More Is More? The Inquiry of Reducing Greenhouse Gas Emissions in the Upstream Petroleum Fields of Indonesia

Abstract

Global dependence on fossil fuels remains high despite the rapid expansion of renewable energy initiatives, and fossil fuels extracted from the earth’s crust are major contributors to greenhouse gasses. Unlike greenhouse gas studies in the downstream area, currently, few studies have investigated greenhouse gas in the upstream field, and there is no published study related to carbon emission influencing factors in Indonesia’s upstream field. A short panel data analysis is used to investigate the influence of oil and gas production and energy usage on greenhouse gas emissions by using data from 25 upstream fields (including offshore and onshore fields) collected from 2015 to 2018. The results show that maintaining a constant energy usage leads to increased oil and gas production and reduced greenhouse gas emissions. This pattern implicitly indicates that improving energy efficiency during oil and gas production is critical for ensuring production stability and further reducing greenhouse gas. This study may contribute significantly toward the industrial decarbonization approach that includes upstream processes to achieve net-zero carbon emissions. We recommend further research to study the carbon mitigation pattern in the upstream petroleum fields.

1. Introduction

According to international trends, several countries have argued against the possibility of economic expansion without increasing CO₂ emissions. Specifically, it has become a hot topic whether it is genuinely possible to achieve steady economic growth without increasing energy consumption or greenhouse gas emissions [1]. Significant growth in infrastructure, agriculture, and industry boosts GDP (gross domestic product) and increases energy usage [2]. With a rise in GDP, the economy tends to grow, leading to significantly increased energy consumption [3]. Since energy is recognized as the primary source of a country’s production and industrial sectors, it plays a prominent role in national economic growth. As a result, understanding energy policy and regulations are essential for understanding the connection between economic growth and energy usage [4]. Furthermore, Indonesia aspires to be a developed country by 2045, and GDP must be increased to achieve the targeted status [5]. Currently, rapid developments in different sectors have been boosted to support the national income target. However, this development requires tons of energy, mainly sourced from fossil fuels. The industrial and transportation sectors are projected to consume 333.93 million GJ of diesel oil and 8430.74 million GJ of gasoline annually by 2030, respectively [6]. Therefore, sustainable and cleaner production needs to be implemented to minimize the risk to the environment [7,8].

Since the 1990s, carbon emissions generated by energy usage have increased dramatically in newly industrialized countries compared to developed countries. As a result, environmental degradation has reached alarming proportions, prompting fears of climate change and global warming [9]. Global warming is currently occurring at an alarming rate across the planet, and Southeast Asia is one of the world’s most climate-vulnerable regions. Although it is not the world’s largest producer of carbon dioxide (CO₂), its emissions will become significant if no action is taken [10]. The growth of sustainability forces many countries to take action on climate change mitigation. Due to international pressure and its vulnerability to the effects of climate change, Indonesia needs to undertake significant reforms. The use of fossil fuels for energy must be decreased, and CO₂ emissions in Indonesia must be minimized [11]. Meanwhile, Indonesia, a developing country with abundant natural resources, is the world’s sixth-highest emitter of greenhouse gases. These natural resources, including both fossil and renewable resources, could promote the country’s economic development. [12]. However, the impact of greenhouse gas emissions as a consequence of economic growth needs to be assessed, particularly since CO₂ emissions increase exponentially [10].

Currently, the global petroleum industry is still considered a valuable industry for fulfilling global energy demand. Petroleum-based energy is still considered the primary energy source in global economic growth. As a result, several efforts to reduce greenhouse gas emissions have been introduced, including replacing coal-burning with natural gas burning as an energy transition fuel [13]. Nevertheless, the dependency on petroleum-based energy is projected to remain high since it is the backbone of the global energy supply [14,15,16]. In addition, the drastic growth of the global economy both increases the usage of fossil fuel in the downstream sector (household, transport, and other human activities) [17,18,19,20] and drives industries’ development in every sector for fulfilling the economic trend, leading to an excess of CO₂ emissions [21]. Thus, elementary industries, such as heavy industries (the iron and steel industry), generate tons of CO₂ emissions while fulfilling the demand for infrastructure growth [22,23]. This trend will trigger an increase of CO₂ emissions in the upstream (exploration) operations of the petroleum industries [24,25]. The petroleum industry is one of Indonesia’s most important industrial sectors as it produces the country’s greatest revenue and is one of the world’s largest exporters [26]. Indonesia’s petroleum industry has a history of more than 100 years and a well-understood regulatory structure for supporting the national energy demand. Indonesia has been a global leader in several policies, including the production sharing contract (PSC) model and LNG (liquefied natural gas) industrialization. Indonesia’s petroleum industry continues to be an essential sector contributing to national income (including 12 percent of the average national budget) and national economic growth [27]. Since energy availability and supply are essential factors for maintaining Indonesia’s economic development, the energy and petroleum industry is listed as a national strategic industry. The government aims to increase domestic natural gas usage in response to Indonesia’s rapid economic growth, as stated in the 2014 Energy Law. It targets a 22% share of natural gas in the national energy mix by 2025, equivalent to 8300 MMSCFD, or a 3000 MMSCFD increase in gas supply above the current domestic supply [26]. However, law enforcement related to sustainable production and net-zero carbon policy enforce the upstream petroleum field to reduce their carbon emissions.

