Optimisation of post-combustion CO2 capture for ...

Available online at www.sciencedirect.com

ScienceDirect Energy Procedia 00 (2013) 000–000 www.elsevier.com/locate/procedia

GHGT-12

Optimisation of post-combustion CO2 capture for flexible operation N. Mac Dowella,b* and N. Shaha a: Centre for Process Systems Engineering, Imperial College London, South Kensington, London SW7 2AZ UK b: Centre for Environmental Policy, Imperial College London, South Kensington, London SW7 1NA UK

Abstract

In order to accommodate the increasing penetration of intermittent renewable electricity generation capacity, it is becoming increasingly clear that decarbonized fossil-fired power plants will have to operate in a highly flexible fashion. In this study, using detailed mathematical models of a coal-fired power plant integrated with a MEA-based post-combustion CO2 capture plant, we present a technical and economic analysis of several distinct modes of flexible operation. Using multi-period dynamic optimization techniques, we evaluate solvent storage, exhaust gas venting and time-varying solvent regeneration using average carbon intensity and profitability as key constraints and objective functions, respectively. Load following operation of the power plant with a tightly integrated capture plant is used as our base case scenario. We find that solvent storage is 4% more profitable than the base case, whereas exhaust gas venting is 17% more costly and appears incapable of meeting our decarbonisation targets. Time varying solvent regeneration is 16% more profitable than the base case and the electricity generated has an average carbon intensity which is approximately 5.5% lower than that of the base case. © 2013 The Authors. Published by Elsevier Ltd. Selection and peer-review under responsibility of GHGT. Keywords: CCS; CO2 capture; flexible CCS, SC-PCC, amine-scrubbing, SAFT, rate-based modelling, dynamic optimisation

1. Introduction As the world’s energy landscape changes, so too will the role of fossil fuels within the energy system. With increasing penetration of intermittent renewable energy generation, the ability of decarbonized fossil fuels to operate

* Corresponding author. Tel.: +44(0) 20 7594 9298. E-mail address: [email protected] 1876-6102 © 2013 The Authors. Published by Elsevier Ltd. Selection and peer-review under responsibility of GHGT.

N. Mac Dowell and N. Shah/ Energy Procedia 00 (2013) 000–000

2

in a flexible manner will become increasingly important. The requirement for decarbonized power plants to operate in a flexible manner has been understood for some time [1] and several options for the flexible operation of power plants integrated with post-combustion CO2 capture have been identified; typically flexibility has entailed either solvent storage or exhaust gas venting. This is usually proposed as way to allow the power plant to operate in a profit maximizing manner, exploiting peaks in power demand and electricity prices. However, it is important to ensure that this flexible operation does not increase the average carbon intensity of the electricity generated by the power plant over a given period. Thus, we distinguish between the instantaneous Degree of Capture (DoC) and the Integrated Degree of Capture (IDoC). The DoC is, as usual, given by

 CO2Generated  CO2Emitted  DoC  100.   CO2Generated   where

(1)

CO2Generated is the CO2 generated by the power station and CO2Emitted is the CO2 emitted to atmosphere.

However, from the perspective of the dynamic operation, it is the average carbon intensity of the electricity generated by the decarbonised power plant that is of interest, and therefore we define the Integrated Degree of Capture to be: tf

IDoC   DoCdt

(2)

t0

where t0 and t f are the start and end times of the period of interest. This is obviously general and can be specified to be any relevant time-period, e.g., a day, week, month or even a year. Thus, any optimization problem associated with evaluating different modes of flexible operation must include an end-point constraint on the average carbon intensity of the electricity generated. It is generally agreed that the CO 2 intensity of the power sector must be reduced by 90% [2, 3]. We therefore adopt this as our end-point constraint in this work. We consider that decarbonized power plants will be required to operate in a load following manner as illustrated in Figure 1. In the remainder of this paper, we apply the theory of Grossmann and Sargent [4] to dynamic, non-equilibrium models of a decarbonized power plant[5-12] and evaluate three distinct options for flexible operation:

