I reversed several decisions, one of my mistakes cracked the pier, and I never resolved two questions. This chapter collects them.
The one that damaged something
Day one after the pour, I took the top rotor off to clean concrete splash from both discs. That part went fine. Putting a nut back on, a thread picked up some concrete and the nut seized, and I torqued it anyway.
The rod turned in concrete that was about 24 hours old. Set, but nowhere near strength.
What followed: a short vertical crack at the top of the pier beside that rod, and a larger one roughly two feet below, about where the buried nut sits. Both consistent with a rod rotating and splitting concrete that could not yet resist it.
Why it was recoverable
The buried nut and fender washer on each rod provide mechanical anchorage: they resist pull-out by bearing against concrete, not by bond along the shank. Twisting the rod breaks the bond. It does not remove the nut.
That detail is why this was a setback and not a demolition job. It is also the argument for specifying it in the first place: it costs a nut and a washer, and it limits the damage to one rod instead of the whole pier.
It is also one rod of four, and the rods do not carry the mount's weight; the concrete does. The rods clamp the adapter down.
What to do instead
Never torque against a seized nut; cut it off. Never put a wrench on a rod that has twisted in green concrete, because more torque only enlarges the damage. And do not fill the cracks immediately. Filler adds no strength, and while the cracks are open they are your instrumentation, the only way to tell whether anything is still moving. Mark and date both ends, measure the width, and re-check at 14, 21 and 28 days. Static cracks are cosmetic; growing ones are not.
Reading the two cracks
The two cracks have different causes, and you should be able to tell them apart.
The lower one is ordinary shrinkage. It wanders, changes direction, crosses the form's spiral seam without following it, and branches. Structural cracks run straight and follow stress. Shrinkage cracks meander. Every 12″ column does this.

The upper one is the rod. Vertical, running down from the top edge, tapering as it goes. But it is hairline, and there is no cone-shaped pattern radiating from the rod, which is what an anchor pulling out would produce.

The design reversals
Ring blocking on the wrong lines
My first framing plan put the dome's ring blocking on convenient grid positions. It looked fine in a solid view. When I rendered a framing plan, it was obvious the blocking was sitting in the field carrying nothing, while the 88″ circle where the dome's load lands had wood under it only by accident.
The bay that fit by 0.6″
An early 10 × 10 layout put the bay at 45° with 0.6″ of clearance, and I called it a fit. Sweeping every angle showed 45° was the only angle that worked. What fixed it was noticing that the dome did not have to be centered at all.
An adapter that flexed more than the pier
My first adapter design used stacked washers as spacers across a 3¾″ gap. Modeling it showed the adapter accounting for 68% of total system compliance, flexing more than twice as much as the concrete pier under it.
None of that was visible by looking at it. Four ½″ rods look substantial. Bending stiffness goes as diameter⁴ and compliance as gap³, and those two exponents were doing all the work.
Seven washers as a spacer
After the rod incident, I bridged the damaged thread section with seven stacked washers. That is a springy load path: washers dish under clamp load and the joint loosens over time. A solid steel sleeve cut square to the gap does the job properly.
The trammel that could not exist
I proposed three methods for centering the wall ring before one worked.
The first was a trammel pinned at the pier center, which is impossible because the center point is inside the pier.
The second measured from the pier's surface. It is geometrically sound, but it produced a 1″ spread because a concrete surface formed in a cardboard tube is not a datum.
The third measured the ring against itself using chords, and worked immediately.
Hanging the cage from the template bar
An early sequence had the rebar cage hanging on wires from the same crossbar holding the adapter. It does not work: the bottom rotor is 11″ inside a 12″ form, leaving no clean path for hanger wires, and anything threaded through gets pinched when the rotor beds.
Two plywood templates
Another version of my plan called for two plywood templates 12″ apart to keep the rods plumb. The rods only project about 5″ above the pour, so there is nowhere to put a second template.
The smaller corrections
| Claimed | Corrected |
|---|---|
| Diagonal is 169¾″ | 169-11/16″ |
| Cut footing cross ties to 22″ | 11″ (a 13″ hole cannot take a 22″ bar flat) |
| Paint the whole adapter against rust | Leave embedded steel bare: concrete passivates it, and paint kills the bond |
| Use cutting fluid on cast iron | Drill it dry: graphite is self-lubricating; oil makes abrasive paste |
| Anti-seize every thread | Never on anything the concrete touches |
| Trim-head screws for decking | Bugle head: trim heads pull through |
What I never resolved
Two questions are still open. If you adapt this build, do not close them by assumption.
The NexDome door swing. I never confirmed with the manufacturer whether the door opens inward, outward, or slides. The 24″ apron is comfortable for a sliding or inward door. An outward swing needs clearance equal to the door's width, and 24″ falls short of that. This is a question for NexDome.
Wind anchoring on a floating deck. A 250 lb dome is a sail, and a floating deck resists uplift only by its own weight. Four ground anchors is a reasonable floor, but it has no manufacturer specification or engineering behind it.
The common cause
Almost every error above came from a number that looked authoritative and that I had not checked against the thing it described. Where I rendered the view, ran the arithmetic or measured the reference before committing, the fix was cheap; the expensive cases were the ones where I did that only afterward.