Albert Park Keeper’s Cottage with Dave Olsen of Mitchell Vranjes and Egbert Koekoek of Cape

Your correspondent keeps a sharp eye out for old buildings under wraps. When one is spotted, this is usually followed by a spate of calls and emails requesting a site visit—what you might call (I hope!) a charm offensive. In the case of the Albert Park Keeper’s Cottage, it was almost as though the building was taunting me: for one thing, it’s right there at the University gates. And for another, it was rising into the air. Come and get me!

Albert Park Keeper’s Cottage. The Cottage has been jacked up off the ground to allow repiling work to take place. The Cottage is an 1882 timber structure, with brick piers supporting the floors and a brick chimney. One unusual feature of the building is its slate roof. The slates are an added mass high up on the structure, and have some effect on its predicted seismic behaviour.

A sign at the site explained that the building was undergoing seismic strengthening: and so, a few phone calls later, we went behind the fence to check out the project and how it was progressing. Guiding us were the project engineer Dave Olsen from Mitchell Vranjes (regular site visitors may remember him from the Melanesian Mission) and Egbert Koekoek from the construction contractors Cape.

Albert Park Keeper’s Cottage. Site visitors assemble to hear from Dave Olsen (left) about the project.

Going up

So, why was the house in the air? As with many older buildings, the basic problem is that the Cottage is not strong enough to resist horizontal loads. A structure can be fine holding up its own weight, but if it’s shoved sideways, it falls off its foundations, and that’d be that. One part of the project is to strengthen and renew the Cottage’s piles, and to brace it horizontally against loads from a future earthquake.

But buildings, especially old houses, are re-piled all the time, right? And they don’t get lifted into the air, do they? The reason for this gets at some of the differences between heritage jobs and regular engineering. In a conventional repiling, holes are cut in the floor, and those are used to dig out and place the new piles. At a heritage building, one of the first principles is to try to avoid damaging the original fabric, and to minimise any necessary damage. Rather than cut the floor to pieces, it was deemed better to lift the building—a technology more commonly associated with house removals.

Albert Park Keeper’s Cottage. Steel lifting beams support the Cottage off the ground. Note weatherboards have been removed to allow the beam to be inserted. As can clearly be seen, the beam is lifting from above the floor.

Lifting the building had other benefits. It gave enough headroom for the workers to install some larger timber piles, the deepest of which extend 900 mm below the surface. Further, because of the heritage-listed trees which surround the site, it was not permitted to use screw-piles, so workers (and the arborist!) had to be able to see where they were digging.

Albert Park Keeper’s Cottage. Lifting beams seen in the interior of the structure, photographed at the southern corner through a convenient gap. Note on the left the timber ribbon beam, which runs longitudinally through the building and is fastened to the studs.

How do you lift a house? I’d’ve imagined that this was done from the bearers, or maybe the joists. But it was plain to see that this was not the case at the Cottage. The orange steel beams running through the house are clearly above the floor level. Egbert Koekoek explained that the house lifters installed timber ribbon beams running the length of the cottage, which were attached to the studs. Weatherboards, and the internal timber lining boards (sarkings), were removed to allow the ribbon beams to attach directly to the studs. The orange steel lifting beams were inserted. Then, fourteen hydraulic jacks lifted the Cottage up into the air, a little at a time, over the course of a couple of hours. Each jack can be individually switched on and off, leading to a certain amount of racing around with a tape measure to make sure that everything’s lifting at the same rate!

Albert Park Keeper’s Cottage. Note the new timber (lighter colour) added either side of existing bearers. The line of brick piers below the bearer, still able to carry gravity loads but with no horizontal capacity, has been augmented and partly replaced by timber posts. Diagonal braces attach to new timber piles. At the right, a concrete block wall replaces bricks which have rotated outwards due to expansive soil.

With the house lifted in the air by its studs, it’s not safe to go inside, for fear that the floor might simply fall away under your feet. However, the raised house also provided the opportunity to strengthen the bearers. The need for strengthening is in part due to the new use of the building as public space, requiring a design for 3 kPa floor loads. This has been done by adding timber either side of the existing bearer—once again, unconventional practice, but in keeping with the heritage principle of retaining original material.

