2021 International Conference on Technology and Policy in Energy and Electric Power (ICT-PEP)

Long-Term Electricity Demand Forecast Using

Multivariate Regression and End-Use Method:
A Study Case of Maluku-Papua Electricity System
2021 International Conference on Technology and Policy in Energy and Electric Power (ICT-PEP) | 978-1-6654-1641-2/21/$31.00 ©2021 IEEE | DOI: 10.1109/ICT-PEP53949.2021.9601144

Tumiran Sarjiya Lesnanto Multa Putranto

Department of Electrical and Department of Electrical and Department of Electrical and
Information Engineering Information Engineering Information Engineering
Universitas Gadjah Mada Universitas Gadjah Mada Universitas Gadjah Mada
Yogyakarta, Indonesia Yogyakarta, Indonesia Yogyakarta, Indonesia
[email protected] [email protected] [email protected]

Edwin Nugraha Putra Rizki Firmansyah Setya Budi Candra Febri Nugraha
System Planning Division Department of Electrical and Department of Electrical and
PT PLN (Persero) Information Engineering Information Engineering
Jakarta, Indonesia Universitas Gadjah Mada Universitas Gadjah Mada
[email protected] Yogyakarta, Indonesia Yogyakarta, Indonesia
[email protected] [email protected]

Abstract— One of the most important stages in electric According to the national electricity portrait, Indonesia's
power system planning is load forecasting. An accurate demand electricity condition varies from one region to the next. This
forecast model is required to create the most optimal plan. picture is also supported by comparing the development in
Long-term demand forecasts are developed in this study using eastern Indonesia to that in Java. With its superior access to
regression and end-use models. Simulation is carried out to electricity compared to other areas, Java has emerged as a hub
determine the load demand of the Maluku-Papua system of economic activity. Meanwhile, the inability to mobilize
through 2050. The variables that have the greatest influence on investment in the productive sector in other regions, such as
demand forecasting, according to data analysis, are gross Maluku and Papua, is due to a lack of electricity supply.
regional domestic product (GRDP), population, electrification
ratio, and electricity price. The electricity demand in Maluku- One of the challenges in planning the Maluku-Papua
Papua is expected to rise by 5.7% in the business as usual (BaU) electricity system is the geographical nature of the
scenario and 6.7% in the High scenario. The peak load is archipelago and the uneven population distribution [6].
expected to increase by 5.6% in the BaU scenario and 6.5% in According to historical data, the electrification ratio in the
the High scenario. Furthermore, the results of the demand area was only 70.5% in 2020 [7]. Furthermore, the household
forecast can be used to determine policies and the planning of sector continues to dominate electricity demand in Maluku-
the electric power system. Papua, accounting for 65% of total electricity sales [7].

Keywords—electricity demand forecasting, multivariate Even though the electricity load is still low, demand in
regression, end-use method, Maluku-Papua Maluku-Papua experienced an average of 10.1% growth rate
from 2002 to 2019. In addition, according to the Indonesian
I. INTRODUCTION Ministry of Industry and the Coordinating Ministry for the
Electrical power is critical for improving people's well- Economy, there will be several potential large electricity
being and promoting economic growth. Electricity customers in Maluku and Papua. These potential customers
consumption per capita is a good indicator of a country's can be found in the Special Economic Zones (SEZ) and
electricity availability. Indonesia ranks sixth in the ASEAN Industrial Estates (IE) designated by the National Strategic
region, trailing Singapore (8,542 kWh/capita), Brunei (8,206 Project (PSN). To drive economic growth, it is necessary to
kWh/capita), Malaysia (5,097 kWh/capita), Thailand (2,485 provide an adequate electricity supply.
kWh/capita), and Vietnam (2,745 kWh/capita) [1], [2]. Load forecasting is an important part of the initial process
Meanwhile, in 2020, Indonesia's electricity consumption is of power system planning. Planning for additional capacity
expected to be around 1,039 kWh/capita [1]. As a result, of the electrical system is impossible without the data of
Indonesia must boost the growth of its electricity sector to electricity demand forecast. The accuracy of load forecasting
spur the economic growth. has an impact on electricity generation and transmission
The gross domestic product (GDP) and electricity planning. If the results are too high, there will be excess
consumption are inextricably linked. From 2009 to 2019, capacity and, as a result, excess investment. On the other
Indonesia's GDP increased by 5.3% per year on average [3]. hand, If the forecast is too low, rolling blackouts will occur,
On the other hand, electricity consumption per capita has causing customer dissatisfaction [8]. Consequently, selecting
increased at a similar rate [4], [5]. As a result, the government the appropriate method is critical to producing the most
has been attempting to accelerate the electrification ratio and accurate forecast possible.
equity in recent years through the State Electricity Enterprise There are several demand forecasting methods, including
(PLN). The provision of better electricity that reaches all of parametric methods, artificial intelligence, and end-use [8],
Indonesia is hoped to increase electricity consumption in the [9]. The parametric method is a forecasting method expressed
productive sector. As a result, the GDP will grow. in a mathematical model. It is based on the relationship

