Sizing a Swale or Roadside Ditch

One Ditch, Two Sums, and Only One of Them Is Yours

A plant hire depot laid 2,400 m² (about 26,000 sq ft) of new concrete hardstanding across what had been a grazed paddock and took the runoff into a grass ditch along the boundary fence — roughly 600 mm (2 ft) across the bottom, banks left at whatever angle the bucket walked out, running on the 3% fall the paddock already had, out to a connection into the highway authority's ditch that had been permitted before any of it was dug. It held through a wet autumn without a mark on it. The following July a short, violent cell sat over the yard for twenty minutes, and by the time anyone walked the fence line the flat bottom had a gully 300 mm (12 in) deep cut straight down the middle of it and a fan of the depot's own fines was spread across the authority's ditch, which is the sort of thing that gets a permitted connection reopened as a subject.

Nothing was wrong with the capacity. That section on that grade would have carried several times the flow the storm produced without the water coming close to the top of either bank, and if the only question asked had been whether the flow fits, the ditch passed comfortably. What it had never been given was the second sum. Water moving fast enough to cut through an established sward does not have to be deep, and a section that is generous in area and steep in grade is exactly the arrangement that produces speed rather than depth. Capacity and velocity fail in opposite directions, so a channel checked only for capacity has been checked for one of the two ways it can go wrong and cleared on that basis.

The shape of the work follows from a division that is worth stating plainly, because it decides which numbers are worth arguing about. Peak flow is not a design decision. It is handed to you by the area that drains to the point being sized, the surfaces inside that area, and the storm the jurisdiction has told you to design against — you can measure it more carefully or less carefully, but you cannot make it smaller by drawing something different. Everything downstream of that number is design: bottom width, side slopes, longitudinal grade, lining. So the job is a loop rather than a calculation. Fix the flow, take a trial section, test it for capacity and again for velocity, move whichever geometric variable is actually free on this site, and go round again until both checks pass at once.

One Channel, Two Values of n

Roughness for a grass lining is not a property of the ditch; it is a property of the sward on the day, and the sward changes. Turf mown a week ago behaves nothing like the same channel in late summer with a long, dense, partly laid cover in it, and the gap between them is a factor of several on n rather than a rounding. One value therefore cannot be conservative for both checks, because the two want opposite errors. Capacity is worst when the channel is roughest, since that is when the flow runs deepest and nearest the top of the bank; velocity is worst when it is smoothest, since that is when the same flow runs fastest across the soil. Run capacity at the high end of the range the cover will occupy and velocity at the low end, and accept two runs rather than one.

There are two established ways to finish the erosion side of that, and they are not interchangeable. The older one, which a great many local manuals still use and which the calculator on this page follows, compares the computed mean velocity against a permissible velocity for the lining and soil combination. Those permissible figures, and the roughness values that go with them, descend from USDA Natural Resources Conservation Service work on grassed waterways — Conservation Practice Standard 412, Grassed Waterway, and the Grassed Waterways chapter of the National Engineering Handbook Part 650, Engineering Field Handbook, are the documents that carry the practice. In that tradition n is not a single tabulated number at all: it is read from a retardance curve against the product of velocity and hydraulic radius, which makes the roughness implicit in the answer and the solution iterative even before you have thought about the season.

The newer approach, which most highway work now runs on, replaces permissible velocity with permissible shear stress. FHWA's HEC-15, Design of Roadside Channels with Flexible Linings, works from the shear the flow applies to the boundary — a function of depth and slope — and compares it against a shear the lining and soil can take, with the grass roughness treated as varying with flow depth rather than fixed. The physical argument for it is straightforward: it is force on the soil particle that moves the particle, and two channels can share a mean velocity while applying very different shear to their beds. If the manual your submission is reviewed against names HEC-15, do it that way and do not present a velocity check instead; if it names a permissible-velocity table, use the table. What you must not do is take the permissible figure from one framework and the demand from the other.

Whichever framework applies, the case that governs is usually the first large storm after construction, and it is the one least often checked. A ditch shaped and seeded in October has no cover in November, and the permissible figure for a newly seeded sward on disturbed soil is a fraction of the one for a dense established cover on the same ground. That is the whole justification for a rolled blanket in a channel that will eventually be perfectly happy as plain turf: temporary reinforcement bought to cover a gap in time, not a permanent upgrade to the lining. Where the design velocity sits close to the permissible even at maturity, the temporary product becomes a permanent one and the conversation moves to turf reinforcement mats or a hard lining for that length.

