Chamberlain IEEE Computer 2013WETLAND

8/10/2019 Chamberlain IEEE Computer 2013WETLAND

1/14

IEEE TRANSACTIONS ON COMPUTERS, SPECIAL ISSUE COMPUTATIONAL SUSTAINABILITY, TCSI-2012-10-0740 1

A Decision Support System for the Designand Evaluation of Sustainable Wastewater

SolutionsBrent C. Chamberlain, Giuseppe Carenini, Gunilla berg, David Poole, and Hamed Taheri

Abstract The drive toward sustainable wastewater management is challenging the conventional paradigm of linear end-of-pipe solutions. A shift toward more sustainable solutions requires that information about new ideas, systems and technologiesbe more readily accessible for addressing wastewater problems. It is commonly argued that decision-making needs to involveengineers and other community representatives to define values and brainstorm solutions. This paper describes a decisionsupport system (DSS) prototype that is designed to help community planners identify solutions which balance environmental,economic and social goals. The system is designed to be scalable, adaptable and flexible to allow fair assessment of new ideasand technologies. It supports the exploration of consequences of various alternatives and visualizes the trade-offs betweenthem. Our DSS takes in modular descriptions of components and a description of a community context, automates the design ofalternative wastewater systems, and facilitates evaluating how well each design satisfies the given context. It provides anadaptable platform from which new solutions can be designed without having to predefine how a single component fits within aspecific system. Our DSS facilitates the exploration of alternative solutions by visualizing the effect of various trade-offs and

their consequences in relation to the communitys sustainability goals.

Index Terms Logic Programming, Decision Support, Design, Environment, Interoperability. Wastewater

!

1 INTRODUCTIONlinear, end-of-pipe infrastructure design has domi-nated wastewater management in the western worldsince the industrial revolution [1], [2]. Population

growth and urbanization in combination with concernsrelated to resource scarcity and global change havesparked an interest in more sustainable and cyclic ap-proaches [3], [4], [5]. As a consequence, the past decades

have seen a rapid growth of innovations based on theidea of waste as a resource rather than a liability, with afocus on water, energy and nutrients. The uptake of tech-nical and institutional innovations is, however, slow. Thisis in part because the liability costs of public and envi-ronmental health may be significant. The slow uptakemay also partly be due to the siloed institutional frame-works which are geared to augment supply (e.g. by build-ing larger pipes) rather than to manage demand (e.g. byintroducing low flush toilets). The challenge of providingrobust management of domestic and industrial sewage isbecoming increasingly urgent as the majority of sewageinfrastructure in the industrialized world will require

retrofitting and replacement in the near future and morethan half of the people living in megacities in the devel-oping world lack access to centralized sanitation services

[6], [7]. Growing costs in combination with environmentalconcerns and the challenges involved in securing thequality and quantity of water heighten the urgency of theissue.

It has been repeatedly shown that successful imple-mentation of robust wastewater management solutions isintricately tied to environmental, social, economical and

political aspects at different scales and thus requires ac-tive engagement of a variety of experts, in addition towastewater engineers [6], [8], [9]. Identifying the mostsustainable solution involves finding solutions that min-imize negative effects, while maximizing benefits for localand global environments. The challenge is considerable; itis context-dependent and multi-dimensional in whichcompeting objectives must be identified and trade-offsmade. Decision makers are scrambling to identify thebest solution for their specific context: but they simplydo not have sufficient resources to carry out an integratedanalysis, and they generally settle on the traditional solu-tion [10], [11], [12].

Guest et al. [12] discuss the challenges with sustainablewastewater management, and that it is necessary to bringin multiple perspectives when identifying possible solu-tions. Decision support systems (DSSs) can, for example,be designed to allow input from different parties in-volved in the decision making process as planners navi-gate through complex problems [13]. Several DSSs havebeen developed to aid decision making in wastewatermanagement [14], [15], [16], [17], [18], [19]; see [20] for areview. According to Hamouda et al. [20], most DSSsfocus almost exclusively on the technical and economicaspects of wastewater, while what is needed is a more

B.C. Chamberlain, G. berg and H. Taheri are with the Institute for Re-sources, Environment, and Sustainability at the University of British Co-lumbia, Vancouver, B.C. Canada, V6T 1Z4 E-mail:[email protected], [email protected],

