When creating a production scheduling model it could be tempting to consider the production process as simply a series of processes.
However, within real production processes there will be dependencies between resources, queues, availability of materials and varying processes.
A useful model will need to have these relationships in the model.
This article will show you how to create a fundamental production flow simulation and what considerations you should make as a developer and industrial engineer.
- Production Process
Identify the key processes involved in the production.
Simplified this could look like:
Material Input
|
v
Preparation
|
v
Machine A
|
v
Machine B
|
v
Inspection
|
v
Finished Products
This is a very simplistic representation. In reality production may have parallel processes, shared resources and alternate routing.
- Time Taken at Each Process
Decide how to represent the time taken to perform each part of the production.
Where possible try to gather operational data to understand how long each process will take. Do not make assumptions that each process takes the same amount of time to perform. Consider any uncertainty and assumptions made.
- Queues
Queues can occur when there is a waiting period for resources.
Consider the following for each queue:
Where does the queue occur? What jobs go into the queue? What resource is used to process jobs? How long do jobs wait in queue? How does this affect downstream activities?
Queue times can be an important metric to consider when evaluating your model.
- Shared Resources
Shared resources can cause dependencies between processes.
For example two processes may need to use the same machine. The simulation should include when this resource is available and how the conflicting processes are scheduled.
- Set Up Times
Where set up/change over times are affecting production times they should be included in the model. A schedule which doesn’t consider these transitions may not give an accurate indication of how long it would take to complete a series of jobs.
- Material Availability
Some production processes may depend on the availability of certain materials. A useful model will consider these dependencies.
- Scheduling Policies
Once you have created a model which includes the correct constraints you can investigate different scheduling policies.
This isn’t to assume one policy will be better but to see how different options compare based on clear metrics.
Metrics could be:
Metric Question
Queue time Is the model considering queue times?
Cycle time How long does work take to complete?
Throughput How much work is being completed?
Work in Progress How much work is incomplete?
Schedule Can the proposed schedule be implemented?
- Model Validation
The production flow model should be validated against operational data.
This could be done by creating a small production flow twin and calibrating it using historic event logs. Then comparing the proposed schedules to the historic replay. If appropriate a limited controlled trial should be performed to investigate the results.
Don’t assume a model is accurate just because the results seem reasonable.
Final Thoughts
Production scheduling is a system wide issue. There are interactions between queues, resource contention, set up times and material availability.
By considering these relationships engineers can investigate different production policies more effectively.
Aperture Venture Studio investigates these issues in their Industrial Digital Twins and Simulation research.
Find out more:
https://apertureventurestudio.com/physical-ai-aiot-research/industrial-digital-twins-and-simulation/
Top comments (1)
DIGITAL TWIN