|Energy Efficiency | Data CenterModel | Junior Staff Development|
Strategies for increasing energy efficiency in HVAC systems in Data Centers
These days, everyoneï¿½s talking about energy efficiency and green IT. But what exactly is behind it all? Well in most cases, it is the attempt to reduce operating costs while maintaining the necessary and required levels of operational reliability. Climate control technology and the corresponding instrumentation and control technology offer particu-larly significant savings potential here.
When we talk about energy efficiency in Data Centers, the first thing that springs to mind is server room HVAC, in other words, optimum airflow management, use of water-cooled racks, hot and cold aisles (with enclosures if necessary) etc. But climate control is not limited to server rooms alone. There are other rooms (the UPS room, for example) which also need to be cooled and which, because they have the same redundancy re-quirements, can be connected to the same cooling system.
Designing the right cooling system
When designing a cooling solution, it always makes sense to aim for a centralized cool-ing system. One which will provide optimum energy efficiency for the entire Data Center, including all ancillary rooms and spaces.
The decision on which system is to be used should be finalised at a very early planning stage of the project. To aid the decision-making process, the different systems must be compared ï¿½ this includes an analysis of investment costs and all operational costs (power consumption, maintenance and service costs etc.).
It is important to consider the cooling system as a whole, before any commitment to in-dividual, highly efficient modules within the overall system is made. Experience from various projects has shown that such an ï¿½individualï¿½ approach can have negative conse-quences for the overall system, which could otherwise have been avoided.
The early planning stages of a Data Center project have a major influence on operating costs. It is here that the crucial groundwork is laid for a truly energy efficient solution.
Factors influencing operating costs
Apart from examining the computer room, we also need to take a closer look at the hy-draulics of the chilled water piping. Here, our aim must be to optimize both the primary (chilled water generation loop) and the secondary circuits (distribution loop).
In addition to the pipe dimensions and pumping capacities, the choice of hydraulic con-trol system needs to be considered. Through parallel operation of all the devices in the cooling circuit (including the redundant ones), it is possible to improve the systemï¿½s effi-ciency. However, this needs to be calculated in advance, as the effect is not always economical.
Other factors influencing operating costs are:
1. The geographic location of the Data Center
2. The location of the computer room within the building
3. Potential short-circuiting and turbulences between intake and exhaust air
4. Suboptimal system temperatures for air and water mediums
5. Stringent ambient humidity requirements
Changing the geographic location is often very difficult as this is dictated by factors such as property ownership, existing Data Centers which have to be adapted, etc. But even here, it is possible to use structural modifications to make suboptimal geographic loca-tions more favourable. The physical location of the computer room within the building should, as far as possible, be free from outside influences such as external walls and windows and warm adjacent rooms. Air short-circuits can be avoided by using the famil-iar enclosed hot aisle/cold aisle rack arrangement.
If temperatures in the water circulation system are too low, the opportunities for free cooling are reduced. Here, there are often significant savings to be made by simply rais-ing temperatures. A possible temperature increase needs to be checked ahead of time, so that operational reliability can be guaranteed. And nominal ambient temperatures are frequently set too high. In such cases, reductions in operational costs are possible by making the appropriate adjustments and by ensuring the highest possible exhaust air temperatures.
Now we have to tackle the question of how to minimize the above influences and how to assess the necessary structural and technical infrastructure modifications in order to facilitate targeted intervention.
Measurement, Monitoring and Control
Thereï¿½s only one rule for improving the efficiency of your facility: measure, measure, measure! Without measuring, it is impossible to control or regulate. And of course, itï¿½s simply not enough to obtain a single set of measurements from the systems and then adjust their settings once and for all time. The measurement data must be analyzed and the system parameters reviewed on a regular basis during day-to-day Data Center opera-tions.
Pro-active energy management with energy data analysis, benchmarking and sophisti-cated reporting procedures will go a long way to making the job easier. But these re-sults must be regularly interpreted and conclusions drawn if the systems are to be oper-ated in an energy optimized and energy efficient way. An action plan can be a valuable aid to implementing these insights right away. Which is why investing in efficient moni-toring systems will always guarantee a rapid return on investment (ROI).
The following rule applies to Data Centers: Thereï¿½s nothing as consistent as change! All Data Centers are continually changing ï¿½ sometimes within very short time spans, and that means that the corresponding technical infrastructure has to be adapted to meet growing and changing IT specification frameworks.
But what should we measure ï¿½ and why?
There are a whole lot of things to be measured and monitored. The diagnostic possibili-ties extend far beyond simple temperature, humidity and power consumption measure-ments.
Apart from measuring, recording and storing consumption and status data, it is also pos-sible to monitor behaviour patterns within the Data Center and all its ancillary and service rooms. Improvements can be achieved by monitoring door positions, for example, as they affect the air-conditioning. In this way, it is possible to monitor operational proc-esses and thus show staff how operational cost-savings can be achieved by adopting different behavioural codes and structures.
If a monitoring system is to be used in a rational and results-oriented way, it is vital that data relating to all types of consumption be collected, i.e. electric power and water as well as other mediums such as gas or oil, where applicable.
