Vacuum Box Testing for Tank Floors and One-Side-Access Welds

Most leak testing depends on pressurising one side of the boundary, but tank floors, roof plates and many weld seams cannot be pressurised at all, because there is no closed volume to fill. itmad provides vacuum box testing for storage tank floors, shell-to-bottom welds and one-side-access seams, by bubble emission to ASME V Article 10, ASTM E515 and API 650, carried out by ISO 9712 qualified technicians. The box reverses the differential: it is sealed over the wetted weld and evacuated, so atmospheric pressure on the far side drives air through any through-wall defect and forms bubbles in the solution. itmad operates across the Middle East, Asia and Europe. Last reviewed: October 2026.

How the test is performed

The weld and adjacent surface are first cleaned of scale, slag, oil and coating that would trap solution or obscure bubbles. A leak indicating solution is applied to the test area, thick enough to form and hold a bubble but not so thick that it masks a small leak. The box, fitted with a transparent window and a compliant gasket, is placed over the area and evacuated to the partial vacuum that the procedure specifies, typically 20 to 35 kPa below atmospheric.

The area is observed through the window for the full hold period, with the gauge reading verified as stable. Boxes are overlapped along the weld by at least 50 mm so that no length is left unexamined between placements, indications are marked on the plate for repair, and the area is retested after rework.

Vacuum level and hold time

Both are set by the governing code or procedure and not by convenience. A partial vacuum below the specified level will not draw air through a tight defect, and an observation period shortened to keep pace with production is the most common reason a leak is missed, because small leaks take time to form a visible bubble. Gauge readings and hold times are recorded for each placement, and the gauge is verified against a calibrated reference.

Where it applies

Storage tank floors are the classic application: bottom plate lap welds and butt welds, where the underside is inaccessible and a leak means product to ground, and the test is required under API 650 for new construction and API 653 for in-service inspection. The annular shell-to-bottom weld at the tank periphery is tested during construction and at out-of-service inspection, and fixed and floating roof welds are tested where vapour tightness matters. The method also verifies repair welds after plate replacement or weld repair before the tank returns to service, and is used on barges, decks and bulkheads where one side is inaccessible or cannot be flooded, and on seams in secondary containment and lined surfaces.

Standards applied

ASME Boiler and Pressure Vessel Code Section V, Article 10, applies, including the appendix covering the vacuum box technique, with ASTM E515 for leaks using bubble emission techniques, API 650 for welded tanks for oil storage (construction testing) and API 653 for tank inspection, repair, alteration and reconstruction (in-service work). Client and project specifications apply where they set tighter requirements.

Advantages and limits

The method is inherently safe, with no pressurisation, no stored energy and no radiation source, and it locates the leak precisely and does not merely confirm that one exists. It needs no power at the test location beyond the vacuum pump, and the results are immediate and visible.

Its limit is sensitivity. Bubble methods detect leaks orders of magnitude larger than tracer gas techniques can resolve, so where the acceptance criterion is a small quantified leak rate, helium leak testing is the appropriate method. The box also needs a reasonably flat surface for the gasket to seal, which restricts it at nozzle penetrations, corners and heavily distorted plate, and those areas need an alternative technique; the report identifies them.

Personnel

Testing is carried out by technicians qualified to ISO 9712 and ASNT SNT-TC-1A Level II, at tank farms, terminals, fabrication yards and marine facilities.

What the client receives

A report that identifies the item, the weld seams tested and the total length examined, with the vacuum level achieved and the hold time applied at each placement. Indications are located and marked with photographic evidence, areas that could not be tested by this method are listed with the reason, and retest results are provided following repair. The report includes gauge calibration traceability and the technician's qualification.

Frequently asked questions

What is vacuum box testing?

A bubble emission leak test for weld seams accessible from one side only, such as tank floors, roof plates and shell-to-bottom welds that cannot be pressurised. A sealed transparent box is placed over the wetted weld and evacuated, and atmospheric pressure on the far side drives air through any through-wall defect, forming visible bubbles.

What vacuum level and hold time are required?

Both are set by the governing code or procedure, typically 20 to 35 kPa below atmospheric. A shortened observation period is the most common reason a leak is missed, because small leaks take time to form a visible bubble.

What is it used for on storage tanks?

It is the standard method for floor welds under API 650 for construction and API 653 for in-service inspection. It covers bottom plate welds, shell-to-bottom annular welds, roof plate seams and repair welds before the tank returns to service.

Which standards apply?

ASME Section V Article 10 including its vacuum box appendix, ASTM E515, API 650 and API 653. Client and project specifications apply where they set tighter requirements.

What are the limitations?

It detects leaks orders of magnitude larger than tracer gas methods can resolve, so helium leak testing is more appropriate where a small quantified leak rate is the criterion. The box also needs a reasonably flat surface, which restricts it at nozzles, corners and distorted plate.

How much do the box placements overlap?

By at least 50 mm along the weld, so that no length is left unexamined between placements. Gauge readings and hold times are recorded for each placement.