How to Design a Retaining Wall — A Complete Step-by-Step Guide (ACI 318-25 & IBC 2024)

Most structural members fail by breaking. A retaining wall usually fails by moving — it slides forward, or it tips, or one edge of its footing pushes into the ground and the whole wall rotates. Nothing cracks first to warn you. That is why the building code hands retaining walls their own acceptance criterion before any concrete design starts: a safety factor of 1.5 against overturning and against sliding, at unfactored loads (IBC 2024 §1807.2.3). Only after the wall has proved it will stay put do you size the reinforcement.

This guide walks that whole method on one wall — a 12 ft cantilever concrete retaining wall — and it does something most worked examples avoid: the first trial fails. Proportioned by the standard rules of thumb, it comes out at a sliding safety factor of 0.86 against the required 1.5, its bearing resultant falls outside the middle third of the footing, and the toe bars cannot develop. That failure is the most useful part of the page, because fixing it is where retaining wall design actually lives. You will see the three levers available, what each one buys, what each one costs, and why the cheapest fix here is not concrete at all — it is a line in the backfill specification.

By the end you will have: the finished wall, every check with its substituted formula and code citation, the reinforcement drawing, the drainage detail that makes the whole design legal, the construction mistakes that undo it on site, and the numbers for what happens if the same wall is built in Seismic Design Category D.

Every number below was produced by the calculation engine that runs StructSuite’s Retaining Wall module, and then recomputed independently by hand — the full side-by-side comparison is in Verification near the end, and both halves are locked by automated regression tests.

#The wall we are going to design

A hillside residence needs a 12 ft grade change held back along the uphill side of a new yard. Above the wall is landscaping with an allowance for light equipment and stored material — a uniform surcharge of 100 psf. The geotechnical report classifies the site soil as SM silty sand, gives an allowable bearing pressure of 3,000 psf, a coefficient of base friction μ = 0.35, and a lateral bearing (passive) value of 200 psf per foot of depth. There is no basement, no slab tying the top of the wall to anything: the wall is free to move and rotate at the top, which is what will let us use active earth pressure.

Geometry — 12 ft stem of uniform 12 in thickness, retained height 12 ft (backfill to the top of the wall), toe 3.0 ft, heel 4.5 ft, footing 18 in thick, giving B = 3.0 + 1.0 + 4.5 = 8.5 ft, 1.0 ft of soil cover over the toe, and a 12 × 12 in shear key under the stem.

Soil — imported GW well-graded clean gravel as structural backfill, γ = 120 pcf, uniform surcharge q = 100 psf (variable), coefficient of friction μ = 0.35, passive lateral bearing 200 psf/ft, allowable bearing 3,000 psf.

Concrete — f′_c = 3,000 psi, Grade 60 (f_y = 60,000 psi; the field flags anything above the 100,000 psi design maximum of ACI 318-25 Table 20.2.2.4(a) as it is typed), normalweight 150 pcf. Covers per ACI 318-25 Table 20.5.1.3.1: 3 in at the footing bottom (cast against earth), 2 in at the stem earth face and the footing top (formed, exposed to earth). Reinforcement: stem verticals #6 @ 8 in at the earth face, stem horizontals #4 @ 9 in, heel top #6 @ 10 in, toe bottom #5 @ 9 in, footing longitudinal #5 @ 9 in.

Site seismicity — Seismic Design Category B, so no seismic earth pressure increment is required (IBC 2024 §1807.2.2 triggers it only in SDC D, E or F with more than 6 ft of backfill). What changes in SDC D re-runs this identical wall with the increment switched on, so you can see the cost of shaking on a wall you already understand.

The design assumes drained backfill. That is not a footnote — it is a load case. IBC 2024 §1610.1 otherwise requires full hydrostatic pressure, which on this wall would be catastrophic; the drainage section shows the number.

#Which wall type — and when a cantilever is the right one

Before any calculation, one decision sets everything else: what kind of wall. Four types cover almost all site work, and the choice is mostly about height and economics — how much concrete versus how much formwork and steel.

At 12 ft, a cantilever is the obvious answer, and it is the type this guide designs. The rule that decides it: a gravity wall needs a base roughly 0.5–0.7 × its height of solid concrete, so its cost grows with the square of the height, while a cantilever recruits the weight of the backfill sitting on its heel to do the same job for free. Above roughly 20–22 ft a cantilever stem gets thick enough (and its steel congested enough) that counterforts start winning again.

Practical note. For walls under about 4 ft of retained height with no surcharge, check whether your jurisdiction's prescriptive tables cover it — IRC 2024 §R404.4 applies the same 1.5 safety factor, and many building departments will accept a prescriptive detail without a stamped calculation. Above 4 ft (or with any surcharge), you are designing.

#Where earth pressure comes from — at-rest, active and passive

Everything downstream rides on one number, the design lateral soil load in pounds per square foot per foot of depth. It is worth 60 seconds to understand what it is, because the single most expensive mistake in retaining wall design is picking the wrong one of the three states below.

Soil is not a fluid, but it pushes like one, and the amount it pushes depends on how much the wall lets it move:

The classical (Rankine) coefficients behind the curve, for level backfill with a smooth vertical wall face:

You do not have to compute those to use the code. IBC 2024 Table 1610.1 tabulates the design lateral soil load directly by USCS backfill class, as an equivalent fluid pressure w in psf per foot of depth, and §1610.1 permits active values for walls free to move and rotate at the top. But the classical formulas are worth carrying, because they tell you what the table is quietly assuming — and they let you sanity-check a geotechnical report. Divide the table's value by the soil unit weight and you recover the coefficient the table implies:

K = w / γ = 30 / 120 = 0.250 → φ ≈ 37°, which is exactly what a well-compacted, well-graded gravel gives.

Run the same check on the site's silty sand: K = 45/120 = 0.375 → φ ≈ 27°. Also right for an SM. The code table is not arbitrary; it is Rankine with sensible friction angles baked in. When a geotechnical report hands you an equivalent fluid pressure that implies a φ your soil description cannot support, that is a question worth asking before you design to it.

#Cantilever or restrained — the decision that changes the load and the load path

Everything above assumed the wall is free to lean. That is a decision, not a given, and it is the one input on this page that can change the answer by a factor of two. Both codes settle it in a single sentence, and IBC 2024 §1610.1 and ASCE 7-22 §3.2.1 use very nearly the same words:

Foundation walls and other walls in which horizontal movement is restricted at the top shall be designed for at-rest pressure. Walls that are free to move and rotate at the top, such as retaining walls, shall be permitted to be designed for active pressure.

Read the two verbs again, because they answer the hardest version of this question before we even get to it. At-rest is what you shall do. Active is what you are permitted to do — a permission you have to earn by establishing that the top can move. The burden of proof sits entirely on the active side, and that asymmetry is the code's, not a house rule.

#Three different things get called "restrained"

Most of the confusion here is vocabulary. Three separate judgments travel under the same word, they are decided by different documents, and StructSuite treats them very differently:

That last cell is the important one, and the rest of this section is mostly about its consequences.

#How much movement are we actually talking about?

Less than you would guess, which is why most free-standing walls genuinely do qualify for active. FEMA P-750 Part 3, Resource Paper 12 (p. 355) puts a number on it:

A displacement at the top of the wall of 0.002 times the wall height is typically sufficient to develop the minimum active pressure state.

On this 12 ft wall that is 0.29 in — a movement no one would notice, and one a cantilever on soil makes without complaint. The same page gives the rule of thumb for sorting walls:

Generally, free-standing gravity or cantilever walls are considered to be yielding walls (except massive gravity walls founded on rock) whereas building basement walls restrained at the top and bottom often are considered to be nonyielding.

#Which one gives you a stronger wall?

The question has three honest answers, and only the third is useful.

(a) Same wall, twice the load — this is what StructSuite computes, and for this backfill the factor is exactly 2. IBC Table 1610.1 gives GW gravel 30 psf/ft active and 60 at-rest, so the implied coefficient goes K = 0.250 → 0.500 and every lateral quantity doubles with no residue. The finished 12 ft wall from this guide, switched to at-rest and otherwise untouched:

The two rows that do not move are the instructive ones. The heel carries the soil column and the surcharge sitting on top of it, and the key is sized by passive pressure in front of it — neither is a function of the pressure state behind the wall. Everything the backfill pushes with doubles; everything the wall stands on does not. ASCE 7-22 C3.2.1 gives the mechanism in one line: "when wall horizontal movement is restricted at the top, the earth pressure is increased from active pressure toward earth pressure at rest."

(b) A real prop is a different load path — and the base moment goes down, not up. This is the half that surprises people. A wall held at the top no longer cantilevers; it spans between its footing and its prop, and the moment diagram changes shape entirely:

For a triangular pressure reaching q at the base of a wall of height H, the base moment falls from qH²/6 to qH²/15 — a factor of 2.5 — and a sagging peak of 0.030·qH² appears at 0.45H, putting tension on the exposed face, where a cantilever never has any. The price is a horizontal reaction at the top of qH/10, which is 20% of the total thrust, and it has to go somewhere.

#And the base fixity decides where the steel goes

Part (b) propped the top and left the base fixed. That is one of two choices, and the other one — a pinned base — is not a rounding difference. It barely moves the size of the peak moment and it moves its location by half the wall.

Three things in that table are worth saying out loud.

The peak barely changes; its address does. 0.0667 against 0.0642 qH² is a 4% difference — you would size nearly the same bar either way. But the fixed-base wall wants it at the footing and the pinned-base wall wants it 0.42H above the footing, on the opposite face. Get the fixity wrong and you have bought the right amount of steel and put it in the wrong half of the wall.

The prop reaction is where fixity really costs you. Going from a fixed base to a pinned one raises the horizontal force delivered into the floor by 67% — qH/10 becomes qH/6. That force is a design load on somebody else's diaphragm, and it is the number most likely to be missing from a set of drawings entirely.

A real base is neither, and you still have to pick one. A stem cast into a spread footing that can rotate in soil is not the textbook fixed end; a wall on piles with a thick mat is close to it. There is no defensible way to interpolate — the two answers put the steel in different places, so a half-fixed result would put it in neither. The honest treatment is to choose the fixity deliberately, state it, and detail for the case you chose.

(c) Put the two together, and the answer stops being "stronger" or "weaker". For this wall's soil term:

So a restrained wall attracts twice the load and, if the prop is real, still ends up with a peak moment 20–23% below the free cantilever's — whichever base fixity you choose. Restraint is not a stronger wall or a weaker one. It is a stiffer boundary condition that raises the load and shortens the span at the same time, and the span wins.

Notice that the two propped rows land within 4% of each other while their peaks sit 5 ft apart. The fixity barely changes how much steel; it changes which half of the wall gets it.

The only configuration that is unambiguously worse than everything else is the one second in that list: at-rest pressure on a cantilever load path — a wall that is held at the top by something nobody designed. It pays the full pressure penalty and collects none of the span relief.

(d) Under earthquake, nonyielding really is bigger — by 8/3. Here the two states are not a factor of two apart, and both numbers come from the same document. At the same k_h, γ and H, and applied at the same 0.6H:

• yielding, Seed-Whitman (RP 12 Eq. 7): ΔP_AE = (3/8)·k_h·γ·H²
• nonyielding, Wood 1973 (RP 12 Eq. 9): ΔP_E = k_h·γ·H²

a ratio of 2.67. RP 12 (p. 358) states the bracket in its own words: "the M-O solution and the Wood (1973) solution represent a 'lower' and an 'upper' bound, respectively."

#When should you consider the wall restrained?

Statically, these are mandatory triggers, not preferences:

1. The top movement is restricted — §1610.1 / §3.2.1. In practice: a slab or diaphragm bearing on the top of the wall, a stair or landing framing into it, a pilaster braced by a building, or rigid returns at both ends close enough that the wall cannot rotate between them.
2. It is a foundation or basement wall — named in the same sentence, with one carve-out below.
3. The geotechnical report says so. The code tables are only "the minimum design lateral soil loads" and only apply "if lateral soil loads are not given in a geotechnical report approved by the Authority Having Jurisdiction" (§3.2.1). Plenty of reports require at-rest for cohesive backfill as a matter of policy.

There is exactly one code-defined middle case, and it is easy to miss: §1610.1 and §3.2.1 both carry an Exception permitting active pressure for "foundation walls extending not more than 8 ft (2.44 m) below grade and laterally supported at the top by flexible diaphragms." ASCE 7-22 C3.2.1 says precisely what "flexible" buys you there: "wood-framed floors and cold-formed steel joists without a cast-in-place concrete floor attached." Note how narrow it is — foundation walls, not more than 8 ft, flexible diaphragms. A 12 ft retaining wall tied to a concrete slab misses on all three counts.

Seismically (SDC D, E or F only — that is §15.6.1's own scope), the choice is taken away from you for this wall type:

Earth-retaining walls are permitted to be designed for seismic loads as either yielding or nonyielding walls. Cantilevered reinforced concrete or masonry retaining walls shall be assumed to be yielding walls and shall be designed as simple flexural wall elements.

RP 12 (p. 357) gives the test for the other side, and it is a demanding one: nonyielding "requires a very stiff wall in combination with a rigid base condition. Most nonyielding walls will be located on rock or very stiff soil… where structural analyses determine that the wall flexibility is such that deformations will not develop seismic active earth pressures (i.e., deformations < 0.002H)."

One case that is neither, worth naming so it does not get mis-filed: a counterfort or buttressed wall. ACI 318-25 §13.3.6.2 makes its stem "a two-way slab in accordance with the applicable provisions of Chapter 8" because it spans horizontally between the ribs. That is a third load path, and it is a wall-type question, not a restraint one.

#If you genuinely do not know, which way do you assume?

For the pressure state: assume at-rest. It is always code-compliant, because at-rest is the "shall" and active is the permission. It is not free — on this wall it fails two checks at once and the fix is real concrete: toe 3.0 → 3.5 ft, key 12 → 60 in and stem bars #6 @ 8 → #6 @ 6 brings it back to FS_sliding 1.57, FS_overturning 2.19 and q_max 2,510 psf, all passing. But "expensive" is the right failure mode; the other one is not.

For the structural system, there is no safe default — and this is the trap. Choosing at-rest in StructSuite gives you a cantilever carrying at-rest pressure. Against a genuinely propped wall that is conservative at the base by 2.5×. It is also, at the same time:

1. A prop nobody has designed. A real propped wall pushes qH/10 into whatever holds its top — for this wall at at-rest, about 860 lb per foot from the soil alone, before the surcharge's share. StructSuite models a free cantilever, whose top reaction is zero, so that number is never computed and never reaches the drawings. A conservative wall bearing on a slab edge that was never checked is not a safe outcome; it is a hidden one.
2. Reinforcement on the wrong face. A propped stem develops its sagging peak at 0.45H with tension on the exposed face. A cantilever model never puts tension there, so a wall detailed from it has earth-face steel only.
3. Sliding and bearing that are over-conservative — 20% of the thrust leaves through the prop and never reaches the footing, which is how you end up buying a 5 ft key for a wall that did not need one.

So: at-rest is the safe default for the pressure, and it is not a substitute for deciding whether the wall is propped. If it is propped, say so on the drawings, size the prop for its reaction, and understand that this module is handing you a conservative stem and no prop force at all.

For the seismic classification you are not free to be conservative either. §15.6.1 mandates yielding for a cantilevered reinforced concrete or masonry wall in SDC D–F. Assuming nonyielding would be conservative on load but contradicts a "shall"; assuming yielding for a wall that cannot actually displace 0.002H is unconservative, and RP 12 (p. 357) says what to do instead: "If movement cannot be tolerated or if the wall may not move enough to mobilize the yield condition, then either the full a_max should be used… or more rigorous procedures… should be used." StructSuite takes the same position rather than guessing: select a restrained wall and it refuses to run without the report's own k_h, because a rigid-wall pressure model is outside its computed methods.

#Choosing the input parameters — where each one comes from

Every soil number in a retaining wall design has a pedigree, and a plan checker will ask for it. Here is each one, its source, and what happens if you get it wrong.

