The Shadow Economic Engine
How we turn public market data into recoverable value — every hour, on every asset.
Data pipeline
Asset telemetry (SoC, generation, consumption) is ingested via API or CSV upload. PREDAIOT connects to your existing SCADA or EMS — no hardware replacement required.
- › Battery State of Charge
- › Solar generation (MW)
- › Grid import/export
- › Asset operating mode
We align your asset data against official Oman market signals — System Marginal Price (SMP), scarcity pricing, and demand patterns — published by APSR and Nama PWP.
- › SMP (OMR/MWh) — APSR published
- › Scarcity price signals
- › Hourly demand profile — Nama PWP
- › Grid saturation indicators
The Shadow Economic Engine replays every dispatch decision your asset made — and calculates what the economically optimal decision would have been at each hour, given actual market prices.
- › 8,760 hourly evaluations per year
- › Up to 287 decisions per day in live mode
- › Price-signal-first decision logic
- › Curtailment signal integration
We calculate the Economic Decision Gap — the difference between what your asset earned and what it could have earned. Expressed in OMR. Cited to official data. Independently verifiable.
- › Economic Decision Gap in OMR
- › Economic Efficiency Index (0–100%)
- › Top dispatch timing opportunities ranked by OMR
- › Curtailment-to-revenue conversion estimate
Built for the dispatch floor
The engine sits next to your control room — not in place of it. Latency, traceability, and operator safety are non-negotiable. Numbers below are measured on internal benchmarks against the Nama PWP 2022 hourly dataset.
< 800 ms
Decision latency
From signal ingest to recommended action — end-to-end. Plenty of headroom for the 15-minute dispatch cadence used in real systems.
8,760 / yr
Hourly evaluations
Every hour of the year is replayed against the actual market price. No sampling. No shortcuts. Full-resolution audit.
100% traceable
Audit trail
Every action carries the price, SoC, and signal that justified it. Replayable in court — or in a board meeting.
Read-only by default
Operator safety
PREDAIOT recommends; the operator commits. Closed-loop control is opt-in and gated by configurable guardrails.
How the engine decides
Patent-pending. The core logic, simplified:
FOR each hour in operational year:
price_now = get_market_SMP(hour) // APSR published signal
price_future = forecast_peak(hour, +8hrs) // Forward price expectation
soc = battery.state_of_charge
curtail_flag = grid.saturation_signal
// Economic value of each action
V_charge = price_future × efficiency - price_now
V_discharge = price_now - avg_charge_cost(soc)
V_absorb = price_future × efficiency // Curtailment = free energy
// Priority decision
IF curtail_flag AND soc < 95%:
action = ABSORB // Convert waste to arbitrage
ELIF V_charge > threshold AND soc < 90%:
action = CHARGE at optimal rate
ELIF V_discharge > threshold AND soc > 10%:
action = DISCHARGE at optimal rate
ELSE:
action = HOLD // Wait for better price window
execute(action)
log_economic_outcome(hour, action, price_now, revenue_delta)Financial_Loss = Σ (P_expected − P_actual) × Market_Price
Economic intelligence vs. standard EMS
| Metric | Standard EMS | PREDAIOT |
|---|---|---|
| Decision basis | Fixed schedule (time-based) | Real-time price signals |
| Charge timing | 02:00–06:00 every night (fixed) | Cheapest available hours each day |
| Discharge timing | 17:00–21:00 every evening (fixed) | Highest-value window each day |
| Curtailment response | None — energy is wasted | Absorbed and resold via arbitrage |
| Arbitrage capture | ~62% of theoretical maximum | Up to 100% of theoretical maximum |
| Revenue on 500 MW (sim.) | 9,441,490 OMR / year | 10,304,393 OMR / year |
| Annual difference | — | 862,903 OMR / year |
Plugs into what you already run
No rip-and-replace. PREDAIOT reads existing telemetry and writes recommendations into the channels your operators already trust.
OPC UA, Modbus TCP, IEC 60870-5-104. Read-only telemetry pull on a 1–15 min cadence — no PLC code changes.
