**Creating Statistics for China's Building Energy Consumption Using an Adapted Energy Balance Sheet**

#### **Mingshun Zhang 1,\*, Xuan Ge <sup>1</sup> , Ya Zhao <sup>2</sup> and Chun Xia-Bauer <sup>3</sup>**


Received: 2 October 2019; Accepted: 8 November 2019; Published: 11 November 2019

**Abstract:** China's regular energy statistics does not include the building sector, and data on building energy demand is included in other types of energy consumption in the Energy Balance Sheet (EBS). Therefore data on building energy demand is not collected based on statistics, but rather calculated or estimated by various approaches in China. This study aims at developing and testing China's building energy statistics by applying an adapted EBS. The advantage of the adapted EBS is that statistical data is from the regular statistical system and no additional statistical efforts are needed. The research result shows that the adapted EBS can be included in China regular energy statistical system and can be standardized in a transparent way. Testing of the adapted EBS shows that China's building energy demand has shown an annual increase of 7.6% since 2001, and a lower contribution to the total energy demand as compared to the developed world. There is also a close link to lifestyle and living standard while industrial energy demand is mainly driven by economy and decoupling of building energy demand with increasing of building floor area, this is due to a considerable improvement of building energy efficiency. The adapted EBS creates a method for China conducting statistics of building energy consumption at the sector level in a uniform way and serves as the basis for any sound building energy efficiency policy decisions.

**Keywords:** building energy statistics; building energy consumption; energy balance sheet; building energy efficiency; China

#### **1. Introduction**

China's energy consumption has increased dramatically since 1980. Around 2011 China became the largest energy-consuming country, replacing the USA. In 2015, China had a 28% share of the global end-energy demand while USA had a 22% share [1]. Building, transportation and industry are the three key energy demand sectors worldwide. The building sector is responsible for more than 25% of China's total primary energy consumption and this figure will increase to 35% by 2030. The GHG emissions contributed by the building sector are about 25% of China's total emissions [2]. Internationally, buildings consume about 30–45% of the global energy demand [3]. Although China's current building energy consumption share is significantly lower than the international level, the fast urbanization, the fast development of the building sector, rising of living standards and increasing consumption will increase the energy demand in the building sector. China is now facing the challenges of both a fast growing building energy demand and low building energy efficiency. It is estimated

that more than 30%–50% of the existing building energy consumption could be saved [4] by adopting various energy efficiency solutions.

Statistics on building energy consumption at the sector level are essential to assess the building energy situation and are indeed used as the basis for any sound building energy efficiency policy decisions. Authorities, building developers, building owners and the public need information about where the building energy demand is and the impact of policy enforcements. The building sector needs information about progress achieved or reasons that prevent progress. The key to meet the information needs mentioned above is to provide complete, timely and reliable data on building energy demand.

In China's national energy statistical system, the final consumption is composed of seven sectors: (1) farming, forestry, animal husbandry, fishery & water conservancy, (2) industrial, (3) construction, (4) transport, storage, postal and telecommunications services, (5) wholesale, retail and catering services, (6) others, and (7) residential consumption. The energy consumption of buildings is mainly included in the consumption sector and also included in other sectors. There are various studies on statistics and monitoring of energy consumption at a single building level, however, there is no specific sectoral building energy statistical system in China and data on building energy consumption at the sector level is calculated or estimated by various approaches [5]. There are many office buildings in industrial sectors, and those buildings' energy consumption is included in the energy demand statistics of the industrial sectors. There are similar cases in other sectors of transportation and construction as well. The lack of reliable and accurate data on building energy consumption at a sector level has been a major barrier for policy making at the national and sectoral levels. The establishment of a national statistical system of building energy consumption in China will create a method for determining the statistics of building energy consumption at the sector level in a uniform way and serve as the basis for any sound building energy efficiency policy decisions.

There are many studies aiming at getting data of building energy demand at the sectoral level. The Building Energy Research Centre of Tsinghua University (THUBERC) has developed the China Building Energy Model, CBEM), based on building energy intensity and building floor area, to estimate China's total building energy consumption [6]. By applying this model, THUBERC publishes an annual report on China's building energy efficiency, which has been one of the key national sources for different stakeholders to get energy consumption data for the building sector. Wang calculated the total building energy demand based on international EBS, surveys and expert workshops and concluded that China's building energy demand was 370 million tons-coal-equivalent (tce) in 2006 [7], which accounts for 21.7% of the total energy demand in China. Long [8] developed a model, based on an analysis of China's energy consumption by sectors and by comparing its industrial structures with USA and Japan, to estimate building energy consumption. According to this model, China's total energy consumption in the building sector was 330 million tce in 2003, which accounts for 20% of the total energy demand of China. The Ministry of Housing and Urban & Rural development of China estimated the existing building energy consumption based on building stocks, climate zone characters and relevant energy efficiency standards, and concluded that China building energy consumption accounts for 27.5% of China total energy demand. From the literature reviews, there are basically four methods to get information on building energy demand at a sector level:


The methods mentioned above have the disadvantages of needing the additional data collection efforts or surveys in addition to regular energy statistics, due to limited samples, using assumptions in modelling, and making estimations by historical experiences. None of the methods mentioned above are standardized and thus it is difficult for China to build up its regional comparisons and benchmarking of building energy performance and compare it to international data.

There have been several studies on China building energy consumption statistics using adapted EBS and those studies are all on a local level [18]. The main disadvantage of these local studies is that the data from EBS is directly used and lacks necessary corrections. This study, built on those local-level studies, focuses on national level and has made necessary amendments to the EBS data. The justified amendments to the data ensure that the study results are closer to the real energy consumption of buildings in China.

EBS is the key resource for all sectors to get energy information. However, building energy consumption is not listed separately in China's EBS and it is divided among the energy consumptions of different sectors. Therefore the first step of this research is to identify and analyze the data sources and statistical definitions of energy consumption of all sectors included in China's EBS. China's EBS has defined the following seven sectors of final energy consumption. This research builds on the following analysis of the final energy consumption of the seven sectors included in China's EBS:

(S1) Farming, forestry, animal husbandry, fishery & water conservancy. This sector is the primary industry and energy consumption of this sector included in EBS covers all production and production related services energy consumption. Thus final energy consumption of this sector does not include building energy consumption. Energy consumption of transportation of this sector is included in EBS.

(S2) Industry. Industries in China's EBS include mining and quarrying, machinery, and power, heating, fuel gas and water production and services. The industrial energy consumption in China's EBS is comparable to but different from the International Energy Agency (IEA) data. There are mainly three differences between China's and IEA's industrial energy consumption: (1) China's industrial energy consumption includes energy consumption of energy industry itself. In IEA energy statistics, energy consumption of energy production is not included in industrial energy consumption but rather listed in the energy loss of energy processing. (2) China's industrial energy consumption includes the transportation energy consumption of this sector. (3) Industrial building energy consumption is included in industrial energy consumption in China's EBS.

(S3) Construction. The energy consumption of construction in China's EBS is mainly the energy consumption of construction process. In IEA's statistics, construction energy consumption is included in industrial energy consumption. China's EBS has separately listed construction energy consumption. Similar to the industrial energy consumption of China, construction energy consumption in China's EBS also includes building energy consumption of construction sector, e.g., office buildings in this sector.

(S4) Transport, storage, postal and telecommunications services. This sector's energy consumption is mainly from transportation enterprises, and it does not include the transportation energy consumption of other sectors as well as citizens' transportation energy consumption. Energy consumption of this sector covers building energy consumption of this sector, e.g., the building energy consumptions of airports, railway stations, bus stations and post office buildings.

(S5) Wholesale, retail trade and catering services. These sectors are mainly tertiary industry and main energy consumption of these sectors are building energy consumption. These sectors' energy consumption is comparable to the IEA's statistics of energy consumption of commercial and public services.

(S6) Others. Energy consumption of "others" in China's EBS is defined as the energy consumption of the tertiary industrial energy consumption excluding the abovementioned sectors (S4) and (S5). "Others" includes software and information technology services, financing, real estate industry, education, science, culture and public health and public administrations. Energy consumption of these sectors are mainly from building performance.

(S7) Residential consumption. Energy demand in residential consumption includes the living energy consumption of citizens and is mainly from building performance. It also includes the transportation energy consumption of citizens.

Another issue is how to amend the data of energy consumption of central heating systems from EBS. The existing data of energy consumption of central heating systems from EBS is significantly low and it is necessary to correct this data. As an example [19], the energy consumption of central heating in the sectors of (S5) Wholesale, retail trade and catering services, (S6) Others and (S7) Residential consumption was 31.11 million tce in 2011 and the floor area of the central heating is 5.18 billion m2. Thus the energy efficiency of central heating is 6 kg coal-equivalent per M2, which is incorrect given to the fact that energy efficiency of the most efficient central heating by a large-scale cogeneration project is 9 kg coal-equivalent per m2, and the national average energy efficiency of central heating by regular boilers is 20 kg coal-equivalent per mM2 [19]. There are three reasons that data of energy consumption of central heating system from EBS is significantly low:

(1) Heat metering is not well installed in China, in particular there is almost no heat metering in the buildings built before 2010.

(2) China's EBS statistics targets enterprises that are on the scale of 20 million Chinese Yuan turnover or energy consumption of above 10,000 tce. There are many SMEs' heat-generators or heat-suppliers (small and medium enterprises) that do not meet this scale and are therefore excluded from the EBS statistics. Therefore heating generated by those SMEs are not included in the EBS.

(3) Heating energy consumption of cogeneration systems is excluded from the central heating energy consumption, but included in energy consumption of power generation (energy industry energy consumption).

For correcting the data of energy consumption of central heating system from EBS, we use the data of central heating from national and local statistics yearbook. In China, statistics yearbooks include detailed data on central heating and this data covers all heat-generators or heat suppliers regardless of their scale.

Since EBS is recognized as a reliable, timely and complete data source of both energy supply and demand, our focus is to develop and test a method of adapted EBS for providing reliable data of China's energy consumption at sector level. Specifically this paper makes the following contributions to develop China's energy consumption statistics by the adapted EBS:


#### **2. Methods**

*2.1. Procedures and Methods of This Research*

The following Figure 1 presents the procedures and methods applied in this research.

