Insights into Small-Scale LNG Supply Chains for Cost-Efficient Power Generation in Indonesia

Abstract

This study demonstrates that small-scale liquefied natural gas (SS LNG) is a viable and cost-effective alternative to High-Speed Diesel (HSD) for power generation in remote areas of Indonesia. An integrated supply chain model is developed to optimize total costs based on LNG inventory levels. The model minimizes transportation costs from supply depots to demand points and handling costs at receiving terminals, which utilize Floating Storage Regasification Units (FSRUs). LNG distribution is optimized using a Multi-Depot Capacitated Vehicle Routing Problem (MDCVRP), formulated as a Mixed Integer Linear Programming (MILP) problem to reduce fuel consumption, CO₂ emissions, and vessel rental expenses. The novelty of this research lies in its integrated cost optimization, combining transportation and handling within a model specifically adapted to Indonesia’s complex geography and infrastructure. The simulation involves four LNG plant supply nodes and 50 demand locations, serving a total demand of 15,528 m³/day across four clusters. The analysis estimates a total investment of USD 685.3 million, with a plant-gate LNG price of 10.35 to 11.28 USD/MMBTU at a 10 percent discount rate, representing a 55 to 60 percent cost reduction compared to HSD. These findings support the strategic deployment of SS LNG to expand affordable electricity access in remote and underserved regions.

1. Introduction

Indonesia’s power generation capacity increased from 50.4 GW in 2014 to 85.1 GW in 2023, with a compound annual growth rate (CAGR) of approximately 5.86% [1]. In 2022, electricity production reached 183,819.03 GWh, supplied by coal (66%), natural gas (13.6%), hydropower (7.24%), bioenergy (6.41%), biofuels (6.3%), geothermal (4.43%), and oil (2.20%) [2,3]. Between 2015 and 2021, the national average levelized cost of electricity (LCOE) was 0.0786 USD/kWh, with specific values of 0.048 USD/kWh for coal, 0.087 USD/kWh for natural gas, 0.026 USD/kWh for hydropower, 0.099 USD/kWh for geothermal, and 0.29 USD/kWh for oil [4]. Oil-based or High-Speed Diesel (HSD) power plants, such as diesel engines, gas engines, and open-cycle gas turbines, tend to have higher LCOEs due to costly fuel prices. To reduce generation costs and improve energy sustainability, the Indonesian government, through the Electricity Supply Business Plan (RUPTL) 2021–2030 [5], aims to convert HSD-fueled power plants to natural gas, particularly in central and eastern regions of the country.

Natural gas is a fossil fuel in gaseous form, primarily composed of hydrocarbons, and is commonly found in oil fields, natural gas reservoirs, and coal seams [6]. Its composition varies, with methane as the main component, accompanied by paraffinic hydrocarbons like ethane, propane, and butane, as well as minor fractions of C5+ hydrocarbons and aromatic compounds such as benzene, toluene, and xylene [7]. Widely used in the petrochemical industry and power generation, natural gas is transported via pipelines, compressed natural gas (CNG), or liquefied natural gas (LNG), with the choice depending on supply locations and demand volumes. LNG is the preferred method for long-distance transport, especially where pipelines are unavailable [7,8]. It is produced by cooling natural gas to −160 °C at atmospheric pressure, converting it into liquid form after removing impurities like carbon dioxide, water, sulfur, and mercury. LNG consists of 70–90% methane (CH₄), up to 20% propane and butane, and traces of carbon dioxide, nitrogen, and hydrogen sulfide. The liquefaction process enhances transport efficiency by reducing gas volume up to 600 times, making it a cost-effective solution for large-scale transportation [9].

The LNG supply chain begins with transporting natural gas from gas fields to liquefaction plants, where it is processed and stored in tanks before being loaded onto LNG carriers. These carriers deliver LNG to receiving terminals, where it undergoes regasification, converting it back into gas for distribution and consumption [8]. LNG terminals store and process LNG, converting it from liquid to gas. They supply power plants and industrial users. LNG terminals can be onshore or offshore. Offshore terminals have different configurations, including the Floating Storage Unit (FSU) and Floating Storage and Regasification Unit (FSRU). Other types include FSRU with Floating Storage Power Plant (FSPP) and Floating Storage, Regasification, and Power Plant (FSRPP). FSPP is a floating facility that stores LNG and generates electricity onboard, then transmits power to the grid, eliminating the need for onshore infrastructure. FSRPP combines storage, regasification, and power generation in one floating unit, offering a compact and fully offshore LNG-to-power solution. Both are ideal for remote or coastal areas with limited land access. Each configuration has unique technical and economic benefits. Over the past decade, FSRUs have emerged as the preferred choice for new LNG import markets due to lower investment costs, faster deployment, and higher flexibility. As of February 2024, 16 markets fully rely on floating terminals, 10 use a mix of floating and onshore facilities, while 21 depend solely on onshore terminals. However, established markets favor onshore terminals for their larger storage capacity, better scalability, and lower exposure to weather risks and vessel leasing costs [10].

In 2022, Indonesia produced 5444.81 BBTUD of natural gas, with 65 percent used domestically for power generation, petrochemicals, and other industries. LNG production reached 34.52 million tons, of which 70 percent was exported [11]. While pipelines remain the main distribution method, areas without access rely on LNG infrastructure of varying scales. Small-scale LNG (SSLNG) includes compact liquefaction units, storage, and receiving terminals, offering flexible supply to remote and island regions. LNG is transported by small carriers to onshore or offshore terminals. Floating solutions like Floating Storage Regasification Units (FSRUs) are well-suited for SSLNG due to lower costs and faster deployment [12,13,14,15].

In deploying SSLNG systems, optimizing LNG distribution from multiple supply points to scattered demand locations is critical, especially in archipelagic regions like Indonesia. The Vehicle Routing Problem (VRP) provides a useful framework for modeling these distribution logistics by optimizing fleet routes to minimize total travel distance and delivery costs. Given the high fuel costs associated with LNG transportation, route efficiency plays a key role in reducing overall supply chain expenses [16,17]. A widely used variant, the Capacitated Vehicle Routing Problem (CVRP), involves a fleet of vehicles with limited capacity serving multiple customers from a depot. Each vehicle must complete a route starting and ending at its assigned depot while ensuring that all deliveries are fulfilled without exceeding vehicle capacity. Travel costs are typically calculated using inter-node distances (d_ij), where node 0 represents the depot and node n is the number of customers [16,18]. CVRP is further categorized into Single Depot CVRP (SDCVRP) and Multi Depot CVRP (MDCVRP). While SDCVRP is applicable to centralized distribution systems, MDCVRP is more suitable for SSLNG applications involving multiple supply points and widely scattered demand locations [16].

Various studies have explored SSLNG supply chain optimization using different modeling approaches. One of the most widely used is the CVRP, which focuses on optimizing distribution routes from depots to multiple demand points under vehicle capacity constraints. To solve CVRP and related logistical problems, many researchers have employed Mixed Integer Linear Programming (MILP) as the underlying mathematical formulation. Jokinen et al. [19] used MILP to optimize SSLNG distribution along coastal areas, minimizing LNG prices, operational costs, and investment. Bittante et al. [20,21] extended this by integrating maritime and land transport, optimizing fleet configuration and terminal locations, while Bittante and Saxén [22] developed a multi-period MILP model for inventory and network design. Pratiwi et al. [23] and Surury et al. [24] applied MILP to include cost efficiency and emissions in LNG logistics planning.

