An assessment of sub-standard water pressure in South African potable distribution systems

Sub-standard residual water pressures in urban water distribution systems (WDS) are a prevalent phenomenon in developing countries – South Africa being no exception. The phenomenon of substandard pressure is poorly understood, with intermittent supply ultimately resulting when there is no residual pressure left in the system. This research addressed the prevalence and extent of sub-standard pressures by using hydraulic models of potable WDS for 71 South African towns, located in 17 different South African municipalities geographically spread over the country. The hydraulic models included 539,388 modelled nodes, which were analysed to determine the number of nodes with sub-standard pressure heads during peak hour flow conditions. The results show that the residual pressure headwas <24 m at 16.5% of the model nodes under peak hour flow conditions, with 6.7% of the nodes having pressure heads<12 m. In contrast, the results also report relatively high pressures in certain parts of the systems, far in excess of the minimum requirement, underlining the need for better pressure management at both high and low ranges. It was also noted that the South African design criterion is relatively stringent compared with some other countries and could potentially be relaxed in future. This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/). doi: 10.2166/washdev.2017.227 s://iwaponline.com/washdev/article-pdf/7/4/557/202165/washdev0070557.pdf Louis Strijdom Heinz Erasmus Jacobs (corresponding author) Department of Civil Engineering, Stellenbosch University, Private Bag X1, Matieland 7602, South Africa E-mail: hejacobs@sun.ac.za Vanessa Speight Department of Civil and Structural Engineering, University of Sheffield, Sir Frederick Mappin Building, Sheffield S1 3JD, UK


INTRODUCTION Background
In South Africa, there is a gap in coverage of providing basic water services to poor and disadvantaged communities which is exacerbated by rampant urbanisation, as is also the case in other developing countries. Over time, water networks have been expanded with the incorporation of previously unserviced consumers as well as new consumers, often without upgrading the main supply pipes. Peak flow rates increase over time and residual pressures decrease, often to sub-standard pressures.
One of the factors that drives the cost of potable water provision is the criteria used for design and hydraulic analysis of the water distribution system (WDS). A well known criterion for steady state analyses is the residual pressure head. The use of steady state demand-driven analysis with minimum pressure head (MPH) under peak hour demand remains a common criterion for system design (Jacobs & Strijdom ), despite the availability of more advanced reliability-based methods and head-dependent methods for distribution system analysis.
Minimum and maximum pressure heads can be obtained from steady state hydraulic model simulation results and are quantifiable, making them an obvious choice as performance indicators for water providing authorities. The MPH during peak flow is used worldwide as a criterion for system design. Most service providers also stipulate maximum Water Industry Act (UK Parliament ) further requires that water be supplied constantly and at such a pressure as will cause the water to reach to the top of the topmost storey of every building within the distribution system. Ofwat also uses a performance indicator regarding MPH called 'Properties at risk of low pressure' to evaluate system performance.
Performance is measured by testing that 10 m head of pressure is provided at the customer's external stop tap at a flow of nine litres per minute, which should be sufficient to fill a one-gallon container in thirty seconds (Ofwat ). Compliance with this pressure performance measure does not override the utilities' duty to comply with the Water Industry Act standard for pressure at the topmost storey.
In the USA, design criteria guidance manuals recommend that the minimum pressure should be 14.1 m (20 psi) at all times, even during a fire event superimposed on peak demand conditions (AWWA , ; GLUMRB ). Furthermore, the US Environmental Protection Agency lists maintenance of positive pressure in all parts of the distribution system as a best practice to avoid microbial contamination   Table 1, with a focus on the pressure during maximum hourly demand. The minimum pressure during

APPROACH
A quantitative theoretical approach was used to assess the MPH during peak flow in this study, based on an analysis of available hydraulic models for South African distribution systems. Actual system pressure and peak flow rates were not recorded as part of this research.

Limitation regarding fire flow
The South African design guidelines, discussed earlier  to receive a given supply at a given head. If this head is not attainable, supply at the node is reduced.' An extension of the standard EPANET solver exists that directly includes pressure driven analysis, the data structures and algorithms within EPANET source code are modified in such a way that fixed demand is assumed above a given critical pressure, zero demand is induced below a given minimum pressure (typically near zero) and some proportional relationship between pressure and demand is provided for intermediate pressures (Cheung et al. ). The EPANET extension was not used directly in this research, but instead a similar procedure was applied in WADISO for zones where the demand-driven analysis resulted in near-zero or negative nodal pressures.

