What the Twenty-Foot Line Changes About Who Designs It

Why this matters

Everything a competent person is allowed to select from in 29 CFR 1926 Subpart P comes out of values somebody else published in advance, and every one of those sets has the same edge. Past 20 feet of depth, the sloping configurations, the timber shoring tables and the aluminium hydraulic shoring tables all stop, and the standard hands the entire selection to a registered professional engineer. Crews meet that edge without noticing, because the depth they quote each other is the depth on the plan, and the depth the rule is written about is the tallest wall on the site after bedding. A job everyone has been calling nineteen feet all week is regularly a twenty-two-foot excavation, and the person who selected the protective system had no authority to select anything.

What the tables deliberately do not cover

The useful way to read Subpart P's appendices is by their exclusions, because that is where the reader's job is.

• Appendix B gives maximum allowable slopes and benching configurations by soil type, for excavations less than 20 feet deep. Type A at 3/4 horizontal to 1 vertical, Type B at 1:1, Type C at 1.5:1, stable rock vertical. Above that depth there is no line in the table to read.
• Appendix C (timber shoring) and Appendix D (aluminium hydraulic shoring) publish member sizes and spacings against depth and soil class, and their depth columns end at 20 feet.
• None of the tables carry a surcharge term. They are keyed to soil class and depth. A spoil pile at the edge, a plated road, a crane outrigger or a stack of pipe is a load the configuration was never selected for.
• None of them carry an adjacent-structure term. A footing inside the influence zone is a separate duty under 1926.651(i), not something the slope table resolves.
• Appendix A excludes soils from a class rather than adjusting them. Fissured soil, soil subject to vibration, previously disturbed soil, and layered systems dipping into the excavation at 4 horizontal to 1 vertical or steeper cannot be Type A. Water freely seeping from a face puts the material in Type C outright.

Read that way, the 20-foot line is not an arbitrary cutoff. It is the edge of a set of pre-approved answers, and past it there are no pre-approved answers at all.

How many routes survive past twenty feet

For sloping and benching, 1926.652(b) offers four options: allowable configurations under (b)(1), Appendices A and B under (b)(2), other tabulated data approved by a registered professional engineer under (b)(3), and a design by a registered professional engineer under (b)(4). For support, shield and other protective systems, 1926.652(c) mirrors it: Appendices A and C, the manufacturer's tabulated data, other approved tabulated data, and an engineered design.

Past 20 feet, the first two routes are gone by their own text. The third survives only if the tabulated data set you are using actually extends that deep and an engineer approved it that deep, which the Subpart P appendices do not. What is left is one route, and it is a person, not a document.

Measuring the depth the rule is about

Worked example. A pumped force main replacement in a back corridor. The crew has called it a nineteen-foot job since the pre-bid.

• Plan depth to top of bedding, measured from the plan datum: 19.00 ft
• Over-excavation below that for the bedding section: 10 in, which is 0.83 ft. Running depth 19.83 ft
• The plan datum was set at the low edge of the corridor. The corridor cross-falls 2.5 ft across the trench.
• Depth on the tallest wall, from the highest adjacent grade: 19.83 + 2.5 = 22.33 ft

22.33 is greater than 20, so Options 1 through 3 are unavailable and the protective system for this excavation is a registered professional engineer's design under 1926.652(b)(4) or (c)(4). Nobody on the crew, including a properly designated competent person, may select it.

Two corrections got applied there, and both had to run in the same direction. The bedding over-excavation is an addition to the plan depth, because the plan depth was published to the top of bedding and did not contain it. The cross-fall is a re-basing, not an addition: it does not make the trench deeper, it changes which edge you measure from, and the rule is about the tallest wall because that is the wall with the most stored energy in it. Measuring from the low edge is the flattering direction and it is the one that gets used, because that is where the crew stands.

Why the extra 2.5 feet matters more than 12.6 percent. The depth rose 2.5 ft on 19.83, which is 12.6 percent. But at constant soil unit weight, constant lateral earth pressure coefficient, no surcharge and no water table, the total lateral force on a wall goes with the square of its height, so the force went up by a factor of (22.33 / 19.83) squared, about 1.27, or 27 percent. That exponent is 2 only with those four things held constant. Put a water table partway up the wall or a load at the edge and the relationship changes shape, which is precisely why the tables refuse to carry either term.

