Economics of energy-efficient electrical systems

Energy efficiency is achievable in all building types, including mission critical environments.

By Steven L. Orscheln, PE, LEED AP and Juan Cardona, LEED AP December 18, 2014

Learning objectives

  • Understand the types of electrical products and systems that can lose energy.
  • Learn how to measure energy loss.
  • Understand how to select and size equipment for the application.

Today’s production, distribution, and consumption of electrical power have become a study in economics. With the ever-increasing efficiencies of electrical system products, energy losses from these products have decreased. This efficiency has a secondary benefit of reducing the cooling required due to the associated parasitic heat load. Energy efficiencies are achievable in all types of facilities such as health care and corporate buildings, but are most apparent in systems and products within data center electrical systems due to the power intensity of this industry. This includes power capacity products such as uninterruptible power supply (UPS) systems and power distribution unit (PDU) transformers, as well as cooling control products for compressor, pump, and fan components of HVAC equipment, such as computer room air conditioning (CRAC) units.

The cost of electrical power as measured in kilowatt-hour (kWh) units of energy, or the amount of energy delivered in 1 hour at a 1 kW power level, is the cost of overcoming the following losses and inefficiencies in delivering the required energy to power and cool a facility:

  • Losses due to power conversion from ac-ac, ac-dc, and dc-ac
  • Varied efficiencies due to technology differences of power conversion products
  • Varied efficiencies of power distribution transformers
  • Losses caused by inefficiencies due to lightly loaded equipment
  • Losses caused by inefficient sizing of power distribution equipment
  • Capital cost inefficiencies due to distribution of power at lower voltages rather than higher voltages.

Currently these energy loss and inefficiency factors may be overlooked on projects outside of the data center industry, but they will work on any project and efforts should be made to use them. Green industry guidelines such as the U.S. Green Building Council’s LEED program have indicated a desire to allow these as innovation credits if specific savings can be quantified. The reality is that modern-day projects do have data-intensive areas that require special consideration; therefore, consideration of these factors is warranted.

Power capacity products

The first example for improving energy efficiency in data centers is uninterruptible power supply (UPS) systems. UPSs are electrical apparatus that provide uninterrupted power to a load when the input power source fails. The first UPS systems used to support legacy data centers were very reliable, but were inefficient rotary-type UPS systems. They supported mostly inductive type mainframes having low power factors, lower than 80%. As computer system technology advanced, so did UPS technology.

The next generation of UPS was a double conversion, static type with battery backup. These UPS systems comprised both input and output transformers, and used silicon controlled rectifiers (SCRs) for high-power switching. They typically operated at an 80% power factor. Early static UPS systems tended to be less reliable than rotary, but because they were also less expensive than rotary technology, many UPS system designs included paralleled modules to increase reliability. This helped the mean time between failure (MTBF) rating of the static UPS come closer to that of the rotary, but would sacrifice efficiency because of the resulting low load conditions realized on each module of a paralleled system.

In a continued effort to increase reliability as well as energy efficiency, the UPS industry steered from the use of SCR transformer-based technology to transistor-based transformerless technology. Transistors ease of on/off switching as compared to SCRs resulted in a focus on the development of transistors with high switching speeds and high power-handling capabilities. Initially, at comparable power-handling capabilities to SCRs, this focused development resulted in the use of the isolated gate bipolar transistor (IGBT). IGBT UPS systems have very high full-load efficiencies, but their notoriety came in the higher efficiencies at part load.

As an example, the efficiency curve of a 500 kVA transformerless based design over a 500 kVA transformer design is on the order of 97% to 92%, respectively, at 40% load. The annual energy cost represented by this comparison at $0.10/kWh corresponds to $16,000 to $44,000, respectively, for an annual energy savings of $28,000 for the transformerless design.

A second example of power capacity products improving efficiency are PDUs. They are an electrical apparatus that consists of a voltage changing transformer and a means by which to distribute conditioned and continuous power from a UPS to the end-use computer loads of a data center. The PDU transformers are the same as a typical stand-alone transformer used in most facilities. They are energy-efficient TP-1 compliant transformers that meet or exceed the requirements of the Energy Policy Act of 2005.

The TP-1 standard dictates that power distribution transformers be more efficient at their typical loading level of 30% to 50%. The higher efficiency transformers required by the energy policy are 2% to 3% more efficient overall as compared to pre-TP-1 lower efficiency transformers. The improved efficiencies of power distribution transformers improve on the overall losses to the electrical system.

Standard pre-TP-1 transformers convert approximately 95% of the electricity received into usable output voltage. With a typical transformer being energized continuously, the 2% to 3% improvement in efficiency can provide significant energy savings. The TP-1 transformers use higher grade electrical steel to lower flux density and reduce losses, especially at the average 35% loading where TP-1 efficiency measurements apply. Beyond TP-1, TP-1S transformers are available to surpass the TP-1 standards to better accommodate nonlinear load profiles at K13 and K20 levels.

For reference and comparison purposes, the efficiency of a pre-TP-1 150 kVA transformer is 96.5%, as compared with 98.3% for TP-1 and 98.8% for TP-1S transformers. The annual energy cost represented by this comparison for a transformer loaded to 65%, at $0.10/kWh, corresponds to $9,000 for pre-TP-1 and $2,000 for TP-1S, for an annual energy savings of $7,000.

