Home / Tools / COMTRADE viewer
Online COMTRADE viewer
Open the fault record of a relay or a disturbance recorder (.cfg + .dat, or a single .cff) and review the waveforms, the digital channels and the fundamental RMS with a cursor. The automatic analysis estimates the fault type, the relay and breaker times and the distance to the fault. The file never leaves your computer.
IEEE C37.111COMTRADE 1991, 1999 and 2013IEC 60255-24common format for transient data exchangeIEEE C37.114single-ended fault location
Results
—
—
— ms
—
—
— ms
Want an engineer to review this case? Our team reviews your data and results and quotes the study or the settings work you need.
Charts
- Selected channel
- Other channels
- Fundamental RMS
- Trigger (t = 0)
Automatic analysis
—
— ms
— ms
— ms
— ms
— km
- Fault
- Current through the pole
- Events
Calculation and details
Values at the cursor
Cursor: —
| Channel | Instantaneous | Fundamental RMS (1 cycle) |
|---|
What the record shows
Pre-fault and fault phasors
—
| Quantity | Pre-fault | Fault |
|---|
Distance, DC component and X/R
What it is for
To look at a fault record without installing the manufacturer's software: confirm which phase faulted, how much the current rose and the voltage dropped, when the protection picked up and tripped, and how long the breaker took to open. It works in the field on a phone and in the office when preparing an event report.
The sample record is synthetic, created by our team: a 13.2 kV feeder with a resistive phase-A-to-ground fault. The phase A current rises to about 600 A and the residual to 420 A; the 51N picks up, trips at about 276 ms and the breaker opens three cycles later, each phase interrupting at its own current zero. You can download the .cfg and the .dat to try them in other programs.
How it is calculated
IEEE C37.111a, b, primary, secondary and P/S factorsIEC 60255-24
value = a · x + bsecondary value = primary value · secondary / primaryx is the number stored in the .dat, and a and b are the channel factors in the .cfg. The channel's P or S flag says whether the result is in primary or secondary values; to convert to the other side, the primary and secondary factors of the same channel are used.
Time of each sample
t = Σ 1 / samp, segment by segment, or t = timestamp · timemultIf the .cfg has sampling rates (nrates ≥ 1 and samp > 0), each sample follows the previous one with the period of the segment the previous one belongs to: with 1000 Hz up to sample 2 and 500 Hz after it, the times are 0, 1, 2 and 4 ms. The standard does not detail that transition and other readers use the new rate; the difference is one sample at the rate change. If the .dat also has timestamps that depart more than half a sample from that base, the viewer warns. If samp = 0, the timestamp in microseconds is used (or in nanoseconds when the date of a 2013 record has nanosecond resolution). Zero on the plot is the trigger time minus the time of the first sample.
The sampling skew of each analog channel (skew field, in µs) is added to its time axis: the plot and the cursor read each channel at its own sampling instant.
Fundamental RMS over the last cycle up to the cursor
x(tk) ≈ a · cos ωτk + b · sin ωτk + cRMS = √(a² + b²) / √2It is a least-squares fit over the samples of exactly one period 1/f0, with τk = tk − tcursor. When the sampling rate is an integer multiple of the fundamental (N = fs/f0 samples per cycle), it gives the same result as the one-cycle discrete Fourier transform used in numerical relays, X = (2/N) · Σ x(k) · e−j2πk/N (Phadke and Thorp, Computer Relaying for Power Systems), and it rejects the DC component and the integer harmonics. When it is not (1000 Hz at 60 Hz is 16.67 samples), a DFT with a rounded N would have up to 1 % error and would oscillate; the fit stays exact for a sinusoid with a DC component.
Automatic analysis: fault inception and fault type
Δi(t) = i(t) − i(t − T)ΔI = Ifault − Ipre-faultThe fault starts at the first sample where the current of any phase (or the residual) departs from that of the previous cycle by more than 5 % of its peak, or six times the channel noise, for at least 1/16 cycle; the voltages refine that instant. The pre-fault phasors are computed over the cycle that ends 1/16 cycle before inception, and the fault phasors over the cycle that ends one and a half cycles after it, with a fit that absorbs the DC component. The superimposed current ΔI removes the load: a phase is faulted if its |ΔI| reaches 50 % of the largest one, and ground is involved if 3|ΔI0| exceeds 10 % (criteria of this viewer, not of a standard). For ground faults the angle of I2 relative to I0 is also checked: close to 0° for A-ground and B-C-ground, −120° for B-ground and C-A-ground, and +120° for C-ground and A-B-ground.
