Ashrae Guideline 14

2018 Building Performance Analysis Conference and
SimBuild co-organized by ASHRAE and IBPSA-USA

Chicago, IL
September 26-28, 2018
CALIBRATION OF A BUILDING ENERGY PERFORMANCE SIMULATION MODEL

VIA MONITORING DATA
Basak Gucyeter
Department of Architecture, Faculty of Architecture and Engineering, Eskisehir
Osmangazi University, Eskisehir, TURKEY
consumption on building level through energy

ABSTRACT conservation measures (ECMs) for existing building
Energy performance gap is considered as one of the most envelopes such as insulation, better-performing glazing,
significant issues associated with the assessment of solar shading, etc. and integration of renewable or clean
energy consumption in the built environment. In order to energy technologies within the building services (Hens
narrow this gap, simulation approach for energy et al. 2010; Diakaki et al. 2010). In such a framework,
performance assessment are requires comprehensive evaluating the effects of ECMs became crucial in
calibration procedures. Calibrated energy performance achieving decreased levels of energy consumption in
models facilitate a baseline representation of existing buildings. Simulation software, machine learning,
building performance patterns, thus, further accuracy in compliance systems (De Wilde 2014) gained importance
diagnosis, operation and energy conservation measures due to their capability to replicate real world phenomena,
(ECMs) become possible through the use of calibrated and especially simulation tools were considered reliable
models. The present study presents an iterative approach when results were within error margins that were set via
for calibrating a building energy model using full year standards such as International Measurement and
monitoring data. The methodology focuses on disclosing Verification Protocol (IPMVP 2001), ASHRAE
the steps in calibrating the simulation model and the Guideline 14 (2002) and the Federal Energy
relative sensitivities of the assumed and monitored Management Program Monitoring and Verification
parameters used in calibration. The magnitude of the Guide (FEMP 2008). Simulation modeling, the widely-
alteration in different levels of calibrated simulation anticipated energy performance assessment
models are evaluated with Mean Bias Error (MBE) and methodology, was distinguished with its capability to
Root Mean Square Error (RMSE) and the model replicate the thermal behavior and energy performance
accuracy is controlled through benchmarks defined by of a building (Crawley et al. 2008). Validation and
ASHRAE Guideline 14, International Performance testing became of utmost importance to accurately assess
Measurement and Verification Protocol (IPMVP) and the realistic energy performance of the buildings.
Federal Energy Management Program (FEMP). However, starting from the mid 1990s, strong indications
INTRODUCTION of a “performance gap” were evident between the
predicted and actual energy consumption, and the
Given the fact that 40% of the world energy consumption exhibited discrepancies, in some cases, were more than
originates from energy use in buildings for space 100% (De Wilde 2014; Bordass et al. 2004; Menezes et
conditioning, ventilation, hot water, lighting and al. 2012). Consequently, building energy performance
appliances (DoE 2008), providing environmentally gap turned out to be one of the widely-discussed issues
sensitive and efficient measures became a priority for associated with energy use in the built environment.
researchers and professionals involved in the production Despite the national/international standards that
of the built environment. Assessing building energy recommend accurate assessment of building energy
performance, decreasing fossil fuel resource performance, the discrepancy between the design
consumption and endorsing the utilization of predictions and as-built energy performance of buildings
technologies that support integration of non-renewable was still significant due to an array of reasons related to
energy sources became significant emphases (Fumo factors affecting energy consumption (such as occupant
2014; Ahmad and Culp 2006). Furthermore, regulatory behavior, simulation model simplifications, poor
approaches encouraged the decrease in energy
© 2018 ASHRAE (www.ashrae.org) and IBPSA-USA (www.ibpsa.us). 542

