Jimmy A regression approach for assessment of building energy performance A Licentiate Thesis in Energy Technology
Motivation Picture: Tomas Mejtoft, 2014.
Assessment of building performance First step to implement measures. Enables a learning process. Can put public pressure to reduce the energy use. Picture: The Swedish National Board of Housing, Building and Planning, Boverket.se, 2014. -> Need for reliable and easy to use evaluation methods.
Ålidhem - a sustainable district Existing buildings, built during the years 1970/71. Energy efficiency goal to decrease the total energy use by 30-40% Goal for new buildings 65 kwh/m 2 /yr. - How can we estimate the savings with the implemented measures? - How can do we assess that the developer has delivered the promised performance?.
What is the problem?
Investigation Problems: The commonly used indicator (kwh/m 2 /yr) is sensitive to other factors than the buildings performance. Difficult to estimate the actual savings from measures Often difficult to perform accurate simulations Objectives: Practical methods to be used for: - Evaluation - Calibration of simulation models, (Is it worth the additional effort?)
Data and case studied buildings Building no 1 Building no 2 Buildings A-E (Pilot-building) (Reference-building) (New buildings) District heating - Radiators - Ventilation system (b.1&2) - Domestic hotwater Electricity - Domestic - Non-domestic Temperatures - Indoor - Exterior - Technical systems (b.1&2) jimmy.westerberg@umu.se Jimmy 2014-09-07
Initial status, Building no. 1 & 2 1. 10 apartments, heated floor area 875 m 2. 2. Two glazed windows 3. Attic floor, 25 cm insulating sawdust. 4. Gable and long side exterior walls, 10 cm mineral wool 5. Mechanical ventilation system, without heat recovery 6. Similar energy characteristic
Taken measures, Building no. 1 1. Additional wall insulation (interior) and new vapor barrier. 2. New roof and improved insulation 3. New windows (triple glazed) 4. Heat recovery on the exhaust air 5. (Photovoltaic panels) 6. (Individual metering of domestic hot water usage)
Energy flows P CV = P entering -P leaving = P airh + P rad + P p + αp elec + P sun - P tr - P vent - P G P rad + P p + αp elec - Q s (T i -T s )ρc p = (AU t +Q L ρc p )(T i -T o ) + P G Controlled ventilation losses Supplied heat to the air handling unit Stored and released thermal energy Outdoor air temp Indoor air temp. Occupancy Airflow convergence to indoor temp. Elec. appliances Radiators Transmission losses and air leakage Solar irradiation Domestic hot water Heat loss to ground
Evaluation of taken measures 2010/2011 P rad +P p +αp elec - Q s (T i -T s )ρc p = (AU t +Q L ρc p )(T i -T o ) + P G 2011/2012 Robust and good agreement with theory Comparison can be made with previous design stage calculations
Usage as a calibration tool Resulted in a simulation accuracy of calibrated level.
But did we get better predictions with the Calibrated model? Simulated individual energy conservation measures (ECMs): 1. Additional attic room. 2. Heat recovery on the exhaust air 3. Improved roof insulation 4. Window upgrade 5. Adjustment of domestic hot water circulation losses 6. Additional interior insulation of exterior walls Sets: #1). ECM 1 and 2 #2). ECM 3, 4 and 6 #3). All individual ECMs implemented
Used two different models for energy saving predictions 1. A basic model based on inputs from current available standards and as-built drawings. (19.1% deviation) IDA-ICE(ver.4.5) model 2. A calibrated model based on input from regression analysis and actual operating data. (3.7% deviation)
Results Basic & calibrated model predictions Individual ECMs 1. New attic room (1.8<%) 2. Heat recovery on exhaust air (ca.14.2%) 3. New roof insulation (1.8<%) 4. Window upgrade (1.8<%) 5. Adj. of dhwc losses (ca.1.6%) 6. Additional interior insulation of exterior walls (1.8<%) Sets: 1. ECM 1 and 2 (ca.12.2%) 2. ECM 3,4 and 6 (2.6%) 3. All individual measures implemented (ca.12.4%)
Results Comparison with actual outcome in a neighboring building Postretrofit demand space heating (elec. and DH) Heat recovery of the exhaust air 2014-09-07
Performance verification
Different normalization approaches 2014-09-07
Ventilation and transmission losses Regression approach Performance control Sidfot Datum 19
Ventilation and transmission losses Regression approach Slopes The largest buildings Small buildings have to be insulated more
Overall conclusions Regression method exhibits high robustness and good agreement with theory. Extracted parameters can be used for performance verification as well as for calibration. Calibration is primarily important for assessing post-retrofit energy demand and savings due to heat recovery of the exhaust air.
Forthcoming research: 1. Include data from additional buildings to further establish the consistency and accuracy 2. To investigate the geographical dependence of the regression method (solar conditions) 3. Further analyze of using the regression method for BES calibration (comparison with other methods)
Thank you for your attention! The Industrial Doctoral School at Umeå University and AB Bostaden are gratefully acknowledged for their financial support.