A Cooperative Co-Evolutionary Approach to

Large-Scale Multisource Water Distribution

Network Optimization
 Abstract
• Potable water distribution networks are important infrastructures of
modern cities. A good design of the network can not only reduce the
construction expenditure but also provide reliable service.
• Nowadays, the scale of the WDN of a city grows dramatically along with
the city expansion, which brings heavy pressure to its optimal design.
• First, an iterative trace-based decomposition method is specially designed
by utilizing the information of water tracing to divide a large-scale network
into small sub-networks.
• Second, a new fitness function is devised to handle the pressure constraint
of the problem. The function transforms the constraint into a part of the
objective to punish the infeasible solutions.
 Introduction
• If we take the design of WDNs as an optimization problem, it can be
defined as finding appropriate settings of water network components for a
pre given network layout, so that the capital cost is minimized subject to a
number of hydraulic, physical, and standardized constraints .
• The problem has been proven nondeterministic polynomial-time hard
metaheuristic methods.
• Among deterministic methods, linear programming and its variants were
first proposed in earlier time, attempting to solve the problem with low
computational burden . However, due to the nonlinear essence of the
problem, LP methods tend to fall into local optima easily .
• Then, some non LP methods , were proposed. But these methods are still
not effective enough to find the optimal or near-optimal solutions, since
the final solution generated by an NLP method highly depends on the
initial status of the method.
• Recently these metaheuristic methods have shown competitive
performance on small-scale networks which contain less than 200 pipes.
WDNCC to solve the large-scale WDN optimization problem. Rooted in the
divide-and-conquer strategy, cooperative coevolutionary methods solve
large-scale optimization problems by decomposing them into a number of
small-scale subcomponents.
• With the assistance of EPANET which is a well-known WDN simulation tool
, so here is an iterative trace-based decomposition method to divide a
largescale WDN into subnetworks in this project.
• Meanwhile, there are many constraints in the problem, and most of them
can be satisfied by utilizing EPANET except one, the minimum pressure
constraint.
• To handle the minimum pressure constraint, the objective function of
WDNCC is specially designed with a penalty function. The applied
evolutionary algorithm(EA) in WDNCC is the self-adaptive DE algorithm.
 WATER DISTRIBUTION NETWORK
• In the first place, a small-scale instance,
Anytown network, is shown in Fig. 1 to
facilitate the explanation of WDN.
• The Anytown network has one reservoir
and one pump station serving as the
main water source.
• Two tanks are set in the city for
auxiliary supplement.
• Each junction represents a consumer.
• Links between junctions are pipes
which are the variables of the problem.
• Assuming that the demands of all
consumers are known in advance, the
goal of the WDN optimization is to
reduce the constructing expenditure by
selecting an appropriate type for each
pipe, while satisfying all consumers’
demands and the hydraulic constraints.
 RELATED WORK
• Simulated Annealing (SA) algorithm to solve the WDN design
problem. The hydraulic network equations are solved by a Newton
search method during the execution of the algorithm. However,
there is not an efficient way to handle the constraint of user
requirements.
• Tabu Search (TS) method which incorporated a simulator to handle
the hydraulic constraints. To satisfy the consumers’ requirements,
they always initialize the solution by setting all pipes to the type of
the largest diameter. This initialization method may greatly
degrade the search ability of the algorithm.
• Genetic Algorithms (Gas) are one of the most popular algorithms
that have been investigated. Simpson et al. first applied a
canonical GA method to optimize a very small network. A penalty
function was designed to punish infeasible solutions which could
not satisfy the minimum head constraint.
• Particle Swarm Optimization (PSO) algorithm with
EPANET, developing a tool called PSONET. Infeasible
solutions in PSONET are also punished by adding a big
number which is larger than the largest cost of the
network to their objective values.
• Two-stage DE (TSDE) method is used to optimize
multisource WDNs. In the method, a WDN is partitioned
into several sub networks in advance. Then, during the
first stage of optimization, the sub networks are
optimized separately. Afterward, the whole network is
handled according to the result obtained in the first stage.
The defect of the proposed decomposition method is that
it only fits some simple single-loading networks, where
pumps or valves do not exist.
COOPERATIVE CO-EVOLUTION FOR
WATER DISTRIBUTION NETWORK
OPTIMIZATION
 Trace-Based Decomposition
• The principle of decomposition is to put interactional variables into the same
group and divide independent variables into different groups. Although there are
already some decomposition methods proposed for large-scale continuous
functions like differential grouping, usually these decomposition methods cannot
be directly applied to WDN optimization problems.
• Based on the results of node partition, we can divide the pipes into different
groups.
• Before showing the trace-based decomposition method, three principles are
stated :
• 1) If the water consumed by a node comes from only one source, the node
definitely belongs to that source.
• 2) If the water consumed by a node comes from multiple sources, it belongs to the
source which provides more water to it.
• 3) Pipes should take charge of the nodes that they can affect. It means that a pipe
belongs to the source which its downstream node belongs to.
• An example is shown in Fig. 2 to illustrate the trace-based decomposition method.
There are two reservoirs, R1 and R2, and four consumers {a, b, c, d}, linked by six pipes
which are ordered from 1 to 6. First, the hydraulic simulation is made. We can find
that the flow direction of pipe 2 is from a to b, and the flow direction of the pipe 6 is
from c to d. It means that only the water from R2 cannot satisfy the requirements of b
and d. Some water flows from R1 to b and d to support them. Nevertheless, the trace
simulation shows that most of the water consumed by b and d still comes from R2.
Thus, b and d are grouped with R2, meanwhile a and c are grouped with R1. According
to the result of node partition, pipe 1 and 4 are partitioned into a group, pipe 2, 3, 5,
and 6 are partitioned into the other group.
 CONCLUSION
• A novel approach called WDNCC to handle the WDN optimization
problem with multiple sources. With the help of the simulation
tool, EPANET, an effective decomposition method is designed,
which needs little hydraulic domain knowledge.
• Meanwhile, the cooperation of the decomposition process and the
optimization process makes the algorithm efficient enough to find
near-optimal solutions. Experimental results on both single-loading
cases and multiple-loading cases have verified the effectiveness of
WDNCC.
• In furure, Considering different structures of the WDNs, more
decomposition methods should be designed specifically. This is also
a key point to promote the usage of EAs on the problems with
higher dimensionality.
Thank You