[email protected] G. Carenini and D. Poole are with the Department of Computer Science, at

the University of British Columbia, Vancouver, B.C. Canada, V6T 1Z4 E-mail: [email protected],[email protected]

anuscript received October 1, 2012

A


2/14

2 IEEE TRANSACTIONS ON COMPUTERS, SPECIAL ISSUE COMPUTATIONAL SUSTAINABILITY, TCSI-2012-10-0740

comprehensive approach, which considers social, legal,environmental and other aspects of wastewater manage-ment [21]. We fully support the argument that there is aneed for a DSS that allows input from a variety of expertsand key-stakeholders when analyzing the impact of val-ues, trade-offs and consequences

The purpose of this paper is to present a DSS proto-type that we developed in order to help communities ef-fectively explore the design space of sustainablewastewater solutions that is relevant for their particularcontext, and make it possible for them to identify solu-tions that balance environmental, economic and socialneeds. The system has been initially developed for plan-ners and their consultants with the intent to be developedfor use by the public. We hypothesize that such a systemmust be scalable, adaptable and flexible to allow fair as-sessment of new ideas and technologies. We also hypoth-esize that it would be most beneficial if the DSS allowsusers to explore the consequences of various alternativesunder different scenarios and visualize the trade-offs be-tween them. To enable the different parts to work to-

gether, the DSS should be based on a common languagethat allows the translation of different types of infor-mation and concepts between different users of the sys-tem. To accomplish this, we have developed a model-based group DSS (GDSS), which is designed to supportbrainstorming and idea evaluation, and to facilitate theidentification of sustainable solutions to challenging mul-ti-dimensional multi-stakeholder problems [13], [22], [23].Our DSS is designed to accept three types of informationprovided by three separate groups:

1. Information about the physical components of asystem that can be arranged to create a sustaina-

ble sewage management system. This infor-mation is provided by engineers, inventors, tech-nology firms, et cetera.

2. Context specific information (regulatory, demo-graphic, geographical, etc.) provided by, for ex-ample, planners (municipality, city, region, etc).

3. Information on values, preferences and predic-tions provided by various stakeholder represent-atives (e.g. elected officials, the public, special in-terest groups, NGOs).

The architecture of our DSS is based on three keypremises. First, that the system is built on an open plat-

form with an explicit vocabulary and taxonomy of thevarious technical and non-technical aspects of sustainablewastewater management. Second, that the system has thecapacity to automatically generate alternative solutions.Third, that the system is able to effectively communicatethe trade-offs between these alternatives and simultane-ously allow users to explore how their values influencethe outcome, i.e. which alternative is assessed to be thebest solution and why.

The architecture of our DSS, along with a descriptionof our methodology, are presented in the following sec-tion, accompanied by an explanation of how these threepremises were addressed with specific examples of the

DSS. In Section 3, we describe our approach in detail. Sec-tion 4, discusses the system in the context of sustainablewastewater management, possible improvements andnext steps.

2 ARCHITECTURE OF THE DSS

The architecture of our DSS, as shown in Fig. 1, comprisesseveral data structures and two software modules. Thedata structures are built on an ontology, derived from anumber of imported domain-independent and domains-specific ontologies, which provide an explicit vocabularyand taxonomy for sustainable wastewater management.These structures include system components, the com-munity context, user values, and the properties and rela-tionships each of these have with each other. The designgeneration module can automatically generate a largenumber of alternative wastewater system designs. Thismodule is useful for exploring the range of possible de-signs and encouraging brainstorming between technicaland non-technical users. The decision aid module is anintuitive and interactive visualization system which al-lows users to easily select their preferred design(s) bycomparing trade-offs between a subset of the solutionsgenerated by the previous module.

Fig. 1 The data structures (ovals) and two software modules (rectangles) ofthe sustainable wastewater decision support system.

2.1 OntologyPlanning the renewal, retrofitting or expansion of awastewater system requires input from a variety of peo-ple with diverse knowledge and expertise. It is a well-known phenomenon that efficient communication amongdiverse groups is hampered by their use of different vo-cabularies and language constructs. Efficient communica-tion is facilitated by the creation of a common vocabulary,which allows different groups to understand and com-municate with one another. Computer-based ontologies

are designed to specify the meaning of the vocabularyused in an information system [24]. Ontologies enableinformation sources to inter-operate at a semantic leveland to facilitate the adherence of different informationsources to a common terminology for the same things[24]. Such ontologies are expected to be defined by thedomain community and evolve as a new vocabulary isdefined. In many areas of science, scientists are develop-ing ontologies for their field. Two examples include theOpen Biological and Biomedical Ontologies1[25] for med-

1


3/14

CHAMBERLAIN ET AL.: A DECISION SUPPORT SYSTEM FOR THE DESIGN AND EVALUATION OF SUSTAINABLE WASTEWATER SOLUTIONS 3

icine, and OneGeology2 to define the vocabulary of geol-ogy and provide open access to geospatial map data.

