DCN December 2016 | Page 24

modular systems additional work that is required includes more terminations for more electrical feeds; more units to set in place, more units to start up and test; and more control wiring and monitoring points. With a modular UPS in which multiple internal modules are used to increase redundancy, this work can be done in a factory setting, saving time and improving the predictability of the result. Risk tolerance themselves, the bypass cabinets, the back up battery system, assembly service and cost of installation. Installation, assembly service and battery costs were identical or almost identical for the non-redundant and Internally Modular options but more expensive for the Parallel Redundant system. This can be explained by the fact that the latter has a separate battery system for each of the three modules in its setup, whereas the other two options only have one battery system. Furthermore, the costs of the extra cabinets required for the Parallel Redundant option and its larger UPS modules also increase material and labour expenses. 24 Deployment time Speed of deployment also varies considerably between the N+1 options. A typical installation of a 1MW UPS can last between six and eight weeks, taken up by preparation of the room, including the concrete housekeeping pads; delivery and rigging of the UPS; running conduits for the UPS; pulling wires and making terminations and scheduling the start-up and testing. These installation steps are the same for the non-redundant design and the Internally Modular configuration, apart from having to add an extra power module for the latter option. But for a Parallel Redundant configuration, additional time is required, which can last for another week or two. The Fault tolerance varies between the three different configurations. Consider the following downtime risks: failure of the inverter or other power module, failure of the DC bus, or the need to bypass the main UPS for maintenance purposes. All else being equal (same UPS design, just different configurations), the Parallel Redundant configuration is the most resilient as the remaining UPSs support the load in those conditions. Conversely, the non-redundant UPS transfers to static (or maintenance) bypass, there will be no battery back up available, and in the event of a mains blackout, the IT under load would be brought down. In the case of the Internally Modular configuration, a maintenance event or a DC bus failure would also leave the IT load vulnerable to a mains event, although in the case of inverter or power module failure, the remaining UPS module would continue to support the load. The choice of power back up strategy must be contingent on many things, including criticality of IT systems, likelihood of disruption to mains power and the cost of any downtime. Nevertheless the Internally Modular approach often provides the needed level of redundancy for a very small cost premium over having no redundant back up for the power supply at all.