Acme Space — Flight Log

Why hyperdrives fail at 0.5 past lightspeed

2412-07-18T00:00:00Z

We have now torn down 214 failed motivators from customer hulls — ours and, more often, other people's. The pattern is consistent enough that I want to write it down plainly, because the folklore around hyperdrive failure is wrong in a way that costs crews money and occasionally costs them more than that.

The failure is not at the transition

Almost nobody's drive fails at the moment of transition to superluminal. Transition is violent, brief and extremely well instrumented; if something is going to break there, it breaks on the test stand. Of our 214 teardowns, eleven failed at transition. The other 203 failed between thirty and ninety minutes into a sustained run.

That timing is the clue. Thirty to ninety minutes is how long it takes a marginally mis-calibrated field lattice to walk itself out of true.

What actually goes wrong

Here is the sequence, every time:

One lattice node sits slightly outside tolerance — often from a hard shutdown months earlier.
At sustained 0.5, that node runs hotter than its neighbours.
Hot node drifts further. Neighbours compensate, so the aggregate field reading stays green.
Compensation exceeds the neighbours' authority. Aggregate reading drops in seconds.
The drive drops you to realspace, hard, wherever you happen to be.

Step three is the dangerous one. Every conventional motivator reports aggregate field integrity, which is exactly the number designed not to move until it is too late to matter.

Read the node data, not the summary

If your drive publishes per-node telemetry, look at the spread between the strongest and weakest node, not the average. A spread above 3% is a scheduled maintenance item. A spread above 6% means you should not begin a long jump.

Node spread	What it means	What to do
Under 2%	Healthy	Nothing
2–3%	Early drift	Note it, watch the trend
3–6%	Real	Schedule a re-trim
Over 6%	Pre-failure	Do not jump long

If your drive does not publish per-node data, that is itself a finding.

The alluvial damper is not optional

The second most common cause in our teardowns — 47 of 214 — was a damper left in service well past its erosion limit. Dampers are consumables. They cost a fraction of a percent of the drive, they are the cheapest insurance in the propulsion bay, and crews skip them because nothing appears to be wrong. Nothing appears to be wrong right up until step four.

A hyperdrive that has never told you bad news is not a healthy hyperdrive. It is a hyperdrive that is not talking to you.

What we did about it

The HD-7 trims its weakest node against its strongest 400 times a second, which converts the failure mode above from a cliff into a slope. It also refuses to spin up on a badly seated damper, because the alternative — spinning up and finding out at minute forty — is worse.

None of that makes the drive immortal. It makes it honest, which in this business is the more useful property.

The case for modular nacelle architecture

2412-06-30T00:00:00Z

The argument against segmented warp coils has always been field continuity. A single unbroken winding produces a cleaner subspace field than twelve segments with eleven joints between them. This is true. It is also, at this point, irrelevant, and I want to explain why.

The continuity penalty is now small

When segmented coils were first proposed, the field discontinuity at each segment boundary cost roughly 2.4% of nominal field efficiency. That was a real penalty and the objection was fair.

Modern boundary geometry has brought that figure down to 0.11%. On a warp 8.4 assembly, the difference between continuous and segmented is a rounding error in your transit time. Meanwhile the difference in maintenance is not a rounding error in anything.

What a continuous coil costs you when it fails

A single degraded section in a continuous winding means:

Unship the nacelle. This is a drydock operation, not a starbase operation.
Ship the coil to a facility that can rewind it. There are eleven in known space.
Wait. Median turnaround on a rewind is 74 days.
Re-certify symmetry against the opposite nacelle, which has aged differently in the meantime.

Total: roughly a hundred days out of service, and a coil pair that no longer matches.

What a segmented coil costs you

Pull the affected segment through the maintenance hatch.
Fit a new segment from stores.
Re-verify. Half a day.

The nacelle never leaves the ship. The pair stays matched, because the replacement segment is wound to the original specification rather than to whatever the rewinding facility's mandrel happened to be that season.

