Data Center Design Consideration: Redundancy

So far we have discussed about the Cooling and Space Planning issues for data center design. We all note that data centers are very special purpose-built facilities providing critical and uninterrupted services. In this section we are going to explore one of the key considerations - Redundancy.

Redundancy entails providing system beyond the minimum capacity to ensure these systems continue to operate even if part of the system fails. Under the TIA-942 Standards, data centers are classified into different tiers based on their availability and redundant designs. What exactly are these designs?


For more information of Tier Levels: N, N+1, 2N, 2(N+1), please visit our previous articles:
http://green-data.blogspot.com/2014/07/data-center-tier-levels-and-uptime.html
http://green-data.blogspot.com/2014/09/more-about-data-center-tier-levels.html


N Design

An N design means the number of components and paths is exactly what is required to meet the data center requirements. There is neither spare capacity nor any standby unit. Basically, there is no redundancy and the system will totally fail when a component or path fails.

N+1 Design

An N+1 design requires the number of components installed to exceed the requirements by one. In this case, when one of the components fails, the standby unit will take over and the system would continue to operate.

2N Design

As the name mentions, a 2N design has 2 independent working systems supporting the data center. In the event a primary system fails, a secondary system would take over and continue the uninterrupted operations.

2(N+1) Design

A 2(N+1) design is an extension of the 2N design. Both the primary and secondary systems are equipped with a single group of (N+1) facilities, e.g., each system has an extra component. This design is more resilient than a 2N design and would be able to withstand a concurrent path and component failure.


It is important to understand that in 2N or 2(N+1) systems, the secondary system should be located away from the primary system which minimizes the primary and secondary system from being damaged by any incidents, such as fire, floods, etc.

The following examples illustrate the design of a UPS (Uninterruptible Power Supply) system supplying 200kVA of power to a data center:

Figure 1: Various redundant design for a UPS system providing 200kVA of power

In an N Design, one UPS of 200kVA is installed. This design would meet the 200kVA requirements of the data center. Normal operation is disrupted when the UPS fail.

An additional UPS is installed in an N+1 Design. This brings the number of UPS system to 2 which allows for normal operation if one UPS fails. E.g., the remaining UPS would still able to support the minimum 200kVA power. However, the system can be still failure if the path (feeder cable) linking the UPS to the equipment is damaged.

A 2N Design offers protection for both path and component failures. This requires 2 UPS systems (Primary and Secondary) each with 1 unit of 200kVA UPS. Should a UPS fails to start or a cable feeder is damaged, the Secondary System would take over and continue to supply the required load.

The 2(N+1) Design is the most resilient of the 4 designs discussed. 2 systems with 2 units of 200kVA UPS each are installed. This design would continue to provide the necessary 200kVA power even if there is a simultaneous component and path failure. It should be noted, however, no design is 100% fail-proof and this design would still fail in the event of multiple failures.

The pictures below show the resiliency of various systems under different failure conditions:

Figure 2: Different redundant design under different failure conditions

The Blogger

Strategic Media Asia (SMA) is one of the CPD Course Providers of the Chartered Institution of Building Services Engineers (CIBSE).

SMA, a critical infrastructure training and event organizer based in Hong Kong, provides an interactive environment and opportunities for members of IDC industry and engineers to exchange professional views and experience on critical infrastructure and E&M facilities.

For details of other data center courses and seminars, please visit our website at http://www.stmedia-asia.com/trainings.html.

Data Center Design: Battery-based / Static UPS or Rotary UPS?

UPSs vary greatly in physical size, weight, form factor (e.g., standalone vs. rack-based), capacity, supported input power source (e.g., single phase vs. 3-phase), technological design, and cost.

