I wanted to make a shelter for a travel trailer (which is only about 98% waterproof) alongside my house. The house is wood-frame built on a concrete slab on compacted sand, but the original surface is peat; the trailer stands over the transition from sand to peat. The implications for a structure are that the two sides would move independently over time - a gatepost in the peat has sunk about a foot in a couple of decades. A cantilevered structure that had no supports bearing on the peat seemed a possible solution, but the weight would be too much for a pure cantilever - at least with snow loading taken into account. So I decided on a partially cable-stayed structure.
The structure was intended to be very light, and relatively low cost. In the end the roofing material came in at some $500, with a few hundred more for wood, glue, screws, brackets and cable. The spruce purlins for instance are made from three 8ft sections each costing about $1.70, while the box beams needed two sheets of OSB costing about $12 each.
Step 1: Analysis
In reality, I built the structure first and did the analysis later to justify my guesswork, but it really ought to be done first. So here it is.
The roofing material is Palruf, a corrugated PVC solution available locally at reasonable cost. The total area of the structure was originally intended to be 25ft by 13ft, but Palruf comes in 12ft x 26" panels, intended to be overlapped on 2ft centres, so I downsized slightly to 24ft by 12ft - about 26.75sq.m. Palruf panels weigh about 2.8kg, or 33.6kg for 12.
Palruf recommends a maximum purlin spacing of 2ft. I used 6 purlins of nominal 1x3" spruce bearing on 6 box beams. The spruce comes in 8ft lengths, and 18 lengths weigh 25.2kg.
The box beams are constructed of 1x2 spruce with verticals of 3/8" oriented strand board. Each weighs 9kg assembled. The weight of the entire structure is thus about 113kg.
The location is fairly sheltered from wind by the house and by trees. I did not consider wind loading. Winters are generally mild with little snow. I considered a snow load of 20cm of fresh snow, with a density of 250kg/m3. The total weight of snow at 50kg/m2 is thus 1337kg or 13123N.
The first panel shows an analysis of the spruce purlins, evenly weighted by snow on the roof. The beams are spaced every four feet. Half a purlin is shown; the full purlin includes a mirror image on the right. The analysis suggests that the leftmost beam should be moved some 6 inches to the left so that there is less overhang. However, the beams are constrained to the 2ft pitch of the supporting wall studs in order to be well secured, and the stress is not unreasonable.
The second panel shows an analysis of one beam, loaded with the maximum purlin weights from the first step. The total stress in the supporting cable is some 1940N, while the working load of the cable is 900lbs or 2448N. I used 3/32" steel cable, though in retrospect a thicker cable would have less stretch and a greater safety margin.
The third panel shows, for comparison, simulation of the beam without cables and without a buttress. In all cases the displacement shown is scaled by a factor of 20 to make it easier to see.
Step 2: Flashing and Wall Support
The house wall is made of plywood over 2x6 studs on 2ft centres, covered with cedar siding. Even though the wall is protected by a wide soffit, adding flashing seemed like a good ideas as it was not very expensive, and fairly easy to fit. This is L-section aluminium flashing, pushed under the lip of a cedar siding board, which dictates the vertical position of the support. The horizontal support is 2x3 spruce, screwed through to the wall studs. The vertical supports to take the beams are also 2x3 spruce, centred on the studs.
It is important to ensure that the eyebolts to support the cables are centred in the hidden wall studs for maximum strength. Though the siding is nailed through to the studs, there is no guarantee that the nails are centred. Accordingly I drilled a series of pilot holes to locate the edges of the studs using a stiff wire, then drilled an hole to prevent splitting and screwed the eyebolt dead centre. The cables are fitted with thimbles and secured with two cable clamps (only one is shown). The pilot holes were later filled with sealant.
Step 3: Assembling the Beams
The support beams are each 12ft long box beams, made from 2x1 spruce and 3/8" OSB. The OSB comes in 8x4ft sheets, while the spruce is in 8ft lengths, hence there are joints. The spruce members are joined with a nailing plate, and with a 1ft length of spruce glued and screwed across the join. The OSB join is a simple butt joint with no tensile strength. The OSB sides are glued and screwed to the spruce at regular intervals, and a section of 2x3 spruce is added internally at the wall end to provide support for the screws securing the beam to the vertical members. There is an additional piece of 2x3 spruce internally to provide more support for the cable eyebolt. In retrospect, there could have been another internal support to take the buttress screws. Total weight, 9kg (easily lifted in one hand).
