I hope you have read about my previous article on the necessity of data center energy efficiency and the factors that affect efficiency. The second thing to be considered is the power system efficiency, various methods to be followed and how can we achieve this.

The entire world is aware of environmental issues like global warming, melting glaciers, rising sea levels, ozone depletion, etc. due to technological advancements without any concern about its adverse impact on the environment. However, lately, everyone is awakened towards its disastrous impacts now and therefore effective measures are being taken to minimize carbon emissions and to make the earth a green place. Data center providers host multiple servers, cooling units, power supply back up, etc. causing too much heat emissions and radiations. Therefore, there should be environment-friendly green data centers that use renewable energy resources and other alternative sources of power to make the server farms energy-efficient.

The power system is considered the most critical element of the data center. Effectively it is a chain of key electrical sub-systems and components that are the essential elements for delivering uninterrupted, conditioned power to the IT equipment. The typical power chain(power train) consists of a Utility Transformer, Automatic Transfer Switch, Back-Up Generator, Distribution Switch Gear, Uninterruptable Power Supply (UPS) and the downstream Power Distribution System going to the IT equipment cabinets. This is a simplified list. Like a break in a link in a real chain, the failure of any component in the power chain could cause the loss of power to the IT equipment. In order to allow for continuous availability, there must be redundant power system components, as well as power paths, to prevent downtime in the event of any equipment failure. Moreover, there are also bypass systems to allow for proper maintenance of equipment (known as concurrent maintainability). This redundancy effectively means that there are typically twice as many power components in Tier 3 or 4 data centers. Innovation in the power train is very much rooted in control and optimization, providing more real-time and software-based interfaces that open up new possibilities for data center operations. This allows closer alignment between demand and provision to give greater visibility of system performance so you can react to the issues that will inevitably arise before they happen!

The relentless pursuit of new cooling methods, IT equipment and DC facility devices has led to great improvements in data center energy efficiency. Hence one of the first primary things to be considered is to upgrade to new technological advancements. Here are a few things you can do today in your data center that was not possible just a few years ago,

1. Circuit breakers just got clever

Some circuit breakers now have native metering capabilities for power quality and consumption measurement, and a network interface supporting many communications protocols to allow you to remotely manage loads at a much more granular level with the ability to query and control the breaker from your smartphone. For example, ABB’s EMAX2 offers “Ekip” functionality which lets the user decide which loads are important when the system goes to emergency power. It’s a breaker Jim, but not as we know it!

2. Switchgear is digital

In the past, any significant change in load characteristics for site upgrades would require a physical replacement of the sensor and control elements in the switchgear to satisfy the new specifications. This is no longer the case. It’s all digital now. Just log in and modify the parameters. Couple this with IEC61850 communications across the powertrain and you reduce your physical connections requirement by up to 80 percent. Digital also reduces project execution time, the physical size of the equipment and makes the installation more reliable and safe. With Digital you can make configuration changes with no downtime, no human access, and no cost. 

3. Control from a smartphone

It’s now possible to control some infrastructure functions from a smartphone app. For instance, ABB’s Drivetune app can start and tune ABB drives. 

4. New UPS approaches

You know the size of a UPS room in a data center and the importance of this. Due to this, the innovations and changes in this mechanism are much higher than any other components involved in power train structure.

While each of the components in the power chain incurs a relatively small loss, the overall efficiency of the power system is primarily affected by the efficiency of the Uninterruptible Power System (UPS) as well as the overall smaller downstream power distribution losses. In the case of redundant power, its greatest impact is on the energy efficiency of the UPS systems. Virtually all type of UPS, and in particular the “double conversion” online UPS system (which is the most commonly used type of UPS used in a data center), will be less efficient when operated at very low load levels (i.e. under 30% of rated load capacity). Nonetheless, this is the normal and common operational range for most 2N or 2(N+1) power systems which are the heart of tier III and tier IV data centers. It is not uncommon to see older UPS units in a redundant configuration operating at only 65 – 75% energy efficiency. This severely impacts older data centers especially those with older UPS systems. It is not uncommon to see older UPS units in a highly redundant configuration operating at only 15-25% of their rated capacity. Their efficiency at these lower loads may be as low as 65 to 75 percent. Moreover, since the 25 to 35percent of the lost energy is directly converted to heat, this requires that an additional amount of cooling system energy must be used to cool the UPS waste heat.

