MSE Wall Design: Step-by-Step with a Worked Example

A reinforced soil wall holds up an embankment with the embankment itself. That idea is simple, and it is also why the design looks nothing like a gravity wall calculation. There is no heavy concrete block pushing back against the soil. There are layers of reinforcement tying the fill together until it behaves as one coherent mass with enough weight to stay put. Get the reinforcement and the backfill right and the wall is cheap, fast and forgiving of settlement. Get the drainage or the fill gradation wrong and the same wall bulges, settles behind a bridge abutment, or pulls a connection apart at the face.

This guide covers the design of a mechanically stabilized earth (MSE) wall from the input data to the final checks, with the arithmetic shown at every step. The example is a 6 m (19.7 ft) geogrid wall carrying a surcharge, which is about as common a case as the type gets, and every number in it can be recomputed from the tables printed here. The companion piece on retaining wall design works the gravity and cantilever cases; this one stays on the reinforced soil side.

What a mechanically stabilized earth wall is

An MSE wall is a composite structure. Four parts do the work, and three of them are easy to overlook until they fail:

• The reinforcement. Geogrids, geotextiles or steel strips, laid horizontally in layers, connected to the facing and extending back into the fill. This is what carries the tensile force the soil cannot carry on its own.
• The select fill. A granular, free-draining backfill that surrounds the reinforcement and provides the friction and interlock the reinforcement pulls against. Its properties are part of the design, not a contractor’s convenience.
• The facing. Precast concrete panels, modular blocks, welded wire, or wrapped geosynthetic. It is a facing, not a retaining structure: it holds the fill between the reinforcement layers and takes only the load of the connection.
• The drainage and the leveling pad. Water is the enemy of every retaining wall, and an MSE wall with a blocked drain loses its backfill strength before it loses anything else.

The design consequence of that list is the thing gravity wall designers find counterintuitive. In a gravity wall, the soil behind the wall is a load. In an MSE wall, the soil between the reinforcement layers is the structure. A gravity wall fails by sliding or overturning as a rigid block. An MSE wall fails internally, by a reinforcement layer pulling out of the fill or tearing, and it can fail that way while the external stability margins still look comfortable.

Why MSE design is not gravity wall design

Two questions decide the whole calculation, and both are answered before a single number is computed.

Is the reinforcement extensible or inextensible? Geosynthetics (geogrid, geotextile) stretch more than the soil around them during construction; steel strips do not. That single distinction changes the lateral earth pressure coefficient you use and the shape of the failure surface for pullout. The two families get different equations in every code, and mixing them is the most common error in hand calculations.

Is the reinforced block treated as rigid? For external stability, yes: the reinforced soil mass is a rigid body sitting on the foundation, and the checks are the ordinary sliding, eccentricity and bearing checks you would run on a gravity wall. For internal stability, no: the reinforcement carries the load layer by layer, and each layer is checked separately.

Codes nudge the details of those two answers. AASHTO uses the simplified method with a lateral earth pressure coefficient that varies with depth for inextensible reinforcement and is close to active for extensible reinforcement. BS 8006 and AS 4678 take their own routes to the same physics, and a design that has to satisfy a national highway authority usually has to follow that authority’s manual rather than the one in the textbook.

Metallic strips or geogrid: the choice that fixes the method

For inextensible reinforcement the soil is stiffer than the strip, so the strip restrains it, and the horizontal stress near the top of the wall is higher than active. AASHTO handles that with a ratio $K_r/K_a$ that starts around 1.7 for steel strips (2.5 for bar mats and grids) at the top of the wall and falls to about 1.2 at 6 m depth. For extensible reinforcement the reasoning runs the other way: the geogrid deforms with the soil, so the soil mobilizes active conditions and $K_r$ is taken as $K_a$ over the full height. That is the case worked below.

