Bulk electrical power extraction through tunnel transmission technology from renewable energy concentrated zones

Abstract

Mountainous regions, such as the Alps, Appalachians, Andes, and Qingzang Plateau, offer vast renewable energy potential, but geographical and ecological challenges limit conventional overhead transmission lines. Underground alternatives, including gas-insulated lines and cables, provide viable solutions but require comprehensive techno-economic evaluation for sustainable deployment. This study integrates multi-physics analysis, combining electromagnetic, thermal, and fluid dynamics, with lifecycle cost assessment per unit transmission task to compare overhead and underground tunnel transmission systems. Optimized conductor designs and ventilation strategies enhance cable capacity and improve economic performance, particularly in regions with high external costs. Here, we show that underground tunnel based power transmission systems, especially gas-insulated lines, deliver cost-effective, high-capacity transmission solutions for mountainous renewable energy concentrated regions, outperform overhead power transmission infrastructure when external costs, including ecological, landscape, and infrastructure constraints, are significant, thus advancing the goals of sustainable renewable energy integration and net-zero.

The study evaluates underground tunnel transmission for renewable energy in mountains, using multi-physics analysis and lifecycle cost evaluation. It finds gas-insulated lines are cost-effective in areas with high environmental impact costs, optimizing conductor layout and ventilation for increased capacity.

Introduction

Facing the global energy crisis and increasing urgency of transitioning towards a low-carbon future^1,2, effective utilization of renewable energy has become crucial. By 2027, the global renewable energy generation capacity is projected to increase by approximately 2400 GW³. The capacity and type of renewable energy construction expected to be completed by 2027 in countries/regions around the world is provided in Supplementary Fig. 1. A critical challenge in realizing this growth lies in transmitting power from remote, resource-rich regions to demand centers, ensuring feasibility, stability, and alignment with environmental policies and public expectations⁴.

Untapped renewable energy resources are particularly concentrated in mountainous and highland regions^5,6, where elevated terrain enhances solar irradiance⁷, wind speeds⁸, and hydropower potential. Other areas, such as offshore wind sites, also represent globally significant resources, but the present study focuses on mountainous and plateau regions where corridor scarcity and ecological sensitivity are especially pronounced. However, these areas frequently overlap with ecologically sensitive or scenic regions^9,10. For example, the Appalachian Mountains in eastern North America hold an estimated 1859 GW of wind power generation capacity potential¹¹, while the Andes in South America possess about 243 GW of combined wind and geothermal generation capacity potential¹². The European Alps offer 19.9 GW of wind generation capacity potential, 45 GW of solar generation capacity potential, and 8.56 GW of hydropower generation capacity potential⁸. In China, the Qingzang Plateau is targeted for the development of 227 GW of hydropower generation capacity and 96 GW of photovoltaic generation capacity^13–15. Indonesia’s Central Range (New Guinea Highlands) contains approximately 26 GW of renewable generation capacity potential, most of which is hydropower, leveraging the area’s rugged terrain¹⁶. These figures, which are shown in Table 1, are based on regional studies that report only partial, technically proven reserves rather than the full exploitable potential of each mountain range. Although the methodologies differ—ranging from geographic information system-based spatial analysis to hydrological modeling—they consistently highlight the vast untapped renewable potential of these regions^17,18.

Table 1.

Proven renewable energy reserves in mountainous regions

Region	Location	Energy type	Proven capacity (GW)	References
Appalachian Mountains	Eastern North America	Wind	1859	11
		Solar (PV)	5	11
Andes	South America	Wind	227	12,17
		Geothermal	16	18
Europe Alps	Central Europe	Wind	19.9	8
		Solar (PV)	45.0	8
		Hydropower	8.56	8
Qingzang Plateau	China	Hydropower	227	13,15,20
		Solar (PV)	96	13,19
Central Range (New Guinea Highlands)	Indonesia / Papua	Hydropower*	~26	16

^*Most of the ~26 GW capacity consists of hydropower leveraging the area’s rugged terrain. PV, photovoltaic. Values represent technically proven reserves and do not reflect full exploitable potential.

However, transmitting renewable energy from these areas faces significant challenges: Geographical and Corridor Constraints: Mountainous regions require long-distance exports to lowland load centers, yet feasible overhead corridors are sharply limited by rugged terrain and regulatory siting restrictions. On the Qingzang Plateau, for instance, four planned transmission pathways (northern, southern, central, and central-branch) intersect with multiple nature reserves, leaving only ~16 feasible HV circuits compared to ~37 required for the planned renewable buildout (≈40% realizable via OHL)^19,20. Comparable potential corridor scarcity exists in the Alps, where Natura 2000 and UNESCO sites restrict routing options, and in the Appalachians, where national forests and the Appalachian Trail limit siting flexibility. Ecological and Tourism Protection: Beyond technical siting limits, these mountain regions also carry exceptional ecological and scenic value²¹. Overhead lines fragment habitats and impair landscapes, triggering strong public opposition. In the Alps, willingness-to-pay (WTP) studies indicate more than $900,000 per kilometer per year to underground lines, reflecting the premium placed on preserving biodiversity and visual integrity ^22,23.

For decades, traditional overhead transmission lines (OHL) have dominated power delivery, yet they inadequately address these challenges in renewable-rich mountainous and plateau scenarios. This has spurred exploration of alternatives, with underground tunnel transmission emerging as a viable solution for long-distance, high-capacity power transport. Unlike OHL, tunnels bypass surface constraints and ecological impacts, enabling delivery through challenging terrains while preserving landscapes^24,25. The undergrounding alternatives include direct-buried and tunnel installations; while direct burial is generally cheaper, in mountainous, corridor-scarce, and environmentally sensitive segments, continuous trenching is often infeasible, and operators therefore typically adopt accessible cable tunnels for bulk transmission²⁵.

However, applying underground transmission tunnel technologies—such as cross-linked polyethylene (XLPE) AC or DC cables or higher-voltage gas-insulated lines (GIL)—to renewable-rich regions presents multiple unresolved challenges²⁶. First, economic costs are a primary concern. Although studies have quantified the additional ecological and landscape costs of overhead transmission lines (OHLs) and the economic benefits of undergrounding, clear criteria and guidelines for determining which regions should prioritize underground solutions versus OHLs, based on regional characteristics, remain absent. Second, the design of current-carrying capacity for lines within tunnels has been evaluated by finite element analysis, accounting for electromagnetic and thermal coupling^27–30. However, few studies have investigated how the spatial arrangement of lines within tunnels impacts their current-carrying capacity³¹. Third, ventilation is critical for maintaining environmental stability in long, confined tunnel spaces. Previous research has proposed ventilation strategies for cable tunnels using finite element analysis³², and a recent study optimized ventilation by constraining conductor temperatures and factoring in energy loss costs^33,34. However, this work was limited to a single-tunnel scenario and focused solely on reducing ventilation energy losses, without exploring the coupled relationship between ventilation and current-carrying capacity in long-distance tunnels. Furthermore, current research has not sufficiently elucidated the interplay between multi-physics fields—encompassing electromagnetic, thermal, and fluid dynamics—and economic benefits in critical tunnel design aspects, such as line layout and ventilation. This gap hinders the development of comprehensive design guidelines and economic evaluation frameworks for long-distance tunnel transmission in mountainous regions.