For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without
ASHRAE or IBPSA-USA's prior written permission.
assumptions etc.). In addition to efforts that facilitate benchmarks defined in ASHRAE Guideline 14 (2002),
post-occupancy evaluation for buildings to bridge the IPMVP (2001) and FEMP (2008). The present study,
performance gap, simulation modeling was as well therefore, both underscores the significance of
designated as an assessment methodology that requires a comprehensive calibration procedures in building energy
certain degree of confidence. Hence, to holistically performance simulation and interprets the research
address whole-building energy performance assessment outcomes in terms of their impact on building
through the utilization of a simulation model, it became performance gap.
significant to implement a calibrated building energy
simulation approach. Although intended to function as a METHODOLOGY
design phase tool, building energy simulation (BES) Building Information and the Monitoring Process
models were developed into tools that allowed complex
The case building, located in the main campus of
calculation of the energy performance of existing
Eskisehir Osmangazi University, Eskisehir, Turkey,
buildings mainly to evaluate the effects of ECMs
predominantly accommodates office functions (Figure
(Royapoor and Roskilly 2015; Coakley et al. 2014). The
1). Further information on the building is presented in
forward approach in modeling and simulation briefly
Table 1. Situated on a flat and open lot, the spaces are
emphasizes the importance of acquiring (1) climate data
oriented towards a central corridor aligned to the
for the case building, (2) building design, (3)
north/south. The building has a reinforced concrete
geographical data (location, orientation, obstructions
structure with filled in brick walls and no insulation
etc.), (3) construction data, (4) building installation
despite the cold/snowy climate in winters. Measured
characteristics, (5) building operations, occupancy and
thermal characteristics of opaque building envelope
schedules (Harish and Kumar 2016), yet inadequacy in
components are presented in Table 2. Transparent
abovementioned data could result in a discrepancy
envelope parts of the building consist of aluminum and
between the simulation results and actual thermal
PVC frames without thermal break and double-pane
behavior of the building. Ahmad and Culp (2006)
clear glass with U-values of 3.0 W/m2K and 3.2 W/m2K,
established that uncalibrated simulation models produce
respectively (TS2164, 2000). The building is
discrepancies between the monitored and calculated
conditioned only with an old non-condensing boiler
consumption levels in the range of ±30% and suggested
using natural gas as the primary energy source. Indoor
that the discrepancies even rise to a range of ±90% for
temperatures for office and classrooms were designed as
end uses such as chilled water, hot water, and electricity
23 ºC and 20 ºC for circulation spaces during the heating
consumption. Therefore, it is possible to assert that
season. Approximate discrepancies of ±1 to 3◦C in
employment of uncalibrated simulation models is an
indoor temperatures were observed during the
important factor in the emergence of building
monitoring of the building. The building is used for
performance gap and simulation models should be
administrative and teaching purposes between 8AM and
calibrated in order to decrease the effect of modeling
5PM on workdays.
errors, insufficient inputs, imprecise assumptions, and
uncertainty related to design and operation on the
simulation outcomes.
Calibrating building energy models based on monitoring
data for existing buildings and from feedback data from
various field studies for new designs could facilitate
performance predictions with high accuracy (Raftery et
al. 2011; Zhao and Magoulès 2012). In this framework,
this study focuses on disclosing six distinct steps in
calibrating the simulation model of an existing building
through employment of monitored indoor temperatures,
calculated/assumed infiltration rates, monitored
occupant presence within the simulation model with an
iterative approach. The outcomes are expected to
provide sensitivities of the assumed and monitored
parameters in calibrating building energy simulation
models. The magnitude of the alteration in presented
calibration steps are evaluated through Mean Bias Error
(MBE) and Root Mean Square Error (RMSE) values and Figure 1 (a) Typical floor plan of the case building (b)
the model accuracy is inspected with respect to the South façade of the case building