• Design lateral soil load (30 psf/ft here) — IBC 2024 Table 1610.1 by USCS backfill class, unless a geotechnical investigation per §1803 determines otherwise. The table values are minimums for level backfill: a geotech report may require more (a sloped backfill takes the tables off the menu entirely, compaction-induced pressure) and occasionally allows less. ASCE 7-22 Table 3.2-1 is the parallel table and it disagrees with the IBC for several classes — clayey sand (SC) is 60 psf/ft in the IBC and 85 in ASCE 7-22. StructSuite offers both verbatim; use the one your jurisdiction enforces.
• Soil unit weight γ (120 pcf) — geotechnical report; 110–130 pcf is the usual band for compacted granular backfill. It does double duty: it sets the weight of soil on the heel (most of the resisting moment) and the implied K used for the surcharge, so it appears on both sides of the stability equation.
• Coefficient of friction μ (0.35) — IBC Table 1806.2 presumptive values run 0.25–0.70 by class, multiplied by the dead load per footnote a. Clays get cohesion instead of friction, capped at one-half the dead load (§1806.3.2). μ is the single most influential number in the sliding check — see the sensitivity table.
• Passive lateral bearing (200 psf/ft) — the "lateral bearing pressure" column of IBC Table 1806.2, or the geotech report. Count it only where the soil in front of the toe is permanent, compacted and undisturbed for the life of the wall. A future utility trench erases it. StructSuite has a checkbox for exactly that judgment — Neglect passive resistance in Step 2 — and this wall is one where it matters.
• Allowable bearing pressure (3,000 psf) — geotech report, or Table 1806.2 presumptive used with ASD combinations (§1806.1). Note the presumptive class-3 value for gravel happens to be 3,000 psf too.
• Surcharge (100 psf, variable) — your own project loading: sidewalk and light-traffic allowances, stored material, permanent fill. Its nature matters more than its size. A variable surcharge (traffic, parking, storage) pushes but is not allowed to help hold the wall down; a permanent surcharge (an adjacent footing, permanent fill) is dead load and does both. IBC §1807.2.3 is what forces that asymmetry, and StructSuite asks the question rather than assuming. A building footing standing in the backfill is not a uniform surcharge and has its own input — A footing behind the wall shows why the stand-in gets the wrong members wrong.
• Seismic Design Category — from the project's seismic design (ASCE 7-22 Chapter 11). SDC D, E or F plus more than 6 ft of retained backfill makes the seismic earth pressure increment mandatory (IBC §1807.2.2). This site is SDC B, so it is not required; the last major section shows what would change if it were.

The parameter engineers most often get wrong is not a soil property — it is the condition. Active pressure is legal only if the wall can rotate at the top. Tie the top of that wall to a slab, a stair, a pilaster braced by a building, or a rigid corner return, and it becomes an at-rest wall. For this backfill that is 60 psf/ft instead of 30 — a doubling of the lateral load. Section What this wall is sensitive to shows what that alone does: the sliding safety factor falls from 1.62 to 0.81.

#Step 1 — Proportion the wall (the first pass)

You cannot check a wall you have not drawn, so retaining wall design starts with a guess — proportions from rules of thumb that have survived a century of practice — and then converges. Here is the vocabulary and the rules.

That is the first trial: B = 7.5 ft, toe 2.5 ft, heel 4.0 ft, 18 in footing, 12 in stem, no shear key, with the site's own SM silty sand re-used as backfill (45 psf/ft) because that is the cheapest thing to do with the spoil you already excavated. Now check it.

#Step 2 — Build the pressure diagram

Two loads push on the wall, and they have different shapes. The soil pressure grows linearly with depth (a triangle); the surcharge produces a uniform horizontal pressure over the full height (a rectangle), because a load spread over the whole ground surface adds the same vertical stress at every depth, and the soil converts a fraction K of it to horizontal stress.

For the stability checks the pressure acts on the vertical plane at the back of the heel, from the backfill surface all the way down to the bottom of the footing:

H_p = H_s + t_f = 12.00 + 1.50 = 13.50 ft

Resolving each block into its resultant, per foot of wall:

• Soil: P_a = w·H_p²/2 = 45 × 13.50²/2 = 4.101 klf, acting at H_p/3 = 4.50 ft above the footing base (the centroid of a triangle).
• Surcharge: P_q = K·q·H_p = 0.375 × 100 × 13.50 = 0.506 klf, acting at H_p/2 = 6.75 ft (the centroid of a rectangle).

The surcharge is only 11% of the total push but it acts 50% higher up, so it contributes 16% of the overturning moment. That leverage is why a surcharge you were tempted to ignore keeps showing up in the governing case.

Where the K for the surcharge comes from. The code table gives you an equivalent fluid pressure, not a coefficient — but the surcharge needs a coefficient. Recover it from the table itself: K = w/γ = 45/120 = 0.375. Surcharge-induced lateral pressure is part of the load H (ASCE 7-22 §2.2), added per IBC §1610.1, which matters later: it takes the same load factors as the earth pressure, not the factors of whatever is sitting on the ground.

#Step 3 — Overturning: will it tip?

IBC 2024 §1807.2.3 is explicit and easy to get wrong: the §1605 load combinations do not apply to stability. Stability is checked at 1.0 × nominal loads, with variable loads investigated set to zero. That cuts both ways, and it is the reason the surcharge appears on only one side of the equation:

• The surcharge's lateral push is driving, so it stays in
• The surcharge's vertical weight over the heel would help resist — so it is dropped.

Resisting weights and moments about the toe (per foot of wall):

Look at row 4: the soil sitting on the heel supplies 72% of the resisting moment. The concrete is almost incidental. That is the whole idea of a cantilever wall, and it is why the heel length is the most powerful geometric variable you have.

Overturning moments about O:

• M_OT(soil) = P_a × H_p/3 = 4.101 × 4.50 = 18.453 kip-ft/ft
• M_OT(surcharge) = P_q × H_p/2 = 0.506 × 6.75 = 3.417 kip-ft/ft

Two cases have to be investigated, because §1807.2.3 says variable loads are examined set to zero:

• Case D + H (no surcharge): FS = 43.783 / 18.453 = 2.37
• Case D + H + surcharge (driving only): FS = 43.783 / 21.870 = 2.00 ← governs

FS_OT = 2.00 ≥ 1.5 ✓. Overturning passes comfortably, which is typical: a wall with a heel full of soil rarely tips. Do not let that reassure you.

#Step 4 — Sliding: the check that actually fails

Sliding is defined by §1807.2.3 as available soil resistance at the base ÷ net lateral force. Resistance may combine three things (IBC §1806.3.1 permits them to be added):

1. Base friction — μ × the dead load, per Table 1806.2 footnote a. Friction = 0.35 × 9.547 = 3.342 klf.
2. Cohesion — c × contact area, capped at one-half the dead load (§1806.3.2). Zero for a granular soil.
3. Passive resistance in front — a triangular block over the depth D_p from the ground surface in front down to the lowest concrete. With no key, D_p = 1.00 (soil over the toe) + 1.50 (footing) = 2.50 ft, so P_p = ½ × 200 × 2.50² = 0.625 klf.

R = 3.342 + 0 + 0.625 = 3.967 klf

Against the driving side:

• Case D + H: ΣH = 4.101 → FS = 3.967 / 4.101 = 0.97
• Case D + H + surcharge: ΣH = 4.101 + 0.506 = 4.607 → FS = 3.967 / 4.607 = 0.86 ← governs

FS_sliding = 0.86 < 1.5 ✗ FAIL — and not narrowly. The wall needs its sliding resistance to nearly double. Notice it fails even with no surcharge at all (0.97): this wall would slide under its own backfill.

This is the normal outcome of a first pass, and it has a structural reason. The rules of thumb that set B were calibrated against overturning, which is a moment problem and therefore rewards width. Sliding is a force problem: widening the base adds friction only in proportion to the extra soil weight it picks up, while the driving force does not care about B at all. The two checks respond to geometry completely differently, which is why FS_OT = 2.00 and FS_sliding = 0.86 can coexist on the same wall.

Where the driving plane stops — and why two programs can disagree on the same wall. Once a shear key is added (the final design has one) there are two legitimate free bodies for sliding. StructSuite uses the footing-base free body of the classical hand method (Das, Principles of Foundation Engineering, and most textbook examples): the active pressure acts on the heel-edge plane down to the bottom of the footing, the sliding surface is the footing base, and the key contributes by mobilising passive pressure over its depth. The other convention in use takes the key-bottom free body for the sliding check instead: the active pressure is carried down to the bottom of the key, and friction is soil-on-soil. On the finished wall below (12 in key) that convention adds 15% to P_a (½ × 30 × 14.5² = 3.154 klf against 2.734) and turns the published FS_sliding = 1.62 into 4.971/3.516 = 1.41. Neither is a code requirement — §1807.2.3 fixes the factor, not the free body — and StructSuite states the one it uses inside the sliding step whenever a key exists. Adopt one, and adopt it consistently: crediting the key-bottom passive depth while stopping the active pressure at the footing bottom is the footing-base free body, not a hybrid.

#The four ways a retaining wall fails

Before fixing it, it is worth seeing the full set. Three of these are in the code checks above; the fourth is not in ACI or the IBC at all, and it is the one that shows up in photographs.

Mode 4 deserves a warning. A wall can pass overturning, sliding and bearing with room to spare and still ride a slip circle down the hill, taking the soil beneath its footing with it. Nothing in ACI 318 or IBC Chapter 18 checks that. If your wall sits at the crest of a slope, above a descending grade, on fill, or over soft or layered soils, global stability is a question for the geotechnical report — ask for it explicitly.

#Step 5 — Bearing pressure, and the middle third

The vertical load lands on the soil through the footing, and how it lands depends on where the resultant of all forces crosses the base. Take moments about O again, this time to locate the resultant:

x̄ = (M_r − M_OT) / ΣV, measured from the toe, and e = B/2 − x̄

If the resultant falls within the kern — the middle third, |e| ≤ B/6 — the whole base stays in compression and the pressure is a trapezoid:

q = (ΣV/B)(1 ± 6e/B)

If it falls outside, the heel lifts. Soil cannot pull, so the contact shortens to 3x̄ and the pressure becomes a triangle with a much higher peak:

q_max = 2ΣV/(3x̄), contact length = 3x̄

Both cases occur on this wall — one in each trial:

For trial 1:

q_max = 2,736 psf ≤ 3,000 psf ✓ (D/C = 0.91) — it passes on numbers, and it is still a problem. Note the asymmetry with the stability checks: for bearing, the surcharge's vertical weight is unfavourable (it raises the pressure), so it is included; on the resisting side of stability it was excluded. Same load, opposite treatment, because §1807.2.3 asks for the worst case each time.

Why "it passes" is not good enough here. Partial contact means the heel has lifted off. Three consequences follow: the pressure peak is far more sensitive to any extra load or any error in your weights; the wall's tilt under service load grows, because settlement concentrates under the toe; and the strength-level distributions get much worse — under 0.9D + 1.6H the trial wall bears on just 1.54 ft of its 7.5 ft base at a factored 11,155 psf. Most experienced designers treat "resultant inside the kern under service loads" as a design requirement even though the code does not spell it out, and StructSuite flags it: "Resultant falls outside the middle third — partial base contact. Consider widening the footing."

#Three levers: how to fix a wall that fails sliding

Trial 1 needs its sliding safety factor to go from 0.86 to 1.5 — a 74% increase in resistance, or an equivalent cut in the driving force. There are only three places that number can come from, and it is worth knowing what each is worth before you start guessing.

Lever 1 — geometry. More base width picks up more soil on the heel, which adds friction (μ × ΣW) and helps overturning enormously. It costs concrete and excavation.

Lever 2 — a shear key. A block cast below the footing that pushes the passive pressure block deeper. Passive resistance grows with the square of depth, so a 1 ft key on this wall takes D_p from 2.50 to 3.50 ft and nearly doubles P_p — for about 0.04 cubic yards of concrete per foot of wall. It is the cheapest structural fix there is.

Lever 3 — the backfill specification. Change what goes behind the wall. Swapping the site's SM silty sand for imported GW well-graded gravel takes the design lateral soil load from 45 to 30 psf/ft — a 33% cut in the driving force, applied to both the soil and the surcharge term. It costs the price difference on the fill, and you were going to place free-draining material behind the wall anyway.

Here is what each is actually worth on this wall — every row is a separate run of the engine, changing one thing at a time from trial 1:

Three lessons are hiding in that table, and they are the reason it is worth running the study rather than guessing:

A thicker footing made the wall worse. Going from 18 to 24 in dropped the overturning safety factor from 2.00 to 1.89 and pushed the bearing pressure past its allowable to 3,089 psf. The reason is that footing thickness enters both sides: it adds weight, but it also lengthens the pressure height H_p from 13.5 to 14.0 ft, and the driving force grows with H_p². The moment arm of the added weight is at mid-width, where it does relatively little. More concrete is not automatically more safety.

The passive term is doing a lot of work for very little concrete — which should make you cautious. Deepening the soil cover over the toe from 1 to 3 ft bought more sliding resistance than a full 2.5 ft of extra base width, and cost nothing at all. But it only exists if that soil is there, compacted, forever. See the traps.

The cheapest structural fix was written in the specification, not drawn on the section. Imported gravel behind the wall drops the driving force by a third for the price difference on the fill — and it is the same material that makes the drainage detail work. That is why the final design takes it.

#What the geotechnical report can take back

The final wall clears sliding at 1.62 — but only if it gets full credit for everything at its base. Geotechnical reports routinely say it does not. Three restrictions turn up again and again:

• "Use 50% of the passive pressure." A factor of safety on a resistance that only exists if the soil in front stays in place and stays compacted.
• "Neglect the upper 1 ft of soil in front of the wall." Landscaping, utility trenches and loose backfill all live in that zone.
• "Use two-thirds of the base friction." Less common, but it appears where the subgrade may be disturbed during construction.

Enter each one separately in Step 2 and the final wall looks like this — every row is a fresh engine run against the final design, nothing else changed:

Every one of them fails the wall — except the second reading of the disregarded foot. That is the point of the table: a design that passes at 1.62 has about 8% of margin to give away, and each of these restrictions takes two to three times that. If the report carries one, you find out at the start, not after the drawings are done.

Notice that halving the passive and excavating a foot of cover land on the same 1.42 by different routes. They are not the same operation. The percentage is a discount on the credit — the depth is unchanged and the soil is still there, so the key is still detailed for the full pressure it could see. The disregarded depth has two readings, and Step 2 asks which one the report means. Excavated — the default — takes the soil as removable (utility trenches, landscaping cuts, a future driveway): the module lowers the toe-side ground surface, the pressure restarts from zero at the new level, D_p drops from 3.50 to 2.50 ft and P_p = ½ × 200 × 2.50² = 0.625 klf. Left in place takes it as loose soil that stays but may not be relied on: only its own slice of the pressure diagram is forfeited, its weight still presses on the soil below, so the diagram keeps its origin at the surface and P_p = ½ × 200 × (3.50² − 1.00²) = 1.125 klf — the credited block is the same 2.50 ft tall, but it runs from 200 to 700 psf instead of from 0 to 500. On this wall that is the difference between FS 1.42 and 1.59, between failing and passing. Read the report literally: may be removed is excavated, not to be relied on is left in place, and when it does not say, excavated is the safe side.

One thing neither restriction touches: the weight of the soil over the toe. It is there today, resisting overturning and adding to the normal force under the footing. What may not be relied on is its passive resistance years from now — so the module keeps the weight and discounts only the push.

#The final design

Three changes from trial 1, chosen from the study above:

Everything else — 12 ft stem at 12 in, 18 in footing, 1 ft of soil cover over the toe, f′_c = 3,000 psi, Grade 60, #6 @ 8 stem verticals — is unchanged. The rest of this guide checks that wall, and the numbers below are the finished design.

Re-running the stability checks with w = 30 psf/ft and K = w/γ = 30/120 = 0.250:

• P_a = 30 × 13.50²/2 = 2.734 klf at 4.50 ft; P_q = 0.250 × 100 × 13.50 = 0.338 klf at 6.75 ft; ΣH = 3.071 klf

• Overturning: M_OT = 12.302 + 2.278 = 14.580 → FS = 55.993/14.580 = 3.84 ≥ 1.5 ✓ (4.55 without the surcharge)
• Sliding: friction = 0.35 × 10.703 = 3.746; D_p = 1.00 + 1.50 + 1.00 = 3.50 ft → P_p = ½ × 200 × 3.50² = 1.225; R = 4.971 klf → FS = 4.971/3.071 = 1.62 ≥ 1.5 ✓ (1.82 without the surcharge)
• Bearing: ΣV = 11.152 klf, x̄ = 3.966 ft, e = 0.284 ft ≤ B/6 = 1.417 ft → inside the kern, full contact, q_toe = 1,576 psf, q_heel = 1,049 psf. 1,576 ≤ 3,000 ✓ (D/C = 0.53)

The wall is stable. Now it needs reinforcement.

#Where the critical sections are — and why they differ

Before the strength checks, one detail that separates a correct retaining wall calculation from a plausible one. ACI 318-25 does not put the critical section in the same place for the stem, the heel and the toe, and the reason is mechanical, not arbitrary.

Getting this backwards is a common and consequential error: taking the heel shear at d here would remove d = 15.63 in = 1.30 ft of loaded length, dropping V_u from 9.711 to 2,158 × (4.50 − 1.30) = 6.90 klf — a 29% understatement, on the member that is usually working hardest.

