Frequently asked questions

Meter files are processed locally in your browser. The analyzer does not upload the meter file or the analysis results to a server, and normalized CSV exports are generated locally.

Loading this website itself still involves ordinary web requests for the page and its assets, as with any site. See how the analyzer handles your data.

The analyzer reads CSV and XLSX meter exports. Timestamps can be a single combined column or separate date and time columns, with 5, 10, 15, 30 or 60-minute readings.

Yes. A kW figure is treated as the average demand over its interval, while kWh is energy per interval. When the file holds kWh, the analyzer converts it using the interval duration to give the equivalent average demand. See demand and energy.

A start stamp marks the beginning of the interval the reading covers, while an end stamp marks its end and is normalized backward by the interval duration. Select the convention your file actually uses — the wrong choice shifts every reading by one interval. See interpreting intervals.

For local timestamps without an explicit offset, choose the timezone of the meter or site the data comes from. The timezone determines how local times — including daylight saving transitions — are resolved. See timezones and daylight saving time.

Nonexistent spring-forward times are excluded and reported, not shifted to a nearby time. Deterministic complete fall-back double passes can be resolved automatically; other ambiguous repeated hours require you to choose how they are resolved. Ambiguous local times are never guessed. See timezones and daylight saving time.

A sustained cadence change blocks whole-file analysis and the detected segments are reported. Each segment can then be analyzed individually using its own cadence. See cadence changes.

Missing intervals are reported as missing and are never interpolated or filled in. The analysis uses the observed intervals only. See data quality.

Identical duplicates are removed and reported. Conflicting duplicates — the same timestamp with different values — block analysis, because the analyzer will not guess which reading is correct. See data quality.

A small number of unreadable timestamps or values can be excluded and reported while the remaining usable rows are analyzed. Larger problems block analysis, and incomplete date or time mappings may also block it. See data quality.

Load factor is average demand divided by maximum interval demand. Higher values mean demand runs more evenly relative to the maximum; neither a high nor a low value is universally better. See summary metrics.

P95 is the demand level that 95% of analyzed intervals are at or below, and P99 the level 99% are at or below. They are less sensitive to one isolated maximum than the absolute peak. See percentiles and the load duration curve.

It is the same interval demands sorted from highest to lowest, showing how often demand stays at or above different levels. It does not preserve chronology. See percentiles and the load duration curve.

You choose a demand limit and the analyzer reports the intervals and events above it, including the largest overshoot and the energy above the limit. Nothing is shifted or reduced. See peak-limit analysis.

It adds a constant load in kW over selected weekdays and a local-time window; overnight windows are supported and energy is calculated from the real elapsed overlap. Only observed meter intervals are changed — missing intervals are never filled — and the coverage figure shows how much of the requested schedule actually has meter observations. See additional-load scenarios.