In this work, we present a comprehensive framework for evaluating underground tunnel transmission in renewable-rich mountainous and plateau scenarios, delivering three key contributions: (1) A multi-physics coupling methodology integrating electromagnetic-thermal-fluid finite element analysis with annualized lifecycle cost (LCC) per unit transmission task (MW-km) evaluation, enabling rigorous comparison of GIL, UGC AC, and UGC DC under realistic constraints. (2) A techno-economic model that accounts for ecological and landscape costs, demonstrating that underground solutions, particularly GIL, achieve lower annualized LCC per unit transmission task (MW-km) than overhead lines when external costs exceed $550,000/km/year, redefining economic feasibility for renewable energy transmission in mountainous regions worldwide. (3) Quantitative insights into spatial and ventilation optimizations, demonstrating up to 57% capacity enhancements (e.g., AC cables from 1273 A to 2003 A) and 20% LCC savings in annualized LCC per unit transmission task (MW-km) via parameterized simulations, providing actionable design guidelines for sustainable energy delivery. This framework equips policymakers and engineers with a decision-support matrix, accelerating deployment of underground tunnels to unlock mountainous renewables, foster net-zero transitions, and balance economic, environmental, and social imperatives.

Results

Comparative techno-economic evaluation of transmission technologies for renewable energy concentrated zones

To evaluate the economic viability of various high-capacity power transmission methods, we analyzed alternating current (AC) and direct current (DC) overhead lines (OHL), underground tunnel AC cables (UGC AC), and gas-insulated lines (GIL).

We selected 500 kV AC and ±600 kV DC overhead lines as benchmarks, as these are widely adopted and technologically mature transmission systems. Cost data for these lines were sourced from the 2024 MISO handbook, which provides U.S. power infrastructure construction costs, with the tower type specified as an angled dead-end structure, suitable for mountainous and high-altitude regions rich in renewable energy³⁵. While China has implemented mature 1000 kV AC and ±1000 kV DC overhead lines with estimated costs of approximately $3,802,580/km and $1,801,444/km, respectively, these data stem from specific projects and the technologies are less commonly applied globally³⁶. Thus, they are excluded from the main analysis but provided in Supplementary Table 1 for reference.

For underground tunnel AC transmission, construction costs were derived from the 2025 IET report on underground cable systems³⁷. Because published cost data for long tunnel GIL systems remain scarce, we explicitly note that the Su-Tong GIL corridor is a short river-crossing tunnel (Yangtze crossing) and use it as a conservative reference, while Supplementary Tables 1–2 compile additional project data (including an Chinese GIL tunnel) to broaden the evidence base. Operation and maintenance costs were estimated based on industry averages and multiple research reports. The Su-Tong GIL project has a rated current of 6300 A. However, based on a State Grid Corporation of China report and our subsequent studies³⁸, dual-circuit lines operate at a reduced current of 3150 A under normal conditions to maintain stable tunnel temperatures. Therefore, the capacity calculations are based on 3150 A. Energy loss data for AC and DC overhead lines were obtained from EIA report³⁹. For UGC AC and GIL, where empirical loss data are scarce, we conducted coupled electromagnetic-thermal-fluidic finite element analysis, estimating energy losses of 11.91%/1000 km for UGC AC and 6.71%/1000 km for GIL (Calculation details are shown in the “Methods” Section). These numbers align closely with industry expectations and also with the data we collected from State Grid Corporation of China, which is 15.2%/1000 km and 5.96%/1000 km for UGC AC and GIL, respectively. The data are given in Supplementary Table 2.

Transmission lines in renewable-rich mountainous regions often traverse ecologically sensitive or scenic areas, where OHLs significantly impact ecosystems and esthetic value⁴⁰. To quantify these costs, we adopted the willingness-to-pay (WTP) metric, using the Alps as a case. Two studies reported residents’ WTP for undergrounding overhead lines in the Alps at $984,000/year/km and $892,043/year/km, respectively. We used their average, $938,022/year/km, for cost estimation^22,41.

Although we also investigated underground DC transmission, limited case studies and data availability constrained the cost analysis. However, we posit that its costs are comparable to underground AC cables, as both primarily incur additional expenses from tunnel construction and equipment installation compared to overhead lines.

Table 2 summarizes the economic cost data for these transmission methods. Although the data are drawn from diverse countries, projects, and studies, it still serves a reliable reference for comparative analysis.

Table 2.

Comparative analysis of construction costs between overhead transmission lines and underground tunnel transmission

Construction Information	OHL AC	OHL DC	UGC AC	GIL
Rated voltage	500 kV	±600 kV	420 kV	1000 kV
Rated power	2598MVA	2400 MW	7482MVA	10,000MVA
Lifecycle	30years	30years	XLPE cable: 30years Tunnel: 50years	GIL: 50years Tunnel: 50years
Capital cost	407,450$/km	294,643$/km	72,164,405$/km (Cable:42.5% Tunnel:57.5%)	100,812,125$/km
Station type	AC substation	DC converter station	AC substation	GIL incoming station
Station cost (at both ends)	333,949,072$	750,800,000$	436,226,676$	138,098,896$
Operation cost(line)	5% capital cost/year	1% capital cost/year	0.5% capital cost/year	0.5% capital cost/year
Operation cost(station)	2% construction cost/year	1% construction cost/year	2% construction cost/year	1% construction cost/year
Energy loss	6.70%/1000 km	3.50%/1000 km	11.91%/1000 km	6.71%/1000 km
Ecological and landscape cost	938,022 $/(year·km) (WTP)	938,022 $/(year·km) (WTP)	14,000 $/(year·km)	17,000 $/(year·km)

OHL overhead line, UGC underground cable, GIL gas-insulated line, AC alternating current, DC direct current, XLPE cross-linked polyethylene, WTP willingness to pay (applied to ecological and landscape cost valuation).

To determine the most suitable transmission method for renewable energy-rich regions, we evaluated lifecycle unit capacity costs, analyzing total and per-unit-length costs as a function of transmission distance, which is shown in Fig. 1.

Fig. 1. Lifecycle cost (LCC) analysis of power transmission methods.

a Annualized LCC per MW capacity on a logarithmic scale, comparing overhead line (OHL) AC, OHL DC, underground cable (UGC) AC, and gas-insulated line (GIL). b Annualized LCC per unit transmission task (MW·km) on a linear scale, highlighting the impact of ecological costs; Ceco denotes the annualized ecological and landscape cost ($/yr/km). AC alternating current, DC direct current. Source data are provided as a Source Data file.

Ecological and landscape costs constitute a significant portion of overhead line expenses, particularly in sensitive regions like the Alps. These costs vary by region, so we plotted cost curves with and without ecological considerations. When including ecological costs, overhead lines exhibit the highest lifecycle unit costs: approximately $430/year/MW/km (DC) and $400/year/MW/km (AC) at a 500 km distance. Underground AC cables follow at $321/year/MW/km, and GIL at $263/year/MW/km.

Excluding ecological costs, overhead lines are significantly cheaper, with costs of $33/year/MW/km (DC) and $28/year/MW/km (AC). Thus, without ecological costs, GIL costs are roughly 8 times higher than overhead lines, and costs for underground AC cables are approximately ten times higher. When ecological costs are excluded, overhead AC lines are more cost-effective for distances below 800 km, while DC lines become preferable beyond 800 km. This aligns with trends in prior studies⁴².

