Generator Dielectric Loss Analysis: Tan Delta Testing for Stator Winding Insulation

Generator dielectric loss analysis, commonly known as tan delta (dissipation factor) testing, is one of the most effective diagnostic techniques for evaluating the condition of stator winding insulation in turbine generators, hydro generators, and other large rotating machines. By measuring the dielectric dissipation factor and its voltage tip-up characteristic, this analysis reveals thermal aging, moisture ingress, delamination, and internal void development long before catastrophic insulation failure occurs. This article explains the underlying principles, applicable standards, test procedures, acceptance criteria, and practical applications of dielectric loss analysis for generator insulation systems.

Technical Overview

Generator dielectric loss analysis measures the dielectric dissipation factor, or tan delta (tan δ), of the stator winding insulation system. When an alternating voltage is applied to a generator winding, the insulation does not behave as a perfect capacitor: the total current consists of a capacitive component and a small resistive (loss) component caused by ionic conduction, dipole polarization, and internal partial discharges. The dissipation factor is defined as the ratio of the resistive current to the capacitive current, and it quantifies how much electrical energy is converted to heat within the insulation per cycle.

For a healthy modern epoxy-mica insulation system, the dissipation factor at low voltage is typically below 1–2%. As the insulation ages or degrades, the dielectric loss increases due to moisture absorption, thermal deterioration, contamination, delamination of the tape layers, and the growth of internal voids. A particularly valuable diagnostic parameter is the "tip-up," defined as the increase in tan delta as the test voltage is raised, for example from 0.2 times rated voltage to 0.6 or 1.0 times rated voltage. A large tip-up indicates significant internal discharge activity caused by voids or delamination, because partial discharges only ignite once the voltage across the gas-filled cavities exceeds their inception level.

Dielectric loss analysis is covered by several international and national standards, including IEC 60034-27-3 for dielectric dissipation factor measurements on stator winding insulation of rotating electrical machines, IEC 60894 for loss tangent measurements on individual coils and bars, IEC 60250 and GB/T 1409 for dielectric property determination of insulating materials, and IEEE Std 286 for power factor tip-up measurements on stator winding insulation. Compared with partial discharge testing, dielectric loss measurement provides an averaged, global picture of the insulation condition, making it well suited for trending the state of an entire winding over the service life of the machine.

Testing Samples

Individual Stator Bars and Coils

New stator bars and coils removed from the manufacturing line, or samples taken during a rewind project, are the most common test objects. Individual bar testing in the factory or at an accredited third-party laboratory provides the most reproducible results because the test geometry is well defined, the bar can be fully shielded, and guard electrodes can be applied to eliminate surface leakage effects.

Complete Stator Windings

For generators already in service, dielectric loss measurements can be performed on the complete stator winding on site. Each phase is tested separately while the remaining phases and the stator core are grounded. Although whole-winding measurements are less sensitive than individual bar tests because the losses of many bars are averaged together, they are extremely valuable for periodic trending of insulation condition during preventive maintenance outages.

Insulation Material Specimens

Raw insulation materials, such as mica tape composites, epoxy resins, and laminate plates, are tested as flat specimens using standardized electrodes to qualify materials and characterize their dielectric properties. This supports insulation system design, supplier quality control, and research into new insulation formulations.

Model Coils and Insulation System Components

Model bars, mock-up coils, and supporting insulation components such as bushings and inter-phase barriers may also be evaluated to validate an insulation system design before it is applied to a full production machine.

Testing Items

The scope of a dielectric loss analysis program typically includes the following measured and derived parameters:

• Dissipation factor tan delta at a low reference voltage, normally 0.2 times rated phase-to-ground voltage, which reflects the baseline conduction and polarization losses of the insulation.
• Tip-up Δtan delta, defined as the difference between the dissipation factor at an elevated voltage (commonly 0.6 or 1.0 times rated voltage) and the value at the reference voltage, which quantifies partial-discharge-related losses in internal voids.
• Per-step tan delta increment, that is, the increase in dissipation factor between adjacent voltage steps, used to detect the onset of intensive internal discharges.
• Capacitance of the test object and its voltage dependence, since a significant increase in capacitance with voltage is another indicator of void ionization.
• The complete tan delta versus voltage characteristic, plotted from approximately 0.2 up to 1.0 times rated voltage.
• Optional temperature-dependence measurements to evaluate the thermal behavior of the insulation and to correct results to a reference temperature.

For new epoxy-mica stator bars, typical acceptance criteria require the dissipation factor at 0.2 times rated voltage to be no greater than approximately 2%, the tip-up between 0.2 and 0.6 times rated voltage to be no greater than approximately 0.5%, and the increment between adjacent voltage steps to be no greater than approximately 0.3%. For aged machines, higher absolute values and tip-ups are tolerated, but a clear upward trend between consecutive outages is considered a warning sign. The exact limits should always be agreed between the manufacturer, the purchaser, and the testing organization based on the applicable specification.

