外文.pdf

1、972 KSME International Journal, Vol. 18 No. 6, pp. 972 978, 2004 Control Scheme Using Forward Slip for a Multi-stand Hot Strip Rolling Mill Young Hoon Moon*, I Seok Jo School of Mechanical EngineeringEngineering Research Center for Net Shape and Die Manufacturing, Pusan National University, Pusan 60

2、9-735, Korea Chester J. Van Tyne Department of Metallurgical and Materials Engineering, Colorado School of Mines, Golden, CO 80401, USA Forward slip is an important parameter often used in rolling-speed control models for tandem hot strip rolling mills. In a hot strip mill, on-line measurement of st

3、rip speed is inherently very difficult. Therefore, for the set-up of the finishing mill, a forward slip model is used to calculate the strip speed from roll circumferential velocity at each mill stand. Due to its complexity, most previous researches have used semi-empirical methods in determining va

4、lues for the forward slip. Although these investigations may be useful in process design and control, they do not have a theoretical basis. In the present study, a better forward slip model has been developed, which provides for a better set-up and more precise control of the mill. Factors such as n

5、eutral point, friction coefficient, width spread, shape of deformation zone in the roll bite are incorporated into the model. Implementation of the new forward slip model for the control of a 7-stand hot strip tandem rolling mill shows significant improvement in roll speed set-up accuracy. Key Words

6、: Finishing Mill, Forward Slip, Hot Strip Mill, Neutral Point, Coefficient of Friction Nomenclature Fi f, hi Ahi h : Shear force between roll and strip of the i th roll stand : Forward slip of the i th roll stand : Plane-strain forward slip of the i th roll stand Improved value for forward slip with

7、 adjustment for width variation : Exit strip thickness at the i TM roll stand : Draft of the i th roll stand : Strip thickness at the neutral point of the i th roll stand * Corresponding Author, E-mail : yhmoon pusan.ac.kr TEL : +82-51-510-2472; FAX : -+-82-51-512-1722 School of Mechanical Engineeri

8、ng/Engineering Re- search Center for Net Shape and Die Manufacturing, Pusan National University, Pusan 609-735, Korea. (Manuscript Received December 12, 2003: Revised March 4, 2004) Lp :Projected contact length between the roll and the strip Ki i TM roll adjustment factor for deter- mining the frict

9、ion coefficient K* : Factor for the i th roll stand used in the determination of Ki. Km : Mill deformation resistance N : Number of coils Pm : Mill load P; : Roll force normal to roll surface for the i th roll stand Pi : Roll pressure for the i th roll stand R; : Radius of roll in the i th roll stan

10、d Zm : Mill torque VLactual Actual strip speed (measured at steady- state conditions) for the i TM roll stand V.,error : Strip speed prediction error for the i th roll stand Vi, preaict: Predicted strip speed for the i th roll stand Control Scheme Using Forward Slip for a Multi-stand Hot Strip Rolli

11、ng Mill 973 Ui vo UN, Wi WN X /3, yi o 7 q)i : Exit strip velocity at the i th roll stand : Circumferential velocity of the i th roll stand Strip velocity at the neutral point of the i th roll stand Exit strip width at the i th roll stand Strip width at the neutral point of the i th roll stand : Hor

12、izontal position in deformation re- gion : Bite angle of the i th roll stand Neutral angle of the i TM roll stand ; Plane-strain prediction accuracy para- meter for the i th roll stand : Angular position in deformation region ; Friction coefficient between the roll and the strip of the i th roll sta

13、nd : Parameter based on torque, load and roll radius which reflects friction condi- tions in plane strain Parameter to account for width changes during deformation in the i th roll stand 1. Introduction In a finishing mill, the roll gap and the roll speed for each stand must be preset in advance bef

14、ore the front end of strip reaches the entry side of finishing mill. Hence, the accuracy of the set-up values directly influences quality of the front end of the hot strip and the subsequent performance of the mill control system. A roll speed model is used to control the mass balance between adjace

15、nt mill stands. So the roll speed control is responsible for attaining stability of the strip velocity and accuracy of the strip thickness. In general, a forward slip model (Koncewicz, 1991 ; Bakhtinov, 1988 ; Zhang, 1995 ; Lenard, 1997 ; Lee, 2002; Zhang, 1989 ; Hum, 1996; Seregin, 1989) is used fo