Albert Park Keeper’s Cottage. The original perimeter bricks are largely being retained and reintegrated into the load-bearing system.

At ground level

Geotechnical testing of the site revealed that the soil is expansive, meaning that it shrinks and swells a lot. Perhaps as a result of this, some of the original perimeter brickwork under the walls has moved around quite a lot over time. On the park side of the house, the wall had rotated about ten degrees, and had to be replaced with a concrete block wall. (The concrete will be faced with brick so that it looks much the same as the original.)

With some new perimeter walls, and with sturdy timber diagonal brace piles taking effect, the underfloor of the Cottage is now going to be fairly stiff. A site visitor asked about stiffness compatibility between the underfloor and the timber structure of the house, which can be expected to be pretty floppy by comparison to its supports. The answer to this came in several forms, if I’ve understood it correctly!

In part, the Cottage itself is getting some increased stiffness. The sarking on the internal walls is going to be renailed in a number of places, making the internal boxes of the rooms considerably firmer. The front room, in the northeastern corner, contains the chimney, about which more later. It requires extra horizontal bracing to restrain the brickwork, so a brace Gib is being added over the sarking. (The ceiling of that room gets an enhanced diaphragm, too.) But, in the main, the answer to questions of stiffness compatibility between structure and substructure is this: it doesn’t matter. The Cottage isn’t overly large. And the inherent flexibility of the timber makes it unlikely to transfer loads very far across the structure, meaning that deflections at the interface between floors and piles shouldn’t be too much for the connections to handle.

Albert Park Keeper’s Cottage. The fireplace and chimney were (naturally) not lifted with the rest of the building. The connections between chimney and structure had to be carefully broken away to allow the house to be lifted

Catching the flue

In the seismic assessment of the Cottage, the chimney was identified as the weakest link, scoring around 15% NBS. Think of the chimney as a freestanding pile of bricks. It’s supported on its foundation, and again at the ceiling level. Then there is a decent length of chimney between the ceiling and the roof, and still more again where the chimneystack protrudes into the sky. So what we have is a long brick column, with a point of restraint at the base, another at the ceiling, and a long unrestrained section above the ceiling. It’s this top section, above the ceiling, that needs extra support. In a quake, it could rock itself right off the rest of the flue, causing collapse.

Albert Park Keeper’s Cottage. A section showing the props bracing the chimney. Sturdy connections are made to the rafters. A plywood diaphragm at ceiling level increases the stiffness of the ceiling restraint. Image courtesy Dave Olsen/Mitchell Vranjes, all rights reserved.

The solution that Dave has chosen is to use timber props, creating a collar around the chimney just below the height of the roof. This creates a firm diagonal bracing for the chimney, meaning that the unrestrained section will be restrained at approximately half height. By changing the unsupported length, the period of the rocking motion expected in the chimney changes, and the resulting forces experienced by the chimney are reduced. In accordance with the NZSEE guidelines, the mortar of the chimney-bricks is assumed to have basically no tensile strength. In a quake, the chimney is expected to form cracks, breaking at predictable points into short but intact sections which will rock but not topple.

With the limited clearance beneath the Cottage’s floors, smaller workers are preferred

Local gossip

A couple more newsworthy points to share with you. Regular visitors to Albert Park will have noticed that the Band Rotunda is also under wraps. Egbert explained that, although there’s plenty to do at the Rotunda, there’s nothing structural happening: the job is mostly maintenance and repair. He also shared a few things about the work that’s been happening at Pembridge House, which is the southernmost Merchant House in the lineup along Princes St. I did make an attempt to get a site visit to Pembridge up and running, but it was too complex because the floor was taken up for a lot of the time and the site was hazardous. (Hazardous = interesting, though, doesn’t it!) A major feature of the job, structurally, was the insertion of two big two-storey steel K braces in the stairwell, which were then concealed. Nothing to see now, folks! Never mind: other opportunities will surely arise.


Sincere thanks to several people for helping to organise this one. We put this together against time pressure, with the Cottage due to be lowered early next week. A number of people set aside other (real) work to make this happen for us, including Richard Bland, Antony Matthews, and Stacy Vallis. Thanks to Auckland Council. We’re also most grateful to Dave Olsen and Egbert Koekoek for their time and their willingness to answer questions and discuss the project.


