Chapter 2 — Track Substructure: Ballast, Subballast and Subgrade

How fast the load spreads: contact stress at each layer

Profillidis follows a 22-kip static wheel load down through the track. The contact area grows from about 0.2 in.² at the wheel–rail contact to about 1,550 in.² at the track bed–subgrade interface, and the stress falls from about 110,000 psi to about 6 psi — a reduction of 1,000 to 5,000 times. The calculation assumes the tie under the wheel carries 40% of the wheel load (finite-element and field results, see §2.2) and ignores dynamic effects.

He defines the subgrade as extending to about 6.5 ft below the top of subgrade, the depth to which train loading has a mechanical effect. That depth is a useful minimum for exploration and for the zone over which subgrade parameters are needed.

Figures · click to zoom

After Profillidis (2014), §7.2, Fig. 7.3. Full reference under Supplemental Sources.

Field view: typical ballast section and how load spreads

Stress under passing trains is about 6,000 psf just below the ties and falls to about 200 psf at 5 ft depth, before any dynamic allowance. At the rail, a loaded tank car’s weight spreads over about 19 ft (roughly 11 ties), and it continues to spread through the ballast and subgrade. A thin or degraded ballast and sub-ballast section concentrates stress into a smaller area of subgrade than the design assumed. The course states that substructure condition is the leading cause of track-geometry degradation, and that the substructure is what fails about 80% of the time when track has a problem.

Figures · click to zoom

After Geotechnology for Railroaders Program, Module 3; §02–03, §13. Full reference under Supplemental Sources.

Load sharing between ties, dynamic increment and subgrade stress (finite-element results)

The classical assumption is that the tie under a wheel carries 50% of the wheel load and each neighbor 25%. Stress measurements and finite-element analysis give a different split: 40% on the tie under the load, 23% on each adjacent tie and 7% on each second tie, with negligible effect beyond.

Track and wheel defects add a dynamic load, so analysis should use Q_tot = Q_stat + Q_dyn, where Q_dyn can reach 50% of the static wheel load. For a 11-ton wheel at 124 mph the measured increment was 4.4–6.6 tons. The book separates three frequency bands: 0.5–15 Hz (sprung mass, rolling-stock dependent), 20–100 Hz (unsprung mass, governed by track quality and stiffness) and 100–2,000 Hz (rail corrugation). Empirical dynamic impact factors close to measured values range from 1.35 to 1.6 at 125 mph, and 1.5 is suggested near that speed. Compare these with the AREMA impact allowance used in this section, which remains the design basis.

Elastoplastic finite-element runs (Figure 2-14) show that stress at the top of subgrade depends mainly on subgrade quality and only secondarily on track-bed thickness. With all else equal, moving the subgrade up one class (S1→S2, S2→S3, S3→rock) increases the stress developed in the subgrade by about 50%: a stiffer subgrade attracts more load, but carries it with far less settlement.

Figures · click to zoom

After Profillidis (2014), §7.11.2, §8.4.7–8.4.8, §8.6–8.7, Figs. 8.7, 8.10, 8.14, 8.15. Full reference under Supplemental Sources.

How the track spreads a wheel load, and why no component can be designed alone

The track is a load-spreading system. It takes a concentrated wheel load at the rail head and spreads it along the track, across it and down through the section until the subgrade can carry it. Track that is well designed, built and maintained shares the load evenly, so each component carries its part. The designer’s job, on new track or on an upgrade of existing track, is to select a section that carries the design wheel load without overstressing any of the four main components: rail, ties, ballast and subgrade.

Along the track. When a wheel sits directly over a tie, that tie carries less than half of the wheel load. The rest goes to the two ties on each side (Figure 2-S5). For a 30,000-lb wheel on good track the UFC example split is 40% / 20% / 10%. That is close to the 40 / 23 / 7 finite-element split given above.

