A field study of manhole cover leakage

Sanitary sewer systems always carry extraneous flow other than the actual sanitary wastewater flow they are intended to transport (ASCE 2007). This is referred to as infiltration and inflow, or I/I. Infiltration is water that enters a sewer system from the ground through defective pipes, pipe joints, connections, or manholes (EPA 2014). Inflow is water that enters a sewer system from sources such as roof leaders, cellar/foundation drains, yard drains, area drains, drains from springs and swampy areas, manhole covers, cross connections between storm sewers and sanitary sewers, and catch basins (EPA 2014).

At the terminal point of the sanitary sewer system is the wastewater treatment plant. The treatment plant continually measures the flow entering the plant. This flow record will show a spike during a rain event, and elevated flows after the event ends. These flow records provide hard evidence of the presence of I/I in every sanitary sewer system.

I/I receives the engineer's attention for various reasons. One is the cost of treating this unwanted water (Sola et al. 2020). This is an expense that comes in two parts. First, the treatment plant has to be built with the capacity to handle the I/I flow. Second, there is the cost of treating the additional water. I/I also represents the state of repair of the collection system – high I/I implies a system in poor condition (EPA 2014), although this is an overly simplistic way of assessing one system against another. For example, a collection system in a high-water table environment will always look worse than a comparable system in a low-water table environment. I/I is also focused on because of the connection between sanitary sewer overflows (SSO's) and excess I/I (Day & Seay 2020).

While the hydraulic rules for sizing sewer systems are universal, the approach to estimating I/I varies by town. The two most common approaches are:

• not explicitly account for I/I but incorporate it into the design flows from buildings or areas (e.g. City of Brisbane, CA 2024).

• combine I/I into one area-based flow rate, e.g. City of Pueblo, CO (2024), uses 0.0003 cfs per acre as its I/I flow.

Another approach to account for I/I is:

• set I/I as a fixed percentage of the design flow, e.g. Southgate Sanitation District, CO (2024) uses 10% of the average flow as its I/I flow.

The continuous flow record at the treatment plant provides an opportunity to attempt to numerically separate the three contributing flows: sanitary, inflow, and infiltration. Over the years there have been a large number of papers published on this topic. In 2008 the EPA itself published a report on sewer design criteria and rainfall-derived infiltration and inflow (RDII) (EPA 2008).

Estimates of how much inflow is typically found in a system vary widely. In one study in Provo, Utah, Sowby & Jones (2022) estimated that direct inflow was only 1% of the annual volume. In Vila Real, Portugal, Bentes et al. (2022) estimated that inflow was 38% of the total flow. In a study of 10 sewersheds in Miami-Dade County, Florida, Chin (2023) found that inflow was significant in 9 out of the 10 sewersheds.

Many studies of inflow have concluded that flow from lateral connections, on private property, is the major source of inflow (Field & O'Connor 1997; EPA 2008). Systems that have historical building drainage connections therefore have large inflow rates. For example, a study by Jiang et al. (2019) in London, Ontario, estimated that foundation drain connections resulted in inflow being 85% of the RDII amount.

Another source of inflow is leakage from manhole covers (ASCE 2007). Although the inflow per manhole may be small, sanitary sewer systems do have a large number of manholes. In 2008 ASCE estimated that there were over 20 million manholes in the US (ASCE 2009). So, there is definitely the potential for manhole inflow to add up to a significant amount. While the maximum spacing for manholes is typically 100 m, the average spacing in a sanitary system will be much less. This is due to curved roads, intersections, grade changes, lateral connections and all the various other reasons that require manhole placement.

Winter et al. (2022) measured approximately 20 MH/ha in five municipalities in South Africa. New York City averages one manhole for every 33 m of pipe (estimate by the author based on NYCDEP numbers). Chicago and Boston average one manhole for every 48 m of pipe (estimate by author based on Chicago DWM and Boston BWSC numbers). Akron, Ohio, a smaller city, averages one manhole every 73 m (based on City of Akron Sewer Maintenance Division numbers).

Reducing I/I is a very expensive undertaking (Smith et al. 1991). In large part, this is due to the inaccessibility of the sewer system. The system is below ground and with a large portion of the system on private property. As most sanitary sewers are under streets, any rehabilitation of a section of the sewer will involve not only replacing the pipes but repairing the pavement. Working on private property to alleviate inflow is legally difficult, unless, for example, it can be shown that an illegal connection exists.

The only part of the system at ground level, and entirely on a public right-of-way, is the top of the manhole. This makes for easy access, and potentially, a cost-effective way to reduce some portion of the inflow.

However, given the large number of manholes in a sanitary sewer system, even a low-cost repair per manhole will still add up to significant cost, and considerable effort. As with any infrastructure program, planning is needed, and data regarding inflow rates are key to this planning effort. Data on manhole cover inflow rates are rare and are discussed further in the following. With this in mind, a field study was undertaken to measure manhole cover leakage rates in New York City and surrounding towns. The study took over two years to complete and tested 250 manhole covers.

One of the motivations to conduct this present study was the fact there is very little data on manhole cover leakage rates. ASCE (2007) noted that a laboratory study by Rawn (1937), who submerged the covers under 2.5 cm of water, found rates of 75.8–284.3 L/min.

A study by Johnson County Wastewater in Kansas in 2016, submerged covers under 7.6–15.2 cm of water and found leakage rates of 0–77.8 L/min (Vergara et al. 2018; Beck 2022).

