Technique-based commissioning testing to optimize vs. oversize mission-critical system design
This article discusses some best practices for commissioning electric and thermal systems in a facility, such as a data center, in addition to introducing test data that suggests certain methods are more appropriate when optimizing the performance of a facility. These methods present an important new opportunity for engineers, contractors and commissioning agents, as proper testing and commissioning of data centers can lead to improved efficiency, and optimizing thermal performance of the facility. The need to test a facility, like any other complex system, is generally accepted and well supported for many reasons, including:
- Predictability of life
- Validate capital expenditure needs
- Lower operating expense
- Improved availability and reliability
Technique-Based Commissioning Testing
As the practice of commissioning mission-critical facilities has become more robust, a wide variety of approaches to commissioning the electric power system and power distribution, as well as HVAC within such a facility have emerged. Yet, there remains a vital need to engage in technique-based versus quantity-based commissioning and testing. A technique-based approach means that more precise simulation of the electrical performance and thermal performance of the equipment in the facility leads to an opportunity to optimize the facilities’ design for performance.
Through the work of ComRent and Engendren, providers of load banks and data center expansion solutions, respectively, a methodology for commissioning has been studied and suggested for mission-critical facilities. This new methodology supports a testing approach that permits optimizing versus oversizing facility designs and operations.
Data Center Containment Systems
For those unfamiliar with data center thermal management containment systems, the following paragraphs are meant to be a summary of the main approaches. This will form the basis for a discussion on testing practices.
In-Row Cooling Configuration
In-row cooling offers capacity and efficiency gains by moving the air conditioner from the perimeter of the room closer to the actual load. Installed in an array of industry-available options, in-row cooling arrangemenents provide local, focused cooling at the rows of server cabinets, which fill the data center. As implied, the cooling mechanism is generally situated between the equipment racks as they are arranged in single or multiple rows.
Overhead Cooling Configuration
Similarly, an overhead cooling arrangement, as implied, comprises a cooling mechanism situated over the racks and/or rack rows. Here again the strategy is to manage the heat loads at the source – an approach that is well published and proven to yield higher efficiencies.
Hot Aisle Containment Scheme
Most IT equipment flows cooling air from front to rear. The conventional approach is to align equipment racks in a side-to-side "row" arrangement – forming a physical separation between the cooler intake air on the fore side (cold aisle) and the warmer exhaust air on the aft side (hot aisle). Hot Aisle Contaiment entails capturing, containing and managing the "hot aisle" air before it migrates to other areas of the data center. This zone controlled thermal management strategy results in heightened HVAC system efficiencies and therefore reduced cap-ex and op-ex costs.
Load Bank Testing in Data Centers
Bulk Room Testing
Many data centers are tested and commissioned using bulk-room power and thermal load simulation. In these cases, floor mounted load banks, sometimes referred to as Suitcase Load Banks, are dispersed throughout the room to simulate the power and heat generated by the rack-level equipment as specified in the data center design.
This method is generally straight forward due to the plug-and-play nature of suitcase units. Although they can be deployed quite rapidly, the efficiencies gained in set-up time are outweighed by the deficiencies that result from inaccurate emulation of rack-level thermal loading. Here, disproportionate temperature readings between that of macro-level load simulation and that of the actual and more discrete rack-level equipment may lead to inappropriately sized HVAC equipment and miscalculated total cost of ownership.
Discrete Zone Testing
A preferred alternative to bulk-room data center simulation is discrete zone simulation. In many cases testing can be performed with rack-mounted Load Banks that define a ‘discrete’ zone. Rack-mounted units can more precisely simulate the power and heat load distribution of the actual rack-level equipment as specified in the data center design.
Although the set-up of rack-mounted load banks is not as expedient as suitcase units, the advantages gained through closer simulation of rack-level architectures are without limit. Closer modeling of airflow throughout the rack for instance yields areas where hot or dead zones must be addressed by containment or other HVAC strategies. Actual rack-level equipment can be reconfigured if rack-mounted load bank testing reveals extreme temperature gradients on the intake and/or exhaust sides of the racks. Moreover, closer emulation of the actual rack architecure allows HVAC equipment to be right-sized versus under- or over-sized.
The Engendren / ComRent Evaluation
Although discrete zone simulation testing was generally considered to be a more accurate method for commissioning, there was little data available to support this. In order to capture this data, Engendren and ComRent configured a two-rack, single row test cell of 4-post open face racks. A containment zone was established with a footprint of 5′ Width x 9′ Length x 8′ Height.
Thermal management was by in-row cooling per Engendren’s Thermodul technology. The variable-speed DC fans incorporated in Engendren’s Thermodule were controlled by the cold aisle temperature; increasing and decreasing as a function of high and low cold aisle temperature readings, respectively.
