Power Quality (PQ) monitoring historically involved portable monitoring equipment used in a temporary setup to identify power quality problems. Once the problems were identified and corrected, the monitor was removed. For critical facilities the need for PQ monitoring is more involved. These applications require continuous monitoring as part of their PQ assurance program. These facilities typically require multiple monitors installed permanently on the electrical distribution system. Problems soon developed as to the notification of PQ disturbances as well as management of several stand alone devices. To address these needs, a concept was developed for networking the portable monitors together. A system was engineered and installed at a client's facility.

This paper presents the concept and methodology of a PQ network and gives a detailed description of the system installed at the client site. This paper also discusses the technical requirements of the equipment chosen as well as the performance of the system.

Power quality monitoring is an important part of any PQ assurance program. Many times the equipment vendors blame the power as the cause of failures. In most cases the power is assumed "guilty" and it is up to the facility's personnel to prove otherwise. It is not uncommon to install power mitigation equipment to solve unknown problems! Many facility managers have found out the hard way that the power mitigation equipment they just installed did not solve their PQ problems, or worse yet is causing a new problem. This type of finger pointing only causes expense and frustration for all parties involved!

Power quality monitoring provides the diagnostic tools required to eliminate such scenarios. The needs of large or critical facilities go beyond the capabilities of portable monitoring. These facilities have identified the needs as:

1. continuous PQ monitoring of utility, standby, and critical power,
2. centralized notification and logging of any power quality disturbances,
3. daily summary reporting of maximum/minimum values,
4. archiving of data for trend analysis and vendor disputes,
5. data logging of circuit breaker operations for event tracking,
6. centralized control and command of all systems, preference for Graphic User Interface (GUI),
7. fault tolerant design, must continue to record even under catastrophic failure,
8. provisions for data analysis and the ability to condense a large quantity of data into usable, understandable formats,
9. and use an open, non-proprietary architecture to allow for technology changes.

The development of a PQ network was broken down into two main areas, PQ monitor selection and network integration. The PQ monitor selection included the following items:

1. determine areas to be monitored,
2. develop PQ monitoring criteria,
3. investigate locations of equipment,
4. evaluate commercially available monitors,
5. evaluate network/communication capabilities,
6. demonstrate ease of use and graphic abilities,
7. and gather cost estimates.

Based on the these steps, the power quality monitors were required to have true RMS voltage and current measurement, AC and DC voltage sampling, abilities to record voltage disturbances with durations from 1 microsecond to seconds, provide disturbance data in a graphic format, provide a 24-hour summary of steady state voltages, and have communication capabilities. Auxiliary data input channels were used for circuit breaker data logging. In addition, the monitors were required to have internal battery backup protection and automatic default back to stand alone operation. The capacity to add total harmonic distortion (%THD) analysis was deemed desirable.

Based on the developed criteria, the Basic Measurement Instruments, BMI Model 4800 was chosen for the PQ monitoring portion of the system. This unit includes 4 main monitoring channels and 8 auxiliary channels.

The network integration comprised the second area of the PQ network development. This included the following items:

1. determine size of system,
2. calculate network traffic,
3. reliability of system and MTTR,
4. evaluate cabling topology,
5. select network protocol,
6. compare commercial hardware vs. industrial hardware,
7. operating system and software selection,
8. peripheral selection including printers and tape backup,
9. designate custom software development,
10. and gather cost estimates.

This section proved to be the most challenging part of the design. The final system chosen was an Ethernet, Net BIOS based network that operated under a MSDOS 5.0 environment. The cabling topology used was a combination of thin Ethernet (RG58 coax) backbone and 10BaseT star for each network node. For reliability and distance considerations the network was divided into three segments. This arrangement allowed for nodes to be added or removed from the system without taking the system down (Figure 1).

Figure 1. PQ Network Block Diagram

Optically isolated RS232 line drivers were used to connect each PQ monitor to its node. This prevented ground loop interference and allowed the monitors to be remotely mounted throughout the electrical distribution system.

Concurrent to the hardware integration, custom software was developed to provide alerting and logging functions. This software operates under Windows 3.1 and can operate in the background while other programs are running. The basic function of the software was to scan all disturbance and report information from the PQ monitors. The information was then filtered and separated into report and disturbance categories. Any disturbance was immediately flagged via a monitor warning box and an audible alarm. Both disturbance and report logging are displayed on the monitor status box and printed on a dot matrix printer (Figure 3).

Each monitor had its own archive directory for data storage. Data files were organized by creating a new file for each day. The operator could view any report or disturbance from any of the PQ monitors by selecting the appropriate data file. The data graphs could then be imported into word processors for reports or presentations as shown in Figure 2.

The hardware/software design also allowed for multiple display stations as well as remote dial in capabilities. Command and control functions were provided at each display station. Special communication software allowed the operator to connect and control any of the PQ monitors through the network system. Automated data backup was provided by tape cartridge and scheduling software.

Once all of the hardware and software was selected, a prototype system was completed and tested. Testing procedures included 48 hour hardware burn-in, software compatibility tests, and system de-bugging. Fine tuning the system included configuration changes, software modification, and enhancements.

Figure 2. Sample graph of power disturbance

3.0 CASE STUDY

This study discusses the installation of a PQ network at a large communication company data center.

The client's facility was a six story building that includes over 200,000 sq. ft of computer room, three utility service feeds, 14 megawatts of standby generators, 7,000 KVA of UPS modules, and over 3,000 wet cell batteries. Primary voltage was supplied by the utility at 13,800 volts and stepped down to 4,160 volts for each service. The generator plant was divided into three sections, each providing 4,160 volts for each service. The service switch gear included closed transition switching (no break) between the utility and generator plant.

