Modeling Water Demand in Droughts (in England & Wales) - IMA 2016

Anderson, B. & Nagarajan, M (2016) Modeling Water Demand in Droughts (in England & Wales). Paper presented at the International Microsimulation Association European Meeting 2016, Budapest, Hungary. 22nd September 2016.

Transcript

1. Modeling Water Demand in Droughts (in England & Wales) (Estimating Scenarios for Domestic Water Demand Under Drought Conditions in the UK: Application of an Agent-Based Microsimulation Model) Magesh Nagarajan & Ben Anderson Sustainable Energy Research Group Energy & Climate Change Division, Faculty of Engineering & Environment
2. @dataknut: Modeling Water Demand in Droughts Contents  The problem  Model Framework  Concepts & Implementation  Preliminary Results  Next Steps 2
3. @dataknut: Modeling Water Demand in Droughts The problem: water 3 205 0  With no ‘behaviour’ change and no flow controls: Source: DEFRA, 2011
4. @dataknut: Modeling Water Demand in Droughts The problem: water 4  Supply:  Locally/regionally scarce  Climate change effects?  Demand:  50% used by households  Drivers not well understood  Climate change effects?  Demographic  Population growth  Increasing single person households Source: Environment Agency, 2008
5. @dataknut: Modeling Water Demand in Droughts The problem: drought is normal 5 Source: Water UK (2016) Water resources long term planning framework (2015-2065) Water Resources Long Term Planning Framework Water UK Atkins | Mott MacDonald | Nera | HR Wallingford | Oxford University Technical Report | Final | 20 July 2016 53 5. Is there a problem? 5.1. Analysis of Drought Coherence, patterns and severity 5.1.1. Evidence from Historic Droughts By using the aridity indices described in Section 4.3.1.2, it was possible to examine the spatial nature of the significant droughts that occurred within the 20th Century. Nearly all water companies now plan their resources to be able to meet these events, with a ‘median’ allowance for expected climate change impacts. However, it is important to understand the nature and patterns of the droughts within the historic record in order to create ‘plausible’ Drought Configurations for the portfolio evaluation and resilience testing. A summary of some of the most informative findings from the analysis of historic droughts is provided below. It should be noted that these representations sometimes contain different years in the same plot – e.g. the 1932-34 and 1995/96 ‘worst’ point in time varied across the country. This is commented upon where appropriate in the figures. Drought Event: Short Duration Aridity Index (12 months ending summer or late autumn) Drought Event: Longer Duration Aridity Index (24 months) Notes/Comments 1901-03: worst point for short duration was December 1901; worst point for long duration was December 1902. Short duration not severe enough to challenge resources. 1921-22: worst point for short duration was December 1921, worst point for long duration was December 1922. Long duration not sufficient to challenge resources – drought stress was exacerbated by the ‘extension’ of the 1921 event into early 1922. 1932-34: Multi-dry winter event; worst short duration occurred at different points spatially (hence apparent coherence). Short duration not sufficient to cause stress – 2 year event in all areas, but varying between 1932/33 in some areas versus 1933/34 in others. Water Resources Long Term Planning Framework Water UK Figure 6-19 Demand growth under upper population scenario, BAU Base strategy, 2040 (left) and 2065 (right) – by Supply Area (top) and Region (bottom) 1901-03 1921-22 1932-34 1976 1995/6 + others (including 2011/12) And it may get worse…
6. @dataknut: Modeling Water Demand in Droughts The problem: current practice 6 Sources: Water UK (2016) Water resources long term planning framework (2015-2065), Essex & Suffolk Water, Daily Mail Water Resources Long Term Planning Framework Water UK Figure 3-4 Illustration of typical sequence of drought interventions (taken from the Affinity Water Drought Plan) Example diagram of a drought trigger-response system. The purple and blue lines represent theoretical monitored groundwater levels during a two or three year event respectively. The green, yellow, orange and red bands represent ‘thresholds’ that are based on an analysis of historic records, and are used to help inform the company when it is making decisions on the level of demand restrictions and supply side interventions to take. The y-axis in this indicative diagram, presents the groundwater level (in metres above ordnance datum, mAOD).