In 2016 the San Antonio Water System tested manhole covers under 7.6–30.5 cm of water and reported leakage rates of 1.5–462 L/min (Lloyd et al. 2018). The San Antonio Water System now requires that sanitary sewer manhole covers pass a leakage test with leakage of no more than 3.8 L/min under 30.5 cm of water.

The cases above represent the only studies this author could find where manhole cover leakage rates have been measured, and the results published in a scientific journal or as a paper contained in a conference proceeding. All of the studies mentioned had two common components:

• they were conducted in laboratories or workshops and

• the manhole covers were submerged.

The Township of Scio in Michigan rehabilitated 48 manholes in 2006 in areas prone to flooding. This resulted in an estimated 56.3 L/s reduction in flow, or 70 L/min per manhole (Cox et al. 2008). The program involved replacing perforated manhole covers with solid manhole covers, improving the cover seal, repairing the manhole below-grade components, and in some cases adjusting the manhole elevation.

As manhole cover replacement/modification is growing as an I/I reduction strategy, there is now a commercial industry that manufactures various manhole sealing devices. Manhole cover leakage rates found on several industry websites range from 0.76 to 19 L/min, although the origin of these rates is not clear. Some websites report leakage rates an order of magnitude higher than this.

A priority of this project was to study manhole cover performance in-situ. As noted, previous measurements were made in laboratories or workshops. It was considered important that the covers were not disturbed by the measurement process, as this would significantly alter the results. This rationale came from observations made during the planning stages of this project, discussed in the following.

While sealing the entire rim would have been ideal, the road camber would make this difficult, as there will be a 2.5–5 cm elevation drop across the cover. This would have meant the lower portion of the rim would have to submerged under 2.5–5 cm of water.

Each manhole was recorded in one of three categories – ‘no leak’, ‘slow leak’ and ‘fast leak’. If water just ponded in the gap and remained, this was classed as ‘no leak’. If water went through the gap faster than it could be added, this was a ‘fast leak’. The rate at which water was added was limited to about 120 mL/min. Adding water faster caused issues from spillage. Leakage less than 120 mL/min was recorded as ‘slow leaks’.

During a rain event, there would be water entering through these holes. No specific testing was performed on the vent holes in this study. Unless a cover was submerged the amount of water entering these holes would be small, given the amount of runoff they could physically intercept, i.e. a 25 mm wide flowpath.

Of 14 manholes in the ‘fast leak’ category, seven were either new manholes or manholes identified as recently opened. The average leakage of the ‘slow leak’ category was 24 mL/min. The fast leak category had leakage rates that exceeded the ability to replenish the supply water, which was a maximum of 120 mL/min. Hence these are recorded as >120 mL/min.

Of the 250 manholes tested, two were tested more than once. These two manholes were new manholes that were tested when they were just installed, and again later. One manhole was on Morris Avenue in the Bronx, and one was on Skyview Drive in Stamford, CT. When initially tested these covers leaked significantly, and were in the ‘fast leak’ category. There was no grit or dust in the gaps of these new installations.

When Morris Avenue, Bronx, was visited a month after installation there was some dust in the gap and it remained in the fast leak category. Another visit occurred 1 year later, and the cover gap and pick holes were full of grit and dust. When re-tested this cover was now listed as a slow leak, leaking at 30 mL/min.

When the new manhole in Skyview Drive, Stamford, was re-tested 1 year later it remained in the ‘fast leak’ category, but it was still slower than when new. This rim-cover gap in this manhole showed signs of dust and grit accumulation but was not completely full.

The build-up of dust and grit in Skyview Drive was slow compared to Morris Avenue. Skyview Drive is a suburban street with much less traffic. The street surface also has visibly less dust and grit compared to Morris Avenue.

This manhole in Skyview Drive was actually tested again two years after installation, with the intention of tracking the progress of the grit seal. However, it was clear that the manhole had been recently opened and there was no dust or grit in the rim-cover gap.

In situ testing of 250 manholes was conducted in New York City and surrounding towns. A key observation was a dust and grit seal that built up between the cover and the rim. This made testing in-situ a necessity, and the test method had to preserve this seal. The test method involved filling a 30 cm section of the cover-rim gap with water and then recording the leakage rate.

Due to the presence of the dust and grit seal, 78.8% of the manhole covers showed no leakage, and 15.6% had slow leaks. Only 5.6% had significant leaks.

The average leakage rate for the slow leak group was 24 mL/min (172.4 mL/min when pro-rated for the full circumference). The fast leak group was noted as >120 mL/min because the leak rate exceeded the rate water could be added during the test. Half of the fast leak group were either new manholes, or recently opened manholes.

The build-up of the dust and grit seal from zero to complete took less than 1 year in NYC, based on the observation of a new manhole. At a suburban street, the grit seal was only partially built-up in a 1-year span, again based on observation of a new manhole.

Street sweeping over manholes was not observed to disturb the grit seal.

This natural sealing phenomenon is not limited to the NYC area. The author has observed the presence of grit seals on manhole covers in numerous cities, both throughout the US and around the world. This is, in fact, the typical condition observed for undisturbed manholes.

The main conclusion from this study is that the contribution of inflow from leaking manhole covers is minimal. In the overall context of I/I this can be considered a minor, or negligible contributor. From an operations and maintenance perspective this a strange case where doing less actually helps the situation.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.