Maintenance is the question every technical review ends on, and it deserves a straight answer. The short version: debris stays on the surface where a sweeper takes it. Nothing wears – the tray and the filter clog slowly by design, lift out, and are expected to run far longer between services than a side-entry pit takes to fill with sediment, debris and organic matter. A blocked node does not block the street, and a whole run failing at once is remote. And in a true flood, the road reserve does what it has always done: convey floodwater downhill, along the relief path we design in.
The system is routinely mis-read as a tank with a tree on top. It is the other way round, and almost every maintenance question answers itself once the arrangement is clear. The full flow path, step by step, is on How it works.
On frequent rainfall – the great majority of events – very little storage is called on. Flow passes through the manifold into the media and soil cells and keeps moving down. The tanks are barely engaged.
On larger events the soil cells may saturate, and only then do the tanks take up the volume the profile can no longer accept. The relief line engages later still, and only once storage and substrate are both satisfied. On an ordinary event it does not engage at all.
Meanwhile the tree transpires and the surrounding profile draws moisture. The store is drawn down slowly and continuously between events rather than held. What sits under the street most of the year is moist media and moist soil – not water.
Lined tanks are a variant, not the default. Tanks are unlined unless there is a reason – retention for harvesting and reuse, or a site where infiltration to the underlying profile is not permitted. That is the only configuration in which the system holds standing water for a sustained period, and it is a design election made deliberately.
Kerb-line tree pit inlets are not new, and most asset teams have some history with them. That history is usually the reason for the scepticism, and it is fair – so it is worth being precise about what is different here.
Runoff enters through an open letterbox throat or a pipe connection from the kerb. Hydraulically it works. But an open throat admits whatever is at the kerb – leaf litter, wrappers, butts, silt – straight into the subsurface structure. Retrieving it means opening the asset.
Water enters through a void structure, not an aperture. The ETI is a sealed, removable tray; the High Flow Inlet admits water through a porous concrete body. Gross pollutants and coarse material do not enter the subsurface at all. They stay at the kerb line, on the surface.
Debris on the surface is collected by a machine already scheduled on that street. Debris inside a subsurface structure needs a crew, an opening operation and a confined-space consideration.
The closed inlet is what moves the dominant maintenance action from the second category into the first. Everything else in the regime – the consumable filter, the removable tray, the modular unit – deals only with what gets past that first line.
This is a design position, not a comparative performance claim: we have not run side-by-side testing against other inlet products, and we name none. The mechanism is straightforward, and the difference in what reaches the subsurface follows from it.
"What is your redundancy provision, short of block?" It is the right question, and the answer is not that the system will not block. Redundancy operates at three different scales, and the third one is the one authorities ask about.
Each node serves its own micro-catchment. Flow a blocked node cannot take continues along the kerb line to the next node. There is no single point of entry whose failure fails the run – which is the defining hydraulic difference from a conventional side-entry pit.
Above the design storm, the system is meant to be exceeded. Frequent events are captured in full; larger events pass to the designed overflow route once storage and substrate are satisfied. Because the nodes are distributed, that overflow is never concentrated at a single point.
For the case where every unit in a run is out of service at once, a surface relief path is provided as part of the design. Each site is unique and the provision is specific to it – sized so the corridor has somewhere to send water if the whole system were unavailable, a case that is already remote with several nodes in a run. And in a true flood, the road reserve does what road reserves have always done: convey floodwater downhill.
We do not argue against this requirement. A relief provision that never operates costs a fraction of the conveyance network it displaces, and it turns an unresolvable argument about failure probability into a bounded design item. We would rather include it than win the argument.
"Everything breaks" is the correct engineering assumption. What matters is whether getting to the broken thing needs a specialist.
The ETI is a removable galvanised tray, Class D rated and fully trafficable, that lifts out on a locking arrangement. No excavation, no proprietary tooling.
The ARI Water Filter beneath the tray consumes itself protecting the void structure below. It is replaced by removal, not cleaned in place – a part swap, not a procedure.
Where a continuous porous kerb run is renewed as a length, a spaced node is lifted and replaced individually. Worst case at any single point is one unit.
Installations can be detailed with inspection access on request, so the manifold and storage can be inspected by camera – giving asset teams a direct view of condition and root interaction over time rather than an inference from flow data.
On private land the obligation sits with the Owners Corporation or landholder. In a road reserve it sits with the authority, as it does for any drainage asset. There is no ambiguity to resolve later and no residual obligation to BGI. Owning the obligation does not mean doing the work, either – servicing contracts out like any drainage function, and Owners Corporation installations already run exactly that way: serviced under contract by a company that carries the plant.
