Ready for Takeoff

In a recent article, we covered the need for having a “pre-flight check” performed before moving existing or installing new information technology (IT) equipment into a newly constructed computer room. Since the contractor was not aware of the requirements for such a room, we volunteered our power quality services in exchange for interesting data for this article.

The same company occupied the building for more than 25 years, and the computer facilities had undergone many changes over that time. Starting with a minicomputer (PDP-11) for performing manufacturing and accounting tasks and a second for engineering development, a handful of terminals offered all the access needed. A room full of servers with 100 networked personal computers distributed throughout the building eventually replaced this.

The electrical requirements for this configuration have changed as much as the network architecture. It started with a single 220 outlet for the minicomputer (no UPS either). As the gradual switch-over to local area networks occurred, the breaker and wiring got “up-sized,” an extra outlet and AC unit were installed, and more.

Because change happened a little at a time, things ended up a bit haphazard and created many power quality problems-from improper grounding that caused PCs to randomly reboot, to computers sharing power lines with copiers, resulting in erratic behavior and data corruption. Not an uncommon situation in many industrial and commercial facilities.

The electrical infrastructure for a new computer room was designed based on the existing UPS wattages (8,500W); an additional 40 percent of circuits and outlets were added for future expansion.

Besides the concern over the power quality, the HVAC system needed to remove the heat and keep the equipment within its operating temperature and humidity range.

The voltage sag at an outlet under load is caused by the current flowing through high-source resistance, causing a voltage drop “upstream” from the load and leaving less voltage for the load.

Loading each outlet in turn and measuring the current and the voltage drop allows the source resistances of the individual circuits to be calculated. To accomplish this, the outlet was instrumented and a 1.5kW load (resistive space heater) was connected. The instrument recorded the voltage drop. An acceptable source resistance is typically 0.5 ohms or less.

The hot or phase conductor is not the only cause for concern. Check the neutral conductor (or other phase conductor for 230V). Using the “delta V/delta I” calculations discussed in previous articles, nearly all the outlets ranged from 0.1 to 0.2 ohms for resistive loads at 60Hz-quite good.

However, we did find a couple of original outlets in the room from before the reconstruction that were close to 0.5 ohms. These were subsequently marked for “noncritical load use only.” In addition, the N-G voltage on single-phase outlets should be less than 1V. The room checked out all right.

The next concern was the source resistance at the distribution panel. This is determined in the same manner; the voltage is monitored and a load applied. In this case, the load should be much higher. A total load of 9kW was applied, and the voltage drop was measured at less than a 3 percent reduction.

However, with 9kW being dissipated in a 13-by-5 foot room, it became apparent that the HVAC system was going to be important. The temperature was rising faster than 1 degree per minute. The heat load calculated out to almost 31,000 BTU or about 2.6 tons of air conditioning. A split unit had been installed and the evaporator had been mounted high on the wall.

The capacity of the unit was not marked on the nameplate, but a quick lookup of the model number on the manufacturers Web site indicated a capacity of 3 tons. Though the current IT load is not this high initially, there will be additional heat radiating off the roof and from an uninsulated cockloft area above in the summer.

These tests were conducted with resistive load banks, which only test at the power frequency. IT equipment typically has significant harmonic current of the odd harmonics, from third to 15th (or higher). The harmonic impedances are different and usually higher than at the power frequency.

Simulating such loads was deemed not necessary at this site, since the 60Hz impedances were so low. Once the computer room is fully loaded, the voltage harmonic levels, as well as amount of “flat topping” of the voltage waveforms, will be checked.

The third harmonic can reduce the peaks of the waveforms, which gives less charging voltage to the capacitor banks of the AC-DC converter found in most electronic equipment, and in turn, reducing ride-through time.

Though we do not expect to find a problem here, it is a quick check that can save aggravation later, as well as provide a baseline as additional equipment is added to the room. EC

BINGHAM, a contributing editor for power quality, can be reached at 732.287.3680. KINDER is principal engineer at a power-quality equipment manufacturer. He performs new product research and development and product maintenance.



About the Author

Richard P. Bingham

Power Quality Columnist
Richard P. Bingham, a contributing editor for power quality, can be reached at 732.287.3680.

Stay Informed Join our Newsletter

Having trouble finding time to sit down with the latest issue of
ELECTRICAL CONTRACTOR? Don't worry, we'll come to you.