Water businesses have always sought to plan for drought-resilient urban water supply systems, especially in areas with extreme climate variability. The recently completed Lower Hunter Water Plan was developed to support population and business growth, and ensure reliable water supplies in drought. As well as introducing Water Wise Rules to encourage water conservation in normal conditions, the plan identified a portfolio (or combination) of drought response measures to be deployed progressively in a drought as water storage levels drop. The keys to the success of the Plan were thorough consultation with stakeholders, and collaborative analysis facilitating transparent evaluation of trade-offs among options and portfolios. A multi-criteria decision analysis process guided the assessment of the drought response options against both quantitative and qualitative criteria, and the assembly and analysis of portfolios. The process integrated the assessment of social and environmental criteria with cost effectiveness analysis, along with analysis of contextual uncertainties and risks, to determine the resilience of the options under different scenarios. The outcome was a portfolio of drought response options that best met the weighted criteria and satisfied their objectives.

## INTRODUCTION

Since the turn of the century there has been growing emphasis on involving stakeholders and the public in Australian water resource planning. Transparency in multi-criteria decision analysis (MCDA) is valued by stakeholders. Building trust in data inputs, assumptions and outputs across all stakeholders is important in the MCDA process (Kain et al. 2007). Decision making in the water sector can be complex due to trade-offs between socio-political, environmental and economic factors, especially with competing objectives and drivers. Complex and ‘black box’ analytical tools prevent meaningful links between output and assessment by a non-expert (Lai et al. 2008). Hence the approach adopted in the work described here attempted to simplify MCDA as much as possible, so that it was transparent to all stakeholders (Chowdhury & Rahman 2008; Paneque Salgado et al. 2009; Hyde 2011).

This paper draws on the development of a water plan for the Lower Hunter region, on the eastern seaboard of Australia in New South Wales (NSW). The primary aim of the Lower Hunter Water Plan (LHWP) was to ensure reliable water supplies both in drought and in the longer term, to support population and business growth.

Development of the LHWP was managed by the NSW Metropolitan Water Directorate (MWD). Long-term supply and demand modeling indicated that current water sources can meet the region's needs under typical climate conditions for around 20 years. However, the water sources are susceptible to severe droughts, so the emphasis of the first iteration of LHWP was on drought response measures.

This paper describes the MCDA process developed to guide the assessment of a wide range of drought response measures against multiple criteria (quantitative and qualitative), and to assemble and analyze drought response portfolios in preparing the LHWP (MWD 2014). It specifically outlines the engagement approach adopted and the role that community values played in the decision process.

## METHODOLOGY FOR THE ASSESSMENT FRAMEWORK

There are two fundamental inputs for an effective MCDA (Mukheibir & Mitchell 2011):

• A good process: Process inputs should be designed to ensure engaged participation from an appropriate range of stakeholders, who are well-supported in identifying meaningful criteria and diverse options. The MCDA process for the drought response strategy set out to bring together divergent views in a thorough and transparent process. It guided participants in considering many alternative supply- and demand- side options, and enabled transparent trade-offs between options and sets of options. Inputs from the community engagement were taken into account by the Lower Hunter Water Senior Officers Group (SOG), who had to advise the NSW Government.

• Good data and transparent analysis: Data inputs informed the performance assessment of each option against the agreed criteria. Studies of the potential options and their likely responses in various scenarios were undertaken by independent consultants. The outcomes were used in the modeling and analysis of future demand and supply measures in assorted drought scenarios. The assessments of social, environmental and risk/resilience criteria with economic analysis were integrated, and analysis of contextual uncertainties and risk was included.

