CC BY
19
OPTIMIZATION AND ADVANCED CONTROL FOR THERMAL
CRACKING PROCESSES Pulotova Moxira Raxmatilloyena, master (e-mail: [email protected]) Bukhara engineering technological institute, city Bukhara, Uzbekistan Bozorov Prim Raimovich, doss. Informational communication system in management of technological processes Bukhara engineering technological institute, city Bukhara, Uzbekistan
Advanced control is a good alternative solution to obtain the desired product yield and still maintain the safety conditions, without changing the system's configuration or replacing the existent components with new ones. A mathematical model of a thermal cracking coil was developed, based on free radical mechanism and some basic assumptions. In order to predict the coil's behavior (temperature, product yields and pressure), a simulator for the dynamic process was used. This paper represents a preliminary dynamic behavior study for implementing advanced process control and optimization.
Keywords: thermal cracking, coking plant, ethylene furnace, advanced control.
Расширенный контроль является хорошим альтернативным решением для получения желаемого выхода продукта и по-прежнему поддерживать условия безопасности, без изменения конфигурации системы или замены существующих компонентов новыми. Разработана математическая модель тепловой катушки крекинга, на основе свободного радикального механизма и некоторых основных допущений. Для того чтобы предсказать поведение катушки (температура, выход продукта и давление), был использован тренажер для динамического процесса. Эта статья представляет собой предварительное исследование динамического поведения для осуществления расширенного управления и оптимизации процессов.
Ключевые слова: термический крекинг, коксовый завод, этилена печи, расширенный контроль.
The highly used thermal cracking furnaces in today's industry are the cracking furnace from the petrochemical domain (cracking of naphtha, ethane, etc.) and the furnace from the coking plant.
The pyrolysis furnaces can be used for any type of raw material; the only condition is the boiling point that must be under 600oC. The furnace represents the centre of the cracking plant and the consumption of energy is also concentrated in this area. The desired products from this reaction are the light olefins, like: ethylene, propylene and butadiene.
Ethylene is industrially obtained through thermal cracking of hydrocarbons. In order to obtain a larger amount of ethylene and decrease the energy and material loss, the process parameters must be held between certain limits according to C.M. Tham as follows:
• Residence time inside the coil, between 0.08 and 0.25 [s] - as short as possible. In order to reduce the residence time, tubes' diameters are reduced; the fabrication materials improved and the burners tend to be much more efficient.
• Dilution steam measured as the ratio S/Hc (steam/hydrocarbon), between 0.3 and 0.6 - high quantity of steam. The dilution steam is introduced in the process to reduce the coke production and to decrease the gas pressure (minimize the undesired secondary reactions).
coke
Gin
Oout
E>
dz
Figure 1. Differential element
Figure 2. Process simulator
• Reaction pressure, between 175 and 240 kPa - as low as possible. Coil output pressure is indirectly controlled by the aspiration pressure of the gas compressor, placed downstream.
• Reaction temperature - at least 900oC - as high as possible. The pyrolysis is an endothermic reaction so a high temperature generates smaller hydrocarbons molecules. A low temperature favors the production of coke and shortens the tubes' "life".
The other process represented by the coking plant is able to upgrade high or low quality vacuum residue, in order to obtain different types of coke and small petroleum fractions.
The main concern is the feedstock composition (the amount of sulphur and metals) when the goal is to obtain high quality coke (needle coke), but when the aim is to produce lighter, more value-added products, the process temperatures and pressure play the most important role. In order to obtain a larger amount of desired products and also decrease the energy and material losses, the process parameters must be held between certain limits.
The free radicals mechanism represents a universal accepted explanation for the hydrocarbons pyrolysis. Once the conversion and the olefin concentration increases, the secondary reactions become more frequent. A small segment of pipe, of infinitesimal volume is represented in Fig. 1.
According to the free radicals mechanism (S. Raseev, 2003), the considered reactions are:
kl
Initiation: C2H6^2CH3 (I)
k2
C2H6 + CH3 ^ CH4 + C2H5 (II)
k3
Propagation: C2HS ^ C2H4 + H (III)
fc4
C2H6 + H^H2 + C2H5 (IV)
fc5
Interruption: C2H5+H ^ C2H6 (V)
fc6
2C2H5 + CH3 ^ C4H10 (VI)
fc7
CH3 + C2H5 ^ C3H8 (VII)
fc8
2CH3 ^ C2H6 (VIII)
fc9
CH3+H ^CH4 (IX)
The mathematical description of a one-dimensional plug-flow reactor tube is present
below, with the following assumptions: laminar regime, axial dispersion is neglected,
ideal gas behavior and inert steam diluents.
