Food for thought
Erdinger Weissbräu is one of Germany’s largest and most successful private breweries. The Munich-based brewery exports its speciality beers to more than 70 countries. And this figure keeps growing. Erdinger brews its beers only in its plant in Erding, six days a week around the clock, and without ever deviating from its high quality standard.
Erdinger is known for its strict quality policy and its consistent brand awareness. Its production processes, from selection and acceptance of raw materials to filling and maturing, are strictly monitored. Consequently, Erdinger Weissbräu maintains a permanent chain of security controls for its keg filling and packing process.
Simulation and risk analysis have shown that, however careful the production, it was conceivable that customers could take delivery of a keg filled with lye (a caustic solution used for cleaning kegs) instead of beer. In order to prevent accidents of this kind, HACCP (Hazard Analysis and Critical Control Point), a quality assurance programme for the food industry, stipulates the installation of critical control points.
However, cases of this kind are not unknown in the industry. Drawing on its exceptional quality and brand awareness, Erdinger Weissbräu decided to ensure a 100 per cent protection of its keg cleaning and filling process. Risk analysis demonstrated the need to set up an additional control point to ensure that a keg filled with lye could not slip through the conventional control points.
As a keg containing lye has a considerably higher temperature than a keg filled with beer, continuous monitoring of temperatures proved to be the optimum solution. In this way, it was easy to identify wrongly-filled kegs by measuring their temperature.
Non-contact temperature measurement using spot pyrometer thermometers proved not to be sufficiently reliable for this application. Accordingly, an infrared camera was the chosen option. When correctly adjusted and interpreted, the camera’s images enable a virtually error-free measurement.
A Flir Systems ThermoVision A20-M camera (pictured above) now measures the temperature of every keg before it leaves the conveyor belt of the filling installation. If the infrared camera identifies a keg with a different temperature, it sets off an alarm and the conveyor belt is automatically halted. The keg is then manually removed.
An infrared image of a beer keg, using the Flir system
The monitoring system with the infrared camera has been in use for six months at the Erdinger brewery, and it has proven to be extremely successful. The system is regularly inspected, but even in error simulations the camera has been reliable – any deviation has been identified immediately, triggering the appropriate alarm stage. ‘Not one keg filled with lye has left the filling shed. The infrared camera solution is now an established part of our plant,’ says a spokesman for Erdinger Weissbräu.
The ThermoVison A20, installed in a protective housing, plays a crucial role in this solution. Through FireWire, the camera is connected to a touch-screen monitor that shows which size and type of keg is currently going through the cleaning and filling process. A thermal image of the keg going through is simultaneously shown on the monitor screen in real time. The system is not incorporated into the washing and filling machinery, which means that it can be used on all, new or existing, filling installations.
Simple as child’s play
Supplying packaging for a global baby food manufacturer cannot be taken lightly. Huhtamaki France was faced with potential technical and public relations issues when dealing with a seemingly straightforward challenge.
Each element for the packaging of a baby food container is produced individually by Huhtamaki to be sent back to the client for filling. The transparent lid of each container being produced at very high rates each day has a label affixed declaring the contents and flavour of the food. It is vital that each delivery to the client contains the right label. The initial supplier of the label cannot guarantee that each batch contains the correct labels.
Thus, the risk that presents itself is that of a label mix-up, which could result in a consumer allergic reaction. Considering the high standards demanded in this kind of market, Huhtamaki found itself faced with a straightforward yet consequential challenge requiring a zero fault tolerance, to protect not only the end user but also the reputation of both Huhtamaki and its client.
Each production line at the plant is fully automated, high-speed and flexible. Production is in operation 24 hours a day, seven days a week. Each line is capable of producing different products and any changeover needs to happen quickly. All of these factors came into play last year for Huhtamaki, when producing packaging for a global baby food manufacturer.
Huhtamaki’s membership of the HACCP, which implies strict compliance with hygiene regulations, must also be taken into consideration. In this context, nothing that could jeopardise its reputation or compromise quality control can be tolerated.
