Ecmweb 1840 306ecm39fig1
Ecmweb 1840 306ecm39fig1
Ecmweb 1840 306ecm39fig1
Ecmweb 1840 306ecm39fig1
Ecmweb 1840 306ecm39fig1

Ask the Experts

June 1, 2003
They may be experts, but that doesn't mean they won't listen to dissenting opinions. Got a comment about a previous topic? Send it to [email protected], and we'll have the applicable expert — whether it's Mark McGranaghan, vice president of consulting services for EPRI-PEAC, John DeDad, editorial director of EC&M, or Mike Lowenstein, president of Harmonics Limited — address it. This month, two readers respond to McGranaghan's discussion of leading power factor at a data center, and DeDad responds to a query about the point of common coupling.

They may be experts, but that doesn't mean they won't listen to dissenting opinions. Got a comment about a previous topic? Send it to [email protected], and we'll have the applicable expert — whether it's Mark McGranaghan, vice president of consulting services for EPRI-PEAC, John DeDad, editorial director of EC&M, or Mike Lowenstein, president of Harmonics Limited — address it. This month, two readers respond to McGranaghan's discussion of leading power factor at a data center, and DeDad responds to a query about the point of common coupling.

Reader response: We recently experienced leading power factor when we put our data center online, similar to the situation you discussed in a previous column (Ask the Experts, March 2003). The culprit in our case was power factor correction in the UPS. When lightly loaded, the power factor capacitors in the UPS pushed the power factor to a leading condition. To correct the problem, we had the UPS manufacturer install a disconnect switch on the capacitors. This allows us to keep them de-energized until we have adequate loading on the data center to require their need.
Gary W. Bolin, P.E., Northern Colorado Workplace Services, Agilent Technologies, Inc.

Reader response: Mark McGranaghan's response to the question about leading power factor in a data center was only partially correct. The incorrect portion refers to his doubt that computer equipment has PF corrected power supplies. Starting in 1996, a CE Directive issued by the European Committee for Electrotechnical Standardization was established that drove an EMC Directive, which in turn, generated EN61003-2, a European product performance standard. This directive went into effect Jan. 1, 2003, and requires all power supplies sold in Europe to be equipped with power factor correction components. Given the fact that nearly all data processing equipment manufacturers sell their hardware worldwide, power factor correction equipped power supplies has become the norm for data processing equipment. While it may not change the overall outcome of Mr. McGranaghan's response, it's important to note that equipment manufacturers are sensitive to this issue and continuously strive to improve the customer experience.
Ken Baker, ISS Rack and Power Solutions, Hewlett Packard Co.

McGranaghan's response: I appreciate this clarification. EN61003-2 establishes harmonic limits for the load currents for power supply loads, called Class D loads in the standard. Computer manufacturers, such as HP, must design their power supplies to meet these harmonic limits if they're to be sold in Europe. This often requires some power factor correction. This again refers to true power factor because the harmonic components don't have any impact on the displacement power factor. The confusion between displacement power factor and true power factor was the main point of the original question and discussion. Nevertheless, the implications of EN61003-2 make for a good discussion.

Reader comment: I read John DeDad's answer about the point of common coupling (PCC) and have to disagree to some extent (Ask the Experts, May 2003). The goal of applying the harmonic limits specified in IEEE Standard 519-1992, as I understand it, is to prevent one customer from causing harmonic problems for another customer or for the utility. If you have high harmonics within your system, they may be detrimental to your system, but as long as they don't cause problems for another customer, they do not violate IEEE Standard 519. Certainly one would want to voluntarily limit harmonics within one's own system to avoid operational problems, perhaps to the levels specified in the IEEE standard, but the standard only applies at the point where you affect your neighbor, which is the PCC, as I understand it. Only if you have multiple feeds from the utility would you have multiple PCCs.

