Call for a Dynamic Approach to GSI Maintenance

Communities around the world are moving to green stormwater infrastructure (GSI) approaches to restore the urban landscape and meet regulatory requirements. In all cases, it is essential that GSI systems are maintained to enable long-term functionality (Field et al. 2005; NRC 2009; Roy-Poirier et al. 2010; Houle et al. 2013; Blecken et al. 2017; DelGrosso et al. 2019) and meet regulatory requirements to ensure their intended design and permitted capabilities. Maintenance was recognized early on as an obstacle in using GSI (USEPA 2000), and it remains a key concern (DelGrosso et al. 2019). As an example, the ASCE Environmental and Water Resources Institute (EWRI) hosted the first Operation and Maintenance of Stormwater Control Measures Conferences in 2017.

Maintenance is a concern across all GSI. Privately owned GSI, whether voluntary or regulated, presents a particular challenge because the maintenance is often executed by the owner and not part of a systematic maintenance program as with a municipality. When privately owned GSI contributes to a municipal stormwater management plan, there is concern about the performance and lifecycle on assets not owned and operated by the municipal utility. This concern is amplified when a private GSI is transferred to a new owner that may not understand what the system is, the needs for maintenance, and the necessary access for inspection. Publicly owned GSI maintenance also presents challenges as these systems tend to be more complex and may have bigger impact on mitigating localized flooding. Concurrent with development plans for GSI, there needs to be community and municipal engagement and enforcement to ensure the sustainability of a stormwater management plan.

Anthropogenic and climatic processes can jeopardize the continued functionality of GSI (Taguchi et al. 2020). These processes are not constant in space and time, and the necessary maintenance to efficiently preserve and restore system performance is often difficult to predict. There is considerable variation among GSI practices in terms of type, design, and site conditions (DelGrosso et al. 2019) that leads to variability in performance, as observed by postconstruction inspection (DelGrosso et al. 2019; Houle et al. 2013; Hunt et al. 2005). Further, the distributed nature of GSI networks and their complex governing processes make it difficult to isolate a precise yet economical maintenance response to this variable demand. The dynamic issues around maintenance require a similarly dynamic maintenance diagnostic response. The need for a comprehensive, adaptive, risk-based, and cost-effective inspection and maintenance program becomes increasingly critical as municipal GSI programs expand to incorporate growing amounts of privately owned GSI. There is also a need to ensure that adequate maintenance is included in annual budgets. These needs present an opportunity for learning across peer communities on best practices and transferable knowledge of GSI performance, operation, and maintenance.

Private parcels represent the majority of a city’s impervious cover, and their location often coincide with strategic hydrologic and geographic features (Clements et al. 2018). For example, in Philadelphia, 66% of the land area is considered private property and 72% of the installed GSI projects between 2011 and 2019 are associated with private development through incentivized retrofit programs in the combined sewer system area (PWD 2019). Privately owned GSI may be considered a strategic asset to municipal stormwater management programs and can alleviate economic, space, and personnel constraints (Valderrama 2017). Many cities have GSI requirements for stormwater compliance or offer incentive voluntary programs for GSI development [e.g., San Francisco (SFPUC 2018), Chicago (CoC 2019), and WRF (2018)]. Popular incentive types include grants, cost sharing, recognition, development incentives, stormwater fee discounts, and credit trading (Malinowski et al. 2020), which focus on design and construction, but provide less support for GSI maintenance over time. Maintenance is critical to ensure continued performance of fundamental reduction and removal GSI functions, such as infiltration and managed outflow (Erickson et al. 2010; Roenigk et al. 1992) and contribution to municipal stormwater goals. The maintenance of privately owned GSI is normally the responsibility of the property owner, and permanent easements are required to support municipal inspections conducted to evaluate compliance. Further, municipalities are moving toward treating stormwater like a utility, with an associated fee to the rate payer to resource stormwater management programs including inspections. For example, 84% of surveyed asset managers have an inspection and/or monitoring GSI program (Stormwater BMP Maintenance TC 2019). An inspection program can ascertain what maintenance practices are accomplishing, long-term GSI functionality, how current protocols compare to site demands, and how the overall maintenance program can become more efficient and cost-effective (Miya and Grobbelaar 2015).