A data-driven approach which relies on factual information can help to predict the future pattern, avoiding false assumptions and increasing the model’s reliability. Compared to the analysis-based approach which relies on anecdotal and observation evidence, the data-driven approach promises excellent and reliable result analysis. There is no sufficient empirical research discussing possible inquiries to reduce greenhouse gas in the upstream field. Numerous studies have determined greenhouse gas emissions in different parts of the petroleum industry [24,25,28], but the inquiry for reducing greenhouse gas in the upstream petroleum field is rarely discussed. This research aims to identify the inquiry needed for reducing greenhouse gas emissions in the upstream petroleum field using panel data analysis. Based on the emission–energy–production relationship found in this study, policies that could efficiently lead to greenhouse gas reductions are further recommended. The focus is on the upstream petroleum industry and the sector pertaining to the primary extraction process. The contribution is extended toward industrial decarbonization strategies involving upstream processes to fulfill the net-zero carbon policy. This finding implies that energy usage in the main process is the key factor in maintaining production stability and further reducing greenhouse gases.

2. Literature Review

Greenhouse gas is a major global issue, and many scholars are studying greenhouse gas from different perspectives. The existing literature provides extensive studies of greenhouse gas patterns, which have been examined using various approaches and statistical methods.

The major environmental hazard posed by the current energy system is climate change due to carbon dioxide emissions from the use of fossil fuels [29]. Carbon dioxide (CO₂), hydrofluorocarbons (HFCs), nitrous oxide (N₂O), perfluorocarbons (PFCs), methane (CH₄), and sulfur hexafluoride (SF₆) are greenhouse gases (GHGs) that contribute to global warming, which is regarded as the most pressing concern of the twenty-first century, according to the 1998 Kyoto Protocol [10]. Global average heat and fossil fuel emissions continue to climb, and global CO₂ emissions from fossil fuels increased by roughly 2% from 2017 to 2019 [30]. In developing countries, including Indonesia, greenhouse gases are mainly produced from fossil fuel usage [31]. Specifically, greenhouse gas emissions in Indonesia are generated from electricity production, manufacturing industries and construction, transportation, domestic buildings (commercial and public services), and other sources, which contribute 37.7%, 34%, 19.7%, 6.3%, and 2.3%, respectively [10].

The contribution and connection between energy usage and greenhouse gas emissions have been researched among countries and industries [2,3,9,23,24,29,32,33,34]. In several countries, it is widely accepted that fossil fuel energy usage is correlated with economic productivity growth, while natural gas consumption has been identified as showing bi-directional causality correlations and strong casualty effects in Brazil, Russia, and Turkey [35]. Energy consumption has a long-term positive and statistically significant impact on greenhouse gas emissions, although greenhouse gas emissions and economic growth have a non-linear relationship, which is consistent with the environmental Kuznets curve [32]. Say and Yücel [36] found a strong relationship between energy usage and greenhouse gas emissions in Turkey. Research using data collected from 1975 to 2011 showed that Indonesia’s economic advancement and energy consumption influenced greenhouse gas emissions, and the variables cointegrated in a long-run relationship [37]. Another empirical result showed that a 1% increase in crop production index and livestock production index caused a 28% proportional increase in greenhouse gas emissions [38].

Several methods have been developed to study the relationship between energy usage, oil and gas production, and greenhouse gas emissions, including the life cycle assessment (LCA) and statistical models [29,32,39,40]. The statistical multivariate regression model involves multiple data variables from different scales and factors and is commonly used to define the relationship between independent and dependent variables for analysis [41]. A panel data model is an empirical model that is mainly used to identify the trend involving time series and cross-sectional data that analyzes behaviors across individuals and over time. It is suitable for holistic environmental policy analysis in different types and scales of countries and industries [42]. In terms of estimating methods, the panel data model aids in producing more efficient estimates by considering both the time and cross-section dimensions of the dataset [22,33,43,44,45]. For example, energy consumption and activity added by the industrial sector in 30 different areas across China resulted in a rise and a vast difference in regional variations of carbon emissions caused by industrial structure differences [44]. Therefore, this method is strongly recommended since it can explain the multiple traits of carbon emission data [45]. Recent studies have examined the factors contributing to CO₂ emission using a multivariate regression model including the panel data model run at national [9,21,35] and industry levels [23,33]. Several panel data studies at the country level showed that investment, urbanization, trade, and economic growth significantly contribute to greenhouse gas emissions [21,46,47]. Other panel data studies have discussed the impact of various primary industries on greenhouse gas emissions in different regions [23,33,48,49]. The factors contributing to greenhouse gas emissions were strongly related to fossil fuel energy usage [50] due to higher oil and gas production in the upstream extraction fields.

