ISA Transactions 47 (2008) 241–246 www.elsevier.com/locate/isatrans Research note An online non-intrusive method for alignment between actuators and their response centers on a paper machine R. Bhushan Gopaluni a,* , Michael S. Davies a , Philip D. Loewen b , Guy A. Dumont a a Department of Electrical and Computer Engineering, University of British Columbia, Vancouver, BC V6T 1Z4, Canada b Department of Mathematics, University of British Columbia, Vancouver, BC V6T 1Z4, Canada Received 18 May 2007; received in revised form 9 October 2007; accepted 30 January 2008 Available online 25 March 2008 Abstract Good cross-directional control of sheet variables in a paper machine depends on a number of factors. One such factor is the correct alignment of actuators with their corresponding responses. The alignment changes with time either locally or across the whole paper sheet due to a variety of reasons. In this paper, we provide an online mapping technique based on estimating the cross-correlation between actuator movement and the measurements. This procedure involves two steps – in the first step (called alignment monitoring) we determine if the current mapping is wrong at any CD position, and in the second step (called re-mapping), the alignment of actuators is corrected. The main advantage of this method is in its ability to re-align under closed loop conditions with minimal operational disruption. c  2008, ISA. Published by Elsevier Ltd. All rights reserved. Keywords: Paper machines; Actuator alignment; Closed loop identification 1. Introduction High performance cross-directional control of paper machines is vital for maintaining the quality of paper produced, and for reducing production costs and/or wastage. In the last two decades, a lot of effort has been spent on providing good cross-directional control through a number of robust controller design methods [1–3]. The performance of these methods depends on the accuracy of the estimated model and its uncertainty region. Even though these controllers are robustly stable, ensuring robust performance is difficult due to the varying process conditions. On a paper machine, the process operating conditions change frequently. As a result, the shape and position of the cross-directional response to actuator movement also changes. This change in response position and/or shape leads to degra- dation of robust controller performance. In fact, change in re- sponse position often leads to ‘picket fencing’ or instability [4, 5]. One obvious way to overcome the loss of performance is to * Corresponding author. Tel.: +1 604 822 2482; fax: +1 604 822 8563. E-mail addresses: gopaluni@ece.ubc.ca, gopaluni@chml.ubc.ca (R.B. Gopaluni). identify a new process model repeatedly [6]. This type of a full- blown identification exercise is expensive and time consuming. Moreover, it is not necessary to perform such an exercise often. Instead, the focus in this paper is on finding the correct align- ment of the actuator response under operating conditions. Both the response shape and alignment with respect to the actuators are of concern, and uncertainty in either of these aspects of the model leads to performance degradation. In this paper, the focus is on identifying the center position or alignment of the response, also known as response mapping. We would like to track alignment by minimal intrusion into the normal operation of the paper machine. In the last two decades a number of approaches for online mapping have been developed [5,7,8]. In most of these methods, mapping is adjusted by performing open loop bump tests on the actuators and obtaining a CD model. Bump tests provide a very good mapping solution. However, these bump tests are time consuming, expensive and require human intervention. Moreover, frequent mapping changes (for instance due to grade changes) require frequent bump tests, which would lead to disruption of the smooth operation of the paper machine. In [5] a new optimization based approach to alignment using local and global variability of the CD profile was presented. 0019-0578/$ - see front matter c  2008, ISA. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.isatra.2008.01.001