Test and Measurement Methods
Partial Discharge (PD)
This page describes partial discharge testing details, including causes of partial discharge and partial discharge measurement techniques. For a general description of partial discharge testing using the iTIG, see the Partial Discharge (PD) Test Summary.
Partial discharge (PD) is a phenomenon in which a discharge bridges only part of the space between two conducting elements. PD can occur due to a fault in electrical insulation. PD can also occur along the boundary between different types of insulating materials. Partial discharge was probably was probably first observed in rotating equipment when higher voltages came into use.
PD Observed in Both High and Low Voltage Equipment
In recent decades, PD was mainly a concern for operators of high-voltage rotating equipment. But when inverter drives were introduced to power electric motors instead of 50/60Hz sinusoidal systems, PD started to be observed in low voltage motors as well.
Today, inverter drives, also called variable frequency drives (VFDs) or variable speed drives (VSDs), adjustable speed drives (ASDs), etc. are commonplace and their use is growing. When VFDs/ASDs are misapplied, partial discharge is often generated in motor windings with serious or catastrophic results.
When using inverter drives, good system design is an important aspect of ensuring long motor life. However, many cases arise when there are limitations, historical or otherwise, to the system design and re-design options available.
PD Testing for Early Diagnosis
Systems often degrade in an unknown manner. In most cases, early changes in motor condition will be the first indication that there is a problem. These changes are best detected by conducting a surge and partial discharge test on the motor with a motor analyzer. If the surge test fails, the motor condition is critical. If the surge test passes, a PD test will provide valuable information about the condition of the motor and the motor’s power system, and provide early warning of insulation breakdown.
Modern motor analyzers such as the Electrom iTIG Winding Analyzer can include PD measurements, an important and cost-effective tool to detect insulation weaknesses early and diagnose motor systems that use inverter drives. With the iTIG, partial discharge measurements can be part of a fully automatic sequence of tests.
Why is partial discharge bad and how does it happen?
Partial discharge is one of the predominant drivers of long-term degradation and eventual failure of electrical insulation. The PD energy dissipates mostly as heat, but also as sound and light. Insulation materials degrade when overheated. Although the partial discharge energy can be small, the PD can take place in the same location hundreds or thousands of times per second.
In the process of PD, ozone that attacks the insulation is created. As a result, cracking associated with a chemical breakdown of the insulation can occur, and the electrical conductivity of the insulation surrounding voids in the insulation system can increase. PD can also cause “treeing” in cable insulation. Treeing is the term for when multiple partly conducting discharge channels expand over time, like the branches growing from a tree.
With inverter drive power systems are applied incorrectly, the localized partial discharge can occur several thousand times per second. In severe cases, PD can cause a motor to fail within months, weeks, or even hours.
When and where does partial discharge take place?
Exposed to electrical fields, a transient gaseous ionization occurs in tiny voids in the insulation system when the electrical stress exceeds a critical value. This ionization combined with a sufficiently high electrical stress produces partial discharges.
As shown in the top picture at right, the voids can be in any location. With layers of insulation, the opportunities for voids between layers grow.
Partial discharges on the surface of the insulation can also occur. Used rotating equipment usually has some layer of contamination on the outside of the insulation. If this “dirt” or moisture is not uniform or continuous across the insulation surface, discharges can take place on the surface. This is especially true if there is a contaminated insulation crack or high conductivity path from the conductor to the surface of the insulation as shown in the bottom picture at right.
Are there other measurable discharges?
Small discharges or spewing of electrons from a sharp point to air can be measured when the voltage is high enough. The humidity in the air influences at what voltage this happens. A sharp point can, for example, be the end of a single cable strand. Electrom Instruments has also observed and can measure discharges from small scratches in magnet wire for low voltage motors.
Partial discharge can cause progressive deterioration of insulating materials, ultimately leading to electrical breakdown. However, note that especially in high voltage equipment, low levels of PD may be fine.
Partial Discharge Generated by Inverter Drives
Inverter drives, usually called variable frequency drives, variable speed drives, or adjustable speed drives, are very popular and growing in use. There are many advantages to using inverter drives, including that they make it possible to replace DC motors with AC motors, which are typically smaller in physical size and reduce operating cost.
However, a key disadvantage to using an inverter drive is that if the power system is not applied correctly, there can be large voltage spikes or voltage “overshoot” on the motor terminals. These spikes can cause partial discharge that eventually results in a failure in the motor insulation system. Several users have told us they just “threw a VFD on it,” meaning they bought a VFD and powered an older motor with it. Usually, the VFD would be added using existing power cables with little regard to cable lengths, cable type, and other important considerations. As a result, the motor lasted only a short time.
How Inverter Drives Work
The inverter drives used to control speed come in many varieties with different types of outputs. Pulse width modulated control, or PWM, is the most commonly used. It takes a fixed AC input voltage, converts it to DC with rectifiers, and from the DC voltage produces square waves or rectangular pulses with variable pulse width, variable frequency, and sometimes variable voltage. See picture below.