Down through the section. Pressure at the wheel–rail contact, about 60,000 psi over roughly ½ sq in., falls to about 300 psi on the tie plate, about 40 psi at the base of the tie and about 10 psi (1,440 psf) at the top of subgrade (Figure 2-S6). That is the same order as the Table 2-1 values for 18 to 24 in. of ballast and subballast, so it is a quick reasonableness check on a Talbot calculation. Actual pressures vary with wheel load, track section and track condition.

Figures · click to zoom

Design point. Track is simple to build, but its behavior under load is not. Strengthening one component can increase the load on another. When a rehabilitation changes one component (rail section, tie type or spacing, ballast depth), recheck the others, including the stress at top of subgrade in §2.2.1.

After UFC 4-860-01FA (2004), ch. 6, §5, Figs. 6-10 and 6-11. Comparison with §2.2 and Table 2-1 added by the authors. Full reference under Supplemental Sources.

Selecting the design wheel load from the car mix

Set the design wheel load from the heaviest loaded car that commonly runs over the track, not the rare heaviest movement. A locomotive usually puts down the heaviest single wheel. On most track, though, loaded-car wheels far outnumber locomotive wheels, so the car wheel load governs in most cases. Check both before the design is final.

The design wheel load is the static wheel load of that car plus an allowance for speed and for track and wheel irregularities. Table 2-S8 gives planning values for slow-speed track:

• 10–25 mph: connecting track between the site and the serving railroad.
• ≤ 10 mph: yard, industry-lead and terminal track.
• Light use: track used mainly by a switch engine, for empty-car storage or only occasionally.

Authors’ check: the Light use column equals the static wheel load (gross rail load ÷ 8 for a four-axle car; e.g., 286,000 ÷ 8 ≈ 35,800 lb). The ≤ 10 mph and 10–25 mph columns are the static value × 1.10 and × 1.25.

Limits on use. These speed allowances are lower than the 1.2 to 1.5 dynamic load factors used elsewhere in this manual because they assume slow operation. Use Table 2-S8 only for industrial, yard, terminal and light-density track. For mainline work, use Cooper E-80 or 286,000-lb loading with the client’s dynamic load factor (Part I, §I.4).

After UFC 4-860-01FA (2004), ch. 2, §4, Table 2-2. Car and locomotive designations as given in the source; factor check added by the authors. Full reference under Supplemental Sources.

Dynamic loads, rare heavy impacts and the fourth-power rule

How large the dynamic load is depends on the traffic. North American static wheel loads range from about 12,000 lb (light rail) to 39,000 lb (heavy haul). Measured dynamic loads compare as follows:

• FAST heavy-haul unit trains with good wheels: dynamic loads were close to static.
• Northeast Corridor mixed traffic: dynamic loads ran well above static at the 2% exceedance level. Almost all of the excess came from passenger cars (23,000-lb static wheels) with wheel flats at 80–110 mph.
• Continuous welded rail with good maintenance: dynamic loads typically run 10–20% above static.
• Wheel and rail defects: loads near 3× static, with vibration above 100 Hz.

Rare heavy loads can control settlement. In box tests the order of loading did not change the final settlement. Once the largest load had been applied, smaller loads added almost nothing. A small share of heavy impacts, even 1 in 1,000, can therefore govern ballast settlement. Removing wheel flats, battered joints and rail defects is a substructure measure as well as a rail measure.

Uplift. The rail lifts the tie ahead of the wheel and then drives it back down. Where water is present, this pumps it.

Axle load and maintenance cost. German (DB) experience puts ballast maintenance cost roughly in proportion to the fourth power of ballast pressure, and British Rail concurs. By that rule, a 20% increase in axle load (for example, 100-ton to 125-ton cars) about doubles ballast maintenance. In box tests, raising the wheel load from 36 to 50 kips doubled the fines produced and tripled coarse breakage. DB responded with these changes:

• 24-in. tie spacing.
• 8.5-ft ties.
• 60 kg/m (121 lb/yd) rail.