Heuristic and clustering-based methods have also been applied to solve CVRP. Budiyanto et al. [25,26] used CVRP and greedy algorithms for SSLNG routing in Sumatra and Eastern Indonesia, incorporating economic indicators such as Net Present Value (NPV) and Internal Rate of Return (IRR). Cahyo et al. [27] implemented saving matrix and nearest neighbor methods for terminal routing, while Arif et al. [28] and Armyn et al. [29] applied clustering and genetic algorithms to improve route efficiency and reduce distribution risks in archipelagic settings.

Most previous studies analyze SSLNG transportation and terminal costs separately. This study addresses that gap by developing an integrated cost optimization model that simultaneously considers LNG transport routes, vessel allocation, and FSRU terminal handling costs. The model is formulated as a Mixed Integer Linear Programming (MILP) problem based on the Multi-Depot Capacitated Vehicle Routing Problem (MDCVRP) and incorporates CO₂ emissions as an external cost through a carbon tax mechanism. The novelty of this research lies in the unified modeling of SSLNG logistics and regasification infrastructure within a single optimization framework, which enables a more accurate assessment of total system costs and supports scalable deployment in archipelagic power systems.

2. Methodology

2.1. System Description

This study uses LNG demand projections for power plants in 2030 as outlined in the Electricity Supply Business Plan (RUPTL) 2021–2030 [5], published by PT PLN (Persero) and approved by the Ministry of Energy and Mineral Resources of the Republic of Indonesia. The RUPTL serves as an official national planning document, making its forecasts widely recognized and used by both government and industry. In line with this plan, the ministry has also mandated the expansion of LNG infrastructure and the conversion of oil-fueled power plants to LNG as part of Indonesia’s long-term energy diversification strategy [30]. To ensure implementation, the government guarantees gas supply for power generation through formal regulations, including the Decree on the Allocation and Utilization of Natural Gas for Electricity Supply by PT PLN (Persero), which secures gas volumes for long-term power needs [31]. The overall research methodology applied in this study, including the analysis of the SSLNG supply chain, is illustrated in Figure 1.

Based on this context, LNG demand points in the study are served by 50 receiving terminals, with some supplying multiple power plants, potentially requiring short-distance pipeline investments. Major large-scale LNG terminals such as FSRU Jawa 1, FSRU Jawa Barat, and FSRU Lampung are excluded due to their centralized infrastructure and dedicated supply chains, which do not align with the decentralized SSLNG focus of this model. Including them would create redundancy and reduce the model’s relevance to regions lacking established infrastructure. Their exclusion ensures that the model remains representative of SSLNG deployment scenarios in remote or underserved areas. LNG supply depots are assumed to be fed by existing and planned plants, including Badak NGL, BP Tangguh, Donggi Senoro, and Masela [32,33].

For nationwide SSLNG optimization with numerous demand locations, this study applies K-Means clustering to group supply and demand points into four geographic clusters based on Indonesia’s regional segmentation in the RUPTL 2021–2030: Sumatra, Central Indonesia, Maluku, and Papua [5]. This method supports efficient route planning by linking demand points to their nearest LNG supply plant. K-Means effectively partitions data into K clusters with high intra-cluster similarity and clear inter-cluster separation [34] and is widely used for geospatial logistics due to its simplicity, speed, and scalability [35]. Despite its advantages, K-Means has known limitations, such as requiring a predefined K value and sensitivity to initial centroid placement [34]. To validate clustering quality, this study uses the Silhouette Index, an unsupervised metric that evaluates cohesion and separation, with values ranging from −1 to 1; higher scores indicate better-defined clusters, and no labeled data are required [36]. A Silhouette Score analysis across K = 2 to 9 showed the highest score at K = 3 (0.4836), with K = 4 scoring comparably high (0.4718), supporting its selection for alignment with national planning. The clustering was implemented using the KMeans class from the scikit-learn library in Python 3.10, which enables efficient model training and prediction. K-Means is used in logistics to cluster dangerous goods transport data by volume and location, enhancing risk analysis and route planning. It improves accuracy through adaptive K selection and spatial feature extraction [37].

This study assumes the use of FSRU as the receiving terminal for LNG. FSRU offers a more cost-effective, faster-to-deploy, and flexible solution compared to permanent and more expensive onshore terminals. With costs 50–60% lower than onshore terminals and a construction time of only 27–36 months, FSRU provides a practical and efficient alternative [14,38]. Figure 2 shows the breakdown of the Capex comparison between the FSRU and onshore LNG terminal. The total Capex for the FSRU is USD 450 million, while the onshore LNG terminal amounts to USD 750 million.

The next step is to conduct a sensitivity analysis on key variables affecting the SS-LNG supply chain cost, particularly LNG demand and FSRU storage capacity. These variables significantly influence both distribution costs and FSRU investment. The total SS-LNG supply chain cost is calculated as the sum of (i) LNG distribution and ordering costs and (ii) FSRU handling and storage costs.

Using a one-at-a-time (OAT) method, the analysis varies FSRU storage capacity from 1 to 100 times the daily LNG demand at each terminal or power plant. This range enables the identification of the storage capacity level that results in the lowest total supply chain cost. The approach captures the trade-off between higher capital investment for larger storage and lower distribution frequency, helping define a cost-optimal SSLNG configuration.

2.1.1. Mathematical Model for SSLNG Distribution Cost

The Multi-Depot Capacitated Vehicle Routing Problem (MDCVRP) approach is used to optimize LNG distribution routes in Indonesia. This mathematical model includes several indices, parameters, and variables:

Sets
N	Set of all nodes or locations
D ⊆ N	Set of depot nodes
T ⊆ N	Set of terminal nodes
K	Set of ships
Indices

Indices i and j represent location points in the distribution network. Index k represents the identity of a ship within the set of ships involved in LNG transportation.

i, j	Location points (i, j = 0, 1, 2, …, N)
k	Ship k (k = 0, 1, 2, …, K)
Notation
dij	Distance from point i to point j (km)
fk	Fuel consumption of ship k (tons per km)
fc	Fuel cost (USD per ton)
ef	Emission rate (tons of CO2 per ton of fuel)
ct	Carbon cost or carbon tax (USD per ton of CO2)
vk	Speed of ship k (km per hour)
ts	Stoppage time for loading/unloading (hours)
rk	Charter cost of ship k (USD per hour)
qj	Demand at point j (BBTU per day)
sf	Safety stock factor (days)
capk	Capacity of ship k (BBTU)
ts	LNG unloading time (hours)
Decision Variables
xijk	Binary variable, 1 if distribution occurs from iii to j using ship k, 0 otherwise
yk	Binary variable, 1 if ship k is used, 0 otherwise
ui	Continuous variable used for sub-tour elimination (MTZ formula)