Tesla Megapack, Wärtsilä GEMS, Fluence Mosaic, Sungrow PowerTitan. REST/MQTT bridges or CSV exports both work.
APSR SMP, scarcity price, Nama PWP demand. Public sources only — no licensed feed required to start.
PDF audit, CSV decision log, Webhook to your data lake. JSON-LD schema on every line for downstream tooling.
What happens when reality breaks
Energy systems fail in messy ways. Signals stall. Forecasts diverge. Operators override. The engine is designed to never lie about what it knows.
| When this happens | The engine does this |
|---|---|
Market price feed drops | Engine falls back to the last 24-hour price curve plus seasonality offset. Confidence flag drops from HIGH to MEDIUM. Decisions continue. |
Asset telemetry stalls | Engine flags STALE on any signal older than 15 min and pauses dispatch recommendations until fresh data returns. No silent guessing. |
Forecast disagrees with realized price | Engine logs the variance per hour and auto-recalibrates the next-day forecast. Variance > 20% triggers an operator alert. |
Operator overrides the recommendation | Override is logged with reason code. The opportunity cost is calculated and shown in the next report — no judgment, just numbers. |
Honesty as a feature, not a footnote
- We do not predict the future. We replay decisions against signals you could have seen.
- We do not control your hardware unless you explicitly enable closed-loop mode with guardrails.
- We do not require new sensors, new PLCs, or new SCADA. Existing data is enough.
- We do not store private operational data outside your tenant. Read-only by default, scoped by contract.
- We do not sell anonymized data, market signals, or insights derived from your asset.
Where the numbers come from
Every published PREDAIOT figure ties back to one of four sources. Public sources are cited on the page they appear; client telemetry never leaves the tenant.
| Source | Cadence | Format | How we use it |
|---|---|---|---|
| APSR — System Marginal Price + scarcity price | Daily (2024 archive) | CSV / HTML from opendata.gov.om | Calibrates the price forecaster and the arbitrage thresholds |
| Nama PWP — MIS hourly demand | Hourly (2022 archive — 8,760 points) | CSV from omanpwp.om | Calibrates the 24-hour price profile used by the replay engine |
| Ministry of Energy & Minerals — renewable capacity register | Quarterly | PDF / OGD portal | Sizes the asset population and validates penetration scenarios |
| Client SCADA / EMS telemetry | 1–15 min live poll | OPC UA · Modbus TCP · IEC 60870-5-104 · CSV export | Feeds the AssetTelemetry stream in production |
How the replay actually works
Algorithm class
A priority-hierarchy heuristic with a value-of-action calculation per hour. Safety limits first, then curtailment absorption, then arbitrage windows (charge / discharge above configurable spread thresholds), then hold. Each branch is a closed-form expression on published SMP, round-trip efficiency, and cycle-life degradation — no black-box ML.
Every decision carries a reason string, a citation-ready price, and the exact equation branch that produced it. That is by design — regulated buyers cannot underwrite a model they cannot audit line-by-line.
Validation
Back-tested on the full 2022 Nama PWP hourly dataset — 8,760 points, no sampling. Every output number is independently reproducible from the methodology published on opendata.gov.om.
Senior technical review by a renewables IPP CTO in Oman (June 2026). Corrections — DC/AC ratio handling, bifacial + SAT gain factors, standalone BESS trajectory to 3 GW by 2030 — were incorporated and published.
Validated on official Oman data
All PREDAIOT simulations use official, publicly available Oman government data. Our methodology is published on the Oman National Open Data Portal. Every assumption is stated. Every source is cited. Every number is independently verifiable.
APSR Annual Report 2024
SMP: 9.120 OMR/MWh · Subsidy: 602.3M OMR · Economic cost: 1,306.3M OMR
opendata.gov.omPublished Methodology
PREDAIOT case study — Oman National Open Data Portal — June 2026
opendata.gov.om/ar/use-cases/1d4a8d55-1b2a-4b72-baba-e1a3763e842fOn a 10 MW solar / 15 MWh battery system: up to 25% uplift / ~$108,000 per year — peak-case scenario. Always labeled separately from the published 500 MW Nama PWP simulation.