When considering the procurement of a UPS for a data center or other mission critical facility, there are a number of design and acquisition decisions to make such as:

1. The size of the load to be protected
2. The battery runtime required
3. The proper input and output voltages
4. The right type of system (i.e. on-line, line-interactive, etc.)
5. Pricing and performance seen within manufacturer product portfolios
6. The advances in technologies
7. The ideal level of redundancy (i.e., N, N+1, 2N, 2N+1, etc.)
8. The required output distribution

This time our focus will be limited to comparing 3-phase battery-based UPS and rotary UPSs that support data centers. Some of the common 3-phase UPS architectures are listed below:


- Double Conversion On-Line UPS
- Delta Conversion On-Line UPS
- Engine-coupled Rotary UPS
- Flywheel UPS


Other Single phase UPS systems (will be discussed later):

• Standby UPS (single-phase load ranges from 100 to 1,500 VA)
• Line Interactive UPS (single-phase load ranges from 0.5 to 10 kVA)
• Standby Ferro UPS (single-phase load ranges from 3 to 15 kVA)

Double Conversion On-Line UPS

This design is by far the most common type for loads above 10 kVA. A double conversion UPS is considered ‘on-line’ because its nearly ideal output waveform is derived completely from battery power through its inverter. As such, this isolates the output from the input. The input primarily serves to continually charge the battery.

Therefore, during an AC power failure, on-line operation results in no transfer time. Since 100% of the load power is converted twice, once from AC to DC (to charge the battery) and a second time from DC to AC (for the output), a double conversion UPS is inherently less efficient than offline UPS types.

These UPSs are available in a broad range of sizes for three-phase loads from 10 to 1,000 kVA. Furthermore, these UPSs can be arranged in a vast array of configurations to achieve high availability for very large three-phase data center loads.

Delta Conversion On-Line UPS

That design is a more recent topology that was originally patented and utilized exclusively by APC for its Silcon-series 3-phase UPS line.

The advantage of this topology is its energy efficiency. It achieves high efficiency by not processing 100% of the power, 100% of the time, as is the case with a double conversion UPS.

Rather, it processes only the portion of the power that is outside of the ‘window’ of an acceptable power waveform, while at the same time having an output that is on-line with load such that there is only a load step change from steady-state to full battery-load operation. As such, this technology is best suited for the highest power applications and well above 1,000 kW.

Engine-coupled Rotary UPS

A rotary UPS is one that unlike the prior approaches does not rely on electronics to pass through or recreate an output AC waveform. Rather, a rotary UPS relies on a mechanical motor and generator to create a pure sine wave output without the need for filter capacitors. Additionally, the resulting low input impedance allows the ability to handle any type of load no matter how ‘dirty’ it may be.

Also, this approach yields an ability to handle high fault currents and provide isolation from high harmonic load input currents. Finally, a rotary UPS can be configured to use direct diesel bypass. By comparison, a static (or battery-based) UPS, can only source its load from a generator in a ‘break before make’ fashion. Due to the nature of its ability to handle ‘difficult’ loads, rotary UPSs are large in size and expensive. As such, they are reserved for large loads in excess of 1,000 kW.

The advantages of Rotary UPS:

1. Most energy efficient UPS/CPS system
2. Use of stored kinetic energy
3. No batteries required, that means no chemical waste!
4. Save energy for battery room's ventilation and cooling 
5. Lowest Total Cost of Ownership (TCO)
6. Highest power factor
7. No conditioned battery room required, that means saving space
8. Long life time
9. Efficiency about 96%

Flywheel UPS

Flywheel UPS system is similar to the settings of battery-based UPS system. The rotary UPS is called “rotary” because rotating components (such as a motor-generator) within the UPS are used to transfer power to the load. The true definition of a rotary UPS is any UPS whose output sine wave is the result of rotating generation. Therefore, the UPS in Figure below, although it utilizes a flywheel as a rotating temporary energy storage source in case the utility fails, is not, by definition, a rotary UPS.

About The Blogger


Strategic Media Asia (SMA, www.stmedia-asia.com) is a leading technical training and event organizer for corporations specialized in data center design & build, E&M facilities, telecom, ICT, finance and colocation. Currently, SMA delivers a series of data center trainings and qualification programs in Hong Kong, Taiwan and Macau.

All these events / training seminars are designed to support the leadership needs of senior executives (Chief Information Officers, IT Directors / Managers, Facilities Managers, company decision makers, etc.) and to provide useful and applicable knowledge.

For detail, please visit our data center courses & training seminars at http://www.stmedia-asia.com/trainings.html.