Acknowledging that infrastructure costs—particularly tunnel excavation—are highly project-specific and dependent on local geological conditions, we conducted a comprehensive sensitivity analysis to validate the robustness of our economic model. We evaluated the impact of cost variations across three key dimensions: tunnel civil engineering costs (ranging from −50% to +100% to cover geological complexities), electrical equipment costs, and operation and maintenance expenses⁴³. The detailed parameter settings and the resulting sensitivity analysis (presented as a tornado diagram) are provided in Supplementary Table 3 and Supplementary Fig. 2. The results demonstrate that the economic advantage of GIL is resilient to cost uncertainties; even under scenarios of significantly higher tunnel construction costs (e.g., in complex terrain), the GIL system remains a cost-competitive alternative to overhead lines when high ecological externalities are factored in.

The choice between overhead and underground solutions hinges on ecological, landscape, and other external costs, such as additional expenses for constructing new lines in fully occupied corridors, forest clearing, ecological fees, and steep terrain construction. Further analysis reveals that when these external costs reach $550,000/km/year, overhead lines’ unit costs approach those of GIL; at $720,000/km/year, they align with underground AC cables. Therefore, we conclude that underground GIL systems become more cost-effective when external costs exceed $550,000/km/year in target regions, offering a viable alternative for sustainable, high-capacity power transmission.

Electromagnetic safety analysis of transmission technologies in underground tunnels

Electromagnetic safety is a critical consideration for underground tunnels housing high-voltage equipment. The confined geometry and proximity of conductors, maintenance personnel, and surrounding infrastructure necessitate rigorous evaluation of electric and magnetic field distributions to ensure operational safety and compliance with regulations⁴⁴.

Following IEEE standards, occupational exposure limits for 50 Hz AC magnetic fields are 2710 μT and electric fields are 5 kV/m; for DC, magnetic fields are limited to 353,000 μT and electric fields are limited to 5 kV/m (based on stricter head-exposure criteria)^45,46. ICNIRP standards limit electric fields to 10 kV/m and magnetic fields to 1000 μT for AC 50Hz⁴⁷. For this analysis, we adopt the stricter standards: AC (50 Hz): 5 kV/m, 1000 μT; DC: 5 kV/m, 353,000 μT.

For electric fields, both XLPE cables (AC and DC) and gas-insulated lines (GIL) incorporate grounded shielding layers or enclosures, resulting in external electric fields near zero during normal operation. However, it is crucial to distinguish normal operation from fault conditions. While the aforementioned shielding is effective under steady-state scenarios, insulation breakdowns—though relatively infrequent with typical failure rates of approximately 0.1–0.5 failures per 100 km·year for modern high-voltage systems—pose a different set of risks. In the event of an insulation fault, short-circuit currents (often reaching tens of kA) will flow through the cable sheath. Without proper mitigation, this massive current induces significant longitudinal voltages along the sheath, creating life-threatening touch potentials at the cable surface. To mitigate this risk, we recommend the installation of Sheath Voltage Limiters (SVLs) within the system—typically at cross-bonding points—to clamp potential overvoltages to safe levels. Additionally, implementing an equipotential grounding grid throughout the tunnel is advised. These measures ensure that the tunnel environment remains safe for maintenance personnel access.

A more common scenario involves partially grounded shielding due to distant grounding points. Simulations indicate that with a shielding layer voltage of approximately 1 V, the electric field at 0.1 m from the conductor is about 3.76 V/m, well below safety thresholds. These simulation results are detailed in Supplementary Fig. 3. For these simulations, excitation voltages were 500 kV (AC XLPE), 500 kV (DC XLPE), and 1000 kV (GIL), with currents of 2000 A (AC), 2000 A (DC), and 6000 A (GIL). Thus, electric field intensities and touch potentials in tunnels are controllable and safe through the combination of conductor shielding and fault protection measures.

Magnetic field analysis is complex due to the influence of multi-phase conductor arrangements and shielding interconnection methods. We conducted magnetic field simulations in tunnel environments, evaluating AC and DC cables in circular and square tunnels with stacked and wall-mounted configurations (see Supplementary Fig. 4 for schematics and dimensions). For GIL, we analyzed the tunnel configuration of the Su-Tong GIL corridor ³⁸ (Supplementary Fig. 5).

Regarding shielding grounding methods, DC cable shielding does not produce induced voltage due to the absence of time-varying currents, allowing solid bonding⁴⁸. GIL enclosures, designed as continuous conductors to carry the return current, significantly reduce external magnetic fields and transmission losses, typically employing solid bonding or multiple-point bonding rather than cross-bonding. In contrast, long-distance AC XLPE cables, due to high shielding losses, generally use cross-bonding. In our tunnel model simulations, DC cable shielding and GIL enclosures use solid bonding, while AC cable shielding employs cross-bonding.

Magnetic field intensity distributions in tunnel environments are shown in Fig. 2.

Fig. 2. Spatial magnetic field distribution in tunnel environments for AC, DC Cables, and GIL.

a Alternating current (AC) cable magnetic field in a circular tunnel, with the outermost contour at 1000 μT, matching the AC safety limit. b Direct current (DC) cable magnetic field in a square tunnel, with the outermost contour at 500 μT, well below the DC safety limit of 353,000 μT. c Gas-insulated line (GIL) tunnel magnetic field, with the outermost contour at 50 μT, below the AC safety limit of 1000 μT. Source data are provided as a Source Data file.

In tunnels, GIL exhibits the lowest external magnetic field intensity, posing no health risks unless directly contacting the conductor surface. DC cables produce higher magnetic fields, approximately 500 μT within 0.6 m, but these remain well below the DC safety limit of 353,000 μT. AC cables show significantly higher magnetic fields, exceeding 1000 μT within a 0.3 m radius for both stacked and wall-mounted configurations. Simulation results suggest that wall-mounted configurations allow maintenance personnel to avoid high-intensity magnetic fields in the tunnel’s central walkway. For stacked configurations, placing conductors entirely along tunnel walls is preferable to minimize exposure.

Spatial configuration effects on power loss and thermal performance in underground tunnels

The operation of transmission systems in tunnel environments differs significantly from overhead lines due to the confined, elongated space with limited heat dissipation⁴⁹. A primary challenge is balancing increased current-carrying capacity—enhancing economic value—with the pronounced thermal effects it induces. Excessive heat, causing conductor temperatures to exceed 90 °C for XLPE cables (or 105 °C for GIL)³⁸, reduces conductor lifespan. Variations in conductor spatial arrangements influence power loss and temperature through magnetic field coupling and altered heat dissipation structures. To investigate these effects, we conducted finite element analysis on the impact of conductor spatial arrangements on temperature and power loss for three tunnel configurations.

We defined spatial parameters based on installation methods: for stacked cable configurations, $L$ represents the distance between the centers of two circuits; for GIL, $L$ denotes the horizontal distance between circuits. In square wall-mounted and GIL scenarios, $H$ indicates the vertical distance between adjacent conductors within the same circuit. In circular wall-mounted scenarios, $\theta$ represents the angle between the line connecting the topmost conductor to the tunnel center and the horizontal axis. These parameters are illustrated in Supplementary Fig. 4. The resulting conductor temperature variations and tunnel thermal losses are shown in Fig. 3.

