On almost every glass-lined reactor datasheet, high-voltage test appears as a single row in the parameter table: one number, usually 20 kV. It says nothing about what was tested, over what area, with what instrument, or what happened when a defect was found. That gap is what this article fills.
The test itself is a non-destructive electrical check. It takes minutes, leaves no permanent mark on a sound lining, and leaves no visible trace at all on the finished vessel. That last point is why the paper matters more than the number.
1. What the spark test actually measures
A glass lining is a dielectric bonded to a conductive steel substrate. The test set applies high voltage to a probe — normally a conductive brush or a flexible electrode swept across the surface — while the vessel shell is earthed. Where the glass is continuous, no current flows. Where there is a break that reaches the steel, the voltage jumps the gap and you get a spark and an alarm.
So the test answers exactly one question: is there any point on the tested surface where a conductive path runs from the surface through the lining to the steel? A pore, a crack, a chip, a thin spot at a weld toe, a burst bubble from firing, a scratch from fitting a baffle — all register the same way.
It does not measure lining thickness, adhesion, uniformity or chemical resistance. Those are separate questions with separate tests. A spark test is not a substitute for the jacket hydrostatic test, and it is not a lining quality certificate on its own.
2. Why 20 kV, and what the number does not mean
The chosen voltage has to sit above the level needed to bridge a defect that penetrates the full lining thickness, and below the level that would damage sound glass. For standard process vessels that band is where 20 kV falls. The margin is the whole design: a pinhole down to the steel conducts, a continuous lining does not.
Two conclusions follow, and buyers get both wrong.
- A higher number is not a better vessel. Asking for 30 kV or 40 kV "to be safe" can create damage that was not there, and can turn an acceptable lining into a rejected one.
- A lower number is not automatically acceptable. Too low a voltage passes over a defect in the thickest part of the lining. The correct figure is the one tied to the lining thickness on your drawing — confirm it against the applicable standard, state it in the purchase order, and do not accept it as a verbal understanding.
3. What the spark test cannot find
An honest list, because this is where projects get caught after commissioning.
- Sub-surface bubbles and blisters that do not yet reach the steel. These can open later under thermal cycling.
- Poor adhesion, where the lining is continuous but not bonded to the shell.
- Lining thickness variation, which drives thermal shock performance and service life.
- Chemical incompatibility. A lining can pass at 20 kV and still be the wrong glass for the medium.
- Damage that happens after the test — in handling, packing, transit, or during agitator and baffle installation on site.
On chemistry, the limits are fixed regardless of any certificate: standard glass lining is not suitable for hydrofluoric acid, fluoride-bearing media, phosphoric acid at 30 % or above at temperatures above 180 °C, or strong alkali above pH 12 at 100 °C. A passing test does not widen that envelope.
Thermal shock is the same story. A lining that passes at 20 kV still carries a cold shock limit of 110 °C and a heat shock limit of 120 °C. Exceed the quench and you crack a lining that was electrically perfect in the factory.
4. Test the small parts separately
The shell is only part of the wetted surface. Every glass-lined item that goes into the vessel is its own test object and needs its own line in the record.
- Agitator shaft and impeller, including the weld where the impeller meets the shaft.
- Baffles and baffle supports.
- Dip pipes, spargers and thermometer pockets.
- Underside of the top cover and the manway neck.
- Bottom outlet and valve seating area, where tool damage is most common.
- Any repaired zone, re-tested over the repair and a margin around it.
A certificate reading "vessel lining tested at 20 kV, passed" covers less than most buyers assume.
5. When the test is valid, and when it is void
Sequence decides everything. The test is valid only after the final firing, after all welding and machining on the vessel is complete, and after the lining has cooled. If any welding or mechanical work happens afterwards, the test is void for the affected area.
The normal order is: final firing, nozzle and manway finishing, visual glass inspection and repair, full-surface spark test, jacket and shell hydrostatic test, then packing. Ask for that sequence in writing, then cross-check the dates in the record. A test date that precedes the final weld date is a defect in the paperwork, not a rounding error.
One question worth asking explicitly: whether a second test is carried out after final assembly, and at what voltage. If a re-test is done at site, the voltage and procedure should be agreed with the manufacturer rather than copied from the factory sheet.
6. What a real test record contains
This document is the actual deliverable. It must be traceable to one vessel and impossible to confuse with another.
- Vessel serial number or model, matching the nameplate and the drawing.
- Test date and the name of the person who performed it.
- Instrument make, model, and the date of its last calibration.
- Test voltage applied and the electrode or probe type used.
- The areas tested — shell, heads, cover underside, nozzles, and each separate part, itemised.
- Any defect found, with its location, size and description.
- The repair method and the re-test result for that zone.
- Photographs of any defect and repair, and of the test in progress on the vessel itself.
- A signature, plus a third-party witness where one is written into the order.
Nine items. A one-page certificate with a voltage printed on it is a different document.
7. Why a verbal pass is worth nothing
The test takes minutes and leaves no trace on a good lining. If neither you nor your representative is in the factory, the paper is the only evidence that will ever exist. Twelve months later, when a pinhole at a nozzle weld opens and a batch discolours, the question becomes whether the record shows that point was tested, at what voltage, with an instrument in calibration.
Request the record before the balance payment, not after. A supplier who holds the record sends it the same day. A supplier who does not will offer reasons, and that is also information.
8. What to write into the purchase order
- Test voltage: 20 kV for standard glass lining, or the value confirmed on the drawing for non-standard thickness.
- Acceptance criterion: no sparking across the full wetted surface at the specified voltage.
- Scope: shell, heads, cover, and every glass-lined internal listed by name.
- Sequence: test after final firing and all welding, before packing.
- Record: the nine items above, signed, in English, traceable to the serial number, delivered before final payment.
- Re-test after any repair, with the repair documented in the same record.
Takeaway
The 20 kV figure is not the deliverable. The record behind it is. Before releasing the balance payment, confirm three things: the test voltage matches the lining on the drawing, the record items each glass-lined part rather than the shell alone, and the test date falls after the last weld and before packing. If any of the three is missing, you have bought a number, not a lining.
When you cannot be in Zibo on the day of the test, put someone in the room who is. That inspection, witnessed and documented, is the difference between a vessel that arrives with evidence and one that arrives with a claim.