Table 1 Building information Modeling and Calibration
Building Information
The calibration approach employed in the present study
Floor Area (m2) 3402
intends to adjust simulation parameters iteratively, until
Floor Height (m) 3.50 to 4.50
certain degrees of accuracy between the monitored and
Volume (m3) 13261
the simulated hourly indoor temperatures and the
Façade Surface Area (m2) 2678
monitored and simulated monthly heating consumption
Roof Area (m2) 561
patterns were achieved. The model accuracy is
Glazing Area (m2) 666
controlled through benchmarks provided by the IPMVP
Glazing Ratio (%) 25
(2001), ASHRAE Guideline 14 (2002) and FEMP
Compactness (Atot/Vtot) 0.23
(2008). EDSL Tas was used for energy performance
modeling of the case building. A multi-zone simulation
Table 2 Building Elements and U-values model was developed with respect to the spatial
Building Elements U-Value (W/m2K) divisions of the building, since the calibration of the
Reinforced Concrete Walls 2,633 model would be conducted with hourly comparisons of
Basement Retaining Walls 0.352 monitored and simulated data for 37 zones. Figure 2
Exterior Walls 1,852 presents the steps in the iterative process in calibrating
Concrete Floor on Ground 0,866 the energy simulation model. R01, the initial model,
Flat Roof 2,740 was created with basic information that was collected
Sloped Roof 3,068 through building audit including as built information,
Roof Slab 3,480 measured envelope characteristics through
Interior Walls 1,726 thermocouple U-value measurements, monitored full
Interior Floors 2,566 year micro-climatic data, calendar and schedules for
Indoor temperature and humidity, gas and electricity occupancy, heating season design temperatures and
consumption and weather data were measured in the heating installation properties.
building during 2016. Electricity consumption was
monitored with a power analyzer data logger on 10 min
intervals. Hourly gas consumption for 2016 was
retrieved from the remote monitoring system of the gas
provider company. Outdoor temperature, outdoor
humidity, global horizontal solar radiation, wind speed,
and wind direction were monitored with 10 min interval
with a weather station. Cloudiness (0-1) data was
retrieved from the macro-climatic weather station in
Eskisehir, Turkey. Heating installation efficiency and U-
value measurements for the opaque building envelope
were completed during the monitoring process (Table 3).
Table 3 Monitored building energy performance

parameters
Monitored Building Energy Measurement
Performance Parameters Interval
Indoor Temperature (°C) 10 min.
Indoor Relative Humidity (%) 10 min.
Occupant Presence (%) 1 min.
U-value (W/m2K) Multiple Figure 2 The iterative calibration process (Partly
Gas consumption (m3/h) 1 h. adopted from Raftery, Keane and Costa 2009)
Electricity consumption (kWh) 10 min. Once the simulation outcomes were retrieved, hourly
Outdoor temperature (°C) indoor temperature results of the R01 simulation model
Outdoor relative humidity (%) were compared to the hourly monitoring data, in
Global horizontal solar radiation (W/m2) 10 min. addition to the comparison of monthly simulated and
Wind speed (m/s) monitored heating consumption data. The initial model
Wind direction (◦)
was not expected to yield an acceptable accuracy;
Cloudiness (0–1) 3 h.
however, the results were extremely discrepant from the
Boiler Performance (CO2) Once

actual monitored indoor temperature and monthly (RMSE) and the mean bias error (MBE). Equations (1)
heating consumption data. Therefore, the calibration and (2) present the formulas employed for RMSE and
process was initiated to match the simulation outcomes MBE, where, n is the number of observations, Tm,av. is
with the monitored data as accurately as possible. As the average of the monitored data for n observations, Ts
presented in Figure 2, the process in running iterative is the simulated data for n observations, and Tm is the
simulation models, followed the procedure of obtaining monitored data for n observations.
run results, comparing these results with monitored *++ * ;
data, identifying the discrepancy and the possible 𝑅𝑀𝑆𝐸(%) = ) 2 × [ × ∑(𝑇8 − 𝑇: ) ]+.= (1)
,-,/0. 5
source of discrepancy, adjusting relevant parameters, 𝑀𝐵𝐸(%) = (,
*++
)×
∑(,?@,-)
(2)
and running the next iterative model. This process was -,/0. 5
repeated until the model calibration was completed on The results obtained with the linear correlation (R)
the 15th run. analysis and the RMSE and MBE analyses were used to
Figure 3 presents the integrations/adjustments in the evaluate the accuracy of the iterative simulation runs
abovementioned process of iterative runs. In the present with respect to the benchmark values provided by the
study, simulation outcomes of the 6 of 15 runs are IPMVP (2001), ASHRAE Guideline 14 (2002) and
compared to monitored hourly indoor temperatures and FEMP (2008). In Results and Discussion section, the
monthly heating energy consumption data and the iterative model characteristics, the nature of
results of these comparisons are presented in detail to integrated/assumed parameters and the magnitude of
disclose the effect of the integration of monitored the alterations in the outcomes due to the calibration
indoor temperatures, occupant presence, and adjustment attempts are discussed in detail.
of calculated/assumed infiltration rates within the
energy performance simulation of an existing building.
RESULTS AND DISCUSSION
R01, created as the initial model with information
collected through building audit and partly during the
monitoring process, was setup with heating season
design temperatures with the intention to demonstrate
the effect of monitored indoor temperatures on the
simulation outcomes. Simulation outcomes for R01 were
correlated with the hourly monitoring data (37 spaces x
8760 hours=324120 hours) and the comparison of
monthly simulated and monitored heating consumption
data was carried out using RMSE and MBE analyses.
The indoor temperature errors between the monitored
and the simulated hourly data (Tm – Ts) were found to be
normally distributed as presented in Figure 5. However,
heating energy consumption comparison yielded a
discrepancy of -58.55% (MBE) (Table 5), which is
highly unacceptable when compared to the hourly
calibration benchmarks provided by ASHRAE
Guideline 14, IPMVP and FEMP (Table 4).
Monitored and Simulated Temperatures for G03
25
20
Figure 3 Iterative runs and calibration attempts 15

