Analyzing the Carbon Performance Gap and Thermal Energy Performance Gap of School Buildings in Osijek-Baranja County, Croatia

Abstract

This study examines the Carbon Performance Gap (CPG) and Energy Performance Gap (EPG) of school buildings in Osijek-Baranja County in Croatia. The variance between the predicted energy efficiency of a building, as indicated by the energy performance certificate (EPC), and its actual performance in terms of energy consumption, is often referred to as the EPG while the variance between the predicted carbon emission of a building from the EPC and its actual emission is referred to as CPG. This study aims to determine the extent of CPG and EPG between actual energy consumption/carbon emission and the calculated, which is presented in EPCs of school buildings. The average EPG among the analyzed schools was found to be 71.73% while the average CPG was found to be 78.77%. The analysis also revealed a substantial average annual difference in heating costs attributable to the EPG. By addressing EPG and CPG while optimizing energy usage, educational institutions can achieve substantial cost savings and contribute significantly to sustainability goals.

1. Introduction

Buildings play a crucial role in energy consumption and are significant contributors to environmental pollution [1]. Globally, they are responsible for over 40% of energy use [2,3,4], a trend that holds true in Croatia, where the Energy in Croatia 2022 report states that buildings account for more than 47% of the nation’s total energy consumption [5]. Considering the long lifespan of buildings, it is essential to recognize their lasting and continuous impact on the environment [6].

According to findings from the European Commission, most EU buildings, precisely 85%, were constructed before 2000 [7]—furthermore, a substantial 75% exhibit inadequate energy efficiency. Taking action to improve the energy efficiency of buildings is crucial for conserving energy and reaching the goal of having a building stock that produces no emissions and is entirely decarbonized by the year 2050. By 11 October 2025, European Union (EU) member states will be required to create and publicly disclose a comprehensive list of buildings owned or used by government entities with a total useful floor area exceeding 250 m². This inventory must be updated at a minimum frequency once every two years [8]. Furthermore, as mandated by the Energy Performance of Buildings Directive (2010/31/EU), as modified by Directive (2018/844/EU), EU nations must create national renovation plans that incorporate measures and policies aimed at all public buildings [8]. A similar approach has been adopted in Croatia, where statistics show that non-residential buildings made up a sizeable portion of the building area in 2018—28%. It is interesting to note that a fairly large proportion of the buildings in this category—10% of all non-residential building areas—are educational. In 2030, 2040, and even 2050, educational buildings are expected to maintain a steady 10% share [9]. Furthermore, areas that require energy efficiency improvements are highlighted by the division of building stock type, particularly the high proportion of non-residential structures [10]. The forecast that educational buildings would continue to take up a significant percentage in the future emphasizes the long-term impacts of patterns of energy usage. It suggests that implementing sustainable concepts will have a lasting impact on Croatia’s environmental effects and energy usage [9].

Buildings are one of the biggest individual energy users and a major contributor to harmful greenhouse gas (GHG) emissions, particularly carbon dioxide (CO₂) [11]. Therefore, increasing energy efficiency in buildings is one of the best ways to reduce harmful environmental emissions and energy costs [12]. Energy efficiency refers to reducing energy use while achieving appropriate thermal comfort levels, indoor climate conditions, and lighting. Furthermore, energy efficiency encourages using economically viable technologies, materials, and services [13]. This helps to decrease the release of GHG that contribute to global warming while also increasing public awareness about efficient energy usage and leading to financial savings [14].

The energy properties, characteristics, and energy consumption of buildings largely depend on the building’s period of construction and the existence and application of regulations on thermal insulation [15]. The absence or inadequacy of regulatory measures is the primary reason why a considerable proportion of older buildings constructed many years ago have a substantial negative impact on the environment and exhibit high energy consumption levels [16]. The Republic of Croatia has a high thermal energy consumption due to the large number of buildings constructed during the rapid population growth and building construction in the early to mid-20th century, particularly in the 1960s, 1970s, and early 1980s. During this period, there were insufficient laws regarding energy efficiency and thermal insulation in buildings. Therefore, a prevailing proportion of buildings were constructed before and during the 1970s of the 20th century (54%), prior to the systematic implementation of legislation regarding the thermal insulation of buildings [17]. This can be seen from Figure 1.

The energy properties of a building can be represented by an energy performance certificate (EPC) to enable comparison between buildings and, more importantly, to undergo energy efficiency retrofitting [18,19]. In Croatia and the EU, the obligation of EPCs stems from EU directives. The foundation for the establishment of an energy certificate system in EU buildings was specifically Directive 2002/91/EC of the European Parliament and the Council on 16 December 2002, on the energy performance of buildings. This regulation established the criteria for assessing a building’s energy efficiency and mandated that an EPC should be provided when a building is being sold or rented [20]. In Croatia, the implementation of the obligation of EPC has been incorporated into the national legislation due to harmonization with EU directives. The Energy Efficiency Act (Official Gazette 127/14, 116/18, 25/20, 41/21), the Ordinance on energy audits of construction works and energy certification of buildings (Official Gazette 81/12, 29/13, 78/13), and the Ordinance on energy audits and energy certification of buildings (Official Gazette 88/17, 90/20, 1/21, 45/21) are important laws in Croatia that regulate EPC matters in compliance with European directives [21].

