PID TUNING FOR CASCADE CONTROL SYSTEM DESIGN R. Vilanova, O.

Arrieta Telecommunication and System Engineering Department, ETSE, Universitat Aut` onoma de Barcelona, 08193 Bellaterra, Barcelona, Spain. Ramon.Vilanova@uab.cat, Orlando.Arrieta@uab.cat
ABSTRACT This communication provides an approach for the application of PID controllers within a cascade control system conguration. Based on considerations about the expected operating modes of both controllers, the tuning of both inner and outer loop controllers are selected accordingly. This fact motivates the use of a tuning that, for the secondary controller, provides a balanced set-point / load-disturbance performance. A new approach is also provided for the assimilation of the inner closed-loop transfer function to a suitable form for tuning of the outer controller. Due to the fact that this inevitably introduces unmodelled dynamics into the design of the primary controller, a robust tuning is needed for this. Index Terms PID Control, Cascade control. 1. INTRODUCTION The use of cascade control can be widely found within the process industries in order to improve the reduction of possible disturbances. The introduction and use of an additional sensor that allows for a separation of the fast and slow dynamics of the process results in a nested loop conguration as it is shown in Fig.(1). Each loop has associated its corresponding PID controller. The controller of the inner loop is called the secondary controller whereas the controller of the outer loop as the primary controller, being the output of the primary loop the variable of interest. The rationale behind this conguration is that the fast dynamics of the inner loop will provide faster disturbance attenuation and minimize the possible effect disturbance before they affect the primary output. This set up involves two controllers. It is therefore needed to tune both PIDs. The usual approach involves the tuning of the secondary controller while setting the primary controller in manual mode. On a second step, the primary controller is tuned by considering the secondary controller acting on the inner loop. It is therefore a more complicated design procedure than that of a standard single-loop based PID control
This work has received nancial support from the Spanish CICYT program under grant DPI2007-63356. Also, the nancial support from the University of Costa Rica and from the MICIT and CONICIT of the Government of the Republic of Costa Rica is greatly appreciated.
CCECE/CCGEI May 5-7 2008 Niagara Falls. Canada 978-1-4244-1643-1/08/$25.00 2008 IEEE

system. Some existing studies provide approaches that help in the design of a cascade control system. In [1] a relay-feedback based autotuning method has been used. The procedure still needs of a sequential application of the usual relay based autotuning approach consuming. Other results provide tuning rules for the primary and secondary controller [2]. However there are no clear guidelines on how to automate the process and what should be the rationale behind both tunings. In this paper a design issue that has not been addressed is considered: the tradeoff between the performance for setpoint and load-disturbance response. When a load-disturbance occurs at the primary loop, the global load-disturbance depends on the set-point tracking performance of the secondary loop. In addition, good load-disturbance performance is expected for the secondary controller in order to attenuate disturbances that enter directly at the secondary loop. Also, it is well known that when the controller is optimally tuned for setpoint response, the load-disturbance performance can be very poor [3]. Based on this observation this paper proposes the use of a balanced performance tuning [4] for the secondary loop. Furthermore, an approximation procedure is provided in order to assimilate the dynamics seen by the primary controller to a First-Order-Plus-Time Delay model such that usual tuning rules for PID control can be applied. However, here a robust tuning is suggested, because the primary controller will need to face with unmodelled dynamics coming from the model approximation used for the secondary loop. Note that this kind of approximation is always needed if simple-model based tuning rules are to be applied. The rest of the paper is organized as follows. Next section presents the cascade control conguration and control setup to be used. Section 3 provides the main contribution of the paper as the design approach involving tuning of the controllers and approximation method. Section 4 presents an application example whereas section 5 ends with some conclusions and suggestions for further research. 2. CASCADE CONTROL A typical conguration for cascade control is shown in gure (1), where an inner loop is originated from the introduction of an additional sensor in order to separate, as much as possible,

001775
Authorized licensed use limited to: Norges Teknisk-Naturvitenskapelige Universitet. Downloaded on December 4, 2009 at 08:26 from IEEE Xplore. Restrictions apply.

