Figure 1. Schematic flow diagram of FCCU
Fluid Catalytic Cracking Unit (FCCU) is an important processing unit in an oil refinery. Fluid catalytic cracking is a process used to convert heavy petroleum products to light products such as gasoline, light fuel oil, and petroleum gas. In the fluid catalytic cracking reactor, heavy gas oil is cracked into more valuable lighter hydrocarbon products. The reactor input (the gas oil fed to the reactor) is a mixture of hydrocarbons that makes the reaction kinetics very complicated due to the involved reactions. The process is highly nonlinear and multi variable with severe interactions. For the simulated dynamic model of FCCU plugged with yield model, sensitivity studies has been carried out to study the effects of independent variables such as feed preheat temperature, feed flow rate and air flow rate in maximizing gasoline yield. With the optimized model, the closed loop studies have been carried out for reactor and regenerator temperature control of FCCU with minimum overshoot as the performance criteria.
Fluidized Catalytic Cracking Unit (FCCU) of gasoil is one of the most important processes in the petroleum refineries. As shown in Figure 1, FCCU receives multiple feeds consisting of high boiling point components from several other refinery process units and cracks these streams into lighter and more valuable components [6]. After further processing, the FCCU product streams are blended from other refinery units to produce a number of products, e.g. distillate and various grades of gasoline. Economic operation of FCCUs (a large refinery may have more than one) plays an important role in the overall economic performance of the refinery. Gasoline yield is an important product from FCCU process since it has a very high market value. Gasoline yield in FCCU depends on various parameters like reactor temperature, regenerator temperature, air flow into regenerator, catalyst circulation rates, etc. Therefore, sensitivity analysis is needed to determine the variable which has strong influence on gasoline yield.
Figure 1. Schematic flow diagram of FCCU
The model captures the major dynamic effects that occur in a real FCCU plant. It is multivariable, strongly interacting and highly non-linear. The non-linear model was developed for the following sub models and integrated.
The simulated model captures the major dynamic effects that occur in a real FCCU plant. The complete FCCU model was simulated in MATLAB Simulink and is validated [1].
Many complex reactions occur during the FCC process and the product consists of a mixture of many compounds. The description of complex mixtures by lumping large number of chemical compounds into smaller groups of pseudo-components has been widely used by researchers to provide number of kinetic equations. The simulated 4- lump model plugged with FCCU model is similar to the 3- lump model of Weekman, the main difference being that coke is independently considered as a lump. Other lumps are feed, gasoline and gas.
It is assumed that the gas oil is cracked into the most desired gasoline, and the by-products of gas and coke. The reaction scheme of 4- lump yield model is shown in Figure 2 [5,7,8,9]. Since the FCC reactor is operating at high temperature, the secondary cracking reaction occurs for gasoline to form coke and gas. There is no inter-reaction between coke and gas[2].
For the simulated model, the sensitivity studies have been carried out i.e., the effects of the following independent variables on gasoline yield have been analyzed [3].
Figure 2. Reaction scheme of 4 lump yield model
To carry out these sensitivity studies, these independent variables were operated between -30% and +30% of their nominal values. The nominal values of temperature of fresh feed entering the furnace is 461˚F, flow of fresh feed entering the reactor riser is 126 lb/s and the stack gas valve position is 0.64.
The values of those variables at which the maximum yield has been obtained were fixed as the optimum values. For the optimized model, the closed loop studies have been carried out with PI controller at the regulatory level.
Feed preheat temperature plays an important role in controlling the temperature in the riser reactor and hence the cracking reactions.
When the temperature of the fresh feed entering the furnace increases, the gasoline yield gets decreased, keeping feed flow into reactor riser (126 lb/s) constant. When the flow of fresh feed to the reactor riser increases, the gasoline yield gets increased because increase in flow rate will lead to more amount of feed taking part in the reaction inside the reactor riser unit, keeping the temperature of fresh feed entering the furnace (461˚F) constant.