The framework for developing the LHWP was based on well-known MCDA processes (Belton & Stewart 2002; White et al. 2006; Hajkowicz & Collins 2007; Lundie et al. 2008; Dodgson et al. 2009) and helped in decision-making. The aim of the framework was to balance the stakeholders' and community's multiple objectives. The key objectives were to (MWD 2014):

• provide water security during drought (evaluated in terms of the probability of water storages falling to 10% of available capacity)

• ensure reliable water supplies to meet growing demand

• help protect local aquatic ecosystems and the environment

• maximize net benefits to the community

The MCDA criteria used to support decisions for LHWP were assessed against seven principles, to ensure a successful options analysis process (Mukheibir & Mitchell 2011):

• Contextual: A criterion must be relevant to both the problem's context and the analysis at hand, i.e. in this case only criteria relevant to drought should be considered.

• Discerning: A criterion should distinguish between options, i.e. if all options score the same, the criterion has no meaning.

• Assessable: A criterion should be operationally meaningful, i.e. it is important that option performance is assessable, either quantitatively through physical measures or qualitatively through judgment.

• Consequential: Criteria must focus on the consequences of options and sets of options.

• Independent: Double counting must be avoided, e.g. by not counting reductions in greenhouse gas emissions and renewable energy generation as two separate positives.

• Life cycle oriented: Each criterion should take account of the options' entire life cycles and/or whole decision-making timeframe. Consistent boundaries are to be applied for each of the criteria across all options and sets of options, i.e. the same steps in the production cycle should be considered for all options.

• Distinguishable: The analysis involved comparisons of criteria pairs to apply weightings to them. Criteria should differ sufficiently for useful comparisons to be made between them.

### MCDA consultation process

Consultation and engagement were undertaken at three levels:

• community and stakeholders;

• government, via SOG, which included representatives from relevant government organizations, with overview by the Metropolitan Water Chief Executive Officers' Committee; and,

• review by an Independent Water Advisory Panel, comprising experts in urban water planning, water resource management, hydrology, economics and community engagement, before submission to the state water minister for approval.

#### Community engagement

The principal aim of community engagement for the LHWP was to include the community at each step of the process. It involved a series of workshops and online engagement, and built participants' knowledge and understanding of water planning, while providing opportunities for input as the plan was developed. It increased awareness, facilitated discussion, and mitigated the risks associated with poorly informed debate on water planning. It also allowed the planning team to understand and take account of community preferences and priorities, as well as understand the values, attitudes and behaviors that might affect option choice. The outcome was an increase in planning process transparency, plus common understanding and final acceptance of the plan (Gamble & Cole 2014).

Community and stakeholder workshops were structured to engage at four stages in the process:

1. Defining community values

Defining the set of values was important, firstly to understand community perspectives about water planning and later as a basis for community input to prioritizing options. Consistency with community values was an important criterion for assessing options. In addition to the overarching value of ‘a process we can trust’, the community values developed through the first series of workshops were (MWD 2014):

• sustainable solutions and water conservation;

• a fair and affordable system;

• safe, healthy water for all uses;

• protecting the natural environment;

• a secure, reliable supply for all;

• a strategic, balanced and adaptable plan;

• investing dollars wisely – i.e., ‘on the right things’; and,

• respecting the Aboriginal cultural value of ‘life water’ (added following a subsequent workshop with Aboriginal community representatives).

• 2. Understanding the supply-demand balance and the broad range of options

The second round of workshops involved discussion about the demand forecast, and the broad categories of supply and demand options. These workshops were designed primarily to share information with community and stakeholder participants, to prepare for the next phase of deliberation.

• 3. Assessing whether the proposed options were consistent with community values

A third round of workshops was held to discuss the options in more detail, introduce the concept of portfolios (combinations of options), and obtain feedback on how well the options aligned with community values. The options' key features were presented a few at a time, and participants selected those that they considered best reflected community values. The most favored options were combined into a sample portfolio, thus demonstrating this concept. The number of times each option was selected was counted, and the total count from three workshops used as an input into the MCDA for the criterion ‘consistency with community values’. This approach applied feedback from the community directly to the MCDA in a manner that avoided risk of bias or misinterpretation.