Relay and breaker times
trelay = ttrip − tinceptiontbreaker = tfirst pole open − ttripThe trip is the first rising edge of a digital channel named TRIP, DISPARO, PTRC, 94 or 86 (pickup channels are not used). A pole is open when its current stays below 3 % of its peak for at least three quarters of a cycle; the fault is cleared when all faulted phases are interrupted. If a pole conducts again, the viewer marks a reclose and checks whether the current returns to fault level. If the fault type changes and holds for three quarters of a cycle, it is marked as an evolving fault.
IEEE C37.114single-ended fault location
x = Im(V / I) / X1x = Im(V · ΔI*) / Im(Z1 · I · ΔI*)The first is the simple reactance method and the second the Takagi method (Takagi et al., IEEE Trans. PAS-101, 1982), with Z1 and X1 per kilometer and ΔI the phase (or phase-to-phase) current minus its pre-fault value. For a ground fault V = VA and I = IA + k0 · 3I0, with k0 = (Z0 − Z1) / (3Z1); for a phase-to-phase fault, with or without ground, V = VB − VC and I = IB − IC; for a three-phase fault the three loops are averaged. These are the loops of the IEEE C37.114 guide as presented by Zimmerman and Costello (SEL, 2004).
DC component and X/R
i(t) ≈ Ip · cos(ωt + φ) + D · e−t/τX/R = ω · τIt is fitted over the first fault cycles of each faulted phase, searching τ by least squares, like the DC component of IEC 60909-0. If D is below 10 % of the AC peak, the X/R is not reported.
Missing data
−32768 in BINARY, −2147483648 in BINARY32, the minimum float −3.4028235 · 1038 (or a non-numeric value) in FLOAT32, an empty field or 99999 (outside the channel range) in ASCII, and the timestamp 0xFFFFFFFF. The viewer leaves those points undrawn, and the RMS of any cycle that includes them is left empty.
Limits of the calculation
- The automatic analysis is indicative. Deciding whether a protection operated correctly takes the settings, the relay logic and the system study.
- Single-ended fault location assumes a homogeneous, well-modeled line. It is shifted by fault resistance combined with load flow (more for the simple reactance than for Takagi), zero-sequence mutual coupling with parallel lines, infeed from the remote end or from taps, a non-homogeneous system, and errors in the CTs, the VTs and the line parameters. Use it to direct the line patrol, not as an exact location.
- The classification needs one cycle of pre-fault and one and a half cycles of fault. With CT saturation, a large DC component or faults shorter than one cycle, the phasors and the fault type lose accuracy; the viewer says so in the confidence note.
- Pole opening is detected from the current, not from the 52a contact. In a phase-to-phase fault without ground, the fault current is interrupted by the first faulted pole that opens. A reclose is only visible if the line carries current again.
- The one-cycle RMS needs a full cycle of data: right after fault inception or a pole opening it shows a transition. If the actual frequency differs from the one you enter, the value has an error, and a decaying DC component is not fully rejected.
- The normal state of the digitals and the .hdr and .inf files are not used.
- Very large records, with millions of samples, can be slow to open on a phone.
Frequently asked questions
How do I open a COMTRADE file without installing software?
Open this page, tap «Open files» and select the .cfg and the .dat of the same record together, or a single .cff file. The viewer reads the file in your browser and plots the analog and digital signals; the relay manufacturer's software is not needed.
What is the difference between ASCII and binary COMTRADE?
Both hold the same samples. In ASCII each sample is a line of comma-separated text; in binary each value takes 2 bytes (BINARY) or 4 bytes (BINARY32 and FLOAT32), so the file is several times smaller. The format is declared in the .cfg, and the viewer reads all four.
Is it safe to open my substation records here?
Yes. The file is read with the browser's FileReader API and processed on your computer: the page does not send it to any server or store it. If you close or reload the page, the record is gone.
If the setting is wrong, we correct it remotely
When the record shows a misoperation or a slow trip, we review the relay settings and logic and load the correction remotely, whatever the relay make and model.