The failure mode. The crew that measures from the low edge does not commit an obvious error. They select a Type B configuration from Appendix B at 19 feet, install it competently, and it fails to be legal and fails to be sized, in that order. The wall that comes in is the one they never measured.

What does not move the line

• Benching the upper portion does not reduce the depth. The excavation is as deep as it is; benching is part of an answer, not a change to the question.
• Digging it in two lifts on two days does not. While the cut stands open at 22.33 ft, it is a 22.33-ft excavation, whatever the sequence was.
• Where the worker is standing does not. A person laying pipe at 6 ft of depth in a 22-ft trench is an employee in a 22-ft excavation.
• A shield rated deeper than the box is tall does not. The rating is a load basis for the full trench depth, not a permission to leave face unsupported above the box.

What "registered professional engineer" means in this standard

29 CFR 1926.650(b) defines it as a person registered as a professional engineer in the state where the work is to be performed, with one carve-out: an engineer registered in any state qualifies when approving designs for manufactured protective systems or tabulated data to be used in interstate commerce.

That split is worth holding onto. The tabulated data sheet that came with a rented box may legitimately carry an out-of-state engineer's approval. Your site-specific design for a 22.33-ft cut may not. If a supplier offers to have their engineer stamp your excavation, confirm the registration is in the state the trench is in.

What has to be on the site, not in an inbox

Both engineered routes carry the same documentation requirement, at 1926.652(b)(4) for sloping and benching designs and 1926.652(c)(4) for support and shield systems: the design shall be in writing, shall include at least the identity of the registered professional engineer approving it, and a copy shall be maintained at the jobsite while the protective system is being constructed. Afterward it may be stored off site, but it must be made available on request.

The reason it lives on site is operational rather than clerical. The competent person's daily inspection under 1926.651(k)(1) has to check the installed system against something, and a design nobody can read is a design nobody can check. Two things make that check possible in practice: the drawing shows member sizes, spacings and elevations rather than a general arrangement, and someone has confirmed the installed system matches it before the first entry rather than after.

Alongside it, 1926.652(a)(2) applies at every depth: the protective system shall be capable of resisting without failure all loads that are intended or could reasonably be expected to be applied or transmitted to the system. That clause is what makes "the engineer did not know about the plated road" the contractor's problem as well as the engineer's.

How to verify before anyone is exposed

Do all of this from grade, outside the influence zone, with nobody in the cut.

1. Shoot the highest adjacent grade along the whole run, not one station. Cross-fall and crown both move the answer, and so does a spoil pile that has become part of the grade.
2. Add every over-excavation below the plan invert - bedding, a sump, a bell hole, a thrust block pocket. Each is depth.
3. Compare the deepest measurement to 20 feet, and treat anything within about a foot of it as if it will cross, because trench bottoms get deeper during construction and never shallower.
4. If it crosses, stop selecting. The competent person's remaining job is to verify installation against a design, not to choose one.
5. Confirm the written design is physically on site and matches the state the work is in, before the system goes in the ground.
6. Re-run steps 1 and 2 whenever the invert changes, including a field change made to clear a utility, which is the single most common way a compliant 19-ft job becomes an uncontrolled 21-ft one mid-shift.

References

• 29 CFR 1926.652(b) and (c), the four design options for sloping and benching and for support, shield and other protective systems, including the engineered-design routes at (b)(4) and (c)(4) and their written-design and on-site requirements
• 29 CFR 1926.652(a)(2), capacity of protective systems to resist all reasonably expected loads
• 29 CFR 1926.650(b), definitions, including registered professional engineer and competent person
• 29 CFR 1926 Subpart P Appendices A, B, C and D, soil classification, allowable slopes, timber shoring and aluminium hydraulic shoring tables
• See related: universal-why-shoring-and-shielding-are-not-the-same-decision; universal-what-a-competent-person-is-required-to-be-able-to-do