Development of Control Performance Diagnosis System and its Industrial Applications

Transcription

1 Development of Control Performance Diagnosis System and its Industrial Applications Sumitomo Chemical Co., Ltd. Process & Production Technology Center Hidekazu KUGEMOTO The control performance diagnosis system, PID Monitor, and the PID tuning tool, PID Tune, have been developed. These systems are useful in improving the control and maintaining high productivity of plants. PID Monitor observes the performance of all controllers in the plant, and it picks out the loops which have problems. The diagnosis report is displayed as a Web page on the intranet. The operator and the staff can efficiently improve the control by supervising it. PID Tune is used to tune extracted loops. It is able to do the tuning safely without process changes, as it does not need specific plant tests. This paper introduces the technical background of these systems and some applications in a real plant. PID 90 PID PID ) TPM Total Productive Maintenance 2), 3) PID 4) PID 4) DCS TPM 5) 6) PID Monitor PDCA Plan-Do-Check-Action TPM PID PID PID Tune 7) VRFT FRIT 8) 9), 10) PID PID Tune PID 2010-

2 IT PIMS Plant Information Management System Fig. 1 PID Monitor Web Web PDCA1 Web server Site r t Fig. 2 yr(t) = 0 y(t) = D 1 CP w(t) F z = d G w(t) 1 z d CP = F z d Controller C(z 1 ) Block diagram G P 1 z d CP u t w(t) Process P(z 1 ) w t D(z 1 ) y t PIMS = Fw(t) Hw(t d) (4) Tag list Web file Fig. 1 Control performance diagnosis system 11) P (4) 1 2 Fw(t) Hw(t-d) Fig. 2 C PD yu y(t) = P(z 1 )u(t) D(z 1 )w(t) (1) u(t) = C(z 1 )(r(t) y(t)) (2) wr d 1 F G D(z 1 ) = F(z 1 ) z d G(z 1 ) (3) z d d Var {y(t)} = Var {Fw(t) Hw(t d)} = Var {Fw(t)} Var {Hw(t d)} Var {Fw(t)} = σ 2 MV (5) Var σ 2 σ 2 MV (5) y σ 2 y σ 2 MV (4) 2 Hw(t-d) (5) 2 Var{Hw(t-d)}= 0 σ 2 MV y σ 2 y η(d 1) = σ 2 MV (d 1) σ 2 y (6) 2010-

3 η 0 1η 1 0 w ARMA y w y y yufig. 3 yσ y η Data1 Data1 Data2 Data1 white noise Fig. 3 y(t) = w(t) Data2 sine curve & small white noise 4.0 y(t) = sin(2πt/60) 0.2*w(t) Standard deviation Control performance index Standard deviation Control performance index Bad loop! Comparison between standard deviation and control performance index η η 0.7 γ PIMS yr u r > ȳ 3σ y r < ȳ 3σ y (8) u = Constant DCS γ γ = η exp( N 24/100) (7) N 1γ η η = 1 2 5), 12), 13) 2010-

4 Flow MVl Level Flow rate Manipulated variable Liquid level Fig. 4 Fig. 4 x(t) = 4 π P x = X X * Power With flowmeter Flow rate Identification using backlash function Manipulated variable Without flowmeter Frequency analysis Frequency 1 1 sin wt sin 3wt sin 5wt 3 5 X x(t)x* P x (9) 1/(2n1) 2 Using the relationship between manipulated variable and flow rate Using the harmonics of power spectrum Methods for detecting valve failure (example of liquid level control) 14) (9) Fig. 4u F F(t) = max [min {F(t 1) Δu(t), F max }, 0] F F max u F max F max F 0.7F max Fig. 5 Cause loop Fig. 5 Method for detecting root cause 11) 2 C xy (τ) = E{ x(t)y(t τ)} R xy (τ) = C xy (τ) C xx (0)C yy (0) (10) (11) x(t), y(t)c xy R xy 2 C xx (τ) = E{ x(t)x(t τ)} C R xx (τ) = xx (τ) Cxx (0) Period 9min propagating (12) 2010-

5 C xx R xx PID 2 PID Fig. 6 PID PID Tune GA PID Fig. 7 GA 2 PID r t Case1 Case2 Case3 Case4 K (1 Ts) e Ls e Ls K (1 T 1 s)(1 T 2 s) 1 Ts Controller Controller model Fig. 7 Identification structure based on GA 10) K s(1 Ts) e Ls e Ls (13) (14) (15) (16) s K, T, L Case1 Case4 u t û t Process Process model ξ t y t ŷ t y(t) = a 1 y(t 1) a 2 y(t 2) b 0 u(t d 1) b 1 u(t d 2) ξ(t) Δ (17) Fig. 6 PID tuning tool ξδ a, b Case1 a 2 =0 Case2 Case3 a 1 = 1, a 2 =0 Case4 a 2 = (a 1 1) (17)CARIMA Controlled Auto-Regressive and Integrated Moving Average PID I-PD GA 1 GA GA Δu(t) = k c T I e = r(t) y(t) e(t) k c T D Δ Δ 2 y(t) k c, T I, T D PID e (18) 2010-

