Analysis of PID Control Design Methods for a Heated Airow Process

Size: px
Start display at page:

Download "Analysis of PID Control Design Methods for a Heated Airow Process"

Transcription

1 Helsinki University of Technology Faculty of Information and Natural Sciences Department of Mathematics and Systems Analysis Mat-.8 Independent research projects in applied mathematics Analysis of PID Control Design Methods for a Heated Airow Process April 7, 8 Tapani Hyvämäki Instructors: Lic. Tech. Lasse Eriksson M. Sc. Antti Pohjoranta

2 Contents Introduction The Case Process. Process Identication Parameter Estimation PID Controller Tuning Methods 6. Ziegler Nichols Method Cohen and Coon Method AMIGO Method Lambda Tuning Method Internal Model Control Method Practical PID Controller Design Aspects Tools for Analysis of Tuning Methods. Performance of Tuning Methods Sensitivity to Disturbances Results. Process Model Parameter Estimation Controller Parameters Comparison of Controllers' Performance Comparison of Controllers' Sensitivity Applying Methods to the Heated Airow Process Educational Discussion 7 Summary and Conclusions 6 A PID Controllers 8 A. Formulations of Controllers A. Lambda Tuning Method Parameters B Performance Measure Formulations 9 C Additional Tables and Figures C. Controller M-values and Performances C. Closed-loop System Output and Control Signals for Category One Controllers C. Performance and Robustness Comparison

3 Introduction The Department of Automation and Systems Technology is responsible for organizing a laboratory course on automation and control engineering for undergraduate students at the Helsinki University of Technology. One of the fundamental objectives of the course is to cover the basics of PID controllers and PID controllers' tuning methods. The current laboratory exercise on PID control included in the course has been found to be in need of improvements. The aim of this special assignment is to examine methods to renew and modernize the contents of the laboratory exercise and to give students more illustrative and practical exercise to systems' control. The case process under study is a heated airow process. It is described more precisely in Section. First we introduce the system model and the principles of parameter estimation. Dierent PI and PID control methods and also more complicated control methods as well as the tools for analysis and comparison of the controllers are described. Also, some aspects that have to be taken under consideration when implementing controllers are discussed. After familiarization with the used methods, the parameter estimation, simulations with the model and calculations of performance and robustness are performed. As a result, a thorough analysis of the performance and suitability of the controllers to the case process is performed. Comparison of the methods in the sense of performance and robustness is considered and a short discussion of the results from the educational point of view is included. The results obtained here are utilized in the development of the laboratory exercise and in its software implementation. However, the nal development and software implementation are not in the scope of this research project though the model implementation and scripts connected to the derivation of the results of this work will be used in the development of the laboratory course material and software. The Case Process The heated airow process that has been and will be still used as the case process in the laboratory exercise consists of a fan element, heater sink, tube and temperature sensor. The fan produces an airow owing through the tube with constant speed. Air is heated in the beginning of the tube with the heater whose heating power can be adjusted with a control signal. The air temperature is measured with a temperature sensor at dierent distances along the tube. The airow speed inside the tube can be manually controlled by a throttle control. The throttle control angle can be adjusted between 7. In this work we use constant throttle control value of. The temperature sensor position is also preserved at the distance of. cm. A schematic illustration of the process is shown in Figure. The control signal of the process is bounded to interval..v. It is approximately linearly proportional to heater power and can produce a temperature dierence up to 6 C between the inowing and Proportional Integral Derivative

4 Figure : A schematic gure of the case process [9] out-owing air. The measurement signal is also bounded to interval..v and is linearly proportional to the airow temperature. [9]. Process Identication The heated airow process can be modeled starting from heat conducted from heater to owing air and again conducted from air into the temperature sensor. This leads to a linear distributed parameter model of rst order represented in [9]. This model can be simplied into a second order linear process, which has the transfer function G (s) = T s(s) P (s) = K (T a s + )(T b s + ) e sl, () where parameters T a and T b are system constants, L is output delay and K is static gain. Output T s is the simulated temperature and it is the temperature above the room temperature. From now on we denote the set of these model parameters as P = {K, T a, T b, L }, the subscript referring to the model order. In the future, we will also denote the system input P (t) with u(t) and output T s (s) with y(t). The model presented above does not take into account, for example, little turbulence in the airow or heat conduction from owing air into the tube before the temperature measurement and vice versa. However, these eects can be assumed to be minimal if the system is operating at some small temperature interval, i.e., varying input signal only at some small interval (and keeping the airow speed constant) keeps also the temperature at bounded interval and the heat conduction minimal. In many cases a rst order approximation of the system is accurate enough to model the process and as will be seen later, many PID tuning methods base on this approximation. Such a system has a transfer function G (s) = K T s + e L s, () where T is the constant, L is delay and K is the static gain of the system. Later on we will denote the set of these parameters as P = {K, T, L }.

5 .. K.8 y(t) / u(t).6. 6.% K K. L T. 6 8 Figure : Tangent method for rst order model parameter determination, static gain K = K /K.. Parameter Estimation In order to obtain proper results when using the system models introduced in previous section, the model parameters P and P need to be estimated. For our purposes it is sucient to estimate the parameters at each operating point at a or use the same estimated parameters in an operating interval as wide as possible. Estimation of parameters P is performed by tting the model output curve to measured process output values for a step input. The curve is tted using the least squares method. So, we minimize the sum of squares min P E(K, T a, T b, L ) = n k= ( Tm,k T s,k (K, T a, T b, L ) ), () where T m,k is the measured temperature value, T s,k is the estimated temperature value, k is the measurement index and n the total number of measurements. The minimization is performed with a Matlab optimization algorithm. Parameters P of the rst order model are determined with slightly dierent and simpler method called tangent method (explained in [] and [6]). The tangent method also utilizes process step response and it is based on sketching a tangent to the step response curve where the slope has its maximum value. Delay L is determined from the intersection point of tangent and t-axis and constant T is determined from the value where step response reaches 6,% of it's nal value. Static gain K is the proportion of input and output signals at stationary state K /K. Illustration of the rst order model parameter estimation is shown in Figure.

6 PID Controller Tuning Methods PID controllers are the most widely used controllers in industry. One reason to such popularity is that usually PID controllers are very simple and easy to implement and tune. A basic PID controller consists of proportional, integral and derivative terms and it can be represented as u(t) = K p [ e(t) + T i t e(α)dα + T d d dt e(t)], () where e(t) is the error, u(t) the control signal and K p, T i and T d the controller parameters. The error signal e(t) can be written in terms of the reference signal y r (t) and the system output signal y(t) as e(t) = y r (t) y(t). Several other representations of () exist and certain of them are listed in Appendix A.. It is worth noting that in this report upper case letters K p, T i and T d are used when parameter K p aects all terms of the PID controller and lower case letters k p, k i and k d are used when each parameter aects only certain term of the PID controller. Most of the PID controller tuning methods assume that the controlled process behaves as a linear rst order system with constant output delay, see (). If the system is non-linear it can be linearized at dierent operating points. Similarly, higher order systems can be approximated with lower order models, especially when the system has dominating poles. PID controllers are mainly used for two purposes in systems' control. The rst is load disturbance rejection which means that the controller is used to compensate the sudden and unexpected disturbances that may occur in process input. The second is reference signal tracking, which means that the controller is used to speed up the process ability to follow changes in the reference signal y r (t). In this work both cases are studied, using dierent PID tuning methods described in following sections. We ended up choosing the following methods mainly for two reasons. First of all, the methods have to be the simplest, because the course is the rst practical course of systems' control for its participants. And secondly as important is that the methods need to clearly illustrate the dierent characteristics as well as advantages and disadvantages of PID controllers. These subjects will be discussed more from the laboratory exercise point of view in Section 6. Other reasons for method selection are that they include the most well known methods (the Ziegler Nichols Methods) and also one fairly recent method (AMIGO) proposed by Åström and Hägglund, who are world-wide recognized researchers and pioneers of modern process control.. Ziegler Nichols Method Ziegler Nichols tuning methods (ZN tuning methods) are the principal methods used in PID controller tuning. The two methods are called Step Response Method and Ultimate Frequency Method. The step response method is based on the open-loop step response of the system. The needed system parameters, delay L, constant T and gain K are determined similarly as explained in Section.. Tuning parameters are calculated using formulas in Table. Ultimate frequency method is based on determining the highest gain K kr that keeps the proportionally controlled closed-loop system stable. By determining 6

7 T kr from the system output, as seen in Figure, we can calculate the PID controller parameters using formulas in Table. Both of these methods are designed using the quarter amplitude damping design criterion. [6] Table : PID parameters according to the Ziegler Nichols step response method K p T i T d P PI PID T KL - -.9T KL L -.T KL L.L Table : PID parameters according to the Ziegler Nichols ultimate frequency method K p T i T d P.K kr - - PI.K kr.8t kr - PID.6K kr.t kr.t kr. (t) T kr Figure : Ziegler Nichols Ultimate Frequency Method parameters determination. Cohen and Coon Method The Cohen and Coon tuning method is based on the same assumptions on the rst order system as the Ziegler Nichols method. Nevertheless, the tuning criteria and parameters are determined in a slightly 7

