1 MW Pilot Plant Plan

The 1 MW pilot is not designed to earn NPV — it exists to convert a paper model worth RM 92.7 M of NPV into bank-grade measured data. RM 5–10 M buys a year of measured data plus TNB process experience plus an EPC/supplier relationship network — the cheapest insurance available before scaling.

Why a pilot first

The +82.7 M NPV / 13% IRR figure from our BTM economics report is a model output, predicated on:

5-year NASA POWER 9-point grid-mean GHI → real factory rooftops may have shading, soiling, sub-optimal orientation
Industry benchmark PR 0.78 → actual could be 0.70–0.83
TNB E3 industrial tariff + 16 sen ICPT → actual settlement may have definitional differences
BESS RTE 0.88 → actual 80–90%, with degradation over time
LP dispatch assuming perfect foresight → actual forecasts have error

A 10% deviation on any one of these moves NPV by RM 20 M+.

From an IC perspective:

Skip the pilot, go straight to 60 MW: trade RM 5–10 M of “insurance premium” for RM 200 M+ of construction and a 5-year site lock-in. If the model is materially off, the cost to redo is enormous.
Run the 1 MW pilot first: spend RM 5–10 M for 12 months of measured data and a calibrated financial model. If NPV validates within ±10% → proceed to replication. If the deviation is large → recalibrate and run another year of pilot.

A 1 MW PV-only pilot is roughly NPV-neutral (full budget below), so the pilot is not financially painful in itself — its main value is lowering execution risk at scale.

Five concrete propositions to validate

#	Proposition	Validation method	Pass threshold
1	Actual annual PV yield ≈ model forecast	12 months of hourly generation vs NASA POWER + model PR	Measured PR ∈ [0.73, 0.83]
2	BTM bill reduction ≈ LP-model forecast	TNB monthly invoice vs LP output	±10%
3	BESS RTE and degradation ≈ vendor warranty	Charge/discharge MWh ratio, capacity-warranty test	Year-1 RTE > 0.85, capacity fade < 5%
4	EPC pricing ≈ model assumption	Actual quote vs RM 3.5 M/MW (PV) + RM 5 M/MWh (BESS)	±15%
5	TNB / NEM 3.0 process can be executed	Application → interconnect → metering → settlement, total time	< 9 months

Pass 4/5 → proceed to replication; pass 2–3 → diagnose and run another year; pass 0–1 → halt, redo the model.

Site selection criteria

Ideal pilot host:

Dimension	Criterion	Why
TNB tariff tier	E2 or E3 (medium-voltage industrial)	These are the tariffs the model is built on; replication uses the same definitional basis
Annual consumption	≥ 8 GWh/yr (~1 MW average power)	1 MW PV won’t oversupply (avoids large NEM 3.0 export)
Operating pattern	24/7 continuous, or single-shift with 14:00–22:00 active window	Aligns with the ETOU peak window — maximises BESS arbitrage value
Roof / land	≥ 5,000 m² unshaded area	Roughly the footprint a 1 MW PV array needs
Lease term	At least 7 years remaining	PV asset life is 25 years, but year-7+ allows relocation/renewal
Location	Within the JS-SEZ corridor (Iskandar / Sedenak / Kulai / Pasir Gudang)	Geographic representativeness; future anchor tenants are also in this corridor
Cooperation	Willing to share hour-level load data and assist with interconnect paperwork	The hidden determinant of pilot success

Sourcing channels (run in parallel):

Direct outreach to existing manufacturing customers — food processing, electronics assembly, textiles (these are the most ETOU-exposed factories)
MIDA / IRDA introductions — government agencies maintain industrial-park rosters
EPC referrals — local PV EPCs (e.g. Solarvest, Cypark, Pekat Group) typically have warm leads

System architecture

flowchart LR
    classDef grid fill:#1e3a5f,stroke:#3b82f6,color:#e0f2fe;
    classDef equip fill:#0e8a8a,stroke:#1eb6b6,color:#f0fdfa;
    classDef monitor fill:#5b21b6,stroke:#a78bfa,color:#ede9fe;
    classDef load fill:#7c4a02,stroke:#d97706,color:#fef3c7;

