Solvent post-combustion capture pilot testing for partially* de-risking commercial projects must include reclaiming and last at least 12 months

* The title says ‘partially' de-risking because pilot testing alone cannot fully de-risk the solvent-related aspects of an amine post-combustion capture project and the ability to change solvents, primarily through avoiding contractual blocks on doing so, should still be retained to give valuable additional de-risking options.? Another article on facilitating a solvent change option is coming soon.

_________________________________________________________________________

Pilot test references in the table above:

? 1. (Endo, 2011; Hirata, 2013 & 2014; Hill, 2014);

? 2. (DECC, 2013)

? 3. (Hallerman, 2013)

? 4. (Radgen, 2014; Reddy, 2013 & 2017)

? 5. (ROAD, 2019)

? 6. (Hitachi, 2012; MIT, 2013)

? 7. (Campbell, 2016; Cotton, 2017)

? 8. (Knudsen, 2014, Norcem, 2019a & 2019b)

? 9. (FOV, 2020b; Fagerlund, 2021)

__________________________________________________________________________

Discussion of pilot testing

A practical warning of what can happen without adequate pilot testing to help give advance warning of solvent-related issues on a commercial plant comes from experience at the Boundary Dam 3 (BD3) project.? In 2018 it was reported at GHGT14 that:

‘The capture facility at Boundary Dam has been operating since 2014, almost four years. During this time, there have been difficulties with the plant being able to supply the contracted CO2 to its off-taker. There were a significant number of design deficiencies and construction quality issues to manage. In addition, the Capture Plant continues to experience significant issues with the amine absorbent chemical that is fundamental to the process.

These issues were, and continue to be, tackled in order of priority: 1) safety, 2) reliability, and 3) efficiency and cost-effective operation. As SaskPower implemented projects to correct the issues of which it was aware, the process was further complicated by the emergence of previously undetected issues that required further corrective action. At times, this involved long lead times to procure and install specialized equipment. This, coupled with amine-related issues, has contributed to lengthy outages and underperformance of the plant.

Since the facility entered service, major work has been undertaken to:

1. Address safety issues and construction deficiencies;
2. Mitigate unanticipated reactions of the fly ash with the amine process;
3. Investigate and in some cases improve the systems designed to remove fly ash from the plant;
4. Increase thermal reclamation capability;
5. Mitigate increased degradation of the amine solvents;
6. Improve temperature control on various process units to meet design specifications;
7. Implement ongoing measures to clean fouling from heat exchangers;
8. Add isolation valves and selected redundancy and capacity increases on heat exchangers and process units to enable “on the-fly” cleaning of fouling; and
9. Incorporate instrumentation to measure fouling within the process control system.’

Except for 1 and possibly 3 (my numbering added) all of the issues above are related to the solvent and the equipment handling it.? It was subsequently noted in 2020 that:

‘……. the research currently available on post-combustion amine-based carbon capture is insufficient for adequately understanding interactions between amines and flue gases.

“Long-term testing of amines was quite often limited in duration around the time that BD3 was built. The data we have on the behaviour of the amine used on this particular facility does not reflect the accelerated degradation that occurred closer to 3,000 or 4,000 hours of run time.”

In the presence of the common components and undesirable particulates present in a flue gas stream, amines degrade and must be replaced with fresh amine solution for the capture process to continue optimally. Degradation products and operational challenges are unique to each of the different amines in combination with various flue gas streams. As such, piloting must adequately emulate the conditions of the final, full-scale process.

Pilot tests to help de-risk commercial amine PCC deployment need to verify that solvent-related problems either don’t occur over an extended period of realistic operation or, if they do occur, demonstrate fixes that can be implemented on the full-scale plant.? This differs from most pilot-scale PCC plant tests to date, which have been for R&D purposes, or for estimating short-term solvent performance relative to other solvents as a stage in solvent development.? Key potential solvent-related issues include:

a) Potential showstoppers:

• Emissions to atmosphere preventing plant operation until addressed (at worst possibly requiring months to years of modification)
• Emissions to water
• Operator health and safety
• Uneconomic project – higher-than-expected running costs that are in excess of revenues
• Uneconomic project – unexpected maintenance costs making capital recovery or even operation unviable