Information about the building, its energy class, its validity period, the person who issued and created the energy certificate, the people who contributed to producing it, the energy certificate’s label, information about thermomechanical systems, the building’s energy needs, data on the use of renewable energy sources, proposed measures, more detailed information, and an explanation of the energy certificate’s content are all included in the EPC, which is valid for ten years after it is issued [20,22,23]. There may be a substantial difference between actual energy consumption and standard user occupancy profiles, standard heating periods based on the climatic zone, traditional space heating switch-on, operating settings, average climatic conditions, and limited energy services [24,25]. It is crucial to remember that the validity of EPCs can be questioned because of this. The disparity between calculated and actual energy usage has been widely acknowledged as a major obstacle to effectively assessing building performance and the expected benefits of energy-saving measures in the context of a comprehensive assessment of a country’s building stock [24]. Also, it has been noticed that this problem occurs not only in existing buildings but also in new ones [26].

This research deals with the gap in EPCs, which refers to the variance between the predicted energy efficiency of a building, as indicated by the certificate, and its actual performance in terms of energy consumption, often referred to as the Energy Performance Gap (EPG). In this study, we define the Carbon Performance Gap (CPG) as the difference between predicted and actual carbon emissions of school buildings, analogous to the EPG, which refers to discrepancies in energy consumption. Despite the aim of EPCs to provide accurate information regarding a building’s energy efficiency, there is often a significant disparity between predicted and actual energy performance outcomes [27]. This disparity, known as the Energy Performance Gap (EPG), can be attributed to two primary factors: inaccuracies in the physical representation of buildings, and the behavior of the building’s occupants [28,29]. The physical discrepancies refer to issues such as poor insulation, outdated construction materials, or discrepancies in the building’s actual thermal performance compared to what is assumed in the EPC calculation. On the other hand, behavioral causes stem from how occupants interact with the building’s energy systems, such as deviations in heating usage patterns, equipment operation, and maintenance practices. Furthermore, operational failures of a building’s energy systems, such as HVAC malfunctions or inefficient controls, also contribute to the EPG [30]. Numerous studies have found that actual energy consumption is often two to five times higher than what is predicted by EPCs [31,32,33,34]. Also, regarding the carbon emission gap, it was found in the UK that the average actual carbon emissions were 3.8 higher than the calculated [35].

Many studies have dealt with the energy performance gap worldwide [18,19,28,31,32,36,37,38,39,40]. However, such studies have rarely focused on educational buildings [41,42,43] and also very rarely focused on CPG. Also, research has observed that variations between countries necessitate carrying out country-specific studies on the EPG [18,44,45]. From this perspective, there has been only one research on this topic in Croatia [46], and none deals with educational buildings. Therefore, this study aims to determine the extent of the EPG and CPG between actual energy consumption/carbon emission and the calculated one, which is presented in EPCs in school buildings in Osijek-Baranja County in Croatia. The remained of the paper is structured as follows: Section 2 presents a literature review in the field of EPG analysis in various types of buildings, specifically educational buildings, and a short review of research on CPG. Section 3 presents the methodology used; Section 4 provides the results of the study and a discussion; Section 5 presents the limitations of the study, and in Section 6 conclusions are drawn based on the analysis.

2. Literature Review

2.1. EPG Analysis in Various Types of Buildings

Palladino measured the EPG by studying Italian residential buildings. The climatic zone, building type, consumption profile, and degree of thermal insulation of the buildings were all taken into account in this analysis. The author also looked into any prebound or rebound effects and how these characteristics affected the EPG [25]. The prebound effect is when actual consumption is lower than the calculated consumption. Conversely, the rebound effect occurs when the actual consumption exceeds the calculated consumption [47,48]. The study found a wide range of variability in the effects of prebound (ranging from 0% to +80%) and rebound (ranging from −30% to 0%) on EPG. The study measured the average energy consumption per heating degree day in the specified site, ranging from −3 to +16 kWh, according to the usage profile [25]. Hernandez-Cruz et al. [30] conducted a study examining the monthly energy usage of 481 residential units located in six buildings of the social housing inventory in the Basque Country, Spain, over three years. The statistical analysis revealed a notable variation in consumption among dwellings within the same building. The study revealed that user behavior significantly impacts the efficiency of centralized building systems, with an average efficiency rate of 65%. The energy usage in these social housing complexes is much below the regional, national, and European average metrics. In addition, the EPG values deviated from the typical values reported in the literature. Specifically, the actual energy consumption of these buildings ranged from 0.70 to 2.28 times higher than the calculated consumption [30]. Several authors also studied EPG in energy-efficient buildings. Authors Padey et al. reported an EPG of 20–30% in an energy-efficient building in Switzerland [49]. Coyne and Denny [18] conducted a study on the EPG of 9923 households, analyzing the actual energy consumption data from 2014 to 2017. The findings indicate that the EPC has little impact on actual energy consumption, and the study observed a range of 457 kWh/year across different EPC levels in the total sample. The researchers also discovered that the most energy-efficient dwellings had an average surplus of 2998 kWh/a, 39% more than the calculated value. In contrast, less efficient homes had lower actual energy usage than predicted values, while the average difference varied from 24% for homes with a D rating to 56% for homes with F and G ratings [18]. In order to evaluate the impact of EPC on energy consumption, capacity growth, and potential savings measures in the construction sector and associated utility services, Anđelković et al. [50] carried out a case study in Novi Sad, Serbia. The case study encompassed buildings that were linked to the local district heating system, as well as those that were only connected to the municipal gas network. The results have indicated a substantial disparity between actual and calculated energy consumption. Furthermore, the authors observed more pronounced disparities in buildings utilizing gas heating systems than district heating ones [50]. Authors Motuziene et al. [51] analyzed the EPG for high-energy performance buildings in Lithuania. The analysis has revealed that the EPG exhibits a range of variation from −101% to +77% for class A. Furthermore, the A+ and A++ classes have a more limited range, ranging from +18 to 76% and +23 to 77%, respectively. The author’s findings agreed with several other studies, verifying that high-energy performance buildings utilize more energy than initially calculated. It was also emphasized that Lithuania faces the same issue despite variations in national certification techniques, and hence, adjustments are necessary for EPC schemes [51]. Cozza et al. [36] conducted an analysis of the EPG in residential buildings in Switzerland by utilizing the EPC database. The researchers discovered that the median EPG was −11%. However, it exhibited variation among different classifications, ranging from 12.4% for class B to −40.4% for class G. It was determined that buildings with low energy ratings utilize much less energy than anticipated. Conversely, buildings with high ratings tend to consume slightly more than anticipated. The analysis of class A buildings revealed an EPG of −6.2%, indicating that highly efficient buildings may be more resistant to the effects of the EPG [36].