the process fast and slow dynamics. As a result, the control system conguration has an inner controller C2 (s) with inner loop process G2 (s) and outer loop controller C1 (s) with outer loop process G1 (s). Disturbance can enter at two possible distinct points: d1 and d2 . The rationale behind this conguration is to be able to compensate for the best, the possible disturbance d2 , before it is reected to the outer loop output. In order to accomplish that purpose it is essential that the inner loop exhibits a faster dynamics that allows for such early compensation.
d2 r C1 (s) u1 C2 (s) u2 G2 (s) d1 G1 (s) y

K1 eL1 s (4) 1 + T1 s The FOPTD process model is widely used and constitutes the starting point for many of the existing PID tuning rules. G1 (s) = 3.2. Inner loop controller tuning According to the role of C2 (s) within the control system conguration it is important to bear in mind that a good disturbance rejection is expected for the inner loop (in order to accommodate the possible disturbance entering at d2 ) as well as good set-point tracking capabilities. Effectively, when a disturbance enters at d1 or when a reference change occurs, the outer loop controller C1 (s) will generate the corresponding reference change for the inner loop controller C2 (s). On the other side it is well known [3] that if we tune C2 (s) for good disturbance rejection (set-point response) the set-point (loaddisturbance) response can degrade considerably. Therefore, a balanced tuning is needed for the inner loop controller where a tradeoff for both operation modes is considered. These kind of issues have been introduced in [4] and provided a tuning approach for suboptimal PID design called -tuning. This -tuning is based on the denition of a performance degradation index that takes into account how the response degrades with respect to the optimal one. As one controller will be optimal just for one of the situations (set-point or load-disturbance) the search for the suboptimal controller is performed on the basis of getting a tradeoff among the corresponding degradations with respect to both optimal controllers. The derivation of the method is not shown here for space constraint reasons. However, more detailed information is available at [3] and [4]. 3.3. Model for outer loop tuning Once the tuning of the secondary loop has been completed, the effective system, Ge (s), seen by the outer loop can be determined. Ge (s) = H2 (s)G1 (s) = C2 (s)G2 (s) G1 (s) 1 + C2 (s)G2 (s)

Fig. 1. Cascade Control Conguration According to this, the overall process G(s) = G1 (s)G2 (s) is split into the two parts G1 (s) and G2 (s) and the associated controllers. The two controllers are standard feedback controllers that are assumed to take the usual ISA-PID form as: T d1 s 1 + Ti1 s 1 + (Td1 /N1 )s T d2 s 1 + Ti2 s 1 + (Td2 /N2 )s

C1 (s) = Kp1 1 +

(1)

C2 (s) = Kp2 1 +

(2)

where Kp1 and Kp2 are the proportional gains, Ti1 and Ti2 the integral times, Td1 and Td2 the derivative times and nally N1 and N2 the derivative time noise lter constants. 3. APPROACH FOR CASCADE CONTROL DESIGN The proposed approach for cascade control design is presented according to the different design stages needed in order to completely determine all the control system components. 3.1. Inner loop and outer loop process models A description of the inner process is assumed to be available as a First-Order-Plus-Time-Delay (FOPTD) model. Well known procedures [5] can be applied in order to provide such approximation. Therefore G2 (s) is assumed to obey to: K2 eL2 s G2 (s) = 1 + T2 s (3)

(5)

where the complete expression for H2 (s) takes the form: H2 (s) = with, p1 (s) = (T2 s + 1)Ti2 s( Td2 s + 1) N2 (7) p2 (s)eL2 s p1 (s) + p2 (s)eL2 s (6)

Along similar lines, considering the output of G2 (s) as the input to G1 (s) a model of the same characteristics can be obtained for the slow dynamics part of the system as:

p2 (s) = K2 Kp2 1 + (Ti2 +

T d2 T i2 T d2 )s + (N2 + 1)s2 N2 N2 (8)

