Minimizing set-up times in production has been a complex and time-consuming issue up to now. However, there is a new way to achieve shorter set-up times more easily and quickly: article classification.
In order to optimize set-up times, matrices were previously set up in which each article appears as a predecessor and as the successor of each other. For example, with 100 articles, such a matrix comprises 100 × 100, i.e. 10,000 cells. The numerical value in each cell represents how much effort it takes to switch from one article to another.
The first filling of the setup time matrix is usually very time-consuming. Later, the maintenance effort is very high, because after a relatively short time it is hardly possible to understand how the individual values came about. The reason for this is that setup times usually depend on several factors, which are summarized in a single numerical value in the setup time matrices.
Only the cause counts
FLS uses a fundamentally different approach for setup time optimization in the production planning system FEKOR – article classification. FLS works with article characteristics, which are individually created depending on the respective process.
The core idea here is always to work out the actual causes for the occurrence of a setup time, because the total time for the setup consists of several parts. Therefore, when changing between articles, the times for converting the machines, cleaning or heating up to a different processing temperature, for example, must be considered separately. FLS illustrates these causes by means of features.
Few features are sufficient
Characteristics can be, for example, the material, its color or the shape of the packaging. The times for changing from one characteristic value to another are entered in a list of characteristics. Here, for example, it is stored that the change of material from A to B or C takes five minutes. For the characteristic “color level”, for example, it is entered that the change from light material to dark material takes eight minutes, but the change from dark material to light material takes twenty minutes.
Important characteristics in an extrusion line are often the mass (PVC or polyethylene), the colour or the temperature profile during processing. In the chocolate industry, characteristics such as “bar colour” (white or not white), “bar mass” (whole milk, bitter, mocha, nut) or “format” (large or small package) are used.
For each production stage, any number of characteristics can be created that are tailored precisely to the respective application. Usually two to five are sufficient.
Characteristic lists are typically created for a specific manufacturing level and there for all affected articles. In this way, you do not have to assign values to hundreds and thousands of individual article combinations. Instead, the values from the corresponding characteristic lists are selected and defined only once for each article.
Changes are quickly implemented
If the time required for setup is reduced, for example, by using a quick-clamping device, only the time for this characteristic is adjusted. All other times for other characteristics remain unchanged. The change then applies immediately to all articles that pass through this process step – without having to adjust each individual article.
Further maintenance is limited to selecting the corresponding characteristic values for new articles. The effort to think through and define all combinations with other articles is completely eliminated.
These are two of the main reasons why work with article classification is significantly faster, simpler and more transparent than with setup time matrices. The author does not know of any other supplier who approaches the topic of setup time optimization in this way.
Once the articles have been classified, it is left to the system to calculate the different combinations and decide which one is the optimum – a classic example of how existing expertise is anchored in a system and the monotonous computing work is then left to the computer. In this way, employees gain time to concentrate on making important decisions or preparing them, for example talking to suppliers about partial deliveries or agreeing on additional shifts or overtime in the event of bottlenecks.
Decisive changes of direction
The article classification takes into account the different duration of the individual work steps: In an extrusion process, for example, where there are different temperature profiles, the temperature is 180 °C for some products and 130 °C for others. Since heating is always faster than cooling, changing from 130 °C to 180 °C takes less time than changing in the opposite direction. When filling beverages, the system takes into account that changing from “Cola” to “Non-Cola” requires more cleaning than changing in the opposite direction.
To minimize setup times, the system analyzes all characteristics for each change and adds the individual times. For simultaneous operations, it uses the respective maximum. It also uses the longer time if a long cleaning process includes a shorter one.
Of course, the system also takes into account that there is no setup time if successive articles have the same characteristic values. For optimization, this means that as many articles as possible with the same characteristics are planned one after the other, and then changes with short setup times are carried out.
Traceable at any time
A further advantage is that all characteristic values are transparent and traceable at all times, because there is a concrete time behind each one. If you later want to check how a setup time came about, you can display the corresponding characteristics and control them easily.
Articles newly included in the user’s production program are only classified once and then incorporated into the entire setup scheme. Only if it becomes necessary to define completely new characteristics does the classification have to be updated once with this new characteristic.
An example from practice
A project at a customer in the rubber industry shows the high benefits of classification: The mixing process for 100 kg of rubber takes six minutes. If the mixer has to be cleaned, it also takes six minutes. If the sequence is not optimally planned, the worst-case scenario is 50 percent plant utilization. If the sequence is planned optimally, for example, cleaning is only necessary after every tenth mixture. The utilization rate can then reach almost 90 percent. In this case, the investment in a new mixer may become superfluous.
When analyzing the set-up times, the most important thing was to determine when the machine only had to be cleaned. One feature was the change from soft rubber to hard rubber, because soft rubber sticks to the dough hook, hard rubber does not. Another criterion was that certain oils, which may be contained in successive compounds, react with each other. Depending on the combination, cleaning had to be carried out when changing from one grade to the other.
Five further criteria were developed in the project. After a few test runs, two more were added and the optimum sequence was then achieved. The development of the characteristics took about half a day, after which the characteristic values were assigned to the individual articles. All in all, the analysis reduced setup time by 35 percent.
Article classification is a valuable tool for optimizing set-up times: In contrast to the set-up time matrices previously used, it is much simpler, faster and more transparent. A major reason for this is that changes in the production process do not require the evaluation of hundreds and thousands of individual article combinations, but rather the selection of values from previously defined feature lists only once for each article. On some systems, this method was used to halve set-up times simply by optimally combining the production sequence.
Author: Dr.-Ing. Hanns Jürgen Hüttner, Managing Director, FLS FertigungsLeitSysteme GmbH & Co. KG, Eschweiler, Germany