The internal failure surface begins at the toe of the wall in both cases, and this is worth stopping on because it is easy to get wrong. What changes between the two systems is the shape of that surface, not where it starts. With extensible geogrids it is approximately a straight line rising at $45° + \phi/2$ from the horizontal, the Rankine active plane. With inextensible steel strips the soil restrains the stiff reinforcement and the surface is approximately bilinear, steeper near the top of the wall and flatter near the toe. The two shapes leave very different embedment lengths behind the surface for the upper reinforcement layers, and the pullout check is where that difference shows up.

What has to be checked

The complete set for an MSE wall is longer than most hand calculations show:

External stability of the reinforced block

• Sliding on the base
• Eccentricity of the resultant, and bearing pressure under the base
• Global (overall) stability of the slope that contains the wall
• Bearing failure or excessive settlement of the foundation soil

Internal stability

• Tension in each reinforcement layer, against its long-term design strength
• Pullout of each layer from the fill behind the failure surface
• Connection strength between the reinforcement and the facing
• For metallic reinforcement, loss of section to corrosion over the design life

Serviceability

• Movement of the facing and of the fill at the top of the wall
• Differential settlement where the wall meets a rigid structure such as a bridge abutment

Skipping the last item is how a wall ends up structurally sound and still a complaint, because a 50 mm step at the end of a bridge deck is a serviceability failure that no factor of safety catches.

Worked example: a 6 m geogrid wall

The rest of this guide designs one wall in full: a 6 m (19.7 ft) high geogrid-reinforced wall with a modular block facing, level backfill, and a uniform surcharge on the retained ground.

Input data

The backfill properties are a design assumption and a specification. If the contractor supplies a plastic clay fill with 30 percent fines, the wall designed on these numbers is not the wall that gets built, and the difference is not small.

Step 1: geometry and reinforcement layout

Reinforcement length is set first, because everything internal depends on it. For extensible reinforcement the minimum length is $0.7H$, with an absolute floor of 8 ft (2.4 m) for constructability:

$L = 0.7H = 0.7 \times 6.00 = 4.20\ \text{m (13.8 ft)}$

With a vertical spacing of 0.60 m and a wall 6 m high, the reinforcement goes in 10 layers, at depths measured from the top of the wall of 0.3, 0.9, 1.5 m and so on down to 5.7 m. The spacing is set by the facing system and by the fill: 0.60 m (2 ft) is typical for a block facing, and going much above 0.80 m (2.6 ft) puts more load into each layer than most geogrids want to carry.

The active earth pressure coefficient for $\phi = 34°$ and level backfill, after Rankine:

$K_a = \tan^2\left(45° - \frac{\phi}{2}\right) = \tan^2(28°) = 0.283$

That number is used twice: as $K_r$ for the internal stability of the geogrid (extensible reinforcement) and for the external pressure behind the reinforced block.

Step 2: external stability

For external stability the reinforced mass is a rigid block 4.20 m wide and 6.00 m high. The soil behind it pushes on the back face, and the block’s own weight resists.

Driving force. The active pressure behind the block, including the surcharge:

$P_H = \frac{1}{2}K_a\gamma H^2 + K_a q H = \frac{1}{2}(0.283)(20)(36) + (0.283)(12)(6)$

$P_H = 101.8 + 20.4 = 122.1\ \text{kN/m (8.4 kip/ft)}$

This uses the Rankine coefficient with no wall friction on the back face, which is the conservative choice. Taking advantage of wall friction with $\delta = \tfrac{2}{3}\phi$ reduces the driving force, and many offices do it, but it is a refinement worth leaving out when a check is already passing.

Weight of the block. The reinforced fill weighs

$W = \gamma L H = (20)(4.20)(6.00) = 504\ \text{kN/m (34.5 kip/ft)}$

and the surcharge sitting on top of the reinforced fill adds another $qL = 12 \times 4.20 = 50.4\ \text{kN/m}$, so the total vertical load at the base is

$V = 504 + 50.4 = 554.4\ \text{kN/m (38.0 kip/ft)}$

Sliding. The base friction coefficient is $\tan\phi_b = 0.45$, so

$FS_{sliding} = \frac{V \tan\phi_b}{P_H} = \frac{554.4 \times 0.45}{122.1} = 2.04$

against a usual minimum of 1.5. The check has room.