#Step 6 — Stem: a one-way slab per ACI 318-25 §13.3.6

ACI 318-25 §13.3.6.1 treats the stem of a cantilever retaining wall as a one-way slab designed per Chapter 7, and §13.3.6.3 puts the critical section for a uniform-thickness stem at the stem–footing interface. The pressure on the stem acts over the retained height H_s = 12 ft only — the footing depth belongs to the stability model, not to the stem.

Service moment at the interface:

M = w·H_s³/6 + K·q·H_s²/2 = 30 × 12.00³/6 + 0.250 × 100 × 12.00²/2 = 10.440 kip-ft/ft

Lateral earth pressure adds to the load effect, so H takes a load factor of 1.6 (ASCE 7-22 §2.3.1; ACI 318-25 §5.3.8):

M_u = 1.6 × 10.440 = 16.704 kip-ft/ft

Capacity with #6 @ 8 in at the earth face (A_s = 0.44 × 12/8 = 0.660 in²/ft; d = 12 − 2 − 0.375 = 9.63 in):

• a = A_sf_y/(0.85 f′_c b) = 0.660 × 60,000/(0.85 × 3,000 × 12) = 1.294 in
• φM_n = 0.90 × 0.660 × 60,000 × (9.63 − 1.294/2)/12,000 = 26.664 kip-ft/ft

D/C = 16.704 / 26.664 = 0.63 ✓

Three Chapter 7 requirements are not optional and are checked with it:

• Tension-controlled (§7.3.3.1; Table 21.2.2 — mandatory for one-way slabs, not just a way to get φ = 0.90): β₁ = 0.850, c = a/β₁ = 1.522 in, ε_t = 0.003(d − c)/c = 0.01597 ≥ ε_ty + 0.003 = 0.00207 + 0.003 = 0.00507 ✓
• Minimum steel (§7.6.1.1): A_s ≥ 0.0018bh = 0.0018 × 12 × 12 = 0.259 in²/ft ✓
• Maximum spacing (§7.7.2.3): s ≤ min(3h, 18 in) = 18 in ✓

Stem shear. §13.3.6.1.1 permits the simple form V_c = 2λ√f′_c·b_w·d for cantilever retaining wall stems — no size-effect factor, no axial term:

• V = w·H_s²/2 + K·q·H_s = 30 × 12.00²/2 + 0.250 × 100 × 12.00 = 2.460 klf; V_u = 1.6 × 2.460 = 3.936 klf
• φV_c = 0.75 × 2 × 1.0 × √3,000 × 12 × 9.63/1,000 = 0.75 × 12.652 = 9.489 klf

D/C = 0.41 ✓ — checked at the interface, not at d above it.

Practical note on the stem. At 63% the stem has room, and that is deliberate: the stem is the cheapest part of a retaining wall to over-reinforce and the most expensive to repair. Note also what cannot be relaxed — the 12 in thickness is set by placement and cover, not by strength. Two inches of cover on the earth face plus 2 in on the formed face plus a #6 bar leaves very little in a 10 in stem, and the vibrator has to get down 12 ft.

#When the stem keeps going above the soil — wind, and the moment that reverses

Plenty of walls do not stop at the backfill. A 3-ft garden wall with a 6-ft fence on top, a screen wall at a property line, a parapet above a basement wall — the stem stands clear of the soil, and that clear face catches wind the buried part never sees. Enter it as Wind on the stem above the soil in Step 2 and the module puts it on the projection at its midheight, 1.0W for the strength checks and 0.6W for the stability ones (ASCE 7-22 §2.4.1).

What goes in that box is a net pressure, and the module applies it rather than deriving it. For a free-standing wall it comes out of ASCE 7-22 §29.3.1 — F = q_hK_dGC_fA_s, Eq. (29.3-1) — with C_f from Figure 29.3-1, which depends on the wall's clearance ratio s/h and its aspect ratio B/s. A wall sitting on the ground and a wall raised clear of it do not get the same coefficient, which is why there is no single default worth offering.

On a low wall with a tall projection it is not a correction — it is the design load. Here is a 3-ft retained wall with a 6-ft projection, 10-in stem, GW gravel backfill, nothing else changed but the wind:

At 20 psf the stem moment is more than four times what the retained soil produces, and the overturning safety factor has fallen by 60%. A wall like this is a cantilevered sign post that happens to retain a little dirt, and designing it on earth pressure alone is designing the wrong structure.

The part that catches people is the direction. Earth pressure only ever pushes one way — toward the toe — so a cantilever retaining wall carries flexural steel on the earth face and nothing on the other. Wind blows both ways. Toward the toe it adds to the earth pressure and the earth-face bars take it. Toward the heel it opposes the earth pressure, and once the projection is tall enough the net moment at the base reverses: tension on the toe face, where there is no steel.

The module checks it as its own step. The earth pressure is permanent and is resisting here, so it enters at 0.9H (ASCE 7-22 §2.3.1); the surcharge push is dropped, because a variable load that happens to help cannot be relied on:

M_rev = 1.0 · M_wind − 0.9 · M_earth

On the wall above at 20 psf that is 0.720 − 0.9 × 0.135 = 0.599 kip-ft/ft, and it reverses. Raise the retained height to 10 ft and drop the projection to a 2-ft parapet and it does not — 0.440 against 0.9 × 5.000 — and the module says so rather than demanding steel nobody needs.

Where it does reverse, the module asks for a toe-face curtain and fails the wall until one is named. The flexural demand itself is usually tiny, because the reversal is the difference of two moments; what matters is that the face carrying it has no reinforcement at all. In practice the toe-face curtain mirrors the earth face, and the shrinkage-and-temperature requirement of §24.4.3.2 sets the bar size long before the flexure does.

If what stands above the soil is a fence, say so. The table above is a concrete projection — a 9 ft stem on a 3 ft backfill — and that is the wrong structure for most of the walls that carry something above the soil. A 42 in guardrail, a 6 ft wood fence, a screen: none of them is concrete, and modelling one by raising the stem does two things that are not true. It adds the projection's own weight — 6 ft of 10 in stem is 750 lb per foot of wall — to the resisting side of every stability check, and it ties the wind input to that fake projection. Enter it instead as a fence or guardrail on the stem top in Step 1 (its height), with its own Wind on the fence (the net pressure) and its Fence weight in Step 2 — left blank for an open rail. The module hangs the wind at the fence mid-height above the whole stem, carries it through the stem, the stability checks and the reversal exactly as it carries wind on a projection, and adds no stem weight for it. The same 6 ft at 20 psf, three ways:

The stem does not care — 0.94 kip-ft/ft either way, because the wind is the same wind at the same height. The stability checks care a great deal: the concrete version was borrowing 750 plf it never had, and the overturning factor that read 7.32 is 5.58 once the fence is a fence. On this wall both numbers pass; on a wall near 1.5 the difference is the design. The guardrail row is the other half of the point — an open rail catches wind on its pickets and posts only, so the pressure over its height is a fraction of a solid wall's (ASCE 7-22 gives the open-sign coefficients by solidity ratio in Figure 29.4-2), and the module takes that net number rather than inventing one.

#A stepped stem — the second critical section

The stem moment is a cubic in the height: at half-height it is an eighth of what it is at the base. So a wall that needs #6 @ 8 and 12 in at the interface needs nothing like that near the top, and the two standard economies are to stop every second bar partway up and to thin the stem above a step. Both are one input in Step 1 — the height of the step above the footing and the thickness above it — with the continuing bar pattern picked in Step 3 (the same size, every bar or every second or third bar; the rest stop). The earth face stays plane and the step is on the toe face, so the bars that continue run straight through it.

Two things change in the calculation. ACI 318-25 §13.3.6.3 puts the critical section of a uniform stem at the interface and says of a wall "with a tapered or varied thickness" that "shear and moment shall be investigated throughout the height of the wall" — so the module adds a second critical section at the step and checks the upper section there against every load above it, collected from the same derivations the base uses: earth, water, surcharge, the adjacent footing's strip-load push, wind on a projection or a fence, and (with seismic on) the increment integrated down to the step. And the bars that stop are held to the termination rules of §7.7.3: they must run past the point where they are no longer required by the greater of d and 12d_b (§7.7.3.3); the continuing bars must be developed beyond that point (§7.7.3.4 — they are the base bars, anchored at the footing in Step 10, so they are); and a bar may stop in a tension zone only if V_u ≤ ⅔φV_n at the cutoff, or the continuing bars give twice the area the moment there needs and V_u ≤ ¾φV_n (§7.7.3.5). The point where the stopped bars are no longer needed is not read off a chart: the module solves M_u(y) = φM_n of the continuing pattern on the full moment envelope.

The final wall of this guide, stepped several ways (the base stays 12 in with #6 @ 8 in every row):

Read the last column before the D/C column. In the 1.5 ft row the upper section has strength to spare — D/C 0.83 — and the wall still fails, because the half of the bars that stop are no longer required only from 0.78 ft up and §7.7.3.3 makes them run 9.63 in (d, which governs over 12d_b = 9 in here) beyond that: the cutoff may be no lower than 1.58 ft. Strength is not the rule that places a cutoff; the extension rule is. Thinning the stem moves that line too — at 8 in above the step the continuing bars are needed higher, and the cutoff may be no lower than 3.62 ft.

What the step buys is modest on this wall and honest about it: 10 in above 6 ft saves 150 lb of concrete per foot of wall, and the stability factors drop by about 1% because the stem is a small part of what holds the wall down. On a taller wall with a heavier base pattern the saving is real; what the module makes sure of is that the thing it cannot see from the base — a cutoff too low, a thinned section too thin — is checked where it happens.

#A concrete masonry (CMU) stem — TMS 402-22 allowable stress design

Most residential retaining walls in the western states are not cast concrete at all: they are 8 or 12 in concrete block, solid grouted, with one bar in every cell or every other cell, on a cast concrete footing. The module designs that stem too. Pick Concrete masonry (CMU) in Step 1 and the thickness becomes a block — 6, 8, 10 or 12 in nominal — because a masonry wall is not any thickness you like; the design reads the specified thickness, nominal less the 3/8 in mortar joint (TMS 402-22 §4.4.1.1). Step 3 then asks for what the standard needs and nothing else: f′_m (the specified strength of the masonry — TMS 602 Table 2 gives it from the unit strength and the mortar type, 2,000 psi units with Type S mortar being 2,000 psi), the bar grade, the unit density (ASCE 7-22 Table C3.1-1a gives the weight of a fully grouted wythe), where the bar sits in the cell, and the Seismic Design Category. The bars go in at a cell spacing. The footing underneath is still ACI 318-25: its hooks, mats and shrinkage steel do not change.

What changes is the stem check, and it is a different kind of check. TMS 402-22 Chapter 8 is allowable stress design: service-level loads (ASCE 7-22 §2.4.1 — H at 1.0 where it adds, 0.6W, 0.7E), linear-elastic stresses on the cracked transformed section with the masonry's tension neglected (§8.3.2), and three limits — the masonry compressive stress from flexure with the axial load may not exceed 0.45 f′_m (§8.3.4.2.2), the steel stress may not exceed F_s (32,000 psi for Grade 60, 20,000 psi for Grade 40 or 50 — §8.3.3.1), and the axial force alone may not exceed P_a of Eq. 8-16 or 8-17, with h the effective height of an unbraced cantilever (§8.3.4.2.1). The commentary is explicit that the unity interaction equation of §8.2.4 is not used for reinforced masonry (C8.3.4.2.2), so the module does not sum stress ratios: it solves the neutral axis with the axial load present and reads the two stresses off that section. Shear is f_v = V/A_nv against F_v = (F_vm + F_vs)·γ_g with F_vm from Eq. 8-23 and the M/(V·d_v) cap of §8.3.5.1.2 (§8.3.5) — a stem that would need shear reinforcement fails, because none is modelled. Each bar may count on a compression width no wider than its spacing, six times the nominal thickness or 72 in (§5.1.2.1). The dowel lap in the masonry is Eq. 6-2 (§6.1.6.3.1), not less than 12 in — and K in that equation is the least of the masonry cover, the clear spacing between splices and 9d_b, which is why a bar pushed to the earth face laps longer than a centred one. A stepped CMU stem gets the same second critical section as the concrete one, with the stopped bars held to §6.1.10.1.3 to .5. Chapter 7's prescriptive reinforcement for a nonparticipating element is checked by SDC (§7.4.3.1, §7.4.4.1), and the calculation states the Quality Assurance Level of Table 3.1.

The sample site of this guide, with a 6 ft masonry stem on a 15 in footing with a 24 in key (the 10 ft concrete stem is more than an 8 in block wall carries on this backfill): f′_m = 2,000 psi, Grade 60, 125 pcf units (81 psf), the sloped backfill and the Mononobe-Okabe increment of the sample, SDC D. The one design variable that matters in a thin masonry stem is where the bar sits:

Read the two #5 @ 8 rows together. The same bars, moved from the middle of the wall to the earth face, take d from 3.81 to 5.81 in and the allowable moment from 2.42 to 4.91 kip-ft/ft — twice the capacity for no more steel — and that is the whole difference between a wall that fails its earthquake case at 1.44 and one that passes at 0.71. Centred bars are what the masons prefer and what the standard's cover rule permits either way; on a wall this thin, under this seismic increment, they are not enough. The price of the earth-face position is the lap: with the bar 1½ in from the face, K of Eq. 6-2 is that 1½ in, and the lap stretches from 20 in to 45 in. Doubling the bar size helps the stresses and then fails a check that has nothing to do with masonry — the standard hook of a #6 dowel needs more than the 15 in footing has (ACI 318-25 §25.4.3.1). The masonry rows also show what the ASD stresses mean: at D/C 0.31 the sample stem works at 282 psi of masonry stress against 900 allowed and 7,170 psi of steel against 32,000 — a wall designed to remain far below cracking stresses under service loads, which is exactly what Chapter 8 intends.

Two walls from a commercial calculation package were run through this stem as a check — an 8 in wall with #5 @ 8 below 2 ft and #5 @ 16 above, and a 12 in wall stepped to 8 in: with the same bar depth entered, the allowable moments agree to the decimal (1,812.8, 3,403.2, 3,132.6 and 6,167.2 lb-ft per foot), so the Chapter 8 section model is the same one that package uses. The demands differ only where this guide's other sections already explain they should — the adjacent footing's push and the wind on a fence, which the module models as what they are.

#A footing behind the wall

The most common load on a residential retaining wall that the surcharge box cannot describe is a building footing standing in the backfill — the garage or house wall footing a few feet behind the stem. It is not a uniform pressure. It is a strip of 1,000–3,000 lb per foot bearing on a 1.5–2 ft wide footing, and the soil does not spread that into a flat K·q band: the push it puts on the wall is concentrated near the top of the stem, peaks a short way below the footing's bearing level, and tails off with depth. Enter it in Step 2 under Adjacent footing behind the wall — the footing's total service load per foot of wall (D + L), its width perpendicular to the wall, the distance from the stem back face to its centreline, and, if it is buried, how far its bearing level sits below the backfill surface.

What the module does with it is the classical method, stated as such. The footing is a strip load of intensity p = Q_f/b on the surface of an elastic half-space, and the horizontal stress it induces on a vertical plane a distance a from its near edge is the Boussinesq line-load solution integrated across the strip:

σ_x(z) = (p/π)·[atan((a+b)/z) − atan(a/z) − z(a+b)/((a+b)² + z²) + z·a/(a² + z²)]

Both the resultant and its moment have closed forms — ∫σ_xdz = (p/π)·[z·atan(ξ/z)] and ∫z·σ_xdz = (p/π)·[(z²/2)·atan(ξ/z) − ξz/2 + (ξ²/2)·atan(z/ξ)], each evaluated between ξ = a and a + b — which is what lets the calculation print numbers you can reproduce with a pocket calculator. The push is evaluated twice, because the wall has two planes: on the stem it acts over the stem's height from the footing's near edge; on the heel-edge plane that the stability checks use, only the part of the strip beyond the heel edge pushes, from its own shorter distance, while anything bearing over the heel enters the free body as a vertical load on it instead — dead or live, by your call, exactly as the uniform surcharge does. The code's part is short and unambiguous: lateral pressure from surcharge loads shall be added to the soil load (IBC 2024 §1610.1; ASCE 7-22 §3.2.1), it is H by definition (§2.2), and it takes 1.6 in the strength combinations even where the surcharge's own weight is well defined (C2.3.1).

One judgment sits in the middle of this, and the module ties it to a choice you have already made. Measurements on walls that cannot yield found roughly twice the elastic pressure, and the classical rigid-wall correction doubles it. So the at-rest condition doubles the push; the active condition — a cantilever wall free to rotate, this wall — uses the elastic value as it stands. The calculation says which it did, and why.