Objective Function

$$\mathrm{m}\mathrm{i}\mathrm{n}\mathrm{i}\mathrm{m}\mathrm{i}\mathrm{z}\mathrm{e}\mathrm{ }\mathrm{Z}\mathrm{ }=\mathrm{ }\sum _{i∊N,i\ne j}\sum _{j∊N}\sum _{k∊K}{x}_{ijk·}[d_{ij·}{f}_{k·}(c_f+e_f·c_t)+({\displaystyle \frac{d_{ij}}{v_k}}+2·t_s)·r_k]$$

(1)

With several conditions as follows

$$x_{ijk}= 0\forall i,j\in D:i\ne j,\forall k\in T$$

(2)

This constraint ensures that there are no direct routes from one depot to another depot without first passing through a terminal.

$$\sum _{i\in \mathrm{N},i\ne j}\sum _{k\in K}{x}_{ijk}=1 \forall j\in T$$

(3)

This constraint ensures that each terminal is visited exactly once by vehicle k, either from a depot or another terminal.

$$\sum _{i\in N, i\ne h}{x}_{ihk}=\sum _{j\in N, j\ne h}{x}_{hjk} \forall h\in T, \forall k\in K$$

(4)

The continuity constraint ensures that the LNG inflow to each terminal is equal to the LNG outflow from that terminal.

$$x_{ijk}\le y_k \forall k\in K, \forall i, j\in N,:i\ne j,$$

(5)

This constraint ensures that ship k is only used if there is a route (arc) assigned to it.

$$\sum _{i\in D}\sum _{j\in T}{x}_{ijk}=y_k \forall k \in K$$

(6)

This constraint ensures that if ship k is used, it must start its journey from a depot to a terminal.

$$\sum _{i\in T}\sum _{j\in D}{x}_{ijk}=y_k \forall k \in K$$

(7)

This constraint guarantees that if ship k is in operation, it must complete its journey by returning to a depot from a terminal.

$$\sum _{j\in T}{q}_j.s_f.\left(\sum _{i\in N, i\ne j}{x}_{ijk}\right)\le {cap}_k.y_k \forall k\in K$$

(8)

This constraint ensures that the total demand, including safety stock, on a single route does not exceed the capacity of ship k.

$$u_i-u_j+\left|T\right|\ast x_{ijk}\le \left|T\right|-1 \forall k\in K,\forall i, j\in T:i\ne j$$

(9)

The subtour elimination constraint is essential to prevent invalid cycles (subtours) in the solution. This constraint ensures that the LNG flow forms a valid route without disconnected loops. The formulation follows the Miller–Tucker–Zemlin (MTZ) [39,40].

$$\sum _{j\in D}{x}_{ijk}-\sum _{j\in D}{x}_{jik}=0 \forall i\in D, \forall \mathrm{k}\in \mathrm{K}$$

(10)

This constraint ensures that a ship must return to the same depot it departed from and cannot end its route at a different depot.

$$\sum _{i\in I}\sum _{j\in J}{x}_{i,j,k}\ast \left({\displaystyle \frac{d_{ij}}{v_k}}+2\ast t_s\right)\le s_f\ast 24 \forall \mathrm{k}\in \mathrm{K}$$

(11)

The objective function of this model is to minimize the total distribution cost, which consists of fuel cost, CO₂ emission cost, and LNG carrier rental cost. The CO₂ emission cost is calculated by multiplying the total CO₂ emissions from LNG carrier operations with the carbon tax (ct, in USD/tCO₂). Emissions are estimated using the direct emission factor of ship engines (ef, in g CO₂/kg fuel), allowing the model to internalize environmental externalities within the cost structure.

The model includes a constraint to ensure that the total LNG delivery time, including transportation and handling, does not exceed the available operating stock time. It also applies non-negativity constraints to all continuous decision variables, which are inherently handled in the formulation. However, real-world operational constraints, such as port capacity limits and boil-off gas (BOG) losses, are not considered. These elements are excluded to simplify the model, which presents a limitation that may reduce the accuracy of detailed operational planning. The model is implemented in Python 3.11.5 using Gurobi 10.0.3 and executed on a system equipped with a 12th Gen Intel^® Core™ i7-1255U (1.70 GHz) processor and 16 GB RAM.

2.1.2. Handling and Storage Cost Formulation

The FSRU investment cost is highly influenced by its LNG storage capacity [14]. As storage capacity increases, capital expenditure (Capex) also rises. The total capital expenditure (Capex) for FSRU is formulated as follows [41]:

$$Total capex terminal={Capex}_{FSRU}\ast LNG storage+{Capex}_{Jetty\&Facility}\ast LNG Storage$$

(12)

where:

Total capex terminal	= Total capital expenditure of the FSRU terminal (USD)
CapexFSRU	= Capital expenditure of the FSRU (USD/m3 LNG)
LNG storage	= Total inventory, including operating and safety stock (m3 LNG)
CapexJetty&Facility	= Cost of jetty, pipeline, and onshore facilities (USD/m3 LNG)

The project implementation cost covers the team, management, specialists, consultants, and administrative expenses before the final investment decision. The total project capital cost is calculated using the following equation [42]:

$$ICC=Con+Own+Cont$$

(13)

where:

ICC	= The total Initial Capital Cost during the construction phase or the total Capex of the LNG terminal;
Con	= The construction cost;
Own	= The project implementation expenses covered by FSRU project stakeholders;
Cont	= The contingency costs for unforeseen events.

2.1.3. Total Unit Cost Formulation

Unit cost is the cost incurred per unit of LNG, expressed as the total cost per MMBTU. In this case, total cost refers to the expenses generated during the distribution/transportation process and handling or storage at the LNG terminal. The total unit cost in this study is formulated as follows:

$$Total unit cost=unit transportation cost+unit handling and storage cost$$

(14)

$$Total unit cost={\displaystyle \frac{Transportation cost}{Total LNG transported}}+\left({\displaystyle \frac{\left(\frac{ICC-Salvage value}{Project lifetime}\right)+Opex LNG terminal}{Daily LNG demand\ast 365}}\right)$$

(15)

where:

Total unit cost	= the combined cost of transportation, storage, and regasification at the FSRU (USD/MMBTU);
Transportation cost	= the total transportation expense (formulated in Equation (1)) based on LNG demand or inventory (USD);
Total LNG trans.	= the total amount of LNG, measured in energy units, delivered from the depot to the LNG terminal (MMBTU);
ICC	= the total Initial Capital Cost (USD), as formulated in Equation (13);
Salvage value	= the residual value of the LNG terminal assets at the end of their operational lifetime (USD);
Project lifetime	= the operational lifespan of the LNG terminal (years);
Opex LNG terminal	= the operational costs of the LNG terminal (USD/year);
Daily LNG demand	= the amount of LNG required by the terminal (MMBTU/day).

2.2. Economic Feasibility

Economic feasibility analysis is calculated based on financial viability parameters, which assess an investment’s ability to generate returns. In financial economic analysis, several key evaluation criteria are used, including NPV (Net Present Value), IRR (Internal Rate of Return), and payback period.