**Figure 1.** Procedures and methods of this research.

#### *2.2. Data Collection and Analysis*

Based on the analysis of the existing EBS of China, we develop the following Formula (1) for statistics of complete building energy consumption in China:

$$\mathbf{E\_c = E\_b - E\_t + E\_h + E\_o} \tag{1}$$

where Ec: complete building energy consumption, Eb: total energy consumption in the sectors of (S5) wholesale, retail trade and catering services, (S6) others and (S7) residential consumption, which are available from the existing EBS, Et: Transportation energy consumption of (S5) wholesale, retail trade and catering services, (S6) others and (S7) residential consumption, Eh: Energy consumption of corrected central heating and Eo: Building energy consumption from the sectors of (S2) industry, (S3) construction and (S4) transport, storage, postal and telecommunications services.

To gather data on Et, we conducted a survey aiming at getting information on the consumption of different types of energy among the sectors of (S5) wholesale, retail trade and catering services, (S6) others and (S7) residential consumption, which are available from the existing EBS. The survey conducted by this study took place between October–December 2018 and it included three phases: 1) collecting data of different types of energy consumption of the sectors of S5, S6 and S7 from national

EBS and local EBS of the four megacities of Beijing, Shanghai, Tianjin and Chongqing and seven provinces of Liaoning, Hebei, Shandong, Guangdong, Henan, Sichuan, Ningxia and Gansu. These four megacities and seven provinces are recommended by the China Association of Building Energy Efficiency, given to the facts that: (1) these 11 cities and provinces have quite good energy consumption databases and good EBS, (2) they are good representatives of China's climate zones and (3) they were willing to cooperate with this research and to answer the questionnaire; 2) sending 11 questionnaires to the energy administrations (local Development and Reform Commission) of the four megacities and abovementioned provinces. All 11 questionnaires were answered and have been sent back for this study. The questionnaires aimed at getting information on allocation of energy consumption among different types of energy in the sectors S5, S6 and S7; 3) analyzing the data and providing survey results. The survey results shows that 95% of gasoline and 35% of diesel are used by transportation of the sectors of (S5) wholesale, retail trade and catering services and (S6) others. Almost 100% of gasoline and 95% of diesel consumption are made by transportation in the sector of (S7) residential consumption. This survey results are in line with the study conducted by Wang [7]. Therefore, in this research, we calculated the Et by using Equation (2):

Et = (95% gasoline consumption + 35% diesel consumption) of Sector (S5) wholesale, retail trade and catering services and (S6) others + (100% gasoline consumption + 95% diesel consumption) of Sector (S7) residential consumption

In Equation (2), data on gasoline and diesel consumption in all sectors is available from the existing EBS in China. Eh is calculated by the following Equation (3):

Eh = total energy consumption of central heating − energy consumption of central heating in sector (S5) wholesale, retail trade and catering services, (S6) others and (S7) residential consumption (3)

The total energy consumption of central heating in Equation (3) is available in the China Statistical Yearbook. Energy consumption of central heating of the sectors (S5) wholesale, retail trade and catering services, (S6) others and (S7) residential consumption is available from EBS.

Eo is calculated by the following Equation (4):

$$\mathbf{E\_{o}} = \mathbf{E\_{b0}} + \mathbf{E\_{bi}} \tag{4}$$

(2)

Ebt is the energy consumption of buildings in the Sector (S4) transport, storage, postal and telecommunications services. An assumption of this research is that coal consumption is only for building energy performance, given to the fact that coal is not used as power fuel in transportation in China. Therefore, Ebt is the sum of coal consumption and electricity consumption of buildings in Sector (S4) transport, storage, postal and telecommunications.

Ebi is the energy consumption of buildings of the sectors (S2) industry and (S3) construction. Those are mainly office buildings and buildings used for production. Energy consumption of those buildings has a limited contribution to the total energy consumption of buildings. In this research, we organized an expert workshop aiming to estimate Ebi. Sixteen experts participated in this workshop. Twelve experts were energy managers from energy-related sectors. Two experts were from Sectors (S4) transport, storage, postal and telecommunications services and the remaining two experts were from universities. Participants were selected based on criteria like: 1) at least seven years of working experience in statistics or estimation of building energy consumption; 2) good knowledge of building energy efficiency; 3) members of the Expert Committee of China Association of Building Energy Efficiency. The workshop, based on the fact that total floor area of buildings of the sectors of (S2) industry and (S3) construction is about same as the floor area of buildings of Sector (S4) transport, storage, postal and telecommunications services, concluded that Ebi is comparable to Ebt.

#### **3. Results**

#### *3.1. China Building Energy Consumption 2001–2015*

Figure 2 shows that China's total building energy consumption increased from 310 million tce in 2001 to 860 million tce in 2015, with an average annual increase of 7.6%, which is in line with China's total energy consumption that increased from 1560 million tce to 4300 million tce with an average annual increase of 7.5%. Table 1 shows that the increase of both energy demand and building energy demand is different in the three five-year periods of 2001–2005, 2006–2010 and 2011–2015. During the 2001–2005 period (known as the China National 10th Five-Year Plan), building energy consumption was increasing by 11.9%, while the total energy consumption was increasing by 13.9%. However, the increase of building energy consumption slowed down to 5.3% in the National 11th Five-Year Plan of 2006–2010 and to 5.5% in the National 12th Five Year Plan of 2011–2015. The main reason is that China's national and local governments launched various building energy efficiency initiatives in 2005–2006, and their key initiatives were energy retrofits for existing buildings, promoting green building & low energy building development and compulsorily energy efficiency improvement programme for large-scale public buildings.

**Figure 2.** China's Building Energy Consumption 2001–2015. (Source: 1) Data on total energy consumption, China EBS 2001–2015; (Source: 2) Data on total building energy consumption, primary).


**Table 1.** China Energy Consumption growth rate in the three five-year periods.

Increased building energy consumption is mainly due to the increase of building floor areas and improvement of living standard. However, as shown in Figure 3, the annual growth of building energy consumption has stabilized, while the annual increasing of building floor area has stabilized at about 4%. This decoupling of building energy consumption with increasing building floor area is due to the improvement of building energy efficiency.

Although EBS is recognized as the most reliable energy information source and the original data of this research is from EBS, it is still necessary to compare the result of this research to other sources. Table 2 presents a comparison between this research and building energy consumption data from China Building Energy Model (CBEM). CBEM was developed and being updated by Tsinghua University Building Energy Research Centre (THUBERC). CBEM is based on a sampling of the energy consumption of different types of buildings in different climatic zones. Thus necessary sampling surveys are needed to support CBEM calculation, while the EBS approach of this research is based

on official statistics and no additional data collection or survey efforts are needed. Since there is a systematic data verification, data quality control and application of standardized data collection, data from official statistics is highly accepted and well applied by policy makers as well as various stakeholders, while the CBEM approach is widely used by academic institutes and researchers. Table 2 shows a 15-years comparison of the building energy consumption data results from this research and CBEM.

**Figure 3.** Changing annual growth of building energy consumption 2001–2015. Data source: data on building floor area [19].


**Table 2.** Data comparison between this research and CBEM, in 100 million tce.

Table 3 presents a comparison of energy consumption of three building types calculated by this research and the CBEM.

**Table 3.** Data comparison of energy consumption of three building types, in 100 million tce.


Source: CBEM data [6].

CBEM has been the most popular tool in China for estimating building energy consumption at the sector level, and this model is revised regularly. Table 2 shows that energy consumption data generated by CBEM in years of 2001 and 2002 is about 20% higher than the data calculated by this research. The difference is getting smaller in 2003–2005, then the difference has been less than 10% since 2006. The reason of the difference that appeared at the earlier stage is the poor-functioning of CBEM and the fact the model needs to be revised to meet the practical conditions. It is interesting that the CBEM results and the results of this research are aligned perfectly after 2007, which could be evidence that CBEM is now working properly and the data calculated by this research is reliable. Table 3 also shows that energy consumptions per category calculated by this study and CBEM are well aligned. We therefore suggest that both methods can be used at the same time for cross-checking and the method developed by this research can be standardized, since it is based on EBS that is dependent on the regular energy statistical system.

#### *3.2. Energy Consumption of Building Types in China*

Types of buildings are categorized into public buildings, urban residential buildings and rural residential buildings in China. Floor areas of public buildings and urban residential buildings are available from the statistics yearbooks. However, there are no statistics on the floor area of rural residential buildings. In this study, the expert workshop organized by this study suggested that the floor area of rural residential buildings could be estimated by the average floor area per farmer and the total population of farmers. Average floor area per farmer is available from both national and local housing authorities and the population of farmers is available from statistics yearbooks. The public buildings (so called commercial buildings internationally) are a mix of various buildings like offices, schools and universities, hotels, theaters, warehouses, airports, train stations, retail stores, etc.. Figure 4 provides information on the building energy consumption by the three building types in China, which shows that energy consumptions of all three types are increasing steadily. The main causes of the increasing energy consumption are the increase of building floor areas and improvement of office conditions and living standard. Energy consumption of public buildings represents a 37%–41% share of the total building energy consumption, the energy consumption of urban residential buildings represents a 36%–39% share and the energy consumption of rural residential buildings has a 23%–25% shares.

**Figure 4.** Building Energy Consumption by Categories in China.

As shown in Figure 5, the energy intensity of public buildings is above 30 Kgce/m2 (Kg coal-equivalent), which is the highest among the three building types. The energy intensity of urban residential buildings is almost double that of rural residential buildings. Among the three

building types, the energy intensity of rural residential buildings is increasing slightly, due to the significant improvement of living standards in rural areas. Figure 5 shows that energy intensity of public buildings is the highest. It was increasing in the period of 2001–2005 and then stabilized in the period from 2006–2010. It was decreasing in the period of 2011–2015, given to the successful energy efficiency solutions adopted in this period. The energy intensity of urban residential buildings is almost stabilized and is not increasing in response to the significant improvement of office conditions and living standards. This is also due to the achievements of energy efficiency efforts made by the building sectors. Figure 5 shows that energy intensity of rural residential buildings is increasing significantly, due to the significant improvement of living standards in rural areas.

