1 Introduction

The energy efficiency of household products is enhanced by the rapid progress in science and technology. When energy efficiency is improved, the total energy consumption decreases. However, many studies in the field indicate that the reduced energy costs due to energy efficiency improvements lead to increased energy use, i.e. the so-called rebound effect [1,2,3]. In recent years, smart homes have attracted a substantial amount of investment in China. According to the survey by Richter (2015), in 2020, the number of smart homes will be seven times what it was in 2015, i.e. an increase from 0.3 million in 2015 to 2.1 million households in 2020. Smart homes have a significant influence on both the lifestyle of their occupants and their energy use patterns. Currently, many studies on the rebound effect have been published in energy journals. However, the existence of rebound effect in smart homes, its impact on the lifestyle of occupants, the size of such rebound effect, and measures to reduce it are still unknown.

Therefore, this study investigates the rebound effect in smart homes and measures the size of such rebound effect. This research study also explores the methods to reduce the rebound effect size, and investigates the energy-saving strategies to minimize the rebound effect.

2 Literature Review and Research Question

2.1 Energy Saving in Smart Homes and Rebound Effect

With the rapid development of smart homes, the smart home products improved the energy efficiency of home appliances, providing a new way for energy saving. Lu et al. carried out a study in 2010, using simple sensor technology in a household to automatically sense the human positions and sleep modes. Those modes were applied to automatically turn off the air conditioning system to save energy. It was found that the adopted method saved an average of 28% of the energy with a cost of only 25-dollar for the sensor [4]. It has been reported that installing an automated system to manage the energy consumption of household appliances and other equipment could reduce the energy consumption by 18.7% [5]. Smart homes use automatic sensitivity to automatically adjust home systems, ensuring both higher comfort levels and less energy consumption. However, energy efficiency improvements offered by smart homes brings about the rebound effect. In more advanced energy-saving environments, people tend to use more energy [6, 7]. Kavousian, Rajagopal, and Fischer (2013) carried out a survey on 1628 American households with smart electric meters [8]. The study results indicate that the existence of energy-efficiency-promoting appliances (such as intelligent temperature controllers and heat insulating materials) leads to a slight increase in electricity consumptions by those households.

The size of the rebound effect is defined as the offset between the percentage of improved energy efficiency and the percentage of increased energy consumption [2]. Some studies indicate that the size of the direct rebound effect is not greater than 50% [2, 6], and that lower electricity bills are one of the main factors for the rebound effect [3]. In 2016, Wang et al. evaluated the rebound effect in seven major industries in Beijing. They found that the short-term rebound effect was 24%–37% and the long-term rebound effect was up to 46%–56%, indicating that the rebound effect on energy use should not be underestimated [9]. According to the estimates by Energy Information Administration (EIA), household electricity consumption accounts for 37% of the total electricity consumption [10]. With the growing demand for smart homes and the significance of household electricity consumption, the rebound effect in smart homes and its effect size need to be further discussed. Hence, the first research question investigates the existence and the size of smart homes rebound effect, as follows:

Research Question 1:

Does the rebound effect exist in smart homes and what is the size of the rebound effect?

2.2 Energy Saving Strategies in Smart Homes

Afterward strategy, as an important energy-saving strategy, is established on key premises upon which the hypothesis that “the existences of positive and negative consequences affect behaviors” is derived [11]. The literature on afterward strategy is mostly focused on “feedback” mechanisms. Some studies indicate that real-time feedback can effectively reduce household energy consumption by 4%–9%, which can significantly enhance household energy saving [12, 13]. Smart homes use various technologies that offer effective energy consumption feedback. Such technologies enable monitoring household appliances at any time and supports real-time view. Chetty et al. (2008) pointed out that families are most likely to ignore their energy use patterns, and they hope to acquire real-time information about their energy consumption in order to save money and make their homes comfortable and environment friendly [14]. Moreover, the timing of providing feedback is crucial [15]. Some studies show that providing proper feedback and real-time information at the appropriate time can reduce the electricity consumption of each household by more than 10% on average [16, 17]. It is worth considering that whether and how users will adjust their energy use when they are informed of the existence of the rebound effect. It is also confirmed that using the proper information-providing strategy contributes to energy saving [11], and information feedback is the simplest and direct way to inform users of the rebound effect. Will providing feedback to the users and recognizing the increase in their electricity consumptions change the size of the rebound effect? Thus, the second research question will explore the strategies to reduce the rebound effect, as follows:

Research Question 2:

How will the informed users adjust their electricity use in order to reduce the rebound effect in smart homes?

