvent residence time; however, the beforementioned reboiler skin temperature of about 130 °C might also contribute to significant thermal degradation as degradation increases exponentially with the temperature.

The solvent inventory exposed to oxygen and the corresponding oxygen exposure time is also presented in Table 2. It is expected that the largest effect of oxygen exposure is seen in the absorber packing, where the actual inventory and exposure time is estimated to about 8 m³ and 12 minutes, respectively, considering CHP baseline operating conditions. This abovementioned exposure time is also relevant for flue gas contaminants when operating on CHP/RFCC gas mixture.

In order to minimize solvent degradation it is clearly of interest to perform plant design such that the exposure times to oxygen and elevated temperatures are limited. For scale-up purposes it is therefore of specific importance to minimize solvent hold-up in hot parts of the plant.

3.3 Monitoring of solvent degradation

Gå til overskriften

Solvent degradation was observed and monitored by a number of parameters during the test campaign. Lean and
rich solvent samples were frequently withdrawn for solvent analysis. The analytical methods are described by
Morken et al (2017) [10]. Firstly the physical properties of the solvent changed during the campaign as shown by the increase of solvent viscosity in Figure 3. The viscosity was measured in TCM DA lab and reported at two different temperatures (30°C and 60°C) and a clear increase of about 50% is observed from the test campaign start until reclaiming started on 12/10/2015.

A clear observation of solvent degradation was also the change of solvent color during the test campaign. The fresh 30 wt% MEA solvent started out as a clear liquid, which changed color quite fast after contact with flue gas. The solvent became gradually darker during the campaign, until it reached the dark brown color illustrated by the third sample glass from 11/10/2015 in Figure 4.

Further, the level of volatile degradation products in the gas phase increased significantly during the period of Mist testing. Morken et al (2017) presents detailed results regarding ammonia emissions, which is associated with presence of ammonia in the solvent originating from solvent degradation [10]. Emission of ammonia is also highly dependent on operating conditions; however the observed build-up of ammonia in the solvent is regarded as a clear sign of solvent degradation.

Heat stable salts started building up in the solvent as shown in Table 3 before it reached a maximum of 0.203 mol/kg just before reclaiming started on 12/10/15. More detailed results concerning HSS analysis are presented by Morken et al (2017) [10]. The concentration of main degradation products was also monitored continuously and shows a significant increase as the test campaign progressed. It is referred to Morken et al (2017) for details [10].

Additional parameters which are important to monitor during operation of the amine plant are solvent foaming tendency and metal ion concentration. The latter gives indications of plant corrosion and was also monitored during the test campaign. The results are presented by Hjelmaas et al. (2017) [11].

4. Reclaiming procedure and operational experience

Gå til overskriften

The reclaimer was operated in a semi-continuous operation mode, meaning that solvent was continuously fed to the reclaimer vessel, while the reclaimer waste was allowed to accumulate and was only disposed at the end of the test campaign. The process was operated continuously for 3 days with exception of one unexpected plant stoppage for about 3 hours on the 13/10/2015.

The reclaimer vessel was initially filled with water. Water circulation and steam heating was started before the solvent feed to the reclaimer vessel. The rather large volume of initial water evaporated during the reclaiming operation and resulted in dilution of the solvent as shown in Figure 5.

The reclaimer liquid was circulated in the reclaiming system loop through the steam heat exchanger at a circulation rate of approximately 165 m³/h. No boiling occurs in the steam heater, but the liquid flashes when it enters the evaporator vessel. The evaporating level was controlled by adjusting the steam rate supply. As the liquid became more concentrated, its boiling temperature increased and the rate of evaporation was reduced. The percentage of degradation products in the reclaimer, and the resulting temperature were slowly increasing. Upon reaching high temperature, high viscosities and high amounts of precipitates, the reclaimer feed was stopped.

4.1 Solvent and water feed rate

Gå til overskriften

The reclamation unit was fed with a continuous slip stream of the lean amine solvent from downstream the stripper. The reclaimer was also fed simultaneously with water in order to control the boiling temperature of the reclaimer fluid below 160 °C. Figure 6 presents the solvent and water flow rates along with the reclaimer liquid temperature.

The solvent slip stream corresponded to 4 – 5 % of the lean solvent circulation and was up to a maximum of about 3 000 kg/h as illustrated in Figure 6. A total accumulated amount of 46 000 kg solvent was fed to the reclaimer during the whole period of 3 days. This corresponds to about 110 % of the total solvent inventory.

