Skip to content
Giorgi Gogitidze

Engineering case study

IEEE 14-Bus N−1 Contingency Analysis and Grid-Risk Dashboard

AC outage analysis, islanding detection, transparent severity ranking, and PTDF/LODF screening on a benchmark network.

  • Executed simulation
  • Automated test

Problem statement

Identify which single-branch outages most threaten the IEEE 14-bus benchmark while keeping topology failures, voltage criteria, synthetic thermal assumptions, and approximation error auditable.

A reproducible reliability workflow that validates the base case, removes every line and transformer, records violations, and compares fast DC screening with full AC results.

20

N−1 outages

Every line and transformer evaluated exactly once

1

Islanding event

Transformer 7-8 separates bus 8

1.174 MW

DC screening MAE

Five selected connected outages versus AC branch flows

13.393 MW

Base real losses

Validated AC base case on a 259 MW load

Project scope

  • 14 buses, 15 lines, five transformers, 11 loads, and 20 single-branch outages
  • Newton-Raphson AC analysis with a fast-decoupled validation solve
  • Graph-based islanding detection and deterministic severity ranking
  • Independent PTDF/LODF implementation and selected-case AC comparison

System or processing architecture

A validated AC base model feeds a contingency engine. Each outage is checked topologically before solving; connected cases produce bus and branch results, while islanded cases are reported without inventing a full-network solution. A separate DC screening path estimates active-flow redistribution.

My engineering contributions

  • Built the branch catalog, topology checks, AC contingency loop, score components, and detailed result tables.
  • Implemented PTDF and LODF matrices independently and cross-checked them against PYPOWER.
  • Created network figures, ranking charts, a self-contained dashboard, an executed notebook, and a technical report.
  • Kept source-case voltage-limit conflicts and missing physical branch ratings visible rather than suppressing them.

Important design decisions

  • Performed graph connectivity checks before AC solution so bridge outages cannot be misreported as ordinary convergence failures.
  • Used transparent stress-test ratings because the canonical case does not contain physical continuous branch ratings.
  • Stored every severity component so the prioritization index can be reconstructed and challenged.

Engineering methods

  • AC Newton-Raphson and fast-decoupled power flow
  • Complete line and transformer N−1 enumeration
  • Connected-component and unsupplied-island analysis
  • Voltage-breach and scenario-loading severity components
  • DC PTDF/LODF factor calculation and AC error comparison

Selected evidence

Readable results, with context

IEEE 14-bus network showing transformer 7-8 out of service and bus 8 islanded
Figure 1Top-ranked event: transformer 7-8 islands bus 8, so no valid full-network AC solution is claimed.
Bar chart ranking the ten highest-severity IEEE 14-bus contingencies
Figure 2Deterministic ranking from visible islanding, solution, voltage, loading, unsupplied-energy, and loss components.
Comparison of DC LODF estimates and full AC post-contingency branch flows
Figure 3Selected connected cases show where lossless DC screening is useful and where nonlinear AC effects remain important.

Verification and testing

  • Two AC algorithms agree within 7.12×10⁻¹⁰ p.u. voltage and 3.74×10⁻⁸ degrees angle.
  • Generation minus demand minus losses closes to 3.55×10⁻¹⁵ MW in the verified run.
  • Tests cover contingency completeness, boundary logic, stable ranking, bridge behavior, and factor matrices.
  • Independent PTDF and finite non-bridge LODF entries match PYPOWER near machine precision.

Honest limitations

  • The IEEE 14-bus model is a teaching benchmark, not a current utility system.
  • Thermal ratings are transparent stress-test values, not utility or manufacturer limits.
  • Generator reactive limits and PV-to-PQ switching are not enforced in this comparison configuration.
  • DC screening cannot establish voltage security, reactive behavior, losses, convergence, or islanded operation.

Tools and technologies

  • Python
  • pandapower
  • NetworkX
  • PYPOWER
  • pandas
  • Plotly
  • Matplotlib

Skills demonstrated

  • Power-system reliability
  • Contingency analysis
  • AC power flow
  • PTDF/LODF
  • Topology analysis
  • Technical visualization

Downloadable proof of work

Review the evidence directly

Technical reportPDF · AC, topology, and screening resultsDownload →Interactive dashboardHTML · self-contained result explorerDownload →Source and evidence bundleZIP · source, tests, results, and figuresDownload →

Continue in Energy & Grid Systems

Small Commercial Facility Power Distribution Design and AnalysisView case study →Solar + Battery Microgrid Energy-Management SimulatorView case study →