Learnings

Outcomes

The Volta – Advanced Scheduling Adviser project delivered a set of concrete outputs during the Alpha phase that together provide clear evidence of what the project achieved and why it is valuable for NESO. These outputs moved the concept of automated scheduling beyond analysis alone and into a demonstrable Proof of Concept supported by tested scenarios, user-facing artefacts, technical documentation, and actionable learning. 

A working Proof of Concept was delivered and demonstrated. 
The Alpha phase produced a functioning optimisation-based Proof of Concept centred on the Dynamic Selection Engine. This tool accepted historical system inputs and generated BMU scheduling recommendations across the 24-hour to 4-hour-ahead horizon at minute-level resolution. The solution was exercised using multiple complex historical days provided by NESO, including scenarios with high demand and concurrent system constraints. Across these test cases, the model generated feasible schedules, demonstrating in practical terms that the scheduling challenge can be solved algorithmically within timescales relevant to strategic scheduling workflows. 

Specific analytical and user-facing capabilities were developed. 
In addition to the optimisation engine, the project delivered defined front-end capabilities through the System Overview and Strategy Workspace components. These enabled users to view system pinch points, margin analysis curves, scheduled BMUs, and cost indicators, and to explore alternative scheduling options within a human-in-the-loop workflow. This is an important tangible outcome of Alpha because it showed not only that recommendations could be produced, but that they could also be interrogated, understood, and adapted in a form relevant to Control Room users and internal stakeholders.

Explainability and traceability were embedded into the delivered solution. 
The project did not simply produce automated recommendations; it also demonstrated a method for explaining them. Through the hierarchical solving approach developed in Alpha, BMU actions could be linked back to the primary system need they were addressing, such as margins, voltage, inertia, or network constraints. This created a clearer audit trail between system conditions and the resulting recommendations, providing a tangible improvement in transparency compared with an opaque optimisation output and supporting the project's objective of trustworthy decision support in a safety-critical environment.

Technical documentation and knowledge transfer outputs were produced. 
Deliverable 7 and the wider Alpha outputs captured the architecture, optimisation methodology, data processing approach, modelling assumptions, technical debt, and known limitations of the Proof of Concept. A knowledge exchange session was also undertaken to walk through the code and solution approach. These are important project outcomes in their own right, because they provide NESO with documented assets that can be used to inform any future MVP phase rather than leaving learning as implicit or dependent on supplier knowledge alone. 

The project generated evidence-based learning to inform future investment decisions. 
Through delivery and testing, the project identified where simplifications were required, what data constraints limited realism, where alignment with the Open Balancing Platform would be needed, and which capabilities would require further development before operational deployment. This learning is a tangible outcome of Alpha because it reduces uncertainty for future decisions. NESO now has evidence not only that automated scheduling is feasible in principle, but also which technical, architectural, and operational issues would need to be addressed to progress the concept responsibly.

Overall, the project delivered more than a set of conclusions: it produced a working optimisation Proof of Concept, demonstrated it against select historical scenarios, created user-facing workflow artefacts, documented the underlying methods and assumptions, and generated a clear body of learning which would be required to move towards an MVP. This provides a stronger, evidence-based account of project outcomes and a robust foundation for future review and decision-making. 

Lessons Learnt

Several important lessons were identified through delivery of the Volta – Advanced Scheduling Adviser project, which are relevant for future innovation projects of similar complexity and strategic importance. 

Early data readiness and access planning is critical. 
While the project successfully demonstrated feasibility using static and synthetic datasets, access to complete, production‑grade operational data remained a constraint throughout Alpha. Future projects of this nature would benefit from careful consideration regarding whether live operational data is truly required. In many cases, historic static data will be sufficient at proof-of-concept stage. If this is not deemed sufficient, then early engagement on data access, security, and governance will be required. Early clarity on what data can be used, at what fidelity, and under what conditions would reduce the need for abstraction and accelerate progress towards operational relevance. 

Strong user engagement materially improves solution credibility. 
Close and continuous engagement with Control Room engineers and SMEs was essential in shaping a solution that aligned with real operational practices. Iterative user feedback influenced both the optimisation logic and the emphasis on explainability and human‑in‑the‑loop design. This reinforced the importance of embedding user research and validation throughout delivery, rather than treating it as a one‑off discovery activity. 

Explainability should be treated as a core requirement, not an enhancement. 
The project demonstrated that optimisation performance alone is insufficient for adoption in safety‑critical operational environments. Clear, intuitive explanations of why actions are recommended were fundamental to building trust and ensuring auditability. Future projects should treat explainability and governance alignment as primary design constraints alongside technical performance. 

Progressive technical validation reduces delivery risk. 
Adopting a phased approach that prioritised feasibility and learning over full production readiness proved effective in managing complexity and risk. Simplifying modelling assumptions and deferring full architectural compliance allowed the team to demonstrate value quickly while still generating actionable insight for future phases. This reinforces the benefit of staged innovation pathways where complexity is increased incrementally.

Architecture and platform alignment require early and sustained coordination. 
Engagement with the evolving target platform (e.g. OBP) highlighted the importance of close alignment between innovation delivery and enterprise architecture direction. While deferring full compliance was appropriate for Alpha, future phases will require early, sustained coordination between innovation teams, platform architects, and delivery governance to avoid rework and ensure smooth transition from Proof of Concept to operational systems.

Overall, the project confirmed that technical innovation, user trust, data readiness, and governance considerations must progress together. Addressing these elements holistically from the outset will improve delivery efficiency, adoption readiness, and value realisation in future projects.