Urban stormwater harvesting – sensitivity of a storage behaviour model

https://doi.org/10.1016/j.envsoft.2007.09.006Get rights and content

Abstract

The harvesting of urban stormwater to supply non-potable water demands is emerging as a viable option, amongst others, as a means to augment increasingly stressed urban water supply systems. This paper investigates the sensitivity of an urban stormwater harvesting system's capacity-yield-reliability relationship to variations in the behaviour modelling method used, focusing on the storage and demand components of a single reservoir system. The aim is to enhance our understanding of the appropriate computational method for assessing such volumetric reliability/storage capacity relationships. Four reference scenarios were developed, based on two different climates and two different water demand patterns. A sensitivity analysis was conducted, which considered the following computational, storage and demand parameters: yield-spillage order, modelling time-step, length of rainfall record, initial storage volume, open/closed storage surface, dead storage volume, diurnal and weekly pattern of water demand, and inter-annual variability of seasonal water demand. It was found that several parameters had an insignificant impact on the estimation of volumetric reliability for the scenarios tested, whilst the three most significant parameters were: length of rainfall record, inter-annual variability of seasonal demand, and storage surface type. Recommendations about the minimum length of rainfall record used and the inclusion of both the inter-annual variability of seasonal demand and net evaporative losses in the case of an open store are made.

Introduction

There has been a fundamental shift in the approach to the provision of urban water services in many countries around the world, mainly due to water stress. This has led to numerous investigations into the feasibility of utilising non-traditional water sources, such as stormwater, for the supply of non-potable urban water demands. During the conceptual planning and feasibility analysis phase of a project, questions such as “how much stormwater can be harvested?”, “how reliable is this supply source?” and “how large a store is required?” are posed. Due to the likelihood that there are numerous options under consideration, the analysis will often be conducted using a “what-if” analysis approach. Therefore, the analysis method should achieve a reasonable level of accuracy, but, whenever possible, not require an undue amount of input data or computational effort.

In sizing storages, two types of stores are considered; within-year storages that go through the full-empty-refill-spill cycle several times a year, and over-year storages that go through this cycle over a much longer time period, often in the order of years. Large urban water supply dams, in locations where stream flow is highly variable or the draft is high and high levels of supply security are required, will be over-year storages, due to their need to buffer against the inter-annual runoff variability. In comparison, urban stormwater stores are often within-year storages, due to lower variability of urban runoff or storage capacity often being limited by space constraints; the epitome being a small household rainwater tank which might fill within a single rain event and empty during a single garden watering session. Due to the within-year nature of urban stormwater stores, the within-year inflow variability, in addition to inter-annual inflow variability, has the potential to influence storage capacity requirements. Other differences are that non-traditional water supply systems predominantly have a single store, rather than more complex multi-reservoir systems; and have a standard, or use-until-empty, operation policy (McMahon and Adeloye, 2005) to maximise long term yield.

Within-year storages are best analysed using sequential methods, capable of considering the temporal pattern of inflows (supply) and outflows (demand and losses). Also, modelling must be undertaken at shorter time-steps (much less than annual) if the reservoir is within-year because the ‘critical period’ will be less than one year (Adeloye et al., 2003). The most appropriate method is behaviour analysis (McMahon and Adeloye, 2005) (also called an operational study), particularly given that the supply security is negotiable (due to the back-up supply and the end-use being less critical), with the design being able to make tradeoffs between storage capacity (thus cost) and yield. Translating the mass balance equation into a behaviour analysis simulation model of a single store requires decisions about computational time-step, the ordering of processes within a time-step (such as yield-spillage order), the total length of simulation, the level of detail for processes such as diurnal demand patterns and inflow patterns, and the inclusion of terms such as incident precipitation and open surface evaporation (Mitchell, in press).

Many authors (such as McMahon and Mein, 1978, Klemes et al., 1981, Vogel and Stedinger, 1987, Pretto et al., 1997) investigated the impact of computational method on reservoir storage-yield predictions, focusing on sizing large, over-year, dams or reservoirs for high levels of supply reliability. More recently, due to the interest in incorporating rainwater tanks into the urban water system throughout the world, a number of researchers investigated how the predicted capacity-yield relationship is influenced by the conceptual representation within a rainwater tank simulation model (Fewkes, 1999, Fewkes, 2000, Fewkes and Butler, 2000, Fewkes and Warm, 2000, Liaw and Tsai, 2004, Mitchell, in press). However, a rainwater tank harvesting system is somewhat different from an urban stormwater store in the sense that the former has a 100% impervious catchment and is a closed small store.