16、r the roll speed because it converts roll circumferential velocity into strip speed at each mill stand. In a tandem hot strip rolling mill, forward slip is difficult to measure due to high temperature and the harsh environ- ment surrounding the mill. Hence, the lbrward slip in hot strip mill is ofte

17、n obtained from a prediction model but high accuracy is difficult to achieve due to many process variables. The pur- pose of the present study is to develop a better model for forward slip. Since, in practice, the velocity of the strip is calculated from the circumferential velocity of the roll, it

18、is important to have a reliable roll velocity model. The roll velocity model influences not only the thickness of the strip at the front and back ends but also the thickness profile of the entire coil and the operational stability of the mill. Accuracy of the roll velocity model is directly dependen

19、t on the accurate prediction of forward slip. 2. Basic Model for Forward Slip Parameter A schematic of the roll bite geometry for the i th roll stand is shown in Fig. 1. The relative slip between the strip and the roll surface is one of the important factors in the control models for the rolling pro

20、cess. In general, forward slip (fi) is considered as a measure of the relative slip and defined as the difference between strip veloc- ity (vi) and roll circumferential velocity (v ) divided by the roll circumferential velocity, or Vi - V 0 f- o (1) From the volume flow balance between adjacent mill

21、 stands, hi_tVi_lWi_l= hiviwi (2) where h: and w; are the strip thickness and width, respectively, on exit from the i th stand. P Fig. 1 Schematic drawing of roll bite geometry 974 Young Hoon Moon, I Seok Jo and Chester J. Van Tyne From Eq. (I), the exit strip velocity can be expressed as Vl = V ( 1

22、 + f). (3) Substitution of Eq. (3) into the volume flow balance equation yields h,-lvwi-l(l+fi-1) =hivwi(l+f,). (4) The volume balance is satisfied at any point within the roll bite. For the neutral point of the i th stand (defined by the subscript N,-), h-i v/-1 wi-1 = hiviwi = h, vA, w,. (5) Rearr

23、angement of Eq. (5) gives hN, v, wv, (6) Vi= hiwi Substitution of Eq. (3) into Eq. (6) yields hN, vt WN, fi- hivOwl 1 (7) From the geometry, as shown in Fig. l, one can show that hu,=h,+ 2R,( l -cOO 13,) (8) where Ri is the roll radius of the i th roll and ,8i is the neutral point angle. There is no

24、 slip at the neutral point. Hence, v, = vi cos ,8i (9) and substituting into Eq. (7), the basic equation for forward slip can be obtained as h2v WN f/= h,w, coo ,si-l. (10) The strip thickness at the neutral point (h,) can be calculated from Eq. (8) if coo ,8i is known. Likewise, if the neutral angl

25、e (,8/) and strip width at the neutral point (w,) are known, the forward slip can be calculated from Eq. (10). From a force balance in the roll bite region, as shown in Fig. 2, one can find t a t,ei _ PiwiRi sin O dO+ J_ PiWiJliRi cos O dO dO dO (11) - J e P , w a - / R / : coo 0 dO=O where a; is th

26、e bite angle, p; is the roll pressure i I I 1 I P F I I Fig. 2 Forces in the roll bite and /zi is the coefficient of friction for the i th stand. If the friction coefficient, the strip width and the roll pressure are constant in the roll bite region, then Eq. (11) can be solved (Koncewicz, 1991; Bak

27、htinov, 1988) for the neutral angle. This gives sin sin a; / . The bite angle (ai) can be derived from geomet- rical relationships (see Fig. I) as -COS ai) =l (hi-l-hi). (13) R/(1 From Eq. (lO) and assuming a constant strip width (wi= w,) during rolling, the forward slip for this plane-strain condit

28、ion is calculated as f/* =/COO ,8i- 1 (14) where, hN,=hi+2R(1-cOO,8i), and cOO,8i = /l-(sin,8i) 2. The value for fli can be deter- mined from Eq. (12) if the friction coefficient is known. 3. Determination of Friction Coefficient in the Forward Slip Equation To calculate forward slip from Eq. (14),

29、the friction coefficient must be known. In a produc- tion strip rolling process, the friction coefficient depends on a large number of variables, which may vary during the operation, such as bite angle, Control Scheme Using Forward Slip for a Multi-stand Hot Strip Rolling Mill 975 rolling speed, lub

