Confidential — internal TECHNICAL analysis — separate from the legal file

The Bearing Housing Failure & Retrofit — Technical Timeline

Second analysis: management record, decision timeline, and the stick-slip causation hypothesis · NOT part of the legal position

Months before any module cracking was reported, the Skyline II fleet suffered a distinct, documented failure: plastic bearing housings slipping and "exploding" across the portfolio. Arctech's corrective action — riveting/screwing the bearing housings on ~444 units — was completed in January–February 2026. The first cracked modules were found on 11 May 2026, roughly three months later. This page reconstructs the full chronology from the mailbox and the Dropbox 9. Warranty Issues record, and assesses whether the retrofit itself could be the root cause of the cracking.

9 Oct 2025
Bearing failures formally reported by iEnergia (2 Falcon + 1 Angol)
444
Bearing housings in Arctech's retrofit scope
≈9 Feb 2026
Retrofit reception (Angol + LS1; Dinamo/Falcon done by 3 Feb)
11 May 2026
First cracked modules found — ~3 months after the retrofit

Two damage waves — keep them distinct

WAVE 1 Oct 2025 – Feb 2026 · Direct breakage from exploding bearing housings

Failed housings physically destroyed adjacent modules. iEnergia, 10 Feb 2026: "Las piezas explotadas provocaron la rotura de una cantidad significativa de módulos… Estos no fueron reemplazados por el contratista." Photos in 9. Warranty Issues\Broken Modules (Trina-Arctech) dated 7 Oct 2025.

WAVE 2 From 11 May 2026 · Recurrent edge-cracking pattern

A different signature: "cracked modules at the short-side edges", mainly rear glass, some with electrical failure/smoking — portfolio-wide, on trackers with "wave-like movement" and documented torsional deflection. Photos dated 12 May 2026, in folders named Cracks in modules, Gap between modules, Deflection by torsion.

The distinction matters for any RCA: Wave 1 is mechanical impact damage from a failed component (undisputed, Arctech's contractor left modules unreplaced). Wave 2 is a stress-pattern failure whose cause is the open question — and it appeared only after the fleet-wide retrofit.

Full chronology — decisions, pushes, and management

Arctech's technical substantiation — the Side Rivets letter and the TÜV test

The full engineering basis Arctech provided is one page: the "Engineering Confirmation Letter — Side Rivets Remedy Solution", REV 00, dated 22 October 2025, from Arctech Solar Holding Co., Ltd (China HQ), plus the referenced test report CN25K8OP001 (delivered as a signed .rar). Recovered from Octavio Caro's email of 30 Jan 2026 and reproduced below. Key passages, verbatim:

"For Arctech Skyline II D166 tracker, spherical plastic bearings were designed as upper and lower half snapped together. Due to installation tolerances or errors, Arctech observed that additional unbalanced friction forces on the bearings may cause the upper and lower half of plastic bearings to separate gradually and trackers may be stuck due to separated plastic bearings."Arctech Engineering Confirmation Letter, 22 Oct 2025 — the root cause, in Arctech's words
"Arctech engineering team propose to add side rivets to fix the plastic bearings to the torque tube, thereafter keep bearings synchronized with torque tube's rotation to minimize the pulling force on snaps."Same letter — the remedy: the bearing no longer rotates relative to the tube
"Arctech did accelerated durability test on the sample tracker mounted with 4 side rivets, typical mechanical loads, installation tolerances and heaters were used… with TUV Rheinland as independent third party witnessing… (Report ID: CN25K8OP001)… it is convinced that 4 rivets tracker having operated 28 days smoothly which equals to 30.68 years operation life without damage." … "due to self-healing performance of ZAM material, the upper coating layer at the cutting edge will be dissolved to cover the cross-section… to prevent corrosion." … "side rivets remedy solution can be the robust solution with verified performance. Tracker warranty remains intact."Same letter — the validation