The rectangular pulses are typically produced by integrated gate bipolar transistors (IGBTs). By controlling the on/off state of the IGBTs, the DC input voltage is essentially chopped up into DC output pulses. Since the voltage pulses have different widths, they produce a sinusoidal wave pattern for the voltage and current of the load to which they are connected. How smooth and sinusoidal the wave pattern will be depends on several factors, including the number of pulses generated per cycle. The pulse rate per cycle is called the carrier frequency or the switching frequency.
To provide “clean” power with an overshoot voltage the motor can handle, the system that includes the inverter drive, the motor, and the cable between the two, must be matched properly from an impedance point of view. If this is not done, there will be large voltage spikes at the motor terminals. These spikes can be up to about two times the intended voltage, and in severe cases much higher than that.
NEMA Standards Publication ICS 7.2-2015 states that the peak voltage at the motor terminals has a typical maximum of twice the VFD’s DC bus voltage, provided the pulses are far enough apart to allow the ringing that is produced to decay before the next pulse comes. When this is not the case, the peak voltage can be higher. The pulse output voltage, called the DC bus voltage, is typically equal to or less than the peak rated voltage for the motor, RMS x 1.414 volts. Some VFDs can reduce the pulse output voltage when full power is not required by the motor.
Example of 460V general-purpose and definite-purpose motor voltage overshoot tolerance:
- DC bus voltage and peak sinusoidal voltage: 1.414 x 460V
- The spike voltage for a 460V motor with a power cable length of 50ft/15m or more can be: 2x (1.414 x 460 V) ≈ 1,300 V.
- NEMA MG1 Part 30 states that a general-purpose motor operating at this voltage should be capable of withstanding a repetitive peak voltage of 1,000 volts. This will not be good enough in the example above.
- Furthermore, NEMA MG1 Part 30 states that definite-purpose, inverter-fed motors are designed to withstand maximum repetitive voltage peaks at the motor terminals equal to 1.1 x 2 x 1.414 x VRated = 3.1VRated. This is two times the peak sinusoidal voltage plus a safety factor of 10%.
In this example, 460V x 3.1 = 1,426V. The definite-purpose motor should hold up if the system is set up correctly. If not, spikes can be more than 2x the DC bus voltage + 10% and cause problems.
Reasons for Voltage Spikes
The spikes are a response to the excitation of the RCL circuit that includes the cable and the motor. The rise-time of each pulse exciting the circuit can be as low as about 100nsec, so very fast. Transmission line theory says that under these circumstances reflected waves will occur at transition points, where there is an impedance mismatch. These transition points may occur at several locations in the system. The reflected waves are high-frequency ringing as indicated in the picture. If the impedance in the system, or the C and L components of the impedance, have certain values, the voltage of the initial spike can be high and added to the pulse voltage.
In reality, the fast rise-time rectangular pulses produced by an inverter drive are a combination of a broad spectrum of frequencies. As the cable length between the inverter drive and the motor increases so does the distributed capacitance and inductance along the cable. This causes the resonant frequency of the cable to drop. When this frequency and the frequencies contained in the inverter drive pulses are in the same range, they can become additive and result in high voltage spikes. As the cable length increases above 50ft/15m, the overlap in frequencies may start to happen depending on the setup and cable type.
Inverter drive carrier frequencies can be as high as 20kHz. This means that each pulse lasts a very short time. In some setups, the ringing of the reflected wave will not have enough time to dissipate to zero before the next pulse comes. When that happens, the reflected wave will build on top of what is left of the previous reflected wave, and the voltage spikes can get very high.
Partial Discharge Effect on the Motor
The front-end voltage spikes or overshoot come at a very high rate, one per pulse produced by the inverter drive. As mentioned, PWM drives can produce pulses with a frequency from 500Hz to 20kHz. When the voltage spikes are high enough they cause partial discharge in the motor windings because of the strength and concentration of the electrical fields produced by the spikes. The inverter drive may also trip. At even a few hundred pulses per second, let alone 20,000 pulses per second, the stress on the winding insulation can be high enough to damage the insulation in short order.
There are many ways to cut down the size of the spikes and their effect on the motor. These include filters and reactors, lower cable lengths, different types of cable, lower carrier frequency, slower rise time pulses, and more. Each has advantages and disadvantages.
Motor analyzers with PD measurement like the iTIG can help diagnose inverter drive power system problems.
Partial Discharge Measurement
There are many techniques to measure partial discharge in electric motors.
Partial Discharge Measurement Techniques
What is measured, and how large is a partial discharge?
The partial discharge measurement made by PD testers, using the surge test as the driver, is the voltage of the PD spikes. This voltage is proportional to the amount of discharge and measured in mV. The range measured is typically from a few mV to a few volts.
