Improving the Energy Performance of Public Buildings Equipped with Individual Gas Boilers Due to Thermal Retroﬁtting

: The article assesses an impact of thermal retroﬁtting on an improvement of the energy quality of public buildings in terms of their heating. The analysis covered a group of 14 buildings, including schools, kindergartens or ofﬁces, while energy audits were carried out for 12 of them. The indications of individual gas meters were the source of actual data for the assessment of changes in energy consumption indexes in operating conditions. The analysis showed a clear improvement in the energy quality of buildings; however, the actual effects were much lower than forecasted. The average forecasted decrease in energy consumption was supposed to be 64.3%, but the measured data showed only 37.1%. The investigation conﬁrmed that the most complex refurbishing provided the most satisfactory decrease in energy consumption (51.4% of real decrease in energy consumption), while objects with partial thermal refurbishing reached an efﬁciency of only 21.8%. It was stated that in operating conditions, special attention should be paid to the manner of energy use, since different indicators of energy consumption can be obtained with the same parameters of building’s balance cover. The results obtained can be further utilized in thermal-refurbishment implementation procedures. Follow-up investigations on the impact of selected parameters on energy consumption are planned.


Introduction
The construction sector is an important field of economy, responsible for approximately 10% of Gross Domestic Product in European Union member states. Due to the high energy consumption, this sector is responsible for over 36% of the world's carbon dioxide emissions, which signals a high reduction potential [1,2]. A major part of the environmental costs related to buildings is connected to their operational stage [3]. With the increased attention paid to greenhouse gases, many reduction policies were addressed to this branch, including a thermal-refurbishment actions support.
Financial support programs implemented in Poland since 1998 [4] included and still include public buildings owned by local government units, which are used to carry out the tasks of these units [5,6]. This applies to buildings regardless of their location, i.e., in urban and rural areas. First of all, local government units indicated educational Poland, maximal allowable values of the primary energy index were successively decreased by the Regulation of Polish Ministry of Transport, Construction and Maritime Economy [23]. In case of the newly erected non-healthcare public buildings PE H+W (heating and hot water production) limits changed from 65 kWh·m −2 ·year −1 in 2014 to 45  Recent studies and publications show that there are differences between the calculated and actual values obtained in operating conditions [24,25], which is called the performance gap [26]. In energy audits prepared for the purpose of carrying out an investment consisting in comprehensive thermal retrofitting of a building, it is necessary to indicate the scope of works and technologies for their implementation, determine the projected level of energy savings for heating, estimate the amount of investment costs and financial operating effects, and determine environmental benefits [2,3,27]. Building owners and managers should be aware that at the stage of auditing a building for thermal retrofitting purposes, forecasted values are obtained, and the actual ones depend on many factors, e.g., indoor temperature, ventilation intensity, method of use, and energy management in building before and after thermal retrofitting [28]. The energy quality of buildings after thermal refurbishment is assessed primarily by comparing the heat transfer coefficients of building partitions to the limit values imposed by legal acts and the applied technical solutions of the heating system, which must at least meet the standards in terms of efficiency of heat generating devices, thermal insulation of pipelines and fittings, automatic central and local regulation influencing the most effective heat collection in heated rooms. Therefore, after thermal refurbishment, transparent and solid partitions should be characterized by low values of the heat transfer coefficient, and the heating system should be characterized by the highest overall efficiency [29]. However, unrealistic theoretical efficiencies of heating systems and insulation materials properties are the factors causing the performance gap, resulting in lower energy savings than expected [30,31].
Buildings located in cities are most often supplied from centralized municipal heating systems through heating transmission networks. On the other hand, buildings located in rural areas are most often equipped with local heat sources, equipped with boilers fired with a given type of solid or liquid fuel. In Poland, in rural areas, where there is an access to natural gas supplied via networks, individual gas boiler houses are built to produce heat for the needs of a given building. The decrease in energy demand for heating as a result of thermal retrofitting translates into a decrease in gas consumption; therefore, the operating costs decrease, and the additional effect is a reduction in carbon dioxide emissions to the atmosphere. In fact, except for the effects related to energy savings, which translate into ecological ones, thermal retrofitting also includes economic aspects (which is involved in energy audits). Other areas of thermal retrofitting are also social and technological aspects including Planning, Quality and Administration, which is discussed in the article [32], constituting the Key Performance Indicators (KPIs). According to the literature, energy savings are one of the most important and widely used KPIs [33,34]; therefore, this indicator was the basis of analysis in this study.
The article presents a long-term evaluation of thermal retrofitting effects based on an extensive and complex dataset including monitoring data, which brings an important contribution to the knowledge on the thermal retrofitting of buildings in Poland. A literature review finds a dearth of empirical studies that illustrate the correlation between the performed retrofitting and the predicted and measured gas consumption, in particular in the case of performance gap observed in buildings with poor energy standard and reduced conditions of thermal comfort. Although similar studies were previously published, they usually referred to other locations and involved simulation data on the energy performance before and after refurbishment [16,19], included shorter observation periods and limited number of buildings [35,36] or discussed other types of objects [37].
The results of this study aimed at providing the comprehensive set of data on the thermal retrofitting of public buildings, including long term monitoring data gathered for 14 objects with common features. Furthermore, on the basis of the breakdown of the analysis of the energy reduction performance, some critical remarks and detailed explanations were stated in view of planning a thermal retrofitting investment.