Eccentricity and bearing. The two parts of the thrust act at different heights, and keeping them apart is worth the extra line. The soil component is a triangle, so it acts at $H/3 = 2.00$ m above the base; the surcharge component is a rectangle acting at $H/2 = 3.00$ m. Their moments about the centre of the base add up to

$M = 101.8 \times 2.00 + 20.4 \times 3.00 = 203.6 + 61.1 = 264.6\ \text{kN}\cdot\text{m/m (59.5 kip}\cdot\text{ft/ft)}$

The vertical load acts at the centre of the 4.20 m base, so the eccentricity is

$e = \frac{M}{V} = \frac{264.6}{554.4} = 0.48\ \text{m (1.6 ft)}$

The middle third of the base ends at $L/6 = 0.70$ m, and 0.48 m sits inside it, so the resultant keeps the base pressure trapezoidal with no tension. AASHTO LRFD is less strict here: for a foundation on soil it only asks that the resultant stay within the middle two-thirds of the base ($e \leq L/3$), a wider band than the classic middle-third rule. Designing to the middle third is the safer habit, and this wall passes it.

The base pressures follow:

$q_{max} = \frac{V}{L}\left(1 + \frac{6e}{L}\right) = \frac{554.4}{4.20}\left(1 + \frac{6 \times 0.48}{4.20}\right) = 132.0 \times 1.68 = 222\ \text{kPa (4,637 psf)}$

$q_{min} = \frac{554.4}{4.20}\left(1 - \frac{6 \times 0.48}{4.20}\right) = 132.0 \times 0.32 = 42\ \text{kPa (877 psf)}$

This is the check that most often sends an MSE wall back to the drawing board. A 6 m wall carrying a surcharge demands a foundation soil with an allowable bearing pressure of at least 222 kPa (4,637 psf) at the base, which is more than many residual soils can offer at a shallow depth. If the foundation cannot supply it, the options are to lengthen the reinforcement (a wider block spreads the same load), to excavate to a better stratum, or to improve the ground under the base.

Global stability. Not computed here, but it cannot be skipped, and there are two surfaces to look at. The deep-seated surface passes below the reinforced block, and the compound surface runs partly through the reinforced mass and partly behind it. The compound case is the one an MSE wall adds to an ordinary slope stability analysis, and it is the one a plain bearing check misses. Shallow-seated global failure is the mode that makes every internal check irrelevant.

Step 3: tension in each reinforcement layer

The horizontal stress at a depth $z$ below the top of the wall is $K_r$ times the vertical stress, and each layer carries that stress over its own tributary height:

$T_{max} = K_r \left(\gamma z + q\right) S_v = K_a \left(\gamma z + q\right) S_v$

The $K_a$ here is the same 0.283 computed above, because geogrid is extensible reinforcement. Take the deepest layer, at $z = 5.7$ m:

$\sigma_v = (20)(5.7) + 12 = 126\ \text{kPa (2,632 psf)}$

$\sigma_H = 0.283 \times 126 = 35.6\ \text{kPa (744 psf)}$

$T_{max} = 35.6 \times 0.60 = 21.4\ \text{kN/m (1,464 lbf/ft)}$

Repeating that for all ten layers gives the tension profile:

The tension grows to the bottom because $K_r$ is held constant at the active value. That is the characteristic of extensible reinforcement. In a steel strip wall, the ratio $K_r/K_a$ would decline with depth and the peak tension would sit somewhere near mid-height instead, which is one reason the two systems do not get the same reinforcement layout from the same wall height.