NPV is commonly used to determine the profitability of an investment, indicating whether the investment generates a profit or a loss. If NPV > 0, the project is financially viable, while NPV < 0 indicates a loss. The formula for calculating NPV is as follows:

$$NPV=\sum _{t=0}^n{\displaystyle \frac{C_t}{{(1+r)}^t}}$$

(16)

where:

NPV	= Net Present Value (USD);
Ct	= Cash flow in year t (USD);
r	= Discount rate (%);
t	= Time period (years);
n	= Total project lifetime (years).

The Internal Rate of Return (IRR) is the discount rate at which the NPV becomes zero. The IRR indicates the annualized return on investment. If the IRR exceeds the required return, the project is considered financially feasible. It is determined using the following equation:

$$\sum _{t=0}^n{\displaystyle \frac{C_t}{{(1+IRR)}^t}}=0$$

(17)

where:

IRR	= The rate of return that equates to cash inflows and outflows;
Ct	= The cash flow at time t (USD);
t	= The time period (years);
n	= The total project lifetime (years).

The payback period (PBP) refers to the time required for an investment to recover its initial cost through accumulated net cash flows. It is calculated using the formula:

$$PBP={\displaystyle \frac{Investment}{Annual net cash flow}}$$

(18)

where:

PBP	= The duration (in years) needed to recoup the investment.
Investment	= The total capital expenditure of the project (USD).
Annual net cash flow	= The yearly net income generated by the investment after deducting key expense (USD/year).

2.3. General Assumptions in SS LNG Supply Chain Modeling

Table 1. General assumptions in the economic simulation of the SS-LNG supply chain.

Item	Unit	Value/Remark	Ref.
Demand size	m3 LNG/day	Based on ref.	[5,30,31]
Safety stock	days	3	[43]
Operating stock	days	Varied	Assumption
Shipping size	m3 LNG	Daily LNG demand x Operating stock	Simulated
Ship fuel (HSD B35) price	USD/ton	1163	Assumption
Direct emission factor of ship engine	g CO2/kg fuel	3.140	[44]
Carbon tax	USD/t CO2	2	Adapted from: [45]
LNG storage capacity	m3 LNG	Daily Demand x operating stock + Daily demand x safety stock	[46,47]
Unloading capacity	m3 LNG/hours	10,000	[48]
Unloading duration	hours	4	Assumption
Pipeline capex (including 2 compressors)	USD/km	40,000	[49]
Capex FSRU	USD/m3 LNG storage capacity	1748.23	[41]
Capex jetty, pipeline, and on shore facility	USD/m3 LNG storage capacity	611	[14]
Owner cost	% from capex terminal	15	[14]
Contingency cost	% from capex terminal	10	[14]
Sales tax	% from sales	12	[50]
Corporate tax	% earning	22	[50]
Opex	% capex terminal	2.5	[41]
Salvage value	% capex terminal	25	[42]
Depreciation	straight method for 25 years		Assumption
Project lifetime	years	25	[42,51]
Equity	% total investment cost	30	[52]
Debt	% total investment cost	70	[52]
Interest rate	%	10	[53,54]

summarizes the key assumptions used in the SS-LNG model, including transportation and terminal construction parameters. A 10% interest rate is applied, reflecting the peak of Indonesia’s benchmark rate over the past 15 years, and serves as a conservative basis for evaluating economic indicators such as NPV, IRR, PBP, and the gas selling price margin (USD/MMBTU). A sensitivity analysis is conducted by varying the discount rate from 5% to 20% to assess its impact on NPV and pricing outcomes.

The carbon tax is set at USD 2/tCO₂, based on Indonesia’s initial carbon tax plan for power generation. This value is relatively low compared to carbon taxes in many European countries, which often exceed USD 50/tCO₂. The technical and cost assumptions for LNG carriers used in the model are detailed in

Table A1. LNG carrier data (adopted from reference [29]).

LNG Carrier	Capacity (m3)	Capacity (BBTU)	Speed (Knot)	Speed (km/Hour)	Fuel (Ton per km)	Rent (USD/Day)	Fuel (Ton/Day)
Shinju Maru	2500	52.97	13	24.08	0.013331	11,679	7.7
WSD59 3K	3000	63.56	12	22.22	0.019487	12,730	10.4
WSD59 5K	5000	105.93	14	25.93	0.026533	16,933	16.5
WSD59 6.5K	6500	137.71	13	24.08	0.023547	20,085	13.6
Coral Methane	7500	158.90	14	25.93	0.032935	22,187	20.5
Norgas	10,000	211.86	14	25.93	0.042730	27,441	26.6
WSDS55 12K	12,000	254.24	14	25.93	0.030042	31,643	18.7
Coral Energy	15,600	330.51	15	27.78	0.065083	39,207	43.4
WSD50 20K	20,000	423.73	15	27.78	0.037653	42,538	25.1
Surya Satsuma	23,000	487.29	15	27.78	0.094204	44,808	62.8
WSD50 30K	30,000	635.59	16	29.63	0.048937	69,464	34.8

.

3. Results and Discussion

3.1. Clustering Supply and Demand

Figure 3 and

Table A2. Details of location and demand capacity of LNG terminal.

