Meeting a Stormwater Runoff Condition on a Tight Site

A Rate Is Not a Volume, and the Condition Only Names One of Them

Q = C i A gives a peak. It is the flow at the top of the hydrograph, and it is the right number to hold against a condition expressed as litres per second, because both are rates and they are directly comparable. What it cannot do — and this is where small submissions lose a fortnight — is tell you how much water is behind that peak. The Rational Method has no time axis. It is a snapshot at the moment of maximum flow, and a snapshot of a flood does not tell you the size of the tank.

The volume comes from routing: an inflow hydrograph for a storm of a given duration, an outflow fixed at the permitted rate, and the difference between them accumulated over time. Run it once and you have the storage for that duration. Run it for a spread of durations and the largest answer is the one you build, which is the whole reason the critical duration for storage is almost never the critical duration for peak flow. The short violent burst makes the biggest instantaneous flow; the two-hour storm makes the biggest tank, because the outflow only drains 5 litres per second no matter how long the rain goes on and a longer storm gives it more to get through. USDA NRCS Technical Release 55, Urban Hydrology for Small Watersheds, and the rainfall-runoff methods published with the Flood Estimation Handbook are two of the routes to the hydrograph; the local ordinance will name one.

The permitted rate itself is worth understanding before it is designed to. Where a condition sets the cap at the greenfield runoff rate, that rate is estimated by a named method — in the United Kingdom, Institute of Hydrology Report No. 124, Flood estimation for small catchments, is the usual one for a plot this size — and on 0.19 hectares it produces a fraction of a litre per second. Nobody builds a flow control that small, because the opening it needs is smaller than the debris that will arrive at it. So sewerage undertakers and lead local flood authorities apply a minimum practicable discharge rate instead, and it is that minimum, not the greenfield estimate, that usually appears in the condition. The 5 litres per second on this site is a floor, not a calculation, and knowing that changes how much argument is available later.

Two things belong in the file before any of the arithmetic. Whether water crosses onto the site from outside it, since offsite catchment has to be carried through the development whether or not the condition mentions it — here the railway embankment behind the yard falls away, which is worth recording as a finding rather than leaving as an assumption. And the discharge hierarchy: Approved Document H, Drainage and waste disposal, puts infiltration first, a watercourse second and a sewer last, so a condition permitting a sewer discharge does not excuse you from showing the ground was tested.

What a Vegetated Roof Is Actually Credited With

The roof is 900 m², and about 620 m² (6,700 sq ft) of it can be vegetated once the plant deck, the rooflights, the perimeter fire and access zones and the outlet surrounds are taken out. Take a manufacturer's rated retention of 30 mm at saturation and that footprint holds a little under 19 m³. Against 80 m³ of required storage, the roof on its own is not the answer to this condition and no amount of specification improves that, because the yard is 700 m² of impermeable surface that does not drain across the roof and never will. This is worth establishing in the first hour rather than the third week, and it is the single most common misreading of what a green roof does for a drainage submission.

The 19 m³ is also a ceiling rather than a design value. A rated retention depth is what the medium and the composite hold starting dry and wetted to saturation, and the design storm has no obligation to arrive on a dry roof. In a wet autumn the medium may already be at capacity when the rain starts, at which point the credited storage is zero and the roof behaves as a flat roof with a rougher surface. CIRIA C753, The SuDS Manual, treats interception — the volume lost before any runoff leaves the site at all — as a design objective in its own right, and how much a vegetated roof may claim against it is set there rather than in the product literature. The component testing sits elsewhere again: ASTM E2398/E2398M for water capture and media retention of geocomposite drain layers, the FLL Guidelines for the Planning, Construction and Maintenance of Green Roofing, and BS 8616 for substrate performance parameters.

None of which makes the roof pointless. It reduces the volume that reaches the tank, which shrinks the tank; it delays the roof's contribution so that it arrives after the yard's has passed, which flattens the combined hydrograph; and on a site being assessed against biodiversity or urban greening policy it is doing three jobs at once. What it does not do is give you a discharge rate. A vegetated roof has no throttle in it. Water leaves it when the medium is full, at whatever rate the composite and the outlets pass, and if the condition is a rate cap then something downstream still has to hold the rate — which is the subject of a later section and the thing most often missing from a submission that leads with a green roof.