**Figure 5.** Energy Intensities of Three Types of Buildings in China.

#### *3.3. China Building Energy Consumption Against Economic Growth*

As shown in Figure 6, the share of building energy demand in the total energy demand varies from 17% to 21%. The share of building energy consumption is generally decoupled from GDP growth. A lower share of building energy consumption appears when there is higher GDP growth in the period of 2001–2015. During the period of 2002–2007, the GDP growth rate increased annually and reaches its peak of 18.8% in 2007, while the share of building energy consumption decreased from 20.26% in 2002 to 17.86% in 2007. During the period of 2007 to 2014, the GDP growth rate was fluctuating, while the share of building energy consumption is reversely fluctuating. After 2010, the GDP growth is slowing down, while the share of building energy consumption is increasing.

Contrary to the varying share of building energy consumption against GDP growth, the share of industrial energy consumption is coupled with GDP growth, as shown in Figure 7. During the period of 2001 to 2007, the GDP growth rate increased, while the share of industrial energy consumption also increased and reached its peak of 69% in 2007. After 2007, DGP growth is slowing down, and the share of industrial energy consumption is decreasing.

Figures 6 and 7 provide evidence that building energy consumption and industrial energy consumption have different features. Building energy consumption is consumption-related and it is driven by lifestyles and living standards. However, industrial energy consumption is more production-related and it is driven by market and economic activities. Therefore, industrial energy consumption is mainly a consequence of economic development and building energy demand is mainly a consequence of the growing living standards. This can explain why higher GDP growth results in an increasing share of industrial energy demand and a decreasing share of building energy demand in Figures 6 and 7.

**Figure 6.** Share of Building Energy Consumption vs. GDP Growth. Data source: GDP data is from [19].

#### *3.4. China Building Energy Consumption: an International Comparison*

After four decades of high economic growth since 1978 when China started its Reforming and Opening Policy, China has been the largest country in terms of total energy consumption and greenhouse gases emissions. China has a share of 28% of the global energy requirements, followed by USA with a share of 22% and by European Union with a share of 15% [1]. However, USA is the largest country in building energy consumption, with a share of 17% of global building energy consumption, while China has a share of 14%. Table 4 provides information that building energy consumption internationally accounts for as much as 30–40% of the global energy requirements. However, China is exceptional and its building energy consumption accounts for only 20.5% of total energy consumption, given the fact that China is still in its high-speed industrializing process and its industrial sectors

are more energy-intensive and have a major contribution to total energy consumption, as shown in Figure 2.



#### **4. Discussion**

The establishment of building energy consumption statistics by adapted EBS creates a method for China to establish statistics of building energy consumption in a uniform way. An advantage of the adapted EBS is that statistical data comes from the regular statistical system and no additional statistical efforts are needed. However, the adapted EBS method has its limitations. First, all data related to building energy demand is extracted from existing EBS and thus various assumptions are made. Those assumptions are only related to the items (e.g., office building energy consumption in the sectors of transport, storage, postal and telecommunications services, industries and construction) of building energy consumption that have less than 5% contributions to the total energy consumption of buildings. This may cause errors in the final calculation of building energy demand, although we can be sure that those errors are not more than the errors generated by methods of sampling surveys, incomplete statistics, modelling and estimations. Further research is needed for identifying the error percentage and for assuming that the error can be negligible. Second, bio-energy has been widely used in Chinese rural residential buildings and this bio-energy consumption is not included in this adapted EBS system. This means that energy consumption of rural residential buildings calculated by this research is lower or significantly lower than the actual energy consumption of rural buildings. Third, our research finds that data on China building floor areas from various sources are quite different, since not all buildings are registered in building departments. The China Association of Building Energy Efficiency estimates that about 15%-25% of the existing buildings are not registered [19]. Thus building energy intensity calculated by this research is about 20% higher than the actual value. Therefore we suggest that building energy statistics and a complete building registration system should be established together. Lastly, there are more and more clean energy used at a single building scale in China [20,21] at this moment. In the existing China EBS, large-scale renewable energy production (e.g., hydropower) is already included. Clean energy production at a single building level (e.g., solar) is not included in the national EBS and thus clean energy uses at a single building scale is not included in this study. Thus further studies are needed for incorporating clean energy uses into the building energy consumption data. By doing so, the adapted EBS methods can be used to calculate CO2 emissions. Further research is also needed to test and ascertain whether this adapted EPS approach can truly help China in establishing building energy consumption statistics at both national and local levels.

#### **5. Conclusions**

China's energy statistical system is different from the IEA system. Building energy demand is not a statistical sector in China's EBS and thus data on building energy demand is not available directly from China's EBS [22]. To gather data on China's building energy demand at the sector level, various methods have been developed, tested and applied [6,19]. All the methods are based on limited sampling, incomplete statistics, modelling and estimations and thus those methods are not standardized and it is difficult to conduct regional comparisons and benchmarking in the building energy sector. This study explores a possibility where data on energy demand of the building sector can be made available from an adapted China EBS. Since EBS is the most reliable energy data source, our method can be standardized and thus will enable regional and international comparisons and

benchmarking. Our study contributes to build up China building energy statistics and our key findings include six perspectives:


The main conclusion from this study is that the existing EBS of China can be adapted to provide a more reliable and complete building energy demand information at the sector level. Thus, the underlying target is not to build up a new building energy statistics that is costly and separated from the existing regular energy statistics, but instead a better option is to apply the adapted EBS developed by this research. In addition, we suggest that the adapted EBS can be used with the support of the existing CBEM to ensure accurate data cross-checking.

**Author Contributions:** Conceptualization, M.Z., Y.Z. and C.X.-B.; methodology, M.Z. X.G. and C.X.-B.; validation, M.Z. X.G., Y.Z. and C.X.-B.; formal analysis, M.Z. and X.G.; data curation, X.G., Y.Z. and C.X.-B.; writing—original draft preparation, M.Z. X.G. and C.X.-B.; writing—review and editing, M.Z. and X.G.

**Funding:** This research is funded by the European Commission's Switch Asia II programme: Promoting Sustainable Consumption and Production (contract number: DCI-ASIE/2015/368-399).

**Acknowledgments:** We would like to thank all the SusBuild project staff involved in this research for their valuable contribution and comments. We would like to give special thanks to China Association of Building Energy Efficiency, who provides us necessary data and supports us for data collection.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Impact of Heating Control Strategy and Occupant Behavior on the Energy Consumption in a Building with Natural Ventilation in Poland**

#### **Aniela Kaminska**

Faculty of Electrical Engineering, Poznan University of Technology, ul. Piotrowo 3a, 60-965 Pozna ´n, Poland; aniela.kaminska@put.poznan.pl

Received: 10 October 2019; Accepted: 7 November 2019; Published: 12 November 2019

**Abstract:** This study aims to provide an experimental assessment of energy consumption in an existing public building in Poland, in order to analyze the impact of occupant behavior on that consumption. The building is naturally ventilated and the occupants have the freedom to change the temperature set point and open or close the windows. The energy consumption is calculated and the calculation results are compared with the experimental data. An analysis of occupants' behavior has revealed that they choose temperature set points in a wide range recognized as thermal comfort, and window opening is accidental and difficult to predict. The implemented heating control algorithms take into account the strong influence of individual occupant preferences on the feeling of comfort. The energy consumption assessment has revealed that the lowering of temperature set point by 1 ◦C results in an energy saving of about 5%. Comparisons of energy consumption with heating control and without any controls showed that the potential for energy reduction due to heating control reached approximately 10%. The use of windows control, which allows to turn off the heating after opening the window and its impact on energy savings have been discussed as well.

**Keywords:** building automation systems; building energy efficiency; heating control; energy savings

#### **1. Introduction**

Currently, the global building sector has been the main consumer of world energy [1]. Energy consumption in the existing buildings accounts for 40% of the total energy consumption in the United States [2] and in Europe [3] where 75% of buildings are energy inefficient [4]. Therefore, the European Commission has published a series of recommendations on the modernization of buildings including guidance on the automation and controls of buildings [5]. However, despite the large number of building retrofit technologies [6] and the management of heating, ventilation and air conditioning (HVAC) systems, the implementation of these recommendations is a difficult and costly challenge.

In making any decisions regarding the modernization of a building, estimating energy consumption in the building is of key importance. This consumption is influenced by many factors such as ambient weather conditions, building structure and characteristics, the operation of HVAC systems and occupancy. One of the most important factors is climate data, which plays a fundamental role in the building design. Results presented in [7] show that an improvement of around 15% in energy consumption in buildings can be achieved due to changes in building design such as space area, exterior openings and material thickness and the choice of building envelope in all climates. An overview of measures and policies adopted by different countries, allowing the monitoring, management and reduction of energy consumption in buildings is given in [8]. The energy consumption related to HVAC systems in different types of buildings (office, commercial and residential) is analyzed in [9]. It is widely expected that building occupancy is of great importance for energy efficient control of

buildings. Therefore, a large number of works have been developed for the estimation and detection of building occupancy. A comprehensive review on this problem is presented in [10]. However it should be noted that new buildings are mostly controlled by a building management system (BMS) where building occupants have minimal access to the controls. In these buildings energy consumption is not strongly correlated with occupancy patterns [11].