3 Experiment 1

Experiment 1 examined the changes of participants’ electricity use when electricity bills were reduced. The aim of Experiment 1 is to investigate whether the rebound effect exists in smart homes and measure the size of the rebound effect. A within-design experiment was conducted.

3.1 Experimental Material

In order to examine the changes in the electricity use of the participants, each participant was provided with a situational simulation. There were three rooms in the situational simulation of household electricity use, i.e. living room, kitchen, and bedroom. Each room was configured with luminaires, air conditioning and three kinds of household appliances (selected by the participants). The researcher collected 19 commonly used household appliances (see Table 1), and their powers was evaluated based on existing products in the market for the reference. Participants rated the use frequency of each appliance with a 5-point scale, in which 1-point means rare use and 5-point means the appliance is in use at the time. According to calculations, participants would see monthly electricity bills of each appliance.

$$ \begin{aligned} {\text{Monthly electricity bills of each appliance}} & \, = \,{\text{appliance power}} \times {\text{use frequency}} \\ & \, \times \,0.48 \, \left( {\text{Yuan, the unit price of electricity}} \right) \\ & \, \times \,30 \, \left( {\text{days of a month}} \right). \\ \end{aligned} $$
(1)
Table 1. Calculation of the power consumed by household appliances
Full size table

3.2 Experimental Procedures

Sixteen students (seven males and nine females) were invited to participate in Experiment 1. All of them were aware of the electricity use conditions and bills at their own homes. The average age of the participants was 23.8. Participants individually took part in the experiment. The experimental procedures were shown in Fig. 1. First, the researcher introduced the experimental procedure to the participants. The experiment involved three phases of electricity use. Then, the participants selected three household appliances commonly used in the living room, kitchen, and bedroom, respectively. In the first phase of electricity use, the participants conducted electricity use (i.e., rated the use frequency of each appliance with a 5-point scale) of six appliances (two illustration devices, one temperature control device, and three household appliances) in each room. Then, the participants were informed that there was an energy-saving assistant in their smart home which can help reduce electricity bills. The energy-saving assistant could automatically adjust the brightness of lamps at home and the temperature of air conditioning system. In the second phase, the participants conducted electricity use when there was an energy-saving assistant in the home. After the second phase, the participants were shown the monthly electricity bills of each appliance. The participants were provided electricity use feedback by diagrams, which compared the difference in electricity bills between the first and second phase of the electricity use. Then, the participants conducted the third phase of electricity use. At last, the participants were interviewed about the opinions about the functions of the energy-saving assistant and the driving forces to reduce the rebound effect.

Fig. 1.
figure 1

Experimental procedure of Experiment 1

Full size image

3.3 Results

The data of the three phases of electricity use was normally distributed. A paired t-test was conducted to analyze the data. The second phase of electricity use (M2 = 64.50, SD2 = 13.16) were significantly higher than those of the first phase (M1 = 56.81, SD1 = 10.00, t(15) = −4.14, p < .001) which confirms the existence of the rebound effect. The size of the rebound effect was 13.5%.

$$ \begin{aligned} {\text{Size of the rebound effect}} = & \left( {{\text{second phase of electricity use}} - {\text{first phase of electricity use}}} \right) \\ & /{\text{first phase of electricity use}})) \\ \end{aligned} $$
(2)

This indicates that the electricity use increased by about 13.5% when the electricity bills were decreased. The third phase electricity use (M3 = 58.44, SD3 = 14.35) were significantly lower than those of the second phase (M2 = 64.50, SD2 = 13.16; t(15) = 5.33, p < .000), which indicates that when the participants were provided electricity use feedback (i.e., recognized rebound effect), they significantly reduced their electricity use. The size of the rebound effect was 9.4%, i.e., the size of counter rebound effect is 9.4%. During the interview, participants were willing to reduce their electricity use when electricity use feedback was provided in the third phase of electricity use. Half of the participants thought the electricity use system should have notification function. Without the notification functions, they would not take actions to reduce their electricity consumption rates.

4 Experiment 2

The aim of Experiment 2 was to examine whether the rebound effect exists in more realistic electricity use condition of smart homes. Experiment 2 examined the changes of participants’ electricity use when electricity use suggestion was provided.