The asymmetry nobody talks about

Here is the part that convinced me, and it is not about downtime. A continuous coil ages as one object. When it degrades, it degrades unevenly — and you cannot address the bad section without disturbing the good sections. Every rewind resets the whole coil to a new baseline that no longer matches its partner.

Segmented coils let you replace exactly the part that is worn. After six years, a WC-9 pair that has had four segments replaced between them is closer to matched than a continuous pair that has had one full rewind.

The one real objection

Segments introduce eleven more joints, and joints are where things leak. This is a legitimate concern and we take it seriously: every WC-9 segment boundary is pressure-tested to four times operating plasma pressure, individually, with the serial number stamped on the test record.

If your operation genuinely never intends to service a coil in the field — some short-haul system operators do not — a continuous coil is still defensible. For everyone flying beyond a starbase's reach, the maintenance argument has won.

Cold start: ion thruster warm-up in deep space

2412-06-11T00:00:00Z

Every operations manual has a thruster warm-up procedure. Almost all of them assume you are within a few hours of a station and can afford to be conservative. Deep space is a different problem, and the standard procedure quietly makes it worse.

Why the standard procedure exists

Conventional warm-up ramps propellant flow gradually over roughly four minutes to avoid thermal shock at the nozzle throat. Sound reasoning: a cold throat hit with full flow can crack, and a cracked throat is a nozzle you no longer have.

The assumption underneath is that you have four minutes. Near a station, you always do.

Where it breaks

Out past the beacons the situation that calls for thrust is usually the situation that does not give you four minutes: a drifting mass on a converging track, a docking approach going wrong, a tumble that started for reasons you have not diagnosed yet. Crews trained on the gradual ramp will either wait out the ramp they do not have time for, or override it and crack a throat.

The deep-space procedure

The IT-12 supports an eleven-second cold start because the throat is pre-conditioned rather than gradually eased into service. The procedure:

Keep two nozzles warm. Idle two of the twelve at minimum flow continuously. Cost: a negligible amount of propellant per day. Benefit: two nozzles always available instantly.
Pre-condition on watch change. Run the full array to 20% for ninety seconds at each watch change. The throats stay above the thermal shock threshold for roughly six hours afterwards.
Full cold start only when pre-conditioning has lapsed. Then, and only then, use the eleven-second ramp — and log it, because the throats accumulate cycles.
Never override the ramp on a genuinely cold array. There is no emergency that is improved by losing a nozzle.

What this costs

Idling two nozzles continuously uses about 0.3% of your propellant budget on a ninety-day transit. Every operator I have talked to who has needed instant thrust in the deep has concluded that this is the cheapest 0.3% they spend.

A note on propellant

Refined tibanna conditions differently from xenon — it holds throat temperature longer but is less forgiving of a hard start. If you are running tibanna, extend the pre-conditioning burn to two minutes and shorten the interval to four hours. The array's own telemetry will tell you when it is outside the window; it is worth putting that readout somewhere the watch officer can see it without asking.

A field guide to deflector shield harmonics

2412-05-22T00:00:00Z

On most ships, exactly one person understands the shield harmonics panel, and when that person transfers off, the ship runs whatever settings were left behind for the next several years. This is a bad way to operate a safety system, so here is the entire subject, compressed.

What a harmonic is

A deflector shield is a standing field. Like anything standing, it has a frequency. Anything arriving at the hull is either absorbed by that field or passes through it, and which one happens depends on how the arriving energy's frequency relates to the field's.

That is the whole concept. Everything else is bookkeeping.

The three numbers that matter

Base frequency. Where your field sits. Higher base frequencies absorb energetic particle impacts better; lower ones handle broad-spectrum radiation better. Most hulls run mid-band and should.

Rotation rate. How often the field shifts frequency. Rotation exists because anything that adapts to your frequency has to keep re-adapting. The DSG-4400 rotates every 340 ms.