Fig. 3. Impact of conductor spatial arrangement on temperature and thermal loss in tunnels.

a Relationship between total power loss (histogram) and conductor temperature (curve) for alternating current (AC) and direct current (DC) cables across four tunnel scenarios, with AC current at 1600 A (root mean square, RMS) and DC current at 2000 A. b Maximum allowable current-carrying capacity for AC and DC cables under a 90 °C constraint, plotted against normalized spatial parameters (0–1, corresponding $L$ to from 0.4  to 1.4 m in circular stacked, $L$ from 0.4  to 1.5 m in square stacked, $\theta$ from 20° to 90° in circular wall-mounted, and $H$ from 0.4  to 1.8 m in square wall-mounted). c Gas-insulated line (GIL) power loss and maximum allowable current under a 105 °C constraint, plotted against $L$ and $H$. Source data are provided as a Source Data file.

For AC cables in stacked configurations with fixed intra-phase distances, increasing the inter-circuit distance ($L$) has minimal impact on power loss but reduces conductor temperature. In circular stacked scenarios, total losses increase slightly from 345 W/m ($L$ = 0.4 m) to 347 W/m ($L$ = 1.4 m), while temperature decreases from 131.02 to 126.42 °C. In square stacked scenarios, losses rise from 338 W/m ($L$ = 0.4 m) to 344 W/m ($L$ = 1.5 m), with temperature dropping from 125.41 to 121.86 °C. This indicates that increasing the distance between heat sources slightly increases losses (primarily shielding layer losses, with conductor losses unchanged) but enhances heat dissipation, lowering conductor temperature and slightly increasing current-carrying capacity from 1273 A to 1304 A (circular) and 1310 A to 1335 A (square). In wall-mounted configurations, increasing intra-circuit conductor distance significantly reduces shielding layer losses, from 64 W/m to 2 W/m (circular) and 88 W/m to 3 W/m (square), boosting current-carrying capacity by over 200 A in both scenarios. Further analysis attributes this to the cross-bonding shielding method used, where larger intra-phase distances allow balanced three-phase induced currents to neutralize shielding layer currents, reducing losses. For AC cables, I²R losses dominate across all configurations. In stacked configurations, shielding layer losses are significant (64–88 W/m, ~18–25% of total losses of 338–347 W/m), despite mitigation by cross-bonding. In wall-mounted configurations, shielding layer losses drop to 2–3 W/m, making I²R losses even more dominant. AC cables exhibit higher I²R losses than DC cables due to two factors: (1) shielding layer losses increase conductor temperature, raising resistance, and (2) skin effect increases effective conductor resistance. Simulations using a uniform multi-turn conductor model (single turn) estimate I²R losses without skin effect at 168.37 W/m, compared to 235.63 W/m with skin effect, indicating a skin effect contribution of ~67.26 W/m (~19% of stacked losses, ~32% of minimum wall-mounted losses). Note that skin effect is a component of I²R losses, manifesting as increased resistance.

For DC cables, the static magnetic field eliminates shielding layer induced currents and associated thermal losses, as well as AC-specific skin effects, resulting in lower losses and temperatures compared to AC cables, unaffected by spatial arrangement. However, increasing heat source distance improves dissipation, slightly raising current-carrying capacity from 2605 A, 2658 A, 2655 A, and 2668 A to 2663 A, 2684 A, 2714 A, and 2710 A across the four scenarios, with increases ranging from 26 A to 59 A. For DC cables, I²R losses are the sole loss mechanism, as shielding layer losses and skin effect are absent.

For GIL, prior magnetic environment analysis shows low external magnetic fields and minimal inter-phase coupling due to larger line distances, resulting in negligible power loss changes with spatial arrangement. Capacity increases, driven by improved heat dissipation from increased $L$ and $H$, are modest, from 4411 A to 4466 A. For GIL, I²R losses dominate (976–991 W/m, ~70–72% of total losses), with enclosure losses (386–394 W/m, ~28%) also significant, aligning with the data from a Sutong GIL related reference⁵⁰. Skin effect contributes ~11% of total losses, as estimated using the same multi-turn conductor model.

Overall, increasing inter-conductor distance and adopting wall-mounted configurations significantly enhance AC cable current-carrying capacity, reducing power loss and allowing increases from 1273 A to 2003 A under temperature constraints. For DC cables, the absence of induced currents limits the impact of spatial changes, and for GIL, low external fields and large distances minimize loss variations, with capacity gains primarily from improved heat dissipation. These findings provide a clear guidance for tunnel based power transmission design for renewable energy dispatch from energy concentrated zones.

Simulations assumed cross-bonded grounding and balanced three-phase currents. With unbalanced currents, AC cable shielding losses increase. In circular stacked and wall-mounted scenarios, with phase angles of 0 °, ±90 °, and 0 °, ±60 °, shielding losses in stacked configurations rise from 109 W/m to 144 W/m (0 °, ±90 °) and 248 W/m (0 °, ±60 °), and in wall-mounted configurations from 22 W/m to 37 W/m (0 °, ±90 °) and 149 W/m (0 °, ±60 °), which is shown in Supplementary Fig. 6. This significantly reduces current-carrying capacity, particularly in stacked scenarios.

Cost-optimized ventilation design for enhanced transmission

Beyond reducing power losses, active ventilation is a critical method for lowering conductor temperatures and maintaining tunnel thermal stability. However, in long-distance tunnels, both ventilation shaft construction and air velocity control directly impact the economic costs of transmission^51,52. This study employs finite element analysis to optimize ventilation shaft spacing $L_{\mathrm{venti}}$ and tunnel air velocity $v_{\mathrm{wind}}$ under a minimum-cost objective for AC, DC, and GIL tunnels.

To address the computational complexity of three-dimensional (3D) simulations, we developed a two-dimensional (2D) model using a negative heat source to approximate the complex electro-magnetic-thermal-fluid coupling in 3D simulations. This approach maintains consistency in key results, such as air heat flux (represented as negative heat source power in 2D), while reducing simulation time from 5,283 seconds to 47 seconds per scenario (Fig. 4a). Specifically, during the validation process for converting the 3D model to a 2D one, a 3D model with a length of 10 meters along the tunnel direction was used; and the air temperature at the inlet was set to a constant 20 °C during the simulation. Notably, the simplified 2D model is capable of simulating scenarios with any ventilation distance. Details are provided in the “Methods” section.

Fig. 4. Impact of tunnel ventilation on current-carrying capacity and cost per unit transmission task.

a Comparison of temperature distribution and key simulation parameters between 3D and simplified 2D models. b–d Maximum current-carrying capacity and annualized lifecycle cost (LCC) per unit transmission task (MW-km) as functions of ventilation shaft spacing $L_{\mathrm{venti}}$ and air velocity $v_{\mathrm{wind}}$ for: b alternating current (AC) circular wall-mounted tunnel (θ = 30°); c direct current (DC) square stacked tunnel (L = 0.5 m); d gas-insulated line (GIL) tunnel. “Opt.” denotes the optimal operating point that minimises LCC under the corresponding ventilation conditions. Source data are provided as a Source Data file.

Building on the techno-economic analysis and reliable industry cost references, we established key cost factors, including baseline tunnel construction, ventilation shaft construction, and ventilation operation costs, to support the cost-optimization analysis (see Methods for cost assumptions). The results, based on these assumptions and parametric finite element analysis, are presented in Fig. 4.

