Collapse: Household Decks and Balconies are not Designed for Parties

So at the beginning of the year I was working on a project which whilst in the scope of the South Australian development act, and potentially the SA Building rules, it was otherwise beyond the scope of the Building Code of Australia (BCA) and National Construction Code (NCC). This in turn led me to take a closer look at human factors data, barrier heights and barrier loads, along with associated floor loads. I haven’t yet turned the associated spreadsheets and discussion into a blog post, though such will come later. However, the recent balcony collapse in Melbourne, brought this back to the forefront of my mind.

It seems balcony collapses, are an increasingly common occurrence, and no one seems to be learning.

  1. Balcony collapse mum dies [November 2008]
  2. Nine people taken to hospital following St Lucia balcony collapse [April 2017]
  3. Balcony collapse at Doncaster East Christmas party kills two people, injures 17 others [December 2017]

Owner-builders, builders and building designers alike are often declaring that the design loads are too high. How do they know this?  I don’t know. For the load to be declared too high, I would have to know what load was suitable and the strength of the structure. I don’t know this, I only know what the legislated load is: and such loads are either uneconomically high, or too low to be “safe”. Similarly the resistance of the structure is determined by the maximum strength permitted by legislation, not the actual strength. The actual strength cannot be known until we load it to collapse.

The legislated loads for decks and balconies do not allow for large parties. Not only is a large gathering of people in a house a problem for the strength of the floors and balustrades, it is also a problem for the neighbourhood with respect to car parking, traffic congestion and noise, and general nuisance and inconvenience to the neighbours. These latter issues are planning issues, land usage issues, issues which pose major problems for business catering to the public. Businesses and other organisations require permits and/or licenses to cater to large public gatherings. This is before even consider other operational requirements concerning health and hygiene. Inviting people round to a building which is not fit-for-purpose, and preparing and serving food possibly in less than hygienic conditions. Impose rules on one then impose rules on all.

Land Use Permit

Rather than call for mandatory annual inspections, just extend existing rules and require permits for large household gatherings, and as part of getting such permit an inspection can be carried out. But what represents a large gathering?

A large gathering would be one in which car parking extends from the house property into the street. Given average car ownership is 2 vehicles per household, and average household occupancy is 3 persons per house, then assuming typical driveway can accommodate no more than 4 vehicles. Then there is only space for 2 guest vehicles. Assuming family of 4, in the house, and 4 persons per vehicle, then the maximum household occupancy is 12 persons. Any more than 12 persons then a permit is required, the permit being based on a plan. The plan would cover the time and duration of the event, how parking is to be managed, and how guests drinking and driving is to be avoided. It would also cover the rooms to be used, and how the party organisers are going to control guests movements between the rooms. Additionally also need plan to control uninvited guests. If don’t want all this aggravation, then use a building suitable for the intended purpose, and that building is not your house.

Suitability of Design Actions

So I’ve pulled a number of 12, out of thin air. Is that number too high or too low? What is the weight of the group? How is weight distributed through the group? If children weight less than adults what percentage of the people are adults?

The legislated loads, for South Australia are determined by the South Australian Development Act & Regulations (SADA). This in turn calls up the Building Code of Australia (BCA) as the South Australian Building Rules. The BCA then references the residential timber framing code AS1684.2, which is dependent on the structural model in AS1684.1: 1999. The structural model is then dependent on the loading code AS1170.

Location Load Factored
kPa kPa
Houses 1.50 2.25
Balcony/Decks 3.00 4.50

The base loads which are given are typically multiplied by a partial load factor of 1.5 to get the factored loads. How do these loads compare with 12 people? First up, is if we are dealing with extreme value loads, such as 95th percentile or higher than no load factor should be required.

Looking at Australian Bureau of Statistics (ABS) data indicates that 97% of the population has a weight less than 110 kg. So our 12 people would have an extreme weight of 1320 kg (1.32 tonne, or 12.9 kN). That seems like a pretty unreasonable weight for a poorly maintained timber balcony. So 12 people is looking like too many. But assume each person occupies a space no less than 300mm by 600mm, so the total area occupied is 2.16 sq.m, and therefore the pressure is 6 kPa. The estimated load thus exceeds our factored design load. Now I typically cannot go increasing this because people will just revert to a higher authority: the council will grant approval if use 4.5kPa: so why should I pay more for a  stronger floor? Well because the stronger floor is more suitable for your purpose, than the minimum legislated value is.

