Energy 77 (2014) 153e163

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Towards measurement and verification of energy performance under

the framework of the European directive for energy performance of
buildings
Esfand Burman a, *, Dejan Mumovic b, Judit Kimpian c
a
Centre for Urban Sustainability and Resilience, Chadwick Building, University College London, Gower Street, London WC1E 6BT, UK
b
Bartlett School of Graduate Studies, University College London, Central House, 14 Upper Woburn Place, London WC1H 0NN, UK
c
Aedas Architects Ltd, 5-8 Hardwick Street, London EC1R 4RG, UK

a r t i c l e i n f o a b s t r a c t

Article history: Directive 2002/91/EC of the European Parliament and Council on the Energy Performance of Buildings
Received 12 December 2013 has led to major developments in energy policies followed by the EU Member States. The national energy
Received in revised form performance targets for the built environment are mostly rooted in the Building Regulations that are
22 May 2014
shaped by this Directive. Article 3 of this Directive requires a methodology to calculate energy perfor-
Accepted 29 May 2014
Available online 26 June 2014
mance of buildings under standardised operating conditions. Overwhelming evidence suggests that
actual energy performance is often significantly higher than this standardised and theoretical perfor-
mance. The risk is national energy saving targets may not be achieved in practice. The UK evidence for
Keywords:
EPBD
the education and office sectors is presented in this paper. A measurement and verification plan is
Buildings proposed to compare actual energy performance of a building with its theoretical performance using
Energy performance gap calibrated thermal modelling. Consequently, the intended vs. actual energy performance can be estab-
Measurement & verification lished under identical operating conditions. This can help identify the shortcomings of construction
process and building procurement. Once energy performance gap is determined with reasonable accu-
racy and root causes identified, effective measures could be adopted to remedy or offset this gap.
© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/3.0/).

1. Introduction Article 3 of the EPBD requires all EU Member States to adopt a

methodology to calculate energy performance of buildings. Such
Energy efficiency is one of the key objectives of the European calculation shall include, as a minimum, energy use related to
policies to address the challenges of energy security and climate heating, hot water, cooling, ventilation and lighting under stand-
change. Substantial steps have been taken towards increasing en- ardised conditions prescribed by national regulations [3, p. L 1/67].
ergy efficiency, notably in the appliances market and construction This Article is the cornerstone of the Directive as the calculation
sector [1]. Energy consumption of buildings in the European Union methodologies developed are used to ensure the energy perfor-
accounts for around 40% of total final energy use and 36% of total mance requirements set out for new and existing buildings are met.
CO2 emissions of the EU Member States [2]. Therefore, it is vital to These calculation methodologies also underpin the energy perfor-
devise appropriate policies to improve energy efficiency of the mance certificate schemes implemented in the EU countries to
existing and new building stock. The Energy Performance of indicate energy efficiency of building stock.
Buildings Directive (EPBD), European Directive 2002/91/EC, came The recast of the EPBD, Directive 2010/31/EU, aims to extend the
into force in 2003 [3]. This Directive underpins the majority of scope of the original Directive by reducing area thresholds that
policies and regulations adopted by the EU Member States to make the EPBD requirements applicable to new and existing
improve energy performance of buildings in the last decade. buildings. It also mandates the Member States to set minimum
cost-optimal requirements for energy performance of buildings to
ensure there is a right balance between the investments involved
and the energy costs saved throughout the lifecycle of a building.
* Corresponding author. Tel.: þ44 020 3186 3042.
The Member States must draw up national plans to increase the
E-mail addresses: esfandiar.burman.10@ucl.ac.uk, es.burman@gmail.com
(E. Burman), d.mumovic@ucl.ac.uk (D. Mumovic), judit.kimpian@aedas.com number of ‘nearly zero-energy’ buildings. The nearly zero or very
(J. Kimpian). low energy performance could be achieved by a combination of

http://dx.doi.org/10.1016/j.energy.2014.05.102
0360-5442/© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/3.0/).
154 E. Burman et al. / Energy 77 (2014) 153e163

