(187d) Multiscale Dynamics System Identification of Time Series of Riser Reactor Temperature in FCC Process Based on Hilbert-Huang Transform

Authors: 
Cao, D., China University of Petroleum, Beijing
Wu, Y., China University of Petroleum
Xu, C., China University of Petroleum, Beijing
Gao, J., China University of Petroleum
Lan, X., State Key Laboratory of Heavy Oil Processing, China University of Petroleum
Fluid catalytic cracking (FCC) is a major process for converting heavy oil into high octane gasoline, diesel and liquefied petroleum gas (LPG) in the oil refinery. This process exhibits strong nonlinear, high dimensions and multiscale features to which enough attention should be paid when describing its complex dynamics. Due to the diversity of physicochemical phenomena occurred in FCC process, an important task in industrial practice is to express the nonlinear effects and multiscale interaction among variables in FCC reactor-regenerator model. In recent years, many refineries propose the aim of smart factory and establish big data platform which provide a chance to use data more efficiently.

In this work, considering the strong nonlinear and multiscale features in FCC process, both Hilbert-Huang Transform (HHT) and phase space reconstruction techniques are applied to identify the multiscale dynamics characteristics of FCC process. Temperature in middle position of riser reactor is an indicator of reaction thermal state, so this time series data was collected from a real industrial FCC process with DCS. HHT method perform well in nonlinear and nonstationary data processing. Eleven intrinsic mode functions (IMF) which represent eleven subscale features and one residence series was decomposed by HHT from the original temperature series (OTS). Each subscale series was analyzed by phase space reconstruction, Grassberger-Procaccia (G-P) algorithm, Largest Lyapunov exponent (LLE), Kolmogorov entropy and correlation dimension (D2) algorithm. Through analyzing, the first subscale IMF(1) with highest instantaneous frequency show the noise behavior while the original temperature series and IMF(2)~IMF(9) show the chaotic behaviors. other series express steady state or period state without deterministic chaotic behaviors. These results indicate that the multiscale dynamics behaviors which caused by different temporal scales exist in the FCC process, and on the other hand, provide a multiscale way to analyze FCC process by using the industrial data.

Table: Result of multiscale dynamics system identification

OTS

IMF(1)

IMF(2)

IMF(3)

IMF(4)

IMF(5)

IMF(6)

IMF(7)

IMF(8)

IMF(9)

IMF(10)

IMF(11)

R(11)

Ï„

3

7

2

4

8

15

30

35

30

26

14

9

12

m

17

∞

16

16

13

12

19

12

6

5

4

3

3

D2

7.945

∞

7.115

7.504

5.803

5.425

8.763

5.027

2.407

1.566

1.127

0.9961

0.9869

K2

0.0797

∞

0.0906

0.0535

0.0141

0.0092

0.0035

0.0061

0.0033

0.0037

≈0

≈0

≈0

LLE

0.0478

∞

0.064

0.0487

0.0386

0.0241

0.0035

0.0066

0.0329

0.0123

≈0

≈0

≈0

Type

Chaos

Noise

Chaos

Chaos

Chaos

Chaos

Chaos

Chaos

Chaos

Chaos

-

-

-


Checkout

This paper has an Extended Abstract file available; you must purchase the conference proceedings to access it.

Checkout

Do you already own this?

Pricing


Individuals

AIChE Members $150.00
AIChE Graduate Student Members Free
AIChE Undergraduate Student Members Free
Non-Members $225.00