Terminal	Cluster, Terminal Number	Latitude	Longitude	BBTUD	m3 LNG/Day	Remark
Ambon and Seram 1	A1	−3.54956	128.3335	6.51	307.27	Proposed LNG terminal facility
Bacan	A2	−0.63148	127.4862	0.73	34.46	Proposed LNG terminal facility
Bula	A3	−3.0994	130.4926	0.93	43.9	Proposed LNG terminal facility
FSRU Gorontalo	A4	0.464411	122.0072	5.11	241.19	Existing terminal facility
Halmahera, Sofifi, and Tidore 2	A5	0.703122	127.5269	3.52	166.14	Proposed LNG terminal facility
Halmahera Timur	A6	0.838249	128.2552	16.08	758.98	Proposed LNG terminal facility
Minahasa	A7	1.679776	125.0795	10.72	505.98	Proposed LNG terminal facility
Morotai and Tobelo 3	A8	2.037213	128.2987	2.35	110.92	Proposed LNG terminal facility
Namlea	A9	−3.23225	127.1099	1.2	56.64	Proposed LNG terminal facility
Namrole	A10	−3.84501	126.7478	0.81	38.23	Proposed LNG terminal facility
Raja Ampat	A11	−0.53295	130.5775	0.71	33.51	Proposed LNG terminal facility
Sanana	A12	−2.07	125.9798	0.87	41.06	Proposed LNG terminal facility
Tahuna	A13	3.615112	125.4952	2.42	114.22	Proposed LNG terminal facility
Ternate	A14	0.768278	127.3065	5.65	266.68	Proposed LNG terminal facility
Arun PAG	B1	5.2156	97.08788	16.2	764.64	Existing terminal facility
Belitung	B2	−2.8905	107.5653	2.1	99.12	Proposed LNG terminal facility
Bintan	B3	1.064211	104.2365	0.7	33.04	Proposed LNG terminal facility
Kalteng	B4	−2.8532	111.6955	4.86	229.392	Proposed LNG terminal facility
Nias	B5	1.210076	97.67568	6.6	311.52	Proposed LNG terminal facility
Pontianak	B6	0.063473	109.2001	36.28	1712.416	Proposed LNG terminal facility
Alor	C1	−8.2479	124.5489	1.35	63.72	Proposed LNG terminal facility
Baubau	C2	−5.39652	122.6252	5.27	248.744	Proposed LNG terminal facility
Bima	C3	−8.40963	118.6968	7.35	346.92	Proposed LNG terminal facility
Flores	C4	−8.30431	120.4959	0.77	36.344	Proposed LNG terminal facility
FSRU Karunia Dewata	C5	−8.74677	115.2107	29.6	1397.12	Existing terminal facility
Jeranjang Lombok 4	C6	−8.65981	116.0716	20.86	984.592	Proposed LNG terminal facility
Kalsel	C7	−3.37403	114.4828	10.08	475.776	Proposed LNG terminal facility
Kupang peaker	C8	−10.3526	123.459	2.92	137.824	Proposed LNG terminal facility
Makassar	C9	−5.4033	119.37	9.6	453.12	Proposed LNG terminal facility
Maumere	C10	−8.62013	122.3366	1.54	72.688	Proposed LNG terminal facility
Sambelia	C11	−8.42181	116.7109	2.6	122.72	Proposed LNG terminal facility
Selayar	C12	−6.04827	120.4582	1.66	78.352	Proposed LNG terminal facility
Sulbagsel	C13	−2.7564	122.053	42.46	2004.112	Proposed LNG terminal facility
Sulselbar	C14	−4.06503	121.6202	5.03	237.416	Proposed LNG terminal facility
Sultra	C15	−3.89476	122.5393	2.1	99.12	Proposed LNG terminal facility
Sumbawa	C16	−8.44733	117.333	19.75	932.2	Proposed LNG terminal facility
Tanjung selor	C17	2.813909	117.3644	0.66	31.152	Proposed LNG terminal facility
Waingapu	C18	−9.47733	120.1521	3.48	164.256	Proposed LNG terminal facility
Biak	D1	−1.14437	135.9474	3.45	162.84	Proposed LNG terminal facility
Dobo	D2	−5.81997	134.2446	1.01	47.672	Proposed LNG terminal facility
Fak-fak	D3	−2.91949	132.2193	1.46	68.912	Proposed LNG terminal facility
Jayapura	D4	−2.61653	140.7867	7.09	334.648	Proposed LNG terminal facility
Kaimana	D5	−3.66659	133.7623	0.6	28.32	Proposed LNG terminal facility
Langgur	D6	−5.55358	132.7717	1.59	75.048	Proposed LNG terminal facility
Manokwari	D7	−0.93527	134.012	7.39	348.808	Proposed LNG terminal facility
Merauke	D8	−8.47729	140.3374	4.41	208.152	Proposed LNG terminal facility
Nabire	D9	−3.37871	135.455	2.3	108.56	Proposed LNG terminal facility
Saumlaki	D10	−7.93842	131.2967	0.83	39.176	Proposed LNG terminal facility
Serui	D11	−1.88209	136.3736	1.14	53.808	Proposed LNG terminal facility
Timika	D12	−4.75358	136.7653	7.16	337.952	Proposed LNG terminal facility
Bontang NGL	S1	0.100036	117.4962	0	0	Existing LNG Plant
Donggi Senoro	S2	−1.24916	122.5889	0	0	Existing LNG Plant
Masela	S3	−8.16121	129.8702	0	0	Proposed LNG Plant
Tangguh	S4	−2.44209	133.1205	0	0	Existing LNG Plant

¹ FSRU Ambon and Seram serves two demand points, requiring an additional 82 km of pipeline. ² The FSRU located in Halmahera, Tidore, and Sofifi is assumed to serve multiple demand points, necessitating an additional 15 km + 15 km of pipeline (total 30 km). ³ FSRU Morotai and Tobelo serve two demand points, requiring an additional 68 km of pipeline. ⁴ The FSRU located in Jeranjang Lombok is assumed to serve multiple demand points, requiring an additional 15 km of pipeline.

present the refined LNG demand potential in Indonesia, categorized into four clusters using Scikit-learn’s K-Means algorithm. The four LNG supply points serve all clusters. The next step involves calculating the distances between supply and demand points and within each cluster, storing these values in a distance matrix. This matrix represents the shipping routes for LNG distribution. The distance matrix is generated using Netpas Distance 4.1, a maritime distance calculator covering 12,000 ports, and extensive route data [55].

Table A3. Distance matrix (in km) for Cluster A.

	Location	A1	A2	A3	A4	A5	A6	A7	A8	A9	A10	A11	A12	A13	A14	S1	S2	S3	S4
A1	Ambon and Seram		442	414	886	674	717	747	824	159	206	660	340	939	594	1648	796	612	605
A2	Bacan	442		468	649	440	484	432	528	326	488	455	242	597	252	1322	580	989	719
A3	Bula	414	468		1036	553	561	834	668	429	560	430	527	999	653	1724	938	636	303
A4	FSRU Gorontalo	886	649	1036		1022	1002	472	805	770	758	1038	551	695	599	1394	366	1376	1295
A5	Halmahera, Sofifi, and Tidore	674	440	553	1022		492	805	452	558	690	424	373	877	625	1694	954	1222	904
A6	Halmahera Timur	717	484	561	1002	492		647	222	601	733	370	642	646	475	1464	997	1180	812
A7	Minahasa	747	432	834	472	805	647		450	630	711	821	503	321	295	1020	494	1294	1085
A8	Morotai and Tobelo	824	528	668	805	452	222	450		708	897	457	649	449	278	1267	821	1287	919
A9	Namlea	159	326	429	770	558	601	630	708		174	544	224	823	478	1532	653	706	699
A10	Namrole	206	488	560	758	690	733	711	897	174		675	284	926	620	1562	594	654	774
A11	Raja Ampat	660	455	430	1038	424	370	821	457	544	675		599	882	640	1700	696	1049	681
A12	Sanana	340	242	527	551	373	642	503	649	224	284	599		708	373	1407	443	887	796
A13	Tahuna	939	597	999	695	877	646	321	449	823	926	882	708		389	969	717	1487	1250
A14	Ternate	594	252	653	599	625	475	295	278	478	620	640	373	389		1155	589	1141	904
S1	Bontang NGL	1648	1322	1724	1394	1694	1464	1020	1267	1532	1562	1700	1407	969	1155		1416	1935	1975
S2	Donggi Senoro	796	580	938	366	954	997	494	821	653	594	696	443	717	589	1416		1200	1209
S3	Masela	612	989	636	1376	1222	1180	1294	1287	706	654	1049	887	1487	1141	1935	1200		816
S4	Tangguh	605	719	303	1295	904	812	1085	919	699	774	681	796	1250	904	1975	1209	816

,

Table A4. Distance matrix (in km) for Cluster B.

