Oscillations In Control Loops Research Articles

Experimental and modeling investigation on dynamic response of sticky control valves

The dynamic response of control valves directly affects the safe and efficient operation of industrial control loops. Valve stiction is a common and persistent issue that causes oscillations in control loops. The stiction behavior of valves has received widespread attention from researchers worldwide in the past two decades. The developed valve stiction models have been widely applied in the detection, quantification, and compensation research of sticky control valves. However, how to more accurately characterize stiction behavior still requires efforts. Most data-driven models do not consider the effects of dynamic response on the stiction behavior. In this paper, the inconsistency of the representative stiction models is discussed during the unidirectional motion of the valve stem, and potential improvements are revealed. An experimental device for valve stiction has been designed. This device can replicate valve stiction caused by tight packing, and measure valve position and friction through smart positioner and force sensor. The second-order dynamic system for the sticky valve is constructed by the physical model, and the response time of the valve is calculated and verified combined with the experiment. The effects of sampling interval on stiction behavior are discussed. On this basis, an improved valve stiction model considering the dynamic response of the valve is proposed, which supplements the situation of stiction during the unidirectional motion of the valve stem. The proposed model can be applied in control systems with fixed sampling intervals that involve sticky valves. It can also be extended to control systems with variable sampling intervals. This work contributes to evaluating the performance of control systems with sticky valves and providing reference value for the detection, quantification, and compensation of valve stiction.

Automatic oscillations detection and quantification in process control loops using linear predictive coding

Oscillations in control loops are very common and primarily responsible for product quality variations and therefore may reduce profitability of the plant. Further, there can also be multiple oscillations at the same time due to multiple sources such as improper tuned controllers, disturbances and nonlinearity etc. Detection and estimation of oscillation is desired for their effective compensation which becomes a challenging task particularly in presence of noise. This paper presents a novel oscillation detection method based on linear predictive coding (LPC). In the proposed technique roots of the linear predictive polynomial are used to detect the oscillations. Further, for the quantification spectrum of cross-correlation of linear predictive residual and original data is used. The effectiveness of the proposed oscillation detection technique has been demonstrated with several simulation examples and actual plant data. The comparative investigation with the empirical mode decomposition method, generally employed for this task, revealed that proposed technique is simple, fast, robust and capable of handling slow varying trends.

Oscillation Detection and Diagnosis in Process Industries by Pattern Recognition Technique

Abstract Oscillation in control loops is a frequent problem faced in process industries. It deviates the process variables from their desired condition, affecting negatively plant productivity. To guarantee profitability, oscillation must be detected, diagnosed, and, finally, eliminated. Dozens of automatic detection and diagnosis techniques have been proposed over the last years. However, the application to real industrial data reveals low efficiencies, which indicates that the techniques require improvement. This work presents a new method for oscillation detection and diagnosis. The technique classifies the loops based on the shape of the PV(OP) diagram by a pattern recognition approach, where the model is trained with simulated examples that cover a large range of processes and different conditions found in industrial data, such as noise and disturbance presence. The performance of the proposed technique is compared to well-established oscillation detection and diagnosis methods, returning better results.

Friction Compensation in Pneumatic Control Valves Through Feedback Linearization

Control valves are very important actuators in the process industry. Due to friction in the stem, they present a nonlinear behavior, which can generate errors in the stem position and oscillations in control loops. This paper presents five control algorithms based on feedback linearization technique, to deal with pneumatic control valves affected by high friction in their stems. These control algorithms measure the valve position and manipulate the actuator pressure, acting as a valve positioner. The paper also shows practical results with a control valve operating in a real flow control loop.

Extraction of Plant-wide Oscillations using Fast Multivariate Empirical Mode Decomposition

This paper proposes a novel time-frequency method for plant-wide oscillation analysis based on the multivariate extension of standard empirical mode decomposition (EMD). The raised fast multivariate empirical mode decomposition (FMEMD) is generalized from EMD by solving an overdetermined system of linear equations. Due to its capability to analyze multiple channels data, FMEMD is especially suitable for characterizing plant-wide control loop oscillations. Unlike traditional methods, both the regularity of oscillations (in frequency domain) and evolution of local characteristics (in time scale) can be well captured via FMEMD. Validity of the raised approach is demonstrated on simulations as well as an industrial case.