Together they stretched tamping intervals from 1–2 years to 4–5 years. British Rail found lab settlement roughly proportional to axle load and to tie spacing. Going from 113 to 131 lb/yd rail (56 to 65 kg/m) cut settlement about 20%. Use these figures when a client asks what a heavier car fleet will cost in substructure maintenance (Chapter 1, §1.11).

After Selig & Waters (1994), chs. 2, 6 and 8. Full reference under Supplemental Sources.

European ballast categories (EN 13450) for comparison with AREMA

The European standard designates ballast by a pair of sieve sizes, 1¼ in. lower and 2 in. or 2½ in. upper, and lets the supplier declare a category for each property. The Los Angeles test is the reference hardness test (in the EN version: 22-lb sample, 1,000 revolutions at 31–33 rpm). Many older regulations simply require LA < 25.

Deval requirements in older regulations range from a coefficient greater than 12 (limestone) or 14 (hard rock) to greater than 16, which may exclude limestone and push a railway toward granite. French practice requires the LA and Deval results to plot together inside an acceptance band (Figure 2-18).

Figures · click to zoom

After Profillidis (2014), §12.2–12.4, Tables 12.2–12.9, Figs. 12.1, 12.5; after EN 13450 (2002). Full reference under Supplemental Sources.

Ballast shoulder, crib fill and lateral track resistance

Lateral (transverse) track resistance has three parts: friction under the tie (proportional to tie weight), friction along the tie sides from crib ballast, and passive resistance at the tie ends from the shoulder. The crib component is about 40–50% of the total for timber ties, 30% for monoblock concrete and 15–25% for twin-block ties, so under-filled cribs cost timber track the most.

• Widening the shoulder and heaping it above tie level gives more resistance than widening alone; the shoulder slope matters less (Figure 2-20).
• Immediately after tamping, lateral resistance is about 50% of its stabilized value. It is almost fully recovered after about 2 million tons of traffic; dynamic stabilization and crib compaction shorten the recovery (Figure 2-21).
• Fully stabilized twin-block concrete track resists more than twice as much as timber; monoblock concrete sits between. Lengthening ties from 8 ft to 8.5 ft (German practice) added 15–20%. Hard, smooth tropical-hardwood ties give about 15% less than other timber; grooving their seat adds 20–25%.
• Closer tie spacing slightly lowers resistance per tie but raises it per unit length of track.

These numbers explain why slow orders follow surfacing and undercutting, and why shoulder width and crib fill should be checked wherever a ground-hazard repair has disturbed the ballast section.

Figures · click to zoom

After Profillidis (2014), §13.3–13.6, Figs. 13.2, 13.3, 13.6, 13.7. Full reference under Supplemental Sources.

The six functions of ballast and what good ballast requires

Ballast does six jobs: drainage, load distribution, vertical stiffness, longitudinal hold, lateral hold, and making surfacing practical. Failure of a function shows in the geometry: 2.5 in. or more of rail deflection at a joint means load distribution has failed, uneven support shows as cross-level, and ballast plowed in the direction of travel means it is too fine to hold the track. Judge ballast by digging, not by the surface.

Ballast shear strength comes from friction, which depends on size, angularity and interlocking. Prefer hard, dark, fine-grained extrusive rock (andesite, basalt, micro-diorite). Limestone, quartz, metamorphic rock and shale are unsuitable, and soft rock can grind into a mud spot within 2–3 years. Ballast life is roughly 200–900 MGT depending on rock type: some limestone lasts about 300 MGT, andesite about 900 MGT.

Shoulder width is a standard to enforce, not tolerate. A minimum 12-in. shoulder is what holds the ties laterally, and inadequate shoulders can contribute to derailments.

Figures · click to zoom

After Geotechnology for Railroaders Program, Module 3, Module 2; Module 3 §04–05, §11; after Selig & Waters (1994). Full reference under Supplemental Sources.

UIC subgrade quality classes and the formation layer

UIC Code 719R classes the subgrade by its expected behavior: S3 good (low settlement, very good support), S2 medium, S1 poor (large settlement), and S0 very poor, plus R for sound rock, which later UIC editions fold into S3. The class follows from fines content, plasticity, Los Angeles value for rock, and whether hydrogeological conditions are good or bad.