30、rication, strip surface condi- tion, etc. The selection of a unique value for the friction coefficient is very difficult task. One possible approach to selecting a numerical value for the friction coefficient is as follows. It can be shown that the more the strip width expands during rolling, the fo

31、rward slip becomes less. It was assumed in derivation of Eq. (13) that the strip width does not expand. It is asserted that sin fl is at a maximum value for this condition. Hence, as the friction coefficient approaches a minimum value, the value for sin fl,., as calculated from Eq. (12) increases an

32、d approaches a more realistic value because it can compensate width variation during rolling. For the strip to enter the roll bite, F cos aPi sin a (15) where Fi is the shear force due to friction at the interface between the roll and the strip and /9,. is the normal force (see Fig. 2). Rearrangemen

33、t yields F; _ sin a; (16) Pi COS Oli From the definition of the friction coefficient Fi=,uiPi, one finds 1 Ahi 2 lz,tan ai 7- . (17) The minimum friction coefficient is the value satisfying the equality of Eq. (17). The minimum friction coefficient is i / Ah; ,t2/,min = / ) . (18) The minimum fricti

34、on coefficient of Eq. (18) can be easily used for the determination of forward slip and hence for the roll-speed control model, but it should be refined in order to compensate for the characteristics of individual mill stands. The refinement can be done through the use of mill processing data. The r

35、olling torque is strongly correlated with the friction coefficient. Hence, a refinement based on the roll torque is developed to obtain more realistic value for the friction coefficient in the roll-speed control model. The mill load (Pro.i) for each stand can be obtained from the mill deformation re

36、sistance (Kin,i) and the projected contact length (Lpa) between the roll and the strip by Pm.i Lp.i r . -.o lXm,iax (19) The roll torque (Tma) can be calculated as Tma Lp., - fo (,ui Km, Mx) R Wi (20) , ffLPd =12 RJo Km,dx By substituting Eq. (19) into Eq. (20), one can obtain Tin.-MR Pro. (21) Wi W

37、i From Eq. (21), a new parameter that reflects the frictional conditions is determined as T=a (22) IZ* - RiPm, i Instead of using the minimum friction coefficient of Eq. (18) in the roll-speed model, one can select a friction coefficient based on the operating parameters as /z = K,u * . (23) The val

38、ue of/2,.* can be obtained from Eq. (22) and the value of K; can be determined from averaging across all the mill stands because the exact value for each stand is difficult to obtain in actual tandem mill. For a seven-stand tandem mill, Ki is - , 1 7 * (24) with K= TmJ(emR) .= , where N is the numbe

39、r of coils that have been rolled and K,.* is found by averaging over th e number of coils that have been rolled. There- fore, forward slip for the plane-strain condition can be obtained from Eq. (14) if the value of the friction coefficient from Eq. (23) is used to de- termine the neutral point angl

40、e in Eq. (12). 976 Young Hoon Moon, 1 Seok do and Chester Van Tyne 4. Consideration of Strip-width Variation During production rolling, plane-strain condi- tions are not maintained within the roll bite and the characteristics of each roll are different from one another. Therefore, the forward slip v

41、alue calculated under the plane-strain condition should be revised according to the width varia- tion by using operational rolling data. The basic forward slip model, given in Eq. (10), can be re-written using the parameter , which compensates for width variation as h, f,=- cos fl,- 1 (25) if one us

42、es a predicted friction coefficient, which determines the most accurate value for forward slip, f, then the most desirable situation would be that the predicted friction coefficient is equal to real friction coefficient, leading to f,=f, (26) From Eqs. (14), (25) and (26) one can obtain, f,- (f*+ 1)

43、 ,-1 (27) Therefore, the parameter that compensates for width variation can be defined as i- fi+ l (28) f*+l Hence if the predicted friction coefficient, which allows calculation of the most accurate value for forward slip (f), can be determined, then the average value of parameter a)i shown in Eq.

