Building Heat Load Research Articles

A predictive model of long-term performance assessment of Ground Source Heat Pump (GSHP) systems in Japanese regions

This study predicts the long-term performance of ground source heat pump (GSHP) systems in different climatic regions of Japan. The research model calculates the dynamic heat load of a typical Japanese building in various cities using their meteorological data, considering different schedules for heating and cooling seasons. The model aims to analyze the performance degradation of the GSHP system after 15 years of operation due to the imbalanced heating and cooling loads. Additionally, it compares ground temperature changes, coefficient of performance (COP) variations, and annual operation times based on different thermal imbalance ratios (TIR) for the case studies. The results demonstrate that TIR serves as an appropriate factor for comprehensively evaluating system aspects, including GHE length, ground temperature, and COP degradation, across various climates. It was found that ground temperatures increased by a maximum of up to 4 °C in cooling-dominated regions due to higher thermal imbalance accumulative loads, while it decreased to around 1 °C in heating-dominated regions. The model also revealed that the COP decreased in the dominant operation for each case study due to induced thermal imbalance and soil temperature change. The thermal imbalance caused the system cooling COP decreased by 0.015/year in the worst case (Okinawa City) and 0.006/year in average for other cases.

Impact of future climate change on energy consumption in residential buildings: A case study for representative cities in Japan

This study examines the impact of future climate change on energy consumption in residential buildings in Japan, taking into account both the current and future typical meteorological year (TMY) database. TMY data is widely used in the building sector for energy performance assessments, system sizing, and optimization. To investigate the impact of current and future TMY databases on energy consumption of buildings, the study predicts the cooling and heating loads of residential buildings in four cities of Japan under the current (2020 ver.) and future (2086 ver.) TMY databases. Results show that the use of future TMY database leads to significant changes in air-conditioning heat loads of residential buildings in Japan. It finds that the annual cooling load in residential buildings of four cites of Japan increases by approximately 12% for Tokyo, approximately 9% for Toyohashi, approximately 8% for Osaka, and approximately 7% for Nagoya, respectively, however the annual heating load decreases by approximately 8% for Tokyo, approximately 9% for Toyohashi, approximately 6% for Osaka, and approximately 5% for Nagoya, respectively, compared to the current TMY database. This finding suggests that climate change will have a significant impact on future building energy consumption, and proposes energy-efficient building design as a solution to mitigate the impact of climate change on building energy consumption.

Development and characterization of a reflection paint with nano-ilmenite material for energy-saving in buildings

Low-rise buildings are suitable for cooling roofs to save energy in hot climates. Thermal reflection paints using titanium dioxide micronized particles, nano P-25, and nano-ilmenite were compared with commercial thermal reflection paints to evaluate thermal performance. nano-ilmenite material was synthesized using a hydrothermal technique. The chemical composition, size, and structure were characterized by XRF, SEM, and TEM. The color was measured according to ISO 1164–4. The reflectance in the UV–VIS–NIR region was measured according to ASTM E 903. The thermal performance was determined using a simulation test with an IR lamp and a field test. The results showed that they had been synthesized to nanofiber structure, increased titanium dioxide content (62.1%), and demonstrated the presence of hematite, metal oxides, and ceramics. In all thermal reflection paints, the results showed that there was a similarity in the lightness mode (L*) and blue-yellow mode (b*), which presents yellowness (0.98, 0.94) in thermal reflection paints using nanoparticles. The NIR and TSR reflectances of thermal reflection paint with nano-ilmenite were highest (56.8, 59.8%). The thermal performance properties measured using an IR camera for both samples (nano P-25 and nano-ilmenite) showed substances with low surface temperatures, indicating that they can block IR radiation. Therefore, thermal reflection paints using nano-ilmenite, which also acts as a barrier, can reduce the surface temperature. Improving thermal performance, which inhibits heat transfer or sunlight entering a building, can reduce the building's heating load demand.