The U.S. EPA now also has an Energy Star Program for UPS systems. This was just finalized and goes into effect in August 2012. New UPS units are more efficient overall (typically 92-95% at full load), but more importantly, are far more efficient at the lower load ranges. In comparison, a new UPS can still operate at 85-90% efficiency even when operating at only15-25% of its rated capacity.

In addition, a feature of the newest generation of UPS systems is the so-called “hybrid” or “eco” mode of operation. This permits the industry standard double-conversion UPS to operate in an internal bypass mode, which allows it to operate at 98 to 99 percent efficiency, even at very low loads. The UPS will still automatically revert nearly instantaneously (typically less than 4 milliseconds) to full double-conversion mode if there are any power anomalies, to ensure that the computing equipment will not be impacted. Not everyone in the data center community is completely comfortable with this mode of operation. However, it should be noted that while this feature will be seen on most of the newest units, it does not have to be operated in that mode, the UPS can be configured to operate in the more conservative full double-conversion mode at all times.

New UPS units are also now being offered without transformers (which were required for older technology UPS systems), to improve efficiency. Besides eliminating transformers from the UPS, the use of new higher efficiency transformers in the downstream distribution can also improve the overall power chain efficiency.

5. Measure to Control

An old maxim of operational efficiency is “You can’t control what you can’t measure”. Efforts to reduce power inefficiencies need to begin with baseline measurements. Because utility bills don’t tell you where your power is going, you need a way to measure power consumption in your data center. We measure power consumption for each of these components. There are various areas where you can measure power usage such as IT systems, UPS, Chillers, Lighting, etc. How much of the data center’s power budget goes to IT systems? How much goes to support systems? For every kilowatt-hour of power being fed to IT systems, how much real IT output do you get? The answers to these questions provide a picture of how much power is consumed for every unit of data-center productivity. Measure your data center PUE values and identify the exact areas where can you improve against the total power drawn against the IT system usage. So make sure to have a real-time inspection on the power usages at various stages and compare this if the results are as expected or power is overdrawn by devices. 

6. Renewable energy programs

Data centers are utilizing a huge amount of power when it’s compared to many other industries. One of the methods that can be adopted is to utilize renewable energy programs such as the utilization of solar energy. Many of the data centers are setting green energy goals but their ambitions vary widely in how they plan to reach their targets. Some are in a position to take a long-term industrial approach and provide capital and resources to build new off-site generation. Others are looking to utilize onsite roof space for incremental solar generation, while again others are looking primarily for long term price certainty and opt for using PPAs from renewable power plants.

Facebook, Microsoft, and Google are a few of the technology companies that have committed to go ‘100% renewable’ through the RE100. Here are some of the articles if you would like to know different initiatives from Google, Microsoft, Facebook and many more can be explored on your searches too.

6. Adopt power distribution at 208V/230V

To satisfy global markets, virtually all IT equipment is rated to work with input power voltages ranging from 100V to 240V AC. The higher the voltage, the more efficiently the unit operates; however, most equipment is run off of lower-voltage power, sacrificing efficiency for tradition.

Just by using the right power cord, you could save money. An HP (Hewlett-Packard) ProLiant DL380 Generation 5 server, for example, operates at 82-percent efficiency at 120V, 84-percent efficiency at 208V, and 85-percent at 230V. You could gain that incremental advantage just by changing the input power (and the power distribution unit in the rack).

