[Author's Note - By popular demand, I'm posting this Builder's Engineer article. It was previously published in NAHB's Nation's Building News, and can be found in my book, CRACKS, SAGS, AND DIMWITS - LESSONS TO BUILD ON, available at Amazon or this website. The concept of ridge beams and collar ties continues to perplex.]
Dear Builder’s Engineer,
My house was built in the early ‘70s, using 2×6 rafters spanning 13 feet between ridge board and wall. I live in New York State with a good-sized snow load. The roof has developed a sag between the ridge line and the rafter tails. I attribute that to the rafters being undersized and sagging in the middle. The sag is not horrible; I would guess it’s about an inch, maybe a little more. Still, I would like to do something now rather than wait to see if it gets worse.
A few details:
- roof pitch is 4 on 12
- a plain ranch style house…no hips or valleys…just front and back
- rafters are nailed directly to the ceiling joists
- rafters and joists are 2×6
I’m wondering if collar ties would help. I could take 12-foot 2×6’s and position them underneath the paired, opposing rafters, miter the tie ends to fit under the rafters and then use tie plates or gang nail plates to attach the ties to the rafters. I’ve read that is stronger than just nailing the collar tie to the side of the rafter. I thought that it would be best to get the collar ties supporting the midpoint of the rafter or as close as possible, that’s why I thought 12-foot length collar ties would be better.
Any thoughts or suggestions would be appreciated.
Daniel in NY
Dear Daniel,
Following is a sketch of your roof framing as I understand it. I analyzed this using a computer and the following assumptions:
- Snow load = 35 psf (pounds per square foot)
- Dead load from comp roofing and self-weight of framing = 15 psf
- Spacing of rafters = 2 feet
First, a little theory:
Truss vs. Rafter. There are big differences between rafters and trusses. A rafter bears at both ends; typically on a wall at the low end and on a ridge beam at the high end. There is no outward thrust at the low end of a rafter.
But what about a rafter that really isn’t one? I’m speaking of a sloping “rafter” with a pitch of 2:12 or greater, with no ridge beam—usually a puny 1x or 2x instead—and no connection to a ceiling joist, truss bottom chord, or other horizontally-restraining member at the low end. If you’ve got this, you’ve got trouble. Without the support of a sturdy ridge beam at the high end, there is nothing to keep that high end from going down. The low end can’t go down because it’s sitting on a wall providing vertical support, so when the ridge sags, the low end must move outward. Bad, bad situation. At the end of this chapter is a case study exploring a fix for such a “non-rafter” system.
The absence of a beefy ridge beam is fairly common, but in such a case, the low end of each (rafter) must be connected to some other member providing horizontal restraint, such as a ceiling joist. We call this connection the heel, and the overall system a truss. In this case, the rafter is no longer called a rafter, but a top chord.
A triangular truss has very large forces at each heel. It is the heel connection that keeps the top chord’s low end from moving outward. Also, the top chord of a truss has two kinds of stresses: bending and compression; whereas a rafter has only one: bending. Basically, a triangular truss has to work very hard because it is paying the price of a long span with no interior support.
So whenever you come upon a distressed truss, the best remedy is to add interior support(s), thus lessening the span and member stresses. You can bolster trusses, but without additional support, you’re simply spreading stress around—stress that must still be dealt with, and which becomes troublesome, particularly at connection points.
Let’s look at the truss in question. Note the 2×6 top chord is 217 percent overstressed and will sag 1.6 inches under a full snow load. The force at each heel is 1,700 pounds which requires twelve 16d nails to keep the top chord from moving outward. It is a miracle this roof hasn’t imploded.
Can anyone tell me how to cram twelve 16d’s into the small overlap space where the top chord and bottom chord (ceiling joist) come together? I think if you tried, you’d massacre the wood so badly, none would be left to hold the nails. It is for this very reason that gang-nail plates were invented.