6 (17)(18) (18) ŷ(t) = y(t 1) a 1 Δy(t 1) a 2 Δy(t 2) b 0 Δû(t d 1) b 1 Δû(t d 2) û(t d 1) = u(t d 2) k c Δy(t d 1) k c e(t d 1) T D T I k c Δ{ y(t d 1) y(t d 2)} (19) (20) C(z 1 )y(t) = Δu(t) C(1)r(t) = 0 (23) C(z 1 T ) = k c 1 s T D 2T 1 D z 1 T D z 2 T I GMVC 16) J = E [{P(z 1 )y(t d 1) λδu(t) P(1)r(t)} 2 ] (24) GA τ f = Σ [{ ŷ(t) y(t)} 2 {û(t d 1) u(t d 1)} 2 ] t=d1 (19)(20) Fig. 8 a 1, a 2, b 0, b 1, d, k c, T I, T D (21) (1) Initial individual population P 1 P 2 P 3 P N (21) (5) Mutation child P k P(z 1 ) P(z 1 ) = 1 p 1 z 1 p 2 z 2 p 1 = 2e p 2 = e μ ρ ρ 2μ ρ = /α cos μ = 0.2(1 δ) 0.51δ λ α μ α μ δ δ 0.0Diophantine 4μ 1 2μ P(z 1 ) = ΔA(z 1 )E(z 1 ) z (d1) F(z 1 ) ρ (25) (26) Fitness (2) Sorting/(3) Selection P 9 P 6 P N P 3 (4) Crossover child P 9 P 9 P N P 1 P 2 child E(z 1 ) = 1 e 1 z 1 e d z d F(z 1 ) = f 0 f 1 z 1 f 2 z 2 (22)(24) Fig. 8 Process of evolutionary identification using GA 10) F(z 1 )y(t){e(z 1 )B(z 1 )λ}δu(t) P(1)r(t)= 0 (27) (27) 2 GMVCPID (17) A(z 1 )y(t) = z (d1) B(z 1 )u(t) ξ(t)/δ (22) A(z 1 ) = 1 a 1 z 1 a 2 z 2 B(z 1 ) = b 0 b 1 z 1 F(z 1 )y(t){e(1)b(1)λ}δu(t) P(1)r(t)= 0 (23)PID 1 k c = ( f 1 2 f 2 ) E(1)B(1) λ T I = T D = f 1 2 f 2 f 0 f 1 f 2 f 2 f 1 2 f 2 (28) (29) 2010-

7 eδ u I(λ) λ K eδ u H 2 (30) (34)λE(z 1 ) F(z 1 )PID I(λ) = E[e 2 (t)] K 2 E[Δu(t) 2 ] e(t) = Δu(t) = 1 T(z 1 ) ξ(t) C(z 1 ) ξ(t) T(z 1 ) T(z 1 ) = ΔA(z 1 ) z 1 B(z 1 )C(z 1 ) 1 T(z 1 ) I(λ) = ǁ ǁ K 2 ǁ ǁ C(z 1 ) 2 T(z 1 ) 2 PID Tune Fig (30) (31) (32) (33) (34) PID 3372Fig. 10 PID Fig. 11 Good Bad Good Bad Control performance index Control performance index Fig Plant A (only tuned loops) After tuning Before tuning Control loop Plant B (only tuned loops) After tuning Before tuning Control loop 33 loops 72 loops Comparison of control performances in plants A and B 14) SV, PV (%) Before tuning k c = 0.455, T I = 400s, T D = 0s PV SV Control performance index Product SV, PV (%) 54 After tuning k c = 1.667, T I = 900s, T D = 0s 46 Fig. 9 Result of tuning (liquid level control) 10) TC Fig. 11 TC Improved control performance in distillation process 14) PID 2010-

8 SV, PV (%) SV, PV (%) Before Fig. 12 Fig. 12 Time (h) After installing positioner SV% PV% MV% Time (h) SV% PV% MV% Improvement of control performance by installing valve positioner 14) PID MonitorPID PID Tune MV (%) MV (%) positioner 143 NO.25 PID 1) R. Miller, Ind. Eng. Chem. Res., 44, 6708 (2005). 2),, (1999), p ),, 44(2), 135 (2005). 4),,,, 52 (8), 270 (2008). 5),, 41 (2005). 6) T. J. Harris, Canadian Journal of Chemical Engineering, 67, 856 (1989). 7) T. Yamamoto, K. Takao and T. Yamada, IEEE Trans. on Control Systems Technology, 17 (1), 29 (2009). 8),,,,,, 22 (4) 137 (2009). 9) H. Seki and T. Shigemasa, Journal of Process Control, 20, 217 (2010). 10),,,, 47 (11), 937, (2008). 11),, S2-5-1 (2006). 12),,, 44 (2), 143 (2005). 13) M. Jelali, B. Huang (eds.), Detection and Diagnosis of Stiction in Control Loops, Springer (2010), p ),, 61 (), 230 (2009). 15) T. Yamamoto, K. Kawada, H. Kugemoto and Y. Kutsuwa, 15 th IFAC Symposium on System Identification, 729 (2009). 16) D.W. Clarke and P.J. Gawthrop, IEE Proc. of Control Theory and Applications, 126 (6), 633 (1979). PROFILE Hidekazu KUGEMOTO 2010-