How Soil Is Classified and What Each Class Permits
Why this matters
The classification is made before anybody enters, from outside the cut, by a competent person, and it is re-made whenever conditions change. Never classify soil by climbing into an unprotected excavation to look at the face; the sample comes out with the bucket or off the spoil, and the face is read from the surface.
A wall that has stood open for three days is not evidence that it is stable. It is evidence that it has not failed yet, which is a different statement, because the mechanisms that fail an excavation wall are time-dependent: pore pressure equalising, a fissure opening as the face dries, water finding a path, a truck route starting up on Monday that was not running on Friday. The soil class is not a property the ground has permanently. It is a decision about what the ground is entitled to be treated as today, and the standard is built so that decision defaults downward.
The four classes and the strength line
Classification lives in Appendix A to 29 CFR 1926 Subpart P. Four categories, and the two numbers that separate the cohesive ones are stated in tons per square foot of unconfined compressive strength.
Unconfined compressive strength is the load per unit area at which a cylindrical sample of the soil fails with no lateral confinement. One ton per square foot is 2,000 lb per square foot, so the two thresholds below are 3,000 psf and 1,000 psf.
| Class | Core definition | What else lands here |
|---|---|---|
| Stable rock | Natural solid mineral matter that can be excavated with vertical sides and remain intact while exposed | Nothing else. This is a narrow category and most weathered or jointed rock is not in it |
| Type A | Cohesive soil with unconfined compressive strength of 1.5 tsf or greater | Clay, silty clay, sandy clay, clay loam, and in some cases silty clay loam and sandy clay loam. Cemented soils such as caliche and hardpan |
| Type B | Cohesive soil greater than 0.5 tsf but less than 1.5 tsf | Angular gravel, silt, silt loam, sandy loam, and in some cases silty clay loam and sandy clay loam. Previously disturbed soils except those that would otherwise be Type C. Soil meeting the Type A strength or cementation test but which is fissured or subject to vibration. Dry rock that is not stable |
| Type C | Cohesive soil of 0.5 tsf or less | Granular soils including gravel, sand and loamy sand. Submerged soil, or soil from which water is freely seeping. Submerged rock that is not stable. Material in a sloped, layered system where the layers dip into the excavation on a slope of 4 horizontal to 1 vertical or steeper |
The strength test alone never promotes a soil. It only makes a class available. What decides the class in most real cuts is the list below.
The conditions that force the class down
This is the section worth memorising, because it is where nearly every misclassification happens. Appendix A states plainly that no soil is Type A if any of the following applies, and each one on its own is enough.
It is fissured. A face with cracks, seams or separations. Drying clay opens fissures as it loses moisture, which means a face that qualified when it was cut may not qualify by mid-afternoon on a hot day. The direction is one way: exposure makes a clay face more fissured, never less.
It is subject to vibration from heavy traffic, pile driving or similar effects. A road, a rail line, compaction equipment, a nearby demolition. This one is a schedule condition as much as a site condition, because the arterial road that was quiet at 6 a.m. is not quiet at 4 p.m.
It has been previously disturbed. Any ground that has been excavated and backfilled before: a utility service, an old cut, a repaired main, imported fill. Previously disturbed soil is Type B at best and Type C where the backfill is loose or granular, and the reason is that backfill was placed rather than deposited, so it has none of the structure that gives undisturbed soil its cohesion.
It is part of a sloped, layered system where the layers dip into the excavation on a slope of 4 horizontal to 1 vertical or steeper. A bedding plane running downhill toward your cut is a ready-made failure surface.
Any other factor requiring classification as a less stable material. This is the catch-all, and a competent person is entitled to use it.
Two of these - fissuring and vibration - do not send the soil all the way to Type C. Appendix A places a soil that meets the Type A strength or cementation test but is fissured or subject to vibration into Type B. Previous disturbance also lands in Type B unless the material would otherwise be Type C. Two go straight to the bottom: submerged soil, or soil from which water is freely seeping, is Type C regardless of what it would otherwise have been, and so is material in a sloped, layered system whose layers dip into the excavation at 4 horizontal to 1 vertical or steeper.
Layers, and the one case where you may split them
A layered profile is classified as the weakest layer present, with one exception: where a more stable layer lies below a less stable layer, the excavation may be classified on a layer-by-layer basis.
Read the direction carefully, because it is the opposite of the intuitive one. Granular fill sitting on top of stiff clay may be split, because the strong material is underneath and the weak material's failure does not undermine anything. Stiff clay sitting on top of loose sand may not be split, because the weak layer is at the bottom holding up everything above it, and its failure takes the strong layer with it as a block.