Channel separation. How far apart your ray and particle channels sit. Too close and a single event degrades both. Too far and you have gaps in coverage. The correct answer is a function of your emitter count, and the generator calculates it — but only if you have told it how many emitters you actually have fitted.

The three mistakes

Setting rotation to zero to "stabilise" the field. The field is not unstable. Rotation looks like noise on the readout and is not. Crews turn it off to make the display calm and turn their shield into a fixed target.
Running channels at the same frequency. This produces a beautiful number on the aggregate readout because both channels reinforce. It also means one event takes both.
Never re-running the emitter survey after hull work. Add a sensor pod, weld a patch, fit a comm relay — the field geometry changed. The generator does not know unless you tell it.

A workable standing procedure

Re-run the emitter survey after any hull modification, without exception.
Leave rotation on automatic. Always.
Check channel separation quarterly; it drifts as emitters age at different rates.
Log your base frequency and review it when your operating region changes. A ship that moves from inner-system freight to deep survey should not be running inner-system settings.

Reading the degradation report

When a shield degrades gracefully it will tell you which stage it is in. Stage 1 is loss of margin — everything works, the reserve is gone. Stage 2 is loss of a channel. Stage 3 is reduced sphere coverage, and the report will name the sector. A ship at Stage 3 with the gap facing the thing it is worried about is in a much worse position than the aggregate percentage suggests. Read the sector, not the number.

Tractor beam safety: ten rules for cargo retrieval

2412-05-04T00:00:00Z

I spent twenty-two years pulling things out of places they should not have been, and I have watched most of these mistakes happen in person. None of them were made by careless people. They were made by competent crews under time pressure who had not thought about the specific failure in advance. That is what a rules list is for.

The ten

1. Survey before you grab. Two minutes of sensor sweep tells you where the structural members are. A capture that loads a spar is a recovery. A capture that loads a hull plate is a hole.

2. Match the gradient to the damage, not the mass. A 900-tonne intact pod takes hard gradient happily. A 200-tonne hull with a compromised frame needs feathered. Mass is not the variable that matters here.

3. Cancel the tumble before you translate. Every time. Pulling a spinning mass toward you converts angular momentum into a problem that arrives at your hull.

4. Never capture through your own structure. If the beam path clips a pylon, you are loading that pylon. The TBP-8 needs a clear 40° cone and will warn you, once, before it scatters.

5. Assume the target has stored energy. Pressurised compartments, charged capacitors, an intact power cell that has been baking for nine years. Approach the side you would rather be on if it lets go.

6. Watch your own attitude. A capture applies a reaction to your hull. On a small ship against a large mass, you are the one that moves. Have thrusters ready.

7. Two operators, always. One flies the approach. One watches the beam load and the target's structural telemetry. On a good day the second person is bored. Keep them.

8. Release is a manoeuvre, not a button. Dropping the beam instantly leaves the mass with whatever momentum you gave it. Ramp down or you have just launched something.

9. Log the capture profile. Which gradient, what load, how long. When an insurer asks — and they ask — the log is the difference between a claim and an argument.

10. If the target is coming apart, stop. A hull actively shedding structure is not a recovery job. It is a debris containment job, and those are done with nets, not beams.

The rule behind the rules

Every entry above is a specific case of one general principle: a tractor beam applies force to a structure you did not design and cannot fully inspect. Treat the target as fragile until it has proven otherwise, and treat your own hull as the thing you are actually protecting.

On automatic tumble cancellation

Modern projectors will cancel a tumble without operator input, and this is genuinely good — it cuts a typical derelict capture from twenty minutes to four. It has also produced a generation of operators who have never cancelled a tumble manually and do not know what the beam load looks like while it happens. Practice it dry. The automatic system will fail eventually, and it will fail while you are busy.