Previous analyses constrained conductor core temperatures to 90 °C for cables and 105 °C for GIL to estimate losses and temperature variations. However, to account for practical ventilation requirements, this section further limits tunnel air temperature to 40 °C and GIL enclosure temperature to 70 °C⁵⁰. Consequently, unventilated scenarios yield lower current-carrying capacities than previously estimated. For round wall-mounted AC tunnels, without ventilation, the capacity is limited to 1575 A to maintain air temperature below 40 °C. With shaft spacing of $L_{\mathrm{venti}}$ = 6 km and air velocity of $v_{\mathrm{wind}}$ = 0.5 m/s, the capacity increases to 2002 A; at $L_{\mathrm{venti}}$ = 1 km and $v_{\mathrm{wind}}$ = 3 m/s, it reaches 2081 A. While higher air velocities could further increase capacity, our analysis, based on cost assumptions, shows that ventilation energy costs scale with the cube of air velocity, significantly increasing overall costs. Nonetheless, tunnels can be designed with high-velocity ventilation potential to accommodate peak transmission demands. At the optimal point, the annualized LCC per unit transmission task is $405/year/MW/km, higher than the $321/year/MW/km for UGC AC in the economic analysis, due to a lower simulated capacity (3,464 MVA) compared to the reference case. Cost assumptions are detailed in Methods, with optimal cost surfaces shown in Supplementary Fig. 7.

For DC tunnels, assuming costs comparable to AC tunnels, the optimal configuration occurs at $L_{\mathrm{venti}}$ = 7 km and $v_{\mathrm{wind}}$ = 0.75 m/s, yielding a capacity of 2754 A and a cost of $251/year/MW/km, significantly lower than AC tunnels. This is attributed to the absence of shielding layer losses and skin effect, resulting in lower total losses and higher capacity under the same temperature constraints, enhancing economic efficiency. Thus, DC cable tunnels are recommended for superior cost-effectiveness under equivalent conditions.

For GIL tunnels, the optimal configuration is at $L_{\mathrm{venti}}$ = 3 km and $v_{\mathrm{wind}}$ = 1 m/s, with a capacity of 5902 A and a cost of $211/year/MW/km, notably lower than the $263/year/MW/km from the economic analysis. Unlike AC and DC simulations, GIL parameters align with the Su-Tong GIL project. However, the Su-Tong tunnel, crossing a riverbed, cannot accommodate mid-tunnel ventilation shafts, restricting shaft spacing to 5.5 km (full length). This limits operating current below the rated 6300 A due to thermal constraints. Our simulations demonstrate that optimizing shaft spacing and air velocity can reduce effective costs by approximately 20% compared to the Su-Tong project.

Discussion

This study evaluates the feasibility of long-distance underground tunnel transmission for large-scale renewable energy delivery in mountainous regions rich in renewable resources. Using annualized lifecycle cost (LCC) per unit transmission task (MW·km) as the primary metric, we compared underground gas-insulated lines (GIL), AC cables (UGC AC), and DC cables (UGC DC) against traditional overhead lines (OHL). Our findings reveal that underground solutions outperform OHL when ecological and landscape costs—often significant in scenic, renewable-rich mountainous areas like the Alps ($938,022/year/km based on willingness-to-pay)^22,41—exceed $550,000/km/year for GIL or $720,000/km/year for UGC AC. In regions like Southwest China, where saturated transmission corridors necessitate costly OHL construction (e.g., forest clearing, steep terrain, ecological compensation), these thresholds are often surpassed, rendering GIL and UGC economically competitive. These thresholds are segment-applicable, enabling mixed corridors that underground only the constrained spans while retaining OHL elsewhere. Multi-physics coupling analyses (electromagnetic, thermal, and fluid dynamics) optimize conductor layouts and ventilation, boosting AC cable capacity by 57% and reducing GIL LCC by ~20% compared to existing projects. These results underscore that OHL’s hidden costs, such as landscape degradation and additional construction expenses, are frequently underestimated, positioning underground transmission as a viable or superior alternative in mountainous scenarios.

This study presents a finite element-based framework that integrates multi-physics coupling (electromagnetic, thermal, and fluid dynamics) with economic cost analysis to optimize conductor spatial arrangements and ventilation settings. This approach surpasses empirical estimates by parametrically adjusting conductor spacing to minimize shielding layer losses and enhance heat dissipation, maximizing capacity within temperature constraints (90 °C for cables, 105 °C for GIL). Ventilation optimization, using a 2D negative heat source model to ensure air (40 °C) and conductor temperature compliance, efficiently identifies cost-optimal shaft spacing and air velocity combinations. While this framework successfully elucidates the primary physical interactions—such as the inverse relationship between conductor spacing and inductive heating—it employs simplified fluid dynamics assumptions (e.g., uniform airflow) that may not fully capture localized turbulence or complex 3D thermal gradients in irregular tunnel geometries. Future iterations could integrate more granular computational fluid dynamics (CFD) to address these higher-order physical phenomena. This scalable framework provides practical design guidance for diverse tunnel environments and transmission technologies, informing global sustainable energy projects from the Alps to the Andes.

We also delineate the strengths and limitations of the three underground transmission methods. GIL excels with low magnetic fields (<50 μT) and high capacity (>5900 A), achieving the lowest LCC ($211/year/MW·km), ideal for long-distance, high-capacity transmission, though initial investment is high. UGC DC follows, with low losses (no shielding layer losses or skin effect) and strong economic performance ($251/year/MW·km), with safe magnetic fields. UGC AC has lower capacity (≤2000 A) and higher losses (due to shielding layer induced currents), but its technological maturity suits medium-to-short distances.

We recommend prioritizing GIL or DC for long-distance, high-capacity, ecologically sensitive mountainous regions, with AC cables as a supplementary option for shorter distances, enhanced by optimized layouts (e.g., wall-mounted with increased spacing) and ventilation to mitigate drawbacks. This guidance informs engineering decisions, balancing economic, safety, and environmental considerations.

Although our baseline results are presented at the route level, the LCC per unit transmission task (MW·km·year) metric is modular and can be applied segment-wise to support hybrid corridors in mountainous settings. In planning studies, ecological/landscape external costs should be assigned only to the affected length (e.g., crossings of scenic or protected zones). Technology choice is then made per segment (OHL for unconstrained spans; tunnel-based GIL or UGC where constraints dominate), and the route-level objective is obtained by length-weighted aggregation. The parity thresholds reported here—approximately $550,000/km/year for GIL vs. OHL and $720,000/km/year for UGC AC vs. OHL—should therefore be read as local criteria for sensitive sections, indicating when selective undergrounding of a corridor segment is economically justified while maintaining OHL elsewhere.

