Seismic retrofit of URM structures (US)

Why URM buildings fail in earthquakes - and what the data shows

Robert Hudson opened by showing two buildings in Christchurch, New Zealand, photographed shortly after the 2011 earthquakes. One was unretrofitted and no longer standing. The other had been retrofitted with concrete shear walls and steel braced frames and was still standing. He noted that when retrofits are designed well and installed correctly, they do work - but the Christchurch data also contained hard lessons about what goes wrong.

On parapets, Robert showed collapse statistics from the Christchurch earthquakes. His key finding: as-built parapets should not be expected to perform well under any meaningful earthquake, regardless of height. Two mechanisms explain why code analysis can be misleading. Damp-proof membranes concealed inside mortar joints at the base of parapets act as sliding surfaces, causing failure under sliding rather than rocking at much lower accelerations than codes predict. And short, light parapets can be ejected by vertical ground accelerations - something most design codes do not address.

On the performance of retrofitted buildings, he noted that roughly 50 percent of failed retrofitted masonry buildings in Christchurch had poorly installed epoxy anchors, meaning the bracing systems were connected to the wall by anchors that had not fully bonded. The anchor failure, not the bracing system itself, was the governing factor in most of those cases.

Retrofit techniques: from parapets to full wall systems

Robert organized retrofit techniques roughly by life-safety priority and covered several systems his team has tested or deployed.

For parapets, steel bracing systems are the most common approach across North America and generally perform well. Testing showed approximately 25 percent of braced parapets in Christchurch failed, and in most of those cases the failure was not the bracing system itself but either an unretrofitted wall below collapsing and taking the parapet with it, or anchor failures at the brace connection. He also described a post-tensioning technique using a threaded rod run vertically through the parapet, with a new concrete pad poured at the base tied into the surrounding masonry. A test specimen retrofitted this way survived accelerations exceeding 2g. He noted the method works well for straight chimneys and parapets where the retrofit can remain hidden, with the practical limitation that the parapet or chimney cannot be used afterward if detailed as tested.

For out-of-plane wall restraint, timber strongbacks - typically 2x4 members at regular spacing connecting the wall face to the floor or roof diaphragm - proved highly effective in testing. An assembly of 2x4s at two-foot spacing held at 1.3g without failure. A full-scale building test running 1,000 cycles at 1.2g showed good life-safety behavior with some damage accumulation. Robert said he prefers timber for its cost, field adaptability, and the fact that local builders can source it from a hardware store. Cold-formed steel or hot-rolled steel become preferable when fire-rating requirements apply or when spans are too long for timber to perform adequately.

For concrete shear walls, Robert described these as having a very high seismic success rate. The main quality-assurance concern he flagged is ensuring no voids form behind reinforcing bars when a shotcrete technique is used. He also noted a cost insight from a contractor: on a project using epoxy dowels as shear studs between new concrete and existing masonry, the ratio of anchor and installation cost to concrete cost was approximately nine to one. Switching to mechanical anchors substantially reduced installation time and overall project cost.

Anchor behavior in existing masonry - what lab testing shows

Robert devoted a substantial portion of the presentation to how different anchor types actually behave in existing masonry, which he described as meaningfully different from new materials like concrete.

On epoxy anchors, he said there is nothing fundamentally wrong with them when installed correctly. The three installation failures that cause problems are: dust remaining in the hole (pressurized air guns are especially counterproductive in soft masonry because they generate more dust rather than removing it - vacuuming is required), insufficient epoxy volume, and loading the anchor before the epoxy has cured. He showed a close-up photograph from the Christchurch earthquakes where a wall-to-diaphragm connection with seven epoxy anchors had only three anchors still holding brick - not the failure pattern you would expect from properly installed anchors.

On anchor diameter in clay brick masonry, he showed a graph from large-scale University of Auckland testing comparing anchor diameter against tension capacity. The finding runs counter to the intuition that larger diameter means more capacity: in existing clay brick, as diameter increases past approximately half an inch (12mm), the failure mode shifts from combined cone pullout to splitting of the masonry unit, and capacity drops. A 20mm (roughly 3/4 inch) bar had lower capacity than smaller bars in many test specimens. The recommendation that followed is to use longer, thinner, fully threaded anchors rather than larger diameter anchors when working in existing masonry.

He also discussed the 22.5-degree downward installation angle required for epoxy anchors in the IBC. University of Auckland testing showed this angle has a negative effect on anchor performance in existing masonry, because loading a bent anchor in tension causes local crushing under the bend, reducing capacity and increasing elongation before the anchor fully engages. Robert said this requirement likely originated when epoxy was more flowable and may have been intended to prevent it from running out of the hole - a problem that no longer applies to modern epoxy products.

ICC certification changes and what they mean for practice

Robert described Python Fasteners' work with ICC to update acceptance criteria for post-installed mechanical anchors in existing masonry. He noted that the resulting changes currently apply only to ICC-certified products but are publicly available standards he expects will eventually extend to other manufacturers as they seek certification.

The five changes he highlighted: installation angle is no longer required to be 22.5 degrees downward, with horizontal installation now permitted for certified products. Edge distance requirements have been reduced from 12-16 inches to approximately 6 inches, justified by test data showing that large conical breakout failures - the basis for the original concrete-derived requirements - do not occur in existing masonry. Minimum embedment depth is now the lesser of engaging every wythe or 10 inches, making anchors usable in one-, two-, or three-wythe walls without requiring embedment nearly through the full wall thickness. Tension and shear interaction is addressed by a combined actions formula. And capacity values for hollow clay tile were added, which Robert described as a significant practical improvement given how common hollow clay tile is in existing US masonry construction.

Quality assurance and site testing on URM projects

A consistent theme across the Q&A was the importance of site-specific anchor testing. Robert and a practitioner from Portland both described early testing as one of the highest-value steps on a URM retrofit project.

Robert explained that masonry quality varies substantially from building to building and even within a single building. Site testing establishes actual anchor capacity in the specific material on your project, which can result in needing meaningfully more anchors or substantially fewer. He said the cost of an anchor testing program often pays for itself through a more efficient design, and recommended testing as early as site access allows so results can drive design decisions rather than simply confirm construction-phase work.

Portland requires pull testing during construction for URM retrofits as a matter of policy. Robert described this as a straightforward formality when early testing has already established expected capacity - the construction-phase testing becomes confirmation rather than a first look at an unknown material.

For walls with suspected voids - a question raised by a practitioner working on a multi-wythe brick wall more than 100 years old - Robert suggested running a site-specific anchor test program rather than defaulting to grouting. He noted that mechanical anchors tend to be more reliable in walls with voids than epoxy anchors, and that testing at the actual site conditions provides a defensible basis for design capacity values.