The only study found to focus on urban stormwater storages was by Wanielista et al. (1991), reporting on a sensitivity analysis of a stormwater harvesting daily time-step behaviour analysis model. The authors defined “efficiency” as the proportion of stormwater runoff that was harvested rather than the more common definition, which is the proportion of the demand actually supplied. Wanielista et al. (1991) found that the inclusion of incident evaporation and precipitation reduced the efficiency by 2%–4%, while the modification of operational rules (no irrigation on days of heavy rain) reduced the efficiency by 5%. They also concluded that varying the length of record between 15 and 24 years did not produce any significant difference in efficiency, although shorter periods of simulation did (less than 15 years).

Wanielista et al. (1991) did not investigate the stormwater storage behaviour model's sensitivity to several key computational processes represented in a behaviour analysis model, such as time-step, initial storage volume, yield-spillage order, dead storage volume, and the pattern of demand.

The relationship between inflow pattern and the predicted yield from the stormwater store is not explored within this study. Although, as noted earlier, behaviour analysis accounts for both the within-year and inter-annual variability of inflow when a monthly (or smaller) time-step is used (McMahon and Adeloye, 2005). This paper's purpose is to provide guidance on the trade-off between computational effort and the accuracy of behaviour analysis estimates assuming the stormwater harvesting site has already been selected. Therefore, the only aspect of stormwater inflow variability which is included in the scope of this paper is due to the length of rainfall record, which, in turn, generates the stormwater inflow series used in the behaviour analysis. This is the first stage of an ongoing investigation in the performance and modelling of urban stormwater systems involving the monitoring of several operational stormwater systems, which will, amongst other outcomes, provide data to verify this desktop based analysis. The outcome of this work can be used to enhance the capabilities of low impact stormwater drainage models (e.g. those reviewed by Elliott and Trowsdale (2007)) and, in particular, those focused on urban water reuse (e.g. UVQ (Mitchell and Diaper, 2006)).

Section snippets

Behaviour model analysis tool – RAT

A tool for the behaviour analysis of stores for harvested stormwater was developed and named ‘Reuse Analysis Tool – RAT’. RAT has two separate modules: (i) a rainfall-runoff module that estimates the inflows into the store; and (ii) a behaviour storage module that routes the water through the store.

Results and discussion

Fig. 4 presents the volumetric reliability verses storage capacity relationship for the four Reference cases: (i) Melbourne with non-seasonal demand; (ii) Melbourne with seasonal demand; (iii) Brisbane with non-seasonal demand; and (iv) Brisbane with seasonal demand. In these Reference cases the influence of demand type is clearly evident; a seasonal demand results in lower volumetric reliability for a given storage capacity, because the store must provide seasonal carry-over storage to

Conclusions

This study examined how the behaviour model of a single urban stormwater store is sensitive to the variation in a range of computational parameters and processes, using volumetric reliability as the performance index. This analysis was done to: (a) develop a greater understanding of the relationship between computational detail and accuracy of simulation results; and (b) knowledge required to provide stormwater harvesting scheme designers with guidance on the compromise between simulation

Acknowledgments

The support of the research project's industry partners, Brisbane City Council, Department of Environment and Conservation (NSW), Melbourne Water, Queensland EPA and Victorian EPA are gratefully acknowledged. Russell Mein, Tom McMahon and Rob Argent are thanked for their constructive comments on the manuscript.

References (24)

  • N.M. Fennessey

    Sensitivity of reservoir-yield estimates to model time step surface-moisture fluxes

    Journal of Water Resources Planning and Management

    (1995)
  • A. Fewkes et al.

    Simulating the performance of rainwater collection and reuse systems using behavioural models

    Building Services Engineering Research and Technology

    (2000)
  • Cited by (83)

    • A multi-parameter method to quantify the potential of roof rainwater harvesting at regional levels in areas with limited rainfall data

      2020, Resources, Conservation and Recycling
      Citation Excerpt :

      A daily time-step interval was selected because results are not significantly different than hourly, simulations (Fewkes 1999), but more accurate than weekly or monthly data series (Liaw and Tsai 2004; Campisano and Modica 2014). The YAS release rule was selected because it is preferred for design purposes based on the reduction of the effective storage capacity at each time step (Mitchell et al. 2008; Fewkes and Butler 2000). In addition, prolonged rainy days were observed in all the sites at the beginning of the simulation, which decreases the uncertainty of assuming an initial empty tank.

    View all citing articles on Scopus
    View full text