APPLIED INDUSTRIAL CONTROL SOLUTIONS LLC

HOME PAGE Case Studies Published Articles Control Tips/Tricks Training/Mediation

Inter-Office Memo

To:

James Hanna Sr.

C.C.

Brian Boulter
Robert Lockhart

ApICS LLC
ApICS LLC

From:

David Morell

RA. RCS

Date:

April 8th 2000

(PROPRIETARY) Diaper Line Control Improvement

Subject:

Jim,

I would like to bring your attention the outstanding work that the ApICS LLC consultant Brian Boulter did on a recent project for one of our larger customers, (PROPRIETARY). Registration control (PROPRIETARY) process lines has traditionally been done by their in-house engineering staff with centralized, custom developed, solutions for each particular process. Recently, Rockwell Automation was given the opportunity to retrofit a diaper line with an RA AutoMAX DPS based registration and inspection control system. Since this was our first opportunity to prove ourselves, we needed to implement a superior design for the registration control portion of the system. We commissioned Brian Boulter from ApICS LLC to take the lead in designing the system architecture, and to project manage the engineering and installation. He proved instrumental in leading us to supply (PROPRIETARY) with a superior design and system to their current in-house designed system.

After discussing the problems with our current registration control design, and presenting a description of the process, Brian came back with a method of dead-time compensation called the "(PROPRIETARY)". Brian brought us up to speed on how it works and the kinds of processes it is best suited for. When it came time to present the design to the customer, Brian participated as a technical resource, and was instrumental in convincing the customer to adopt the proposed control system.

After the meeting Brian worked closely with (PROPRIETARY), and his engineering team to manage the completion of the new registration system, and oversaw the installation on the pilot diaper line at the customer site. The production statistics have shown significant improvement in performance, in some cases by more than a factor of four. I have pasted into this memo a study performed by the customer comparing a registration loop using the in-house engineered PID regulator to that of the Rockwell Automation proportional only regulator with the "(PROPRIETARY)" design. The differences are impressive.

Brian’s efforts proved highly beneficial. I believe the success of this project has left our customer with a very favorable impression concerning Rockwell Automation’s resourcefulness and capabilities. I appreciate ApICS LLC’s tremendous contribution to this project, and recommend that we consider them as a "First-Choice" resource for future projects of this type.

David Morell

Consumer Products Engineering Director

DISTRUBUTION

TO: (PROPRIETARY)
Dave Morell – RA
Brian Boulter – ApICS

Research Files #100173
To: Distribution
Date: February 6th, 2000
From: (PROPRIETARY)
Subject: (PROPRIETARY)

Summary

Using proportional control with feedback delay compensation the (PROPRIETARY) control scheme improved surge registration performance when compared with the standard PID control method. Improvements in settling time and overshoot were especially significant.

Introduction

The Surge Placement registration algorithm that rolled out with the Capsule project uses a standard PID control scheme to give registration commands to the actuator. During the N6 Rockwell Automation Registration project a different method of registration loops were converted to RA control. This study was done to compare PID control to the new method, known as the (PROPRIETARY) control scheme.

Results

Two types of situations were observed to evaluate the performance of the two methods: 1) A manual phase by the operator so that the patch was 50 mm out of position and 2) a machine start-up from 0 to 550 DPM.

Table 1. 50 mm Disturbance to Patch Placement

Time to come within 5mm of setpoint

 

(PROPRIETARY)

PID

Trial #1

8.6 sec

36.8 sec

Trial #2

8.9 sec

39.8 sec

 

Table 2. Startup Response, 0 to 550 DPM

Time to come within 5mm of setpoint

 

(PROPRIETARY)

PID

Trial #1

3.5 sec

15.7 sec

Trial #2

3.1 sec

12.1 sec

The two color figures attached illustrate the differences for the 50 mm upset to the process. These graphs were produced with a Sigma data collection station. Figure 1 illustrates the performance when running with the PID control enabled. Time is shown in seconds on the x-axis and position error is measured on the y-axis. Note that time increases from right to left.

The green line is the position error calculated by the control system for the surge patch placement. The error starts at 1140 encoder counts (50 mm) and then goes towards 0 after the automatic registration loop is enabled. Again the green line is the surge patch position error. The other data lines shown were used for troubleshooting the control algorithm.

When running with (PROPRIETARY) control scheme (Figure 2) the placement error came under control without the overshoot that was seen with the PID control. This is the result of tracking requested registration moves, before the move is actually seen at the feedback device. On the N6 diaper machine there is a 12 product delay between the applicator and the feedback sensors that detect the position of the patch.

Theory

(PROPRIETARY).

Infant Care Engineering

HOME PAGE Case Studies Published Articles Control Tips/Tricks Training/Mediation