    GRID["TNB grid<br/>E2 / E3 medium-voltage"]:::grid
    METER["Main meter<br/>(TNB MD billing)"]:::equip
    BUS["Factory distribution bus<br/>~1-3 MVA peak"]:::equip
    PV["1 MW PV array<br/>~5,000 m² rooftop<br/>String inverters"]:::equip
    BESS["BESS<br/>250 kW × 4 h = 1 MWh<br/>LFP chemistry"]:::equip
    LOAD["Factory load<br/>(continuous shift)"]:::load
    MON["Monitoring SCADA<br/>15-min telemetry<br/>+ historian"]:::monitor

    GRID <-->|"Bidirectional metering<br/>(pilot import-only)"| METER
    METER --> BUS
    PV --> BUS
    BESS -->|"Discharge"| BUS
    BUS -->|"PV surplus + off-peak arbitrage"| BESS
    BUS --> LOAD
    BUS -.->|"Data"| MON
    PV -.-> MON
    BESS -.-> MON
    METER -.-> MON

Key design principles:

BTM-only (no NEM 3.0 export): simplifies compliance and focuses validation on the BTM economic model itself. The pilot is not designed to earn NEM export revenue.
PV self-consumption + BESS internal arbitrage: all generation directly offsets factory load; BESS discharges in peak, charges in off-peak, charges from PV surplus.
15-minute telemetry: far finer than the TNB monthly invoice grain — enough to calibrate the LP model precisely.

Three budget tiers

The IC should pick a tier after seeing the trade-offs:

Option A: PV-only (lowest risk)

Item	Qty	Unit (RM)	Subtotal (RM)
1 MW PV EPC turnkey	1	3,500,000	3,500,000
MV interconnect + factory wiring	1	500,000	500,000
Monitoring SCADA	1	200,000	200,000
Project management + compliance	1	200,000	200,000
Contingency (10%)	—	—	440,000
Total			4,840,000

Economics: ~RM 532k/yr savings; simple payback ~9 years; 20-yr NPV @ 8% ≈ +RM 384k (slightly positive, bankable solo).

Validation coverage: propositions 1, 2, 4, 5 (4 of 5). BESS dispatch logic is not validated.

Option B: PV + small BESS (recommended) ⭐

Item	Qty	Unit (RM)	Subtotal (RM)
1 MW PV EPC turnkey	1	3,500,000	3,500,000
BESS DC block (LFP, 4-h)	250 kWh	5,000 / kWh	1,250,000
BESS PCS inverter	62.5 kW	(included above)	—
MV interconnect + factory wiring	1	800,000	800,000
Monitoring SCADA + BESS BMS integration	1	300,000	300,000
Project management + TNB / NEM 3.0 / SEDA compliance	1	300,000	300,000
Contingency (10%)	—	—	615,000
Total			6,765,000

Economics: ~RM (532 + 25) = 557k/yr savings; simple payback ~12 years; 20-yr NPV @ 8% ≈ −RM 297k (slightly negative — roughly RM 0.3 M of “tuition”).

Validation coverage: all 5 propositions. This is the default recommendation — RM 0.3 M of tuition buys full-stack BESS dispatch validation, the cheapest insurance available before scaling.

Option C: full 1 MW PV + 1 MWh BESS

Item	Qty	Unit (RM)	Subtotal (RM)
1 MW PV EPC turnkey	1	3,500,000	3,500,000
BESS DC block (LFP, 4-h)	1,000 kWh	5,000 / kWh	5,000,000
BESS PCS inverter	250 kW	(included above)	—
MV interconnect + factory wiring	1	1,000,000	1,000,000
Monitoring SCADA + BESS BMS integration	1	400,000	400,000
Project management + full compliance	1	400,000	400,000
Contingency (10%)	—	—	1,030,000
Total			11,330,000

Economics: ~RM (532 + 100) = 632k/yr savings; simple payback ~18 years; 20-yr NPV @ 8% ≈ −RM 5.1 M (clearly negative).

Validation coverage: all 5 propositions, plus full-topology BESS capacity utilisation and degradation curve — a complete engineering reference for the 60 MW anchor site.

When this fits: if the institution is willing to treat the RM 5 M shortfall as engineering-grade R&D investment.