Here is what a 3,000 plf footing, 2 ft wide, does to the final wall of this guide — the one that cleared sliding by 8%:

Three things in that table are worth more than the numbers. First, sliding fails at every distance tried, and the push barely decays with distance: a 13.5 ft plane collects the stress from a footing 12 ft away almost as fully as from one at 6 ft, because the pressure it loses near the top it gains lower down. A wall with 8% of sliding margin has no room for a footing behind it at any distance a house is likely to stand. Second, the uniform-surcharge stand-in gets the wrong members wrong. 250 psf finds the sliding failure, roughly, but it puts the push at mid-height, where a uniform band belongs — the real push from the 6 ft footing acts at 11.5 ft, and that is why the stand-in says the stem is at 0.79 when it is at 0.93 and the toe at 0.37 when it is at 0.67. Third, the at-rest doubling is not a refinement, it is a different wall — overturning at 1.10, bearing at three and a half times allowable. That is the reason the module ties the factor to the pressure condition and prints which one it used, rather than leaving a ×2 to be remembered or forgotten.

Move the same footing closer — centreline 4 ft behind the stem, so 1.5 ft of its 2 ft width bears over the heel — and the picture inverts. Only a 0.5 ft sliver stands beyond the heel edge, so the push on the heel-edge plane drops to 239 lb and the stability checks improve (declared permanent: FS_OT 4.15, FS_sliding 1.74 — the 2.25 klf bearing on the heel is now resisting weight); but the stem does not care which side of the heel edge the footing is on. Three feet from the back face the strip puts 858 lb on the stem plane at 7.8 ft above the interface, and the stem fails at D/C 1.03. A far footing fails the wall's sliding; a close one fails its stem; neither is the uniform surcharge's failure.

Two limits, both stated on the calculation. The half-space solution assumes a level surface, so a footing on a sloped backfill is refused and sent to the geotechnical report; and a footing bearing at or below the top of this wall's footing is not a surcharge on the backfill at all — it is a foundation beside a foundation, which is a different problem.

#Step 7 — Heel: it hangs, and it carries the whole soil block

The heel is loaded downward by everything above it and hangs off the stem. Upward bearing pressure under the heel is conservatively neglected — a simplification that is on the safe side and is what makes the heel the usual governing member.

Two gravity combinations of ASCE 7-22 §2.3.1 compete here, because lateral earth pressure puts nothing on the heel — it is the one member of the wall where plain 1.4D can govern:

• Dead pressure D = γ·H_s + w_c·t_f = 120 × 12.00 + 150 × 1.50 = 1,665 psf
• Combination 1a — 1.4D: w_u = 1.4 × 1,665 = 2,331 psf
• Combination 2a — 1.2D + 1.6L: w_u = 1.2 × 1,665 + 1.6 × 100 = 2,158 psf

1a governs: the extra 0.2 on 1,665 psf of soil and concrete (333 psf) outweighs 1.6 on a 100 psf surcharge (160 psf). The crossover is a variable surcharge of 0.2·D/1.6 ≈ 208 psf — above that, 2a takes over. Both critical sections are at the back face of the stem:

• M_u = w_u·L²/2 = 2,331 × 4.50²/2 = 23.601 kip-ft/ft
• V_u = w_u·L = 2,331 × 4.50 = 10.489 klf, taken at the face (§13.2.7.2 / §7.4.3.2)

Capacity with heel TOP bars #6 @ 10 in (A_s = 0.528 in²/ft; d = 18 − 2 − 0.375 = 15.63 in):

• a = 1.035 in → φM_n = 35.895 kip-ft/ft → D/C = 0.66
• φV_c = 0.75 × 2√3,000 × 12 × 15.63/1,000 = 15.405 klf → D/C = 0.68 ← governs the heel
• ε_t = 0.03549 ≥ 0.00507 ✓ tension-controlled; A_s = 0.528 ≥ 0.0018 × 12 × 18 = 0.389 in²/ft ✓; s = 10 ≤ 18 in ✓

✓ PASS at D/C = 0.68. Note where the surcharge went: at 1.6L it is a live load on top of the heel in combination 2a, its lateral push entered the stability checks as H — and in the governing 1a combination it is absent altogether. Same 100 psf, three different treatments across this page, each one dictated by which effect it is producing and which combination is being checked. (Until 2026-09-20 StructSuite checked the heel at 2a only — 21.850 kip-ft/ft and D/C 0.63 on this wall. The 1.4D check is an 8% correction here, and a light, permanent or absent surcharge always exposes it.)

#Step 8 — Toe: designed from the factored bearing pressure

The toe is loaded upward by the bearing pressure, less the weight of the toe itself and its soil cover. The pressure distribution has to be rebuilt at strength level for each combination, because the load factors change both the vertical load and the overturning moment, and therefore move the resultant:

The 0.9D + 1.6H combination is the one people forget. It strips the wall to 90% of its dead load while keeping the full 1.6 on the earth pressure, and it is often what pushes the resultant outside the kern at strength level even when the service case is comfortably inside — exactly what happens here.

The 1.4D + 1.6H row is combination 1a with the H rule applied: no live load, the full 1.4 on the dead weights, and still 1.6 on the earth pressure because H adds to the toe's load effect (§2.3.1; ACI §5.3.8). More vertical load at the same push pulls the resultant back toward the centre but lifts q_toe to the highest of the three — and on this wall it governs the toe. (StructSuite added it on 2026-09-20; 2a had governed at 7.969 kip-ft/ft, 3% under.)

Net of the factored toe self-weight and soil cover (q_u − γ_D × (225 + 120) psf at each combination's own 1.2 / 0.9 / 1.4), with moment at the stem front face and shear at d from the face:

• Governing (1.4D + 1.6H): M_u = 8.222 kip-ft/ft, V_u = 3.278 klf — against 7.969 / 3.181 for 2a and 7.667 / 3.080 for 0.9D + 1.6H
• Toe BOTTOM bars #5 @ 9 in (A_s = 0.413 in²/ft, d = 18 − 3 − 0.313 = 14.69 in): a = 0.810 in → φM_n = 26.565 kip-ft/ft (D/C = 0.31), φV_c = 14.480 klf (D/C = 0.23)

D/C = 0.31 ✓. The toe is nowhere near critical — but it cannot borrow the heel's bars. Its tension face is the bottom; the heel's is the top. Two separate mats, and mixing them up is one of the few retaining wall detailing errors that is unambiguously a failure.

Notice also that the toe bars are not sized by strength at all. A_s = 0.413 in²/ft is set by the shrinkage-and-temperature minimum of 0.389 in²/ft, and the bar size was set by development length (Step 10). That is normal for footings on this scale.

#Step 9 — The shear key: what it resists, it must also carry

Step 4 leaned on the key for 0.600 klf of the 1.225 klf of passive resistance. That force does not come from nowhere: the soil in front bears on the key's front face and the key has to carry it back into the footing. It is a short cantilever hanging off the base, and it gets designed like one.

The passive pressure over the key depth is the trapezoid between the footing bottom and the key bottom (200 psf/ft lateral bearing, IBC Table 1806.2):

• p_top = 200 × 2.50 = 500 psf; p_bot = 200 × 3.50 = 700 psf
• P_p,key = ½(500 + 700) × 1.00 = 600 plf, acting ȳ = (1.00/3)(500 + 2 × 700)/(500 + 700) = 0.528 ft below the junction

Passive bearing is lateral earth pressure, so it enters at 1.6H — the same factor the active side carries on the stem:

• M_u = 1.6 × 0.600 × 0.528 = 0.507 kip-ft/ft; V_u = 1.6 × 0.600 = 0.960 klf

The section is 12 in of wall length by the key width, with the stem #6 @ 8 pattern continued into the key and the bars toward the toe face, where the junction moment puts the section in tension. (The module places the key's toe-side face at the stem front face unless you offset it — Step 1 takes an offset, positive toward the heel, for the details and programs that set the key back under the heel. The passive block it mobilises and its own design do not change; what changes is that the stem bars can no longer be carried through, so an offset key gets separate dowels at the stem pattern, and the reinforcement drawing says so.) Cast against earth on both faces, so d = 12 − 3 − 0.375 = 8.63 in:

• φM_n = 23.694 kip-ft/ft → D/C = 0.02; φV_c = 0.75 × 2√3,000 × 12 × 8.63/1,000 = 8.503 klf → D/C = 0.11 ✓

Three details decide the drawing rather than the numbers:

Shear stays at the face. The footing reaction does not put the key's end region in compression, so §7.4.3.2 gives no d relief — the same rule as the heel, the opposite answer from the toe.

The dowels need a hook. Straight development of a #6 is ℓ_d = 32.9 in and an 18 in footing offers only 18 − 2 = 16.0 in. A standard hook does fit: ℓ_dh = 11.5 in ≤ 16.0 in ✓. Detail the key dowels hooked into the footing; a straight bar would not be developed whatever the D/C says.

The key gets its own shrinkage and temperature steel. Its flexural reinforcement runs one way only, so §24.4.3.2 requires steel perpendicular to it — bars running along the wall inside the key. Sized from the key's own gross section, A_s ≥ 0.0018 × 12 × 12 = 0.259 in², and placed where S+T steel does its work, at the faces: (4) #5 continuous — 2 levels down the depth × a bar at each face (1.240 in²), at the footing longitudinal bar size. Sizing from the key’s own section is what makes a deep key come out right. A 12 × 36 in key needs A_s ≥ 0.0018 × 12 × 36 = 0.778 in², and the §24.4.3.3 spacing cap of min(5w, 18 in) = 18 in forces three levels down its depth rather than two. Width matters the same way: the cap applies across the key as well as down it, so a key wider than 24 in picks up a third bar across each level.

A plain-concrete key is not what this is. ACI 318-25 §14.5.2.1 allows only 5λ√f′_c on the gross section at φ = 0.60, which an ordinary key does not satisfy on its own moment. If your detail shows an unreinforced key, it is decoration.

#Step 10 — Development, hooks and laps: where the toe length came from

Three development questions decide whether the bars actually work — and one of them is why the final design has a 3.0 ft toe instead of 2.5 ft.

Stem dowels into the footing (§25.4.3.1, Table 25.4.3.2). With ψ_e = 1.0 (uncoated), ψ_s = 1.00 (≤ No. 9), ψ_cc = 0.7 (side cover normal to the hook plane ≥ 2.5 in — automatic in a continuous wall — and ≥ 2 in over the 90° hook) and ψ_r = 1.0:

ℓ_dh = (60,000 × 1.0 × 1.00 × 0.7 × 1.0)/(50 × 1.00 × √3,000) × 0.750 = 11.5 in ≥ max(8d_b = 6.0, 6 in)

Available = t_f − cover_bottom − d_b(toe mat) = 18.0 − 3.00 − 0.625 = 14.4 in ✓. The 90° hook extension is 12d_b = 9.0 in, placed near the bottom of the footing with the free end toward the toe (R13.3.6.3).

Dowel-to-stem lap (§25.5.2.1, Class B): ℓ_d(#6 vertical, ψ_t = 1.0) = 32.9 in → lap = 1.3ℓ_d = 42.7 in of dowel projection above the interface. Round that up on the drawing — 3 ft 7 in, as the reinforcement detail calls it — and remember the dowel has to be that long before the stem is formed.

Heel and toe mats (§13.2.8.3; Table 25.4.2.3). The heel TOP bars sit above 15.3 in of fresh concrete — more than 12 in — so they take ψ_t = 1.3 (Table 25.4.2.5):

• Heel: ℓ_d = 42.7 in vs 4.50 × 12 − 3 = 51.0 in available ✓
• Toe: ℓ_d(#5) = 27.4 in vs 3.00 × 12 − 3 = 33.0 in available ✓

This is where the geometry came from. Trial 1's 2.5 ft toe with #6 bars needed ℓ_d = 32.9 in and offered only 27.0 in — a straight-bar development failure that no amount of extra area fixes. Two things solved it: a smaller bar at a tighter spacing (#5 @ 9 instead of #6 @ 12, same area class, ℓ_d down to 27.4 in because ℓ_d is proportional to d_b), and a longer toe (3.0 ft), which the sliding check wanted anyway. When a footing projection is too short to develop its bars, your options are exactly those two, or a hook at the end — or, as the next section shows, the other equation the Code offers.

#The general equation — what the Eq. 25.4.2.4a path buys

Every straight length above comes from the simplified Table 25.4.2.3, which is what the module runs unless you say otherwise. ACI 318-25 §25.4.2.1(a) offers a second path — the general development length equation, Eq. (25.4.2.4a) — and its commentary R25.4.2.1 says exactly what the table is: the general equation evaluated at a preselected confinement term. The equation uses the wall's own. c_b is the lesser of the cover to the bar centre and half the bar spacing (R25.4.2.4), the confinement term (c_b + K_tr)/d_b is capped at 2.5, and K_tr = 0 is permitted outright — a slab or a wall has no ties crossing the bars anyway. ψ_s = 0.8 for No. 6 and smaller bars enters here (Table 25.4.2.5); the simplified table had already folded it in

On this wall every mat clears the cap — the stem bars have c_b = min(2.00 + 0.375, 8/2) = 2.375 in, so (c_b + K_tr)/d_b = 3.17 → 2.5 — and the ratio to the table is simply (3/40) × 25 × 0.8/2.5 = 0.60. Switching Step 3 to the general equation gives:

The development demand-capacity ratio moves from 0.84 (the heel mat, 42.7/51.0) to 0.80, and it is now the hook that governs. The key dowels still need theirs: 19.7 in straight does not fit the 16.0 in of footing above the junction.

Re-read trial 1 in this light. Its 2.5 ft toe with #6 @ 12 offered 27.0 in and the table demanded 32.9 — but the equation demands 19.7, and the toe would have developed. The 3.0 ft toe and the #5 bars were the price of the simplified table, not of the Code. That is a fair trade on a wall where sliding wanted the longer toe anyway; on a wall where it did not, the general equation is 6 in of footing you do not have to pour. What it never buys is stability — trial 1 still slides at 0.86 and still lands outside the kern whichever equation develops its bars.

#Step 11 — Minimum and interface reinforcement

The bars nobody calculates and every plan checker looks for:

• Wall–footing interface (§16.3.4.2 → Table 11.6.1): A_s,min = 0.0015 × 12 × 12 = 0.216 in²/ft; the #6 @ 8 dowels provide 0.660 ✓
• Stem horizontal shrinkage + temperature (§24.4.3.2): 0.0018 × 12 × 12 = 0.259 in²/ft required; #4 @ 9 provides 0.267 in²/ft ✓, spacing 9 ≤ min(5h, 18) = 18 in (§24.4.3.3) ✓. A single curtain is expressly permitted in cantilever retaining wall stems regardless of thickness (§11.7.2.3 exception) — which is why the horizontals sit inside the verticals on the earth face and there is no second mat on the front face.
• Footing longitudinal S+T (§24.4.3.2) — the bars running along the wall: 0.0018 × 12 × 18 = 0.389 in²/ft required; #5 @ 9 each face — a curtain against each mat, as the drawing shows — provides 2 × 0.413 = 0.827 in²/ft ✓, spacing ≤ 18 in ✓. These are a real selection with real limits, not a note to the detailer. (Until 2026-09-20 the module counted a single curtain, 0.413 in²/ft — half of what its own reinforcement drawing had always shown. The E.F. count is what the detail builds.)

#The reinforcement drawing

Every number above resolves into one drawing. This is the detail a contractor builds from:

StructSuite draws this detail live from the same numbers the checks used, so the hook length, the lap and the embedments on the drawing can never disagree with the calculation — open the Retaining Wall module and it is Step 5.

#Practical detailing — the details engineers actually draw

The drawing above is a correct detail, not the correct detail. On real sheets the same wall gets detailed several different ways, and the differences are about buildability and risk, not about strength. Here is the menu, with what each one costs.

#Getting the stem bars into the footing

For this wall detail A is the only one of the three that works: a straight #6 needs ℓ_d = 32.9 in of embedment and an 18 in footing offers 14.4 in, so the dowel has to be hooked (§25.4.3.1) — which is why the hook shows up in the calculation and not just on the drawing. Detail B is real, but only on short walls: a 12 ft bar cage standing unbraced while the footing is poured is a safety and tolerance problem, and it forces the stem forms over a full-height cage. Detail C stays a constructability choice: ACI permits every bar to be lap-spliced at one section for a Class B splice, so staggering buys you room for the vibrator, not a shorter lap. (ACI does allow a shorter Class A splice under Table 25.5.2.1, but only when the section has at least twice the steel it needs and no more than half the bars are spliced there — StructSuite never assumes either, so it reports the Class B length.)

Three things a detailer gets wrong here, in order of frequency. The lap drawn as two bars meeting rather than overlapping — see the caption above. The hook turned the wrong way: R13.3.6.3 puts it near the bottom of the footing with the free end toward the toe, so the bend works against the direction the stem wants to rotate. And the dowel projection dimensioned to the bar length instead of the lap length: what the drawing must state is 42.7 in above the interface, not the overall dowel.

