تحلیل رویکردهای فشارمحور و تقاضامحور در بهینه‌سازی شبکه‌های آبرسانی

Abstract
In recent decades, growing pressure on water resources-particularly in arid and semi-arid regions-has highlighted the need for intelligent management of urban water distribution systems. One of the major challenges in this field is water leakage, which leads to resource loss and decreases the hydraulic reliability of the network. This study presents a comparative analysis of the Demand-Driven Analysis (DDA) and Pressure-Driven Analysis (PDA) approaches for the simulation and optimization of a water distribution network. The hydraulic model was developed in EPANET, while the optimal scheduling of pressure reducing valves (PRVs) was achieved through the Bat Optimization Algorithm (BOA) implemented in MATLAB. Simulations were conducted over a 24-hour period for two reservoir head levels (185 m and 205 m) and three leakage rates (0%, 25%, 45%). The results revealed that both approaches exhibit a decline in the reliability index as leakage increases; however, the PDA model more accurately represents real hydraulic behavior due to its pressure–flow dependency. Moreover, increasing the reservoir head significantly improved network reliability. Therefore, integrating pressure-driven modeling with intelligent optimization algorithms can provide an effective framework for pressure control and leakage management in urban water distribution systems.
 
Extended Abstract
Background and Objective
In recent decades, increasing pressure on global water resources-particularly in arid and semi-arid regions-has emphasized the necessity for efficient and intelligent management of urban water distribution systems. Leakage in water supply networks remains one of the most critical challenges in this field. It not only leads to the waste of valuable resources but also undermines the hydraulic stability and reliability of the network, imposing substantial economic, environmental, and social consequences. Studies have shown that leakage can account for up to 30% of the total supplied water in some regions, highlighting the importance of accurate modeling and optimal control strategies for minimizing losses.
Among the available hydraulic modeling frameworks, two principal approaches are widely used for network simulation: The Demand-Driven Analysis (DDA) and the Pressure-Driven Analysis (PDA). The DDA model assumes that each consumer node receives its full water demand, regardless of the prevailing pressure conditions. Although this approach simplifies computations, it often fails to represent real network behavior under abnormal conditions such as pressure drops, pipe bursts, or leakage events. Conversely, the PDA model considers the nonlinear dependency between pressure and flow, thereby providing a more realistic representation of system behavior under variable hydraulic conditions.
Given these considerations, the present study aims to conduct a comparative analysis of DDA and PDA approaches in modeling the hydraulic performance of an urban water distribution network. The objective is to evaluate how the two approaches differ in estimating network reliability and leakage behavior under different operational scenarios and reservoir head levels. Furthermore, this research integrates a metaheuristic optimization method-the Bat Optimization (BO) algorithm-to determine the optimal time-based control schedule for pressure reducing valves (PRVs), aiming to improve pressure regulation and minimize leakage in the network.
 
Methodology
The hydraulic simulations were conducted using the EPANET software, which provides a robust environment for modeling steady and unsteady flow in pressurized pipe networks. The study network, a benchmark system originally introduced by Alperovits and Shamir, includes two storage reservoirs, two PRVs, eight demand nodes, and nine connecting pipes. The network configuration was selected due to its capability to represent realistic dynamic conditions such as head loss, pressure fluctuations, and flow reversals.
Two head levels for the reservoirs-185 m and 205 m-and three leakage rates-0%, 25%, and 45%-were considered to examine the influence of energy level and leakage intensity on network reliability. The demand pattern was defined for 24 hours, with hourly variations reflecting typical domestic and commercial water consumption behavior.
To optimize the valve control schedule, the Bat Optimization (BO) algorithm was implemented in the MATLAB environment. This algorithm, inspired by the echolocation behavior of bats, iteratively searches for optimal solutions by updating frequency, velocity, and position parameters of each agent based on the best current solution. The objective function was designed to maximize the nodal pressure reliability index (NPRI) while minimizing total network leakage and energy loss. The coupling between EPANET and MATLAB allowed for automated simulation runs and evaluation of network performance metrics during each iteration of the optimization process.
For performance evaluation, both DDA and PDA models were run under identical conditions. The PDA approach was implemented using pressure-dependent emitter equations to simulate the relationship between nodal pressure and outflow, ensuring a realistic assessment of leakage and flow distribution across the network.
 
Findings
The simulation and optimization results revealed significant differences between the DDA and PDA models in capturing the network’s hydraulic behavior under varying leakage and pressure conditions.
Under the DDA framework, the reliability index showed a consistent decline with increasing leakage. At a reservoir head of 205 m, the reliability decreased from 0.797 (no leakage) to 0.776 (45% leakage), while at 185 m, it dropped from 0.562 to 0.473. This pattern indicates that the DDA model tends to overestimate network performance because it neglects the pressure–flow relationship, leading to unrealistic stability predictions during low-pressure events. Additionally, leakage volume was found to rise considerably with increasing leakage percentage, reaching up to 143 units at 205 m head.
In contrast, the PDA model demonstrated a more robust and realistic response. Although the overall reliability index also declined slightly with higher leakage rates, the reduction was much less pronounced compared to the DDA results. For instance, the reliability index for the205 m head decreased only marginally from 0.797 to 0.776, while for 185 m, it remained nearly constant around 0.702–0.701. This behavior suggests that the PDA model more accurately represents the hydraulic resilience of the network under variable pressure conditions, as it inherently accounts for the interdependence between flow and pressure at each node. The comparative analysis confirmed that the PDA approach provides a more reliable and physically consistent representation of water distribution systems, particularly when modeling leakage and pressure fluctuations. The model’s ability to dynamically adjust flow rates based on actual pressure values resulted in improved accuracy of leakage estimation and reliability evaluation. Furthermore, optimization using the Bat Algorithm effectively enhanced the operational performance of PRVs by minimizing unnecessary pressure peaks during off-peak demand hours and ensuring adequate supply during high-demand periods. The optimized control schedule contributed to a noticeable reduction in total leakage and improved network energy efficiency without compromising service reliability. Overall, the study highlights that combining the Pressure-Driven Analysis (PDA) approach with intelligent metaheuristic optimization techniques such as the Bat Algorithm can significantly enhance the decision-making process in pressure management, leakage reduction, and operational reliability of urban water distribution systems.
 
Conclusion
This study develops an optimal PRV scheduling framework to compare PDA and DDA approaches in urban water networks. Using the Bat Algorithm in MATLAB and EPANET simulations, various leakage (0–45%) and reservoir head (185–205 m) scenarios were tested. PDA consistently showed higher pressure reliability, confirming its superior hydraulic realism. Increased head improved performance; leakage reduced it. Findings highlight the value of pressure control and smart optimization for efficient network management.