Transporter buffer integrity: reading the pattern logs

2412-04-16T00:00:00Z

Transporter faults are famously dramatic and, in practice, almost never sudden. The system degrades for weeks, records the degradation faithfully, and waits for somebody to look. This is a guide to looking.

What the log actually contains

A full-fidelity pattern log stores, for each transport: confinement beam drift, buffer occupancy time, biofilter rejection events, and the per-stage error correction load. Summarised logs — which is what most units keep — store a cycle count and a fault code. The fault code tells you what finally broke. The full log tells you what was breaking.

The TR-7 keeps four hundred transports at full fidelity for exactly this reason.

The four numbers, in order of usefulness

Error correction load. How hard the system worked to reconstruct a clean pattern. This is your leading indicator. A unit whose average correction load has climbed 15% over fifty cycles is telling you something is wrong months before anything trips.

Confinement beam drift. Should sit near 0.0003%. Drift rises with emitter age and rises sharply with mechanical disturbance — if it jumped after a hull repair, the emitter array moved.

Buffer occupancy. How long each pattern sat in the buffer. Occupancy creeping up means the reconstruction stage is slowing, which usually means correction load is rising too.

Biofilter rejections. Interesting for what they catch, but also for the rate. A filter suddenly rejecting more is either seeing more or has drifted into false positives. Both matter.

A monthly routine that takes fifteen minutes

Pull the correction load average for the last fifty cycles and compare to the previous fifty.
Anything above a 10% increase, look at drift.
Rising drift plus rising correction load means the emitter array wants a survey.
Rising correction load with flat drift means the reconstruction stage, not the emitters.
Write the two numbers in the maintenance log even when nothing is wrong. The trend is the whole value; a single reading tells you almost nothing.

Why buffer hold time is the specification that matters

Everyone compares range. Range is easy to build and rarely the constraint. The number that determines what happens on your worst day is how long the buffer will hold a full pattern on its own power while somebody deals with a fire.

Ask any vendor for their measured floor across destructive testing, not their typical figure. A vendor who only quotes a typical figure has a floor they would rather not discuss. Ours is ninety-four minutes, and the tests that produced it were not kind.

Astromech socket wiring standards, revised

2412-03-29T00:00:00Z

The Droid Interface Standard moved to Rev. 4 this quarter. Most of the changes are clarifications, but three are substantive and one will bite anyone running older cable.

Change 1: the permission layer is now mandatory

Rev. 3 described a permission layer for droid access to the systems bus and made it optional. In practice this meant most installations gave the droid unrestricted read/write access to everything, because that was the default and it worked.

It worked until an astromech with a corrupted task queue re-routed power away from life support on a survey vessel, correctly, according to instructions it had been given nine days earlier by a technician who had since left the ship. Nobody was hurt. It was close.

Rev. 4 makes the permission layer mandatory and specifies four access tiers:

Tier	Access	Typical use
0	Read-only, non-critical	Passenger hulls
1	Read all, write non-critical	Freight, standard
2	Read all, write with crew confirm	Survey, long range
3	Full autonomous write	Unmanned only

Tier 3 on a crewed hull is now a certification failure. The AS-3 ships defaulted to Tier 1.

Change 2: droid-initiated ejection is standardised

Rev. 3 allowed droid-initiated ejection but did not specify the signalling. Three manufacturers implemented it three ways. Rev. 4 specifies a single two-wire interlock with a defined failure state — loss of signal means do not eject, which is the correct default and, notably, not what one of the three implementations did.

Change 3: the pinout

Pins 7 and 9 on the primary connector have swapped function. Pin 7 is now the permission-layer handshake; pin 9 carries the ejection interlock.

This is the change that will bite you. Rev. 3 cable physically mates with a Rev. 4 socket. It will present a droid that appears to enumerate correctly, sits at the wrong access tier, and has a non-functioning ejection interlock — a socket that looks fine on every panel readout and is not.

