In this paper, a deterioration model is created and used to simulate the life cycle of a water distribution network (WDN). Then, two strategies – leakage fixing and pipe cleaning – are evaluated to rehabilitate its capacity to attend the demand. In order to implement the deterioration model, the following parameters were considered: growth of the consumer population, increase in leakage rate, functional pipe deterioration and reduction of the hydraulic capacity of the pumps. For the leakage fixing, a fixed reduction rate in water losses was considered throughout the entire WDN until a minimum reference value was reached. For pipe rehabilitation, leaning was considered at a rate of 1% of the total length of the network per year. In each of the rehabilitation strategies, a cost–benefit analysis was carried out using the net present value. The results showed that both alternatives can restore the capacity of the WDN, with the pipe cleaning presenting a better economic impact.
This study presents a methodology to understand the impact that the deterioration of a water distribution network (WDN) has on its operation.
This study brings an estimate of the time when a WDN should be rehabilitated as it deteriorates throughout its useful life.
This study presents two ways to rehabilitate a WDN and the economic impact of each of these implemented measures.
The optimal operation of a water distribution network (WDN) can be achieved through the implementation of different strategies or a combination of them. Pump scheduling is commonly combined with the use of tanks and pressure reducing valves (PRVs) to save energy and reduce leakages (Shao et al. 2019). In addition, Brentan et al. (2018) proposed the use of demand forecasting to improve the reliability of the operation for near real-time operation. District metering area (DMA) creation can also improve the operation of both pumps and valves since smaller districts are easier to manage (Di Nardo et al. 2014; Campbell et al. 2016). Advanced strategies could also seek the improvement in water quality during the operation (Kang & Lansey 2010; Brentan et al. 2021) and energy recovery in PRVs (Hamlehdar et al. 2022).
The more efficient the operation, the better the performance of the WDN. As highlighted by Mala-Jetmarova et al. (2018), the design of WDNs can take into account, in addition to the implementation costs, several operational parameters, such as water quality and resilience index. Multi-objective optimization and multicriteria procedures are used to contemplate all the relevant aspects of a WDN design and operation and achieve a reliable solution (Farmani et al. 2006; Carpitella et al. 2019). Even so, considering the long life cycle expected for these infrastructures, the deterioration of its components will constantly reduce the operation efficiency. The rate in which this decline occurs depends on several factors, such as water quality, soil conditions, pipe material, pressure surges and water demand (St. Clair & Sinha 2012). In extreme conditions, this can lead to an intermittent supply, which can lead to economic, social and health problems (Klingel 2012; Simukonda et al. 2018). In order to minimize the impacts of intermittent operation, Souza et al. (2022) proposed an optimal operation in these conditions, with the rehabilitation of main pipes.
Other problems can arise with the deterioration of WDN and the intermittent operation, such as water contamination due to the intrusion of pathogens through small cracks that appear in the pipes (Mora-Rodríguez et al. 2015), and pipe collapse during the filling process (Martins et al. 2017). In addition, Liu et al. (2016) described the impact of biofilm growth on the pipe wall, potentially reducing water quality for consumers, affecting the taste and smell of water and possibly enhancing bacterial contamination.
Several approaches can be used to rehabilitate a WDN operating under intermittent conditions, such as leakage control, pipe replacement and pumping station reinforcement (Haddad et al. 2008; Creaco & Pezzinga 2015). In order to establish the best interventions in the infrastructure of a WDN throughout its life cycle, a cost–benefit analysis is necessary to evaluate the feasibility of each alternative. This is not simple because costs and revenues will occur at different times throughout this period and can significantly vary, as observed for example in energy tariffs. Covelli et al. (2016) used a genetic algorithm to propose a methodology to control water losses in a WDN using PRVs. The optimization was performed considering the total costs of acquisition, installation and management of the valves, in addition to the avoided costs with leakages. Creaco & Walski (2017) reinforced this approach with an economic analysis of different pressure control solutions to reduce leaks and pipe bursts. They pointed out that the most appropriate control method should consider the net costs throughout the WDN life cycle, comparing the investment with the leakage reduction economy.
Even in reliable WDN, economic analyses are necessary to try to improve the operation of pump stations (Brentan et al. 2018; Briceño-León et al. 2023) or even to recover the excess energy in gravity systems (Meirelles et al. 2017). Finally, as highlighted by Tscheikner-Gratl et al. (2016), a maintenance schedule is also required to preserve the WDN conditions through its life cycle.
In this paper, a deterioration model is used to identify the optimal period to start investing in rehabilitation strategies and avoid water shortages. The model is composed of four algorithms to simulate pipe encrustation, population growth, leakage behaviour and pump deterioration. Each parameter is set with a specific deterioration rate, and 20 years of operation are simulated. A typical operation week is used to calculate the yearly costs, and after each year, the deteriorated parameters are updated. With the simulated life cycle, it is possible to identify at which period the WDN began to have problems. At this point, two strategies for rehabilitation are studied separately: leakage fixing and pipe cleaning. The D-Town network is used as a case study, and the results have shown that both strategies for rehabilitation are successful, with pipe cleaning having a better economic impact.