Materials and Methods
In the following section, assumptions for the calculation of the analyzed indexes and the objects of study are presented.

Characteristics of the Analyzed Buildings
The analysis covered 14 public buildings, including 10 primary schools, 1 kindergarten and 3 office buildings. All buildings are owned by local government units and are located in Eastern Poland in rural areas. With the assumed indoor air temperature of 20 • C in heated rooms, four of them are assigned to a weather station with the number of degreedays 3963.4 (days·K)·year −1 (marked with M1 group, and buildings with numbers from 1 to 4), while ten are assigned to stations with the number of 3825.2 (days·K)·year −1 (marked by the M2 group and buildings by numbers from 1 to 10). In eight cases, boiler rooms are single-function and cover the thermal needs of buildings only for heating purposes. In the remaining six cases, boiler rooms are dual-function and cover the needs for heating and hot water.
Data characterizing buildings from group M1 and group M2 are presented in Table 1, Table 2, Table 3, Table 4, respectively.

Evaluation of Heat Consumption Indexes
Gas consumption was measured using certified gas meters belonging to the natural gas supplier common to all investigated objects. In three cases, the meters measure only gas consumption for heating purposes, in five cases for heating purposes and for cooking food (gas cookers) in total and in six cases for heating and hot water in total. The individual gas meters provided monthly indications of gas consumption in the heating period (winter) and two-month indications outside it (summer). The analysis of gas consumption measurements in winter and summer allowed for the estimation of gas consumption for individual purposes.
In case of the objects with gas consumption for many purposes, total annual gas consumption for heating was estimated in the following way:

•
In the case of school buildings that had a common gas consumption measurement for heating and the gas cookers, consumption by cookers was subtracted from the total annual consumption readouts in the following way: readouts from the months outside the heating season (May-September) were subtracted in full, and for the months of October-April, the average monthly gas consumption for cookers for three months was calculated (May, June, September, when the schools were still running) and subtracted for each month.

•
In the case of school buildings with common gas consumption for heating and hot water preparation, a similar method was applied as for the gas cookers.

•
In case of the office building, average consumption for all months outside the heating season was evaluated, because they perform similarly during summer period as during the rest of the year. • There was no circumstance within the investigated objects where gas consumption measurements would cover three purposes (heating, hot water preparation, gas cookers).
It should be mentioned that in the case of common measurements of the gas consumption for heating and gas cooker operations, the share of gas consumption by gas cookers was in the range of 1.5% to 8.0% of the annual gas consumption in the building. However, in the case of the joint measurement of consumption for heating and hot water purposes, the share of consumption for hot water preparation ranged from 2% to 8%, and in one case equaled approx. 17%.
The annual gas consumption for heating readouts came from the years 2001-2010 (the period before and after thermal retrofitting) and needed to be adjusted to so-called standard conditions that would cover differences in the harshness of the winter season in the particular year. To that aim, ϕ 1 and ϕ 2 correction factors were used and were calculated using Equations (1) and (2).
For the buildings belonging to the M1 group assigned to a meteorological station with the number of the degree-days SD 1 = 3963.4 (days·K)·year −1 : For the buildings belonging to the M2 group assigned to a meteorological station with the number of the degree-days SD 1 = 3825.2 (days·K)·year −1 : where SD 01 and SD 02 are values calculated for a particular year on the basis of the monthly average outdoor air temperatures (θ em ) measured in an appropriate meteorological station, L D is the number of heating days in particular month and indoor air temperature θ i = 20 • C. The difference between SD 01 and SD 02 results from the fact that in standard conditions the stations have different average monthly temperatures θ em , while the number of heating days is the same, which equals 222 days for the whole period (Table 5). ϕ 1 and ϕ 2 correction factors enable us to adjust energy consumption to the standard year. SD 1 and SD 2 are the standard values from the long-term measurements. SD 01 and SD 02 values represent the number of degree-days in a given year, and they were calculated using the following equation: With the data obtained from the meteorological stations, values of the ϕ 1 and ϕ 2 correction factors for the following years were as follows ( Table 6): Table 5. Climatic data for the analyzed meteorological stations [38,39]. In the case of the M1 group of the buildings, adjusted (corrected) gas consumption equals V 01 in case of M2 group V 02 and can be calculated using the following equations:

Group of Buildings
where V P1 and V P2 are the measured values of gas consumption in a selected building in a particular year (m 3 ·year −1 ) decreased for the consumption by gas cookers or hot water preparation.
In order to evaluate the adjusted volume of the gas consumption (m 3 ·year −1 ) into heat consumption (GJ·year −1 ), the average calorific value of medium given by gas supplier was assumed: It could then be assumed that Final Energy Consumption for heating purposes Q FH1 (for M1 group of buildings) and Q FH2 (for M2 group of buildings) expressed in GJ·year −1 can be evaluated using the following: or Knowing the value of useable heating area of particular building, Annual Final Energy Consumption index for Heating (FE H ) was evaluated according to the following equation: or where: w H -primary energy input factor, according to Polish regulations [20,21] assumed 1.1 for all objects (heating system powered by gas boiler located in the investigated building).
For the aims of building efficiency verification, the boundary value of the Annual Primary Energy Consumption Index for Heating (PE H,0 ) was calculated according to Polish regulations from the period when thermal refurbishment was conducted, according to the following equation [20]:

Energy Audits
Together with gas consumption readouts, energy audits were carried out according to the methodology presented in the following Polish and European documents regulation [5,[40][41][42], which was up-to-date during the facilities refurbishment. Conducted audits enabled additional evaluation of both final energy consumption (Q FH ) and annual primary energy consumption index for heating (PE H ).

Statistical Analysis
The analyzed material was tested for differences between energy consumption (expressed by Q FH , FE H and PE H indexes) before and after thermal refurbishment. Basic descriptive statistics such as minimum, maximum, mean, median and standard deviation were estimated. Boxplots illustrating the distributions of the replacement were also used. The significance of the differences was verified using the Wilcoxon test for dependent variables [43], which is a non-parametric equivalent of the Student's t-test. The choice of the test was dictated, firstly, by the relationship between the measurements before and after thermal refurbishment and, secondly, by the lack of meeting the assumptions about the normality of the distribution of the studied features in the groups and the lack of homogeneity of variance. In such a situation, it was reasonable to use this test [44].
The evaluation of energy consumption was carried out by comparing both absolute and relative values. Relative measures were determined as the percentage of energy yield caused by thermal refurbishment according to the following formula: where x before , x after are the readings, respectively, before and after thermal retrofitting.
The actual values of energy consumption were also compared with the levels determined on the basis of the audit of buildings and indications obtained from technical conditions contained in legal acts in force. This comparison is shown in bar charts.

Results
Using the measurement readouts, the described methodology and statistical research, the calculation results presented below were obtained. First, in each building, on the basis of the corrected averaged values of heat consumption before and after thermal-refurbishment (Q FH ), the percentage decrease in final energy consumption for heating was calculated. The trends are shown in Figure 1, while the average values before and after and the percentage decrease (savings) in Table 7. In the M1 group, an average decrease in final energy consumption was achieved at the level of 35.2%, while in the M2 group it was 42.7%. Then, with the usable area heated in each building known, the index of the annual final energy consumption for heating was determined, expressed in kWh·m −2 ·year −1 and denoted as FE H . The decrease in FE H index value is presented in Figure 2, while its averaged values before and after thermal refurbishment are presented in Table 8.   The average value of the FE H index obtained after thermal-refurbishing in the M1 group was 95.5 kWh·m −2 ·year −1 , while in the M2 group it was 107.9 kWh·m −2 ·year −1 . With the assumed value of the non-renewable primary energy input index of w H = 1.1 (energy production in the building, energy carrier in the form of natural gas), the consumption rate of non-renewable primary energy was calculated in each building. It was expressed in kWh·m −2 ·year −1 and denoted as PE H . The values obtained in the condition before and after thermal refurbishment in each building are presented in Table 9. The average, minimum, maximum and median values of the PE H index divided into M1 and M2 groups before and after thermal-refurbishment are presented in Table 10. the obtained results of improving the energy quality of buildings in operational conditions to the requirements contained in legal acts in force during the period of thermal-refurbishment works. Table 11 provides a summary of the obtained calculation results, data contained in energy audits of buildings and the required PE H values resulting from legal acts.     On the other hand, Figure 5 compares the actual results obtained with the calculation values contained in the audits after thermal refurbishment and with the required values resulting from the Polish technical regulations.