2.2. EPG Analysis in Educational Buildings

Authors Kim et al. highlighted that the actual energy consumption in schools can be 60–70% higher than the calculated and even over 85% for universities [52]. In addition, authors Van Dronkelaar et al. have reported a 67% average EPG for schools and the same average EPG for university buildings [38]. Authors Herrando et al. [53] analyzed the EPCs of 32 faculty buildings in Zaragoza, Spain. They found that the actual energy consumption of the buildings is around 30% higher than the one in the EPCs [53]. The authors, Kim et al. [52], conducted a case study on a university building in the UAE. An energy audit, post-occupancy evaluation, and dynamic simulation were conducted to identify the causes contributing to the dynamic energy performance gap and assess the effectiveness of strategies in minimizing this gap. Moreover, the building energy audit and post-occupancy review data were utilized to verify and adjust a dynamic simulation model. The findings indicated that the building’s systems were not functioning according to their intended design, resulting in over 25% of the cooling energy wastage. This was attributed to the building facilities management’s failure to oversee the mechanical systems [52].

From the literature review on EPG, it is evident that previous research on educational buildings has primarily concentrated on university buildings. In contrast, studies that focus exclusively on school buildings have not yet been conducted, at least to the best of the authors’ knowledge. This gap highlights the need for targeted research on school buildings to address any specific issues and requirements distinct from those of university buildings.

2.3. CPG Analysis in Various Buildings

Cozza et al. in their previously mentioned research on residential buildings in Switzerland [36], analyzed the actual and theoretical CO₂ emission but without explicitly quantifying CPG. Innovate UK analyzed the CPG on multiple public buildings and found that the average carbon emission was 3.8 times higher than the design estimate [35]. A lot of research has emphasized the importance of reducing CO₂ emissions, yet as can be seen from the very short literature review on this topic, almost none have dealt with the carbon performance gap of buildings.

To enhance the clarity of the literature review and facilitate comparisons across different studies, a summary table has been provided below (

Table 1. Summary of literature review.

Study	Building Type	Location	EPG/CPG Range	Key Findings
Palladino [25]	Residential	Italy	Prebound: 0% to +80%; Rebound: −30% to 0%	EPG varies by climate zone, insulation, and consumption profile.
Hernandez-Cruz et al. [30]	Residential	Spain	0.70 to 2.28 times higher than the calculated consumption	Large variation in energy consumption across identical buildings.
Padey et al. [49]	Energy-Efficient	Switzerland	20–30%	EPG exists even in highly efficient buildings.
Coyne & Denny [18]	Residential	Various	+39% (most efficient homes); −24% to −56% (less efficient homes)	EPCs have little effect on actual energy consumption.
Anđelković et al. [50]	District/Gas Heating Buildings	Serbia	Significant disparity (Gas heating has a higher gap)	EPC underestimated actual consumption, more so in gas-heated buildings.
Motuziene et al. [51]	High-Energy Performance	Lithuania	−101% to +77% (Class A); +18% to 76% (A+); +23% to 77% (A++)	EPG higher in high-energy performance buildings.
Cozza et al. [36]	Residential	Switzerland	Median: −11%; Class B: +12.4%; Class G: −40.4%	Low-rated buildings consume less, high-rated consume slightly more.
Kim et al. [52]	Educational	UAE	25% cooling energy wasted	Systems mismanagement caused high consumption.
Van Dronkelaar et al. [38]	Educational (schools and universities)	United Kingdom	67%	EPG consistent across schools and universities.
Herrando et al. [53]	Faculty Buildings	Spain	+30%	Actual energy use exceeded EPC predictions.
Innovate UK [35]	Public Buildings	UK	CPG: 3.8 times higher than design estimates	Large carbon emission discrepancies.

