1 ThermalThrift: Cost Effective Thermal Energy Storage for Load Shifting with Water HeatersErin Griffiths

2 Peak Energy Demand - Peak is people turning on stuff

3 Peak Energy Demand Peak Hours and Peak Prices reduce consumeRequires use of peaking plants To prevent the use of these inefficient and expensive plants

4 Energy Storage (ES) use store

5 Thermal Energy Storage (TES)use 100 million homes: 7-19 GWh storage ~12GWh combined thermal, battery, compressed air, and flywheel currently worldwide store

6 hot water usage (liters)

7 tank temp (⁰C) - About x degrees hot water usage (liters)

8 tank temp (⁰C) hot water usage (liters)

9 Standby loss Normal Heating tank temp (⁰C) hot water usage (liters)

10 Peak Hours tank temp (⁰C) hot water usage (liters)

11 Peak Hours tank temp (⁰C) hot water usage (liters)

12 Peak Hours Higher Standby Loss NOT COST EFFECTIVE! tank temp (⁰C) hot water usage (liters)

13 Peak Hours tank temp (⁰C) hot water usage (liters)

14 Peak Hours High Cost Heating tank temp (⁰C) hot water usage (liters)

15 Peak Hours Wasted Energy due to Standby Loss tank temp (⁰C) hot water usage (liters)

16 Minimize: high cost heating + standby loss tank temp (⁰C)Peak Hours Minimize: high cost heating + standby loss tank temp (⁰C) 8 minutes hot water usage (liters)

17 Outline Approach In-situ Evaluation + Results Historical DataTank Modeling: Predicting Usage Cost Prediction and Heating Decision In-situ Evaluation + Results

18 Historical Hot Water Usage For each iteration: Water Tank Model Historical Hot Water Usage For each iteration: Decide whether or not to heat for TES Apply decision Peak Hours and Pricing

19 Historical Data Collected as flow data from the hot water tankNo user input required Collected from the whole home

20 Tank Model: Predicting UsageStandard Water Heater Standby loss Heating Usage Predicting Usage + = Modeled Hot Usage Modeled Cold Usage Historical Hot Usage

21 Cost Prediction Historical Day 1 Tank temp Peak Hours Current Time𝝉 Time Peak Hours Historical Day 1 Current Time

22 Cost Prediction Store thermal energy now Historical Day 1 Tank temp𝝉 Time Peak Hours Historical Day 1 Store thermal energy now Current Time

23 Cost Prediction Store thermal energy nowTank temp 𝝉 Time Peak Hours Historical Day 1 Store thermal energy now Store thermal energy next iteration Current Time

24 Cost Prediction Store thermal energy nowTank temp 𝝉 Time Peak Hours Historical Day 1 Store thermal energy now Store thermal energy next iteration Never store thermal energy Current Time

25 Heating Control DecisionOutput: matrix of costs for all historical days of data Is average predicted_cost(store now) < predicted_cost(store later) ? Is average predicted_cost(store now) < predicted_cost(never store) ?

26 Outline Usage Based Approach In-situ Evaluation + ResultsHistorical Data Tank Modeling Heating Control Decision In-situ Evaluation + Results

27 Seametric Flow meter and Hobo datalogger – 2 week collection time

28 Evaluation TRNSYS simulation software12 real world peak pricing schemes Baselines: Conventional water heater with setpoint of 49⁰C Usage Agnostic TES water heater always storing 93⁰C before peak hours ThermalThrift Perfect Prediction where “historical” data is oracle data

29 Usage Agnostic Pricing Scheme Peak Energy (%) Cost (%)Total Energy (%) P1 P2 P3 P4 P5 P6 P7 P8 9.8885 P9 P10 P11 P12 Average

30 Usage Agnostic Pricing Scheme Peak Energy (%) Cost (%)Total Energy (%) P1 P2 P3 P4 P5 P6 P7 P8 9.8885 P9 P10 P11 P12 Average

31 Thermal Thrift Pricing Scheme Peak Energy (%) Cost (%)Total Energy (%) P1 P2 P3 P4 -0.327 0.8835 P5 0.6214 P6 P7 P8 -1.852 1.3105 P9 P10 4.0731 P11 0.2812 P12 5.9649 Average 8.8559

32 Thermal Thrift Pricing Scheme Peak Energy (%) Cost (%)Total Energy (%) P1 P2 P3 P4 -0.327 0.8835 P5 0.6214 P6 P7 P8 -1.852 1.3105 P9 P10 4.0731 P11 0.2812 P12 5.9649 Average 8.8559

33 Thermal Thrift Pricing Scheme Peak Energy (%) Cost (%)Total Energy (%) P1 P2 P3 P4 -0.327 0.8835 P5 0.6214 P6 P7 P8 -1.852 1.3105 P9 P10 4.0731 P11 0.2812 P12 5.9649 Average 8.8559

34 Savings Comparison Approach Peak Energy (%) Cost (%) Total Energy (%)Usage Agnostic ThermalThrift 8.8559 ThermalThirft + Perfect Prediction

35 Savings Comparison Approach Peak Energy (%) Cost (%) Total Energy (%)Usage Agnostic ThermalThrift 8.8559 ThermalThirft + Perfect Prediction