proposed by | Barbara Smith bms@scs.leeds.ac.uk |
The following example is based on data from an order for cartons for different varieties of dry cat-food.
Variation | Order Quantity |
Liver | 250,000 |
Rabbit | 255,000 |
Tuna | 260,000 |
Chicken Twin | 500,000 |
Pilchard Twin | 500,000 |
Chicken | 800,000 |
Pilchard | 1,100,000 |
Total | 3,665,000 |
Because in this example there are more slots in each template (9) than there are variations (7), it would be possible to fulfil the order using just one template. This creates an enormous amount of waste card, however. We can reduce the amount of waste by using more templates; with three templates, the amount of waste produced is negligible. The problem is therefore to produce template plans which will minimize the amount of waste produced, for 1 template, 2 templates,... and so on.
It is permissible to work in units of say 1000 cartons, so that the order quantities become 250, 255, etc.
A variant is to allow up to 10% under-production of some designs, if this allows the overall over-production to be reduced. This is not a sensible option for the catfood problem, because it leads to under-production of all the designs.
The optimal solutions for the catfood problem are shown below. For each template, the table gives a list of the number of slots allocated to each design, e.g. [1,1,1,1,1,2,2,] means that 1 slot is allocated to each of the first five designs and two each to the last two.
No. of | Layouts | No. of Pressings | Total pressings |
templates | of each template | ||
1 | [1,1,1,1,1,2,2] | 550,000 | 550,000 |
2 | [0,0,0,0,0,2,7] | 158,000 | |
[1,1,1,2,2,2,0] | 260,000 | 418,000 | |
3 | [0,5,3,0,0,1,0] | 51,000 | |
[0,0,1,0,0,7,1] | 107,000 | ||
[1,0,0,2,2,0,4] | 250,000 | 408,000 |