	Location	Arun PAG	Belitung	Bintan	Kalteng	Nias	Pontianak	Bontang NGL	Donggi Senoro	Masela	Tangguh
B1	Arun PAG		1588	969	2105	848	1537	3060	3844	4141	4538
B2	Belitung	1588		669	723	2203	411	1678	2463	2760	3157
B3	Bintan	969	669		1185	1766	617	2140	2924	3221	3618
B4	Kalteng	2105	723	1185		2065	781	1130	1922	2219	2616
B5	Nias	848	2203	1766	2065		2261	2995	3749	4022	4444
B6	Pontianak	1537	411	617	781	2261		1736	2520	2817	3214
S1	Bontang NGL	3060	1678	2140	1130	2995	1736		1416	1935	1975
S2	Donggi Senoro	3844	2463	2924	1922	3749	2520	1416		1200	1209
S3	Masela	4141	2760	3221	2219	4022	2817	1935	1200		816
S4	Tangguh	4538	3157	3618	2616	4444	3214	1975	1209	816

and

Table A5. Distance matrix (in km) for Cluster C.

	Location	C1	C2	C3	C4	C5	C6	C7	C8	C9	C10	C11	C12	C13	C14	C15	C16	C17	C18	S1	S2	S3	S4
C1	Alor		429	682	501	1065	1017	1319	281	791	329	911	613	777	632	652	843	1778	511	1360	888	702	1225
C2	Baubau	429		557	4545	909	861	1032	811	447	371	759	309	557	280	432	693	1474	1061	1055	668	971	1363
C3	Bima	682	557		252	428	380	758	1093	690	437	274	337	1012	594	887	206	1392	730	972	1123	1310	1766
C4	Flores	501	4545	252		635	586	913	912	608	256	481	263	863	505	738	413	1421	937	1002	974	1129	1585
C5	FSRU Karunia Dewata	1065	909	428	635		100	676	972	1007	820	237	665	1364	911	1239	277	1546	579	1065	1476	1677	2147
C6	Jeranjang Lombok	1017	861	380	586	100		627	950	958	772	195	617	1316	863	1191	229	1497	556	1017	1427	1645	2098
C7	Kalsel	1319	1032	758	913	676	627		1506	1114	1095	656	799	1504	1018	1379	670	755	1112	822	1615	1912	2309
C8	Kupang peaker	281	811	1093	912	972	950	1506		1180	740	882	1010	1148	1021	1066	963	2175	457	1757	1235	812	1448
C9	Makassar	791	447	690	608	1007	958	1114	1180		675	864	391	804	420	782	801	1556	1320	1138	1000	1340	1733
C10	Maumere	329	371	437	256	820	772	1095	740	675		666	402	767	535	642	598	1571	990	1152	878	958	1437
C11	Sambelia	911	759	274	481	237	195	656	882	864	666		517	1213	768	1088	114	1454	489	996	1325	1539	1992
C12	Selayar	613	309	337	263	665	617	799	1010	391	402	517		780	295	655	454	1242	973	824	892	1188	1586
C13	Sulbagsel	777	557	1012	863	1364	1316	1504	1148	804	767	1213	780		663	114	1148	1556	1398	1527	325	1117	1231
C14	Sulselbar	632	280	594	505	911	863	1018	1021	420	535	768	295	663		641	705	1461	1224	1042	860	1181	1574
C15	Sultra	652	432	887	738	1239	1191	1379	1066	782	642	1088	655	114	641		1023	1620	1316	1402	296	1038	1225
C16	Sumbawa	843	693	206	413	277	229	670	963	801	598	114	454	1148	705	1023		1436	570	982	1260	1471	1924
C17	Tanjung selor	1778	1474	1392	1421	1546	1497	755	2175	1556	1571	1454	1242	1556	1461	1620	1436		1910	588	1403	2183	1962
C18	Waingapu	511	1061	730	937	579	556	1112	457	1320	990	489	973	1398	1224	1316	570	1910		1452	1485	1123	1698
S1	Bontang NGL	1360	1055	972	1002	1065	1017	822	1757	1138	1152	996	824	1527	1042	1402	982	588	1452		1416	1935	1975
S2	Donggi Senoro	888	668	1123	974	1476	1427	1615	1235	1000	878	1325	892	325	860	296	1260	1403	1485	1416		1200	1209
S3	Masela	702	971	1310	1129	1677	1645	1912	812	1340	958	1539	1188	1117	1181	1038	1471	2183	1123	1935	1200		816
S4	Tangguh	1225	1363	1766	1585	2147	2098	2309	1448	1733	1437	1992	1586	1231	1574	1225	1924	1962	1698	1975	1209	816

present the distance matrix for clusters A to D.

3.2. Cost Optimization and Economic Feasibility

The lowest-cost configuration is identified through a sensitivity analysis of LNG inventory levels and their impact on total supply chain costs. In the model, the volume of LNG transported is determined by the required operating stock (in days), while the FSRU storage capacity is based on the combined operating and safety stock, as defined in Equations (14) and (15). Once the inventory level that minimizes the total unit cost—the sum of transportation cost and handling and storage cost (USD/MMBTU)—is identified, the corresponding LNG distribution routes and economic feasibility indicators (NPV, IRR, and payback period) are evaluated.

The optimal inventory level is influenced by factors such as transport distance, ship turnaround time, and fuel consumption, as reflected in Equation (1). Lower inventory levels result in smaller, more frequent shipments, leading to increased fuel usage and vessel rental costs. In contrast, higher inventory levels support larger batch deliveries and fewer trips, which reduce transportation costs but require greater storage capacity and, thus, higher capital expenditure. The trade-off between operational and capital costs determines the most cost-effective inventory strategy.

3.2.1. Cluster A

Figure 4 illustrates a decreasing trend in LNG distribution costs as the inventory level (operating stock) increases in Cluster A. However, the model remains feasible only within the range of 4 to 39 days. Beyond this range, the model becomes infeasible due to capacity constraints, such as the limited storage capacity of the available LNG ship carriers (as listed in Table A1) and the delivery time window, where the required transportation time exceeds the available inventory stock time. These technical limitations prevent the model from generating valid solutions at extremely low or high inventory levels.

As shown in Figure 4, the lowest total unit cost in Cluster A is achieved at an operating stock level of 20 days, resulting in a cost of 1.15 USD/MMBTU. Therefore, the most cost-effective LNG distribution configuration in Cluster A occurs when the operating stock is set at 20 days. The corresponding LNG distribution routes between supply and demand points are detailed in

Table 2. The optimal LNG distribution route in Cluster A (at 20-day inventory).

LNG Carrier Ship	Shipping Route	Demand (m3 LNG/Day)	Total Demand (m3 LNG)	Shipping Time (Days)
Shinju Maru	Donggi Senoro (S2) ⟶ Tahuna (A13) ⟶ Donggi Senoro (S2)	114.22	2284.48	3.15
WSD59 3K	Donggi Senoro (S2) ⟶ Namrole (A10) -> Namlea (A9) ⟶ Sanana (A12) ⟶ Donggi Senoro (S2)	135.94	2718.72	4.02
WSD59 5K	Tangguh (S4) ⟶ Raja Ampat (A11) ⟶ Halmahera, Sofifi, and Tidore (A5) ⟶ Bula (A3) -> Tangguh (S4)	243.55	4871.04	4.48
WSD59 6.5K	Donggi Senoro (S2) ⟶ Bacan (A2) ⟶ Ternate (A14) ⟶ Donggi Senoro (S2)	301.14	6022.72	3.46
Coral Methane	Donggi Senoro (S2) ⟶ FSRU Gorontalo (A4) ⟶ Donggi Senoro (S2)	241.19	4823.84	1.84
Norgas	Tangguh (S4) ⟶ Ambon and Seram (A1) ⟶ Tangguh (S4)	486.16	9723.20	2.61
WSDS55 12K	Donggi Senoro (S2) ⟶ Minahasa (A7) ⟶ Donggi Senoro (S2)	505.98	10,119.68	2.25
WSD50 20K	Tangguh (S4) ⟶ Halmahera Timur (A6) ⟶ Morotai and Tobelo (A8) ⟶ Tangguh (S4)	981.29	19,625.76	3.93
Total		3009.47	60,189.44	-

and Figure 5.