The strength side. A geogrid’s ultimate strength is not a strength you can design to. Creep under sustained load, damage during installation and long-term durability all eat into it, so the design strength is the ultimate strength divided by three reduction factors:

$T_{al} = \frac{T_{ult}}{RF_{ID} \cdot RF_{CR} \cdot RF_{D}} = \frac{90}{1.10 \times 2.60 \times 1.20} = \frac{90}{3.432} = 26.2\ \text{kN/m (1,797 lbf/ft)}$

The most loaded layer demands 21.4 kN/m against 26.2 kN/m available, a utilization of 0.82, so the geogrid chosen passes with about 18 percent in hand. Those reduction factors are not conservative filler: the creep factor is usually the largest of the three and depends on the polymer, the temperature and the design life, and it comes from the manufacturer’s test data, not from a table in a textbook.

Step 4: pullout of each layer

A layer can carry its tension and still come out of the fill. The resistance is the friction and interlock along the length of reinforcement that sits behind the failure surface. For extensible reinforcement the failure surface is the straight line from the toe at $45° + \phi/2$, so the distance from the face to the surface at a height $d$ above the base is

$x = d \tan\left(45° - \frac{\phi}{2}\right) = d \tan(28°) = 0.532\,d$

and the embedment length is what is left of the reinforcement behind that line:

$L_e = L - x$

The resistance is written the way AASHTO writes it, as a factor times the vertical stress times the embedment length:

$P_r = F^* \alpha \sigma_v C L_e$

where $F^* = \tfrac{2}{3}\tan\phi = 0.45$ is the default pullout resistance factor for a geosynthetic, $\alpha = 0.8$ is the scale correction for a geogrid (0.6 for a geotextile), $C = 2$ counts both faces of the reinforcement, and $R_c = 1$ because the geogrid covers the full width. Those values give $F^* \alpha C = 0.72$, so

$P_r = 0.72\,\sigma_v L_e$

Top layer, $z = 0.3$ m, so $d = 6.00 - 0.30 = 5.70$ m:

$x = 0.532 \times 5.70 = 3.03\ \text{m}, \qquad L_e = 4.20 - 3.03 = 1.17\ \text{m (3.8 ft)}$

$P_r = 0.72 \times 18.0 \times 1.17 = 15.1\ \text{kN/m}, \qquad FS_{pullout} = \frac{15.1}{3.1} = 5.0$

Bottom layer, $z = 5.7$ m, so $d = 0.30$ m:

$x = 0.532 \times 0.30 = 0.16\ \text{m}, \qquad L_e = 4.20 - 0.16 = 4.04\ \text{m (13.3 ft)}$

$P_r = 0.72 \times 126 \times 4.04 = 366\ \text{kN/m}, \qquad FS_{pullout} = \frac{366}{21.4} = 17$

The pattern is the opposite of the tension check. The upper layers have little tension but also little embedment, and the lower layers have heavy tension but almost the whole reinforcement length behind the failure surface. The upper layers are the ones that run short of pullout resistance, so if this check ever fails, the fix is usually a longer geogrid in the top two or three layers rather than a stronger one. AASHTO also sets a minimum embedment length of 3.0 ft (0.91 m) behind the failure surface, which the top layer’s 1.17 m clears.

A note on conservatism in $P_r$: the factors above are AASHTO’s defaults. In a final design, $F^*$ for a specific geogrid comes from pullout tests on that product (ASTM D6706) rather than from an assumed friction angle, and it is often higher than $\tfrac{2}{3}\tan\phi$, because a geogrid interlocks with the fill and mobilizes more than pure interface friction. Using the default here is conservative, and on the upper layers it is the difference between a factor of safety of 5 and one in the double digits.

Step 5: the connection at the face

The reinforcement is only as good as its attachment. The connection load at the facing is the same $T_{max}$ computed above, so the facing connection has to carry 21.4 kN/m (1,464 lbf/ft) at the most loaded layer. Codes frame that as a resistance check rather than a factor of safety: AASHTO requires the applied connection load $T_o$ to satisfy $T_o \leq \phi T_{ac} R_c$, where $T_{ac}$ is the nominal connection strength from a connection test and $R_c$ the coverage ratio. If you want a margin on top of the code check, a factor of 1.5 on the peak tension is a reasonable design choice rather than a code requirement, which would ask for 32.1 kN/m (2,196 lbf/ft) here. Either way, the connection strength is a property of the facing system and the geogrid together, measured in a connection test, and it is the number most often quietly exceeded when the geogrid is changed to a stronger product without revisiting the facing.