Fines = particles smaller than 60 µm. Some railways place the mixed/organic soils in S1.

S0 soils settle heavily, are inhomogeneous, change with time and let ballast punch in. Avoid them in alignment where possible or replace them; where neither is possible, especially under high embankments, combine ground improvement with thicker ballast and subballast and a geotextile.

On S1 and S2 subsoil, UIC practice adds a formation layer of better material compacted to 100% Standard Proctor, against the 95% usually specified for embankment subsoil. It only works if the subsoil has a low water content (otherwise fines migrate up and spoil the crossfall) and the layer is homogeneous, without pockets of fine material.

Figures · click to zoom

After Profillidis (2014), §9.5–9.7, Tables 9.2–9.3, Fig. 9.5; after UIC Code 719R (1994). Full reference under Supplemental Sources.

UIC track-bed thickness formula with a worked example

UIC sizes the total track bed (ballast + subballast) by adding corrections to a base value for subgrade class. The formula comes from the finite-element stress results in Figure 2-14:

Book example. UIC group 4 line, 8.5-ft monoblock concrete ties, S2 subgrade, 22-ton axles, medium maintenance, 124 mph, 0.2-in. geotextile: e = 21.7 − 2.0 − 2.0 + 0 + 0 + 2.0 + 0.2 = 19.9 in. (about 20 in.). With the usual 6-in. subballast, the ballast is 14 in., close to the 15 in. that British practice would give. The same site with a frost index of 540 °F-days needs 31 in. of track bed for frost protection (Figure 2-29), and the larger value governs.

North American check. A 286k car on 4 axles is about 35.7 tons/axle, beyond the formula’s 33-ton calibration point (d = +9.8 in.). Treat the formula as a cross-check on the AREMA/Talbot thickness in this chapter, not as a substitute.

UIC subballast is usually 6 in. of well-graded gravel with Micro-Deval < 15–20 and LA < 20–25 depending on the line’s importance; some railways require at least 30% crushed stone on new lines. Some railways omit subballast and use a thicker formation layer instead.

After Profillidis (2014), §12.1.2, §12.5.1–12.5.5, Eq. 12.8. Full reference under Supplemental Sources.

Sub ballast in practice

Sub ballast is a compacted, well-graded sand–gravel layer 6–12 in. thick between ballast and subgrade. It is often missing on century-old mainline. It drains as part of the lateral path, adds vertical strength, and separates ballast from fine-grained subgrade in both directions. Course specification: crushed or screened pit-run gravel graded 0–3 in. or 0–¾ in., less than 8% fines, no organics, well graded, and ideally at least 50% fractured faces. Watch for contractors offering fill with more fines than specified: it compacts well dry but performs poorly once wet.

Figures · click to zoom

After Geotechnology for Railroaders Program, Module 3, Module 2; Module 3 §08; Module 2 §09. Full reference under Supplemental Sources.

Cyclic loading: when subgrade plastic strain stops stabilizing

For soils, fatigue shows up as accumulating plastic strain rather than sudden fracture. Repeated-load triaxial tests on clays (Figure 2-25) use R, the ratio of the first-cycle deviator stress to the deviator stress that causes failure. Above R ≈ 0.9, plastic strain increases very rapidly with the number of cycles. Below it, strain grows roughly with log N and levels off. The runaway behavior was observed in S0 and S1 subgrades.

For the effect of traffic on subgrade stress the book uses Dormon’s rule from highway practice: σ₁/σ₂ = (N₂/N₁)^λ, with λ ≈ 0.2. For two lines with the same axle load, the allowable subgrade stress therefore scales with (T₂/T₁)^0.2: doubling tonnage lowers the allowable stress by about 13%.

Subgrade fines can also migrate upward. Figure 2-26 shows the combinations of plasticity index and liquid limit for which strong infiltration of subgrade fines into a gravel subballast was observed. It is a useful screen when deciding whether a separation geotextile or sand blanket is needed.