As a major perpetrator of greenhouse gas emissions, and even though petroleum is a non-renewable energy source, many petroleum industries are constantly improving procedures for more efficient exploration of oil resources [51]. Previous studies identified greenhouse gas emissions in different sectors of the petroleum industry [24,25,28,52]. Johansson et al. (2012) [25] found that petroleum oil refineries account for approximately 8% of all industrial greenhouse gas emissions in the European regions (EU). Another study conducted by Abella and Bergerson [53] found that greenhouse gas emissions in petroleum refineries range from 4 to 18 g CO₂eq/MJ of petroleum product (23–110 kg CO₂ eq/bbl of crude oil). The potential greenhouse gas emissions from liquid gas transportation was studied in the Indian Himalayas. It was estimated that hauling a 14.2 kg cylinder produces 60 g of greenhouse gas emissions per kilometer [39]. In the upstream petroleum industry, natural gas and crude oil extraction in China is projected to generate 9.65 and 94.50 million tons of greenhouse gas emissions by 2030 [54]. An impact assessment study conducted by Sulistyawanti et al. [55] on the oil and gas exploration fields showed that greenhouse gas emissions were nearly 48,947.35 tons CO₂eq in 2015.

The current study examines the relationship between CO₂ emissions and significant contributing factors such as oil and gas production and energy consumption in the oil and gas exploration industries. By adopting the panel data model, which is widely used in environmental policy research, the statistical pattern of contributing factors and CO₂ emissions in the oil and gas exploration industries may be explained by describing the supporting and constraint variables with more precise correlation and low deviation.

3. Materials and Methods

3.1. Data Collection and Treatment

Exploration and exploitation of natural resources are primary activities in the upstream petroleum industry. The main process in this study consists of the main activity during oil and gas production in the exploitation stage. Natural resources exploration focuses on finding oil and gas reserves, while exploitation focuses on oil and gas extraction from the subsurface [56]. In Indonesia, natural resource extraction in the upstream petroleum industry is categorized according to its onshore and offshore associations. The onshore petroleum industry is located on the island, and the offshore petroleum industry is in the ocean [27]. Figure 1 presents the difference between the onshore and offshore oil and gas extraction processes. The products of the upstream petroleum industry are crude oil and natural gas, which are distributed to different industries. Crude oil is transferred to the refinery unit for further processing while natural gas is mostly used for power plants and the petrochemical industry.

Data variables in this study included greenhouse gas emission load and total production of oil and gas in the upstream petroleum field. As shown in

Table 1. Data description for the panel data model for investigating greenhouse gas emissions in the upstream petroleum industry.

Data Specification	Description
Variable	Data variables included: energy usage (gigajoule) greenhouse gas load (ton CO2eq) total production of oil and gas (ton oil equivalent)
Data sampling period	Data spans from 2015 to 2018
Emission sampling procedure	CO2, CH4, and N2O emissions sampling regulated by Indonesia’s government
Greenhouse gases	Calculated from CO2, CH4, and N2O concentration
Industry type	Offshore and onshore upstream petroleum fields
Number of companies	25 companies, including 16 offshore fields and 21 onshore fields, and 12 of the companies owned both offshore and onshore fields.
Field notation	Each field was given a code of C1, C2, C3, etc.

, the data were collected from 25 different upstream petroleum fields over four years, from 2015 to 2018. Greenhouse gas emission data consisted of CO₂, CH₄, and N₂O emissions monitored in their respective emission stacks, individual power plants, and funnels during crude oil and natural gas production.

All data collected from the upstream petroleum fields were raw daily data from automatic emission and production sampling. Energy consumption data of the main process and the petroleum fields, including the warehouse and other supporting facilities, were collected monthly. At several oil and gas fields, emission sampling was taken manually. After data were collected, they were transformed (Figure 2). Data were converted to a universal unit and categorized as a panel dataset. The oil and gas production were monitored at the valve before storage, while energy usage was monitored based on electricity usage in the fields during oil and gas production. Data were from 25 companies of various scales of production and management located in different regions of Indonesia.

As previously mentioned, the greenhouse gas emissions were calculated from CO₂, CH₄, and N₂O emissions monitoring data. In addition, the greenhouse gas emissions calculation in a single field was determined using Equation (1), as per Indonesia’s Ministry of Environment and Forestry [57]:

$$GHG=\frac{\left(EDC{O}_2\times n\right)+\left(EDC{H}_4\times n\right)+\left(ED{N}_{2}O\times n\right)}{r }$$

(1)

where GHG is yearly greenhouse gas emissions load (ton CO₂eq/year) and n is the annual stack operation (days) and r is the conversion factor from kilograms to tons (1000 kg/ton). EDCO₂, EDCH₄, EDN₂O are the daily emission loads of CO₂, CH₄, and N₂O (kg/day), which were calculated using Equations (2)–(4), respectively:

$$EDC{O}_2=CC{O}_2\times EF\times V_{av}\times A\times h\times t$$

(2)

$$EDC{H}_4=CC{H}_4\times EF\times V_{av}\times A\times h\times t$$

(3)

$$ED{N}_{2}O=C{N}_{2}O\times EF\times V_{av}\times A\times h\times t$$

(4)

where CCO₂, CCH₄, and CN₂O are the concentrations of carbon dioxide, methane, and dinitrogen oxide emission, measured at the pollutant sources (mg/Nm³). EF is the CO₂, CH₄, and N₂O emission factor valued at 1, 25, and 298, respectively, for greenhouse gas emissions [58]. V_av is the flow rate (m/s) (normal flow rate was used for manual sampling and the average flow rate was used for automatic sampling), A is the stack cross-sectional area (m²), h is the constant number of 0.0036 indicating mass/time conversion factor for 1 mg/second (kg/hour), and t is the daily stack operation (hour).

