Energy-Aware Appliance Scheduling for Peak Price Savings

ISEF Category: Engineering Technology: Statics and Dynamics

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

The Hook

Electricity prices can jump fast, and your washer does not care. A smart scheduler can decide when to run appliances so you pay less without changing how you live. That sounds like a tiny software trick, but it connects to power grids, household behavior, and operations research. You can test whether your model really saves money on real price data.

What Is It?

This project asks a simple question with a complex answer, when should home appliances run so you spend less on electricity? You are not building a smart appliance. You are building a decision system that picks times for tasks like laundry, dishwashing, or EV charging based on price and predicted household demand.

Think of it like planning a road trip around traffic. The appliance load is your car, the electricity price is traffic, and the scheduler tries to avoid the worst congestion. Bayesian forecasting means you update your guess as new information comes in, instead of pretending tomorrow is identical to today. That makes the project feel realistic, because home energy use changes with habits, weather, and schedules.

Why This Is a Good Topic

This is a strong science fair topic because you can measure a clear outcome, cost savings, and compare it across different scheduling rules. The problem matters to real households and the electric grid, especially during peak pricing. You can learn forecasting, optimization, and data validation without needing a wet lab. You also get a clean way to judge success, since your model can be tested against real price data and actual appliance logs.

Research Questions

• How does a Bayesian forecast of household demand change the total electricity cost compared with a fixed schedule?
• What is the effect of using real CAISO or PJM price data instead of average daily prices on estimated cost savings?
• Does adding appliance priority rules reduce peak-time energy use without increasing total runtime?
• To what extent does forecast error change the scheduler's performance on cost and load shifting?
• Which scheduling rule saves more money, a price-only rule or a price-plus-demand rule?
• How does the number of controllable appliances affect the best possible cost reduction?

Basic Materials

• Laptop or desktop computer with internet access.
• Spreadsheets or Python for data analysis.
• Smart plugs with energy monitoring capability.
• INA219 current sensor or similar power-monitoring module.
• Microcontroller such as Arduino, ESP32, or Raspberry Pi for logging.
• Household appliance or appliance simulator setup.
• Access to CAISO or PJM historical price data.
• Notebook for experimental logs and decisions.

Advanced Materials

• Data logger with timestamped voltage, current, and power readings.
• Microcontroller development board with external storage.
• Relay module rated for the appliance test load, used only under proper supervision.
• Calibration load or reference device for sensor checking.
• Access to multiple smart plugs for parallel appliance logging.
• Weather data source for demand modeling.
• Local electricity tariff records for validation.
• Optional EV charging simulator or controllable resistive load.

Software & Tools

• Python: Cleans price data, fits the forecast, and simulates scheduling rules.
• Pandas: Organizes time series data from prices and appliance logs.
• NumPy: Handles calculations for cost, load, and summary statistics.
• Matplotlib: Plots price spikes, demand forecasts, and savings results.
• Jupyter Notebook: Keeps code, notes, and charts in one reproducible file.

Experiment Steps

1. Define which appliances you will model first and which ones you can safely monitor or control.
2. Collect historical price data and match it to your own household demand or logged appliance use.
3. Choose one forecasting approach, then compare it with a simple baseline so you know if it adds value.
4. Build a scheduling rule that translates forecasted demand and price into a run-time decision.
5. Validate the schedule with smart-plug or INA219 logs, then compare predicted cost to measured cost.
6. Analyze how forecast error, appliance priority, and tariff spikes change the results.

Common Pitfalls

• Using appliance labels without measuring real power draw, which makes the cost model drift from reality.
• Mixing price zones or tariff types, which creates fake savings that do not match one market region.
• Ignoring startup spikes and cycle lengths, which makes the schedule look better than it works.
• Training the forecast on the same period you test, which hides overfitting and inflates accuracy.
• Forgetting to align time stamps between smart-plug logs and market prices, which breaks the cost calculation.

What Makes This Competitive

A competitive version goes beyond a simple cost calculator. You would compare several scheduling strategies, test them on separate time periods, and report both savings and forecast error. Strong projects also quantify uncertainty, not just the average result, so you can say when the scheduler helps and when it fails. If you include real logs from multiple appliances and a careful baseline, your work starts to look like a real industrial engineering study.

Project Variations

• Focus on one flexible load, such as EV charging, and compare charging windows under different tariff structures.
• Replace Bayesian forecasting with a simpler baseline model, then test how much accuracy you lose.
• Study how weather, weekday patterns, or school calendars change household demand and scheduler performance.

Learn More

• CAISO Open Access Same-Time Information System: Search the CAISO site for historical prices, demand, and market data tables.
• PJM Data Miner 2: Search PJM for historical energy prices and load data for schedule testing.
• U.S. Energy Information Administration, Electricity Data Browser: Use this government database for tariff context, load patterns, and price history.
• MIT OpenCourseWare, Introduction to Operations Research: Find lecture materials on scheduling, optimization, and decision models.
• PubMed: Search review articles on household energy use, demand response, and user behavior for background reading.