Eight things I learned by reading Section C8

I know, I know: a listicle. How 2011! But this is a site about heritage, after all… Here’s the pitch. I’ve been reading Section C8 Unreinforced Masonry Buildings from The Seismic Assessment of Existing Buildings, written by a team of researchers, academics and professionals and published at (I mentioned it in the post on bed joint sliding shear.)

You might think it sounds tedious, but it’s not. It’s well-written, well-illustrated, and provides some really useful and clear categorisation of structures and the ways they can fail. I needed to read it: this is the work I want to do. Some of you might be in the same position. But even if you’re not interested in doing seismic retrofit, if you’re involved in any way with building works on heritage buildings made from URM, you should probably leaf through the first few sections.

It would be presumptuous and preposterous for me to write a “review” of Section C8. I don’t understand it well enough to do so. But I thought I’d share my own highlights, as a way of enticing others to have a look. We’ll see how it goes.


On a more serious note, this article includes images of damaged and collapsed buildings from Christchurch. The following is presented with sincere respect to the 185 people who died in the 2011 earthquake and to their families. Let’s work to try to stop something like that happening again. Kua hinga te tōtara i Te Waonui-a-Tāne. HT


Section through a cavity wall. Note also the change in thickness at the first floor, creating a ledge for the joist to sit on. From Section C8 Unreinforced Masonry Buildings, in The Seismic Assessment of Existing Buildings, at Attributed to Holmes Consulting Group.

1.  Spot cavities with algebra

It was common practice to have a vertical cavity in brick walls. The cavity provided a barrier to moisture, and the outer wall could be made from higher-quality bricks that gave a finer appearance to the building. (The cavity is the black line running down the wall in the picture above.)

There are various complexities pertaining to the ties that were used to hold the layers of wall together across the cavity, how those ties have lasted, and how to assess the relative motion of the outer wythe (the veneer of good bricks) and the inner wythe(s). I won’t go into those here beyond noting that they exist. The question is, how do you know whether there’s a cavity, if you haven’t made a hole in the wall?

Algebra to the rescue! Well, multiplication, anyway. Brick are standard sizes. They’re usually 110mm thick (taking “thickness” as the dimension going into the wall). With some mortar in between wythes, that means that two wythes (layers) ≈ 230mm, three wythes ≈ 350mm, and four wythes ≈ 450mm. So if your wall thickness isn’t pretty close to one of those numbers—check for a cavity.

Insufficient connections between (floor and ceiling) diaphragms and walls leading to out-of-plane collapse. Section C8, as above,

2. Loose diaphragms fall out (sorry)

We’ve heard on site about floor joists sitting on ledges and maybe falling off when the ground shakes. We’ve also heard, on almost every site we’ve visited, about how tying the diaphragm into the walls at floor height, ceiling or both, can improve the structural performance of the building. What Section C8 makes clear is that the diaphragm can act to redistribute loads from out-of-plane to in-plane walls. If the diaphragm deforms too much, though, it won’t be able to support the walls effectively. In fact, too much diaphragm deformation can actually shove a face-loaded wall right off the edge of the building.

In-plane sliding on a damp-proof course. Section C8, as above,

3. Slip-‘n’-slide on a damp-proof course

For walls loaded in-plane (along their long axis), one of the possible failure modes is in-plane sliding. I wrote about one version of this in my post on bed joint shear testing. Section C8 points out that damp-proof courses can also be vulnerable to sliding.

To stop moisture wicking upwards through porous masonry, a layer of damp-proof non-porous material was commonly included in brick walls. Bitumen, slate, lead, or similar waterproofing was laid down in a continuous layer, usually not far above the foundations. (At a building I saw this summer, the DPC is granite, but that’s exceptional.)

Turns out, those damp-proof materials are softer than the masonry, or perhaps bond less well to the surrounding materials. Or maybe the change of material just provides a stress concentration. I don’t know! Still, I’d never seen or heard about DPCs as a site for sliding until I read C8.