What the letter establishes

  • Arctech admits the defect causes tracker jamming — "trackers may be stuck". Binding of the torque tube is Arctech's own documented failure mode, not our speculation.
  • Arctech attributes separation to "unbalanced friction forces" — precisely the mechanism a sloped layout produces. This is direct support for the slope→friction line of investigation.
  • The remedy changes the rotation architecture: riveting the bearing to the tube means all relative rotation moves to the bearing-outer ↔ housing interface, and tube↔bearing axial sliding is eliminated.
  • Letter dated 22 Oct 2025, generic "Dear Customer", REV 00 — 13 days after our report, remedy already engineered and TÜV-tested. The defect was known at fleet level; shared with us only on 30 Jan 2026, three months later and after the retrofit contract was already tendered.
  • ZAM self-healing coating answers the corrosion question at the rivet cut.

What the validation did NOT cover

  • Modules are never mentioned. The accelerated test validated that the tracker operated "smoothly… without damage" — it did not measure module glass stress, clamp loads, or torsional shock transmitted to modules. The module–tracker interaction post-rivet is unvalidated.
  • One sample tracker, 4 rivets, 28 days — extrapolated ×400 to "30.68 years". No fleet variability (slopes, tolerances across 444 units, rivet workmanship by a local contractor).
  • No analysis of axial thermal expansion after pinning the bearing to the tube — the mechanism that would produce wave-like bowing and module gaps.
  • Field contradiction: at reception (10 Feb 2026) two further housings had failed "ambas sin remache" — and Wave-2 module cracking began 3 months after the fleet retrofit the letter warrants.
  • Warranty extension requested by the Owner (5 yrs): denied. Motor/gearbox stress investigation: dismissed as "no evidence".

Inside the TÜV report CN25K8OP001 — now fully read

ItemFact (verbatim from the report)
Testing period2025-09-08 → 2025-10-06 · sample received 8 Sep · order date 17 Sep · issued 20 Oct 2025 (Peter Lai / Constant Zhao, TÜV Rheinland Guangdong)
Who tested"The testing conducted by Arctech Solar Holding Co., Ltd. under the witness of TUV Rheinland staff" — at Arctech's own Kunshan facility, spec "According to client's requirement", result field "See Other" — a witnessed client test, not a Pass/Fail TÜV certification
BearingSpherical plastic bearing, upper GFZ-D166-SR93.75-01 / lower -02 · R93.75 ext / R83 int / 60 mm · 318.3 g · material PA66 + 30% GF + PTFE
RivetsΦ4.8 × 31.5 mm, 4 rivets total (26.7 g)
SetupOne 10 m span between two columns; columns at ±2° vertical AND horizontal deviation (misalignment simulated); 450 kg distributed + 150 kg cantilever; water + sand added daily into the spindle/bearing/seat gaps; bearing heated to 70 °C 08:30–17:30; slew cycle 3.6 min; 28 days ⇒ "30.68 years"
Result"no deformation or abnormalities… operated normally" — but bearing mass 318.3 g → 311.4 g: 6.9 g lost (~2.2%) over the simulated life

Fair to Arctech

  • Misalignment (±2°), abrasives (sand+water) and heat (70 °C) were all simulated — not a trivial test.
  • PTFE-filled PA66 is a self-lubricating bearing material — a reasonable choice.
  • The rivet assembly itself survived the accelerated life without deformation.

What it still never measured

  • No modules mounted — dead weights only (450+150 kg). Module glass stress, clamp loads and torsional shock: unmeasured.
  • No friction/torque instrumentation reported — no breakaway torque, no drive current, no stick-slip detection. "Operated normally" is observational.
  • One bearing pair, one 10 m span — no full-length tube, so no axial thermal-expansion behaviour of a real 40–60 m riveted tube.
  • The 6.9 g wear finding cuts our way: the bearing measurably abrades at the new sliding interface. In the field, months of wear-in (debris, roughening, changing friction) is a candidate explanation for the ~3-month lag between the Feb retrofit and the May cracking.
  • Chronology: samples were at test on 8 Sep 2025 — a month before iEnergia's first failure report (9 Oct). Arctech was already validating the remedy for a defect it had not yet disclosed to this portfolio.

