Operations & Management
Increasing read rates to 99.5% results in 75 more instrument panels per day
February 27, 2017 by Cognex
Feb. 27, 2017 – Marking barcodes on parts is a must in the automotive industry in order to diagnose problems quickly and accurately and maintain high levels of quality.
One major automobile OEM uses laser marking to apply a 2D DataMatrix barcode to each instrument panel it makes, then immediately reads the code to ensure it is legible. In the past, about 5 per cent of the codes could not be read, resulting in about 15 minutes of downtime over two shifts per day on the instrument panel line while associates investigated and manually corrected the problem. The OEM worked with Voelker Controls Company to implement Cognex Explorer Real Time Monitoring (RTM) which captures and analyzes no-read images from the ID reader. RTM provides a continuous feed of no-reads and their causes on a large television mounted near the line. The OEM’s associates keep an eye on the screen and are immediately alerted any time there is a no-read and also provided with the cause. Associates immediately fix the problem, such as by adjusting the laser marking machine if it is producing faulty barcodes. Read rates are now consistently at 99.5 per cent or higher, reducing downtime due to no-reads to only two minutes per day and increasing production by about 75 parts per day.
Traceability critical to automotive industry
Automotive OEMs are reducing lead times and optimizing their supply chains by documenting the genealogy of parts and subassemblies as they move through the supply chain. Data such as source manufacturer, part number, manufacturing or assembly facility, model number, place of origin, and serial number is accumulated and combined into a barcode and marked on the part. Barcode readers then decode the data throughout the part’s lifecycle and transfer the information into an Enterprise Resource Planning (ERP) software system, ensuring full traceability for each part of a finished vehicle. Real-time tracking systems can help reduce lead times and optimize the supply chain by providing information about work movements from zone to zone. Traceability can improve quality by ensuring that the appropriate processes are performed in the correct sequence on the right parts. In the event that bad parts enter the supply chain, traceability can reduce the time and cost involved in diagnosing what went wrong, fixing the problem, and finding and fixing all of the affected parts.
To address full lifecycle traceability, automakers and suppliers permanently mark parts with 2D barcodes using methods such as dot peening, laser marking, inkjet printing, and electro-chemical etching. Reading 2D barcodes and text can be challenging in real-world applications because of variations in the size, shape, position and orientation of the code, the potential for degradation in printing and marking of the codes, the wide range of surfaces on which the codes may be marked, the potential for degradation in the equipment used to print or mark the codes, variations in ambient lighting, etc.
Challenge of minimizing no-reads
The automotive OEM primarily uses Cognex DataMan 302 ID image-based readers to read the codes. By capturing an image of the code and using algorithms to read it, these devices locate and read codes even when the codes exhibit significant damage to or complete elimination of the finder pattern, clocking pattern, or quiet zone. The ID readers use texture-based location algorithms that take a unique, inside-out approach to reading 2D DataMatrix codes. While conventional feature-based algorithms start by locating the finder pattern, the new algorithm looks for a pattern of alternating light and dark modules within the code. This technology increases read rates in 2D barcode-reading applications where a part’s geometry, poor lighting, occlusion, or print-registration errors make it difficult to capture an image of the entire code.
With the improvement in barcode reading algorithms, the vast majority of no-reads occur because there is no barcode to be read for one of several reasons. The readers in this application are triggered manually by the associate and sometimes the associate may trigger the reader when the part is not present in the inspection machine. Or the part may be present but improperly positioned in the fixture so that the barcode does not appear in the vision system’s field of view. Finally, the laser marking machine may have failed to properly mark the part.
When a no-read occurs, the line stops and an associate checks the part. If the part is properly marked the associate manually runs it through the validation station to record the code. If there is no code present, the associate prints a paper label and attaches it to the part. The OEM’s associates perform this process quickly and efficiently but some downtime is inevitable. The OEM’s instrument panel line includes two molding machines, each with a 105 second cycle time and each producing two parts per cycle. Total production from the line is about 120 parts per hour so each minute of downtime costs four parts. In the past, with a 95-per-cent read rate, about 15 minutes of production was lost per day.
Implementing a monitoring solution
The automotive OEM has recently been focusing on implementing procedures to identify the root cause of process errors so they can be quickly corrected. The OEM’s engineers asked Voelker Controls Company, a Cognex distributor, if they might have a solution to the lost production problem. Voelker proposed using the RTM to track and analyze no-reads in order to identify the root cause. RTM is installed simply by connecting it to the networking with the ID readers and assigning it with an IP address. RTM automatically detects the ID readers on the network and begins collecting data. When a process error results in an unreadable code, the image captured by the barcode reader is transferred to RTM. RTM uses machine vision technology to classify each image and categorize them into defect groups such as “no code present,” “no part present,” and etc. Two other features that appealed to the OEM were RTM’s ability to measure data over specified timeframes and its ability to measure readers for specified groups of readers, such as those in one area of the plant.
A key to the success of this application is that the read rates and causes for the no-reads are displayed in real time on 55-in. television screen. Associates can see the screen from their work stations and are immediately alerted to any sudden spike in no-reads as well as the cause. For example, the monitoring system showed in its early days of operation that the majority of no-reads occurred when associates triggered the ID readers when there was no part in the inspection station. Making associates aware of this problem helped to reduce the number of no-reads. Likewise, associates are immediately alerted when the laser etching machine stops working or starts producing faulty codes. These alerts enable associates to fix problems that cause no-reads much faster than in the past.
Implementation in a second plant
Based on the successful reduction in no-reads on the instrument panel line, the RTM was also implemented in another plant that produces aluminum cylinder heads. The plant runs ten lines to process these parts including six machining lines, two assembly lines and two diecasting lines. These lines have 27 ID readers and run 2,000 parts per day, resulting in 54,000 potential reads per day. In the past, no-reads were tracked by associates on each line using paper forms. These forms were provided to an engineer who entered the information into a spreadsheet. Little information was available on the cause of no-reads. Most of the no-reads were caused when casting suppliers failed to mark the parts but there was no way in the past to track which suppliers were responsible. RTM was configured to determine the cause and the casting supplier for each no-read. The end result was that no-reads were reduced from 90-95 per cent to above 99.5 per cent.
The automotive OEM’s two plants that have implemented RTM increased their read rates consistently to 99.5 per cent and above by providing associates with immediate feedback on no-reads and their causes. At these read rates, the instrument panel line experiences only about two minutes per day of downtime so it produces an extra 75 parts per day. At a value of about $50 per part, the result is an additional $3,750 per day in revenue. The company has freed up a considerable amount of associate time that was previously spent on correcting no-reads. The savings on the aluminum cylinder lines have not yet been quantified but they are most likely equal or greater in magnitude.
Submitted by Cognex