001776
Authorized licensed use limited to: Norges Teknisk-Naturvitenskapelige Universitet. Downloaded on December 4, 2009 at 08:26 from IEEE Xplore. Restrictions apply.

where the presence in the denominator of the irrational term eL2 s difcultly the obtention of the effective process model Ge (s) and posterior outer loop controller design. In this paper it is proposed to perform an approximation of the denominator of H2 (s) on the basis of a new polynomial p(s) and delay term es such that: p1 (s) + p2 (s)eL2 s d(s) = p(s)es (9)

1 b , p =A

min

p1 (s) + p2 (s)eL2 s p(s)es

(15)

Once the values for and p are got, the determination of Te in (11) is right straightforward. 3.4. Outer loop controller tuning Once the model for the effective process (11) is available we can proceed with the tuning of the outer loop controller. As the inner loop will provide compensation for local disturbances, we can think of the outer loop controller to be tuned in order to accommodate good performance for the set-point response. Bearing in mind that the process model used for design comes from an approximation of an higher order dynamics, aggressive tunings should be avoided. A very simple and FOPTD model based tuning rule that guarantees some degree of robustness is provided in [6]. 4. EXAMPLE The presented approach is now exemplied by means of a simulation example. Consider the following denitions for the process models: 5 e4s 20s + 1 (16) The tuning of the secondary controller has been performed by application of the optimal ISE tuning rules [7] for set-point sp sp sp tracking operation: Kp 2 = 0.89, Ti2 = 17.83, Td2 = 2.34 ld ld ld and load-disturbance: Kp 2 = 1.40, Ti2 = 5.34, Td2 = 2.39. Application of the tradeoff -tuning [4] with a weighting factor that gives a 25% extra weight to the load-disturbance performance degradation with respect to the set-point performance degradation provides: Kp 2 = 1.15, Ti2 = 11.59, Td2 = 2.37. For all these tunings the value of the derivative time lter is taken as N2 = 10. Consequently, for each one of the three secondary loops, sp e (s) is computed: Ke = an effective model approximation, G sp sp ld ld ld 1, Te = 99.78, Le = 44; Ke = 1, Te = 96.58, Le = 44; = 1, Te = 97.73, L Ke e = 44; and corresponding outer sp sp loop tunings generated as: Kp 1 = 0.87, Ti1 = 101.10, sp sp ld ld ld ld Td 1 /N1 = 99; Kp1 = 0.84, Ti1 = 97.90, Td1 /N1 = 102; Kp1 = 0.85, Ti1 = 99.05, Td1 /N1 = 101. Figure (2) shows that the performance with respect to a step reference change is almost identical for the three scenarios. However, the disturbance attenuation to a load disturbance entering at d2 is clearly superior for the load disturbance optimally tuned secondary controller. Note for the -tuning the load disturbance is also clearly better than that of the set-point tuning (without loosing performance with respect to a step change). G1 (s) = G2 (s) = 1 e40s 100s + 1

First of all the required order of p(s) is determined. Note the effective transfer function Ge (s) can be rewritten, assuming (9), as: p2 (s) K1 e(L1 L2 +)s p2 (s)eL2 s K1 eL1 s = p(s)es 1 + T1 s p(s) 1 + T1 s (10) The purpose here is to have an approximation to Ge (s) on the basis of a FOPTD. Therefore, a second order polynomial p(s) is needed. Note this way Ge (s) will behave, with a 20dB roll-off, as a rst order system. On the other hand, the approximation with a rst order system will make no sense. On this basis, the rst step is to determine the coefcients of p(s) = p0 + p1 s + p2 s2 and the value of . Secondly, to nd: Ge (s) =
Le s e (s) = Ke e Ge (s) (11) G Te s + 1 as the FOPTD model the design of the outer loop controller will be based on. Note that Le = L1 + L2 and Ke = K1 (integral action in the inner loop will assure H2 (0) = 1). Therefore it only rest to determine the vale of Te . The approximation (9) is performed according to the following procedure. First of all, the rst three terms of the McLaurin expansions of both p(s)es and p1 (s)+p2 (s)eL2 s are performed. Equating corresponding terms provides the following set of equations: p0 b0 1 0 0 1 0 p1 = b1 (12) p2 b2 2 2 2 A p b