energy efficiency improvements and renewable energy produced predicted energy performance. Subsequently, a methodology is
on-site or nearby [4]. Furthermore, there is more emphasis in the presented for measurement and verification of energy use under
recast of the EPBD on quality assurance requirements to improve the EPBD framework. An example is also provided as proof of the
the accuracy and robustness of energy performance assessments concept. Finally, it is explained how this measurement and verifi-
[5]. cation plan could be scaled up and integrated into the EPBD. The UK
To implement Article 3 of the EPBD, the EU member states education sector and the National Calculation Methodology (NCM)
developed various calculation methodologies. Most countries implemented in England and Wales have been used to demonstrate
opted for whole-building simulation, using thermal modelling the concept. However, in principle, the framework outlined in this
software developed in accordance with national calculation paper could be applied to other countries within the European
methodologies, to determine energy performance of their building Union.
stock [6].
Thermal modelling is a useful method to calculate energy per- 2. Background literature
formance of a building through mathematical equations that relate
physical properties of the building such as external envelope's The drastic increase in oil prices in 1990s, following the world
thermal conductivities, air permeability, type and efficiency of energy crisis in late 1970s, raised governmental concerns regarding
heating, ventilation and air conditioning systems, and intensity of energy security. European countries' dependence on imported en-
lighting to the building's energy use under specific climatic con- ergy resources and the large contribution of building stock to na-
ditions. There are also a number of operating conditions that need tional energy use prompted the European governments to
to be defined and used in these calculations. Examples include introduce energy regulations. Europe developed regulations
building occupants' density, temperature set points, occupancy related to air tightness and building fabric in the late 1970s. Energy
profiles and operation schedules of building services. These oper- regulations related to building services were subsequently intro-
ating conditions are often unknown in the design of new buildings duced in various European countries [10].
or subject to a lot of uncertainty in existing buildings. Furthermore, Before 1970s post-occupancy studies that compare actual per-
the extent of small power loads such as plug-in and ICT equipment formance of buildings with design intents were predominantly
is often unknown prior to building completion. These uncertainties focused on architectural aspects of building performance, envi-
can compromise the accuracy and effectiveness of energy perfor- ronmental psychology and human behaviour in buildings [11e13].
mance calculations. This justifies the use of standardised operating With the ever-increasing significance of energy, the focus of post-
conditions under the EPBD. This is helpful for regulators as they can occupancy evaluations gradually shifted towards energy and in-
assess energy efficiency of construction projects under standard door environmental quality. In the United Kingdom, this trend was
conditions, defined in the national calculation methodologies, and accelerated by the Latham report which was commissioned by the
decide whether they meet minimum requirements. However, government to investigate the root causes for the poor quality
calculating energy performance under standard conditions means offered by the UK construction industry [14]. Over the last two
the outcome of such calculation cannot be directly compared with decades, this has led to a large body of empirical evidence that
the actual performance as actual operating conditions often differ could be used to investigate the root causes of shortfalls in opera-
from standardised conditions. An unintended consequence of this tional energy use.
policy set-up is where actual energy use of a building is higher than
what is calculated under the EPBD framework, it is often very 2.1. Energy performance gap: the context
difficult to identify what proportion of this discrepancy is due to
deviations from standardised operating conditions and what pro- The Post-occupancy Review of Buildings and their Engineering
portion is down to specific procurement issues associated with the (PROBE) was the first systematic post-occupancy evaluation pro-
building construction. Another unintended consequence is that gramme carried out in the UK with special focus on energy and
energy efficiency measures are tailored to comply with the Building indoor environmental quality [15]. Detailed study of 16 non-
Regulations' requirements under standardised conditions only, and domestic buildings that perceived to be exemplar designs over
do not necessarily reflect procurement and operational risks [7]. the period 1995e1999 revealed energy was often poorly specified
In theory, the calculated energy performance of a building after in briefing and design criteria. Actual energy use of most buildings
completion must reflect the as-built status including any procure- in the sample was higher than expectations. It was also found that
ment issues. However, as the supply side of the construction in- there was very little connection between the values assumed in
dustry is fragmented [8], it is often not feasible to check all design design estimations and computer models and actual values found
intents have been met in the immediate aftermath of building in the completed buildings [16].
completion. Furthermore, the evidence suggests there are short- More recent studies confirm the findings of the PROBE pro-
comings in complying with the EPBD requirements, enforcement of gramme, and reveal that despite technological advances there is still
the regulations, and the existing quality control schemes in all EU a significant gap between actual performance and design intents. For
Member States [9]. As actual energy use is not directly comparable example, a long-term post-occupancy study of five secondary
with the calculated energy use, there is a significant risk that schools in England that were intended to be low-energy buildings
energy-related procurement issues go unnoticed with any found that 80% of these buildings use more energy than expected.
discrepancy between actual and calculated energy being justified The energy performances of these buildings were between the 35th
solely on the basis of expected differences in operating conditions. and 82nd percentile of the national building stock. The introduction
This can seriously compromise energy efficiency of building stock of IT into the schools' curriculum, improved internal environmental
in the EU. standards, extended operating hours for extracurricular activities,
The aim of this paper is to show how a measurement and and poor control of building services were identified as major rea-
verification plan can be integrated into the existing EPBD frame- sons for higher than expected energy consumption [17].
work to ensure actual energy performance is in line with the Another study carried out on 28 new-build properties in the UK
theoretical assessment carried out after completion of a building. that used the EBPD compliant software for energy calculation found
First, a brief overview of the background literature is provided to that 75% of the case studies did not perform as well as expected.
highlight the significance of the discrepancy between actual and The projects covered a variety of building types including retail,
 E. Burman et al. / Energy 77 (2014) 153e163 155

education, offices and mixed use residential buildings. The identi- 2.2. Policy gap: measurement and verification of performance in-
fied root causes for this discrepancy were: reliance on calculated use
performance under standardised conditions rather than perfor-
mance in-use, inadequate prediction of energy use during design Excess in energy use over regulatory limit is often attributed to
stages, complexity of control strategies, poor construction prac- actual operating conditions and human behaviour that are not
tices, inadequate commissioning, insufficient means of managing adequately predicted at design stages [19,27]. However, a review of
the building systems' performance once operational, and lack of the implementation of the energy-related Building Regulations
designers/contractors involvement in fine-tuning buildings after across all EU Member States, Switzerland and Norway points out
completion. In the worst case scenario, combination of these fac- there is little attention to enforce these sustainable regulations. It
tors led to operational energy use being almost five times higher also highlights the shortage of qualified people with appropriate
than design estimate [7]. level of technical expertise to undertake the building control
CarbonBuzz is a collaborative research platform that aims to function in most European countries [9]. This finding is reinforced
share information about calculated and actual energy use of by another study on 404 new-build dwellings in England and Wales
buildings with a view to narrow this so-called energy performance that revealed only a third of these buildings were compliant with
gap. As of June 2013, energy performance records for 600 projects the energy efficiency requirements set out in the Building Regula-
had been reported to CarbonBuzz with the largest contingents tions. The study also pointed to the lack of adequate knowledge
being offices (around 40%) and educational buildings (around 30%). about energy efficiency requirements of the Building Regulations
Table 1 provides the mean data for design and actual energy use in among the supply and building control side of the construction
office and education sectors based on the latest audit on Carbon- sector [28]. Therefore, it is expected that part of excess in actual
Buzz data. energy use over regulatory limit is related to construction and
Most construction projects registered in CarbonBuzz are based building procurement process.
in the UK. Case studies carried out for non-domestic buildings in The discrepancy between actual and calculated energy perfor-
other European countries indicate discrepancies of up to 30% be- mance of buildings could be perceived as a sub-set of a more
tween measured performance and energy performances derived generic problem called energy efficiency gap first formulated in
from the EPBD compliant software [19,20]. 1990 [29]. Energy efficiency gap indicates the discrepancy between
In Housing sector, studies carried out in Europe often report actual and optimal energy use and is essentially a market failure. An
energy savings less than expected for retrofit projects and energy effective energy policy must be able to translate investments in
use higher than expected for new builds. Indoor temperatures in energy savings to economic value [30]. To bridge energy efficiency
heating season are often higher than modelling assumptions as gap in each sector, various structural and behavioural barriers must
building occupants expect to be more comfortable in new build- be identified and addressed. This paper deals with a specific
ings. This type of behavioural response to energy efficiency im- structural barrier related to policy making in the context of the
provements which leads to shortfall in expected energy savings is EPBD that is the lack of a requirement to verify energy performance
called the rebound effect [21]. An investigation into the impact of in-use.
occupant behaviour on energy consumption of dwellings in Austria The current energy assessment framework prescribed by the
provided evidence for a rebound effect between 20% and 30% in EPBD is overwhelmingly based on theoretical performance [6]. For
space heating [22]. A more recent review of the empirical evidence example, the UK Building Regulations require that total energy
for the direct rebound effect in household heating that covers ev- performance of every new building, calculated based on annual CO2
idence from the UK, Austria, Norway in Europe in addition to emissions, be no greater than total energy performance calculated
Canada and the US reveals a shortfall in expected energy savings of for a notional building that possesses minimum acceptable speci-
up to 68% with most UK studies reporting a mean shortfall above fication. The minimum specification is updated with every new
50% [23]. Also it is likely that any perceived financial saving on version of Building Regulations to set out ever more stringent CO2
energy is spent on appliances that in turn increase energy use targets [31]. The updates in minimum specification should pave the
[24,25]. A review of the English House Condition Survey (EHCS) way to a low carbon future that is in line with the national energy
consisting of 2531 unique cases found that homes with better en- saving targets. The risk is national energy saving targets may not be
ergy ratings often consume more energy than less efficient homes. achieved in practice if actual performance of building stock is
It is suggested that, while energy efficiency upgrades must be significantly higher than this theoretical performance [32].
adopted for homes with poor energy ratings, a combination of Discrepancy between actual and theoretical performance may
behavioural strategies and economic incentives should be used to be attributed to four major sources:
ensure energy efficiency measures already implemented in housing
stock lead to energy saving [26].  Inaccuracies and uncertainties associated with modelling inputs
Overall, there is strong evidence for energy performance gap in [27,33]
European non-domestic sector and housing stock that must be  Inadequacies of modelling methods and tools [34,35]
addressed to ensure ever stringent energy policies such as nearly  Procurement issues including construction process and building
zero-energy buildings, which will be enforced in the EU for all new commissioning [7,36]
buildings by 31 December 2020 [5], are effective.  Building management & operational inefficiencies [36,37].