• 4. Ranking the portfolios of options based on the set of values

A final set of workshops was held to engage the community and stakeholders in evaluating potential portfolios (combinations of options). A discussion paper publicly released prior to these workshops provided supporting information and invited feedback, via the workshops or as written submissions. Workshop participants considered trade-offs between cost, drought security and environmental features, with respect to six potential portfolios, and then ranked them, recording the reasons for preferences. This provided both quantitative data (rankings) and qualitative information (reasons), which were combined with expert evaluation to select the recommended portfolio for the LHWP.

For each stage, one workshop was held with a Representative Community Group, and others with representatives of stakeholder organizations and self-selected community members, who responded to an open invitation in newspaper advertisements. The Representative Community Group was recruited to represent a cross-section of the region's population. Sixty participants were recruited for the first workshop, with the intention that the same group would participate throughout the process, while recognizing that there would be losses over time. Young people, those from culturally and linguistically diverse backgrounds, and Aboriginal people proved to be the hardest groups to recruit from. An Aboriginal facilitator was engaged to assist with a separate workshop to involve Aboriginal community members.

The activities at each set of workshops were designed to integrate with the planning framework, so that community feedback provided data for direct incorporation in decision-making. Figure 1 illustrates this by showing what was asked, what the community said, and how feedback was used from each set of workshops. The need for openness and transparency was considered paramount. Participants were always advised how their input would be used and the consolidated feedback from each set of workshops was reported back to the next set of workshops. This helped to validate, and where necessary amend, the input (for example, on community values), while also demonstrating that participants at different workshops might express a range of views, but that all were important and being heard.
Figure 1

Operating structure of the community and stakeholder workshops.

Figure 1

Operating structure of the community and stakeholder workshops.

An online engagement platform called ‘Have Your Say’, established by the NSW Government, provided opportunities for the community and stakeholders to participate via discussion forums, surveys and submissions. The planning team used this resource throughout the process to invite community feedback on values, water supply and demand options, and portfolios. The specific web page for this project recorded 8,651 visits over the project's life, with 3,166 document downloads.

#### Inter-agency engagement

The MWD led a collaborative, whole-of-government process to develop the LHWP. The SOG was established as a cross-agency committee of representatives from key government organizations to contribute to all stages of the process. Some of its main tasks included:

• screening a long list of options;

• assisting in community and stakeholder engagement workshops;

• agreeing on and weighting the MCDA criteria used to rank the options;

• evaluating option portfolios to derive a preferred portfolio of drought response measures; and,

• endorsing the LHWP before it was submitted to the NSW Government for approval.

Within the overall governance structure, the SOG advised the Metropolitan Water Chief Executive Officers' Committee, ensuring high-level interagency agreement before the plan was submitted to the NSW Government.

### MCDA

The MCDA process was designed to encourage discussion and debate around the qualitative issues relating to options and combined-option (portfolio) implementation. It also introduced a step in which the investment approach was defined, to guide portfolio building based on option ranking. Several steps were involved (Figure 2):
1. undertaking a risk and uncertainty analysis to inform the development of the options assessment criteria, and for future sensitivity analysis;

2. developing criteria for option and portfolio assessment, and assigning weightings to them;

3. assessing the options against the relevant criteria;

4. ranking the options by assigning aggregate scores to them as weighted sums of ratings against criteria;

5. defining the drought response investment approach to guide the construction of drought portfolios;

6. assembling portfolios of options;

7. modeling many possible hydrologic scenarios, to assess portfolios against objectives and arrive at a preferred portfolio of measures;

8. community engagement on values, options and portfolios (this was integrated across the steps above);

9. assessing the portfolio sensitivity to the significant risks identified in Step 1; and,

10. ranking the portfolios on cost, drought security, environmental impacts, community preferences and risk.

Figure 2

MCDA process diagram.

Figure 2

MCDA process diagram.

#### Analyzing risks and uncertainties

To assess the effectiveness of the drought response options and portfolios, SOG identified risks and uncertainties that could affect them. Once identified, each risk was addressed in one of four ways:

• use as a basis for assessing option resilience in scenarios;

• use as a criterion for ranking options;

• undertaking a portfolio risk sensitivity analysis at the end of the process; and,

• ignoring them, if they only related to the broader supply-demand balance, which would be dealt with through the regular demand-supply strategy plan.