Testing Methods

Schering Bridge Method

The classic Schering bridge balances the test object capacitance against a reference capacitor using adjustable resistive and capacitive arms. The dissipation factor and capacitance are read directly from the balance settings. With a guard circuit connected to a guard electrode on the specimen, surface leakage currents are excluded from the measurement, which significantly improves accuracy, particularly in humid environments.

Digital Dielectric Loss Analyzer Method

Modern digital instruments measure the capacitive and resistive current components directly using high-resolution phase-sensitive detection and fast Fourier transform algorithms. Digital analyzers offer automatic bridge balancing, wide measurement ranges, built-in safety interlocks, and direct data logging for trending purposes. Both the Schering bridge and digital methods are accepted by IEC 60034-27-3 and IEEE Std 286 provided the instruments are regularly calibrated and the stray capacitance of the test setup is compensated.

Test Procedure

A typical dielectric loss measurement on a generator winding or bar proceeds as follows:

• Isolate the test object, disconnect external connections, discharge and ground all conductors, and clean the insulation surface of the measuring ends.
• Record the ambient conditions and the winding temperature; if the temperature deviates significantly from the reference temperature defined by the specification, apply the appropriate temperature correction factors.
• Apply a starting test voltage of approximately 0.2 times rated phase-to-ground voltage and record the tan delta and capacitance readings once they stabilize.
• Increase the voltage in defined steps, for example 0.2, 0.4, 0.6, 0.8, and 1.0 times rated voltage, recording the tan delta and capacitance at every step. For service-aged machines, the maximum test voltage is often limited to the normal operating voltage.
• Reduce the voltage through the same steps to check for hysteresis, which may indicate discharge-induced changes during the test.
• Calculate the tip-up values and per-step increments, compare the results with the acceptance criteria, and archive the data for long-term trending.

On complete windings, each phase is tested separately with the other two phases and the core solidly grounded. For directly water-cooled windings, the cooling circuit should be drained and dried according to the machine specification before measurement, because residual water in the insulation hoses can distort both the capacitance and the dissipation factor readings.

Applications

Generator dielectric loss analysis serves a wide range of engineering purposes across the life cycle of a machine:

• Manufacturing quality control: routine tan delta and tip-up tests on every production bar verify the uniformity of the taping and impregnation process and confirm that the void content within the insulation is within specification.
• Acceptance testing: new generators and rewound stators are subjected to dielectric loss measurements before energization to document the initial insulation condition as a baseline for future comparison.
• Preventive and predictive maintenance: periodic whole-winding measurements, typically every major overhaul, allow maintenance engineers to trend insulation aging and plan repairs before failures occur.
• Post-incident investigation: after events such as cooling water leaks, overheating, fire proximity, or contamination ingress, dielectric loss testing provides a rapid quantitative assessment of insulation damage.
• Aging assessment and residual life estimation: combined with partial discharge measurements, insulation resistance tests, and visual inspection, dielectric loss data supports end-of-life decisions for large hydro and turbo generators.
• Insulation system research and development: manufacturers of insulation materials use dielectric loss characterization to optimize resin formulations, mica content, and curing processes for improved thermal endurance.

FAQ

What does a high tan delta value indicate in a generator?

A high dissipation factor at low voltage usually indicates increased conduction losses caused by moisture absorption, surface contamination, or advanced thermal aging. A high tip-up between voltage levels specifically points to internal voids, delamination, or poor impregnation, because these defects generate partial discharges whose energy loss adds to the measured tan delta only at higher voltages.

Can the test detect a single defective bar in a complete winding?

Whole-winding measurements average the losses of all bars in the phase, so a single severely degraded bar may be masked by the healthy remainder. Individual bar testing during a rewind, or complementary methods such as partial discharge mapping, capacitor coupler testing, and visual inspection, are used when localized defects must be pinpointed.

How often should dielectric loss analysis be performed?

For large generators, measurement at every major overhaul, typically at intervals of three to eight years depending on the maintenance strategy and machine criticality, is common practice. More frequent testing is recommended after abnormal operating events, cooling water leaks, or when trending data show a significant deterioration compared with previous results.

What factors can distort the measurement results?

Ambient humidity and surface contamination, winding temperature, residual water in directly cooled hoses, incomplete isolation of adjacent phases, and uncompensated stray capacitance of test leads can all bias the readings. Surface cleaning, guard techniques, temperature correction, and instrument calibration are therefore essential parts of a reliable test protocol.

Is dielectric loss testing destructive to the insulation?

No. The test voltages are limited to the normal operating range of the winding, and the applied stress is comparable to, or lower than, everyday service conditions. The test is considered nondestructive and can be repeated safely throughout the service life of the machine.

How does dielectric loss analysis compare with partial discharge testing?

Dielectric loss testing gives a global, averaged figure of merit that is easy to trend and interpret, making it ideal for tracking overall insulation aging. Partial discharge testing provides location-sensitive information about individual defect sites. The two methods are complementary, and comprehensive condition assessments of generator insulation, as recommended by IEC 60034-27-3 and IEEE standards, normally combine both techniques along with insulation resistance and polarization index measurements.