Wood Construction and The Risk of Fire

Whenever the topic of wildfires comes up, someone always mentions [1] that US homes are built almost entirely made from wood, that it’s senselessly risky to build homes out of a material that burns, and homes should be built from some non-combustible material like steel or concrete like they do in civilized countries. The steel industry is especially fond of playing up the inherent risks of using burnable building materials. But the data presented with these arguments is usually fairly thin. So I thought it would be worth looking into how much of a fire risk wood construction actually is.

(note: this post will chiefly focus on light framed wood construction, as that’s what's used in the overwhelming majority of US construction. Heavy timber construction, such as CLT, has very different combustion properties.)

Fire and Building Codes

First, it’s worth pointing out that one of the central principles of US building codes has always been mitigating the risk of fire. This goes back to the US’s first model building code, the National Building Code in 1905, which was created by fire insurance underwriters to try to prevent increasingly destructive (and expensive) urban fires. Fire risk shapes everything about a building, from what material it can be built from, to where windows need to be, to how many stairways you need, to where you’re allowed to put hand sanitizer dispensers.

The focus of these code provisions is often on ensuring occupants can evacuate the building safely, rather than preventing property damage. They’re based around the number of people who will be in one place, the number of exits, the length of path to safety, etc. One reason that single family homes have comparatively less strict fire protection provisions (for instance, they don’t require fire sprinklers) is that it’s comparatively easy to exit them (and the code ensures this via provisions such as mandating every bedroom must have an exit window).

The US and Fire Risk

To start, how does the US do on fire risk in general?

(via NFPA)

According to NFPA, there were a total of 3704 civilian fire deaths in 2019, or about about 1.15 deaths per 100,000, a figure which has been declining long term, but rising slowly since around 2013. For reference, in 1953 the fire death rate was around 7.5 per 100,000. This makes fire responsible for around 0.13% of the annual deaths in the US, similar in magnitude to MS or Asthma. These deaths overwhelmingly occur in homes: 77% of fire deaths occur in residences, and 65% occur in single family homes.

Compared to other developed countries, the US does worse than average in fire safety - our rate of death by fire is somewhere around 3x what most other countries achieve, though it’s very similar to the rates in the Nordic countries (more on this later). The US also has more fires-per-capita than average, though there doesn’t appear to be much correlation to fire deaths - Japan, for instance, has among the highest fire deaths per capita but the among the lowest fires per capita.

(via CTIF)

Property damage in the US in 2019 was $14.8 billion, or $45 per capita. It’s hard to find good data on property damage caused by fire, but US numbers seem to roughly track other developed countries [2]:

• In the UK in 2008, this value was around $50 per capita (adjusted to 2019 dollars).

• In Australia in 2005, it was around $67 per capita.

• In Finland in the late 90s it seems to have been around $40 per capita.

However, catastrophic events can dramatically increase this - in 2001, the events of 9/11 caused more fire damage than the US experienced in the previous three years combined. And the wildfires of 2017 and 2018 doubled the amount of fire property damage in the US for those years.

In terms of fire prevention expenditure, the US spends about $132 dollars (111 euros) per capita per year on fire departments. This is around twice as much as the European average, though similar to what the Nordic countries spend.

Within the US, these averages mask a huge amount of state to state variation. Fire death rates by state, for instance, vary from 0.46 per 100,000 (Utah) to 2.27 per 100,000 (Mississippi).

One possibility is that fire losses are influenced by something like population density - the longer it takes for firefighters to reach you, the worse the outcome. I wasn’t able to track down fire response rates by state, but Buffington 2019 found that increased response time increased the likely property damage.

The US (along with most of the Nordic countries) has much lower population density than most of Europe. And within the US, we see fire death rates correspond very strongly to percent population living in urban areas:

We also see substantial correlation by community size in the US: large cities have significantly better fire outcomes than small towns, for both property damage and deaths:

However, we don’t see this correlation extend to percent urbanization between countries. It’s also highly confounded by income - denser areas and larger cities are substantially wealthier than small towns and rural areas. The correlation between median income and fire deaths is not quite as high as urbanization, but it’s still high. So at best this data is suggestive.

So, overall the US does worse than most comparable countries with respect to fire prevention spending, number of fires, and fire deaths (though it performs similarly to Nordic countries). It may do about the same in terms of property damage per capita. But within the US, there’s a huge amount of variation - death rates and property loss rates will vary by a factor of 4 or more depending on where in the country you live.