WDN DETERIORATION MODEL
Fixing leakages in a WDN affect the whole water production chain: less water needs to be withdrawn from natural sources and treated, and less energy is consumed in pumping stations, as the system will require lower flows and consequently lower hydraulic grade, as the head losses are also reduced. However, despite its great benefit, it is not easy to detect and locate leakages in a WDN (Li et al. 2015). According to Zaman et al. (2020), two main methods can be used to identify leakages: (i) direct, which requires field inspection using specific equipment such as acoustic devices, that rely on the detection of vibration or noise signal created by leakages and (ii) indirect, which uses hydraulic models or data mining techniques to identify anomalies in the monitoring data. The major drawback of the direct method is the related cost, whereas, for the indirect methods, the uncertainties are high for the existing techniques.
In this paper, it is considered that any method can be used to detect and locate leakage. Therefore, an average value of 7.27 R$/m³/year of water loss reduction is used (European Commission 2013). The applied investment strategy in this case is reducing 10% of the actual water losses each year until this index reaches the benchmark value of 10%. It is important to highlight that, to maintain the new index level, the investment must be maintained for each of the following years. This leakage reduction is modelled by reducing the emission coefficient Ce in all nodes by the same amount. Thus, no geographical influence of leakages can be observed in this study, i.e., there is no prioritization to fix high pressure zones, where leakage rates are higher.
The rehabilitation of pipes can be done simply by cleaning their wall, aiming to restore, or at least improve, its hydraulic capacity by reducing its roughness. Another alternative is to replace part or the entire length of the pipe. In addition to the roughness reduction, this alternative allows us to improve the pipe capacity by increasing its diameter, and, indirectly, it is possible to fix existing leakages that are not detected. The drawback of this alternative is its elevated cost. In this paper, only the cleaning strategy is studied, with a cost of 9.13 R$/m (CASAN 2021). Therefore, each rehabilitated pipe remains with the same diameter and the same leakage rates. Thus, the only benefit considered is roughness reduction.
Two main aspects should be considered for the pipe replacement strategy: which pipes are more suitable for replacement and which replacement rate should be used. To select the pipes for replacement, the proposal of Campbell et al. (2015) to identify the trunk network was used, in which the WDN is modelled as a graph to identify the shortest path between each node and the water source. As these pipes transport higher flow rate and are connected to a higher number of nodes, it is expected that by reducing their headlosses, the pressure in the entire WDN will improve. For the rate of replacement, 1% of the length of the WDN is replaced each year, following the recommendations of the European Commission (2013). It is important to highlight that, as the mains pipes are replaced, they start the deterioration process again, and could be reconsidered for replacement in the following year if their headlosses significantly increase again.
The paper presented two different strategies for the rehabilitation of a deteriorated WDN. Pipes incrustation, leakage increase and pump deterioration were considered to model the WDN conditions during its life cycle. In addition, the water demand increase was also considered, using an estimation of the population growth. For the D-town case study, a significant increase in operational costs was observed when compared to the ideal situation of no deterioration. More important, after the ninth year of operation, the minimum pressure dropped below the required value. Thus, if no maintenance is made, the WDN would operate under intermittency or be unable to supply some consumers for half of its life cycle. Pressure management could be an important strategy to improve the water supply conditions at this point. However, this is not a direct rehabilitation of the network, since the orifices of leakages remain and the water losses are just better controlled. Thus, the real problem was not effectively solved, only postponed. Both strategies for rehabilitation studied – leakage fixing and pipe cleaning – were capable of reestablishing the required pressure conditions. From the hydraulic point of view, fixing leakage is more attractive, as it reduces both pump flow and head, and also the water production volume. This is clearly observed when the operational costs and SEC of the two strategies are compared, with leakage fixing significantly better. However, the investment necessary for leakage fixing is much higher, and by using the NPV as an economic indicator, it was observed that pipe cleaning resulted in a much better option. In addition, it is much easier to effectively implement a maintenance schedule of pipe cleaning according to its lifetime and relevance to the WDN compared to fixing leakages, which can be hard to identify – especially the small ones. It is important to note that the case study presented here contains two singular aspects, namely its design, with pipe diameters capable to attend the population growth, and five pump stations, capable to adjust the schedule to operate during more hours per day, also assisting the demand increase. In other case studies, with poorer hydraulic conditions, a pipe replacement strategy may be required for rehabilitation.
The authors acknowledge the financial support from the National Council for Scientific and Technological Development (CNPq) through the Productivity Scholarship PQ-2 (CNPQ No 305256/2021-1) and Universal Demand Project (CNPQ N. 404605/2021-4), and Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG) through the project APQ-01320-21.
DATA AVAILABILITY STATEMENT
Data cannot be made publicly available; readers should contact the corresponding author for details.
CONFLICT OF INTEREST
The authors declare there is no conflict.