Why are deck joists supposed to be on top of posts, not bolted to the side?

Question

I remember reading that deck joists shouldn't be mounted to the side of posts, but to the top, and this is a common mistake in deck construction. The reason I found was that you'll need a bigger bolt for the shear load, and that even then, the wood in the post might not be strong enough, and the bolt would pull through the post, almost like a cheese cutter.

That makes sense, but I saw something on bolts in metal. If properly torqued and pre-loaded, there isn't any shear load on the bolt; all the shear load is resisted by friction in the joint.

Does this not apply to wood? Is it because wood changes so much over the seasons and over the years? Is it because people are unlikely to properly pre-load? Is it in case of failure, so that when the frictional force is no longer enough, the joist is still safely supported?

Answer 1

For a steel bolt in steel beams there is no differential thermal expansion (the beam and bolt expand and shrink the same with temperature changes) and no hygroscopic swelling and shrinking, since steel does not change size with humidity.

NEITHER of those apply to steel bolts in wood, which has a different rate of thermal expansion and is hygroscopic and changes size with the amount of moisture in it. Both of those things vary wildly throughout the year for a deck outdoors.

If you try to "fix" it by just cranking up the bolts, you'll break it by crushing the wood.

You could, of course, use steel framing for your deck. But it will likely cost you more, at least in the short run.

Answer 2

This is an engineering thing, and I don't pretend to totally understand it. But there are plenty of instances, beyond just decks, where how things are connected/positioned can make a huge difference.

The first example that comes to mind is the Hyatt Regency Walkway Collapse. In this particular case, there was supposed to be a single rod going through multiple levels and instead it was split into two. That resulted in double the expected force on some nuts, which eventually led to a truly catastrophic failure. It was a relatively minor shortcut in the construction process and not something that would be an inherently obvious problem to a non-engineer.

Similarly, the stresses on different pieces of hardware and lumber can vary quite a bit just due to positioning - side-by-side vs. top/bottom, etc. Decks are notorious for problems because they seem easy to build, they may be built totally incorrectly and yet feel quite sturdy initially, and many of the potential problems (connection of joists to building, connection of posts to joists, etc.) are hidden once construction is complete.

As a result, there are a lot of very specific rules regarding deck construction, often including local variations to deal with earthquake zones, different soil types, etc. There are a ton of places where you can (and many do) cut corners with little or no ill effect, but deck construction more than a few feet above the ground is definitely not one of them.

Answer 3

In addition to the good answers by others, wood is slippery, especially when pressure-treated and juicy. You cannot count on the same coefficient of friction as you'd get with steel, and it varies wildly between dry and wet conditions. It can be that simple.

Answer 4

Contrary to other answers, here in NZ, they can be. There are various ways to mount and fasten the joists, which define the dimensions and quantity of timber and fasteners.

1. On top. This is the safest, especially if dealing with poor tradesmen.

2. Fastened to the side, can be one, or one on each side. If done this way, we have to notch the post, so that the joist sits on timber and cannot sag/move. I have done this on my last project which was a deck on the side of a lake, and cantilevered into it. This also requires washers on the bolt heads and nuts, so that they don't sink into the wood.

3. Using joist hangers. They exist for a reason. These are hanging brackets meant to fasten joists to the side of the post.

Answer 5

I'm not able to use side-bolts for my decking because I am using steel posts, and am required to put electrical insulation between the wood and the steel. This is because the wood has been treated with a copper compound to resist decay, and when in contact with steel causes accelerated electro-chemical corrosion.

The isolation layer prevents effective clamping between the wood and the steel. A cylindrically isolated bolt can be used to "pin" the wood and steel together, but it won't "clamp" the way a torqued bolt is used to replace a rivet on a riveted bridge.

There are other reasons as well, but they don't come into it, because the standard I have for wooden decks requires that isolation layer.

Answer 6

Consider the stress on the wood AND on its supports, as well as the characteristics of wood. Wood is crushable and splittable; it swells and shrinks, twists, warps, cups, and rots. I know a retired building inspector; he pointed out that most building code rules are there because someone died. For some strange reason, they try to prevent recurrences...

Try imagining a proposed structure with the wood gone soft. How and where does it fail?

• When there is a large bearing area, the stress is low. Joists sitting on a post have a large supported area. This does not promote crushing, splitting, warping, etc.

• When there is a small bearing area (e.g. the joist is being supported by the side of whichever one bolt is slightly higher than the other), especially with a varying load: that can cause crushing or splitting. Noting that a splitting axe, wedge, etc. works by applying high force in one spot.

• When a load is supported on the side of a post instead of the top, there is torque: force is applied to (on average) the joist center, however the joist is supported on the side where it is against the post. This will:

  1. apply a sideways bending force to the post, and
  2. a twisting force to the joists, tending to warp them, and
  3. try to twist the joist off of the post, and
  4. with shrinking, swelling, time and rot, the head of the upper bolt will sink into the wood; the joist will begin to twist off and that increased distance (top of joist to side of post) will increase the torque.

The people writing the rules have seen decks after 10, 20, 30 years, and how they fail.

Answer 7

There are a lot of reasons for this, but there are ways to deal with them, like a pre-engineered hanger or bracket.

Short answer is Physics and Engineering principles.

Longer answer is this: Using a bolt, vs setting the joist edge on a hanger or top surface of the post means that the forces are transferred through the bolt. for the joist that means that the effective depth of the joist is at whatever depth you have drilled the hole. If it's the center, then your 2x10 just became a 2x5 effectively. (Think similar to notching the end of a joist) Also the bearing surface became MUCH smaller, instead of having 8 square inches of surface for all that force to transfer through (width of joist on top of the width of the post) you have (I am estimating here based on a 1/2" thick bolt) .25 square inches (thickness of joist x bearing surface of bolt) now all your forces are being transferred through a surface that is 1/32 of the size. Now you are getting into the crush limits of the wood fiber. It also results in forces bieng altered. If you have the bolt in the center of the joist then there might be an induced twist, and bolts can and will bend, reducing their strength. Also, twist on the joist moves the load off axis rfirther reducing the capacity of the joist, and inducing a different loading at the end of the joist. All of which reduce the capacity of your joint, and joist.

So the chances of your joist splitting at the bolt, and the hole getting bigger because of crushing, you no longer have a stable joint. Wiggle means dynamic forces and shock loads. Its just a matter of time before it fails.

Caveat: I estimated the bearing surface of the bolt due to the surface being curved, and forces transfer normal to the surface. So your vertical force component won't be evenly distributed across the entire curved surface. So it could be more, or it could be less. For the purposes of the answer the exact amount isn't entirely important, just the fact that it results in a much smaller bearing surface.

Longest answer gets into static and dynamic load analysis and involves a crap ton of math that you need an engineering degree (and a cheat sheet) to remember, and involving a material with hugely varying characteristics (type of lumber, moisture content, age, grain, cut)