Advances in insulation properties have necessitated
changes to a longstanding industry standard
The types of insulation used in rotating machines have continued to evolve and precipitate changes to insulation resistance testing since 1974 when IEEE Std. 43, Recommended Practice for Testing Insulation Resistance of Rotating Machines was introduced, but it wasn't until 2000 that the standard received a long overdue facelift. The revised standard, IEEE Std. 43-2000, drastically changed a number of traditional testing programs for insulation resistance that had been in place for more than 50 years, including polarization index (PI), insulation to ground tests, and DC versus AC testing of insulation systems. This article will discuss each of the changes to IEEE Std. 43-2000, beyond the standard re-affirmations from 1974 to 2000.
The purpose of an insulation resistance (IR) reading is to evaluate the condition of the insulation between conductors and ground. This is done by applying a direct voltage between the conductors (windings) and the casing of the electric motor (machine) and measuring current leakage across the insulation system. The readings are applied to Ohm's law (R=V ÷I), which provides a resistance.
In the case of an insulation system, current may be measured in milli- or micro-ohms. The lower the current reading, the higher the insulation resistance reading. These IR readings change over time because of dielectric absorption. Basically, the insulation system consists of polarized atoms that line up (polarize) with the applied DC voltage. As they polarize, the insulation resistance will increase.
Polarization index. Prior to 1974, the industry relied on nonstandardized systems to evaluate the polarization of the insulation system. This led the IEEE to form a standards committee to address this situation. It was discovered that many insulation systems would polarize in anywhere from 10 minutes to several hours, so it was possible to produce a ratio using the insulation readings at the one-minute and 10-minute marks. This ratio could be trended or compared to a simple table that would provide an indication of the condition of the winding.
It was then possible to analyze PI by either providing the direct ratio or plotting the readings in steady increments over time. In the past, a PI ratio of less than 2.0:1.0 would indicate a problem with the insulation system; usually the insulation would be carbonized (burned) or would have absorbed contamination. A poor reading would indicate that additional testing was necessary.
With the release of IEEE Std. 43-2000, new limitations on the use and evaluation of PI were issued. This test is no longer as easy and straightforward as it once was because most new insulation systems start with test results in the Giga-ohm and Terra-ohm (billions and trillions of ohms) range. As a result, leakage current between the windings below the micro-ohm range require the instrument and test methods used, including how to position the conductors, to be extremely accurate. The readings must also be taken at the machine itself.
Readings taken from the motor control center (MCC) or disconnect may pick up high leakage currents in the cables caused by the cable surface area and produce errors. The testing at this point is so critical that the standard states, “If the one-minute insulation resistance is above 5,000 Meg-ohms, the calculated PI may not be meaningful. In such cases, the PI may be disregarded as a measure of winding condition.”
Additional concerns with PI testing are outlined within IEEE Std. 43-2000, including temperature limitations, dew-point limitations, trending tests, surface conditions, types of windings to be tested, and more. For instance, the tests must be performed above the dew point and corrected for temperature to 40°C. This is a change from the 1974 version of the standard, where it was assumed that PI wasn't a temperature-dependent test.
The main change has to do with the polarization of new insulation systems based upon temperature because unlike conductors where the resistance increases with temperature, insulation resistance is inversely proportional to temperature so insulation resistance decreases with temperature. Therefore, a PI performed on a hot insulation system may be dramatically different than a PI performed on a cold one (i.e., in storage). The only true way to trend the PI is to perform each test with the winding at about the same temperature, over the dew point. This may be very difficult in a production environment, where extended inactivity of the machinery will have adverse effects on corporate profitability; testing one machine can take as long as 15 minutes.
In addition, PI can only be used on form wound or random wound machines. Open winding machines, transformers, and certain insulation materials won't provide appropriate ratios and will fail a one-time test. In most cases, PI is now used for the “estimation of the suitability of a machine for the application of appropriate overvoltage tests…” according to the standard.
Insulation resistance. Insulation resistance readings have been used to troubleshoot and evaluate the condition of electric motors since the test method was first introduced in the 1800s. Unfortunately, the test often failed to detect many incipient and existing faults.
