*3.4. Forecast Analysis*

Figure 7 compares the predictions of the building energy stock model to the actual total energy consumption in Kuwait from 2005 to 2017 after a systematic calibration procedure. The actual energy consumption data were obtained from the Ministry of Electricity and Water (MEW) [2]. For the calibration analysis, three main input parameters were adjusted as follows:


Good agreement between the predictions of the building energy stock model and the actual energy consumption was obtained with a relative error of less than 5%, as shown in Figure 7.

To predict a business-as-usual case (i.e., the baseline scenario), the forecast model mainly relies on the *UEC* and *stock*. With the projected values of *UEC* and *stock* for each equipment, we can use Equations (1) and (2) to predict the energy consumption and peak demand. The forecast of the equipment stock is mainly driven by the diffusion rate *D*(*y*) and new housing construction. For Kuwait, the diffusion rate is higher due to the high levels of income and electrification rate.

In addition to the population data mentioned in the Introduction, Table 4 shows the amount of housing subsidies provided by the Public Authority of Housing Welfare (PAHW) each year that has been projected until 2034. The housing subsidy values are another driving variable used to estimate the stock included in the model by *First purchases*(*y*) in Equation (3).

Even with the base case scenarios, the efficiency of equipment and appliances tends to improve over the years. This was estimated by assigning an efficiency improvement rate for each equipment and appliance in the model. Depending on the equipment type, *UEC* was assumed to improve 1–5% in efficiency per year based on [31,32] to account for the technology changes and code requirements. In addition, some new technologies will diffuse into the market and replace old ones that can be less efficient. Light-emitting diode (LED) lighting is a good example since it was introduced in the Kuwaiti

market a couple of years ago. A Bass model was used to estimate the adoption rate of LED lighting and was incorporated into the baseline model. The Bass model defines the fraction of sales *F(y)* in year *y* to represent the adoption rate of a new technology or product as follows:

$$\frac{dF(y)}{dt} = (p + qF(y))(1 - F(y))\tag{12}$$

where *p* represents the external factors that drive the market to adopt a new technology such as advertisement, and *q* is often referred to as the "word-of-mouth" effect from the early adopters to encourage the "imitators" to adopt the new technology [50]. To use the Bass model to forecast the adoption of a new product or technology, the parameter *p* (innovators), *q* (imitators), and the potential market size need to be estimated. Since no historical sales data of LEDs are available for Kuwait, the Bass model parameters were estimated by an analogy to the compact fluorescent lamps (CFL) that have past shipment data and similar diffusion characteristics with LEDs [51]. The ordinary least squares (OLS) method was used to estimate the Bass model parameters (i.e., the coefficient of innovation (*p*) and imitation (*q*)), as shown in Table 5.

**Figure 7.** Comparison of the actual and modeled annual residential sector energy use from 2005 to 2017.

Several researchers have analyzed the effect of weather on energy consumption [52–54]. For the case of Kuwait, and based on [55–57], the influence of weather in the form of cooling degree-days (CDD) on long-term electricity demand forecasting is only statistically significant at 20% due to the low year-to-year weather variation in Kuwait. Therefore, the effect of annual weather variation was not considered in the forecasting model.


**Table 5.** Coefficients of innovation and imitation of the Bass model.


## **4. Results and Discussion**

Based on the specified inputs explained in the previous section, Figure 8 shows a bubble plot of the unit energy consumption of home appliances against the total stock to reveal the energy usage. The additional dimension, the size of the bubble, represents the total annual energy consumption. Household appliances included in the analysis consist of televisions (TV), personal computers (PC), washers, irons, microwave, refrigerator, freezer, water cooler, and dryers.

It can be observed that two major data clusters emerged with similar *UEC*s. One contained the following household appliances: televisions (TVs), personal computers (PCs), washers, irons, and microwaves. The *UEC* for this group ranged from approximately 200 kWh/year to 400 kWh/year. Despite the relatively low unitary electrical consumption of TVs, the quantity of the stock raised the level of impact. With approximately a thousand sets at a *UEC* of roughly 250 kWh per year, TVs represent a significant portion of the domestic energy use in Kuwait.

