Process simulation using Aspen HYSYS

2.1. Process description

The process flow diagram of the conventional PCC process is shown in Fig. 1. Before the flue gas enters the absorber, a blower and a direct contact cooler (DCC) or flue gas scrubber is used to set temperature and pressure, remove any sulfur oxides in general, and condense most of the water vapor in the flue gas. In the absorber column, the cooled flue gas transfers CO₂ to the solvent while passing through the three packed beds in a counter-current flow direction (Nexant Inc., 2016) and the clean flue gas is vented from the top of the absorber. This stream contains nitrogen, oxygen, water, a small amount of CO₂ and traces of solvent. A rich-lean heat exchanger, typically a plate and frame exchanger, is located between the absorber and the stripper columns for heat recovery. Inside the stripper column (also known as the desorber or solvent regenerator), the rich amine passes through the packed sections of the column to release the CO₂. At the bottom of the stripper, steam is used in the reboiler to boil-up part of the solvent and return it back to the stripper. Finally, the majority of the lean solvent is recycled back to the absorber after cooling in the rich-lean exchanger and subsequently in the lean solvent cooler.

After the top of the stripper, the overhead (a mixture of CO₂ and water) is cooled to condense most of the water that is returned to the stripper. Then, there is a compression section to compress the CO₂ and make it ready for transportation via pipeline. The captured CO₂ is compressed to 15,700 kPa (157 bars) in multi-stage compressors with inter-stage coolers to cool down the compressed CO₂ and remove as much moisture as possible between each stage. At 45 °C and 15,700 kPa (157 bar), CO₂ is a supercritical fluid and less energy is required for its transportation through pipeline (Nexant Inc., 2016). The process specifications of the major equipment in this process configuration are listed in section 2.3.

2.2. Process modifications

Different modifications have been investigated separately and in combinations to minimize the energy demand of the process. It is recommended to use the sensible heat or the latent heat of the solvent streams by flashing a hot liquid stream at the top or at the bottom of the stripper column (Le Moullec and Kanniche, 2011a). This produces a hot gaseous stream (boiled-up) that mainly contains steam, which is then recompressed to the corresponding operating pressure of the stripper and is injected below the condenser stage or above the reboiler stage. In the case of the reboiler, steam recompression increases the temperature of the boil-up stream and therefore decreases the boil-up flow rate, which results in less energy consumption in the reboiler.

In principle, in absorber intercooling (AIC) modification, a part of the solvent is taken out of the absorber, cooled and then pumped back into the absorber with a typical pump-around arrangement (see Fig. 2).

Lean vapor recompression (LVR) is a modification, presented in Fig. 2, through which the boil-up liquid from the reboiler at the bottom of the stripper is sent to a flash drum with lower pressure to create lean vapor. The vapor is then compressed and sent back to the stripper column. Compression increases both the temperature and pressure of the lean vapor and decreases the reboiler duty. The lean liquid stream from the flash drum is recycled back to the absorber column. In the parallel exchanger arrangement (PEA) modification, shown in Fig. 2, the rich solvent stream is split into two streams and the divided streams exchange heat with the lean solvent stream and the overhead vapor of the stripper before the condenser. Despite often being considered plug-in units, like any other carbon capture, amine-based PCC process units require some attention when integrated into an existing plant (Feron, 2016). In order to maximize the synergies between the two operations and reduce the cost associated with carbon capture, the following points should be considered:

i)

Heat integration from the hot flue gas: this will provide some of the regeneration energy required in the PCC unit, and the hotter the flue gas, the larger the amount of regeneration energy that can be provided by heat recovery. Otherwise, all the required energy has to be provided by a separately installed steam generation system.

ii)

If low-carbon electricity is available from the grid, then use electricity-driven pumps and blowers rather than steam-driven equipment.

iii)

The PCC unit uses the control system of the parent plant that is intended for capturing CO₂ from its flue gas.