Parapets tied back to a roof with raking braces — but are they sufficiently restrained in the vertical direction? Section C8, as above, Attributed to Dmytro Dizhur.

4. Vertical tie-down for parapets

It’s a common sight around the traps to look up and see a steel brace holding back a parapet. Job done, right? Maybe not. Section C8 points out that such braces may not have enough capacity to deal with vertical displacement, especially when shaking of the roof plane is amplified by the brace and transmitted to the parapet. Quoth C8: “The danger of non-robust strengthening is that the parapet still fails, but collapses in larger, more dangerous pieces.” Not good. The parapet may need to be drilled and post-tensioned onto the top of the wall below.

As axial load increases, masonry walls gain strength from confinement. From Section C8, as above, Attributed to Dunning Thornton.

5. Look out above

I suspect if you’d asked me whether masonry buildings suffered more damage at the top or the bottom in earthquakes, I’d’ve said the bottom. Makes sense, right? The walls crush and they turn over? Wrong. Generally speaking, the more axial load that is on the masonry walls, the stronger they are and the better they resist disintegration. Of course, this depends on the building form, the shaking, and other things, but the relationship between axial load and strength is useful to know.

Failure modes. Left, out-of-plane failure: instability of wall insufficiently tied in. Right, in-plane failure: spandrel failure, diagonal tension cracking, toe-crushing of piers. [Right, Section C8, as above, Attributed to Sharpe. Left, CCC Heritage, licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 New Zealand License]

6. We don’t talk about failure here

Section C8 enumerates the failure modes of URM buildings. I won’t write a list or a summary, but I will say that I found it really clarified my thinking about how buildings work to consider a finite list of failure modes. When I look at a structure now, I feel much better informed about how to break it into chunks and think about how each chunk might move and how it might fail. From that perspective, reading C8 was like taking your American mate to the cricket and telling him where to look.

Hierarchy of vulnerability, Section C8, as above, from

6b. And what to look for

An afterthought to the above. In assessing the structure, the C8 guidelines suggest working from left to right on this hierarchy chain, with the idea that the most vulnerable components of the building, those which are likely to fail first, are at the left. No good wasting time and money diagnosing a complex in-plane failure mode if the parapet’s not secured.

Ten steps. The assessment procedure, Section C8, as above, from

7. Ten simple steps…

I suppose I haven’t much to add to the image above. These are the guidelines which C8 provides to engineers as advice on how they should approach assessing a building. As with #6 above, for me this helped to understand how engineers divide the building into a set of observations and parameters which allow a model to be created—and don’t bother creating one that exceeds the complexity of the structure! I will be trying to hold these ideas in my head on our next site visit (St Paul’s Church on the 13th of March) and to think about how I’d approach the task of assessing the building. Thankfully we’ll have Peter Liu from EQ STRUC with us to show us how it’s really done!

A video from the Uminho Research Group on Historical and Masonry Structures. This is apparently a strengthened model. Still, watch the upper sections of the wall crack at the floor line and rock.

8. Walls rock

Walls under face load can be modelled as rocking. This means if the load is perpendicular to the wall, the wall can be “assumed to form hinge lines at the points where effective horizontal restraint is assumed to be applied… At mid height between these pivots… a third pivot point is assumed to form.”

When I read those words I recalled John O’Hagan talking about this at Hopetoun Alpha, but I think I understand it a little better now. To me, it feels different to think about a masonry wall in an earthquake as two rigid panels teetering one atop the other, as opposed to thinking about n bricks shuffling about independently, or as one rigid surface.

And this is the note on which I’ll leave this post. C8 offers quite a lot of guidance about how to make simplifying assumptions that allow analyses to be made, the rocking walls being one. It also offers suggestions for how to calculate important parameters if you can’t or haven’t tested them—things like tensile strength of the masonry. The impression I had, on reading these guidelines, was that the task of doing this kind of work myself someday seemed not impossible. Surely that’s the most sincere praise I can offer.

Testing mortar on-site with EQ STRUC, February 2018

What’s this then?

