When designing a multistory  building, there’s quite a few different options to choose for floor framing systems. Floor framing is more loosely coupled to the rest of the building than other structural elements - it’s often possible to swap out flooring systems while leaving the rest of the building unchanged. There’s some limitations - you can’t use a wood framing system in type I or type II construction for instance. And some systems have more natural synergies than others - if the rest of your building is cast concrete, it probably makes sense for your floor to be concrete too. But generally for a given building, there are at least a few different options for floor framing.
Because of this, the market for flooring systems is fairly competitive. Unlike some other aspects of construction, it’s not uncommon for old systems to die out and for new ones to appear. Cinder slabs or tiled arch construction are virtually unheard today, while some popular systems like I-Joists were introduced comparatively recently.
At the level of flooring components, we see an interesting spread of systems that exist on all points of the spectrum between factory built, requiring little site assembly, and completely site constructed. So it’s interesting to compare flooring systems directly to see how they stack up.
First, some terminology. A floor framing system system is what it says on the tin - a set of structural elements that supports a floor. This can be a truss spanning between two walls, a concrete slab supported by columns, a series of 2x6s with wood sheathing on top, or any of a dozen other systems. Some of the most common (and a few uncommon) systems are listed below:
This isn’t every possible flooring system, but will probably cover 95% of new construction.
The comparative ease by which floor systems can be switched out makes it fruitful to compare them directly to see how they stack up. We’ll take a look at three different measurements of floor system performance: depth, weight, and cost, and see how these systems perform .
Span vs Depth
All else being equal, a shallower flooring system is better. Shallower means less building height, which means it requires less structure, less material, and less labor to get a given amount of building area. A deep flooring system creates a lot of wasted building volume, so a shallow system that can span a great distance is a very useful thing indeed.
Unfortunately, the principles of structural engineering push in the opposite direction - the easiest way to get additional capacity out of your beam or slab is to increase its depth. Floors (and other spanning systems) navigate this by putting as much material as possible on the top and bottom edges, where it can be used most efficiently, and eliminating as much material on the interior.
Below is a graph of flooring system span vs system depth. The span is based on a 75 pound per square foot superimposed load, enough to cover most sorts of residential building design. (click to embiggen)
The datapoints are coded by color and shape. Green indicates wood systems, blue steel systems, and grey concrete systems. Circles indicate systems based on individual members like joists or trusses, squares indicate systems based on panels, and triangles indicate composite systems that require field-poured concrete. Spacing of individual member systems is assumed to be 16” on center.
Right away we see that wood systems require more depth for a given span than steel or concrete systems do. Wood is in the neighborhood of 30 to 60 times weaker than steel, so this shouldn’t come as a surprise.
We also see that composite concrete systems (composite deck, dovetail deck, deep dek) perform extremely well in the depth vs span department. These systems are basically a concrete slab with a thin layer of steel at the bottom edge, where it has the greatest impact. But as we’ll see later, these systems have to make some compromises to achieve this efficiency.
It’s also obvious from this what the most frequent floor spans buildings require are - the vast majority of the systems have a maximum span between 10 and 30 feet, with very few options once you go beyond 50 feet or so. It’s rare to have floors spanning longer than this (though for roofs it’s not uncommon). Large, uninterrupted spans are expensive to frame, so are generally limited to buildings that require them, like parking garages or some industrial spaces.
Interestingly enough, the cluster of low-span panels (SIPS, 3-ply CLT, AAC) are very uncommon systems compared to wood 2x members, despite spanning the same distances at shallower depths. The benefits of shallower systems are real, but aren’t necessarily enough.
Span vs Weight
In general, the lighter your building system, the easier and cheaper it is to use. Heavier floor framing puts a lot more load on the structure, both via gravity and from earthquakes, which requires a larger, heavier, more expensive structural system to resist. And heavy elements are more difficult to build with. They require more equipment to move around, they’re harder to put a fastener into or drill through, and they have less load-path flexibility.