Signal Preprocessing for Stiction Detection Methods

Oscillation in control loops is a problem of high incidence, stiction in control valves being the most frequent cause. In the last two decades, many automatic stiction detection methods were proposed. These methods require certain signal preprocessing for proper use, where the choice of the appropriate preprocessing technique and parameters depends on the detection method and signal characteristics and, due to the lack of information about its automatic application, are mostly led manually. The need for user interaction turns an automatic into a nonautomatic detection method, which makes the application unfeasible to an entire plant with hundreds or even thousands of control loops. This work proposes techniques for automatic preprocessing according to the demands of each detection method. The techniques cover the main problems related to stiction identification: removal of noise excess, identification of peaks and valleys, and removal of variable mean and zero-crossing identification. Satisfactory results...

Detecting non-linearity induced oscillations via the dyadic filter bank property of multivariate empirical mode decomposition

Non-linearity induced oscillations in control loops are characterized by the presence of higher order harmonics. In this paper the dyadic filter bank property of the multivariate empirical mode decomposition (MEMD) is exploited to reveal the harmonic content of the oscillatory signal to indicate the presence of non-linearity. Once the harmonics are identified the extent of non-linearity is evaluated automatically using degree of non-linearity measure (DNL) introduced in our previous work [11]. Although detection of non-linearity via harmonics is an old concept; any automatic method has still not been reported. Moreover, the existing methods suffer from the restrictive assumption of signal stationarity. The proposed method is more robust in identifying the non-linearity induced oscillations and is adaptive and data driven in nature and thus requires no a priori assumption about the underlying process dynamics. The proposed method can also differentiate among the different sources of multiple oscillations, for example combinations of nonlinearity and linear sources or two nonlinear sources. Apart from detecting the non-linearities the proposed method can also contribute in locating the source of non-linearity, thereby reducing the maintenance time to a considerable extent. The robustness and effectiveness of the proposed method is established using industrial case studies and results are compared with existing methods based on higher order statistics [7] and surrogate based methods [8].

A Delay Vector Variance based Approach for Detecting and Isolating the Non-linearity Induced Oscillations in Control Loops

Non-linear time series analysis based methods are a popular choice for industrial control loop data analysis. In this paper a delay vector variance (DVV) based approach is presented to analyze the source of oscillations in an industrial control loop. The method is capable of differentiating between the linear and non-linear causes of oscillations and can also help in isolating the source of non-linearity. The automatic determination of embedding dimensions is augmented with the DVV analysis to make it more robust and reliable. The efficacy of the proposed method is established using simulation as well as industrial case studies.

The Effect of the Sampling Period on Stiction Detection Methods

Among the causes of oscillation in control loops, stiction in control valves has the highest incidence. There are several works on stiction detection, but few of them discuss problems related to industrial data. Fast process dynamics combined with low sampling periods cause problems in the detection, since the number of data per cycle may be less than that required for the analysis. This study evaluates the effect of sampling period on five stiction detection methods. The evaluation was lead using standard signals and industrial data. The results present the minimum signal sample rate required by the selected methods and the lack of techniques for stiction detection in signals with low sampling.

Time-varying oscillation detector based on improved LMD and robust Lempel–Ziv complexity

A novel method based on improved local mean decomposition (LMD) for oscillation detection is proposed. Due to its capability to analyze amplitude and frequency modulated signals, LMD is especially suitable for characterizing time-varying control loop oscillations. In contrast to empirical mode decomposition (EMD), the improved LMD performs better in the following aspects: (i) ability to extract both single/multiple oscillations in process output, (ii) robustness to noise, and (iii) capability to handle non-stationary trends. In addition, improved LMD can precisely characterize the time-varying oscillations without distortion and frequency leakage even for short time series. Finally, a robust Lempel–Ziv complexity based statistic for time-varying oscillation detection is presented. Simulation examples and industrial applications are provided to demonstrate the effectiveness of the proposed LMD oscillation detector.

Fault propagation analysis of oscillations in control loops using data-driven causality and plant connectivity

Oscillations in control loops are one of the most prevalent problems in industrial processes. Due to their adverse effect on the overall process performance, finding how oscillations propagate through the process units is of major importance. This paper presents a method for integrating process causality and topology which ultimately enables to determine the propagation path of oscillations in control loops. The integration is performed using a dedicated search algorithm which validates the quantitative results of the data-driven causality using the qualitative information on plant connectivity. The outcome is an enhanced causal model which reveals the propagation path. The analysis is demonstrated on a case study of an industrial paperboard machine with multiple oscillations in its drying section due to valve stiction.