Furthermore, all the status data outlined above must be recorded. This includes:
1. Water volume
Necessary for the adjustment and fine-tuning of the systems and hydraulics. Cor-responding control features within the piping system and on the pumps must be provided.
2. Air volume
Also required for adjustment and fine-tuning. Here too, mechanisms for adjust-ing the air volume in the air duct network and on the fan units are essential.
3. Water system temperatures
Temperatures are recorded for evaluation purposes, so that the nominal tem-peratures can be raised, if required. They also enable conclusions to be drawn with regard to the utilization levels and effectiveness of the cooling supply.
4. Cold energy meters
Measurement of actual cold energy consumption in individual sectors for monitor-ing system utilization and, where necessary, billing of Data Center subtenants.
5. Air temperatures
These measurements provide data which can be evaluated in many different ways. They offer considerable opportunities for adjustment and energy efficient control across the board. The following temperature data should be collected:
a. External temperatures
b. Exhaust air temperature from the computer room
c. Computer room intake air temperature
d. Air temperatures in all racks
e. Ambient temperature
6. Humidity readings
Here it is particularly important to ensure that the prescribed ambient humidity spectrum is not too narrowly defined. More sensitive IT equipment requiring a narrower control band could, for example, be placed in a separate computer room with more precise humidity control. This cuts down on excessive operating ex-penses for cost-intensive humidification and dehumidification for the remaining IT equipment.
7. Power consumption
Monitoring of power consumption for both, the entire infrastructure and the IT equipment. The relationship between infrastructure and IT equipment must be in-dividually reflected in the measurement data. In this way, values such as PUE (Power Usage Effectiveness), DEC (Data Center Efficiency) or DI-EER (Site In-frastructure ï¿½ Energy Efficiency Ratio) are continually recorded, to provide a pic-ture of the Data Centerï¿½s overall efficiency. One way of obtaining power data from the racks is to deploy rack-mount Smart Power Strips. Care should be taken to measure all system components individually, for better analytical results. If power consumptions are recorded individually and represented as load curves, the load curves can then be overlaid to reveal reciprocal influences between individual components.
8. Monitoring door positions (or window positions, where relevant), room lighting (on or off), number of persons present in the room etc., facilitates better analysis of the gathered data.
When choosing control features, care must be taken that they do not use increased en-ergy input to throttle down the systems, as this would result in increased pressure losses.
One of the prime objectives of monitoring is to detect load peaks, which enables correc-tive action to be taken to counteract these peaks. High load requirements at high exter-nal temperatures, for example, can be cushioned by providing additional storage tanks or by reducing the external air temperature before it hits the cooling systemï¿½s roof-mounted heat exchanger. This can be achieved by using a hybrid chiller plant, for ex-ample. The cost of wetting the cooling surface is less than the savings gained from the resulting improved efficiency levels of the heat exchanger.
Logging the external temperature is particularly important for the subsequent evaluation and appraisal of a systemï¿½s efficiency. Such diagnostic procedures can explain vari-ances in system performance, as considerable deviations become manifest, particularly when free cooling is utilized.
Decisions on measurement methods and parameters must be taken before a monitoring system application is planned. Monitoring allows complex processes to be simplified and represented in a uniform way. Rapid access to all the relevant data means that all necessary adjustments can be made faster, more flexibly and more productively. => Improved Performance
Improved efficiency is also achieved because the monitoring system can serve all a buildingï¿½s technical systems. For this reason, it is advisable to take a holistic view of the building and all its different systems and spaces when designing a monitoring system.
Further strategies for improving efficiency
One of the currently more popular methods of performing a Data Center cooling audit is computer-simulated airflow modelling. The technique employs software tools (CFD software) to provide graphic simulations which allow potential hot-spots to be detected. It also enables you to examine, in advance, the effects of different settings configura-tions in technical systems. But it must be said that this procedure is quite complex and thus cost-intensive. And the system cannot be used without specialist skills ï¿½ it is advis-able to call in an expert, well-versed in its operation.
Thermal imaging is a simpler and more cost-effective method of graphically recording thermal conditions in existing Data Centers. Thermal imaging cameras can either be purchased or leased and learning to work with them is easy. The images they produce show up weak spots in the HVAC system, and allow them to be analyzed.
We can conclude that is only possible to achieve an optimum improvement in the energy efficiency of a Data Center when all types of data are both measured and recorded. Without the evaluation of this data, any adjustments to technical systems will only be piecemeal measures and when only a few, individual areas are scrutinized, it can even have negative effects in other areas. Which is why it is always necessary and sensible to take a holistic viewpoint.
Furthermore, there are significant savings to be made by checking, reassessing and if necessary, correcting boundary values.
Another important factor is the constant monitoring of energy consumption and continual comparison with the framework parameters (all internal and external temperatures). An intelligent monitoring system is ideally suited to this.
It is advisable to communicate all analysis information to personnel having day-to-day contact with the Data Center, so that they may gain a better understanding of the techni-cal systems and influential factors. In this way, it is possible to avoid errors caused by the lack of knowledge and ingrained behavioural patterns of those dealing with the Data Center.