7. @dataknut: Modeling Water Demand in Droughts Contents  The problem  Model Framework  Concepts & Implementation  Preliminary Results  Next Steps 7
8. @dataknut: Modeling Water Demand in Droughts IMPETUS: joined-up modelling… 8 RCUK Funded under the UK Droughts & Water Scarcity Programme 2014-2017 IMPETUS: Improving Predictions of Drought for User Decision-Making Meteorological Models Hydrological models Demand models
9. @dataknut: Modeling Water Demand in Droughts The Water Demand Model 9 •‘Normal’ demand •Drought phase Inputs •Impact of ‘drought’ •Impact of ‘interventions’ Microsimulation Model •Estimated demand under drought Outputs For a given catchment… Drought phase Label Interventions Normal All indicators normal Developing Heightened risk of water deficit Voluntary abstraction restriction & efficiency measures Drought Stress on water supply Temporary use bans & efficiency measures Severe Failure of water supply Restrictions on non- essential use Recovery Returning to normality Efficiency measures
10. @dataknut: Modeling Water Demand in Droughts The Water Demand Model 10 Q1 2011 •Drought phase X -> Estimated demand Q2 2011 •Drought phase X -> Estimated demand Q3 2011 •Drought phase X -> Estimated demand Q4 2011 •Drought phase X -> Estimated demand … •… Q4 2030 •Drought phase X -> Estimated demand For a given catchment… Drought phase Label Interventions Normal All indicators normal Developing Heightened risk of water deficit Voluntary abstraction restriction & efficiency measures Drought Stress on water supply Temporary use bans & efficiency measures Severe Failure of water supply Restrictions on non- essential use Recovery Returning to normality Efficiency measures
11. @dataknut: Modeling Water Demand in Droughts Contents  The problem  Model Framework  Concepts & Implementation  Preliminary Results  Next Steps 11
12. @dataknut: Modeling Water Demand in Droughts Framework: Water Cultures 12 Water Cultures (after Stephenson et al, (2010) Energy Cultures - doi: 10.1016/j.enpol.2010.05.069) Materiality Cognitive/Cultural norms Water practices Cleanliness Cooling Greenness Income Occupancy Metering Price Garden infrastructure Appliances Washing & laundry habits Garden watering Cleaning habits
13. @dataknut: Modeling Water Demand in Droughts ‘Normal’ usage model  Ownership of devices (Pullinger et al, 2013)  Appliance litres/day (Parker, 2014) – With household attributes – With weather – By season  Water efficiency installations (EST, 2013) – 41% of HH have dual-flush toilet. – 25% of HH have efficient shower heads. 13 Data source: “At home water needs” EST (2013, p13) So what’s the point of an external use ban??
14. @dataknut: Modeling Water Demand in Droughts Intervention ‘impact’ model Impact of efficiency & temporary use bans, UKWIR 2013 report 14 Water-use appliance Usage Water- use saving % switch per year Dual-flush toilet 5 l/flush 47% 2% Other-flush toilet 9.5 l/flush Eco-shower 5 l/min 61% 1% Power-shower (Others) 13 l/min Initial values: (EST, 2013) • Key assumptions: • No change in practices (the user experience is unchanged) • Efficiency does not degrade over time • Water efficiency uptake can be varied • Key parameters: • External use ~= 11% households • TUB compliance can be varied Type of household % Water-use saving Complianc e High flow user 14% 10-18% 44% (6% of total) Other 86% ? ?
