Pull-Out Testing vs Proof-Load Testing: Choosing the Right Method for an Occupied Roof

Proof-load testing and pull-out testing are not interchangeable names for the same procedure. They answer different engineering questions, they carry different risk profiles for the structure and its occupants, and applying one where the other is warranted produces results that are either meaningless or dangerous. For structural engineers and height-safety designers specifying anchor testing on occupied roofs, the distinction matters at every stage: brief writing, mobilisation, interpretation, and certification.

What Each Test Actually Does

Proof-load testing applies a defined load to an installed anchor point and holds it for a specified duration, typically two minutes under AS/NZS 1891.4 or as nominated by the design documentation. The anchor either holds without exceeding a displacement threshold or it does not. The test is non-destructive by intent. The load applied is a fraction of the anchor's theoretical capacity, chosen so that a compliant installation survives without damage while a deficient one reveals itself. A point that passes a proof test is certified as fit for the service load it was designed to carry. Nothing more.

Pull-out testing, sometimes called ultimate load testing or characterisation testing, takes the anchor toward or to failure. The objective is to measure actual capacity: the load at which the anchor, the substrate, or the connection between them ceases to perform. This is destructive or near-destructive work. The anchor tested is not returned to service. The data produced informs design, validates substrate assumptions, or investigates a failure mode. It does not certify a point for ongoing worker use.

These two test types sit at opposite ends of the purpose spectrum. Proof load is a serviceability check on a specific installed point that a designer has nominated for a specific duty. Pull-out is a characterisation exercise, often applied to sacrificial anchors or to a statistical sample where the design itself is under investigation.

When Pull-Out Testing Is Appropriate

Pull-out testing is the right tool in several situations that arise regularly on Australian projects.

Where a substrate's actual properties are unknown or disputed, testing a sample of installed anchors to failure gives the designer real capacity data rather than conservative assumed values. This is common in older concrete structures where compressive strength records are absent, in masonry substrates where bond strength varies across a facade, and in rock anchors where geological variability affects grout-to-rock interface strength.

Where a new anchor system is being qualified for a substrate type not covered by existing approval documentation, characterisation testing under EOTA TR 054:2016 or AEFAC TN05 methodology establishes the statistical basis for a design resistance value. The simplified method and statistical method described in those documents both require pull-out data from a defined sample size.

Where an anchor has failed a proof test and the engineer needs to understand whether the failure was installation-specific or substrate-wide, pull-out testing of adjacent sacrificial anchors can distinguish between a localised installation defect and a systemic problem with the base material.

ATA's ultimate load testing service covers all three scenarios, with displacement recorded continuously throughout the load ramp so that the load-displacement curve is available for analysis, not just the peak load value.

When Proof-Load Testing Is Appropriate

Proof-load testing applies to anchors that have been designed, installed, and nominated for a specific duty. The test verifies that the installation meets the design intent. It is the standard method for certifying height-safety anchor points under AS 5532 and for confirming post-installed fasteners under AS 5216:2021.

The test load is typically 6 kN for single-person fall-arrest points under AS/NZS 1891.4, though the design documentation may specify higher loads for multi-person systems, horizontal lifelines, or anchors carrying both fall-arrest and restraint duties simultaneously. The hold period and displacement acceptance criteria come from the applicable standard or the project specification, not from the testing contractor's discretion.

ATA's proof-load testing service applies calibrated hydraulic equipment with continuous displacement monitoring throughout the hold. The displacement record matters because a point that holds the load but creeps 8 mm during the two-minute hold is telling the engineer something that a pass/fail load cell reading alone does not capture. Displacement behaviour under sustained load is particularly informative for chemical anchors in warm or wet conditions and for anchors installed in low-strength substrates near their design limit.

The Point ATA Will Not Proof-Load

This is where the distinction between the two test types has direct consequences for how a testing brief should be written.

ATA will not apply a proof load to an anchor point that was never nominated in a design. The reason is not procedural caution. It is that proof-load testing a point without a design basis gives the client a result they cannot use and may give them false confidence in a point that was never engineered for the duty they intend to assign it.

