FROM MEASUREMENTS TO LIFETIME

Understand how performance fades.

Explore aging data from devices and materials. Find measured thresholds and compare simple lifetime models.

Local processing · Your data stays in your browser

01 Your data

Workflow

Open CSV

Two columns: time / cycle, performance

Optional header. Comma, tab or semicolon separated. Use a decimal point. Maximum 20,000 rows / 2 MB.

X-axis unit

Threshold (%)

Normalize toFit starts at (x-axis value)

02 Analysis

Example dataset

Measured T80

—First downward crossing

Measured T90

—First downward crossing

Linear decay slope

—percentage points / h

Performance retention

MeasuredLinear fitExponential fitDashed: threshold

LINEAR MODEL

—

EXPONENTIAL MODEL

—

Exploratory fits, not validated lifetime predictions. A good fit does not establish a degradation mechanism. Curves are drawn only over the fitted measurement interval.

Know what the numbers mean.

Measured lifetime

Performance is normalized to the first measurement or the maximum observed value. The first downward threshold crossing is linearly interpolated between neighboring points. “Not reached” means it was not observed.

Two transparent fits

Linear: y = mx + b, ordinary least squares. Exponential: y = A exp(−kx), least squares on ln(y). Fit start selects points; it does not reset time or the normalization reference. R² and RMSE are reported on the original normalized scale.

Scope & privacy

All calculations run locally, without data uploads, analytics or cookies. Values must be nonnegative with unique, nonnegative time/cycle values. This MVP does not implement burn-in detection, TS80, uncertainty intervals or field-specific certification standards.