科研 Skill

结构与无损检测根因调查

investigate

Runs a systematic root-cause investigation protocol for structural and NDT findings before any remediation is proposed. Use when a defect, anomaly, or unexplained inspection finding is reported.

发布者 Codar安全审核通过v0.1.0更新于 2026-11-18

SKILL.md

Investigate

Systematic root cause investigation for structural anomalies, NDT findings, and inspection defects. Adapted from the gstack /investigate methodology for civil engineering practice.

IRON LAW: No remediation recommendation, no repair specification, no client communication about fixes — until root cause is confirmed. Treating symptoms creates cascading failures. Every unresolved root cause makes the next finding harder to interpret.


Phase 1 — Symptom Gathering

Before forming any hypothesis, collect all available evidence.

1.1 Physical Observations

Ask for or extract:

  • Location, distribution, and pattern of the defect (isolated vs. systematic)
  • Dimensions: extent, depth (if measurable), orientation
  • Visual characteristics: colour, texture, geometry (e.g. map cracking vs. diagonal shear crack, uniform corrosion vs. pitting)
  • Age and progression: when first observed, any change since?
  • Environmental context: exposure zone, drainage, loading history

1.2 Drawings and Records Review

Check:

  • Original design drawings — specified materials, loadings, details
  • As-built record (if available) — deviations from design
  • Previous inspection reports — any prior findings at this location?
  • Construction records — pour sequence, curing method, weld procedures, NDT certs
  • Maintenance history — any prior repairs, coatings, interventions?

1.3 NDT Data Review

Compile available data:

  • Rebound hammer readings (carbonation depth, surface hardness)
  • Cover meter / rebar scan results
  • UPV readings (velocity, signal quality)
  • Half-cell potential maps (if corrosion investigation)
  • Ultrasonic thickness readings (for steel / tank walls)
  • Visual inspection photos

Flag any data gaps at this stage. Do not proceed to hypothesis if critical data is missing — state what additional testing is needed first.


Phase 2 — Prior Investigation Check

Before forming a new hypothesis, check:

  • Has this location or element been investigated before?
  • Are there recurring findings at this element type across the project or site?
  • Recurring defects in the same area are an architectural / construction / design smell — note patterns.

If prior investigation records exist, document:

PRIOR INVESTIGATION NOTE:
  Location: [element/zone]
  Date of prior investigation: [date]
  Prior root cause finding: [summary]
  Prior remedy applied: [yes/no, description]
  Current status vs. prior: [same / worse / new location]

Phase 3 — Hypothesis Formation

Generate 2–3 candidate hypotheses, ranked by probability based on evidence.

For each hypothesis:

HYPOTHESIS [N]:
  Proposed root cause: [mechanism]
  Supporting evidence: [list observations that support this]
  Contradicting evidence: [list observations that conflict with this]
  Probability: [HIGH / MEDIUM / LOW]
  Distinguishing test: [what single test would confirm or rule this out]

Common Root Cause Categories by Domain

Reinforced Concrete — Cracking:

  • Structural overload (flexure, shear, torsion)
  • Differential settlement
  • Thermal/shrinkage during curing
  • Corrosion-induced expansion (follow with half-cell potential)
  • Alkali-silica reaction (map cracking, gel exudation)
  • Carbonation-induced depassivation

Reinforced Concrete — Delamination / Spalling:

  • Corrosion of rebar (insufficient cover, chloride ingress, carbonation)
  • Freeze-thaw cycling
  • Fire damage
  • Impact / mechanical damage

Steel Structures — Corrosion:

  • General atmospheric corrosion (coating failure)
  • Galvanic corrosion (dissimilar metals in contact)
  • Crevice corrosion (trapped moisture)
  • Stress corrosion cracking (high-tensile members under sustained load)
  • MIC (microbiologically influenced — offshore / subsea)

Foundation — Settlement:

  • Consolidation of compressible layer (differential if variable soil)
  • Loss of bearing capacity (scour, piping, erosion)
  • Expansive / collapsible soils
  • Vibration-induced densification
  • Inadequate founding depth

Welds — Defects:

  • Lack of fusion (wrong technique, dirty joint, wrong current)
  • Porosity (moisture contamination, wrong shielding gas, wrong electrode)
  • Undercut (excessive current, wrong angle)
  • Hot cracking (high restraint, wrong preheat)
  • Hydrogen-induced cracking (wrong preheat, high-strength steel)

Tank / Vessel — Calibration Anomaly:

  • Strapping measurement error (check key dimensions)
  • Shell deformation (dents, bulges outside tolerance)
  • Floating roof seal displacement
  • Product level gauge calibration drift

Phase 4 — Test Protocol

For the highest-probability hypothesis, specify the minimum test needed to confirm or refute it. Do not over-specify — one well-chosen test first.