The Engineering Letter — source

Open full screen

The TÜV report — source

Open full screen

The mechanism — no longer a hypothesis

✅ Independently confirmed by the Owner, 7 August 2026

This page originally set out the constrained-bearing mechanism as an unproven working hypothesis. On 7 August 2026 Infrastrutture's Technical Director described the identical mechanism in writing, reached independently: rivets fix the bearing to the tube → the tube's thermal contraction drives the bearing out of position inside the housing → it "cannot slide freely" → additional friction → loud noises and jerky movement. He also reports that lubrication temporarily eliminates the problem — which is itself diagnostic: a structural or geometric fault would not respond to oil. The mechanism is interfacial friction created by the retrofit.

What remains open is only the final link — from stick-slip shock to module glass fracture — which is the subject of the joint inspection protocol in the claim. The spatial cross-map (crack positions vs the 444 retrofitted bearing positions) remains the decisive test, and is deliberately capable of disproving it.

Stick-slip / constrained-bearing mechanism

The question: the added rivet/screw fixes the plastic bearing to (or through) its housing. If the fastener restricts the motion the bearing was designed to allow, two things follow:

  1. Rotational binding → stick-slip. Arctech's letter confirms the rivets make the bearing rotate with the tube — so all relative rotation now occurs at the bearing-outer ↔ housing interface, a surface that was not the primary design rotation interface. If that interface binds (dry plastic-on-steel, dirt, tolerance stack-up, slope side-loads), the synchronous multi-slew drive keeps winding elastic torsion into the tube until friction releases suddenly. Each release is a torsional shock propagating along the tube — into the module clamps.
  2. Axial constraint → thermal buckling. Before the retrofit, the tube could slide axially through the bearing. Riveting pins the bearing to the tube, so axial thermal expansion must now drag the bearing through the housing seat or bow the tube — presenting exactly as the observed "wave-like movement along the tracker" and the "gap between modules" photographed in the warranty folder.

Note that both halves of this hypothesis now rest on Arctech's own documents: the jamming failure mode ("trackers may be stuck") and the friction driver ("unbalanced friction forces") are stated in Arctech's 22 Oct 2025 letter; the change of rotation architecture ("keep bearings synchronized with torque tube's rotation") is the letter's own description of the remedy. What remains unproven is only the link from that changed architecture to the module glass — the exact thing the TÜV test never measured.

CONSISTENT with the hypothesis

  • Sequence: retrofit completed ≈9 Feb 2026 → first edge-cracks 11 May 2026. No Wave-2 cracking is on record before the retrofit.
  • Failure signature: cracks at the short-side edges — where torsional/bending stress concentrates at the outer clamps — not random impact damage.
  • Field observations post-retrofit: jerky, uneven slew movement; wave-like motion; torsional deflection at tracker 56; module gaps (axial displacement indicator); misalignment vs as-delivered.
  • Wave 2 is portfolio-wide on the retrofitted fleet, same configuration everywhere.
  • Arctech's own precedent cases (Chile/Trina, Colombia/LONGi) were on Skyline II — recurrence on this product line.

AGAINST / UNKNOWN — must be tested

  • ~3-month lag Feb→May needs explaining (candidates: accumulating fatigue cycles; autumn/winter wind loads; larger thermal swings; detection lag — thermography only began May).
  • No cross-map yet of cracked-module positions vs retrofitted-bearing positions. If cracks cluster on non-retrofitted trackers, the hypothesis weakens badly.
  • LS2 retrofit status unclear — the documented scope was Dinamo/Falcon (444 u) + Angol/LS1 reception; LS2 had 3 failed bearings but its retrofit record hasn't been located.
  • Alternative causes uncontrolled: inherent module fragility (dual-glass 2.0 mm on 33 mm frame), purlin/fastener interface (bolt-washer vs rivets — the Owner's own open question), tracker control behaviour, O&M practices.
  • Arctech's 28 Jun position: "no significant structural abnormalities" — an interested party's conclusion, but on record.

