TorqueScope · Investment Proposal

From signal to certainty

We are looking for £20,000 grant funding to complete our proof of concept validation and move from TRL 5 to TRL 6.

The technology

What TorqueScope does

TorqueScope detects wind turbine drivetrain faults by analysing existing SCADA data — no new sensors, no hardware, no historical failure data required. The platform applies the Ab Astris Lomb-Scargle periodic signal detection framework, originally developed for variable star discovery with NASA TESS photometry, to temperature and operational sensor streams from turbine SCADA systems.

The core insight: physics-constrained systems — rotating machinery, tidal constituents, stellar orbits — produce periodic signals with coefficient of variation (CV) below 2%. Faults disrupt that stability. TorqueScope detects the disruption before it becomes a failure.

SCADA data stream
Lomb-Scargle periodogram
Multi-window CV validation
Normal behaviour model (NBM)
Hybrid anomaly score + criticality counter
⚠ Early warning alarm
Technology readiness

TRL roadmap

1 basic principles 2 concept formulated 3 experimental proof 4 lab validation 5 relevant environment ← WE ARE HERE 6 relevant env. (full) Levenmouth gate 7 operational prototype 8 qualified system 9 market deployment

What we have (TRL 4–5)

  • CARE benchmark: 36 turbines, 3 farms, 89 years of SCADA
  • Hill of Towie: 20/21 turbines, zero code changes, Siemens SCADA
  • Cross-domain Ab Astris validation (6 domains)
  • CARE score 0.588 (detection without failure history)

What Levenmouth unlocks (TRL 5–6)

  • First categorised fault-alarm ground truth for POD measurement
  • Offshore-class 7MW turbine (Samsung), 574 SCADA channels
  • Earliness measurement by fault category vs alarm log
  • Probability of detection (POD) per fault type

What comes next (TRL 6–9)

  • Paid pilot with operator (Banks Renewables / Fred Olsen)
  • Multi-farm SaaS deployment
  • OEM-independent market positioning
  • Global operator licensing
The opportunity

Levenmouth Turbine POD Proposal

574
SCADA channels
7MW
offshore-class turbine
Categorised
fault alarm log

The Levenmouth Demonstrator Turbine is the only publicly accessible wind turbine dataset with a categorised fault alarm log. Every other public dataset — including the CARE benchmark, which carries the strongest ground truth labels of any open dataset — records turbine stop events, not fault categories. Levenmouth lets TorqueScope measure probability of detection per fault type: gearbox, bearing, yaw system, pitch system. That is the number operators and insurers ask for.

Dataset Turbines Fault log POD by category
CARE benchmark 36 Stop events only
Kelmarsh / Penmanshiel 25 Stop events only
Hill of Towie 21 Stop events only
Levenmouth 1 (7MW) Categorised faults
The gating question: Can TorqueScope detect the onset of real categorised drivetrain faults — bearing damage, gearbox anomaly, yaw system failure — with sufficient lead time to justify intervention? Levenmouth answers this. If POD ≥ 70% with mean lead time ≥ 24 hours across fault categories, TorqueScope has a defensible commercial claim. If not, the detection architecture needs iteration before pilot deployment.

View the full validation protocol →

Competitive environment

Competitive landscape

early warning zone specialist / high cost reactive expensive, low ROI Setup & data friction → Detection earliness → SCADA thresholds standard OEM config ML anomaly platforms requires failure history OEM-locked CMS Siemens, Vestas, GE Vibration CMS SKF, Azima — £5–15k hardware TorqueScope

vs OEM CMS

  • No OEM contract required
  • Works with existing SCADA
  • Manufacturer-agnostic

vs Vibration CMS

  • No hardware installation
  • Works day one
  • Zero capex

vs ML platforms

  • No historical failure data
  • Zero-shot detection
  • No training period
Investment

How the £20,000 is deployed

£5,000
£5,000
£3,000
£3,000
£2,000
£2,000
Data acquisition
Core development
Technical report
Route to market
Pilot preparation
Contingency
Milestone Deliverable Gate Timeline
M1 Data ingest Levenmouth 574-channel SCADA ingested, format validated Zero-code-change format test passes Month 1
M2 POD validation POD per fault category measured against alarm log POD ≥ 70%, mean lead time ≥ 24h Month 2–3
M3 Technical report Validation report for ORE Catapult, suitable for funder submission Report accepted by ORE Catapult innovation lead Month 3
M4 Pilot prospectus Commercial pilot pack for operator outreach First meeting with Banks Renewables or Fred Olsen Renewables Month 4
Commercialisation

Route to market

Near term

Direct to operator

Target: Independent operators (Banks Renewables, Fred Olsen, Statkraft UK)
Model: Annual SaaS licence per farm
Why: No OEM dependency, direct O&M budget, short procurement cycle
£15,000–35,000 / farm / year
12–24 months

O&M contractor channel

Target: Independent O&M providers (Visualwind, RES O&M)
Model: White-label or embedded tool licence
Why: Multiplier effect — one O&M contract covers multiple farms
£50,000–120,000 / year
24–36 months

Insurer / lender data product

Target: Wind energy underwriters, project finance lenders
Model: Independent condition report / CMS certification
Why: Insurance and debt covenants increasingly require independent CMS evidence; OEM CMS is not independent
£3,000–8,000 / turbine / year

Sequencing logic: Banks Renewables as the reference customer (accessible, innovation-receptive, no OEM lock). Greencoat UK Wind as the institutional prize (1.8GW+, independent O&M, institutional buyer). ORE Catapult as the parallel credibility accelerator — LDT validation strengthens all three routes simultaneously.