Step 4: Raising the Beams
It proved surprisingly easy to raise the beams into place.
With the eyebolts in place in the wall, I rigged a block and tackle from the eyebolt. Then with one screw through the beam to act as a pivot, I raised the beam into position with the rope and then fitted the buttress. The diagonal member is a temporary support to hold the beam correctly at 90 degrees to the wall horizontally. The buttress is 2x3 spruce, secured to a steel hanger screwed to the wall and screwed and glued to the beam. The bottom end of the buttress is also secured with a lag bolt into the wall stud.
Step 5: Tensioning the Cables
With the beams in place, and cables fitted to the upper eyebolts, I threaded the cable through the lower eyebolt and tensioned the cable with the block and tackle before securing it with cable clamps.
The photo shows an early test; in the actual assembly I drilled clearance holes in the Palruf material to take the tensioned cable and threaded it through before securing it. I passed the cable through the roof, throug ha cable clamp, through the eyebolt, back through the clamp, back through the roof and made a loop in the end secured with another cable clamp. This loop is joined to a pulley to that the whole cable can be tensioned by pulling on the rope, and then the first cable clamp tightened before the rope is removed, and the loop and second clamp removed along with the pulley. Later, I added a second cable clamp to both ends of all the cables for extra security. In retrospect, a turnbuckle could have been used for final adjustment, or a come-along (with greater mechanical advantage) used to tension the cable more effectively.
Step 6: Adding the Roof
The spruce purlins are secured to the top of the beams with L-brackets. The three 8ft members comprising one 24ft purlin are joined with nailing plates, more to keep them aligned than to provide significant mechanical strength. The nailing plates were pressed into the spruce with a G-clamp rather than hammered in.
The Palruf sheets are secured to the purlins with roofing screws, which have rubber washers pre-fitted. The material is quite brittle and cracks easily, so clearance holes are drilled. At the wall end, the flashing is secured to the horizontal support through the Palruf sheet.
Diagonal members prevent the assembly from skewing, though in retrospect I think the roofing sheets are entirely adequate for this purpose.
Step 7: Modifications
I found I had miscalculated the height of the roof, and had to raise it by decreasing the slope - the wall end was constrained by a window, so I could not raise the entire structure (which in any case would have been more difficult). This meant I had to move the buttresses and re-tension the cables, while accessing them from the roof rather than from the side while I fitted each roof sheet. It was relatively easy - the assembly is strong enough to walk on with duckboards across the purlins (I weigh about 80kg, rather less than the designed snow load). To support the extensions, I used 1x2 spruce, glued-and-screwed to the beams (in part since I had a surplus of 1x2 due to an earlier miscalculation), with a seventh 1x3 purlin.
I also decided to extend the width to about 14ft from 12, i.e. slightly more than the original design of 13ft. That meant adding an extra section of Palruf, by sawing 2ft lengths. As stated, the PVC material is brittle and hard to cut, and I did not do a perfect job. Cutting with a circular saw with fine teeth at high speed seems better than cutting by hand, but I still ended up with some splits. Creating a design that matches an available length seems better than planning on making cuts.
As stated earlier, I actually did the analysis after constructing most of the shelter. The cables are adequate to bear the entire weight of the structure plus snow load, but I had not allowed for the elasticity of the steel. Also, the upper eyebolts for the end pair of cables are lower than the rest to allow for the house roof. This means that the tension on the end cables is more than on the others, and the analysis suggests that the stress on the wall supports is actually outwards from the wall. The cables are actually pre-stressed (but to an indeterminate amount), so that may not be entirely accurate, but I decided to add cross-bracing cables to transfer some of the load at the ends to the next, higher, eyebolts so that the final configuration is like in the sketch. In retrospect, I should probably have oversized the cables from 3/32 inch to 1/8 inch or larger.