Demands
The actual monthly water use per individual consumer, as recorded via the consumer water meter (in kL/month), formed the basis of the peak flow calculation in the hydraulic models. The monthly water meter readings are used for billing consumers in South Africa, with consumers typically billed for water monthly, based on actual water use.
Monthly water meter readings, used for billing, are recorded in the municipal financial billing systems, also called treasury systems. Jacobs & Fair () described a software tool called SWIFT that was also used in this research to extract monthly metered water use from treasury systems while maintaining spatial integrity of the data, meaning that each water meter could be plotted on a map and could thus be linked to hydraulic model topology. SWIFT has been employed for numerous research studies in Southern Africa over the past two decades (Jacobs & Fair ).
Hydraulic models were populated with the hourly peak flow rate, which is derived from the average annual daily demand (AADD). The AADD is widely used for problems relat- For the majority of the analysed municipal areas, SWIFT was used to calculate each consumer's AADD. For SWIFT to work, reliable monthly water meter readings need to be available in the treasury system, as used for billing by the municipality. For a few of the smaller municipalities, reliable treasury data was not available, so a manual process had to be performed to assign theoretical AADD values to consumers, based on available land-use and plot-size information. Each consumer was thus assigned a theoretical unit water demand (UWD). The UWD allocated to each of the different consumer types included in the study, is summarised in Table 2. In order to allocate the AADD to a consumer for which no water meter readings were available, the analyst would identify the land use code from the town planning records and then identify the corresponding type of consumer in

Peak factors
Flow rates in a WDS vary throughout the day, resulting in peak flows during times of high usage. Various methods are available for determining the peak flows and representing these within hydraulic models. In South Africa, it is common practice to multiply the AADD with a corresponding peak hour factor in order to estimate the peak hour flow rate. In this research, the peak factors by Vorster et al. () were employed, which is in line with the practice used by all municipalities reported on in this study; the peak hour factors are between 3.0 and 4.6 times the AADD.
The peak flow rate of each consumer was allocated to the model node nearest to the centre of the consumer's GISparcel, representing a property. As part of the procedure, a cross-reference was made between each individual customer's GIS-parcel and each node in the water model to geographically allocate the peak flow rate for each consumer to the nearest hydraulic model node with an automated GIStool, as explained by Jacobs & Fair ().
The AADDs of about 4.9 million individual consumer records, of which 3.5 million represented occupied homes,  were determined as part of this project with SWIFT (the remaining 1.4 million records were either vacant plots or unoccupied homes with no water use). The calculated peak flows for each of the consumers were subsequently cross-referenced to the appropriate model node. The total peak flow rate was thus determined for each model node.
The existing operational scenarios were used for each model to simulate the current status quo as closely as possible.

Statistical analysis
After performing the hydraulic analyses, the nodal result tables

RESULTS
The results, summarised in Table 3    events; are they merely inconvenient or are they seriously compromising system performance and service delivery?
Relatively low system pressures may be intentional (e.g., for leakage reduction), or unintentional (e.g., due to problems such as financial constraints that prevent system upgrades). The authors are of the opinion that sub-standard pressures in the study area are unintentional and are the result of various challenges faced by water service providers in South Africa. It would be necessary to further research and better understand the reasons for sub-standard pressures in the systems reported on in this paper.
However, a relatively stringent MPH requirement, such as the 24 m currently used, may lead to overdesign and overspending on infrastructure when compared to a reduced MPH value. The results of this study suggest that the MPH criteria of 24 m may possibly be too conservative for South African systems. In contrast, the results also report relatively high pressures in certain parts of the systems, far in excess of the minimum requirement. The results show the need for better pressure management at both high and low ranges, but how low could the MPH requirement possibly be set?
A system pressure head of 10 m is needed for operation of some typical household appliances. Lowering the standard to 10 m, in line with Ofwat (), may be acceptable and would lead to some advantages, but customer outreach would be needed. If the standard were lowered to 10 m, proactive management would be needed because even small reductions below 10 m may have a larger risk in terms of system performance and effective service delivery than reduction to just under <24 m. The consequences of MPH between 10 m and the current minimum requirement of 24 m are limited to longer waiting times for filling of containers (baths, basins, water bottles, etc.) and less efficient irrigation systems. The consequences of MPH values decreasing to below 24 m, but not below 10 m, are therefore not considered to be insurmountable.
Reduced criteria for MPH have some clear advantages.
In a South African case study, a cost saving of 32.5% on required upgrading of infrastructure was found when reducing the design standard from 24 m to 15 m in a particular urban system (Strijdom ). Future research is needed to investigate the financial benefits of dropping to (say) 15 m or 10 m, such as avoided or postponed infrastructure cost, reduced operations and maintenance cost, lowered leakage and lower pressure-driven demand.
In contrast, the negative impacts also need to be well researched. At the extreme when no residual pressure remains in the system (or parts of the system), intermittent