ECSS Brazing — Re-brazing and Repair (space-systems/ecss/q7040-rebrazing-and-repair)
Use when the task is the repair clause of ECSS-Q-ST-70-40: deciding
whether a brazement that failed inspection may be put back through a
braze cycle, how many such cycles it has left, and what the repaired
joint owes before it can be offered again.
Domain quick reference
- A rejected brazement is not automatically a repair. The defect has to
be one a second heating can actually cure.
- Short fillets, entrapped flux residue, fillet porosity, incomplete
wetting and filler run-out are wetting and flow problems. Re-melting
the filler re-forms the fillet, so these are the repairable set.
- Base-metal erosion, filler depletion into the parent, a crack that has
run out of the fillet into the parent metal and thermal distortion are
losses of parent material or geometry. A second heating deepens them.
They are remake findings, not repair findings.
- The repair budget is counted twice. A joint category allows a number
of re-braze cycles, and a base-metal and filler pair allows a
cumulative dwell above the filler liquidus. A joint can sit inside the
cycle count and outside the dwell budget, or the reverse, so both are
evaluated and neither substitutes for the other.
- The dwell budget exists because every excursion above liquidus
dissolves more parent metal into the filler and diffuses more filler
into the parent. That is progressive and it is not undone by cooling.
- Aluminium pairs have the tightest budget: the filler and the parent
melt within a narrow interval, so a long dwell takes the parent with
it.
- A repair is a process in its own right and needs its own qualified
procedure. Working a repair to the original braze procedure is an
unqualified process, whatever the joint looks like afterwards.
- The repaired joint owes the original inspection again. A repair that
consumes the last permitted cycle owes a destructive metallographic
check as well, because nothing else evidences what the accumulated
heat did to the parent.
Workflow
- Take the defect and decide whether it is reflow-curable or a loss of
parent material. A loss of parent material ends the assessment at a
remake, whatever the cycle count says.
- Count the re-braze cycles already taken from the thermal history and
subtract them from the limit the joint category allows.
- Sum the dwell each recorded cycle spent above the filler liquidus,
add the dwell the planned repair will spend, and compare the total
with the budget the material pair tolerates. Absorb representation
error at the boundary with a named tolerance rather than by rounding
the budget.
- Confirm a qualified repair procedure covers the joint. An absent
qualification is a nonconformance route, not a remake.
- Set the disposition: a remake when the defect is a parent-material
loss or the dwell budget is spent; a nonconformance board when the
cycle count is exhausted or the procedure is unqualified; a repair
otherwise.
- Report every finding, not only the deciding one, so the board sees
the whole picture rather than the first rule that fired.
- Hand a permitted repair back with the re-inspection set for its
category and its cycle number.
Pitfalls
- Reading the cycle count as the whole limit. A joint on its first
repair can already be out of dwell budget if the original braze ran
long, and the cycle count will happily say there is room.
- Repairing an eroded joint because the fillet still looks short. The
short fillet is a symptom; the filler went into the parent and a
further heating takes more of it.
- Working the repair to the original braze procedure. The thermal
condition of a repair is different from that of a first braze, which
is exactly why the repair procedure is qualified separately.
- Widening the dwell budget to let an exact-equality case through. An
equality at the limit is a representation question, handled by the
tolerance inside the comparison; the budget stays as qualified.
- Offering a repaired critical joint on a visual inspection alone. The
defects a repair cycle introduces are internal, and the inspection
that found the original defect is the minimum the repair owes.
- Treating a missing repair-procedure qualification as a scrap decision.
It is a paperwork route with a real part at the end of it; scrapping
the part closes the finding by destroying the evidence.
Behavior contract (gate 3)
The defect curability split, the cycle limit by joint category, the
cumulative dwell accumulation with its liquidus check, the material-pair
dwell budget with its boundary tolerance, the re-inspection set and the
repair, remake and nonconformance dispositions are exercised by the gate
3 contract test:
scripts/test_q7040_rebrazing_and_repair.py against
scripts/q7040_rebrazing_and_repair_logic.py (stdlib unittest, offline).
Run:
python3 scripts/test_q7040_rebrazing_and_repair.py
Compliance
- ECSS standards are freely downloadable (ESA); cite the source and
paraphrase per standards-map.yaml.
- compliance: STANDARDS-REF, gated: false.