8 dierent way. The PID controller parameters are calculated using the gain K, the constant T and the delay L of the rst order approximation of the process (), but using the rules shown in Table.[] Table : PID controller parameters according to the Cohen and Coon method K p T i T d P PI PID ( T +. ) - - K L ( T ). +.(L /T ) L K L +.(L /T ) ( T. +. ). +.6(L /T ) L K L +.6(L /T ) -.7L +.9(L /T ). AMIGO Method AMIGO (Approximate M-constrained Integral Gain Optimization) tuning method uses a dierent PID control algorithm compared to the previous methods. Here ltering is applied on all terms, not only on the derivative part. The AMIGO controller is given as u(t) = k p ( byr (t) y f (t) ) + k i t y r (α) y f (α)dα + k d ( c dy r (t) dt y f (t)), () dt where b and c are the set-point weights and y f is the ltered process output signal. y f is obtained from the process output signal y by ltering, i.e. Y f (s) = G f (s)y (s). The lter transfer function is G f (s) = ( + st f ) n, (6) where T f is the lter constant and n =,,.. is the lter of order. The AMIGO PID controller actually has six tunable parameters k p, k i, k d, T f, b and c. The AMIGO tuning rules proposed by Åström and Hägglund in [8] are K p = K (. +. T L ) T i = L.L +.8T L +.T T d =.L T.L + T where parameters K, T and L are determined by the rst order model in (). Filter constant is set to T f.l and the set point weights are set to c = and b =, if L T +L <., if L T +L. Note that k p = K p, k i = K p /T i and k d = K p T d.. Lambda Tuning Method Lambda tuning method is principally a pole placement method. The process model is assumed to be the rst order transfer function (). The closed-loop transfer function of the process is desired to be of rst 8 (7)

9 y r + - Q(s) u G(s) G ( s) d + + y - y + Figure : Block diagram of IMC controller order G cl (s) = K + λs e L s (8) where λ is the tuning parameter that determines the pole location. Smaller λ values increase the controller performance but take it closer to the instability region. Now by using the rst order Taylor approximation on the delay and by setting T i = T we get the following PI controller tuning parameters (see Appendix A.) T K p = K(λ + L ), T i = T. (9) Similarly using Padé approximation of the delay we obtain tuning parameters for a PID controller as [] K p = T + L / K(λ + L /), T i = T + L /, T d = T L L + T. (). Internal Model Control Method Internal Model Control (IMC) is a model based tuning method that utilizes a rst or higher order model of the process. It is the only model based tuning method we use that allows applying higher than rst order models. IMC controller consists of an estimator block G, a process G and the controller Q. A schematic block diagram of an IMC controller is illustrated in Figure. The system model G considered in this study was introduced in Section.. The model G(s) can be divided into two parts and written in form G(s) = G + (s) G (s), () where G + (s) is the non-invertible part of the transfer function and G (s) is the invertible part. Thus, all unstable zeros and delays of G(s) are included in G + (s). Now the controller Q(s) can be written in form Q(s) = where f is a low-pass lter of desired order n. For f we can write G (s)f(s), () f(s) = ( + T l s) n, () where T l is the constant of the lter and n is the lter order. The IMC control loop can also be represented in a classical control loop form with controller C(s) as illustrated in Figure. 9

10 y r + - C(s) u G(s) y Then, the controller transfer function becomes Figure : Block diagram of IMC controller Q(s) C(s) =. () G(s)Q(s) Dividing the heated airow process model transfer function () into invertible and non-invertible parts yield to K G + (s) = e sl, G (s) = (T a s + )(T b s + ). () Now, () can be written in the form Q(s) = Furthermore () can be written as G (s)f(s) = (T as + )(T b s + ) K( + T l s) n. (6) C(s) = (T a s + )(T b s + ) K ( + T l s) n, (7) sl e where n should be selected such that the degree of denominator is greater or equal to the degree of numerator. So, here n should be set to n. By approximating the delay with the Padé approximation we can write (7) in form e sl = L s + L s C(s) = L T a T b s + (T a T b + L (T a + T b ))s + (T a + T b + L )s + K L Tl s + (Tl + T l L )s + (T l + L )s (8). (9) It should be noted at this point that T l is the only tunable parameter in the controller. So the tuning of the controller reduces into nding the optimal value for parameter T l. The delay e Ls could also be approximated with a rst order Taylor polynomial which leads into Lambda PID controller introduced in previous Section. [].6 Practical PID Controller Design Aspects When embedding the PID controller into a physical process such as our case process, the heated airow system, some design restrictions must be taken into consideration. We focus on three of those restrictions called integrator windup, measurement noise and nonlinearity.

11 Integrator windup occurs because the control signal is in practice bounded. In our process the heater control signal is bounded to the interval..v. If the closed-loop process is controlled with a reference signal y r that exceeds the process maximum output y max, a constant deviation remains in the error signal e(t). This causes the integral term, integrating the error signal to continuously increase, which directly causes the controller output signal to increase. While the actuator remains bounded, the integral term continues increasing and nally when the reference signal y r is set back to a feasible value it takes a relatively long for the controller integral term to get back to normal value. Integrator windup can be prevented using integrator anti-windup, which measures the dierence between the PID controller output and the bounds of output if the signal is out of bounds. Upon actuator saturation the integral term is aected by the anti-windup mechanism to prevent the undesired behavior. Measurement noise always occurs in physical measurements. The temperature measurement output signal of our process has substantially large noise. Measurement noise, especially noise with high frequency, inuences mostly on the PID controller derivative term, because it has an increasing frequency response. Hence, ltering needs to be applied at least on the controller derivative term as explained in Appendix A.. Nonlinearity in a system can be seen, for example, if the parameters of the process changes as a function of the operating point. In our process it can be seen that the static gain, for example, changes dramatically at high temperatures. Generally, the nonlinearities can be avoided by approximating the system to be linear at a small interval around certain operating point. All computer-aided controllers are eventually discrete- controllers. Discrete- controllers are discrete approximations of the corresponding continuous- controllers and thus contains always some error compared to the continuous controllers. However, with modern microprocessors and measurement electronics the sampling interval can be so small that the error is very small. [6] Tools for Analysis of Tuning Methods The educational purpose of the PID control laboratory exercise is to give a clear and illustrative picture of the basic behavior and characteristics of PID controlled systems, to familiarize students with advantages and disadvantages of PID controllers and to introduce the tuning methods of PID controllers. To meet these requirements some comparison and analysis of the PID control methods needs to be performed. Our target is to choose a few of the methods introduced in Section and use them rather than using all methods. It is also desired that the application area that is originally intended for each controller is illustrated. In order to be able to compare the methods, the performance criteria need to be set. In this work we focus on performance and robustness. This section briey reviews the chosen criteria.

12 . Performance of Tuning Methods There are several criteria that are used to measure controller performance. These methods more or less measure the total error between the system reference path and the real system path. The formulation of a performance measure can be represented in form J i (T ) = T h(e(t), u(t), t)dt, () where e(t) is the deviation from the path error, u(t) is the control signal and h is a weight function giving positive values. Dierent representations of performance measures are shown in Appendix B []. Another measure indicating system performance is the system settling. Generally, settling is determined as the from which the error remains less than % of the reference value after the step change in reference. Nevertheless, somes it can be quite dicult to determine the settling accurately. Especially, if the error signal oscillates or noise is present in the measurements, determining the instant when error is less than some xed value is very dicult and sensitive to variation of parameters. We measure the performance of each control method introduced in Section and compare the methods using the performance measures presented here. It has to be noted that we compare only PI controllers with other PI controllers and PID controllers with other PID controllers. In general, it is reasonable to compare only controllers of same order. This is because theoretically controllers with higher order can be tuned to give at least as good results as controllers of lower order.. Sensitivity to Disturbances The tuning methods might give good results in sense of the performance measures discussed in previous section, but no method is useful if the controlled system is sensitive to modeling errors, disturbances in control signal or measurement noise. System sensitivity analysis is another useful method to measure the quality and adequacy of the controllers. In this case we perform a sensitivity analysis based on the maximum sensitivity of the closed-loop system. First we determine the sensitivity function and complementary sensitivity function for a closed-loop system. Let the controller transfer function be G c (s) and process transfer function G(s). The sensitivity function of the system is then S(s) = + L(s), () where L(s) = G c (s)g(s) is the transfer function of the open-loop system. The complementary sensitivity function can thus be formulated as T (s) = L(s) + L(s). () It should be noted that by denition T (s) + S(s) =.[] The sensitivity function describes the system response sensitivity to reference signal changes, whereas the complementary sensitivity function describes the system sensitivity to measurement noise. Moreover, they describe the system sensitivity to modeling errors which is important to notice in this study.