The nature of the challenge prompted those responsible at Huhtamaki to turn to machine vision for a solution. It was the only possibility for such high-speed production where human inspection would not be viable.
Didier Lemaire, technical manager at Huhtamaki, had had prior experience with vision, but this time he needed a faster, more robust solution that was, above all, easy to use. Lemaire contacted Cognex and, following a consultation, began to work with Caire, a Cognex partner integrator who ensured the implementation of a vision system with a dedicated user interface, adapted in such a way as to ensure continued ease of use and possibilities for evolution in the future.
Each production session involves two lines of lids passing along a conveyor belt, each being labelled with the same sticker indicating a particular flavour and set of ingredients. The lids are constantly turning at a 360° angle, adding to the complicating factors facing the vision system. Once labelled, the lids need to be verified before being packed and sent to the client.
To be without faults and fully reliable, the system needed to be capable of acquiring and memorising quite a large amount of information in order to achieve the objective of zero faults required.
The solution provided needed to be flexible and scalable in order to allow for future changes of graphics, colours and products. In addition to all of this, environmental conditions were prone to change in terms of lighting and temperature factors.
The system also needed to be simple, insofar as personnel with limited vision experience could be trained easily to work independently with the program and have the possibility to make simple changes without compromising the reliability and success of the system.
Caire based its solution on Cognex InSight, equipping both lines on the conveyer with a camera and connecting the Huhtamaki interface simply by OPC on one PC. The use of Cognex advanced vision tools in the form of PatMax provided for a simple interface, with the product images clearly displayed, and played a decisive part in recognising the label and acquiring the image to be assessed.
Using PatMax, the program memorises and recognises an area of text and certain aspects of the label, as well as its positioning on the lid. There is no need for any complex parameter configuration, which ensures ease of use and long-term flexibility. An item that does not correspond correctly to the image acquired is immediately rejected. The capacity of PatMax to acquire a sufficient amount of detailed information ensures not only zero defects, but also a negligable rate of false rejects.
The simplicity and reliability of the solution permits Huhtamaki to maintain a high level of production without compromising quality, to the extent that the company is considering applying the technology to other areas of their production.
Cognex cameras in action at Huhtamaki
There are other issues with labeling. For example, EAN codes on the outer packaging are not allowed to be used for the identification and control of the production facilities – since they would cause problems with the merchandise management system of the customers. For this reason, matrix codes represent a convenient alternative to bar codes.
With about 1,600 employees, Zentis in Aachen is not only one of the largest jam and marmalade producers in Germany, but it also supplies cereals, raw marzipan, candy bars and marzipan eggs to retailers and processing companies. At the ‘Plant II’ in Aachen-Eilendorf, 190 employees make approximately 20,000 tons of confectionery per year.
To be able to definitively identify the origin and course of the products within the production, as well as for retracing purposes, the intraplant logistics was reorganised at the beginning of 2005. In doing so, a special solution was called for at the end of the production lines, where the goods are packaged in automatically-erected covering boxes and conveyed via various tracks to the palletising stations. For the reliable identification of the cartons prior to the discharging to the proper station, the manufacturer decided on Simatic VS130-2 vision sensors from Siemens Automation & Drives.
‘Many of our customers, in particular discounters, demand that the outer packaging is no longer marked with conventional EAN or UPC codes, in order to avoid potential conflicts with the codes of the respective merchandise management system and corresponding problems at their scanner checkouts,’ explains Achim Minten, who, as European maintenance manager at Zentis’ Plant II, is responsible for the automation. ‘We therefore print a data matrix code onto the cartons, which is read/ decoded and analysed via camera systems.’
This not only meets the requirements of the retail partners, but offers functional advantages as well, since the data matrix code (DMC) can very conveniently be adjusted in shape, content and size to the individual needs. Thus, at Zentis, the production line, palletising station assignment and article number are accommodated in a dot matrix measuring just 12 x 12mm.
