A Behavioural Approach to Stochastic End Use Modelling

A Behavioural Approach to Stochastic End Use Modelling Mark Thyer, Tom Micevski, George Kuczera Matt Hardy, Hugh Duncan, Peter Coombes and Bill Pascoe

Outline • • Motivation Model Overview and Results Applications and Implications Future Research

Urban Water Use is Changing (Beatty and O'Brien, 2008) • Increase in efficient appliances/prices/restrictions • Increase in IUWM to reduce mains water demand • Requires a flexible approach that can adapt

Integrated Urban Water Management • Requires robust models of household end-water use at small temporal and spatial scales

Behavioural End-Use Stochastic Simulator (BESS) Features: • Simulate multiple individual houses (1 -1000’s) – Differences in appliance type and household size between houses • Indoor water use time series – Shower, WM, Toilet, Tap etc. (Duncan and Mitchell, 2008) - Sub-daily time steps • Outdoor water use time series – Probabilistic behavioural response to rain and temp. (Micevski et al, 2011) Benefits: • Flexible approach adapt to changes in future water use behaviour • Utilise new datasets as they become available • Model scenarios of predicted changes in future

BESS: Indoor Water Use - Stochastically simulates differences in house size and appliances between houses • • Different appliances have different water use patterns and volumes Parameters based on Yarra Valley Water smart metering study of 100 homes (Roberts, et al, 2007)

Model Evaluation Washing Machine For each indoor water use event type simulated matches observed daily totals

BESS: Outdoor Water Use • Probabilistically model daily outdoor water use – Simulates individual households at daily time step – Behavioural approach to capture response of outdoor water use occurrence and volume to weather – Based on concepts of Coombes et al (2001) Weather Drivers Daily Rainfall Will outdoor water be used today ? Avg. Behaviour Max. Temperature Monthly Avg. Outdoor Water Use Yes How much water will be used today? Vol (watering) P (Watering) No Daily time series of outdoor water use

Outdoor Water Use: Calibration Results • Hunter Water Data set – Outdoor water use for 130 homes over 10 years • Existing approaches – Underestimate observed variability (56%) – Over-parameterised => too many un-identifiable parameters • Enhanced Behavioural Approach – Underestimates variability by only 8% – Parsimonous => parameters well-identified Existing Models BESS

Insights on Drivers of Outdoor Water use • Drivers of P(Watering) – For 80% of houses is increases in response to the long dry periods (days with rainfall) – For 20% of houses increases response to long hot periods (degree days) • 30% of houses, delay watering after a significant rainfall/watering event • Parameters are site specific and vary with climate • More research/data is needed to understand variability in outdoor water use

Applications: Optimising subsidies for reducing domestic water consumption • Final year engineering honours project • Little basis for setting of current rebate levels • Is a $10 rebate on showerhead, better than $200 rebate on washing machine? • $100’s millions spent on rebate schemes in Australia in past decade Previous rebate programs to forecast uptake BESS + Urban Developer to estimate water savings Multi-objective optimisation to identify Pareto optimal rebate policy

Applications: Optimising subsidies for reducing domestic water consumption • Current estimates wide variation our ACT estimate ACT^ our WA estimate WA* 0 5 10 15 Water Savings (k. L/household/year) 20 25 • Water Savings from BESS + Urban Developer

Identifying Optimal Rebate Policies: Multi-objective optimisation Current schemes Improved Program Cost/Water Savings between of 40 -70%:

Applications: Undergraduate Teaching • Env Eng: Fourth Year Course on WSUD/IUWM • MUSIC and Urban Developer used as a tool to apply key concepts to real world project • Positive student feedback • Students commented learning and using MUSIC and UD was best aspect of course

BESS: Practical Implications • Impact of water use variability on mains water saving estimates from tanks – Water savings from tanks will be sensitive to variations in demand – BESS can model increased in outdoor water use variability at end of hot, dry periods (when tank is low) – Current approaches are likely to over-estimate water savings • Simultaneously model changes in appliances with rainwater tanks and grey-water re-use scenarios • Impact of changes in water use on peak demands – Water infrastructure is designed to service peak demands => driven by water use variability – As BESS captures variability, potential to evaluate changes in peak. – Further research is needed on drivers of peaks demands – Strategic approaches to reduce peak and defer infrastructure costs

Future Research • Understanding behavioural drivers of water use – – – Outdoor Behavioural Change Appliance Uptake Price Attitude/Demographics • Combining end-use monitoring and behavioural surveys • Proposed monitoring programs in Adelaide and Newcastle – 300 - 350 homes with hi-resolution smart meters

Summary • Behavioural End-use Stochastic Simulator – – Stochastically simulates end uses for individual houses Differences/Changes in Appliance type and household size Indoor water use events subdaily time steps Outdoor water use – probabilistic behavioural response to rainfall and temperature – Simulations capture observed statistics (variability) • Flexible approach – Adapted to changes in water use behaviour – Predict impact of changes – Utilise new datasets in the future - promote data collection • Integrated into Urban Developer – e. Water product for cluster scale urban water management solutions • Future Research – Behavioural drivers of water use variability