The Layer That Decides Whether the Roof Drains At All

Beneath the medium sits a drainage and water-retention composite: a dimpled core that keeps a reservoir standing in its cups and passes everything above that to the outlets, under a filter fabric that stops the fines washing down and blinding it. It is the least visible layer in the build-up and the one that decides whether the roof is a retention device or a bath. Get it wrong and the medium sits permanently saturated, the planting drowns, the dead load climbs to the case the structural engineer was told was the worst case, and the retention volume you claimed in the submission is not available on any storm because it was never empty.

Ordering it is the simplest arithmetic on the page and still gets undercooked. At 620 m² of vegetated roof against a common roll coverage near 9.3 m² (about 100 sq ft), the base count is 67 rolls before a single lap or cut. Manufacturer laps between rolls, cuts around every outlet, upstand, rooflight kerb and inspection chamber, and the trim at the perimeter all come off the top of that, so the delivered figure is always higher than the area sum and the product's own installation guide is what sets by how much. Two site details are worth fixing at the same time: the falls have to be confirmed on the deck rather than assumed from the drawing, and the waterproofing under all of it gets a flood test to ASTM D5957, Standard Guide for Flood Testing Horizontal Waterproofing Installations, before a single roll goes down — because it is the one layer nobody can reach again once the overburden is placed.

The Yard as a Tank, and the Clock Running Underneath It

Which leaves the ground. Eighty cubic metres of water in clean open-graded stone at a permitted void ratio of 0.35 needs about 230 m³ of stone, and spread over the 340 m² of yard that is actually available that is a reservoir roughly 675 mm (26.5 in) deep below the pavement construction. Stone is not the only way to build a void — geocellular units are rated by their manufacturers near 95 per cent void against roughly a third for clean aggregate, and CIRIA C737, Structural and geotechnical design of modular geocellular drainage systems, is where the loading and long-term serviceability of that choice are dealt with. Either way the depth is buildable on this site, and either way compliance turns on a number nobody has yet measured: what the subgrade at the bottom of that excavation will take.

The measurement has to happen at the elevation the excavation will finish at, not at the surface, because the horizon two feet down owes nothing to the topsoil above it. ASTM D3385, Standard Test Method for Infiltration Rate of Soils in Field Using Double-Ring Infiltrometer, is one accepted route; in the United Kingdom the soakage pit procedure in BRE Digest 365, Soakaway design, is the conventional one, and it is worth knowing that the two methods are not interchangeable in what they credit. BRE 365 sizes on the internal surface area up to half the effective storage depth and conventionally takes no credit for the base, on the reasonable ground that the base is where the silt ends up; the drawdown check below is a one-dimensional vertical sum that credits only the base and takes nothing from the sides. They are different conservatisms, not different accuracies, and a submission that mixes them without saying so invites the reviewer to pick whichever is worse.

Whatever the test returns, it gets divided before it is designed to. Every manual applies a safety factor to a measured infiltration rate, and the factor is larger where the testing was thin, where the soil is variable, or where the contributing area is large relative to the infiltrating footprint — which on this site it is, because 1,600 m² of hard surface is draining into a 340 m² reservoir. Suppose the test comes back at 8 mm/hr on a silty clay and the manual asks for a factor of three on a single test. The design rate is 2.7 mm/hr, the water depth to be lost is 675 mm at 0.35 voids, which is 236 mm (9.3 in), and the drawdown runs to something close to 88 hours. Against a permitted maximum of two days, that is not a marginal fail; it is not a scheme.

There are two ways out and only two. Spread the reservoir shallower over a bigger footprint, which this site does not have. Or fit a perforated underdrain at a raised invert, so that only the stone below the pipe has to infiltrate and everything above it leaves through the pipe — at which point the depth on the clock is 200 mm (8 in) of stone rather than 675, the water in it is 70 mm, and the drawdown comes back around 26 hours. That is the standard remedy for a marginal soil and it is a legitimate one, but it converts the scheme from full infiltration to partial infiltration, and the underdrain now needs an outfall that is itself throttled to the 5 litres per second in the condition. The check has not gone away. It has moved to a different part of the section, and a different part of the submission.

1. Fix the formation level on a section before testing, and test at that level — a rate measured at 300 mm down describes soil you are going to dig out.
2. Test more than one location across the footprint, and record the coordinates and the date with each result rather than reporting a single averaged figure.
3. Note the depth to the seasonally highest groundwater table at each hole; the vertical separation the manual requires below the reservoir is a hard constraint on depth, independent of drawdown.
4. Apply the safety factor the governing manual specifies for the number of tests you actually did, not the one that applies to a full investigation.
5. Protect the formation once it is exposed: keep tracked plant off it, excavate the last lift with a toothless bucket, and accept that a smeared subgrade is not recovered by scarifying it afterwards.