978-1-6654-1641-2/21/$31.00
Authorized licensed use limited to: ©2021 258on December 22,2022 at 07:50:20 UTC from IEEE Xplore. Restrictions apply.
IEEE GADJAH MADA. Downloaded
UNIVERSITAS
between electricity demand and the factors that influence it.
The regression method is a subcategory of the parametric
method that is commonly used in electrical load forecasting 1,800

Electricity Sales (GWh)

[10]. Several studies on demand forecasting have been 1,500
conducted using the regression method. According to 1,200
researchers [11], GDP and GDP per capita affect the amount
900
of electricity consumed in Italy. Meanwhile, the researcher
[12] discovered that electricity price, income, and lifestyle 600
variables impact household electricity consumption. 300
The end-use method is used to forecast the energy 0
consumption by considering the electricity usage of 2002 2004 2006 2008 2010 2012 2014 2016 2018
individual equipment [13]. As a result, this method Year
necessitates very specific data [9]. For example, researchers
in [14] forecasted building electricity demand by measuring Residential Commercial Public Industrial
energy use in its facilities and connecting it to other
influential variables. Fig. 2. Statistics of energy sales by sector of Maluku-Papua

This study employs a multivariate regression method, TABLE I. SEZ AND IES IN MALUKU-PAPUA
which is widely used in many electric power companies
Province SEZ and IE
worldwide, in combination with the end-use method. It aims North Maluku SEZ of Morotai
to generate output that is more precise and closer to reality. IE of Buli
The Maluku-Papua system is being used as a test system for West Papua IE of Sorong
demand forecasting until 2050. This is the first study to IE of Bintuni Bay
conduct long-term demand forecasts in both regions to the The government of Indonesia through PSN policy aims to
best of the author's knowledge. promote growth and equitable development to improve
The following is how this paper is organized: The current people's well-being [17]. To achieve this, SEZs and IEs were
state of the Maluku-Papua electricity demand is described in established in several areas, some of which are located in
Section II. The concept of linear regression is discussed in eastern Indonesia. Maluku-planned Papua's SEZ and IEs are
Chapter III. The research methodology, including workflow, located in the provinces of North Maluku and West Papua, as
scenario, and data, is illustrated in Section IV. The forecasted shown in Table I.
demand results are presented in Section V. Finally Section VI
summarizes the conclusions and implications. III. DEMAND FORECAST: REGRESSION MODEL
Aside from representing the foundation for the electrical
II. MALUKU-PAPUA ELECTRICITY DEMAND system planning process, electricity demand forecasts are
Maluku-Papua is an area located in the eastern part of also used to prepare distribution planning, personnel
Indonesia. Maluku, North Maluku, Papua, and West Papua planning, financial projections, and other similar planning.
are the four provinces that make up its territory. Maluku has Regression is widely used for forecasting electricity demand.
the characteristics of a large number of islands. Papua, on the A simple linear regression model is depicted in (1).
other hand, is a large island in Indonesia's easternmost region.
= + + + ⋯+ + (1)
The industrial sector in Maluku-Papua is underdeveloped.
The gross regional domestic product (GRDP) in the region is
only 370,901 billion IDR. It is much lower than the GRDP in Alternatively, the regression equation can also be an
Java, which is 25 times larger [15]. It is also reflected in the exponential or polynomial model, as in (2) and (3).
location's electricity demand, as illustrated in Fig. 1. In 2019,
⋯ (2)
Maluku-energy Papua's sales were only 2,624 GWh, far less = . .
than other regions [16]. Historically, residential use = . . .… . . (3)
dominated electricity consumption as depicted in Fig. 2.
With,
Kalimantan
Sumatra
10.7 TWh : dependent variable
Sulawesi
36.7 TWh
10.8 TWh Maluku-Papua ,…, : independent variables
2.6 TWh
: constant (intercept)
,…, : regression coefficients
: error

Java
174 TWh
The variable to be searched is the dependent variable:
electrical energy consumption or energy sales. The
independent variables, on the other hand, are the factors that
Fig. 1 Electricity sales in Indonesia will influence the dependent variable. GDP, population,
electricity price, and other factors are among them.