Tier comparison

Dimension	Option A (PV-only)	Option B (recommended) ⭐	Option C (full)
Capex	RM 4.84 M	RM 6.77 M	RM 11.33 M
20-yr NPV @ 8%	+0.4 M	−0.3 M	−5.1 M
Propositions validated	4/5	5/5	5/5 + capacity fade
Information per RM of capex	medium	high (best trade-off)	high, but capex doubles

BESS chemistry: LFP vs NMC trade-off

Dimension	LFP (lithium iron phosphate)	NMC (nickel manganese cobalt)
Safety	⭐⭐⭐⭐⭐ Thermal runaway at 270 °C, hard to ignite	⭐⭐⭐ Thermal runaway at 210 °C, requires stronger BMS
Cycle life	6,000–8,000 cycles @ 80% DoD	3,000–5,000 cycles @ 80% DoD
Volumetric energy density	90–160 Wh/L	200–280 Wh/L
Capex (2024)	RM 5.0 M/MWh (incl. PCS)	RM 6.0–7.0 M/MWh
Cycle cost	low (longer life amortises)	medium
Supply chain	China-dominated, CATL / BYD spot	Korean/Japanese suppliers, more diversified
Best fit	Large capacity, stationary, long discharge duration	High power density, mobile applications, space-constrained
JB VPP pilot recommendation	✅ Default choice	❌ Not recommended

Why LFP by default:

BTM applications are insensitive to volumetric density — factory yards or building-adjacent enclosures fit a 40-ft container easily; a 1.5× volume penalty is irrelevant.
LFP 6–8k cycle life at 250–365 cycles/yr translates to 16–30 years of theoretical life — same horizon as the PV.
LFP capex is already 20% lower and will continue to widen the gap (resource side: Li is scarce, but Fe + P are not).
LFP safety is critical for factory deployments — an insurance denial after a fire would be catastrophic.
LFP’s China supply chain plugs directly into CATL, BYD, EVE Energy — local prices are transparent and lead times are short.

The only case for NMC: high C-rate (≥ 1C) short-burst applications (FFR, frequency regulation). Our pilot is energy-arbitrage (C-rate ≤ 0.25), where LFP is the perfect fit.

Timeline

gantt
    title 1 MW pilot end-to-end timeline (~21 months)
    dateFormat YYYY-MM
    axisFormat %Y-%m
    section Prep
    Site selection + sign MOU         :a1, 2026-06, 2M
    System design + EPC tender         :a2, after a1, 2M
    section Build
    NEM 3.0 / TNB interconnect filing  :crit, b1, after a2, 4M
    EPC procurement + delivery         :b2, after b1, 3M
    Install + commissioning + grid-tie :b3, after b2, 2M
    section Operate
    12 months of operating data        :c1, after b3, 12M
    section Evaluate
    Model calibration + validation report :d1, after c1, 2M
    IC scale-up decision               :milestone, d2, after d1, 1d

Critical-path risk: TNB / NEM 3.0 interconnect approval (highlighted red). Industry experience is 4–9 months, depending on factory voltage class and the speed of the TNB regional office. Selecting an EPC with a live NEM 3.0 project can compress this to 4 months.

Monitoring and instrumentation plan

🎯 Design principle: what we measure determines what we learn. Monitoring spend of RM 0.2–0.4 M is a critical input to the scale-up decision.

Measurement	Granularity	Instrument	Purpose
PV DC + AC output	15 min	Inverter built-in + revenue-grade meter	Calibrate PR, detect module degradation
Factory total load	15 min	TNB main-meter SCADA tap	Calibrate LP model load shape
BESS charge/discharge MWh	15 min	BESS BMS + revenue-grade meter	RTE calculation, capacity-fade tracking
BESS SoC profile	1 min	BESS BMS	Validate LP dispatch execution accuracy
Grid import / export	15 min	TNB main meter (bidirectional)	Reconstruct LP settlement, reconcile against invoice
Rooftop weather station	1 min	GHI + temperature + wind sensors	Correct NASA POWER grid-mean bias
Per-string current	1 hour	String monitor	Early detection of soiling, shading, module faults
ICPT / ETOU actual invoice	monthly	TNB monthly invoice scan + OCR	Validate LP settlement basis

Data stack:

Edge gateway: single-board computer + Modbus / IEC 61850 ingest
Historian: InfluxDB or TimescaleDB (open source)
Dashboard: Grafana
Monthly report: automated Python script comparing LP model vs measured

All data persisted locally and mirrored to cloud S3, ensuring 12 months of complete operating records.