A maintenance management plan accompanies every installation and is provided with the approval documentation, whatever pathway the project takes. It is site-specific, not a generic manual – the regime is built from the project's own catchment, sweeping regime and node arrangement. Asset teams see the servicing regime and the deviation thresholds before the asset is handed over, not after.
Nothing proprietary is required to service it. Every maintenance tier uses plant a council or contractor already owns and operates – a vacuum sweeper, a pressure unit, small plant for a module lift. One vehicle can carry the lot: a vacuum sweeper truck with a pressure unit and spare trays on board services a whole run in a single pass. That was a design constraint from the outset. A maintenance regime that depends on a manufacturer's service contract does not resolve a liability; it transfers one and adds a supplier dependency on top.
Gross pollutant traps, sediment basins and the vacuum-truck cycle that serves them exist because the network concentrates and conveys. Retain runoff at source and the load reaching them falls with it.
Establishment watering of street trees is a recurring, labour-intensive cost with a poor success rate. Passive irrigation at every frequent rainfall replaces the truck run.
A tree that fails in a compacted pit is replaced, and charged to the same budget. Root-zone water, oxygen and soil volume are the cheapest tree-asset insurance available.
The dominant action is vacuum street sweeping – already scheduled on the street for reasons unrelated to drainage.
Entering a stormwater pit is confined-space work under the WHS Regulations. A clean is not a task, it is an operation:
Because debris never enters the subsurface, the dominant action is the vacuum sweeper run the street already receives. The tiers above it – a tray lift, a filter swap, a pressure wash – are one-person surface tasks with no confined-space entry, no eductor truck and no extraction load to dispose of.
The comparison is not a percentage we have invented. It is a category difference in what the work is.
Each Matter SensAI unit sits behind the water ramp of the inlet it serves, configured with a baseline flow curve calibrated to that inlet unblocked. Measured inflow is compared against the baseline continuously, and a fall below the curve raises an inspection notification – on a tablet or a phone – naming the location and the indicated action, before it becomes a blockage.
The more useful property is spatial. One node deviating while its neighbours hold baseline points to a local cause – a bag over the tray, dumped material, a damaged unit. Upstream, subject and downstream deviating together points to a street-scale condition. That cross-reference is the difference between lifting every unit in a run to find the blocked one, and being told which one it is.
| Deviation from baseline | Action | Equipment | Crew |
|---|---|---|---|
| Minor | Vacuum street sweeping through the zone | Council sweeper | Existing sweeper run |
| Moderate | Lift and pressure-wash the ETI tray | Pressure unit | One |
| Significant | Replace the ETI tray insert or ARI Water Filter | Replacement consumable | One |
| Major | Pressure-wash the EnviroKerb™ High Flow Inlet | Pressure unit | One |
| Critical | Lift out and replace the EnviroKerb™ unit | Replacement module, small plant | Two plus plant |
Where this framework stands. The tiers are calibrated from the hydraulic behaviour of the inlets, not from a field-validated fault-classification model. Threshold calibration and the deviation-to-cause mapping need instrumented field data, and that pilot is part of the research programme below.
"How often does it need servicing" assumes a scheduled inspection round – every unit checked so the blocked one is found. This system does not work that way. Each node is monitored against its own baseline, and the measured deviation flags which node needs attention and how urgently. Servicing is by exception, not by schedule – that is what makes the regime proactive rather than periodic. What varies from street to street is how often flags occur, and three things drive it.
Land use, tree litter, traffic and construction activity upstream. This dominates everything else, and loss of capacity tracks cumulative load rather than elapsed time.
Fine-particle capture is what matters, and sweeper type changes it materially: published removal efficiencies sit near 47% for mechanical broom against 74% for regenerative air. The same street on different plant implies a different node interval. Vacuum or regenerative air plant only – a rotary broom pushes fines down into the voids, and the manufacturer's schedule excludes it.
Closer spacing puts less load on each node and lengthens the interval between services, at higher capital cost. It is a dial, and we set it per street.
The only field test a crew needs. Tip a bucket of water in. If it disappears, the inlet is working. If it is slow, bring the next clean forward. That is the whole test, straight from the manufacturer's maintenance schedule.
Hold a few spare trays. A flagged tray is then exchanged on the spot in minutes and flushed back at the depot at a convenient time – no traffic management, no wet work in the road reserve. The manufacturer rates this the single most effective way to keep maintenance cost down.
What you get, per project. An indicative service expectation derived from your catchment and sweeping regime, a baseline calibrated at commissioning, and thereafter a measured record from the monitoring data rather than a modelled one. The number improves the longer the asset runs.
Three findings hold consistently across the permeable pavement research. Capacity declines along an exponential curve with cumulative sediment loading, so most of the loss happens early. Clogging progresses from the upgradient edge of a permeable surface toward the downgradient edge, so a continuous run does not fail evenly – it fails at one end and is renewed as a length. And restoration is partial and method-dependent: combined pressure washing with vacuum suction restores a substantial share, surface vacuuming alone often does not, and where fines have migrated below the surface no surface method works.