Many factors influencing energy consumption mentioned above make the estimation of this consumption very difficult. In [12] recently developed models for solving this problem, including elaborate and simplified engineering methods, statistical methods and artificial intelligence methods are reviewed. Quantitative energy performance assessment methods are described in [13]. To simplify the calculation of energy in the building, a steady-state model was developed as CEN standards, i.e., energy performance of the building—calculation of energy use for space heating and cooling [14]. In this model the predicted energy consumption consists of heat transfer through the building envelope, heat losses for ventilation, heat gain from solar radiation and internal heat gain from people and equipment. In cold climates, such as in Poland, the energy used for heating is predominant, therefore, knowing the thermal characteristic of the building envelope and ventilation is crucial [15]. In old buildings, natural ventilation with operable windows is usually used. In new buildings, this type of ventilation also becomes increasingly popular as a solution with lower energy consumption compared to mechanical ventilation and air conditioning. Over the past decades, the impact of various parameters on the performance of natural ventilation has been studied [16] and many models have been developed. Important natural ventilation models and simulation tools as well as the comparisons of their prediction capabilities are reviewed in [17]. The analysis shows that these models are generally only applicable to specific geometries and driving forces. Furthermore, the most accurate models are developed for cases with small and simple openings. To investigate the air flow pattern inside a building, computational fluid dynamics (CFD) models are developed. The model based on the finite volume numerical solution of the Navier–Stokes equations presented in [18] shows that different positions and shapes of an opening can determine the behavior of the flow stream inside the building. It allows to determine the condition of natural ventilation efficiency of the building. Another fluid dynamics (CFD) model allows to investigate a wind-driven ventilation system in a building with multiple windows [19].

The study mentioned above shows the complexity of a phenomenon that has a decisive influence on thermal comfort and energy consumption. In a naturally ventilated building, thermal comfort can be improved and adapted to individual preferences when occupants have the freedom to change the temperature set points and open or close the windows.

Various case studies [20,21] have shown that occupants tend to adapt to changing environmental conditions in such a way as to achieve their individual comfort. Research on such behavior is called the adaptive approach. The application of this approach to thermal comfort standards is considered in [22] and an equation for naturally ventilated buildings in hot-humid climates is developed in [23]. It was found that acceptable comfort ranges showed asymmetry and leaned towards operative temperatures below thermal neutrality for all climates. However, other results, inter alia in [24], based on the data of surveys conducted in a naturally ventilated building found symmetry of comfort ranges. Many studies also confirm it is difficult to use defined comfort ranges in the real conditions because it depends on the occupants' physiology and subjective perception [22]. The thermal sensations of occupants inside buildings are influenced by many factors such as air temperature and velocity, humidity, concentration of CO2, building microclimate, as well as age, activities, preferences, etc. [25,26]. Occupants have various means of interacting with the indoor environment: they can interact directly with a given built environment by changing the temperature set points (or adjusting thermostats), operating the windows, shading, or they can adjust themselves to the existing environmental conditions by changing their clothing or activity [27]. As regards the theory of thermal comfort in buildings, a large impact of clothing and activity on the level of comfort is represented by the most extended predicted mean vote (PMV) index [22,25,28]. This index described the statistical response about thermal sensation of a large group of people exposed to specific thermal conditions. Six variables, namely metabolic rate, clothing insulation, air and mean radiant temperatures, air velocity and relative humidity affect the PMV index. Four of them can be recorded during the experiment, while clothing insulation and metabolic rate are not easily measurable and their values are most often taken from [27]. For a typical office the values of clothing insulation are 1.0 and 0.5 clo for winter and summer respectively, whereas a typical value used for metabolic rate is 1.0 met. It is also worth noting that the occupant-building interaction is bidirectional, which means that the building environment and interior also affect the occupants' behavior [25], but this interaction requires additional research to identify and describe.

The behavior of occupants is a key issue in the design of the HVAC system and its integration with other control systems in the building as well as in the assessment of energy efficiency [29]. Various methods of occupant behavior estimation and detection are used in [10] and models of occupant behavior can be an efficient means to be implemented into building energy modeling programs [30]. Detecting the presence and absence of occupants allows to determine the operation time of HVAC systems in the building. Potential annual energy savings are estimated at around 10–40%. It has been shown in [31] that the HVAC system can save up to 9% of energy if occupancy-based HVAC schedules are used. In [32], an algorithm for adjusting temperature set points with various indicators of occupant discomfort tolerance has been proposed and energy savings are estimated at 20% while maintaining the building comfort requirements. In [33], based on the detection of the instantaneous number of occupants in the building and related behaviors, it was demonstrated that the energy consumption of the building could be reduced by 40% without compromising the thermal comfort and air quality. However, although there are many methods for detecting and describing occupant behavior to achieve energy savings, their limitations are revealed when applied to real HVAC systems, and they are mainly related to the difficulty of tracking occupant-provoked changes by the HVAC system.

The use of information about occupant behavior to control the HVAC system and estimate possible energy savings depends on the thermal behavior of the building, which determines the heating and cooling time of the building. Several studies have been carried out to investigate the building thermal behavior and model predictive control (MPC), which allow better tracking of changes in the operating mode and temperature set points [34]. The knowledge of building thermal behavior and the popular gray box model approach are the basis for designing an HVAC control system and estimating the energy savings potential [35,36].

As stated above, because the potential of energy savings depends on various parameters, its estimation shows large discrepancies. This paper deals with the experimental and theoretical evaluation of energy consumption in an existing public building in Poland. The building is naturally ventilated and the occupants have the freedom to change the temperature set point and open or close the windows. The effect of occupant behavior as well as heating control and window operation on energy consumption is investigated. The main purpose of the work is to determine the impact of window opening and the range of temperature set point chosen by the users on energy consumption.

The temperature set points in the heating zones of the building and the outdoor temperature are measured and recorded by the KNX automation system and for these temperatures the energy consumption is calculated taking into account heat transfer through the building envelope and heat losses for ventilation. The calculation results are compared with the experimental data. A heating control strategy has been implemented in the building and the energy saving potential is estimated for this strategy.

#### **2. Methodology**

The main purpose of the work was to determine how much energy could be saved in a real building by using heating control. It is also important to determine what factors affect the energy savings in a building. In order to achieve this goal, energy consumption for heating was first calculated. The calculations took into account temperatures outside the building and inside the rooms as they occurred during the one-month period. These temperatures were recorded in the KNX system implemented in the building. It was noted that occupants chose temperature set points corresponding

to their thermal comfort, which differed by several degrees. In order to verify the calculations, calculated energy values were compared with measured values. Then, it was assumed that the temperature in the whole building was constant during the analyzed period and that the outside temperature was as in the previous experiment. Two temperature values were selected, namely 20 ◦C and 21 ◦C. Energy consumption for these conditions was referred to as "consumption without heating control". The next task was to calculate energy consumption for the same external conditions, but taking into account the control method used in the building. This consumption was marked as "energy consumption with heating control". However, this required determining the temperature changes in the rooms of the building after lowering the temperature set point. This problem was investigated experimentally and discussed. Attention was also paid to the temperature change when the window is tilted from the top by 30◦ from vertical. This method of window opening is often used by occupants.

#### **3. Building and Experimental Installation**

#### *3.1. Construction of the Building*

This study deals with the activities of the Laboratory of KNX System and Evolution of Installation Energy Efficiency (SKNX and EIEE Laboratory) at Poznan University of Technology in Poznan, located in the north-western part of Poland (Figure 1). The building was built in the 1980s and is representative of existing Polish buildings from that period considering building envelopes. In 2010 the building was retrofitted and its energy efficiency improved significantly. It is a three-story building with a height of 11.48 m and the external outline surface of 236.8 m2. On the south the building adjoins another facility up to the level of one story.

**Figure 1.** External view of the KNX System and Evolution of Installation Energy Efficiency (SKNX and EIEE) Laboratory building.

Figure 2 shows the thickness and the value of the thermal conductivity coefficient of each layer that constitutes part of the building envelope. The thermal conductivity coefficients are taken from PN-EN ISO 6946 [37].

The external walls (Figure 2a) with a thickness of 380 mm were built of full ceramic brick and covered with 15 mm lime and cement-lime plasters. In the ground, the walls were made of cement blocks and covered with two 15 mm layers of cement-lime plasters. As a thermal insulation, a 120 mm layer of styrofoam was used on the external walls. At a height of 50 cm below and above the ground, extruded polystyrene with a thickness of 90 mm was placed.

The roof (Figure 2b) is multi-layered and consists of 240 mm channel slabs, 100 mm layer of Supreme, a void of 210 mm, 20 mm cement plaster and the final layer of 45 mm roofing felt. Thermal isolation was achieved by blowing Rockwool granules into the air void. The laboratory floor was not thermo-modernized, and the layers in contact with the ground in the part corresponding to heating zone 1 are presented in Figure 2c, and those corresponding to zones 2 and 3 are shown in Figure 2d. The main layers of the floor in heating zone 1 are a 150 mm layer of concrete debris and a 300 mm layer of granulated blast-furnace slag. Insulating roofing tar on a layer of waterproof asphalt and cement-bonded wood fiber are used as the insulation. In heating zones 2 and 3 the floor forms layers of concrete debris, leveling concrete and terrazzo. The whole floor in all the zones is covered with floor gres laid on cement-plaster.

The thermal resistance of a component layer *i* of a building envelope is defined as *Ri* = *di*/λ*i*, where *di* is the thickness of the layer and λ*<sup>i</sup>* is the thermal conductivity coefficient. The thermal resistance *R* of a multi-layer building envelope is determined as the sum of the thermal resistance of the component layers and the conventional internal surface thermal resistance *Rsi* and the external surface thermal resistance *Rse*. The values of *Rsi* and *Rse* resistance depend on the type of building envelope and the direction of heat flow. For external walls and the horizontal direction of heat flow *Rsi* = 0.13 m2 K/W and *Rse* = 0.04 m2 K/W, for flat roof *Rsi* = 0.10 m2 K/W and *Rse* = 0.04 m2 K/W [38]. The heat transfer coefficient, by definition, is calculated as *U* = 1/*R*.

**Figure 2.** Cross-section of: (**a**) the external wall; (**b**) the roof; (**c**) the floor in heating zone 1 and (**d**) the floor in heating zones 2 and 3.

In the walls, there are window jambs, lintels and wall connections, which result in the formation of thermal bridges that increase heat transfer. They are taken into account by introducing a correction of Δ*U*. For external walls with windows Δ*U* = 0.05 W/m<sup>2</sup> K is assumed.