Green stormwater infrastructure networks strain inspection frameworks due to their distributed nature, large number of sites, and multifaceted functions (Moser 2016). The City of Philadelphia, for example, relies on 676 GSI projects (which may include more than one individual GSI per project, as of 2019) with paved and vegetated, surface and subsurface, and infiltrating and noninfiltrating configurations (PWD 2019). There is an urgent need for dynamic, data-driven approaches to more effectively manage large, heterogeneous GSI networks. GSI inspections range in complexity, frequency (i.e., typically every 1–4 years, although some programs or GSI types may be inspected more frequently) (CWP 2013; CoS 2019a; MCWD 2019; K. Flynn, personal communication, 2019; L. Sherman, personal communication, 2019; T. Saldutti, personal communication, 2019; K. Vacca, personal communication, 2019), and can be performed by property owners, private contractors, or city inspectors (CoS 2019b; MCWD 2019; Waickowski et al. 2019). Research is needed on appropriate inspection frequency and protocols (DelGrosso et al. 2019). Simple visual inspection is most common (Erickson et al. 2013; Stormwater BMP Maintenance TC 2019), but has low sensitivity to maintenance issues that are not visibly apparent (Erickson et al. 2013). While visual inspection, field testing, and continuous monitoring can produce a more comprehensive understanding of GSI performance status, quantitative performance monitoring of field tests and continuous monitoring is not often employed because of cost and personnel resources (e.g., only 11% of surveyed asset managers use these quantitative techniques) (Stormwater BMP Maintenance TC 2019). A national survey concluded that there is a willingness to pay for smart monitoring systems when shown to reduce maintenance costs (Meng and Hsu 2019).

A GSI system’s hydrologic performance is affected by regional climate, soil and vegetation characteristics, land use, site conditions, and specific design. Recognizing and understanding these dependencies provides opportunities for improving maintenance procedures (Traver and Ebrahimian 2017). This performance of a GSI is impacted by the hyperlocal environment, such as fall leaf litter or land-use specific influent sediment particle size. Present maintenance plans tend to be based on standard, set practices and the resources available (Taguchi et al. 2020), instead of a more dynamic and site-specific maintenance plan for a particular GSI type. The existing knowledge on the dynamic performance of GSI systems (Ebrahimian et al. 2020a, b) can be leveraged to improve maintenance procedures for different GSI systems in a dynamic way. For example, a GSI with surface inflow or outflow that may get clogged by leaf litter could be scheduled to be more frequently maintained during the local leaf drop period, targeting clogging maintenance before hurricane season, or even maintenance in response to community events such as parades.

Maintenance issues have a range of overlapping drivers, are specific to different GSI types (Erickson et al. 2018), and can be addressed with different maintenance techniques. However, it is unwieldy and costly for an organization to custom fit the maintenance needs for each GSI type and site. The best time to do maintenance is when the maintenance cost is lowest and just before there is an exponential increase in risk of failure (Peng et al. 2010). Finding this sweet spot has yet to be identified in GSI operations. Accordingly, there are multiple strategies for how limited resources can be deployed most effectively to manage large GSI networks. These strategies link inspection with maintenance. Stormwater BMP Maintenance TC (2019), Erickson et al. (2013), and Debo and Reese (2002) outline maintenance frameworks.

Reactive maintenance occurs in response to complaints or failures, which would exist even with sophisticated planning tools (e.g., vandalism or major storms), and often generates high long-term maintenance costs (Debo and Reese 2002). This nonroutine maintenance approach is widely employed, with one survey of cities in the United States and Canada reporting nearly half of GSI as managed with reactive maintenance (Jin et al. 2016). Technology use in GSI is increasing, and when continuous monitoring is employed, the monitoring system creates alerts for reactive or preventative maintenance in a more targeted way (Ho and Vacca 2018).

Predictive maintenance is potentially cost-effective, as it informs a nonroutine maintenance need(s) by inspection (Debo and Reese 2002). However, this framework requires a thorough inspection program, and both frequent or infrequent inspections can compromise the efficiency and cost-effectiveness. As a contrast, periodic maintenance is performed according to predetermined, routine maintenance frequencies, which can be more effectively planned for in a municipal budget. Existing maintenance guides normally suggest frequencies for various tasks, but these schedules are generalized and not always based on actual data of how the GSI systems function over time (Erickson et al. 2010; Houle et al. 2013). Further, the maintenance guides do not consider potential GSI performance and maintenance differences that arise from different design choices or site-specific conditions.

Instead of responding to site failures, inspection results, or a defined schedule, proactive maintenance attempts to address issues before they occur. This framework requires data and analysis tools to make connections between a site’s maintenance needs and the contributing factors. Geographic information system (GIS) tools are emerging for spatial decision-making techniques to determine risk level in urban stormwater networks due to social, environmental, and technical factors (Shariat et al. 2019) and for statistical techniques to analyze urban flood risk at the city scale (Hosseiny et al. 2020), which could aid this framework.