Proposing hybrid prediction approaches with the integration of machine learning models and metaheuristic algorithms to forecast the cooling and heating load of buildings

Accurate prediction of heating and cooling loads in residential buildings is crucial for both researchers and practitioners. This study employs advanced forecasting techniques, utilizing support vector regression and extreme gradient boosting methods, enhanced by state-of-the-art metaheuristic algorithms. The metaheuristic optimizers applied include the Satin Bowerbird Optimizer, Ant Lion Optimizer, Artificial Ecosystem-Based Optimizer, Slime Mold Optimizer, Moth-Flame Optimizer, and Particle Swarm Optimizer. These optimizers are strategically used to refine the forecasting models, optimizing their parameters for increased precision. To evaluate the models, a suite of statistical indices is used, including mean square error, root mean square error, Mean Absolute Percentage Error, Mean Absolute Error, Relative Absolute Error, coefficient of correlation, Coefficient of Determination, and Normalized Mean Square Error. This comprehensive analysis assesses the effectiveness of the forecasting methods. The study's findings highlight that the combination of the Satin Bowerbird Optimizer and extreme gradient boosting is particularly effective, achieving high coefficients of determination for cooling and heating load predictions (0.95826 and 0.938048, respectively). This hybrid algorithm shows remarkable performance, with minimal error values and consistent convergence across various datasets. In summary, this research underscores the efficiency of integrating sophisticated machine learning models with metaheuristic optimization techniques, particularly identifying a potent hybrid algorithm for the accurate prediction of heating and cooling loads in residential buildings.

Effects of urban vegetation on microclimate and building energy demand in winter: An evaluation using coupled simulations

As an important landscape, the evergreen vegetation is prevailing, even in cold region cities. However, the impacts of promoting vegetation on microclimate and building heating load (BHL) are still inexplicit. This study validated the feasibility of a new-developed coupled model that linked ENVI-met and EnergyPlus, for predicting the microclimate and BHL in winter. The impacts of vegetation configuration changes on the performances of three typical urban blocks in Xi'an, a high-density city in the cold region of China, were then evaluated. When the leaf area index of green lands increased from 0.79 to 4.77 m2/m2, a decrease of 3.27 °C on mean radiant temperature, a decrease of 0.48 m/s on wind speed, and an increase of 0.47 °C on air temperature at the block scale could be observed. To the three blocks, the BHL of thermal zones in the height range of tree canopy increased by 7.16 %, 1.52 % and 3.57 % during a sunny day, and decreased by 2.80 %, 0.55 %, and 0.06 % during a cloudy day maximally. Overall, the evergreen vegetation produced negative impacts on the building energy efficiency of cold region cities. For urban planners, it's advisable to concentrate evergreen plants in high-rise blocks to mitigate such disadvantages.

Steady- and Transient-State CFD Simulations of a Modified Barra–Costantini Solar System in Comparison with a Traditional Trombe–Michel Wall

Passive solar systems are one of most important strategies to reduce the heating loads of buildings. The Trombe–Michel (TM) wall and its variants are some of the better-known structures in the field of solar systems. An alternative to the TM wall is the Barra–Costantini (BC) system. In the present paper, CFD numerical simulations, both in steady and transient states, of modified BC and TM walls were carried out in the winter season. Different interspace thicknesses were simulated in order to evaluate their effects on the temperature field and air velocity, and the numerical results were compared among them. It was found that the BC system offers greater hot air flow compared with the TM wall; the mass flow rate increased up to 43% in the BC system and up 28% in the TM system when the interlayer thickness was increased by 500%. The transient simulations (100 h simulated) demonstrated that the dynamic response of the BC wall was shorter than that of the TM wall, even when the TM wall was simulated with initial thermal conditions that were more advantageous than those for the BC wall. The BC system reached a periodic stabilized regime within 24 h, whereas the TM system failed to stabilize in 100 h. The results show that for both TM and BC structures, the interlayer thickness scarcely influenced the temperature of the environment reached (the temperature peak increased up to 3–4% as the interlayer thickness was increased by 500%), while larger air speed changes were observed in the BC system in the transient state compared with the TM system. Thus, in the TM system, the outlet air velocity was practically constant as the interlayer thickness was increased; in contrast, the outlet velocity peak increased up to 50% in the BC system. Moreover, the BC wall presented a quicker response to satisfy the ambient thermal loads.

Numerical investigation of the deep borehole heat exchanger in medium-depth geothermal heat pump system for building space heating