I’ve mentioned here before that I’m working for Salmond Reed Architects this summer. Recently, I had the opportunity to observe an on-site test of the material properties of some mortar at a heritage building in Auckland, and I thought I’d share some of the details here. The site is a late-nineteenth century two-storey brick building. (To clarify some jargon used henceforth:  bricks = unreinforced masonry = URM). I’ve been given permission to share the test procedure, but I haven’t sought permission to disclose exactly where the testing was being carried out, so that is a deliberate omission from the post. Still, I thought the process was novel and interesting enough that some of you might enjoy hearing more about it.

Borderline punching shear failure, wall anchor. From Section C8 Unreinforced Masonry Buildings, in The Seismic Assessment of Existing Buildings at Attributed to Dymtro Dizhur, et. al.

So you think you’re pretty tough

If you’re going to assess an existing building, you have to decide how you think the building might fail. For URM buildings, there is a hierarchy of failure modes from most likely to least likely, and an engineer needs to work down through  the hierarchy, examining all of the possible modes. At a certain point, the assessment will find a mode that causes something unacceptable to happen under the predicted load. That doesn’t mean that the assessment stops there: but certainly, something needs to be done about the potential failures and their resulting risk to life.

So far, so tidy. But there’s a problem. Heritage materials are far from homogenous. The 1920s concrete at the St James Theatre  is soft and drummy, whereas on another site I recently saw 1920s concrete that was described to me as “rock-hard”. Bricks can be low-fired and soft, or fired at high temperature and hardened. Stone’s anisotropic. And mortar’s really idiosyncratic. How much lime went in? Were there shells in the sand? How wet or dry has it been throughout its life? It’s hard to predict the strength of hand-mixed materials from a time before the standardisation of products. The true in-situ strength of the material, the actual number, will make a difference to which failure modes come out of the analysis as critical weaknesses, and how much work you have to do to the building. Hence also the dollars involved. So how do you find out how strong the materials at your specific building truly are, and hence, how they will fail? You test them.

Sliding shear along a defined plane. From Section C8 Unreinforced Masonry Buildings, in The Seismic Assessment of Existing Buildings at Attributed to Dunning Thornton.

In due course

Mortar’s not brick glue. Its primary purpose is not to stick bricks together. Instead, it provides a slightly compressible joint between the bricks as they sit in their stacks, allowing them to expand and contract without cracking each other to shards. Generally, mortar is softer than the bricks, especially lime mortar, and this is a good thing.

Notwithstanding the above, the mortar is the thing that stops the bricks sliding across each other if the wall gets shoved along its length, for example by an earthquake. The test that I observed, a bed joint shear test, examines how well the mortar prevents the bricks sliding across each other. It seeks to establish a cohesion value for the mortar. Key takeaway: it’s a mortar test, not a brick test.

Diagonal tension cracking, piers. From Section C8 Unreinforced Masonry Buildings, in The Seismic Assessment of Existing Buildings at Attributed to Dmytro Dizhur.

The cohesion value influences several important failure modes: embedded anchor pullout; punching shear for plate anchors; (diagonal tensile strength leading to) diagonal tensile cracking and spandrel shear due to flexure; and bed-joint sliding. In this post, you’re seeing  pictures of several of these modes. For the building that was being tested, the  EQ STRUC engineers told me that the cohesion value was going to be used for calculating bed joint sliding shear.

Bed joint sliding, stair-step crack sliding in low axial load walls. From Section C8 Unreinforced Masonry Buildings, in The Seismic Assessment of Existing Buildings at Attributed to Bothara.

The bed joint shear test is carried out by taking a brick out of the wall and using a hydraulic jack to push on the bricks either side of the hole. When the brick that’s being pushed starts to move, that means that the mortar has failed. Simple. In a moment, I’ll show some images of the steps. But before that… this.

Bed joint sliding shear

Bed joint sliding shear is calculated thusly:

Bed joint sliding shear, equation C8.33, from Section C8 (Unreinforced Masonry Buildings) of The Seismic Assessment of Existing Buildings, from

I hope it won’t totally destroy my credibility if I tell you that my first reaction to seeing the above was yuck. But on closer inspection, the equation is pretty straightforward, and considering its terms really helps to understand what is being measured.