#### Developing criteria for assessing the options and portfolios

Choosing relevant criteria was iterative. A preliminary list was developed, drawing from the available literature and project team knowledge. The SOG considered how to assess performance against each criterion, and how to undertake transparent and meaningful assessment that maintained important distinctions. The draft criteria set was also checked for consistency and alignment with:

• the LHWP objectives;

• the problem to be addressed by the LHWP – dealing with vulnerability to drought; and,

• the key community values discussed earlier.

Later the criteria were revisited with the options in mind, to ensure that they differentiated sufficiently between options. If a criterion would achieve similar scores across the range of options, it was not considered useful for the MCDA.

The analytical criteria needed to be kept to a minimum – i.e. the set of criteria should include only those needed to solve the problem. Assessment against the criteria should not require too many assumptions about the future or lead to second-guessing. A key issue at this stage was consideration of the metric (and its range) assigned to each criterion, i.e. how performance against each criterion would be assessed.

The multi-criteria analysis combined expert and community input to help assess the options by comparing:

• the cost of improving drought security;

• how well each option reflected community values;

• the level of certainty of implementation of the option;

• potential to impact on the natural environment; and,

• flexibility to respond to drought in stages, without locking out other options in future.

#### Assigning criteria weightings

Since not all the criteria are equally relevant in MCDA, they were assigned relative weightings (Table 1). The SOG ranked them by weighting each relative to the others. Each was ranked according to its importance by comparison, i.e. each was compared individually with all other criteria. This avoided the manipulation that can occur when percentages are simply assigned.

Table 1

Assessment criteria and relative weightings

Criterion definition and description units and weighting
Cost of water supplied or saved per unit of improved drought security Risk-adjusted cost per unit of contribution to drought security over the assessment period (15 years). The contribution to drought security is an index based on the risk of the water storages reaching minimum operating capacity (10%). Maximum, minimum and mean values were reported. Costs exclude externalities (e.g. community value placed on unsatisfied demand) Units: A/unit of risk reduction. Relative Weighting: 27% Consistency with community values Degree to which the option is consistent with community values based on outcomes of community engagement, where participants selected options that aligned with the six community values. (three workshops and six values = potential score of 18) Units: Cumulative score out of 18 Relative Weighting: 13% Controllability The degree of certainty with which implementation or uptake can be guaranteed when an option's performance depends on third party discretion – e.g. water recycling opportunities not taken when needed. This criterion should not to be confused with reliability, which is considered as part of the drought security modeling. Units: relative ranking Relative Weighting: 19% Natural environmental impact Consideration of the environmental impact during a drought, e.g. loss of land use, and/or damage to fauna, flora and aquatic life (ecosystem disturbance). This assessment should not include the cost of off-sets or GHG emissions, which belong in the cost analysis. Heritage concerns should not be considered here, since they are site-specific. Units: relative ranking Relative Weighting: 19% Flexibility to change Assessment is based on ability to supply/save water in an incremental (modular) manner, i.e. the degree that it can be scaled up (staged) if necessary when further supplies/savings are needed. This criterion also reflects whether or not up-scaling can be delayed, when new technology or other changes require that another option be pursued (i.e. whether or not it locks out any other options in future). Units: relative ranking Relative Weighting: 22% Criterion definition and description units and weighting Cost of water supplied or saved per unit of improved drought security Risk-adjusted cost per unit of contribution to drought security over the assessment period (15 years). The contribution to drought security is an index based on the risk of the water storages reaching minimum operating capacity (10%). Maximum, minimum and mean values were reported. Costs exclude externalities (e.g. community value placed on unsatisfied demand) Units: A/unit of risk reduction. Relative Weighting: 27%
Consistency with community values Degree to which the option is consistent with community values based on outcomes of community engagement, where participants selected options that aligned with the six community values. (three workshops and six values = potential score of 18) Units: Cumulative score out of 18 Relative Weighting: 13%
Controllability The degree of certainty with which implementation or uptake can be guaranteed when an option's performance depends on third party discretion – e.g. water recycling opportunities not taken when needed. This criterion should not to be confused with reliability, which is considered as part of the drought security modeling. Units: relative ranking Relative Weighting: 19%
Natural environmental impact Consideration of the environmental impact during a drought, e.g. loss of land use, and/or damage to fauna, flora and aquatic life (ecosystem disturbance). This assessment should not include the cost of off-sets or GHG emissions, which belong in the cost analysis. Heritage concerns should not be considered here, since they are site-specific. Units: relative ranking Relative Weighting: 19%
Flexibility to change Assessment is based on ability to supply/save water in an incremental (modular) manner, i.e. the degree that it can be scaled up (staged) if necessary when further supplies/savings are needed. This criterion also reflects whether or not up-scaling can be delayed, when new technology or other changes require that another option be pursued (i.e. whether or not it locks out any other options in future). Units: relative ranking Relative Weighting: 22%