Figure 6 illustrates that the NPV value at the 25th year (lifetime operation) increases as the LNG sales margin rises. The NPV values are varied based on different discount rates. A higher discount rate results in a lower NPV. According to Table 1, the assumed discount rate is 10%. At this rate, the LNG sales margin must exceed 2.6 USD/MMBTU for the NPV to become positive. Figure 7 shows that the payback period decreases as the LNG sales margin increases. A higher LNG sales margin also enhances the investment return throughout the project’s operational period. At an LNG price margin of 2.6 USD/MMBTU, the payback period occurs after 10 years of operation.

3.2.2. Cluster B

Figure 8 shows a declining trend in LNG distribution costs as inventory levels (operating stock) increase. However, the model produces feasible solutions only within the range of 10 to 17 days. Outside this range, the model becomes infeasible due to technical constraints, particularly the limited capacity of available LNG ship carriers and the delivery time requirement. When the required delivery time exceeds the allowable operating stock time, the model cannot satisfy both delivery and demand constraints, resulting in no valid solution.

At an inventory level of 17 days, the total unit cost of LNG reaches its lowest value at 1.745 USD/MMBTU. This indicates that the most optimal and cost-effective LNG distribution setup for this case is achieved at 17 days of operating stock. The corresponding LNG distribution routes from supply to demand points are shown in

Table 3. The optimal LNG distribution route in Cluster B (at 17-day inventory).

LNG Carrier Ship	Shipping Route	Demand (m3 LNG/Day)	Total Demand (m3 LNG)	Shipping Time (Days)
WSD59 6.5K	Bontang NGL(S1) ⟶ Belitung (B2) ⟶ Kalteng (B4) ⟶ Bontang NGL (S1)	328.51	5584.70	7.11
WSD50 20K	Bontang NGL (S1) ⟶ Bintan (B3) ⟶ Arun PAG (B1) ⟶ Nias (B5) ⟶ Bontang NGL (S1)	1109.20	18,856.40	11.76
WSD50 30K	Bontang NGL (S1) ⟶ Pontianak (B6) ⟶ Bontang NGL (S1)	1712.416	29,111.07	5.55
Total		3150.13	53,552.18

and Figure 9.

Figure 10 illustrates that the NPV value at the 25th year (lifetime operation) increases as the LNG sales margin rises. The NPV values are analyzed under various discount rates, where a higher discount rate results in a lower NPV. According to Table 1, the assumed discount rate is 10%. At this rate, the LNG sales margin must exceed 2.80 USD/MMBTU for the NPV to become positive. Figure 11 demonstrates that the payback period decreases as the LNG sales margin increases. A higher LNG sales margin also enhances the investment return over the project’s operational period.

3.2.3. Cluster C

Figure 12 illustrates that the total unit cost of LNG (USD/MMBTU) reaches its lowest value when the operating stock/inventory is set at 11 days, amounting to 1.05 USD/MMBTU. Consequently, the most cost-effective and optimal LNG distribution model or delivery route occurs at an operating stock level of 11 days.

Table 4. The optimal LNG distribution route in Cluster C (at 11-day inventory).

LNG Carrier Ship	Shipping Route	Demand (m3 LNG/Day)	Total Demand (m3 LNG)	Shipping Time (Days)
Shinju Maru	Bontang NGL (S1) ⟶ Tanjung selor (C17) ⟶ Selayar (C12) ⟶ Maumere (C10) ⟶ Flores (C4) ⟶ Bontang NGL (S1)	218.54	2403.90	7.71
WSD59 5K	Masela (S3) ⟶ Alor (C1) ⟶ Waingapu (C18) ⟶ Kupang peaker (C8) ⟶ Masela (S3)	302.08	3322.88	5.32
WSDS55 12K	Donggi Senoro (S2) ⟶ Baubau (C2) ⟶ Makassar (C9) ⟶ Sulselbar (C14) ⟶ Sultra (C15) ⟶ Donggi Senoro (S2)	1038.40	11,422.40	4.64
WSD50 20K	Bontang NGL (S1) ⟶ Bima (C3) ⟶ Sumbawa (C16) ⟶ Kalsel (C7) ⟶ Bontang NGL (S1)	1754.90	19,303.86	5.34
Surya Satsuma	Donggi Senoro (S2) ⟶ Sulbagsel (C13) ⟶ Donggi Senoro (S2)	2004.11	22,045.23	1.32
WSD50 30K	Bontang NGL (S1) ⟶ Sambelia (C11) ⟶ Jeranjang Lombok (C6) ⟶ FSRU Karunia Dewata (C5) ⟶ Bontang NGL (S1)	2504.43	27,548.75	4.48
Total		7822.46	86,047.02

and Figure 13 present the LNG distribution routes, highlighting the connections between supply and demand points.

Figure 14 demonstrates that the NPV at the 25th year (lifetime operation) increases as the LNG sales margin rises. The NPV values are evaluated under different discount rates, where a higher discount rate results in a lower NPV. As stated in Table 1, the assumed discount rate is 10%. Under this assumption, the LNG sales margin must exceed 1.85 USD/MMBTU for the NPV to become positive. Figure 15 indicates that the payback period decreases as the LNG sales margin increases, implying that higher sales margins lead to a faster return on investment.

3.2.4. Cluster D

Figure 16 shows that the lowest total unit cost of LNG (USD/MMBTU) is achieved when the operating stock/inventory is maintained at 19 days, with a cost of 1.53 USD/MMBTU. This indicates that the most efficient and cost-effective LNG distribution model or delivery route occurs at an operating stock level of 19 days.

Table 5. The optimal LNG distribution route in Cluster D (at 19-day inventory).

LNG Carrier Ship	Shipping Route	Demand (m3 LNG/Day)	Total Demand (m3 LNG)	Shipping Time (Days)
Shinju Maru	Tangguh (S4) ⟶ Fak-fak (D3) ⟶ Tangguh (S4)	68.91	1309.33	1.40
WSD59 3K	Tangguh (S4) ⟶ Langgur (D6) ⟶ Dobo (D2) ⟶ Kaimana (D5) ⟶ Tangguh (S4)	151.04	2869.76	4.22
WSD59 5K	Masela (S3) ⟶ Merauke (D8) ⟶ Saumlaki (D10) ⟶ Masela (S3)	247.33	4699.23	4.81
WSD59 6.5K	Tangguh (S4) ⟶ Timika (D12) ⟶ Tangguh (S4)	337.95	6421.09	3.42
WSD50 20K	Tangguh (S4) ⟶ Nabire (D9) ⟶ Jayapura (D4) ⟶ Serui (D11) ⟶ Biak (D1) ⟶ Manokwari (D7) ⟶ Tangguh (S4)	1008.66	19,164.62	7.38
Total		1813.90	34,464.02

and Figure 17 illustrate the LNG distribution routes within Cluster D, highlighting the linkages between supply and demand points.