Everyone wants to fix their sagging trusses with collar ties, presumably because they’re relatively easy to install. Here is the subject truss with a collar tie at the top chord mid-span.
Now the top chord is in better shape: only 70 percent overstressed with 0.82-inch sag. But look at the connections. Each collar tie connection point must resist 2,100 pounds of force, which would require fourteen 16d nails. Can’t be done. Also, the collar tie is in such compression that a 2x won’t cut it; a 4x is needed. But even more troubling is what happens at the heel. The top chord/bottom chord force has ballooned to a whopping 2,850 pounds. No way to make this connection. In short, the collar tie took load from the top chord and shuffled it around to other places, but those other places can’t take it.
When this column originally ran, I suggested the following upgrade: sister 2×6’s on to the existing rafters, like so:
This reduces the stress and sag of the top chord by half, but more importantly, does not increase the load on the heel connection. Note the sistered 2×6’s need only cover the middle 70 percent +/- of the rafter, not the entire span. This is because the sag is a bending problem (as opposed to shear, tension, or compression), which occurs in the middle two-thirds or so of the top chord. Bending stresses go to zero at the top chord ends; thus, no bolstering is needed there. I like this solution better than the collar tie. For more on beam theory, see my book, Basic Structural Concepts for the Non-Engineer, available at www.constructioncalc.com.
If we really want to solve the problem, we’ll find additional interior support. Like so:
Note that this completely solves the sag and top chord overstress problems, and it hugely reduces the heel connection force. The rub is making sure the interior support can truly take the load. In our example we’re adding about 600 pounds per foot (that comes from 1,200 pounds per truss, which are spaced every two feet) to the interior supporting wall—no small amount. If this wall has a continuous footing below it, it is probably okay. If this wall has large openings in it, those must be spanned with a beam, and the ends of said beam(s) need proper support all the way to a proper footing.
I once was involved in an old schoolhouse with a badly sagging roof. It was a hip system, about 8:12 pitch. There were no trusses, it was a rafter system; but amazingly, there were no beams at ridges, hips, or valleys. So the outward thrust at rafter heels had to be taken by the walls, but they couldn’t, so there were big outward bulges in the exterior walls. It is a miracle this building didn’t implode in one of our snowy northwest winters—a testament to the toughness of wood. Here is a sketch showing the problem and the fix (shaded).
We did our best to winch the exterior walls back to plumb and jack the roof back up. Then we installed a series of beams directly supporting the rafters, ridges, hips, and valleys. Of course, the new beams had to bear on something, so we positioned them over existing walls below and then retrofitted footings in the crawl space where the new loads came down. Interestingly, the contractor’s first suggestion was to install collar ties as a fix. No, no, no.
In summary, there is no easy fix for an improperly designed roof framing system. Collar ties are almost never recommended. Rather, find a way to add interior support, taking loads all the way to a good footing. And certainly, the best alternative is to design it right the first time.
how to go about sizing the purlin as a mid-span beam when there are only bearing walls at the gable ends and not interior bearing walls for kickers?
thanks
Hi kjward,
I don’t understand your question. Are you sure you mean “purlin”? That’s a 2′ OC member attached to the top chord of a truss, perpendicular to the truss, common with post-frame buildings.
hi tim-
it is a term also used to denote an intermediate mid-span-ish support for over-spanned rafters. the mid-beam added as shown in the last sketch above could be considered a purlin. they are usually supported by braces to an interior bearing wall, much like the kickers in the next to last sketch above.
my question is how to size the beam to support the rafters mid-span when it is only supported at its ends?
thanks tim
To size a beam that supports the middle or so of rafters, I’d use ProBeam by ConstructionCalc. The span is the length of the beam, which could be long if it goes from gable to gable. The load is a uniform load over the length of the beam and the tributary width is half the distance to the wall on one side + half the distance to the ridge beam on the other. If there is no ridge beam then the trib width is half the distance from exterior wall to exterior wall where the rafters bear. Dead load will be about 10 and snow load will be what it is in your neck of the woods – around here it’s 25 psf. That’s all there is to it.