Figures · click to zoom

After Profillidis (2014), §9.8, §9.10, §9.15.2, Figs. 9.7, 9.19. Full reference under Supplemental Sources.

Frost index and frost-protection thickness

The frost index is the time integral of below-freezing temperature (°F × hours or °F × days). A railway must choose whether to design for the coldest winter or for an average winter and accept occasional frost penetration. Silt is very frost-susceptible, clay is susceptible, and sand and gravel are not.

The required frost-protection layer under 14 in. of ballast increases with frost index and depends on the material (gravel needs the most, peat/bark the least; Figure 2-27). A foam-plastic insulation board 2–4 in. thick sharply reduces the required thickness (Figure 2-28). On existing track, frost protection is normally installed at renewal using one of four details: a gravel or cinder layer, stone with a peat filter, foam plastic, or insulation over a frost-storage bottom layer.

Figures · click to zoom

After Profillidis (2014), §9.11, §12.5.2, Table 9.6, Figs. 9.8–9.13, 12.6. Full reference under Supplemental Sources.

Mud spots, cavities and silt pumping: diagnosing what is below

Before touching a mud spot, ask what kind it is and how deep the problem goes. The cause is usually ballast degradation, subgrade liquefaction and pumping, plastic deformation, subgrade failure, or poor drainage. The fix ranges from tamping (temporary) to undercutting or reinforcing the subgrade (durable). One railroad spent nearly $10M in a year on mud spots and wasted about 90% of it treating symptoms. On another system, over 95% of a $200M+/year ballast budget went into the upper 20 in. of the structure.

On a steep embankment, uncorrected ballast pockets can become a slope failure. In one silt-pumping case, an investigation trench beside sinking track showed silt rising as the track went down. The fix was filter cloth, gravel backfill and deeper ditches, which addressed the water path rather than the sinking.

Figures · click to zoom

After Geotechnology for Railroaders Program, Module 3; §02, §09, §11. Full reference under Supplemental Sources.

Frost heave and spring silt liquefaction: field mechanism and remedies

Frost heave needs five conditions at once: saturated soil, frost-susceptible soil (silt), sub-zero temperature, a slowly advancing freezing front, and a water supply from below. Remove any one and it cannot occur. About 75% of heave problems trace to poorly maintained ditches and about 25% to poor ballast drainage (ballast pockets), even on high fills. In spring, ice lenses can push the rail up several inches over 10–40 ft of track, usually by different amounts on each rail. Shimming is reactive; treat the cause.

Marking and salting is the most practical remedy once drainage alone is not enough. Mark the start and end of the heave tie by tie in spring, then salt in the fall; salting earlier lets rain carry the salt too deep. Provide a transition between salted and unsalted ground so a depression does not form. Other options are drainage and subdrainage, shimming, excavating and replacing silty soil, insulation and injection.

Spring silt liquefaction happens when the thawed top of a silt layer is squeezed by traffic against still-frozen ground below. Pore pressure cannot drain down, so it fractures upward and carries silt onto the shoulder. In one case cross-level moved 3–4 in. under a train. The condition clears once the frost is out. A lateral trench drain speeds recovery:

• excavate at the point of bulging silt, down to the soft silt at the frost front, and drain any water pocket;
• place nonwoven geotextile in direct contact with the silt;
• backfill with free-draining angular stone compacted in lifts, which acts as a shear key; wrap the fabric over the trench about 2 ft below base of rail; finish with clean ballast sloped to a clean ditch;
• space trenches about 15–20 ft along the weak zone; use a vacuum truck where cables are buried; record locations, since trenches may need redoing in 2–6 years.

Figures · click to zoom

After Geotechnology for Railroaders Program, Module 6.6; §02–05. Full reference under Supplemental Sources.