While our results establish a strong economic case for underground transmission, we emphasize that these findings serve as a strategic decision-making baseline rather than a substitute for site-specific financial auditing. To account for regional heterogeneity, we incorporated a sensitivity analysis (detailed in Supplementary Fig. 2) confirming that GIL remains competitive even when tunnel civil costs double (+100%). However, we explicitly acknowledge that in hostile geological environments—such as zones with severe water inflow, unstable faulting, or ultra-hard rock—excavation costs could escalate precipitously, potentially exceeding even the upper bounds of our sensitivity analysis. Beyond economic parameters, certain idealized model conditions warrant consideration; for instance, perfectly balanced currents may not fully hold in practice, where three-phase imbalances could increase AC cable losses. Similarly, while geological impacts like seismic-induced fault displacement were not explicitly monetized in our cost model, successful projects in seismically active zones⁵³ (e.g., California’s metro tunnels) demonstrate that engineered solutions like isolation techniques and flexible joints can mitigate these risks⁵⁴. Thus, our framework is extensible, but future research should further quantify the cost implications of such extreme geological countermeasures.

Furthermore, outage statistics provide context for availability. One source reports average annual downtime of ~2.3 h for OHL, ~156 h for XLPE cables, and ~34.2 h for GIL (400-kV class)⁵⁵. While underground segments reduce exposure to surface hazards, longer time-to-repair can offset failure-rate advantages. Other system-level evidence (covering 163 U.S. cooperatives) associates a + 1% underground share with −0.211% SAIDI (and an additional −1.84% in forested territories), and with ~4.4% lower annual O&M when the underground share doubles²⁵. Given these divergent findings and our sensitivity checks, we treat availability as context rather than a monetized cost driver in the LCC. Instead, geographical corridor scarcity and ecological/landscape protection emerge as the dominant constraints shaping the feasibility and competitiveness of underground tunnel transmission. By highlighting these factors, our study provides a more balanced perspective on where underground solutions are most justifiable and impactful.

Moreover, the reported 57% increase in AC cable capacity (via optimized layouts) and ~20% reduction in GIL LCC per transmission task (via ventilation optimization) should be viewed as theoretical upper bounds derived from simulations under fixed tunnel geometry and uniform airflow. In practice, constraints such as tunnel cross-section geometry, maintenance access, and non-uniform ventilation may reduce the achievable benefits. Thus, our framework should be interpreted primarily as revealing performance trends and design trade-offs, with quantitative values serving as references. Future work will incorporate these engineering constraints to deliver more practical guidelines.

The environmental impact of GIL’s insulating gas, SF₆, is a growing regulatory concern. The EU’s F-gas regulation bans SF₆ in new medium-voltage equipment from 2026 and phases it out in high-voltage equipment by 2030–2035 to reduce greenhouse gas emissions⁵⁶. The U.S. EPA’s CARB regulations mandate SF₆ phase-out in gas-insulated equipment (GIE) from 2025, requiring emission reporting and restricted procurement of high-GWP gases⁵⁷. These regulations drive GIL technology toward sustainability. Emerging alternatives include GE’s g3 gas (CO₂, O₂, C₄F₇N mixture), used in 145 kV circuit breakers, gas-insulated substations, and 420 kV GIL, with ~99% lower GWP than SF₆ and comparable performance^58,59. Siemens’ Blue series, using clean air (80% nitrogen, 20% oxygen) with vacuum insulation, supports 72.5–145 kV equipment with zero harmful emissions⁶⁰. These advancements ensure GIL’s future viability, transitioning to sustainable solutions for urban and renewable energy transmission, aligning with global net-zero goals. However, cost and technical validation challenges must be addressed during this transition.

Looking forward, our framework has broad global applications. Future research should incorporate geomechanical analyses (e.g., finite element seismic simulations), higher voltages (±1000 kV), and climate change impacts on ventilation. These advancements will accelerate the net-zero transition, enhance renewable energy utilization, reduce carbon emissions, and improve grid resilience.

Methods

Lifecycle cost analysis

To assess the economic viability of power transmission systems, we calculated the construction costs of transmission lines and substations separately. For new lines, we assumed the need for substations or converter stations at both ends to match the line’s capacity.

For 500 kV AC and ±600 kV DC overhead lines, construction and substation/converter station costs, including operation and maintenance, were sourced from the 2024 MISO report³⁵. For AC substations, the report requires additional parameters such as land area and circuit breaker counts. We estimated these costs using typical parameters for substations of this voltage class (see Supplementary Table 4 and Supplementary Table 5 for billing items, unit costs, and assumptions).

Construction and substation costs for UGC AC and their corresponding capacities were derived from the 2025 IET report³⁷. For gas-insulated lines (GIL), we referenced the construction parameters of China’s Su-Tong 1000 kV GIL corridor, with a total cost of 4.763 billion CNY. As specific cost breakdowns were unavailable, we estimated substation/terminal station costs at 16–26% of the total, based on similar projects. This portion covers terminal station construction, auxiliary systems, and project management. Terminal stations ensure seamless integration between GIL and ultra-high-voltage overhead lines. We used a mid-range estimate of 1.013 billion CNY for these costs.

Lifecycle costs were evaluated over the full operational lifespan of each system. AC and DC overhead lines were assigned a conservative lifecycle of 30 years, reflecting the minimum expected duration in challenging mountainous regions⁶¹. Underground AC cables have a 30-year lifecycle, while the tunnels themselves have a minimum lifespan of 50 years. Thus, cable and tunnel costs were calculated separately, with cables and tunnels accounting for 42.5% and 57.5% of total costs, respectively, as per the IET report³⁷. For GIL, both the lines and tunnels have a 50-year lifecycle. Although lifetimes differ by component (OHL: 30 years; UGC cable: 30 years; tunnels and GIL: 50 years), all costs are annualized consistently over their respective service lives within the LCC framework.

Operation and maintenance costs for AC overhead lines were estimated at 5% of construction costs annually⁶². For DC overhead lines, this was set at 1% per year⁴³. Underground tunnel maintenance costs are lower relative to construction costs, at approximately 0.5% annually⁶². Another study suggests AC overhead and UGC maintenance costs are 1.5% and 0.15% of construction costs, respectively⁶³, confirming UGC costs rate are roughly one-tenth of AC overhead lines. Due to limited data on GIL maintenance, we assumed the same 0.5% of construction costs as UGC AC.

Energy loss data for AC and DC overhead lines were sourced from EIA report³⁹. For UGC AC and GIL, where data are scarce, we relied on finite element simulations coupling electromagnetic, thermal, and fluidic effects. For UGC AC (500 kV, dual-circuit, six lines), when $v_{\mathrm{wind}}$ is 0.5 m/s and $L_{\mathrm{venti}}$ is 6 km, the annualized LCC per unit transmission task is optimal with the total loss was approximately 371.3 W/m at a current of 2000 A. Assuming a power factor of 0.9, the total power is 3117 MW, yielding an energy loss of 11.91%/1000 km. For GIL, when $v_{\mathrm{wind}}$ is 1 m/s and $L_{\mathrm{venti}}$ is 3 km, the loss was 556.8 W/m at a current of 5902 A, with a dual-circuit power of 18,400 MW (power factor 0.9), resulting in an energy loss of 6.71%/1000 km.

Ecological and esthetic costs are critical for transmission projects in renewable-rich mountainous regions, encompassing ecosystem restoration, scenic value degradation, biodiversity loss, and land reclamation. Quantifying these costs lacks a universal standard, so we employed the willingness-to-pay (WTP) and hedonic pricing methods to estimate societal costs of avoiding visual and ecological impacts from overhead lines⁶⁴. A discrete choice experiment in Italy’s Alps found that the social marginal benefit of undergrounding per kilometer far exceeds construction costs, with residents’ WTP at $984,000/year/km²². Another study reported WTP values ranging from €229,796/km/year to €1,555,223/km/year across different regions, averaging €771,464/km/year ($892,043/year/km)⁴¹. Given the consistency between these studies, we used their average, $938,022/year/km, for estimation.