#### Assessing potential options

To assess the options against the qualitative criteria, the SOG determined their relative effectiveness or optimal performance against each criterion, and options were ranked from best to worst. The ranked options were assigned scores according to the number(s) of options considered: the highest-ranked receiving the highest score and the lowest-ranked "zero". Whilst the focus in this step was on assessing the options against the criteria, the conversations about ranking were equally important in determining the outcomes.

At the options stage, Hunter Water's ‘Drought Portfolio Evaluation Model’ (DPEM) was used to determine the expected present value cost and the probability distributions of option costs by simulating 10,000 different 15-year hydrologic sequences, including a wide range of possible droughts, using a hydrologic model and recording the costs reported. These were then characterized statistically.

Hunter Water's ‘Source Model’ (SoMo) was used to determine the contribution of the options to drought security, using 250,000 synthetic climate sequences. Drought security was assessed based on the average risk of storage levels dropping below 10% over the analysis period. The risks estimated using SoMo were determined by running the model with a constant underlying demand for the 15 years. The demand was set at that estimated for 2029, i.e. 70.9 GL/year (the forecast demand in 2029, 15 years ahead of the time of the study). The demand simulated for any day is a function of the time of year and climate, and whether or not restrictions are imposed.

#### Prioritizing options by assigning aggregate scores to them

Based on the process inputs described, the aggregate score for each option was calculated using a weighted sum of the criteria ratings. A simple spreadsheet was used to allow for transparency and enable stakeholders to examine the data. The outcome is shown in Figure 3.
Figure 3

Ranking of options from multi-criteria decision analysis.

Figure 3

Ranking of options from multi-criteria decision analysis.

A sensitivity analysis of the criteria was undertaken by sequentially increasing the magnitude of the weights for each criterion by a factor of two and adjusting the others down proportionately, so that the sum of the weights was still 100%. The new weights were then checked to see whether the option order had changed significantly.

The performance of each option against the criteria was used in developing the option portfolios.

#### Defining the drought response investment approach

The assembly approach for each portfolio was defined in order to construct the portfolios in a sensible manner. The highest ranking options were considered in priority order. It was evident from the rankings that the demand side measures (DSMs) were clustered together at the top and it was considered important to include them as a set, because they ranked consistently highly across all criteria in all portfolios. All portfolios also included the existing set of water supply sources, water efficiency programs and recycled water initiatives that were part of the ‘base case’.

• Approach 1: consider only the highly ranked DSMs

• Approach 2: in addition to the highly ranked DSMs, include inter-regional transfers under existing agreements with neighboring regions, alone and in combination with the next best supply side options

• Approach 3: in addition to the highly ranked DSMs, include alternative inter-regional transfers from under-utilized irrigation dams.

#### Assembling option portfolios

Option portfolios were assembled based on these approaches (see Table 2). The options were selected by giving priority to those with the highest weighted aggregate scores.