Table 1
Calculated vs. actual energy performance for offices and educational buildings in the UK [18].

Category Mean desgin total Mean actual total Performance gap Mean desgin total Mean actual total Performance gap
heat consumption heat consumption factor change electricity consumption electricty consumption factor change
(kWh/m2/annum) (kWh/m2/annum) (actual to design) (kWh/m2/annum) (kWh/m2/annum) (actual to design)

Office 46 73 1.59 71 121 1.71

Education 57 84 1.48 56 106 1.90
156 E. Burman et al. / Energy 77 (2014) 153e163

Unless a like-for-like comparison is made between calculated EPBDverified: energy performance derived from a calibrated
and actual energy performance, it would be very difficult to thermal model under the EPBD settings
differentiate and address these root causes. In this paper, a method EPBDintended: EPBD calculation carried out following completion
is proposed to make such comparison feasible by reverting a cali- of a building
brated thermal model developed post-occupancy to the EPBD Actual Energy: measured energy performance based on meter-
standardised conditions. ing or utility bills

In practice, project teams have thermal models developed dur-

3. Methodology ing design stages, which are used to demonstrate compliance with
the EBPD requirements following completion of buildings. In the
First, a theoretical discussion is presented to establish how a United Kingdom, the main construction contractor is often liable
measurement and verification plan for energy performance of for building defects for at least one year after building handover.
buildings could be drawn up under the EBPD framework. Next, a Consequently, introducing a new requirement for measurement
brief description of the case study building used for proof of the and verification of energy performance within the first year of
concept is provided along with the criteria used for calibrating the building operation would not be onerous and the benefits achieved
thermal model with the actual performance. during the life-cycle of a building far outweigh the incurred costs.
Depending on the project type and contractual arrangements, the
measurement and verification plan could also be extended to allow
3.1. Measurement and verification of energy performance a building achieve its steady mode of operation before verification
of energy performance is carried out.
The results derived from a thermal model compliant with the Fig. 1 depicts the principle of using calibrated thermal models to
EPBD are not directly comparable with actual performance for the verify the performance calculated under the EBPD standardised
following reasons: conditions. The forward path shows how actual energy perfor-
mance could be significantly higher than calculated performance
 The use of standardised conditions/assumptions under the EBPD conditions. The backward path shows how a cali-
 A number of loads are not included in energy performance brated thermal model could be used to verify the EPBD calculation
calculations (e.g. actual small power and equipment load). and establish if there is any procurement gap. Procurement gap in
this context represents shortcomings in building design, con-
The International Performance Measurement & Verification struction process, system installation, implementation of control
Protocol (IPMVP) provides a framework to develop calibrated strategy, and building commissioning.
thermal models for energy saving projects where whole-building
simulation is required [38]. Calibration is achieved by adjusting
3.2. Case study building
the thermal model of a building to reflect the as-built status and
actual operating conditions. In the context of IPMVP, whole-
The case study building, used for proof of the concept, is a
building calibrated simulation after one year of steady post-
secondary school in North West England with total useful floor area
refurbishment occupancy could be used to establish energy sav-
of 10,418 square meters and nominal occupancy of 1150 pupils (see
ings achieved when pre-refurbishment energy performance is not
Fig. 2).
available or difficult to establish (e.g. multiple buildings on one site
It was constructed as a low carbon building in accordance with
without sub-metering). Once the thermal model is calibrated with
the UK Building Regulations 2006 and was completed in 2008. The
actual performance post-refurbishment, systems and settings may
building is located under the air path of Manchester airport.
be changed to pre-refurbishment conditions to establish the initial
The building is designed as a sealed envelope to screen the
baseline. The energy saving achieved is the difference between
ambient noise from the airport. Full mechanical ventilation with
energy performance derived from calibrated thermal model under
heat recovery is provided to all teaching and office spaces. Variable
pre-refurbishment conditions, and the actual energy performance
speed fans were part of the design strategy to save energy when the
measured after refurbishment work.
The principle and techniques outlined by the IPMVP for cali-
brating whole-building thermal models could be used to draw up a
measurement and verification plan under the EPBD framework. A
thermal model that reflects the steady post-occupancy operation of
a building for at least 12 months could be developed and calibrated
with actual energy use. Once calibration is achieved under actual
operating conditions, the model could be reverted to the EPBD
standardised settings to establish the verified performance under
the EPBD conditions. The following definitions can help separate
different aspects of actual vs. theoretical energy performance:

