Basic function for setting the priority

in the Grenzebach-optimization-system

1. Job parameters
The Grenzebach job manager provides and assigns several parameters to each job. Example:
- Length
- Width
- Quality
- Value, corresponding to a priority
- Stacker destination
- ...

2. Optimization principle
- A new optimization run is triggered when the rear edge of the last optimized glass plate passes a
defined mark in glass flow direction (located at approximately 1 m upstream of the longitudinal cutting
bridge).
- Starting from the rear edge of the last optimized glass sheet and viewing in the direction of the defect
detection system, the optimization system generates different combinations out of all enabled jobs.
- Whilst doing this it respects, under all circumstances, the required glass quality.
- It first attempts to produce the job having the highest value.
- In the event this job cannot be produced (for example due to quality requirements) the optimization
system attempts to assign this queue position to the job having the next lower value.
- Moreover, the system also attempts to assign specific queue positions to other jobs of less significant
value, striving to obtain the best possible overall result.
- This process yields a number of different combinations with respect to the jobs to be produced,
combinations that are in the end competing with each other (up to a defined combination-comparison
length).
- The best combination is selected and the data of the first glass plate or cap in this winning
combination are transmitted to the cutting control system.
- The optimization system now waits for the rear edge of this plate or cap to pass the above described
mark in glass flow direction. Then a new optimization run is initiated.

3. Value affecting production

The assignment of a value / priority to each job to be produced has a decisive impact on the results of the
Grenzebach optimization system. The entered value can range between [10 ... 999]. Closely ranging
values signify, that the corresponding jobs are nearly equivalent. If, on the other hand, the single values
largely differentiate, this signifies to the optimization system, that one job is significantly more important
than the other.
The optimization system applies the following formula to compute the glass yield:
yield = sum of all plates (length * width * value)
(This implies that the glass yield is surface-related.)
The yield for cullet plates is 0.

Principle_Of_Optimization.doc page 1 of 4
4. Example
The following, simple example is to clarify the effect different values take on the optimization result.
The following glass qualities are assumed having been defined:
quality: ADF: absolute defect free: no defect allowed
TST: defect class 1, cl 2, cl 3 are allowed

The following jobs are assumed to be available:

order length width quality
#1 2000 3000 ADF
#2 3000 3000 TST

For a better understanding of the process, we now exemplary consider 3 possible and different
combinations the optimization system could generate (refer to the graphics on page 3).

Comb. A:
Position 1: Job #1 can be produced
Position 2: #1 = OK
Position 3: because of defect, job #1 not possible
#2
Position 4: #1 possible
Position 5: #1 not possible
#2

Comb. B:
Position 1: Job #1 can be produced
Position 2: #1 = OK
Position 3: because of defect, job #1 not possible
#2 could theoretically be produced, but
also cullet plate possible
Position 4: #1 = OK
Position 5: #1 = OK
Position 6: #1 not possible
#2

Comb. C:
Position 1: Job #1 can be produced
Position 2: #1 = OK
Position 3: because of defect, job #1 not possible
#2 could theoretically be produced, but
also cullet plate possible
Position 4: #1 = OK
Position 5: #1 = OK
Position 6: #1 not possible
#2 could theoretically be produced, but also cullet plate
possible
Position 7: #1 = OK

Principle_Of_Optimization.doc page 2 of 4
 scanner line

Comb. Comb. Comb.

A B C

#2 #2 #1

cullet

#1 #1 #1

#1 #1
#2 cullet cullet

#1 #1 #1

#1 #1 #1

glass flow direction combination compare length

optimization start

order #1 order #2

cullet defect class 3

Principle_Of_Optimization.doc page 3 of 4
5. Evaluation of the various combination
Using the formula
yield = sum of all plates (length * width * value)
an overall assessment and a comparison of all combinations may be effected.

5.1. Values slightly differing

The system operator assigns the following relatively close values to the 2 jobs:
order length width quality value
#1 2000 3000 ADF 300
#2 3000 3000 TST 290

This yields the following overall assessment of the different combinations:

Comb. A: total value = 3 plates *(2m*3m*300) + 2 plates *(3m*3m*290) = 10620 <- best
Comb. B: total value = 4 plates *(2m*3m*300) + 1 plate *(3m*3m*290) = 9810
Comb. C: total value = 5 plates *(2m*3m*300) = 9000

5.2. Values largely differing

The system operator assigns the following relatively wide spread values to the 2 jobs:
order length width quality value
#1 2000 3000 ADF 300
#2 3000 3000 TST 120

This yields the following overall assessment of the different combinations:

Comb. A: total value = 3 plates *(2m*3m*300) + 2 plates *(3m*3m*120) = 7560
Comb. B: total value = 4 plates *(2m*3m*300) + 1 plate *(3m*3m*120) = 8280
Comb. C: total value = 5 plates *(2m*3m*300) = 9000 <- best

6. Summary
Assigning two closely ranging values (300 - 290) as in case 5.1 signifies to the optimization system, that
both jobs are of near equal importance. Consequently, the optimization system will prefer combination A
for production in this exemplary case.

Assigning two largely differing values (300 - 120) as in case 5.2 suggests to the optimization system that
job #1 is much more important that job #2. Choosing combination C allows the optimization system to
produce 5 good quality plates (and 2 cullet plates) instead of only 3 good quality plates. Consequently,
combination C is preferred.

This clearly shows the importance of the value assignment by the operator considering the decisive
effects it has on the optimization results.
The major advantage compared to simple prioritization (e.g. priorities ranging between [1 ... 10]) is the
possibility to weight single priorities. The weighting is obtained by the different relations between the single
values.

Of course, the optimization result is subject to still other influences such as for example:
- Position of the defect in glass ribbon
quality
direct effect on the selection of the job to be
produced
- Variation of the mixed jobs with respect to their length and width
- Restrictions due to bottlenecks in the processing line such as
- Angle transfer conveyors or other conveying bottlenecks (cycle times)
- Unfavorable ratio between annealing lehr speed and conveyor line speed
- Stacking bottlenecks
- Restrictions due to bottlenecks in the cutting equipment such as
- Number of available cross cutting bridges
- Cutting tool settings / repositioning at the longitudinal cutting bridge
- ...

Principle_Of_Optimization.doc page 4 of 4