The results at a glance

Everything internal passes with room, and the binding constraint is the bearing pressure under the base. That distribution is normal. MSE walls are internally generous and externally demanding, because the same narrow block that makes them cheap is what concentrates the load on the foundation.

The details that decide whether the wall stands

The calculation above is the part that can be checked with a spreadsheet. The part that decides how the wall behaves over twenty years is in the specification and the construction sequence.

Backfill gradation. AASHTO and FHWA practice wants the reinforced fill granular, free draining and non-plastic: no more than 15 percent passing the 0.075 mm (No. 200) sieve, and a plasticity index of 6 or less. Fines hold water, water raises the horizontal pressure and softens the fill, and a wall designed on $\phi = 34°$ does not get 34° out of a plastic silt.

Drainage. A perforated collector pipe at the base of the reinforced fill, wrapped in drainage aggregate, discharging to a positive outlet. The wall face is not a drain.

Compaction of the reinforced fill. Every lift has to be compacted to the specified density, and the compaction has to be done with a roller whose path does not damage the geogrid. Fill placed behind the facing with a plate compactor and fill placed in the wide zone with a roller are compacted to the same requirement. The mechanics of how that density is specified and verified on site are the subject of the guide on soil compaction.

Durability of the reinforcement. Steel strips need a corrosion allowance and, more importantly, an electrochemical check on the fill: resistivity, pH, chloride and sulfate contents decide whether the galvanizing and the steel itself will last the design life. FHWA’s limits for a metallic reinforcement are a resistivity above 3,000 ohm·cm, a pH between 5 and 10, chlorides under 100 ppm and sulfates under 200 ppm, with an organic content under 1 percent. Geosynthetics do not corrode, so those electrochemical limits do not apply to them; what matters there is installation damage, creep and chemical compatibility with the fill over the design life.

Movement. MSE walls settle with their fill and the facing moves outward slightly at the top as the soil mobilizes. The scale of that movement depends on the reinforcement: FHWA’s estimates put lateral deformation at roughly 0.2 to 0.4 percent of the wall height for inextensible metal reinforcement, and around three times that, so up to about 1 percent of $H$, for extensible geosynthetics. On a 6 m wall that is up to 60 mm of outward movement at the top, which is harmless against soil and a problem against anything rigid. The settlement that meets a bridge abutment is the classic version of this problem, and it is controlled by extending the reinforcement, stiffening the foundation, or detailing the approach slab to absorb the step.

Facing. Panels are a cladding system with connections and joints, not a structural slab. The panel thickness, the joint detail and the coping all exist to keep the fill in and the water out, and none of them add capacity to the wall.

Factors of safety, and where the codes part company

The worked example uses the classic allowable-stress method with factors of safety on the external checks and reduction factors on the reinforcement strength. AASHTO LRFD, which governs most highway work in the United States, reaches the same physics through load factors and resistance factors, so the arithmetic does not transfer line for line. A wall designed under LRFD and one designed under allowable-stress methods will not have identical reinforcement layouts, and neither is wrong.

The internal checks transfer more directly than the external ones. The LRFD route applies a load factor of $\gamma_{EH} = 1.5$ to the horizontal stress and resistance factors of about 0.9 to the tensile and pullout resistance, which lands close to the allowable-stress result on a granular fill. The external sliding and bearing checks are where the two methods diverge most, because LRFD loads the earth pressure and the surcharge with its own factors and then resists them with factors on the base friction and the bearing resistance.

The practical rule is to design to the manual of the authority that will accept the work, and to hand the reviewer the manual you used. Mixing a reduction factor from one code with a factor of safety from another produces a wall that satisfies nobody’s check, and the arithmetic will not reveal it.