13 It is shown in [7] and [8] that if the Nyquist Diagram of the open-loop system L(s) and a circle with radius r R and center on negative real axis in c r do not intersect, the system sensitivity function S(s) and complementary sensitivity function T (s) are less than or equal to M, where and r R = M M(M ) c R = M M +, () M(M ) where parameter M >. Thus, the parameter M can be interpreted as a parameter of sensitivity or equivalently robustness that expresses the maximum value of system sensitivity. Smaller values of M make the circle radius r R greater, which takes the Nyquist curve further o from the critical point -+i and makes the system more robust. On the contrary, greater values of M decrease the radius r R and take the Nyquist curve closer to instability region. Typically M values of a robust system are at interval. < M <..[8] Results. Process Model Parameter Estimation The parameter estimation process of the system models described in Section. is performed with Matlab. Because of the nonlinearity of the real process and the inaccuracy of the utilized models it is desired that we choose a small bounded interval as operating interval as explained in Sections. and.. We start the parameter estimation by determining the model parameters' P dependence on the operation point. In the experiment setup we apply a step input of V into process and measure the response. This step response is determined at points in the input range and the parameters P are determined by solving (). The results are shown in Figure 6. Parameter values seem to vary quite signicantly at both ends of the range. At 9V there was a lot of drifting in the measurement data which explains the deviation in the parameter values. However, between input range -7V the parameter values seem to be quite stable and constant. So, it is reasonable to tune the controllers to operate at this interval. Thus, if we do not want varying parameters or operation point dependent parameters, this is the best interval to perform controller tuning with methods that utilize the linear process model. The average values of process parameters at interval -7V are shown in Table. These values are used in the controller tuning that is discussed in the following section. The parameters P of the rst order model (see () ) are estimated utilizing the second order model. Because the second order model approximates the process very well and the parameter values P are quite stable at the chosen interval, it is reasonable to assume that the rst order model also behaves in such manner. To perform the estimation process a suitable Matlab script was written. The script determines automatically the maximum of derivative function and calculates the parameters as described in Section.. The values for parameters P are shown in Table. ()

14 . Controller Parameters Table : Parameters P mean values and parameter P values P P parameter value parameter value K.888 K.888 T a.6 T.68 T b.76 L.87 L.67 Now, parameters for dierent controllers introduced in Section can be calculated. This is performed with the rst and second order models of the process and the values of parameters P and P determined in the previous section. For convenience, we divide the controller tuning methods into two categories. The rst category contains the tuning methods that have constant parameters and the second category contains the methods that have an adjustable tuning parameter. Thus, into the rst category we include Ziegler Nichols step response method and ultimate frequency method, Cohen and Coon method and AMIGO method. Into the second category we include the Lambda tuning method and the IMC tuning method. The controllers in the rst category are tuned rst. We concentrate on PI and PID controllers. The parameters for the controllers in the rst category are shown in Table. To illustrate the eciency of the tuning methods we performed simulations with the process model. This gives a rst hand information of the controller eects to the behavior of the actual process. We perform a more deeper comparison of the methods in the following section. We study the controller behavior in two. K T a [s] T b [s] L [s] Parameter value input operation point [V] Figure 6: P parameter values as function of operation point

15 Table : Parameter values of category one controllers ZN step ZN freq Cohen&Coon AMIGO PI: K p T i PID: K p T i T d...8. cases; the load disturbance and reference signal tracking as mentioned in Section. Response of the closed-loop models to step change in reference signal are shown in Figure 7. Similarly, the closed-loop system responses to a load disturbance are shown in Figure 8. In the second category, parameter values are determined only for Lambda controller. For IMC controller explicit parameter values cannot be determined, because the controller structure and parameter values depend signicantly on the process model structure. Parameter values for Lambda controller are determined such that we use four predetermined values for λ-parameter. These values have been selected with graphical inspection from system step response. The parameter values and corresponding λ values can be seen in Table 6. Table 6: Values of Lambda controller parameters Lambda λ =. λ =. λ =. λ =.6 PI: K p T i.68 PID: K p T i.796 T d.98 Determining the parameter values for IMC controller is more complicated. Even though the IMC controller reduces to PID controller when rst order process model is used, it is not practical to calculate the parameter values explicitly. The closed-loop response of the model when a step reference signal is applied with Lambda tuning and IMC controller is shown in Figures 9 and (a). Here we also use few predetermined values for parameter T l value, which can be seen in the gures. Similarly, the behavior of the system, when a load disturbance (V step change in input signal) is applied is shown in Figures and (b).

16 . PI controllers reference ZN step ZN freq Cohen&Coon. PID controllers reference ZN step ZN freq Cohen&Coon AMIGO (t) (t) u(t) u(t) Figure 7: Output and control signals of the closed-loop systems for dierent tuning methods when a step in reference signal is applied. Category one controllers. (t)..... PI controllers reference ZN step ZN freq Cohen&Coon (t)..... PID controllers reference ZN step ZN freq Cohen&Coon AMIGO u(t) u(t) Figure 8: Output and control signals of the closed-loop systems for dierent tuning methods when a load disturbance is applied. Category one controllers. 6

17 .. =. =. =. =.6 y ref.. =. =. =. =.6 y ref (a) (b) Figure 9: Closed-loop system output and control signals with Lambda tuned (a) PI (b) PID controller when a step of V is applied to reference at t=s..6.. =. =. =. =.6 y ref.6.. =. =. =. =.6 y ref (a) (b) Figure : Closed-loop system output and control signals with Lambda tuned (a) PI (b) PID controller when a load disturbance of V is applied at t=s. 7

18 T l :. T l :. T l :. T l : t t (a) y ref T l :..8 T l :..6 T l :.. T l :.6. y ref 6 8 t t (b) Figure : Closed-loop output and control signals of the IMC controlled system in case of (a) reference tracking (step at t=s) (b) load disturbance rejection (step at t=s). Comparison of Controllers' Performance We determine the controllers' performance with the methods introduced in Section. and Appendix B. As mentioned earlier, it is reasonable to compare only PI controllers with other PI controllers and PID controllers with other PID controllers. It is good to keep in mind that the performance measures determined here, loosely speaking, 'correlate' with the area between the reference signal and the output signal of the process seen in Figures 7 -. First we determine the performance measures for category one PI controllers in the case of reference signal tracking and load disturbance rejection. Results are shown in Figures (a) and (b). The performance values are relative such that the best tuning method has value. and the other values are relative to this. Comparison of PI controllers in the case of reference signal tracking shows that Ziegler Nichols ultimate frequency method gives the best results in our performance tests. As seen in Figure (a) Ziegler Nichols ultimate frequency method beats all other methods with all performance measures. In case of load disturbance the Ziegler Nichols step response method is the best in sense of all performance measures except J opt. The performance measures for category one PID controllers in case of reference signal tracking and load disturbance rejection are shown in Figures (a) and (b). The best PID controller tuning methods in case of reference signal tracking are Ziegler Nichols ultimate frequency method and step response method as shown in Figure (a). They give almost equally good performance measures in case of all other performance measures except J opt. The AMIGO PID controller is the best in case of the J opt performance measure. In case of load disturbance rejection the situation is more complicated as seen in Figure (b). A clearly dominating controller in sense of all performance measures is dicult to determine among Ziegler Nichols and Cohen&Coon methods. Ziegler Nichols Step response method is best in case of absolute 8

19 J_itae:,,8 6,7 J_itae:,,,6 J_iae:,6,, J_iae:,,, J_ise:,,,78 J_ise:,,, J_itse:,,6, J_itse:,,8,6 J_opt:,,79,8 J_opt:,,,,,,,,, 6, 7, ZN step ZN freq Cohen&Coon (a),,,,,,, ZN step ZN freq Cohen&Coon (b) Figure : Relative performance values for category one PI controllers in case of (a) reference signal tracking (b) load disturbance rejection J_itae:,,,6,6 J_itae:,,,8, J_iae:,,,,6 J_iae:,,8,,79 J_ise:,,,,76 J_ise:,,,,99 J_itse:,,,77, J_itse:,,,6,68 J_opt:,,7,9 8,68 J_opt:,,,,,,,,,, 6, 7, 8, 9, ZN step ZN freq Cohen&Coon AMIGO (a),,,,,, 6, ZN step ZN freq Cohen&Coon AMIGO (b) Figure : Relative performance values for category one PID controllers in case of (a) reference signal tracking (b) load disturbance rejection 9

20 error performance measures, but Cohen&Coon seems to be the best in case of squared error performance measures, although, the dierences are small. AMIGO PID controller is not very good in sense of performance measures based on absolute and squared error, but it is dominating in case of the J opt performance measure. Settling s t s of the category one controllers are listed in Appendix C.. The shortest settling s seem to be achieved with Ziegler Nichols methods. In the worst case, the Cohen & Coon method seems to have twice as long settling s as Ziegler Nichols methods. For category two controllers we determined the performance measure values using dierent values for the tunable parameter. Thus, the performance measures were determined with small separation of tunable parameter and then plotted as function of the parameter to illustrate the performance behavior. Performance values for Lambda PI controller in case of reference signal tracking are shown in Figure (a) and the corresponding Lambda PID controller performance values are seen in Figure (b). Performance of the PID controllers seems to be slightly better at all values of λ in sense of all performance measures. Performance values for Lambda PI controller in case of load disturbance are shown in Figure (a) and the corresponding Lambda PID controller performance values are seen in Figure (b). Similarly Lambda PID controller performance values are slightly better for all values of λ. Performance values of the IMC controller are shown in Figures 6 (a) and (b). Compared to the Lambda controllers the performance values seem to increase very rapidly when parameter T l increases but at the same the decrease in the M-value is very gradual. This indicates that the IMC controller is more sensitive to variations of the tunable parameter T l. There also seem not be a similar minimum values for performance measures as the Lambda controllers have in case of reference signal tracking. A summary of performance numbers and settling s is shown in Appendix C... J itae J itae. J iae J iae J ise J itse. J ise J itse. J opt M value J opt M value M / J. M / J λ (a) λ (b) Figure : M values and cost function values (in case of reference signal tracking) as functions of λ for (a) PI controller (b) PID controller