Discussion
The research included in this publication aimed at: The analysis of data on individual buildings shows that 12 buildings were covered by comprehensive measures aimed at improving the efficiency of the heating system and improving the insulation of building partitions. Out of the group of 14 buildings, only in two, marked as M1-2 and M2-3, did the heating system not require refurbishment. Before the start of the thermal refurbishment, the respective building partitions differed in the heat transfer coefficients (U expressed in W·m −2 ·K −1 ), while in the final state, they were similar to each other. The calculations and analyses carried out show that, as a result of thermal retrofitting in each building, a decrease in final energy consumption in operating conditions was achieved, ranging from 21.8% to 60.5%. The lowest value in this range applies to building M1-2, in which only building partitions were thermo-modernized, while the largest one concerns building M2-10, which among all the buildings had the highest U coefficients of partitions before the retrofitting. The obtained average value of final energy savings for heating in the M1 building group was 35.0%, while it equaled 42.7% in the M2 group. The comparison of final energy consumption before and after thermal refurbishment showed significant differences (p = 0.002) for buildings from the M2 group. This is confirmed by a noticeable decrease in final energy consumption resulting from the performed investments.
The most favorable configurations of the thermal refurbishing are the most complex ones. Additionally, the most satisfactory values of the energy consumption decrease are achieved for the objects that were characterized by the lowest energy quality before refurbishing. As an example, data contained in the Tables 1, 2 and 7 can be given. The M1-2 object before refurbishing previously had an insulated roof, which is why after refurbishing a lower level of savings was achieved (only 21.8% of energy consumption decrease). On the other hand, the M1-3 object had a thermally poor external envelope (high U values of the walls and the roof). After their thermal properties were aligned to the other objects (by thermal refurbishment), the level of the energy savings increased (51.4% decrease in energy consumption). The level of energy savings in the case of the M2-2 object was higher (47.5%) because of the non-heated flat roof compared to the M2-3 and M2-4 buildings, with thermally insulated flat-roofs before refurbishing and comparable values of the overall heat transfer coefficients.
Although in the M1 group the significance of the differences was not statistically significant (p = 0.12), which most likely results from the fact that a small sample size was used for testing, in all analyzed cases, there was a decrease in energy consumption. The estimated values of energy savings for heating obtained under operating conditions are in each case lower than the forecast calculated in the energy audits. The decrease in final energy demand calculated in audits based on commonly used European standards and in conjunction with the methodology in force in Poland was much greater (Figure 3) and ranged from 37.6% to 77.1%. It should be noted that these values are forecast values and are calculated assuming the standard boundary conditions of the internal environment in rooms and the external environment surrounding the building's balance zone before and after thermal retrofitting. This is one of the basic assumptions of the methodology of preparing an energy audit of a building. The results obtained in this study confirm the general trend observed in thermal restoration of buildings. According to the literature concerning multifamily buildings in Poland, the actual energy savings range between 8.8% and 74.8% of energy savings calculated in audits, depending on the various renovations [35]. In another paper, the economic effects resulting from thermal refurbishment of schools were compared after several operation seasons and were calculated (59-71%), and real savings (33%) differed significantly [37].
To try to explain this, the PE H index obtained on the basis of measurements in operating conditions with the same index calculated in the building energy audit was compared in 12 buildings (in two the audit results were not available), in the condition before thermal refurbishment (Figure 4). The comparison of the values of the PE H index shows that in eleven cases examined in this respect, they were higher in the audit than in the operating conditions. These differences ranged from 9.4% to 47.7%, and the average for both groups was 26.1%. It turned out that in the examined group of buildings, only in the case of M2-3 were the theoretical values of the indicator describing the energy performance of the building lower than the results obtained on the basis of measurements. In the remaining cases, these results were much higher, as presented in Figure 4. It should be assumed that before the thermal retrofitting, the buildings were heavily underheated, and the standard indoor air temperature was not reached in the rooms. This means that the thermal needs of the building were not fully covered in the conditions before thermal refurbishment, so the level of final energy savings achieved in operating conditions was lower than the forecast calculated in the audit. The final effects of energy savings are influenced by user behavior, as evidenced by literature reports [51], especially the use of control fittings of radiators (thermostatic valves) and programming of internal temperature. However, this influence of individual behavior is greater in residential buildings than in public buildings, where one person is responsible for energy management.
The analysis of the FE H index values shows that its decrease in each building was significant. The mean value of the index in the M1 group of buildings decreased from 152.1 kWh·m −2 ·year −1 to 95.5 kWh·m −2 ·year −1 , which means a decrease by 37.2%, while that in the M2 group decreased from 190.2 kWh·m −2 ·year −1 to 107.9 kWh·m −2 ·year −1 , which is a decrease of 43.3%. The same percentage reductions were obtained for the PE H index, because in each building, the input coefficient w H was equal to 1.1, which in this case means a linear relationship between the two indexes. Additionally, since FE H and PE H values are derivatives of the Q FH index as its linear combinations, the comparisons of energy consumption decreases expressed by the above-mentioned indexes are also characterized by the same level of significance of differences. The study comparing the PE H value obtained in real conditions with the one calculated in the audit after thermal refurbishment shows that the relations between them developed in different ways. In three buildings, the PE H value resulting from gas consumption measurements was lower than the one specified in the audit, from 6.1% to 11.4%; in the remaining nine buildings, it was greater than 6.8% up to 86.8%, and on average, for both groups, it is a 28.4% higher value. A comparison of the PE H index values obtained on the basis of measurements with the values required in legal acts as for reconstructed buildings shows that after thermal retrofitting in nine buildings the requirement was met with an excess of 0.4% to 29.4%, while in four buildings, 1.4% to 19.1% was missing. However, on average for both groups in total, the value was 6.4% lower. There may be several reasons for this, e.g., inadequate adjustment of the central heating system, maintaining higher temperatures in the rooms than required, not using the heating weakening outside the building's working hours or worse than assumed thermal parameters of the building partitions and lower overall efficiency of heating systems. Considering the impact of thermal refurbishment on the improvement of the energy performance of a building, it should be clearly stated that it is diversified but always leads to a reduction in energy consumption and thus its carriers. Figure 6 aims to better illustrate the relationship between the required PE H,0 values depending on the building shape factor (A/V), and the values obtained on the basis of measurements. The continuous red line indicates the required values of the index at a given value of the A/V ratio, while the points on the graph show the average values of the PE H index obtained after the thermal refurbishment of the building. Additionally, the relation between the building shape ratio and Annual Primary Energy Consumption Index for Heating PE H was checked. From the diagram, it can be noticed that after the refurbishment, the values of PE H index were in most cases lower than required. It was noticed that in the case of low A/V ratio values, the influence of building shape was not significant for PE H . Hence, it can be interpreted that analyzed buildings functioned differently, and some of them could be underheated before the retrofitting. In the case of the high value of A/V ratio (building M2-5), energy consumption expressed as Annual Primary Energy Consumption Index for Heating PE H was higher than in other cases and exceeded the required PE H,0 index value, which may result from both building shape ratio and the manner of operation. These relations are similar to those previously described in the literature [7].