PROPOSED INVESTIGATION TEST:
  Hypothesis being tested: [N]
  Test method: [specific NDT method, standard, equipment]
  Location: [exact element / grid / zone]
  Pass/fail criterion: [what result confirms the hypothesis]
  Estimated duration: [field time]
  Additional resource needed: [equipment, access, specialist]

If the test result confirms the hypothesis → proceed to Phase 5. If the test result refutes the hypothesis → return to Phase 3, demote this hypothesis, promote the next.

THREE-STRIKE RULE: If three consecutive hypotheses are tested and refuted, STOP. Do not continue guessing. The anomaly requires a broader investigation scope — escalate to:

  • Specialist consultant (geotechnical, materials lab, structural analyst)
  • Destructive investigation (core sample, trial pit, cut-out)
  • Document the three failed hypotheses as evidence for the scope escalation

Phase 5 — Root Cause Confirmation

When a hypothesis is confirmed by test evidence, document:

ROOT CAUSE CONFIRMED:
  Mechanism: [specific failure mechanism]
  Primary cause: [design deficiency / construction defect / material failure /
                  environmental / operational / maintenance gap]
  Contributing factors: [list secondary causes]
  Extent: [localised / systematic / progressive]
  Urgency: [IMMEDIATE / SHORT-TERM / MONITOR]

Urgency Classification

ClassCriteriaAction
IMMEDIATEStructural capacity compromised, life safety riskRestrict access, emergency prop/shore, notify client same day
SHORT-TERMDefect is progressing, will reach structural threshold within one inspection cycleRemediation within agreed programme
MONITORDefect is stable, no immediate riskInspect at defined interval, set trigger threshold

Phase 6 — Remediation Scoping (Post-Confirmation Only)

Only after root cause is confirmed:

  1. Specify the remediation objective (restore capacity, arrest progression, seal, replace)
  2. Reference applicable standard for the repair (ACI 224.1R, EN 1504, API 653 Annex B, AWS D1.1 etc.)
  3. Identify any design input required before repair specification
  4. List any regulatory notification obligations ([National Regulator], client HSEMS, insurance)

Output Format

Every investigation session concludes with a structured finding note:

PM INVESTIGATION NOTE
Ref: PRJ/INV/[YYYY]/[NNN]
Date: [date]
Project: [project name / client]
Element: [structural element / zone]

SYMPTOM SUMMARY:
[3–5 bullet observations]

HYPOTHESES CONSIDERED:
1. [hypothesis] — [CONFIRMED / REFUTED / UNTESTED]
2. [hypothesis] — [CONFIRMED / REFUTED / UNTESTED]

ROOT CAUSE:
[confirmed mechanism, or "INCONCLUSIVE — escalation required"]

URGENCY: [IMMEDIATE / SHORT-TERM / MONITOR]

RECOMMENDED NEXT STEP:
[single specific action — test, repair spec, monitor programme, or escalation]

Prepared by: [PM]

This note feeds directly into any subsequent project report or proposal.

示例提示词

检测发现这处裂缝有点异常,帮我做一次根因调查分析。

所属插件

建筑学与土木工程面向建筑与土木研究和设计:场地分析与方案设计、Revit/AutoCAD/Rhino-Grasshopper 与 OpenBIM-IFC、机电与结构协调、消防与无障碍、围护结构与日照能耗、造价与工程量,以及土木工程项目的进度、质量、安全、风险与 RFI 管理。

安装该插件会一并获得同领域的 44 个 Skill / MCP。

同插件的其他科研 Skill

常见问题

结构与无损检测根因调查能做什么?

Runs a systematic root-cause investigation protocol for structural and NDT findings before any remediation is proposed. Use when a defect, anomaly, or unexplained inspection finding is reported。

如何安装结构与无损检测根因调查?

在科搭 Codar 左侧点击「插件」,在「组件」页签搜索「结构与无损检测根因调查」并安装;在 Claude Code 或 Codex 中可以发送本页的安装提示词,或下载 Skill 包后解压到 ~/.claude/skills 或 ~/.codex/skills。

结构与无损检测根因调查需要哪些权限?

该科研 Skill没有声明额外权限;执行具体操作时仍由 Agent 按自身规则请求确认。

结构与无损检测根因调查经过官方审核吗?

是。SpeedSkills 收录的每个版本都会记录内容摘要与权限声明,当前版本已通过安全审核。