How Does PD–DGA Correlation Reveal Transformer Health Risks?

Combining partial discharge monitoring with dissolved gas analysis gives transformer teams two independent views of the same insulation problem: electrical activity now and chemical evidence accumulated over time. When PD pulse behavior, hydrogen trends, operating load, and oil condition point to the same location and defect mechanism, maintenance decisions become faster, safer, and more defensible.

Multi-Modal Diagnostics within Predictive Maintenance through PD Monitoring

What Is the Value of Combining PD and DGA?

Partial discharge (PD) detects active insulation defects through electrical pulses, while dissolved gas analysis (DGA) identifies gases created by electrical and thermal degradation in transformer oil. Used together, they distinguish a real internal defect from noise, a temporary disturbance, or a misleading single-test result.

DGA is often the first warning that something inside an oil-filled transformer has changed. Hydrogen commonly rises with low-energy electrical activity, while acetylene is a more serious signal associated with arcing or high-temperature discharge conditions. PD testing adds the missing electrical signature: pulse magnitude, phase position, repetition rate, and probable defect location.

In field service, neither method should be treated as a stand-alone verdict. A transformer can show rising hydrogen with no measurable PD because the discharge is intermittent, deeply attenuated, or no longer active at the time of test. Conversely, an online PD system can report apparent pulses caused by corona, poor grounding, radio-frequency interference, switching noise, or cable reflections.

A combined diagnostic approach answers three practical questions:

• Is an electrical insulation defect active now?

• Has the defect produced chemical damage over weeks or months?

• Is the risk increasing quickly enough to justify outage, inspection, oil treatment, or internal repair?

For utility operators, industrial plants, transformer OEMs, and third-party test providers, correlation reduces costly false alarms. For a China manufacturer, supplier, or wholesale buyer, it also creates a clearer equipment specification: the PD system and DGA platform must share time stamps, alarm logic, and data-export capability.

How Do PD Pulses and Gas Patterns Correlate?

PD and DGA correlate when electrical pulse activity and gas-generation trends indicate the same fault mechanism within a similar operating period. Strong correlation does not require a perfect one-to-one match; it requires a technically credible relationship between pulse pattern, gas type, gas rate, load, temperature, and insulation condition.

The key is to compare trends rather than isolated readings. In our production support work, we have seen operators overreact to one elevated hydrogen value while ignoring a stable five-year trend. A single value can be influenced by sampling technique, oil processing, topping-up oil, load changes, or gas migration. The gas generation rate is often more useful than the absolute number.

For PD, look beyond apparent charge. A stable 300 pC pattern at the same phase angle can be less urgent than a 90 pC pattern that doubles in pulse count every week. Pulse repetition, phase-resolved distribution, polarity symmetry, and frequency content help separate internal void discharge from surface activity or external interference.

A useful correlation sequence is:

1. Confirm DGA sampling integrity and compare at least three historical samples.

2. Check whether hydrogen, methane, ethylene, acetylene, or carbon oxides are rising.

3. Conduct synchronized PD measurement under representative load and voltage.

4. Compare PD behavior with bushing tap results, moisture data, temperature, and recent switching events.

5. Escalate only when two or more independent indicators support the same defect hypothesis.

Which Gas and PD Combinations Require Immediate Action?

The most urgent combination is sustained or escalating internal PD activity together with rapidly increasing hydrogen and acetylene, especially when pulse location suggests windings, leads, tap-changer interfaces, or high-field oil gaps. This pattern may indicate progression from low-energy discharge toward arcing and demands immediate engineering review.

The following decision table is designed for practical dual-fault confirmation rather than automatic trip logic. Actual limits must be based on transformer voltage class, oil volume, design, service history, baseline gas values, and site risk.

In factory acceptance and commissioning work, the most expensive mistake is treating acetylene as a “wait and see” gas when the trend is accelerating. A small acetylene concentration may be historical, particularly after prior switching events or repair. But a repeated increase across two closely spaced samples, combined with new PD pulses, changes the risk category immediately.

HV Hipot recommends configuring alarms in stages: advisory, investigation, and urgent action. A single threshold alarm is too crude for a critical transformer fleet because it cannot distinguish stable legacy gas from active fault growth.