The heat transfer coefficient for windows is determined as:

$$
\Delta L\_{\mathcal{W}} = \frac{A\_{\mathcal{S}} \cdot \mathcal{U}\_{\mathcal{S}} + A\_f \cdot \mathcal{U}\_f + l\_{\mathcal{S}} \cdot \mathcal{V}\_{\mathcal{S}}}{A\_{\mathcal{S}} + A\_f},
\tag{1}
$$

where: *Ug* and *Uf* are the heat transfer coefficients in the middle part of double glazing and the frame, respectively, *Ag* and *Af* are the surfaces of the glass and the frame, Ψ*<sup>g</sup>* is the linear heat transfer coefficient of the thermal bridge at the interface between the glass and the frame and *lg* is the length of the thermal bridge. According to the technical approval for windows *Ug* = 0.5 W/m<sup>2</sup> K and *Uf* = 1.2 W/m<sup>2</sup> K. The surface of the glass is 0.4544 m<sup>2</sup> and that of the frame is 0.7781 m2. The length of the thermal bridge amounts to 2.3 m and the linear heat transfer coefficient is taken as 0.06 W/m<sup>2</sup> K.

The main entrance to the building leads through two doors from the west. The surface of the single door is 3.494 m2. There is an additional door with a surface of 3.478 m2 on the east of the building, occasionally used for moving heavy equipment. According to the technical approval the heat transfer coefficient is 2.6 W/m2 K.

#### *3.2. Heating Zones*

The building was divided into heating zones shown in Figure 3, differing in use, size and separation walls. The division into zones determined the pipeline system, in particular the number of heating circuits supplying hot water to panel radiators.

**Figure 3.** Heating zones in the case study building: (**a**) the ground floor plan; (**b**) the first floor plan and (**c**) the second floor plan.

On the ground floor, there are three heating zones, namely zone 1 and 2 including high-current laboratories and zone 3 including a workshop, sanitary facilities and a corridor. People staying in these rooms do not perform sedentary work and the operation of the devices causes an increase in temperature. The first floor consists of four heating zones. These zones are the most stable in temperature, due to the floor being closed with a staircase door and because of its location between the heated floors of the building. The second floor was divided into five heating zones corresponding to the rooms. The height of all zones is the same and amounts to 2.8 m.

#### *3.3. Control System and Data Acquisition*

The heating system in the SKNX and EIEE Laboratory building is designed in such a way that it is possible to estimate the heat consumption in each room and implement various control algorithms as well as to measure, record and visualize useful data [39]. Panel radiators are used as the heating devices. In this system heat is carried by water supplied from the city heating network. The scheme of the pipeline system is shown in Figure 4. In order to force the water flow through the installation, circulation pump (P) is used. At the inflow, a control valve (CV) has been mounted and heating water parameters are measured using a heat meter. Then, the hot water flows into three main circuits assigned to each story and the heating water parameters are also measured at the inflow to each circuit. The water feeds heating circuits assigned to heating zones (Figure 3): on the ground floor—three circuits, on the first floor—four circuits and on the second floor—five circuits. Water from heating devices returns through the pipelines on the stories and then the main pipeline to the city heating network. Each water circuit is equipped with a heat meter and a KNX servo drive. The servo drives are controlled by signals sent directly from the KNX bus. The KNX multi-function push-button with a room temperature control unit is located in each heating zone. In addition, the KNX Laboratory (heating zone 5) is equipped with a KNX touch panel that visualizes the states and parameters of the system. A valve controller at the heating system inflow and heat meters is connected to the ControlMaestro controller with a SCADA (superior control and data acquisition) system using an M-Bus network (Figure 4). This system allows the visualization and acquisition of values measured in the building heating system.

**Figure 4.** Heating system pipeline scheme in the SKNX and EIEE Laboratory.

To control the heating system KNX devices mentioned above and KNX BACS field network are used. In the KNX system other devices are integrated, including a weather station, brightness and temperature sensor, presence detectors and Gira HomeServer. KNX is an open standard for public, commercial and domestic buildings [40], which allows the integration of many devices from different manufacturers. KNX devices are most often connected by a twisted pair or RF bus and programmed with the use of ETS software. It is worth noting that the system used in the laboratory building can be easily expanded with new devices, and in addition, it allows testing various control algorithms through reprogramming using the available ETS software. Two networks, M-Bus and KNX, are integrated using a M-Bus/KNX converter (Figure 5), which enables the acquisition of all measured values and events in the form of telegrams (standardized KNX messages) by the KNX HomeServer. The HomeServer visualizes the results on-line, archives them and, once a day, sends the results as a csv file to specified e-mail addresses. The recording format allows further processing of the results by external tools and programs.

The following data were recorded by the HomeServer:


**Figure 5.** Integration of M-Bus and KNX networks.

In order to determine the position status of the windows and take it into account in the heating control, the intruder alarm system (IAS) in the building was integrated with the KNX system. In window frames, reed switches are mounted and signals from these devices are sent to the alarm control unit, which transmits them to the KNX binary input.

#### *3.4. Temperature Set Point*

The temperature set points for the heating seasons are established based on ISO (International Standard Organization) Standard 7730 [41], which defines the comfort ranges according to the specificity of Europe [42]. However, it should be noted that thermal sensations differ between persons sharing the

same environment, because there are many factors that affect the perception of human beings [26,28]. The thermal sensations experienced by a human being result mainly from the overall thermal balance of the body. This balance includes two components, namely heat generated by a human being and heat transferred to the environment. The first depends on the physical activity and the second depends on clothing, as well as on environmental parameters such as air temperature, radiant temperature, air velocity and air humidity [43].

The American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) standard [44] specifies the conditions in which a fraction of occupants find the environment thermally acceptable. The predicted mean vote (PMV) and the predicted percentage dissatisfied (PPD) are defined in ISO 7730 [41]. The thermal comfort index PMV-PPD reflects the degree of human thermal balance deviation and is a comprehensive comfort indicator that represents the feelings of most people in the same environment. PMV scales constitute seven thermal sensation points ranging from −3 (cold) to +3 (hot), where 0 means a neutral thermal sensation [45]. The PMV index involves activities (expressed through the metabolic rate index), clothing corresponding to the total thermal resistance from the skin to the outer surface of the clothed body and the four environmental parameters mentioned above [41,46].

Depending on the admissible ranges for PMV and PPD, three kinds of comfort zones or categories of thermal requirements are defined by ISO 7730 as: category I (or class A; PPD < 6%, i.e., −0.2 < PMV < 0.2), category II (or class B; PPD < 10%, i.e., −0.5 < PMV < 0.5) and category III (or class C; PPD < 15%, i.e., −0.7 < PMV < 0.7). The ranges of recommended air temperatures for different types of buildings depending on the previous categories are shown in Table 1 [41]. Thus, in the study case building the range of temperature set point was set from 19 to 25◦C and the occupant had some freedom to choose the preferred temperature during their presence in the room. It should be noted that this value was a subjective decision of the occupant and the prediction of occupant behavior was a factor of considerable uncertainty in the analysis [47].


**Table 1.** The range of recommended air temperatures for offices and classrooms, according to ISO7730 [41].

#### *3.5. Building Use and Heating Control Algorithm*

The analyzed information about the occupancy, opening windows, operation mode of the heating system and changing the temperature set point in each room of the building is derived from the data recorded by Gira HomeServer. On weekdays, the building is usually occupied from 8 a.m. to 6 p.m. In this time, the heating system operates in comfort mode with the various temperature set points in the rooms set by the occupants. From 6 p.m. to 8 a.m. the system operates in night mode with the constant temperature of 16 ◦C. In practice, lowering the temperature set point to 16 ◦C results in closing the KNX servo drive and switching off the heating system. This control algorithm is considered below and the experimental results were compared with the calculation. To assess the energy saving potential due to heating control the same algorithm was assumed, but the temperature was constant in comfort mode (21 or 20 ◦C). This case was referred as "with control".

In a real heating control other functions are implemented. One of these functions is the detection of window opening (or tilling from the top by 30◦ from vertical). This function is essential because the occupants have free and easy access to open the windows in their own office and laboratory rooms. Opening the window by the user in the room results in a transition of the heating system to the anti-frost mode with a temperature of 7 ◦C. In addition, the heating control system was integrated with the intruder alarm system. It is not possible to arm this system when a window in the building is open. Occupants leaving the building arm the system and they must close all the windows.

Another function is presence detection in the off time, between 6 p.m. and 8 a.m. and on weekends. If users start work earlier, finish later or work on weekends, information about the events is transmitted from the presence sensor to the heating control system, which changes the operating mode to comfort mode in the room where such presence is detected.

#### **4. Calculation of Energy Consumption**

The energy consumption *Qsmj* in the time interval Δ*tm* in the *j*-th heating zone is estimated taking into account heat transfer through the building envelope and heat losses for ventilation according to the following formula [14,15]:

$$Q\_{\rm smj} = \sum\_{i=1}^{n} Q\_{Tij} + Q\_{Vj} \tag{2}$$

where: *QTij* is heat losses for transmission through the *i*-th barrier in the *j*-th heating zone, *QVj* is heat losses for ventilation in the *j*-th heating zone and *n* is the number of partitions.

The heat losses (or gains) for transmission through the *i*-th barrier are estimated as:

$$Q\_{T\bar{i}} = \mathcal{U}\_{\bar{i}^\*} (\mathfrak{\vartheta}\_{\text{im}} - \mathfrak{\vartheta}\_{\text{em}}) \cdot \mathcal{A}\_{\bar{i}^\*} \Delta t\_{\text{m} \star} \tag{3}$$

where: *Ui* is the heat transfer coefficient through the *i*-th barrier in W/m<sup>2</sup> K, ϑ*im* is the air temperature in ◦C, in the room, in the time interval Δ*tm*, ϑ*eim* is the air temperature in ◦C, outside the *i*-th barrier, in the time interval Δ*tm*, *Ai* is the surface of the *i*-th barrier in m2 and Δ*tm* is the time interval in hours.

The heat loss for ventilation in the *j*-th heating zone in Wh is calculated as follows:

$$Q\_{Vj} = 0.333 \cdot (\theta\_{im} - \theta\_{eim}) \cdot V\_j \cdot \Delta t\_{m\prime} \tag{4}$$

where: *Vj* is the ventilation air stream flowing into the *j*-th heating zone in m3.