Since the installation of commercial-scale PCC units with distinctive design improvements is costly, process simulation plays a pivotal role in system optimization and evaluation of the various process modifications. Among the explained process design modifications, AIC, LVR, and rich solvent split are known to be less complex and more applicable. AIC and LVR were among the most recommended modifications in the literature (Feron, 2016) and are available in some commercial applications. For the sake of comparison with the study performed by Nexant as reference (Nexant Inc., 2016), we chose similar modifications (i.e., AIC and LVR). PEA is preferred over other modifications due to the minimal amount of additional equipment required for its implementation. Nonetheless, a thorough analysis is necessary to conclude whether more advanced modifications are justified for a special amine-based solvent considering carbon capture efficiency, reboiler heat intensity, capital investment, and operating costs. In this study, the determining factors and key operational variables were varied over a feasible range in order to have a grasp of the overall performance of the process with respect to the key performance indicators (KPIs), including carbon capture efficiency, reboiler heat demand, and cooling requirement.

A base case simulation model was developed according to Nexant’s study for a small-scale PCC unit using MEA-based absorption capturing 20 tonnes per day (TPD) from a natural gas-fired power plant (Nexant Inc., 2016). The model was validated against plant data. Then it was expanded to a large-scale model to capture 2000 TPD, and the results were again validated against existing plant data. Finally, the simulation model was used to capture the CO₂ from flue gas produced by the once-through steam generators (OTSGs) of a typical in-situ oil sands extraction facility. The flue gas characteristics are presented in Table 2. All configurations aimed for a 90% CO₂ capture rate from the flue gas. The proposed process modifications considered different combinations of AIC, LVR, and PEA as follows: 1) AIC only, 2) AIC and PEA, 3) AIC and LVR, 4) AIC, PEA and LVR. All simulations were performed for two aqueous solvents: a benchmark MEA (MEA 30% by wt.) and an activated a-MDEA (30% MDEA by wt. with 12.5% PZ by wt.). In addition, we tried to fill in the gap on the advantages of using a-MDEA as an advanced amine solvent for carbon capture. AIC, LVR, and PEA (combined with split flow arrangements) were simulated and compared with respect to the mentioned KPIs and cost. Including the base case design with no process modifications, a total of five different scenarios were simulated for each solvent, with each scenario representing a separate configuration.

2.3. Simulation assumptions

All of the simulation cases were created in Aspen HYSYS® v10. The amine (Chemical Solvents) fluid package was used in the simulation environment.

Several aspects and assumptions were considered for the simulation of the different scenarios, as follows:

•

Aqueous solutions of MEA (30 wt%) and a-MDEA (30 wt% MDEA and 12.5 wt% PZ) were used.

•

Special adjustments were applied to the numeric solver to expedite the convergence of several reactions in the process and the recycle loops with significant amine flow rates.

•

No heat loss to ambient was taken into account.

•

A rate-based stage-wise model was created to simulate the multi-component reaction and mass transfer efficiencies in the absorption and desorption sections of the process.

•

For both absorber and regenerator (stripper or desorber) columns, a rigorous non-isothermal (advanced rate-based modeling with internals in the column settings) set of reactions with mass transfer in the gas-liquid phases was applied to achieve accurate simulation predictions.

•

Heat-stable salts, as products of amine degradation, were not considered to moderate the long run times.

•

Make-up streams for water and amines were included to compensate for the carryover in the clean flue gas vented from the top of the absorber.

•

Spreadsheet variables and formulas were implemented to calculate special (user-defined) indices.

•

The packed sections of the absorber were divided into 25 packing stages. The top 10 stages were allocated to the two water wash sections and the bottom 15 stages were allocated to the absorption section. We considered intercooling at the bottom of the absorber column since Gomez increased column efficiency by considering this modification. For the cases with AIC modification, the pump around was located on top of the absorption section (Gomez, 2015).

•

The stripper was divided into 20 packed stages with the top 5 stages operating as a reflux wash section.