Automatic detection of multiple oscillations by wavelet analysis

This paper proposes an automatic method based on wavelet analysis to detect multiple oscillations in control loops. Continuous wavelet transform provides a way to represent the signal as a superposition of wavelets. The redundancy of the representation can be exploited to precisely indicate the location of oscillatory behaviors in the time-frequency plane. Oscillatory components are separated by the adaptive design of filter bank. It uncovers all the zero-crossings in the frequency bands which possibly include oscillations to be confirmed by Auto-covariance Function (ACF). The method is applicable to non-stationary and noisy data due to the multi-resolution analysis of orthonormal wavelet bases. This paper ends with demonstrating the effectiveness of the proposed method with benchmark industrial data.

Robust Oscillation Detection Index and Characterization of Oscillating Signals for Valve Stiction Detection

Detection of oscillation is a necessary step for determining valve stiction, which is a common problem in process industries. More specifically, many stiction detection methods assume persistency of the oscillation pattern in the control loop signals, which is a necessary condition to achieve reliable detection results. However, the existing oscillation detection methods do not study the persistency of the oscillation patterns directly. Instead, these methods try to discover the presence of specific periodic characteristics in the signals, their autocorrelation, or power-spectrum. This paper aims to propose an oscillation detection method that directly evaluates the similarity of the shapes of subsequent oscillation periods by means of correlation coefficient. The proposed oscillation detection method is compared against five other methods reported in the literature. Furthermore, the stiction detection methods assuming oscillation in control loops have different robustness to the disturbances corrupting t...

An Overview of Impact of Nonlinear Friction in Oscillating Control Loops: from Diagnosis to Compensation

Among the prominent reasons for oscillations in control loops such as poor controller tuning, inappropriate process design, oscillating load and disturbances with a high-frequency, the presence of static friction or stiction in the control valve is the most common reason. As a consequence of the friction, the control valves may present an oscillatory behavior, affecting the regulatory control performance and therefore it increases the variability of the control loop, causing loss of product quality and increasing energy consumption. Modeling of stiction using clear mathematical definition is hard due to its nonlinearity properties such as hysteresis, deadband or deadzone. This comparative study identifies the relative strengths and weaknesses of the different approaches of diagnosing of stiction in control loops and the compensation techniques to mitigate this effect. The aim is to determine the appropriate and precise model of stiction in order to design a proper controller to prevent control loops from oscillations. Numerous algorithms and methods which are either qualitative or quantitative in the literature show that despite these attempts to modeling, diagnosing and compensation of this phenomenon, lack of a reliable methodology to predict occurrence and properties of such oscillations and therefore lack of proper model is quite sensible. Furthermore exclusive of lack of precise model, even solving the problem were limited to some and different assumptions and applications and there is not one method which can cover all cases reliably. From this study too, it is seen that Data Driven model because of providing a simpler solution with fewer parameters is the most common method in contrast to physical based models with the presence of a number of unknown physical parameters. It is our view that hybrid approaches where different methods work in conjunction to solve parts of the problem are attractive and more effective methods. Lastly, numerous methods of stiction detection are available. Hence, it is suggested that the future research emphasis should be more on compensation of stiction especially for nonlinear processes.

Quantification of Valve Stiction and Dead Band in Control Loops Based on the Harmonic Balance Method

The presence of stiction in control valves often causes oscillations in control loops, with negative effects on quality and cost of goods. To address this issue, it is necessary to quantify this stiction to decide about maintenance or to implement compensators that can improve control loop performance until the next plant stop. The describing function (DF) method is a well-known scheme to predict the period and amplitude of limit cycles in control loops, requiring the knowledge of linear and nonlinear parameters of the system model. In the present study, a method has been proposed to estimate these nonlinear parameters using the parameters of linear model and amplitude and period of limit cycle produced by nonlinearity. A procedure is proposed to overcome the case of unknown process model. In addition, generalization of this method to the case of parametric uncertainties in the linear part of the control loop has also been presented. The result is a simple and efficient algorithm that can be easily extended to other nonlinearities. Furthermore, the conditions for existence and uniqueness of solution for dead band and stiction estimations have been obtained. The usefulness of the proposed method has been demonstrated through simulations and its applications to a pilot plant and to real industrial data.

