Reed marks and temple marks are the maintenance-driven weaving defects — a worn reed dent, a bent reed wire, a temple with sticking rollers leaves a marking pattern that repeats across every metre of fabric off that loom until the part is changed. Grey inspection catches them at the table but rarely attributes them to the specific reed or temple; fabric downgrades, loom keeps producing marked fabric, and by then the reed has ruined a full shift's production. Vision AI at the loom detects the pattern within the first affected metres, attributes it to the specific reed section or temple assembly, and raises a CMMS work order.
iFactory / Reed & temple mark vision
Attribute Reed and Temple Marks to the Specific Part — Not "Loom Under Investigation"
Loom-vision reed and temple mark detection with attribution to the specific reed section or temple assembly and CMMS work order raised with the part identifier — so maintenance replaces the actual worn part rather than swapping components on suspicion.
Loom 34
Reed A-7
28 marks / 4 hr
Loom 12
Temple L
14 marks / 8 hr
Loom 19
Reed B-3
9 marks / 12 hr
Loom 34
WO-4471
Reed replace scheduled
Reed + temple
mark detection
CMMS work order
with part ref
The Problem in Weaving Maintenance Diagnosis
A typical weaving shed's maintenance response to grey-inspection-flagged reed and temple marks is slow triage. Inspector flags a loom for chronic reed marks; shift foreman puts a note on the maintenance board; day-shift engineer investigates when there's a gap; diagnosis walks the reed looking for the specific section, sometimes finds it, often replaces the whole reed on suspicion. Meanwhile the loom has been running for hours producing marked fabric. Temple marks are worse — the assembly has needles or rollers that wear differentially, and diagnosis requires running the loom while watching the temple engagement. Vision AI shortcuts both.
Where Reed & Temple Maintenance Actually Slips
Reed and temple mark failure modes are consistent across weaving operations. Each is a specific part-level attribution the loom vision resolves.
Diagnosis delay
Grey inspector flags loom for reed marks. Maintenance investigation queued behind other work. Loom runs 8-16 hours producing marked fabric before intervention.
Whole-reed swap
Diagnosis can't localise the specific reed section causing the mark. Maintenance replaces the whole reed on suspicion. Waste of parts life on the sections that were fine.
Temple assembly guess
Temple marks traced to a temple assembly but the specific needle or roller within it not identified. Whole temple swapped. Root cause not attributed for spare life planning.
Chronic loom accumulates
Loom flagged for reed or temple issue every few weeks. Root cause never fully closed. Chronic status not systematically identified until quarterly OEE review.
What Good Looks Like in Reed/Temple Vision
A working reed/temple vision system holds four disciplines together — pattern detection at loom, part-level attribution, CMMS work order with part ID, and chronic-loom identification.
Pattern Detection
Vision at loom detects reed mark and temple mark patterns within the first affected metres. Warp-way stripe pattern indicates reed; edge-marking pattern indicates temple; both distinct signatures.
First metres, not first shift
Part Attribution
Reed mark localised to specific reed section (by warp position). Temple mark attributed to specific temple assembly (left, right, position number). Part identifier ready for CMMS.
Specific part, not general area
CMMS Work Order
Work order raised in CMMS with loom, part identifier, mark pattern history, and recommended action. Maintenance targets the specific part on the scheduled slot, not on firefight.
Targeted, not firefight
Chronic Attribution
Recurring reed or temple issues per loom trend for chronic-loom identification. Reed replacement cadence tracked; temple assemblies with early wear surfaced for supplier or specification review.
Chronic surfaces early
How iFactory AI Fits
iFactory AI overlays your looms, CMMS (IBM Maximo, SAP PM, eMaint, Fiix), and parts inventory system — providing the vision-based part attribution that turns maintenance from firefight into planned intervention.
Mark Vision
Vision Layer
Loom vision trained on reed mark and temple mark patterns. Warp-way stripe and edge-marking signatures distinguished. Position localised to reed section or temple assembly.
Part Registry
Vision + Parts
Parts registry with reed sections, temple assemblies, and their install date, life history, and replacement schedule. Vision attribution links to specific part identifier.
CMMS Writer
Vision + CMMS
Work order raised in CMMS with loom, part identifier, mark pattern history, recommended action, and priority. Maintenance planning integrates with scheduled slots.
Chronic Tracker
Vision + Analytics
Recurring reed/temple issues per loom trended for chronic-loom identification. Reed and temple life patterns feed spare life planning and supplier evaluation.
