Most of Unilever’s food processes consist of a large number of ingredients with limited storage capacity, a small number of process plants, a large number of intermediate product storage facilities and a smaller number of packing lines. The practical production scheduling inside the vast majority of these factories focusses on scheduling the packing lines on the production floor. The schedule is ‘thrown over the wall’ to the process department, in which a schedule is being made to satisfy the packing demand. This schedule is also "thrown over the wall" to the incoming materials department, etc. This way of scheduling poses three problems: 1. There is no clear insight in where the bottlenecks in the whole process are, resulting in a reduced production capacity. 2. Any change in the packing schedule might lead to an infeasible schedule in the upstream departments. As a result, packing lines may not run due to lack of intermediate products, wrong intermediates being made in the process plant, etc. 3. Each department will strive to ensure that their department is not to blame for not packing products, hence less available capacity will be communicated to the plant. The challenge is to reduce the impact of these problems in order to increase the capacity of the factory and reduce the product cost/tonnes.
Problem This paper describes a methodology is to reduce the impact of these problems by modelling the factory operations. This is done by building a multi-stage scheduling model which describes the infra-structure of the factory, which products are being produced and how the plant is operated. The key challenge is to translate the complexity of the plant (and the operations) into a simplified, but realistic, multi-stage scheduling model. This model of the whole factory, including the constraints, is used to schedule the whole plant, maximising the production capacity and minimising the impact of the above described problems.
Main results This paper describes a methodology to translate the complexity of the plant into a simplified model that can be used to schedule all relevant plant operations. The methodology consists of six steps: 1. Based the process flow diagrams, interviews and standard operating procedures a factory structure model is built. 2. Including the bill of materials and product routing a material flow structure is built. 3. Combining the above structures and taking into account the change-over structure the factory model is built. 4. This factory model is used to specify which data is to be retrieved from the existing factory systems into the data model. 5. The simulation model is implemented by combining the factory model, data model and the operational inputs. 6. The model is firstly verified with the operators in the plant, followed by validation by running the plant by the model.
Conclusions & recommendations The above methodology was applied to an ice-cream factory. The resulting model was validated and shown to describe the plant with sufficient accuracy. Operational use of the model proved a production capacity increase of 10 – 30 %
|Titel||Proceedings of the 19th European Symposium On Computer Aided Process Engineering, (Escape-19) 14-17-June 2009, Cracow Poland|
|Redacteuren||J. Jezowski, J. Thullie|
|Plaats van productie||Poland, Cracow|
|ISBN van geprinte versie||978-0-444-53433-0|
|Status||Gepubliceerd - 2009|
|Naam||Computer Aided Chemical Engineering|
|ISSN van geprinte versie||1570-7946|