Figure 1: Load following behaviour of a decarbonised power plant. In this graph, the symbols represent the power plant capacity factor (blue squares) and the electricity price (green triangles). The continuous red and dark yellow curves represent the simulated load factor and electricity price respectively. The dashed grey lines represent period boundaries. From this, it can be observed that there are 6 distinct periods of operation across the 24 hours in this study, each denoted by roman numerals. The question which we are addressing in this study is what design and operating procedure of the CO2 capture plant will provide optimal, i.e., profit maximising, behaviour across all of these periods.


1. 2. 3.

3

Solvent storage – a portion of solvent is stored during periods of peak electricity demand and regenerated during off-peak periods Exhaust gas venting – a portion of the exhaust gas is vented during periods of peak electricity demand Time-varying solvent regeneration – CO2 is allowed to accumulate in the working solvent during periods of peak electricity demand and the solvent is more thoroughly regenerated during off-peak periods

In all cases, we use a load-following power plant with post-combustion CO2 capture as a reference case. We evaluate each option for flexible operation based upon the IDoC and cumulative profit realized by the power plant over the course of the simulation. Our economic considerations are based on both fuel (coal and gas) prices and a cost of CO 2 emission based on the carbon price floor proposed by the UK’s Department of Energy and Climate Change †. The time period we are considering is the early 2030’s.

2. Evaluation of flexible operation In this section, we present the results of our optimization problem. For all scenarios, the optimization problem solved was the maximization of profit subject to the end point constraint of the IDoC being greater than or equal to 90% ‡. 2.1. Load-following This was our benchmark scenario. Here, the capture plant was designed and the operating parameters specified via a steady state optimization as described in our previous paper [13]. We evaluated the effect of a multi-period dynamic optimization had on the end design of the plant. However, as the duration of the period for which the power plant is operating at full load is long relative to the period for which the power plant is operating at part load, the multiperiod design was essentially identical to the steady state design, and the solution period of the steady state design problem was an order of magnitude faster than that of the multi-period problem. Thus, in this case, all operating parameters in the CO2 capture plant, i.e., the L/G ratio, lean loading, θlean, solvent inlet temperature, 𝑆𝑜𝑙𝑣 𝑇𝐼𝑛 etc., were held constant throughout the simulation. It can be observed from Figure 2Error! Reference source not found. that, as the capacity factor of the power plant changes, so too does the instantaneous degree of capture (DoC), albeit by a very small amount. This is associated with the changing gas and liquid flowrates in the absorption column leading to a variation in Figure 2: Degree of CO2 capture varying with power plant capacity factor (CF). the effective surface area in the packed section. This implies that if the duration or frequency of periods of dynamic operation are long or great relative to

†

DECC: www.gov.uk/government/organisations/department-of-energy-climate-change In the case of exhaust gas venting, this constraint had to be loosened to IDoC ≥ 89% as it was not possible to solve the optimisation problem with end-point IDoC constraint ≥ 90% ‡


4

the duration or frequency of periods of steady state operation, the integrated degree of capture (IDoC) could be reduced as a result. Illustrated in Figure 3, we provide an evaluation of the various financial streams associated with the power plant. It can be observed that the fuel cost per MWh is slightly increased during periods of operation at a low load factor in comparison with periods of operation at a high load factor. This is commensurate with the decreased efficiency of the power plant under this mode of operation. The cost associated with CO2 emission is similarly elevated during this period, commensurate with the slightly reduced DoC as illustrated in figure 9. Finally, as the plant is operating at an essentially constant degree of capture of approximately 90%, it can be observed that the CO2 price exerts an important influence on the profitability of the plant.

Figure 3: An analysis of the various financial streams associated with the generation of low carbon electricity in the case of conventional load following.

