Before going to study a real model of an existing refinery, it is necessary to validate the model built in Aspen. To do this, the two models described in Chapter 3 are analyzed. The two models use the values of the example refinery found in the literature and very precisely reflect these values. The ELECNRTL model is commonly used for the aqueous amine-based CO2 capture (including MEA) and for mixed solvent systems. The ELECNRTL model is
Figure 4.4: Process flow diagram of the MEA process for post-combustion CO2 capture. [33]
Chapter 4
also used for the calculation of the properties of electrolytes and can handle mixed solvent systems at any concentration from low to very high; for the gas phase, on the other hand, the Redlich-Kwong equations of state are used. The solubility of critical gases is modeled with Henry's law. The ELECNRTL method offers the advantages of being the most versatile in the calculation of properties. Furthermore, this method is consistent with the ENRTL-RK method, in fact the interactions between the molecules are calculated in the same way. The second model chosen is the ENRTL-RK one. This model is based on unsymmetric electrolyte NRTL property model. It uses the Redlich-Kwong equations of state for the computation of steam properties, the unsymmetric reference state for ionic species (infinite dilution in aqueous solution), henry's law for the computation of the solubility for supercritical gases and unsymmetric electrolyte NRTL method of handling zwitterions. The ENRTL-RK model is identical to the ELECNRTL model for systems with a single electrolyte, however for systems with multiple electrolytes, it uses the mixing rules only to calculate interaction parameters, instead of calculating both interaction parameters and Gibbs free energy from mixing rules.
The Table 4.9 compares the values found by the two simulations with the reference values of the literature.
Table 4.9: Comparison between the literature values and those calculated in Aspen Plus.
The difference between the literature values and the calculated ones are shown in the Table 4.10 and commented.
It is evident that the values are close to those in the literature, in fact in all cases the relative error is minimal. Both models reflect well what is literature; the only values that differ enough from those of the lean MEA flow coming out of the stripper. This is because the flow rate is high and in fact it can be seen in the Table 4.11 that the relative error is in any case very low. Again, with reference to the relative error, the reboiler temperature is different by 4 °C which however does not exceed 5% of error and therefore is an acceptable value.
Literature ENRTL-RK ELECNRTL
CO2 in Flue Gas kg/h 41754,4 41754,4 41754,4
CO2 in Clean Gas kg/h 4174,4 4107,2 4217,0
Clean Gas Temperature °C 60,7 64,9 64,7
Reboiler Temperature °C 119,7 121,1 121,5
Lean MEA @ Bottom Stripper kg/h 477477,4 471609,2 473765,7859
CO2 captured kg/h 37568,7 37504,7 37389,0
Heat Duty Stripper MW - 48 48
Table 4.10: Difference between the literature values and those calculated by the two models.
Δ Values
ENRTL-RK ELECNRTL CO2 in Clean Gas kg/h 67,2 -42,6
Clean Gas Temperature °C -4,2 -4,0 Reboiler Temperature °C -1,4 -1,8 Lean MEA @ Bottom Stripper kg/h 5868,2 3711,6
CO2 captured kg/h 64,0 179,7
Table 4.11: Relative errors with respect to the literature.
Relative Error %
ENRTL-RK ELECNRTL
CO2 in Flue Gas - -
CO2 in Clean Gas 0,80 0,51
Clean Gas Temperature 3,46 3,29
Reboiler Temperature 0,58 0,75
Lean MEA @ Bottom Stripper 0,61 0,39
CO2 captured 0,09 0,24
Heat Duty Stripper - -
Error average 1,11 1,04
Going then to analyze the relative errors, it is found that both models find very similar results and this shows that both work well however if you go to calculate the average of the errors, the ELECNRTL model wins slightly compared to the ENRTL-RK model. Except for the error on the reboiler temperature calculation, all other values are less than 1%. In the analysis of the real case, however, the model ENRTL-RK will be considered as it has been developed with the rate-based model and consequently it is possible to build an optimization model for the dimensioning of the stripper and of the absorber. In fact, the rate-based model is much more precise and reflects reality because it considers the reactions that occur in the columns.
