In the food industry, a change may happen every few hours and
cause scheduling nightmares for producers. Special offers and rapid
changes in demand force manufacturers and distributors of food and
other short shelf-life and fast turnover products into hectic rescheduling
of their operations. A single food line might have the capacity to
manufacture 400 types of products, and every change of product requires
rescheduling the supply of numerous raw materials.
This project developed a software toolbox that enables companies to
be much more efficient in scheduling such production, which is a better
use of resources. In addition, ScheduleZ allows them to be more responsive
to their customers' changing demands.
Shop floor managers in food manufacturing
plants spend much of their time trying to juggle customers' rapidly
changing demands for an often random range of products. Because
many producers of fast turnover foods promise to meet customers'
orders within 24 hours, they have to rapidly clean their production
lines and switch them to different products before they can supply
the correct raw materials and packaging materials from a vast range
The problems of such batch production are compounded by the need
to cater to frequent special offers, such as the addition of a free
packet of crisps in a multi-pack of different flavours, and for
the use of special packaging. Customers often end up dissatisfied.
ScheduleZ was set up to produce a software tool box that would allow
manufacturers and distributors to create plans and schedules for
more efficient production. Although there were software systems
already on the market, they did not take account of the particular
conditions of the food and other fast moving manufacturing sectors.
ScheduleZ, coordinated by the systems company MJC2 of the UK, combined
several commonly used approaches to the development of software
in order to exploit their best features. The different approaches
may be suitable for different scheduling circumstances. For example,
rule based or artificial intelligence systems, can be useful in
some circumstances, but in practice they are slow and it is difficult
to generate enough rules.
A second approach involves the use of operational research based
on matrixes, but not every aspect of a system can be defined by
mathematical models. The ScheduleZ team included two large end users
of scheduling systems, Compal of Portugal and KP Foods of the UK.
Together with the system houses MJC2 and Syco, the Portuguese research
organisation Uninova, and Italy's main engineering university PoliTo
in Turin, the partners looked at existing models and generic algorithms
and developed their own software systems. (Eventually KP largely
withdrew from the project because of internal problems.)
The collaborators tested two demonstrators. The system at Compal's
fruit juice and tomato pulp factory was based on the G++ programming
language developed from C++, and the system installed at KP's potato
and cereal snacks factory combined Prolog, C++ and a G++ based Scada
The Compal system initially combined the annual planning at the
factory and short-term planning for the following month, but this
was found to take too much computer time and was replaced by a heuristic
approach based on 'greedy' software rules. The KP system was developed
with the emphasis on relating staffing to the use of machines and
production at the factory, although it was general enough for many
other factories manufacturing fast-moving consumer products.
Four largely independent relationships were established which could
be built up into an accurate staffing model by summing their contributions.
The resulting system can define the number of shifts needed each
day, the lengths of shifts and the start times, and is seen as a
major benefit of ScheduleZ. For example, shift configurations can
be related to the changes that sales and production make in the
stock levels of different products, so that the programme can take
account of the effect of a schedule on stocks as it works out the
best production plan. As a result, the system can also produce stock
ScheduleZ can also cope with the manufacturing technique of specifying
acceptable stock levels rather than specific production levels.
The user of ScheduleZ has been given more control by the introduction
of schedule frameworks which allow factories to continue with their
existing work patterns and practices within ScheduleZ systems. This
should ensure that there can be a smooth switch over to automated
A generic optimising algorithm to allow efficient changeover and
delivery deadlines has also been developed based on constraints.
The constraints include timings such as speed of bagging and the
earliest possible start of production, changeover targets, unavailability
and the pre-assignment of resources. The resulting ScheduleZ can
take into account scheduling parameters, machine configuration,
manufacturing constraints and product information.
The project was also very successful in its use of OOD reusable
code libraries and software based on frameworks. A distribution
framework, Colombo, was adopted to allow a prototype of an application
to be turned into a distributed working version. Typically this
means that a system can be set up throughout a factory on a client/server
Pepsico's Walkers crisps factories in the UK uses a PIMSS software
system based on ScheduleZ to schedule production at four plants.
Prior to the use of PIMSS, the Walkers factories were achieving
a customer satisfaction of about 75%; this rose to 98% within four
weeks of adopting the scheduling package. Similar improvements have
been recorded elsewhere. To ease the use of the software, the collaborators
developed a strategic planning editor which allows an operator to
provide inputs and see the production plan graphically, split into
periods as short as one hour over up to a year, along with measures
such as stocks and accumulated production.
The operator can also choose to display data as lines, bars and
area charts. Tied in with ScheduleZ is a simulation framework for
Scada management of production plants, able to model and then monitor
events such as machine start-ups and failures.
The success of this project has led to MJC2 considering applying
the software to the petrol and oil industries as well as the textile
sector. The company also believes it can apply the ScheduleZ approach
to the manufacture of paper to optimise production. MJC2 believes
that it could produce up to 20 ScheduleZ systems a year worth about
2.5 MECU and generate a profit of about 1.5 MECU. This project has
also led to improvements in the courses run by Uninova and the main
engineering university of Turin.