21 6 6 J itae J itae J iae J ise J iae J ise J itse J itse M / J cost J opt M value M / J cost J opt M value lambda (a) lambda (b) Figure : M values and cost function values (in case of load disturbance rejection) as functions of λ for (a) PI controller (b) PID controller. 6 J itae J itae J iae J ise J iae J ise M / J cost.. J itse J opt M value M / J cost J itse J opt M value T f (a) T f (b) Figure 6: M-values and cost function values of IMC controller as functions of parameter T l in case of (a) reference signal tracking and (b) load disturbance rejection

22 . PI controllers. PID controllers Imag[L(jω)]. Imag[L(jω)].. ZN step, M=. ZN freq, M=. Cohen&Coon, M=.... Real[L(jω)] (a). ZN step, M=. ZN freq, M=. Cohen&Coon, M=.6 AMIGO, M= Real[L(jω)] (b) Figure 7: Nyquist diagrams and M-circles for (a) PI and (b) PID controllers. Comparison of Controllers' Sensitivity Sensitivity of controllers is also analyzed separately for category one and category two controllers. For category one controllers the Nyquist diagrams and the M-circles are plotted in Figure 7. Controllers with greater M values (smaller circles) are more sensitive to disturbances and measurement noise than ones with smaller M values (larger circles). For category two controllers the M-values were also determined as a function of tunable parameter. In Figures - the M-value is plotted as a function of λ in the same gure with controller performance function values that were determined in last section. M-values are naturally equal for PI controllers in both cases, the reference signal tracking and the load disturbance rejection. Similarly M-values are equal for PID controllers in both cases as the controllers and the plant are the same. M-values of the IMC controller are plotted together with the controller performance costs as a function of the parameter T f in Figure 6 (a)-(b). The M-curve is same in both gures (a) and (b) but the costs are dierent in the two cases. By plotting the controllers M-values and settling s into same gure gives good picture of the trade-o between robustness and performance. All controllers plotted in M-t s plane are shown in Figure 8. In case of tunable parameter controllers we have chosen the ones with best values by weighting M-value and settling with equal weights. It should be noted that only tuning methods that reach the area of good robustness ( < M ) are the AMIGO method and tunable parameter controllers. However, as mentioned earlier the settling can be considered only as an approximative performance indicator. Furthermore, to compare the theoretical goodness of tuning methods we have plotted the J iae

23 6 Cohen&Coon PI Cohen&Coon PID M-value ZN freq PID IMC Lambda PID ZN step PI ZN step PID ZN freq PI AMIGO Lambda PI Settling [s] Figure 8: Controllers' M-value versus settling. Better controllers are near origo measure in load disturbance rejection case versus the J ise in reference tracking case in Figure C. in Appendix. J iae and J ise are typically used to measure the performance in the case of load disturbance rejection and reference tracking, respectively. Thus, Figure C. illustrates if the controller is more suitable to reference tracking or load dsturbance rejection. The performance measures J iae, J ise and J opt are also plotted versus the M value in Figures C. - C.6 in Appendix. These gures show that in this case the Ziegler Nichols methods seems to be very good in both PI and PID controllers. On the contrary, the Cohen and Coon methods seem to be quite poor in all cases. The tunable methods seems to be better than other methods in case of reference tracking, but in load disturbance rejection they can only reach the performance of the Ziegler Nichols methods. The ranking of AMIGO method is quite tricky. In case of J opt performance measure it is superior and in load disturbance rejection it reaches quite good results, but in reference signal tracking it is very poor. A summary of the M-values of the category one controllers is shown also in Appendix C... Applying Methods to the Heated Airow Process After a thorough study of tuning methods and their applicability to the process model we lack only the knowledge of how the actual process can be controlled with the methods. Next, we perform tests with the process that are based on the results we have found with the process modeling. Implementing rst the controller algorithms into Matlab using xpc Target toolbox, a real- process control and measurement system, we can test the behavior of the controlled process. The response given by the simulation of the second order model and the actual process are plotted in same gure in Appendix

24 C. for all controllers. For tunable parameter controllers we have chosen controllers that are good representatives or compromises of controllers with good performance and robustness. The actual process and simulation results seem to meet each other relatively well, especially in the case of load disturbance rejection. The gures show that the static gain of the system and the output delay were estimated very well. However, in case of reference signal tracking the actual process seems to settle to steady state value faster than the estimated model predicts. It is also noticed that in the case of PID controllers the measurement noise is not ltered very well in the feedback loop and hence the noise in the control signal u(t) is signicantly large. The noise could be reduced with increasing the measurement signal ltering but this would decrease the controller performance. 6 Educational Discussion In this section we discuss the essential guidelines and practical results we have found to be important in developing a laboratory course exercise on PID control for undergraduate students. From the laboratory exercise point of view it is important to give students knowledge about the basic behavior of the closed-loop system dynamics and PID controllers and to give illustrative examples of their behavior. Furthermore, explaining the background and preconditions for usage of tuning methods as well as giving reasons for their usage is necessary to make students understand the conditions in which the usage of PID controllers is useful. The aim is to advise students to use the methods properly and to learn the advantages and drawbacks of the PID controllers. However, it should be kept in mind that introducing too many methods and controllers might be confusing and make it dicult to distinguish relevant issues from rather irrelevant issues. Students are expected to know the basics of the process dynamics and they most probably also know something about the PID controllers. For example the eects of measurement delay, especially in a closed-loop process and the shortcoming of a simple P controller which leaves a constant error between process output and reference signal. Some student might also be familiar with the eects that appear in the closed-loop system output when the controller parameters are varied. Particularly, greater P-term makes the closed-loop system more rapid, the I-term can be used to overcome the problems related to simple P controller, but it increases the oscillation of the output signal and the oscillations (caused by I-term and measurement delay) can be reduced with the D-term. It is important to distinguish the variation of parameters K p, T i, T d and N and terms (P, I and D) because varying K p aects all terms as explained in Section. However, these eects can be demonstrated with the computer controlled process very easily and the eects of changing each parameter separately can be powerfully visualized. Also the possibility to vary certain process parameters, measurement delay and airow speed, makes the demonstration of process dynamics more versatile. The process identication is the rst part of the tuning methods' application. In this work we have studied rst and second order models. These models explain the process behavior suciently for

25 controller tuning purposes. But, it should be noted that these models are quite simple and do not take into account for instance process parameter variation. This means that the process model must be re-estimated in case some of the process parameters change. The parameter tuning of a basic PID controller can be performed with the methods introduced in this work. These tuning methods provide a good starting point for students to learn the PID controller. However, it should be noticed that the PID tuning methods give only a good approximative values for controller parameters. These values should be used mainly as a good initial estimate for parameters rather than optimal or nal values. Hence the manual adjustment of parameters is at least as important part of the controller tuning as the application of the tuning methods. Actually the manual adjustment could be considered as the eventual ne tuning method that can be used to set the controller to behave as desired. Nevertheless, incorporating the robustness measures into the analysis in the simulation phase, as was done in this study, provides deeper understanding of dierent methods and may even decrease the need for manual ne tuning when implementing the controllers. As we found out in this project work, none of the basic tuning methods is superior to others in sense of both robustness and performance. Thus the relation between these properties requires clarication and students should understand the trade-o between the robustness and performance. In other words, it has to be emphasized that increasing the performance of a controller always makes the system more sensitive. This is because the performance is achieved by strengthening the feedback which makes the system behavior more rapid. With tunable parameter controllers it is easy to demonstrate the change in controller behavior, because the tuning can be reduced into changing one parameter. One of the most noticeable characteristics of the case process is the noise in temperature measurement. This is mainly caused by the turbulence in the owing air which is not included into the system model as mentioned in Section.. Noise is of course an undesired property in any part of a process, because it decreases the measurement accuracy and it has a destabilizing eect especially if the process includes feedbacks. Thus, dierent noise ltering methods can be used in the case process and the eects are very observable because the measurement noise of the case process is substantially large. The students are expected to have some knowledge of the concept of noise and probably the means to reduce it. At least they should know that noise is a high frequency variation in the measurement signal and they should be able to distinguish noisy measurement signal from a less noisy measurement. In the case of a simple PID controller () the noise aects the D-term signicantly. It is easy to visualize that high frequency in signal means also high frequency in the derivative signal. So, including the D-term into a well functional PI controller can cause a large noise in the controller output signal u(t) and can lead even to instability of the process. Low-pass ltering of the measurement signal provides a solution to this problem. The controllers introduced in this work contain the low-pass lter. However, the tuning methods do not determine any values to lter parameters except the AMIGO method that determines an approximative value for T f. This is because the strength of ltering depends on the existing noise and should be determined in all situations according to the characteristic process noise.