10
Temperature (◦C)
The comparison between monitored and simulated data 5

was carried out with two analyses. The first approach is 0
00:00…
00:00…
00:00…
00:00…
00:00…
00:00…
00:00…
00:00…
00:00…
00:00…
00:00…
00:00…
00:00…
00:00…
00:00…
a linear correlation (R) analysis based on hour-to-hour

12:00
12:00
12:00
12:00
12:00
12:00
12:00
12:00
12:00
12:00
12:00
12:00
12:00
12:00
12:00
-5
correspondence of simulated and monitored indoor -10
temperatures for a full year, for each of the 37 zones Exterior Temperature (◦C)
Hours
Monitored Temperature (◦C)
monitored in 2016. Second approach is an error R01 Simulated Temperature (◦C)
analysis that intends to check the deviation of simulated Figure 4 Monitored and simulated temperatures for the
hourly temperatures and monthly consumption patterns unoccupied and unconditioned space G03
from the monitored data with root mean square error (February 1st to 16th)

Simulation model R01 was checked for modeling errors in the simulation model resulted in a 38% improvement
and building envelope characteristics via indoor in annual heating energy consumption in comparison to
temperature comparison of unoccupied and the initial run. However, R07 could not predict the
unconditioned spaces, since such data could easily monthly heating energy consumption with an accuracy
reveal the errors in thermo-physical characteristics of that would meet the calibration benchmarks, rather the
the building envelope in simulation. Figure 4 prediction was inaccurate with a MBE of -42.66%. In
demonstrates the exterior, monitored, and simulated order to accept a simulation model as calibrated, both
temperature fluctuations for ground floor entrance indoor temperatures and consumption patterns should be
space G03, which is an unoccupied and unconditioned within the acceptable calibration values presented in
space. This evaluation indicated that building envelope Table 4. Hence, the calibration process was continued
with the integration of infiltration rates in the next
characteristics were modeled with a certain level of
simulation run, R08. Since blowerdoor tests could not be
accuracy since the average errors for eight unoccupied
completed during the monitoring period, the infiltration
and unconditioned spaces between the monitored and
rates were calculated based on the effective leakage area
the simulated hourly data (Tm– Ts) were found to be and volume of each zone, and to the ATTMA standard
between 0.85 and 1.13◦C. Such discrepancy did not TSL2 (2010) benchmark for normal levels of building air
necessarily have to be the result of errors in the thermo- permeability 0.7 m3/h.m2 @50Pa.
physical characteristics of the building envelope
Table 5 Case building calibration results
integrated in simulation. Evaluation of the building
Heating
envelope characteristics would be more substantial
Energy
consequent to the calibration of other parameters that RUNS Indoor Temperature Calibration
Consumption
might be causing the model discrepancies. Identified Calibration
sources of discrepancies for R01 were interpreted as RMSE MBE Eav RMSE MBE
follows: (1) absence of realistic indoor temperature R
(%) (%) (◦C) (%) (%)
profiles for the heating season, (2) omitted infiltration R01 10.60 0.74 2.35 0.86 72.41 -58.55
rates and (3) assumed occupant presence instead of the R07 7.32 0.16 1.62 0.91 54.27 -42.66