Procurement gap ¼ EPBDverified  EPBDintended (1)

Operational gap ¼ Actual Energy  EPBDverified (2)

Total energy performance gap ¼ Procurement gap

þ Operational gap (3)

where, Fig. 1. Measurement & verification plan for energy performance of buildings.
 E. Burman et al. / Energy 77 (2014) 153e163 157

Fig. 2. Axonometric view of the thermal model developed for the case study building.

building is not fully occupied. Four closed-loop ground source heat The monthly calibration method was used for this building. The
pumps with vertical boreholes act as the lead system for heating following criteria were used for calibrating the model with the
and are supplemented by gas fired condensing boilers. Domestic actual performance:
hot water is provided by the boilers via a calorifier. The ground
source heat pumps also provide limited amount of cooling to ICT  Calibration based on monthly utility data to achieve a CVRMSE
enhanced classrooms via chilled beams. Classrooms are designed to of 15% or better and an NMBE (Normalised Mean Bias Error) of
have a minimum 2% daylight factor. Electrical lights installed in 5% or better for gas and electricity use [38,39]. The Coefficient of
teaching and office spaces are high efficacy fluorescent lighting Variation of the Root Mean Square Error (CVRMSE) and Nor-
with efficiency better than 2.5 W/m2/(100 lux). The building fol- malised Mean Bias Error (NMBE) are defined as follow:
lows the normal England and Wales secondary schools' calendar
with some extracurricular activities and a night school that runs
two days per week during term time.
The building emissions rate calculated for the building following " #1=2 ,
X
n
its completion, using dynamic simulation with IES Apache soft- CVRMSE ¼ 100  ðyi  b
2
y i Þ =ðn  1Þ y (4)
ware, was 27.2 kg CO2/m2/annum. This accounts for all fixed i¼1
building services including heating, hot water, cooling, lighting and
auxiliary energy use under the EBPD standardised conditions Pn
defined in the National Calculation Methodology (NCM) for En- b
i¼1 ðyi  y i Þ
NMBE ¼  100 (5)
gland and Wales. ðn  1Þ  y
The authors performed a post-occupancy evaluation on this
building over the period 2011e2013. Total annual energy perfor- where:
mance of the building, based on the latest utility bills and metered
data, was 93.6 Kg CO2/m2.1 This accounts for all fixed building yi: measured monthly gas or electricity use
services, small power, server room load, external lights, lifts and b
y i : monthly gas or electricity use derived from thermal
other miscellaneous loads not regulated under the EPBD. modelling
A thermal model was developed for the building based on post- y: average monthly gas or electricity use for the measurement
occupancy information and observations with the same software period
used for the original EPBD calculation. The model was calibrated n: number of data points (n ¼ 12 for calibration based on 12
with the actual performance including all loads. Where it was not months of data)
possible to define some energy end-uses in the model, the results
were adjusted to ensure a like-for-like comparison is made be- In addition to the criteria set out by the IPMVP, the following
tween modelling results and actual total performance. Notably, criteria were set to ensure reasonable consistency between actual
external lights and lifts' energy use were added to the modelling and calculated annual performance is achieved:
results. Under the IPMVP framework, the data used to calibrate a
thermal model shall contain, as a minimum, monthly utility data  Total annual gas within 5% of the measured performance per
for 12 months [38]. It is also possible to use hourly calibration data. kWh/m2
 Total annual electricity within 5% of the measured performance
per kWh/m2
1
Carbon emissions conversion factors used for gas and electricity are 0.19 kg  Total annual energy performance calculated per kg CO2/m2
CO2/kWh and 0.55 kg CO2/kWh respectively. within 5% of the measured performance
158 E. Burman et al. / Energy 77 (2014) 153e163

Table 2
The inputs used for the calibrated thermal model based on post-occupancy evaluation.

Building characteristics Calibrated thermal model inputs

Heating 21% of heating demand is satisfied by the Ground Source Heat Pumps (sub-metered); gas
fired boilers supplement the GSHPs.
Coefficient of Performance for the GSHPs: 4.1
Gross efficiency of gas-fired boilers in condensing mode: 95.2%
Gross efficiency of gas-fired boilers in non-condensing mode: 88%
Ventilation Overall system Specific Fan Power of 4.02 W/L/s based on the commissioning results, no demand
control ventilation enabled. All main air handling units have thermal wheels installed for heat recovery.
Air conditioning Ground source heat pumps Energy Efficiency Ratio: 5.2
Server room DX units Energy Efficiency Ratio: 3.27
Hot water Hot water tank capacity: 2000 L with 0.0026 kWh/L/day loss.
Lighting All lighting wattages based on as-built drawings; average lighting density is 12.2 W/m2. Automatic
daylight sensing with an average daylight factor of 2% within 6 m of the building perimeter, absence
detection sensors in classrooms and presence detection sensors in circulation areas
External envelope The building external wall is brick block with insulated cavity.
Average U value for the external envelope including glazing: 0.48 W/m2 K
Impact of thermal bridges on average U value: 7.9%
Air permeability 9.2 m3/h/m2 @ 50 Pa (based on the pressure test results)
Equipment and other miscellaneous Sub-metered non-regulated energy; all electric unless stated otherwise (kWh/m2/annum):
loads not regulated by the Building Regulations ICT equipment (including servers) : 18.6
Small power: 8.9
Central catering (gas): 7.9
Central catering : 5.6
Distributed catering: 2.8
External lights: 4.6
Lifts: 0.2

