3D Printer Job Sequencing to Cut Filament Waste

ISEF Category: Engineering Technology: Statics and Dynamics

Subcategory: Industrial Engineering-Processing  ·  Difficulty: Advanced  ·  Setup: Home Setup  ·  Time: Full Year

The Hook

Every print swap on a 3D printer can dump fresh filament into the trash. That waste adds up fast when you run many small jobs. You can treat each print like a scheduling puzzle and test whether smarter sequencing saves material. This turns a home printer into a real operations research project.

What Is It?

A Markov decision process is a way to make choices when today’s decision affects tomorrow’s options. In this project, each print job is one choice, and each choice can change how much purge filament, setup time, or failed material you need next. Think of it like packing a backpack for tomorrow’s classes. What you choose first changes how much room you have later.

FDM 3D printers often waste filament during color changes, nozzle swaps, or material changeovers. That waste comes from purging old material so the next print starts clean. Your model asks a simple question with real math behind it, can you order jobs in a smarter way so the printer spends less material on cleanup? You can test that idea on a home printer by comparing your model’s schedule with a random schedule or a first-come, first-served schedule.

Why This Is a Good Topic

This makes a strong science fair topic because you can measure the main outcome, filament waste, with clear numbers. You also get to test a real industrial engineering idea on equipment many students can access at home. The project connects to manufacturing efficiency, sustainability, and cost savings. You can learn scheduling, basic optimization, data logging, and model validation from one project.

Research Questions

• How does a Markov decision-process schedule change total purge filament compared with first-come, first-served scheduling?
• What is the effect of grouping similar material or color jobs on total filament waste?
• Does prioritizing short jobs before changeovers reduce waste more than prioritizing long jobs first?
• To what extent does the number of material swaps predict total purge waste in a home print queue?
• Which scheduling rule best lowers waste while keeping average job completion time low?
• How does printer state, such as loaded material or nozzle condition, affect the best next print choice?
• To what extent does a model trained on past print jobs predict the waste from future print queues?

Basic Materials

• Household FDM 3D printer with one or more filament types
• Digital kitchen scale with 0.1 g accuracy
• Spool of standard PLA filament
• Notebook or spreadsheet for print logs
• Phone or camera for documenting setup and outputs
• Calipers for checking print dimensions
• Simple labels or tags for tracking job order
• Basic cleaning tools for the printer bed and nozzle
• Computer with spreadsheet software
• Access to slicer software used by your printer.

Advanced Materials

• Household FDM 3D printer with logging access or firmware data export
• Multiple filament types or colors for changeover testing
• Digital scale with 0.01 g accuracy
• Calipers or micrometer for more precise dimensional checks
• Computer with Python for decision modeling
• Data file export from the slicer and printer queue
• Optional filament sensor data if your printer supports it
• Controlled sample set of repeat print jobs
• Camera setup for consistent photo documentation
• Statistical software for comparing schedules.

Software & Tools

• Python: Builds the Markov decision-process model and compares scheduling rules.
• Google Sheets: Organizes print logs, waste measurements, and schedule results.
• Jupyter Notebook: Keeps code, notes, and graphs in one place for analysis.
• ImageJ: Measures print area or photo-based output traits if you document results by image.
• Excel Solver: Tests simple optimization rules for job order if you want a lower-code path.

Experiment Steps

1. Define the printer states you care about, such as material loaded, print size, or changeover count.
2. Choose one outcome metric, such as purge mass, total waste, or waste per completed job.
3. Build a baseline queue rule, then write a second rule based on your Markov decision model.
4. Plan how you will log every print so each job has the same fields, including material changes and waste measurements.
5. Decide how you will test both schedules on the same printer so you can compare them fairly.
6. Set up an analysis plan that compares model predictions with actual waste from the print campaign.

Common Pitfalls

• Measuring only finished print mass, which hides purge waste and makes the savings look smaller than they are.
• Mixing different filament brands or colors without tracking them, which confuses the effect of material changeovers.
• Letting print settings change between jobs, which adds noise that the schedule model cannot explain.
• Using too few print jobs, which makes the comparison too weak to trust.
• Comparing schedules with different kinds of prints, which makes waste differences come from model mismatch instead of sequencing.

What Makes This Competitive

A stronger project goes beyond one simple before-and-after comparison. You can make that happen by defining states carefully, testing more than one scheduling rule, and checking whether the model still works on new print queues. Strong analysis comes from clear controls, honest error bars, and a fair baseline. If you can show that your model predicts savings across different job mixes, the project looks much closer to real industrial planning.

Project Variations

• Use PLA, PETG, and TPU as separate job classes to test whether material type changes the best sequence.
• Compare color-based sequencing with size-based sequencing to see which rule saves more purge filament.
• Add print priority as a constraint and test whether the model can still reduce waste when urgent jobs jump the queue.

Learn More

• NIST Engineering Statistics Handbook: Search the handbook for process improvement, scheduling, and statistical comparison methods.
• MIT OpenCourseWare Operations Research: Find free lectures and notes on optimization, decision models, and queueing.
• USGS Data and Software: Browse examples of data logging and analysis workflows that use repeatable measurement.
• NASA NTRS: Search for papers on manufacturing efficiency, additive manufacturing, and resource planning.
• PubMed: Search for review articles on additive manufacturing waste, sustainability, and process optimization.
• Computers & Industrial Engineering: Search the journal for papers on production scheduling, optimization, and decision models.