4.2 Steam consumption

Gå til overskriften

The reclaimer heat duty variations were according to the changing amount of the lean solvent slip stream directed to the reclaimer vessel. As shown in Figure 7, in order to vaporize MEA in the reclaimer a significant amount of heat was required. At times, the amount of heat used for reclaiming was almost equal to the heat used to regenerate the solvent in the stripper. As reclamation of MEA is energy intensive, it is important to optimize the amount of lean amine slip stream sent to the reclamation unit. However, as shown in Figure 6 the flow of slip stream varied due to the fluctuations in the process conditions and it was not possible to achieve a constant flow during the reclaiming procedure.

The reboiler heat duty increased significantly when the reclaimer was brought on stream and then plateaued at about 2 500 kW. This was due to the large amount of water that was initially added to reclaimer unit, which evaporated from the reclaiming vessel and caused dilution of the solvent. The concentration of MEA was consequently reduced to about 21 wt% as shown in Figure 5. Thus the amount of water to be boiled off in the stripper was much larger, causing higher energy numbers.

The reclaimer liquid circulation and steam heating continued for 2 days after the solvent feed was stopped in order to evaporate as much as possible of the useful MEA solvent and concentrate the waste.

4.3 Dosage of NaOH

Gå til overskriften

Aqueous solution of 50 wt% NaOH was added to the reclaimer vessel via the reclaimer liquid circulation loop in order to stabilize anions of amine heat stable salts by converting them to sodium salts and liberating the amine according to Reaction (1). The recovered amine and water vapor was returned to the stripper sump.

A dosage rate of 3 L NaOH/m³ solvent was applied during reclaiming based on previous experience at TCM DA. In total 227 liters 50% NaOH was added, which corresponds to 4299 mol Na⁺.

According to Reaction (1), the stoichiometric ratio of NaOH to HSS should ideally be 1:1. This is a very rough estimate since the actual ratio depends on the electrical charge of the anions. The concentration of HSS components was 0.203 mol MEA-eq/kg solvent at the point of reclaiming start 12.10.15 (see Table 3). With a total solvent inventory of 40 800 kg in the plant at the time, this corresponds to 8282 mol HSS. A stoichiometric check shows excess HSS compared to NaOH, which might cause additional MEA loss in the reclaimer waste.

4.4 Reclaimer waste

Gå til overskriften

After the reclaiming operation was concluded the majority of the concentrated waste was drawn off to the flushing line and passed through the sea water cooler to the IBC (Intermediate Bulk Container) drainage system. The reclaimer fluid was quite concentrated and viscous at the time, thus some water was added in order to dilute the waste and enable unloading of the vessel. The total concentrated waste was collected in IBCs and added up to a total of about 6 m³. This corresponds to about 1.3 kg reclaimer waste/ton CO₂ captured during the overall campaign, which is well below reported numbers in the literature. Further, the reclaiming process was initiated when HSS concentration reached 0.203 mol/kg, as beforementioned. The actual necessity of reclaiming at this level of HSS must be considered based on the actual solvent condition and potential plant corrosion issues, i.e. at this moment the reclaiming campaign was not necessary but rather conducted for demonstration purposes in the test campaign. The waste/ton CO₂ capture would thus be even lower in an actual necessary reclaimer case. The reclaimer vessel and piping was afterwards flushed with water.

5. Efficiency of thermal reclaiming

Gå til overskriften

In order to investigate the reclaiming efficiency and demonstrate how the solvent quality is recovered and maintained by the reclaiming process, samples were frequently taken from the lean amine solvent, the reclaimer liquid and reclaimer vapor. The samples were analyzed for MEA, degradation products, HSS and metals, and the results are summarized in Table 4.

The concentration of degradation products in lean amine was analyzed throughout the test campaign and the results are presented by Morken et al (2017) [10]. Figure 8 below shows the concentration of degradation products in the lean amine solvent during the reclaiming operation. It is seen that the degradation products is efficiently cleaned from the lean amine and about 95% percent of the degradation products was removed. A small increase in concentration from day three indicates that degradation is significant during reclaiming, likely due to thermal degradation due to operation at elevated temperatures inside the reclaimer vessel.

A very similar trend is seen for the concentration of metal elements iron (Fe), Nickel (Ni) and Chromium (Cr) in
Figure 9 below. The concentration is reduced by more than 95% after reclaiming.

The trends for heat stable salts in the lean solvent, reclaimer liquid and the reclaimer vapor are shown in Figure 10. Again, the concentration of HSS in lean amine is rapidly reduced to less than 5% of the start concentration, as shown by the blue columns in the graph. The accumulation of HSS in the reclaimer liquid is also clearly seen by the red columns. HSS could not be detected in the reclaimer vapor return to stripper, as expected. Figure 11 presents the concentration of MEA, NaOH and HSS in the reclaimer liquid during the reclaiming process. Most of the MEA is evaporated during the period as seen in the figure. HSS and Na⁺ is accumulated, however MEA seems to be in excess, also at the end of the reclaiming.

