CO Processing Simulation

Next: Conclusions Up: Application and Results Previous: Experimental Four-Tank Process

CO $_{2}\protect$ Processing Simulation

The CO₂ case study simulated here is taken from the work of [4]. The simulation considers mass and energy balances, vapor-liquid equilibrium data, chemical reactions, and mass transfer relations. Monoetanolamine (MEA) is used in the system to react with the CO₂ in the with are considered in the simulation model. The controlled variable is the CO₂ concentration in the product vapor stream leaving the condenser. The manipulated variable is the reboiler steam flow rate. Disturbances include the feed CO₂concentration, the feed CO₂ flow rate, the steam quality to the reboiler, the condenser duty, and variations in the MEA concentration.

**Figure:** CO $_{2}\protect$ Simulation Schematic
$\resizebox*{3in}{!}{\includegraphics{abs.ps}}$

The following assumptions are made in the modeling of the absorber-desorber system.

The absorbtion column is adiabatic
The absorbtion column has no pressure drop
Plug flow gas phase throughout the the columns
Well mixed liquid phase on each tray
The gas phase is pseudo steady state
The gas phase is neglected in energy balance
There are no radical temperature gradients exist throughout the column
Liquid and gas phase temperatures are equal
Constant physical properties for the columns
The MEA in desorption column is non-volatile, no MEA in product stream
No heat loss in the columns
No accumulation of mass on desorption trays
The condenser condensate contains no CO₂
Condenser model is assumed steady state

Using the Aspen simulation model, linear disturbance models were developed for the six different faults. A sample time of 60 seconds was used, and 60 coefficients were use in the step response models. Table 3 shows the parameters used in the control and estimation problems. In figure 12, the controlled and manipulated variables for a step disturbance in the steam flow rate are shown for two cases. In the first case, the estimated disturbance is not used by the control algorithm. In the second case, the estimated disturbance is used in the control move calculation. Use of the disturbance estimate clearly improves control performance in the simulation environment.

Table: Control and estimation algorithm parameters for CO $_{2}\protect$ case study

m	2 samples	p	40 samples
$\Gamma _{y}$	[1]	$\Gamma _{u}$	[0.1]
m₁	10*[11111111]	m₂	0.8*[11111]
m₃	0.2*[111111]	h	30 samples

Figure 13 shows the residuals for the eight different process measurements. The measurements selected include the absorber top tray mole fraction, the absorber top tray gas leaving mole fraction, the absorber bottom tray gas mole fraction, the reboiler temperature, the desorber top tray mole fraction, the top tray desorber temperature, the desorber bottom tray mole fraction, and the desorber bottom tray temperature. As stated, a linear model for the effect of the manipulated variable (steam flow rate) on the measured outputs is compared to the actual process measurements for calculation of process residuals.

**Figure:** Controlled and manipulated variables for CO $_{2}\protect$ absorber model, steam flow rate disturbance from t=1000 to t=5000 sec.
$\resizebox*{4in}{!}{\includegraphics{co2inout.ps}}$

Figure 14 shows a typical horizon estimate for the system. In this case, there are small estimation errors at the onset of the disturbance. This can be attributed to model error due to nonlinearitites and initiall offset in the process residuals.

**Figure 13:** Process variables for steam flow rate disturbance
$\resizebox*{4in}{!}{\includegraphics{co2residuals.ps}}$

**Figure 14:** Horizon estimates for steam flow rate disturbance
$\resizebox*{4in}{!}{\includegraphics{co2horizonfig.ps}}$

Next: Conclusions Up: Application and Results Previous: Experimental Four-Tank Process

Edward Price Gatzke
1999-10-27