Post-Repair Monitoring and Follow-Up

Purpose

Every foundation repair this shop sells comes with a claim that movement has stopped. Monitoring is how that claim gets tested, and it only works if the baseline, the benchmark, the interval and the movement threshold were all fixed on the day the crew left.

Get that wrong and a year later you have a re-shoot with nothing to compare it to. A customer says the floor moved, a tech shoots elevations, the numbers differ from a survey taken from a different reference point in a different season with a different instrument, and nobody in the room can say whether that is movement or noise. The argument that follows is not about engineering, it is about who remembers what, and the shop loses it every time. This procedure exists so that a reading taken in year two is decidable by whoever is standing there.

Scope

Covers commissioning the baseline, setting the movement threshold against measured repeatability, installing and reading crack gauges, the visit schedule, the seasonal comparability rule, the escalation decision, and what the customer is told at each visit.

It does not cover the elevation survey built for an engineer's design package, which belongs to the engineer involvement SOP, and it does not re-derive how to shoot one. It does not cover diagnosing the cause of confirmed movement, which routes to the settlement and recurrence troubleshooting cards. It does not cover sump pump service beyond reading the cycle counter.

Roles and handoffs

Procedure

1. Take the commissioning baseline the day the crew demobilizes. Acceptance: a full elevation survey on the same grid and from the same physical benchmark used before the work, with the instrument model recorded; crack gauges installed and zeroed; the sump cycle counter reading; and photographs from the marked positions used in the pre-condition record. Wrong looks like a baseline taken a few weeks later once things have settled, which quietly folds the first weeks of movement into the reference and makes it invisible forever. Stop rule: no baseline, no closed ticket, no final invoice. If the benchmark from the original survey was destroyed by the work, set a new one on a permanent feature, photograph it, and record that the series restarts here rather than pretending it continues.

2. Measure your repeatability before you set a threshold. Read one grid point three times, breaking the instrument down and re-setting it between reads, and record the spread. That spread is the combined noise of instrument, setup and operator, and it is what any reported movement has to exceed to mean anything. Acceptance: a stated threshold equal to at least three times the measured spread, written on the record with the spread it came from. Wrong looks like a threshold inherited from another shop with no derivation under it. Stop rule: where the spread has not been measured, use 1/4 in as an interim and measure it on the next job. When rounding the threshold, round it down rather than up, because rounding a detection limit upward makes the system less likely to catch the thing it exists to catch.

3. Install crack monitors where a crack is the metric, and state the gauge's resolution. Acceptance: a tell-tale gauge bonded across the crack at a marked, photographed location, zeroed on day zero, with the gauge's graduation recorded; a common grid gauge is graduated in 1 mm and read to roughly 0.5 mm by eye. Wrong looks like a pencil mark and a date on the wall, which records that something was there and nothing about how much. Stop rule: a change smaller than the stated resolution is not movement and is recorded as no change, and a single reading in one direction is not a trend; two consecutive readings moving the same way is the shop's rule for calling movement.

4. Book the whole visit series at close-out. Shop default: 30 days, 90 days, 365 days, then annually through the warranty term. The shape has a reason. Most post-repair movement that matters appears in the first season, and the 365-day visit is the one that lands in the same season as the baseline, which is the only reading that compares like with like. Acceptance: every visit on the schedule as a real appointment with a date, not a note to call later. Wrong looks like a shop that monitors by waiting for a complaint, which guarantees the first data point is an angry one. Stop rule: a customer who declines the series has that declination recorded in their own words, and the shop's warranty terms then apply as written rather than being argued about later.

5. Annotate every out-of-season comparison, and let the in-season reading govern. Expansive clay sites move seasonally by design, and so do sites with a shallow water table; the same house reads differently in February than in September for reasons that have nothing to do with the repair. Acceptance: any comparison to a baseline taken in a different season is marked as such on the record. Wrong looks like an autumn baseline compared to a late-winter re-shoot on a clay site and reported to the customer as movement. Stop rule: an out-of-season reading over threshold does not by itself open a warranty claim; it triggers an added reading and, where possible, one taken in the baseline's season before anyone draws a conclusion.

6. Run every visit to the same script, or it is not a series. Acceptance: same instrument type, same benchmark, same grid points in the same order, every crack gauge read, sump counter recorded, plus a fixed set of questions to the customer about doors, windows, new cracks and water. Wrong looks like a different tech shooting a shorter grid because the far rooms looked fine, which puts a hole in the series exactly where nobody was worried. Where the visit needs the crawl space, the full entry and atmosphere check runs every time, not just on the first visit. Before hand-testing a sump float, unplug the pump at the receptacle; a sump found on an extension cord or a non-GFCI receptacle gets photographed and referred to an electrician rather than corrected by a foundation tech.