Figure 18 demonstrates that the Net Present Value (NPV) at year 25 (operational lifetime) grows as LNG sales margins increase. The NPV is analyzed across various discount rates, where a higher discount rate leads to a lower NPV. As indicated in Table 1, the assumed discount rate is 10%. At this rate, the LNG sales margin must surpass 2.75 USD/MMBTU for the NPV to remain positive. Figure 19 highlights that the payback period shortens as LNG sales margins rise. An increase in sales margins also improves the return on investment throughout the project’s operational lifespan.

3.3. Investment Value and Fuel Costs

Table 6. LNG supply chain investment value per cluster.

Parameter	Cluster A	Cluster B	Cluster C	Cluster D
Daily demand (m3 LNG)	2678.13	3150.13	7886.18	1813.90
Number of supply point (LNG Plant)	2	1	3	2
Number of LNG terminal	14	6	18	12
Optimum transported LNG stock (days)	20	17	11	19
Number of LNG ship carrier	8	3	6	5
Total distance covered (km)	11,267	13,955	14,070	9512
Trans. Cost (USD/MMBTU)	0.77	1.43	0.83	1.16
FRSU cost (USD/MMBTU)	0.41	0.30	0.23	0.37
Total unit cost (USD/MMBTU)	1.18	1.75	1.06	1.53
Total inventory LNG (m3 LNG)	53,563	53,552	86,748	34,464
Total FSRU storage capacity (m3 LNG)	61,597	63,003	110,406	39,906
Total ICC (million USD)	165.49	147.76	272.75	99.30

and Figure 20 present the total LNG demand and investment values across all clusters. As shown, investment cost is strongly influenced by LNG demand, where higher demand leads to higher investment. Several existing FSRUs in the supply chain are assumed to have zero investment value. Among the four clusters, Cluster C records the highest investment at USD 272.75 million, aligned with its status as the highest-demand cluster.

Despite having the highest daily demand (7886.18 m³ LNG) and the longest distribution distance (14,070 km), Cluster C achieves the lowest total unit cost at 1.06 USD/MMBTU. This is due to its well-developed infrastructure, which includes three supply points and 18 LNG terminals, allowing for efficient routing and optimal vessel utilization. Its large FSRU storage capacity (110,406 m³) demonstrates that high demand paired with strategic infrastructure planning enables economies of scale that reduce per-unit costs.

In contrast, Cluster B shows the highest unit cost at 1.75 USD/MMBTU despite a relatively moderate demand level (3150.13 m³ LNG/day) and a lower investment value of USD 147.76 million. With only one supply point, six terminals, and three LNG carriers, it lacks logistical flexibility. The long transport distance (13,955 km) and limited routing options lead to higher fuel use and operational inefficiencies, emphasizing that limited infrastructure coverage in geographically dispersed regions can drive up distribution costs.

Cluster A represents a balanced configuration. It has moderate demand (2678.13 m³ LNG/day), two supply points, and 14 terminals, along with eight LNG carriers. With a total transport distance of 11,267 km and an investment of USD 165.49 million, it achieves a unit cost of 1.18 USD/MMBTU.

Meanwhile, Cluster D has the lowest daily demand (1813.90 m³ LNG) and the shortest transport distance (9512 km) but still incurs a high unit cost of 1.53 USD/MMBTU. Although it has two supply points and 12 terminals, the limited demand results in underutilized infrastructure and reduced economies of scale. Its total investment is the lowest at USD 99.30 million, but high per-unit costs suggest inefficiencies caused by small-scale operations.

Table 7. Detailed comparison of natural gas fuel costs per cluster with HSD at the plant gate.

Cost Components	Cluster A	Cluster B	Cluster C	Cluster D	HSD
LNG price (USD/MMBTU)	8.50	8.50	8.50	8.50
Trans. Cost (USD/MMBTU)	0.77	1.43	0.83	1.16
FSRU cost (USD/MMBTU)	0.41	0.30	0.22	0.37
Capital cost and profit (USD/MMBTU)	1.42	1.05	0.79	1.22
HSD price (USD/MMBTU)					25.48
Fuel price	11.10	11.28	10.35	11.25	25.48
Difference to HSD (%)	56.44%	55.71%	59.38%	55.85%

and Figure 21 present a comparison of natural gas and HSD prices at the plant gate. The LNG price is set at 8.5 USD/MMBTU, representing the upper price range [56]. The HSD price is based on Rp 13,492.12 per liter [57], with an assumed exchange rate of 15,000 Rupiah per USD. The SS LNG supply chain simulation in Indonesia indicates that natural gas fuel prices at the plant gate are lower than HSD. The natural gas price at the plant gate ranges from 10.35 to 11.28 USD/MMBTU, while HSD is priced at 25.48 USD/MMBTU. This means that natural gas is 55–60% cheaper than HSD. Therefore, the use of natural gas has great potential to reduce fuel costs for power plants that still rely on HSD.

4. Conclusions

This study presents a techno-economic analysis of the small-scale LNG (SS LNG) supply chain for power generation in Indonesia, based on a projected LNG demand of 15,528 m³/day distributed across 50 terminal locations utilizing Floating Storage Regasification Units (FSRUs). The optimized model estimates a total investment requirement of USD 685.3 million across four regional clusters (A to D). At a 10% discount rate, the resulting natural gas price at the plant gate ranges from 10.35 to 11.28 USD/MMBTU, offering a 55–60% reduction compared to High-Speed Diesel (HSD) fuel prices.

The results demonstrate that SS LNG is not only economically viable but also strategically positioned to support Indonesia’s energy transition, particularly in remote and island regions with limited access to pipeline infrastructure. The model optimizes inventory levels, vessel routing, and infrastructure sizing to identify cost-effective configurations. These findings can support key stakeholders, including government energy planners, state-owned enterprises, and private investors, in making informed decisions related to LNG terminal development, carrier fleet planning, and regional fuel-switching strategies for power generation.

Although the economic potential of SS LNG is clear, this study does not incorporate a formal risk analysis. Several real-world uncertainties—such as fuel price volatility, capital cost variability, regulatory changes, delivery disruptions due to extreme weather, and supply–demand mismatches—may significantly affect project feasibility. Future research should integrate these uncertainty factors using scenario-based analysis, Monte Carlo simulations, or sensitivity modeling to evaluate the robustness of SS LNG systems under dynamic conditions.

In conclusion, SS LNG presents a compelling alternative to diesel-based power generation by offering both cost and emissions advantages. While this study focuses on system optimization under fixed assumptions, further development of the model to include environmental and financial risk factors will enhance its utility as a decision-support tool for policymakers and project developers in shaping Indonesia’s cleaner and more resilient energy future.