What to do if you are on Rev. 3 cable

Check the harness stamp behind the retention collar. If it does not read DIS-4 or later, replace the harness before your next droid rotation. It is a two-hour job and about ninety credits of cable. The alternative is a droid bay whose safety interlock is decorative.

We are supplying Rev. 4 harnesses at cost through the end of the year. Contact field service with your socket serial and we will ship one.

Navicomputer calibration after a gravitic anomaly

2412-03-08T00:00:00Z

Ships pass through gravitic anomalies more often than the incident statistics suggest, because most passages produce no immediate symptom at all. The navicomputer keeps working. That is the problem.

What an anomaly does to stored data

A steep gravitic gradient induces state changes in dense storage. It does not usually damage the core, the interfaces or the display. It flips a small number of bits in a very large number of stored lane solutions.

A lane solution with a few flipped bits is still a syntactically valid solution. It will load, it will display, and it will plot a course that is subtly wrong in a way that a plot review will not catch — because the plot review is reading the same corrupted data.

The verification procedure

After any transit through a gradient above 5 g, before your next jump:

Run a full almanac verification, not a spot check. On the NC-1701 this takes about forty minutes and re-verifies every stored solution against its checksum.
Read the repair report. Self-healing storage will silently fix what it can. You want to know how much it fixed. A handful of corrected solutions is routine. Several thousand means the exposure was worse than you thought and you should verify twice.
Cross-check one known lane by hand. Pick a route you have flown, plot it, compare to your own records. This catches the case where the corruption reached the checksums themselves.
Check the secondary core independently. Hot standby cores are exposed to the same gradient. Two cores agreeing is not confirmation if both were corrupted the same way.
Log the exposure. Peak gradient, duration, corrections applied. The next crew will want it.

Why spot checks are worse than nothing

A spot check verifies a sample and returns clean, which produces confidence rather than information. Corruption from a gradient is not uniformly distributed — it concentrates in whichever physical storage regions were oriented unfavourably. A sample drawn from a clean region says nothing about the affected one, and the crew now believes the almanac is verified.

On not having local storage

Some operators respond to all of this by preferring a relay-dependent navicomputer that fetches solutions on demand and therefore cannot hold corrupt ones. This trades a manageable problem for an unmanageable one: the relay is unavailable exactly where anomalies are common, which is exactly where you most need to re-plot.

Carry the almanac. Verify it after events. The procedure above has caught real corruption on customer hulls eleven times that we know of, and in at least two of those the crew had already plotted a jump.

Life support recyclers: scrubber maintenance intervals

2412-02-19T00:00:00Z

Every atmosphere recycler has a scrubber cartridge interval stamped on the plate. That number assumes nominal crew, nominal activity and nominal atmosphere composition. If your ship matches all three, you may stop reading.

Crew count is not linear

The intuition is that twenty crew on a forty-crew system means double the interval. It does not, for two reasons.

First, the cartridge has a shelf life independent of load. A cartridge at half load does not last twice as long; it lasts about 1.6 times as long before the sorbent degrades on its own.

Second, and less obvious: low crew counts often mean higher per-crew activity. A skeleton crew on a long transit is doing more EVA, more maintenance and more physical work per person than a full complement. The LSR-6 measures actual CO₂ load rather than counting heads, which sidesteps this entirely — but if your recycler counts heads, apply the correction manually.

Crew (of 40 nominal)	Naive interval	Real interval
40	1.0×	1.0×
30	1.33×	1.2×
20	2.0×	1.6×
10	4.0×	1.9×

Below about a quarter load the sorbent shelf life dominates completely and the interval stops extending at all.

Surge crew is worse than it looks

A recycler rated for 40 nominal and 56 surge will handle 56 crew. What the surge rating does not tell you is that cartridge consumption at surge is not 1.4× nominal — it is closer to 1.7×, because the system runs the scrubbing stage harder and hotter, and heat degrades sorbent.