The decision for employing the VS130-2 sensors to detect the cartons prior to them being discharged to the terminal stations was backed by three major factors: the convincing performance of the camera system at high-throughput speeds, the support from Siemens in the development of an industry-suitable solution, and its time-saving and cost-saving integration concerning compatibility with existing automation technology.
During the planning stage, special attention was paid to the interplay between DMC encoding and the automatic reading station, since the system had to operate, error-free, at speeds of up to 0.7m/s. On account of this basic requirement alone, several competitive systems were ruled out from the start, since they only could read the matrix code while it was stationary. The vision sensor VS1302 also convinced, with further performance advantages. Thus, the system detects DMCs on the most varied surfaces and – depending on the lens – offers variable reading distances from 75 to 3000mm.
Another plus point was the availability of a VS1302 version with C or CS mount sensor head, which opened up a flexible solution that passed the restrictive guidelines of the International Food Standards (IFS) agency with regard to the use of glass in the food production. In this case, the lens of the camera is covered by a protective housing with a front shield made of unbreakable plastic. The image field of the lens used with the VS1302 can be adjusted to the respective application.
Zentis has always preferred to employ automation technology from Siemens, and it has again proven itself: ‘We know most of the systems very well, and the required know-how can usually be attained with a single training course. The integrated Siemens approach with regard to configuration, data management and communication makes all areas of the work easier for our engineering department – from the setup and changes, up to the quick error correction, even without external help.’
The conveyor operation is controlled by a highly available Simatic S7-400H controller and visualised via a WinCC user interface. Besides the data acquired by the camera, a status count is also displayed here, i.e. how many cartons of a palletising job have already been detected. In doing so, the counting criteria can be user-defined.
‘With the new system, we could improve our logistics and quality control equally, since a retracing of the products down to the individual machine and corresponding batch is possible via the DMC marking. Should complaints arise, the error cause and the path of the affected products can thus quickly be reconstructed.’ Before the introduction of the new system, the carton data had to be acquired by hand and written down on a pallet label; for an integrated data acquisition, this was associated with correspondingly more work.
The success of the automatic reading, with the DMC encoding of 22 x 22 matrix dots employed at Zentis, depends on the perfect interplay between all components. After all, the edge length of the dots amounts to only 0.54mm. The Simatic VS130-2 compensates for all deviations of the DMC that might occur during its printing.
The CCD sensor, with a resolution of 640 x 480 pixels, provides a display area that is sufficient to optimally acquire the data matrix code to be read in spite of system-dependent fluctuations such as positioning and trigger inaccuracies or print position. The lens employed, and the automatic adaptation of the read algorithms, deliver a reliable read result – even with the sharpness and size fluctuations.
The best possible illumination of the DMC is ensured by a ring light. The ring light is controlled by the analysis unit of the VS130-2, synchronous to the image taking. This enables an exposure time at maximum light intensity with a duration from 20µs to 10ms. The image-taking and the illumination are triggered by a light barrier.
‘A significant advantage of the image processing system is that the analysis device possesses Profibus, as well as Ethernet interfaces,’ adds Achim Minten. With them, typical interface problems caused by additional interface boxes or converters are ruled out from the beginning. Furthermore, the configuration technician can directly access the analysis device from any networked terminal for changing the parameterisation, or for calling and activating one of the 15 storable parameter records. For this, the analysis device receives a fixed IP address on the network. It goes without saying that the access can be password-protected.
A special feature of the Simatic VS130-2 is also that it does not have to be programmed or parameterised like a conventional image processing system, since it automatically configures the illumination and learns the detection algorithms without user instructions by means of a code pattern. This self-parameterisation can take place while the conveyor is running under real operating conditions, and can either be started on-site at the device or remote-controlled via the integrated, web-based user interface. It also activates automatically, if a read process fails while in operation. Thus, the system offers maximum reading reliability independent of operator inputs.
Already planned is the use of additional systems in the area of the four carton erectors. A DMC will then be printed onto the cardboard blanks according to the requirements of the production lines. This DMC is then automatically read and triggers the distribution of the erected cartons to the respective production line. This increases the productivity and reduces the risk of errors.