Wood Construction and Fire Risk

In the US, generally the shorter and smaller the building (ie: the easier it is to evacuate), the fewer restrictions there are on building materials. Residential buildings up to 5 stories in height are allowed to be built from wood in the US, so long as they meet egress requirements.

Compared to other countries, the US does much more wood construction. 93% of single family homes are wood construction, and 70% of new multifamily buildings are wood. This makes the US unusual - few places build as many wood single family homes, and outside of Canada no other developed country builds multifamily buildings out of wood in any significant quantities.

To what extent can wood construction be attributed to increased fire risk?

We do see reasonably strong correlation (r^2 of about 0.38) between wood construction market share in a country and fire risk, though our dataset is pretty sparse, and there’s a few big outliers:

In the US, we also see that the uptick in fire deaths roughly corresponds to a big uptick in the number of multifamily units built from wood. Steel industry advocates point to code changes allowing for taller wood buildings as responsible for both of these. But it seems hard to place the blame on increased multifamily construction, as it makes up a relatively small fraction of fire deaths. And we also saw a huge variation in the fire death rate from state to state, despite the fact that wood residential construction is fairly ubiquitous across the country.

I couldn’t find actual numbers, but it seems like insurance rates for wood buildings are higher than for noncombustible buildings as well (though some sources say otherwise).

Our best bet is probably to look directly at fire data to see what types of structures survive.

Garis 2014 looked at 11,875 fires that occurred in British Columbia between 2008 and 2013. It was found that wood construction performed worse than noncombustible construction (concrete, steel) in terms of deaths and injuries, BUT that this difference mostly disappeared once the presence of fire sprinklers was controlled for. NO residences with fire sprinklers, wood or otherwise, had any recorded fire deaths.

Odeen 1985 found that in single family homes in the US (the majority of which are unsprinklered), construction type had no impact on fire survivability (though this data is fairly old at this point):

He also found that in Sweden, combustible construction had similar frequency of damage as non-combustible, but the damage was much more extensive in combustible construction:

Richardson 2001 found similar results to Garis, that fire spread had little variation by construction type in single family homes. She found a slightly higher death rate by fire in wood single family homes, and a substantially higher one multifamily apartments, as well as a higher average property damage, but a lower injury rate. No effort was made to control for the presence of fire sprinklers, which likely confounds this somewhat.

Fire Sprinklers

One thing the data makes clear is that fire sprinklers are amazingly effective, though almost all residential construction in the US (even multifamily) is unsprinklered. Hall 2010 found that sprinklers reduced fire damage by 40-70%. NFPA found that 87% of fires, 90% of building damage, and 98% of deaths, occurred in unsprinklered buildings. Adding a sprinkler to a building drops the death rate by fire by over 80%:

In Scottsdale AZ (where fire sprinklers are more common in single family homes), the fire sprinklers were responsible for preventing 8 deaths and reducing the average property damage by fire by over 80%. The NFPA in general strongly advocates more fire sprinklers.

Other Confounding Sources of Risk

An interesting aspect of the problem I hadn’t previously considered is that fire risk isn’t necessarily associated with wood construction, but rather lightweight construction - buildings built from thin and light structural elements. This would include typical wood frame construction in the US, but also metal stud, open web steel joist, and unprotected structural steel. These members fail quickly in a fire, as they have a high surface area to volume ratio, and thus lose strength quickly when exposed to high heat.

Advocacy by firefighters for safer buildings is often focused on lightweight construction in general, not wood construction specifically. Li 2014 found that in tests metal stud construction in several ways performed substantially worse than light framed wood construction (the studs buckled due to the heat, causing the protective gypboard to fail in less than half the time as on the wood structure). Textbooks on fire safety lump all lightweight construction together, and describe metal stud and wood stud as having similar fire performance characteristics (ie: they’re both dependent on protective gypboard for any sort of fire resistance).

A variety of sources also noted that modern houses are a greater fire risk due to their larger size, open floorplans, and more combustible home goods. Kerber 2012 notes that modern homes burn much quicker than older homes due to these factors:

This suggests that maybe the US has higher fire risk not due to wood construction, but due to our enormous, open floor plan homes. This would also explain why multifamily buildings are a relatively small fraction of fire deaths. But cross country comparisons don’t really bear this out - average living space per person in a country doesn’t correspond especially well with fire risk.

Overall it seems that all else being equal, wood construction is associated with increased fire risk - fires that occur in wood homes are slightly deadlier, and more destructive than fires in noncombustible construction. However, this risk seems like it can be more or less completely mitigated by adding fire sprinklers.

Wood Construction and Wildfire

How does wildfire risk factor into this?