Once these criteria were satisfied, the model was reverted to the occupancy pattern which determine domestic hot water re-
EPBD standardised settings to establish the procurement and quirements. However, sensitivity to items that will be standardised
operational gaps. for the EPBD calculations is not a major concern as long as the
While calibration based on monthly data was used for this average error is within acceptable limits. Calculated electricity is
building, hourly electrical power demand data available from the generally very close to the measured electricity. However, per-
utility supplier were also compared with the electrical power de- centage of error grows in summertime when the building occu-
mand derived from thermal modelling to assess the level of infor- pancy and use are highly erratic and difficult to fully capture within
mation required to make calibration based on hourly data feasible. the model. Again, this poses no problem for verification of the EPBD
This would provide an insight about the level of monitoring and the calculations as long as the overall error is within the limits set out
data points required to get accurate hourly predictions if higher for calibration.
level of accuracy is targeted in future projects. Annual performance: The outcomes of the model satisfy the
criteria set out for calibration. Therefore, following the backward
4. Results path of Fig. 1, the model is reverted to the EPBD settings and con-
ditions. This process involves removing actual small power and
Input data: Table 2 includes the inputs of the calibrated thermal equipment load that are not regulated under the EPBD and
model for building services, building fabric, equipment and replacing them with the EPBD default loads, using standard occu-
miscellaneous loads based on post-occupancy observations and pancy density and profile, standard heating and cooling set points,
measurements. Table 3 compares the actual operating conditions standard airflow rates for the ventilation system, and the stand-
observed in the building with the standardised operating condi- ardised schedules of operation.
tions used for the EPBD calculations. Lower occupant density, Most commercially available software for the EPBD calcula-
higher heating set points in classrooms, lower cooling set points, tions in the UK are capable of replacing actual settings with the
higher ventilation rates, and longer hours of operation in the actual EPBD standardised settings automatically. Therefore, once the
building are among the major differences between actual and model is calibrated based on the measured performance,
standardised operation. following the backward path of Fig. 1 is not time or resource
Monthly calibration: Fig. 3 and Fig. 4 show the calculated vs. intensive. Fig. 5 compares the measured performance with the
measured monthly gas and electricity use respectively. The outcomes of the calibrated thermal model, the verified EPBD
measured data is based on utility bills and the calculated data is calculation, and the initial EPBD calculation. All energy end-uses
based on thermal modelling. The heating components of gas con- are also sub-metered and compared with the modelling results.
sumption derived from dynamic simulation using Test Reference Table 5 reports the annual performance for gas, electricity and
Year (TRY) [40] weather file for Manchester have been weather total energy.
corrected based on actual heating degree-days experienced over Comparison between the verified and intended EPBD calcula-
the measurement period. The electricity consumption derived from tions reveals that the verified auxiliary energy use associated with
modelling has been adjusted to allow for external lights and lifts. fans, pumps and control under the EPBD conditions is significantly
The coefficient of variation of the root mean square errors and higher than the intended performance. Auxiliary energy use is also
the normalised mean bias errors for gas and electricity are listed in the highest energy end-use in the measured performance. Post-
Table 4 and are all within the acceptable limits set out for the study. occupancy studies revealed that poor implementation of the con-
Calculated gas is reasonably close to the measured gas except in trol strategy specified for the mechanical ventilation system led to
June and July. As the heating consumption is very low in these failure of demand-controlled ventilation (a procurement issue).
months, the modelling outcome is sensitive to slight changes in This was in turn compounded by poor building management (an
 E. Burman et al. / Energy 77 (2014) 153e163 159

Table 3
Standardised vs. actual operating conditions.

Operating conditions Standardised EPBD conditions for schools in England & Wales Actual operating conditions for the case study building

People density (pers./m2) Classrooms: 0.55 Classrooms: 0.50

Open office space: 0.11 Open office space: 0.09
Cellular office space: 0.07 Cellular office space: 0.06
Heating Set point ( C) Classrooms: 18 Classrooms: 21 ± 2
Offices: 22 Offices: 21 ± 2
Cooling Set point ( C) Classrooms: 23 Classrooms: 21
Offices: 24 Offices: 21
Ventilation rate (L/s/p) Classrooms: 5 Classrooms: 8
Offices: 10 Offices: 14
Schedules of operation: Occupancy: 7:00e18:00 Occupancy: 7:00e18:00; extended to 21:00 on Tuesdays & Thursdays
Weekdays; term time for night school
(standard diversity factors applied) (diversity applied based on post-occupancy studies)
Heating & Cooling: 5:00e18:00 (weekdays; term time) Heating, Cooling and Mechanical Ventilation: 6:00e18:00 weekdays;
Mechanical Ventilation:7:00e18:00 extended to 21:00 on Tuesdays & Thursdays for night school
(weekdays; term time) (Weekdays and school holidays)

operational issue) and led to excessive auxiliary and heating energy resolution for hourly calibration. There is also evidence of unnec-
use. This shows the knock-on effect of procurement gap on oper- essary plant room operation in early hours of the day during the
ational gap and the necessity to address it in the early stages of free running season (Fig. 8). Depending on the level of accuracy
post-occupancy. required, appropriate sensors could be installed and data points
To assess the effect of procurement issues on operational gap, defined within the Building Management System to capture detail
the identified root causes for the procurement gap were addressed information about building operation on an hourly basis. However,
in the thermal model. Fig. 6 illustrates that addressing the root it is important to strike the right balance between calibration cost
causes of the procurement gap in the case study building would not and accuracy. The analysis carried out on the case study building
only bridge the procurement gap but also narrow the operational demonstrates monthly calibration method can achieve acceptable
gap by one forth. level of accuracy and uncertainty with reasonable amount of effort
Hourly electrical demand profiles: Fig. 7 and Fig. 8 show the that is scalable for wider application in the construction industry.
calculated vs. measured electrical power demand curves for typical The monthly calibration method is also the preferred option under
days in heating and free running seasons respectively. The the IPMVP [38, p. 35].
measured data is based on hourly electricity data provided by the
utility supplier. The calculated data is derived from the thermal 5. Discussion
model and adjusted to allow for external lights and lifts. The
baseline demands, peak demands, and the shape of the demand First, the root causes for procurement and operational gaps in
curves predicted by the model reasonably match the measured the case study building are briefly reviewed. Next the implications
data. However, these graphs reveal that further information is of the proposed measurement and verification plan and the ways to
required to achieve better consistency if a whole-building calibra- integrate this plan to the existing policy framework are discussed.
tion method based on hourly calibration is targeted. Occupancy
profiles after normal school occupancy hours are very erratic and 5.1. Energy performance gap in the case study building
require detail attention for hourly calibration. For monthly cali-
bration, on-site observations during night school and extracurric- The total measured energy performance of the case study
ular activities along with semi-structured interviews with building building is 93.6 kg CO2/m2/annum. This is almost twice the energy
users were used to determine the occupancy pattern for out-of- performance of the median stock reported in CarbonBuzz and in
hours' activities. Using school attendance sheets (if available and the 90th percentile of the energy performance of secondary schools
reliable) or occupancy sensors can help collate data with finer in England and Wales [41]. Therefore, this school is one of the worst
performers in this building category. The measurement and