The deep borehole heat exchanger (DBHE) is a vital part in medium-depth geothermal heat pump system (MD-GHPs). In comparison to other forms of heat exchangers, the DBHE has the following outstanding advantages: the small footprints, the little damage to rock as well as the high efficiency. However, numerous DBHE models were fixed variable values such as flow rate or inlet fluid temperature and there were few studies combining the user heating load variation with the temperature fluctuation on the DBHE side. In order to fill this gap, a DBHE heat transfer model with dynamic building heating load constraint is developed. In this heat transfer model, the load boundary condition is the hourly flow rate and the difference in inlet and outlet water temperature, and the outlet water temperature is obtained through iteration. In addition, this model is validated by a building space heating experiment data in Harbin and guaranteed its correctness. The average heat exchange per meter at the coldest heating moment also ensures the viability of MD-GHPs, which is 141.5 W/m. A series of typical coordinates are used to analyze the effect of DBHE on the surrounding rock, and the concept of DBHE recommended insulation length are proposed. One single DBHE with 2000 m burial depth possesses a temperature effect radius of 13.0 m for one heating period. Besides, some critical conclusions are as follows: For the structural parameter, the recommended insulation length is 178 m. For the operating parameter, there is a minimum value for the total power consumption and 18 m3·h−1 is the recommended optimal flow rate in this situation. For the geological parameter, raising the geothermal gradient can dramatically improve inlet fluid temperature (IFT) and outlet fluid temperature (OFT), and enhance the ability of DBHE to carry the building heating load. When maintaining the same IFT at the coldest moment, the geothermal gradient of 0.04℃·m−1 is 1.33 times more capable of carrying loads than the 0.03℃·m−1. This paper can provide basis for a series of DBHE thermal performance in practical space heating engineering applications of MD-GHPs.

A Hyper Parametrized Deep Learning Model for Analyzing Heating and Cooling Loads in Energy Efficient Buildings

The huge increase in energy consumption in recent decades, has made it cumbersome to anticipate energy usage in the residential sector. However, despite substantial advancements in computation and simulation, the modelling of residential building energy use is still in need of improvement for efficient and reliable solutions. To this end, the overarching objective of this research study is to construct a self-adaptive model (HBO-DL) for predicting the amounts of heating and cooling loads in residential buildings. The developed HBO-DL model is envisioned on coupling Bayesian optimization with deep learning neural network. Five statistical metrics of mean absolute percentage error (MAPE), root mean squared error (RMSE), root mean squared logarithmic error (RMSLE), mean absolute error (MAE) and normalized root mean squared error (NRMSE), are leveraged to measure and test the accuracies of the developed HBO-DL. Analytical results explicated that the developed HBO-DL model can endorse informed decision-making and foster energy conservation in built environment.

Theoretical cost and energy optimisation of a 4th generation net-zero district heating system with different thermal energy storage technologies.

A model was used to simulate and optimise a speculative district heating (DH) system for an existing urban area in Loughborough, UK. Heat was supplied using only renewable heat sources (RHS) with long and short duration thermal energy storage (TES) systems used to address mismatch between heat generation and demand. For an 18-month simulation period, a) building heat loads, b) heat generated, c) heat flow to and from stores, and d), heat losses from the stores and network were calculated using hourly weather data, network and building characteristics. RHS included evacuated-tube solar thermal collectors (ETSTCs), and ground and air source heat pumps powered by electricity generated by solar-photovoltaic (PV) and Wind turbines (Wind). The effect of long-term TES system volume and charging temperature on the calculated DH system cost per dwelling, and the calculated DH system energy efficiency was assessed. The minimum cost of the DH system was obtained by modifying the capacity of installed Wind and PV, while ensuring that heat demands were met in every hour of the simulation. A minimum cost per dwelling of £12,278 was predicted with an energy efficiency of 84.84 % for a long term TES volume of 15,000 m3 and temperature of 60 °C.

Investigation of energy performance of nanotechnological material and double-skin facade system in office buildings in Turkiye

ABSTRACT Due to the increasing energy demand around the world, new concepts have emerged in order to reduce it in the building sector. In this context, double-skin facade systems have increasingly been used due to their effectiveness in reducing the heating load, cooling load, and primary energy demand of buildings. In this study, an office building located in the industrial zone in Kocaeli/Gebze was examined, and the effects of nano-technological materials to be added to the building and double-skin facade systems on the energy performance of the building were determined using DesignBuilder and EnergyPlus simulation programs. As a result; with the thermochromic glass and vacuum insulation panels used in the nano scenario, the heating load was reduced by 70% when compared with the base model. In the double skin scenario implemented on the east–west facade, heating load was decreased by almost 25% and cooling and primary energy demand were decreased by nearly 20% as compared to the base model. Whereas, an efficiency of 20% in heating load, 40% in cooling load and almost 30% in primary energy demand was observed in the nano double skin scenario when compared with the base model.

