Ground Mount Racking Installation Technique

Why this matters

A ground-mounted array trades roof penetrations for a structure built from scratch in dirt, and every mistake a roof mount hides under shingles a ground mount exposes at eye level for the life of the system. A footing poured short of frost depth heaves a fraction of an inch every winter until the rail line is visibly out of plane. A post set out of plumb loads its footing off-center in a way the engineering never accounted for. None of it is correctable after the concrete cures; a ground mount's structural mistakes get fixed by demolition, not by a service call. The technique is unglamorous and entirely about sequence, and the cleanest way to run it is to build one continuous site record as the structure goes up, filling in each field only once its own check has passed.

Record one: the utility locate and the layout check

Before the first stake goes in, get utilities located and confirm the marks against the site plan; this is the first physical action on any excavation, not a formality specific to solar, and skipping it on a residential yard is exactly how a buried gas or communications line gets struck. Where the site carries its own private lines, irrigation, a septic field, get those located separately, since a public utility locate does not cover them.

With the locate marked, stake post locations from the engineering letter's spacing and stringline the whole run before digging a single hole. Log the check two ways: measure the diagonal of each rectangle the posts form and confirm both diagonals match within a small tolerance, and sight down the stake line from one end to confirm it reads straight rather than a shallow arc. A frame laid out from one fixed point without the diagonal check can drift an inch or more by the far end, and nobody notices until the rails refuse to seat flush, by which point every post in the run is already set. Record: locate ticket number, both diagonal measurements, and the sight-line result, before any hole is dug.

Record two: footing depth against both numbers, not one

Footing depth comes from two figures that rarely agree: the structural depth the racking manufacturer's engineering letter calls for to resist this site's design wind uplift, and the frost depth published by the local building department. Dig to whichever is deeper. A footing that satisfies the structural depth but sits above the frost line still heaves, because frost heave is a function of how much saturated soil freezes and expands beneath the footing, not of how much concrete resists uplift. Record the letter's structural depth, the jurisdiction's frost depth, and the depth actually dug, so a later inspection can see which number governed rather than assuming the deeper hole was an accident of soil conditions.

Where footing depth approaches or passes five feet, the excavation itself becomes a protective-system trigger under 29 CFR 1926.652(a)(1), not just a deep hole to pour into, and nobody enters it without that protection in place; keep any open excavation barricaded or covered on an occupied residential site until it is backfilled, since an open hole at grade is a trip and fall hazard to a household that was not there when it was dug. Wet concrete is caustic to skin for the duration of a pour, so gloves and eye protection stay on through mixing and placement, and where dry mix is cut or poured in a way that raises dust, that is a silica route controlled under a respirator program per 29 CFR 1910.134, not a nuisance dust mask.

Record three: plumb at set, verified again before it leaves your hands

Set each post and check plumb in two directions with a level held against the post face, not eyeballed against the rest of the row. An out-of-plumb post carries its design load off the post's own axis and into bending the structure was not engineered for, and the error compounds with height: a post a degree or two off plumb at the base reads inches off at the top, exactly where the rail has to seat true to its neighbors. Handle posts and rail stock as a two-person lift past a length either person can control alone, and record plumb readings in both directions before bracing, not after.

Brace every post front and back before backfilling or pouring, and do not pull the braces until concrete reaches the cure the mix supplier specifies for load, not a calendar date remembered from the last job. Do not shove a braced post to test it; a post that moves under hand pressure before cure was going to move under wind anyway, and finding that out is not worth a foot injury from a falling post. A rack loaded onto concrete before it reaches working strength can settle unevenly under its own dead load alone, before wind ever tests it.

Where the design uses ground screws instead of concrete piers, record torque instead of cure time: drive to refusal or to the torque the manufacturer's installation chart correlates to the required holding capacity for this soil class, logging the reading at each screw, and check plumb the same way as a set post. A screw driven fast and shallow reads a torque number that looks fine on the driver and holds nothing once wind loads the frame; a reading short of the chart's correlated capacity means the screw is backed out and redriven at a new point, or reinforced per the manufacturer's stated remedy, not left in place because it stopped turning.

Record four: rail attachment across the whole run

Once posts are cured or torqued in and confirmed plumb, set the beam brackets or post caps that carry the rail, leveling across the full run rather than post to post; a run leveled in short sections can drift gradually from one end to the other with every individual step passing its own local check. Torque bracket bolts to the manufacturer's spec and paint-mark each one once torqued, the same discipline used on the module clamps that go on later. Leave the thermal expansion gap the rail splice hardware specifies for the run's length; ground-mount rail runs are typically longer than a roof row and see the same daily temperature swing with nothing overhead to shade them, and a splice closed up tight in cool morning air can bind and buckle by afternoon. Record bracket torque and the splice gap used.

Record five: the bonding continuity reading

Bond each post and rail section with the listed grounding lugs or jumpers specified for the racking system, and confirm the path is continuous end to end with a resistance meter across the two farthest points in the array once assembly is complete, not by eye. Anodized or painted post and rail surfaces do not make a bond just because a lug is clamped onto them; the lug has to bite through that coating to reach bare metal, and a bond that looks tight but sits on a painted surface reads high resistance or open on the meter even though it is physically fastened. A near-zero reading end to end confirms a continuous metal path; anything materially higher points at one joint with an unbroken coating, and that joint is found and corrected before the array is energized, not noted for a later visit. Record the reading and the two points it was taken between.

The completed record, and the one field that failed

Ten-post frame, two rows of five, concrete piers, engineering letter calling for a footing depth deeper than the local frost line and a post spacing of eight feet on center.

Record one: locate confirmed, both rectangle diagonals within tolerance, the stake line reading straight end to end. Record two: excavation to the letter's structural depth, which exceeds the published frost depth for the site, so the single deeper number governs both requirements at once; depth logged at all ten holes.

Record three fails at post six. It reads roughly two degrees out of plumb along the row's axis, caught with the concrete still workable rather than after it sets. The post is pulled, reset, and rechecked plumb in both directions before that hole is poured; had this been caught after cure, the fix would have been demolition and a new footing rather than a reset. Posts one through five and seven through ten set plumb on the first try, braced, and held through the mix supplier's stated cure period before any load goes on them.

Record four: rail brackets go on across the full ten-post run, leveled end to end rather than post to post, torqued to spec and paint-marked, one splice set to the kit's expansion gap for this run's length. Record five: grounding lugs bite through the anodized rail finish to bare aluminum at every post, and the resistance check between the two farthest posts in the array reads near zero, confirming one continuous bonded path across all ten posts before the rails carry any rows of modules.

References

• See related: the ground mount versus roof mount reference for when this configuration is the right call for a site.
• See related: the attachment point spacing and load verification technique for translating the engineering letter's span and uplift numbers into a checked layout.
• ASCE 7, in the edition adopted by the local building department, for wind design loads feeding the racking manufacturer's engineering letter.
• 29 CFR 1926.652(a)(1) for excavation protective-system requirements; 29 CFR 1910.134 for respirator selection where dry mix handling raises dust.
• Racking manufacturer engineering letter and ground-screw or pier installation manual for footing depth, post spacing, torque and bonding hardware.