Fault Location

Overview

Six impedance-based fault location algorithms are available. Each has different data requirements and handles fault resistance and system non-homogeneity differently. The table below summarises the methods.

Choosing the Right Algorithm

The tab auto-selects the best available algorithm based on what data you have. However, understanding the selection logic helps you interpret results and decide when to override.

A practical decision tree:

• If the recording has pre-fault data (at least one cycle before inception) and you do not know the source impedances, Takagi is the best choice. It uses the change in current from pre-fault to fault to eliminate load-flow error.
• If no pre-fault data is available and the fault is unbalanced (single-line-to-ground, line-to-line, or double-line-to-ground), Modified Takagi can substitute negative-sequence current for the superimposed quantity. It does not require pre-fault data but is limited to unbalanced fault types.
• If you know the local and remote source impedances, Eriksson or Novosel provide the most accurate results because they model infeed through the fault resistance. Use these when source impedance data is available from the system study or short-circuit model.
• As a fallback, Reactance is always available. It requires no pre-fault data and no source impedance, making it the simplest and most universally applicable method.
• For radial feeders with low fault resistance, Radial provides a quick apparent-impedance estimate. It does not separate fault resistance from line impedance, so it is best suited for bolted or near-bolted faults on simple radial circuits.

Using the Algorithm Strip

Below the hero result banner, six algorithm pills are displayed in a horizontal strip. Each pill shows the algorithm name and its median distance result.

• Click a pill to select that algorithm. The hero banner, chart highlight, and readout update to reflect the selected method.
• Active pill -- the currently selected algorithm is visually highlighted with a coloured border and bold text.
• Warning icon -- a caution indicator appears on a pill when the algorithm produced results but with low confidence (wide uncertainty band). The result is still shown but should be interpreted carefully.
• Greyed-out pill -- the algorithm could not run because its data requirements were not met (e.g., Takagi is greyed out when no pre-fault data is available, Eriksson is greyed out when source impedances are not configured). Hovering over a greyed-out pill shows the reason.

Line Parameters

Line impedance parameters define the electrical characteristics of the protected transmission or distribution line. These values are essential for converting measured loop impedance into a physical distance.

R + jX Per-Unit-Length Input Mode

As an alternative to entering Z1 and Z0 as magnitude and angle, you can switch to the R + jX / km input mode using the toggle in the line parameters section. In this mode, you enter the per-unit-length resistance and reactance values directly:

When R + jX values are entered, the total line impedances Z1 and Z0 are computed automatically as Z = (R + jX) × line length. The Z1/Z0 magnitude and angle fields update to reflect the computed values. This mode is convenient when line parameters are specified as per-kilometre constants from a line data book or relay settings file.

X₁ Reactance / Angle Input Mode

Some numerical relays (e.g. Schneider P437) expose Line Reactance (X₁)as their primary fault location setting rather than total impedance magnitude. To support this, you can switch to the X₁ reactance / angle input mode.

In this mode, you enter the total line reactance X₁ and the line angle. Detego automatically computes the total Z₁ magnitude using the mathematical relationship:
|Z₁| = X₁ / sin(angle)

This avoids manual conversion errors, particularly on cables with shallow line angles (e.g., 15°) where entering X₁ as the Z₁ magnitude would cause massive distance errors.

Zero-Sequence Compensation ($k_0$)

For ground fault loops the zero-sequence compensation factor $k_0$ adjusts the current denominator to account for the difference between positive-sequence and zero-sequence line impedance. Three input modes are provided to match the convention used by your relay manufacturer:

Regardless of the input mode, the computed $k_0$ value is displayed as a read-only field so you can verify the conversion.

Ground Loop Method

The ground loop method controls how zero-sequence current is included in the phase-ground measurement loop denominator. Two options are available:

• Standard -- $I_{ph} + k_0 \cdot I_N$ where $I_N = I_A + I_B + I_C$ is the residual (neutral) current. This is the conventional approach used by the majority of relay manufacturers.
• Phase-only -- $I_{ph} \cdot (1 + k_G)$ where $k_G$ is a phase-referenced ground compensation factor. This form uses only the faulted phase current and avoids using the residual current. Some manufacturers (e.g., Schneider P437) use this convention.