But is 6 kPa any more reasonable than 4.5 kPa? What if 50% of the persons were children and they weigh no more than 50 kg, and each can squeeze into a space 150mm x 400mm? Then our total weight drops to 960 kg (9.4 kN), and the area drops to 1.44 sq.m, giving a pressure of 6.5kPa. That didn’t help, with lighter smaller persons, they can now all squeeze into a smaller area. So what if all 12 of the occupants were children, the weight drops to 600kg (5.9kN), but they can now crowd together more compactly, occupying an area of 0.72 sq.m, and giving a pressure of 8.2kPa. This  isn’t going in the right direction, we want lower loads don’t we, cheaper lower cost structures?

What if we use averages? We can use averages, but what if our situation isn’t average? What is the probability of our usage exceeding the average value? We need to ensure a structure is fit for use, therefore every time we propose to change its use we should assess the suitability of the structure for the alternative use. But if our structure is poorly maintained then it is unlikely to prove suitable for its current use. Thus poorly maintained structures need to be downgraded. If came to sell the house and the buyers got an independent survey done, then the poorly maintained structure would be downgraded.

As I’ve mentioned elsewhere, houses don’t increase in value, they deteriorate and decay and cease to be suitable for current and future purposes. Location increases in value, and current availability has value compared to wait for new construction. If the house has deteriorated below suitability-for-purpose, then its value has also diminished.

So the design load is difficult to determine, and 12 people are too many if they are crowded together in one space to get a group photo. So if want a group photo, take it standing on solid rock, standing on solid ground. [to paraphrase]

… to be continued


The above assessment considered the weight of people crowded into a small space, producing a high local pressure. it could be argued that the people can be crushed into an even smaller space which would push the floor pressures up even higher. For that matter could consider a single person and the pressure they exert on the individual spaces allowed, for the adult the pressure is 6 kPa, and for the child the pressure is 8.2 kPa. {which is good, it matches the previous result for 12 people as it should. } These pressures would be higher still if took the area of the feet supporting a persons weight rather than an allocated space into which the whole body can fit. If a person lies down and spreads their weight over a larger area, then the floor pressures decrease. For design purposes and an individual person, we tend to ignore the pressure and represent them as a point load, for the adult that is 1.1kN.

People are generally not crowded together they tend to prefer some clear space around them, and spread out. So if have a balcony which extends 1.2m from the house and is 5 m long, we get an area of 6 sq.m, and for the 1.32 tonne of adults, the pressure drops to 2.16 kPa, which is less than the code of 4.5 kPa given in the code reference above. Similarly if we increase to an elevated deck extending 4m from the house and 5m long, then get an area of 20 sq.m, and the pressure drops to 0.65 kPa. As we increase the size of the decking the 12 people have less and less impact on the over all area of the structure. But the larger we make the floor deck the more people can fit in the space.

When designing any structural element in the deck it should be able to support the weight of at least one person. The problem to resolve is how many more people should it be able to support? Should the floor be designed for the maximum number of people which can be squeezed onto the floor? Or should it be designed for a comfortable arrangement of people and furnishings?

Whilst architecturally can design the floor space for people to spread out and have comfortable space around them, the people will still be congregated into clusters. The clusters would be spread out on the floor with clear space between, whilst, within each cluster each person is separated from each other by a comfortable distance. In a restaurant for example people would be seated around a table, and placing a ring load on the floor, rather than a load uniformly distributed over the floor. Such ring can be placed any where on the floor: so how to determine the forces on the floor decking, the joists and the bearers, and any support posts? There are too many possible patterns, so it is not practical to consider. We therefore need an equivalent uniformly distributed load over the floor area (pressure), which gives similar results to the actual pattern loadings which a structural element may encounter.

