The question after the blow
On the morning of 25 October nobody was aboard, the wind was still gusting near 30 knots, and the night before had brought a gust of 36. Every boat owner asks the same thing after a night like that: are the lines still on? Enja has a wind sensor, a barometer, three GPS receivers and a compass whose heading the recorder keeps four times a second. None of them is attached to a mooring line.
The compass is the useful one. Alongside, a boat is held by her lines the way a weathervane is held by its spindle, and what the lines allow shows up as heading. In calm air Enja's heading holds to about a tenth of a degree for a whole day. A crosswind pushes her round by roughly 0.4 to 0.5 degrees for every 100 kn² of side load (wind speed squared, times the sine of the angle it comes from), so a 20-knot wind on the beam turns her almost two degrees. When it eases, she comes back. If a line parts or slips, she should settle at a new angle, or swing further for the same wind.
Checking the night by hand
So we fitted her heading against the side load, minute by minute, for three stretches: the blow of 22 to 23 October, the evening of the 24th up to the 01:40 gust, and everything after it until midday on the 25th. Lines that had let go would show either as a different heading at zero wind or as a steeper response.
| Period | Heading at zero wind | Yaw per 100 kn² | Heel per 100 kn² |
|---|---|---|---|
| 22–23 Oct blow | 95.19° | 0.40° | −0.20° |
| 24 Oct, before 01:40 | 95.06° | 0.38° | −0.20° |
| After 01:40 | 95.15° | 0.31° | −0.22° |
Nothing had moved. After the peak gust she yawed slightly less per unit of wind than before, which is the wrong direction for a missing line. The heel column is there as a check on the method: heel per unit of wind doesn't depend on the lines, and it stayed the same, so the wind numbers were comparable across the three periods.
Doing this by hand once was easy. We wanted the boat to do it after every blow, and that meant knowing what normal looks like over months, not one night.
There is no right angle
The first idea was to store her angle at the berth and alarm when she strayed from it. The season's data killed that quickly.
One dot per day: the median true heading over that day's calm minutes at the home berth. Until she left on 25 August she lay between 93.5 and 94.9 degrees, wandering by a few tenths from day to day. After the trip on 25 and 26 August she came back about half a degree lower and lay between 92.7 and 93.6 for four weeks. Between 19 and 21 September, across the blow of 19 and 20 September, she stepped up by 0.84 degrees, from 93.37 to 94.21, and stayed above 93.8 until she left on 8 October. That step is the one the line check, replayed over the season, would have reported. After 8 October she came back at about 95.3 and settled between 94.9 and 95.2. The last point is 26 October, after the 36-knot night: 94.93 degrees, the median of the check's own first four calm hours, against 94.95 the day before the blow. The bracket on the left shows 1 degree, the size of change the check alerts on.
Enja's own recorder: heading from the trusted compass, apparent wind from the masthead, 1-minute bins; the last point from enja-alarms.
Her calm-air heading at the same berth has ranged from 92.7° to 95.5° since July. Inside a week it wanders by a few tenths, for reasons the data doesn't show. Each trip away resets it, because the lines never go back exactly as they were: she came back from the trip at the end of August half a degree lower, and from the one on 8 October three quarters of a degree higher. A fixed angle would have been wrong for most of the season.
The yaw response wasn't usable on its own either. Fitted over three-hour windows it ranged from almost nothing to over one degree per 100 kn² in ordinary blows. Part of that is the fenders: with the wind from port she rests about a degree lower than with the wind from starboard, as if she lies against them on one side and hangs on her lines on the other.
Two checks
What stays steady is her heading compared with her own recent past, so both checks compare her with herself.
The first runs in calm air, under three knots. Each hour with at least half an hour of calm gives a resting angle, and the check keeps the last six. If a new calm hour is more than one degree from the median of those six, and the next calm hour agrees, it posts to Slack that she is lying at a new angle and a line may have parted, slipped or been retied. The new angle then becomes the reference, so a retie is reported once and not every hour after.
The second runs during a blow, when the calm check can't. It takes her heading, subtracts the push the wind should be giving her, and compares the last half hour with the three hours before. A line parting under load should look like a step that stays. If the step is more than two degrees with at least 15 knots blowing, it says so at once and asks for somebody to go and look.
After any wind alert, the first calm hour also reports, so the crew hears a result and not just silence: unchanged if she is within half a degree of where she was, nudged if she is between half a degree and a whole one. When she goes underway, or is found more than 200 metres from where the reference was learned, the check forgets everything and quietly learns her new berth.
Replaying the season
The thresholds came from replaying the whole season, about 145,000 one-minute bins of heading, apparent wind, position and speed, through the same code that now runs aboard. The swing check never fired: the largest step in any blow since July was 1.3 degrees. The calm check fired once at the home berth.
That was 21 September, after the blow of 19 and 20 September. Before it her calm hours sat at 93.2°; afterwards at 94.2°, and she stayed above 93.8° until she left on 8 October. Something about her lines changed that weekend, slowly or all at once. The logs can't say whether someone was aboard and adjusted them, and that's the first thing to ask before calling it a slipped line.
Looser settings were noisier. A threshold of three quarters of a degree would also have fired on 26 July and 11 September, on changes she drifted back from within days.
What it can't see
A line that was slack and holding nothing can part without changing anything the compass measures, until the wind comes from the side where it mattered. Chafe that hasn't yet cut through looks like a healthy line. A change smaller than her normal drift, a few tenths of a degree, is invisible. Her GPS can't help: at the berth the receiver that keeps talking scatters its fixes over three to five metres, much more than she can move on her lines. The check is a reason to go and look, and it doesn't replace looking.
The first answer
The check went aboard on 26 October, a day after the blow. Its first four calm hours came in between 94.91° and 94.97°. The day before the blow she lay at 94.95°. After thirty-six knots she lies where she lay before them.