Increase in the stack gas valve position will lead to a decrease in pressure inside the regenerator, which demands more amount of air flow into regenerator, keeping the feed flow rate (126 lb/s) and feed temperature (461˚F) constant. Increase in air flow rate into the regenerator will cause an increase in the regenerated catalyst flow rate due to which the reactor temperature gets increased. When the reactor temperature crosses its optimum temperature, it will lead to decrease in the gasoline yield.
It has been observed from the sensitivity analysis that optimum yield (Figure 3) is obtained at the temperature of fresh feed entering the furnace = 401.2˚F, the flow of fresh feed to reactor riser = 100.3 lb/s and the valve position at 0.448 (air flow into regenerator = 2.1236 moles/s).
The following Table 1 shows the operating ranges of the independent variables used for the study of the sensitivity effect on the gasoline yield and their impact on it.
The Figures 4 and 5 show the responses of the yield and their variations before sensitivity analysis and after sensitivity analysis.
Figure 3. Optimum yield based on the sensitivity analysis
Table 1. Effect Of Independent Variables On The Gasoline Yield
Figure 4. Yield response before sensitivity analysis
Table 2 lists the response of the gasoline and coke yield before and after sensitivity studies.
It has been observed from Table 2 that, there is a considerable increase in gasoline yield and decrease in the coke formation after fixing the values of independent variables for maximizing gasoline in the simulated model.
Each manipulated variable depends on only a single controlled variable i.e., a set of conventional feedback controllers. In most of the chemical industries, the processes are basically MIMO systems. The controlled variables have been selected to provide, through control, a safe and economic operation. The reactor temperature and regenerator temperature are selected as a controlled variable. The manipulated variables are flow of regenerated catalyst and flow rate of air[4].
Figure 5. Yield response after sensitivity analysis
Table 2. Yield Of Gasoline And Coke Before/After Sensitivity Analysis
The control structure selection was done based on Relative Gain Array (RGA) analysis and is shown in Figure 6.
The input and output relationship are given as The above transfer functions were found out by using the process reaction curve method.
The steady (gain) model is expressed as,
Figure 6. Control structure based on RGA analysis
Thus, the Relative Gain Array for a 2 × 2 system can be expressed as given below
Pair the controlled and manipulated variables so that, the corresponding relative gains are positive and are as close to one as possible. From this RGA matrix, we can conclude that pairing of manipulated variables with control variables (i.e., flow rate of regenerated catalyst is used to control reactor temperature and flow rate of air is used to control regenerator temperature) is best.
Figure 7 shows the schematic diagram of the closed loop system with PI controller.
The reactor temperature must be maintained at a certain level to provide a desired maximum conversion of feed oil. The regenerator temperature must be maintained at a certain level in order to allow a stable coke from the catalyst. Permanent catalyst deactivation is produced by exceeding the high temperature limit. The proportional integral (PI) controller is used at regulatory level. The specified reactor temperature is maintained by using a PI controller to adjust the flow of regenerated catalyst. Also the specified regenerator temperature is maintained by using a PI controller to adjust the air flow into the regenerator.
Before performing closed loop studies on any dynamic model, the best performance criteria for the specific control variable need to be studied. As per the conversation with experts working in CPCL (Chennai Petroleum Corporation Limited), minimum overshoot was chosen as the performance criteria for the reactor and regenerator temperature control of FCC unit.
Figure 7. Schematic diagram of closed loop system with PI controller
With minimum overshoot as the performance criteria, the following tuning rules listed in Table 3 have been used to tune the PI controller [10].
where Kc = Proportional gain
τi = Integral time constant
λ = Adjustable parameter
Ku = Ultimate gain of the sustained oscillations
Tu = Ultimate time period of the sustained oscillations
The control signal from the tuned PI controller has been applied to the whole simulated FCCU model and not to the transfer function that has been obtained through process reaction curve method. The transfer function was used only to establish the input-output relationship between the controlled variable and the manipulated variable.