Why Can DGA and PD Results Disagree?

DGA and PD results can disagree because they measure different physical timelines: PD records present electrical activity, while DGA reflects gases generated, dissolved, transported, and sampled over time. Disagreement is not failure; it is diagnostic information that requires disciplined interpretation.

A PD event may be too weak, too brief, or too infrequent to produce a measurable gas response. This commonly occurs with surface discharge in air, external corona, or low-energy electrical noise. Conversely, gases can remain in oil after a past event even when the electrical source has stopped.

Oil temperature and circulation matter. In a large forced-oil-cooled transformer, gases from a winding fault may take time to mix through the tank. A sample taken from a low-flow valve can lag behind the active zone. In one service scenario, hydrogen appeared almost flat for six weeks despite increasing PD activity; the root issue was poor sample-point representativeness during low-load operation, not an absence of gas generation.

Common causes of mismatch include:

• Inadequate oil sampling, air ingress, sample-bottle leakage, or laboratory variation

• Online DGA sensor cross-sensitivity or delayed calibration

• PD sensor coupling limitations, electrical interference, and insufficient bandwidth

• A thermal fault producing gases without measurable discharge pulses

• A discharge occurring in a region inaccessible to the installed PD sensor

• Oil degassing, filtration, topping-up, or recent maintenance masking the natural gas trend

Do not force agreement by discarding the inconvenient result. Investigate the mismatch, document the operating state, and repeat targeted measurements.

How Should a Dual-Fault Decision Table Be Built?

A dual-fault decision table should combine fault evidence, trend direction, data quality, asset criticality, and required response. It must tell operators what to verify next, not merely assign a fault label from one gas ratio or one PD threshold.

For B2B transformer owners, the decision table should be agreed before an alarm occurs. Maintenance teams, control-room staff, test contractors, and management need consistent actions. A wholesale instrument package is only useful when its measurements enter an operational workflow.

Start with four evidence grades:

• Grade A: Validated data from calibrated PD and DGA instruments.

• Grade B: Repeated trend over two or more meaningful intervals.

• Grade C: Agreement between electrical, chemical, thermal, or acoustic indicators.

• Grade D: Asset consequence, including transformer redundancy, loading, and outage impact.

For example, a 110 kV transformer with moderate hydrogen but no PD may remain under enhanced observation if it has a redundant parallel unit and stable gas rate. The same readings in a heavily loaded generator step-up transformer with no spare capacity may justify immediate off-line diagnostics.

China-based manufacturers and OEM customers should also define data ownership, protocols, and report templates at the purchase stage. Specify Modbus, IEC 61850 gateway requirements where applicable, CSV export, time synchronization, alarm history retention, and access permissions. These details prevent a common commissioning failure: PD and DGA devices working independently but producing reports that cannot be correlated.

When Should Online Monitoring Replace Periodic Testing?

Online monitoring should supplement or replace periodic testing when transformer criticality, fault consequence, loading volatility, or historical defect evidence makes delayed laboratory data unacceptable. Periodic sampling remains suitable for lower-risk units with stable history, but it cannot capture fast-developing discharge events.

A practical selection rule is based on risk and fault speed. Critical generator transformers, intertie transformers, urban substations with limited redundancy, traction transformers, and units with prior abnormal gases benefit most from continuous monitoring. Lower-voltage distribution transformers may need only scheduled laboratory DGA unless their service duty is unusually severe.

At HV Hipot, we advise customers not to purchase every available sensor by default. More channels create more installation points, more calibration obligations, and more potential nuisance alarms. Instead, select the measurement architecture that addresses the known risk.

For example:

• A fleet with recurring bushing and cable-end issues may prioritize high-frequency PD sensing and location capability.

• A transformer fleet with thermal overload history may prioritize multi-gas DGA, top-oil temperature, cooling status, and load correlation.

• A transformer following internal repair may require both systems, with tighter initial trend intervals for the first 90 days.

A capable China factory partner should support OEM, custom, and private-label configurations, including sensor count, enclosure protection level, communications protocol, installation layout, and language requirements for reports.

Who Should Specify PD–DGA Equipment for a Project?