The problem with AS1170.1 is that it is descriptive and doesn’t quantity the situations for a given load in terms of the actual function of the space, and also lacks guidance on floor vibration and/or bounce. If consider a house the main thing which occupies the floor is the furniture. The rest of the space is largely empty, with maybe 4 people scattered in various places.  The family comes to leave the house and they congregate in a closely packed cluster at the front door, possibly with luggage or shopping bags. The floor in front of the door potentially experiences higher loading more frequently than any other part of the floor. But it is still intermittent, whilst the load from the refrigerator, washing machine, and chest freezer are relatively permanent. Whilst these appliances can be placed any where in the house, they are most likely stored in specific rooms, of the house: the kitchen and laundry. These rooms however are not designed for higher floor loads. They could be, but then we would also have to consider the movement of the appliances into the house through other rooms. Only balcony is designed for higher load than the rest of the house floor. Why?

Well one thing to note is that AS1170 has been updated a few times since 1999, whilst the structural model AS1684.1 for timber framed construction has not, though AS1684.2 has been updated many times. The base load of 3 kPa in AS1684.2 is no longer required for self-contained dwellings, and the load can be reduced to 2 kPa, or be 1.5 kPa similar to the house if less than 1m above ground. Why is self-contained dwelling of any relevance? What does height have to do with the load?

The height of the floor above ground level doesn’t affect the number of people who can fill the space on the floor. Height may be considered as increasing the severity of injury experienced should the floor fail and people fall: but a person could trip up at ground level and receive fatal injuries. Increasing the magnitude of the design load is not tackling the real problem. No matter what reasonable value we pick for the design load it always has the potential to be exceeded. When the design load is exceeded the structure will collapse. The installed resistance of the structure deteriorates with time and poor maintenance: the structure will therefore fail at less than the design load.

Machines are typically designed to be “failsafe”. The machines can be overloaded, but otherwise designed so that such overload fails the machine with minimum hazard or acceptable hazard. We cannot get rid of hazard, though we can avoid it altogether.  We have decided to have an elevated floor therefore there is an hazard we have to accept, we therefore need to minimise the consequences of the hazard occurring. Therefore mode of failure is important to structural design, not just the magnitude of the load: in practice mode of failure and controlling it is largely ignored: its too complicated {no one has figured it out and provided a methodology}.

Machines also experience fatigue. Roughly speaking each operating cycle reduces the strength of the material, therefore the components need to be replaced before they fail. So operating cycles are monitored and recorded, and then parts which look perfectly good are replaced by parts that have less fatigue.

Buildings accumulate fatigue during tropical cyclones, and earthquakes. Bridges accumulate fatigue with the loading and unloading of traffic flow. So fatigued building components also require replacement.

Potentially one of the problems we have is cultural: the notion that buildings last for a hundred years and should increase in value. Buildings decay, their internal space becomes less useful, and the buildings themselves hinder the efficient operation of a much bigger machine: the city.

The knee jerk reaction to structural failure is typically to increase the magnitude of the design load. Relatively easy thing to do, if cannot defend the magnitude of the original load, which was obviously too low, otherwise the structure wouldn’t have been over loaded and failed.

However increasing the magnitude of the load for all, to accommodate poor maintenance on the part of some people, is not really acceptable: it is an unwarranted expense to the many, and otherwise a waste of material.

Design Issues

We have two issues:

  1. The design load can be exceeded due to an infrequent event.
  2. The design resistance decreases with time and poor maintenance.

Whilst the design load is uncertain, the past and current design loads are suitable the majority of the time, and few people experience a problem. Those who do experience a problem may die as a result. We do not want the failure to be newsworthy, we want it to be little more than an inconvenient nuisance. To achieve that, it is not the magnitude of the design load which needs to change but the structural form of the balconies and decks: more attention to detailing.

Different materials and structural forms, require different levels of maintenance. Development approval controls what gets built. Occupational health and safety controls maintenance of buildings used as a workplace, but there is no obvious public and/or domestic health and safety legislation. Most accidents are considered to occur in the home, so we need to make houses and domestic activities safer.

Possible Solutions

How to make domestic activities safer without unwarranted expense to home owners? I remember reading in the late 70’s early 80’s that in the UK grants were given to improve the appearance and amenity of houses. Australia could adopt something similar and probably better than specific funding of solar power, and household insulation. Most of our houses exist already, so BCA/NCC criteria are largely irrelevant, as only affect new houses. Of course someone buying an established house, could get a survey done to compare against the current BCA criteria and determine cost to upgrade, and so argue for lower price. So a first requirement is a cultural change which hammers house prices down, so that it becomes more viable to bulldoze the buildings and make land available closer to established facilities: thus reducing the need to rezone and open up new land for development. Thus prepurchase building surveys becomes the norm.