A charge or discharge is measured in coulombs. 1 coulomb = 1A for 1 second. It is also the charge in a 1 farad capacitor with a potential of 1 volt.
Since most people do not relate to these numbers, here are some rough comparisons:
- Typical lightning strike: 15C to 350C
- Static discharge from a human hand to a doorknob: µC range
- Partial discharge: pC and nC range (represented by mV to volts in the testers)
- 1 pC = 0.000 000 000 001C
It is important to note that all PD measurements are relative, and do not measure the absolute PD at the site of discharge. The PD voltage spikes attenuate between the discharge sites and the measurement location. The amount of attenuation depends on a list of factors and will vary from motor to motor. It will depend on the frequency range used by the instrument to detect PD, and so forth. Despite the relative nature of the measurement, it is an excellent way to track the condition of a motor and get early warnings of problems.
Partial Discharge Measurements by Electrom Instruments
The voltage stress required to generate partial discharge (PD) comes from the high-frequency surge test pulses produced by the iTIG Motor and Winding Analyzer. These pulses are similar in rise-time to the pulses produced by inverter drives and simulates what the motor sees during operation very well.
The iTIG measures repetitive partial discharge inception voltage (RPDIV), repetitive PD extinction voltage (RPDEV), and maximum PD during the partial discharge test.
Partial discharge is a broad-spectrum signal and can be measured in a wide range of frequency bands. Electrom’s iTIG Motor Analyzer measures PD in the 1MHz to 30MHz range. This means the partial discharge measurement is done in a range where the signal has relatively high energy compared to measurements done in higher frequency ranges, such as in the GHz range.
No Accessories Needed
No accessories are required for partial discharge measurement with the iTIG. The PD coupler is internally in the iTIG, and the normal output leads used for other tests are also used for partial discharge measurements.
The partial discharge measurements can be part of a fully automatic sequence of tests, including everything from inductance, impedance and low resistance measurements to hipot step voltage tests, surge tests, and everything in-between.
All Electrom iTIG models can be upgraded to include partial discharge measurements. The upgrade may involve hardware and software, or only software.
How Partial Discharge Measurement Results Are Used to Analyze Motor Conditions
Low Voltage Motors
For low voltage motors, there should be no partial discharge at the normal surge test voltages of 2E+1000V or less. This makes the analysis very easy. If PD is found, the motor has insulation that is breaking down, or the motor is contaminated with a significant amount of “dirt” and/or moisture. The contamination level should already be known from the Insulation Resistance test, and if very “dirty” the motor should be reconditioned before conclusions are made about partial discharge levels.
When a low voltage motor is used with an inverter drive, PD indicates that the system has problems that eventually will fail the motor. Actions should be taken to improve the system to extend the life of the motor. These improvements can include those mentioned in Partial Discharge Effect on the Motor. Re-impregnating the stator can also extend the life of the motor.
If the surge test is done at normal surge test voltages, the detection of PD is an early warning because it is done at a voltage above the peak sinusoidal voltage. If action is not taken, the motor will eventually suffer from turn to turn arcs, and/or arcing to ground with complete breakdown.
Medium Voltage Motors and Inception Voltage
Medium voltage motors may have some partial discharge that is acceptable at normal surge test voltages but may also have none. If PD is detected, the question is, will it change over time? Since the PD amount is somewhat variable, it is not the best measure to track, at least not as the only value tracked.
The best value to track is the repetitive partial discharge inception voltage, or RPDIV.
The RPDIV is defined by the IEC 61934 standard as the voltage at which PD is detected in 5 of 10 consecutive surge pulses.
To find out what the inception voltage is, the surge test voltage is automatically ramped up in small steps, and partial discharge is measured at each voltage step. This is done for each phase in a 3-phase motor. The RPDIV is likely to be different from phase to phase. Maximum PD is also displayed and stored during this test.
If PD inception is reached, the key is to track it over time. If it does not decrease significantly, the PD is not harming the insulation or has not started to weaken it yet.
If the partial discharge inception voltage starts dropping, it is a warning that the insulation is breaking down, or that motor contamination is increasing. By its nature, there is some inherent randomness to the PDIV value so it is important to make sure the results are statistically significant by looking for a consistent, multi-point trend or a substantial decrease.
It is important to note that surface PD tends to increase with contamination and moisture on the windings. Therefore, it is best to do the PD tests during similar humidity conditions. Megohm results from insulation resistance tests should be taken into consideration in the analysis.
The iTIG motor analyzer with PD will automatically store all test results including PD, RPDIV and RPDEV so the results are easy to compare and track over time. This can be done with reports that are generated, by scrolling through past tests in the iTIG or in the TRPro report program on a PC, and by reviewing test summary data generated by the iTIG in an Excel spreadsheet or database program.