44、(28) can be obtained at each stand, and the optimal forward slip can be obtained from Eq. (25). 5. Determination of f To determine the friction coefficient that pro- duces most accurate value for forward slip (f), the following procedure has been used. For a plane-strain condition, the circumferen-

45、tial velocity of the i th roll can be determined from Eqs. (4) and (14) as VO, NEWl (l +fi,*NWl ) h7 0 _ (29) U i, NEW1 - (l + f,*E) h, which uses the exit velocity, exit thickness and forward slip from the last mill stand (i-7). The subscript NEW l notes the calculations are based on the plane-stra

46、in forward slip model. For the revised situation where width variation in the roll bite can occur, the forward slip based upon the refined model (f), yields 0 o _ vT,. (l +f) h7 vi, (30) ( l +fiT.ue) hi where the subscript NEWT denotes the forward slip model after refinement (f). To estimate the pre

47、diction accuracy of the plane-strain forward slip model, the parameter (Vi) is defined as 0 0 v, r,cr - vi.mwi (31 ) ti- 0 V i, NEW1 where the target circumferential speed (viO, rARcEr) is the actual measured speed after strip stability has been achieved in the mill. Therefore, when the (Vi.NEa) is

48、the predicted circumferential speed 0 ( Vi, raEr), Ti same as target circumferential speed 0 would be zero. For the refined model, the following must be satisfied. o o (1 + ,) (32) V i, NEW2 = V i, NEW1 Eq. (32) can be expressed as V,vEvez (1 +-7,vew2) (1 +mw2) (33) _ VO.NEWl ( 1 + f T*,NEW1 ) (l+yi

49、). (1 +/5, ) In the mill set-up, speed for the final (7 th) stand does not change significantly, Hence YO.UEWl is 0 nearly the same as VT.CE. Therefore, f-,.,E = ( l +/:.*,E ) ( l +f.NE.) I. (34) ( I +/7,%. ) (1 + ri) TO calculate fi.m from Eq. (34), fT,UE must be known. Control Scheme Using Forward

50、 Slip for a Multi-stand Hot Strip Rolling Mill 977 The value of 7,NEW2 in the right side of Eq. (34) can be initially approximated by f,*NEVa. Then, Eq. (34) becomes 0.03-0.09 and shows maximum at the 3 rd and 4 th mill stands. Figure 5 shows an implementation of the newly ( 1 + f*VEV, ) ( 1 + fT*,U

51、E ) f i, NEW2 ( +feWl ) (l + 7/) - f NEW1 - ?i ( I + ?) (35) From Eq. (35), fT,vE can be obtained as fr, jVEW2 = f,*EWX - ?7 (36) (1+77) By substituting Eq. (36) into Eq. (34), the final expression for the f; can be obtained as f i f i,vEw2 ( 1 q- f-.*,vEwa ) ( 1 + 7;) (37) 6. Calculation of Forward

52、 Slip Figure 3 shows the predicted friction coefficient at each stand from operational data. The friction coefficients obtained by Eq. (18) show a steeply decreasing trend with increasing stand number. On the other hand, friction coefficients obtained by Eq. (23) do not show a steep decrease for the

53、 later stands. Figure 4 shows the predicted forward slip at each stand from operational data. The forward slip obtained by Eq. (25) lies in the range of 0.5 0.4 1 0.3 0.2 E. 0.1 0.0 -.-e- u = (,:/ja/R) 2 -.u =K -,u I | I I I I I F1 F2 F3 F4 F5 F6 F7 Stand number Fig. 3 Friction coefficient at each s

54、tand I old model i I new model 0.12 ,4.- 0.00 . . l F1 F2 1:3 F4 F5 F6 FT Stand number Fig. 4 Predicted forward slip at each stand . I old model J- new model . 2 io F1 F2 F3 F4 F5 Standard number (a) Standard deviation of Vi, error F6 a o -1 I / old model -1 new model Fig. 5 F1 F2 F3 F4 F5 F6 Stand

55、number (b) Average of Vl.ero, Comparison of strip speed prediction error (V . ) 978 Young Hoon Moon, I Seok Jo and Chester Van Tyne Fig. 6 Summary of forward slip calculation developed model. The operational data obtained by existing forward slip model, designated as old model, had been compared wit

56、h newly developed model. Although not specifically ex- pressed in this paper, the semi-empirical old model had only considered reduction ratio and strip thickness at each stand in the calculation of forward slip and the prediction accuracy had not been so good. As seen in the figure, the roll speed set-up accuracy is significantly increased when using the newly developed model. In Fig. 5, the strip speed prediction error, Vi, error can be obtained by Eq. (38). Vi, error= Vi, actual- Vi, prectict (38) Vi, predict w