). This table outlines the key findings related to the EPG and CPG in various building types, locations, and study contexts. It highlights the range of discrepancies between actual and calculated energy consumption, along with insights into the factors influencing these variations. By presenting the data in a structured format, this table allows for a better understanding of how EPG and CPG manifest in different settings and underscores the need for further research, particularly in educational buildings.

3. Methodology

3.1. Energy Rating of Buildings in Croatia

In Croatia, based on the calculation of the specific annual required thermal energy for heating Q_H,nd,ref, the building is classified into an energy consumption class, from A+ class with the lowest consumption of thermal energy for heating (Q″_H,nd ≤ 15 kWh/m²·a), to G class building with the highest energy consumption (Q″_H,nd > 250 kWh/m²·a) (

Table 2. Energy classes of buildings [54].

Energy Class	Q″H,nd—Specific Annual Required Thermal Energy for Heating [kWh/m2·a]
A+	≤15
A	≤25
B	≤50
C	≤100
D	≤150
E	≤200
F	≤250
G	>250

) [54].

From 2017, buildings are also classified into one more consumption class based on the calculation of specific annual primary energy for reference climate data and the Algorithm-prescribed regime of use of space and regime of operation of technical systems [55]. However, since many EPCs analyzed in this research are older than 2017, this criterion was not taken into consideration and the analysis is conducted only for thermal energy demanded for heating. Most of the schools obtained EPCs in 2011 when it became mandatory.

The calculation of EPCs in Croatia is based on standardized usage profiles defined in the Algorithm for Calculating the Required Energy for Heating and Cooling according to HRN EN 13790 [56]. For educational buildings, the usage period is defined as 12 h per day, from 8:00 to 20:00, with a total of 14 h of heating per day (starting 2 h before the building is in use) and a five-day workweek, Also, the Algorithm defines the projected indoor temperature in heating season to be 20 °C and the internal gains for non-residential buildings as 6 W/m². Regarding the ventilation, the number of air changes at a pressure difference of 50 Pa, measured during the air permeability test, n₅₀ [h⁻¹] is defined to be ≤3.0 h⁻¹ for buildings without mechanical ventilation systems and ≤1.5 h⁻¹ for buildings with a mechanical ventilation system [56]. The most often used software tools in Croatia for EPC calculations are KI EXPERT PLUS, Thorium A+ and Energetski certifikator [57,58,59].

3.2. Study Area: Osijek-Baranja County, Croatia

The study is located in Osijek-Baranja County in the Republic of Croatia. It is a county in the east of Croatia characteristic for its fertile fields, which are the basis of a rich agricultural tradition (highlighted dark orange in Figure 2).

As an important economic center of eastern Croatia, it has a significant influence on energy consumption in the region. Given the diversity of its industries, agriculture, and urban areas, energy consumption varies and spans a wide range of sectors. Considering all these factors, Osijek-Baranja County faces the challenge of sustainable energy consumption and the need to introduce energy-efficient technologies and practices. Developing renewable energy sources, improving energy efficiency, and educating the public about the importance of rational energy consumption are key steps towards a more sustainable future for the region.

Osijek-Baranja County is in Croatia’s continental region based on its geographic position. When selecting the average monthly outdoor air temperature of the coldest month at the building’s location (θ_e,mj,min), it is necessary to determine it based on the nearest meteorological station. If θ_e,mj,min ≤ 3 °C, the building is located in the continental part of Croatia. Cities and towns in continental Croatia are defined as having 2200 degree days or more of heating, and their yearly energy requirements are determined using the continental Croatian reference climatic data. The sum of the temperature variations between the average daily outside temperature and the inside design temperature for every day of the heating season is known as the number of heating degree days [54]. The County’s Heating degree day values were publicly published in 2008 and have not changed since. Accordingly, Osijek as the County’s capital has an average of 3057 heating degree days which confirms its belonging to the continental Croatia region with 2200 degree days or more of heating [60].

The emission factor for calculating CO₂ emissions for gas heating is 1.095, while for district heating the emission factor is 1.529 (average for Osijek—the capital of the analyzed County). These are the national averages for Croatia and in case of district heating, for region capitals [61].

3.3. EPC Data Collection

The EPCs were collected for 81 primary and 13 secondary schools in Osijek-Baranja County in Croatia. Most EPCs were obtained from the Croatian Information System of Energy Certificates (IEC) [62]. The IEC application is used for issuing and storing EPCs of buildings, entering data on the energy status of buildings, collecting, storing, and managing data on buildings, listing authorized persons for conducting energy audits and energy certification of buildings, etc. [62]. However, only those EPCs issued since 1 October 2017 are available on the IEC. Therefore, in December 2023, a request was sent to the Ministry of Physical Planning, Construction, and State Assets with an official request for the data on EPCs of school buildings in the observed County whose energy certificates are not in the IEC, i.e., those that were issued before the specified date [63]. In addition, to fill the gap even further, schools were contacted by email for the purpose of research and asked to provide the EPCs of their building. Using all three approaches, a total of 94 EPCs were obtained for the schools in the county.

3.4. Actual Data Collection

The Ordinance on Systematic Energy Management in the Public Sector (Official Gazette 18/2015), which was adopted in 2015, describes the framework for systematic energy management in the public sector, as well as the responsibility to control energy and water consumption, analyze consumption, and report on energy and water consumption [64]. Furthermore, the Energy Management Information System (EMIS) was developed to facilitate strategic energy planning and sustainable energy resource management in buildings owned or used by the Government of the Republic of Croatia, counties, cities, and other government budgetary and extra-budgetary users, as well as public authorities [65]. An official request for permission to access the EMIS was made in December 2023 for research reasons, and access to primary and secondary school data for the Osijek-Baranja County area was granted.

