Value-Based Web Inspection for Brown Board Machines
Traditional web inspection has been cost-prohibitive for many brown board machines due to the requirement of using dedicated camera and light beams to eliminate the optical ‘noise’ in the environment. Line scan-based inspection systems require the camera to be placed perpendicular to the sheet which restricts camera placement options.
This white paper explains an alternative software and hardware solution from ECS that uses area scan cameras placed outside the sheet using 100% existing mill lighting to provide critical process data to increase the paper machine’s OEE. This may be the final inspection solution for some machines while others will require dedicated LED and camera beams that ECS also manufactures.
- Web Inspection System (WIS)
- Line Scan camera vs. Area Scan camera
- Acute Angle Web Inspection
- Optical Noise
- Template Based Imaging
- Dynamic vs. Static Based Template Algorithms
- Filtering False Defects - Directional Analysis
- Transmitted and Reflected Lighting Systems
- Winder Control
- IDMR (Inspect, Database Mark and Read)
Web Inspection Background and Traditional Limitations
WIS (web inspection system) is a comparative engine. Cameras bring in images (by line scan or area scan) and compare these images (pixels) to what the system has determined as the ‘correct’ set of pixels (also called the ‘template’). Incoming images that contain a pattern of pixels that differ from the template over a set of thresholds are called defects. Critical to a successful inspection system are two main criteria (1) does not miss defects and (2) does not detect normal operating conditions as defects (false defects). These conditions can also be called ‘noise’. There are plenty of conditions in a paper mill that cause noise:
- Changing shadows from cranes, sun and other variations
- Mill lighting
- Paper grade changes
- Natural variations in the makeup of the sheet
- Vibration and sheet movement
Traditional WIS systems typically isolate the camera from optical ‘noise’ by using highly engineered cross directional beams with dedicated lighting systems. Many web inspection systems also use line scan cameras that require the camera to be perpendicular to the sheet and must have an encoder signal to ‘reconstruct’ the image from the single ‘line slices’ of the sheet. And in the case of dust – both area scan and line scan cameras may consider dust as defects as the dust creates concentrated round shadows on the sheet when using reflective lighting sources.
What is OEE?
OEE stands for Overall Equipment Effectiveness. It has three main criteria: 1 – Uptime, 2 - performance (process speed) and 3 - product quality. Correctly operating web inspection systems can positively influence all three aspects of OEE. But the data resulting from WIS must be free of false triggers and not miss defect for the data to have value. The cost of this solution should not restrict it to only the more expensive paper grades.
All of these factors can contribute to expensive and complex solutions to deploy web inspection. Or poorly performing systems that may generate large amounts of data with only a portion of that data as being useful to make critical machine running decisions. A large portion of brown board machines use cameras only for event capture. While event capture is an extremely valuable tool – it’s a reactive tool and requires the combination of an operator and video to determine a course of corrective action. Web inspection systems that are properly designed and installed can proactively provide an entirely new set of data to decrease machine downtime and increase paper quality.
The ECS Solution
The ECS solution is fundamentally different in its approach to web inspection. Instead of eliminating the noise before it is captured by the cameras – ECS uses what is called Real-time Dynamic Threshold Template Imaging to remove the noise by software – not in the environment. Every frame (and corresponding pixel of the frame) that is captured by the ECS program is compared to a new and unique template image. One or more cameras mounted outside of the sheet with ambient mill lighting and no encoder can capture, classify and display critical process data. This includes edge cracks that will break the sheet at winder, slime spots, holes, water drops, etc. This means actionable process control data can be collected in noisy environments without costly beams and dedicated lighting systems. ECS does use beam structures for cameras and LED’s in many grades of paper that require finer defect detection and precise classification. The emphasis of this paper is to address the considerable number of machines that still need a more value based automated defect detection to proactively alarm and reduce unwanted defects, costly downtime and customer quality complaints.
The five illustrations below show the difference between static and dynamic threshold applications. Even if the software resets the template every couple of minutes or some other time interval – the system is still considered static.