2.2. Solvent storage In this scenario, the power plant ramps up and down as illustrated in Figure 1, but in order to decouple the operation of the power and capture plants, we consider the option of storing a fraction of the rich solvent during periods of peak electricity prices and subsequent regeneration of the solvent during periods of off-peak electricity prices. Here, the extra parameter to be determined is the quantity of solvent stored or regenerated in each time period. Therefore, this is a multi-period, piece-wise linear dynamic optimisation problem. As can be observed, when electricity prices are lowest, the flowrate of solvent to regeneration is greatest. Similarly, during periods of peak Figure 4: An analysis of the solvent storage and regeneration is presented here. electricity price, rich solvent is directed to storage, bypassing the regeneration process.


5

The advantage of the solvent storage strategy is that it reduces the quantity of steam required for solvent regeneration during key periods of peak electricity demand, as illustrated in Figure 5.

Figure 5: Steam requirement for solvent regeneration under the solvent storage option.

However, the availability of steam is a major constraint on the amount of solvent which can be stored during periods of peak electricity demand. It was a constraint in this problem that whatever solvent was stored during periods II and IV had to be regenerated during periods I, III, V and VI. This meant that it was only possible to store approximately 15% of the total solvent flow. This corresponds to a cumulative volume of stored solvent of approximately 11,750 m3 of solvent over both periods. 2.3. Exhaust gas venting In this scenario, the power plant ramps up and down as illustrated in Figure 1, but in order to decouple the operation of the power and capture plants, we consider the option of venting a fraction of the exhaust gas during periods of peak electricity prices. Here, the extra parameter to be determined is the quantity of exhaust gas to be vented in each time period. Therefore, this is a multi-period, piece-wise linear dynamic optimisation problem. The key result of this optimization problem is illustrated in Figure 6 opposite.

Figure 6: Exhaust gas venting. In this scenario, approximately 23% of the exhaust gas was vented during periods of peak electricity prices.


6

As can be observed from Figure 6, during periods of peak electricity prices, approximately 23% of the total exhaust gas was vented. In this scenario, we departed from the conventional procedure of capturing 90% of the CO 2 at all times. Rather, of the exhaust gas which was introduced into the absorption column, 95% of the CO 2 was captured. However, during the periods of exhaust gas venting, the DoC fell to approximately 74%, increasing the carbon intensity of the electricity generated to approximately 225 kg/MWh. This is illustrated in Figure 7. It is noted that the venting of this quantity of CO2 had the primary consequence of imposing a significant cost penalty on the profitability of the plant and the secondary consequence that it was not possible to solve the optimization problem with the end-point constraint of IDoC ≥ 90%. Here, this had to be loosened to 89%.

Figure 7: Variation in DoC with exhaust gas venting. It can be observed here that the absorber DoC remains approximately constant.

2.4. Time-varying solvent regeneration In this section, we evaluate the option of using the working solvent as means to provide flexibility to the power plant. This is achieved by allowing the CO2 to accumulate in the working solvent during hours of peak electricity prices and regenerating the solvent more completely during off peak periods. This is not storing solvent separately to the capture plant; this is storing CO2 in the solvent which is circulating within the capture plant. This means that the lean loading of the solvent is no longer a time-invariant process parameter. Rather, this control vector is now parameterised such that it is expressed as k  Lean   kt 2   kt   k

(3)


7

where  Lean is the function describing the way in which  Lean varies across a given period k,  ,  and  describe this function in each period k and finally t is the time in each period k. Here, the variable t is set to zero at the beginning of each new period. The only additional constraint we have imposed upon the optimisation problem by this formulation is the quadratic nature of the parameterisation. This could have equally been a cubic or higher order polynomial, however a quadratic polynomial was chosen in the interest of simplicity. Further, we did not constrain the values of the coefficients of this polynomial, i.e., the magnitude of the subsequent behaviour was strictly a function of the process response to the time-varying electricity prices. The results of this optimisation problem are presented in Figure 8. k

k

k

k

Figure 8: Solvent regeneration as a function of time and electricity price is shown here.