Ventilation of the rooms is provided by ventilation ducts (Figure 3) and window ventilators integrated in the frames. Each ventilator is equipped with a regulator allowing different air flow rates. Due to the impact of various parameters on the performance of natural ventilation and the complexity of the phenomenon [16–19], the volume of ventilated air in the room was estimated based on the difference between energy consumption measured with open window ventilators and this energy measured with completely closed ventilators and ventilation duct. This difference determines the heat loss for ventilation and the volume *Vj* is estimated using Formula (4).

Energy consumption in the analyzed period is estimated as the sum of heat losses calculated in time intervals *m* in which various temperature increases ϑ*im* − ϑ*eim* occurred, therefore:

$$Q\_{sj} = \sum\_{m} Q\_{smj} \,\tag{5}$$

#### **5. Results and Discussion**

#### *5.1. Ambient Weather Temperature and Daylight Illuminance*

The analysis of energy consumption was carried out for the month of January 2017. January is usually the coldest month of the year in Poland. The calculations were performed taking into account the actual ambient temperature measured by the weather station installed on the south-eastern facade of the building. In calculation it is ϑ*eim* temperature. However, the temperature was also measured by the external brightness and temperature sensor mounted on the northern facade. It should be noted that the values measured by these two sensors on a sunny day differ from each other. Two phenomena are responsible for the measurement discrepancies. The first one is the insolation of the building walls, which is stronger for the south-eastern wall than for the north wall. On cloudy days there is no difference in the heating of the walls by sunlight and the measured temperatures are close to each other. The second is the direct impact of sunlight on the weather station. This effect is mainly observed on a

sunny day with high variability of daylight. In this case, variations in illuminance and temperature occur simultaneously. The lowest and highest temperatures on each day of January measured by the weather station and the brightness sensor are shown in Figure 6a. The lowest temperature in the month was about −13 ◦C and the highest about 4 ◦C. The temperature difference during the day reached 10 ◦C.

**Figure 6.** Data measured in January 2017 by the weather station WS and the brightness and temperature sensor BS: (**a**) the lowest and highest temperature and (**b**) the highest daylight illuminance level.

When the illuminance levels measured by the weather station and the brightness sensor are the same (Figure 6b), it means that the day is overcast and the walls are not heated by sunlight. This is from 18 to 21 January. Obviously, there is a time shift between variations in the illuminance level and the temperature. On 18 January, the wall was still warmed up by daylight and there was a difference in the measured temperature values. Due to these temperature differences, on a sunny day, the temperature values measured by the brightness and temperature sensor are represented as ϑ*eim* temperature in the calculation. Time intervals Δ*tm* are determined, in which the temperature ϑ*eim* differs by 1 ◦C. The air temperature ϑ*im* is taken as the current temperature in the heating zone, measured by the push-button with room temperature control unit and recorded by the HomeServer.

#### *5.2. Temperature Changes Inside the Building*

The implementation of heating control algorithms must take into account temperature changes in the rooms as a result of lowering the temperature set point, switching off the heating, opening the window and other events. Anyway, heating control usually consists in lowering the temperature at night and on weekends and turning off the heating after opening the window. The change in temperature will depend on the thermal properties of the building and the ambient conditions, i.e., temperature, wind speed, rainfall and daylight. Figure 7a shows the temperature inside and Figure 7b the temperature outside the building during three days, i.e., from 0:00 on 10 April to 24:00 on 12 April, which is during 4320 min. To investigate temperature changes in the building, the temperature was first lowered by fully opening (on 9 April) one window in zones 4 and 5. The heating system switched to the anti-frost mode and until 9:40 on 10 April (in 580 min) the temperature in these zones decreased to 20.9 and 21.8 ◦C, respectively. At that time, the windows were closed and the temperature increased to the temperature set points. Further temperature changes were forced at 18:40 (in 1120 min) by turning the heating system off and then at 9:30 on 12 April (in 3450 min) by turning this system on. Temperature changes in zone 4 prove the high thermal inertia of this zone and it may take several hours to reach a higher temperature set point or comfort mode temperature after earlier turning off the heating. On the other hand, the temperature reduction after switching off the heating is small when the windows are closed. For the considered conditions it was approximately 1 ◦C for about 39 h. It is worth noting that the temperature changes in zone 4 were even smaller than in zone 5. In zone 3 the window

was not open and the temperature increased around 750 and 2100 min as a results of insolation and an increase of temperature outside the building. A slight effect of the outside temperature on the inside temperature could also be seen in zones 4 and 5.

**Figure 7.** Impact of changing the heating system operating mode on the temperature inside the building. Temperature from 0:00 on 10 April to 24:00 on 12 April: (**a**) in the heating zones and (**b**) outside the building.

In order to estimate the effect of window operation on energy consumption, the temperature changes after tilting the window from the top by 30◦ from vertical were analyzed. It is worth noting that such window operation was often used by occupants. The window was tilted on 6 April at 16:15, 975 min from 0, which corresponds to 0:00. Figure 8a shows that after tilting the windows the temperature in both zones dropped to about 23 ◦C, then due to the increase in the outside temperature (Figure 8b) the temperature inside the zones increased too. However, the temperature increase in the two zones was different due to the difference in insolation of these rooms. In zone 4 the windows were located on one wall of the room on the north-east side, while in zone 5, the windows were on two sides of the room, i.e., north-west and north-east. The illuminance level of daylight is shown in Figure 8c.

**Figure 8.** *Cont.*

**Figure 8.** Impact of window operation on the temperature inside the building. (**a**) Temperature change due to one window tilting from the top inside zones 4 and 5; (**b**) temperature outside the building and (**c**) daylight illuminance level.

On 8 April, the daytime temperature dropped below 10 ◦C and then to around −1 ◦C at night. This resulted in a lower room temperature, more significant in zone 4. On 9 April at 9:40 (4900 min) the windows were closed in both zones, the heating system turned on and the temperature started increasing to the set point value. It should be noted that tilting of only one window in the room led to a temperature decrease of around 3 ◦C during the considered time, which corresponds to the weekend time.

#### *5.3. Energy Consumption Experiment and Calculation*

Energy consumption in the heating zones on days of January 2017, measured by the heat meters, is shown in Figure 9, and Table 2 presents the measured and calculated energy consumed over the whole month. The results were obtained with no heating control and the actual room temperatures were equal to the temperature set points. The occupants had the freedom to choose the temperature set points and, as can be seen in Table 2, the range of the selected set points was wide: from 19 to 24.5 ◦C. It reveals a strong influence of individual occupant preferences on the feeling of comfort.

**Figure 9.** *Cont.*

**Figure 9.** Energy consumption on days of January 2017 measured with heat meters: (**a**) in heating zones 1–3; (**b**) in heating zones 4–7; (**c**) in heating zones 8–12 and (**d**) on the stories of the building.

The highest energy consumption (Figure 9a) was in heating zone 1 due to the large volume of air to be heated, which results from the fact that this zone includes not only the laboratory room but also the entrance of the building and the open space of the staircase. The energy consumption in heating zone 2 was higher than in zone 3 due to the heat transfer through the door in zone 2 and a lower temperature in zone 3. The comparison of energy consumption on the three floors of the building (Figure 9d) shows that the highest consumption was on the ground floor due to the poor thermal insulation of the floor and the volume of heated air. The lowest energy consumption occurred in the rooms on the first floor (Figure 9b). On the second floor, where there was heat transfer through the roof, energy consumption was higher than on the first floor.

**Table 2.** Experimental and calculated energy consumptions in the heating zones, in the month of January 2017.


Energy consumption in each room depended on their volume, temperature set point and insolation, therefore it was better to compare the energy consumption per unit of room area, at which the above conclusions were rather obvious. Another good example is the comparison of energy consumption per unit of room area in heating zones 4 and 8 with the same volume, which showed that the consumption in zone 4 was lower despite a higher temperature. It is worth noting that the calculated value was always larger than the measured value, and it seemed to be the case for two reasons. Firstly, heat gains from insolation, people and equipment were not included in the calculations. Secondly, at small energy values measured, heat meter indications were burdened with significant errors, namely the values were underestimated. As the measurement of thermal energy by the heat meter was carried out indirectly, on the basis of measuring the volume of the water and the temperature difference at the inflow and return, the measurement error could be relatively large (±5%). However, the difference

between the calculated and measured energy consumption values in relation to the measured value did not exceed 15% and the estimation of energy consumption could be considered sufficiently accurate.

#### *5.4. E*ff*ect of Room Temperature and Heating Control on Energy Consumption*

In the heating season 2017/2018, between the end of September and the beginning of May, time control of the heating was implemented. Due to different weather conditions, the experimental results of two heating seasons could not be compared in order to estimate the effect of heating control on the reduction of energy consumption. Therefore, the calculations were carried out for the weather conditions in January 2017: first, without heating control, assuming that the room temperature was 21 and 20 ◦C. The value of 21 ◦C corresponds to the recommended indoor air temperature in education rooms of category A and 20 ◦C in rooms of category B (Table 1). Then, based on the observation, it was assumed that after the transition of the heating system to night mode, the temperature dropped in the rooms located on the ground floor by an average of 1.5 ◦C at night (during 14 h). At weekends (during 62 h), the reduction was about 4 ◦C. These temperature drops were, respectively, about 0.5 ◦C and 1.5 ◦C in the rooms on the first floor and 1 ◦C and 3 ◦C on the second floor. The calculation results are given in Table 3. This calculation shows that reducing the temperature set point by 1 ◦C gives an energy saving of about 5% compared to energy consumption at 21 ◦C.

**Table 3.** Energy consumptions in the heating zones without and with control, calculated considering the weather conditions in the month of January 2017. \* The energy saving potential is determined in comparison with the energy consumption at the temperature of 21 ◦C.


The comparison of energy consumption with and without heating control reveals that the energy saving potential mainly depended on the temperature drop after the set point lowering. The greater the decrease, the greater the potential for energy savings. In the study case, in the rooms with a poorly heat-insulated floor, the energy reduction due to heating control reached about 10%. A slightly lower reduction of about 7.5% was estimated for the rooms on the second floor, where heat was transferred through the roof, and the smallest reduction of less than 4% was estimated for the rooms on the first floor. This proves that in well-insulated rooms with a low energy consumption for heating the implementation of the control system gave relatively little benefit.