4. Results and Discussion

4.1. Descriptive Analysis

The analysis for both EPC data and actual data was performed using TIBCO Statistica^® 14.1.0 [66] and Microsoft Excel workbooks [67] compatible with the aforementioned software. To ensure comparability between EPC and actual data, the actual thermal energy for heating consumption data were collected from 2018 to 2022 for 94 school buildings on an annual basis and divided by the provided useful area of the school buildings to obtain the same unit of measure of Q″_H,nd—specific annual required thermal energy for heating [kWh/m²·a] as in the EPCs. The carbon emissions data were also divided by the provided useful area of the school buildings to obtain the same unit of measure [kgCO₂/m²·a] as in EPCs.

Table 3. Descriptive statistics analysis of EPC and actual thermal energy consumption and carbon emission.

Variable	N	Mean	Median	Min.	Max.	Lower Quartile	Upper Quartile
EPC Q″H,nd [kWh/m2·a]	94	82.06	67.13	17.00	238.00	39.00	116.06
Actual Q″H,nd [kWh/m2·a]	94	105.52	102.02	0.46	239.79	75.36	136.70
EPC CO2 [kgCO2/m2·a]	84	27.50	20.13	2.00	101.53	13.50	40.35
Actual CO2 [kgCO2/m2·a]	84	33.09	28.20	2.46	130.83	21.48	40.77

shows the descriptive statistics analysis for the EPC and actual data where it can be seen that the mean and median values for the actual consumption and carbon emission are notably higher than those for the EPCs. This suggests that, on average, the actual energy consumption and actual carbon emission exceed the expected values. However, the minimum EPC energy consumption is higher than the actual ones, which is very likely caused by faulty measurements or incorrectly entered data by the user.

Also, the lower and upper quartiles for the actual energy consumption are notably higher than the expected values, indicating a significant portion of instances where the actual consumption exceeds the upper expected threshold. Overall, this analysis suggests notable discrepancies should be analyzed between the expected and actual energy performance.

Figure 3 displays the analysis of the overall count of each energy class in the dataset for both the EPCs and the actual data. The energy classes for the actual data were established based on Table 1, which categorizes the buildings according to their specific annual required thermal energy for heating. From the actual consumption statistics, it is evident that there are three buildings classified as energy class A+. However, according to the EPCs, no buildings belong to this category. This is likely due to incorrect consumption measurements, leading to excessively low consumption values, such as the aforementioned unrealistic minimum value of the actual consumption. The detailed consumption analysis of these buildings showed that even though the average yearly consumption was observed, there were no significant deviations in individual consumption per year. This also highlights the importance of determining the EPG and suggests that either the actual consumptions were recorded incorrectly or there is an inaccuracy of the EPCs.

Furthermore, it can be seen that the majority of buildings are rated in the mid-range categories of EPC ratings (B, C, and D). In terms of actual performance, the highest number of buildings falls within the C and D classes. Besides the absence of class, A+ in EPC ratings, there is an absence of class A in the actual performance data, suggesting a potential discrepancy between expected and actual performance for this category. The most notable difference with regards to class B where the expected number of buildings according to the EPCs was supposed to be 22, while the actual number of such buildings is only five. There is also a large discrepancy in class D where the expected number of buildings according to the EPCs was supposed to be 17 while the actual consumption classified 33 buildings in this class. It is probable that most of the buildings which were supposed to belong to class B actually showed class D performances and therefore created this high discrepancy. A positive sign for both EPCs and actual consumption is that there are no buildings in class G so it can be concluded that none of the schools are of very low energy performances. These discrepancies between EPC ratings and actual performance suggest potential energy performance gaps. Buildings may not be achieving the expected energy efficiency levels as indicated by their EPC ratings.

The analysis of school heating systems revealed that 80% (75 of 94) of schools use a gas central heating system while 20% (19 of 94) use a district heating system (Figure 4).

Since the sample includes also small local schools in the County, the absence of district heating systems could be influenced by factors such as limited access to district heating infrastructure. District heating infrastructure is more commonly found in urban or densely populated areas, where it is economically feasible to establish and sustain. Furthermore, the implementation of district heating infrastructure in rural or sparsely inhabited regions can be excessively costly. Small schools with limited finances may find it more practical to invest in gas central heating systems, despite their lower long-term efficiency and environmental friendliness. Furthermore, in regions characterized by limited and scattered populations, the need for district heating may not reach a level that justifies the allocation of resources towards the development of the necessary infrastructure.

4.2. EPG and CPG Analysis

The EPG shows whether the expected energy consumption from the EPCs is achieved in actual consumption. Thus, it is described as the discrepancy between a building’s measured and calculated energy consumption [32,47]. Notwithstanding the different methods for analyzing the EPG, the following formula is frequently used and also employed in this research to calculate the difference in consumption as a percentage of the EPC consumption for a particular building [36,48]:

$$EPG \left[\%\right]={\displaystyle \frac{Actual consumption-EPC consumption}{EPC consumption}}·100$$

(1)

In this perspective, Figure 5 presents the EPC consumption and actual consumption of thermal energy for heating of all 94 schools, while Figure 6 presents the EPG calculated according to Formula (1).