None of that is a reason to avoid porous infrastructure. It is a reason to put the porous element where accumulation is concentrated at reachable points, where the part that consumes itself is a cheap consumable, and where the element that eventually cannot be restored is a module that lifts out.
Porous surfacing has been on United States highways since the 1950s, and the early performance problems that soured some road authorities in the 1970s were materials problems since engineered out. The porous asphalt car park at Walden Pond, laid in 1977, is still in service after more than forty-five years without full repaving.
But an urban street is not a highway. Part of why highway porous surfacing stays open is that high-speed traffic scours fines from the voids – which is why it is expressly not recommended for low-speed pavements. Australian streets are low-speed, high-deposition environments. They get the sediment without the self-cleaning.
That is the whole argument for the kerb line: put the porous element where it can be lifted, serviced and replaced – not spread across a trafficked surface where it cannot.
Each site is unique and the relief provision is specific to it – the surface relief path is designed in from the start, taking whatever form the corridor allows. With several nodes in a run, simultaneous failure is remote; we size the relief path anyway rather than argue probability. And in an extreme flood the road reserve does what it always has – conveys floodwater downhill.
Those devices generally admit water through an open letterbox throat, which lets gross pollutants into the subsurface structure. These inlets admit water through a void structure – a sealed removable tray and a porous concrete kerb body – so debris stays on the surface where a sweeper collects it.
The tray lifts out on a locking arrangement, the filter beneath it is a consumable that is swapped rather than cleaned, and a whole module can be lifted and replaced individually. Installations can be detailed with inspection access on request, so the subsurface can be camera-inspected.
Whoever owns the land, as with any drainage asset – Owners Corporation, landholder or authority. Which is precisely why it is designed to be serviced with plant that owner already operates. The work itself is routinely contracted out – Owners Corporation installations are already serviced this way, under contract by a maintenance company with the plant.
No. We provide the regime, the thresholds and crew training. No ongoing obligation to us attaches to the asset. Owners who prefer not to self-perform simply contract the regime out, as Owners Corporations already do.
Distributed. Each node has its own manifold, tanks and soil cells serving its own micro-catchment – there is no central chamber. On most rainfall the water is not sitting in the tanks at all; it is moving down through the filter media and structural soil cells toward the root zone.
In the default arrangement the tanks are unlined, so there is no sustained standing water to go anaerobic – they fill on a large event and drain down through the profile afterwards. Where tanks are deliberately lined for harvesting or reuse, water is retained and low dissolved oxygen should be expected; that condition is engineered on purpose in stormwater biofilters to drive denitrification. In either arrangement the root zone is not the saturated zone: water arrives oxygenated from above and filters down through the root mass, rather than pooling beneath it as it does at the base of a conventional tree pit.
Moisture in the profile, not litres in a tank. Water held in media and soil is drawn down at transpiration rates rather than conveyance rates, and moves slowly downward and laterally, perching where the geology allows. The absence of an obvious volume is not the absence of available moisture – this is the same principle that lets rehydrated rural landscapes carry through dry cycles.
There is no calendar interval to defend. Each node is monitored against its own baseline, and the measured deviation flags which node needs attention and how urgently – service by exception rather than by schedule. How often flags occur depends on sediment supply at the kerb line and on your sweeping regime.
It substitutes rather than adds. Source retention reduces the load reaching end-of-line assets, passive irrigation displaces manual watering, and the dominant action is street sweeping already scheduled for other reasons. A five-year whole-of-life costing against the displaced asset comes with every installation.
Before handover. It is provided with the approval documentation, whatever pathway the project takes.
The positions on this page rest on published permeable pavement, biofiltration and smart stormwater research, set out with full citations in the technical paper below.
Where the literature stops is the specific question of how to space, monitor and renew a discretised porous interface at the kerb line at least cost. We are quantifying that – service-spacing optimisation, restoration efficiency across repeated cycles, and sensor threshold calibration against observed obstructions – as an industry partner of the ARC Training Centre for Climate-Resilient Water, a national centre led by The University of Queensland with Griffith University and the University of Technology Sydney.
The practical consequence for an asset owner is straightforward: the service interval you are given is measured on instrumented assets rather than assumed, and it improves with every season of data.
Clogging, serviceability and sensor-driven O&M in kerb-line source control. Fully referenced, with every claim labelled by evidence status – established, BGI position or open question – so the assessing engineer can see exactly what is literature and what is our judgement.
We email the paper rather than host it, so we know who is assessing what and can follow up with the detail that matters to your project.
Free to request and share intact. Not for reproduction, adaptation or derivative use – see terms.
