newsroom

The colocation business is highly competitive with everyone looking for cost-efficient designs without any compromise on reliability and efficiency. And of course, speed of deployment remains a key requirement and focus for larger colocation data centers. We believe the critical factor in achieving that is close collaboration between the different domains of expertise and players (e.g. end users, engineering companies, panel builders, vendors etc.) involved. This is the reason why at Schneider Electric we have a cross-functional, cross-organizational team of senior power engineers dedicated to Large Data Center electrical distribution applications to bring expert support and optimized solutions to end users, engineering companies, panel builders and contractors.

Since we are focused exclusively on large data center applications, we are continuously learning from our customers and partners. This includes not only the specific technical needs they have in terms of design and operation, but also the business constraints and trends to drive adaptations on our solutions. We bring our expertise in power systems and electrical distribution design. Additionally, we understand how to take advantage of all the potential performance and functionality of our products and equipment – that spans from the connection to the utility, all the way down to the rack PDU.

Reliable Reference Design for Colocation Electrical Distribution

When we developed Reference Design RD107, we wanted to create a cost and energy efficient design (~12MW IT) for colocation companies who needed to build fast, step by step modular approach, with a flexible electrical infrastructure that could adapt to their customers’ needs.

Access Reference Design RD107 now >

For example, the design features a mix of IT data halls, some supporting standard racks or OCP-type racks with integrated battery systems. Of course, while these are implicit requirements, safety and reliability are always the top priority. RD107’s electrical architecture is designed to be concurrently maintainable and fault tolerant. In this Reference Design, we implemented several architecture-level optimization ideas we developed, using the right equipment at the right place to be cost efficient and reach a high reliability level.

Standard Design, Yet Adaptable to Unique Requirements

Of course, there is never a single answer and any design needs to be adapted to each case’s unique requirements. Let’s talk about some of the optimal design choices we made on the low voltage (LV) side:

For the power train redundancy, we choose to use a distributed-redundant system instead of the more traditional “2N”, as it is not really cost effective for larger data centers. And instead of using a block redundant architecture which requires Static Transfer Switch downstream of the UPS, distributed redundancy is the more efficient design. But this can lead to having to shut down one path to the server racks in case of maintenance operation. This would not be suitable especially for retail colocation. In order to minimize this “redundant path” downtime we added an external UPS maintenance bypass for performing UPS maintenance without downtime and we added a backup transformer for the overall site to keep the same redundancy level even during maintenance operation of one MV/LV transformer or MV switchgear.

For the power train size, we chose to create a smaller block size made up of one single UPS and 1600kVA transformer, because it allows us to:

• simplify the LV architecture (single UPS)
• install the main LV equipment closer to the load to reduce wiring
• reduce the short-circuit level even at the IT rack distribution level
• to put the overall power train (MV switchgear, MV/LV transformer, LV switchboard and UPS) on one single skid which can be very helpful in reducing installation and onsite commissioning time.

Of course, using small transformers obliges to use more MV equipment as there are more transformers to supply, but in the design, we have also made some optimization on the MV distribution as well. Again, there is never one single right answer, but the optimum power train size is a balance between the cost of the MV distribution and the cost of the LV distribution up to the rack.

Your Resource for Planning an Optimized Data Center

The Reference Design RD107 is one good answer to address the needs of large colocation data centers and a way for our expert team to share with you optimization solutions – contributing  to a reliable, efficient and cost-effective data center. Explore Reference Design RD107 for an overview, or email ReferenceDesigns@schneider-electric.com to receive the Engineering Package which contains the electrical one-lines, piping diagrams and layouts.

The post Schneider Electric Power Engineers Introduce Highly Optimized 11,2MW Reference Design for Colocation Application appeared first on Schneider Electric Blog.