Energy usage data were collected from electricity and fuel consumption in the upstream petroleum fields during oil and gas production. The electricity consumption was monitored daily during the production process and for the supporting systems. Fuel consumption in the petroleum fields consists of energy usage but excludes the centralized electricity system. Activities that consume fuel include crane operation, emergency pumping, running a remote diesel generator for inspection, and other process activities not covered by the electricity system.

The yearly energy usage (EU) (gigajoule (GJ)/year) was calculated using Equation (5):

$$EU=\left(ec\times ef\times d\right)+\left(fc\times t\times ff\times d\right)$$

(5)

where ec is electricity consumption (kWh), fc is fuel consumption (liter/hour), t is operation hours per day (hour), d is the annual energy or fuel usage in a year (days), ef and ff are energy unit conversion factors for electricity and fuel to gigajoule (0.0036 GJ/kWh and 0.0342 GJ/liter · hour⁻¹), respectively.

Oil and gas production data were monitored before the main collecting station or storage for daily product measurement. The petroleum fields produced different products based on subsurface characteristics. Several fields produce only crude oil, some only natural gas, while others produce both natural gas and crude oil. Therefore, the total production, presented as an oil equivalent, was calculated from natural gas and crude oil production in the upstream petroleum fields (Equation (6)).

$$P=\left(DNG\times w\times m_n\right)+\left(DCO\times j\times m_o\right)$$

(6)

where P is the total production of crude oil and natural gas in one year (TOE/year), DNG is daily natural gas production (MMcf/d), DCO is daily crude oil production as barrel oil equivalent (BOE/day), m_n and m_o are daily production for crude oil and natural gas, respectively, in one year (days), w is the natural gas conversion factor (25 TOE/MMcf/d) to ton oil equivalent TOE, and j is the crude oil coefficient (0.136 TOE/BOE) to TOE [59].

3.2. The Statistical Model

Multiple events observed over various periods for the same individuals are studied using panel data analysis [60]. This analysis technique identifies cross-sectional and time-series dimensions based on the overall information available in the observation sample. [61].

The statistical model purposely identified greenhouse gas patterns in the upstream oil and gas field and the behavior between energy usage and production toward emission. Using three variables closely related to production and generated emission, the emission behavior during oil and gas production may be studied to formulate policy recommendation. Current data covered 4 years from 25 upstream petroleum fields, providing time-series and cross-sectional data. Therefore, the short panel data model was suitable for determining the greenhouse gas emission patterns in the petroleum fields. Since the petroleum fields are managed by different companies, measured at different scales, and have different productions, a natural logarithm was used to accommodate the data scales of different fields as percentage variables. The use of energy in the petroleum industry increases GHG emissions and is also required for the production of natural gas and crude oil. Consequently, the model was constructed using these three influential factors. We used a short panel data model to examine the impact of energy usage and oil and gas production on greenhouse gas emission in the upstream petroleum fields and their main process, which is summarized below (Equation (7)):

$$FGH{G}_{it}={\alpha }_{it}+{\beta }_{1i}FE{U}_{it}+{\beta }_{2i} F{p}_{it}+{\epsilon }_{it}$$

(7)

where FGHG is petroleum field greenhouse gas, FEU is petroleum field energy usage, Fp is petroleum field production, α is intercept, β_1-2 are the coefficients of the independent factors containing individual and time data, i is individual dimension of different upstream field data in panel dataset, t is time-period data sampling in panel dataset, ε is the error term describing any differences across groups to enter the model. The field-scale model covered data from the main process, supporting processes, and supporting facilities, including theoretical greenhouse gas emissions and energy usage in all upstream areas. The field panel data model (Equation (7)) was developed to understand the relationship between greenhouse gas emissions during oil and gas production and production activity in the remote upstream fields. Since the field data covered whole field areas, there is the possibility that energy usage and greenhouse gas emissions data indirectly relating to oil and gas production (staffing and material transport, field housing, and other activities) were included. However, another model was developed to determine the energy usage of greenhouse gas emissions during oil and gas production in the upstream fields. Considering the detailed oil and gas production activity, energy usage may also influence production. Therefore, it was essential to specify a model for the main process. Equation (8) presents the panel data model for the main process of the upstream petroleum fields.

$$MPGH{G}_{it}={\alpha }_{it}+{\beta }_{1i}MPE{U}_{it}+{\beta }_{2i} F{p}_{it}+{\epsilon }_{it}$$

(8)

where MPGHG is main process greenhouse gas, MPEU is main process energy usage, Fp is petroleum field production, β₁ and β₂ are the coefficients of the main process’s energy usage and of oil and gas production, respectively, α is intercept, i is individual dimension of different upstream field data in panel dataset, t is time-period data sampling in panel dataset, ε is the error term describing any differences across groups to enter the model. Given the data limitation, the short panel data model may give an insight into greenhouse gas and production of oil and gas activity in the petroleum fields. The effect of random and fixed variables and the panel’s cross-sectional dependence were investigated using both models’ Breusch–Pagan Lagrange multiplier test.