26 In tunable parameter controllers the ltering is included into controller and changes when adjusting the tunable parameter. So, including a proper lter increases the controller robustness but at the same it makes the system response slower. In Section. we already divided the studied controllers into two categories. This division is very good starting point in teaching the methods, although it is not yet necessary to decide the methods that should be nally used in the laboratory exercise. Category one tuning methods are very good to illustrate the behavior and dierences between the tuning methods. By illustrating both of the cases, the reference signal tracking and load disturbance rejection, it is evident for students that none of the methods is superior to others in sense of both performance and robustness. This could also be demonstrated with the robustness-performance maps shown in Appendix C.. Category two controllers have a little more complex structure, but they are very useful in demonstrating the eects of controller tuning on the robustness and performance. This is because the controller tuning nally reduces to varying one parameter which determines the controller position in robustness-performance map. 7 Summary and Conclusions In this research project we have performed a comprehensive study over the PID controller tuning on a heated airow process. Motivation for the study was the demand to develop a laboratory course exercise for undergraduate students. Several controllers and controller tuning methods described in this work were found out to be suitable for controlling the case process. Comparison of controllers performance and robustness proved that none of the controllers were superior to others. Thus, dierent controllers and tuning methods should be used in dierent situations. Measurement noise and process parameters variation also aect the controller performance and robustness. Controllers with high performance are usually sensitive to noise. Trade-o between robustness and performance makes the decision making among controllers more dicult. The prospective development of the laboratory exercise will concentrate on writing the exercise instructions and implementing the measurement software. Results of this work will be used to guide the determination of the exercise instructions. Especially this study gives basis for examination of the controllers robustness and performance. During the study of the tuning methods and process model estimation we have also implemented several Matlab scripts and Simulink models to perform the model parameter estimation automatically. These will be utilized in the measurement software development. 6

27 References [] Dorf R. C. and Bishop R. H., Modern Control Systems, Ninth Edition, Prentice Hall, [] Eriksson L., Control Design and Implementation of Networked Control Systems, Licentiate thesis, Department of Automation and Systems Technology, Helsinki University of Technology, May 7. [] Garcia C. E. and Morari M., Internal Model Control.. A Unifying Review and Some New Results, Industrial & Engineering Chemistry Process Design and Development, vol., pp. 8-, 98 [] O'Dwyer, A., Handbook of PI and PID Controller Tuning Rules, Imperial College Press, [] Visioli A., Practical PID Control, Springer, 6 [6] Åström K.J. and Hägglund T., Automatic Tuning of PID Controllers, Instrument Society of America, 988 [7] Åström K.J., Panagopoulos H. and Hägglund T., Design of PI Controllers based on Non-Convex Optimization, Automatica, Vol., No., pp. 8-6, 998 [8] Åström K.J. and Hägglund T., Revisiting the Ziegler-Nichols step response method for PID control, Journal of Process Control () 6-6, Elsevier, [9] Process Control Workshop Advanced Teaching Manual 7--M, Feedeback Instruments Ltd. 7

EECE 460 : Control System Design

EECE 460 : Control System Design EECE 460 : Control System Design PID Controller Design and Tuning Guy A. Dumont UBC EECE January 2012 Guy A. Dumont (UBC EECE) EECE 460 PID Tuning January 2012 1 / 37 Contents 1 Introduction 2 Control

More information

SAMPLE CHAPTERS UNESCO EOLSS PID CONTROL. Araki M. Kyoto University, Japan

SAMPLE CHAPTERS UNESCO EOLSS PID CONTROL. Araki M. Kyoto University, Japan PID CONTROL Araki M. Kyoto University, Japan Keywords: feedback control, proportional, integral, derivative, reaction curve, process with self-regulation, integrating process, process model, steady-state

More information

A simple method to determine control valve performance and its impacts on control loop performance

A simple method to determine control valve performance and its impacts on control loop performance A simple method to determine control valve performance and its impacts on control loop performance Keywords Michel Ruel p.eng., Top Control Inc. Process optimization, tuning, stiction, hysteresis, backlash,

More information

Dr. Yeffry Handoko Putra, S.T., M.T

Dr. Yeffry Handoko Putra, S.T., M.T Tuning Methods of PID Controller Dr. Yeffry Handoko Putra, S.T., M.T yeffry@unikom.ac.id 1 Session Outlines & Objectives Outlines Tuning methods of PID controller: Ziegler-Nichols Open-loop Coon-Cohen

More information

PID Control. Chapter 10

PID Control. Chapter 10 Chapter PID Control Based on a survey of over eleven thousand controllers in the refining, chemicals and pulp and paper industries, 97% of regulatory controllers utilize PID feedback. Desborough Honeywell,

More information

Positive Feedback and Oscillators

Positive Feedback and Oscillators Physics 3330 Experiment #6 Fall 1999 Positive Feedback and Oscillators Purpose In this experiment we will study how spontaneous oscillations may be caused by positive feedback. You will construct an active

More information

stable response to load disturbances, e.g., an exothermic reaction.

stable response to load disturbances, e.g., an exothermic reaction. C REACTOR TEMPERATURE control typically is very important to product quality, production rate and operating costs. With continuous reactors, the usual objectives are to: hold temperature within a certain

More information

PID Control. 6.1 Introduction

PID Control. 6.1 Introduction 6 PID Control 6. Introduction The PID controller is the most common form of feedback. It was an essential element of early governors and it became the standard tool when process control emerged in the

More information

Autotuning of a PID-controller

Autotuning of a PID-controller ISSN 0280-5316 ISRN LUTFD2/TFRT--5728--SE Autotuning of a PID-controller Camilla Andersson Mirjam Lindberg Department of Automatic Control Lund Institute of Technology October 2004 Department of Automatic

More information

Σ _. Feedback Amplifiers: One and Two Pole cases. Negative Feedback:

Σ _. Feedback Amplifiers: One and Two Pole cases. Negative Feedback: Feedback Amplifiers: One and Two Pole cases Negative Feedback: Σ _ a f There must be 180 o phase shift somewhere in the loop. This is often provided by an inverting amplifier or by use of a differential

More information

AC 2012-4561: MATHEMATICAL MODELING AND SIMULATION US- ING LABVIEW AND LABVIEW MATHSCRIPT

AC 2012-4561: MATHEMATICAL MODELING AND SIMULATION US- ING LABVIEW AND LABVIEW MATHSCRIPT AC 2012-4561: MATHEMATICAL MODELING AND SIMULATION US- ING LABVIEW AND LABVIEW MATHSCRIPT Dr. Nikunja Swain, South Carolina State University Nikunja Swain is a professor in the College of Science, Mathematics,

More information

QNET Experiment #06: HVAC Proportional- Integral (PI) Temperature Control Heating, Ventilation, and Air Conditioning Trainer (HVACT)

QNET Experiment #06: HVAC Proportional- Integral (PI) Temperature Control Heating, Ventilation, and Air Conditioning Trainer (HVACT) Quanser NI-ELVIS Trainer (QNET) Series: QNET Experiment #06: HVAC Proportional- Integral (PI) Temperature Control Heating, Ventilation, and Air Conditioning Trainer (HVACT) Student Manual Table of Contents

More information

HITACHI INVERTER SJ/L100/300 SERIES PID CONTROL USERS GUIDE

HITACHI INVERTER SJ/L100/300 SERIES PID CONTROL USERS GUIDE HITACHI INVERTER SJ/L1/3 SERIES PID CONTROL USERS GUIDE After reading this manual, keep it for future reference Hitachi America, Ltd. HAL1PID CONTENTS 1. OVERVIEW 3 2. PID CONTROL ON SJ1/L1 INVERTERS 3

More information

A Design of a PID Self-Tuning Controller Using LabVIEW

A Design of a PID Self-Tuning Controller Using LabVIEW Journal of Software Engineering and Applications, 2011, 4, 161-171 doi:10.4236/jsea.2011.43018 Published Online March 2011 (http://www.scirp.org/journal/jsea) 161 A Design of a PID Self-Tuning Controller

More information

G(s) = Y (s)/u(s) In this representation, the output is always the Transfer function times the input. Y (s) = G(s)U(s).

G(s) = Y (s)/u(s) In this representation, the output is always the Transfer function times the input. Y (s) = G(s)U(s). Transfer Functions The transfer function of a linear system is the ratio of the Laplace Transform of the output to the Laplace Transform of the input, i.e., Y (s)/u(s). Denoting this ratio by G(s), i.e.,

More information

Controller Design in Frequency Domain

Controller Design in Frequency Domain ECSE 4440 Control System Engineering Fall 2001 Project 3 Controller Design in Frequency Domain TA 1. Abstract 2. Introduction 3. Controller design in Frequency domain 4. Experiment 5. Colclusion 1. Abstract

More information

054414 PROCESS CONTROL SYSTEM DESIGN. 054414 Process Control System Design. LECTURE 6: SIMO and MISO CONTROL

054414 PROCESS CONTROL SYSTEM DESIGN. 054414 Process Control System Design. LECTURE 6: SIMO and MISO CONTROL 05444 Process Control System Design LECTURE 6: SIMO and MISO CONTROL Daniel R. Lewin Department of Chemical Engineering Technion, Haifa, Israel 6 - Introduction This part of the course explores opportunities

More information

Microcontroller-based experiments for a control systems course in electrical engineering technology

Microcontroller-based experiments for a control systems course in electrical engineering technology Microcontroller-based experiments for a control systems course in electrical engineering technology Albert Lozano-Nieto Penn State University, Wilkes-Barre Campus, Lehman, PA, USA E-mail: AXL17@psu.edu

More information

FUZZY Based PID Controller for Speed Control of D.C. Motor Using LabVIEW

FUZZY Based PID Controller for Speed Control of D.C. Motor Using LabVIEW FUZZY Based PID Controller for Speed Control of D.C. Motor Using LabVIEW SALIM, JYOTI OHRI Department of Electrical Engineering National Institute of Technology Kurukshetra INDIA salimnitk@gmail.com ohrijyoti@rediffmail.com

More information

Control System Definition

Control System Definition Control System Definition A control system consist of subsytems and processes (or plants) assembled for the purpose of controlling the outputs of the process. For example, a furnace produces heat as a

More information

DIGITAL-TO-ANALOGUE AND ANALOGUE-TO-DIGITAL CONVERSION

DIGITAL-TO-ANALOGUE AND ANALOGUE-TO-DIGITAL CONVERSION DIGITAL-TO-ANALOGUE AND ANALOGUE-TO-DIGITAL CONVERSION Introduction The outputs from sensors and communications receivers are analogue signals that have continuously varying amplitudes. In many systems

More information

PID Controller Tuning: A Short Tutorial

PID Controller Tuning: A Short Tutorial PID Controller Tuning: A Short Tutorial Jinghua Zhong Mechanical Engineering, Purdue University Spring, 2006 Outline This tutorial is in PDF format with navigational control. You may press SPACE or, or

More information

Op-Amp Simulation EE/CS 5720/6720. Read Chapter 5 in Johns & Martin before you begin this assignment.