actual presence of occupants. In this respect, first R08 8.18 1.39 1.81 0.85 27.94 -17.96
monitored indoor temperature data was used to R13 8.41 2.16 1.86 0.85 23.76 -13.77
calibrate the model. Indoor temperature profiles were R14 7.78 0.95 1.72 0.85 18.96 -5.05
integrated in the simulation model in seven runs and R15 7.87 1.06 1.64 0.88 18.49 -4.70
required changes to schedules, set point temperatures,
calendar days and the heating system operation Calculated infiltration rates for the zones ranged between
schedule. 0.3 and 1.8ach. Since the building was completed in
Table 4 Calibration benchmark values 1992 and underwent no major renovation, infiltration
Calibration Benchmarks Calibration Type rates were accepted and integrated in the simulation R08.
Hourly Monthly Integrating the calculated infiltration rates within the
ASHRAE 14 (2002) MBE ± 10% ± 5% simulation model resulted in decreased correlation and
RMSE 30% 15% increased error values in comparison to the outcomes of
IPMVP (2001) MBE - ± 20% the previous two runs, R01 and R07, with a RMSE of
RMSE 10-20% - 8.18%, a MBE of 1.39%, an average absolute error (Eav)
FEMP (2008) MBE ± 10% ± 5% of 1.81◦C and a correlation coefficient (R) of 0.85 (Table
RMSE 30% 15% 5). The frequency of errors for R08 indicated an
improvement towards obtaining a peak value for 0◦C and
R07 is presented as the next disclosed iterative run, since the errors between the monitored and the simulated
it could represent a level of mid-calibration with respect hourly data (Tm– Ts) ranging between -2◦C and +2◦C
to the accuracy achieved for simulated indoor were found to be 82% of the total hours (Figure 5) for
temperatures. Integrating monitored indoor temperature the 37 monitored spaces for a year (324120 hours).
data in the simulation model resulted in improved Moreover, integrating the calculated infiltration rates
correlation and error values in comparison to the resulted in an improvement in the MBE of heating
outcomes of the initial run R01, with 7.32% RMSE, energy consumption prediction of the model, the result
0.16% MBE, 1.62◦C absolute average error (Eav) and was significantly different, the underestimation of the
0.91correlation coefficient (R) (Figure 5, Table 5). In model decreased to -17.96% when compared to the
addition, integrating monitored indoor temperature data previous two runs R01 and R07, that yielded values of -
58.55% and -42.66%, respectively (Table 5, Figure 6).

Frequency of Errors for R01 Frequency of Errors for R07
90000 90000
74921
80000 80000 71898
70000 57901 70000
Number of Hours
52777
Number of Hours
60000 51189 60000 51785

44821
50000 37624
50000
34433
40000 31066 40000 27552
25195
30000 19405 20085 30000
20000 8588 10001 20000 7961 8398
10000 4275 10000
2 39 578 2588 1233 276 16 1 1 16 141 1634 1550 233 49 5 3
0 0
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
More
More
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
Error (∘C) Error (∘C)
90000 90000
79101 72135 77074 75505
80000 80000
70000 70000 54554
Number of Hours
Number of Hours
60000 48127 47838 60000

50000 50000 43237
40000 27213
40000 31267
30000 17538
30000
15036 14752 15002
20000 20000
4814 6740 4007 5251
10000 25 138 216 608 1096 2729 661 99 6 10000 9 31 47 186 722 1857 545 70 4
0 0
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
More
-8
-7
-6
-5
-4
-3
-2
-1
More
0
1
2
3
4
5
6
7
8