Table 2

Portfolios developed from ranked options

Abbreviations: DSM – Demand side measures, CC – Central Coast water transfers, SW – Stormwater, TD – Temporary desalination, LD – Lostock Dam.

Portfolio 1 - options that work on the demand side of the supply–demand balance. This set of six ranked consistently high across all criteria in the multi-criteria analysis and was considered important to include in all portfolios as they would underpin any drought response.

Portfolios 2–4 - as Portfolio 1 but including water transfer from the Central Coast, with or without other options such as stormwater use and temporary desalination. These portfolios were developed by adding different combinations of measures to boost the supply of water, selecting from those options that had a medium to high ranking in the multi-criteria analysis.

Portfolios 5 and 6 - including accessing water from Lostock Dam as an alternative supply option. Portfolio 5 involved using water available from the existing dam by buying licenses on the water market, and constructing a new treatment plant and pipeline. Portfolio 6 also involved enlarging the dam to make more water available in a drought.

#### Portfolio evaluation

The portfolios were compared, and assessed against cost and the relevant LHWP objectives:

• providing water security during drought;

• helping protect aquatic ecosystems; and,

• maximizing benefits to the community.

In the portfolio phase of modeling, SoMo was used to determine the contribution of the portfolios to drought security using 1,000,000 synthetic climate sequences. The DPEM was used to determine the expected costs of the portfolios and their probability distribution, using a method similar to that used for assessing options. Consideration of lead-in times and triggers was important in this step.

When assessing the portfolios, it was also important to account for interactions between options operating together, to identify possible issues that might compound and/or undermine impacts or benefits.

The environmental impact of the portfolios was assessed qualitatively by the SOG, comparing the portfolios against each other in a pairwise manner.

#### Sensitivity analysis

The significant quantitative risks and assumptions identified earlier were used in a sensitivity analysis of the portfolios based on Hunter Water's models. This gave an indication of each portfolio's resilience to future contextual changes, through either its flexibility or the diversity of its options.

Each portfolio was rated low, medium or high in its sensitivity to qualitative uncertainties. Discussion of portfolio sensitivity against the qualitative uncertainties with the SOG revealed where mitigation actions would be required in each case.

#### Portfolio ranking

Finally, the portfolio that best met the LHWP objectives was identified by ranking. This was based on the qualitative and quantitative inputs from the modeling, the community engagement outcomes, and the SOG's deliberations. A workshop with SOG members evaluated and ranked the portfolios against the planned objectives, focusing on trade-offs between drought security, environment and cost, and taking into account the outcomes of community engagement.

The weighted portfolio scores in the ranking process were established by members of the SOG assigning the maximum score to the highest ranked portfolio, with a score of 1 for the lowest ranked. A portfolio's final score was the aggregate of all its scores, expressed as a percentage of the sum of all the portfolio scores. While the objective was not necessarily for the SOG to arrive at a consensus, discussion of the trade-offs between portfolios was informative.

## OUTCOME

### Cost versus drought security

Hunter Water modeled the portfolios against thousands of possible weather scenarios, using SoMo and the DPEM to determine water storage levels and the costs of implementing the measures in accordance with defined trigger levels. The scenarios included hypothetical extreme droughts. A period of 15 years was adopted as a realistic modeling horizon and drought security was expressed in terms of the probability of storage volumes falling to 10% or lower (expressed as 1 in X years).

The inputs for this included:

• capital and operating costs (as a whole-of-society cost);

• the volume of water supplied or saved;

• the lead times for implementation of the various options in the portfolios; and,

• sensitivity analysis, by varying the trigger levels for some options, to see how the level of security changed and at what cost.

Modeling of the level of security and the portfolios' mean present value is illustrated by Figure 4. Security is expressed as the likelihood of not falling below 10% storage (as 1 in X years). Higher X values represent higher security. A red line connects the portfolios yielding the lowest costs. The base case is provided in Figure 4 as a reference, i.e. no drought investment cost, with a drought security of only 1 in 5,000 years.
Figure 4

Mean present portfolio value versus security of supply achieved.