Underground tunnels may impact groundwater systems or permafrost due to cable heat, generating ecological costs. However, these effects are complex, poorly quantified, and less studied. They are likely far lower than overhead lines’ ecological and esthetic costs. For groundwater, open-pit mine dewatering systems cost $10,000–$20,000 per well annually ⁶⁵. For permafrost, mitigation costs along the Alaska Highway are approximately $11,000/km/year⁶⁶. Adding a conservative $3,000/km/year for drainage system inspection and maintenance, we estimated total ecological costs for underground cable tunnels at $14,000/km/year and GIL tunnels at $17,000/km/year. These costs are modest and likely subsumed within the high construction and maintenance expenses of tunnels, but we include them as a conservative estimate for a holistic evaluation.

To evaluate the economic feasibility of different transmission technologies, we calculated the annualized lifecycle cost (LCC) per unit effective power capacity and per unit effective transmission task, which are shown in Fig. 1. The annualized LCC per effective MW $C_{\mathrm{MW}}$ is given by:

$$C_{\mathrm{MW}}=\frac{A_{\mathrm{L}}+A_{\mathrm{S}}+O_{\mathrm{L}}+O_{\mathrm{S}}+E_{\mathrm{C}}}{P_{\mathrm{eff}}}$$ 1

The annualized LCC per effective MW per km $C_{\mathrm{MW}-\mathrm{km}}$ is derived by:

$$C_{\mathrm{MW}-\mathrm{km}}=\frac{C_{\mathrm{MW}}}{d}$$ 2

$A_L$ represents the annual amortized line capital cost ($/year), the yearly amortized cost of the transmission line capital investment. For UGC AC:

$$A_{\mathrm{L}}=\frac{C_{\mathrm{L},\mathrm{c}}\cdot d}{L_{\mathrm{c}}}+\frac{C_{\mathrm{L},\mathrm{t}}\cdot d}{L_{\mathrm{t}}}$$ 3

For other technologies:

$$A_{\mathrm{L}}=\frac{C_{\mathrm{L}}\cdot d}{L}$$ 4

where $C_{\mathrm{L}}$ is capital cost per kilometer of the transmission line, $C_{\mathrm{L},\mathrm{c}}$ and $C_{\mathrm{L},\mathrm{t}}$ represent capital cost per kilometer for the cable component and tunnel component of UGC AC, respectively. $L$ is the lifecycle (years), $L_{\mathrm{c}}$ and $L_{\mathrm{t}}$ are the lifespans of UGC AC cable and tunnel, which are 30 and 50 years, respectively. $d$ is the transmission distance (km).

$A_{\mathrm{S}}$ represents the yearly amortized cost of the substation or converter station capital investment, calculated as:

$$A_{\mathrm{S}}=\frac{C_{\mathrm{S}}}{L}$$ 5

where $C_{\mathrm{S}}$ is the total capital cost for constructing both end substations or converter stations.

$O_{\mathrm{L}}$ and $O_{\mathrm{S}}$ are the yearly cost for maintaining the transmission line, given by:

$$\left\{\right)\begin{array}{c}\hfill O_{\mathrm{L}}=O_{\mathrm{L},\mathrm{p}}\cdot C_{\mathrm{L}}\cdot d\hfill \\ \hfill O_{\mathrm{S}}=O_{\mathrm{S},\mathrm{p}}\cdot C_{\mathrm{S}}\hfill \end{array}$$ 6

where $O_{\mathrm{L},\mathrm{p}}$ is the line operation and maintenance cost proportion of the capital cost per year, and $O_{\mathrm{S},\mathrm{p}}$ is the station operation and maintenance cost proportion of the construction cost per year.

$E_{\mathrm{C}}$ is the yearly cost associated with environmental and landscape impacts ($/year), computed as:

$$E_{\mathrm{C}}=E\cdot d$$ 7

where $E$ is the annual ecological and landscape cost per kilometer ($/year/km).

$P_{\mathrm{eff}}$ is the actual power delivered after accounting for energy losses, defined as:

$$P_{\mathrm{eff}}=P\left(1-\frac{L_{\mathrm{loss}}\cdot d}{1000}\right)$$ 8

where $P$ is the rated power (MW), and $L_{\mathrm{loss}}$ is the Energy loss rate (%/1000 km).

Tunnel ventilation cost configurations and assumptions

To evaluate the impact of tunnel ventilation on transmission capacity and optimize economic costs per unit transmission task (MW·km), we established cost assumptions for ventilation shaft construction, ventilation equipment, and energy costs associated with air velocity. Below, we outline the key cost assumptions and their sources.

For AC and DC underground cable tunnels, the annualized lifecycle baseline cost was derived from the IET report for a 420 kV, 7482 MVA project, with a total cost of $72,164,405/km, split into 42.5% for electrical cables and 57.5% for tunnel construction, with lifecycles of 30 and 50 years, respectively³⁷. This yields an annualized cost of $1,852,220/km/year. However, our model considers a 500 kV AC double-circuit scenario with a rated current of ~2000 A and power of 3464 MVA, significantly lower than the reference case. Thus, we referenced IET costs for 75 km projects at 2494 MW ($1,226,900/km/year) and 4,988 MW ($1,693,500/km/year), interpolating linearly to estimate a baseline cost of $1,408,407/km/year for 3464 MVA. For DC tunnels, as the tunnel dimensions match those of AC tunnels in our simulations, we assumed identical baseline construction costs, as the primary expense is the tunnel itself, with minimal cost differences between AC and DC cables. Additional cost parameters, drawn from the IET report, Indianapolis water tunnel, HS2 tunnel projects, and related case studies⁶⁷, include: ventilation shaft construction at $7,000,000/shaft and maintenance at $46,000/shaft/year; ventilation energy cost at $28,000/km/year for an air velocity of 1 m/s, scaling cubically with velocity. These costs vary significantly with geological conditions and tunnel depth, and our analysis adopts typical values for methodological clarity. Baseline tunnel costs include ventilation system construction, maintenance, and shaft costs at 5 km intervals. In simulations, we assumed air enters at 20 °C at each shaft, carries heat to the next shaft’s exit, and re-enters at 20 °C in the subsequent segment, allowing the axial tunnel length to be set as the shaft spacing.

For GIL tunnels, the baseline cost was calculated directly from the Su-Tong GIL project, with a 50-year lifecycle, yielding an annualized cost of $2,016,243/km/year. Due to the larger tunnel cross-sectional area, ventilation energy costs increase. Fan shaft power (P) scales as $P\propto {v_{\mathrm{wind}}}^3\frac{\pi }{4}D$, where $D$ is tunnel diameter, indicating power is proportional to diameter. Scaling by tunnel diameter relative to AC/DC tunnels, we estimated GIL ventilation energy cost at $108,889/km/year for 1 m/s. This aligns with Su-Tong GIL data, where a 160 m³/s airflow (equivalent to ~1.85 m/s) requires two 210 kW axial flow fans, consuming ~3.68 GWh/year³⁸. At a global average industrial electricity price of $0.161/kWh^68–70, this equates to $592,480/year, or ~$93,575/km/year at 1 m/s, accounting for airflow losses, closely matching our estimate of $108,889/km/year. For GIL ventilation shafts, based on HS2 project costs for rural tunnels with similar dimensions and shaft spacing, we assumed construction costs of $16,200,000/shaft and maintenance costs of $106,000/shaft/year⁶⁷. The tunnel ventilation cost configurations and assumptions are also shown in Supplementary Table 6.

