I recently overheard a conversation that reminded me of how we take for granted that things will work the way we want them to without doing a few simple checks to verify them. Before every flight, a pilot will go through a checklist, walk around the plane, wiggle this and look into that-all to improve the probability of a successful, uneventful flight. When it comes to successfully and uneventfully powering critical loads, that thought process is often overlooked.
The conversation concerned a facility's computer room relocation. It required knocking down and rebuilding some walls, repositioning doors, running wire up from the service entrance to a new distribution panel, wiring a dozen or so outlet boxes, and making changes to the HVAC system.
The underlying factor is power-enough power of the right quality to operate the servers, routers and other IT equipment properly. But all that power generates heat, hence, the modifications to the HVAC system. The company was all set to move all of the IT equipment into the room overnight until I asked if they verified that the power is good and the heat got out. “Well, no” was the sheepish answer; they assumed everything would work fine. We all know what assuming does.
With a little persuasion, they agreed to have a “preflight check” performed. We encountered a potential problem on the first step of the checklist. There was no inspection by a qualified licensed inspector. They were willing to risk their business by assuming everything was installed according to Code. This does not mean the contractors did anything wrong, but there are reasons why the building codes require inspections on major renovations like this.
The second item on the checklist also surprised them. They did not have a list of how much power was required for all of the IT equipment, which would have dictated the size of the service needed, as well as the HVAC capacity to get the heat out of the room. Computer rooms typically need to have tight temperature and humidity controls for proper operation. Without knowing the total kilowatts, it is not possible to know the size of the air conditioning unit needed, where 1W converts to 3.4 BTUs.
The next step was to check if each outlet was wired properly and if the wires ran back to the service entrance as expected. In a normal single-phase outlet where other loads are on the same circuit, Kirchoff's and Ohm's laws would say the line-to-ground voltage should equal the line-to-neutral voltage plus neutral-to-ground voltage.
Since there is current flowing in the line and neutral (or grounded) conductors, there will be a voltage drop in the line and neutral conductors based on the wiring and connection impedances.
Measuring voltage for L-G will only see one of the drops. Measuring L-N will see both. Hence, L-N should be less than L-G, and N-G should be in the region of a volt or less. In a normally operating circuit, it is unlikely to be exactly 0.0V, again, because of the voltage drop in the neutral conductor. However, if the neutral and ground bond is illegally made in the distribution panel (except where there are separately derived sources involved), then the voltage should not read 0.0V in an operating circuit.
In our case, since there weren't loads on the circuit, there wasn't current flowing in any of the conductors and one expected that the L-N and L-G voltages would be identical and the N-G voltage would be zero at this point in the check.
Once we established that, it was time to see how the new circuits perform under load. Using load banks to approximate the load of the IT equipment, the loads were switched on and off and the voltage sags were measured to verify that during start up and also steady-state operation, the voltage levels are within spec for the IT equipment.
With the load banks on, we could also check the L-N, L-G and N-G voltages to see that they fit the pattern for a normal operating condition. When the actual equipment is plugged into the outlets, the voltages will likely be different, as IT equipment is notorious for harmonic generating currents, whereas the load banks were purely resistive.
With the load banks on and the PQ monitor in place, it was time for the next two items on the checklist: does the HVAC keep the temperature within limits, and does the quality of the voltage in the room remain within specifications while the rest of the loads in the building are operating over a normal business cycle?
For most facilities, a normal business cycle is one week, as different operations usually take place each day of the week and at different times. For a 24/7 facility, it might remain constant in the facility, but since the electric utility has other customers on the same feeders without such consistent loads, it is good to monitor for at least 24 hours.
Once all the checklist items got a passing mark, it was time to put the IT equipment in and repeat the monitoring step. The actual loads were different than the load banks, and what really mattered was that the IT equipment ran properly. Should there be an operational disruption caused by any installation issues, it can be caught right up front and fixed. That way, all the passengers are assured of a smooth flight. EC
BINGHAM, a contributing editor for power quality, can be reached at 732.287.3680.