#Where the footing mats start and stop

This is the question the section drawing earlier in this guide answers with detail B: the heel top mat runs through to the stem front face, and the toe bottom mat continues 27.4 in past the stem back face — each one carried its own development length past its own critical section, exactly as §13.2.8.3 and Table 13.2.7.1 require. That is the least-steel answer and it is what the module reports.

Most production drawings use detail A instead, and it is a defensible choice: on an 8.5 ft footing the extra steel is a few pounds per foot of wall, while a bar mat cut short in the field is a repair. Detail C is not an alternative — it is what you are forced into when a projection is shorter than the bar can develop straight, and it is the reason this wall’s toe grew from 2.5 ft to 3.0 ft rather than getting hooks.

Bars that are not in the calculation but must be on the drawing. Three of them, every time:

• Stem horizontal S+T on both faces of a two-curtain stem. This wall uses a single curtain, which §11.7.2.3 expressly permits in cantilever retaining wall stems regardless of thickness — but as soon as the stem gets a second curtain (usually past about 16 in, or where the designer wants crack control on the exposed face), the horizontals double and the outer face needs its own vertical support bars.
• Corner and end bars. Horizontal bars have to be continuous around a corner or the corner opens. Standard practice is a separate corner bar lapped to the horizontals in each leg, and at a free end the horizontals get a U-bar or a hook so the last vertical is tied into something. ACI gives you the lap length; the geometry is a drafting convention.
• Joints. Long walls need vertical joints or they will crack at intervals of their own choosing. Control joints at roughly 20–30 ft with the horizontal steel continuous through them and a formed groove on the exposed face; full construction or expansion joints at wider spacing and at every abrupt change in wall height, with a shear key or smooth dowels across so the two panels deflect together. Put a joint at every step in the footing.

Curtailing the stem verticals is in the calculation. Cutting alternate bars partway up where the moment has dropped is legitimate and saves steel on tall walls, and StructSuite checks it: enter the cutoff height as the Stem step height above the footing in Step 1 with the same thickness above it, pick the continuing bars in Step 3, and Step 4 checks the section at the cutoff and the extension of the bars that stop (A stepped stem). This guide's D/C of 0.63 is the uncurtailed interface check; do not read it as covering a curtailed detail.

#Drainage — the detail that decides whether any of this is true

Every number on this page rests on one assumption: the backfill is drained. IBC 2024 §1610.1 requires design for full hydrostatic pressure where it is not, and Table 1610.1 footnote a is explicit — submerged or saturated soil pressures shall include the weight of the buoyant soil plus the hydrostatic loads. Drainage installed per §1805.4.2 and §1805.4.3 is what buys the exemption.

Here is what that is worth. Take the same wall and saturate the gravel: the soil's contribution drops to the buoyant unit weight, but full water pressure adds on top of it — K_a·γ′ + γ_w ≈ 0.25 × (135 − 62.4) + 62.4 ≈ 80 psf per foot of depth, against 30 psf/ft drained. That 80 psf/ft is the back-of-the-envelope version; the module's own groundwater model (next subsection), run with the water table at the backfill surface, gives the finished wall:

#When the water table is above the footing — what the module does

Since 2026-09-20 the module takes the design groundwater level as Groundwater above footing bottom, h_w in Step 2 — measured above the bottom of the footing on the retained side — together with the saturated unit weight γ_sat from the report, and applies IBC 2024 §1610.1 and Table 1610.1 footnote a ("submerged or saturated soil pressures shall include the weight of the buoyant soil plus the hydrostatic loads") the way ASCE 7-22 §3.2.1 and §3.2.2 spell it out:

• Buoyant earth pressure below the water table, at the same K: w′ = w·(γ_sat − γ_w)/γ = 30 × (135 − 62.4)/120 = 18.15 psf/ft. The earth triangle kinks at the water table — w·z above it, then w′ below.
• Full hydrostatic pressure, γ_w = 62.4 pcf, as its own triangle on the same plane: P_w = ½·γ_w·h_w² at h_w/3.
• Saturated soil weight over the heel below the water table (135 pcf here), moist above it.
• Base uplift per ASCE 7-22 §3.2.2, with the head measured from the underside of the footing: γ_w·h_w at the heel edge, falling to zero at the toe edge because the front of the wall is taken as drained — U = ½·γ_w·h_w·B, acting upward at 2B/3 from the toe. It drives overturning, comes off the weight that friction and the §1806.3.2 cohesion cap count on, is deducted from the vertical load in the bearing check, and adds to the toe's upward pressure in strength design.

Ground water pressure is H by definition (ASCE 7-22 §2.2), so it enters the stability checks at 1.0 and the strength combinations at 1.6. What the module will not do is combine a water table with the seismic increment: Mononobe-Okabe and Seed-Whitman assume the groundwater below the base of the wall (RP 12 p. 355), and a submerged backfill's seismic pressure is the geotechnical report's to give. Nor does it decide the water elevation for you — the design level, the head at the toe, and whether water can stand on both sides are the report's call.

Run the finished wall with the water rising behind it (γ_sat = 135 pcf; every other input as designed):

Three things to read off that table. Sliding goes first, and it goes early: with the water only 3 ft up, the hydrostatic thrust is a mere 0.28 klf — but the uplift of 0.80 klf has already taken 0.28 klf off the base friction as well, and between the two the wall is under 1.5. The heel barely notices: saturated soil is heavier than moist soil, and that is the whole effect on a member whose load is what sits on it. And the stem is untouched until the water climbs above the footing top, because the stem feels only what stands above the interface. A wall that fails when it floods is the normal outcome, which is the point of the section: the drain is the design assumption, and the water table is what you enter when it cannot be relied on.

The drain is a structural element. Detail it and specify it like one:

• Free-draining material behind the wall, at least 12 in wide, extending from the footing up to near the surface. Where the whole backfill is imported gravel — as here — the drainage zone and the structural backfill are the same material, which is a large part of why that specification was the cheapest fix in the lever study.
• Filter fabric between the gravel and the native soil on the back and top faces. This is the component most often value-engineered out and it is the one that decides whether the drain still works in year five: without it, fines migrate into the gravel and the drain silts up.
• A perforated collector pipe, 4 in minimum, holes down, bedded in the gravel at the base, sloped continuously to a real outfall — daylight, a storm structure, or a sump. A pipe that ends in the backfill is worse than no pipe, because it collects water and stores it.
• Weep holes through the stem, typically 3 in at 5–10 ft on centre with a gravel pocket and fabric behind each one. Treat them as the backup and as the visible evidence that the system works, not as the primary drain — a weep hole that never runs is not proof of a dry wall, it may be proof of a blocked one.
• A low-permeability cap — pavement, a clay layer, or simply grading that falls away from the wall — so that surface water is shed rather than fed into the drain. Most walls that get into trouble are drowned from the top, not from below.
• The active wedge matters. Imported gravel only changes the pressure if it extends past the failure surface, a plane rising at roughly 45° + φ/2 from horizontal at the back of the heel — 63° for this backfill. A 12 in ribbon of gravel against a silty-sand backfill is a drain, not a 30 psf/ft design lateral soil load. If your calculation uses the gravel's number, your section must show the gravel filling the wedge.

#Building it: what actually goes wrong on site

The calculation is the easy half. Here is what to put on the drawings and watch for in the field.

Set the dowels in the footing pour, to a template. Dowels pushed into wet concrete after the screed drift out of position and lose cover, and the hook rarely ends up at the bottom where R13.3.6.3 wants it. Straightening or re-bending a partially embedded bar in the field is not a detailer's decision — it needs the engineer of record. Get the template right and the stem cage drops over it.

Roughen the construction joint at the top of the footing. The interface between footing and stem is a cold joint carrying the full stem shear and the dowel force. Clean it of laitance, roughen it, and wet it to saturated-surface-dry before the stem pour. The code counts on the dowels crossing that joint (§16.3.4.2); it does not count on a smooth, dusty surface.

Do not backfill early, and do not backfill heavy. Two rules that save more walls than any calculation. The stem must reach its specified strength first — a green cantilever has no capacity at its base. And compact in thin lifts with light equipment within about half the wall height of the back face: a heavy roller working right behind a wall induces compaction pressures well above the active values this design used, and it is a classic cause of a wall that leans before it is even in service.

Protect the toe soil, on the drawing and on site. This design counts on 1.0 ft of compacted soil in front of the toe plus the key for 25% of its sliding resistance. Note it on the drawings — "soil in front of the toe to remain undisturbed; do not excavate for utilities within X ft" — because a trench dug there in year three quietly takes the wall from FS 1.62 to 1.22.

Place the two mats the right way up. Heel bars on top, toe bars on bottom. Use chairs, check them before the pour, and photograph them. Inverting the heel mat is a genuine collapse mechanism, not a serviceability issue.

Verify cover, especially at the earth face. The stem's d = 9.63 in assumes 2 in of cover. Lose an inch to sloppy chairs and φM_n drops by about 10% — more than the whole margin some walls have.

Detail the joints. Long walls need vertical joints for shrinkage and temperature; without them the wall will pick its own locations. Control joints at roughly 20–30 ft with the horizontal steel continuous through, and full construction/expansion joints at greater spacing or at abrupt changes in wall height, with a shear key or dowels across so the two panels deflect together. Change wall height in steps, not gradually, and put a joint at each step.

Cure the stem. A 12 ft stem is mostly formed surface with a high surface-to-volume ratio; it cracks if it is stripped early and left to dry. Cracks on the earth face are also the path water uses to reach the reinforcement.

#Expansive soils — what the code lets you assume

Everything above assumed a backfill that behaves the way Rankine says it does. An expansive clay does not. It develops swelling pressure when it takes on water — a pressure that has nothing to do with K_a, K_o or K_p, that can exceed the at-rest value several times over, and that cycles with the seasons rather than settling to a design value. Here is what the code gives you, what it refuses to give you, and what the numbers look like when you try.

1. The code's first answer is: do not backfill with it. IBC 2024 Table 1610.1 tabulates a design lateral soil load for every USCS class it accepts. For CH (inorganic clays of high plasticity), MH (inorganic clayey silts, elastic silts), OL and OH it tabulates nothing — footnote b reads, verbatim, "Unsuitable as backfill material." There is no active value, no at-rest value, and therefore no legal way to design a wall for those materials out of the table. StructSuite does not paper over it: the CH row in Step 2 is greyed out and cannot be picked, and with no class selected the engine stops with "Design lateral soil load is not resolved… No default pressure is assumed."

2. The clays the table does accept still cost you double. C_L (inorganic clays of low to medium plasticity) is tabulated at 60 psf/ft active and 100 psf/ft at-rest — twice and more than three times the imported gravel used here. ML-CL is the same. And both come with the table's blanket footnote a: the values are for moist conditions at optimum density; actual field conditions govern.

3. Cohesive soils have no coefficient of friction. IBC 2024 Table 1806.2 class 5 — "Clay, sandy clay, silty clay, clayey silt, silt and sandy silt (C_L, ML, MH and CH)" — gives no friction coefficient at all. It gives cohesion c = 130 psf instead (footnote b: multiplied by the contact area, as limited by §1806.3.2), with vertical bearing 1,500 psf and lateral bearing 100 psf per foot of depth. In StructSuite you enter μ = 0 and c = 130 psf, and the engine takes c × B capped at one-half the dead load (§1806.3.2).

Put the same finished wall on that soil, keeping the imported gravel backfill, and the arithmetic is brutal:

The wall loses two-thirds of its sliding resistance and half its bearing capacity at once, and it fails both. Swap the gravel for a C_L backfill on top of that and FS_sliding falls to 0.28 with FS_OT at 1.92; take the at-rest condition the way many geotechnical reports require for cohesive backfill and it is 0.17, with a bearing demand of 8,125 psf against a 1,500 psf allowable. There is no reinforcement detail that rescues any of that. On expansive ground the design decision is made in the specification and the drainage, not in the concrete.

4. Give up the passive term first. Passive resistance in front of a toe founded in expansive clay is the least reliable number on the page: shrinkage cracks open a gap at the face in the dry season, exactly when the soil is stiffest and the calculation looks best. Re-run the wall with the passive term neglected and it drops from 0.56 to 0.36. If the toe soil is expansive, neglect it.

5. Swelling pressure is not in the code, and it is not in StructSuite. No table in IBC Chapter 16 or 18 gives a swell pressure — it comes from the geotechnical investigation required under IBC §1803, as a measured value from the specific material at the specific moisture range. If your report states one, enter it as an equivalent fluid pressure in Step 2; that is the only honest way to get it into the model, and the module will tell you it is using your number rather than a table.

6. The seismic screening methods refuse cohesive backfill outright. Ask for the Seed-Whitman increment on a C_L backfill and the engine stops with the scope statement it is quoting: "If soil conditions behind the wall have a cohesive soil component (i.e., a c-φ soil), this simplified approach is no longer appropriate." (FEMA P-750 Part 3, Resource Paper RP 12, p. 356 — the document ASCE 7-22 C11.8 defers to.) For an expansive site in SDC D–F, the seismic earth pressure is the geotechnical engineer's number, full stop.

What to actually do. In order: replace the expansive material inside the active wedge with non-expansive granular fill — the same lever the lever study used, now for a different reason, and the wedge geometry in the drainage detail is what tells you how far back the replacement has to go. Control the moisture: positive surface grading away from the wall, an impermeable cap, no planting beds or irrigation against the stem, and a drain that actually outfalls, because expansive clay damage is a moisture-change problem before it is a pressure problem. Take the report's numbers, not the table's, for the material you cannot replace. And design the base for cohesion, not friction — which on almost every expansive site means a wider footing, a deeper key, or both.

#Sloped backfill — when the ground keeps rising

Everything so far assumed the ground behind the wall is level. Regrade that yard so it keeps climbing — a 3H:1V lawn, a cut at the base of a hillside — and three things change at once, none of them in your favor. StructSuite models all three; this section shows the mechanics on the same finished wall, and what the regrade actually costs.

What the slope does to the statics. The design section for stability is the vertical plane at the heel edge. On level ground that plane runs from the backfill top to the footing bottom: H = 12 + 1.5 = 13.5 ft. On a slope the surface has been climbing since it left the stem, so by the heel edge it stands L_heel·tanβ higher, and the pressure plane is taller:

H′ = H_retained + t_f + L_heel·tanβ = 12 + 1.5 + 4.5 × (1/3) = 15.0 ft

Driving force grows as the square of that height — P_a = ½·w·H′² — so an 11% taller plane is a 23% larger thrust before the pressure itself changes at all. Second, the soil above the heel is no longer a rectangle: the slope adds a triangular wedge — ½·γ·L_heel²·tanβ, here ½ × 120 × 4.5 × 1.5 = 405 plf at ⅔ of the heel from the stem — which resists overturning with the other weights but also loads the heel as a triangular dead load. Third, the pressure coefficient itself rises: Rankine's active coefficient for a sloped surface is larger than the level-ground value the code tables tabulate, and the thrust arrives inclined, parallel to the slope. StructSuite takes the horizontal component and conservatively neglects the slope-parallel vertical component on the resisting side.

Where the pressure number comes from — and why the module refuses the table. IBC 2024 Table 1610.1 and ASCE 7-22 Table 3.2-1 are level-backfill values: their basis is the classical active/at-rest states (Terzaghi and Peck — ASCE 7-22 C3.2.1), and §1610.1 offers them as minimum design loads "unless determined otherwise by a geotechnical investigation in accordance with Section 1803." A rising backslope increases the active coefficient beyond the tabulated state, so the minimum no longer bounds the real load — which is exactly the situation both codes route to the geotechnical investigation. Declare a slope in Step 1 and StructSuite guards the two tables off: the design pressure must be entered as the report's equivalent fluid pressure for the sloped condition. Enter Step 1's β and the geotech EFP in Step 2, and every check downstream — both stability cases, the kern-aware bearing, the heel's wedge terms, the seismic increment height — picks up the sloped geometry automatically. (Sanity-check the report the same way as before, and note that on a slope the answer depends on the friction angle far more than on the level table you are used to. Rankine's sloped-backfill coefficient on 3H:1V is K_A = 0.298 at φ = 36°, 0.270 at φ = 38° and 0.245 at φ = 40°; that coefficient is the resultant parallel to the slope, so at γ = 120 pcf the horizontal pressure the module wants runs 33.9, 30.7 and 27.9 psf/ft — against 30 level. A report giving 30 psf/ft for a sloped yard is therefore neither obviously right nor obviously wrong: it is right for a well-compacted φ = 40° imported fill and light for a φ = 36° one. Ask which φ it assumed rather than which number it printed.)