The color of the solvent changed back to a lighter color after reclaiming as illustrated by the fifth sample glass from 15/10/2015 in Figure 4. Based on analysis of the reclaimer waste and assessment of the total MEA inventory in the plant before and after reclaiming, it is estimated that about 500-550 kg MEA was lost to waste during reclaiming. This corresponds to 4% of the total inventory (according to Table 4) and 0.11 kg MEA/ton CO₂ captured.

6. Solvent performance after reclaiming

Gå til overskriften

After the reclaiming operation had been concluded the plant was operated for another 28 hours at a flue gas flow rate of 47,000 Sm³/h. Two test cases were conducted during this period, and these are used for comparison to other similar tests conducted previously in the campaign with a fresh solvent. The two test cases after reclaiming is designated “T4” and “T5”, while the optimum energy case with the use of anti-foam (case 2B6) from previously in the campaign is used for comparison. The total operating hours at the point in time when case 2B6 was conducted was approximately 950 hours. The overall 2015 MEA campaign and the entire specific test series carried out to investigate the capture plant performance is described by Gjernes et al. (2017) [7].

Figure 12 summarizes the operation before and after reclaiming. T4 and T5 were operated with 24 and 18 m absorber packing height, respectively. During T4 the amine plant was a bit unstable while there were stable conditions during T5. Case 2B6 was operated with 24 meters of packing height. The plant performance after reclaiming was comparable to the optimum performance achieved earlier in the campaign and there were no significant indications of reduced solvent quality.

7. Discussion and future work

Gå til overskriften

Monitoring the amine concentration and CO₂ loading is very important for optimal operation. At TCM DA the solvent concentrations are mainly followed on a daily basis with manual samples and analysis. Further, a number of analyzers are available for real-time online monitoring, i.e. conductivity, density and pH analyzers. These online results can be correlated to enable a closer follow-up of the solvent condition.

As an effect of reclaiming start-up, the solvent in the main process was diluted by water evaporating from the reclaimer vessel. In future campaigns, extra care will be taken not to disrupt the main process during reclaiming. The deviation in solvent concentration could have been corrected at an earlier stage with an online estimate of solvent concentration.

The reclaiming environment is very harsh to the solvent due to high temperatures (up to 160 °C). The elevated temperatures represent a risk of additional thermal degradation. Care must therefore be taken in order to limit the residence time of the reclaimer solvent and thereby unnecessary degradation. Thus frequent manual solvent sampling or online analyses are required in order to monitor the progress of reclaiming and terminate the reclaiming process when the target is reached. In this test campaign it was very successfully demonstrated a 95 % cleaning efficiency when circulation a 4-5 % slip stream through the reclaimer for three days, which added up to an accumulated reclaimed volume of about 110% of the total solvent inventory.

The total HSS analysis indicates that the amount of NaOH added during reclaiming was on the stoichiometric low side to limit the MEA loss in the reclaimer waste. It is therefore reason to believe that additional MEA was lost in HSS to the waste. Thus, the total MEA loss of 4% could be reduced even further by optimizing the NaOH dosage. However, the actual effect of NaOH addition on MEA release from HSS should be investigated more in detail.

There is little information available in the literature that addresses how the build-up of impurities impacts the energy demand for regenerating MEA in the stripper i.e. reboiler heat duty. However, the density and viscosity of the solvent increased with the increasing level of contaminants as discussed in Section 3. This will cause reduction of the solvent heat transfer coefficient and consequently the heat transfer efficiency in the reboiler. The impact of accumulation of the contaminants on the specific heat capacity of amines is also very little addressed in the literature. However, it is expected that degraded MEA has higher specific heat capacity than MEA which in turn could increase the sensible heat needed to regenerate the solvent in the stripper. It is recommended to investigate these effects in the future.

As the amine plant was only operated for 28 hours after solvent reclaiming, a very limited investigation of the effect of removing the aqueous phase contaminants on the energy requirement of the stripper reboiler was performed. In future tests, sufficient time should be allowed to investigate in detail and compare the solvent performance at the beginning of the test campaign to the performance just before reclaiming and just after reclaiming.

In order to further optimize the process and reduce disposal problems both the reclaiming procedure itself and the collection and drainage of the reclaimer waste can be improved. The rapid cleaning of the lean solvent suggests running the reclaimer more frequently for shorter time periods (for example 12 hours a week) as one option to avoid degraded solvent to accumulate in lean amine. In this way the acceleration of degradation reactions could also be minimized. The draining and flushing operation can be improved by using less water or even small amounts of steam for keeping the reclaimer vessel fit for purpose. This will reduce the amounts of waste.