The second group consisting of higher *UEC*s, contained the following household appliances: water coolers, refrigerators, dryers, and freezers. Unlike refrigerators, freezers, and water coolers, dryers have low duty cycles and therefore consume less energy in a year, hence the smaller bubble. Moreover, in contrast, this group had a higher *UEC* range starting from approximately 800 kWh/yr to 1000 kWh/yr. Collectively, despite being less in stock, the overall impact is almost equally relevant due to the higher electrical consumption. This is partly due to the components that require significant power to operate such as compressors in refrigeration systems or resistive heaters commonly found in irons and electrical dryers.

**Figure 8.** Electrical energy consumption of selected household appliances in Kuwait.

Table 6 displays the *UEC*, stock quantity, and the 2017 total energy consumption for specific household appliances for Kuwaiti homes. Due to the differences in energy use patterns between Kuwaiti and non-Kuwaiti homes, a similar analysis was conducted utilizing equal usage parameters, but with different stock quantities. The *UEC* values for the listed household appliances were calculated as outlined in the Methodology section and remain unchanged for both models. The highest *UEC* was noted to be freezers, refrigerators, and dryers, respectively, while PCs, washers, and TVs had *UEC*s that were less than a third that of freezers.

The results indicate that the energy consumption for the listed appliances totaled 1972.36 GWh for Kuwaiti households. Approximately 40% of the total consumption was attributed to refrigerators and freezers. The high energy consumption for these appliances was expected as the *UEC* values were high to begin with. However, due to the relatively high stock quantities, TVs also represented a significant load on the grid. Despite their low *UEC*s, the impact was offset by the volume, adding up to 926,505 TV sets, the highest stock quantity in all the listed appliances.


**Table 6.** The 2017 total energy consumption of the modeled household appliances with corresponding stock quantities for Kuwaiti residential homes (2017).

Utilizing the same list of appliances, along with their *UEC*, Table 7 shows the 2017 total energy consumption of modeled household appliances with corresponding stock quantities for expatriate (non-Kuwaiti) residential homes in Kuwait and displays the total energy consumption for specific household appliances for non-Kuwaiti homes.

**Table 7.** The 2017 total energy consumption of modeled household appliances with corresponding stock quantities for expatriate (non-Kuwaiti) residential homes in Kuwait.


According to the results, the distribution of electricity consumption in residential households in Kuwait differs vastly, since the stock quantity weighs in heavily. Kuwaiti households account for roughly 70% of the total electrical consumption of the modeled appliances, whereas the remaining 30% was attributed to non-Kuwaiti household usage at 721.53 GWh. Parallel to the Kuwaiti profile, the results governing the expatriate households indicated that the top two energy-consuming appliances were refrigerators and freezers. The energy consumption of these two appliances make up approximately 40% of the overall energy usage for the expatriate household appliances.

From a broader perspective, the electrical consumption and demand distribution in residential households in Kuwait is broken down by the following main usage categories: lighting, air conditioning, space heating, water heating, and miscellaneous loads. Electrical consumption patterns remain heavily dependent on air-conditioning, as it represents the biggest slice within the pie charts shown in Figure 9. Air conditioning accounts for two thirds of the residential household consumption. However, although it is as little as 4.5%, space heating still accounts for a small load.

**Figure 9.** Distribution of on-site residential energy use and demand in Kuwait.

As seen in the distribution for residential electricity consumption, air conditioning makes up the bulk of the demand for Kuwaiti households at 66%, whereas the rest of the categories (miscellaneous loads, water heating, and lighting) range from 5% to 13%.