The risk of wildfire is closely tied to what’s known as the Wildland-Urban-Interface (WUI) - the area where houses are in or near wildland vegetation. There are a LOT of houses in the WUI - according to Radeloff 2018, around 1/3rd of houses are in the WUI, despite the fact that WUI is less than 10% of land area. This value is increasing over time - 43% of new homes are built in the WUI.

However, simply being in the WUI isn’t synonymous with wildfire risk. Most of the WUI is on the east coast, but the vast majority of wildfire risk is in the western states (along with some risky areas in Texas and Florida). And though climate change is increasing the risk of wildfires, the total wildfire hazard potential in the US actually decreased slightly between 2012 and 2018 [3].

The design requirements for a wildfire resistant building are distinct from normal fire resistant construction, and there’s a variety of building codes and design guides for construction in wildfire prone areas. The ICC has the WUI building code, which California adopted as chapter 7A in their residential construction code. Colorado and Montana (and perhaps other states) have standards based on the “Ignition Resistant Construction Guide”. FEMA has a Homebuilders Guide to Construction in Wildfire Zones. NFPA has Standard 1144, Standard for Reducing Structure Ignition Hazards From Wildfire.

Unlike normal building codes (which focus on evacuation), these codes are much more focused on preventing the buildings from catching fire in the first place. They all make relatively similar recommendations:

• Fire-resistant (Class A) Roofs: The roof is the most vulnerable portion of the house in a wildfire, and the roofing should be an assembly that will resist the spread of flame. This can be materials like tile, concrete, or metal, but asphalt or even wood shingles can be made class A as well. Anything combustible should be either fire-retardant treated or protected with a layer of gypsum.

• Non-combustible or fire resistant exterior elements: Things like exterior siding, windows, doors, and decks should all be of noncombustible construction, and designed to be fire resistant. Brick or cement siding instead of cedar siding, double-paned windows, etc. If there’s a deck, it should be either noncombustible or covered with protective gypboard or fire retardant.

• Construction details that won’t capture flying brands: This means things like not having exterior vents (through which a brand could enter), and having fire resistant soffits in case they capture a flying brand.

• Removing combustible vegetation: this means clearing an area 30ft from the house of vegetation, trimming overhead branches, and making sure gutters and downspouts are clear.

Surprisingly, wood construction merits very little mention in any of these design guides - there’s no prohibition on wood construction, no discussion of it’s risks outside of exterior use, and almost no mention of the benefits of noncombustible structural materials. The overwhelming focus is on eliminating combustible elements on the outside of the building.

We can also look at structure survival rates in wildfires to get a sense of what factors are the most important.

Syphard 2017 studied over 1500 buildings exposed to wildfire between 2003 and 2007. While certain materials and details had significant effects on building survivability, wood in general wasn’t a significant factor. Interestingly, the most significant factors were structural density (number of houses per square mile) and structural age. Young, densely packed houses all survived. Old houses spread far apart were all destroyed.

A follow up study by Syphard looked at 40,000 buildings exposed to wildfire found similar results, that the greatest impact of survival had to do with specific construction details (such as eaves and window panes) and whether the exterior-facing elements were non-combustible.

So for wildfire, wood construction seems to be even less of a factor than it is for “conventional” fires. By far the most important factors are things like how dense your neighborhood is, and specific, fire-resistant construction details which are mostly independent from structural materials.

Conclusion

So where does this leave us?

• Overall, it seems like wood construction does somewhat increase the potential risk of fire, mostly by allowing fires that do occur to be somewhat deadlier and more destructive. However, this effect is mostly swamped by other factors such as what state and city you live in, or whether you live in a house or an apartment. For a wood apartment in Salt Lake City, the risk of fire is vanishingly small; for a wood single family home in a tiny town in Arkansas, it’s much larger.

• The most important factor for fire risk in a home is whether or not it’s sprinklered. Fire sprinklers reduce the risk of fire by an enormous amount, and sprinklered wood construction seems to perform about as well as sprinklered non-combustible construction. And sprinklers are cheap, costing about $1-2 per square foot (much less than it would cost to say, change a wood house to concrete).

• For wildfires specifically, we see something similar - construction details such as fire protected eaves and class A roofs, along with things like community density, matter far more than whether your home is wood or steel.

[1] - Ironically, the picture of the house destroyed by wildfire in this article is a masonry house:

[2] - This might be what we’d expect to see if the US had more fires destroying less expensive buildings.

[3] - The authors of the wildfire hazard potential maps caution that changes in methodology and input data from year to year make comparing maps directly inappropriate.