?b) Marginal improvements (scope for cost reduction):

• Reduced equipment sizing / lower-cost materials
• Reduced solvent make-up and waste disposal costs
• Improved equipment performance – e.g. better capture rates, reduced energy requirements
• Reduced financing costs since lower project risk premium

?c) In-between issues:

• Process upsets triggering closure – could exceed annual emission allowances in a short period
• Corrosion – reduced equipment lifetime or increased waste disposal costs – catalytic effect on degradation
• Solvent inventory or waste streams may be hazardous material
• Biological fouling

In particular pilot-scale PCC solvent testing needs to replicate key features of the proposed full scale PCC plant to produce relevant data:

• The flue gas and its variations during extended operation, with realistic cleaning
• The long term time/temperature/gas composition histories that will be experienced by the solvent,
• The ways that solvent degradation and the solvent reclaiming and other management techniques that will be used on the full-scale plant will interact to evolve the solvent inventory to something different, and probably much more complex, than the fresh solvent initial charge and make-up.

The inclusion of the proposed full-scale solvent reclaiming or cleaning, and having it successfully treating solvent that is fully ‘dirty’ rather than fresh solvent, is the key to having a relevant test for commercial deployment. Test periods therefore need to be long enough to allow this, and in particular to be able to demonstrate the ability to maintain a STABLE solvent composition (or composition range) that gives satisfactory PCC performance (i.e. with respect to operational issues such foaming, fouling, corrosion etc. as well as overall performance aspects such as capture rate, energy consumption and emissions).?

It should be obvious to engineers making solvent assessments that it must be possible to remove ALL degradation products and flue gas impurities at the average rates at which they are formed in, or added to, the solvent in a PCC system and still have satisfactory performance.? If this is not demonstrated, under fully realistic conditions, in pilot tests that are aimed at de-risking full-scale deployment then not only has satisfactory solvent management not been demonstrated but ALL aspects of PCC plant performance have not been properly assessed, because the solvent mix in the plant will not be representative.? Partial cleaning of the circulating solvent may slow down accumulation of some solvent impurities but only full cleaning/reclaiming, or bleed and feed, can remove everything.

Examples of some pilot tests undertaken to date specifically to de-risk commercial deployment (i.e. not for research and with a particular project and hence flue gas and plant configuration in mind) are given in the figure above.? It must be emphasized, however, that no comprehensive examination of the correlation between pilot plant and long-term full-scale PCC behaviour for coal-fired power plants has yet been published, and also, since no full-scale plants are in service, cannot yet have been undertaken for biomass or natural gas power plants nor for most other PCC applications.? A number of the de-risking pilot tests in the figure above are reported to have included solvent reclaiming or cleaning.? Examples of what has been said regarding two solvent de-risking test campaigns that apparently were aimed at commercial applications but that did not include reclaiming or cleaning are included as Annex 1.

Regarding the scale for testing, a larger scale is needed when measurements of absorber and stripper performance (i.e. in particular, required packing heights) are required to dimension the commercial unit more precisely and limit construction ‘safety’ margins.? MHI pilot testing at Plant Barry was undertaken at a scale of 500 tpd (~25MWe equivalent) but such a large-scale capture pilot is the exception, possibly because in this case the CO2 was also being used for a storage pilot.? Most other pilot tests on PCC for coal flue gases undertaken as the intended precursor to full-scale commercial units have been of the order of 50-100 tpd (and have vented the captured CO2). It therefore seems that this scale has been considered to be an adequate compromise between cost and realism, although obviously it would still be desirable to ensure that flow passage dimensions, velocities, temperatures etc. match those planned for a full-scale plant.? Subsequently, though, in a conceptual study for a pilot plant to support a PCC retrofit on a CCGT power plant it was stated that the minimum packed column diameter for adequate performance estimates was 0.4 m for the stripper, with a corresponding absorber diameter of 1 m and a capture rate, on gas turbine flue gas, of 11 tpd.? The conclusion seems to be that the minimum diameter should be checked with the specific packing supplier for the packing type that is planned to be used.

