As distributed energy resources are increasingly integrated into low voltage grids, there is growing interest in preventive control strategies to anticipate and mitigate voltage and line loading violations. Recent approaches combine probabilistic forecasts with corrective actions to trigger preventive interventions before violations occur. However, residual violations persist even in advanced simulation studies and preventive control frameworks. When planning system improvements, it is often unclear whether failures stem primarily from forecast uncertainty or from insufficient flexibility, as both factors interact and their individual contributions cannot be directly observed.

This paper introduces a diagnostic method that systematically analyses residual violations through controlled counterfactual replay. The approach acts as a post-operational analysis layer on top of an existing preventive control framework. For each actionable residual violation, an oracle test replaces forecast information with perfect future information, while a high-flexibility test increases available corrective flexibility. Their individual and combined effects are used to determine which tested interventions are sufficient to remove the residual violation. Delayed reactive control is analysed separately using flexibility and timing interventions. This enables a systematic assessment of operational limitations under different operating conditions.

The method is evaluated using Monte Carlo time-series simulations of a realistic German low voltage distribution grid with photovoltaic generation, residential demand, and controllable curtailment. The analysis is based on more than 24,000 simulated operating points from probabilistic preventive control studies and 32 held-out operating cases. Prior results from the same framework demonstrate that probabilistic triggering benefits are strongly regime-dependent, motivating the need for systematic attribution. The results show distinct attribution profiles across the investigated control strategies, with actionable ML-risk residuals being more sensitive to the tested forecast-information intervention and prospective power flow screening (PF-only) residuals more frequently flexibility-limited.

The framework can provide operators with a simple and interpretable tool for post-operational evaluation and evidence-based planning based on original simulation studies using a realistic low voltage grid model.