Op-Amp Simulation EE/CS 5720/6720. Read Chapter 5 in Johns & Martin before you begin this assignment. Op-Amp Simulation EE/CS 5720/6720 Read Chapter 5 in Johns & Martin before you begin this assignment. This assignment will take you through the simulation and basic characterization of a simple operational

More information

Frequency Response of Filters

Frequency Response of Filters School of Engineering Department of Electrical and Computer Engineering 332:224 Principles of Electrical Engineering II Laboratory Experiment 2 Frequency Response of Filters 1 Introduction Objectives To

More information

SINGLE-SUPPLY OPERATION OF OPERATIONAL AMPLIFIERS

SINGLE-SUPPLY OPERATION OF OPERATIONAL AMPLIFIERS SINGLE-SUPPLY OPERATION OF OPERATIONAL AMPLIFIERS One of the most common applications questions on operational amplifiers concerns operation from a single supply voltage. Can the model OPAxyz be operated

More information

Vi, fi input. Vphi output VCO. Vosc, fosc. voltage-controlled oscillator

Vi, fi input. Vphi output VCO. Vosc, fosc. voltage-controlled oscillator Experiment #4 CMOS 446 Phase-Locked Loop c 1997 Dragan Maksimovic Department of Electrical and Computer Engineering University of Colorado, Boulder The purpose of this lab assignment is to introduce operating

More information

Transistor amplifiers: Biasing and Small Signal Model

Transistor amplifiers: Biasing and Small Signal Model Transistor amplifiers: iasing and Small Signal Model Transistor amplifiers utilizing JT or FT are similar in design and analysis. Accordingly we will discuss JT amplifiers thoroughly. Then, similar FT

More information

Understanding Poles and Zeros

Understanding Poles and Zeros MASSACHUSETTS INSTITUTE OF TECHNOLOGY DEPARTMENT OF MECHANICAL ENGINEERING 2.14 Analysis and Design of Feedback Control Systems Understanding Poles and Zeros 1 System Poles and Zeros The transfer function

More information

Matlab and Simulink. Matlab and Simulink for Control

Matlab and Simulink. Matlab and Simulink for Control Matlab and Simulink for Control Automatica I (Laboratorio) 1/78 Matlab and Simulink CACSD 2/78 Matlab and Simulink for Control Matlab introduction Simulink introduction Control Issues Recall Matlab design

More information

Chapter 19 Operational Amplifiers

Chapter 19 Operational Amplifiers Chapter 19 Operational Amplifiers The operational amplifier, or op-amp, is a basic building block of modern electronics. Op-amps date back to the early days of vacuum tubes, but they only became common

More information

Measurement Products. PID control theory made easy Optimising plant performance with modern process controllers

Measurement Products. PID control theory made easy Optimising plant performance with modern process controllers Measurement Products PID control theory made easy Optimising plant performance with modern process controllers 1. What are the main types of control applications? Introduction Accurate control is critical

More information

Process Control Primer

Process Control Primer Process Control Primer At the onset of the Industrial Revolution, processes were controlled manually. Men turned valves, pulled levers or changed switches based on the need to turn devices on or off. As

More information

Student Performance Q&A:

Student Performance Q&A: Student Performance Q&A: 2008 AP Calculus AB and Calculus BC Free-Response Questions The following comments on the 2008 free-response questions for AP Calculus AB and Calculus BC were written by the Chief

More information

PID control - Simple tuning methods

PID control - Simple tuning methods - Simple tuning methods Ulf Holmberg Introduction Lab processes Control System Dynamical System Step response model Self-oscillation model PID structure Step response method (Ziegler-Nichols) Self-oscillation

More information

dspace DSP DS-1104 based State Observer Design for Position Control of DC Servo Motor

dspace DSP DS-1104 based State Observer Design for Position Control of DC Servo Motor dspace DSP DS-1104 based State Observer Design for Position Control of DC Servo Motor Jaswandi Sawant, Divyesh Ginoya Department of Instrumentation and control, College of Engineering, Pune. ABSTRACT This

More information

Chapter 9: Controller design

Chapter 9: Controller design Chapter 9. Controller Design 9.1. Introduction 9.2. Effect of negative feedback on the network transfer functions 9.2.1. Feedback reduces the transfer function from disturbances to the output 9.2.2. Feedback

More information

EE 402 RECITATION #13 REPORT

EE 402 RECITATION #13 REPORT MIDDLE EAST TECHNICAL UNIVERSITY EE 402 RECITATION #13 REPORT LEAD-LAG COMPENSATOR DESIGN F. Kağan İPEK Utku KIRAN Ç. Berkan Şahin 5/16/2013 Contents INTRODUCTION... 3 MODELLING... 3 OBTAINING PTF of OPEN

More information

Eigenvalues, Eigenvectors, and Differential Equations

Eigenvalues, Eigenvectors, and Differential Equations Eigenvalues, Eigenvectors, and Differential Equations William Cherry April 009 (with a typo correction in November 05) The concepts of eigenvalue and eigenvector occur throughout advanced mathematics They

More information

Implementation of Fuzzy and PID Controller to Water Level System using LabView

Implementation of Fuzzy and PID Controller to Water Level System using LabView Implementation of Fuzzy and PID Controller to Water Level System using LabView Laith Abed Sabri, Ph.D University of Baghdad AL-Khwarizmi college of Engineering Hussein Ahmed AL-Mshat University of Baghdad

More information

Roots of Equations (Chapters 5 and 6)

Roots of Equations (Chapters 5 and 6) Roots of Equations (Chapters 5 and 6) Problem: given f() = 0, find. In general, f() can be any function. For some forms of f(), analytical solutions are available. However, for other functions, we have

More information

INTERACTIVE LEARNING MODULES FOR PID CONTROL

INTERACTIVE LEARNING MODULES FOR PID CONTROL INTERACTIVE LEARNING MODULES FOR PID CONTROL José Luis Guzmán, Karl J. Åström, Sebastián Dormido, Tore Hägglund, Yves Piguet Dep. de Lenguajes y Computación, Universidad de Almería, 04120 Almería, Spain.

More information

Module 2 Introduction to SIMULINK

Module 2 Introduction to SIMULINK Module 2 Introduction to SIMULINK Although the standard MATLAB package is useful for linear systems analysis, SIMULINK is far more useful for control system simulation. SIMULINK enables the rapid construction

More information

Section 5.0 : Horn Physics. By Martin J. King, 6/29/08 Copyright 2008 by Martin J. King. All Rights Reserved.

Section 5.0 : Horn Physics. By Martin J. King, 6/29/08 Copyright 2008 by Martin J. King. All Rights Reserved. Section 5. : Horn Physics Section 5. : Horn Physics By Martin J. King, 6/29/8 Copyright 28 by Martin J. King. All Rights Reserved. Before discussing the design of a horn loaded loudspeaker system, it is

More information

Chapter 10. Key Ideas Correlation, Correlation Coefficient (r),

Chapter 10. Key Ideas Correlation, Correlation Coefficient (r), Chapter 0 Key Ideas Correlation, Correlation Coefficient (r), Section 0-: Overview We have already explored the basics of describing single variable data sets. However, when two quantitative variables

More information

Oscillations of the Sending Window in Compound TCP

Oscillations of the Sending Window in Compound TCP Oscillations of the Sending Window in Compound TCP Alberto Blanc 1, Denis Collange 1, and Konstantin Avrachenkov 2 1 Orange Labs, 905 rue Albert Einstein, 06921 Sophia Antipolis, France 2 I.N.R.I.A. 2004

More information

Industrial Steam System Process Control Schemes

Industrial Steam System Process Control Schemes Industrial Steam System Process Control Schemes This paper was developed to provide a basic understanding of the different process control schemes used in a typical steam system. This is however a fundamental

More information

Drivetech, Inc. Innovations in Motor Control, Drives, and Power Electronics

Drivetech, Inc. Innovations in Motor Control, Drives, and Power Electronics Drivetech, Inc. Innovations in Motor Control, Drives, and Power Electronics Dal Y. Ohm, Ph.D. - President 25492 Carrington Drive, South Riding, Virginia 20152 Ph: (703) 327-2797 Fax: (703) 327-2747 ohm@drivetechinc.com

More information

PC BASED PID TEMPERATURE CONTROLLER

PC BASED PID TEMPERATURE CONTROLLER PC BASED PID TEMPERATURE CONTROLLER R. Nisha * and K.N. Madhusoodanan Dept. of Instrumentation, Cochin University of Science and Technology, Cochin 22, India ABSTRACT: A simple and versatile PC based Programmable

More information

This unit will lay the groundwork for later units where the students will extend this knowledge to quadratic and exponential functions.