Subgrade stability first, and a screening bearing check for light-density track

If the subgrade is not stable, the track will not hold alignment, surface or cross level, however much maintenance is spent on it. Settle the subgrade before the track section. Investigate the alignment (Chapter 5). Design cuts, fills and subgrade preparation to AREMA Chapter 1, Part 2. Confirm soil suitability against the AREMA table of soil types and their suitability as railroad subgrade.

For cohesive subgrades, UFC 4-860-01FA sets the design bearing capacity of the track structure from the unconfined compressive strength, q_u, measured at natural water content:

Use as a screening check only. The UFC rule was written for low-tonnage installation track. The progressive-shear check in §2.5.1 (FS ≥ 3 on (π + 2) s_u) allows a stress of about 1.7 s_u divided by the dynamic load factor, so the UFC rule is less conservative. Use it to screen industry, yard and light-density branch-line track. For mainline and heavy-haul track, design to §2.5.1 and the granular layer methods in §2.4.

Chemical stabilization in frost areas. Use lime or cement stabilized subgrade under track with caution in seasonal-frost areas, and only after laboratory testing that includes frost susceptibility and freeze–thaw durability (Chapter 17, §17.7).

After UFC 4-860-01FA (2004), ch. 6, §8. Comparison with §2.5.1 added by the authors. Full reference under Supplemental Sources.

Reaction coefficients, ballast thickness and the stiffness balance

Profillidis expresses stiffness as a reaction coefficient ρ = load / settlement at the level considered (56 kip/in. per t/mm in the source). Typical component values are: rail 280,000–560,000 kip/in.; timber tie 50–80; concrete tie 1,200–1,500; ballast 10–30; rubber pad 560–1,120 kip/in.. The subgrade is the softest element in the chain:

Existing track with ballast only (no subballast): total ρ = 8.4–56 kip/in., average 17 kip/in.. Bridges: 560–840 kip/in..

Design should satisfy two principles: stresses reaching the subgrade must stay below failure values, and the track must remain flexible. A rock subgrade is more than three times as stiff as a clay subgrade, so even rock cuts need a full ballast and subballast section. In the finite-element breakdown (Figure 2-34), the subgrade supplies about 40–43% of total track elasticity, the track bed 20–22% and the rail pads 25–34%.

Figures · click to zoom

After Profillidis (2014), §8.2.2–8.2.3, §8.4.9, §8.8, Table 8.1, Figs. 8.12, 8.16. Full reference under Supplemental Sources.

Measuring track modulus: method, slack and nonlinearity

Track modulus lumps the fastenings, ties, ballast, subballast and subgrade into one spring. It cannot be calculated from the components. It has to be measured. Selig and Waters describe three methods:

• Single point load (preferred). Measure the maximum rail deflection under one load: u = [P / (2 y_max)]^4/3 / (64 EI)^1/3.
• Deflection basin (Talbot). The load equals u times the area of the deflection basin. This averages over a length of track.
• Multi-axle vehicle. Convenient, but it relies on superposition.

Remove slack. With all three methods, use the difference between a light seating load and a heavy load. This removes gaps and slack under the ties. Load–deflection curves are strongly nonlinear, and more so as track settles, so always report the load levels with the modulus.

Field values and correlation. At eight U.S. sites, u ranged from 2,200 to 7,000 psi. CPT soundings to at least 16 ft below the tie, back-analyzed with a multilayer model, gave a subgrade resilient modulus of about E_r ≈ 14 q_c for all the data, or about 21 q_c with outliers removed. This makes CPT a practical way to estimate track stiffness along a corridor.

Sensitivity. In the model, fastener stiffness had the largest effect on u, but it varies little in real track (about 10%). Ballast modulus, ballast depth, tie spacing and rail section each changed u by 10–20%. The subgrade remains the most influential variable in the field.

After Selig & Waters (1994), chs. 5 and 10. Full reference under Supplemental Sources.