259on December 22,2022 at 07:50:20 UTC from IEEE Xplore. Restrictions apply.
Authorized licensed use limited to: UNIVERSITAS GADJAH MADA. Downloaded
 IV. METHODOLOGY After that, the regression models that satisfy the statistical
tests are used to forecast the electricity sales by province.
A. Scenario Forecasted electricity sales are calculated for each demand
There are two forecasting scenarios in this study: Business sector, namely residential, commercial, public, and industrial,
as Usual (BaU) and High. These two scenarios were created and then added together to form the provincial demand. For
to represent various levels of economic growth. The BaU regular general customers, the electricity demand is
scenario assumes a low economic growth projection based on forecasted using a linear or non-linear multivariate regression
the 2020-2024 National Mid-Term Development Plan method, depending on the model obtained. Then, for the High
(RPJMN) [18]. Meanwhile, no end-use load is taken into scenario, large prospective customers are calculated
account because the potential of large loads in SEZ and IE separately using the end-use method. As a result, the final
are not considered. On the other hand, the High scenario demand is the sum of the forecasted demand based on the
represents optimistic growth by taking into account the regression and the end-use method.
medium growth of loads. The end-use loads considered are
all potential large customers in SEZ and IE that can be served Furthermore, energy production is calculated from energy
by the utility. Table II contains a summary of the scenario. sales by including own use in power plants and substations,
as well as transmission and distribution network losses. Peak
B. Workflow load is calculated for each province based on energy
The demand forecast workflow is depicted in Fig. 3. The production and the average load factor per year.
forecast is based on both historical and projection data. It C. Data
considers the major independent variables such as GRDP,
population, electrification ratio, and electricity prices. Time- Historical energy data, GRDP, total population,
series data for those variables are then generated from the electrification ratio, and electricity price are all required for
input data, ranging from 2020 to 2050. Inflation is assumed electrical energy demand projection. On the other hand, peak
to be the only factor influencing the forecast for the average load calculations necessitate the use of historical peak load
price of electrical energy. data, line losses, and system load factors. Table III and Table
IV contain information about the data that was used.
The model is created by repeatedly running the Simple
Econometric Simulation System, Expanded (Simple-E) with TABLE II. DEMAND FORECAST SCENARIO
different combinations of independent variables. The process Business as Usual (BaU) High
is repeated until the best-fitted regression equation is Low economic growth based on Medium economic growth based
obtained by incorporating multiple statistical test indicators. RPJMN 2020-2024 on RPJMN 2020-2024
The R2 and AR2 (Adjusted R2) tests are two of them. Normal load growth in SEZ and Considering all potential large
IE, without considering potential customers in SEZ and IE
major customers

Historical and
Simple-E Output
Projection Data

GRDP Data
Statistical Test
(~2000-2050)

Population Data Electricity

Variable Modelling

(~2000-2050) Energy
R-squared
Demand
End-Use Peak
Electrification Ratio Data Forecast of
Demand Load
(~2000-2050) Residential,
Projection Forecast
Commerical,
Adjusted R-squared
Electricity Price Data Public, and
(~2000-2050) Industry

Energy Demand Data

...
(~2000-2019)

Fig. 3. Demand forecast workflow

TABLE III. DATA FOR ELECTRICITY ENERGY DEMAND FORECAST

Province
Data
Maluku North Maluku Papua West Papua
Energy sales and Yearly historical data 2000-2019 2002-2019 2001-2019 2001-2019
production Average annual growth 10.34% 11.4% 9.95% 8.84%
GRDP Yearly statistical and forecast data 2000-2050 2000-2050 2000-2050 2000-2050
Average annual growth (BaU; High) 5.05%; 5.36% 5.38%; 5.69% 4.35%; 4.67% 5.49%; 5.81%
Total population Yearly statistical and forecast data 2000-2050 2000-2050 2000-2050 2000-2050
Average annual growth 1.65% 1.76% 2.19% 2.31%
Electrification ratio Yearly statistical and target data 56.15% (2000) 99.66% (2018) 24.3% (2001) 44.03% (2005)
100% (2050) 100% (2050) 100% (2050) 100% (2050)
Electricity price Price 1,017-1,352 (2019) 1,142-1,307 (2019) 1,159-1,393 (2019) 1,208-1,394 (2019)
Annual growth 3.5% 3.5% 3.5% 3.5%

