By Tim Hirschenhofer
From the March/April 2016 Issue
Energy efficiency and conservation are some of the most important topics discussed in data centers throughout the world. In addition to the need to mitigate rising energy prices, data center facility managers know that it simply makes good financial sense to reduce power and operating costs in their facilities wherever and whenever possible. These goals have become even more attractive now that utility companies across the country are offering incentive programs to entice data center owners and operators to reduce their energy consumption. This has led to some of the fastest return on investments (ROIs) for energy efficiency projects in data centers since the industry was founded.
There are a variety of areas ripe for improvement that can help every data center run more efficiently, no matter their size or location, such as: enhancements to cooling systems, better airflow management, deployment of advanced data center infrastructure management (DCIM) systems and software, utilization of proactive monitoring tools, and more streamlined operational procedures for staff. The knowledge and expertise found in firms that work exclusively in data center energy optimization combined with the long list of products designed to decrease power consumption allow managers to realize immediate savings to their operational costs with no data center downtime and minimal impact on staff.
Throughout 2015, Future Resource Engineering conducted a survey of 40 data centers across the United States to see what common improvements could be made to allow for substantial energy savings. The data centers spanned a variety of industries, including: colocation, healthcare, higher education, government and telecommunications. The computer rooms varied from 5,000 square feet to 95,000 square feet, illustrating that energy and operational efficiency are highly sought after goals of data centers regardless of size.
Ultimately, four unifying factors were identified across all sites that we consider best practices employed by data center owners and operators to reduce their energy usage:
- Control cooling resources.
- Monitor cold aisles.
- Understand external environment.
- Evaluate IT load demands.
Control Cooling Resources
Control cooling resources in real-time to provide maximum efficiency, while ensuring environment conditions remain safe for IT equipment.
The vast majority of data centers (including several in this survey) are not providing control of cooling resources to run them more efficiently. They are still typically “flood” cooling the data center—running all required and redundant cooling units at the same time. This practice sometimes provides two or three times the amount of cooling required.
The team worked with a Baltimore area data center that recently upgraded all 16 of its cooling units in one room with variable frequency drives (VFDs). Calculations showed that, to begin with, this data center could have safely run on only nine of the cooling units. With the installation of 16 VFDs, all of the cooling units should have been able to run at around 50% to 60% of maximum fan speed to cool the data center safely. This would have provided an additional benefit of significant energy savings. However, because the data center did not install controls, all cooling units remained running at 100% fan speed. In essence, the staff performed half of the job: they installed a worthwhile efficiency measure but failed to provide adequate controls.
The result was a significant financial investment that did not provide efficiency benefits. Additionally, the local utility provided an incentive but received no reduction in consumption from this facility because proper cooling controls were not installed.
Monitor Cold Aisles
Monitor cold aisles via sensors located in the data center to record temperatures and establish trends as to how cold supply air and hot return air is mixing.
For the facilities in this survey, monitoring cold aisle temperatures is still not being done consistently, at least to the extent that it should be. The survey team was constantly surprised at the number of large colocation and enterprise data centers that are only spot monitoring temperatures in a few locations.
There is starting to be a change in this trend, but typically only for new data center builds. Existing legacy data centers are often neglected when it comes to thorough temperature monitoring. When implemented properly, monitoring cold aisle temperatures allows not only for trending to identify areas for efficiency gains (areas that are being overcooled), but also helps data center managers avoid issues with their customer service level agreements (SLAs) for temperature.
Ideally, cold aisle temperature monitoring should also be integrated with controls to allow for more efficient cooling.
Understand External Environment
Understand the data center’s external environment so economization methods can be employed to lower energy consumption.
In the survey, economization was being implemented much more frequently than in the past. However, this is typically done during equipment refreshes. For example, if a data center is not currently taking advantage of economization and their cooling units are approaching end of life, customers are considering replacing those old units with a system that can take advantage of economization (two of the surveyed data centers are currently working with this goal in mind).
Not only does economization provide reduced operating costs, but many utilities will also incentivize replacing older inefficient equipment with high-efficiency equipment, so capital outlay is reduced.
In many cases existing systems can be upgraded to take advantage of economization without purchasing new equipment. Plate and frame heat exchangers, for example, can often be added to existing chilled water systems in many areas of the country. Utilities will also often incentivize these upgrades. Even when economization is already in place, the cooling systems can often be enhanced with controls to better automate the process and take advantage of more economization hours per year at warmer temperatures.
Evaluate IT Load Demands
Evaluate current and future IT load demands to determine if uninterruptible power supply (UPS) equipment is right-sized and running efficiently.
This type of evaluation can often be difficult for customers to determine. Data centers are typically built with a maximum IT capacity in mind, and UPS systems are often purchased with that limitation in mind. In many cases, this means that the UPSs tend to be underutilized for a significant period of time as the data center fills up. When UPS systems are underutilized they are less efficient, although there are UPS systems that are modular or incrementally enabled and allow for “staged” loading which helps avoid this problem. Another corrective strategy often employed is to build out data centers in stages or zones, bringing new appropriate sized UPSs online as the stages are filled.
Not all of the data centers surveyed used every (or in some cases any) of these best practices, and the cumulative impact on their power consumption and budgets is sobering. The team calculated that in just the 40 sites inspected, there is room for over 24 million kilowatt-hours (kWh) in annual energy savings that these data centers could be recouping through some simple equipment and operational process upgrades. It was calculated that this equates to $2.2 million annually in unnecessary costs.
Identifying efficiency opportunities, overall costs, potential savings, available incentives, equipment strategies, and payback requirements differ widely from one facility to another and can seem overwhelming. Still, considering the opportunities and upgrades available, the cost of beginning an efficiency process is more attractive than ever before. When taken in combination with available utility rebate programs, ROIs are being achieved relatively quickly. In the 40 sites surveyed, the team identified $1.4 million in available utility incentives that these data centers can use to offset retrofit costs, with an average project ROI of 2.8 years. Together with these best practices, the conditions have never been better for data center operators to reduce their power consumption and enjoy a healthier bottom line.
Hirschenhofer is the director of data center engineering for Future Resource Engineering, a data center energy and operational efficiency firm. Based in Fort Collins, CO, Future Resource helps clients implement efficiency plans by using utility rebates and vendor neutral solutions to offset project costs.
Do you have a comment? Share your thoughts in the Comments section below. Or send an e-mail to the Editor at [email protected]