For energy saving, a very important function was to control the opening of a window. As shown in Figure 8 the tilt of the top of only one window in the room led to a temperature decrease of a few degrees. Leaving the window open before night or weekend would result in a significant increase in energy consumption, by about 5% per 1 ◦C drop.

#### **6. Conclusions and Future Work**

In this paper, the potential of energy savings in an existing public building in Poland was estimated. This estimation includes the most important parameters affecting energy consumption for heating. Experimental verification of the building case study showed that the calculation of energy consumption in a cold climate including the heat transfer through the building envelope and heat losses for ventilation were sufficiently accurate. In such calculations, a good knowledge of the thermal characteristics of the building, the volume of ventilated air and the temperature outside and inside the building is crucial.

Using the KNX system implemented in the building, the behavior of occupants was investigated revealing that occupants choose temperature set points in a wide range recognized as thermal comfort, and window opening was also accidental and difficult to predict. The proposed heating control algorithms took into account the strong influence of individual occupant preferences on the feeling of comfort. However, in order to reduce energy consumption, the anti-frost mode was applied after opening the window, as well as integration with the intruder alarm system. Investigation of temperature changes in the building with changes in the temperature set points and after opening the window showed that from the point of view of energy saving, the most important issue is the window opening control.

Finally, detailed comparisons of energy consumption with heating control and without any controls were performed. It shows that the energy saving potential depended on the temperature drop after lowering the set point, and thus on the dynamics of the thermal behavior of the building. The greater this drop, the greater the potential for energy savings. In the case study, in rooms with poorly heat-insulated floors, the energy reduction potential due to heating control reached about 10%. A slightly lower potential of about 7.5% was estimated for rooms on the second floor, where heat was transferred through the roof, and the smallest potential of less than 4%, for rooms on the first floor. This proved that in a well-insulated room with a low energy consumption for heating, the implementation of the control system gave relatively little benefit.

Future work will include an analysis of information from presence detectors to describe occupant behavior, and the implementation of such information to control heating and estimate energy savings. Research associated with the optimal operation of the heat source will also be undertaken.

**Funding:** This study is based upon work supported by the National Centre for Research and Development in the context of the Innovative Economy Program under grant No. POIG.02.02.00-00-018/08. This work was also supported by the 2018 Poznan University of Technology funds transferred from the Ministry of Science and Higher Education.

**Conflicts of Interest:** The author declares no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.

#### **References**


© 2019 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Analysis of the Thermal Behavior of an Earthbag Building in Mediterranean Continental Climate: Monitoring and Simulation**

#### **Lídia Rincón, Ariadna Carrobé, Marc Medrano \* , Cristian Solé, Albert Castell and Ingrid Martorell** †

SEMB Research Group, INSPIRES Research Centre, Universitat de Lleida, Pere de Cabrera s/n, 25001 Lleida, Spain; lrincon@diei.udl.cat (L.R.); acarrobe@diei.udl.cat (A.C.); csole@diei.udl.cat (C.S.);

acastell@diei.udl.cat (A.C.); imartore@diei.udl.cat (I.M.)

**\*** Correspondence: mmedrano@diei.udl.cat

† Serra Húnter Fellow, Generalitat de Catalunya.

Received: 7 November 2019; Accepted: 24 December 2019; Published: 30 December 2019 -

**Abstract:** Nearly 30% of humanity lives in earthen dwellings. Earthbag is a sustainable, cheap, feasible and comfortable option for emergency housing. A comparative monitoring-simulation analysis of the hygrothermal behavior of an Earthbag dwelling in Mediterranean continental climate, designed under bioclimatic criteria, is presented. The dome shape Earthbag dwelling has a net floor area of 7.07 m2, a glass door facing south and two confronted windows in the east and west facades. A numerical model (EnergyPlus v8.8) was designed for comparison. Twenty-four hour cross ventilation, night cross ventilation, and no ventilation in free floating mode and a controlled indoor temperature were the tested scenarios. Comparisons between experimental data and simulation show a good match in temperature behavior for the scenarios studied. Reductions of 90% in summer and 88% in winter, in the interior thermal amplitude with respect to exterior temperatures are found. Position of the glazed openings was fundamental in the direct solar gains, contributing to the increase of temperature in 1.31 ◦C in winter and 1.37 ◦C in the equinox. Night ventilation in the summer period had a good performance as a passive system. Passive solar gains made a reduction of heating energy consumption of 2.3% in winter and 8.9% in equinox.

**Keywords:** earth building; thermal comfort; passive design; monitoring and simulation

#### **1. Introduction**

Earthen architecture historically has been widely used for wall construction around the world. According to Minke [1] earth construction has been used for more than 10,000 years. Today, it is estimated that nearly 30% of the world's population lives in earthen dwellings, not only in developing countries but also in industrialized countries, where using earth as a construction material has raised interest recently, as it is considered an environmentally friendly solution. Particularly, Earthbag (also called Superadobe) is presented as a sustainable, cheap, feasible and comfortable option to improve thermal comfort. Superadobe is a form of Earthbag construction patented and developed by the Iranian architect Nader Khalili, who proposed fundamental rules for the design and building recommendations [2]. Earthbag and Superadobe are building techniques that consist of the use of earth-filled sandbags in order to build structural walls, usually in a dome shape [3]. The dome shape offers more structural integrity and durability than adobe square buildings [4], but also limits its design to 5 m in diameter and one ground floor. The dome shape allows the foundation, load-bearing walls and roof to be built with the same materials and technique. In low-cost buildings, the roof

used to be the part of the building with the highest cost. Thus, this building technique is four times cheaper than conventional techniques [5]. When comparing two low-cost earthen buildings, such as Earthbag dwelling and adobe traditional Burkinabe dwelling, the Earthbag one achieves better thermal performances in hot arid climates. In that case, a combination of night ventilation, roof solar protection, and high-inertia of the Earthbag enclosure lead to an almost total elimination of thermal discomfort during the year [6]. Among their possible uses, the Earthbag building is a good solution to temporary emergency housing, as shown in the construction of 14 Earthbag shelters in the refugee camp of Baninajar [7]. The architect Khalili built them for the displaced Iranians in Iraq after a flood. The project also served to assess the feasibility and cost of building with Earthbag and to evaluate the possibilities of Earthbag shelters in the case of a real emergency [8]. Earthbag building has also been used in cooperation projects, such as the construction of part of the Emsimision Training Medical Center in the Boulmiuogou District, in Ouagadougou, Burkina Faso [9]. After the earthquake of 2010 in Haiti, humanitarian aid of different organizations constructed several buildings with Earthbag and Superadobe techniques, such as numerous houses for those affected inhabitants [10], a medical clinic [11], a school for orphans [12], a community center [13], a shelter for children, and some experimental Earthbag buildings for scientific tests [14]. In Nepal, the Small-Earth partnership built an orphanage with Superadobe [15]. Because of the particular location of the orphanage, in a hilly area with difficult access, the Superadobe system was a good choice, since it allows the use of local resources, such as earth and stones, and it only needed to transport bag rolls to the place.

Previous research includes thermally simulated and monitored raw earthen buildings [16–18], but not Earthbag buildings yet. This research analyzes the hygrothermal performance and comfort of an Earthbag building located in Mediterranean continental climate by experimentation of a real construction and by energy simulation. Passive design strategies are tested, such as the use of high thermal inertia in the enclosure, the collection of direct solar radiation through the glazed openings and the use of natural ventilation. In this research, energy simulation results are also validated with experimental data.

#### **2. Materials and Methods**

#### *2.1. Constructive Characteristics of the Monitored Earthbag Building*

The experimental Earthbag building was designed under bioclimatic criteria. The building was constructed with a dome shape following the Superadobe technique. The dome shape allows for the reducing of the shape factor, which is a smaller surface of the building envelope per same volume compared to a cubical shape. It has a net floor area of 7.07 m2, a circular plant of 3 m of diameter, a height of 3.3 m, an envelope surface of 29.96 m2, an interior volume of 17.67 m3, and a shape factor of 1.7 (Figure 1). The Earthbag walls are 35 cm thick, but the buttress is formed by a double Earthbag (70 cm thick). The Earthbag dome roof has an average thickness of 28 cm. The continuous polypropylene bag contains an earthen mixture of on-site earth and construction sand in a 1:1 proportion. Slaked lime in water was used as stabilizer, in approximately 10% of the total earthen volume. The sieve analysis showed that the earth mixture contained in weight: 0.80% fine gravel, 92.21% sand, 3.42% slime and 3.57% clay. The earth mixture was manually rammed. The building was exteriorly coated with 4 cm thick lime mortar. The floor is made of lime concrete (9 cm thick) and it is directly in contact with the ground, over a waterproofing plastic layer. The main glass opening is the entrance door, which is facing exactly south (exterior window frame of 0.91 × 2 m, with a glazed surface of 1.09 m2) to take advantage of the direct solar gain. Two confronted windows in the east (exterior window frame of 0.8 <sup>×</sup> 0.67 m, 0.25 m<sup>2</sup> of glazed surface) and west (exterior window frame 0.6 <sup>×</sup> 0.35 m, with a glazed surface of 0.06 m2) facades allow crossed ventilation. The position of the windows with respect to the walls is in the interior, which produces small solar protection due to the thickness of the walls. Over the square windows, there is a space that has been insulated with polystyrene (6 cm) and covered with a wooden exterior coating (2.2 cm) (Table 1).

**Figure 1.** Earthbag building, University of Lleida Campus, Spain.


**Table 1.** Materials properties of the Earthbag building.

<sup>a</sup> Source: Catálogo de Elementos Constructivos del Código Técnico de la Edificación (2015) [19]. <sup>b</sup> Source: Measured density taken from the building prototype of the Cappont Campus, Lleida [20]. <sup>c</sup> Source: Estimated thermal conductivity from experimental U-value calculation. <sup>d</sup> Solar heat gain coefficient.