Tim,
I’m in Maine and I’ve discovered a potential disaster in a 24 x24 garage, built in 1991, on my father-in-laws property. The code calls for 50PSF snow load and the garage is built almost exactly as your example above: Stick built 24 foot wide (door on gable end), 24 foot 2×6 ceiling joists on 48″, 13 foot 2×6 roof rafters on 24″, 5:12 pitch, all lumber #2 SPF. Roof has plywood sheathing and 1 layer asphalt, the ceiling is unsheathed. Note the 24′ ceiling joists are every other rafter with 2×4 high collar ties on the odd rafters (useless), rafters meet at peak on (non structural) 2×6 ridge beam. Walls seem OK: 2×4 on 16″, sheathed in T111 with double top plate. Each rafter heel is secured to the joists with only 4 12d nails. Front 4 rafters have been elevated about 6″ to clear GDO drive rail (so they are low collar ties). There are 2×4 wind braces at each gable end. There is sag evident in the joists, a little sag (1″) in the mid ridge but the walls have not bowed out (yet!).
I want to retrofit this structure before our luck runs out. We can’t add any mid-span posts because we need the clear space.
After I add the missing odd rafter ceiling joists, I am considering either retrofitting a stick built Pratt truss as shown here: http://www.public.iastate.edu/~mwps_dis/mwps_web/plans/truss_24.pdf
(I need to know if the current heel connections are feasible in this truss design)
OR would I be better off by sistering 10′ (2x?) onto the rafters? If that is the best way, how should I reinforce the joist heel connections?
Utimately we want to use the ceiling area for LIGHT storage (< 10PSF). But I recognize that we have to support the potential snow load as a first priority!
Thanks, snowman
Hi Snowman,
I would go the truss route. The challenge will be the heel connections because the rafter and ceiling joist are side by side. A way around that is to sister a 2x, say 2′ long to either the rafter or joist to flush out a single face upon which a plywood gusset would be attached. I’d glue the 2x in place and glue (Liquid Nails or similar) the gusset as well, then connect everything with SDS lag screws of appropriate length so that each member is penetrated at least 3/4 of the way through. If you got 5 or 6 SDS screws in each the rafter and joist, you’d be golden.
Tim,
I’ve been considering the truss retrofit discussed above and I’ve become concerned about the gusset at the ridge (Gusset “B” 10″H x 12″W in the link above). Because there is a ridge beam, I will need to notch the ridge gusset to fit around the 2×6 ridge board. I’m assuming this will compromise the truss design. If this does compromise the design, do you see a way around this? My impression is the king post is not under much stress but the webs are under major tension.
The truss was first designed in the 60′s (with revisions thru ’89) and they recommended using Casein, Resorcinol or Epoxy resin as the glue. While I’m willing to use epoxy, it is messy and will require mixing many batches. In view of this, I’m considering using a more modern adhesive such as Loctite PL Premium or PL and I’m wondering if it will be strong enough. I don’t have the calculated stress values. All I have to go on is the recommended gusset size for each joint (in the link), the nailing pattern and the stress assumptions for the plywood (from the MWPS9 publication).
Plywood stress assumption: 250 psi shear and 53 psi for rolling shear.
Nailing pattern: 5d or 6d box galvanized; 4″ spacing along grain, 2″ spacing across grain.
The nails are also used to hold the joint tight while the glue cures.
Thanks for any advice you can provide. – snowman
Hi Snowman,
The gusset at the ridge has almost no stress in it. A moderate notch there isn’t a problem. If the notch gets too big, say more than 25% of the top chord depth, use thicker plywood gussett to compensate.
Regarding that old school glue / epoxy, yeah it’s a mess and I don’t recommend it. Your Loctite is good. Here’s my standard spec on wood glue:
Glue. Where specified wood glue shall be commercial grade with minimum shear strength of 450 psi at 28 days. Use Liquid Nails LN-940 or equivalent. Prep and apply per manufacturer’s recommendations.
