Thermal Energy Flow Balancing for Optimizing
Energy Performance in Public Swimming Pools with
Solar Thermal Micro Generation
C. de Torre, A. Macia, M.A. Garcia-Fuentes, C. C. de Torre, A. Macia, M.A. Garcia-Fuentes, C.
Valmaseda
CARTIF Foundation, Energy Division, Valladolid,
Spain
H.Simonis
University College Cork, Ireland
Purpose of study
• Look at one site in more detail
– Test software and algorithms
• Understand factors influencing project progress
– Process requirements for potential roll-out
– Investment cost/effort of project implementation– Investment cost/effort of project implementation
Demo Site
• Huerta del Rey Sports Arena, Valladolid, Spain
– Indoor swimming pool
– Sports arena
– Gymnasium
– Outdoor courts
– Office space
Use Case
• Heating of swimming pool
• Either by
– Gas fired boilers
– Solar thermal array on roof
• Current control• Current control
– Maintain pool temperature at all times
• Idea:
– Use occupancy and weather data
• Best use of solar thermal array
– Storage of solar thermal energy
• Storage tanks and pools themselves
High-level Schematic
Cost of running the sports centre: Gas
20,000.00
30,000.00
40,000.00
50,000.00
60,000.00Consumos Gas en M3 C. D. HUERTA DEL REY
Enero Febrero Marzo Abril Mayo Junio Julio Agosto Septiembre Octubre Noviembre Diciembre
Año 2005 53,758.00 40,266.00 39,286.00 33,719.00 17,303.00 8,816.00 7,434.00 6,910.00 11,282.00 15,279.00 37,754.00 40,804.00
Año 2006 53,363.00 45,229.00 37,948.00 29,077.00 17,112.00 7,814.00 2,743.00 4,148.00 4,802.00 14,067.00 20,964.00 27,946.00
Año 2007 50,991.00 36,585.00 34,926.00 26,163.00 17,764.00 12,061.00 5,059.00 3,495.00 5,839.00 13,902.00 34,874.00 27,119.00
Año 2008 34,219.51 20,369.00 0.00 0.00 4,319.00 12,539.25 307.75 3,087.00 5,618.00 17,732.22 23,519.78 28,832.00
Año 2009 43,544.00 34,353.00 23,777.00 19,736.00 9,106.00 890.00 116.00 0.00 37.00 1,641.00 14,543.00 22,106.00
Año 2010 27,696.00 22,870.00 17,432.00 14,283.00 9,047.00 3,853.00 581.00 47.00 220.00 7,795.00 20,296.00 23,233.00
Año 2011 21,263.00 21,163.00 20,501.00 10,347.00 6,526.00 5,258.00 2,523.00 1,680.00 3,613.00 4,573.00 12,007.00 23,824.00
Año 2012 29,629.00 22,637.00 24,505.00 13,445.00 9,286.00 5,045.00 4,516.00 3,477.00 2,275.00 11,579.00 24,081.00 19,083.00
Año 2013 31,803.00 23,360.00 18,137.00
0.00
10,000.00
20,000.00
Implementation
• Data collection
– Existing infrastructure for temperature measurements
– Integrated in existing BMS
– Sensors only maintained if essential for control
• New sensors for energy flows• New sensors for energy flows
– Flow meters/Energy meters
• Requirements for high-quality data collection
– Synchronized measurements
– How much redundancy can be provided?
Web-based Data Visualization
• Access to sensor data
– Temperature
– Pump status
– Valve status
– Flow rates
– Energy flows
Data Access from Solver (Solar Irradiance)
Affecting the BMS
• Optimization module suggests operating modes
• Does not replace existing control
• Change set-points or rules of controller
• Access control issues for BMS
– Who can authorize changes?– Who can authorize changes?
– How to avoid run-away behaviour
Energy Flow Model
• Energy Sources
– Solar thermal array
– Boilers
• Storage Components
– Tanks– Tanks
– Pool
• Sinks
– DHW Use
– Pool Hall Environment
• Flows/Heat Transfer
Component-Based Model
• Model is described as combination of
– Components
– Flows connecting components
• Component library
– Extension by inheritance and composition– Extension by inheritance and composition
• Mathematical model is generated automatically
– Code for mixed integer programming (CPLEX)
– Documentation of model
Resulting MIP Model
• Time
– Discrete time periods (15 min)
• Variables
– Flows between components
– Vectors of values– Vectors of values
• Constraints
– Energy conservation
– Heat transfer
• Objective
– Sum of objectives of components
Levels of Abstraction
• Solver module abstracts most details
– Not automated
• Correspondence to detailed control/sensor model
required
– Partially automated
• Translate solver decisions translate into control changes
– Human understanding of design
Forecast Data Required
• Solar irradiance
– Really: Forecast of thermal array output
• Temperature
– Heat-loss of swimming pool
• Occupancy• Occupancy
• Domestic hot water (DHW)
• (Electricity Price Prediction)
• (Electricity Demand Prediction)
Solar Irradiance Forecast for Valladolid
Temperature Forecast Valladolid
Component Reuse
• Key to economic success
• All demonstrators share optimization tools
– Models generated from declarative description
• Reuse of forecasting modules
• Possible through middleware integration• Possible through middleware integration
Occupancy Sports Arena (Handball)
Occupancy Sports Arena
• Use of arena known well in advance
• Two main types of activities
– Games (teams, spectators)
– Training (teams)
• DHW use follows team schedule• DHW use follows team schedule
Occupancy Swimming Pool
Occupancy Swimming Pool (II)
• No sensor data for direct measurement of occupancy
– Turnstile
– Card reader
• Clear calendar for group activities (courses/clubs)
– Majority of users– Majority of users
• Individual users difficult to predict
– No ground truth
– Small numbers make forecasting hard
• Individual decisions have large effect
– Impact on total user numbers not too significant
Contracting Schedules
• The optimization is based on imprecise forecast data
• How accurate is the schedule, compared to hypothetical
solution with perfect knowledge?
• Evaluation Strategy:
Surprising Results
• UCC CHP plant (Impact of price and demand forecast)
Similar Results
• Manufacturing Scheduling
– Ifrim, O’Sullivan & Simonis, 2012
• Home Energy Management System, EV Charging
– Grimes, Simonis, Pratt & Sheridan, 2012
• Good schedules can be found, even if forecasts are
imprecise
• High quality even when compared to hypothetical
schedule based on ex-post data
Conclusions
• Work in progress
– Installation of sensors
– Connection of middleware
– Access through middleware
– Evaluation of predictors
– Integration with solver
• Lead-time requirement for forecasting
– Training data, historical data needs
• Change process issues
– Investment sign-off
– Changes to existing systems
The research presented is supported by a fund
from Seventh Framework Program – ICT
“Control & Automation Management of
Buildings & Public Spaces in the 21st Century”
or CAMPUS 21
(Project-Nr: 285729).(Project-Nr: 285729).