#### *2.2. Location and Climate*

The prototype is located in the Cappont Campus of the University of Lleida, Spain (41.60N, 0.62E; 167 m above sea level). Lleida has a Mediterranean continental climate, classified as BSk by the Köppen climate classification [21]. It is characterized by hot and dry summers and cold and wet winters due to the presence of fog. Rains are low and irregular, with an annual average of 423 mm. The annual average temperature is 15.2 ◦C although there are big differences between summer and winter temperatures and between maximum and minimum daily dry air temperatures.

#### *2.3. Instrumentation and Experimental Campaign*

For the thermal evaluation of the Earthbag building different experiments were performed.

	- -No ventilation is provided to the building, just a base level of infiltrations.
	- - Natural ventilation is provided to the building. Two different scenarios were tested: all-day long and night ventilation.

The experimental setup is instrumented with 18 temperature sensors, 2 relative humidity sensors and the control and data acquisition systems, as it is shown in Figure 2. Monitoring consisted in data collection of interior and exterior air temperatures as well as interior surface temperatures in both south and north walls. The interior temperature and humidity sensor was located in the geometrical center of the dome at 1.50 m high (position 2 in Figure 2). Four additional temperature sensors were located in the center of the prototype at different heights, every 0.80 m (positions 3, 4, 5 and 6 in Figure 2). The north surface wall was monitored with 9 temperature sensors, located in a vertical axis every 0.40 m (positions from 8 to 16 in Figure 2). Moreover, 2 temperature sensors (positions 17 and 18 in Figure 2) were located in the north surface wall, next to the sensor in position 12, covering a triangle surface of 300 cm2. The idea is that sensor 12 follows the 0.4 m distance between all the surface sensors in the north wall. The other two sensors were added next to this 12 sensor drawing a triangle to measure the U-value. The south surface wall temperature sensor was located at 2.10 m (position 7 in Figure 2), above the door. Finally, there was a sensor measuring the external temperature and relative humidity (position 1 in Figure 2), and an energy consumption meter to register the energy consumed by the electric radiator.

**Figure 2.** Sensors location in the monitored Earthbag building.

Temperatures were registered every 5 min by means of a data acquisition system connected to a computer. Air temperature and humidity sensors used were Elektronik device model EE210 (±0.1 ◦C uncertainty for temperature and 0.5% for humidity). PT-100 class B (±0.3 ◦C uncertainty) sensors were used for surface temperatures. The acquisition data equipment consisted of a data logger (model DIN DL-01-CPU), connected to the adapter data logger-computer (model AC-250). The computer software to compile the data was TCS-01. When controlled temperature experiments were carried out in winter, a 1500 W electric radiator was used and its energy consumption was also measured with a Finder E7energy meter.

#### *2.4. Experimental U-Value Calculation*

In order to calculate the U-value of the Earthbag wall, a transmittance test according to [22,23] was performed. The test consists in monitoring indoor, outdoor air and indoor wall temperatures. It is important to locate the surface sensors on the north wall to avoid the solar radiation interfering with the measures, or having a protected sensor. Moreover, for the wall surface temperature reading, three temperature sensors where located in a triangular shape separated about 20 cm from each other, in orderto calculate an average temperature to compute the U-value. It is also important that the indoor air temperature and the outdoor air temperature are as constant as possible, as the importance of the calculation is focused on the heat transfer to the wall. To assure this specification, the experimentation was performed during the indoor controlled temperature scenario and during a fog week. The inner temperature was kept constant with a radiator and the external one due to the presence of all-day-long pervasive fog in Lleida. In this situation, a quasi-steady state hypothesis is justified [24] and the expression to obtain the U-value is the following:

$$
\delta L = \frac{(T\_i - T\_{si})}{(T\_i - T\_c)} \ast h\_{si} \tag{1}
$$

where: *Ti* : Indoor air temperature, ◦C. *Tsi*: Indoor surface temperature, ◦C. *Te*: Outdoor air temperature, ◦C. *hsi*: Heat transfer coefficiewnt of external envelopes, 7.69 W/m2 ◦C [25].

The experimental U-value will allow calculating the thermal conductivity of the Earthbag wall λ1, from the equation:

$$\mathcal{U} = \frac{1}{R\_{\text{si}} + \frac{\mathcal{E}\_1}{A\_1} + \frac{\mathcal{E}\_2}{A\_2} + R\_{\text{sc}}} \tag{2}$$

where: *Rsi*: Interior surface thermal resistance for a vertical facade, 0.13 m2· ◦C/W [25]. *Rse*: Exterior surface thermal resistance for a vertical facade, 0.04 m2· ◦C/W [25]. *e*1: Earthbag wall thickness, m. *<sup>e</sup>*2: Lime mortar coating thickness, m. <sup>λ</sup>1: Thermal conductivity of the Earthbag wall, W/m2· ◦C. λ2: Thermal conductivity of the lime mortar coating, W/m2· ◦C.

#### *2.5. Thermal Lag and Decrement Factor*

The thermal lag (φ) represents the time that elapses between the indoor air temperature maximum value and the outdoor maximum value. The decrement factor (μ) is the reduction of the temperature range of both measures (Figure 3).

**Figure 3.** Definition of the thermal lag and the decrement factor of a sinusoidal heat wave. Source: adapted from Yáñez, 2008 [26].

Assuming an external sinusoidal wave temperature, some formulas are presented for the calculation of the thermal lag and the decrement factor in homogeneous walls, knowing the diffusivity of the material (α), the thickness of the wall (*l*) and the period of the wave.

In this case the heat flow is supposed to be transmitted only in the normal direction to the wall, neglecting the effects of the edge. Likewise, it is assumed that the variation of the temperature inside the wall depends only on the external conditions, neglecting any variation that may be generated inside (semi-rigid solid).

As both parameters depend on the period of the wave, the following formulas correspond to a period of 24 h:

$$\mu = \exp.\left(-0.362 \ast l \ast \sqrt{1/\alpha}\right) \tag{3}$$

$$
\Phi = 1.38 \ast l \ast \sqrt{1/\alpha} \tag{4}
$$

where: *l*: Thickness (m). α: Material diffusivity, α = <sup>λ</sup> <sup>ρ</sup>∗*Cp* [m2/s].

*2.6. Simulation*

A numerical model of the Earthbag prototype designed with EnergyPlus was defined and used for comparisons. Open Studio was used as the graphical user interface. Open Studio does not allow for the creation of dome shapes; this is why a polygonal dome has been drawn for the dome building. Shadow elements were drawn, such as the door buttress and the thickness of the walls producing shadow over the windows. The default heat balance algorithm based on the conduction transfer function (CTF) transformation and 6 time steps per hour for the simulation are applied. CTF is a widely used numerical method to calculate transient heat conduction in Building Energy Simulation tools. It is preferred to the finite difference method thanks to the smaller computational time required [27].

Considerations in the energy simulation:


The initial hypotheses are:


Different scenarios during summer and winter periods were tested. Each test was design to give an answer to the objectives listed in Table 2. In test #2, the simulation of the Earthbag building has been compared in addition with an equal Earthbag building with no glazed openings.


**Table 2.** Testing scenarios of monitoring and simulation of the Earthbag prototype.

#### **3. Results**

In this section, firstly steady-state and dynamic parameters are presented. Experimental data from the monitoring is taken to calculate the thermal transmittance and the conductivity of the Earthbag walls. Secondly, results of the experimental analysis and simulation analysis are presented. The experimental data are presented to analyze the effect of the air stratification. The monitoring and simulation free floating results of temperature and solar radiation data are presented to, in one hand, validate the simulation with the experimental data and, in the other hand, to analyze the thermal inertia and the solar heat gains in winter solstice, equinox and summer solstice. The monitoring and simulation results of power consumption are presented in winter conditions to analyze the energy consumption of the Earthbag building. The monitoring and simulation temperature results of natural ventilation in free floating conditions are presented to analyze its effect and validate the energy simulation.

#### *3.1. Steady-State and Dynamic Thermal Parameters*

Figure 4 shows indoor and outdoor air temperatures. For indoor temperatures, air temperature in the geometrical center of the dome (position 2 in Figure 2) is represented with the average value. The indoor north surface temperature monitored (average of sensors 12, 17 and 18 in Figure 2) in quasi-steady state conditions [24] with the average value is also plotted. Moreover, the U-value calculated with the average of the indoor temperature and the uncertainty of this calculation according

to the sensors accuracy, are also included. The uncertainty for the U-value was determined to be ±4%, applying the standard method for uncertainty propagation [29].

**Figure 4.** Interior air and surface temperatures, exterior air temperature and U-value calculated of the Earthbag wall. Data taken from experimental monitoring.

As shown in Figure 4, indoor and outdoor temperatures are almost constant, with an indoor temperature average of 22 ◦C and an indoor north surface temperature average of 15.5 ◦C. Outdoor temperature is around 3 ◦C in days of persistent fog, oscillating only 2 ◦C throughout the day. The U-value calculated for the Earthbag wall of 35 cm with an exterior lime coating of 4 cm has an average value of 2.7 W/m<sup>2</sup> K. According to this experimentally obtained U-value, and to the Equation (2) described in the methodology section, the thermal conductivity of the Earthbag material is 2.18 W/m·K.

The theoretical thermal lag and decrement factor are calculated considering a homogeneous Earthbag wall of 35 cm (with no exterior coating). According to Equations (1), (3) and (4) and the Earthbag properties (Table 1), the corresponding values are shown in Table 3.

**Table 3.** Theoretical results of the steady-state and dynamic thermal parameters for the Earthbag wall.


#### *3.2. Experimental and Simulation Results*

#1. Air stratification inside the Earthbag dome.

The air stratification testing scenario shows an increase of 1.4 ◦C from the bottom to the top of the dome in summer and 2.8 ◦C in the equinox. The surface temperature keeps more stable, oscillating in less than 1 ◦C (Figure 5).

**Figure 5.** Air stratification inside the Earthbag dome during at solar noon (2 p.m., on 10 June 2018). (Temperature values in ◦C).