Material balance for component j:
dw; Wj-W; 0 ^
-^-v-^xE^xso.j) (1)
where:
kg
Wj
Ikgi
Wjo
- Mass fraction of component j
kg'
- Mass fraction of ethane at the beginning
.kg
z — [m] - Length along coil t — [5] - Scanning rate
S(i,j) - Stoichiometric constant of component j, in reaction i v — [m/s] - Fluid velocity In this
r - reaction rate, evolves with temperature according to an Arrhenius equation:
ri=Aixe~Ei/RxT (2)
where:
Tj — [kmol/ m3 •s] - Reaction rate
And S (i,j) represents the multiplication between the molar concentration of the
involved substances. Energy balance:
dTg _ v(Tg-Tg 0) _ [(Tg-Tw)xAxktg+YliriHi + gx(T^-T■*)] (3)
dt dz pxCp ^ '
dTw Axktx(Tg-Tw)
dt pxCp
where:
Cp — [kcal/kg • K] - Process gas specific heat
Hi — [kcal/kmole] - Heat of reaction
A — [m2] - Transfer tube area
Tw — [K] - Refractory wall temperature
Tw q — [K] - Initial refractory temperature
Tg - Flue gas temperature
Tg 0 — [K] - Initial flue gas temperature
ktg — [W/m • K] - Gas thermal conductivity
kt — [W/m • K] - Tube thermal conductivity
o— [W/m4 •K4] - Boltzmann coefficient referring to radiant energy Mechanical energy balance:
(4)
^ = (5)
dt dt p2xgyD J v '
where:
P — [kgf / m2 ] - Pressure
P ~ [kg/ - Process gas density
G3 — [kg • m/kgf • s2 ] - Dimensional constant
The result is a time and space variation of all parameters, an isobaric process which
depends on the inlet mass flow's variation and considering the inlet gas temperature as a disturbance.
For the entire coil, the implementation was made using ode15s function from Matlab and matrices, solving nine materials balance differential equations: C2H6, CH3, CH4, C2H5, C2H4, H, H2, C4H10, C3HQ, and two energy balance differential equations: gas and wall temperature.
The inlet flow represents 95% ethane. The ethane concentration decreases in time, along with radicals and final products formation. The radicals' concentrations increase rapidly and afterwards they recombine in final products. In comparison with the other products the concentration of ethylene is high. Due to the fact that the cracking reactions are endothermic, one can observe in Fig. 4, the consumption of energy by decreasing the gas temperature. In time, the gas temperature stabilizes. The wall temperature remains constant in time. The variability is insignificant.
Nowadays the real challenge is to implement the dynamic model in a real time control environment, which is usually not designed for rigorous simulation. The best way to solve differential equations in these cases is to use Euler's Method.
A reliable dynamic model for this type of coil could be used for a future advanced control project, based on the following techniques: feed-forward control, state estimators and predictive model controllers.
The operating objectives should be: to maintain the outlet temperature constant, operate within constraints, minimizing the excess air and fuel consumption. The MPC variables could be the one presented in table, for a small process that needs optimal control, defined constraints and certain optimal set points. These rules could be applied also in the case of a coking plant when the actual temperature control point is at the heater outlet. The chemical reactions are en-dothermic and so the coke drum (following the heater) temperature will be lower. If the temperature is too low, the coke will be too soft and the other products' specifications will not be met. As I have mentioned earlier this paper represents a dynamic regime study that can be followed by the implementation of an advanced process control, like the one presented above: MPC. This paper could be the groundwork for the development of a simulator for the entire reactor, with unlimited number of cracking tubes, which could be monitored, and its outlet variables controlled. By using this simulator, a steady state optimization can be tested and even a predictive model can be generated in order to build an MPC preliminary structure, for test purposes and operator trainings.
Figure 3. Advanced control system's structure
This implemented model can also be the basis when one needs to build an equivalent model in a more real time control environment, much harder to follow.
Refernces
1. A. Niaei, J. Towfighi, et al., 2004, "The combined simulation of heat transfer and py-rolysis reactions in industrial cracking furnace" Applied Thermal Engineering 24 (2004) pp. 2251-2265.
2. Program: MATLAB R 2008A, © 1984-2008 The Math Works, Inc. Simulink. Subsystem. Delete Contents, Simulink. Block Diagram. Copy contents to Subsystem, Simulink. Subsystem. Convert to model Reference.
3. B.G. Houk, et al., 1995, "Model predictive control of a delayed coking unit [DMC]: Development Issues", Chem. Technol. Eur., vol.2, pp. 20-24.
