Name: Covenant Systems
Address: US
Price range: $$

Question 1

Does my event file get uploaded anywhere?

Accepted Answer

No. Parsing, phasor extraction, fault classification and every export all run in your browser using local JavaScript. The record itself never leaves your machine, and nothing derived from it is transmitted — no file name, no station or relay identifier, no channel names, no measured values. That matters because fault records are frequently treated as confidential in utility environments. For completeness: this page carries the same anonymous usage analytics as the rest of the site, which records that a record was opened and which format was chosen, along with which tabs and exports get used. It cannot see anything about what is inside your file.

Question 2

Which file formats can I load?

Accepted Answer

Two families. SEL compressed event reports (.CEV) are read natively, including the relay's own settings block — so the analysis uses the line constants of the line it is actually protecting. IEEE C37.111 COMTRADE records are also supported: the .cfg configuration file paired with its .dat data file, in ASCII, BINARY, BINARY32 or FLOAT32 form. If you have a .EVE report you would like supported, send one over.

Question 3

How is the fault type determined?

Accepted Answer

From the samples themselves, not from metadata. The tool takes one-cycle Fourier phasors of each channel before and after the trigger, flags the phases whose current rises meaningfully above load, and checks the residual (3I0) to decide whether ground is involved. That combination gives the classification — for example two faulted phases with residual current is a double line-to-ground fault.

Question 4

How is the distance to fault calculated?

Accepted Answer

The tool selects the correct measuring loop for the fault type, the same way a distance element does: the ground loop V/(I + k0·3I0) for ground faults, or the phase loop (Va−Vb)/(Ia−Ib) for phase faults. The apparent impedance is projected onto the line angle and divided by the positive-sequence ohms per mile. For a .CEV record those constants come from the relay's own Z1MAG, Z1ANG, Z0MAG, Z0ANG and LL settings; the reference cases use the modelled 24-mile, 69 kV line shown in the settings panel.

Question 5

What do the symmetrical components tell me?

Accepted Answer

The Sequence tab resolves the three phase quantities into their zero-, positive- and negative-sequence parts in both the pre-event and event windows. Two ratios do most of the work: negative over positive says the system is unbalanced, and zero over positive says current is returning through ground. A single line-to-ground fault puts the three networks in series, so all three components come out close to equal; a line-to-line fault produces negative sequence with almost no zero sequence; a three-phase fault produces neither. When the positive-sequence quantity has collapsed — a de-energization, an open breaker — the ratios are withheld rather than reported, because dividing one noise floor by another gives a number that looks meaningful and is not.

Question 6

Why does the tool show harmonics?

Accepted Answer

Because the second harmonic is how a transformer differential element tells inrush from an internal fault. Energizing a transformer draws a magnetising current that looks like a fault to the differential, but it is heavily distorted, and a conventional element restrains above roughly 15 per cent second harmonic. The Harmonics tab takes the spectrum of any channel over either measurement window, draws the bars to scale against the fundamental, marks that restraint threshold and reports total harmonic distortion alongside it. The energization reference case reads about 60 per cent second harmonic; every fault case reads under four.

Question 7

What does the impedance plane show me?

Accepted Answer

The apparent impedance a distance element would measure, traced through the whole record rather than reduced to a single number. The line is drawn from the relay at the origin out along the line angle and marked every quarter of its length — that is where a bolted fault sits. Before a fault the measurement sits far out in the load region at a shallow angle; when the fault arrives it swings in toward the line angle and parks near the fault location. Watching it move is how you separate a real fault from a power swing or a CT saturation artefact. All six measuring loops are selectable, and the one implied by the classification is highlighted.

Question 8

Can I get the analysis out of the browser?

Accepted Answer

Two CSV exports, both written on your machine. The analysis export is a sectioned summary — record metadata, classification, measured quantities, sequence components, the harmonic spectrum of every channel, the element evaluation and every relay bit that asserted or changed state, with times in cycles from the trigger. The samples export is the raw waveform, one row per sample with every analog channel and every plotted digital, for when you want to do your own arithmetic. Neither involves a server.

Question 9

Does it check the relay's settings against what actually happened?

Accepted Answer

For the overcurrent elements, yes — this is the part a plain viewer cannot do. A .CEV record embeds the relay's own settings, so the tool reads the pickups, curve types and time dials, converts the measured waveform to secondary quantities through CTR and PTR, and works out whether each element should have picked up and how long its curve would have taken to operate. It then compares that against the relay word and reports whether the two agree. Reading that 51G1T asserted tells you the relay operated; computing that these settings on this current should have operated, and finding the relay agreed, tells you the settings are doing what you think they are.

Question 10

Why are the voltage elements not evaluated?

Accepted Answer

Because they did not survive the check. Across four real SEL-351S records from the same feeder, no single comparison basis reproduced the relay's own 27 and 59 bits: in steady state the 59P1 bit sits clear against a 73.40 V pickup, which only line-to-neutral explains, while the 27P2 bit sits clear against a 58.95 V pickup, which only line-to-line explains. Both cannot be right, so some supervision in the relay logic is not visible in the settings the record carries. An evaluation that quietly disagrees with the relay is worse than no evaluation, so those elements are left out until a record settles it.

Question 11

Can I see relay elements that are not plotted?

Accepted Answer

Yes — all of them. A real SEL event report names every bit in the relay word, commonly five hundred or more, and only a handful can be drawn before the chart becomes unreadable. The Elements tab keeps the rest reachable: search by name, filter to the ones that changed state or ever asserted, see when each first picked up relative to the trigger, and click any of them to add it to the waveform view.

Question 12

Are the reference waveforms real captures?

Accepted Answer

They are simulated, not field records — but they are not drawn by hand either. Each case is solved through the symmetrical component networks of the modelled line, so fault current, voltage depression, residual current and DC offset are all physically consistent. That is what makes them useful for training and for sanity-checking your own analysis.

Question 13

Can you build something like this for our team?

Accepted Answer

Yes — this tool came out of a client project and is the kind of purpose-built engineering software we do. If you have a workflow that off-the-shelf software handles badly, submit a service request and we will scope it.

SEL Event Analyzer

Single Line-to-Ground (A-G)

Everything above is software we wrote, running in your browser

Need it to do more?

Why we built it

Fault type distribution

Questions