The same wall, regraded. Take the finished 12-ft wall — unchanged concrete, unchanged bars — and let the grade above it climb at 3H:1V (β = 18.4°), with the geotechnical report giving 36 psf/ft for the sloped condition:

The wall that carried its level yard with a 1.62 sliding margin fails at 1.14 the day the grade changes. And the failure is not mostly the bigger pressure number: run the thought experiment of keeping the old 30 psf/ft on the sloped geometry and FS_sliding is still only 1.36 — the taller plane and the H′² term sink it before the coefficient moves. A regrade behind an existing retaining wall is a structural event, whatever the soils report says.

Fixing it has a trap in it. The two familiar levers interact with the slope in a way level-ground intuition misses: lengthening the heel raises H′, because the slope keeps climbing over every foot you add. Stretch the heel to 6.0 ft with a 24-in key — a combination that rescues far worse level walls — and sliding still fails at 1.45: the rise grew from 1.5 to 2.0 ft and the thrust grew with its square. Push the key to 30 in with the 6-ft heel and sliding clears at 1.56 — but now the heel bars fail: the longer cantilever picks up the wedge's triangle on top of the rectangle, and #6 @ 10 cannot carry it. The combination that closes everything is more surgical: heel 5.5 ft + key 30 in, with the heel bars tightened to #6 @ 8 — FS_sliding = 1.52, FS_OT = 3.04, bearing 2,135 ≤ 3,000 psf, heel D/C = 0.89 (shear-governed at 13.67/15.405; 0.86 in flexure). Keep the original #6 @ 10 and the 5.5 ft heel is 7% over under 1.4D (D/C 1.07): the extra half-foot of cantilever plus the wedge's triangle is more than the bars sized for the level yard can carry. On a slope, the heel's own strength — not stability — becomes the binding constraint, and one more foot of heel buys less resistance than it adds demand.

If the slope is seismic country too. The backfill slope in Step 1 is the one β the module knows: Mononobe-Okabe reads it into K_AE (mind the validity limit φ − ψ > β — a steep slope plus a strong motion has no equilibrium wedge, and the engine stops rather than extrapolating), and Seed-Whitman refuses a slope outright, quoting its source: RP 12's simplified formulation "is not applicable for sloping ground above the wall" (FEMA P-750 Part 3, p. 356). On a sloped site in SDC D–F your choices are full M-O within its limits, or the geotechnical report's seismic pressure — which the code makes the normative source anyway.

What the module does not do. The backslope model is an infinite uniform slope; β at or beyond 45° is refused. A broken slope (rises, then benches level) is conservative to model as infinite; a slope below the wall at the toe, or global stability of the whole hillside — the failure circle that passes under the footing and through the slope — is a geotechnical evaluation under IBC §1807.2.1/§1803, outside any wall-statics module. If the site is a real hillside, that check is not optional, and no cantilever wall design — hand or software — substitutes for it.

#What changes in Seismic Design Category D, E or F

The wall above is complete for an SDC A–C site. Where the structure is assigned to SDC D, E or F and the wall retains more than 6 ft, IBC 2024 §1807.2.2 requires the additional seismic lateral earth pressure, determined per ASCE 7-22 §11.8.3 — and there the geotechnical investigation is the normative source. Everything below is what StructSuite does when the report does not state a pressure and you need a defensible screening value.

One thing anchors the mechanics, and it comes from ASCE 7-22 C11.8.3: the dynamic increment is an earthquake load E superimposed on the static pressure, which remains an H load — so the two carry different factors in every combination below, and mixing them up changes every number. A cantilever reinforced concrete or masonry retaining wall is a yielding wall designed as a simple flexural wall element (§15.6.1).

Where k_h = S_DS/2.5 actually comes from — because it is not a code provision. This is worth stating plainly, since it is the number the whole increment scales with. ASCE 7-22 prescribes no seismic coefficient for retaining walls at all: §11.8.3 requires the geotechnical investigation report to determine the dynamic pressures, and C11.8.3 says only what the load letter is. What the standard does print is S_DS/2.5 under the name "Effective Peak Acceleration" — it is the column head of Chapter 19 Tables 19.3-1, 19.3-2 and 19.3-3, in the soil–structure interaction chapter. Taking that ratio as the site a_max for a yielding wall is engineering practice, not a code mandate, and RP 12 is the document that works in a_max terms: p. 356 gives the full a_max where movement cannot be tolerated, or k_h = a_max/2 after Elms and Martin (1979) "if the wall is allowed to slide up to 10a_max … displacement is in inches." Using the full a_max, as this guide and StructSuite do, is the conservative end of RP 12's own range. Where the geotechnical report states its own k_h, that governs.

#Where the equations actually come from

This is worth knowing before you put a number on a drawing, because ASCE 7-22 prints no pressure equation at all. C11.8 defers to the 2009 NEHRP Recommended Seismic Provisions (FEMA P-750) Part 3, whose Resource Paper RP 12 carries the method. So:

• The Seed-Whitman simplified increment is printed verbatim in RP 12 as Eqs. 6–7 (p. 356) — not as code text. RP 12 restricts it on that same page to dry, cohesionless backfill (a c-φ soil needs the full methods) and to level ground above the wall.
• The 0.6·H application height and the "inverted trapezoidal pressure distribution" are RP 12’s (p. 356) as well.
• S_DS/2.5 appears in ASCE 7-22 as the effective peak acceleration column head of Chapter 19 Tables 19.3-1/2/3 — not as a retaining-wall provision, and not in RP 12, which works in a_max. Using it as k_h is practice; see above.
• The optional wall self-weight inertia term is the Richards & Elms (1979) displacement-based formulation, which RP 12 names (p. 356) without printing its equations.

StructSuite cites each piece to its actual source rather than passing all of it off as code, and it refuses to run the screening methods outside their scope — see Expansive soils for the cohesive-backfill refusal.

#The increment on this wall

Take the same finished wall, unchanged, and put it on an SDC D site with S_DS = 0.50 g, using Seed-Whitman as a yielding wall:

k_h = S_DS/2.5 = 0.50/2.5 = 0.20 g

ΔK_AE ≈ 0.75·k_h = 0.15, so ΔP_AE = ½·ΔK_AE·γ·H_p² = (3/8)·γ·H_p²·k_h = 0.5 × 0.15 × 120 × 13.50² = 1,640 plf, applied at y_E = 0.6·H_p = 8.10 ft above the footing bottom.

That application height is the whole story of why a modest increment hurts: the static resultant sits at H_p/3 = 4.50 ft, the seismic one at 8.10 ft — 80% higher up, on a wall whose failure modes are all moments.

Where the 1.6 / 0.4 ordinates come from. RP 12 prints the shape and the 0.6·H resultant but not the ordinates. For a linear distribution, a centroid at 0.6·H forces p_top = 4·p_bot, and the total area then fixes p_top = 1.6·ΔP_AE/H and p_bot = 0.4·ΔP_AE/H. That is the unique linear distribution satisfying both printed facts, and it is what the module draws.

Two scope notes the module prints with the derivation. k_v = 0 inside the coefficient: vertical shaking enters the strength combinations separately as E_v = 0.2·S_DS·D (§12.4.2.2), so a k_v in K_AE would count it twice. And wall self-weight inertia (k_h × the concrete weights) is an opt-in, left off here — it is small next to a 1,640 plf soil thrust, and peak soil thrust and peak wall inertia are generally not coincident.

#The six checks the increment adds

Three rules decide every one of those numbers, and each of them is a place designs go wrong:

• Stability at 0.7 × nominal earthquake, FS ≥ 1.1. IBC §1807.2.3’s earthquake exception is an additional check, not a replacement: the static FS ≥ 1.5 checks still stand on their own. The static earth pressure and surcharge stay at 1.0 throughout, because they are H, not E. And the resisting weights are not reduced by −E_v — ASCE 7-22 §12.4.2.2 Exception 2(b) permits E_v = 0 at the soil–structure interface of foundations, so base friction keeps the full dead load.
• Bearing gets no safety factor at all. No IBC factor applies, so the module builds the ASD case from §2.4.5 combination 8 with the static H at 1.0. Here E_v = 0.2·S_DS·D acts downward on the dead weights, because for bearing that is the unfavourable direction. The allowable is q_allow × a transient/seismic increase factor taken from the geotechnical report — entered as 1.0 here, meaning no increase at all.
• Strength keeps the increment at 1.0 and the static pressure at 1.6. ΔP_AE is E, so it enters at 1.0; the static earth and surcharge pressures are H and stay at 1.6 (§2.3.6 H rules; ACI 318-25 §5.3.8, Table 5.3.1). E_v = 0.2·S_DS·D = 0.10·D rides on the dead weights, and E_h = ρ·Q_E with ρ = 1.0 (§12.3.4.1 — nonbuilding structures not similar to buildings).

The wall survives — but every margin it had is gone, and the stem becomes the governing member at D/C = 0.98 where it was the most relaxed at 0.63. Push the site harder and it breaks: at S_DS = 0.60 g the stem fails (D/C = 1.05), and by S_DS = 0.70 g seismic sliding fails too (FS = 1.06 < 1.1). A wall proportioned for static loads has a real but finite seismic budget, and on this wall the first thing to buy is stem steel, not base width.

#Choosing a ΔP_AE method

In order of preference: the geotechnical report’s pressure, whenever it states one — §11.8.3 makes it normative for SDC D–F, and StructSuite accepts either a resultant with its application height or a uniform pressure. Seed-Whitman (used above) is the low-input screening method for level cohesionless backfill: ΔK_AE ≈ 0.75·k_h, no soil angles needed. Mononobe-Okabe computes the full K_AE from φ, δ and β — reach for it when the backfill slopes or when you have documented soil parameters, and mind its validity limit: when φ − ψ ≤ β no equilibrium wedge exists, and the module stops with an error rather than quietly clamping the coefficient. The FAQ below has the full comparison and the scope limits RP 12 puts on both.

#What this wall is sensitive to

Design margins are only meaningful next to the assumptions they rest on. Each row below is the finished wall re-run with one input changed:

Read that table as a list of what your drawings and specifications have to protect. Four of the six failure rows are not engineering errors at all — they are things other people do to the wall after you design it: dig at the toe, restrain the top, or let it flood. Those belong on the drawings as notes, not in your files as assumptions.

The one favourable row is worth noticing too: reclassifying the surcharge as permanent improves every number, because its weight is then allowed onto the resisting side. That is a judgment about the load, not a calculation trick — a parked-car allowance is variable, an adjacent building footing is not — and it is exactly the kind of question StructSuite asks rather than assuming.

#Summary of results — 12 ft cantilever wall

The design is sliding-governed at D/C = 0.93 — the classic cantilever wall outcome, and the reason this guide spent its first half on stability. The concrete members all sit between 11% and 68%, because their sizes were set by stability, cover and development long before strength entered the conversation.

Final schedule, per foot of wall: B = 8.5 ft (toe 3.0, heel 4.5), 18 in footing, 12 in stem × 12 ft, 12 × 12 in key. Stem verticals #6 @ 8 earth face with dowels hooked into the footing and a 3 ft 7 in lap; stem horizontals #4 @ 9; heel top #6 @ 10; toe bottom #5 @ 9; footing longitudinal #5 @ 9; key (4) #5 continuous with the stem pattern continued and hooked. Imported GW backfill filling the active wedge, with drainage per §1805.4.2/.3.

#Verification — hand calculation against StructSuite

Software that agrees with itself proves nothing. Every number on this page has therefore been recomputed by hand, long-hand, from the code equations, and checked against what the module reports. The hand column below is not a rearrangement of the software's output: it is the substituted arithmetic, written out, and it is locked in an automated test (tests/verification/retaining-wall-design-guide-handcheck.test.ts) that deliberately shares no code with the calculation engine. If a future change ever made the guide and the engine agree on a wrong number, that test is what still fails.

#Loads and stability

#Concrete strength — ACI 318-25

#Development — ACI 318-25 Chapter 25

#Seismic increment at S_DS = 0.50 g

Every quantity agrees exactly at the precision the module reports, and where StructSuite exposes a full-precision value the two match to twelve significant figures. Two things are worth saying plainly about what that does and does not prove. It proves the arithmetic — that the module substitutes into the equations it says it does. It does not relieve you of judgment: the equations were chosen by an engineer (active versus at-rest, whether to count passive resistance, whether the surcharge is variable or permanent), and a correct calculation of the wrong load case is still the wrong answer. That is why this guide spends more words on the input parameters and the traps than on the algebra.

#What this module assumes — and what it will not design for you

Every number on this page comes out of one structural model, and the model is worth stating plainly, because most of the ways a retaining wall calculation goes wrong are not arithmetic. They are a wall being pushed through a model that was never meant for it.

#The model

A concrete stem, or a solid-grouted CMU one. The stem is reinforced concrete designed by ACI 318-25, or — since the 2026-10-10 revision — a solid-grouted, running-bond concrete masonry wall designed by TMS 402-22 Chapter 8 allowable stress design (the CMU section); the footing is reinforced concrete either way. Partially grouted or hollow masonry, clay masonry and masonry not laid in running bond are not modelled.

A cantilever. The stem is fixed at the footing and free at the top. Its moment peaks at the stem–footing interface, it carries flexural steel on the earth face only, and the wall is free to lean — which is what earns the active pressure state and what makes overturning, sliding and bearing the questions that govern.

One stem thickness, or one step. For a uniform stem the critical sections for flexure and shear sit at the interface (ACI 318-25 §13.3.6.3). A stem stepped once — thinner above a height, lighter bars above it, or both — gets a second critical section at the step and the bar-cutoff rules of §7.7.3 (the stepped-stem section). A tapered stem is a different section at every height and is not modelled.

Lateral pressure — and, if you enter it, axial load on top of the stem. A house wall, a beam reaction or a deck ledger landing on the stem is carried, as four separate load letters (D, L, L_r, S) with an optional eccentricity. Leave them at zero and the wall is designed exactly as it would have been without the feature. What happens when you leave the axial load out works through which checks that helps and which it hurts.

A per-foot strip. Everything is a one-foot slice: no corners, no returns, no counterforts, and no distribution of load along the wall to a stiffer element.

#The three checks that assume the wall can move

Overturning, sliding and bearing all ask the same underlying question — can this wall go anywhere? Which means they only mean something for a wall that could. Two conditions quietly take them off the table:

The first is the one that catches people, and it has a code test rather than a judgment call.

#At-rest or active — and why a floor is not automatically restraint

IBC 2024 §1610.1 and ASCE 7-22 §3.2.1 settle it in nearly the same sentence: a wall whose horizontal movement is restricted at the top shall be designed for at-rest pressure, while a wall free to move and rotate at the top is permitted active. Note which verb goes where — active is a permission you earn by establishing the movement, and the burden of proof sits on that side.

But a floor bearing on top of a below-grade wall does not automatically restrict it. ASCE 7-22 §3.2.1 carries an exception: a foundation wall extending not more than 8 ft below grade and laterally supported at the top by a flexible diaphragm is permitted active pressure. Its commentary C3.2.1 names what counts as flexible — wood-framed floors and cold-formed steel joists, without a cast-in-place concrete floor attached — and gives the mechanics: restraint at the top pushes the pressure up from active toward at-rest, and a flexible floor does not supply it.

Two things follow, and they pull in opposite directions:

• A shallow basement or crawl-space wall under a wood-framed floor can often still be designed for active pressure, on the code's own exception.
• A deep one cannot. A basement wall for a multi-storey building is normally well past 8 ft below grade, so the exception lapses and at-rest is required — and that wall is genuinely held top and bottom, which means it is no longer the cantilever this module solves. Its moment redistributes between a span moment and the base, depending on how much rotational fixity the base really has; it delivers a reaction into the diaphragm that has to be designed for; and the three stability checks above stop governing, because a wall held at both ends cannot overturn or slide.

The module says this rather than leaving it to be discovered: selecting At-rest adds a note to the Step 4 design notes, and the code text itself sits behind the book icon on the Earth pressure condition heading in Step 2.

#What the geotechnical report can take away

The report does not only supply values — it restricts what the base may rely on, and those restrictions belong in the model rather than in a fudged input. What the geotechnical report can take back works three of the common ones through on the finished wall, including both readings of a disregarded depth of toe soil — excavated or left in place — which the module asks you to choose between. Every one of them fails it except the in-place reading.

#What happens when you leave the axial load out

Sooner or later a retaining wall stops being free-standing. A house wall lands on the stem, or a deck ledger, or a beam pocket — and the question is whether that load is worth entering, or whether leaving it out is a harmless simplification.

It is not one answer. The load pushes four different checks in three different directions, and knowing which is which is the difference between an honest simplification and a wall that is quietly under-designed.

Take the finished wall from this guide — 12 ft stem, 12 in thick, 3 ft toe, 4.5 ft heel, 18 in footing with a 12 in key — and stand a two-storey wood-framed house wall on it: D = 1,800 plf, L = 800 plf, S = 600 plf, centred on the stem. Every number below is the module's own output, with the load in and with it out.

