Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 186 (2017) 76 – 83

XVIII International Conference on Water Distribution Systems Analysis, WDSA2016

Serious Game Approach to Water Distribution System

Design and Rehabilitation Problems
Mark S. Morley, Mehdi Khoury, Dragan A. Savić*
Centre for Water Systems, University of Exeter, Devon, United Kingdom

Abstract

This paper presents an online, web-based Serious Game developed to investigate end-user behaviour when faced with complex
WDS design and rehabilitation problems. SeGWADE (Serious Game for WDS Analysis, Design & Evaluation) couples an
innovative and visually attractive interactive front-end with a server-side modelling engine handling real-time hydraulic simulation.
A multiplayer online game infrastructure is implemented allowing interactions with a model to be instantly broadcast to both
collaborating and competing users. The interactions are recorded and can be subsequently be replayed or analysed through the
interface to explore the decision making process in detail. The Serious Game engine is designed to be extensible and can be
reconfigured dynamically through an interface allowing the upload of EPANET[1] input files and the parameterisation of different
game scenarios permitting a range of games to be developed independently. In addition, the front-end employs adaptive graphics
that can switch between HTML5 canvas or WebGL rendering technologies, depending on the client hardware capabilities, in order
to deliver the best user experience.
The serious game has been evaluated through a classroom-based exercise in which 20 students competed to obtain the best
WDS design solution. The results obtained demonstrate a high degree of player engagement with the game.

©©2016
2016TheThe Authors.
Authors. Published
Published by Elsevier
by Elsevier Ltd.is an open access article under the CC BY-NC-ND license
Ltd. This
Peer-review under responsibility of the organizing committee of the XVIII International Conference on Water Distribution
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Systems. under responsibility of the organizing committee of the XVIII International Conference on Water Distribution Systems
Peer-review

Keywords: Water Distribution System; Serious Game

* Corresponding author. Tel.: +44-1392-723637; fax: +44-1392-217965.

E-mail address: D.Savic@exeter.ac.uk

1877-7058 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the XVIII International Conference on Water Distribution Systems
doi:10.1016/j.proeng.2017.03.213
 Mark S. Morley et al. / Procedia Engineering 186 (2017) 76 – 83 77

1. Introduction
Serious games are an emerging type of cognitive tools designed to train and teach in an entertaining and interactive
manner [2,3,4]. Although serious games have already been used in numerous domains, only a handful of computerised
game simulations have been applied to the water sector [5]. These games tend to focus on Water resource management,
with a strong emphasis on the development of awareness and understanding of problems and trade-offs. Present games
such as Aqua Republica [6], CauxOperation [7], the Climate Game [8], EMOVER [9], Marine Spatial Planning
Challenge [10], Shariva [11], and the UVA Bay Game [12] usually involve multiple stakeholders cooperating to
resolve conflicting priorities in the context of resource scarcity. This paper presents a basic game that allows a number
of users to take part in a competitive game to optimize the performance of a water distribution system. WDS
optimization can be applied to many facets of the topology or operation of the network. The topological design and
layout of the system can be optimized with a view to minimizing the cost of a layout [13] whilst meeting some
minimum design constraint – typically the provision of sufficient pressure to each demand node in the network. The
costs of pumping water through a system are a key consideration in network design and pumping arrangements [14]
and schedules [15] can be optimized to minimize energy consumption and to ensure that off-peak electricity tariffs are
exploited accordingly.

1. Methodology

The SeGWADE (Serious Game for WDS Analysis, Design & Evaluation) game structure owes much to other
multiplayer online games. As such, it has a modular architecture where clients (the players using a web browser
from their machine) visit a web site (the server), log in, and play the game collaboratively.

The first application of this game architecture is to the rehabilitation of an existing WDS by the optional installation
of duplicate pipes in order to alleviate a pressure deficiency. The principal activity of the game is, therefore, to
select the diameters of the duplicate pipes from a predetermined list of commercially available pipe sizes, the cost of
which increases as a function of diameter. The user can select diameters for each available pipe interactively and
then submit the resulting model for hydraulic simulation at which point the resulting pressures are evaluated to
determine whether the model meets the minimum pressure requirements. When a new “best” (i.e. lowest cost)
solution is identified in the session of online gamers, the game banner on each of the currently connected clients is
updated to show the identity of the player and the score that they have achieved – adding a competitive element to
the exercise. At the end of the game, it is possible to summon a high-score table on each of the players’ browsers to
display their relative performance.

2. Architecture

The game engine is designed using a well-known MVC (Model-View-Controller) Client/Server paradigm with a
few modifications so as to make it slightly more efficient for the purposes of gaming over the internet, as seen in
Figure 1. As in a standard Model-View-Controller pattern, the model organises the application from the point of
view of the data, logic and rules, independently from the user interface. The view has the user interface in which the
information is visually represented. The Controller, acts as an intermediary that takes input from the user and
converts it to commands for the model or view. In order to reduce the communications overheads and load on the
web server, the MVC loop is short-circuited by using the angular.js [16] library so that relatively trivial operations,
such as input data validation, are performed on the client-side and not sent to the server-side model.