Conclusions
According to the performed analysis, the following conclusions were stated:

•
After thermal retrofitting, a decrease in final energy consumption in operating conditions was achieved, ranging from 21.8% to 60.5%. The obtained average value of final energy savings for heating in the M1 building group was 35%, while it equaled 42.7% in the M2 group.

•
The estimated values of energy savings for heating obtained under operating conditions are in each case lower than the forecast calculated in the energy audits. The decrease in final energy demand calculated in audits was much greater and ranged from 37.6% to 77.1%. • Despite the planned and implemented similar scope of works aimed at reducing energy consumption in buildings, both the forecasted (resulting from the audit) and the actual (resulting from the measurements) PE H indexes are highly diversified. This is confirmed in the case of forecasting values ranging from 63.6 kWh·m −2 ·year −1 to 148.6 kWh·m −2 ·year −1 , while in real conditions, the values ranged from 72.4 kWh·m −2 ·year −1 to 177.1 kWh·m −2 ·year −1 . • A comparison of the PE H index values obtained on the basis of measurements with the values required in legal acts for reconstructed buildings shows that after thermal retrofitting in nine buildings, the requirement was met with an excess of 0.4% to 29.4%, while in four buildings, 1.4% to 19.1% is missing. However, on average, for both groups in total, the value is 6.4% lower. There may be several reasons for this, e.g., inadequate adjustment of the central heating system, maintaining higher temperatures in the rooms than required, not using the heating weakening outside the building's working hours or worse thermal parameters than assumed of the building partitions and lower overall efficiency of heating systems.

•
Although the decrease in final energy consumption in operating conditions was achieved in all the examined cases, investors cannot expect theoretically calculated savings in the real conditions. At the same time, thermal refurbishment will certainly ensure the additional effects of improving the thermal comfort in rooms and achieving the required internal conditions, which is very important from the point of view of the users.