Optimising protection coordination in critical power systems using a genetic algorithm

Protection coordination plays a critical role in maintaining the resilience of modern critical power systems, particularly in data centres where uninterrupted operation is essential. As electrical networks grow in scale and complexity, ensuring only the protective device closest to a fault operates becomes increasingly challenging. Effective coordination requires a balance of system modelling, technical expertise and optimisation techniques. Genetic algorithms (GAs) can be used to enhance protection coordination, serving as an alternative to conventional coordination methods.

In practice, protection engineers carry out coordination studies by analysing time-current characteristics (TCCs) using software such as ETAP or SKM. Relay, fuse and circuit breaker settings are adjusted through an iterative process guided by engineering judgement, system requirements and industry best practice. While effective, this approach can become increasingly time-consuming as power systems expand and coordination scenarios become more complex.

With increasing demands for reliability and performance, researchers have begun exploring computational methods to support the coordination process. The use of GAs has gained significant attention as a leading optimisation method. 
GAs can search large numbers of possible relay setting combinations and identify solutions that satisfy coordination constraints more systematically than traditional trial-and-error methods. Although GA-based protection coordinate on remains largely a research-focused application, it demonstrates how optimisation techniques can complement conventional engineering practices and help address the growing complexity of modern power systems.

The need for new optimisation techniques

Although manual relay coordination remains the industry standard and has proven to be a reliable approach, it does have several limitations:

• Requires iterative tuning
• Heavily reliant on the engineer’s experience
• Complex systems can lead to multiple borderline or conflicting settings
• Large networks make coordination time-consuming

In research settings, optimisation algorithms offer a way to automate parts of the coordination process and assess whether more effective solutions can be achieved. Among these techniques, GAs are one of the most widely studied due to their ability to handle the nonlinear constraints and complex interactions that are commonly encountered in protection coordination problems. By systematically exploring a large range of possible settings, GAs can identify solutions that may be difficult or time-consuming to find using conventional methods.

Genetic algorithm in the context of protection research

GAs are optimisation techniques inspired by biological evolution. They work by generating a population of potential solutions and improving them over successive generations through selection, crossover, and mutation.

In protection coordination research, GAs are used to automatically evaluate large numbers of relay setting combinations while ensuring coordination requirements, such as the coordination time interval (CTI), are maintained. This allows them to identify optimal or near-optimal settings more systematically than traditional trial-and-error methods.

While GA-based coordination is not yet widely used in industry, it remains an active area of research due to its ability to efficiently search large solution spaces and handle complex coordination constraints. As a result, GAs are increasingly being explored as a potential tool for future protection system optimisation.

Applying genetic algorithms to protection coordination

To evaluate the use of GAs in protection coordination, a data centre electrical system with known coordination issues was selected as the test case. This provided a practical environment for assessing how effectively a GA could optimise protection settings.

The process involved modelling relay and circuit breaker characteristics, defining pickup and time-dial setting ranges, and applying CTIs requirements based on industry standards and manufacturer recommendations. A fitness function was then used to penalise miscoordination and reward correct relay selectivity. MATLAB's GA engine was used to generate optimised settings, which were subsequently compared with manually coordinated solutions.

This approach reflects a growing area of research that explores how optimisation techniques can support protection engineering. While GAs are not yet widely used in industry, they provide a useful framework for investigating more efficient and systematic ways of solving complex coordination problems.

What the GA demonstrated

From the experiments performed on the test case:

• GA consistently achieved optimised settings with fewer CTI violations
• Reduced the time and iterations needed compared to manual tuning
• Manual coordination still produced valid results but required more engineering effort
• GA was able to refine specific relay pairs that were difficult to coordinate manually

These findings align with published literature that positions GA as a promising experimental tool, especially for systems with multiple coordination pairs or nonlinear constraints.

Industry context and practicality

While GAs have shown promising results in research, several factors currently limit their widespread adoption in industry. Protection coordination is a safety-critical activity, and engineers typically favour methods that are transparent, well understood, and easily validated. Most industry-standard software platforms, such as ETAP, SKM, and ProDesign, are built around established coordination workflows, although some are beginning to incorporate more advanced automation features.

Practical implementation also presents challenges. Applying a GA requires the development of custom code, the creation of suitable fitness functions, the preparation of accurate system data, and thorough validation of the resulting settings. Since every electrical system has unique characteristics, relay types, and coordination requirements, GA implementations often need to be tailored to each project, increasing the initial engineering effort compared to conventional approaches.

As a result, GAs should currently be viewed as a research and optimisation tool rather than a replacement for established protection engineering practices. Their value lies in demonstrating how advanced optimisation techniques can support future protection workflows and help engineers address increasingly complex coordination challenges.

Conclusion

This practicum demonstrates how GAs can improve protection coordination by providing a faster and more systematic way to optimise relay settings. While GAs are not yet widely used in industry, they show strong potential for complex systems where manual coordination can be difficult and time-consuming.

The 14 MW critical facility used in this project served as a practical test case to evaluate the approach. The focus was not the facility itself but rather demonstrating how GAs can be applied within a modern relay coordination framework.
As power systems become more complex, optimisation techniques such as GAs may become increasingly valuable. Their ability to automatically assess large numbers of possible settings makes them a useful tool for addressing challenging coordination problems.

In practice, GAs can work alongside established software such as ETAP or Amtech. The power system is modelled and analysed within the coordination software, while the GA runs externally in MATLAB to determine optimised relay settings. These settings can then be imported back into the software to verify coordination using Time–Current Characteristic (TCC) curves and confirm correct relay operation across the network.