Cooling control products

Cooling systems comprise cooling and air movement equipment that consists of individual compressors, pumps, fans, and their respective controls. Though the arrangement of compressors, pumps, and fans can take any number of forms in cooling system equipment, one form that is typically found alongside the data center UPS and PDU power equipment previously mentioned are computer room air conditioning (CRAC) units. CRAC units are self-contained assemblies capable of maintaining precise airflow, temperature, and humidity. Through the years their basic framework has remained essentially unchanged, but technology advances of their components have resulted in more precise and energy-efficient control of computer room cooling air. CRAC unit development has progressed from the use of semi-hermetic to scroll to digital scroll compressors, and from the use of standard centrifugal fans to variable speed, electronically commutated (EC) plug fans.

Early compressors were single-stage semi-hermetic that were either all on or all off. To accommodate varying load conditions, multiple single-stage compressors were employed that would each handle part load. Developing technology resulted in a choice of multiple stage semi-hermetic compressors, scroll compressors, and the most energy-efficient variable stage digital scroll compressors that operate seamlessly from 10% to 100%.

Single-speed motors have been traditionally used to drive the CRAC unit centrifugal fans, but developing technology has provided a choice for variable frequency drive (VFD) control on the motor used to drive the fans, matching the motor speed to the room cooling requirements. This allows the unit to use less motor energy to move room air. The VFD is controlled to match the speed of the fan with the compressor load . This eliminates excessive energy use due to an oversized design.

Alternatively, the use of EC plug fans is the most energy-efficient option. They are inherently more efficient than centrifugal fans with VFDs due to their direct drive design versus the belt drive of a centrifugal fan that adds to its losses.
As an example of energy savings using a VFD, the energy cost (at $0.10/kWh) for an 85% efficient 10 hp motor operating 100% of the time corresponds to $8,000 without a VFD and $4,000 with a VFD, for an annual energy savings of $4,000.

Lighting

Lighting design improvements are another way to increase overall energy efficiencies for facilities. Many facilities are still using fluorescent lighting. With the drop in cost and advances in technology, semiconductor-powered lighting (LEDs) has become a feasible alternative. LEDs typically mimic the spectrum of daylight better than fluorescents. Their life span is 50,000 to 100,000 or more hours, compared to 20,000 hours for fluorescent. LEDs are normally standard dimmable, allowing for even longer lamp life and greater energy savings when full levels of light are not needed. The heat generated by the LEDs is much lower than the heat produced by fluorescents. Table 1 represents a sample comparison of LED versus CFL lamp types.

For example, a typical data center has on/off switching where often the lighting will remain on even after personnel have completed their tasks within the data center. However, occupancy-based lighting can be implemented within data centers for considerable energy savings. Occupancy sensors can ensure lights are on only where personnel need to work. If the design intent is to maintain the lighting on, then the lights outside the zone can be dimmed to an acceptable level to reduce energy consumption.

Electrical distribution equipment sizing

Another area of improvement is electrical equipment that is oversized. Oversized equipment has many losses due to the inefficiencies resulting from lightly loaded equipment. Design loads are typically based on nameplate plus future growth. In reality, loads operate in the range of 25% to 50% of nameplate. When designing, it is common practice to use the manufacturer’s nameplate information as the basis of design for calculated loads. The nameplate includes the following information: rated input voltage, phases, rated input amperage, Watts or kilowatts (kW), sometimes Volt-amperes or kilovolt-amperes (kVA), and frequency information. The nameplate information that has been provided by the manufacturer is used to provide data needed for the preparation of the electrical installation (circuit breaker, wire, receptacle, etc.) and to assure the user that the equipment is suitable for installation at the specific location within the facility.

As an example, many computer server manufacturers bulk-produce different size power supplies and use the same nameplate for the smaller units, which results in a drop in actual energy consumption. Furthermore, the nameplate is the maximum rating for a fully loaded server. These servers are rarely configured to their maximum nor do they run at 100% capacity. A better design practice would be to use the peak measured load or the maximum manufacturer’s load draw reported for the specific equipment to be provided. Many manufacturers have power estimation tools, including Dell, HP, IBM, and Cisco, among others. The power load should be reviewed with the authority having jurisdiction (AHJ) to ensure it meets all applicable local codes and standards. If the measured load can be provided by the end user or determined otherwise, along with the power estimation tools available from some manufacturers, the electrical equipment can be sized more accurately, thereby reducing losses in the system—that is, using higher voltage data equipment (400 V) to preclude the use of additional transformation and thereby the associated losses.

The electrical industry is demanding highly efficient building design. Savings realized for right sizing, dc versus ac, and varying design prototypes and efficiencies for the same product type results in a list of options for economic review.  Less copper, smaller transformers, reduced UPS and PDU size, and effective use all factor in to first cost of buildings.

Sustainability of these considerations is significant if cradle-to-grave carbon calculations are prepared. While this may seem daunting, the availability of this information to make these calculations is increasing.


Steven L. Orscheln is a principal and electrical engineer at ccrd partners. Juan C. Cardona is an associate at ccrd partners. Both are power systems design engineers for data center, corporate, and health care projects.