| Popis: |
The research programme challenge was to establish optimum reliability for the Royal Bank of Scotland’s data centres electrical networks, given such mission critical buildings are vital to business operations, and loss of data centre services in terms of critical equipment outage is likely to negatively affect the UK economy. The Royal Bank of Scotland’s data centres are listed with the Committee for Protection of National Infrastructure (CPNI), and although despite being of national importance the current industry practice is to design and construct electrical infrastructures in-line with the Uptime Institutes Tier Classification table, which provides a design only approach for equipment topologies and expected annual downtime of critical systems. This approach is based on Inherent Availability (Ai) which utilises manufacturers design metrics and does not consider the complexities or issues encountered in an actual mesh connected electrical network i.e., those of an operational state. \ud \ud Therefore, in this investigation Operational Availability (Ao) of the actual data centre electrical network was established through a series of power system model simulations, which included load flow, short circuit analysis, protection device co-ordination and grading, arc flash and load point reliability. Power system models were constructed from the actual installed equipment, site load values and considered all possible operational scenarios and failure modes, with power system simulations being undertaken for every operational scenario that had been outlined in the original design specifications, including symmetrical fault current simulations undertaken at every critical busbar. \ud \ud The findings from each of the power system studies were analysed against the original system benchmarks, with gap analysis undertaken leading to the formation of a new and improved generalised approach for maximising Operational Availability (Ao). Electrical network issues were encountered across a range of power system studies with the more prominent areas including high voltage protection device arrangements and arc flash incidents located on the low voltage switchgear. In fact, over 30% of the electrical equipment investigated did not provide an optimal solution, with each one of these issues being simulated individually and critiqued with power system engineering theories and best practices, allowing subsequent improvements to be achieved. The improved solutions were again simulated for system failure rates thus provided a tangible metric comparison, with estimated outage times being reduced by as much as 45 hours per year. \ud \ud This new approach was formulated into a flowchart, a Generalised Approach to Improving Data Centre Operational Availability which is an advancement to the current Uptime Institutes Tier Classifications methodology, the approach was validated in this electrical network and reduced estimated outage times by over 45 hours per year. This is a substantial improvement in operational iii reliability metrics and the flowchart can be utilised by any other data centre owner i.e., for simulation and improvement of their electrical network with core focus on achieving an actual improvement in Operational Availability (Ao) for the exact installed equipment, settings and selected operational configurations. |
