VoltTwin

Inspiration

I only had my MacBook, so I wanted to build something that still felt like a real engineering project without pretending to use hardware I did not have. That led me to the idea of a software-based digital twin: a virtual version of a smart building that can simulate IoT sensor data, test safety scenarios, and recommend actions before real hardware is deployed.

Smart buildings depend on sensors, automation, and fast decision-making, but deploying physical systems can be expensive and difficult to test safely. VoltTwin explores how simulation can help engineers prototype those systems first.

What it does

VoltTwin is an AI-style digital twin for smart building safety. It simulates virtual IoT sensor data for a smart classroom/building, including:

• CO₂ levels
• temperature
• humidity
• smoke
• occupancy
• power usage
• air quality
• ventilation status
• emergency exit status

Users can trigger different building scenarios such as poor ventilation, overcrowding, fire/smoke risk, winter energy waste, blocked exits, and earthquake emergency mode.

VoltTwin then calculates multiple risk scores:

• Air Quality Risk
• Fire/Safety Risk
• Energy Waste Risk
• Evacuation Risk
• Overall Building Risk

It also generates AI-style safety recommendations, shows live telemetry charts, predicts near-future risk with a T+5 forecast, and records events in an incident replay timeline.

How we built it

VoltTwin was built as a frontend-only web application using React, TypeScript, Vite, and Tailwind CSS. The live dashboard runs entirely in the browser.

The simulation engine generates virtual IoT sensor values over time. Instead of instantly switching numbers, the values gradually move toward scenario targets, making the system feel more realistic. A risk engine then evaluates the current sensor state and calculates weighted scores.

The overall risk score is calculated using a weighted model:

[ OverallRisk = 0.30(AirQuality) + 0.30(FireSafety) + 0.15(EnergyWaste) + 0.25(Evacuation) ]

The recommendation engine is deterministic and rule-based. It does not call an external AI API, but it behaves like an AI safety assistant by analyzing simulated sensor values, risk scores, and the active scenario to produce contextual recommendations.

Challenges we faced

One of the biggest challenges was making a software-only project feel like a real IoT engineering system. Since I did not have physical sensors, I had to be careful and honest about the project’s scope. All sensor data is clearly simulated.

Another challenge was designing risk logic that was understandable rather than random. I wanted the system to react in a way that made sense: CO₂ should rise during poor ventilation, smoke should increase fire risk, blocked exits should increase evacuation risk, and mitigation actions should visibly reduce danger.

I also had to debug TypeScript build errors, Tailwind setup issues, GitHub deployment problems, and Vercel deployment settings during the build process.

What we learned

I learned how digital twins can be used to prototype physical systems before real hardware is deployed. I also learned how to structure a React + TypeScript application with modular simulation logic, risk scoring, recommendation generation, charting, and incident reporting.

Most importantly, I learned that a strong hackathon project does not always need physical hardware. A well-designed simulation can still demonstrate engineering thinking, system design, and real-world impact.

What’s next

Next, I would like to expand VoltTwin into a multi-room building simulation with floor-by-floor risk tracking. I would also add CSV exports, more advanced forecasting, real sensor compatibility, and optional integration with actual IoT devices such as ESP32 or Arduino sensor nodes.

Safety note

VoltTwin uses simulated virtual IoT data only. It is a software prototype for smart-building safety research and is not a certified emergency system, medical device, or official safety authority.

Built With

• ai-style
• css
• digital
• engine
• frontend-only
• github
• local
• lucide
• modeling
• react
• recharts
• recommendation
• rule-based
• simulation
• tailwind
• twin
• typescript
• vercel
• vite