4. Results and Discussions

4.1. Data Summary

Upstream petroleum data collected from several fields in Indonesia were collated into a dataset. From the 25 petroleum fields and four years of data, we obtained 100 successful observations.

Table 2. Descriptive statistics and definitions of the original measured variables and the natural logarithm of the panel data model’s variables (Obs is observations).

Variable	Definition	Unit	Obs	Mean	Std. Dev.	Min	Max
FGHG	Total greenhouse gases	Ton CO2eq	100	236,018.90	325,005.40	165.11	2,096,406.00
MPGHG	Main process greenhouse gases	Ton CO2eq	100	195,727.40	270,744.10	163.87	1,677,124.00
FEU	Total energy usage	GJ	100	1,780,789.00	2,392,510.00	1599.66	11,100,000.00
MPEU	Main process energy usage	GJ	100	1,609,078.00	2,231,300.00	822.57	10,700,000.00
Fp	Production	TOE	100	1,724,020.00	2,407,033.00	43,730.50	10,200,000.00

presents the variables’ definitions and summary statistics of the original and logarithmic variables used in the panel data model. The independent and dependent variables were logarithmically transformed since the different data scales varied among petroleum fields, as shown by a significant standard deviation in the original data. The lowest greenhouse gas emissions for the petroleum fields’ activities and the main processes were 165.11 and 163.87 ton CO₂eq, respectively. A similar pattern was shown for energy usage, with the minimum for the total of the petroleum fields’ activities and the main process being 1599.66 and 822.57 GJ, respectively. Since the data were obtained from the petroleum fields’ automatic monitoring systems, the anomaly activities, maintenance frequency, and well shut down were not reported. Therefore, for one field, it is possible that monitoring data varied greatly.

4.2. Greenhouse Gas Emissions by the Upstream Petroleum Industry

The oil and gas industry in Indonesia is categorized as a national strategic industry. Including upstream petroleum fields, the industry is Indonesia’s main income [27]. Greenhouse gas emissions in the upstream petroleum fields are mainly from energy consumption that supports oil and gas production. The majority of emissions come from power plants located in the field during oil and gas production and from the burning flare [55]. Because the majority of the fuel used in power plants is a fossil fuel, the combustion of the power plant generates carbon emissions. The flare is used to burn excess natural gas created during crude oil and natural gas production. Figure 3 illustrates the greenhouse gas emissions in the petroleum fields and their contributors. In 2018, the petroleum fields generated 5,750,966.71 ton CO₂eq, of which 56.7% of emissions were generated by onshore fields and 43.3% from offshore fields. The contributors of emissions for the onshore and offshore fields are also shown in Figure 3 (denoted as C with a field number (C1, C2, C3, etc.) since the field name is confidential) and are categorized according to onshore (Figure 3a) and offshore (Figure 3b) association. The fields with both onshore and offshore wells were included as contributors to greenhouse gas emissions for both offshore and onshore fields. The greenhouse gas contribution is presented as a percentage to illustrate the share of emissions in the different fields. The largest greenhouse contributor was for the mixed well fields, which have both offshore and onshore production wells (C2 and C4). The most significant contributor of greenhouse gases for the offshore fields was C23, and the onshore fields’ largest contributor was C24.

A considerable portion of the electricity and heat causing the emissions is directly derived from crude oil and natural gas extraction [54]. As a result, upstream petroleum fields’ greenhouse gas emission intensity is 0.18 CO₂eq/ton oil produced, similar to the petroleum refinery emission intensity, which ranges from 0.16 to 0.2 CO₂eq/ton oil produced [62]. According to our data, more than 5 million tons of CO₂eq were released during oil and gas production from offshore and onshore fields in 2018. Releasing greenhouse gas increases climate change potential in Indonesia [11]. Indonesia is a particularly vulnerable region to global warming [10] and, thus, releasing greenhouse gas should be reduced. However, reducing climate change potential in a developing country is challenging since different obstacles occur in the industry. Several obstacles faced by the upstream oil and gas industry include the low green commitment and difficulty in technical aspect when the field is located in a remote area.

4.3. Correlation between Greenhouse Gas, Energy Usage, and Production

To test the extent of multicollinearity in our model, we first conducted correlation tests between the variables. The relationship and potential variable change in value were studied using correlation of the original datas’ total and main process variables to understand the relationship between variables. The results of the correlations between greenhouse gas emission, energy usage, and production are shown in

Table 3. Correlation between variables in the upstream petroleum fields.

	FGHG	FEU	Fp
FGHG	1.0000
FEU	0.5403	1.0000
Fp	0.0380	0.5564	1.0000

and

Table 4. Correlation between variables in the main process of oil and gas production.

