(654g) Detection of CONTROL LOOP Interaction and Prioritization of LOOP MAINTENANCE

In a multiloop control configuration of a chemical process many control loops are used to stabilize the process and ensure smooth operation of the plant. Change in set point and/or controller parameters of one control loop may also affect the variables of other loops. This interaction arises mainly due to a hidden feedback loop between two interacting loops. Control loop interaction in a process plant can cause significant cost, quality and production losses of the plant. It is challenging to measure the degree of interaction between control loops and rank the loops according to the extent of interactions. As per this loop rank, the maintenance or remedial measures priority can be determined, which will ultimately increase the plant profitability.

Over the past few decades, several researchers presented different methods to determine the degree of loop interactions. [1] proposed a method in 1966 to determine control loop interaction and recommend best pairing of manipulated and controlled variable using steady state information of the process. [2] and [3] proposed two different methods which uses the open loop response of the process. [4] calculate the loop interaction using singular perturbation technique. [5] and [6] also used steady state information of the process to determine control loop interaction. [7] proposed a graphical method which determines the loop interaction by considering the dynamic information of the process. [8] analyses loop interaction using sensitivity and complementary sensitivity functions. [9] uses only plant routine operating data to determine loop interaction in both frequency domain and time domain. [10] uses dynamic information of the process to measure loop interaction. [11] gives a new tool to determine loop interaction using partial correlation and PageRank technique. [12] calculates loop interaction by using the variability information of the controlled variables.

All methods except [9], [11] and [12] require the process model information. The exact process model is almost never known. For this reason the calculated interaction has limited reliability. [9], [11] and [12] require only routine operating data of the process but these methods require extensive calculation and sometime they produce erroneous results.

This paper presents a new method that determines control loop interaction effectively and that can also rank the loops according to their importance thus prioritizing the loop maintenance work.

Simple statistical signal processing methods such as data correlation is used to find a technique for determination of control loop interaction. In data correlation, different correlation methods are used such as partial correlation, cross correlation and canonical correlation. The correlation value of 1 represent the data is correlated with itself. Therefore, correlation value close to one for two variables indicates that they are highly correlated. On the other hand, if a variable in one control loop is highly disturbed by change in another control loop then that variable will have high value of integral of absolute error or integral of squared error.

Among the different correlation, canonical correlation analysis is a way of measuring linear relationship between two multidimensional variables. The canonical correlation is optimized such that the linear correlation between two variables is maximized. Interaction index compares the integral of absolute/squared error of one control loop with that of other interacting loops. A new index called Loop Importance Index (LII) has been developed. It calculates the importance of each loop in the order of 0 to 100 %.

The newly developed method is validated by doing extensive simulation work using MatLab Simulink software.

References 1. Bristol, E. H., ?On a new measure of interaction for multivariable process control?, IEEE Trans. Auto. Control, Vol-11, pp 133-134, 1966. 2. Witcher, M. F. and McAvoy, T. J., ?Interacting control systems: Steady-state and dynamic measurement of interaction?, ISA Trans., Vol-16, pp 35-41, 1977. 3. Gagnepain, J.P. and Seborg, D.E., ?Analysis of process interactions with applications to multiloop control system design?, Industrial & Engineering Chemistry Process Design and Development, Vol-21, pp 5-11, 1982. 4. Shimizu, K. and Matsubara, M., ?Singular perturbation for the dynamic interaction measure?, IEEE Trans. Auto. Control, Vol-30, pp 790-792, 1985. 5. Hwang, S. H., ?Geometric Interpretation and Measures of Dynamic Interactions in Multivariable Control Systems?, Industrial & Engineering Chemistry Research, Vol-34, No. 1, pp 225-236, 1995. 6. Zhu, Z. X., Lee, J. and Edgar, T. F., ?Steady State Structural Analysis and Interaction Characterization for Multivariable Control Systems?, Industrial & Engineering Chemistry Research, Vol-36, No. 9, pp 3718-3726, 1997. 7. Meeuse, F. M. and Huesman, A. E. M., ?Analyzing Dynamic Interaction of Control Loops in the Time Domain?, Industrial & Engineering Chemistry Research, Vol-41, No. 18, pp 4585-4590, 2002. 8. Lee, J. and Edgar, T. F., ?Dynamic Interaction Measures for Decentralized Control of Multivariable Processes?, Industrial & Engineering Chemistry Research, Vol-43, No. 2, pp 283-287, 2004. 9. Rossi, M., Tangirala, A. K., Shah, S. L. and Scali, C., ?A data based measure for interactions in multivariable systems?, Proceedings of ADCHEM-2006, pp 681-686, April 2-5, 2006, Brazil. 10. Vilanova, R., ?Closed loop interaction and performance considerations for decentralized control of two-by-two multivariable system?, Canadian Conference on Electrical and Computer Engineering, pp 725-728, May 5-7, 2008, Niagara Falls, Canada. 11. Faenzena, M. and Trierweiler, J. O., ?LoopRank: A novel tool to evaluate loop connectivity?, Proceedings of ADCHEM-2009, Vol-2, pp 1023-1028, July 12-15, 2009, Istanbul, Turkey. 12. Faenzena, M., Trierweiler, J. O. and Shah, S. L., ?Variability Matrix: A novel tool to prioritize loop maintenance?, Proceedings of ADCHEM-2009, Vol-2, pp 705-710, July 12-15, 2009, Istanbul, Turkey.