Application of the hottest soft sensor in inferrin

  • Detail

The application of soft sensing in inferring and controlling the dry point of crude gasoline in fractionator

catalytic cracking unit (FCCU) has a very important economic position in the secondary processing of oil and how it can not be cleaned. Its operation condition is directly related to the recovery rate of light oil products in oil refining, thus affecting the economic benefits of the whole refinery. At present, a major obstacle restricting its quality control level is that the hard instruments for detecting product quality are expensive, complex maintenance, poor accuracy, and large time delay, which cannot provide effective feedback signals for closed-loop control. This paper introduces the soft sensing model establishment, correction and closed-loop control of crude gasoline dry point in FCCU fractionator. The actual operation effect is good and can fully meet the quality requirements

1 problem presentation

fccu fractionator crude gasoline dry point, the conventional operation is to control the distillation temperature to indirectly control the product quality, and take a manual sampling and analysis test value for 4 ~ 8h as a reference for offline adjustment. Once the distribution of cracking products in the reaction oil and steam changes greatly, the control will not be timely and the products will be unqualified. Therefore, in order to ensure product quality, operators often adopt too conservative operation. This reduces production and increases energy consumption

through the analysis of mechanism and field collected data, the relationship between the dry point of crude gasoline and auxiliary variables is obtained:

y (k) = f [X1 (k), X2 (k), X3 (k), X4 (k)]

where: Y (k) is the dry point of crude gasoline, ℃; X1 (k) is the top temperature of fractionator, ℃; X2 (k) is the partial pressure of crude gasoline at the top of fractionator, kPa; X3 (k) is the internal reflux ratio; X4 (k) is the liquid level of oil-gas separator, m; F [] is the function relationship to be evaluated

x2 = p.xg

xg = (R2/mg)/(R2/Mg + R3/M gas + RW/18)

x3 = R1/r2

r1 = (R4 + its lighter weight R5). [1 + CP. (T1-T2)/λ]

where: P is the overhead pressure of fractionator, kPa; XG is the molar component of crude gasoline at the top of fractionator; R2 is crude gasoline flow, t/h; R3 is rich gas flow, t/h; RW is acid water flow, t/h; MG is the average molecular weight of crude gasoline; MGAs is the average molecular weight of rich gas; R1 is the internal return flow, t/h; R4 is the cold reflux flow of crude gasoline, t/h; R5 is the reflux flow at the top of the tower, t/h; CP is the liquid phase specific heat of crude gasoline, kcal/(kg. ℃); λ Is the latent heat of crude gasoline, kcal/kg; T2 is the temperature of top reflux return tower, ℃

2 calibration soft sensing model

oil in the oil refining industry is an extremely complex mixture, and the composition after processing is more complex. The process design and calculation errors based on charts and empirical formulas are large, and the quality indicators of petroleum products are mostly analyzed under specific equipment and specific conditions, so it is difficult to calculate quantitatively by pure mechanism method. Therefore, we use regression method to establish the empirical model of crude gasoline dry point of FCCU unit based on the training data obtained after processing, and predict the calibration data

= 0.6970 (x1 - 113) + 0.0077 (x2 - 469.87) -

1.0666 (x3 - 4.16) + 0.0143 (x4 - 305.53) + 182.1933

model fitting MSE = 3.4744; The model predicts MSE = 1.4597

with the change of operating conditions and other factors, the index reflecting product quality - the dry point of crude gasoline also changes. Figure 1 shows the change curve of the monthly average dry point of crude gasoline in FCCU unit in a certain year (the unit was shut down for maintenance in July and August), and the monthly average dry point of crude gasoline changed by more than 20 ℃ in this year. Obviously, no matter how accurate the original model is, without proper correction, the estimated value of the model will deviate from the actual working conditions. Here, the short-term correction method is adopted

[IMG f [/img

after determining the soft measurement model and correction method of crude gasoline dry point, the soft measurement of crude gasoline dry point is realized with process basic on the DCS of FCCU. When using the soft measurement model to predict, the measured variables are filtered by the 30 min moving average method with the increasing international influence of China's plastic machinery industry.

fccu crude gasoline dry point soft measurement model has been running since December, and the operator inputs the test value every shift. Run for a period The calculation results of the time model are shown in Table 1, which proves that the soft sensing model basically meets the accuracy required by the process

Table 1 Comparison between calculated values of FCCU soft sensing model and laboratory values

number of absolute values of deviation (50) frequency (%)

<1 ℃

<2 ℃

<3 ℃

> 30 ℃



1 60




Quality Control of 3 fractionator

the basic control objectives of fractionator are product quality, product recovery rate and energy consumption. Since the 1970s, it has successively developed decoupling control, self-tuning control, state feedback and adaptive feedforward control of side line product quality, intelligent coordinated control of multi side line product quality and yield, inference control and other advanced control schemes, and achieved obvious economic benefits. At present, various software packages have been formed for the advanced control strategy of fractionator at home and abroad

3.1 basic method of crude gasoline dry point control

the basic schemes of crude gasoline dry point control on the top of fractionator include: tower top temperature control and closed-loop control based on component analyzer

at present, the scheme of adjusting the top temperature of the main fractionator is widely used. Based on the close relationship between the concentration change of the key components of the top product and the top temperature change, the top temperature of the tower is maintained near the given value through PID control. This scheme has the advantages of low control cost, fast response and less failure. However, its control performance is also limited by many factors: the influence of non key components on top temperature change, the influence of tower top pressure on top temperature change, system disturbance, etc. In addition to the lack of real-time crude gasoline dry point data, due to the slow temperature response, the system control quality is poor, so it is difficult to achieve quality edge control

based on the closed-loop control of the composition analyzer, the analyzer provides the measured value of the dry point of crude gasoline to achieve quality control. However, the component analyzer is expensive, complex to maintain and adjust, large measurement lag, low reliability, and the control quality is difficult to meet the process requirements, so it is not widely used

the cascade control scheme composed of composition and temperature control integrates the advantages of temperature control and composition control. The basic control is realized by the temperature controller, and the set value of the temperature controller is slowly adjusted by the output of the component analyzer controller to prevent product quality drift. This control scheme allows the analyzer to have a certain lag, and when the analyzer fails, the temperature controller can still maintain automatic control. Its defect is still composition

Copyright © 2011 JIN SHI