Beside a Road, Hydraulics Is Not the Only Client

A ditch inside a highway reserve answers to a second document with its own opinions about the section, and that document usually wins. The AASHTO Roadside Design Guide treats the ground beyond the edge of the travelled way as a recovery area and sets out which combinations of front slope and back slope let a vehicle that has left the carriageway cross the channel and slow down rather than trip, dig in or roll. A narrow, deep, steep-sided section is often the hydraulically neat answer and one of the least acceptable answers on that measure — the shapes that perform well for traversability are broad and rounded, which pushes you toward width for a second independent reason. Where the section cannot be made traversable the alternatives are shielding or relocation, both expensive enough that redesigning is generally cheaper. The AASHTO Drainage Manual covers the hydraulic side of the same work, and outside North America the equivalent is the road authority's own manual — Austroads Guide to Road Design Part 5, Drainage, in Australia and New Zealand.

Maintenance access is the constraint that quietly decides whether a grassed ditch is still a grassed ditch in five years. The roughness the capacity check was run at assumes a sward somebody is cutting; a channel too steep-sided to mow safely stops being mown, fills with scrub and self-seeded willow, and its real roughness drifts far above anything on the drawing while its capacity drops with it. Banks a tractor can work along, an access point at each end, and a grade that does not leave standing water between cuts are all part of the design rather than an operations afterthought. Standing water is its own failure: below a certain fall a grassed channel stays saturated, loses the sward that was providing both the roughness and the erosion resistance, and becomes a reed bed with a different Manning's n from the one that was submitted.

The last external constraint is that the ditch grade is not a free line between two endpoints. Every driveway culvert, field entrance and crossing along the length fixes an invert, and those inverts were often set by somebody else years ago. The design grade is the line through the pins you have inherited, and where one sits too high the reach above it does not drain at the grade you assumed whatever the profile drawing says. Shoot the existing inverts before drawing that profile: a single legacy culvert 150 mm (6 in) proud can flatten a reach into the range where the sward will not survive the winter.

Freeboard, Bends and the Storm Bigger Than the One You Designed For

The depth that goes into the capacity check is water, not excavation. Bank height has to exceed it by whatever freeboard the governing manual requires, and that figure is a requirement to be read rather than a margin to be judged — a section built exactly to the design flow depth has none at all, which makes the design storm itself the overtopping storm. Two things then eat what freeboard there is without anyone deciding they should: newly cut banks settle and slump through their first wet season, and sediment deposits in the flat bottom of a treatment-shaped swale, which is precisely what that wide flat bottom was drawn to encourage.

Bends and tailwater both break the uniform-flow assumption the capacity check rests on, in opposite directions. Around a curve the water surface superelevates, standing higher against the outside bank than the computed normal depth, and the outside of the bend needs both the extra height and, frequently, a lining upgrade the straight reaches did not — HEC-22 and the road authority's drainage manual carry the treatment. At the downstream end, a receiving ditch, culvert or watercourse that is itself running full in the design storm raises the water level in your reach above normal depth and pushes a backwater curve upstream, so a system sized as though it always discharges into open air will surcharge from the lowest point on the site instead. Where the receiving channel has a known high-water level, use it; where it does not, the assumption of free discharge is an assumption and should be recorded as one.

Finally, decide deliberately where the water goes when it exceeds the section, because it will exceed the section eventually. Every ditch has an overtopping path whether or not anyone chose it, and the difference between a designed one and a discovered one is usually the difference between water crossing a yard and water entering a building. Identify the low point on each bank, confirm that the flow leaving there runs somewhere tolerable, and where it does not, either raise that length of bank or provide a defined spillway at a point of your choosing. This is a five-minute exercise on a profile and it is the single most valuable thing on the drawing the first time a storm arrives that is larger than the one the ordinance nominated.

What Ends Up in the File

A drainage submission is judged on whether a reviewer can reproduce your numbers, not on whether they are right, and those are different tests. The catchment plan wants the divide drawn on it and the offsite area shown separately with its own coefficient. The time of concentration wants its segments listed — length, slope, surface and method for each — rather than a single figure. The intensity wants its source named and dated, because the published curves are revised and a submission against a superseded dataset gets returned. The section wants both values of n written down with which check each was used for, since a reviewer who sees one value knows immediately that only one of the two checks was really run. And the erosion check has to say whether it is a velocity comparison or a shear comparison, and against whose permissible figures.

Once it is built, the file that matters is the as-built: flow line levels at every station and grade break, the bottom width where the machine left it rather than where the drawing put it, the crossing inverts as found, and the dates the cover was seeded and accepted as established. A ditch that silts, scours or ponds three years later becomes an argument about whether it was built as designed, and that argument is settled in ten minutes by a level book and a handful of photographs taken before the first storm — or not settled at all, which in practice means settled against whoever dug it.