Plan surge operations with cartridge stock at 1.7×, not 1.4×. Running out of scrubber cartridges with 56 people aboard is a scenario with no good improvisation.

Hot-swap changes the maintenance culture

When cartridge replacement requires taking the loop down, crews batch it — they wait until several bays need service and do them together during a scheduled window. Batching means some bays run well past their interval.

With hot-swappable bays there is no reason to batch. Replace each bay when that bay is due. This sounds like a small operational detail and it is the single largest factor in real-world recycler reliability that we have measured.

The number to actually watch

Not cartridge age. Not cycles. Watch days of remaining endurance at current crew count, and watch it as a trend. A recycler whose endurance figure is falling faster than the calendar is telling you a bay is degrading early — usually contamination, occasionally a seal.

Percentages hide this. Days do not. If your system only reports percentage, compute the days yourself and write them somewhere the watch officer sees them.

Cloaking fields and the treaty compliance checklist

2412-01-27T00:00:00Z

We sell the CFM-2 to survey operators, salvage crews and research vessels. We sell it with a compliance dossier, and I want to explain why that dossier is not a formality.

The short version

Article 14 governs refraction field systems across the signatory jurisdictions. It does not prohibit them. It prohibits specific configurations and specific uses, and it requires notification for others. The distinction between a legal survey cloak and a serious offence is usually a matter of configuration and paperwork, not of hardware.

The checklist

Before purchase

Confirm your operating jurisdiction is among the fourteen where civil operation is permitted.
Confirm your vessel class is eligible. Several jurisdictions permit research and survey hulls and prohibit commercial freight, regardless of intent.
Hold a current licence before you take delivery. We cannot ship against a pending application.

At installation

Verify the weapons interlock is intact and functional. This is a hard-wired board-level interlock; if an installer offers to bypass it, end the engagement and tell us.
Complete the hull survey. An unsurveyed installation produces a visible field seam, which is both a performance problem and, in two jurisdictions, evidence of an unregistered modification.
File the installation notification within thirty standard days.

In operation

Log every engagement: time, duration, position, purpose. Most jurisdictions require the log on request; all of them treat a missing log badly.
Do not engage inside a controlled traffic zone without prior notification. This is the single most common violation and it is almost always inadvertent.
Drop the field on hail. Always. Failure to answer while cloaked is what escalates an administrative matter into an enforcement one.

Annually

Re-verify the interlock and file the certification.
Confirm your licence still covers your current operating region. Regions change; licences do not follow.

Why we refuse some orders

We decline roughly one order in six. The usual reasons are an operating region outside the fourteen, a vessel class that is not eligible, or a request to modify the interlock. That last category we report, because Article 14 requires us to and because the alternative is being the company whose hardware turns up somewhere it should not.

This costs us revenue. It also means our customers can operate a cloak openly, with paperwork that survives an inspection — which is the entire product, in a way the specification sheet does not capture.

If you are unsure

Ask us before you order. Compliance questions cost nothing and we would rather answer forty of them than process one return from a jurisdiction where the unit should never have been shipped.

The dilithium contraction: what 2387 taught us

2411-12-14T00:00:00Z

Anyone who was building starship components in 2387 remembers where they were when the supply notices went out. What is less remembered is how long the contraction actually lasted — nine years — and how thoroughly it changed the way careful manufacturers buy materials.

What happened

A single region supplied roughly 60% of refined dilithium to the commercial market. When that supply became unavailable, the remaining 40% did not scale. Prices went up by a factor of eleven within four months, and the shortage propagated into everything that used dilithium in any quantity — which, at the time, was most warp-capable systems.

The interesting part is what happened to manufacturers, because it sorted them very cleanly.

Three responses

Manufacturers who had single-sourced stopped shipping. Some never resumed. Their contracts were with one supplier, at a good price, and that price was good precisely because it was exclusive.