with, b0 K2 Kp2 b1 Ti2 + K2 Kp2 (Ti2 + Td2 /N2 L2 ) = 2Ti2 Td2 /N2 + 2Ti2 T2 + b2 K2 Kp2 (L2 2 2L2 (Ti2 + Td2 /N2 )+ 2Ti2 Td2 (N2 + 1)/N2 )

(13) Once a value for is provided, p can be determined by: 1 p = A b (14)

The value of will be determined in such a way that provides the better approximation (9) in the following sense:

001777
Authorized licensed use limited to: Norges Teknisk-Naturvitenskapelige Universitet. Downloaded on December 4, 2009 at 08:26 from IEEE Xplore. Restrictions apply.

However, the noted lower robustness margins of the load disturbance tuning generate a system that may be very sensitive to model errors [4]. This can be noted if we take a look at the generated control signal u2 , shown in gure (3). If we assume, for example, a 5% uncertainty in the secondary process time-constant, the performance for the setpoint and -tuning is maintained whereas for the load disturbance the system is critically unstable as it is shown in gure (4).
Process Output variation 1.4 SP Tuning LD Tuning JTuning

Process Output (5% uncertainty in T2) 1.4

1.2

1 Primary output y 1

0.8

0.6 SP Tuning LD Tuning JTuning

0.4

0.2

200

400 time

600

800

1000

1.2

1 Primary output y 1

Fig. 4. Primary output assuming a 5% uncertainty in T2

0.8 0.6

0.4

0.2

200

400 time

600

800

1000

method has been provided that generates a FOPTD approximation suitable for PID tuning. The success of the approx has been shown by means of an example. However a more deep analysis has to be done specially with respect to the obtention of concrete uncertainty bounds for the secondary loop modelling. 6. REFERENCES [1] C. C. Hang, A. P. Loh, and V. U. Vasnani, Relay feedback autotuning of cascade controllers, IEEE Transactions on Control Systems Technology, vol. 2, 1994. [2] Yongho Lee and Sunwon Park, PID controller tuning to obtain desired closed loop responses for cascade control systems, Ind. Eng. Chem. Res., vol. 37, pp. 18591865, 1998. [3] O. Arrieta and R. Vilanova, Performance degradation analysis of Optimal PID settings and Servo/Regulation tradeoff tuning, CSC07, Conference on Systems and Control, Marrakech-Morocco, 2007.

Fig. 2. Primary output for a step in r and d2

Control signal variation 3 LD Tuning SP Tuning JTuning

Control signal u2

200

400 time

600

800

1000

Fig. 3. Control signal u2 for a step in r and d2

[4] O. Arrieta and R. Vilanova, Servo/regulation tradeoff tuning of PID controllers with a robustness consideration, CDC07, 46th IEEE Conference on Decision and Control, New Orleans, Louisiana-USA, 2007. [5] M.A. Johnson and M. H. Moradi, PID Control. New Identicaton and Design Methods, Springer Verlag, 2005.

5. CONCLUSIONS This paper has addressed the problem of PID controller tuning within a cascade control system conguration. A procedure has been outlined that considers a balanced operation (setpoint and load-disturbance) for the secondary controller and a robust tuning for the primary controller. In order to facilitate the design of the primary controller, an approximation

[6] R. Vilanova, IMC based robust PID design: Tuning guidelines and automatic tuning, Journal of Process Control, vol. 18, pp. 6170, 2008. [7] M. Zhuang and D. Atherton, Automatic tuning of optimum PID controllers, vol. 140, no. 3, pp. 216224, 1993.

001778
Authorized licensed use limited to: Norges Teknisk-Naturvitenskapelige Universitet. Downloaded on December 4, 2009 at 08:26 from IEEE Xplore. Restrictions apply.