Figure 4

Mean present portfolio value versus security of supply achieved.

Analysis of the portfolios against the LHWP objectives reveals that Portfolio 1 (DSM only) has the lowest cost but provides the lowest level of security. The portfolio that includes DSM, transfers from the Central Coast and temporary desalination (Portfolio 4), provides water security during a drought closer to 1:100,000, with an expected mean present value cost that is approximately A$11 million more than that for Portfolio 1. The portfolio that includes DSM and transfers from the Central Coast, and without desalination (Portfolio 2), costs$1.5 million more than Portfolio 4, and provides 20% less security. Note also that Portfolio 3, with stormwater, costs significantly more than Portfolio 2, while achieving negligible additional security.

### Environmental impact

The members of the SOG ranked the portfolios based on their expected levels of environmental impact. Portfolio 6 (including an enlarged dam) was expected to have the highest environmental impact and was followed by Portfolio 4 (incorporating temporary desalination facilities). The Portfolio 1 (with DSM only) was expected to have the least environmental impact.

### Sensitivity analysis of portfolio risks

The portfolios were evaluated by the SOG for sensitivity to the significant risks and uncertainties identified earlier, to determine their robustness (resilience). This was assessed in relation to five significant uncertainties – three qualitatively and two quantitatively. For the qualitative risks, the portfolios were rated as having high, medium or low likelihood of sensitivity to the uncertainty/risk concerned.

1. Reliance on similar options/sources of water:

Reliance on a single measure or measures, limited in their variety, reduces a portfolios' adaptability to respond to droughts. Droughts vary in severity, duration, and rate of onset, and it is advantageous to have a diverse range of response measures available. Portfolio 1 (DSM only) was identified as having ‘all ones eggs in one basket’ and did not provide sufficient diversity. Portfolio 4 (DSM + Central Coast transfers (CC) + temporary desalination (TD)) was viewed as having the widest diversity of response measures.

2. Energy availability during long-term drought:

A severe drought could affect coal-powered electricity generators and hence security of electricity supply. Rolling brown outs, or total black outs, could undermine options that rely on energy to deliver the expected supply or savings. Portfolio 1 (DSM only) was least dependent on energy supplies to achieve its expected water supply savings. Portfolio 4 (DSM + CC + TD) was viewed as being most dependent on energy supplies.

3. Capital cost of options increases during drought:

The cost of securing construction materials or water treatment equipment could be higher during a drought, due to increased demand and/or opportunists seeking to make money out of an emergency. Portfolio 4 (DSM + CC + TD) was viewed as being most at risk.

4. Community responds in unexpected ways:

If public compliance with restrictions is unexpectedly low, this will reduce water savings. A sensitivity analysis was conducted with respect to sensitivity to variations in volumetric savings from water restrictions. The DSM response options are common to all portfolios, so all are sensitive to reduced savings from water restrictions.

5. Increase in demand or a new large user comes on line:

Sensitivity analysis was conducted on the cost and drought security of portfolios, using the LHWP's high composite demand forecast scenario. The demand forecast sensitivity analysis included a risk-based assessment of usage by individual customers and a review of the potential for loss/gain of customers based on macro-economic trends.

### Combined weighted outcome

Using this information and feedback from the community engagement, the SOG ranked the portfolios from most to least preferred (as shown in Figure 5). The outcome showed that Portfolios 2 (DSM and transfers) and 4 (DSM, transfers and temporary desalination) were equally ‘most preferred’. A large proportion of the organizational groupings did not rank Portfolio 1 (with DSM only) highly since they argued that it was included in all other portfolios and on its own provided the lowest level of drought security. The results were similar to the preferences expressed through the community and stakeholder engagement workshops, which ranked Portfolios 2 and 4 the highest.
Figure 5

Weighted MCDA outcome.

Figure 5

Weighted MCDA outcome.