But, if pilot testing is intended mainly to cover solvent management, corrosion and emissions issues, rather than defining equipment sizing more accurately, then smaller pilot test unit scales might, in principle, be used, provided that key factors such as solvent temperatures, residence times, CO2 loadings, liquid to gas ratios, materials of construction and, critically, solvent reclaiming or cleaning arrangements are reproduced.? The logic behind this is that fresh solvent performance on a range of flue gas CO2 concentrations can be readily measured with larger packed beds in existing pilot test plants such as Technology Centre Mongstad and the National Carbon Capture Center, even though these units may not be available for the 12 month dedicated test periods required for full performance verification (and cannot anyway provide the exact flue gas to be used).? The relative change in performance (e.g. capture rate, energy requirement and emissions) between the fresh and the aged solvent, at its stable long-term composition (or composition range, if intermittent reclaiming is used), can still be measured in smaller pilots and then be used to guide any derating that may be necessary.? But for this fresh-to-aged performance change, and for all the other solvent-related checks, there can be absolutely no compromise on applying representative solvent reclaiming/cleaning, which is where a limit on scaling may come in, with the size depending on the solvent to be tested.

For example, in a recent article on reclaiming it was shown that simple thermal reclaiming, just boiling the solvent in a vessel, collecting the vapour and discarding the residue, was very effective at removing impurities from MEA but could not remove all impurities from CESAR1, a piperazine/AMP blend.? Ion-exchange cleaning has been used as an alternative for CESAR1 in pilot testing, but it is not clear that this alone would be desirable for large-scale commercial use.? To reclaim another blended solvent, Shell Cansolv DC-201, in the Peterhead retrofit project in the 2nd UK CCS Competition it was proposed to use an ion-exchange reclaimer combined with a three-stage thermal reclaimer train, operating at 105°C and 1.23 bara, 75°C and about 0.1 bara and 162°C and 0.12 bara respectively, although this arrangement was apparently not built and tested.? The scale at which more complex reclaiming/cleaning arrangement like this can be applied may then be what limits the minimum scale for an effective commercial de-risking pilot test on a particular solvent.

Of course, pilot tests may be run for many months, particularly with more stable solvents and clean, or otherwise benign, flue gases, with steadily-rising impurity levels, but this is obviously an unsustainable mode of operation for commercial operation and none of the transient mixes of impurities will represent the stable long-term circulating solvent mix that will form with adequate cleaning/reclaiming (that will selectively remove or return different components and thereby adjust the proportions of impurities present) in place.? Moreover, solvent degradation, as measured in tests without reclaiming/cleaning, does not equate to solvent consumption, with additional solvent losses in the reclaiming/cleaning processes required to control the impurities arising from degradation likely to be at least as significant.? So these latter losses also need to be reproduced to estimate overall solvent makeup costs.

It is also possible to maintain a stable amount of impurities by semi-continuous (if impurities are catalytic) or intermittent bleed and feed of solvent, but again this is unlikely to be feasible at commercial scale without some measures to return the ’good’ solvent in the stream that is bled off rather than just feeding fresh solvent and, as shown with CESAR1 thermal reclaiming, this return stream may bring back specific impurities that will then accumulate, possibly to quite high levels, in the final stable mix.?

If representative reclaiming/cleaning can, however, be implemented then small-scale testing is quite desirable because it allows multiple parallel tests to be undertaken.? When at least a year at one set of operating conditions is required to establish confidence in the results for solvent-related behaviour then sequential testing of multiple sets of conditions or multiple solvents would necessarily be a slow business.? A team at the University of Sheffield is working on a unit to test solvent management at reduced throughput (SMART) at approximately 0.1 tpd scale with single- or two-stage semi-continuous thermal reclaiming – for MEA - that is fully integrated with the process.?? The same size unit would obviously work with solvents with similar reclaimability to MEA but more complex reclaiming requirements may present a challenge.? It is also easier to route the flue gas from a small pilot test into the combustion zone of a boiler, avoiding the risk of amine and other emissions to atmosphere and so facilitating permitting.

