Picture this: You're on a dream vacation in Hawaii with your spouse to celebrate your anniversary. The week has been great, and it's the last day before the long flight home. You both decide to spend the final day relaxing on the beach, reflecting on your wonderful week in paradise. While you're on the elevator, heading down from the gorgeous, ocean-view hotel room, the lights blink and the elevator grinds to a halt between floors.
As it turns out, that little blink in the lights caused similar problems at about a dozen other high-rise hotels in the area. And because it's Sunday, there's only one maintenance person on call to get the elevators at the affected buildings back online. When he finally arrives, half the day is gone and you and your spouse are irate.
This story isn't as far-fetched as it sounds. A similar situation happened at one of Hawaii's finer hotels on Waikiki Beach. About 10 to 15 times a year, any one of the hotel's elevators would stop and experience a control-card failure, which meant long delays for the stuck passengers. The hotel engineers and managers were frustrated. The elevator service company blamed the utility, citing poor power quality as the cause.
The elevator service company asked the Hawaiian Electric Co. (HECO) for their assistance. HECO installed a power-disturbance recorder at the hotel's service entrance. The results confirmed that voltage sag disturbances did cause the elevator lockups and, in one case, a control-card failure. The recorded sags, averaging about one per month, were attributed to various fault clearings on the utility distribution system.
Despite this data, the elevator service company still had to resolve the problem of the control card's occasional failure and the random system lockups, which required manual resets. This called for a cooperative effort, which was undertaken by HECO, the hotel engineers, the elevator service company, and EPRI PEAC Corp.
The hotel's elevators are delta-connected loads, fed from a 208V, 3-phase service. The control system is a Ward-Leonard type, which means it uses a DC generator mechanically coupled to an AC motor shaft. The generator's field voltage is controlled, producing a variable DC output. The variable DC is necessary to control the speed of the DC motor that drives the elevator car.
The control card delivers the variable DC voltage to the generator's exciter field winding. The field winding is highly inductive and can carry significant current, depending on elevator loading. If the current is interrupted, the field winding can produce a high-energy voltage transient.
Under controlled stop conditions, the exciter field is de-energized quickly, and a snubber circuit absorbs the transient energy. The remaining energy is diverted through a selenium surge suppressor connected across the field winding. These selenium devices are the predecessors to the metal-oxide varistors (MOVs) used today for surge protection and transient minimization.
The primary diagnostic tool used in the investigation was a portable sag-test device or “sag generator.” A sag generator has the ability to induce voltage sags of controlled magnitude and duration on the load undergoing tests. For this particular case, the team enhanced the sag generator with EPRI PEAC's data acquisition software, which allowed them to monitor the responses of selected components (on 16 channels) during the induced sags. This way, the team could determine which components were most susceptible (and exactly how they were susceptible) to voltage sags. It also validated cost-effective solutions on-site. The team didn't have to wait for the next disturbance to find out if a specific solution would actually work.
The team also chose to apply specific beginning and ending phase-angle controls of the induced sag using both the sag generator and the software (see Fig. 1, on page 58). The generator contains the switching devices and the software monitors the voltage to decide when to start the sag. This approach is helpful because, in the real world, a power disturbance's point of initiation is random. In addition, many sensitive devices behave differently when the test is initiated at different phase-angles. If you want to set phase-angle controls, find a sag generator that uses amplifiers or IGBT switches. Ones that use contactors or thyristors for switching or don't have control software will not permit such testing.
After numerous discussions and an analysis of the elevator control schematics, the group devised a plan to sag test one of the three hotel elevators while the car traveled upward — and then downward — at full speed.
The elevator's up and down movements are two completely different electrical scenarios. The DC motor must drive the car downward because the counterweight is heavier than the car. However, the upward movement of the car occurs naturally because of the counterweight. The motor must then act as a brake, regenerating electrical energy into the power system.
The team connected the sag generator at the elevator's main fuse disconnect. This way, they subjected the entire elevator system to the voltage sags. Sag tests began at 90% of nominal voltage for short durations, then longer durations. Sag magnitude was then reduced to 85%, and each duration repeated. Sag tests would continue down to lower voltage levels until a voltage tolerance envelope was developed for the components in the elevator system.
This systematic approach allowed the team to cover the entire range of sags from least severe to most severe, while minimizing the number of interruptions to the equipment. Sag durations of 6, 12, and 24 cycles were chosen to simulate utility switchgear operations. In addition, a series of unbalanced sags were performed. This type of sag simulates the single line-to-ground faults typically experienced on electrical distribution circuits.
The team put the elevator into motion and subjected it to the matrix of voltage sags. They found that the elevator system could ride through both single-phase and 3-phase voltage sags to 87% of nominal for any duration. Voltage sags below 87% of nominal caused the drive to trip and automatically reset. If the voltage sag fell below 80% of nominal, the drive would trip offline and its programmable controller would lock up. This controller lockup required a manual reset of the system.
Fig. 2 shows the complete battery of sag test results in the form of a voltage tolerance envelope. As it turned out, any time the applied sags were significant enough to cause the microprocessor to reset, significant voltage transients were observed across the power supply control card (mainly due to the loss of the controller).
While these transients were not always damaging, it was clear that a properly rated MOV would probably solve the card damage problem.
It also was clear that using some type of power conditioner on the main controller power supply would likely eliminate the nuisance-tripping problems that necessitated manual resets.
Based on the sag test results, hotel management implemented two recommendations. First, technicians installed MOVs on the control-card outputs to replace the selenium diodes and protect the cards from potentially damaging transients.
Second, an uninterruptible power supply (UPS) system was installed on the main elevator controller to eliminate the potential for lockups during voltage sags. As a result, the problems with control card failures and nuisance trips were finally eliminated.
Use of a portable sag generator in conjunction with a cooperative hotel staff and elevator maintenance firm proved invaluable in resolving a serious elevator problem. In the future, we'll describe additional cases where a sag generator helped solve difficult-to-diagnose problems, and we'll discuss some of the investigation methodologies that make this testing technique a valuable piece of the power quality puzzle.