The physical difference that drives every other consideration is the way stress distributes inside insulation under direct voltage. In an alternating field, stress distributes according to the permittivity of the materials, which is the distribution the insulation was designed around. In a direct field, it distributes according to resistivity, which is a different profile and can be markedly more severe at particular interfaces.

Where DC testing is still correct

The clearest place for direct voltage is measurement rather than withstand. Insulation resistance and polarisation index testing applies a defined direct voltage and measures the resulting current, and it is a condition check used on transformers, machine windings, cables and switchgear alike. Nothing about the space charge problem applies, because the applied voltage is low and the measurement is comparative.

Surge arrester testing is a second established use. Arrester reference voltage and leakage current measurement is performed with direct voltage, because the quantities being measured are defined in direct terms and are used to derive the arrester condition. That is a measurement characteristic rather than a stress test, and the method follows from what is being measured.

A third group covers legacy and specific asset types where the applicable standard still defines a direct voltage test. Paper-insulated lead-covered cables behave differently from extruded insulation, and specifications for certain asset classes continue to include a direct voltage test. Where that is the case, the test is defined by the document and the justification is simply that the document requires it.

Why DC stresses some insulation types differently

Composite insulation systems contain interfaces: between oil and paper, between different dielectric materials, between insulation and semiconductive screens. Under alternating voltage, the stress at each interface is set by the permittivity ratio, which is a stable property of the materials. Under direct voltage, the stress is set by the resistivity ratio, and resistivity is strongly dependent on temperature and moisture.

The practical consequence is that a direct voltage test can produce a much higher stress at a particular interface than the same nominal voltage would produce in service, and the excess depends on conditions rather than on design. A test that applies a modest multiple of the operating voltage may therefore stress one part of the insulation far beyond what its design margin assumed.

In a purely resistive insulation that behaviour is manageable and in some cases useful. In a heterogeneous insulation system, and particularly in one that contains an extruded polymer, it is the mechanism behind the damage that follows.

Space charge and its delayed effects

Space charge is what happens when charge injected during a direct voltage test does not leave when the voltage is removed, but remains trapped at interfaces and within the polymer. The trapping is characteristic of extruded insulation, and the charge distribution that forms depends on the test voltage, the duration and the temperature.

The delayed effect appears when the cable returns to service. The trapped charge adds to the field produced by the alternating voltage, and at points where the charge is concentrated the resulting field can exceed the design level. The result is localised degradation that begins after the test rather than during it, which is why a DC test can pass cleanly and the cable can still fail months later.

The effect accumulates across tests. A cable that is given a DC test at every maintenance interval receives a series of charge injections, and each one leaves a distribution that the next test builds on. That is the reason the practice was withdrawn for extruded cable rather than merely limited.

Maintenance testing on legacy assets

Legacy assets present the hardest decisions, because they were installed under standards that are no longer current and are maintained by teams working to specifications that may not have been updated. A cable that has been in service for forty years on a paper insulation system is not the same problem as a new extruded cable, and applying a modern method to it may not be better.

The practical approach is to follow the asset’s own specification and to record the basis. Where the specification calls for a direct voltage test, perform it and record the document. Where the specification is silent, the decision belongs to the asset owner with an explicit statement of the method chosen and why, and that statement should consider what the alternative methods would have shown.

The transition case is the most delicate: a mixed route containing both older paper-insulated sections and newer extruded sections. A single test method applied to the whole route will be wrong for one part of it. Segmenting the route, or choosing the method appropriate to the dominant insulation type and stating the limitation, are the two defensible options.

Legal and contractual contexts for DC

In some contexts the method is fixed by contract rather than by engineering. A specification written years ago may name a direct voltage test, and a contractor who substitutes a different method is in breach regardless of the technical merits. The reverse also happens: a specification that has been updated to require an alternating method may be presented on a project where the equipment available cannot deliver it.

Both situations are resolved the same way, by raising the method question before the test rather than after it. Where a contractual test is technically inappropriate, the request for a variation should be documented with the technical reasoning and with the standard that supports the alternative. Where the alternative cannot be mobilised, the record should state what was done and identify the deviation explicitly, so that the result is not later presented as if it satisfied the contract.

The two sides of this argument are supported by documents rather than opinions. The field test requirements for the cable classes most affected are set out in IEC 60502-2 for the medium voltage range and in IEC 60840 for the higher voltage range, and the general framework for field testing of high-voltage equipment is in IEC 60060-3.

Deciding between DC, VLF and AC

The decision comes down to three questions: what the insulation type is, how capacitive the load is, and what the applicable standard permits. Answering them in that order resolves most cases without further discussion.

For extruded cable, the insulation type rules out direct voltage, and the capacitance decides between VLF and resonant testing. For GIS and machine windings, the requirement for an alternating test is fixed by design, and the capacitance decides how the alternating voltage is generated. For legacy paper-insulated cable, the applicable specification decides, and where it permits direct voltage there is no reason to change the method for its own sake.

Where the answer is genuinely ambiguous, the tie-breaker is what the test is for. A pass-or-fail withstand test on a new installation should use the method that imposes the stress the insulation was designed for. A condition check on an ageing asset should use the method that carries the least risk of accelerating degradation while still producing information. Those two purposes can point to different methods on the same asset at different points in its life.

What the standard actually requires

The standards do not remove direct voltage testing from the toolbox; they define where it belongs. For extruded cable classes, the after-installation and maintenance test requirements are written around alternating methods, which is why a direct voltage test on those assets is difficult to justify regardless of how well it has been performed.

For assets where direct voltage remains in scope, the standard defines the test voltage, the duration and the acceptance criterion. Reading those three values from the document rather than choosing them on site is what makes the test defensible, and it is also what makes the result comparable with the next test on the same asset.

There is a further requirement that is easy to miss: the discharge procedure. After a direct voltage withstand test, the stored energy in the test object and in the test set has to be discharged in a controlled way, and the residual charge has to be allowed to dissipate before connections are touched. The procedure belongs in the plan, not in the operator’s memory, and the general high-voltage test technique framework in IEC 60060-1 is where the definitions the procedure relies on are established.

Recording the justification for the method

Since the choice of method has become the point of contention, the record has to carry the reasoning rather than only the outcome. The minimum set is the asset identification and insulation type, the method used, the standard or specification that permits it, the test voltage and duration, the polarity, the measured leakage current, the discharge procedure and the conditions observed during the test.

Added to that, a short statement of alternatives considered makes the record complete. A line saying that an alternating method was considered and not available within the outage, or that the applicable specification requires the direct voltage test, answers the question that will otherwise be asked later.

Where the test was performed on an asset where direct voltage is not the preferred method, the record should say so plainly and should identify the residual risk in terms of what the method may have done to the insulation rather than only in terms of whether the test passed. That is an uncomfortable note to write, which is precisely why it is the note that protects the asset owner. Method documentation from high-voltage test equipment suppliers such as Haefely describes how direct voltage sets are configured and discharged, and the field experience behind current cable testing practice is coordinated through CIGRE study committees. Cable testing and fault location work is grouped under cable fault testing, within the wider high-voltage insulation withstand test range.