Utilizing the methodology outlined in this paper, the forecast of the residential energy consumption end-use was modeled and plotted in Figure 10. Revealing a similar trend observed in the household electrical consumption distribution and the household electricity demand distribution, air conditioning load is one of the highest loads for households. As per the results of the analysis, it is expected to rise exponentially from the year 2022 onward, reaching an estimated load of 60 TWh. Lighting is predicted to also rise, but much flatter, unlike the trend in air conditioning. The comparison between the actual and forecast points show an accurate model starting from 2005 until 2017. In terms of electrical demand, Figure 11 displays the growth for the air conditioning load, as it comprises a significant portion of the annual power demand. The results are also presented in a tabular form in Table 8. Figure 12 displays the forecast of electrical consumption for Kuwaiti and expatriate (non-Kuwaiti) households until the year 2040. Despite the slow growth in population, the forecast analysis indicates that the Kuwaiti energy consumption per capita was significantly higher than that of the expatriates, reaching levels of 15 MWh. The values for expatriates were almost stagnant, staying well below 1.5 MWh, despite the growing population figures that are expected to reach four million, more than doubling since 2005. The results are also represented in tabular form in Table 8.

**Figure 10.** Forecast of on-site residential energy consumption by end-use until the year 2040.

**Figure 11.** Forecast of on-site residential energy demand by end-use in Kuwait.


**Table 8.** Forecast of on-site residential energy consumption by end-use.

**Figure 12.** Forecast of population and residential energy consumption per capita.

#### **5. Conclusions**

Kuwait has one of the highest energy consumption per capita levels in the world. This large-scale consumption is negatively impacting its natural resources and the environment. The building sector alone accounts for 57% of electrical consumption. It is therefore important to study the driving impacts in a building's energy consumption in Kuwait. Utilizing end-use baseline information for residential loads sets an important foundation to help understand the residential consumption patterns. Based on the specified end-use equipment and certain parameters, a forecasting analysis was conducted to estimate the end-use distribution of electrical consumption for the state of Kuwait until the year 2040. In the model, end-uses were broken down into the following: air conditioning, lighting, miscellaneous loads, and space heating and water heating.

The resulting unit energy consumption (*UEC*) of home appliances was plotted against the total stock, which illustrated the impact of each of the specified home appliances. Refrigeration units, out of all appliances, held the highest *UEC* by far, as they were high in both stock and *UEC* values. A forecast model was then plotted to reveal the end-use energy consumption and peak demand in Kuwait until 2040. The air conditioning loads are expected to rise in the future with an average annual growth rate of 2.9%. Meanwhile, the rise in lighting energy consumption is much flatter due to an expected gradual shift toward more efficient lighting. Furthermore, based on the forecast results, differences between the Kuwaiti and expatriate (non-Kuwaiti) residential loads were observed. To the best of our knowledge, this is the first attempt to estimate the energy consumption of non-Kuwaiti households, where expatriates make up two-thirds of the population. These results provide opportunities for the development of more effective energy policies as well as opportunities for energy efficiency initiatives for the future.

The proposed model in this paper integrates equipment stock and unit energy consumption in order to project energy consumption at a more detailed level than other forecasting models. This level of detail in individual end-use equipment allows for the construction of various and detailed energy efficiency scenarios such as energy efficiency standards and labeling programs. Since the model

accounts for replacement stock of equipment and appliances, this can also be used to evaluate energy retrofit programs. Moreover, this approach allows for data on equipment efficiency, sales, and stock over time to be separately developed, assessed, and incorporated into the model. The result is the ability to evaluate the stock turnover and penetration of energy-efficient equipment to the building stock, and their effect on the energy use and peak demand. It will also make the model more dynamic and updated based on the available sales data.

**Author Contributions:** Conceptualization, T.A. and P.P.; methodology, T.A.; software, T.A.; validation, T.A.; formal analysis, T.A. and P.P.; investigation, T.A. and P.P.; resources, T.A. and P.P.; data curation, T.A.; writing—original draft preparation, T.A.; writing—review and editing, P.P.; visualization, T.A.; supervision, P.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research received no external funding.

**Acknowledgments:** The authors would like to express their gratitude to Barlas Demirciler, Nicholas Fette, and Ali Hajiah for their expertise and assistance throughout this study. The authors are also grateful to the Kuwait Ministry of Electricity and Water for their data and information support, especially Iqbal Al-Tayyar, Director of the Technical Supervision department.

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