This unit will lay the groundwork for later units where the students will extend this knowledge to quadratic and exponential functions. Algebra I Overview View unit yearlong overview here Many of the concepts presented in Algebra I are progressions of concepts that were introduced in grades 6 through 8. The content presented in this course

More information

Lambda Tuning the Universal Method for PID Controllers in Process Control

Lambda Tuning the Universal Method for PID Controllers in Process Control Lambda Tuning the Universal Method for PID Controllers in Process Control Lambda tuning gives non-oscillatory response with the response time (Lambda) required by the plant. Seven industrial examples show

More information

19 LINEAR QUADRATIC REGULATOR

19 LINEAR QUADRATIC REGULATOR 19 LINEAR QUADRATIC REGULATOR 19.1 Introduction The simple form of loopshaping in scalar systems does not extend directly to multivariable (MIMO) plants, which are characterized by transfer matrices instead

More information

A Study of Speed Control of PMDC Motor Using Auto-tuning of PID Controller through LabVIEW

A Study of Speed Control of PMDC Motor Using Auto-tuning of PID Controller through LabVIEW A Study of Speed Control of PMDC Motor Using Auto-tuning of PID Controller through LabVIEW Priyanka Rajput and Dr. K.K. Tripathi Department of Electronics and Communication Engineering, Ajay Kumar Garg

More information

Enhancing the SNR of the Fiber Optic Rotation Sensor using the LMS Algorithm

Enhancing the SNR of the Fiber Optic Rotation Sensor using the LMS Algorithm 1 Enhancing the SNR of the Fiber Optic Rotation Sensor using the LMS Algorithm Hani Mehrpouyan, Student Member, IEEE, Department of Electrical and Computer Engineering Queen s University, Kingston, Ontario,

More information

1 Error in Euler s Method

1 Error in Euler s Method 1 Error in Euler s Method Experience with Euler s 1 method raises some interesting questions about numerical approximations for the solutions of differential equations. 1. What determines the amount of

More information

Root Locus. E(s) K. R(s) C(s) 1 s(s+a) Consider the closed loop transfer function:

Root Locus. E(s) K. R(s) C(s) 1 s(s+a) Consider the closed loop transfer function: Consider the closed loop transfer function: Root Locus R(s) + - E(s) K 1 s(s+a) C(s) How do the poles of the closed-loop system change as a function of the gain K? The closed-loop transfer function is:

More information

Session 7 Bivariate Data and Analysis

Session 7 Bivariate Data and Analysis Session 7 Bivariate Data and Analysis Key Terms for This Session Previously Introduced mean standard deviation New in This Session association bivariate analysis contingency table co-variation least squares

More information

Basic Op Amp Circuits

Basic Op Amp Circuits Basic Op Amp ircuits Manuel Toledo INEL 5205 Instrumentation August 3, 2008 Introduction The operational amplifier (op amp or OA for short) is perhaps the most important building block for the design of

More information

1. s to Z-Domain Transfer Function

1. s to Z-Domain Transfer Function 1. s to Z-Domain Transfer Function 1. s to Z-Domain Transfer Function Discrete ZOH Signals 1. s to Z-Domain Transfer Function Discrete ZOH Signals 1. Get step response of continuous transfer function y

More information

POTENTIAL OF STATE-FEEDBACK CONTROL FOR MACHINE TOOLS DRIVES

POTENTIAL OF STATE-FEEDBACK CONTROL FOR MACHINE TOOLS DRIVES POTENTIAL OF STATE-FEEDBACK CONTROL FOR MACHINE TOOLS DRIVES L. Novotny 1, P. Strakos 1, J. Vesely 1, A. Dietmair 2 1 Research Center of Manufacturing Technology, CTU in Prague, Czech Republic 2 SW, Universität

More information

Correlation key concepts:

Correlation key concepts: CORRELATION Correlation key concepts: Types of correlation Methods of studying correlation a) Scatter diagram b) Karl pearson s coefficient of correlation c) Spearman s Rank correlation coefficient d)

More information

ECE 3510 Final given: Spring 11

ECE 3510 Final given: Spring 11 ECE 50 Final given: Spring This part of the exam is Closed book, Closed notes, No Calculator.. ( pts) For each of the time-domain signals shown, draw the poles of the signal's Laplace transform on the

More information

APPLICATION BULLETIN

APPLICATION BULLETIN APPLICATION BULLETIN Mailing Address: PO Box 11400, Tucson, AZ 85734 Street Address: 6730 S. Tucson Blvd., Tucson, AZ 85706 Tel: (520) 746-1111 Telex: 066-6491 FAX (520) 889-1510 Product Info: (800) 548-6132

More information

Modeling, Analysis, and Control of Dynamic Systems

Modeling, Analysis, and Control of Dynamic Systems Modeling, Analysis, and Control of Dynamic Systems Second Edition William J. Palm III University of Rhode Island John Wiley Sons, Inc. New York Chichester Weinheim Brisbane Singapore Toronto To Louise.

More information

FAST METHODS FOR SLOW LOOPS: TUNE YOUR TEMPERATURE CONTROLS IN 15 MINUTES

FAST METHODS FOR SLOW LOOPS: TUNE YOUR TEMPERATURE CONTROLS IN 15 MINUTES FAST METHODS FOR SLOW LOOPS: TUNE YOUR TEMPERATURE CONTROLS IN 15 MINUTES Michel Ruel P.E. President, TOP Control Inc 4734 Sonseeahray Drive 49, Bel-Air St, #103 Hubertus, WI 53033 Levis Qc G6W 6K9 USA

More information

MATLAB Control System Toolbox Root Locus Design GUI

MATLAB Control System Toolbox Root Locus Design GUI MATLAB Control System Toolbox Root Locus Design GUI MATLAB Control System Toolbox contains two Root Locus design GUI, sisotool and rltool. These are two interactive design tools for the analysis and design

More information

TCOM 370 NOTES 99-4 BANDWIDTH, FREQUENCY RESPONSE, AND CAPACITY OF COMMUNICATION LINKS

TCOM 370 NOTES 99-4 BANDWIDTH, FREQUENCY RESPONSE, AND CAPACITY OF COMMUNICATION LINKS TCOM 370 NOTES 99-4 BANDWIDTH, FREQUENCY RESPONSE, AND CAPACITY OF COMMUNICATION LINKS 1. Bandwidth: The bandwidth of a communication link, or in general any system, was loosely defined as the width of

More information

School of Engineering Department of Electrical and Computer Engineering

School of Engineering Department of Electrical and Computer Engineering 1 School of Engineering Department of Electrical and Computer Engineering 332:223 Principles of Electrical Engineering I Laboratory Experiment #4 Title: Operational Amplifiers 1 Introduction Objectives

More information

The front end of the receiver performs the frequency translation, channel selection and amplification of the signal.

The front end of the receiver performs the frequency translation, channel selection and amplification of the signal. Many receivers must be capable of handling a very wide range of signal powers at the input while still producing the correct output. This must be done in the presence of noise and interference which occasionally

More information

EDUMECH Mechatronic Instructional Systems. Ball on Beam System

EDUMECH Mechatronic Instructional Systems. Ball on Beam System EDUMECH Mechatronic Instructional Systems Ball on Beam System Product of Shandor Motion Systems Written by Robert Hirsch Ph.D. 998-9 All Rights Reserved. 999 Shandor Motion Systems, Ball on Beam Instructional

More information

Creating, Solving, and Graphing Systems of Linear Equations and Linear Inequalities

Creating, Solving, and Graphing Systems of Linear Equations and Linear Inequalities Algebra 1, Quarter 2, Unit 2.1 Creating, Solving, and Graphing Systems of Linear Equations and Linear Inequalities Overview Number of instructional days: 15 (1 day = 45 60 minutes) Content to be learned

More information

Reactive Power Control of an Alternator with Static Excitation System Connected to a Network

Reactive Power Control of an Alternator with Static Excitation System Connected to a Network Reactive Power Control of an Alternator with Static Excitation System Connected to a Network Dr. Dhiya Ali Al-Nimma Assist. Prof. Mosul Unoversity Dr. Majid Salim Matti lecturer Mosul University Abstract

More information

Real-time simulation environment for control loop performance monitoring using Matlab

Real-time simulation environment for control loop performance monitoring using Matlab Real-time simulation environment for control loop performance monitoring using Matlab Riku Pöllänen, Ari Koponen, Manu Huttunen, Olli Pyrhönen Department of Electrical Engineering Lappeenranta University

More information

Introduction to SMPS Control Techniques

Introduction to SMPS Control Techniques Introduction to SMPS Control Techniques 2006 Microchip Technology Incorporated. All Rights Reserved. Introduction to SMPS Control Techniques Slide 1 Welcome to the Introduction to SMPS Control Techniques

More information

Introduction to time series analysis

Introduction to time series analysis Introduction to time series analysis Margherita Gerolimetto November 3, 2010 1 What is a time series? A time series is a collection of observations ordered following a parameter that for us is time. Examples

More information

Sound absorption and acoustic surface impedance

Sound absorption and acoustic surface impedance Sound absorption and acoustic surface impedance CHRISTER HEED SD2165 Stockholm October 2008 Marcus Wallenberg Laboratoriet för Ljud- och Vibrationsforskning Sound absorption and acoustic surface impedance

More information

INTERFERENCE OF SOUND WAVES

INTERFERENCE OF SOUND WAVES 1/2016 Sound 1/8 INTERFERENCE OF SOUND WAVES PURPOSE: To measure the wavelength, frequency, and propagation speed of ultrasonic sound waves and to observe interference phenomena with ultrasonic sound waves.