7. Decide against the threshold on site, in writing, with three named outcomes. Acceptance: every reading resolves to one of three. Within threshold: record and confirm the next visit. Over threshold but out of season, or moving in one direction for the first time: add a reading, do not escalate, and tell the customer exactly that. Over threshold and confirmed by two consecutive readings in the same direction, in a comparable season: escalate to the supervisor the same day, and to the engineer where the original design was sealed. Wrong looks like a tech resolving an over-threshold reading on the spot by telling the customer it is normal. Stop rule: that sentence is not available to a technician, because the whole point of a threshold is that it takes the judgment out of the driveway.

8. Give the customer the numbers at every visit, including the uneventful ones. Acceptance: a short written summary with the actual readings, the threshold, and the next visit date, delivered the same day. Wrong looks like silent visits, so the only time the customer hears from the shop is when something is wrong, which trains them to read any contact as bad news. Stop rule: no summary sent, visit not complete. This is also the cheapest referral the shop gets, because a customer who has four dated readings showing a stable house tells the neighbours something specific rather than something vague.

The record this produces

One monitoring file per repaired address, carrying the whole series rather than one row per visit. Fields: benchmark description and photograph; instrument model; the measured repeatability spread and the threshold derived from it; the baseline survey with its date and season; every crack gauge with its location, resolution and zero photo; and per visit, the date, tech, every grid reading, every gauge reading, the sump counter, the customer's answers, the outcome of step 7, and the summary sent.

It attaches to the original job and follows the warranty term. The warranty administrator reads the series, not the last reading, because a claim turns on direction over time rather than on one number. The supervisor reads the repeatability line before accepting any escalation, since a threshold set below the noise generates escalations that are not real. The engineer, on a sealed job, reads the series against the as-built. And the estimator reads the whole file when the same street calls, because a subdivision on one clay stratum tends to move together.

Worked pass: 9 piers on a rear addition, expansive clay, four readings after baseline

Commissioning in October: 18 grid points, benchmark set on the front stoop and photographed, two crack gauges, one on the rear interior wall and one on the exterior brick. Repeatability check gives a spread of 0.06 in across three re-set reads. Three times that is 0.18 in, and the shop writes the threshold as 0.15 in rather than rounding to 0.20, on the rule from step 2.

Thirty days, November: largest change at any grid point is 0.09 in, under threshold. Both gauges read within resolution of zero. Recorded, next visit confirmed, summary sent.

Ninety days, January: grid point 12 at the rear reads 0.22 in of drop from baseline, over the 0.15 in threshold. The interior gauge reads 1.0 mm, which is above the gauge's 0.5 mm read resolution. Step 7 fires, and it lands on the middle outcome rather than the last one: this is a January reading against an October baseline on a clay site, and it is the first reading in that direction. So the tech adds a visit rather than opening a claim, and tells the customer precisely that, with both numbers and the threshold written down.

One hundred and twenty days, February: point 12 reads 0.26 in, up from 0.22, and the interior gauge reads 1.5 mm, up from 1.0. Two consecutive readings, same direction, both over threshold. The tech escalates from the site that afternoon.

The engineer's answer, read against the as-built log, is that all 9 piers met both their torque and depth criteria, so pier settlement does not explain a localized drop at one grid point. The supervisor sends a crew to look at the outside, and they find the backfill along the rear wall has consolidated into a trench line over the winter and is now ponding against the house. That top-off was scheduled at six months on the original backfill log and never booked. The grade is restored and the downspout extended.

Three hundred and sixty-five days, October, same season as the baseline: point 12 reads 0.11 in from baseline, back under threshold, and the interior gauge has held at 1.5 mm across two consecutive readings. The series says what nobody could have said from any single visit: the piers held, surface water moved a slab edge for one winter, and the fix was a wheelbarrow of soil the shop already owed the customer.

References

• International Residential Code Section R401.3, for the grade fall the winter finding in the worked pass was measured against
• ASTM D4719 and the pier system's ICC-ES evaluation report, for the installation criteria an as-built log is checked against on escalation
• Crack monitor manufacturer's printed instructions for graduation, bonding and reading practice
• 29 CFR 1926 Subpart AA, where a monitoring visit requires crawl space entry
• See related: crawl space entry and atmosphere check; engineer involvement and sign-off; exterior waterproofing and backfill; pier settles again after install