What about power distribution in the data center? Typically, the UPS operates at 480V, and a power distribution unit (PDU) steps down that power from 480V to 208V or 120V. If you could eliminate that step-down transformer in the PDU by distributing power at 400/230V and operating IT equipment at higher voltages (using technology available today), the power chain would be more efficient. Distributing power at 400/230V can be 3-percent more efficient in voltage transformation and 2-percent more efficient in the power supply in the IT equipment. This slight increase in efficiency is still worthwhile; a data center with 1,000 servers could save $40,000.

7. Utilize the Auto-Cut Lights and Devices Hibernate/Sleep features

Data centers use many of the electric lights and devices integrating for multiple functionalities of the data center. Do you really think that all of these have to be actively powered on to achieve its real purpose? In many situations, the answer would be NO. For example, electric light on your UPS, battery and switchgear rooms and even white space areas may not be required to light every time even when no one is available. You can make use of some auto-cut features and sensors to meet its functionality of power when people are working around. Same way, the sleep and power-saving modes on small devices can be activated to save energy. Think of this, when you are utilizing any of these features can save a lot since this will be applicable to many of the electric devices.

8. Reduce Electrical losses

In an ideal electrical circuit, all the power applied to the input terminals would reach the critical load with no energy wasted or dissipated in the wiring or components along the power path. In real circuits, these components always have some resistance, however small. This occurs with both AC and DC supplies, causing electrical losses which are dissipated as heat. When we talk about a single point or circuits this loss can be less, however, when we are considering multiple areas of entire power infrastructure the total loss could be huge which is known as I2R losses. Hence it’s really necessary for us to reduce this loss to the least possible and one of the best techniques to determine this loss is thermal scanning(thermography or infrared scan). We are able to get this due to the heat build-up of any electrical element under stress and predict that it will fail while it is still functional and appears normal to the naked eye or any other test equipment. The heat signature identified with the use of an Infrared camera and the pictures are then analyzed and put into an easy-to-follow report so that they can rectify before a breakdown occurs, preventing loss of production due to unplanned downtime. With the use of this technique, we can even identify various faults such as hot spots(inductive heating) that could indicate a possible risk of short circuits, loose/deteriorated connections, overloads, unbalanced loads, open circuits,  defective equipment and address them. I would also encourage you to have a look at this video to understand the best of this technique.

9. Optimize Power Distribution

In a data center, you pull power in from the electrical grid and you can convert it down to the voltages that are needed for all the components. There is a lot of electric energy conversion stage there, by minimizing those conversion stages you can save money, energy and also make each conversion stage more efficient. Traditionally one of the biggest losses is UPS systems. Oh really !!! When the UPS systems are really necessary to avoid any electricity interruptions, can we eliminate UPS systems or how can we improve this.

Let me explain to you which are the areas mainly power conversion is happening(AC->DC or DC->AC power). The power from a Utility provider is high-voltage AC(Electricity is always transmitted from one point to another in the form of AC due to the inefficiencies with DC for transmission). The type of power computers need is low-voltage DC. All computer parts (the electronic chips on the motherboard and adapters, the electronics on the drives, and the motors in the hard drive and optical drive) need DC power to operate. So the power from Utility will be converted from AC to DC when storing in UPS systems. When this power is drawn from UPS and needs to be transported, it will be converted from DC to AC power. Finally, when the AC power reaches to IT equipment, this AC will be converted into DC power. Here you can see that there are 3 areas where power conversion happens. How can we reduce this conversion stages? One of the best adopted solutions over here is to have on-board batteries and AC/DC converter within the IT equipment itself. Big Tech companies like Google and Facebook have already achieved this technology in their data center and found that the data centers are more efficient than before.

SUMMARY

It should be noted that the greatest risk of an outage in a data center is primarily from a power systems failure. Unlike broader environmental conditions, which may or may not impact long term reliability of IT equipment, any issue that can potentially disrupt power can cause an immediate outage in a data center. Toward that end, the entire power system must always be designed with a mandate for maximum availability and high levels of fault tolerance. While improved energy efficiency is a worthy goal, it should never be done at the expense of overall reliability.

Have a comment or points to be reviewed? Let us grow together. Feel free to comment.