More information

In mathematics, there are four attainment targets: using and applying mathematics; number and algebra; shape, space and measures, and handling data.

In mathematics, there are four attainment targets: using and applying mathematics; number and algebra; shape, space and measures, and handling data. MATHEMATICS: THE LEVEL DESCRIPTIONS In mathematics, there are four attainment targets: using and applying mathematics; number and algebra; shape, space and measures, and handling data. Attainment target

More information

Classification of Fingerprints. Sarat C. Dass Department of Statistics & Probability

Classification of Fingerprints. Sarat C. Dass Department of Statistics & Probability Classification of Fingerprints Sarat C. Dass Department of Statistics & Probability Fingerprint Classification Fingerprint classification is a coarse level partitioning of a fingerprint database into smaller

More information

Best Practices for Controller Tuning

Best Practices for Controller Tuning Best Practices for Controller Tuning George Buckbee, P.E. ExperTune, Inc. 2009 ExperTune, Inc. Page 1 Best Practices for Controller Tuning George Buckbee, P.E., ExperTune Inc. 2009 ExperTune Inc Summary

More information

Review of Fundamental Mathematics

Review of Fundamental Mathematics Review of Fundamental Mathematics As explained in the Preface and in Chapter 1 of your textbook, managerial economics applies microeconomic theory to business decision making. The decision-making tools

More information

APPLICATION NOTE ULTRASONIC CERAMIC TRANSDUCERS

APPLICATION NOTE ULTRASONIC CERAMIC TRANSDUCERS APPLICATION NOTE ULTRASONIC CERAMIC TRANSDUCERS Selection and use of Ultrasonic Ceramic Transducers The purpose of this application note is to aid the user in the selection and application of the Ultrasonic

More information

Time Response Analysis of DC Motor using Armature Control Method and Its Performance Improvement using PID Controller

Time Response Analysis of DC Motor using Armature Control Method and Its Performance Improvement using PID Controller Available online www.ejaet.com European Journal of Advances in Engineering and Technology, 5, (6): 56-6 Research Article ISSN: 394-658X Time Response Analysis of DC Motor using Armature Control Method

More information

Designing Fluctronic Real-Time Systems

Designing Fluctronic Real-Time Systems Journal of Real-Time Systems, Special Issue on Control-Theoretical Approaches to Real-Time Computing Feedback Control Real-Time Scheduling: Framework, Modeling, and Algorithms * Chenyang Lu John A. Stankovic

More information

The Effects of Start Prices on the Performance of the Certainty Equivalent Pricing Policy

The Effects of Start Prices on the Performance of the Certainty Equivalent Pricing Policy BMI Paper The Effects of Start Prices on the Performance of the Certainty Equivalent Pricing Policy Faculty of Sciences VU University Amsterdam De Boelelaan 1081 1081 HV Amsterdam Netherlands Author: R.D.R.

More information

Enhancing Classical Controls Education via Interactive GUI Design

Enhancing Classical Controls Education via Interactive GUI Design C Enhancing Classical Controls Education via Interactive GUI Design Raymond C. Garcia and Bonnie S. Heck lassical control design methods reg- tions is manually; that is. the students plot options available,

More information

So far we have investigated combinational logic for which the output of the logic devices/circuits depends only on the present state of the inputs.

So far we have investigated combinational logic for which the output of the logic devices/circuits depends only on the present state of the inputs. equential Logic o far we have investigated combinational logic for which the output of the logic devices/circuits depends only on the present state of the inputs. In sequential logic the output of the

More information

Loop Analysis. Chapter 7. 7.1 Introduction

Loop Analysis. Chapter 7. 7.1 Introduction Chapter 7 Loop Analysis Quotation Authors, citation. This chapter describes how stability and robustness can be determined by investigating how sinusoidal signals propagate around the feedback loop. The

More information

DCMS DC MOTOR SYSTEM User Manual

DCMS DC MOTOR SYSTEM User Manual DCMS DC MOTOR SYSTEM User Manual release 1.3 March 3, 2011 Disclaimer The developers of the DC Motor System (hardware and software) have used their best efforts in the development. The developers make

More information

Formulations of Model Predictive Control. Dipartimento di Elettronica e Informazione

Formulations of Model Predictive Control. Dipartimento di Elettronica e Informazione Formulations of Model Predictive Control Riccardo Scattolini Riccardo Scattolini Dipartimento di Elettronica e Informazione Impulse and step response models 2 At the beginning of the 80, the early formulations

More information

C21 Model Predictive Control

C21 Model Predictive Control C21 Model Predictive Control Mark Cannon 4 lectures Hilary Term 216-1 Lecture 1 Introduction 1-2 Organisation 4 lectures: week 3 week 4 { Monday 1-11 am LR5 Thursday 1-11 am LR5 { Monday 1-11 am LR5 Thursday

More information

Evaluating System Suitability CE, GC, LC and A/D ChemStation Revisions: A.03.0x- A.08.0x

Evaluating System Suitability CE, GC, LC and A/D ChemStation Revisions: A.03.0x- A.08.0x CE, GC, LC and A/D ChemStation Revisions: A.03.0x- A.08.0x This document is believed to be accurate and up-to-date. However, Agilent Technologies, Inc. cannot assume responsibility for the use of this

More information

Laboratory 4: Feedback and Compensation

Laboratory 4: Feedback and Compensation Laboratory 4: Feedback and Compensation To be performed during Week 9 (Oct. 20-24) and Week 10 (Oct. 27-31) Due Week 11 (Nov. 3-7) 1 Pre-Lab This Pre-Lab should be completed before attending your regular

More information

Proceeding of 5th International Mechanical Engineering Forum 2012 June 20th 2012 June 22nd 2012, Prague, Czech Republic

Proceeding of 5th International Mechanical Engineering Forum 2012 June 20th 2012 June 22nd 2012, Prague, Czech Republic Modeling of the Two Dimensional Inverted Pendulum in MATLAB/Simulink M. Arda, H. Kuşçu Department of Mechanical Engineering, Faculty of Engineering and Architecture, Trakya University, Edirne, Turkey.

More information

The Method of Least Squares. Lectures INF2320 p. 1/80

The Method of Least Squares. Lectures INF2320 p. 1/80 The Method of Least Squares Lectures INF2320 p. 1/80 Lectures INF2320 p. 2/80 The method of least squares We study the following problem: Given n points (t i,y i ) for i = 1,...,n in the (t,y)-plane. How

More information

Multi-variable Calculus and Optimization

Multi-variable Calculus and Optimization Multi-variable Calculus and Optimization Dudley Cooke Trinity College Dublin Dudley Cooke (Trinity College Dublin) Multi-variable Calculus and Optimization 1 / 51 EC2040 Topic 3 - Multi-variable Calculus

More information

Graphical Integration Exercises Part Four: Reverse Graphical Integration

Graphical Integration Exercises Part Four: Reverse Graphical Integration D-4603 1 Graphical Integration Exercises Part Four: Reverse Graphical Integration Prepared for the MIT System Dynamics in Education Project Under the Supervision of Dr. Jay W. Forrester by Laughton Stanley

More information

Part IB Paper 6: Information Engineering LINEAR SYSTEMS AND CONTROL Dr Glenn Vinnicombe HANDOUT 3. Stability and pole locations.

Part IB Paper 6: Information Engineering LINEAR SYSTEMS AND CONTROL Dr Glenn Vinnicombe HANDOUT 3. Stability and pole locations. Part IB Paper 6: Information Engineering LINEAR SYSTEMS AND CONTROL Dr Glenn Vinnicombe HANDOUT 3 Stability and pole locations asymptotically stable marginally stable unstable Imag(s) repeated poles +

More information

PID Tuning Guide. A Best-Practices Approach to Understanding and Tuning PID Controllers. First Edition by Robert C. Rice, PhD

PID Tuning Guide. A Best-Practices Approach to Understanding and Tuning PID Controllers. First Edition by Robert C. Rice, PhD PID Tuning Guide A Best-Practices Approach to Understanding and Tuning PID Controllers First Edition by Robert C. Rice, PhD Technical Contributions from: Also Introducing: Table of Contents 2 Forward 3

More information

CORRELATED TO THE SOUTH CAROLINA COLLEGE AND CAREER-READY FOUNDATIONS IN ALGEBRA

CORRELATED TO THE SOUTH CAROLINA COLLEGE AND CAREER-READY FOUNDATIONS IN ALGEBRA We Can Early Learning Curriculum PreK Grades 8 12 INSIDE ALGEBRA, GRADES 8 12 CORRELATED TO THE SOUTH CAROLINA COLLEGE AND CAREER-READY FOUNDATIONS IN ALGEBRA April 2016 www.voyagersopris.com Mathematical

More information