90000 80351
90000 80346
80000 71951 80000 72336
70000 70000
Number of Hours
57345
Number of Hours
60000 51068
55348 60000 51530
50000 50000
40000 40000
25318
30000 21541 30000 21677 22005
20000 10293 20000 10858
10000 2 13 25 102 643
3121 3032 958 325 29 0 10000 3 12 26 91 625
3067 3008 867 298
25 1
0 0
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
More
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
More
Figure 5 Frequency of errors (Tm -Ts) for iterative simulation runs

These outcomes indicated that the infiltration rate of a run was R13 and yielded improved results for heating
building could be highly influential on its energy energy consumption with respect to the monitored data,
consumption and is an important parameter that should with a RMSE of 23.76% and a MBE of -13.77% (Table
be included in simulation to obtain a higher level of 5, Figure 6). However, due to the changing envelope
accuracy. However, outcomes of the simulation run R08 characteristics, the simulation model yielded decreased
did not yet meet the calibration benchmarks defined in accuracy in indoor temperature data when compared to
Table 4, since RMSE and MBE values for heating energy the previous two runs R07 and R08, with 8.41% RMSE,
consumption calibration were not in acceptable margins. 2.16% MBE, 1.86◦C average absolute error (Eav) (Table
The following four runs focused on the fine tuning of 5, Figure 5). This result indicated that the altered
infiltration rates (adjustments of ±0.1-0.2ach for envelope parameters in the simulation model helped
monitored zones and other zones represented by the further in calibration for improved consumption patterns,
monitored zones) with the aim to reduce the increased however the thermal behavior of the indoor environment
frequency of errors between ±4 C and ±8 C (Figure 5)
◦ ◦ was negatively affected by such an intervention.
and an improvement of 1.12% MBE was achieved for Therefore, the consequent run, R14, was utilized for final
heating energy consumption at run R12. The model was adjustment of infiltration rates and indoor temperature
thenceforth analyzed for indoor temperature fluctuations profiles through an hourly comparison of monitored data
for the unoccupied and unconditioned spaces once more for 37 zones. The adjustments made on indoor
(as in R01) and it was concluded that the U-values temperatures ranged between ±0.2 to 1.0◦C and the
integrated in simulation could contain an inherent error adjustments made on infiltration rates were between ±0.2
from the thermocouple measurements. Additional heat to 0.3 each. The results of the run R14 provided
transfer of 10% was defined for the building envelope acceptable error margins with respect to the benchmarks
elements in contact with the exterior environment. This defined by the IPMVP (2001), ASHRAE Guideline 14
(2002) and FEMP (2008) standards (Table 4). As