	MPGHG	MPEU	Fp
MPGHG	1.0000
MPEU	0.5495	1.0000
Fp	0.0952	0.5797	1.0000

. As expected, a relationship between energy usage and greenhouse gas shown in the tables exist in the field and main process model. The field and main process energy usage have a moderate positive correlation with greenhouse gas emission and total oil and gas production, while oil and gas production have a relatively low positive correlation to greenhouse gas emission. A similar pattern of correlation strength was also found for the main process’s data. The correlation does not show a noteworthy change after excluding the data not directly linked to oil and gas production. The main process’s correlation indicates that the correlation coefficient between greenhouse gas emission and energy usage is 0.5495, while the greenhouse gas and petroleum production correlation coefficient is 0.0952 (Table 4). The correlation value below 0.6 indicates that the variables did not show multicollinearity [46].

A moderate positive correlation is found between energy usage and greenhouse gas emission and indicates that the emissions in the petroleum fields are mainly caused by the energy usage. Similar to findings in recent research, greenhouse gas emission was mainly correlated with energy consumption [63]. Previous regional-based studies conclude that there is a strong correlation between energy consumption and greenhouse gas emission [42,64,65]. They also show that energy consumption is the major contributor to greenhouse gas emission [31,37,40,66,67]. Different studies also indicate the importance of energy usage during industrial production [68]. However, the low positive correlation between oil and gas production and greenhouse gas emission was unexpected. This pattern suggests that oil and gas production may not necessarily influence greenhouse gas emissions.

4.4. Panel Data Model Analysis of Greenhouse Gas Emissions

On a theoretical basis, pooled OLS, fixed-effect, and random-effect methods are used for estimating panel data models. However, we opted against using the pooled ordinary least squares because our dataset included cross-section and time-variance. As seen in

Table 5. Breusch–Pagan Lagrange multiplier test of the petroleum fields and main process models with greenhouse gas from the upstream petroleum fields used as a dependent variable (yearly greenhouse gas emission).

Dependent Variable	p-Value	Model
FGHG	0.0000	Random effect
MPGHG	0.0000	Random effect

, the Breusch–Pagan Lagrange multiplier test summary for the two models of the petroleum fields and the main process of greenhouse gas emission show similar results. This test was conducted to investigate the panel effect and heteroscedasticity within models. The regression assumption is independently and identically normally distributed (IIND, where the first I means identical). If the variances are not the same, the conclusion may be wrong (i.e., underestimated or overestimated). Therefore, a random effect model was used to overcome the heteroscedastic conditions for both models.

This research aims to understand the effect of energy usage and petroleum production on greenhouse gas emissions in the petroleum fields. The panel data model was established with the respective variables to study the greenhouse gas emission pattern. Since the activity in the petroleum fields is in remote areas for both offshore and onshore fields, the panel data model for the total greenhouse gas of the studied petroleum fields was determined. Thus, the variables describe all fields’ activities that support petroleum production. Another model that focused on the main process was established with variables directly linked to petroleum production to validate the total greenhouse model.

Table 6. Panel data model results with greenhouse gas as the dependent variable. The coefficient number is the calculated β₁ and β₂ factors in the panel data model.

Dependent Variable **	FGHG	MPGHG
Obs	100	100
R-square	0.1098	0.1082
EU	0.4422 *	0.4015 *
	(0.1577)	(0.1439)
Fp	−0.7695 *	−0.7365 *
	(0.2045)	(0.1976)

The dependent variables are the upstream petroleum fields’ activities (represented by energy usage) and oil and gas production (represented by production). GHG is greenhouse gas emission and Obs is the number of observations. The robust standard errors are in parentheses; * p-value is <0.05; ** dependent variable (energy usage and total production) is scaled differently based on the model’s scale. If the model’s scope is petroleum field, the energy usage and total production data are on the petroleum field. If the model’s scope is the main process, the dependent variables only include the main process data.

denotes the result of a panel data analysis for the total and main process’s greenhouse gas model. The overall R-square of the total of the petroleum fields model was 0.1097, indicating that approximately 11% of the variation of greenhouse gas emission was explained by field energy usage and total production variables. The p-value of the dependent variables (total energy usage and petroleum production) was less than 0.05 against total greenhouse gas emissions. Therefore, the coefficient of the respective independent variables was significant in determining greenhouse gas emissions. In contrast, a random-effect was used for the main process panel data model. The overall R-square of the main process model was 0.1082, indicating that approximately 11% of the main process greenhouse gas emission variation may be explained by energy usage and petroleum production. Similar to the total of the petroleum fields model, the coefficient of energy usage and production in the main process were significant at a five percent level (p-value < 0.05).

The statistical panel data model for the petroleum fields has been successfully established in this study with 100 observations over a four-year time series and from 25 upstream petroleum fields. However, the original data showed a high deviation caused by the different scales used at the petroleum fields and their different production. Thus, it was necessary to transform the data into natural logarithms for use in the panel data model. The data were obtained from monitoring points regulated by the Indonesian government. Stacks and flares are frequently included in mandatory monitoring during petroleum production, which operates with multiple gas stream compositions [69].