Given the challenges and strengths of each maintenance framework, it is impossible for one framework to be most effective for all GSI programs or in all situations. Rather, municipalities should craft a comprehensive maintenance approach that complements their resources and goals, employing individual maintenance frameworks where they are most effective from a risk-based approach. Identifying risk factors and using them to prioritize maintenance activities allows a maintenance schedule to be customized for a particular GSI site according to its specific needs and as the needs shift over time (William et al. 2018), which may provide cost-efficiency. A proactive maintenance framework can readily employ a risk-based decision-making process that would help focus limited resources where they are most essential or will have the greatest impact in achieving desired goals (Van Auken et al. 2016). Reactive maintenance, or run to failure, may be most economical for low-risk GSI (Van Auken et al. 2016), however further work must be done to determine which GSI is considered low-risk and should consider type, location within watershed, and target storm size and impact. Periodic maintenance may be appropriate for GSI that have a moderate risk for failure, including structural failure (e.g., media clogging or outlet collapse), while predictive maintenance provides the most individualized effort for high-risk GSI.

A challenge to dynamic maintenance is the proper utilization of large quantities of data to make these data-driven decisions. While many municipalities have collected large amounts of data and have access to increasingly fine scale environmental data, they are lacking comprehensive data management tools and experience (Eggimann et al. 2017). The data collected may be viewable but not readily accessible to query, aggregate, and analyze at the network scale. Recently there has been some guidance on collecting maintenance data to be shared publicly (Clary et al. 2018), although there is still an issue of data silos that can form wherein groups are unaware of or unable to relate (e.g., discrepancies in data collection and storage platforms) similar data collected by others, even within the same organization. However, through these new data reservoirs [i.e., municipal data collection, state transportation departments, and national agencies such as the National Aeronautics and Space Association (NASA), National Oceanic and Atmospheric Association (NOAA), and National Weather Service (NWS)], there is great potential for innovation as stormwater program managers begin to incorporate these data-driven techniques.

A key challenge particular to privately owned GSI for a successful inspection and maintenance programs is communication gaps between the owner and the municipality, which varies depending on if the owner is responsible for the maintenance or the municipality bills the owner for maintenance done by the municipality. Owners of private GSI on residential areas or on commercial land may require different communication strategies and maintenance expectations. Confusion develops over ownership and responsibility of maintenance due to change in property ownership or if a property is under lease (AWPD, personal communication, 2019; JMT, personal communication, 2019, OptiRTC, personal communication, 2019). Though GSI sites typically must be disclosed on a site plan or property deed, tenants or new owners are not always aware of a property’s GSI nor the benefits and responsibilities associated with these systems. This is especially true for primarily underground systems, such as infiltration trenches.

A dynamic, data-driven, and risk-based maintenance framework is required to ensure that private GSI meets the design intent and municipal stormwater requirements, provides social cobenefits, and is cost-effective. Inspection and maintenance protocols vary among stormwater management entities, but as urban areas evolve there is a common opportunity and need for improved data management and communication that builds upon the past two decades of maintenance experience. The benefits of performance monitoring data and dynamic hydrologic processes, data management tools, and new technologies in enhancing GSI type selection, operational decisions, and cost-effectiveness of maintenance activities are key factors in these frameworks. For example, Fig. 1 presents a conceptual workflow for a data-driven maintenance framework. This iterative approach opens the door for more efficient and directed maintenance in the ever-changing urban environment. Beyond enhancing system performance, data-driven tools can streamline regulatory reporting and support public outreach efforts. Moving even further, as there is increased discussion over sustainable, resilient, and equitable GSI, having GSI systems that are designed and operated with a more holistic community perspective in mind is required to amplify GSI as a long-term solution strategy. To support this, in addition to community engagement, there must be the political will to resource, enforce, and hold accountable stormwater management plans throughout the design life of GSI systems. Innovative software tools to support data-driven maintenance must be developed to dissolve data silos, improve communication and modeling, learn from existing data, and account for uncertainty. Simply stated, for meeting the regulatory requirements to protecting our waters with GSI requires a dynamic approach to maintenance.

Some or all data, models, or code that support the findings of this study are available from the corresponding author upon reasonable request.

The authors would like to thank Philadelphia Water Department for funding this research. Support from Villanova Center for Resilient Water Systems is also acknowledged. Reference in this article to any commercial product or the use of any trade, firm, or corporation name is for general informational purposes only and does not constitute an endorsement by the authors or the funding agency.