Fault Detection

Fault timing and type are auto-detected from the recording data. You can review and override these values in the settings panel.

Auto-detected values are shown with a green "Auto" badge. Editing any field clears its Auto badge, indicating a manual override. Click the dimmed "Auto" badge that appears next to an overridden field to restore the auto-detected value for that individual field. The detected fault type is also displayed as an amber badge in the bottom recording info bar for quick reference.

Measurement Loop

The measurement loop determines which voltage and current quantities are used to compute loop impedance. It is automatically selected based on the detected fault type using the following mapping:

When you change the fault type, the measurement loop updates automatically. Phase-ground loops include the zero-sequence compensation factor $k_0$ in the current denominator. Phase-phase loops use delta voltages and delta currents, which naturally cancel the zero-sequence component.

For double line-to-ground (DLG) faults, Detego automatically selects the ground loop of the faulted phase carrying the highest fault current. A banner below the loop selector shows the reasoning (e.g., “Phase A has higher fault current (3600 vs 864)”). You can override this by selecting a different loop manually.

Source Impedance

The source impedance section is an optional panel that becomes relevant when you want to run the Eriksson or Novosel algorithms. Source impedances represent the Thevenin equivalent impedance of the power system behind each end of the protected line.

These values are typically available from short-circuit studies or system planning models. If you do not have them, leave the source impedance section collapsed -- the Reactance, Takagi, and Modified Takagi algorithms do not need them. When source impedances are entered, the Eriksson and Novosel pills in the algorithm strip become active.

Channel Mapping

The Fault Location tab requires three voltage channels and three current channels to be assigned:

Channels are auto-detected from the COMTRADE file metadata using phase field matching, name pattern recognition, and unit detection. If auto-detection assigns incorrect channels, you can manually override each assignment using the dropdown selectors in the settings panel.

CT / VT Scaling

CT and VT ratios are synced from the global scaling configuration. When the global CT/VT settings match the Fault Location settings, a "Synced" badge appears next to each ratio field. If you change the global scaling, the Fault Location tab picks up the new values automatically. When the COMTRADE file contains primary-referred quantities (indicated by a "P" flag in the channel header), the tab automatically detects this and bypasses CT/VT ratio scaling — the measured values are already in primary units and no conversion is needed.

Overlaid Algorithm Traces

Each algorithm is drawn in a distinct colour:

The selected algorithm's trace is drawn bold (thicker line) while the other algorithms are drawn thinner and partially transparent. This lets you see all results at a glance while keeping the selected one visually dominant.

All algorithm traces are clipped to the fault window — only samples between inception and clearance are plotted. This eliminates misleading noise from pre-fault and post-fault regions where the impedance calculation produces unreliable values. The Y-axis is clamped to ±2× line length to prevent within-window outliers from compressing the chart scale.

Hero Result Banner

The hero banner at the top of the tab shows the headline result for the selected algorithm:

• Distance -- median fault distance in the selected unit (km or mi) and in per-unit of line length.
• Confidence badge -- a colour-coded label (HIGH, MEDIUM, or LOW) indicating the consistency of the per-sample distance estimates during the fault window.
• R_f -- estimated fault resistance in ohms.
• Loop -- the active measurement loop (e.g., AG, AB).
• σ -- robust spread estimate based on Median Absolute Deviation (MAD), quantifying the consistency of per-sample distance estimates.

Chart Elements

Confidence Levels

Confidence is derived from the ratio of the Median Absolute Deviation (MAD) to the line length. MAD is a robust measure of spread that is resistant to outliers — unlike standard deviation, a few extreme samples will not inflate the uncertainty. When line length is not set, a fallback coefficient of variation (CV) method is used.