•

A cutter block was used to transfer the global fluid package to the Plocker equation of state (EOS) for the CO₂ compression section. This was to maintain the supercritical CO₂ in the liquid phase after the final stage of compression before its pipeline transportation.

•

The steam consumption and the required cooling duty were calculated for each case at a 90% CO₂ capture rate.

The process specifications for the major equipment in the simulation are tabulated in Table 3 below.

As can be observed in Table 3, the main difference is in the absorber diameter between the cases with MEA solvent and the ones with a-MDEA solvent. In order to have a fair comparison between those cases, the same values of design parameters (except for the absorber column diameter, which is different in MEA and a-MDEA case studies) were used in the simulations.

In general, MDEA has a lower heat of regeneration compared to MEA. However, MDEA reaction rates with CO₂ are significantly lower than those of MEA. As a matter of fact, these behaviors stem from the different chemistry and chemical equilibriums of the solvents with carbon dioxide. To promote the slow reaction rates of MDEA and CO₂, piperazine (PZ) is added as a promoter. Although a considerable amount of PZ was added to promote the reaction rates, the size of the absorber columns in the MDEA cases was approximately two times bigger in volume and 38% bigger in diameter. The absorber diameter was set to have the desired carbon capture rate of 90% and, at the same time, feasible hydraulics for the column. This means that the operating point is located in a reasonable pressure drop region with sufficient vapor and liquid flow rates along the height of the column, to maintain the hydrodynamics. The stripper dimensions were kept constant in all cases and the only differences are the reboiler duty, and correspondingly the condenser duty, that were adjusted to reach appropriate CO₂ concentration in the overhead and CO₂ loading in the regenerated/lean solvent. Conventional process configurations were first simulated for each solvent as the base cases. All of the simulated scenarios were then compared to those base cases and ranked with respect to their performance indicators.

Fig. 1 depicts a schematic of the entire PCC process simulation. The entire process is divided into three sub-flowsheets, namely flue gas cooling, CO₂ capture, and CO₂ compression in order to better manage the convergence of the simulation. Fig. 2 represents the simulation model of the carbon capture section with all proposed process modifications that were detailed in the methodology section. In all analyzed simulation cases, the rate-based absorber columns were used for both absorber and stripper columns. In the Column Environment section, Advanced Modeling was selected, hydraulic plots along the height for each column were monitored, and pumparounds were added to take into account the absorber’s internal cooling. The Internals Manager feature was also used to divide each bed into segments with different packing types. Recycle stream flow rates and the outlet temperature of the lean/rich heat exchanger (shown as L/R HE in Fig. 2) were carefully adjusted to avoid a cross temperature effect in the heat exchanger and to maintain a realistic run time as well as tolerance for the convergence of the recycle loops.

Unit operations that exist in an amine contactor unit are specified in the chemical solvents/amines property fluid package in Aspen HYSYS®. A rate-based stage-wise model was created to simulate the multi-component reaction and mass transfer efficiencies in the absorption and desorption sections of the process. Mass and energy balance equations along the height of the columns were solved by means of a rigorous non-isothermal set of reactions with mass transfer in the gas-liquid phases. The base case models were validated against detailed PFD (Process Flow Diagram) data at small and large scales published in our reference study (Nexant Inc., 2016).

In this simulation work, Mellapak+ type 250Y was selected for the bottom bed of the absorber while 3.5″ pall rings were used for the two beds at the top of the absorber and the stripper. This was in accordance with the data presented in Nexant’s study (as listed in Table 3). Regarding the columns’ hydrodynamics, diameters were estimated based on 75% of flooding for the gas phase velocity. Hydrodynamics curves were checked with the latest feature offered in Aspen HYSYS® columns named Advanced Modeling with internals. In the absorber, the diameter and lean-amine flow rate were adjusted to achieve CO₂ recovery of 90%. In the stripper, the condenser temperature and the reboiler duty are specified considering lean/rich loading, lean amine boil-up temperature in the reboiler, and a carbon capture rate of 90%.