Furthermore, the CPG was also calculated based on Formula (1), adjusted to fit the carbon emission of buildings. Therefore, CPG is calculated as follows:

$$CPG \left[\%\right]={\displaystyle \frac{Actual carbon emission-EPC carbon emission}{EPC carbon emission}}·100$$

(2)

In this perspective, Figure 7 presents the EPC emission and actual emission of carbon dioxide of all 94 schools, while Figure 8 presents the CPG calculated according to Formula (2).

Moreover, a descriptive statistics analysis was performed on the EPG and CPG (

Table 4. Descriptive statistics analysis of the EPG and CPG.

Variable	N	Mean	Median	Min.	Max.	Lower Quartile	Upper Quartile
EPG [%]	94	71.73	42.11	−99.67	384.69	−14.96	136.40
CPG [%]	94	78.77	43.19	−85.92	906.41	−25.59	116.57

), taking into account the school type (

Table 5. Descriptive statistics analysis of the EPG and CPG by school type.

Variable	School Type	N	Mean	Median	Min.	Max.	Lower Quartile	Upper Quartile
EPG [%]	Primary	81	71.53	42.16	−99.67	384.69	−20.87	136.40
	Secondary	13	72.93	22.56	−2.66	226.95	6.96	110.65
CPG [%]	Primary	81	84.95	50.43	−85.92	906.41	−26.27	118.32
	Secondary	13	40.21	12.14	−42.50	221.32	−18.24	68.47

) and the type of heating (

Table 6. Descriptive statistics analysis of the EPG and CPG by type of heating.

Variable	Heating Type	N	Mean	Median	Min.	Max.	Lower Quartile	Upper Quartile
EPG [%]	District	19	100.65	84.04	−2.66	384.69	13.49	188.98
	Gas	75	64.40	38.99	−99.67	346.10	−23.53	123.78
CPG [%]	District	19	75.02	58.85	−59.02	309.08	−6.40	132.24
	Gas	75	79.71	40.76	−85.92	906.41	−30.90	116.57

).

Based on the provided results, it can be seen that both EPG and CPG exhibit a wide range of positive and negative deviations, indicating variability in how individual data points differ from their expected values. Positive deviations in the EPG results signify those certain buildings consume more energy than initially estimated by their EPC ratings, suggesting that these buildings or systems use more energy in practice than anticipated based on their energy performance assessments. The CPG results follow a similar trend, where positive deviations indicate that actual carbon emissions exceed predicted values. For both EPG and CPG, positive deviations are much larger and more frequent than negative ones. This suggests that most buildings consume more energy and emit more carbon than initially anticipated according to EPC calculations. The larger positive deviations in CPG compared to EPG imply that carbon emissions may be disproportionately higher than energy consumption gaps. Positive deviations highlight areas where energy efficiency and carbon reduction interventions may be necessary to align actual performance with expected levels. Contributing factors could include inefficient equipment, suboptimal building design, poor insulation, occupant behavior, or changes in usage patterns not accounted for in the initial EPC assessments. Overall, the average EPG (71.73%) aligns with literature findings, such as 60–70% for schools and 70–85% for universities [38,52]. This consistency underscores the widespread nature of energy performance gaps across educational buildings and highlights the need for targeted interventions. The average CPG (78.77%) is slightly higher than the EPG, reinforcing the idea that emissions impact may be even more significant than energy inefficiencies alone. When comparing primary and secondary schools, similar mean EPGs (71.53% and 72.93%, respectively) are observed, but primary schools display a wider range and greater variability. A similar trend is seen in CPG, where primary schools have a mean of 84.95% compared to 40.21% for secondary schools, with primary schools again showing greater variability. This could be due to differences in building age, size, maintenance levels, or occupancy patterns, which lead to diverse energy and carbon performance outcomes. Regarding heating types, district heating exhibits a higher mean and median EPG (100.65% and 84.04%) compared to gas heating (64.40% and 38.99%), indicating that district heating systems experience larger energy performance gaps. A similar pattern is observed in CPG, where district heating systems show a mean of 75.02% compared to 79.71% for gas heating, although gas heating exhibits a wider range in CPG. The variability in district heating performance could stem from differences in system efficiency, network losses, and the mix of energy sources used in district heating systems.

In this context, a t-test was performed for heating system types and school types in regard to EPG and CPG (

Table 7. t-test results.

Comparison	Metric	Group 1	Group 2	Mean Diff	p-Value
Heating Systems	EPG	District: 100.65%	Gas: 64.40%	+36.25	0.207
(District vs. Gas)	CPG	District: 75.02%	Gas: 79.71%	−4.69	0.912
School Types	EPG	Primary: 71.53%	Secondary: 72.93%	−1.40	0.967
(Primary vs. Secondary)	CPG	Primary: 84.95%	Secondary: 40.21%	+44.74	0.366

).