Root Cause Analysis of Linear Closed-Loop Oscillatory Chemical Process Systems

In a single-input single-output (SISO) closed-loop system, under constant or nonoscillatory set-point, oscillations in the output can occur mainly because of one or a combination of the following reasons: (i) presence of stiction in control valve, (ii) marginally stable control loops (because of aggressively tuned controller/changes in process time constant/gain/time delays or a combination of them), and (iii) disturbances external to the loop. The presence of these oscillations can propagate plant-wide and force plants to deviate from optimal operating conditions. Therefore, it is essential to develop techniques that can diagnose the source of oscillations in control loops. Several data-driven methods have been developed to address the diagnosis problem by focusing on only one of the causes for oscillations. In the current study, an off-line data driven approach is developed to identify the root cause for oscillations in control loops using the routine plant operating data. Unlike the existing techniques, this approach identifies and distinguishes between the three major causes for oscillations in linear closed loop systems. The proposed methodology combines both parametric (Hammerstein-based approach) and nonparametric (Hilbert–Huang spectrum) schemes for performing oscillation diagnosis. Simulation and industrial case studies that demonstrate the utility and limitations of the proposed method for root cause diagnosis in closed loop systems are discussed.

Detection of oscillations in control loops in irrigation channels

Oscillation in irrigation channels have many undesired consequences such as increased erosion on channel banks, reduced level of service to farmers, increased wear and tear on mechanical equipments, etc. It is therefore important to detect oscillations early. In this paper computationally inexpensive algorithms for detection of oscillatory control loops are applied to operational data from six consecutive pools of an irrigation channel. The algorithms performed well and detected the control loops which gave oscillatory responses. Moreover, the controller which caused the oscillations was also identified. As there can be many control loops in an irrigation channel, methods which are able to detect and locate the cause of oscillations are of great value in the operation of irrigation channels.

Two-Layer Binary Tree Data-Driven Model for Valve Stiction

Valve stiction commonly exists in process control systems. It generally leads to oscillation in control loops, which affects the product quality, increases energy consumption, and accelerates the equipment weariness. In this paper, current data-driven stiction models are first reviewed. A new data-driven model, with a simple two-layer binary tree logic structure is proposed. The logic behind the new model is carefully explained and illustrated. This model can cope with expanded types of stiction patterns, including some special cases such as linear and pure deadzone. Simulation results for both open-loop and closed-loop configurations will show the practical appeal of the proposed model.

A Curve Fitting Method for Detecting Valve Stiction in Oscillating Control Loops

Many control loops in process plants perform poorly because of valve stiction as one of the most common equipment problems. Valve stiction may cause oscillation in control loops, which increases variability in product quality, accelerates equipment wear, or leads to control system instability and other issues that potentially disrupt the operation. In this work, data-driven valve stiction models are first reviewed and a simplified model is presented. Next, a stiction detection method is proposed based on curve fitting of the output signal of the first integrating component after the valve, i.e., the controller output for self-regulating processes or the process output for integrating processes. A metric that is called the stiction index (SI) is introduced, based on the proposed method to facilitate the automatic detection of valve stiction. The effectiveness of the proposed method is demonstrated using both simulated data sets based on the proposed valve stiction model and real industrial data sets.

QUANTIFICATION OF VALVE STICTION

Oscillations in control loops lead to poor controller performance. Stiction in control valves is one of the major causes of such oscillations. Therefore, the correct diagnosis of stiction is important. There are several methods for detecting stiction, but quantification of stiction still remains a challenge. Two parameters are used to model the stiction phenomenon successfully, namely, deadband plus stickband, ‘S’, and slip-jump, ‘J’. It has been observed that the main cause of valve deterioration is the presence of slip-jump, ‘J’. The higher the value of ‘J’, the more severe is the level of deterioration of controller performance. Thus, in addition to the estimation of ‘S’, an estimate of ‘J’ is the main challenge in monitoring the condition of a control valve. In this work a method is proposed to estimate both ‘S’ and ‘J’ simultaneously unlike existing quantification methods where stiction is quantified as a single parameter.