Fig. 3. Monthly gas use: calculated vs. measured. Fig. 4. Monthly electricity use: calculated vs. measured.
160 E. Burman et al. / Energy 77 (2014) 153e163

Table 4 speed could be manually adjusted. However, contrary to the design

Modelling errors for calibration based on monthly energy use. intent, there is no CO2 sensor installed in classrooms or extract
Fuel CVRMSE (%) NMBE (%) ductwork to trigger automatic regulation of fan speed based on
Natural Gas 14.4 1.4
occupancy level. Therefore, the demand-controlled strategy has
Electricity 8.0 3.9 failed and all fans are running at their full capacity regardless of
actual demand. The commissioning results also reveal that specific
fan powers at full load are higher than the maximum allowable
Table 5 under the UK Building Regulations. Another root cause for the
Annual performance for gas, electricity and total energy. procurement gap is poor actuator control at the sliding header
Annual performance Measured Calculated Difference
interface between ground source heat pumps and gas-fired boilers
which led to low contribution of the ground source heat pumps to
Gas [kWh/m2/annum] 87.2 86.1 1.3%
heating, almost half the design intent.
Electricity [kWh/m2/annum] 140.1 135.1 3.6%
Total Energy [kg CO2/m2/annum] 93.6 90.7 3.1% The most influential factor related to the operational gap is the
schedules of operation set for the heating and ventilation systems.
Schools are seasonally occupied buildings. This means that not all
building services need to serve all zones of a building at all times.
verification plan helped differentiate the root causes. The single The building is open to public in half term breaks and a number of
most influential factor in procurement gap is poor installation of teaching and admin staff may work in the building. However, fa-
the mechanical ventilation system. The default ventilation strategy cilities managers can take advantage of heating and ventilation
for schools in the UK is natural ventilation. Mechanical ventilation zoning to isolate parts of building that are not used. The schedules
is only used if natural ventilation is not feasible. Concerns about the of operation of these systems and the set points could be optimised
ambient noise levels led the design team to opt for full mechanical to save energy. None of these materialised in the case study
ventilation. Due to the power law relation between airflow and fan building, which led to poor energy performance.
power, mechanical ventilation strategy can have detrimental effect Addressing the issues related to demand-controlled ventilation
on energy performance unless demand-controlled ventilation is and ground source heat pumps along with optimised seasonal
specified. The building services' designers specified variable speed operation of heating and ventilation systems would have signifi-
drives for all airflow fans and performed the EPBD calculation cantly improved actual performance of the building.
assuming demand-controlled strategy is adopted. In practice, in- This case study confirms the feasibility of using calibrated
verters were installed on all supply and extract fans, but were only thermal models for measurement and verification of energy per-
used to balance the system at the commissioning stage. The fans' formance under the EPBD framework.

Fig. 5. Breakdown of measured and calculated energy performance. Small power and ICT equipment loads are not regulated under the UK Building Regulations. However, an
allowance is made for equipment load as part of the EPBD standardised settings to estimate building's heating and cooling demand. Energy consumed by this standard equipment
load is reported for the EBPD calculations in Fig. 5 to include all energy end-uses on the graph. The equipment energy reported for the measured and calibrated performance reflect
the actual loads.
 E. Burman et al. / Energy 77 (2014) 153e163 161

Fig. 8. Typical hourly electrical power demand curve: free running season
Fig. 6. Knock-on effect of procurement gap on operational gap.
(summertime).

5.2. Integration of M&V plan into the EPBD  Possibility of using the existing methods and tools with minor
adjustment for measurement and verification,
The Energy Performance of Buildings Directive has been greatly  Growing awareness of energy performance gap (credibility gap)
successful in reshaping the energy policy landscape in the EU and necessity to address it,
Member States. For example, following inception of the EPBD,  Cost effectiveness of the scheme given that thermal models are
whole-building energy performance calculations, energy certifica- already being used for whole-building performance calculation
tion for buildings, protocols for inspection of air conditioning sys- of new buildings and major renovations. Updating these models
tems, and provision of advice for boilers, have led to better after building handover and when steady mode of operation is
understanding of energy efficiency of buildings in the UK. The ev- achieved could be done with reasonable amount of time and
idence available from other Member States also confirms the resources.
pivotal role of the EPBD in achieving a low carbon built environ-  Measurement & Verification of energy performance post-
ment for the future of the EU [42]. However, regulatory frameworks occupancy is already an optional credit in the LEED sustain-
that are based on uncalibrated thermal modelling may hinder the ability rating system [43]. Total energy performance is calcu-
EU countries in achieving their ambitious energy efficiency targets. lated based on predicted operating conditions under LEED rating
One of the main contributions of the EPBD in the last decade has system. The framework presented in this paper makes it
been the development of a growing body of professional practi- possible to use a measurement and verification plan under the
tioners that have been formally trained and are qualified for ther- EPBD standardised operating conditions.
mal modelling. The cornerstone of the framework proposed in this
paper is to integrate a measurement and verification plan into the It is suggested that, depending on contractual arrangements,
EPBD to ensure measured energy performance is consistent with designers or the main construction contractor should take re-
the intended performance under identical operating conditions. sponsibility for implementation of the measurement and verifica-
The requirement of having identical operating conditions is satis- tion (M&V) plan, and report the results to the building control body
fied by a thermal model that reflects actual performance and is at the end of M&V period. In the context of United Kingdom, de-
calibrated in accordance with the International Performance Mea- signers can take responsibility for M&V plan in traditional contracts
surement & Verification Protocol. Enablers for successful imple- where they are supposed to witness the installation of systems and
mentation of this proposal are: confirm the design intents have been met following building
completion. For design & build contracts, where the main con-
 The existing body of energy assessors trained for thermal struction contractor takes full responsibility for building procure-
modelling, ment following the design stages, contractors can take
responsibility for implementation of M&V plans. The experts who
carry out the measurements and calculations must be registered
with the existing EPBD energy assessment schemes, and subject to
regular quality assurance checks. If poor building maintenance
within the measurement and verification period has compromised
the operation of building services, this can be identified by the
designer or contractor and confirmed by the building manager
before final calculations. An adjustment could then be made in final
calculations to reflect this. Otherwise, designers and contractors
should be held accountable for any procurement gap. The opera-
tional gap, on the other hand, is the responsibility of building users.
The measurement and verification process can help differentiate
the procurement and operational root causes of energy perfor-
mance gap. Furthermore, it can lead to a more proactive engage-
ment from the construction team post-occupancy that will help
fine-tune a building and provide effective training to building
users. Therefore, intangible benefits and tacit knowledge gained
from such measurement and verification plan could outweigh the
Fig. 7. Typical hourly electrical power demand curve: heating season. regulatory contribution of it.
162 E. Burman et al. / Energy 77 (2014) 153e163