Modified shape coefficient of building with Thermal considerations: concept, experiment and Application in solar-enriched areas

ABSTRACT The shape coefficient of building (SCB) characterizes the correlation between building shape and building energy consumption. Strictly restricting the SCB was harmful to reducing energy consumption, neglecting the potential benefits of solar radiation. A novel concept and determining method of modified shape coefficient of building (MSCB) were proposed based on the energy dissipating and solar heat gain. Field measurements of the indoor air temperature of model buildings for solar radiation, building orientations, and U-values under natural and heating conditions were conducted in western Sichuan. Results demonstrated that with identical SCB, the indoor thermal environment was substantially affected by the above parameters, proving the one-sidedness of SCB simultaneously. A simulation method verified via the experiments was employed to forecast the dynamic heating load of a real building. The influence of building orientations on the MSCB under actual climatic conditions from different time scales was discussed. Results illustrated that the SCB could be reduced by 34.6% in the present case. The monthly correction factor trend and the heat consumption were consistent. The restriction on SCB could be liberalized moderately even if the heating demand was tremendous. This research could provide a reference for the energy-saving design of passive buildings.

Enhancement Techniques for the Reduction of Heating and Cooling Loads in Buildings: A Review

The building sector is rated as a big consumer of electric energy and emissions, responsible for about 40% of final electric energy consumption. As a result, the Paris Agreement 2015 set a goal for buildings and the construction sector to reach a nearly zero-carbon stage by 2050. This urged most countries to create regulations for the construction sector and invest in energy efficiency programs. The present paper aims to present an updated review of building energy-saving solutions and techniques to contribute to carbon emission mitigation in the building sector. The high energy consumption of a building is mainly due to heating and cooling, which is directly related to the thermal properties of the materials used. Natural ventilation and illumination are other aspects that contribute to the high energy consumption. Considering these issues, the review covers energy-efficient construction materials such as mortars, concrete with PCM, new construction materials with PCM such as 3d printing concrete and geopolymer concrete, and bricks usually used in buildings. Also, the review covers the methods and solutions for energy saving for building heating and cooling. Since transparent windows and façades are essential for structures, their thermal and visual performance is crucial. Established and under-development techniques for windows and façades are presented and discussed. Walls and roofs are usually rated at the top of the weak barriers against a building's heat losses and energy gains. The present paper reviews existing and still under research and development techniques to improve the thermal performance of walls and roofs, such as cool roof and cool walls, walls and roofs with phase change materials (PCM), and ventilated walls and ceilings.Some authors’ comments are presented at the end of each topic. Some possible opportunities for future research and developments are also presented.

Experimental and simulation study on the modified shape coefficient of building integrated thermal considerations for heating buildings in solar-enriched areas

The shape coefficient of building (SCB) is a measure of correlating a building’s shape and its energy consumption. However, strictly adhering to restrictions the SCB may not be conducive to reducing energy consumption, as it disregards the potential benefits of solar radiation. To address this, a novel concept and method for determining the modified shape coefficient of building (MSCB) have been introduced in considerations of heat loss and solar heat gain. Field measurements were conducted in western Sichuan to establish a comprehensive understanding of the factors influencing indoor thermal environments, involving collecting indoor air temperature data for model buildings under natural and heating conditions, including solar radiation, building orientations, and U-values. Results demonstrated that maintaining an identical SCB did not consistently ensure similar indoor thermal environments. The indoor thermal environments were significantly influenced by building orientations, U-values, and solar radiation, highlighting the limitations of relying solely on SCB. A simulation method validated through experiments was utilized to predict the dynamic heating load of an actual building. The influence of building orientations on the MSCB across different time scales and actual climatic conditions was investigated. Results illustrated a potential reduction of the SCB by up to 32.4% in the specific case. This reduction correlated with trends observed in the monthly correction factors and heat consumption, suggesting that moderate relaxation of SCB restrictions could be viable even in scenarios with high heating demands. This research contributes valuable insights into the realm of energy-efficient building design.

Heat load-carrying capacity of surface water source and its heating load matching characteristics

Meteorological parameter changes not only affect the heat load-carrying capacity of the surface water, but also affect the heat load demand of the building, thus changing the heat load matching between the water source heat pump system and the users. In order to study the climatic adaptation of the heat load-carrying capacity, this paper establishes an analytical model of the water body's heat load-carrying capacity and the building's heat load with the solar radiation intensity, the change of the water body and the heat extraction time as variables. Experimental and simulation studies on the heat load-carrying capacity of lake water were carried out under the winter climate conditions in hot summer and cold winter areas. The results show that increasing the depth or area of the lake can improve the heat load-carrying capacity. Solar radiation is an important factor affecting the temperature of surface water body, the stronger the solar radiation the greater the temperature rise of the water body. Changes in outdoor air temperature have a significant impact on water temperature, and a sudden drop in outdoor air temperature will result in a sharp drop in water temperature. When the water body area was 3000 m2 and 50000 m2, the maximum integrated temperature drop of the water body was 15.0℃ and 9.7℃, which was 2.02 and 4.85 times of the temperature drop of heat extraction, respectively. Reduced heat load-carrying capacity of the surface water body or decreased COP of the heat pump is an important reason why the surface water source heat pump heat supply fails to meet the user's heat demand in winter.