Read the table as three findings.

The stability checks get better, so ignoring the load is safe there. The dead part of an axial load is a permanent weight sitting almost directly over the toe. Its moment about the toe resists, and its weight grows the base friction and the §1806.3.2 cohesion cap. Both safety factors rise. A wall you checked without the load and passed will still pass with it.

That cuts the other way too, and it is worth knowing: a wall that fails sliding by a little without the load may pass with it. Trim this wall to a 10 ft stem on a 2 ft toe and 3.5 ft heel with no key, and its sliding factor goes from 1.42 — a fail — to 1.69, a pass, on the strength of the house wall alone. If your wall is marginal in sliding and something real bears on the stem, entering it is not cheating. It is the actual load path.

The footing checks get worse, and that is the one that matters. Bearing pressure rises 31% and the toe demand 30%, because the axial load adds to ΣV at an arm near the toe — it both increases the total and pulls the resultant forward. Nothing about the stem reinforcement protects against this. A wall sized for 3,000 psf allowable at 1,576 psf looks comfortable; at 2,069 psf it is still fine, but a wall that was already at 2,400 psf would not be. This is the check that makes entering the load non-optional.

On the stem itself it very nearly cancels, which is why it is so easy to miss. Axial compression raises the moment strength of the section — 26.66 to 27.80 kip-ft/ft here, because compression delays the concrete from being the limit. But the same load riding on a wall that deflects adds a second-order P-Δ moment, +3.5%. The two effects land within a percent of each other and the demand-capacity ratio barely moves: 0.626 to 0.622.

So the practical rule is short: ignore the axial load if you like and your stem steel is still right, but your footing is not.

Why the stem is checked four times, and why 0.9D governs

A load on the stem does not have one value. ASCE 7-22 gives it several, and they disagree about which is worst — because the axial load helps the capacity and hurts the demand at the same time.

The module runs all of them. On this wall the four strength combinations come out:

0.9D governs — the combination with the least axial load on the stem. That is not an accident. Since axial compression is what raises φM_n, the worst case for the stem is the one where the least of it is present, and 0.9D is exactly the combination ASCE 7-22 writes for "the dead load might be lighter than you think, and nothing variable is acting." A designer who checked only 1.2D + 1.6L would credit a capacity the wall is not guaranteed to have.

Slenderness: the check you cannot skip, and the answer you can usually ignore

Put an axial load on a stem and it stops being a slab in bending. ACI 318-25 §11.5.1.1 requires the axial–moment interaction to be considered, and §11.5.2.1 sends the section to §22.4 for it. The simplified wall method of §11.5.3 is not available — it needs the resultant of all factored loads inside the middle third of the thickness, and on this wall the total eccentricity is 64.0 in against an h/6 of 2.00 in

Slenderness cannot be waived either. §6.2.5.1(a) lets you neglect it when k·ℓ_u/r ≤ 22 for a member not braced against sidesway. For this stem, with k = 2.0 from Table 11.5.3.2 (walls not braced against lateral translation) and r = 0.30h = 3.60 in per §6.2.5.2(b):

Past the limit, comfortably — and it will be for essentially any retaining wall stem, because the ratio is fixed by the wall's proportions, not by how much load you put on it. So the second-order moment gets computed rather than waived.

And then the answer is small. The magnifier of §6.6.4.5.2 comes out between 1.035 and 1.055 here, because a stem stiff enough to carry the earth pressure is already far stiffer than a 5 klf axial load can trouble. The critical buckling load is 128 to 147 klf against a P_u of 3 to 6.

That combination — formally required, numerically negligible — is exactly the kind of thing hand calculations skip and then cannot defend at plan check. The module does the arithmetic and prints it, so the answer to "did you consider slenderness?" is a number rather than a shrug.

There is a limit on how far that can be pushed. ACI 318-25 §6.2.5.3 caps the magnified moment at 1.4 times the first-order moment. Past that the section is simply too slender for the load it is being asked to carry, and the module says so rather than reporting a magnifier of 2.

The eccentricity, and the trap in crediting it

If the load does not land on the stem's centreline, enter the eccentricity. It does two things: it shifts the arm about the toe, and it plants a constant P·e moment down the whole stem.

Which way matters. Toward the heel the moment opposes the earth pressure and relieves the stem. Toward the toe it adds. The relief is real, but it is only there when the load is — so the module credits it per combination, and 0.9D, carrying the least axial load, keeps the least of it. That is the same logic that makes 0.9D govern above, applied to the eccentricity.

Push a relieving eccentricity far enough and the stem bends the other way: tension moves to the toe face, where a cantilever retaining wall carries no flexural steel at all. The module flags that rather than quietly reporting a small moment.

Which letter you enter is not cosmetic

The load is entered as four separate letters — D, L, L_r, S — because ASCE 7-22 factors them separately, and two of the differences are easy to get wrong:

• Only D may resist. §2.3.1 and §2.4.1 both open with "effects of one or more loads not acting shall be considered." Live, roof live and snow can all be absent, and every one of them makes this wall more stable — so none of them is allowed on the resisting side of overturning or sliding. They appear only where they raise a demand.
• Roof load reaches factors live load never does. Combination 3a — 1.2D + (1.6L_r or 1.0S) + (L or 0.5W) — puts the roof load at its own full value with L absent. In 2a the same load is only a 0.3S or 0.5L_r companion. A snow-governed wall reaches its worst bearing in 3a, and a designer who lumped snow in with floor live load would never generate that case. The module builds 3a whenever a roof load is present, and skips it entirely when there is not one.

#Outside the scope entirely

Global stability. A deep-seated failure surface passing beneath the footing and out behind the wall is a slope-stability problem, not a wall problem, and no amount of reinforcement in the stem answers it. It is a geotechnical evaluation — trap 5 below has the detail.

Wall movement. The active state assumed here needs the wall to rotate: a displacement at the top of roughly 0.002 times the wall height is typically enough to develop it. The module does not compute that rotation, nor the tilt from soil compressing under the toe, and neither is checked against whatever a finish, a fence or a slab on top might tolerate.

Loads that are not modelled. A footing behind the wall is modelled, as the strip load it is — its own section — but a footing bearing beside or below this wall's footing is a foundation-interaction problem the module refuses. A surcharge on the toe side is omitted; that omission is conservative, since both its weight and the passive pressure it adds would resist. Seismic on a non-yielding wall, hydrodynamic pressure and the vertical inertia of the soil wedge are not carried either — the seismic section says which, and why.

None of it is hidden. Every one of these appears as a note on the calculation itself, beside the numbers it qualifies. If the wall you are designing sits outside the model, the module says so rather than quietly handing you a number.

#Six traps that sink retaining wall designs

1. Counting the surcharge as resistance. It sits on the heel, so it is tempting to let its weight help. IBC §1807.2.3 forbids it: variable loads are investigated set to zero on the resisting side. Letting its weight resist here — which is what reclassifying it as permanent legitimately does — moves the governing overturning FS from 3.84 to 4.03 and the sliding FS from 1.62 to 1.67. Small, but on a wall that clears 1.5 by only 8% that is exactly the size of error that turns a failing design into a passing one on paper. The question to answer is not "how big is the surcharge" but "is it ever absent" — a parked-car allowance is; an adjacent building footing is not.

2. Taking passive resistance the soil cannot promise. Passive pressure requires the soil in front of the toe to be there, compacted and undisturbed, for the life of the wall. This design needs its passive term — neglect it and the sliding FS collapses from 1.62 to 1.22. If a future excavation at the toe is plausible, either neglect the passive term now and widen the footing, or protect that soil with a note on the drawings and mean it.

3. Forgetting the water. Table 1610.1 values are for moist, drained soil. Undrained backfill means full hydrostatic pressure (§1610.1) plus buoyant soil weights and base uplift, and on this wall that is FS_sliding = 0.52 with the water at the surface — and already 1.43, below the code's 1.5, with the water only 3 ft up the wall (the drainage section's table). The gravel, the fabric and the pipe are not decorations; §1805.4.2/.3 is what makes the 30 psf/ft legal.

4. Moving the heel's shear check to d. §7.4.3.2 allows that only where the reaction puts the end region in compression. Under the toe it does; at the heel–stem joint the top is in tension, so the shear stays at the face (§13.2.7.2). The relief you are not entitled to is worth 29% of the heel shear on this wall (10.489 klf at the face against 7.46 klf at d), on the member that is usually working hardest.

5. Designing the wall and forgetting the ground it sits on. Overturning, sliding and bearing are structural checks on a wall. Global stability — a slip surface passing beneath the whole wall — is a geotechnical analysis, and no clause in ACI 318 or IBC Chapter 18 covers it. On a slope, above a descending grade, or over soft or layered soils, ask for it by name.

6. Helping the seismic case with factors it has not earned. Two show up constantly. The one-third bearing increase belongs to the IBC’s alternative basic load combinations; for the seismic ASD check the allowable is q_allow × a factor from the geotechnical report, default 1.0, and StructSuite never assumes one. And a k_v inside the pressure coefficient on top of E_v = 0.2·S_DS·D in the load combinations counts vertical shaking twice. Behind both sits the asymmetry worth memorising: ΔP_AE is an E load (1.0 for strength, 0.7 for stability and ASD) while the static earth pressure stays H (1.6 for strength, 1.0 for stability).

#Frequently Asked Questions

#When can I use active earth pressure instead of at-rest?

When the wall is free to move and rotate at the top — IBC 2024 §1610.1 and ASCE 7-22 §3.2.1 both say so in nearly identical words. A free-standing cantilever retaining wall qualifies; a displacement at the top of about 0.002·H (0.29 in on a 12 ft wall) is typically enough to develop the minimum active state (FEMA P-750 Part 3, RP 12, p. 355). A basement wall braced by a floor diaphragm, a wall tied into a building, or a wall with a rigid return at both ends does not qualify and must be designed for at-rest pressure. For this backfill that is 60 psf/ft instead of 30 — and it takes this wall's sliding safety factor from 1.62 to 0.81. Note which verb each state gets: at-rest is what you shall design for, active is what you are permitted to design for, so the burden of proof sits on the active side. Cantilever or restrained works the whole decision through, including the case where the wall really is propped and the prop is the thing nobody designed.

One exception is worth knowing before you concede at-rest on a basement wall. ASCE 7-22 §3.2.1 permits active pressure for a foundation wall extending not more than 8 ft below grade that is laterally supported at the top by a flexible diaphragm — and its commentary C3.2.1 names wood-framed floors and cold-formed steel joists, without a cast-in-place concrete floor attached, as exactly that. The reasoning is the mechanics: restraint at the top pushes the pressure up from active toward at-rest, and a flexible floor does not provide enough of it. So a shallow crawl-space or basement wall under a wood-framed floor often still qualifies for active. A deep basement wall does not, on two counts at once — it is past the 8 ft limit, and it is genuinely held top and bottom, which makes it a propped wall rather than the cantilever this module designs.

#Is a restrained wall always heavier than a cantilever? Doesn't at-rest pressure double everything?

No — and the belief that it does is what makes people oversize restrained walls while missing the thing that actually matters about them.

At-rest pressure really is about twice active for a clean gravel (IBC 2024 Table 1610.1 gives GW 30 psf/ft active, 60 at-rest). But a wall is only restrained because something holds its top, and that same something turns it from a cantilever into a beam spanning between two supports. The pressure goes up; the span goes down. For a triangular pressure the span wins:

A genuinely propped wall ends up 20–23% below the free cantilever's active moment, even carrying double the pressure. The arithmetic behind it is just the two beam coefficients: qH²/6 for a cantilever against qH²/15 for a propped one — a factor of 2.5 that more than swallows the factor of 2 on the load.

The row that is genuinely worse is the second one: at-rest pressure on a cantilever load path. That is what you get if you select "At-rest" in a module that only knows how to build cantilevers. It pays the full pressure penalty and collects none of the span relief, so it is conservative — expensively so — but it is not the wall you are building, and its conservatism hides two things that matter more.

What a restrained wall actually costs you is not concrete, it is two details:

1. A reaction nobody designed. 864 to 1,440 lb per foot goes into whatever holds the top. A cantilever model computes zero there, so the number never reaches the drawings. A slab edge or a rim joist quietly carrying 1.4 klf is a worse outcome than a slightly thin wall.
2. Steel on the wrong face. A propped wall puts tension on the exposed face over most of its height, and a pinned-base one puts its only peak there. A cantilever detail has earth-face bars and nothing else.

So the honest summary: restraint does not make the wall heavier, it makes it different. Cantilever or restrained works the whole comparison through with diagrams.

#If the base fixity barely changes the moment, why does it matter?

Because it changes where the moment is, and where a reinforced concrete wall wants its steel is not a detail you can average.

Propped at the top, the same wall gives:

• Fixed base — peak 0.0667 qH² at the footing, earth face; a smaller 0.0298 qH² sag at 0.45H from the top on the exposed face
• Pinned base — no base moment at all, and a single peak of 0.0642 qH² at 0.58H from the top, exposed face

The two peaks are within 4% of each other, so you would size almost the same bar. They sit half a wall apart, on opposite faces. Choose wrong and you have bought the right steel and detailed it into the wrong place — the failure mode is not a thin wall, it is a wall with a plain section exactly where the moment is.

Fixity also drives the prop force: qH/10 for a fixed base against qH/6 for a pinned one, a 67% difference in what the diaphragm has to take.

A real base is somewhere between the two — a stem cast into a spread footing bearing on soil is not a textbook fixed end, while a wall on a thick pile cap is close to one. There is no useful middle answer, though, because the two cases put the steel in different places and a half-fixed result would satisfy neither. Pick the fixity deliberately, say so on the drawings, and detail the case you picked.

#A house wall bears on top of my retaining wall. Do I have to enter that load?

For the footing, yes. For the stem, it barely matters — and knowing which is which saves you from both the wasted effort and the mistake.

An axial load on the stem pushes different checks in different directions:

On the wall worked through in this guide, a two-storey house wall (D = 1800, L = 800, S = 600 plf) leaves the stem demand-capacity ratio essentially unchanged — 0.626 to 0.622 — while lifting soil bearing 31% and the toe demand 30%. What happens when you leave the axial load out has the full comparison and the reasoning.

So: if you are sizing stem bars, the axial load is noise. If you are sizing the footing, it is not optional — and there is no warning built into a stem-only check that would tell you the footing is short.

Two details that catch people when they do enter it:

• Enter the letters separately. Only the dead part may be counted where the load resists — "effects of one or more loads not acting shall be considered" (ASCE 7-22 §2.3.1, §2.4.1), and live, roof live and snow can all be absent. Lumping them into one number would credit stability the wall is not guaranteed to have.
• Snow and roof live are not live load. Combination 3a — 1.2D + (1.6L_r or 1.0S) + (L or 0.5W) — gives the roof load its own full factor with L absent. A snow-governed wall reaches its worst bearing there, in a case that never appears if the snow was entered as L.

#A house footing sits a few feet behind my wall. Is a uniform surcharge enough?

No — and the reason is where the push acts, not how big it is. A uniform surcharge spreads into a flat K·q band whose resultant sits at mid-height; a footing's strip load induces a pressure that peaks a short way below its bearing level and tails off, so its resultant sits high on the wall. On the final wall of this guide, a 3,000 plf footing 6 ft behind the stem puts 942 lb on the heel-edge plane at 11.5 ft above the base and takes the stem from 63% to 93% of capacity; a 250 psf surcharge used as the stand-in finds the sliding failure but leaves the stem at 79% and the toe at 37% when it is at 67%. A footing behind the wall has the whole table.

Enter it as what it is: Adjacent footing behind the wall in Step 2 — load per foot of wall, width, distance to the centreline, bearing depth. The module applies the elastic strip-load solution, adds the push to the soil load as H (IBC 2024 §1610.1; ASCE 7-22 §3.2.1, §2.2), doubles it where the wall is designed for at-rest pressure, and, where part of the footing bears over the heel, puts that weight on the heel as dead or live load per your choice. A footing on the stem itself is an axial load on the stem — enter it as one.

#There is a fence or guardrail on top of my wall. Do I make the stem taller?

No — that is the one thing not to do. A taller stem is more concrete, and the module would credit its weight to overturning, sliding and bearing; a 6 ft fence modelled as 6 ft of 10 in stem borrows 750 lb per foot of wall that is not there, and the overturning factor it reports is 7.32 where the real wall's is 5.58 (the comparison). Enter the fence as what it is: its height in Step 1 under Fence or guardrail on the stem top, and in Step 2 the net wind pressure over that height as Wind on the fence and its weight per foot as Fence weight — left blank for an open rail. The wind lands at the fence mid-height above the whole stem, goes through the stem at 1.0W and the stability checks at 0.6W in both directions, and the reversal check runs on it; the weight, where there is any, is permanent weight on the wall. A concrete parapet above the soil is still stem height — that one is concrete.