The rate-based ELECNRTL model present in the Apen Plus library was not used as the last revision dates to several years ago compared to the ENRTL-RK model and consequently is less updated and less precise than ENRTL-RK model. Recall that in the equilibrium model, the vapor and liquid phases are assumed to be in thermal equilibrium and the Murphree vapor phase efficiency is used to describe the deviation from equilibrium. The equilibrium model
Chapter 4
is relatively simple, but the accuracy of the model depends on the Murphree efficiency prediction. The rate-based model, or in other words non-equilibrium model, is accurate, but more complicated than equilibrium model and it is also difficult to converge; it eliminates the necessity of using efficiencies and is capable of predicting the actual performance of the process.
In equilibrium model, component material balance equations, the equations of phase equilibrium, summation equations, and heat balance for each stage are solved to give product distributions, flow rates, temperatures, and so on. In many cases, the Murphree efficiencies used in equilibrium models are calculated through a rate-based process. In case Murphree efficiency is correctly predicted for the equilibrium model the results are like those found in the rate-based model.
Results and discussion
In this chapter I will use the optimized values found in Chapter 3 and later they will be implemented in Aspen. Then they will be commented on. Subsequently, two technologies will be studied and evaluated to produce the heat necessary for the stripper. The data on the composition of the flue gases derive from a European research project of which Professor Bonalumi is a part and since it is still not completed, it is not possible to go into the details of the analysed refinery. Consequently, I will limit myself to using only the composition of the flue gases of a chimney already described above and then apply the ENRTL-RK model in rate-based mode present in the Aspen Plus template library. This model has been validated using data present in the literature and described previously. Once the plant is merged, it will be possible to consider how the captured CO2 affects the total emissions of the refinery.
The Table 5.1 shows the optimum parameters deriving from the sizing of the absorber and the stripper with the method described in Chapter 3. In the literature there are also other methods and consequently the results may not be the same. In addition, the chosen parameters come into play, such as the inlet temperature in the stripper, the capture efficiency of the system and the choice of the height of the stripper according to the heat required by the reboiler.
Table 5.1: Design parameters optimized for the two columns in the real case simulation.
Units Absorber Stripper
Top temperature °C 53.2 113.5
Bottom temperature °C 31.4 121.2
Condenser temperature °C - 35
Pressure bar 1 1.8
Height m 15 5
Diameter m 3 2.05
Reboiler Duty kW - 16395
Once the optimization of the two columns is finished, it is possible to merge them and build the complete process in Aspen. The system scheme is shown in the Figure 5.1. You can also deactivate the option to calculate the optimal diameter.
The performances of the system are shown in the Table 5.2.
Table 5.2: Apparent flow rates coming from the operation of the plant in optimized conditions.
Stream Name Units CLNGAS FLUEGAS MEA CO2
Apparent component mass flow rate
MEA kg/hr 43,90 0 24593,50 0
H2O kg/hr 8374,87 1615,32 56513,05 94,54
CO2 kg/hr 600,05 7660,04 1504,97 7044,78
N2 kg/hr 66483,92 66488,64 0,00 4,72
Apparent component molar flow rate
MEA kmol/hr 0,72 0 402,62 0
H2O kmol/hr 464,88 89,66 3136,95 5,25
CO2 kmol/hr 13,63 174,05 34,20 160,07
N2 kmol/hr 2373,28 2373,45 0 0,17
The carbon capture efficiency considering all the process is 92.1%. The fact that the efficiency is greater than 90.6% calculated in the sizing of the absorber is because an apparent mass flow rate of CO2 equal to 15.2 kg/h comes out in the purge stream. This is necessary to insert a purge flow rate equal to 1% to avoid problems of encrustations and change the degraded solvent. The heat required by the stripper is equal to 16395 kW. During the process it will be necessary to add MEA and water as it will leave the plant with the clean
Figure 5.1: Flowsheet of the plant on Aspen Plus.
