Case Studies
Data Centres & Commercial Facilities
Background
Older Computer Room Air Conditioners/Handlers (CRAC/CRAH) with belt driven fans require belt replacement and tensioning on a periodic basis to avoid slipping.
CRACs typically have program maintenance schedules. The schedules are prescribed by the product manufacturer or the mechanical services companies.
V-belts can start to slip intermittently in between maintenance schedules. This significantly reduces the efficiency of the cooling system and results in higher power consumption and reduced cooling capacity.
In this case study we look at how NUMEN AI is able to detect this common problem and ensure optimal operation of CRAC units.
NUMEN AI Solution
The chart below shows two identical CRAC units. One unit is operating normally (purple line), the other unit with an intermittent slipping belt (blue line).
NUMEN AI was trained to recognise this pattern and notify staff when this condition is detected.
After maintenance was performed, the unit returned to normal operation.
Conclusion
NUMEN AI is capable of learning specific patterns that represent fault conditions in air conditioning equipment such as a slipping fan belt.
Once the algorithm has been trained, NUMEN AI will continuously monitor equipment for recurrence of this pattern in future. Alerts can be issued to a desktop or mobile phone when the equipment moves outside normal operating patterns.
“NUMEN AI gives me peace of mind that all cooling systems are running optimally and that there are no hidden problem areas.”
Grant Meyers, Data Centre Services Engineer, Insurance Australia Group Ltd
Background
Imperial College London (ICL), like most other organisations, is under increasing pressure to reduce their energy consumption and become more sustainable.
In order to achieve these sustainability targets, Prof. Tim Green, Academic Leader for Sustainability, from the Electric and Electronic Engineering department realised that a technology solution based on high-frequency granular data would be required.
Current data acquisition and monitoring systems lack the granularity and data accessibility to effectively support efforts to improve sustainability. Current solutions were not delivering the required results and a different approach was needed.
NUMEN AI was selected for its ability to record granular, high frequency electrical data, analyse the data with machine learning and other AI algorithms, and produce insightful reports that highlight areas for improvement.
NUMEN AI has the added benefit of making the data readily available to academic staff and students who could further analyse the data and develop innovative sustainability solutions.
Installation
As proof of concepts, specific areas of the Electrical Engineering building were selected for the installation of NUMEN AI.
This initial installation covered mains distribution points and some mechanical services such as air conditioning and lifts.
The installation was performed by local electricians from the Imperial College Estates team under the supervision of a NUMEN installation technician.
Outcome
The installation showed that NUMEN can easily be installed by local electricians and does not disrupt the operation of the buildings.
The NUMEN Internet Gateway provides Internet access to all the NUMEN devices. The gateway required zero configuration.
The NUMEN team was able to remotely configure and set up the NUMEN system to start recording data and making data available via the NUMEN Web Console immediately.
The NUMEN system further monitors each NUMEN device continuously to ensure correct operation.
Initial Findings
Low Power Quality
Power quality refers to how ‘clean’ the electricity delivered to electrical equipment is. The graph below shows the three-phase supply power with a very low power quality (22%) vs high quality power of 92%.
An electrical network with a PQ of 22% only utilises 22% of the energy supplied by the energy utility as electricity and the remaining 78% as heat or vibration.
In the case of ICL, the overall PQ of the circuits monitored by NUMEN has been calculated at 74%. This indicates that 26% of electrical energy consumed is generating heat and vibration and not contributing to the normal functioning of the electrical equipment it is supplying.
In addition to excessive consumption, low PQ also reduces equipment life and causes unexpected breakdowns.
Based on the NUMEN calculations above, a significant energy saving is possible once the source of the poor power quality has been corrected.
NUMEN AI pinpointing the problem
Since NUMEN samples electrical current at high frequency, NUMEN is able to pinpoint the source of the reduced PQ to high current on the 5th harmonic (250Hz) of the Sub-panel Board, AC and Lifts.
The 5th harmonic is well known to produce an effect called "torque fight”. In the case of ACs and lifts (or any other electric motor driven device), this harmonic component causes negative torque that reduces the mechanical output torque, while increasing heat buildup. Consequently, more power is consumed and the motor life is significantly reduced.
Hours of operations
Another significant insight by NUMEN is related to the hours of operation. Since this facility is not in use after-hours or over weekends, it is expected that the consumption should reduce significantly.
Initial data has shown that this is the case in many areas, but that in some areas the difference in consumption between weekdays and weekends is very small. In some cases this is expected, as certain areas of the building have continuous loads, in other areas, however, this is of concern.
Background
Metal die casting machines consume lots of electrical energy to keep metal in a molten state. Due to the unbalanced nature of the machines, it could easily consume significantly more power on one or more phases.
In this case, the manufacturing facility has multiple die casting machines and careful consideration of phase distribution is vital to avoid the issues associated with phase imbalance which can include:
- Wasted energy. When phases are out of balance, current will flow on the neutral cable to maintain Kirchoff’s law. This current flow is simply wasted energy.
- Nuisance tripping due to overloaded phases, especially with switchover to backup power with different phase loadings.
- Fire hazard due to overloading of a single phase or high current on undersized neutral cables
- Higher electrical bills where electrical suppliers simply charge by multiplying the highest consuming phase by three.
If attention is not paid to the phase distribution of equipment, large phase imbalance is very probable.
The graph below shows the ideal situation where all three phases are perfectly balanced and each has a power factor of 1.0. This results in zero current flow to the neutral (N).
As the graph below shows, phase imbalance causes current to flow on the neutral cable. This current is wasted energy and can pose a fire hazard as neutral cables are often undersized. Difference in current draw on the phases could further create unwanted circuit breaker trips.
NUMEN AI Solution
Numen AI provided accurate data on the extent of the problem and clearly showed the machines that were causing the imbalance. In this case, the problem could be corrected by redistributing other loads across the three phases to ensure phase balance.
Numen AI is now continually monitoring to ensure phase balance remains as it should be.