4.1 Experimental Material

Two software prototypes were developed with Axure 7.0 to provide immersive situational simulations, i.e., “Electricity Use Steward” and “Intelligent Electricity Use Steward” (see Fig. 2). The software prototypes were used to remotely control environmental settings (temperature, humidity, and water temperature) and indoor appliances (in living room, kitchen, bedroom, and bathroom). Participants used software prototypes to conduct electricity use. “Intelligent Electricity Use Steward” had higher level of intelligence than “Electricity Use Steward”. “Intelligent Electricity Use Steward” could provide electricity use suggestions which helps participants live more comfortable, and could also automatically help participants adjust each appliance to the recommended setting. Experiment 2 classified the electricity use into four categories: environmental setting (e.g., target temperature and air volume, target humidity), illumination setting (e.g., light in the living room, light in the kitchen), and appliance setting (e.g., speakers in the living room, washing machine in the kitchen).

Fig. 2.
figure 2

Homepage of the two prototype software programs

Full size image

4.2 Experimental Procedures

Fifty-one students (thirty males and twenty-one females) were invited in Experiment 2. All participants were aware of electricity use conditions and electricity bills at their own homes. The average age of the participants was 23.7. The experimental procedures were almost the same as the Experiment 1, whereas the electricity use of Experiment 2 was conducted by software manipulation. The experimental software presented by iPad air 2 had 9.7-inch LED screen, pixel of 2048 × 1536 and resolution of 264 ppi. In the first phase of electricity use, “Electricity Use Steward” was used to obtain the electricity consumption in common conditions. In the second phase of electricity use, “Intelligent Electricity Use Steward” was used, which could provide suggestions to adjust electricity use. Prior to the third phase of electricity use, the participants were provided with electricity use feedback by diagrams to compare their electricity use in the first two phases. Then the “Intelligent Electricity Use Steward” was used in the third phase of electricity use.

4.3 Results

The data about the four categories of electricity use settings in three phases experiment of electricity use was normally distributed. A paired t-test was used conducted to analyze the data. The electricity use in the first phase (T1) and the second phase (T2) were listed in Table 2. Electricity use in the second phase was significantly higher than that in the first phase. The results indicate that when electricity use suggestions were offered, the electricity use would be reduced. Hence, a rebound effect occurred; the rebound effect size of each type of electricity use setting is listed in Table 3.

Table 2. Paired t-test of phase one and phase two
Full size table
Table 3. Sizes of rebound effect
Full size table

There was no significant difference in electricity use between the second and third phase in aspects of the environmental setting, illumination setting, appliance setting, and total power consumption. It was found that the electricity use of the above four settings in the third phase are less than those in the second phase.

During the interview, participants demonstrated living habit is the most important reason to save energy. Most of the participants believed that they could achieve more comfortable living conditions by following offered electricity use suggestions in smart homes. This explains why participants in Experiment 2 slightly adjusted their electricity use when they found that their electric use increased in the third phase of electricity use. Many participants expressed that they would stick to their electric use habits regardless of the offered electricity use suggestions. But the statistical analysis result shows that participants are easily influenced by proposed electricity consumption rates, while they are unaware of it.

5 General Discussion

The stimulation of electricity bills and electricity use suggestions confirmed the occurrence of the rebound effect with an effect size of 12.5% in the study. According to the research by Greening et al. (2000), household rebound effects are about 0–30% [2]; thus, the rebound effect obtained in this study is within the normal range. The research by Greening et al. (2000), which involved four studies, indicates that the rebound effects in illuminating are about 5–12%, while the rebound effect reported in this study is 20%, exceeding the above range. This is possibly due to the software simulation employed in this study.

Reducing electricity bills and offering electricity use suggestions can both cause rebound effect. Even though all participants’ electricity use were reduced when perceiving rebound effect, only those who were provided with immediate feedback about electricity bills had sharply reduced use. Multiple energy-saving functions can be provided for smart homes users to save energy. For example, intelligent learning is the most importation function that helps to change lifestyles. Electricity use suggestions can be offered to users in a personified manner to gradually change the users’ lifestyles. Moreover, immediate personified notifications will be sent when any changes in the electricity use or any abnormality in the performance of household appliances is detected in order to optimize the users’ experience.

6 Conclusion

This study verifies that the rebound effect exists in smart homes. Lower electricity bills and providing electricity use suggestions have similar rebound effect in smart homes. The size of the rebound effect is 13.5%. The rebound effect sizes vary under different electricity use patterns. Under the impact of electricity use suggestions, the reported rebound effect sizes are 6.42% upon appliance settings, 12.91% upon environment settings, and 20.24% upon illuminating settings. Hence, in order to reduce the rebound effect in future smart homes, the following recommendation can be considered: provide immediate electricity bills information combined with electricity use feedback, and provide electricity use suggestions through intelligent learning.