15. @dataknut: Modeling Water Demand in Droughts R NetLogo Model v1 flow diagram 15
16. @dataknut: Modeling Water Demand in Droughts Contents  The problem  Model Framework  Concepts & Implementation  Preliminary Results  Next Steps 16
17. @dataknut: Modeling Water Demand in Droughts Model v1: Retrospective 17 Census 2011 N households Househol d size Age distribution Work status Synthetic survey
18. @dataknut: Modeling Water Demand in Droughts Retrospective: With(out) drought 18 Without drought With drought response Drought phase Interventions Normal Water efficiency Developing Water efficiency Drought Temporary use bans & efficiency measures Severe Temporary use bans & efficiency measures Recovery Water efficiency
19. @dataknut: Modeling Water Demand in Droughts Retrospective: Sensitivity 19 With drought response For every 0.25% increase in “dual flush” uptake, average incremental saving in water was c.55 million litres For every 0.25% increase in “Eco shower” uptake, average incremental saving in water was c.9 million litres For every 10% increase in “Ban compliance”, average incremental saving in water was c. 0.65 million litres
20. @dataknut: Modeling Water Demand in Droughts Model v1: Prospective 20 Census 2011 N households Househol d size Age distribution Work status Synthetic survey Year Jan-Mar Apr-Jun Jul-Aug Sep-Oct 2016 2017 2018 2019 2020 2021 Drought phase Interventions Normal Water efficiency Developing Water efficiency Drought Temporary use bans & efficiency measures Severe Temporary use bans & efficiency measures Recovery Water efficiency
21. @dataknut: Modeling Water Demand in Droughts Prospective: With(out) drought 21 Year Jan-Mar Apr-Jun Jul-Aug Sep-Oct 2016 2017 2018 2019 2020 2021 Without drought With drought response Drought phase Interventions Normal Water efficiency Developing Water efficiency Drought Temporary use bans & efficiency measures Severe Temporary use bans & efficiency measures Recovery Water efficiency
22. @dataknut: Modeling Water Demand in Droughts Prospective: With(out) drought 22 Year Normal consumptio n Adjusted consumptio n % reduction Normal consumption Adjusted consumption % reduction 2016 105.44 104.57 0.82% 105.44 104.42 0.96% 2017 105.07 102.70 2.25% 105.07 102.53 2.42% 2018 105.96 102.02 3.72% 105.96 99.60 6.00% 2019 105.57 100 5.28% 105.57 97.60 7.55% 2020 105.48 98.33 6.78% 105.48 97.31 7.74% 2021 105.54 96.87 8.22% 105.54 97.07 8.02% Sum 633.07 604.48 4.51% 633.067 598.55 5.45% Without drought With drought response Total demand reduction is 28.59 and 34.52 million litres respectively (2016-2021) Year Jan-Mar Apr-Jun Jul-Aug Sep-Oct 2016 2017 2018 2019 2020 2021 Without drought With drought response Drought phase Interventions Normal Water efficiency Developing Water efficiency Drought Temporary use bans & efficiency measures Severe Temporary use bans & efficiency measures Recovery Water efficiency The hosepipe ban ‘effect’
23. @dataknut: Modeling Water Demand in Droughts Prospective: Sensitivity 23 Year Jan-Mar Apr-Jun Jul-Aug Sep-Oct 2016 2017 2018 2019 2020 2021 With drought response Drought phase Interventions Normal Water efficiency Developing Water efficiency Drought Temporary use bans & efficiency measures Severe Temporary use bans & efficiency measures Recovery Water efficiency For every 0.5% increase in “dual flush” uptake, average incremental saving in water was c.13 million litres For every 0.5% increase in “Eco shower” uptake, average incremental saving in water was c.18 million litres For every 10% increase in “Ban compliance”, average incremental saving in water was c. 9 million litres
24. @dataknut: Modeling Water Demand in Droughts Prospective: Robustness 24 Mean 4.55% (4.32 to 4.74%), Std.dev 0.10 Mean 5.43% (5.22 to 5.66%), Std.dev 0.09 Year Jan-Mar Apr-Jun Jul-Aug Sep-Oct 2016 2017 2018 2019 2020 2021 Without drought With drought response Drought phase Interventions Normal Water efficiency Developing Water efficiency Drought Temporary use bans & efficiency measures Severe Temporary use bans & efficiency measures Recovery Water efficiency • Re-run model 100 times Varies over a narrow range
25. @dataknut: Modeling Water Demand in Droughts Contents  The problem  Model Framework  Concepts & Implementation  Preliminary Results  Next Steps 25
26. @dataknut: Modeling Water Demand in Droughts Next Steps 26  Adding: – Practices – Weather – Interactions  Linking to – drought forecasts 1994 2012
27. @dataknut: Modeling Water Demand in Droughts Thank you!  b.anderson@soton.ac.uk (@dataknut)  Come and work on the project! – “Research Fellow in Household Water Demand Modelling”  jobs.soton.ac.uk/Vacancy.aspx?ref=782916AT 27