Maybe this is just semantics. I know that it's helpful to look at various points in a system and see what level of harmonics you might have as compared to the IEEE 519 standard (at least voltage harmonics), but these should not be called PCCs. They are not points in the system where you can adversely affect another utility customer. If you have one connection with the utility, then there is one point where you can affect your neighbors if you inject harmonics. That is the PCC. That is where you must meet harmonic limits — if that is what your utility contract specifies.

The basic problem with your approach, and why I am being so picky, is that if you do call points internal to your system “PCCs,” then you may mislead people into thinking that the IEEE 519 limits apply there, too. While it may be a good idea to minimize harmonics internal to your system, it's not required.
Tom Blooming, P.E., Power Systems Engineering, Cutler-Hammer Engineering Services and Systems

DeDad's response: Your point is well taken in that the IEEE 519 standard's intent is to reduce or eliminate the possibility of generating harmonics onto the utility supply and causing problems for the utility's other customers. I believe Warren Lewis' intent was to clarify and point out other points on a distribution system where the interaction of the served nonlinear load's distorted current waveform can create distortion on the voltage waveform. It's this resultant voltage waveform that's provided to the bus or feeder served by the power source. From here, it can propagate to all served downstream loads as a power quality problem related to the distorted voltage waveform.

IEEE 519 has created a good deal of confusion and misinterpretation in the power quality industry. Our intent certainly wasn't to exacerbate these conditions, but to inform our readers of other possible points of potential problems resulting from the existence of harmonics on power distribution systems.

And after all that, there's still enough gas left in the tank to answer one of your questions. What do you do when the handbook that's supposed to have all the answers is missing the one you need?

Q. Dranetz-BMI's “The Handbook of Power Signatures” mentions several times that high neutral-to-ground voltages cause computers to lock up and to corrupt data files. However, it doesn't mention at what magnitude this becomes a problem. What is the neutral-to-ground voltage tolerance level for computers?

DeDad's answer: The EPRI Power Electronics Applications Center (PEAC) conducted tests to evaluate the effects of steady-state, momentary, and transient neutral-to-ground (N-G) voltage differences on personal computers (PCs) in the early '90s. The results of the steady-state test address your question.

To create a 10-min overvoltage between the neutral (grounded) conductor and grounding conductor, PEAC connected an isolation transformer between the PC and its power source, without bonding the neutral or ground of the transformer secondary (Figure). This replicates what is seen as a real-world miswiring condition. Before the test began, the PC was operated for 10 min to stabilize its temperature. A digital storage oscilloscope recorded the AC voltage between the neutral and grounding conductors of the PC. Also, a diagnostic program was running on the PC and visible on the PC's monitor during the test to verify the continuous operation of the random access memory and the CPU.

According to PEAC, the resulting continuous N-G voltage of approximately 50Vrms didn't upset the PC. PEAC's conclusion is that, contrary to popular belief, off-the-shelf PCs are almost immune to steady-state N-G voltage differences as well as momentary and transient ones.

PEAC does caution that its test results don't indicate how those same N-G potentials affect PC operation with the PC connected to peripherals via communication ports. N-G overvoltages on a PC connected to a network, printer, modem, or any other communications peripheral may cause a voltage difference between reference grounds, which can corrupt the data flowing from one device to another, resulting in a lockup.

About the Author

Mark McGranahan, Electrotek, and Mike Lowenstein, Harmonics Limited

Voice your opinion!

To join the conversation, and become an exclusive member of EC&M, create an account today!

Sponsored Recommendations

Electrical Conduit Comparison Chart

CHAMPION FIBERGLASS electrical conduit is a lightweight, durable option that provides lasting savings when compared to other materials. Compare electrical conduit types including...

Fiberglass Electrical Conduit Chemical Resistance Chart

This information is provided solely as a guide since it is impossible to anticipate all individual site conditions. For specific applications which are not covered in this guide...

Considerations for Direct Burial Conduit

Installation type plays a key role in the type of conduit selected for electrical systems in industrial construction projects. Above ground, below ground, direct buried, encased...

How to Calculate Labor Costs

Most important to accurately estimating labor costs is knowing the approximate hours required for project completion. Learn how to calculate electrical labor cost.