National and regional energy targets set for the built environ- makes it difficult to compare actual performance with theoretical
ment reflect countries' concerns about climate change and energy performance calculated under the EPBD. Integration of an appro-
security. However, the regulatory limits are often not directly priate measurement and verification plan into the EPBD could help
comparable with measured energy performance. The framework separate procurement issues from operational inefficiencies. A
proposed here would enable effective measurement of any excess framework for this integration is proposed, and a case study is used
in energy use over the regulatory limit set out for a building. It to prove the concept. The case study confirms that calibrated
could be argued that the social cost associated with this excess in thermal models could reasonably match actual performance and,
energy use is greater than the cost associated with the regulatory hence, be used to establish energy performance of a building under
limit. For example, if the regulatory limit ultimately stems from the different sets of conditions, including the EPBD standardised
necessity to limit global warming to 2  C in accordance with the operating conditions. It is suggested that, depending on contractual
recommendations of the Intergovernmental Panel on Climate arrangements, building designers or construction contractors could
Change [44], any excess over this regulatory limit could cause take the responsibility of verifying energy performance of their
disproportionate environmental damage and, therefore, should be buildings after completion. This could lead to a more proactive
charged at a different rate or be subject to an environmental tax. engagement from the construction team in fine-tuning buildings
Excess in energy use over the regulatory limit set for buildings is an post-handover. A concerted action from the construction team and
economic negative externality that has not so far been addressed building users could pave the way to improved building perfor-
partly because it cannot be effectively measured under the current mance. Comparing actual energy performance with theoretical
policy framework. The argument put forward here follows the performance under identical operating conditions could also help
notion of Pigouvian tax that is used to reduce or eliminate envi- measure any excess in energy use over the regulatory limit that
ronmental negative externality by imposing a tax on a polluter stem from poor construction practices. In the context of climate
equal to the social cost of pollution [45]. The Stern Review esti- change, the social cost of excess in energy use over the regulatory
mated the social cost of carbon, in 2005 prices, at $85 per tonne of limit is disproportionately high. Therefore, this excess in energy use
CO2 for the business as usual scenario, defined by the Inter- could be considered a negative externality that must be measured
governmental Panel on Climate (scenario A1B) [46]. This review and treated effectively. Integration of measurement and verifica-
adopted a public policy framework that takes an ethical stance for tion plan into the existing policy framework can facilitate this.
the future generations by including low discount rates in its net
present value calculations to estimate the social cost of carbon. This Acknowledgements
price effectively reflects the risk of failure of climate change miti-
gation policies. Therefore, it is justified to impose such an envi- The Engineering and Physical Sciences Research Council of the
ronmental levy on any excess in buildings' energy performance UK supported this study through an Engineering Doctorate pro-
over the regulatory limit under identical operating conditions. gramme (Grant Reference Code: EP/G037698/1). The Technology
Measurement and verification of the energy performance of new- Strategy Board also supported the post-occupancy evaluation part
build and retrofit projects can identify any procurement gap of this study (Project no. 1798-16365).
which could be subject to environmental levies. Operational gap
may not be subject to these levies under the current policy References
frameworks as it is influenced by the way users operate a building.
However, addressing the procurement gap can also help narrow the [1] Carvalho M. EU energy and climate change strategy. Energy 2012;40:19e22.
[2] European Commission. Energy efficiency: delivering the 20% target. Brussels:
operational gap as a result of building fine-tuning and training European Commission; 2008.
provided to building users. [3] The European Parliament and the Council of the EU. Directive 2002/91/EC of
The M&V plan introduced in this paper makes it possible to the European parliament of the council of 16 December 2002 on the energy
performance of buildings. Official Journal of the European Communities 2003.
determine the procurement gap regardless of any potential dispute
L1/65-71.
about its root causes. However, robust protocols that clearly [4] Desideri U, Arcioni L, Leonardi D, Cesaretti L, Perugini P, Agabitini E,
distinguish the root causes for procurement gap from operational Evangelisti N. Design of a multipurpose “zero energy consumption” building
according to European directive 2010/31/EU: architectural and technical
issues would be beneficial to assess the effect of procurement gap
plants solutions. Energy 2013;58:157e67.
on operational gap. To this end, future work will focus on analysing [5] The European Parliament and the Council of the EU. Directive 2010/31/EU of
the root causes for procurement and performance gaps in a number the European parliament and of the council of 19 May 2010 on the energy
of buildings that were investigated by the authors as part of the performance of buildings (recast). Official Journal of the European Commu-
nities 2010. L153/13-35.
Building Performance Evaluation Programme instigated by the [6] Intelligent Energy Europe. Implementing the energy performance of buildings
Technology Strategy Board [47,48]. directive. Country reports 2008. Brussels: Concerted Action EPBD; 2008.
It is also suggested that the M&V plan introduced here could [7] Carbon Trust. Closing the gap e lessons learned on realising the potential of
low carbon building design. London: Carbon Trust; 2011.
first be applied to projects under voluntary and flexible frameworks [8] Cooper I. Post-occupancy evaluation e where are you? Building Research &
such as Soft Landings before wider applications. This provides the Information 2001;29(2):158e63.
opportunity to further improve the M&V plan, and assess cost [9] Building Performance Institute Europe (BPIE). Europe's buildings under the
microscope, A country-by-country review of the energy performance of
implications along with any possible unintended consequences. buildings; 2011.
The Soft Landings framework is focused on performance in-use and [10] Perez-Lombard L, Ortiz J, Gonzalez R, Maestre I. A review of benchmarking,
extends the after-care duties of construction teams up to three rating, and labelling concepts within the framework of building energy cer-
tification schemes. Energy and Buildings 2009;41:272e8.
years post-handover [49]. [11] Markus T, Whyman P, Morgan J, Whitton D, Maver T, Canter D, Fleming J.
Building performance. London: Applied Science; 1972.
6. Conclusion [12] Pol E. Environmental psychology in Europe: from architectural psychology to
green psychology. Aldershot: Avebury; 1993.
[13] Canter D. Beyond building utilisation. In: Powell J, Cooper I, Lera S, editors.
The Energy Performance of Buildings Directive has helped the Designing for building utilisation. London: E & FN Spon; 1984.
quest for energy efficiency and low carbon buildings in the EU. [14] Latham M. Constructing the team, final report of the first government/in-
However, there is no requirement under the EPBD to verify build- dustry review of procurement and contractual agreement in the UK con-
struction industry. London: HMSO; 1994.
ings' energy performance in-use. Furthermore, the use of stand- [15] Cohen R, Standeven M, Bordass B, Leaman A. Assessing building performance in
ardised operating conditions in energy performance assessment use 1: the probe process. Building Research & Information 2001;29(2):85e102.
 E. Burman et al. / Energy 77 (2014) 153e163 163