## DISCUSSION AND CONCLUSION

The approach undertaken in this MCDA met the criteria stated at the start of this paper, i.e. a good process, with good data and transparent analysis.

The process followed ensured a structured means of integrating the multiple objectives of the LHWP, and provided a process for weighing up the performance of the options available against these goals. The consultative process ensured transparency between and within the different levels of stakeholder consultation. The involvement of community-based participants in the planning and review processes improved their water literacy and level of engagement. While the community representatives had no opportunity to vary the assumptions or weightings used in modeling or multi-criteria analysis, they were able to provide informed feedback to the planning team.

The most up-to-date data were used on all relevant options for analysis, and transparently reported. This enabled discussion of the trade-offs between the objectives as levels of achievement changed with different options and option portfolios, rather than relying on answers from a black box approaches. The MCDA approach enabled structured and transparent synthesis of qualitative and quantitative values, and preferences without the need to resort to disputable monetary ‘equivalents’.

While community values were specifically incorporated in the MCDA as a criterion, they received the lowest weighting from SOG, whereas cost criteria were ranked highest. Considering non-financial issues on an equal footing with financial ones remains a challenge when assessing outcomes with social, environmental and liveability benefits.

This MCDA consultation process demonstrated that a well-conceived and integrated community engagement program is critical to robust planning because of the need to respond to broader goals and values, which can lead to a more acceptable outcome.

## ACKNOWLEDGEMENTS

The authors wish to thank the members of the Lower Hunter Water Senior Officers Group, Metropolitan Water Directorate staff, Hunter Water Corporation staff, and the stakeholder representatives and community members who participated in the engagement workshops.

## REFERENCES

REFERENCES
Belton
S.
Stewart
T. S.
2002
Multiple Criteria Decision Analysis. An Integrated Approach
.
,
Massachusetts
.
Chowdhury
R. K.
Rahman
R.
2008
.
International Journal of Environmental Science & Technology
5
(
2
),
195
204
.
Dodgson
J. S.
Spackman
M.
Pearman
A.
Phillips
L. D.
2009
Multi-criteria analysis: a manual. Prepared for the
UK Department for Communities and Local Government
,
London
.
Gamble
R.
Cole
C.
2014
An integrated approach to community engagement
.
Water (AWA)
(February)
,
41
(
1
),
41
44
.
Hajkowicz
S.
Collins
K.
2007
.
Water Resources Management
21
(
9
),
1553
1566
.
Hyde
K. M.
2011
.
Water Practice & Technology
6
(
1
),
1
9
.
Kain
J.
Kärrman
E.
Söderberg
H.
2007
.
Proceedings of the Institution of Civil Engineers – Engineering Sustainability
160
(
2
),
87
93
.
Lai
E.
Lundie
S.
Ashbolt
N. J.
2008
.
Urban Water Journal
5
(
4
),
315
327
.
Lundie
S.
Ashbolt
N. J.
Livingston
D.
Lai
E.
Kärrman
E.
Blaikie
J.
Anderson
J.
2008
Sustainability Framework. WSAA Occasional Paper
.
Water Services Association of Australia
,
Sydney
.
Mukheibir
P.
Mitchell
C.
2011
Planning for resilient water systems – a water supply and demand investment options assessment framework. Prepared for the Smart Water Fund
.
Institute for Sustainable Futures: Institute for Sustainable Futures, University of Technology
,
Sydney
.
MWD
2014
Lower Hunter Water Plan
.
Metropolitan Water Directorate
,
New South Wales
,
Australia
.
P.
Corral Quintana
S.
Guimarães Pereira
Â.
del Moral Ituarte
L.
Pedregal Mateos
B.
2009
.
Ecological Economics
68
(
4
),
990
1005
.
White
S.
Fane
S.
Giurco
D.
Turner
A.
2006
Putting the economics in its place: decision making in an uncertain environment
. In:
Ninth Biennial Conference of the International Society for Ecological Economics
.
New Delhi
,
India
.