shiviancodes/riskwatt

RiskWatt - energy risk intelligence for African renewables. Probabilistic curtailment, counterparty, and settlement risk modelling. GridCap (curtailment risk engine) is the first module # SA Energy Intelligence Platform A multi-module probabilistic energy analytics platform for South African renewable energy. Core shared infrastructure lives in `core/`; domain modules live in `modules/`. --- # GridCap **Probabilistic congestion curtailment risk quantification for South African renewable energy.** --- ## The Problem South Africa's Renewable Energy Independent Power Producer Procurement Programme (REIPPPP) Bid Windows 7 and 8 are connecting into a Western Cape and Eastern Cape grid that is approaching firm capacity limits at five key 400 kV substations: Agulhas, Bacchus, Droerivier, Nuweveld, and Neptune. The NTCSA Grid Connection Capacity Assessment (GCCA, October 2025) allocates curtailment budgets to these substations ranging from 100 MW (Agulhas) to 720 MW (Nuweveld). The industry standard for modelling curtailment risk is a flat 3% annual assumption, applied uniformly across all projects regardless of substation, season, or hour of day. This assumption originates from developer spreadsheets and has become embedded in lender independent reports (IMs) without empirical validation. Primary research across BW7 developer interviews and IM template reviews finds that the P90 curtailment rate at high-exposure substations (Nuweveld, Neptune) is materially higher than 3% flat — with NPV drag estimates of R33m on a 100 MW project at WACC 12% over a 20-year PPA. No publicly available probabilistic tool exists for quantifying this exposure at the substation level. GridCap fills this gap by combining ERA5 reanalysis (35 years of hourly wind/solar), GCCA firm capacity data, and AEMO SA transfer labels into a calibrated probabilistic engine that outputs P10/P50/P90 curtailment distributions with hour-level WEPS TOU revenue impact quantification. --- ## Architecture ``` ┌─────────────────────────────────────────────────────────────┐ │ Layer 1: Ingestion │ │ ERA5 (cdsapi) │ GCCA (PDF) │ NERSA (PDF) │ AEMO …