INTELLIGENT CONTROL OF MATERIAL HANDLING SYSTEMS - Environmentally Conscious Materials Handling - page 99

imagine the increase in complexity if duration of class sessions were allowed to
vary dynamically in length. The argument could be that to take full advantage
of the resources (teachers and classrooms), teachers should only stay as long as
required for the students to understand the topic, but bounded by a minimum and
maximum timeframe.
This case study is based on a project conducted in collaboration with Den-
mark's most well-known manufacturer of high-end audio and video products.
The products are respected worldwide for their extremely high-quality finish and
design, and the investigated production facility is the process that gives the sur-
face of the product the high-quality finish. The process is known as an anodization
process that increases the corrosion resistance of aluminum, but coloring of the
surface is also part of the process.
In a generalized and simplified form, the problem could be described as a
number of chemical baths, which the items have to visit according to a prescribed
recipe. Besides containing information about which baths to visit and in what
order, the recipe also gives an allowed time frame for the item to stay in each
bath. Items are grouped on bars with the same recipe, but a mix of different bars
(that is, different recipes) could be processed at the same time in the production
system.
The system consists of about 50 baths, and a typical recipe would have
roughly 15 to 25 baths to visit. Even though all recipes do not have to visit
all kind of baths, there is still room for additional baths of the same type
to overcome bottlenecks, as the processing times in the baths types vary a
lot. Thus, the recipe contains only bath types, not bath number, and it is the
task of the control software to allocate a specific bath among duplets for
every bar.
Three slightly overlapping cranes move the bars from one location to another
in the array of baths. Here, a simplified notion for the movements will be used,
but in practice, they are more complex than that, because moving between some
specific baths includes subprocesses such as rinsing the bar of items, opening
and/or closing the lid of a bath, and so on, but it comes down to an estimated
travel time of moving a bar from bath u j to bath u j + k . In general, the cranes are
not considered to be a bottleneck in the production system, as they handle the
tasks quite sufficiently.
Apart from the baths and cranes, an important part of the system is the input
buffer, where typically around 30 bars are waiting to be processed. This also is
an important focus point for the control software, because choosing the best bar
to fit the current configuration is the key to optimizing throughput. A general
overview of the system is presented in Figure 3.24, where the C are the cranes
with a domain (how far they can move) and the U are all the chemical baths
into which the bars can be placed.
At first sight, the problem described seems to be a candidate for classic opti-
mization and scheduling principles, but as already mentioned, the allowed time