Frequently asked questions

How long does the reinforcement have to be? The starting point is $0.7H$ for extensible reinforcement, with 8 ft (2.4 m) as an absolute minimum, and both the pullout check and the external stability check can push it longer. It is rare for a wall to need less, and common for one on soft ground to need more.

Can I use the site soil as backfill? Only if it meets the gradation and plasticity limits. Cohesive or high-fines fill changes the friction angle, reduces the interface resistance and holds water, so the design would have to be redone on the real properties. If the fill is suitable, using it saves the cost of importing granular material and is one of the reasons MSE walls are cheap.

What happens if the wall is taller than expected? Everything scales, and not all of it in your favour. The lateral thrust grows with $H^2$ while the block weight grows with $H$ and $L$, so a taller wall needs a disproportionately wider block, and the bearing pressure under it climbs quickly. Tall MSE walls are common on highways, but the foundation check is what usually decides whether they are feasible.

Do MSE walls need a concrete footing? No. They sit on a leveling pad, typically a 300 mm wide by 150 mm deep concrete strip that gives the first facing course a stable base. The reinforced block spreads the load across its whole width, which is the point of the system.

Why do some MSE walls bulge at the face? Usually a connection problem, an over-compacted or under-compacted zone right behind the facing, or a facing system installed to a tolerance the panel joints cannot absorb. Bulging is rarely a global stability issue and often a detailing one.

Is the geogrid strength in the datasheet the design strength? No. The datasheet ultimate strength has to be divided by the installation damage, creep and durability reduction factors before it is a design strength. The creep factor alone is commonly two to three, so a 90 kN/m geogrid may only offer 26 kN/m to the design.

Where to go next

• The external checks here are the same ones a gravity or cantilever wall needs, with the soil pressure worked from first principles: practical retaining wall design.
• Reinforcement connects to load the way piles do, and the tension side of the calculation has a lot in common with pile capacity.
• When the foundation under the base cannot take the pressure, the fix is usually ground improvement: ground improvement techniques.
• Bearing pressure under a foundation, and the difference between gross and net values, is covered in gross and net bearing pressure.
• Run the retaining wall and bearing capacity numbers in the browser with the retaining wall tool.
• Compare the earth pressure coefficients and thrusts of a backfill, Rankine against Coulomb, with the free lateral earth pressure calculator.

References

• Berg, R.R., Christopher, B.R. and Samtani, N.C. (2009). Design and Construction of Mechanically Stabilized Earth Walls and Reinforced Soil Slopes, Volume I. FHWA-NHI-10-024 (GEC-11). Federal Highway Administration, Washington, D.C.
• Taylor, T.P., Collin, J.G., Boyle, S.R., Fishman, K.L. and Han, J. (2023). Design and Construction of Mechanically Stabilized Earth (MSE) Walls. FHWA-HIF-24-002 (GEC-11). Federal Highway Administration, Washington, D.C.
• AASHTO (2020). LRFD Bridge Design Specifications, 9th edition, Section 11: Abutments, Piers, and Walls (MSE walls, Article 11.10).
• British Standards Institution (2010). BS 8006-1: Code of practice for strengthened/reinforced soils and other fills.
• Standards Australia (2002). AS 4678-2002: Earth-retaining structures.
• European Committee for Standardization. EN 1997-1: Eurocode 7, Geotechnical design, Part 1: General rules, and EN 14475:2006: Execution of special geotechnical works, Reinforced fill.
• ASTM International. ASTM D6637/D6637M: Standard Test Method for Determining Tensile Properties of Geogrids, and ASTM D6706: Standard Test Method for Measuring Geosynthetic Pullout Resistance in Soil.
• Elias, V., Fishman, K.L., Christopher, B.R. and Berg, R.R. (2009). Corrosion/Degradation of Soil Reinforcements for Mechanically Stabilized Earth Walls and Reinforced Soil Slopes. FHWA-NHI-09-087. Federal Highway Administration, Washington, D.C.