The reactor temperature must be maintained at a certain level to provide a desired maximum conversion of feed oil. The PI tuning parameters for the reactor temperature control process for different PI tuning rules listed in Table 4.
The reactor temperature was operated for three different set points (990˚F, 993˚F and 995˚F) and their responses for the different PI tuning rules have been shown below.
From Figures 8, 9 and 10, it can be inferred that the same tuning parameters as given in Table 4 can be applied to any operating point. Also it shows that the direct synthesis method is having minimum overshoot but having a sluggish response. So it is better to choose the Atkinson and Davey PI tuning rule for the controller as it gives the next best response with minimum overshoot and quicker settling time.
Table 3. Pi Tuning Rules And The Parameters
Table 4. Pi Tuning Parameters For Reactor Temperature Control
Figure 8. Control of reactor temperature for different PI tuning rules (990˚F set point)
Figure 9. Control of reactor temperature for different PI tuning rules (993˚F set point)
Figure 10. Control of reactor temperature for different PI tuning rules (995˚F set point)
Figure 11 shows the reactor temperature control for different operating points using Atkinson and Davey PI tuning rule. After the steady state was reached, disturbance has been given to the effective coke factor (Ѱf) from 1 to 1.5 at 7000s and from 1 to 0.5 at 12000s and the controller performance was studied and is given in Figure 12.
Figure 11. Reactor temperature control for different operating points using Atkinson and Davey PI tuning rule
Figure 12. Control of reactor temperature with disturbance in effective coke factor
The PI tuning parameters for the regenerator temperature control process for different PI tuning rules is listed in the following Table 5.
The regenerator temperature was operated for three different set points (1265˚F, 1275˚F and 1285˚F) and their responses for the different PI tuning rules have been shown below.
From Figures 13, 14 and 15, it can be inferred that the same tuning parameters as given in Table 5 can be applied to any operating point. Also, it shows that the direct synthesis method is having minimum overshoot but having a sluggish response. So, the Atkinson and Davey PI tuning rule is better to choose for the controller as it gives the next best response with minimum overshoot and quicker settling time.
Table 5. Pi Tuning Parameters For Regenerator Temperature Control
Figure 13. Control of regenerator temperature for different PI tuning rules (1265˚F set point)
Figure 14. Control of regenerator temperature for different PI tuning rules (1275˚F set point)
Figure. 15 Control of regenerator temperature for different PI tuning rules (1285˚F set point)
Figure 16 shows the regenerator temperature control for different operating points using Atkinson and Davey PI tuning rule. Again, when the steady state was reached, disturbance has been given to the effective coke factor (Ѱ ) from 1 to 1.5 at 7000s and from 1 to 0.5 at 12000s and the controller performance was studied and is shown in Figure 17.
Figure 16. Regenerator temperature control for different operating points using Atkinson and Davey PI tuning rule
Figure 17. Control of regenerator temperature with disturbance in effective coke factor
Table 6 describes the variation of gasoline yield and coke yield for different PI control tuning methods when the reactor temperature is controlled at 993˚F and regenerator temperature is controlled at 1275˚F.
The “process” is represented by a simulated dynamic model of a model IV FCC unit which is combined with a steady-state yield model for the FCC reactor. The dynamic model will calculate the time-varying states of the FCC unit at any point in time, while the yield model uses the reactor conditions to calculate the conversion and the distribution of products.
Table 6. Variation Of Gasoline Yield With Different PI Tuning Rules
The sensitivity analysis of certain independent variables performed results show that the gasoline yield can be increased by decreasing the temperature of the fresh feed entering the furnace and air flow into regenerator (by minimizing the valve opening), when the flow of fresh feed to the reactor riser is increased.
From the closed loop performance study, it is concluded that Atkinson and Davey PI tuning rule proves to be the best PI tuning rule when the performance criteria of minimum overshoot was considered.
The first author thanks “Mr.S.Manikandan, CPCL Official, Chennai” and “Mr.P.Kanagasabapathy, Professor Emeritus, MIT, Chennai” for their constant support throughout this work.