Evaluation of renovation strategies for existing residential districts in Shanghai based on coupling calculation of microclimate and energy consumption

This study aims to analyze the impact of renovation strategies on the air-conditioning load in existing residential districts in Shanghai by using the Grasshopper platform coupled with ENVI-met and EnergyPlus. These strategies include increasing the greening rate (G), improving pavement reflectivity and permeability (P), increasing the reflectivity of external wall coatings (C), increasing roof material reflectivity or applying green roofs (R), and improving the thermal performance of windows (W). The results show that microclimate and weather station data cause a 13.9% maximum deviation in energy consumption simulations. W and R4 significantly reduce the air-conditioning load throughout the year. W contributes to a decrease of 34.8 kWh in the cooling load and 110.4 kWh in the heating load. When the application level of P is low, and that of other strategies is high, the cooling load of the target building reaches its minimum. The minimum cooling load scenario is 302 kWh (25.1%) lower than the maximum. When the application level of G and C is low, other strategies have a high application level, and green roofs are used, the heating load of the target building is minimal. The minimum heating scenario is 386 kWh (29.3%) lower than the maximum.

Building energy demand of urban blocks in Xi'an, China: Impacts of high-rises and vertical meteorological pattern

High-rise buildings have been promoted in high-density cities. The impact degree of urban high-rises on the building energy demand at the urban block scale is determined by many factors, including the effects of the vertical meteorological pattern (VMP), which have been discussed insufficiently. In this study, the VMP of urban areas in Xi'an, China was observed, and a building energy simulation model that revealed the “urban block form–outdoor climate–building energy” nexus was developed with the URBANopt plugin on the Grasshopper platform. The impacts of the VMP on the building heating and cooling energy demands of different urban blocks were then predicted. In January, the VMP led to building heating load increases of 0.7–1.6 kWh/m2 (16–52%) and 0.0–1.1 kWh/m2 (0–35%) in the residential and mixed-use blocks, respectively. In July, the ranges of the building cooling load changes produced by the effects of the VMP were 0.0–0.1 kWh/m2 (0–1%) and −0.3–0.2 kWh/m2 (−2%) in the residential and mixed-use blocks, respectively. Furthermore, the VMP affected the relationship between the urban block form and the building energy demand. Based on the relationship, we provided form-optimization advice for different functional urban blocks. The urban high-rise mechanisms in high-density cities and their impacts on the building energy demand were discovered in this study. This can help decision-makers to optimize urban planning while promoting urban sustainability.

Methodological aspects of optimization of energy consump-tion costs of low-rise buildings with a built-in air heat pump

Objective. The rational use of fuel and energy resources is one of the urgent problems. One of the promising ways to solve this problem is the use of environmentally friendly energy-saving air source heat pumps based on the use of non-traditional energy sources. The use of such innovative energy-saving technologies requires not only modern architectural and planning, design and engineering solutions, but also sound methodological recommendations for as-sessing their energy efficiency and feasibility. The article presents a methodology for calculating the consumed electrical energy and heat generation, as well as the seasonal indicator of energy efficiency in the operation of an air source heat pump.Method. The technique takes into account, as input data, experimental data from tests of high-voltage heat pumps in terms of heat output depending on the change in the temperature of the source of low-grade heat and the temperature regime of the coolant.Result. The need to develop a methodology is due to the shortcomings in the regulatory documents of the methodological basis for optimizing energy costs and heat generation by heat pumps. The accuracy of calculating the operating costs that will occur during the heating period of the air source heat pump (AHP) depends on the correct accounting of the required heat load of a low-rise building and the consumption of electricity to ensure it.Conclusion. This technique can be applied not only to calculate the annual operating costs of using a heat pump and determine the structure of energy consumption (thermal and electrical), but also to optimize the energy parameters of HHP with an assessment of the energy efficiency of different options for their configurations.