8. Conclusions

Gå til overskriften

A test campaign with 30 wt% MEA has been conducted for a total of 1960 hours at the CO₂ Technology Centre Mongstad. The present paper discusses main causes of solvent degradation and various parameters for monitoring degradation products. Further, the effect of process design and operating conditions on solvent degradation is discussed, and thermal reclaiming is evaluated as a technique for removal of degradation products and other contaminants in the MEA solution.

The solvent condition was closely monitored during the test campaign and several observations such as increasing solvent viscosity and darker solvent color indicated solvent degradation. Solvent exposure to oxygen and flue gas contaminants in the absorber and operation at elevated temperatures (above 100 °C) in the stripper section are highlighted as main causes for degradation. When performing scale-up to commercial CO₂ capture units it is recommended to minimize the hot solvent residence time in the plant, in order to minimize solvent degradation.

Thermal reclaiming has demonstrated an efficient clean-up of the MEA solvent. The cleaning efficiency was about 95% with respect to degradation products, HSS and metal elements. The solvent viscosity returned to normal values and the solvent color was normalized to a clearer and more yellow appearance. The quality recovery of the solvent was further assessed by an evaluation of the capture process after the reclaiming was concluded by comparing the solvent performance to results obtained at earlier stages of the test campaign and there were no significant indications of reduced solvent quality.

Acknowledgements

Gå til overskriften

The authors gratefully acknowledge the staff of TCM DA, Gassnova, Statoil, Shell and Sasol for their contribution and work at the TCM DA facility. The authors also gratefully acknowledge Gassnova, Statoil, Shell, and Sasol as the owners of TCM DA for their financial support and contributions.

References

Gå til overskriften

1. Dumée L, Scholes C, Stevens G, Kentish S. Purification of aqueous amine solvents used in post combustion CO₂ capture: A review. International journal of Greenhouse Gas Control 10 (2012): 443-455.
2. Abdi MA, Golkar MM. Improve contaminant control in amine systems. Hydrocarbon Processing 63 (2011): 102C – 102-I.
3. IEAGHG, “Evaluation of Reclaimer Sludge Disposal from Post-Combustion CO₂ Capture”, 2014/02 (March 2014).
4. Wang T, Hovland J, Jens, KJ. Amine reclaiming technologies in post-combustion carbon dioxide capture. Journal of Environmental Science 27 (2015): 276 – 289.
5. Kohl AL, Nielsen RB. Gas purification. Gulf publishing Company, Houston, Texas, USA, 5^th edition (1997).
6. Nurrokhmah L, Mezher T, Abu-Zahra MRM. The evaluation of monoethanolamine-based CO₂ post-combustion capture process waste handling approaches considering the regulations in UAE. Energy Procedia 37 (2013): 751 – 758.
7. Gjernes E, Pedersen S, Cents T, Watson G, Fostås BF, Shah MI, Lombardo G, Desvignes C, Flø NE, Morken AK, de Cazenove T, Faramarzi L, Hamborg ES. Results from 30 wt% MEA performance testing at the CO₂ Technology Centre Mongstad. Energy Procedia (GHGT-13), Forthcoming 2017.
8. Lombardo G, Fostås BF, Shah MI, Morken AK, Hvidsten OA, Mertens J, Hamborg ES. Results from aerosol measurement in amine plant treating gas turbine and residue fluidized catalytic cracker flue gases at the CO₂ Technology Centre Mongstad. Energy Procedia (GHGT- 13), Forthcoming 2017.
9. Faramarzi L, Thimsen D, Hume S, Maxson A, Watson G, Pedersen S, Gjernes E, Fostås BF, Lombardo G, Cents T, Morken AK, Shah MI, de Cazenove T, Hamborg ES. Results from MEA testing at the CO₂ Technology Centre Mongstad: Verification of baseline results in 2015. Energy Procedia (GHGT-13), Forthcoming 2017.
10. Morken AK, Pedersen S, Kleppe ER, Wisthaler A, Vernstad K, Ullestad Ø, Flø NE, Faramarzi L, Hamborg ES. Degradation and Emission Results of Amine Plant Operations from MEA Testing at the CO₂ Technology Centre Mongstad. Energy Procedia (GHGT-13), Forthcoming 2017.
11. Hjelmaas S, Storheim E, Flø NE, Thorjussen ES, Morken AK, Faramarzi L, de Cazenove T, Hamborg ES. Results from MEA amine plant corrosion processes at the CO₂ Technology Centre Mongstad. Energy Procedia (GHGT-13), Forthcoming 2017.

Collection overview: Research for more than 10 years