Definition
SSA measures whether the outcomes a digital twin predicts for specific scenarios (cooling failures, power failures, capacity additions, extreme ambient events) match what actually occurs when those scenarios are observed or controlled tests are conducted.
The Measurement Gap
Standard data center KPIs (PUE, DCIM uptime dashboards, cooling system efficiency) measure what happened in actual facility operation but have no mechanism for evaluating what would have happened under alternative scenarios. Incident post-mortems document actual outcomes but cannot be compared to what the twin predicted, because that comparison is never formalized. When management asks "what would happen if we lost a cooling zone?" the twin's answer is accepted at face value with no record of how accurate its previous scenario predictions have been. This blind spot is especially dangerous for resilience planning, where over-optimistic scenario predictions lead teams to believe they have more margin than they do.
Decision Supported:
SSA determines whether a digital twin's scenario simulation outputs are accurate enough to inform resilience planning, capacity decisions, and emergency response procedures, or whether those predictions carry unknown error rates that undermine their reliability.
Formula
Plain language: For each validated scenario (a real event or controlled test where the twin made a prediction and an actual outcome was measured) calculate how far off the prediction was as a percentage of the actual outcome. Average those errors across all S validated scenarios and subtract from 1.0. Track optimistic errors (twin predicted better outcomes than occurred) separately, because optimistic errors are the dangerous kind.
Notation:
SSA = 1 − (1/S) × Σₛ |Outcome_predicted_s − Outcome_actual_s| / Outcome_actual_s
Variables:
| Variable | Definition | Unit |
|---|---|---|
| Outcome_predicted_s | Digital twin's predicted outcome for scenario s (e.g., predicted max inlet temperature, predicted load shedding required) | Varies by scenario type |
| Outcome_actual_s | Actual measured outcome when scenario s occurred or was tested | Varies by scenario type |
| S | Total number of validated scenarios with both predicted and actual outcomes | Count |
| s | Index running from 1 to S across all validated scenarios | n/a |
Track Optimistic_error_rate = count of scenarios where Outcome_predicted was better than Outcome_actual / S.
Confidence Grading (required alongside every SSA value):
| Grade | Basis | Interpretation |
|---|---|---|
| A: Empirical | 5+ validated scenarios per category (cooling failure, power failure, capacity event, extreme ambient) provide direct outcome comparison | Measurement grounded in observed outcomes; highest confidence |
| B: Hybrid | 1-4 validated scenarios per category plus Monte Carlo simulation of twin prediction error distributions | Blends observation and simulation; moderate confidence |
| C: Simulated | Zero real events in a scenario category; SSA derived entirely from synthetic scenario injection | Modeled, not observed; treat as indicative, not validated |
Report the confidence grade alongside every SSA score. An SSA of 0.90 Grade A means the twin's predictions are within 10% of reality across observed events. An SSA of 0.90 Grade C is a simulation that agrees with itself: internally consistent but not reality-tested. Grade C scores should trigger proactive scenario injection tests and cross-site validation rather than acceptance as calibrated.
Proposed formula, pending validation pilots.