Data centre power systems, air conditioning plant: testing and commissioning

System composition – What to test?

Any given emergency power generation unit is a complex system, or series of systems; working together to perform several duties at once and at the system’s heart is the generator. This could consist of several diesel generators. However, various discrete systems and components complete the total package, such as UPS systems, alternators, regulators and switchgear.

These additional components typically come from various manufacturers, and are usually designed to interface with a number of makes, models, and sizes of generators. As with any other mechanical or electrical components, all are potentially subject to failure, and have varying maintenance needs, at the very least requiring regular testing and servicing. However, individually testing a series of components never answers the most important question of all: “How do you know that your system—and not merely its components—will work when it counts?” In an actual emergency, a platform or FPSO’s entire emergency power generation system will be stressed. Unlike in a series of short, component-by-component tests, the system must operate at full power, with all components working together to do their jobs. The stresses introduced by this mode of operation cannot be simulated by discrete tests of a system’s numerous individual components: automatic transfer switches, switchgear, load-sharing centres, voltage regulators, alternators, electrical cabling and connectors, ventilation, cooling systems, and fuel systems.

While the generators may have been tested at the factory, the installation variables of the interaction with other parallel-connected power generation units (UPS), load profile, ambient temperature, humidity, fuel, exhaust, and cooling systems can be significantly affected by the installation. Typically, gensets are cautiously oversized and in many data centre applications they are designed with future expansion in mind. Therefore the available load from the centre is not accurate or large enough for the commissioning of multiple gensets and will not provide a healthy load for maintenance testing in future.

Therefore, a system-wide test is the only way to ensure that the individual components of any power generation system will work together harmoniously, whether for continuous production demands, or in an emergency power outage situation.

The limitations of  resistive-only testing
Sometimes commissioning engineers may only consider testing their genset’s engine, rather than the whole system. The most common form of testing is using a resistive load bank to run the prime mover, connected at the generator’s bus. However, this fails to replicate the actual stresses produced during real-world generator operation.

A resistive-only loadbank provides an electrical load (at unity power factor) which when applied to a generator converts and dissipates the resultant generated power as heat. This electrical loading will highlight individual engine problems. Unfortunately, resistive loads are usually only a small part of any data centre’s total power consumption. Quite often, the influence of a lagging power factor (pf) <0.8 due to reactive loads is underestimated or even ignored.

Generally the only equipment operating on a resistive-only load are incandescent lights and electric heaters; these units draw a steady supply of electricity from a generator, but do not produce the large block loads that truly test a generator’s performance. A resistive load test will verify that a generator’s prime mover is working, but it will not identify how well it will actually perform when exposed to the real reactive load pattern.
R+R – the role of resistive and reactive testing

A reactive load test of an installation’s power system can accurately simulate the system’s response to a changing load pattern, such as would be encountered during a real power failure. Resistive/reactive combination load banks are used to test the genset at its rate pf. In most cases this is 0.8 pf. The reactive component of the load will have a current that “lags” the voltage. The resulting power is described in two terms, the kW, or real power, and the kVA or apparent power.

The combination of resistive and reactive current in the load will allow for the full kVA rating of the generator windings to be tested. Even though the genset is producing more kVA, it is actually not producing more kW. The “real” power (kW) required from the engine is essentially the same.
The inductive loads developed during reactive testing illustrate how any given system will handle the voltage drop in its regulator; paramount when paralleling generators. The test will also verify that this regulator is working properly, if not, its magnetic field could collapse, rendering the generator useless and preventing other generators in the system from operating efficiently in parallel. Resistive/reactive testing can also reveal additional stresses (and predict pending failures) of a system’s switchgear, alternators, and other systems that resistive-only testing cannot.

Loadbanks for data centre heat load testing
In addition to the commissioning of the emergency power systems, loadbanks have a part to play in air conditioning testing. Air conditioning plays a vital role in keeping server halls at stable temperatures and humidity levels and more often than not is designed with extremely high levels of capacity, redundancy and future expansion in mind. With millions invested in the very best new technology the air conditioning plant must be commissioned against stable and traceable heat loads.

Resistive only loadbanks can provide a portable and highly controllable heat source to allow air conditioning systems to be thoroughly commissioned against their design criteria. The loadbanks are small enough to be placed directly inside the server halls and can be controlled independently at 1kW increments, or operated as multiple units at the same or differing kW heat load increase and/or decreases.

The distributed heat can then be measured by various temperature sensor and probes, or by thermal mapping devices, to ensure the air flow and cooling is evenly distributed and that there are no “hot-spots” around any vital equipment enclosures.

DC loadbanks
In addition to AC and heat load testing, DC loadbanks provide the ideal platform with which to independently test UPS battery arrangements. Small, portable and air-cooled, DC loadbanks provide variable loads at a given DC voltage and/or current. A constant current test will accurately evaluate the batteries performance against time and quickly determine that the component is within specification.