presented in Table 5, the indoor temperature comparison 10,00
Monitored occupant presence- Office 305 Jan. 18th to 29th
between the R14 data and the monitored data yielded a
8,00
RMSE of 7.78%, a MBE of 0.95%, an Eav of 1.72◦C, and
6,00
a correlation coefficient (R) of 0.85 and the comparison
4,00
of monthly energy consumption resulted in 18.96%
2,00
RMSE and -5.05% MBE. The last run in calibrating the
0,00
simulation model of the existing building (R15) was
Hour of Day
9
19
5
15
1
11
21
7
17
3
13
23
9
19
5
15
1
11
21
7
17
3
13
23
9
19
5
15
intended to integrate the monitored occupant presence Actual occupancy sensible gain (W/m2 per hour)
Actual occupancy latent gain (W/m2 per hour)
data to determine the discrepancy between the Assumed occupancy sensible gain (W/m2 per hour)
Assumed occupancy latent gain (W/m2 per hour)
assumption of constant presence of office occupants and
their actual presence patterns in an office space (Figure Figure 7. Occupant presence in sample office space
7). Integrating solely the actual presence related CONCLUSION
occupancy gains within the simulation model resulted in
outcomes that yielded a 0.35% improvement in RMSE Simulation efforts that solely rely on predictions and
and 0.47% improvement in MBE for the heating energy assumptions could result in a significant discrepancy
consumption. Further modeling of occupant behavior when compared to the actual energy performance of a
and controls in addition to actual presence could provide building. Although calibration of simulation models via
a more significant improvement in simulation outcomes. monitoring data is considered as a time-consuming and
difficult process in building energy performance studies
In summary, it is possible to assert that; (a) parameters (Yoon, Lee and Claridge 2003), present research
such as building audit data, actual weather data, heating underscores the significance of such calibration
installation efficiency, measured U-values, and design approach in terms of controlling different parameters
temperatures were effective on obtaining acceptable correspondingly and understanding their sensitivities on
indoor thermal environment results from the simulation the overall consumption prediction of an accurate
model R01, however, were insufficient in terms of simulation model.
accurately predicting the heating consumption of the
building, (b) calibrating the simulation model with In order to evaluate the effects of energy conservation
hourly monitored indoor temperatures (R07) resulted in measures (ECMs) on building energy performance, it is
15.89% of improvement in MBE for heating energy necessary to obtain a base case simulation model that
consumption in comparison to the prediction of R01, (c) represents the existing thermal behavior of the building
integrating calculated infiltration rates in the simulation as closely as possible. In this respect, the findings of the
run R08 for each zone resulted in an improvement in the present study suggest that long term monitoring data (full
MBE by 24.7% compared to R07, (d) modeling U-values year) could facilitate an accurately calibrated building
with a 10% of error caused an underestimation of heating energy simulation model through a manual iterative
energy consumption by 4.19% in comparison to R08, (e) method. Such approach could as well be influential on
further adjustment of indoor temperatures and reducing the energy performance gap and the
infiltration subsequent to the integration of additional discrepancy between simulated and monitored energy
heat transfer was conducted and 8.72% improvement performances of buildings. In addition, present research
was achieved in MBE for heating energy consumption could help to underscore the fact that energy
calibration, and finally (f) actual occupant presence was performance modeling and calibration for existing
integrated in the model and a 0.35% improvement in buildings are detailed, yet convenient, since the baseline
RMSE and 0.47% improvement in MBE for the heating model could be developed through a meticulous audit
energy consumption were achieved for the particular and monitoring process. However, modeling and
calibration attempt presented in the study. calibration efforts during design phase could be more
challenging where especially monitoring data does not
600000
exist. The findings of the present study could as well be
500000
significant for design phase energy performance
400000
modeling and calibration, where input parameters such
300000
as occupant presence are considered as probabilistic data
200000
sets instead of assuming as deterministic inputs based on
100000
building schedules.
0
kWh
Jan
R01
Feb
R07
Mar
R08
Apr
R13
May
R14
Sep Oct
R15
Nov
Monitored
Dec
ACKNOWLEDGMENTS
This research was sponsored by the Eskisehir Osmangazi
Figure 6 Simulated and monitored heating energy
University, Commission of Scientific Research Projects
consumption