Below is a graph of our flooring systems showing span vs system weight (click to embiggen).
Here we see the drawback of composite steel deck (and other concrete-based systems). Concrete is heavy, and any system that uses it will necessarily be an extremely heavy one. Some systems, like hollowcore and bubble deck, cast voids into the concrete (removing material where it’s not needed) to reduce the weight. But these will still be heavier than a pure wood or steel system.
However, this advantage isn’t quite as large as it seems. Finishing a floor usually requires some sort of site-cast cementitious topping to provide a level walking surface and reduce potential vibration problems. Systems like site-cast post-tensioned slabs or composite decks, on the other hand, get this for free.
This is a perennial problem with prefabricated systems. Even with extremely low manufacturing tolerances, you’re still putting a building on a site that will have a lot of natural variance in it. And even a system made up of low-variance elements can end up far out of tolerance, as the variation from multiple elements stacks up. Site-built systems, especially ones like concrete that can use gravity as a leveling mechanism, can often accommodate this. But it’s a tricky one for prefab systems to solve.
Span vs Cost
Of course, a performance analysis wouldn’t be complete without considering cost . More often than not, this is the prime consideration in whether a system gets used or not - no matter how many other benefits a system has, it’s fighting an uphill battle if it’s more expensive than the alternative. As always, construction costs are highly variable, so these values should be taken with a grain of salt.
Here we see why wood is so frequently chosen if the construction type permits it - it’s by far the cheapest floor framing option. Outside of wood, steel joists are the least expensive option, and also the one chosen most frequently in practice. Together wood framing and steel joists probably make up 75% of new construction (though it’s very common to use composite deck in conjunction with steel joists). Other systems generally get used when their particular benefits outweigh their higher cost. PT slabs, for instance, allow for extremely flexible interior layouts, as they only require column support rather than walls or beams. Combined with their thin profile, this makes them an attractive option for residential high-rises.
We can also see why certain systems remain niche. SIPs and CLT, despite their benefits, are extremely expensive relative to other options that could be chosen. Products like this, even if they’re part of the original design, often get value engineered away as part of cost-reduction measures when the project is inevitably over budget.
The above costs combine both material cost, and the labor required to install it. If we break it down into labor and material cost, we see exactly what we’d expect - the higher the level of prefabrication, the smaller proportion site labor is of the overall cost:
The ratios here suggest an unfortunate truth about prefabrication - that it mostly shifts where the costs occur, rather than reducing them. Highly prefabricated building components won’t necessarily use less labor to produce, they simply use the labor in factories rather than on-site. This can have some advantages - it lets your site-work proceed faster, as you can start building things long before ground has been broken. And it can reduce costs if you can build in a cheaper labor market than where your building is. But it’s not necessarily more efficient, in the sense of getting a greater output with the same input.
The Slow Advance of Prefabrication
It’s interesting that even at the level of building components, which exist in a competitive market, we don’t see prefab systems dominating. Factory produced, industrialized systems exist along with completely site-fabricated ones, and at most seem to have a modest advantage. Some highly prefabricated systems (SIPs, CLT) are among the most expensive systems available, and some site-built ones (wood dimensional lumber) are among the cheapest. You have to go down to even lower level components - fasteners, hangers, standard structural sections - before you see factory-produced items dominate.
 Flooring systems are usually used only on multistory buildings. In single story buildings, the floor is usually a concrete slab poured against the ground, unless the building has a crawlspace or a basement.
 Data here is from a variety of sources. Span, depth, and weight data is from manufacturers’ provided data, information provided from other engineers, and calculations done in engineering software. Cost data is from personal experience, professional estimating references, manufacturers data, and a few other sources.
 Cost information is often difficult to acquire, especially for less common systems - more often than not, suppliers like to tout other, non-specific benefits like site speed or improved coordination rather than direct cost. So a few systems are omitted here.