Ask your weaving maintenance engineer what percentage of reed and temple replacements last month were made on suspicion (whole-part swap) vs targeted (specific section identified). If the answer is uncomfortable, vision-based part attribution is where parts life and downtime both recover. Book a reed/temple vision review.
10-Week Reed/Temple Vision Pilot on One Shed
One shed section (30-60 looms), ten weeks. The pilot mounts loom vision, activates reed and temple mark attribution, connects to CMMS, and closes the first month of part-attributed work orders.
Weeks 1–2
Vision + Registry
Loom vision mounted per loom in pilot section. Parts registry loaded with reed sections and temple assemblies per loom. Baseline current-state mark-related maintenance cycle.
Weeks 3–5
Attribution Live
Mark detection and part attribution live. Accuracy tested against maintenance ground truth. Model tuning against loom types and specific reed/temple designs.
Weeks 6–8
CMMS Integration
CMMS work orders raised with part identifier. Maintenance response cycle measured against baseline. Whole-part vs targeted-swap ratio tracked.
Weeks 9–10
Chronic Loom List
Chronic reed/temple issue looms identified from vision trends. Reed and temple life data feeds spare planning. Rollout to remaining shed sections scoped.
Who Owns the KPI
Reed/temple maintenance crosses weaving maintenance, production, quality, and parts inventory. Each function owns a specific KPI or diagnosis stays slow and parts life stays under-utilised.
Weaving Maintenance Engineer
Targeted vs whole-part swap %
Owns the maintenance discipline — the share of reed/temple replacements made against specific vision attribution vs whole-part-swap on suspicion. Higher = better parts life utilisation.
Weaving Production Head
Reed/temple mark downgrade metres
Owns the production outcome — grey-inspection metres downgraded for reed or temple marks. Rising signals attribution or response cycle gaps.
Weaving QA
Chronic-loom identification cycle
Owns the pattern outcome — time from first mark occurrence to chronic-loom flagging. Vision trends surface chronic patterns weeks earlier than quarterly review.
Parts Inventory
Reed and temple spare life utilization
Owns the inventory outcome — average life extracted from reed and temple spares. Whole-part-swap-on-suspicion destroys life; targeted swap preserves it.
FAQ
How does vision distinguish reed marks from other warp-way defects?
Reed marks present a specific signature — periodic warp-way stripes at the reed's dent spacing, typically 1-4 stripes per centimetre depending on reed density. Vision AI trained on this signature distinguishes reed marks from other warp-way patterns: warp streaks from sizing (aperiodic, single-warp), tight ends (single-warp), colour warp mistakes (single-warp, chromatic). Where the pattern is periodic at reed spacing, the loom's reed section is attributed; the specific section is localised by warp position on the fabric width. Temple marks are edge-band signatures at the fabric selvedge, distinct from body defects. What the AI provides is high-confidence attribution to the specific mechanical part, not just "warp-way defect present."
What about looms running specialty reeds (variable dent, insertion reed) — does the model handle those?
Yes, with per-loom reed profile configuration. Standard reeds have uniform dent spacing across width; variable-dent reeds (used for specific pattern effects) and insertion reeds (used for jacquard heavy patterns) have non-uniform spacing that changes what "periodic" looks like at each position. The vision model reads the reed profile from the loom's setup data and adjusts its pattern-matching accordingly. Where the same shed runs multiple reed types on different looms (common in tier-1 mills producing both plain and specialty fabrics), each loom's model applies its own reed profile. What doesn't change is the underlying discipline — pattern-based attribution to the specific reed section.
Book a demo to see specialty reed handling.
How does this integrate with our CMMS (Maximo, SAP PM, eMaint)?
CMMS integration writes vision-attributed work orders directly to your validated CMMS through its documented API. IBM Maximo, SAP Plant Maintenance, eMaint, Fiix, and UpKeep all support work order creation via REST API or middleware bridge. The work order carries loom, part identifier, mark pattern description, mark history, and recommended action; CMMS routes to maintenance per its own workflow rules. Where the CMMS is a validated system with strict change control (regulated environments), the vision workflow surfaces recommended work orders to maintenance planner for approval before CMMS write — preserving the validation posture. Where CMMS is standard operational tool, vision writes directly to reduce cycle time.
Stop swapping whole reeds on suspicion.
Attribute Reed & Temple Marks to Specific Parts — Live
Bring one shed section's last quarter of reed and temple replacement records, the current diagnosis-to-swap cycle time, and the parts inventory life trend. We'll walk what vision-based part attribution would have changed, and demonstrate the CMMS work order with part identifier.