Chapter 5
gases and the captured CO2 flow rate. The calculator is used to calculate the make-up flow rates of CO2 and water. The water and MEA make-up flow rate with the purge flow rate are shown in the Table 5.3.
Table 5.3: Flow rates of make-up and purge calculated through the Aspen Plus calculator.
Stream Name Units MKH2O MKMEA PURGE
Apparent component mass flow rate
MEA kg/hr 0 289,40 245,50
H2O kg/hr 7350,54 0 496,59
CO2 kg/hr 0 0 15,20
The CO2 capture plant can capture 7044.78 kg/h of carbon dioxide. Let's now see how it impacts on the global emissions of the refinery. Since the refinery is very large, and since there are many stacks, it is not possible to capture all the flows and it is also difficult to merge more than one. As a consequence it is expected that this quantity of CO2 is not enough to greatly reduce pollution, however it is still a starting point since the analysed chimney is the one that has the greatest flow rate and has the highest CO2 emissions.
It becomes interesting now to go and see how much the reduction of CO2 emissions affects the total refinery. The Table 5.4 shows the emissions from the main stacks of the refinery.
Table 5.4: CO2 emissions from the main stacks of the refinery.
Emission
A2-4 Stabiliser Reboiler Heater F-204 9198 10394 8,1 1264 A2-5 Powerformer Preheater Furnace F201X 17785 20097 4,2 1146 A2-6 Powerformer Reheat Furnace F-2028x/CX 6394 7225 0,2 23 A2-7 Powerformer Reheat Furnace F202AN 9369 10587 8 1272 A2-8 Naphtha Hydrofiner/Debustaniser Reboiler
Furnace F-206-207
24549 27740 10,5 4317 A2-10 Hydrofiner Feed/Hot Oil Heater F-801/F802 19278 21784 7,5 2461 A2-11 Distillate Hydrofiner Furnace F-901 3780 4271 4,5 294
The chimney studied in this work is the A2-1 which corresponds to the emissions of atmospheric distillation. In fact, 33% of the total emissions derive from this process. We have seen in the previous chapters how this refinery process is one of the most polluting as
it treats all the crude oil that arrives at the refinery. In the Figure 5.2 it is possible to see how much it affects the total. Total CO2 emissions amounted to 23595 kg/h which correspond to 33% of the total.
With the use of carbon capture, the carbon dioxide emissions of the refinery become equal to 16535 kg/h; 7060 kg/h of CO2 were removed which was equivalent to 30% of total emissions. The capture process takes place with an efficiency of over 90%. Observing the values in the Table 5.4, not all the chimneys have a CO2 flow rate capable of justifying the installation of a capture system.
11%
33%
5% 11%
5%
0%
5%
18%
11%
1%
A1-3 A2-1 A2-2 A2-4 A2-5 A2-6 A2-7 A2-8 A2-10 A2-11 Figure 5.2: Percentage of CO2 emissions out of the total. Based on mass flow rate.
0 5000 10000 15000 20000 25000
Without CCS With CCS
Figure 5.3: CO2 emissions in kg/h before and after using the CCS in the refinery.
Chapter 5
The other chimney where it would be feasible is the A2-8 where emissions are 18% of the total. The Figure 5.3 shows the emissions before and after CCS installation of the refinery.
The difficulty in capturing the remaining CO2 is due to costs that do not justify the feasibility of the plant. It could be interesting to take advantage of the proximity of some structures to convey more flow rates in a single system. This is only possible if the refinery area allows it; in fact, connection pipes must be added in which the flow rate flows and compressors must also be added to guarantee pressure to overcome pressure drops. It is an expensive process that often does not justify the adoption of CCS.