Dual 4160 volt feeders were run to each floor of the facility where they were stepped down to A side and B side 480/277 volt 3 phase. Each side powered a paralleled redundant 750 KVA UPS system with dual modules. Both UPS systems (total of 4 modules per floor), were connected to an output switchboard that had a common tie breaker. This allowed either side to power the load under maintenance or repair conditions. The A side and B side UPS power was distributed to Power Distribution Units (PDU's) in the computer room. The PDU's stepped the voltage down to 120/208 volt 3 phase and provided electrical distribution to the computer load. Each UPS module had its own rack mounted battery string.

Production and installation of the PQ Network was started in December of 1992. The system was completed and put on-line in February of 1993. Display/control stations were installed in the data processing management area and in the facilities operation and control center. Each display station was equipped with a log printer. One station included a graphic report printer and tape backup system.

The PQ network required to monitor this facility included 23 PQ monitors. Six monitors were installed on the three utility services and three generator plants. Ten monitors were installed on the output of the UPS systems and 6 monitors are installed on the UPS battery strings. One monitor was utilized as a portable field monitor and spare. The monitors were remote mounted in the electrical switch gear enclosures and battery rooms. This minimized sensing lead length and attenuation losses.

Due to the high voltage of the utility and generators, the respective monitors were connected to the secondary side of the switch gear potential transformers (PT's). These PT's had a ratio of 35:1 and stepped the high voltage down to 120 volts for control and monitoring circuits. Power disturbance and steady state voltage data was multiplied by the PT's ratio to calculate the high voltage values. Voltage impulse accuracy was dependent on the frequency response and loading of the PT's [2].

Voltage monitoring points on each UPS output included phases A-B, B-C, A-C, and phase A-Neutral. Initial threshold settings were set using the industry Power Quality Envelope as a guideline [1].

There were 4 UPS battery strings per PQ monitor. Swell voltage thresholds were set between nominal float and equalize levels. Sag thresholds are set 20 VDC below normal float level. The sag setting would trigger a 20 second battery discharge graph when the UPS switches to battery. This information is critical to determining the condition of the battery system. The initial DC voltage drop is recorded for trend analysis. A sudden change of initial DC voltage drop is an indication of a weak battery or poor electrical connections. If the initial DC battery voltage drops below the inverter cut-off point, the UPS will shut down and drop power to the load. By monitoring the battery strings, deteriorating conditions will be discovered before catastrophic failure occurs.

Due to the complex electrical distribution system, data logging of critical circuit breakers was required. A total of 32 circuit breakers were monitored as to open or closed position via the environmental channels in the BMI 4800. When a breaker changes position, the time stamp is used to re-create the event and correlate any disturbance data.

Each PQ monitor was connected to a network node located on the respective floor. Optically isolated RS232 short haul modems and unshielded twisted pair cable were used to prevent signal loss and ground loops.

Long distances between network components in the facility required Ethernet repeaters to extend the coax backbone. The backbone was split into three segments for reliability purposes. One segment was used for the display stations, the other for floors 1-3, and the last for floors 4-6.

At the time of this paper, the PQ network has been operational for 6 months. A number of software changes and enhancements have been made as more is learned about the capabilities of the system. PQ monitoring thresholds have been adjusted to match the facility's nominal levels.

Expansion of the system has been proposed to include monitoring additional battery systems as well as a UPS system in a remote building.

Two software enhancements are being developed to provide automated analysis of the archive data files.

(A) Steady-state voltage histogram graphs will condense data from selected files into one graph. The histogram will indicate the percentage of time that the voltage was inside or outside the nominal voltage limits.

(B) PQ disturbance data from selected files will be plotted on the power quality envelope [1] to give summary information as to the type and number of disturbances over a given period.

This paper has presented the concept and need for a Power Quality network as part of critical power quality assurance programs. It has shown how integrating portable PQ monitors into a total system is both practical and functional. The system provides effective management of PQ resources and information. Alerting and logging functions allow non-technical personnel to monitor critical support systems. The open architecture of the system makes it very adaptable as PQ monitoring needs change.

New software enhancements will extend the PQ network's function to include automated analysis of power quality.

The author is grateful to Mr. Dave Pares of Lone Eagle Systems (Omaha, Nebraska) for technical assistance and software support and to Mr. Gene Sullivan, Mr. Doug Solberg, and Ms. Laura Manning, EE (US West Communications, Omaha, NE). for design and monitoring criteria.

Special thanks go to Mr. Bob Yester, PE, Garth Muirhead (Swanson, Rink Engineers, Denver, CO), and all parties involved in the installation of the PQ network system.

1. "Guideline on Electrical Power for ADP Installations," FIPS PUB 94, US Department of Commerce/National Bureau of Standards, September 1983, pp. 15-16

2. Douglas, D.A. "Potential Transformer Accuracy at 60 Hz Voltages Above and Below Rating and at Frequencies Above 60 Hz, " IEEE Transactions on P.A.S., Vol. PAS-100, No. 3, pp. 1370-1375, March 1991

Martin D. Conroy is CEO of Computer Power Corporation in Omaha, Nebraska.

He founded CPC in 1981 to provide power quality services and equipment to meet the growing needs of clients. Over the last 5 years, he has been extensively involved in providing power quality surveys and consulting services to major accounts. Martin has specialized in area of power quality, grounding, harmonics, and code inspections. He has developed and taught power quality seminars for both commercial and utility accounts.

He was the principal designer of the PQ network presented in this paper.

Previous to founding CPC, Martin worked in the electrical contracting field for 8 years.

Martin is a IAEI electrical inspector and holds a State of Nebraska, Class A Electrical Contractor license.