Consider the situation: a maintenance contractor identifies an existing eyebolt on a roof. No drawings are available. The eyebolt appears to be in good condition. The contractor wants it proof-loaded so workers can clip to it. If ATA applies 6 kN and the eyebolt holds, what has been demonstrated? That the eyebolt survived 6 kN for two minutes. What has not been demonstrated: the substrate beneath it, the embedment depth, the edge distance, the thread engagement, the steel grade, or whether the load path through the fixing plate is adequate. A proof test on an undesigned point is not a substitute for engineering.

The correct process for an anchor with no design basis is characterisation first. Pull-out testing of a sample establishes what the substrate and installation can actually carry. An engineer then reviews that data, determines whether the capacity is adequate for the intended duty, and either nominates the point for service or specifies remediation. Only after that process is complete does proof-load testing of the nominated points make sense.

This sequence protects the engineer, the building owner, and the workers who will eventually clip to those points.

Displacement Records on an Occupied Roof

Occupied roofs introduce constraints that affect both test types. The structure below is in use. Loads applied to anchors travel through the substrate and into the building frame. For proof-load testing, the loads involved are modest and the risk of structural distress is low provided the anchor was designed for the duty. For pull-out testing on an occupied roof, the situation requires more care.

Pull-out tests that approach or reach failure loads can produce sudden energy release if the anchor fails in a brittle mode. Cone pull-out failures in concrete and grout-plug failures in chemical anchors are not slow, graceful events. The testing programme needs to account for this, both in terms of personnel positioning and in terms of the structural consequences of the load being applied to a live floor or roof system.

Displacement monitoring during both test types provides data that goes beyond pass/fail. A load-displacement curve from a proof-load hold that shows progressive creep rather than a stable plateau is a warning that the anchor is working harder than expected, even if it technically passes the acceptance criterion. That information belongs in the certification report and should inform the inspection interval assigned to that point.

For pull-out tests, the full load-displacement curve characterises the failure mode. A curve that shows a long linear phase followed by a sharp drop indicates brittle failure. A curve with a long plastic plateau before failure indicates ductile behaviour. The failure mode affects how the designer interprets the test results and what safety factors are appropriate.

Writing the Testing Brief

For structural engineers and height-safety designers, the testing brief is where these distinctions are locked in. A brief that says only "test all anchors" leaves the testing contractor to make decisions that belong with the engineer.

A well-written brief specifies:

• Which anchors are nominated for proof-load testing and what design load applies to each
• Whether any anchors are to be tested to ultimate load, and if so, which ones are sacrificial
• The applicable standard for each test type
• The displacement acceptance criteria, if different from the standard default
• The reporting format required, including whether RPEQ review is needed for the certification
• Any site constraints that affect mobilisation, access, or sequencing

ATA works from the brief. Where the brief is ambiguous about test type, ATA will query it before mobilising, because applying the wrong test type to an anchor on an occupied roof is not a recoverable situation.

Standards That Govern Each Method

The applicable standards differ by test type and anchor application.

For proof-load testing of height-safety anchor points, AS 5532 and AS/NZS 1891.4 are the primary references. For post-installed fasteners in concrete, AS 5216:2021 sets out the design basis that the proof test is verifying.

For characterisation testing, AEFAC TN05 volumes cover the statistical and simplified methods for determining design resistance from test data. EOTA TR 054:2016 applies where European assessment methodology is referenced in the design. BS 8539:2012+A1:2021 provides additional guidance on installation and testing of post-installed anchors in masonry and concrete.

Referencing the correct standard in the brief is not a formality. The standard determines the test load, the hold period, the displacement criteria, and the reporting requirements. A certification report that references the wrong standard is not a valid certification.

Getting the Sequence Right

The practical takeaway for engineers specifying anchor testing on occupied roofs is this: proof-load testing confirms what a design has already established. It does not replace the design step. Where there is no design basis for an anchor point, characterisation testing comes first, engineering review follows, and proof-load testing of the nominated points comes last.

ATA operates across Australia from its Brisbane base, with mobilisation to other cities by booking. For projects where the test type is uncertain or the existing documentation is incomplete, early engagement with the testing contractor allows the brief to be written correctly before mobilisation rather than revised on site.

For more on how ATA approaches each method, see the proof-load testing service page and the ultimate load testing service page, or contact the team at [email protected] or +61 7 3132 2534.