Ballast cleaning trigger, renewal cycle and reuse

• Cleaning trigger: ballast cleaning machines are used when grains smaller than 0.87 in. make up more than 30% of the ballast volume. The cleaner removes all stone smaller than 1.4 in. to a depth of 10 in. below the tie, at about 1,300 ft of track per hour.
• Cycles on main lines: tamping every 3–5 years; ballast replacement every 15–30 years depending on stone strength and traffic (15–20 years on high-speed lines, against about 50 years for concrete ties and rail). Ballast stone tolerates only a limited number of tamping cycles.
• Longer life: harder stone, a wider ballast section and more uniform compaction lengthened the Swiss tamping interval from 4 to 7 years and cut maintenance cost by 40% over 10 years.
• Reuse: fouled ballast removed from track weighs about 106 pcf against 94 pcf new. If strength remains, washed spent ballast can be reused as subballast or formation layer on secondary lines, or as ballast on lines under 87 mph after treatment. France removes about 1 million tons per year this way; disposal costs (up to €60/t) often make reuse economic.
• Vegetation contaminates ballast with debris that blocks drainage, accelerates tie decay by root expansion and hides defects from visual inspection.

Figures · click to zoom

After Profillidis (2014), §12.7, §16.8–16.9, §16.12, §9.16, Figs. 9.21, 16.10. Full reference under Supplemental Sources.

Fouling sources, the failure zone, and choosing a repair

Fouling comes mainly from particle breakdown under traffic, plus abraded tie material, spilled lading (coal, grain), wind-blown sand and silt, slurry pumped up from the subgrade, chemical weathering, and maintenance tools (tamping accelerates breakdown). Gradation testing shows a failure zone. Once material finer than ¾ in. passes about 25% (the ideal point to undercut) and heads toward 40% (action mandatory), degradation accelerates sharply, typically within 500 MGT. Ballast next to an active mud spot is usually near its own failure zone. Visual inspection is not enough; use sieve analysis or GPR.

Undercut to about 12 in. below the ties, with the bar inclined to leave a 2% slope that drains to ditches or shoulders. Plan depths so every track drains on multiple-track territory, and place waste where it neither blocks drainage nor loads the shoulder. Undercutting without restoring lateral drainage gives minimal benefit. Watch for buried fiber. A documented move from reactive to planned ballast management cut ballast-related slow orders from over 600 to under 20 per year, ballast derailments from about 8 to about 0, and spending from over $250M to under $100M per year.

Figures · click to zoom

After Geotechnology for Railroaders Program, Module 3; §06–07, §13. Full reference under Supplemental Sources.

What fouling does to drainage and settlement, and how to limit it

Settlement. In box tests, settlement increased with fouling index. A small gap under the tie (0.039–0.16 in. of lift-off between loads) increased settlement by about an order of magnitude and accelerated both fouling and tie wear. Highly fouled ballast that settles quickly after tamping is a sign that it needs cleaning.

Sources vary by railway. British Rail attributed its fouling as follows:

• External sources (spillage, airborne): 52%.
• Attrition: 21%.
• Tamping: 20%.
• Delivered with the ballast: 7%.

North American sites were dominated by ballast breakdown (see above). Sample before assuming the source. In BR tests, 20 tamping insertions broke 15–45% of the 1½–2-in. particles, and each tamp produced 4–9 lb of material finer than ½ in.

Not all fines are equal. Sand and gravel fouling is tolerable. Silt forms an abrasive slurry. Clay fouling may require 100% ballast replacement.

Six ways to limit fouling:

• Deliver ballast clean.
• Limit spillage from cars.
• Tamp only when geometry requires it.
• Specify attrition-resistant stone.
• Keep the ballast drained.
• Provide a blanket over fine-grained subgrade.

Undercut or clean at a fouling index of about 40 if the ballast is cleanable. Plow or sled it out if it is fouled with wet, plastic fines.

After Selig & Waters (1994), chs. 8, 13 and 16. Full reference under Supplemental Sources.