Risk register

Risk	Probability	Impact	Mitigation
Cannot find a suitable host site	medium	6-month delay	Negotiate 3 backups in parallel; source via EPC + IRDA dual-track
TNB interconnect approval slips > 6 months	high	Whole project delayed	Pick an EPC with NEM 3.0 experience; engage TNB Johor regional office early
Measured PR < 0.73	medium	Cashflow 10–20% lower	EPC contract with PR performance guarantee + auto LD payments (industry standard 0.78 ± 0.03)
BESS year-1 capacity fade > 5%	medium	NPV deteriorates over time	Choose LFP (fades 2× slower than NMC) + capacity-warranty clause (annual fade < 2%)
Factory terminates lease early	low	Asset relocation / removal	Lock minimum 7-year usage in MOU; rooftop PV is removable, BESS container is mobile
ETOU / ICPT policy reform	medium	Model needs redo	Tariff-floor clause in PPA (see risk report)
EPC quotes ≥ 15% above expectation	medium	NPV deteriorates	Tender ≥ 3 EPCs; fixed-price contract; 10% contingency
Data acquisition system failure / data loss	low	Validation report invalidated	Dual-redundant SCADA + cloud mirror + monthly backup audit
IC turnover, support withdrawn	low	Project shelved	Lock 12-month budget + milestone-gated reviews

Funding structure options

Option 1: Own funds (off-balance-sheet)

100% cash. 100% upside, 100% downside.

Best for: funds with ample cash and a preference for full data control + decision flexibility.

Pros: simplest process, fastest decisions, cleanest data ownership Cons: ties up RM 5–10 M of cash for 12–18 months

Option 2: Project finance (70% debt + 30% equity)

Bank loan covers 70% (5–6% per annum, 10-year tenor); equity covers 30%.

Best for: amplifying leverage and stress-testing the bank diligence process (paving the way for scale).

Pros: equity ROIE roughly doubles via leverage; running the bank process once is dress rehearsal for scale-up Cons: loan covenants, longer diligence (add 3 months), more contractual complexity

Recommended bank outreach: CIMB (green finance practice), Maybank (project finance experience).

Option 3: Third-party PPA / opex model

A third-party developer (e.g. Solarvest, Cypark, TNB-X) finances and builds; the factory buys at a PPA price. We participate as LP investor or advisor.

Best for: fully avoiding capex risk, focusing on model/methodology IP.

Pros: zero capex risk; what we acquire is contract-structure and EPC-process experience Cons: data control diluted; lower long-run economics than self-build; not always possible to negotiate full data-access carve-out

Recommendation

Default recommendation: Option 1 (own funds) — reasoning:

The pilot’s core value is learning + calibration, not ROI. Keep the financing structure as simple as possible.
RM 5–10 M cash is manageable relative to fund size.
Data + relationship network can be 100% internalised.
After 12–18 months, when scaling to the 60 MW anchor site, switch to Option 2 (project finance).

Deliverables (post-pilot)

By approximately 2028 Q1, the pilot should deliver:

Deliverable	Use
12-month hourly operating dataset	Model calibration + downstream academic / commercial citation
Model calibration report	Tightens NPV uncertainty from ±20% to ±5%
TNB process playbook	Application → interconnect → settlement, step by step (timing / forms / contacts)
EPC vendor benchmark pricing	Strong leverage for the 60 MW anchor-site negotiation
BESS chemistry measured performance	Degradation curve, RTE, true O&M cost
Updated risk register	Which risks were disproved, which new risks surfaced
IC scale-up decision memo	Go / no-go / redo — explicitly delivered to the decision body

Next steps

If the IC approves the pilot in principle:

2026 Q2: stand up the pilot project team (project manager + systems engineer + finance liaison)
2026 Q3: kick off site selection (3 factories in parallel) + EPC tender
2026 Q4: sign MOU + open NEM 3.0 application
Quarterly IC reporting on progress + key milestones (interconnect approval, grid-tie, month-6 calibration, month-12 report)

Detailed financial model, contract templates, and technical specifications will follow within 4–6 weeks of IC approval in principle.

Related reports:

BTM economics report — full model for the 60 MW anchor site
BESS investment trigger curve — the window for expanding into BESS once the pilot is complete
Risk-adjusted NPV — how pilot data tightens the risk distribution
PPA term-sheet — contract structure for downstream anchor customers