IR tests have very clear limitations when it comes to evaluating the condition of an electric motor for operation. For one thing, the fault has to have a direct path between the windings and casing of the machine. Air, mica, or any other non-conducting material between the winding and ground will provide a high insulation resistance. Faults on the end-turns of motor windings will also fail to provide a clear path to ground. Most winding faults start out as internal winding shorts and may graduate to insulation faults, although not always.
An excerpt from IEEE Std 43-2000 does a good job of explaining the test's limitations and noting safe minimum values for testing (Table):
“Insulation resistance test data is useful in evaluating the presence of some insulation problems such as contamination, absorbed moisture or severe cracking; however, some limitations are as follows:
“a) IR of a winding is not directly related to its dielectric strength. Unless the defect is concentrated, it is impossible to specify the value of insulation resistance at which the insulation system of a winding will fail.
“b) Windings having an extremely large end arm surface area, large or slow speed machines, or machines with commutators may have insulation values that are less than the recommended value.
“c) A single IR measurement at one particular voltage does not indicate whether foreign matter is concentrated or distributed throughout the winding.
“d) Direct voltage measurements, such as IR and PI tests, may not detect internal insulation voids caused by improper impregnation, thermal deterioration or thermal cycling in form wound stator coils.”
DC versus AC voltage testing of insulation. DC voltage testing is normally accomplished by applying the DC source directly across the insulation system between the windings and stator frame of the machine. But the resistivity values of dirt, oil, and water that often contaminate the end winding areas of rotating machinery are quite low. Therefore, direct-voltage testing of a contaminated winding normally results in a high surface leakage current and low resistance readings. This is what makes DC voltage testing a viable method for determining the extent of contamination to an insulation system.
In addition, if the insulation system uses a cotton-backed tape with mica as the primary electrical insulation, a DC voltage test may or may not reveal if the cotton has absorbed moisture. However, most windings manufactured after 1970 don't have hygroscopic tapes, and a DC voltage test won't normally detect problems internal to the insulation system, such as thermal deterioration.
The standard states, “Since the primary electrical insulation used in the design of form wound stator windings is mica, and mica has virtually infinite resistivity (thus a good insulator), only one layer of mica tape would prohibit any direct current. Therefore, if a void exists within the insulation system due to improper impregnation, thermal deterioration, or thermal cycling, a DC voltage test would be unable to detect it. If, however, there exists a severe crack through the entire insulation system, it's possible that an electrical track would be established between the copper conductors and ground, and would appear as a low resistance.”
It also says, “When a[n] alternating voltage is connected between the terminals of the test specimen [machine]… the capacitance of the test specimen dominates the current. Capacitance is determined by [the following] equation, C = eA/D, where C is capacitance, e is dielectric permittivity of the material, A is the cross-sectional area, and D is the thickness of the material.
“Since the dielectric permittivity of an insulation system is greatly effected by the presence of voids and/or water, an alternating voltage test is more sensitive than direct voltage tests with regard to detection of internal insulation problems associated with all types of insulation systems. Because of the different test capabilities, both a DC and an AC test should be conducted to more completely assess the condition of an insulation system.”
With this in mind, an impedance-based test (impedance includes resistance, inductance, capacitance, and frequency) as opposed to other test results, such as resistance and inductance, could be used as the AC test. A standard IR test could be used to accomplish the DC test. If you take the inductance across three phases of a 3-phase machine and compare it to the impedance, changes to capacitance in one or more phases would have a negative effect on the impedance, causing the value to fall toward inductance. The DC test would identify continuity to ground, while a comparison of impedance to inductance would indicate additional winding problems.
IEEE Std. 43-2000 addresses some new issues regarding modern insulation systems. In particular, it identifies the complexities of performing PI and IR tests and notes their limitations. It also notes an acceptable alternative: combined AC and DC testing.
Penrose is the general manager for the All-Test Pro division of BJM Corp. in Old Saybrook, Conn.