A Numerical Investigation of the Thermal Performance of a Gabion Building Envelope in Cold Regions with a Mountainous Climate

Applying rock-filled gabion to buildings in cold regions with mountainous climates has multiple potentials, such as utilizing rock resources, improving building sustainability and saving building energy. Therefore, it is necessary to analyze the thermal performance of gabion buildings. Based on the CFD method, this paper establishes a numerical model of buildings with gabion enclosure structures, analyzes the influence of the gabion structure on the external convective heat transfer coefficient (CHTC), wind pressure, air infiltration, room temperature and building load, and further uses the building energy consumption simulation method to analyze the heat load of gabion buildings. The results showed that the adverse impact of climate on the building thermal performance is significantly diminished by the gabion. Under different weather conditions, the CHTC, the maximum wind pressure difference on the exterior surface, and the air infiltration rate are reduced by different rates. Further, the room base temperature increases throughout the heating season, and the maximum heat load and the cumulative heat load of the building are, respectively, reduced by 10.6% and 24.8%. This work revealed that the gabion is an eco-friendly and adaptive measure to improve thermal performance and indoor thermal comfort.

Impacts of UHI on Heating and Cooling Loads in Residential Buildings in Cities of Different Sizes in Beijing–Tianjin–Hebei Region in China

The heating and cooling energy consumption levels of urban buildings account for a large and rapidly growing proportion of the total end-use energy consumption of society. The urban heat island (UHI) effect is an important factor influencing the spatiotemporal variations in the heating and cooling energy consumption levels of buildings. However, there is a lack of research on the impact of the UHI on the heating and cooling energy consumption of buildings in cities of different sizes in the Beijing–Tianjin–Hebei urban agglomeration, which is the most urbanized region in northern China. We selected rural reference stations using the remote sensing method, and applied an hourly data set from automatic weather stations, to examine the impact of the UHI on the typical residential building heating and cooling loads in three cities of varied sizes in the Beijing–Tianjin–Hebei urban agglomeration through building energy simulation. The main conclusions were as follows. As the UHI intensity (UHII) increased, the heating load difference between urban and rural areas decreased, while the cooling load difference between urban and rural areas increased in the cities. The average daily heating loads in the urban areas of Beijing, Tianjin, and Shijiazhuang were 8.14, 10.71, and 2.79% lower than those in their rural areas, respectively, while the average daily cooling loads in the urban areas were 6.88, 6.70, and 0.27% higher than those in their rural areas, respectively. Moreover, the absolute hourly load differences between urban and rural areas were significantly larger during the heating periods than during the cooling periods, with the former characterized by being strong at night and weak during the day. During the peak energy load period, the contribution of the UHI to the peak load of residential buildings varied between the cities. During the stable high-load period, from 18:00 to 07:00 the next day in the heating periods (from 18:00 to 05:00 the next day in the cooling periods), the hourly loads in the urban areas of Beijing, Tianjin, and Shijiazhuang were 3.15 (2.48), 3.88 (1.51), and 1.07% (1.09%) lower (higher) than those in their rural areas, respectively. Our analysis highlights the necessity to differentiate the energy supplies for the heating and cooling of urban buildings in different sized cities in the region.

Transfer learning-based adaptive recursive neural network for short-term non-stationary building heating load prediction

Building energy consumption is a non-stationary time series, and its distribution law changes over time. Traditional machine-learning models are prone to model shift, which leads to a reduction of their prediction accuracy when applied to building energy consumption forecasting. Therefore, in this paper, an adaptive neural network framework is proposed for non-stationary building energy consumption prediction based on transfer learning, in which the training datasets are divided into the most dissimilar periods, and based on the transfer learning mechanism, the different periods of data are learned to obtain the minimum overall loss to improve the model generalization. Taking LSTM and GRU models as examples, adaptive long short-term memory (adaLSTM) and adaptive gated recurrent unit (adaGRU) building energy consumption prediction models are established. The models were trained and verified using heating load data from Xi'an, China. The results show that compared with LSTM model, the coefficient of determination, root mean square error, the coefficient of variation of the root mean squared error and mean absolute error of the adaLSTM model were improved by 0.61%, 37.78%, 38.05% and 30.69%, respectively, and the over-fitting degree was reduced by 227.7%. Compared with the traditional GRU model, the corresponding evaluation indexes of adaGRU were improved by 2.50%, 70.58%, 70.64% and 68.83%, respectively, and the over-fitting degree was improved by 505.7% points. The adaptive recurrent neural network framework proposed in this paper is a generalized approach which can be applied to other non-stationary time series prediction models.