An ontology for wastewater systems describes physi-cal components, among many other concepts related towastewater. Examples of physical components wouldinclude: pipe, activated sludge system and energy recov-ery system. A pipe is a type of transportation agent withproperties, such as diameter and material, and relation-ships to other entities described using the same or differ-ent ontologies. Similarly, an activated sludge systemwould be a type of treatment system with properties suchas operating temperature and volume capacity. An ener-gy recovery system would be a type of resource recoverywith properties such as operating temperature and recov-ery efficiency. There could be other ontologies that, forexample, describe processes, community contexts,wastewater constituents and related indicators. A processontology would include, for example, denitrification, dis-infection, and odor removal. A few computer-based on-tologies have been developed for wastewater systems[26], [27], [28], though these focus primarily on the tech-

nical aspects or operation and maintenance of a plant.These ontologies are not adequate by themselves for mu-nicipalities facing the challenge of renewing, extending orretrofitting their systems, as they do not include social,economic and environmental aspects.

Researchers in other domains have, however, devel-oped conceptual models that relate various wastewaterinfrastructure and treatment systems to social, economicand environmental aspects [29], [30] (also see the Sustain-able Sanitation and Water Management toolkit3). Drawingon this work, we have developed some prototype ontolo-gies where the properties, which define physical compo-nents and wastewater products, are formally specified.

Fig. 2 shows a diagram of a simplified example of atraditional linear wastewater system, with specific partsof this system identified (e.g. settling tank).In the ontology, settling tank is a subclass of Component.In the OWL Web Ontology Language4, this would bewritten as:

2http://www.onegeology.org/3http://www.sswm.info/4 -

subClassOf(SettlingTank Component)

Other examples of OWL which describe parts of thewastewater system in Fig. 2, include:

Fig. 2 Simplified example of linear end-of-pipe wastewater system. Thenumbers correspond to a product associated with each component and dis-

posal method.

subClassOf(Component PhysicalObject)

subClassOf(Product PhysicalObject)

Describing the various parts of a wastewater systemusing an explicit definition of the properties, domains andranges, enables other ontologies to interoperate and reusethese descriptions. For our ontology, we import the speci-fication for defining quantities, units, dimensions anddata types from [31], officially called Quantities, Units,Dimensions and Data Types in OWL and XML or QUDT.If there is some quantity we need to express in our ontol-ogy that is defined in QUDT, we use their definitions. Forexample, when we use a flow in our DSS, which is a vol-ume per unit time, we use the QUDT name: Quanti-

ty:VolumePerUnitTime . For other units, such as moneyper time unit, we define our own terminology which wewill publish so others can import these definitions. A listof some properties, their associated domains and rangesfor our DSS can be found in Table 1.

Looking back at Fig. 2 there are three products, whichhave been labeled. The first product (1) represents the

constituents in the settling tank. Say a settling tank con-tains 5% of TS (see Table 1 for a description) in the

TABLE 1:

EXAMPLE PROPERTIES,THEIR DOMAIN AND RANGES FOR THE PHYSICAL COMPONENTS IN THE DSS

!"#$%"&' )#*+,- .+-/%

!"#$#%"&'$()*%+,-+.',/ 1!(-2 34#/5&6 75',6"6*8-+,9"6*

:#6'$;#$"/9 1:;2 34#/5&6 75',6"6*8


4/14

4 IEEE TRANSACTIONS ON COMPUTERS, SPECIAL ISSUE COMPUTATIONAL SUSTAINABILITY, TCSI-2012-10-0740

wastewater. By the time the wastewater reaches the disin-fection system, some of the TSwould have been brokendown, and more would have settled to the bottom of thepond. Thus, when the product reaches the disinfectiontank (product 2), TSmay be less than 1%. The final prod-uct (3) would then be released into a water stream, shouldbiological oxygen demand (BOD) be low enough to meetlocal regulations.

The ontology is meant to facilitate the adherence to acommon terminology. It is assumed that individuals whouse the ontology will abide by the set standards, whilealso making suggestions for improvement. Such a stand-ard ensures that the information being collected and theway it is organized will be consistent.

