Why 24-Hour-Ahead Forecasts Are the Wrong Tool for Real-Time Balancing
Day-ahead forecasts are accurate enough for unit commitment. They're too coarse for real-time imbalance management — and most utilities are using the same model for both problems.
Read more
Dispatch Sequencing Under OpenADR 2.0b: What the Spec Doesn't Tell You
OpenADR 2.0b describes the signaling protocol. It says nothing about which assets to curtail first or how to handle partial responses. That logic is entirely on the utility side — and most implementations get it wrong.
Read more
BAL-001-3 vs BAL-002-2: The Performance Metrics That Actually Get Utilities Cited
Most compliance violations aren't about egregious events. They're about ACE accumulation patterns that look acceptable in isolation but fail the 10-minute average thresholds in aggregate. Here's where operators miscalculate.
Read more
Stuck Registers and Silent Failures: SCADA Data Quality Problems That Break Forecast Models
A stuck register that reports the same MW value for 45 minutes doesn't trigger alarms in most EMS configurations. It also silently corrupts the training data for any model that doesn't validate for it explicitly.
Read more
Net-Load Forecasting When 30% of Your Generation Is Behind-the-Meter Solar
The traditional load forecasting assumption — that generation is dispatchable and load is uncertain — breaks down when distributed solar represents a third of peak capacity. The variable you're actually forecasting has changed.
Read more
Gradient Boosting vs. ARIMA for Short-Interval Load Forecasting: An Honest Comparison
ARIMA handles trend and seasonality cleanly. Gradient-boosted ensembles handle non-linear interactions between weather, time-of-week, and load better. Neither is universally superior — the decision depends on what kind of error you can tolerate.
Read more
The Real Cost of Imbalance Energy: How ISO/RTO Settlement Charges Accumulate
Imbalance energy charges in ISO/RTO markets can represent 8–15% of a utility's total energy procurement costs. Here's how the settlement mechanism works and where forecast accuracy directly affects the settlement statement.
Read more
How to Evaluate a Load Forecasting Vendor: A Technical Checklist for Utility Operators
Vendor marketing emphasizes MAPE benchmarks that may not reflect operational value. This checklist focuses on the technical and contractual questions that predict whether accuracy improvements will translate to reduced balancing costs.
Read more
Electric Vehicle Load Growth Is Not What Your Forecast Model Thinks It Is
EV adoption projections consistently underestimate charging load concentration in time and space. The distribution problem matters more than the aggregate MW growth number for grid operations.
Read more
FERC Order 2222 and Aggregated DER: What Grid Operators Need to Prepare For
FERC Order 2222 requires RTOs and ISOs to allow aggregated DER to participate in wholesale markets. The operational and forecasting implications arrive well before full market participation begins.
Read more
Load Shedding vs. Demand Response: Why Operators Still Reach for the Wrong Tool
The conditions where operators choose load shedding over DR when DR would work better have a consistent pattern. The gap is usually not a technical limitation — it's an information problem.
Read more
How to Build a Load Forecasting Pilot That Produces Defensible Numbers
A poorly designed pilot produces numbers that look good in a presentation but don't reflect operational performance. Here's how to structure a pilot that generates accuracy statistics your team can trust.
Read more