How to test it — data requests and field work

#ActionWhat it proves / kills
1Cross-map: overlay (a) Wave-2 cracked-module locations, (b) the 444 retrofitted bearing positions, (c) Wave-1 failed-bearing locations, on the plant layouts (GA DWGs available per plant)If cracks cluster at/near retrofitted bearings → strong support. If uncorrelated → hypothesis substantially dead.
2Physical inspection of a sample of retrofitted housings: does the fastener contact/penetrate the bearing? Can the tube still slide axially? Measure breakaway torque vs an un-retrofitted unitDirect mechanical confirmation of binding
3Request Arctech's retrofit engineering: the method statement, fastener spec, and the engineering approval for the 444-unit campaign — plus TCU/slew-drive motor current logs (their system measures drive load; ours does not)Their own design docs either allowed for motion or didn't; current logs show binding as elevated/spiky draw
4Thermal test: measure module-gap / tube position morning vs midday on retrofitted rowsAxial-constraint confirmation (gaps should breathe; pinned tubes won't)
5Video + accelerometer on a jerky tracker during a full tracking dayCaptures stick-slip events and shock magnitude

Motor current on the layout — feasibility

Short answer: not mappable from our current SCADA — but there are three practical routes.

The Skyline II TCUs communicate via LoRa/RS-485 to Arctech's own controller; motor/slew-drive current is not exposed in the plant monitoring we operate for these sites (no tracker motor-current channel exists in the configured datapoints). Slopes concentrating friction on specific bearings is exactly the kind of pattern a current map would reveal, so:

  1. Ask Arctech for their TCU logs (route 3 above). Their controller measures drive load; this is the reciprocal of the field evidence they demanded from us, and reasonable to request within the ongoing technical review.
  2. Clamp-meter campaign: measure drive current on a stratified sample of trackers — flat rows vs sloped rows vs retrofitted vs non-retrofitted — during identical tracking movements. One day of field work per plant gives the map.
  3. Proxy map now, at zero cost: we already hold the topographic DWGs (Arctech engineering package, Nov 2023) and the failure locations (bearing failures, cracked modules, jerky trackers). Overlaying slope zones vs failure density tests the same friction hypothesis without current data — and can be built from existing files as soon as the failure positions are tabulated per tracker.

How it was reported — the meetings record

Finding: the bearing issue was never minuted in the shared meeting record. A full sweep of the Dropbox shared folders (PowerTree Common Files, PowerTree, O&M, and all five plant folders) found no weekly meeting minutes at all for the Oct 2025 – Jul 2026 period. The weekly-minute practice that existed during construction (Gabriel Sanhueza's "Minuta de reunión" series; the Monday "Project review meeting iEnergia | PowerTree" Teams series) does not appear in the shared record for the O&M period in which the bearing failure and retrofit occurred.

The actual reporting channel was the email thread "Portfolio Powertree - New Bearing Housings" (28 messages, Oct 2025 – Mar 2026), on which the Owner's representatives were included — plus:

Implication for the record: the Owner cannot claim ignorance of the bearing defect or the retrofit — its Technical Director ran a written technical interrogation of Arctech about it in January 2026 — but equally, no shared minute exists linking the two failure waves; that linkage exists only in this analysis and should be socialised deliberately, not assumed known.

Management record — what this timeline shows about iEnergia's handling

This page is a technical/management record. Its hypothesis is deliberately excluded from the legal correspondence until tested — an untested causation theory cuts both ways.