2.2 ComponentsA database of the various entities was then derived usingthe ontology. The database contains information aboutmany physical components which is used to construct amodel of a potential wastewater system. Here, we definecomponents simply as the pieces that can be organized

together to create a sewage management system. Depend-ing upon an individuals background such pieces maybe referred to as processes, technologies, components,modules, or combinations of these. In the real world, eachcomponent has a set of constraints or limitations, requiredinputs and outputs. Whereas the database uses descrip-tions of these components to specify the types of inputs,restrictions on the inputs, the types of outputs, and howthe output is a function of the inputs and the operatingconditions. An example component is presented in Fig. 3,showing the required descriptions.

Fig. 3 Example of a component with the properties associated with thevarious parts of the component. In the diagram, the Input(s) and Output(s)

are equal to Product(s).

2.3 Community contextComponents are used to build a wastewater system, yetin order to identify the best system for a community, theinformation which defines a community context mustalso be given. A community context provides constraintsthat the system must attempt to satisfy. These constraintsmay depend on a communitys population, climate,amount of land available, regulatory restrictions, etcetera, and would be structured based on the communitycontext ontology. The context specific information wouldlikely be known by the local planning department andvarious government authorities. In the present DSS, a

community context can be based on a variable number ofproperties and constraints, such as the level of BOD (anindicator of effluent quality), capital available for con-struction and energy use. An example of a city communitycontext is presented in Fig. 4.

Fig. 4 Example of a community context with the associated properties.

2.4 Generating alternative wastewater solutionsThe design generation module is one of two softwaremodules in the DSS architecture (see Fig. 1). This moduleuses computational methods to facilitate the automatic

creation of alternative wastewater system designs. Theconcept of computer-based design generation has existedfor some time in wastewater management. In fact, as ear-ly as 1979, Rossman [32] developed a computer-basedsystem for creating an arrangement of processes for a sys-tem. One of the most recent and complex automationmethods uses a genetic algorithm for creating feasiblewastewater systems [18]. Recently, there has been a pushtoward developing sustainable wastewater systems [12],but very few use computational methods for automatingthe design of these systems. One example can be found inBalkema et al. [21], where they use an integer program-ming optimization method.

In the DSS presented, a design is a set of components,arranged together in a way that is physically possible,which satisfy the constraints specified by the communitycontext and treats the communitys output (see Fig. 4)such that there are no remaining outputs from the com-pleted design. A partial design is a set of componentswhere components are connected together, but wheresome components may have some outputs that are notyet connected to a component. Initially we treat the com-munity as a component that has an output (Fig. 5, start).A completed design (Fig. 5, complete) is a partial designwhere all outputs of components are connected to othercomponents. Designs are generated by employing adepth-first branch-and-bound search [33] maintaining a

current partial design and a cut-off bound.

Fig. 5 Abstraction of starting, partial and completed designs. CC = commu-nity context, rectangle = component, arrow = output.

A component can be added to a partial design if theinput type (see Product domain in Table 1) matches thet e of the available out ut and the constraints of the


5/14

CHAMBERLAIN ET AL.: A DECISION SUPPORT SYSTEM FOR THE DESIGN AND EVALUATION OF SUSTAINABLE WASTEWATER SOLUTIONS 5

new and existing components of the design are all satis-fied (constraints include component constraints and pa-rameters, see Fig. 3, and community context constraints,see Fig. 4). As components are added, the DSS maintainsan estimate of the costs (utility and community con-straints) of completing that design. It maintains a currentpartial design such that the actual cost plus the heuristicvalue is less than the cutoff. It then extends the currentpartial design with additional components (or reusescomponents) in all possible ways, in a depth-first manner.When solutions are found, the cutoff is reduced in a wayto find a limited number of solutions which can be com-pared (e.g., if we want to find 10 solutions to compare, itcan be set to the cost of the 10th best solution found sofar). The pseudo code below provides context of how adesign is constructed:

!"#$%"&'()%*+ - ./0112+%$3402$!2$) - ./0112+%$3 02$!2$4

#(!("$5/600)( 02$!2$ !02$!2$)

7 /01!0+(+$ )2/6 $6"$ $3!( 802$!2$9 - $3!( 8%+!2$97 "&& /0+)$#"%+$) "#( )"$%):%('

"'' /01!0+(+$ $0 !"#$%"&'()%*+#(10;( 02$!2$ :#01 02$!2$)%: /01!0+(+$ +0$ "("'3 %+ !"#$%"&'()%*+

Chamberlain IEEE Computer 2013WETLAND

Documents

Transcript of Chamberlain IEEE Computer 2013WETLAND