The game engine part of the application is written entirely in JavaScript on both server and client side. This provides
an unmatched flexibility from a development point of view. The Node.js [17] and Socket.io [18] libraries are used
for connectivity to handle instant messaging between clients and server during game time. Sails [19] being the most
popular MVC (Model–view–controller) framework for Node.js, is being used for deployment. The code is hosted as
open source on GITHub at https://github.com/gentr1/DWS_serious_game.
78 Mark S. Morley et al. / Procedia Engineering 186 (2017) 76 – 83

Figure 1 - MVC pattern of Serious Game

Owing to the use of the Sails framework, the system is database agnostic and can be adapted to operate with any major
database system on the server-side, both SQL or non-SQL. The framework also allows the application to scale so that
multiple servers can be installed to operate together in the event of demand requiring such. User authentication and
password encryption are also provided.

End users access the game with their web-browsers and is thus compatible with a variety of platforms, having been
tested on Windows, Linux, iOS and Android. Users having access to WebGL capabilities can switch between a
medium and a high graphic detail mode which offers a more aesthetically pleasing immersive 3D visualisation and
which is better able to scale in size for rendering large networks. In order to accommodate the gamut of hardware
capability that might be experienced on diverse target platforms, the graphic rendering is designed to fall-back to
HTML Canvas technology in the event that WebGL rendering is unavailable. Interaction with the network takes place
by selecting the element of the network that is to be modified. This pops-up a selector that allows the player to change
the properties of the network element directly. For the example presented in this paper the options are limited to
changing pipe diameters. However, it is possible using the game engine developed to manipulate pumps, valves and
other network components in a similar fashion, according to the needs of the game problem. When the user has
changed a number of elements and is happy with the configuration of the network then a “solve” button is pressed
which runs the modified configuration on the server-side hydraulic solver and returns the simulation results.

The use of socket.io allows messages to be instantly broadcasted between users during game time. The messaging
system between clients and server at game time is organised as seen in Figure 2:
 Mark S. Morley et al. / Procedia Engineering 186 (2017) 76 – 83 79

Figure 2 - Socket.io messaging system between client and server at game time

Underpinning the game is a hydraulic simulation engine based on EPANET [1] which includes the ability to
simulate hydraulic models using a pressure-driven approach [20]. Adopting a pressure-driven model allows more
meaningful results to be shown to the user in the event of an infeasible network solution being proposed.

3. Case Study

The problem that has come to be known as “New York Tunnels” was introduced [21] as an illustration of a (then)
large-scale optimization problem for the reinforcement of the water supply for New York City. The “Tunnels”
name stems from the fact that the pipes are of inordinately large diameter, ranging from 60 inches (1.5 metres) to
204 inches (5.2 metres). As presented, the network is pressure deficient in five nodes. In order to alleviate this
problem, each of the 21 pipes in the network may be duplicated with one of 15 commercially available pipe sizes or
left unduplicated. Despite the apparent simplicity of the problem, this gives a solution space of 1621 = 1.93×1025
possible solutions – far too many to be enumerated exhaustively despite the advances of computer power in the
intervening years.

This network has become a favourite benchmark for optimization applications and has been employed widely in the
literature. From the original solution in [21] of $78.1M, the best known solution for the problem has been advanced
by, among others, [22] (1985 – $39.23M) and [23] (1993 - $38.81M). Savić and Walters [24] used the problem to
illustrate the sensitivity of such optimization problems to small changes in the coefficients used in calculating the
frictional losses observed in the system, demonstrating solutions ranging from $37.14M to $40.45M representing the
best solutions found for the gamut of coefficients used for this problem by other workers in the field. At the time of
writing, the best-known feasible solution to the problem of $38.64M was determined using an Ant Colony
Simulation (ACS) approach [25] in 2003.
80 Mark S. Morley et al. / Procedia Engineering 186 (2017) 76 – 83

The single objective optimization is formulated thus: ࡺ࢒

ࡹ࢏࢔࢏࢓࢏ࢠࢋǣ࡯࢏࢔ࢌ ൌ ࢌ൫ࡰ૚ ǡ Ǥ Ǥ ǡ ࡰࡺ࢒ ൯ ൌ ෍ ࡯൫ࡰ࢐ ǡ ࡸ࢐ ൯

࢐ୀ૚

ࡰ࢐ ‫ࡰ א‬ሺ࢐ ൌ ૚ǡ Ǥ Ǥ ǡ ࡺࢊ ሻ
where: Cinf is the total infrastructure Cost to be minimized, Nl is the number of links in the network for which
reinforcement or installation is an option, C(Dj,Lj) is the cost of the jth pipe with diameter Dj (chosen from a discrete
set of available diameters D) and length Lj. K is the penalty multiplier constant, Hi is the pressure head at node i (as
computed by the hydraulic solver), Hi,min is the minimum pressure head requirement sufficient to fully satisfy the
demand at node i and Nn is the number of nodes in the network. Nd is the number of decision variables in the
optimization.