What the class permits
The class is an input to the protective system decision, and the geometry it buys is owned by a sibling card. The ceilings, for excavations 20 ft or less in depth, are these:
| Class | Maximum allowable slope, horizontal to vertical | Angle from horizontal | Benching |
|---|---|---|---|
| Stable rock | Vertical | 90 degrees | Not applicable |
| Type A | 3/4 to 1 | 53 degrees | Permitted |
| Type B | 1 to 1 | 45 degrees | Permitted |
| Type C | 1.5 to 1 | 34 degrees | Not permitted |
Above 20 ft, the design goes to a registered professional engineer regardless of class. And the class is one input among several: the same Type B soil with a building beside the cut, water in it, or an excavator tracking alongside is a different problem, and the protective system card covers what else the system has to resist.
One classification record, and where it turned
A 320 ft storm lateral in an established commercial area, 9 ft deep, cut in stages over three days.
Strength testing, day 1, station 0 plus 00 to 1 plus 00. Three pocket penetrometer readings on undisturbed samples taken from the bucket at the surface, not from inside the cut: 1.8, 2.1 and 1.9 tsf. The mean is about 1.9 tsf. Use the lowest reading, 1.8 tsf, not the mean, because averaging lets a strong sample carry a weak one and the direction of that error is the direction that puts a wall up steeper than it should be. Even at 1.8 tsf the material clears the 1.5 tsf line, so the Type A strength test passes.
Then the downgrades, applied one at a time and each one printed:
| Condition observed | Where in the standard | Effect on the class |
|---|---|---|
| Strength 1.8 tsf, lowest of three | Appendix A, Type A strength threshold of 1.5 tsf | Type A available on strength |
| Vertical fissures visible in the exposed face by mid-morning | Appendix A, no soil is Type A if fissured | Type A removed. Meets Type A strength but is fissured, so Type B |
| Arterial road with truck traffic 12 ft from the alignment | Appendix A, no soil is Type A if subject to vibration | Type A removed on a second independent ground. Still Type B |
| Station 1 plus 40 to 1 plus 75 crosses a backfilled gas service | Appendix A, previously disturbed soil | Type B at best, and the backfill here is loose granular, so Type C for that segment |
| Overnight rain before day 2, seepage at the toe of the face at station 2 plus 10 | Appendix A, soil from which water is freely seeping | Type C for that segment, regardless of strength |
| Governing classification for the run | Type B for the general run, Type C at the crossing and at the seepage |
What that record actually decided. The strongest reading on the job, 2.1 tsf, did not produce a single foot of Type A anywhere on 320 ft of alignment. Two independent conditions removed it before the tape came out. A crew that classified on strength alone would have cut a 53 degree face where 45 degrees was the ceiling, and at 9 ft deep that is a run of about 6.8 ft per side instead of 9 ft per side, so the cut would have been over 4 ft narrower at the top than the class allowed - narrower being the wrong direction, because the missing width is the safety.
The rain is the part crews skip. 29 CFR 1926.651(k) requires an inspection by a competent person before the shift, as needed through the shift, and after every rainstorm or other hazard-increasing occurrence. Day 2 was a re-classification, not a re-inspection of Monday's classification, and the seepage at station 2 plus 10 dropped a segment two classes from where the day-1 testing had put it. A classification made once at the start of a run and carried for three days is the single most common failure in this whole subject, and it is silent, because nothing looks different from the surface.
What changes the classification later in a job
The face dries. Cohesive soil loses strength at the face as it dries and opens fissures. A cut left open over a weekend in hot weather is not the cut you left.
Something starts up nearby. A compactor two lots over, a delivery route, a rig. Vibration is the downgrade that arrives without touching your site.
The spoil pile grows. Spoil is a surcharge and it is also, when it is placed on ground you will later cut, previously disturbed material.
The alignment moves into a different profile. Two hundred feet of the same trench can cross three materials. The classification belongs to the segment, not to the job.
The excavation gets deeper. Depth alone does not change the class, but it changes what the class permits and it can bring a new stratum into the face, which does.
Confirming the classification was made rather than assumed
Ask three questions and each one has a checkable answer.
Which visual analysis and which manual analysis were run? Appendix A requires at least one of each, and a manual test alone or a look alone is not a classification. The sibling card on running the field tests covers the methods and the sampling.
When was it made and what has happened since? If the answer is older than the last rain, the last stratum change or the last time equipment started working nearby, it is stale.
Who made it, and can that person stop the work? The competent person definition at 29 CFR 1926.650(b) includes the authority to take prompt corrective measures. Someone who can classify the soil and cannot halt a crew has produced an opinion, not a classification.
References
- Appendix A to 29 CFR 1926 Subpart P, soil classification: the four classes, the unconfined compressive strength thresholds, the conditions that disqualify Type A, layered systems, and the requirement for at least one visual and one manual analysis.
- Appendix B to 29 CFR 1926 Subpart P, maximum allowable slopes by soil type.
- 29 CFR 1926.650(b) for the competent person definition, and 29 CFR 1926.651(k) for the inspection and re-inspection triggers.
- See related: the library's cards on running the field tests that classify soil, on sloping and benching by soil type, and on what a protective system is required to do.