The statistical analysis reveals no significant differences in either EPG or CPG across either comparison group. For heating systems, neither the EPG (p = 0.207) nor CPG (p = 0.912) differences between district and gas heating reached significance, despite notable numerical differences. Similarly, school types showed non-significant EPG (p = 0.967) and CPG (p = 0.366) variations. This uniform lack of statistical significance suggests that, within these datasets, neither the heating system type nor school category systematically predicts performance gaps when accounting for sample variability. However, the large effect sizes observed in some comparisons (particularly for school-type CPG differences) may warrant further investigation with larger samples to assess potential practical significance.

Also, a correlation analysis was performed to explore relationships between building age, energy class, heating type, school type, and EPG/CPG (

Table 8. Correlation analysis.

Variable	EPG	CPG
School Type	0.004	−0.094
Building Age	−0.016	−0.047
Energy Class	−0.301	−0.130
Heating Type	−0.131	0.012

).

The correlation analysis reveals that only the energy class shows a statistically significant relationship with EPG (r = −0.301, p < 0.05), indicating that buildings with better energy efficiency ratings tend to have smaller energy performance gaps. All other variables—including school type, building age, and heating type—demonstrate weak correlations (|r| < 0.15) that are not statistically significant with either EPG or CPG. Interestingly, while energy class significantly predicts EPG, it does not show a meaningful relationship with CPG (r = −0.130, p > 0.05), suggesting that different factors may drive carbon performance gaps compared to energy performance gaps.

Additionally, the change in the cost of thermal energy for heating due to the EPG was also calculated. The prices were obtained from official energy distributors for district heating and gas for the location of Osijek (0.0236 €/kWh for district heating and 0.0403 €/kWh for gas) [68,69]. The average annual thermal energy for heating cost is estimated to be 2048.74 € higher due to the EPG. This means buildings are on average spending around 2000 extra € yearly on heating than predicted because of the gap between expected and actual energy use. This highlights a crucial point for building owners and managers. Factoring in the EPG when budgeting for energy costs can help create a more realistic idea of expected expenses. By accounting for this potential difference, facility managers can allocate sufficient funds to cover heating costs and identify areas for potential cost savings through energy efficiency improvements. Also, the analysis showed that 66 of the analyzed schools (70%) exhibited an EPG larger than 30%. When analyzing the year of construction of the schools that showed such EPGs it can be seen in Figure 6 that the majority of buildings; i.e., 45 of the total 66 (68.2%) were built before 1980, prior to the systematic implementation of legislation regarding the thermal insulation of buildings. In Figure 9, the same year spans that were used in Figure 1 are utilized to maintain consistency. This approach ensures that the time periods compared across different figures are uniform, facilitating a clearer and more coherent analysis of the data presented.

This suggests that the EPG is highly more pronounced in older buildings highlighting the importance of retrofitting such buildings since they will be used for many years to come, especially due to their educational purposes and the importance of health, safety, and comfort due to their users. Prioritizing the retrofitting of older educational buildings not only addresses energy inefficiency but also safeguards the well-being and educational experience of their users, aligning with broader sustainability and educational objectives. Additionally, the finding regarding the age of the buildings also highlights the need for improved EPC issuing due to the aging of the building materials. In this context, it is necessary to perform in situ measurements of building elements’ thermal transmittance since the current thermal transmittance is calculated using the standard values for each material as if the building were new, and therefore, it provides inaccurate results. This is significant since it has been demonstrated that even small variations in the U-value, one of the most sensitive metrics used to forecast energy use, cause large variations in the demand for heating [70,71,72]. Additionally, it has been discovered that there might be a 153% discrepancy between theoretical values and U-values observed in situ using various techniques [71,73]. Also, the findings of this research proved in line with several others regarding the fact that buildings with EPCs of higher class showed much larger EPGs than the ones of lower classes [18,71,74]. Namely, buildings that have an EPC of class A and B showed an average rebound EPG of 199.46 and 180.41%, respectively, while buildings with an EPC of class E and F showed an average prebound EPG of 30.53 and 32.28%, respectively. This can also be due to the fact that the manufacturers overestimate new materials in new low-energy buildings, and they actually have worse properties than stated.

4.3. Discussion

The analysis of EPGs and CPGs in the studied school buildings provides key insights into the discrepancies between expected and actual performance. The average EPG of 71% determined in this study aligns with previous research findings for educational buildings, with reported values ranging from 60% to 85%. This consistency underscores the widespread nature of energy inefficiencies in schools and universities and highlights the need for targeted interventions. Similarly, the average CPG of 78.77% suggests that actual carbon emissions exceed predicted values even more significantly than energy consumption, emphasizing the additional impact of emissions factors such as fuel source carbon intensity, heating system efficiency, and operational behaviors. Several factors contribute to the observed performance gaps. Inefficient building design and construction practices, such as inadequate insulation, aging building materials, and inefficient HVAC systems, result in higher-than-anticipated energy use and emissions. Additionally, suboptimal building operation, including poor HVAC scheduling, improper temperature settings, and lack of maintenance, exacerbates inefficiencies. The financial impact of these gaps is also significant. The analysis revealed an average annual difference of approximately 2000 € in heating costs attributable to EPG, highlighting the potential for budgetary strain if inefficiencies are not properly addressed. Since CPG follows a similar trend, additional carbon-related costs, such as penalties or carbon pricing mechanisms, could further increase financial burdens for educational institutions in the future.