Groundwater depth, crossfall and geotextile filters

UIC guidance is that the groundwater level should be at least 2.6 ft below the top of subgrade. Where it is not, it must be lowered with ditches or deep drains. Individual railways set their own minimum depths below rail or formation (Figure 2-37). Even where groundwater is deep enough, hydrogeological conditions are not rated “good” unless drainage devices are provided and the top of subgrade has a 3–5% crossfall. Sites with large seasonal groundwater fluctuation need a separate study, where a sand filter or geotextile may be warranted.

Geotextiles (polypropylene or polyester, 0.016–0.12 in. thick, woven or non-woven) separate, reinforce, filter and drain. They go under the subballast, never directly under the ballast. They do not replace ballast or subballast for load spreading: railways that laid a geotextile and also cut ballast thickness or omitted subballast saw the ballast puncture the fabric and lost the crossfall. A finite-element study found about a 10% reduction in stress at the top of subgrade from the geotextile. Filter permeability criteria quoted by the book:

where k_g = geotextile permeability (ft/day), t_g = geotextile thickness (in.), k_s = soil permeability (ft/day) and d₅₀ = soil 50% passing size (in.).

Figures · click to zoom

After Profillidis (2014), §9.4, §9.14.1, §9.15, Eqs. 9.13–9.14, Figs. 9.3, 9.4, 9.20. Full reference under Supplemental Sources.

Internal drainage on multiple tracks, and why it blocks

On double track, each track drains laterally away from the centerline. On three or more tracks, the outside tracks drain outward and the inner tracks drain into longitudinal trench drains, with the subgrade and sub-ballast surfaces sloped to suit. Ballast-deck bridges need their own designed drainage. Lateral drainage blocks when there is no cross-slope, when ballast pockets have formed, or when material and vegetation build up between the ballast toe and the ditch. Water trapped in the ballast is the leading cause of mud-fouled ballast. Groundwater seepage, where the subgrade sits at or below a fluctuating water table, is the second.

• Move water away from below the ties as quickly as possible.
• Keep ballast and shoulders clean and permeable.
• Keep the subgrade dry; drain ballast water pockets.
• Do not let water pond in ditches, especially in side-hill cuts.

Figures · click to zoom

After Geotechnology for Railroaders Program, Module 4; §03, §05–06. Full reference under Supplemental Sources.

Can the ballast section drain the design storm?

Three requirements. Track drainage works only if all three are met:

• The ballast is clean enough to pass water as fast as it enters.
• The subballast and subgrade surfaces slope to the sides.
• An outlet carries the water away. A section with no outlet is a bathtub.

Size the system for the highest inflow expected over its design life, not the average.

The center of the track ponds first. On a low-permeability subgrade, water flows mostly sideways. About two-thirds of it leaves through the shoulders. The zone within about 12 in. of the track centerline has the longest flow path and saturates first, so it sets the critical rainfall rate.

Table 2-SW5 gives the critical rainfall rate for several ballast conditions. The cases assume 12 in. of ballast below the tie and 6 in. of silty-sand subballast on an impermeable subgrade.

Double track that drains to one side only has about half these critical rates.

How to use it.

• Compare the critical rate with the design rainfall intensity over the planned cleaning cycle to set an allowable fouling index.
• For moderately fouled track, shoulder cleaning beats undercutting.
• For highly fouled track, undercutting is better, because water under the center would otherwise have to pass through fouled ballast to reach the cleaned shoulders.
• Full cleaning is best.

After Selig & Waters (1994), ch. 13. Full reference under Supplemental Sources.

Material parameters and typical European cross-sections

Values converted from kp/cm² by the authors (1 kp/cm² = 14.22 psi). These are model inputs from UIC test programs, not design values for a specific site.

The cross-sections below show the European layering in practice: 3–5% crossfall on the subballast and subgrade, 2:3 ballast and fill side slopes, and 12 ft track centers to the edge of the formation on a double track with drains and signal/electrification foundations at the shoulder.

Figures · click to zoom

After Profillidis (2014), §8.4.6, Table 8.2, §12.6, Figs. 12.7, 12.9. Full reference under Supplemental Sources.