The current panel data model (Table 6) describes the significant influence of both energy usage and petroleum production on greenhouse gas emissions. A positive coefficient is shown for energy usage, and a negative coefficient is shown for petroleum production. Therefore, the greenhouse gas emissions can be minimized by maintaining constant energy usage in the main process to stabilize oil and gas production. Other research using panel data has examined various primary sectors’ impacts on greenhouse gas emissions in various locations [23,33,48,49]. As mentioned previously, production data, energy usage, and greenhouse gas emissions were obtained during petroleum production, but infrastructure data of additional wells or units in the fields were excluded. In several field wells, the natural resource lifting process is still processed using natural pressure from the pressurized sub-surface, thereby minimizing energy usage during the lifting process. Unlike the refinery process requiring tons of energy for crude oil heating and processing, the low energy intake is used for different pumps and utilities. Therefore, the potential greenhouse gas emissions can be minimized by maintaining a constant energy usage thereby supporting further oil and gas production. A similar pattern is shown in the main process panel data model, which has a similar coefficient pattern to that of the petroleum fields model, as expected. The pattern similarity between the models indicates that the main process panel data model validates the petroleum fields model. The correlation test also presents a similar correlation pattern between petroleum field variables and main process variables. Due to increased oil and gas production in upstream extraction fields, all factors contributing to greenhouse gas emissions were closely tied to fossil fuel energy consumption [50].

4.5. Energy Efficiency Mitigation in the Offshore and Onshore Fields

The energy mitigations in the petroleum fields are discussed to elucidate potential energy mitigations in the upstream petroleum fields. As most emissions are generated from power plants and burning flare, the energy mitigation to reduce the power plant load is important to be implemented. The energy mitigation program is classified based on the mitigation types. Five different mitigations were identified for the upstream petroleum fields in 2018 (Figure 4). It is common to optimize field utilities and pumps to maintain a stable oil and gas production. Utility optimization includes facility optimization, forklift replacement, compressor management, stabilizer repair, and other utility-related energy management activities for oil and gas production. The petroleum fields’ main process requires different types of pumps for lifting and transferring the crude oil and this component is considered as a primary factor in oil and gas production. Therefore, keeping the pumps in optimal condition is mandatory for maintaining production. Pump optimization usually consists of pump replacement, maintenance, and management. Pump replacement and management are done to a greater extent in the offshore field, while pump maintenance is commonly conducted in the onshore field. Onshore fields’ energy mitigations are more varied than those of the offshore fields since the onshore fields are more accessible than the offshore fields. Therefore, the number of energy mitigation options for onshore fields is commonly higher than for offshore fields, except for the power plant management category.

Table 7. The cost of different energy mitigations in the upstream petroleum field and potential greenhouse gas (GHG) reductions calculated for 2018. Mitigations are categorized based on the type of energy management.

Energy Efficiency Mitigation	Offshore		Onshore
	Cost (USD)	GHG Reduction (Ton CO2eq)	Cost (USD)	GHG Reduction (Ton CO2eq)
Gas flare reuse and management	-	25,194.23	3514.06	27,078.52
Pump maintenance and optimization	788,182.55	4913.07	2334.38	2503.55
Utility-related optimization	241,440.05	263.13	579,821.62	219,352.83
Power plant efficiency	477,772.01	3.61	89,922.90	285.06
Fuel substitution	239,871.55	3293.35	506,770.89	4400.22
Average	349,453.23	6733.48	236,472.77	50,724.03

presents energy mitigation costs and potential reduction from implementing energy-efficient management activities. The mitigation cost is identified from the annual cost spent on implementing energy-efficient management. However, the detailed cost breakdown and the proportion divided between petroleum fields’ operational costs and mitigation programs are not further investigated since the data are confidential. There is a possibility that the efficient energy implementation cost is incorporated in maintenance or production costs, which is included as greenhouse gas reduction effort since it is found that the petroleum fields’ maintenance also has the potential to reduce greenhouse gas emissions. The cost and greenhouse gas reduction range varies since the upstream petroleum field scales differ.

Potential greenhouse gas reduction data were accumulated from reduction data collected in 2018. Different results and patterns were obtained for the onshore and offshore fields (Table 7). The greenhouse gas reduction for onshore fields was, on average, higher than for the offshore fields, and utility optimization was the major contributor to the greenhouse gas reduction. In the upstream petroleum fields, flare reuse contributes a considerable amount to GHG emissions. The original design of the current offshore platform usually includes gas flare reuse for a petroleum field’s powerplant, while gas flare reuse in onshore fields is not included in the initial production design. Therefore, in the onshore fields, companies modify the gas flare line to reuse it in their powerplant to cut the operational cost of energy usage. The greenhouse gas reduction from utility-related optimization costs more in offshore fields. However, it may result in a significant greenhouse gas reduction. Power plant efficiency is related to the maintenance, management, and additional utility for maintaining a more efficient powerplant. This mitigation excludes fuel-related efficiency. Fuel substitution is not limited to the main powerplant but includes portable powerplants for remote maintenance and for operational uses unrelated to the primary power plant.

Oil and gas production is a primary target of business performance in the upstream industry, and many view it as only self-serving. However, a field’s components for the main process need to be in prime condition to maintain a stable production. Our results suggest that maintaining a constant energy usage during primary production stages by means of implementing energy mitigation at the petroleum fields is advantageous for oil and gas production as well as for greenhouse gas emissions reduction.