The Serious Game uses the cost versus diameter function defined in [21] thus:

‫ ݐݏ݋ܥ‬ൌ ͳǤͳ ή ‫ݎ݁ݐ݁݉ܽ݅ܦ‬ଵǤଶସ ή ‫݄ݐ݃݊݁ܮ‬

Where: Diameter is in inches, Length in in feet and Cost is in US Dollars.

Figure 3 - New York Tunnels Topology - as rendered in the SeGWADE Serious Game

The network topology consists of two loops and two branches supplied under gravity by a single, fixed-head
reservoir. In the original, pressure deficient configuration, nodes 16, 17, 18, 19 and 20, at the periphery of the
network, fall below the required minimum pressures. Figure 3 illustrates how the network is rendered during the
game. The blue (dark) coloured pipes represent the fixed topology of the network with the relative diameters of the
pipes clearly visible. The curved pipes represent the duplicate pipes that may be optionally installed in parallel to
each of the existing pipes. When no pipe diameter is selected for one of these pipes it is rendered in red and is
considered closed in the hydraulic model, whereas green pipes represent open, duplicate pipes. The node
 Mark S. Morley et al. / Procedia Engineering 186 (2017) 76 – 83 81

representations are colour coded to visually highlight whether they are pressure deficient or not. Nodes shown in
green have sufficient pressure, whereas those shaded orange to red are those exhibiting increasing levels of pressure
deficiency.

4. Results

The game was deployed to a class of 20 students in order to gauge how well they would interact with it.
Engagement was good, with the majority of students attempting over 100 evaluations of the network and half in excess
of 200, it is clear that the game maintained the interest of the players. Figure 4 shows the number of attempts made
by each of the players and the high scores they achieved.

Figure 4 - Number of attempts versus best score attained for the Case Study group

Notable is the performance of player #13 who, despite making 224 attempts at the problem, failed to identify a
feasible solution, as did player #1, albeit with markedly fewer attempts.

¡Error! No se encuentra el origen de la referencia. shows the performance of the players as the game progressed
in terms of their proximity to the best known score of $38.64M. The three colour lines in this figure represent the
three highest scoring players in the game. As can be seen, the majority of users converged upon solutions under $70M
after making few attempts with most going on the identify solutions better than $45M. Initially, the players opted to
start from the most expensive and fully compliant solution with all of the duplicate pipes installed. However, it is
clear that the players quickly identified that the optimal solution lay closer to the other extent of the problem (i.e. with
none of the pipes duplicated) and proceeded to investigate the problem from that standpoint.

5. Conclusions & Future Directions

A serious game has been developed to allow users to competitively optimize a simple water distribution system
problem in which network reinforcement is undertaken by selecting pipes to be duplicated to meet a minimum pressure
requirement across the network.
82 Mark S. Morley et al. / Procedia Engineering 186 (2017) 76 – 83

Figure 5 - Player scores during the progress of the game

The game engine has been designed to be extensible and will be deployed on more complex optimization problems.
It is envisaged that these problems will involve multiple objectives and will be undertaken by teams who will be
competing in their own right but also collaborating within teams by allocating different roles to the players on each
team such that they represent different stakeholders and have different priorities in terms of the objectives and how
they might be achieved.

The game has been evaluated in a classroom environment in which a high degree of player engagement with the
game was observed. In addition, a clear shift in the fashion in which players attempted to identify feasible solutions
for the problem was seen with players readily identifying the proximity of the better solutions to the starting, infeasible
configuration of the network rather than starting from the highest-cost, feasible solution.

The game as presented in this paper can be played online at http://water.servegame.org/

Acknowledgements

The authors would like to acknowledge the assistance cooperation of the UNESCO-IHE Institute for Water Education
class of MSc students on the Urban Water Module for their engagement in playing the game and the developers of the
open source libraries Sails.js, Node.js, Three.js and Angular.js which underpin the game engine.