The equipment specification should be written jointly by the asset owner, protection or maintenance engineer, commissioning team, and qualified diagnostic supplier. Purchasing teams should not select PD–DGA systems only by sensor quantity, lowest wholesale price, or nominal measurement range.

The most effective projects involve the people who will interpret alarms. A procurement specification should require:

• Measurement method and sensitivity under defined site conditions

• Sensor compatibility with transformer design and available installation points

• Calibration documentation and traceable functional testing

• Noise-rejection strategy for substations with variable-frequency drives, communications systems, or switching equipment

• Data synchronization between PD and DGA records

• Commissioning acceptance criteria and baseline report

• Training for engineers who must distinguish an advisory event from an actionable fault

• Spare parts, remote support, and long-term service availability

HV Hipot supports these requirements as a high-voltage test equipment manufacturer based in China. From custom instrument selection through safe export packing and after-sales assistance, HV Hipot helps utilities, transformer factories, electrical contractors, and diagnostic service companies build a practical condition-monitoring workflow rather than assemble disconnected devices.

How Can Factories Prevent False PD–DGA Conclusions?

Factories can prevent false conclusions by establishing a clean baseline before shipment, controlling test interference, verifying oil sample traceability, and documenting every abnormal event during factory acceptance testing. The baseline is the reference point that makes later field data useful.

In our production runs, the most reliable baseline reports include ambient temperature, oil temperature, applied voltage, test duration, grounding configuration, sensor position, noise spectrum, oil batch, moisture result, and gas sample chain of custody. Without these details, a field engineer may mistake factory test artifacts for service deterioration.

Before releasing equipment, verify the following:

1. PD sensors are installed with repeatable polarity and labeled signal paths.

2. The measurement system records its noise floor before energization.

3. Oil samples are collected after sufficient mixing time and sealed without headspace contamination.

4. Any vacuum filling, oil filtration, heat run, or switching operation is recorded because it can alter gas interpretation.

5. The acceptance report identifies whether values are baseline observations or confirmed defects.

HV Hipot Expert Views

“A dual-diagnostic system earns its value when it prevents the wrong outage as well as when it detects the right fault. We have seen plants schedule unnecessary inspections because they treated a single PD alarm as proof of internal discharge. After checking phase pattern, grounding, and DGA trend, the source was external interference. We have also seen the opposite: modest PD readings paired with a rising hydrogen rate, where the trend revealed a developing defect before the apparent-charge value looked dramatic. The disciplined approach is simple: validate the signal, compare time-aligned trends, and act on converging evidence—not on the loudest instrument.”

What Are the Key Takeaways for Transformer Owners?

PD–DGA correlation turns two incomplete diagnostic methods into a stronger transformer health decision system. Use DGA to identify chemical evidence and fault progression, use PD to confirm active electrical behavior and likely location, and treat disagreement as a prompt for further investigation.

The most actionable advice is to establish a baseline, trend both data streams on synchronized time stamps, validate measurement quality, and define response levels before an alarm occurs. Select a manufacturer and supplier that can support custom monitoring architecture, commissioning documentation, reliable testing equipment, and long-term technical service.

HV Hipot provides integrated high-voltage testing and diagnostic solutions for utilities, transformer OEMs, industrial facilities, and electrical service teams seeking dependable equipment from a China manufacturer with OEM and wholesale support.

Can PD be detected before DGA gases increase?
Yes. Active PD may appear before enough gas is generated, dissolved, and distributed in the oil for DGA to show a clear increase. This is why repeat DGA and trend analysis are essential after a new PD finding.

Does high hydrogen always mean partial discharge?
No. Hydrogen can be associated with partial discharge, but it can also result from other low-energy electrical or thermal processes. Interpret it with methane, acetylene, moisture, loading history, and PD evidence.

How often should abnormal DGA results be retested?
For a credible rising trend or an unexplained abnormal result, retest within days or weeks rather than waiting for the normal maintenance interval. The exact interval depends on gas rate, asset criticality, and operating conditions.

Can a PD monitor locate the exact defect inside a transformer?
It can estimate the probable defect zone when sensor placement, signal quality, and timing methods are suitable. Confirmation may still require complementary tests, inspection, or outage-based investigation.