Finally, a couple a caveats.? The first is about materials.? The ideal, from a replication point of view,? is for stainless steel construction in the pilot plant, which is easy to reproduce, at least qualitatively, in the pilot plant.? The use of carbon steels, ceramic- or plastic-lined concrete, enameled steel etc. at full scale may be much more difficult to reproduce effectively in pilot plants, especially smaller ones, but obviously could be important for corrosion, catalysis by dissolved metals and physical or biological surface fouling.? Mistakes in materials of construction may also be more serious at pilot scale, in particular the inadvertent (or deliberate, by poorly-supervised or uninformed workers) inclusion of potentially catalytic metals such as copper or zinc (possibly as brass).? This may also occur on commercial plants but the relative quantities are likely to be much smaller if it does.? On a pilot plant even part of a fitting in the wrong material can still represent a relatively large part of the wetted surface.

The second caveat, at any scale of pilot, is not to rely on a very high level of absolute accuracy in reproducing solvent-related issues.? In detail neither the geometry of the pilot plant nor the actual materials of which it is constructed can be guaranteed to be the same as for a future full-scale commercial plant and phenomena such solvent degradation, material corrosion, fouling, foaming etc. have complex origins that are not fully understood – this is why pilot tests are recommended.? The flue gas fed to the plant may also be slightly different in a future commercial application due to fuel changes or differences in upstream treatment, even when a slipstream from the actual unit is used in the pilot tests.? The most valuable thing that can be achieved in a pilot test is to demonstrate that solvent reclaiming/cleaning systems can manage not only what is required under pilot testing conditions but also rather more if necessary; the types of impurities that need to be dealt with will probably be replicated even if not the exact quantities/formation rates.? Similarly, if included in the tests, emission controls such as an acid wash should be able to control higher rates of emissions than encountered in the pilot to allow for more adverse conditions in full-scale operation.

And, as a very last thought, the useful role of a pilot plant does not finish once the full-scale plant design is finalised or even when the full-scale plant is commissioned.? As illustrated at the start, solvent-relate problems in full-scale operation can take years to develop.? If a pilot plant that was set up before the main unit is kept running then it may give advance warning of any problems and, hopefully, be used to develop fixes for these problems before they have serious effects on commercial operation.? Pilot plants can also be used to test alternative solvents that can provide a backup if intractable solvent-related problems are encountered with the design solvent for a commercial plant.? As also noted at the start, even a successful pilot test is not a 100% guarantee that no problems will be encountered in service.

?

Further reading

The UKCCSRC recently published an updated evidence review of emerging techniques for post-combustion CO2 capture using amine-based and hot potassium carbonate technologies for power, energy-from-waste facilities, refining process units, cement and lime and other industries that includes additional information.

The Environment Agency has published ‘Guidance - Waste operations and A1 installations: carrying out research or trials - How to trial new operating models and technologies at regulated facilities’ that cites pilot testing of post-combustion capture as an example.

?

Annex: Examples of what was said regarding two solvent de-risking test campaigns that were aimed at commercial applications but that did not include reclaiming or cleaning

a) Cansolv DC-201 testing at TCM and NCCC for Peterhead 2nd UK CCS Competition Project

(Cotton, 2017): In addition, operational testing of the amine formulated for the PCCS project was undertaken at the world’s largest CO2 capture test plant, Technical Centre Mongstad (TCM). The testing campaign focused on the following elements:

·?????? mimic the Peterhead process conditions;

·?????? confirm the amine degradation rate;

·?????? measure the amine emissions; and

·?????? verify process performance including CO2 removal and energy consumption.

Results from the TCM campaign confirmed that the PCCS project design was fit for purpose and no design change was required in terms of emissions as regulatory requirements would be met.

Prior to the TCM campaigns the DC-201 solvent was also tested at NCCC (Shell, 2017) with results summarized as follows.? ‘In 2013, Cansolv DC201 was successfully tested under simulated CCGT flue gas conditions for 1715 hours of operation at the NCCC piloting facility in Wilsonville, AL, US. 90 (+/‐ 5) % CO2 capture was achieved and energy consumption requirements have not deteriorated before 1200 hours of operation, where the concentration of degradation products in the solvent hindered its performances (no bleed‐and‐feed or reclamation technology was used during the test).’