[16] Bordass B, Cohen R, Standeven M, Leaman A. Assessing building performance [35] Raslan R, Davies M. Results variability in accredited building energy perfor-
in use 3: energy performance of the probe buildings. Building Research & mance compliance demonstration software in the UK: an inter-model
Information 2001;29(2):114e28. comparative study. Journal of Building Performance Simulation 2009;3(1):
[17] Pegg I, Cripps A, Kolokotroni M. Post-occupancy performance of five low- 63e85.
energy schools in the UK. ASHRAE Transactions 2007;113(Part 2). [36] Dasgupta A, Prodromou A, Mumovic D. Operational versus designed perfor-
[18] UCL Energy Institute. Summary of audits performed on CarbonBuzz by the mance of low carbon schools in England: bridging a credibility gap. HVAC&R
UCL energy institute; 2013. Research 2012;18(1e2):37e50.
[19] Tronchin L, Fabbri K. A round robin test for buildings energy performance in [37] Bordass B, Cohen R, Standeven M, Leaman A. Assessing building performance
Italy. Energy and Buildings 2010;42:1862e77. in use 2: technical performance of the probe buildings. Building Research &
[20] Petersen S, Hviid C. The European energy performance of buildings directive: Information 2001;29(2):103e13.
comparison of calculated and actual energy use in a Danish office building. In: [38] Efficiency Valuation Organisation. International performance measurement
Proceedings of IBPSA-England first building simulation and optimisation and verification protocol. Concepts and Options for Determining Energy and
conference (BSO 2012), Loughborough; 2012. Water Savings 2012;1.
[21] Hirst E, White D, Goeltz R. Indoor temperature changes in retrofit homes. [39] Measurement of Energy And Demand Savings, ASHRAE Guideline 14, Amer-
Energy 1985;10(7):861e70. ican Society of heating, refrigeration, and air-conditioning engineers (ASH-
[22] Haas R, Biermayr P. The rebound effect for space heating: empirical evidence RAE). Atlanta: ASHRAE Inc.; 2002.
from Austria. Energy Policy 2000;28:403e10. [40] The Chartered Institution of Building Services Engineers (CIBSE). Environ-
[23] Sorrell S, Dimitropoulos J, Sommerville M. Empirical estimates of the direct mental design, CIBSE guide A. 7th ed. London: CIBSE; 2007.
rebound effect: a review. Energy Policy 2009;37:1356e71. [41] Godoy-Shimizu D, Armitage P, Steemers K, Chenvidyakarn T. Using display
[24] Barker T, Ekins P, Foxton T. The macro-economic rebound effect and the UK energy certificates to quantify schools’ energy consumption. Building
economy. Energy Policy 2007;35(10):4935e46. Research & Information 2011;39(6):535e52.
[25] Herring H. Energy efficiencyda critical view. Energy 2006;31:10e20. [42] Intelligent Energy Europe. Implementing the energy performance of buildings
[26] Kelly S. Do homes that are more energy efficient consume less energy?: a directive featuring country reports 2010. Brussels: Concerted Action Energy
structural equation model of the English residential sector. Energy 2011;36: Performance of Buildings; 2011.
5610e20. [43] US Green Building Council. LEED reference guide for building design and
[27] Menezes AC, Cripps A, Bouchlaghem D, Buswell R. Predicted vs. actual energy construction. LEED v4. Washington DC: USGBC; 2013.
performance of non-domestic buildings: using post-occupancy evaluation [44] Intergovernmental Panel on Cliamte Change (IPCC). Synthesis report, contri-
data to reduce the performance gap. Applied Energy 2013;58:157e67. bution of working groups I, II and III to the fourth assessment report of the
[28] Pan W, Garmston H. Compliance with building energy regulations for new- intergovernmental panel on climate change; 2007.
build dwellings. Energy 2012;48:11e22. [45] Pigou A. The economics of welfare. 4th ed. London: McMillan; 1932.
[29] Hirst E, Brown M. Closing the efficiency gap: barriers to the efficient use of [46] Stern N. The economics of climate change: the stern review. Cambridge:
energy. Resources, Conservation and Recycling 1990;3:267e81. Cambridge University Press; 2007.
[30] Annunziata E, Frey M, Rizzi F. Towards nearly zero-energy buildings: the [47] Kimpian J, Chisholm S, Burman E. Final report for building performance
state-of-art of national regulations in Europe. Energy 2013;57:125e33. evaluations on three new build academies: academy 360, Petchey academy
[31] HM Government. The building regulations 2000, conservation of fuel and and Stockport academy. UK technology strategy board building performance
power, approved document L2A; 2010. evaluation (BPE) programme, Project no. 1798-16365; 2013.
[32] Ekins P, Lees E. The impact of EU policies on energy use in and the evolution of [48] Kimpian J, Chisholm S, Burman E. Final report for building performance
the UK built environment. Energy Policy 2008;36:4580e3. evaluations on two new build schools: Brine Leas Sixth Form and Loxford
[33] Ahmad M, Charles H. Uncalibrated building energy simulation modelling re- secondary school. UK technology strategy board building performance eval-
sults. HVAC&R Research 2006;12(4):1141e55. uation programme, Project no. 1281-16183; 2013.
[34] Lomas K. The UK applicability study: an evaluation of thermal simulation [49] Building Services Research and Information Association (BSRIA). The soft
programs for passive solar house design. Building and Environment landings framework for better briefing, design, handover and building per-
1996;31(3):197e206. formance in-use, BSRIA BG 4/2009; 2009.