LIVE · Lempuyangan ↔ Maguwo · KM 6.945

The rail powers its own
health check.

PiezoRail harvests the vibration of passing trains into electricity and uses it to monitor the very track it sits on — dual-IMU + thermal sensing, on-device anomaly detection, streaming live. No wiring. No grid power. Zero.

Open the live dashboard See how it works ↓
8/8Self-powered nodes online
568mW peak harvest / unit
r=0.8730-day field validation
ONLINEGateway · piezorail.online
Field-validated r=0.87 30-day correlation 568 mW/unit peak Dual-IMU + thermal ESP-NOW mesh
The story

Vibration is the problem and the power.

Three moves take PiezoRail from an unmonitored stretch of track to a live, self-powered structural health feed.

① THE PROBLEM
🚆 ⚠️

Rail faults stay invisible

Structural monitoring needs power and wiring along remote track — so most rail goes unmonitored, and early-stage faults are missed until they become failures.

② THE INNOVATION
〰️ → ⚡ → 🔋

Vibration → power

A PZT piezo transducer + LTC3588 harvests train vibration into a 1F supercap — up to 568 mW/unit. The node runs on zero grid power.

③ THE PROOF
📡 ✓

Live & validated

Dual-IMU + thermal sensing, on-device anomaly thresholds and FFT — streaming live to the dashboard, field-validated at r=0.87 over 30 days.

How it works

From a passing train to a live alert.

Every link in the chain is self-powered. The energy of the train is what carries its own diagnostics upstream.

01
〰️

Train vibration

Rail flexes under each axle — a continuous mechanical energy source.

→
02
⚡

PZT + LTC3588

Piezo transducer rectified by an LTC3588 nano-power harvester.

→
03
🔋

1F supercap

Buffers energy; the node wakes only when it has charge to spend.

→
04
📐

Dual-IMU + thermal

MPU6050 + ADXL345 + DS18B20 sample vibration, shock & rail temp.

→
05
📡

ESP-NOW mesh

ESP32-C6 node bursts a 29-byte packet to the ESP32-S3 gateway.

→
06
📊

Live dashboard

Gateway forwards telemetry to the cloud — charts, FFT & anomaly alerts.

Why it matters

Self-powered, so it scales where wiring can't.

Validated per-node figures, projected across a national rail network using the Indonesia grid emission factor.

3.6Wh/day
Harvested per active node — enough to run dual-IMU + thermal sensing and transmit, indefinitely.
6,000km
Target network length the approach projects across — no trackside power runs required.
>1,300t CO₂/yr
Avoided versus grid-powered monitoring at network scale (Indonesia grid factor).
In the field

Deployed on a working KRL line.

PiezoRail isn't a bench demo — it runs on the Yogyakarta–Klaten commuter line, with eight dual-IMU units watching eight rail segments around Lempuyangan Station.

  • LineYogyakarta — Klaten KRL
  • Anchor stationLempuyangan
  • Instrumented spanLempuyangan ↔ Maguwo · KM 6.945
  • Sensor units8 × dual-IMU + thermal
  • Per nodeMPU6050 · ADXL345 · DS18B20
  • GatewayESP32-S3 N16R8 · ESP-NOW
Lempuyangan ──────●────── Maguwo
8 segments · KM 6.945 · live ESP-NOW mesh
A
B
C
D
E
F
G
H

Each segment hosts one self-powered node. Status reflects the live gateway feed.

Recognition

Built to be judged on the evidence.

The credibility is in the field data.

📈

Field-validated r = 0.87

30-day correlation between harvested-energy signatures and independently observed track activity.

30-DAY STUDY
⚡

568 mW / unit peak

Measured peak harvest per node from real train passes — the budget the whole system runs within.

MEASURED PEAK

See it running, live.

Open the technical dashboard — real-time telemetry, oscilloscope & FFT, 3D track view, and the Advanced Analysis Tools suite.

Launch the dashboard