Building industry develops new technology to improve the potential for building renovation, to minimise need to demolish. This all helps with respect to the rubbish buildings changing ownership, but doesn’t help with poorly maintained buildings when there is no change of ownership.

Above I proposed a land use permit: basically planning and building approval to use the land temporarily for a purpose for which it was not previously approved. The permit could be for an individual event or for a fixed time period. A permit could be issued for 5 years and possibly at the time a building is approved for construction. For example if have a large house, mostly designed for entertainment purposes then the house owners are granted a 5 year permit at the start. Every 5 years the permit needs to be renewed, following a house inspection. The permit can restrict the frequency of large parties and the number of persons who may attend: a variety of other conditions may also be imposed.

It should not require any additional legislation, just a change to the development act and regulations to accommodate: transient land use. So infrequent and intermittent land use requires a permit. The permit is to provide for public safety and to control public inconvenience from the activities of others. The safety issue requires an answer to the question: when is a household gathering a public gathering?

Above I suggested the limit was 12 people, but showed they could overload the structure: for that matter also showed the potential for one person to overload the structure. I suggest a gathering is considered public if it involves people from more than two related families (eg. family gathering:=  the family and two sets of grandparents). So can tell the door to door salesman he cannot come in because need a permit for visitors. Nah! Not going to work. Set a number based on the floor area of the house and the property size, and the ratio between the two, taking into consideration any traffic controls which limit the number of people which can enter a floor space.

Such permits are mostly concerned with gatherings. Poor maintenance and house owner falling from balcony is another issue. Lets see. The car industry recalls vehicles and upgrades them on the basis of: may be, could be, a problem.  the purchase of a new vehicle, now tends to include a service agreement. Buildings are an architect’s brand, make them responsible. If buy a house then also get a service agreement. The city councils grant building approval, make them responsible for maintenance.

Development approval, land use approval, instead of being a one time occurrence, becomes an approval which has to be renewed. Say every 20 years from original approval a building needs to be inspected and granted new permit. Some items may be required to be inspected more frequently. Elevated decks, and balconies for example may be inspected every 5 years. These inspection requirements are placed as conditions on the development approval documents.

If the building designers and builders, offered such service in the first place, then no need for legislation: assuming the owners can afford. {Business is about providing a service not maximising monetary profits}

Thus every house in an area remains the responsibility of designers and builders in the local area. When the original designers and builders are no longer around the building is handed over to some other supplier in the area. Limits can be placed on the number of buildings allocated per qualified person, rather than based on business organisation. Home owners can change who is responsible, but someone has to be responsible.


One problem is the mismatch between authority and responsibility. The council has the authority but doesn’t really have the responsibility. During a coroner’s inquest the council officers are likely to be given a kind of sanctuary because they complied with the code. Engineers and other designers however are not considered to be fools blindly complying with codes: they are supposed to be smart enough to know when the code is inadequate and design systems to be suitable for purpose, not design to the code. The designers need to have the authority as well as the responsibility and not have their decisions over ridden.

{This is where I was going to start: I will delete this when I write the intended article on design actions}

On Structural Design

An amusing definition of Structural engineering:

The art of moulding materials we do not really understand
into shapes we cannot really analyze,
so as to withstand forces we cannot really assess,
in such a way that the public does not really suspect.

Professor E. H Brown, (1967), Structural Analysis, Vol 1, Longmans, Green & Co. [Actually even the origin of this quote and its variants is unknown.]

Contrast this with the definition from the Institution of Structural Engineers (IStructE UK):

… structural engineering, which is the science and art of designing and making, with economy and elegance, buildings, bridges, frameworks, and other similar structures so that they can safely resist the forces to which they may be subjected.

The flaw with the IStructE definition is the word “safely”: what exactly does it mean? I have previously written about limit state design and attempted to explain the uncertainty in the strength of materials.

… to be continued


  1. [20/12/2017]: {DRAFT} Original
  2. [21/12/2017]: {DRAFT} Added new introductory paragraph, and added discussion where I indicated to be continued.