References

[1] Rossman, L.A. 2000. EPANET 2 Users Manual, United States Environmental Protection Agency, Washington, USA, 200pp.
[2] Michael, D. & Chen, S. 2006. Serious games: games that educate, train and inform. Thomson Course Technology, Boston, Massachusetts,
USA. 312pp.
[3] Prensky, M. 2006. Don't bother me Mom, I'm learning - how computer and video games are preparing your kids for 21st century success and
how you can help. Paragon House, Minnesota, USA. 254pp.
[4] Rayburn, E.M. 2007. Applying simulation experience design methods to creating serious game-based adaptive training systems. Interacting
with Computers, 19(2), pp206–214.
[5] Medema, W., Furber, A., Adamowski, J., Zhou, Q. & Mayer, I. 2016. Exploring the Potential Impact of Serious Games on Social Learning
and Stakeholder Collaborations for Transboundary Watershed Management of the St. Lawrence River Basin. Water, 8(175), 24pp.
 Mark S. Morley et al. / Procedia Engineering 186 (2017) 76 – 83 83

[6] Chew, C., Lloyd, G.J. & Knudsen, E. 2013. An Interactive Capacity Building Experience—An Approach with Serious Games. Available
online: http://www.dhigroup.com/upload/publications/mikebasin/Chew_2013.pdf (accessed 28 February 2016).
[7] Souchère, V., Millair, L., Echeverria, J., Bousquet, F., Le Page, C. & Etienne, M. 2010. Co-constructing with stakeholders a role-playing
game to initiate collective management of erosive runoff risks at the watershed scale. Environmental Modelling and Software, 25(11),
pp1359-1370.
[8] Zhou, Q., Mayer, I.S., Bekebrede, G., Warmerdam, J. & Knepfle, M. 2013. The Climate Game: Connecting Water Management and Spatial
Planning through Simulation Gaming. In: Edelenbos, J., Bressers, N. & Scholten, P. [eds.] Water Governance as Connective Capacity,
Ashgate Publishing, Farnham, UK, pp109-127.
[9] Link to EMOVER estuary management game: http://www.hkvk.nl/en/news/bericht:estuary-management-game-online.htm (accessed 6 June 2016)
[10] Mayer, I.S., Zhou, Q., Lo, J., Abspoel, L., Keijser, X., Olsen, E. & Kannen, A. 2012. Integrated, ecosystem-based marine spatial planning:
First results from international simulation-game experiment. In: Proceedings of the Third International Engineering Systems Symposium
(IEEE), CESUN, Delft University of Technology, Delft, The Netherlands, 18–20 June 2012; pp18–20.
[11] Douven, W., Mul, M.L., Son, L., Bakker, N., Radosevich, G. & Hendriks, A. 2014. Games to create awareness and design policies for
transboundary cooperation in river basins: Lessons from the shariva game of the Mekong river commission. Water Resources Management,
28, pp1431–1447.
[12] Learmonth, G.P., Smith, D.E., Sherman, W.H., White, M.A. & Plank, J. 2011. A practical approach to the complex problem of environmental
sustainability: The UVa bay game. Journal of Public Sector Innovation, 16, pp1–8.
[13] Murphy, L.J., Dandy, G.C. & Simpson, A.R. 1993. Design of a Pipe Network Using Genetic Algorithms. Water, 20(4), pp40-42.
[14] Goldberg, D.E. & Kuo, C.H. 1987. Genetic algorithms in pipeline optimization. Journal of Computing in Civil Engineering - ASCE, 1(2),
pp128-141.
[15] Yu, G., Powell, R.S. & Sterling, M.J.H. 1994. Optimized pump scheduling in water distribution systems. Journal of Optimization Theory
and Applications, 83(3), pp463-488.
[16] Link to Angular.js: https://angularjs.org (accessed 21 June 2016)
[17] Link to Node.js: https://nodejs.org (accessed 21 June 2016)
[18] Link to Socket.io: http://socket.io (accessed 21 June 2016)
[19] Link to Sails: http://sailsjs.org (accessed 21 June 2016)
[20] Morley, M.S. and Tricarico, C. 2008. Pressure Driven Demand Extension for EPANET (EPANETpdd) – Technical Report 2008-02, Centre
for Water Systems, University of Exeter, UK. 11pp.
[21] Schaake, J. & Lai, D. 1969. Linear programming and dynamic programming application of water distribution network design (Report 116).
MIT Press, Cambridge, U.S.A.
[22] Morgan, D.R. & Goulter, I.C. 1985. Optimal urban water distribution design. Water Resources Research, 21(5), pp642-652.
[23] Murphy, L.J., Dandy, G.C. & Simpson, A.R. 1993. Design of a Pipe Network Using Genetic Algorithms. Water, 20(4), pp40-42.
[24] Savić, D.A. & Walters, G.A. 1997. Genetic algorithms for least-cost design of water distribution networks. Journal of Water Resources
Planning and Management - ASCE, 123(2), pp67–77.
[25] Maier, H.R., Simpson, A.R., Zecchin, A.C., Foong, W.K., Phang, K.Y., Seah, H.S.& Chan Lim Tan, C.L. 2003. Ant Colony Optimization for
Design of Water Distribution Systems. Journal of Water Resources Planning and Management - ASCE, 129(3), pp200-209.