As can be observed, as opposed to operating at a constant lean loading (the continuous blue curve), the degree of solvent regeneration varies in sympathy with the prevailing electricity price, as might be expected for the solution of this kind of optimisation problem. The solvent is deeply regenerated during periods of low electricity price, whilst CO2 is allowed to accumulate in the working solvent volume during periods of high electricity price. This has the effect of allowing the plant to direct substantially less steam to solvent regeneration operations when the opportunity cost associated with doing so is high. Obviously, following this kind of operating strategy will have the effect of varying the carbon intensity of the electricity generated.


8

Figure 9: Variation in carbon intensity of electricity generated with the variable solvent regeneration approach

As can be observed, from Figure 9, the carbon intensity varies quite significantly over the course of the day. For some periods, when the solvent has been deeply regenerated, the carbon intensity tends towards zero, whereas at other times is closer to 160 kg/MWhre, corresponding to a DoC of approximately 84%.

2.5. Comparison As stated in the introduction, the aim of this work was to identify which operating modes allowed the decarbonised power plant to operate in a profit maximizing manner whilst simultaneously reducing the carbon intensity of its electricity by a minimum of 90% over a given period. Considering first the instantaneous degree of capture, DoC, it was observed that in both the base case and the solvent storage scenario, the DoC remained approximately constant throughout the simulation. This could be of particular importance in the case of an exceptionally restrictive regulatory regime in which the carbon intensity of the electricity was only permitted to fluctuate within very tight bounds. Conversely, the DoC corresponding to the exhaust gas venting and the variable solvent regeneration scenarios varied throughout the simulation. Specifically, the DoC associated with the exhaust gas venting option varied in the range 75% to 95% whereas the DoC corresponding to the varying solvent regeneration varies in the range 88% to 100%. This is illustrated in Figure 10.


9

Figure 10: Comparison of DoC for each scenario

Illustrated in Figure 11 is the performance of each option based on the IDoC. It is apparent that both the base case and solvent storage options provide an IDoC of approximately 90%, the exhaust gas venting option has an IDoC of approximately 89% whereas the variable solvent regeneration option has an IDoC of approximately 96%. This would serve to emphasise the value of an appropriately flexible regulatory regime if flexible CCS is to be available within the energy system.

Figure 11: Comparison of IDoC for each option


10

It is finally instructive to consider the profitability of the power plant – this is illustrated in Figure 12 below. It is apparent that the solvent storage option offers marginally greater profitability (~ £374k) when compared with the base case (~£359.6k), approximately a 4% improvement, whereas the exhaust gas venting option (~£298.6k) is approximately 17% less profitable than the base case. Finally, the variable solvent regeneration option (£417.5k) is 16% more profitable than the base case scenario.

Figure 12: Cumulative profit available for each option

3. Conclusions Under the scenarios considered here, it would appear that exhaust gas venting is unlikely to be a cost-effective scenario for the provision of flexible, low carbon electricity generation. The option of solvent storage would appear to provide a marginal benefit, but this inevitably comes at with the additional cost of a large, on-site, solvent inventory and the associated regulatory burden. The concept of allowing CO2 to accumulate in the working solvent appears to offer the option of increased operational flexibility, improving plant profitability whilst concurrently allowing the generation of low carbon electricity. Obviously these economic conclusions are a function of the prices of fuels, peak and off-peak electricity and the cost of CO2 emission to atmosphere, and the electricity market in which the power plant must operate, with different strategies being more or less viable as the market size, structure, type of CCS technology available and degree of interconnection vary [14-20]. Acknowledgements The research leading to these results has received funding from the European Union’s Seventh Framework Programme (FP7/2007-2013) under CO2QUEST Grant Agreement number 309102. References 1.

2.