Manufacturers who had multi-sourced within the same region discovered that three suppliers drawing from one field is one supplier with extra paperwork. This was the largest group and the most surprised.

Manufacturers who had qualified alternate materials kept shipping at reduced volume. Not comfortably — the alternates were more expensive and performed worse — but continuously.

The lesson, which is not about dilithium

The failure was not "we bought dilithium from the wrong place." It was that qualification programmes are expensive and slow, so nobody runs them for materials that are currently available. Qualifying an alternate material takes two to four years. You must start before you need it, which means spending real money against a risk that may not materialise.

Everything we source now falls into one of three categories:

Dual-qualified. Two materials, both flight-certified, from geologically unrelated sources. This covers every critical path material in our propulsion line.
Single-qualified with an active programme. One material in production, an alternate in qualification. We hold four materials here at any time.
Single-sourced, accepted risk. Documented, reviewed quarterly, with a named person who owns the exposure. Currently three materials, all in non-critical applications.

What it costs

Dual qualification adds roughly 4% to our materials cost and a meaningful amount of engineering time. We have been asked, more than once, why we do not simply carry more inventory instead.

Inventory covers a six-month disruption. It does not cover a nine-year one. The WC-9 uses a verterium cortenide formulation that is dual-qualified across two unrelated sources specifically because its predecessor was not, in 2387, and we could not ship coils for eleven months.

That was thirty-five years before my time here. The policy it produced is still the most valuable thing in our procurement manual.

Inside Drydock 12: refitting a corvette in forty days

2411-11-02T00:00:00Z

In the third quarter we took a 140-metre corvette through a full propulsion and defence refit in forty days. Customers ask what a refit window actually involves, so here is the real schedule, including the part where we lost two days.

The scope

Replace a failing class 1 motivator with an HD-7
Fit a DSG-4400 and eight emitters
Replace four thruster clusters with IT-12 arrays
Full navicomputer replacement and almanac load
Recertify the hull

Days 1–6: survey

Nothing gets cut, welded or unshipped until the survey is complete. Six days sounds extravagant. It is the reason the rest of the schedule holds.

The survey found what surveys usually find: the previous motivator cradle had been shimmed non-standard during an earlier repair, which nobody had recorded. Caught on day four. Had we caught it on day twenty-two, with the new drive already in the bay, it would have cost a week.

Days 7–19: the drive

Unshipping a motivator is the long pole in any propulsion refit. The old unit came out on day nine. The cradle adaptation — the shim problem from the survey — took days ten through thirteen. The HD-7 went in on day fourteen and was aligned by day sixteen.

Days seventeen through nineteen were cold alignment verification, which cannot be rushed and cannot be done in parallel with work that induces hull vibration. This is why the shield work did not start until day twenty.

Days 20–31: shields and thrusters

Eight emitters, four thruster clusters, running in parallel with two crews. Emitter placement followed the hull survey rather than the previous emitter positions — the ship had gained a sensor pod at some point and the old geometry left a gap in the ventral sector that had been there, undetected, for at least four years.

Days 32–33: the two days we lost

A thruster hardpoint on the port quarter had corrosion under the mounting ring that the survey missed because it was not visible without removing the ring. We removed the ring on day thirty-two, found it, and stopped.

Two days to cut out the corroded section, fit new plate and re-certify. We ate the cost, which is our policy when a survey misses something. It also went into the survey procedure: mounting rings on any hull over fifteen years old now come off during survey, not during installation.

Days 34–40: certification

Almanac load and verification, shield emitter survey and harmonic tuning, full-power drive test, hull recertification, and the sea trial equivalent — a short lane run with our engineers aboard.

The ship left on day forty. The schedule had forty-two in it, which is the other reason the survey takes six days: the margin has to come from somewhere, and it is cheaper to find problems early than to build slack into every later phase.

If you are planning a refit window, talk to us before you book the dock. Half of what makes forty days possible is deciding the right scope in the first place — see Services for how we structure it.