‘In general, in an amine‐based post combustion CO2 capture process, with no make‐up added, it is expected that CO2 capture declines over time. This is due to transformation of the main amine component to product(s) which do not have any CO2 capture capacity. This deterioration was not observed during NCCC 2014 campaign. One reason was the length of the test, which was not long enough to build significant amounts of degradation products and/or contaminants and then not long enough to loose significant amount of amine. In addition, CANSOLV DC‐201 transformation product maintains certain capacity for CO2 capture.’

On solvent management it was also stated ‘Depending on the flue gas composition, the solvent in the Cansolv CO2 Capture System can accumulate non‐regenerable salts (also called Heat Stable Salts) as well as various degradation products over time.? These contaminants must be removed from the solvent in order to maintain the guaranteed system performance. During the design stage of the project, Cansolv engineers will design for the removal of these contaminants by circulating a small fraction of the lean solvent flow to an amine purification unit (APU). The APU can be a simple Ion‐Exchange system designed to remove ionic species, or may be a thermal reclaiming stage, or a combination thereof. Validation and confirmation of this requirement is an optimization step to be done during an engineering phase of a project.’

b) Cansolv DC-103 pilot tests at FOV Klemetsrud

(Text below is from Fagerlund (2021), see also (Jemtland, 2019; FOV, 2020) for additional details)

Fortum Oslo Varme ‘decided (in July 2018) to build a 1:350 scale pilot plant to demonstrate that the selected Cansolv capture technology is suitable for cleaning CO? from the exhaust gases of the Klemetsrud [waste-to-energy] plant, and in particular to show that the emissions of amines and solvent degradation products to air are within the set requirements. The duration for a successful demonstration of the first pilot campaign were decided to be at least 2000 operational hours, with total amine emissions lower than 0.4 ppmv on average over the last 500 h of testing (of the first test campaign).’

‘Notable differences to the full-scale plant design [(FOV, 2020)] were that:

?????? No thermal reclaimer unit (TRU) was installed;

?????? No mechanical vapour recompression (MVR) system was installed;

?????? Steam supply was provided from a separate steam generator instead of from the [waste-to-energy] plant;

?????? No caustic was injected in the pre-scrubber;

?????? The cooling system was designed as a once-through system, i.e. no closed loops in the cooling system;

?????? All piping and vessels were made from stainless steel 316 (unlike in the full-scale design which consists of various materials optimized considering area of use)

These differences are mostly linked to cost and complexity considerations and were accepted as they did not affect the relevance of the pilot plant results for the full-scale plant.’

………………

‘Solvent purification

As mentioned above, the pilot plant is not equipped with a solvent reclaiming system. Starting with pure solvent, the degradation products concentration is left uncontrolled and allowed to progressively increase over the campaign. This allows to accurately estimate the degradation rate of the solvent, and also to compare performance at different concentrations of degradation products. The only solvent purification system in place is a mechanical / activated carbon filtration system that can be brought online if and when required’.

………………

‘The [PCC] pilot plant was in operation between March and December 2019 and the total number of successful pilot plant operational hours at Klemetsrud reached about 5100. The purpose of running the pilot plant beyond the original 2000 h was to obtain additional knowledge of operating the plant at higher concentrations of solvent degradation and various upset conditions. …. At the end of the campaign, the [total degradation products – three individual degradation products were shown but not identified] degradation product concentration in the solvent exceeded 5 wt% (i.e. around 10 % degraded solvent) without using a reclaimer. This is above the full-scale plant design envelope (1-2 wt% degradation products), where a thermal reclaiming unit (TRU) will maintain and control the degraded product concentration. The impact of a higher than design degradation product concentration has been negligible and indicates an opportunity to improve/ optimise the TRU/ full-scale CC plant design. An indication of a possible acceleration of degradation product build-up rate was only observed at the most elevated concentrations at the end of the campaign but would have needed continued operation to be confirmed.’

?