Test Data Input — Statistical Scatter
Probabilistic Fracture Life defaults its scatter sliders to illustrative coefficient-of-variation values (30% flaw size, 10% stress, 25% material) — explicitly not measured scatter for any real material. This tool replaces that with your own measured scatter: upload an Excel file of multiple test samples, and it computes both the mean and the coefficient of variation of each property, then sends both directly into that tool's Monte Carlo simulation. Nothing leaves your browser — the file is parsed locally, not uploaded to any server.
Excel file (.xlsx), one row per sample, with any of these columns: Sample ID, E (Msi), Su (ksi), Fcy (ksi), Paris C, Paris m, KIC (ksi√in). Column names are matched flexibly. At least 2 samples per property are required to compute scatter — more (5+) gives a meaningfully more trustworthy estimate.
▾ Methodology, equations, and caveats
What this computes. For each recognized property, this tool computes the sample mean, standard deviation, and coefficient of variation (CV = standard deviation ÷ mean, as a percentage) — the exact same CV definition used by Probabilistic Fracture Life's scatter sliders. Both the mean and CV are sent together, so that tool's Monte Carlo simulation samples around your measured center point with your measured spread, not illustrative defaults.
Why scatter needs more samples than a mean does. A mean can be estimated (poorly) from just 2\u20133 samples. A standard deviation — and therefore a CV — is far more sensitive to sample size; with very few samples, the computed scatter can be wildly different from the true population scatter just by chance. This tool requires at least 2 samples to compute CV at all, and flags anything under 5 as low confidence, but treat any CV from a small sample count as a rough starting point, not a validated measurement.
There is no limit on the number of samples you can upload. Aerospace test datasets in excess of 2,000 samples have been successfully run through the underlying statistics and Monte Carlo algorithms behind these tools without issue.
Only real reliability tool. This feeds Probabilistic Fracture Life specifically, because that's the only tool on this site that uses scatter as an input. If you just want a single representative material value for Parameter Review, use Test Data Input — Material Properties instead.
All parsing and statistics happen in your browser — the uploaded file is never sent to any server. Data sent to Probabilistic Fracture Life is stored only in this browser's local storage, the same mechanism used to carry COR between tools elsewhere on this site.