260on December 22,2022 at 07:50:20 UTC from IEEE Xplore. Restrictions apply.
Authorized licensed use limited to: UNIVERSITAS GADJAH MADA. Downloaded
 TABLE IV. DATA FOR ELECTRICITY PEAK LOAD DEMAND FORECAST

Province
Data
Maluku North Maluku Papua West Papua
Peak load historical Yearly statistical data 2003-2019 2002-2019 2002-2019 2004-2019
data Average annual growth 9.42% 9.58% 10.64% 9.37%
Average power Transmission 1.11% 1.06% 0.17% 0.16%
system losses (2019-2050) (2019-2050) (2017-2050) (2020-2050)
Distribution 10.81% 8.37% 6.96% 8.8%
(2002-2050) (2002-2050) (2001-2019) (2003-2050)
Load factor Yearly statistical and forecast data 2002-2050 2002-2050 2001-2050 2003-2050
Yearly average 0.6 0.6 0.64 0.61

V. RESULT TABLE VI. STATISTICAL TEST RESULT OF BAU MODEL

A. Energy Sales Forecast Province

Statistics
North West
Test Maluku Papua
According to the data analysis, there is a strong Maluku Papua
relationship between electricity sales, GRDP, population, Residential Demand
electrification ratio, and electricity price. The demand -R2 0.979 0.988 0.998 0.994
-AR2 0.969 0.983 0.997 0.992
forecast regression model obtained for Maluku in the BaU Commercial Demand
scenario is shown in (4)-(7). The formulation of the demand -R2 0.995 0.985 0.998 0.982
forecasting model is carried out in each customer sector. -AR2 0.993 0.978 0.998 0.976
Public Demand
= −16.1005 − 101.55 + 0.011721 ∙ %&' -R2 0.977 0.975 0.979 0.979
+ 0.000695254 ∙ *+ − 0.0727851 -AR2 0.967 0.967 0.974 0.973
∙ + − 12.8846 ∙ & - − 29.4912 (4) Industrial Demand
∙ & ./ 0 -R2 0.752 0.896 0.874 0.843
-AR2 0.639 0.861 0.838 0.798
12 = −4.17624 − 42.9419 + 0.0171073 ∙ %&'12
+ 0.0116612 ∙ '45 − 0.0182232
(5) TABLE VII. AVERAGE ANNUAL GROWTH OF ELECTRICITY SALES
∙ +12 − 0.830437 ∙ & / 6
Average Annual Growth
728 = −3.49699 − 19.6244 + 0.0162257 ∙ %&'728 Province (2020-2050)
+ 0.00565794 ∙ '45 − 0.0419265 BaU High
∙ +728 + 2.13676 ∙ & 9, 0, (6) Maluku 5.4% 6.0%
+ 11.7587 ∙ & 9/ 6 North Maluku 6.5% 8.9%
Papua 5.5% 5.9%
/ 0..=60 / . 9 0 .= 6 >
:;< = 1.03011 ∙ ∙ %&':;< ∙ '45 West Papua 5.7% 6.8%
.=>9. 9
∙ +:;< ∙ .6=>00 ∙? @@ A @@B (7) Fig. 4 depicts the electricity sales forecast by province for
∙ / . >= 69-∙? @@C. @@D, @ , @
the BaU and High scenarios in 2020-2050. The High scenario
forecast always produces higher results than the BaU.
With, However, the most notable distinction between BaU and
: electricity energy sales in year y of each sector High scenarios can be found in North Maluku and West
%&' : GDP in year y of each sector Papua. The need to supply SEZ and several IEs planned
*+ : number of customers in year y of each sector resulted in a significant increase of demand in both regions
for the High scenario. As shown in Table VII, this has
+ : electricity cost in year y of each sector
encouraged the growth of electrical energy in North Maluku
'45 : population in year y and West Papua to be higher than in other provinces.
&E : dummy variable, the value is 1 if F = G,
B. Energy Sales Forecast by Sector
otherwise 0
Meanwhile, the type of model for the BaU scenario, as Fig. 5 depicts the energy sales by sector in the BaU and
well as the statistical test results, are summarized in Table V High scenarios in 2020 and 2050. The energy use in the four
and Table VI. According to the statistical test results, the regions is expected to shift. In the BaU scenario, the trend
models obtained are well-suited, as evidenced by the high R2 shows that the percentage of electricity used by the residential
and AR2 values in almost all models. sector is decreasing while the commercial and industrial
sectors are increasing. However, except for North Maluku,
TABLE V. TYPE OF REGRESSION MODEL OF BAU MODEL the changes are minor. The use of energy in North Maluku
and West Papua changes dramatically under the High
Province
Sector
scenario. The primary need for electrical energy in North
North West
Maluku Papua Maluku has shifted from residential to industrial. The same
Maluku Papua
Residential Linear Linear Linear Linear pattern was observed in West Papua, where the percentage of
Business Linear Linear Linear Linear household energy requirements decreased while industry
Public Linear Logarithmic Linear Linear increased. The presence of SEZ and IEs contributed
Industrial Exponential Linear Linear Linear
significantly to increase the industrial electricity needs in
these two areas.