Ecosystem

Potential partners

UK partners

ORE Catapult

active engagement

Research & innovation body, Levenmouth access, grant pathway

Banks Renewables

first meeting target

Independent operator, onshore UK, accessible procurement

Fred Olsen Renewables UK

target

Independent, onshore & offshore, not OEM-captive

Greencoat UK Wind

post-validation

Largest UK wind fund, 1.8GW+, institutional buyer

Longer term

Statkraft UK

target

Norwegian state-owned, significant UK onshore portfolio, innovation appetite

RES Group

target

Major independent developer-operator, technology-forward, internal O&M

Visualwind

channel partner candidate

SCADA software provider, white-label integration opportunity

Market opportunity

Market opportunity at scale

1,100 GW
global installed wind capacity
$15–40k
avg. cost per unplanned failure
$17.5B
global wind O&M market / year

UK addressable market (TRL 7–8): Approximately 11,000 operational wind turbines in the UK. At a conservative £8,000 / turbine / year SaaS licence on 10% penetration, that represents an annual recurring revenue ceiling of ~£88M in the UK alone.

Global addressable market (TRL 9): 450,000+ commercial wind turbines globally. The total addressable market for SCADA-based condition monitoring — a segment currently dominated by OEM-locked proprietary systems — is estimated at $2.1B annually and expanding as turbine counts grow and offshore decommissioning risk increases. Zero-shot, hardware-free detection is a structural advantage in the 80% of the global installed base where independent CMS coverage is absent.

TorqueScope is a cold-start capable, hardware-free wind turbine fault detection platform. Its periodic signal layer begins scoring from day one with no historical failure data; its normal behaviour model calibrates over the first 60 days. No competitor offers useful output without 12-24 months of baseline collection. The Levenmouth dataset is the missing piece: ground truth at fault-category resolution, on an offshore-class turbine, accessible through ORE Catapult. £20,000 buys the data, the validation, and the first commercial conversation - for a platform built to work anywhere on earth.

beyond the UK market

The wider vision

The premium end of the wind O&M market is already served - expensively, by OEM-locked systems with £5-15k hardware installations per turbine and contracts that assume the operator has a maintenance budget, a service team, and years of historical failure data. TorqueScope's real opportunity is the 80% of the global installed base that has none of those things.

conventional CMS

  • ×Vibration sensor installation (£5-15k / turbine)
  • ×OEM service contract
  • ×12-24 months of baseline data collection
  • ×On-site calibration visit
  • ×Proprietary data format (OEM-locked)
  • ×Skilled CMS analyst to interpret output
  • ×Grid connectivity for real-time telemetry

TorqueScope

  • Existing SCADA connection (already present on any modern turbine)
  • No new hardware
  • No historical failure data required - cold start from day one
  • No OEM contract - manufacturer agnostic
  • Adapts to any SCADA format (validated: Siemens, EDP, offshore German)
  • Automated scoring - no specialist analyst required
  • Runs on standard cloud infrastructure at minimal cost

The delta is not marginal. It is the difference between having early warning and having nothing.

Where turbine failure costs most

In mature markets - UK, Germany, Denmark - an unplanned turbine failure is expensive but manageable: a mobilised crew, a spare part, a few weeks of lost generation. In resource-constrained markets, the same failure can mean months of downtime, lost grid capacity in a system with no slack, and communities returning to diesel generation. The economics of prevention are not just stronger - they are transformative.

Sub-Saharan Africa

~5 GW
Nigeria, Kenya, South Africa, Morocco, Ethiopia expanding capacity

Grid penetration of wind is rising but O&M infrastructure is nascent. A single large turbine failure on a constrained grid removes capacity that cannot be quickly replaced. TorqueScope's Kenya mode - already built, validated on Lake Turkana, Kipeto and Ngong Hills climate data - demonstrates the platform's readiness for this context.

Kenya mode built

South and Southeast Asia

~45 GW
India operational; rapidly expanding Vietnam, Philippines, Indonesia

India's wind sector is adding capacity faster than its O&M workforce is scaling. The gap between turbine installation and effective condition monitoring is widening. A zero-hardware, zero-training-data system is not a premium product here - it is the only viable product.

high growth

Island and remote grids

100%
of some island grids run on renewables with no generation redundancy

Small island developing states - Caribbean, Pacific, Indian Ocean - are deploying wind as primary generation. A single turbine failure on a 4-turbine island installation is a grid emergency. The economics of early warning here are not about optimising an O&M budget. They are about keeping the lights on.

grid-critical

What early warning is worth in constrained markets

$40-180k
unplanned failure cost, developing market
Includes logistics premium, import duty on parts, extended downtime on constrained grid
~80%
of global turbines with no independent CMS
Estimated from GWEC installed base vs reported CMS penetration
6-18 months
typical downtime in remote locations
vs 3-6 weeks in mature markets with local supply chains
£800-2,000
target TorqueScope cost per turbine per year
At target SaaS pricing for non-OECD markets - 10-20x cheaper than OEM CMS