Rack loads were simulated with various load bank arrangements, including: a) a single suitcase load bank positioned in the hot aisle, b) two, 9u rack-mounted load banks as located in the racks and c) four, 5u load banks as located in the racks.
Temperature and air flow readings in the hot and cold aisles were recorded from floor to ceiling in accordance with the various load bank arrangements above. Power draw was recorded for the auxiliary HVAC system as it reached equilibrium for each load bank arrangement.
This graph illustrates the equivalent of the "occupant temperature" in a room to that of a "cold aisle temperature" in a data center space. Cold aisle temperatures, like creature comfort temperatures, set the design basis for the HVAC industry (see ASHRAE TC 9.9, 2011 Thermal Guidelines for Data Processing Environments – Expanded Data Center Classes and Usage Guidance : http://ecoinfo.cnrs.fr/IMG/pdf/ashrae_2011_thermal_guidelines_data_center.pdf ). Maldistributed temperatures within the room’s occupancy space will inevitably lead the HVAC designer to up-size the equipment and/or the occupant to adjust the thermostat to compensate for the "hot pockets". The first leads to excessive capital equipment investments whereas the later leads to either higher capital and/or operating costs.
Similarly, this graph demonstrates that improved emulation of load bank size and configuration (to that of the actual server equipment) leads to more realistically profiled cold aisle temperatures. The results: heightened precision of empirical data that allows for more accurate designing, selecting and operating facilities equipment.
Like the cold aisle temperature, the data center industry references Power Use Efficiency (PUE) as a benchmark efficiency target. Power Use Efficiency is the ratio of Total Consumed Power to Total Consumed IT Power [(HVAC auxiliary load + Load Bank Setting) / Load Bank Setting]. The ideal ratio approaches 1. In this case, for the same reasons as above, the maldistributed thermal loads lead to "false-positive" efficiencies of a data center. In this case, the load banks that more closely simulate that of rack-level equipment, the lower the PUE. Bulk-average, suitcase or floor-mounted load banks cause the auxillary equipment to compensate – where input power must increase thereby resulting in higher and effectively "false" PUE values.
Opportunity for Test Equipment Improvement
This data also shows that there may be opportunity to fine-tune the designs of load banks using these testing scenarios. For example, you can see from the thermal image below that the hot aisle air is being pushed into the cold aisle due to the high fan speed. This could be curtailed by more precise resolution of fan speed control.
Commissioning Best Practices
For high-performance building designers who must meet energy consumption targets, installation of integrated and complex systems, and introduction of new technologies and concepts, commissioning in the design phase will help achieve energy performance parameters.
Review of Testing Requirements
Determine which high-performance building standards will require verification through testing and commissioning, which tests will need to be run and which stakeholders in the commissioning process need to be involved. Determine what equipment is required to conduct these tests, how to capture the necessary data, and determine the proper amount of time for reports to be created.
Evaluation of Equipment Requests in Relation to Testing Objectives
Determine if the equipment is the right tool for the job, if it is readily available and cost effective, and if it will arrive on time in proper working order. Determine if the power quality measurement data and reporting are efficient.
Ensure that the equipment is properly maintained and operated in a safe manner.
Development of customized test solutions
Implement a load bank test plan that accurately simulates the electrical and thermal load characteristics of the equipment to be used in the data center. This will result in optimized efficiency and lower maintenance costs.
On Wednesday, April 22, 10:30am PT, at Data Center World, Mark Siira, Director of Technology Strategy, ComRent International; Alan P. Meissner, Chief Technology Officer, Engendren Corporation will present Technique-Based Commissioning Testing to Optimize vs Oversize Mission Critical Facilities, session FAC 5.1. http://www.datacenterworld. com/spring/education/ education-by-track-2015-data- center-world-global/
Mark Siira is a senior member of IEEE and active in IEEE Interconnection and Smart Grid standards development. He is a member of the Standards Coordinating Committee 21, which establishes standards for Grid interconnection and smart grid interoperability, and a member of the UL Standards Technical Panel 1741 (Inverters) and 6141 (Wind farms). Mark is the Director of Technology Strategy for ComRent International, a leader in load testing solutions.
Alan is the Chief Technology Officer at Engendren Corporation, parent company to IEA, LLC (IEA) and Silver Linings Systems, LLC (SLS). Alan spent 13 years with Modine Mfg in various research positions prior to joining Engendren. His efforts at Modine brought realization to the first on-board hydrogen system for Daimler’s Fuel Cell Vehicle, BMW’s carbon dioxide air conditioning system and PACCAR’s off-idle technologies. Prior to Modine, Alan worked at Pella Windows where he developed enhanced thermal resistive media for large architectural windows. Alan holds 16 US patents.