261on December 22,2022 at 07:50:20 UTC from IEEE Xplore. Restrictions apply.
Authorized licensed use limited to: UNIVERSITAS GADJAH MADA. Downloaded
 5,000 6,000

4,000 5,000
Energy (GWh)

Energy (GWh)
4,000
3,000
3,000
2,000
2,000
1,000 1,000

0 0

Forecast BaU Year High

Forecast Statistics Forecast BaU Year High
Forecast Statistics

(a) (b)
8,000 5,000

4,000
6,000
Energy (GWh)

Energy (GWh)
3,000
4,000
2,000
2,000
1,000

0 0

Forecast BaU Year High

Forecast Statistics Forecast BaU Year High
Forecast Statistics

(c) (d)
Fig. 4 Electrical energy demand forecast of (a) Maluku; (b) North Maluku; (c) Papua; (d) West Papua

100% 100%
Energy Use by Sector

Energy Use by Sector

80% 80%

60% 60%

40% 40%

20% 20%

0% 0%
2020 2050 2020 2050 2020 2050 2020 2050 2020 2050 2020 2050 2020 2050 2020 2050
Maluku North Papua West Papua Maluku North Papua West Papua
Maluku Maluku

Residential Commercial Public Industrial Residential Commercial Public Industrial

(a) (b)
Fig. 5 Electrical energy use by sector of (a) BaU scenario; (b) High scenario

C. Peak Load Forecast

The peak load forecast is calculated using the energy sales
TABLE VIII. AVERAGE ANNUAL GROWTH OF PEAK LOAD
forecast and the load factor assumption. Load factor is
defined as the average load in MW divided by the peak load Average Annual Growth
over a given period. Fig. 6 shows the peak load forecast per Province (2020-2050)
BaU High
province and the summary is shown in Table VIII. Because Maluku 5.4% 5.9%
of the projected demand of SEZ and IEs, the peak load of the North Maluku 6.4% 8.3%
High scenario in North Maluku and West Papua differs Papua 5.4% 5.8%
significantly from that of the BaU scenario. West Papua 5.7% 6.6%

262on December 22,2022 at 07:50:20 UTC from IEEE Xplore. Restrictions apply.
Authorized licensed use limited to: UNIVERSITAS GADJAH MADA. Downloaded
 Peak Load (MW) 1,000 1,000

Peak Load (MW)

800 800
600 600
400 400
200 200
0 0
2020 2025 2030 2035 2040 2045 2050 2020 2025 2030 2035 2040 2045 2050
Year Year

Forecast BaU Forecast High Forecast BaU Forecast High

(a) (b)
1,500 1,000
Peak Load (MW)

Peak Load (MW)

1,200 800
900 600
600 400
300 200
0 0
2020 2025 2030 2035 2040 2045 2050 2020 2025 2030 2035 2040 2045 2050
Year Year

Forecast BaU Forecast High Forecast BaU Forecast High

(c) (d)
Fig. 6 Peak load forecast of (a) Maluku; (b) North Maluku; (c) Papua; (d) West Papua