According to the triple bottom line concept, the environment is at the same level as the economy (profit) and society [70]. Therefore, greenhouse gas reduction should ideally be of the same priority as the economy and society. However, profit and expense are significant factors when implementing environmental protection in the upstream industry. Enforcing National law should be supported with strong commitment and implementation, which entails several expenses. Usually, an environmental protection policy includes ecosystem recovery, pollution minimization inside and outside fields, and is separated from production policy, which lacks production benefit. Furthermore, greenhouse gas reduction in the upstream petroleum industry is hampered by serious financing problems, impacting profit for the companies. Our results suggest that reducing greenhouse gas emissions has the potential to benefit the company if it is planned strategically. The greenhouse gas reduction policy should first ensure energy efficiency in the main process thereby also increasing oil and gas production.

Upstream petroleum companies usually face a dilemma when government regulates a mandatory national greenhouse gas reduction. In contrast, the company is encouraged to boost its oil and gas production to fulfill the national income. The oil and gas extraction industry is highly dependent on the natural resource reserves in the earth’s crust. Production will move to other rich reserve fields when the sub-surface no longer contains oil and gas reserves. Usually, a newly established extraction well is excluded when natural gas or crude oil production is inadequate. Therefore, it is important to understand the relationship between the three interconnected factors of oil and gas production, energy usage, and greenhouse gas emissions during their production cycle. Energy usage plays a vital role in greenhouse gas emissions and oil and gas production. Correlation analyses indicated that energy usage was moderately correlated with greenhouse gas emissions and oil and gas production. However, there was a low significant correlation between oil and gas production and greenhouse gas emissions.

5. Conclusions and Policy Implications

There are various greenhouse gas emissions studies in the petroleum industry [25,54,55,66]. However, there has been limited use of panel data investigation in the upstream petroleum industry thus far. Given this limitation, our research aims to understand the relationship between oil and gas production, energy usage, and greenhouse gas emissions at 25 Indonesian upstream petroleum fields and their main processes. The panel data regression shows that energy usage, natural gas, and crude oil production negatively affect greenhouse gas emissions. Similar regression patterns were identified for the petroleum fields and main process models. The correlation test also presents moderate and low significant correlations between variables.

There are several policy implications generated from this study. First, most energy usage occurs during the main oil and gas production process and is attributed to the lifting pump, transfer pump, the natural gas purification process, and other utilities. Since most upstream petroleum fields are located in remote areas, energy performance is an important requirement for crude oil and natural gas production. Therefore, an improved greenhouse gas reduction policy and scenario could be initiated by targeting the main oil and gas production stage. Second, the environmental policy and roadmap for environmental protection in the upstream petroleum industry should include energy management for the target production. In addition, the production policy and greenhouse gas reduction policy should not be separated since the factors influence each other. Optimizing existing energy usage in the production stage at upstream petroleum fields will boost oil and gas production. At the same time, the existing production efficiency will lead to greater greenhouse gas reduction. Third, an indirectly profitable green policy for reducing greenhouse gas would be more appealing to the decision-makers in upstream petroleum companies. Potential greenhouse gas reduction, which can be integrated with production profit, varies based on energy usage management. The main factors associated with oil and gas production performance that are profitable for greenhouse gas reduction in the upstream petroleum fields include powerplant efficiency, pump maintenance, utility optimization, and gas flare reuse. These factors may be included in profitable, sustainable green policies.

From the government’s policy perspective, encouraging the upstream petroleum industry to boost the energy efficiency of their target production is an important policy for minimizing greenhouse gas emissions before making upstream companies cope with general greenhouse gas reduction plans. Moreover, energy efficiency will be a priority since it may lower oil and gas production expenses. Therefore, an improved greenhouse gas reduction policy can be formulated to achieve a well-balanced and more profitable low-carbon implementation.

This study has a few limitations. Firstly, the data collected from the companies are limited to production monitoring and cannot represent all activities in the petroleum fields. Thus, further research could be conducted to investigate GHG mitigation and petroleum field behavior to mitigate GHG. Secondly, the data set may be relatively small compared to that of a national policymaker since only a four-year time series was considered in this study. Therefore, the result and policy implication may apply limited to the studied upstream petroleum field. Thirdly, although our findings suggest the opposite, it is theoretically possible that an increase in production will also increase greenhouse gas emissions. This may be caused by differences in production mechanisms at the petroleum fields. However, to test this theory for national policy evaluation, a longer time series data set is required to validate the panel model’s coefficient and statistical patterns.

Finally, this study contributes to the literature by identifying the relationships between greenhouse gas emissions, energy usage, and oil and gas production in Indonesia’s upstream petroleum fields using short panel data analysis, which has not been previously investigated. The result successfully identifies the interaction of the studied variables at the petroleum fields. The regression model shows a significant coefficient for independent variables, which may be used for greenhouse gas projection in the observed fields. Finally, this study may trigger further studies regarding a similar topic in the upstream petroleum fields since the fields are open for environmental improvement.