1---2name: q7040-rebrazing-and-repair3description: "Determine whether a rejected brazement may be re-brazed or has to be remade. Use when a post-braze inspection has turned a joint down and someone must choose between a repair cycle and a new part: separate the flow defects a second heating can cure from the losses of parent metal it only deepens, count the repair cycles the joint category still allows, add the planned dwell above the filler liquidus into the cumulative erosion budget the material pair tolerates, refuse a repair that has no qualified procedure behind it, and hand a permitted repair back with the re-inspection set it owes. Trigger: ecss, q-st-70-40-brazing, braze-repair-cycle-limit, braze-rebraze-dwell-budget, braze-defect-reflow-curability, braze-repair-procedure-qualification, brazement-post-repair-reinspection."4license: Apache-2.05compliance: STANDARDS-REF6standards:7 - id: ecss8 reference-only: true9gated: false10domain: space-systems11pack: space-systems12compatibility: "agentskills.io SKILL.md; any SKILL.md host (Claude Code, Hermes, OpenClaw)"13metadata:14 source: https://skillmd.com/skills/ashfordeou/q7040-rebrazing-and-repair15 domain: space-systems16 subdomain: ecss17 tags: [ecss, q-st-70-40-brazing, q7040-rebrazing-and-repair, braze-repair-cycle-limit, braze-rebraze-dwell-budget, braze-defect-reflow-curability, brazement-post-repair-reinspection]18 version: 0.1.019 author: Aero Agent Skills20---2122# ECSS Brazing — Re-brazing and Repair (space-systems/ecss/q7040-rebrazing-and-repair)2324Use when the task is the repair clause of ECSS-Q-ST-70-40: deciding25whether a brazement that failed inspection may be put back through a26braze cycle, how many such cycles it has left, and what the repaired27joint owes before it can be offered again.2829## Domain quick reference3031- A rejected brazement is not automatically a repair. The defect has to32 be one a second heating can actually cure.33- Short fillets, entrapped flux residue, fillet porosity, incomplete34 wetting and filler run-out are wetting and flow problems. Re-melting35 the filler re-forms the fillet, so these are the repairable set.36- Base-metal erosion, filler depletion into the parent, a crack that has37 run out of the fillet into the parent metal and thermal distortion are38 losses of parent material or geometry. A second heating deepens them.39 They are remake findings, not repair findings.40- The repair budget is counted twice. A joint category allows a number41 of re-braze cycles, and a base-metal and filler pair allows a42 cumulative dwell above the filler liquidus. A joint can sit inside the43 cycle count and outside the dwell budget, or the reverse, so both are44 evaluated and neither substitutes for the other.45- The dwell budget exists because every excursion above liquidus46 dissolves more parent metal into the filler and diffuses more filler47 into the parent. That is progressive and it is not undone by cooling.48- Aluminium pairs have the tightest budget: the filler and the parent49 melt within a narrow interval, so a long dwell takes the parent with50 it.51- A repair is a process in its own right and needs its own qualified52 procedure. Working a repair to the original braze procedure is an53 unqualified process, whatever the joint looks like afterwards.54- The repaired joint owes the original inspection again. A repair that55 consumes the last permitted cycle owes a destructive metallographic56 check as well, because nothing else evidences what the accumulated57 heat did to the parent.5859## Workflow60611. Take the defect and decide whether it is reflow-curable or a loss of62 parent material. A loss of parent material ends the assessment at a63 remake, whatever the cycle count says.642. Count the re-braze cycles already taken from the thermal history and65 subtract them from the limit the joint category allows.663. Sum the dwell each recorded cycle spent above the filler liquidus,67 add the dwell the planned repair will spend, and compare the total68 with the budget the material pair tolerates. Absorb representation69 error at the boundary with a named tolerance rather than by rounding70 the budget.714. Confirm a qualified repair procedure covers the joint. An absent72 qualification is a nonconformance route, not a remake.735. Set the disposition: a remake when the defect is a parent-material74 loss or the dwell budget is spent; a nonconformance board when the75 cycle count is exhausted or the procedure is unqualified; a repair76 otherwise.776. Report every finding, not only the deciding one, so the board sees78 the whole picture rather than the first rule that fired.797. Hand a permitted repair back with the re-inspection set for its80 category and its cycle number.8182## Pitfalls8384- Reading the cycle count as the whole limit. A joint on its first85 repair can already be out of dwell budget if the original braze ran86 long, and the cycle count will happily say there is room.87- Repairing an eroded joint because the fillet still looks short. The88 short fillet is a symptom; the filler went into the parent and a89 further heating takes more of it.90- Working the repair to the original braze procedure. The thermal91 condition of a repair is different from that of a first braze, which92 is exactly why the repair procedure is qualified separately.93- Widening the dwell budget to let an exact-equality case through. An94 equality at the limit is a representation question, handled by the95 tolerance inside the comparison; the budget stays as qualified.96- Offering a repaired critical joint on a visual inspection alone. The97 defects a repair cycle introduces are internal, and the inspection98 that found the original defect is the minimum the repair owes.99- Treating a missing repair-procedure qualification as a scrap decision.100 It is a paperwork route with a real part at the end of it; scrapping101 the part closes the finding by destroying the evidence.102103## Behavior contract (gate 3)104105The defect curability split, the cycle limit by joint category, the106cumulative dwell accumulation with its liquidus check, the material-pair107dwell budget with its boundary tolerance, the re-inspection set and the108repair, remake and nonconformance dispositions are exercised by the gate1093 contract test:110scripts/test_q7040_rebrazing_and_repair.py against111scripts/q7040_rebrazing_and_repair_logic.py (stdlib unittest, offline).112Run:113python3 scripts/test_q7040_rebrazing_and_repair.py114115## Compliance116117- ECSS standards are freely downloadable (ESA); cite the source and118 paraphrase per standards-map.yaml.119- compliance: STANDARDS-REF, gated: false.