Tim,
Thanks for responding to my truss question.
->You said:The gusset at the ridge has almost no stress in it. A moderate
notch there isn’t a problem. If the notch gets too big, say more than 25% of
the top chord depth, use thicker plywood gussett to compensate.
Unfortunately the ridge board is the same depth as the top cord; both are
2×6. So the notch would be ~100% of the top cord depth. My only hope is
there is no significant stress (assuming equal roof loading) across that
joint. I’m hoping the actual path of the tension force from the web is only
passed to the top cord *above* the web, rather than across to the top cord
on the other side. Do you see it that way too?
Can your structural software calculate Pratt truss stresses? If so, what is
the magnatude of the stress at the ridge gusset and the heels when the roof
is fully loaded with 50psi live + 10psi dead. With that information I can
properly scale the increase I may need in gusset surface contact area, so
the design integrity is maintained. With cured epoxy shear strength at
over 3000psi and the Loctite PL premium at about 500psi, I better check the
numbers.
Thanks, snowman
TIM , you are good.
I have been reading a lot on this site. All of you have a lot to bring to this forum and most important is the help you give to all that want to receive it .
Here in northeast lots of roof structural failures have occurred over the last few years. Some from “contractors want to be” that refuse to get an education of structural engineering and or willing to read or apply any of it . I think some of this is disbelief in how things are in the real world .
The spans in clear openings beams for weight carrying roof design is in shambles. Some buildings people buy a mess or something a little off the edge.
I have noticed some of you are jointing/connecting lumber for structure such as roof design/repair/strengthening . Gluing with pl and gusseting using steel 12 gauge and thicker with thru bolts ,nuts washers is best . This is good for multiple pieces of 2X and greater thickness wood. For 2X and less ;Next to that is the glue and right kind of plywood with deck screws with washers to get more squish . Epoxies and resorcinol have their uses . Epoxy will crack and not bond well to some wood at all conditions that a “wood” may in it’s life time in a structure. Yes there are 3000 to 5000 pound epoxies out there but that is for the epoxies only not the bond or lack there of to wood that is flexing and or changing size from natural environment conditions. You all know that water soaked MDF is like oat-meal. So like the glue is the material good for the intended purpose. May I suggest that you build a test connection then do destructive testing. Like the only sure way to keep roof gutters from freezing is move them close to earths equator. No , heating then with free energy will not cure all. The heating system may fail. Test and prove your designs.
GUTS
Hi Snowman,
I thought the ridge board was under the top chords, but now I understand that it is truly a ridge board and the top chords butt up to it. Can’t get a gusset on that. Here’s a case where a collar tie can actually be useful. Can you use a 2×4 collar tie snug up against the bottom of the ridge board? This can be nailed to the top chords with 16s or SDS screws. This becomes your gusset.
Regarding my software’s ability to analyze trusses, no, it doesn’t do that. ProBeam designs any single beam, joist, or rafter. Truss software is massive, beyond what ConstructionCalc is about. In fact when I have a truss to analyze, I farm that task out to the local truss company who can bang that analysis out in less than an hour. Without the software it takes 4+ times longer.
Hi Guts,
Welcome to the site, good to have you.
I couldn’t agree more that most guys in the field are sorely lacking in basic structural concepts. This site, books, and software is devoted to them.
I find that lack of knowledge in this area generally costs guys $ and jobs. Most builders are aware that they don’t really understand the concepts and so to compensate overbuild. In these tight times, who can afford that?
I was at a big sawmill yesterday helping the plant foreman with a steel cover for a new boiler. They welded roof trusses together with angles and flat bar! The county got wind and red tagged them. Now I have to somehow salvage this mess. I’ll do it, but it’s going to cost double what it would have if they’d have come to me before cranking up the welders. I see this kind of inefficiency every single day. It just tears me up.