#Can the stem be concrete block (CMU) instead of poured concrete?

Yes. Pick Concrete masonry (CMU) in Step 1, choose the block (6, 8, 10 or 12 in nominal), and in Step 3 enter f′_m, the bar grade, the unit density, the bar position in the cell and the Seismic Design Category. The stem is then designed by TMS 402-22 Chapter 8 (allowable stress design) — masonry stress ≤ 0.45 f′_m and steel stress ≤ F_s on the cracked section with the axial load, P ≤ P_a, f_v ≤ F_v, the compression width of §5.1.2.1, the dowel lap of Eq. 6-2 and the Chapter 7 minimums by SDC — while the footing stays ACI 318-25. The CMU section runs the sample site through seven bar arrangements and shows why the bar's position in the cell, not its size, is what makes a thin block wall pass its earthquake case. Solid grouting and running bond are assumed; a partially grouted wall is not modelled.

#Can I stop half the stem bars partway up, or thin the stem above a height?

Yes, and the module checks it where it happens. Enter the step height and the thickness above it in Step 1 and the continuing bar pattern in Step 3 (the same size, every bar or every second or third bar). ACI 318-25 §13.3.6.3 requires a stem of varied thickness to be investigated throughout its height, so the upper section gets its own critical section at the step with every load above it; the bars that stop are held to §7.7.3.3 (run the greater of d and 12d_b past the point they are no longer required — a point the module solves on the moment envelope, not a chart), §7.7.3.4 and §7.7.3.5. A stepped stem runs the final wall through seven variants, including one that has strength to spare at the step and still fails because the cutoff sits below the extension rule's line. The earth face stays plane (the step is on the toe face); a taper, an earth-face step, and a separate bar set lapped at the step are not modelled.

#Does a retaining wall stem need a slenderness check?

Once anything bears on it, yes — and it will fail the screening test every time.

ACI 318-25 §6.2.5.1(a) lets you neglect slenderness when k·ℓ_u/r ≤ 22 for a member not braced against sidesway. A cantilever stem is not braced, so k = 2.0 (Table 11.5.3.2, walls not braced against lateral translation), and r = 0.30h (§6.2.5.2(b)). For a 12 in stem 12 ft tall that is 2.0 × 144 / 3.60 = 80. The ratio is set by the wall's proportions, not by the load, so no ordinary retaining wall stem passes it.

The second-order moment then has to be computed. The good news is that the answer is small: on this guide's wall the §6.6.4.5.2 magnifier lands between 1.035 and 1.055, because a stem stiff enough to hold back the soil is already far stiffer than a few kips per foot of axial load can trouble.

That is the awkward combination — formally required, numerically negligible — which is exactly why it gets skipped by hand and then cannot be defended when a reviewer asks. ACI 318-25 §6.2.5.3 also caps the magnified moment at 1.4 times the first-order moment; past that the stem is genuinely too slender for the load, and thickening it is the answer rather than a bigger magnifier.

A stem with no axial load on it needs none of this. With no P there is no P-Δ, and the section is a one-way slab in pure flexure per §13.3.6.1.

#What safety factors does the code require, and why don't the load combinations apply?

IBC 2024 §1807.2.3 requires FS ≥ 1.5 against both overturning and sliding, checked at 1.0 × nominal loads, and it explicitly displaces the §1605 load combinations for that check. Variable loads are investigated set to zero, which is why a surcharge pushes but never holds down. Where earthquake loads are included, they enter at 0.7 × nominal and the minimum factor drops to 1.1 — an additional check, not a replacement: the static 1.5 checks still stand on their own. The factored combinations (1.2D + 1.6H + 1.6L, 0.9D + 1.6H) come back for the concrete strength design (ACI 318-25 §5.3.8).

#Can I count on passive resistance in front of the toe?

Only if you can guarantee the soil. Passive resistance is legitimate under IBC §1806.3.1 and may be added to base friction, but it needs full mobilisation — a movement of 2–5% of the embedded depth, far more than active pressure needs — and it needs the soil to be permanent. Many engineers neglect it entirely where a future utility trench or landscaping cut at the toe is plausible. Check what your wall does without it: on the wall in this guide, neglecting passive takes FS_sliding from 1.62 to 1.22, so the passive term is load-bearing in the design and has to be protected by a drawing note.

#The report says to disregard the top foot of soil in front of the wall. Is that soil excavated or left in place?

Ask what the report is protecting against, because the two readings give different numbers and the module asks which one in Step 2 (Excavated is preset). If the soil may be removed — utility trenches, landscaping cuts, a future driveway — it is excavated: the ground surface drops by that foot and the passive pressure restarts from zero there, P_p = ½·w_p·(D − n)². If it stays but is loose, uncompacted or simply not to be relied on, it is left in place: only its own slice of the pressure diagram is forfeited, while its weight still presses on the soil below, so the diagram keeps its origin at the surface and P_p = ½·w_p·(D² − n²). Excavated is always the lower number — on the wall in this guide 0.625 against 1.125 klf, FS 1.42 against 1.59 — so it is the safe side when the report is silent. Neither reading changes the weight of the toe soil, and under both the shear key is detailed for the pressure that is actually there. What the geotechnical report can take back runs the numbers.

#Should I develop the bars with Table 25.4.2.3 or with the general equation?

Either is code (ACI 318-25 §25.4.2.1(a)); they differ in how much they know about your wall. The simplified table is the general equation with a preselected confinement term (R25.4.2.1), so it is always at least as long. Eq. (25.4.2.4a) uses the actual cover and bar spacing — c_b is the lesser of the cover to the bar centre and half the spacing — capped at (c_b + K_tr)/d_b = 2.5, with K_tr = 0, which §25.4.2.4 permits even where ties exist, and slabs and walls have none anyway. On ordinary footing mats and wall bars the cap is reached and the equation comes out at 60% of the table: on this wall the dowel lap drops from 42.7 to 25.6 in and the toe bars from 27.4 to 16.4 in. The hooked-dowel length ℓ_dh is the same under both. The module runs the table unless you choose the equation in Step 3, and the printed package names which one it used. The general equation has the full table for this wall.

#Do I need a shear key, and how deep should it be?

Only if sliding needs it. A key is the cheapest way to buy sliding resistance because passive pressure grows with the square of depth: on the wall here, a 12 × 12 in key takes the passive depth from 2.50 to 3.50 ft and nearly doubles P_p, for about 0.04 cy of concrete per foot of wall. Put the key under the stem, aligned with the stem's front face, so its dowels are the stem bars carried through — the module lets you offset it toward the heel if a detail needs that, at the cost of separate key dowels — and remember that whatever resistance it mobilises, it must also carry back into the footing as a short cantilever (Step 9). Make it roughly as wide as it is deep; a thin deep key is a bending problem.

#How thick should the stem and footing be?

Start at H/12 to H/10 for both, then check. For this 13.5 ft overall height that gives 13.5–16 in, and the wall was built with a 12 in stem and an 18 in footing — the stem thinner than the rule because it comes out at only 63% in flexure, the footing at the top of the range because heel shear is what it has to satisfy without stirrups. Practical floors matter as much as the rule: below about 10 in you cannot place a #6 bar with 2 in of cover on the earth face and still get a vibrator down the form, and below about 12 in of footing the stem dowels struggle to develop their hooks.

#What if the bearing resultant falls outside the middle third?

The heel lifts, the base loses contact, and the pressure becomes a triangle over 3x̄ with a much higher peak. The code does not forbid it — the trial wall in this guide still passed on q_max — but treat it as a design failure anyway. Partial contact makes the peak pressure very sensitive to small errors, concentrates settlement under the toe so the wall tilts, and gets dramatically worse at strength level: the trial wall bore on 1.54 ft of its 7.5 ft base under 0.9D + 1.6H. The fix is almost always a longer heel.

#Do I need to check global (slope) stability?

If the wall is on or near a slope, retains fill, or sits over soft or layered ground: yes, and it is not a structural calculation. Overturning, sliding and bearing all assume the block of soil under the footing stays put. A deep-seated slip surface passing beneath the entire wall is analysed by a geotechnical engineer using limit-equilibrium methods, and it frequently governs where the structural checks are comfortable. Ask for it explicitly in the geotechnical scope; neither ACI 318-25 nor IBC Chapter 18 covers it, and neither does StructSuite.

#When is the seismic earth pressure increment required?

When the structure is assigned to Seismic Design Category D, E or F and the wall supports more than 6 ft of backfill (IBC 2024 §1807.2.2). Then the geotechnical investigation is the normative source for the pressure (ASCE 7-22 §11.8.3). Below that threshold — as on this SDC B site — no increment is required. The section above re-runs this identical wall at S_DS = 0.50 g so you can see the cost: the stem goes from D/C 0.63 to 0.98 and sliding from FS 1.62 to 1.18.

#What drainage does the code actually require?

IBC 2024 §1610.1 sets the terms: the tabulated lateral soil loads are for moist conditions, and submerged or saturated backfill must be designed for the buoyant soil weight plus full hydrostatic pressure. A drainage system installed per §1805.4.2 and §1805.4.3 is what lets you use the drained values. Practically that means free-draining material against the wall, filter fabric between it and the native soil, a perforated collector pipe sloped to a real outfall, and surface grading or a cap that keeps rain out of the drain. On the wall in this guide, losing the drainage takes the sliding safety factor from 1.62 to 0.52 with the water at the surface — and below 1.5 as soon as it stands 3 ft above the footing bottom. The module models that case directly: enter the design groundwater level and γ_sat in Step 2 (the drainage section's table shows the whole climb).

#Why is the heel's shear checked at the face but the toe's at d from the face?

ACI 318-25 §7.4.3.2 permits moving the shear check to d only when the support reaction introduces compression into the end region. Under the toe, upward bearing pressure does exactly that, so the toe gets the relief (§13.2.7.2). At the heel–stem joint the top fibre is in tension, so it does not, and the shear stays at the face. The stem follows the same logic at its own joint: §13.3.6.3 puts both moment and shear at the stem–footing interface for a uniform-thickness stem, because that joint opens under lateral load.

#The IBC and ASCE 7-22 lateral soil load tables disagree — which one governs?

Both are legal minimums inside their own documents, and they genuinely differ. Clayey sand (SC) is 60 psf/ft active in IBC Table 1610.1 and 85 psf/ft in ASCE 7-22 Table 3.2-1; well-graded gravel (GW) is 30 versus 35. Where the IBC is the adopted code it governs, but a geotechnical report prepared under IBC §1803 supersedes either table. StructSuite carries both tables verbatim, never unified, and records which one you selected — because a plan checker will ask.

#Can the shear key be plain (unreinforced) concrete?

Not on the strength ACI gives it. §14.5.2.1 permits only 5λ√f′_c on the gross section at φ = 0.60 for plain concrete in flexure, which an ordinary key does not satisfy against its own junction moment. Reinforce it — the simplest detail continues the stem vertical pattern down into the key with a standard hook developed in the footing, and adds shrinkage-and-temperature bars running along the wall inside the key.

#Which ΔP_AE method should I pick?

The geotechnical report’s pressure, whenever the report states one — ASCE 7-22 §11.8.3 makes it the normative source for SDC D–F, and StructSuite accepts either a resultant with its application height or a uniform pressure. Seed-Whitman is the low-input screening method for level cohesionless backfill: ΔK_AE ≈ 0.75·k_h, no soil angles required. Mononobe-Okabe computes the full K_AE from φ, δ and β — use it for sloped backfill or documented soil parameters, and mind its validity limit: when φ − ψ ≤ β no equilibrium wedge exists and the module stops with an error rather than clamping the coefficient.

Both computed methods come from Resource Paper RP 12 of the 2009 NEHRP Recommended Seismic Provisions (FEMA P-750) Part 3 — the document ASCE 7-22 C11.8 defers to — not from code text. Seed-Whitman is printed there verbatim (Eqs. 6–7, p. 356); for Mononobe-Okabe, RP 12 prints the thrust equation P_AE = ½·γ·H²·(1 − k_v)·K_AE (Eq. 4, p. 355) and defers the closed-form K_AE to the soil-dynamics literature it lists (Prakash 1981; Das 1983; Kramer 1996; Ebeling and Morrison 1992).

RP 12 also scopes both computed methods to dry, cohesionless backfill with groundwater below the wall base (pp. 355–356, 359). StructSuite enforces that rather than leaving it to you: Seed-Whitman is blocked for clay-bearing USCS classes (GC, SC, SM-SC, C_L, CH), fine-grained or unknown classes get an advisory, Mononobe-Okabe on a clay class gets an advisory, and a wall retaining more than 15 ft posts a special-studies note (p. 356; NCHRP 611). Saturated backfill is outside both methods and belongs to the geotechnical report.

#How do I model a water table behind the wall?

Enter the design groundwater level (above the bottom of the footing) and the saturated unit weight γ_sat in Step 2. The module then runs the earth pressure on the buoyant unit weight below the water table at the same K, adds full hydrostatic pressure, weighs the heel soil saturated below the water table, and applies base uplift with the head measured from the underside of the footing — γ_w·h_w at the heel edge falling to zero at the toe edge, the front of the wall being drained (IBC 2024 §1610.1 and Table 1610.1 footnote a; ASCE 7-22 §3.2.1, §3.2.2). It is all H (§2.2): 1.0 in the stability checks, 1.6 in the strength cases. Two things it will not do: pick the water elevation for you (that is the geotechnical report's), and combine a water table with the seismic increment — RP 12's methods assume groundwater below the base, so a submerged backfill's seismic pressure has to come from the report.

#Where does the active pressure stop when there is a shear key?

At the bottom of the footing, in this module — the footing-base free body of the classical hand method: the sliding surface is the footing base and the key contributes by mobilising passive pressure over its depth. The other legitimate free body carries the active pressure down to the bottom of the key and takes friction soil-on-soil. On the wall in this guide that alternative turns FS_sliding = 1.62 into 1.41 — a real difference between two correct conventions, and one to settle with the reviewer before comparing software outputs. What is not legitimate is the hybrid: crediting passive to the key bottom while stopping the active pressure at the footing bottom is the footing-base free body; carrying the active pressure deeper while keeping soil-on-concrete friction is neither.

#Why does 1.4D show up on the heel?

Because lateral earth pressure puts nothing on the heel, so the plain dead-load combination of ASCE 7-22 §2.3.1 (1a: 1.4D) competes directly with 1.2D + 1.6L (2a) — and wins whenever the variable surcharge is lighter than 0.2·D/1.6, about 208 psf on the wall in this guide, or is permanent, or is absent. On the toe the same combination appears as 1.4D + 1.6H, because H adds to the toe's load effect and the §2.3.1 H rule puts it at 1.6 in every combination it adds to. StructSuite checks both combinations on the heel and all three on the toe and reports the governing one; on this wall 1a governs both, by 8% on the heel and 3% on the toe.

#What do I assume for an expansive soil

For the backfill, the code mostly declines to answer: IBC 2024 Table 1610.1 marks CH, MH, OL and OH "Unsuitable as backfill material" — no active value, no at-rest value — and the clays it does accept cost double (C_L: 60 psf/ft active, 100 at-rest, against 30 for a clean gravel). For the base, Table 1806.2 class 5 gives no coefficient of friction at all: you get cohesion c = 130 psf instead, limited to one-half the dead load by §1806.3.2, with lateral bearing 100 psf/ft and vertical bearing 1,500 psf. Swelling pressure itself is in no code table — it is a measured value from the geotechnical investigation under IBC §1803, entered as an equivalent fluid pressure. On the wall in this guide, moving from a class 4 base to class 5 takes the sliding safety factor from 1.62 to 0.56 and the bearing from a pass to a fail; see Expansive soils for the full comparison and the practical fixes.

#How do I handle sloped backfill, or a wall taller than about 20 ft?

Sloped backfill is a Step 1 geometry input: enter the slope β and StructSuite raises the pressure plane at the heel edge to H′ = H_retained + t_f + L_heel·tanβ, adds the triangular soil wedge above the heel to the weights and to the heel's design load, and shares the same β with Mononobe-Okabe. What it will not do is price the sloped pressure from the code tables — Table 1610.1 / Table 3.2-1 are level-backfill values, so a sloped wall requires the geotechnical report's equivalent fluid pressure for the sloped condition (the module guards the tables off and says why). See Sloped backfill — when the ground keeps rising for what a 3H:1V regrade does to this exact wall. Above roughly 20–22 ft, a cantilever stem gets thick and congested enough that a counterfort wall is usually cheaper, and the stem stops behaving as a vertical cantilever — it spans horizontally between the ribs, which is a different design entirely. Tall walls also deserve a closer look at global stability, at wall movement and tilt, and at construction sequencing.