VI. CONCLUSION Statistics 2013,” 2013.

[5] M. of E. and M. R. Directorate General of Electricity, “Electricity
A demand forecast model was developed in this study to Statistics 2019,” 2019.
project the electricity demand in Maluku-Papua. Although [6] Tumiran et al., “The Masterplan for Developing Electricity Systems
for Archipelagic Area by Considering Local Energy Resources: A
this paper focuses on the Maluku-Papua system, the Case Study of Maluku Islands,” Proc. 2019 7th Int. Conf. Smart
developed method can be applied to other case. The Energy Grid Eng. SEGE 2019, pp. 290–293, 2019.
characteristics of the results may also differ depending on the [7] Company Secretariat of PT PLN (Persero), PLN Statistics 2020.
system's uniqueness. The variables that have the greatest Jakarta: Company Secretariat of PT PLN (Persero), 2020.
[8] H. M. Al-Hamadi and S. A. Soliman, “Long-term / mid-term electric
influence on demand forecasting, are GRDP, population, load forecasting based on short-term correlation and annual growth,”
electrification ratio, and electricity price. The electrical vol. 74, pp. 353–361, 2005.
energy demand in Maluku-Papua is expected to rise by 5.7% [9] G. R. T. Esteves, B. Q. Bastos, F. L. Cyrino, R. F. Calili, and R. C.
in the BaU scenario and 6.7% in the High scenario, between Souza, “Long Term Electricity Forecast : A Systematic Review,” vol.
55, no. Itqm, pp. 549–558, 2015.
2020 and 2050. Peak load increased by 5.6% in the BaU [10] J. Lin, K. Zhu, Z. Liu, J. Lieu, and X. Tan, “Study on a simple model
scenario and 6.5% in the High scenario. The High scenario to forecast the electricity demand under China’s new normal
has a faster growth rate because the GRDP growth situation,” Energies, vol. 12, no. 11, 2019.
assumption used is higher than the BaU scenario, and there [11] V. Bianco, O. Manca, and S. Nardini, “Electricity consumption
forecasting in Italy using linear regression models,” Energy, vol. 34,
are considerations of large customer potentials. The projected no. 9, pp. 1413–1421, 2009.
special economic zone and industrial estates are expected to [12] S. Zhou and F. Teng, “Estimation of urban residential electricity
increase the energy demand in Maluku-Papua, thus positively demand in China using household survey data,” vol. 61, no. 2013, pp.
impacting the economic growth. Furthermore, the forecasted 394–402, 2015.
[13] S. Ha, S. Tae, and R. Kim, “Energy demand forecast models for
demand can be used to guide policy and planning for the commercial buildings in South Korea,” Energies, vol. 12, no. 12, pp.
electric power system expansion, including generation and 1–19, 2019.
transmission expansion planning. [14] G. Escrivá-Escrivá, C. Roldán-Blay, and C. Álvarez-Bel, “Electrical
consumption forecast using actual data of building end-use
decomposition,” Energy Build., vol. 82, pp. 73–81, 2014.
REFERENCES [15] BPS-Statistics Indonesia, “Gross Regional Domestic Product of
[1] Our World in Data, “Per Capita Electricity Consumption.” [Online]. Provinces in Indonesia by Industry,” BPS-Statistics Indonesia,
Available: https://ourworldindata.org/grapher/per-capita-electricity- Jakarta, 2021.
consumption?tab=table. [Accessed: 16-Jul-2021]. [16] PT PLN (Persero), PLN Statistics 2019. Jakarta, 2020.
[2] U.S. Energy Information Administration, “Electricity.” [Online]. [17] “Presidential Decree No. 109/2020 Third Amendment to Presidential
Available: Regulation 3/2016 concerning the Acceleration of PSN,” 2020.
https://www.eia.gov/international/data/world/electricity/electricity- [18] National Development Planning Agency (Bappenas), “Presidential
consumption. [Accessed: 16-Jul-2021]. Decree No. 18 Year 2020,” Natl. Mid-Term Dev. Plan 2020-2024, p.
[3] PDSI of Trade Ministry, “GDP Expenditure,” 2021. [Online]. 313, 2020.
Available: https://statistik.kemendag.go.id/gross-domestic-product.
[4] M. of E. and M. R. Directorate General of Electricity, “Electricity

263on December 22,2022 at 07:50:20 UTC from IEEE Xplore. Restrictions apply.
Authorized licensed use limited to: UNIVERSITAS GADJAH MADA. Downloaded