Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions Smart control of energy systems with PVT Peder Bacher (pbac@dtu.dk), Linde Fr¨ olke, Rune G. Junker og Henrik Madsen DTU Compute, Dynamical Systems WORKSHOP ON PV-THERMAL SYSTEMS October 9, 2019
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions C HALLENGES WITH R ENEWABLE DRIVEN SYSTEM Make the operation “optimal” using available “flexibility”: Adapt to variation in generation and demand in general (e.g. wind) Help the grid (peaks and congestion)
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions C HALLENGES WITH R ENEWABLE DRIVEN SYSTEM Make the operation “optimal” using available “flexibility”: Adapt to variation in generation and demand in general (e.g. wind) Help the grid (peaks and congestion) Solutions Shift demand: Dish washer, etc. (usually low demand) Store energy: Thermal (hot water tank, DH grids and in building elements) Chemically (batteries) Problem: When to charge battery? when to run heat pump?
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions S EVERAL DEMO CASES OF M ODEL P REDICTIVE C ONTROL (MPC) BIPVT-E project in Stenløse, DK: EMPC for optimizing battery charging Grundfos test house: EMPC for optimizing heat pump heating with hot water tank Swimming pool heating: EMPC for optimizing the heating of swimming pools
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions S EVERAL DEMO CASES OF M ODEL P REDICTIVE C ONTROL (MPC) BIPVT-E project in Stenløse, DK: EMPC for optimizing battery charging Grundfos test house: EMPC for optimizing heat pump heating with hot water tank Swimming pool heating: EMPC for optimizing the heating of swimming pools HVAC in buildings, waste-water treatment, water management, ...
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions E CONOMIC M ODEL P REDICTIVE C ONTROL (EMPC) Need to setup: A model of the system (keep as simple as needed) An objective function with constraints (usually the energy buy and selling costs, but could be any “penalty”) Forecasts of input variables to the model and objective function At every time step Calculate forecasts: 1 Weather, generation, demand and price forecasts Optimize the objective function: 2 Find the sequence ahead of control variables (CV) which optimize the objective function, while keeping constraints Implement optimal CV until next step 3
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions BIPVT-E SYSTEM WITH DTU B YG , COWI, R ACELL
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions PV WITH BATTERY EMPC objective function and model with constraints: N k − τλ k g + λ k g − ∑ � � ( cost sell − cost buy ) Minimize k k = 1 k − b + k − g + d k = p k + b − k + g − ( demand ) subject to 1 ≤ k ≤ N k , b k = b k − 1 + c B b + k − b − ( Simple battery model ) k , ( min. & max. of bat. ) 0 ≤ b k ≤ b max , 0 ≤ b + k ≤ b + ( max. charge rate ) max , 0 ≤ b − k ≤ b − max , ( max. discharge rate ) k , g + g − ( buy & sell positive ) k ≥ 0.
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions BIPVT-E EMPC FORECASTS Abbriviations: FORECAST Abbriviations: INPUTS OUTPUT - a: ambient MODEL - NWP: weather forecast - i: indoor - T: temperature - d: diffuse or direction - P: power - b: beam - G: global radiation - el: electrical - I: radiation - oc: of charge - NWP (Ta,G) - W: wind Electrical load of - Pel_hp Pel_hp_hat - hp: heat pump - S: State heat pump - Pel_pv - pv: photovoltaic - apl: appliances - bat: battery INPUTS OPTIMIZATION OUTPUT - NWP (Ta,G) Power output of Pel_pv_hat - Pel_pv PV Must have: - Pel_hp_hat - Pel_pv_hat - Pel_apl_hat MPC Pcharge_bat - Cel_buy_hat Must have: - Cel_sell_hat - Pel_apl - Power load of Pel_apl_hat - Soc_bat - NWP (Ta,G) appliances Electricity prices Cel_buy_hat ENFOR module (buy and sell) Cel_sell_hat
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions PV WITH BATTERY ( SIMULATIONS )
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions PV WITH BATTERY
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions IPower og CITIES projektet (Jacopo Parvizi): Grundfos Test Facility, the heating system is composed of the following elements: 600 l Stratified Hot Water Tank 7.2 m 2 Solar Thermal Collector Heat Pump - 7 kW with Variable Speed Compressor Domestic Hot Water Grundfos Fresh Water Module District Heating Backup Local Weather Station Kamstrup Multical Heat Meters
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions Necessary forecasts as input to Economic Model Predictive Control (EMPC): Heat demand in the building Solar heating Electricity price
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions Necessary forecasts as input to Economic Model Predictive Control (EMPC): Heat demand in the building Solar heating Electricity price
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions Necessary forecasts as input to Economic Model Predictive Control (EMPC): Heat demand in the building Solar heating Electricity price 9 Water Tank Temp [ C ] 55 8 House Load [ kW ] 7 50 6 5 45 4 3 40 2 y 1 dist 35 0 02−Jan 03−Jan 04−Jan 05−Jan 06−Jan 07−Jan 08−Jan 09−Jan Elspot price [Eur / MWh] Heat Pump Load [ kW ] 5 50 4 40 3 30 2 20 1 10 u p 0 0 02−Jan 03−Jan 04−Jan 05−Jan 06−Jan 07−Jan 08−Jan 09−Jan
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions Savings achieved with EMPC (two scenarios: varying price and flat price) 2013 2014 EMPC var. tariff 11% 16% EMPC flat tariff 3% 8% 9 Water Tank Temp [ C ] 55 8 House Load [ kW ] 7 50 6 5 45 4 3 40 2 y 1 dist 35 0 02−Jan 03−Jan 04−Jan 05−Jan 06−Jan 07−Jan 08−Jan 09−Jan Elspot price [Eur / MWh] Heat Pump Load [ kW ] 5 50 4 40 3 30 2 20 1 10 u p 0 0 02−Jan 03−Jan 04−Jan 05−Jan 06−Jan 07−Jan 08−Jan 09−Jan
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions P OOL SUMMERHOUSES FROM THE PROJECT S MART N ET Control of heating of swimming pools in summer houses, using the pool as the heat storage:
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions P OOLSOMMERHUSE FRA PROJEKTET S MART N ET The EMPC buy electricity when the price is low:
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions C ONCLUSIONS USING EMPC Conclusions EMPC become attractive when economic incentives are strong enough (especially tax schemes have influence) It doesn’t need to be a price signal which drives, any “penalty” can be used We need robust statistical models and good forecasts To be done: hotwater tank, building and battery in one EMPC
Why MPC EMPC PV with battery Hot water tank Swimming pool Conclusions C ONCLUSIONS USING EMPC Conclusions EMPC become attractive when economic incentives are strong enough (especially tax schemes have influence) It doesn’t need to be a price signal which drives, any “penalty” can be used We need robust statistical models and good forecasts To be done: hotwater tank, building and battery in one EMPC Forecasting software (R package) Grey-box modelling (R package (ctsmr: ctsm.info ) Many optimization implementations are available (easy with linear models: linear programming)
Recommend
More recommend