Gibbins, J.R. and R.I. Crane, Scope for reductions in the cost of CO2 capture using flue gas scrubbing with amine solvents. Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy, 2004. 218: p. 231-239. Mac Dowell, N., et al., An overview of CO2 capture technologies. Energy & Environmental Science, 2010. 3(11): p. 1645-1669.


3. 4. 5.

6. 7. 8.

9. 10.

11.

12.

13. 14. 15. 16. 17. 18. 19. 20.

11

Boot-Handford, M.E., et al., Carbon capture and storage update. Energy & Environmental Science, 2014. 7(1): p. 130-189. Grossmann, I.E. and R.W.H. Sargent, Optimum design of multipurpose chemical plants. Ind. Eng. Chem. Process Des. Dev., 1979. 18(2): p. 343-348. Arce, A., et al., Flexible operation of solvent regeneration systems for CO2 capture processes using advanced control techniques: Towards operational cost minimisation. Int. J. GHG Con., 2012. 11: p. 236250. Mac Dowell, N. Optimisation of post-combustion CCS for flexible operation. in The 14th Annual APGTF Workshop - 'The Role of Fossil Fuel Power Plant in Providing Flexible Generation'. 2014. London. Mac Dowell, N., et al., Integrated solvent and process design for the reactive separation of CO 2 from flue gas. Comp. Aid. Chem. Eng., 2010. 28(1231-1236). Mac Dowell, N., et al., Transferable SAFT-VR models for the calculation of the fluid phase equilibria in reactive mixtures of carbon dioxide, water, and n-alkylamines in the context of carbon capture. J. Phys. Chem. B., 2011. 115(25): p. 8155-8168. Mac Dowell, N., N.J. Samsatli, and N. Shah, Dynamic modelling and analysis of an amine-based postcombustion CO2 capture absorption column. Int. J. GHG Con., 2013. 12(247-258). Mac Dowell, N. and N. Shah, The multi-period operation of an amine-based CO2 capture process integrated with a supercritical coal-fired power station. Computers and Chemical Engineering, 2014 (Under Review). Mac Dowell, N. and N. Shah, Dynamic modelling and analysis of a coal-fired power plant integrated with a novel split-flow configuration post-combustion CO2 capture process. Int. J. GHG Con., 2014. 27: p. 103119. Rodriguez, J., et al., Modelling the fluid phase behaviour of aqueous mixtures of multifunctional alkanolamines and carbon dioxide using transferable parameters with the SAFT-VR approach. Mol. Phys., 2012. 110(11-12): p. 1325-1348. Mac Dowell, N. and N. Shah, Identification of the cost-optimal degree of CO2 capture: An optimisation study using dynamic process models. Int. J. GHG Con., 2013. 13: p. 44-58. Wijk, P.C.v.d., et al., Benefits of coal-fired power generation with flexible CCS in a future northwest European power system with large scale wind power. Int. J. GHG Con., 2014. 28: p. 216-233. Oates, D.L., et al., Profitability of CCS with flue gas bypass and solvent storage. Int. J. GHG Con., 2014. 27: p. 279-288. Davison, J., Flexible CCS plants–A key to near-zero emission electricity systems. Energy Procedia, 2011. 4: p. 2548–2555. Nimtz, M. and H.-J. Krautz, Flexible Operation of CCS Power Plants to Match Variable Renewable Energies. Energy Procedia, 2013. 40: p. 294-303. Chalmers, H., M. Leach, and J. Gibbins, Built-in flexibility at retrofitted power plants: What is it worth and can we afford to ignore it? Energy Procedia, 2011. 4: p. 2596-2603. Kalinin, A., et al., CCS Chain Capacity Selection for Flexible Load Power Plant. Energy Procedia, 2012. 23: p. 343-353. Ludig, S., M. Haller, and N. Bauer, Tackling long-term climate change together: The case of flexible CCS and fluctuating renewable energy. Energy Procedia, 2011. 4: p. 2580-2587.