Since I’m on the subject of facility, I might as well cover a few more points on data center facility, measurement and sensors. When I think about sensors, it reminded me of my first job. I used to be with a Systems Integrator, working on Plant Information Management Systems (PIMS) for refineries, gas plants, utilities, etc. I remember once, I was 120km west of Doha, Qatar, and was commissioning the PIMS system together with my customer and main contractor. It was August, in the mid-90’s. We were going through the screens and data points / tags with values collected from field instruments in the plant via their Distributed Control System (DCS).
As we review the data points, we came across a few thermocouples out in the plant indicating measurements of 54′C to 60′C. Although it was a hot summer day, a reasonable reading for those tags should be in the range of 45′C to 50′C. It was an interesting anomaly…
For a gas plant, it is very critical that the plant data collected are accurate as the data are used for planning purposes, production controls, yield accounting, and many other forms of analysis.
For a data center, data accuracy is also equally important.
It is important because the impact of inaccurate data can have an effect on energy consumption. For instance, if, for whatever reasons, our temperature readings of chilled water is inaccurately higher, would lead us to decrease CW temperature and hence uses more energy to maintain that low temperature. There are a few contributors to bad data, namely, inaccurate sensors, need for calibration, inappropriately placed or installed sensors, faulty sensors, etc.
What do we need to measure in a data center facility? Typically, these would include:
- Temperature, flow rate, current, voltage, power draw of Compressors in the Chiller sub-system;
- Flow rate, pressure drop, current, voltage, power draw of Pumps;
- Current, voltage, power draw of blowers of Cooling Towers;
- Current, voltage, power draw of Humidifier and Reheaters of CRAC;
- Contaminants / pollutant levels of the air in the facility;
- Ambient temperature, RH across the facility to provide representative profile of actual conditions;
- Current and voltage of UPSs and Transformers of the critical power path;
- Current, voltage, power draw of IT equipment.
Effective measurement and monitoring that will avoid inaccurate readings or representations of the data center facility must consider the following:
- Ensure appropriate positioning of sensors. For example, if a temperature sensor is placed inside the data center facility, on the ceiling, above a perforated tile, then it may provide a misleading indication that the facility conditions are appropriate. Placing multiple sensors inside a facility, but having the sensors situated close to each other, is also not ideal.
- Avoid using one single point of reference to infer to the facility’s profile.
- Use appropriate type of sensors or measuring device to attain the required accuracy for the specific application, for example, strain gauge over bourdon tube for more accurate pressure measurement, and different types of liquid flow sensors are designed for different flow range, density and viscosity of the fluid, and provide different levels of accuracy.
- Data collected should be stored and used for historical trending and analysis. Data gleaned from such analysis can be quite useful to determine seasonal changes and external influences, and their impact on your data center facility efficiency.
- Alarm capability in a monitoring system, if available, is useful too. At the very minimum, the system should offer an avenue to set threshold and notification alert to your operations staff should conditions move outside of that preset threshold.
- Service maintenance and periodic re-calibration of sensors should be considered and scheduled.
Tags: Data Center, Efficiency, electrical, Metric, Operation, Operations, RH, Temperature