(Project ID 201515037). Special thanks are due to the Fumo, N. 2014. A review on the basics of building
project participant, Asst. Prof. Dr. Orkun Alptekin, for energy estimation. Renewable and Sustainable
his valuable efforts during the monitoring process of the Energy Reviews, 31, 53-60.
building. Harish, V., and Kumar, A. 2016. A review on modeling
REFERENCES and simulation of building energy systems.
Renewable and Sustainable Energy Reviews, 56,
Ahmad, M., and Culp, C. H., 2006. Uncalibrated
1272-1292.
building energy simulation modeling results.
HVAC&R Research, 12(4), 1141-1155. Hens, H., Parijs, W., and Deurinck, M. 2010. Energy
consumption for heating and rebound effects.
ASHRAE Guideline 14, 2002. Measurement of Energy
Energy and Buildings, 42(1), 105-110.
and Demand Savings, American Society of Heating,
Refrigerating and Air-Conditioning Engineers, Inc., IPMVP, 2001. International Performance Measurement
Atlanta. & Verification Protocol—Concepts and Options for
Determining Energy and Water Savings, Volume I,
ATTMA TSL2. 2010. The Air Tightness Testing &
Oak Ridge.
Measurement Association. Technical standard L2.
Measuring air permeability of building envelopes Menezes, A. C., Cripps, A., Bouchlaghem, D. and
(non-dwellings). Buswell, R., 2012. Predicted vs. actual energy
performance of non-domestic buildings: Using post-
Bordass, B., Cohen, R. and Field, J. 2004. Energy
occupancy evaluation data to reduce the
Performance of non-domestic buildings: closing the
performance gap. Applied Energy, 97, 355-364.
credibility gap. In: Proceedings of IEECB’04
Building Performance Congress, Frankfurt, Raftery, P., Keane, M. and Costa A. 2009. Calibration of
Germany. a detailed simulation model to energy monitoring
system data: A methodology and case study.
Coakley, D., Raftery, P. and Keane, M. 2014. A review
Proceedings of Building Simulation 2009, Eleventh
of methods to match building energy simulation
International IBPSA Conference Glasgow, Scotland
models to measured data. Renewable and
July 27-30, 2009
Sustainable Energy Reviews, 37, 123-141.
Royapoor, M. and Roskilly, T. 2015. Building model
Crawley, D. B., Hand, J. W., Kummert, M. and Griffith,
calibration using energy and environmental data.
B. T., 2008. Contrasting the capabilities of building
Energy and Buildings, 94, 109-120.
energy performance simulation programs. Building
and Environment, 43(4), 661-673. TS2164, 2000. Turkish Standard on Principles for the
Preparation of the Projects of Central Heating
De Wilde, P., 2014. The gap between predicted and
Systems.
measured energy performance of buildings: A
framework for investigation. Automation in Yoon, J., Lee, E.J. and Claridge, D.E. 2003. Calibration
Construction, 41, 40-49. procedure for energy performance simulation of a
commercial building. Journal of Solar Energy
Diakaki, C., Grigoroudis, E., Kabelis, N., Kolokotsa, D.,
Engineering, 125, 251-257.
Kalaitzakis, K., and Stavrakakis, G. 2010. A multi-
objective decision model for the improvement of
energy efficiency in buildings. Energy, 35(12),
5483-5496.
DoE, U. S., 2008. Energy efficiency trends in residential
and commercial buildings. US Department of
Energy, Washington, DC.
EN ISO 13791, 2004. Thermal Performance of
Buildings-Calculation of Internal Temperatures of a
Room in Summer Without Mechanical Cooling-
General Criteria and Validation Procedures.
FEMP, 2008. Measurement and Verification Guidelines:
Measurement and Verification for Federal Energy
Projects Version 3.0. DOE, US.


Ashrae Guideline 14

Uploaded by

Copyright:

Available Formats

Ashrae Guideline 14

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ashrae Guideline 14

Uploaded by

Copyright:

Available Formats

2018 Building Performance Analysis Conference and

SimBuild co-organized by ASHRAE and IBPSA-USA

CALIBRATION OF A BUILDING ENERGY PERFORMANCE SIMULATION MODEL

consumption on building level through energy

© 2018 ASHRAE (www.ashrae.org) and IBPSA-USA (www.ibpsa.us). 542

© 2018 ASHRAE (www.ashrae.org) and IBPSA-USA (www.ibpsa.us). 543

Table 3 Monitored building energy performance

© 2018 ASHRAE (www.ashrae.org) and IBPSA-USA (www.ibpsa.us). 544

Figure 3 Iterative runs and calibration attempts 15

The comparison between monitored and simulated data 5

a linear correlation (R) analysis based on hour-to-hour

© 2018 ASHRAE (www.ashrae.org) and IBPSA-USA (www.ibpsa.us). 545

© 2018 ASHRAE (www.ashrae.org) and IBPSA-USA (www.ibpsa.us). 546

60000 51189 60000 51785

60000 48127 47838 60000

Error (∘C) Error (∘C)

80000 71951 80000 72336

Figure 5 Frequency of errors (Tm -Ts) for iterative simulation runs

© 2018 ASHRAE (www.ashrae.org) and IBPSA-USA (www.ibpsa.us). 547

© 2018 ASHRAE (www.ashrae.org) and IBPSA-USA (www.ibpsa.us). 548

© 2018 ASHRAE (www.ashrae.org) and IBPSA-USA (www.ibpsa.us). 549

You might also like