Firstly, the µf(P + Pw) bit. That’s just saying that the load pressing on the bricks makes them rub on each other, which makes it harder for them to slide across each other. More load on top: harder to slide the bricks.

Secondly, the tnomLwc bit. tnom and Lare the thickness and the length, so really that’s an area: the area of the bed joint. And the cohesion value is the stickness. How sticky is the mortar, per area? That’s what this part is describing.

In the real-world bed joint shear test, the engineers try to make the equation even simpler, because the friction bit mightn’t be completely obvious: exactly how much load IS on the wall at the time you test? You can measure the height of the wall, but what about superimposed loads? So, to avoid worrying about this, the EQ STRUC team carried out the testing on bricks just underneath windows, where there wouldn’t be much imposed load. That means the cohesion can be found by knowing how much area of mortar you were testing (the bed joint size) and how hard you shoved the brick.

Opening up and finding a stretcher course

The test, in detail

First, catch your brick. I’ve noted above that the best spot to test is under a window. Another limitation is that the test should only be carried out with stretcher bricks in a stretcher course. If you’re not sure what that means: stretchers are laid along the wall, headers are laid across the wall. Like so:

A brick wall, anatomised.

The tests I saw were all carried out on the internal wythe of the brick wall. (A wythe is a layer of thickness–ie, a wall that’s three bricks thick is laid in three wythes, usually interlocked. There’s more complexity to this but that’ll do for now.) So, to carry out the test, you need to open up the linings and identify the right course to test.

Using a mortar saw to remove a brick; and a closeup of the mortar saw

Secondly, a single brick is removed. This means using a mortar saw to take out all the mortar above, below, and to the side of the bricks. On the picture two up you can see that these are the two bed joints and the two head joints. The brick gets extracted whole and can be replaced later.

A brick is removed. The mark reads “W Hunt, Auckland”

Mortar is removed from the extracted brick, and set aside for future testing—think of it as something like a concrete cylinder test. This testing isn’t part of determining cohesion.

The head joint is removed from the far side of a brick adjacent to the hole

Thirdly, the head joint is removed from the far side of one of the bricks adjacent to the hole. This is to give the brick that will be shoved some room to move. If the mortar wasn’t removed, the test results would be affected by the compressibility of the mortar in the head joint.

A jack exerts force on the brick, shearing the mortar in the bed joints above and below

Finally, the jack is inserted. It’s pumped by hand, exerting a force on the brick. The mortar in the bed joints above and below the brick resists the force. Since we’re testing under a window, there’s basically not much friction from the weight above. So the thing that’s stopping the brick moving is the two bed joints.

When the shove gets strong enough to move the brick—that’s deemed to be the point of failure, and the value is recorded. This is the peak value, meaning how much force would have to be exerted on the bricks by an earthquake to get them to start moving. Once they’re moving, they still need some force to keep them moving. So the engineers reset the jack, and pump it up once more until the brick begins sliding again—at which point, the residual value has been found.

The brick on the other side of the hole can now be tested. The engineers wedge something into the vacant head joint on the first brick—to prevent further sliding—and then cut out the head joint on the second brick, before testing again as described above. With a set of values from paired tests carried out around the building, an average value can be determined. Remember, we know the area of the bed joints, we know how much force we used, we’re ignoring friction, so there’s only one unknown: the cohesion value.

Engineers read the gauge on the jack during bed joint shear testing


There’s a somewhat subjective element to the tests. As for most things, it takes experience to determine exactly when sliding has begun. The tester gets a certain amount of physical feedback from the unloading of the jack as the brick slides, but even then, the exact moment of failure and the associated peak value are not precisely defined.

And this is only appropriate. After all, the mortar isn’t going to be homogenous throughout the whole building. Nor will other conditions be exactly similar. What’s required is something more like a geotechnical value—which is to say, a good rigorous estimate of the cohesion, into which some safety margins can be built. It’d be possible, with more elaborate equipment, to measure the load and the deflection more precisely; but a more precise number wouldn’t really be more meaningful.


Thanks are very much due to Romain Knowles and Antti Wallenius from EQ STRUC. Thanks also to Phillip Hartley from Salmond Reed Architects for getting me in the door at the site.