The recast Energy Performance of Buildings Directive (EPBD 2024/1275) [75] introduces ambitious measures that align with the need to address these performance gaps. The directive mandates that all new public buildings must meet zero-emission building (ZEB) standards by 2028, ensuring very high energy performance and eliminating on-site fossil fuel emissions. Furthermore, buildings owned or occupied by public bodies should serve as exemplary models of environmental responsibility, with regular energy certification and public display of energy performance certificates, particularly in frequently visited buildings such as schools. Given that many educational buildings are over 40 years old and exhibit significant energy performance gaps, targeted retrofitting strategies aligned with ZEB criteria are crucial. Implementing deep renovations, integrating renewable energy sources, and improving building envelope performance will be key to reducing both EPG and CPG. Moreover, the directive emphasizes the role of efficient district heating and cooling systems as part of the transition to zero-emission buildings. This is particularly relevant for schools, where centralized heating systems often exhibit high and variable CPG values. Enhancing the efficiency and carbon footprint of district heating networks through improved energy management and the integration of renewable energy sources can significantly mitigate emissions. Collaboration with local authorities and energy providers to upgrade district heating infrastructure should be a priority. Another critical aspect of the EPBD is its focus on indoor environmental quality. As educational institutions undergo retrofitting efforts, it is essential to ensure that energy efficiency improvements do not compromise occupant comfort and health. The directive mandates the installation of monitoring and control systems for indoor air quality, which should be incorporated into school renovation strategies to promote a holistic approach to energy performance and well-being.

Additionally, occupancy schedules, user behavior, and internal gains are known drivers of EPG and CPG. However, these aspects are often not explicitly considered in EPC assumptions. School operation patterns, including after-school programs, extended facility use, and holiday schedules, may significantly affect actual energy consumption compared to standardized EPC profiles. For example, schools that offer evening classes, extracurricular activities, or community events may require heating, lighting, and ventilation beyond the assumed 12 h daily operation. Likewise, during holiday periods, some schools remain partially open for administrative work, maintenance, or special programs, leading to variable energy use patterns. Seasonal variations in occupancy can also influence internal gains, as student and staff presence contribute to heat generation, potentially reducing heating demand in winter but increasing cooling loads in warmer months. These variations highlight the need for more dynamic EPC models that account for real-world usage scenarios. A qualitative discussion of these factors is necessary to better understand the discrepancies between predicted and actual energy performance and to improve future energy assessments and retrofitting strategies.

In conclusion, addressing EPGs and CPGs in educational buildings requires a comprehensive strategy aligned with the EPBD’s objectives. Retrofitting efforts should not only improve energy efficiency but also ensure compliance with ZEB criteria, enhance district heating systems, and prioritize indoor environmental quality. By adopting these measures, educational institutions can significantly reduce energy consumption and carbon emissions while ensuring sustainable and comfortable learning environments.

5. Limitations

While this study provides valuable insights into EPGs and CPGs in school buildings, several limitations should be acknowledged. The EPC methodology relies on a set of assumptions that may not fully capture real-world building performance. Notably, cooling energy consumption is not currently considered in EPC calculations for educational buildings, which could lead to an underestimation of total energy use, particularly in warmer months. Additionally, for existing buildings, in situ measurements are not mandatory, meaning key parameters such as air permeability and thermal transmittance (U-values) are determined using standardized values rather than empirical testing. U-values are typically calculated based on tabulated thermal conductivity (λ) values, and the lack of comprehensive data on all building layers means that detailed material properties cannot be verified without destructive sampling methods. This introduces potential inaccuracies, especially for older buildings where material degradation or undocumented retrofits may affect actual performance. Furthermore, behavioral and operational variations among building occupants are not systematically accounted for in EPC assessments, despite their significant influence on energy consumption. The absence of direct monitoring data, such as smart meter readings or continuous energy audits, further limits the ability to precisely quantify discrepancies between predicted and actual performance. Future studies could benefit from incorporating real-time energy monitoring, detailed thermal imaging assessments, and more granular occupancy and usage data to refine the accuracy of performance gap analyses.

6. Conclusions

The analysis of EPGs and CPGs in school buildings highlights both challenges and opportunities for improving energy efficiency and reducing carbon emissions in the education sector. This study has identified significant disparities between expected and actual energy consumption and emissions, emphasizing the need for targeted interventions to bridge these gaps. While EPGs indicate inefficiencies in energy use, CPGs further highlight the role of carbon intensity in performance deviations, underscoring the need for carbon-conscious energy strategies. Addressing both EPGs and CPGs presents a major opportunity for energy conservation, emissions reduction, and cost savings in educational institutions. Bridging these performance gaps requires a multifaceted approach that integrates technological advancements, operational improvements, and behavioral interventions. The findings also underscore the importance of retrofitting older buildings to enhance energy performance and decarbonizing heating systems to reduce emissions. Future research should explore the specific drivers of both EPGs and CPGs in educational buildings and assess the effectiveness of various interventions in reducing these gaps. Longitudinal studies tracking energy consumption and emissions patterns, as well as comprehensive energy audits, can provide deeper insights into the dynamics of these gaps and inform evidence-based strategies for improving both energy and carbon performance. Ultimately, addressing both EPG and CPG is critical for creating more sustainable, cost-effective, and environmentally responsible educational facilities. By implementing targeted interventions identified in this study, educational institutions can significantly reduce their carbon footprint, enhance resource efficiency, and contribute to long-term sustainability goals.