Integrated techno-economic multi-physics analysis framework for underground transmission tunnels

This paper presents a comprehensive, finite element-based framework that integrates electromagnetic, thermal, and fluid dynamics analyses with economic cost optimization to design cost-effective underground transmission tunnels, which is shown in Fig. 5.

Fig. 5. Comprehensive electro-magneto-thermo-fluid multi-physics coupled with economic cost analysis framework for underground transmission tunnels.

The figure illustrates the four-step finite element-based framework for optimizing underground transmission systems in renewable energy applications. Step 1: line modeling and simplification of cable and gas-insulated line (GIL) geometry. Step 2: single-conductor electromagnetic simulation and tunnel electromagnetic environment simulation to confirm field safety. Step 3: parametric sweep of line spatial positions to identify configurations with lowest temperature and maximum current-carrying capacity for alternating current (AC) and direct current (DC) scenarios. Step 4: 3D-to-2D model reduction and ventilation optimisation to determine optimal shaft spacing and air velocity, yielding minimum annualized lifecycle cost (LCC) per unit transmission task (MW·km). AC alternating current, DC direct current, GIL gas-insulated line, LCC lifecycle cost, XLPE cross-linked polyethylene, HDPE high-density polyethylene.

This integrated approach models gas-insulated lines (GIL), XLPE AC, and DC cables under varying spatial arrangements and ventilation conditions, providing a unified methodology for sustainable energy infrastructure. The analysis process comprises four interconnected steps, detailed below:

Line and tunnel foundation modeling

Finite element models were constructed based on transmission line types, tunnel dimensions, and material parameters, with a 20 °C boundary condition applied 25 m from the tunnel center to simulate ambient conditions. For GIL, the model includes four layers: inner SF₆ gas, conductor, outer SF₆ gas, and shell, defined by their geometric and material properties (shown in Supplementary Fig. 8). For XLPE cables (model YJLW03-Z-290/500 kV-1, 2500 mm²), the complex internal structure (e.g., copper conductor, XLPE insulation, semiconductive buffer, aluminum sheath, HDPE jacket) was simplified by grouping non-critical layers into composite layers (shown in Supplementary Fig. 9 and Supplementary Table 7), reducing computational complexity while preserving key metallic structures for accurate loss and cost calculations. The stranded copper conductor was modeled as a single circular cross-section, with a correction factor $N_{\mathrm{con}}$ applied:

$$N_{\mathrm{con}}=\frac{A_{\mathrm{con}}}{S_{\mathrm{con}}}$$ 9

where $A_{\mathrm{con}}$ is the area of conductor cross-section which is 2500 mm² in this case, and $S_{\mathrm{con}}$ is the cross-sectional area of the conductor in the modeling, which can be calculated as:

$$S_{\mathrm{con}}=\pi {\left(\frac{D_{\mathrm{con}}}{2}\right)}^2\approx 2941{\mathrm{.66mm}}^2$$ 10

where $D_{\mathrm{con}}$ is the diameter of the conductor.

This factor $N_{\mathrm{con}}$ adjusts thermal and electrical conductivities to ensure precise heat and loss estimates, critical for economic assessments.

Electromagnetic field safety analysis

Electromagnetic field safety was assessed to ensure compliance with tunnel safety standards, minimizing operational risks. For electric field analysis, a voltage source was applied to the conductor, with the shielding layer grounded, typically yielding zero external electric field. High-impedance grounding faults were simulated using a circuit module connecting the shielding layer to a non-zero resistor before grounding. For magnetic field analysis, a current source was applied within the tunnel environment, evaluating spatial magnetic field distributions against safety standards. GIL exhibited low stray magnetic fields, DC cables had higher but safe fields, while AC cables exceeded limits over a larger range, favoring wall-mounted configurations to reduce central tunnel impacts and safety-related costs.

Line spatial position and heat loss

Spatial layout parameters (e.g., $L$, $H$, $\theta$ for stacked, wall-mounted, or GIL setups, as in Results) were defined to optimize capacity and reduce losses. Using a 2D model, parametric simulations varied current (100 A steps) and layout parameters, with current sources applied to conductors. Cross-bonding, common in long-distance transmission, significantly reduced AC cable shielding layer losses. This was modeled by replicating the magnetic field module for three segments (each one-third of the total length), connecting shielding layers via a circuit module for cross-bonding, and grounding through a zero or small resistor. Conductor temperature, conductor losses, and shielding layer losses were analyzed to inform cost calculations. Optimal layouts minimized temperature or total losses for a fixed current. Under temperature constraints (90 °C for XLPE cables, 105 °C for GIL), maximum current-carrying capacity was estimated via interpolation, guiding layout selection to maximize economic value while ensuring electromagnetic safety.

3D-to-2D model simplification for ventilation optimization

Finite element simulations of air, treated as a fluid, and tunnel ventilation typically require three-dimensional (3D) models to capture axial airflow dynamics. However, 3D simulations are computationally intensive, with long run times and complex meshing issues, particularly for long tunnels where transmission line multi-layer structures (e.g., conductor, insulation, sheath) are small relative to tunnel length. Simplifying these structures alters thermodynamic behavior. To address this, we developed a two-dimensional (2D) model that uses a negative heat source to approximate the heat removed by axial airflow, significantly reducing computational complexity while maintaining accuracy.

The negative heat source $q$ (W) is expressed as:

$$q=-\frac{{\rho }_{\mathrm{air}}\cdot \left(\frac{\pi \cdot D^2}{4}\right)\cdot v_{\mathrm{wind}}\cdot c_{\mathrm{p},\mathrm{air}}}{L_{\mathrm{venti}}}\cdot \left(T-T_{\mathrm{mref}}\right)\cdot L_{\mathrm{rep}}$$ 11

where ${\rho }_{\mathrm{air}}$ = 1.2 kg/ m³ is the air density, $D$ is the tunnel diameter, $c_{\mathrm{p},\mathrm{air}}$ = 1005 J/(kg·K) is the specific heat capacity of air, $T$ is the temperature of air, $T_{\mathrm{mref}}$ = 20 °C is the reference inlet air temperature, $L_{\mathrm{rep}}$= 1 m representative unit length for calculation. This negative heat source simulates the cooling effect of forced ventilation. The expression is derived as follows:

1) Air Mass Flow Rate Calculation: The mass flow rate $\stackrel{°}{m}$ of air through the tunnel cross-section is determined by the air density, cross-sectional area $A=\pi D^2/4$, and axial air velocity:

$$\stackrel{°}{m}={\rho }_{\mathrm{air}}\cdot A\cdot v_{\mathrm{wind}}={\rho }_{\mathrm{air}}\cdot \left(\pi D^2/4\right)\cdot v_{\mathrm{wind}}$$ 12

2) Heat Removal over a Ventilation Segment: Over a segment $L_{\mathrm{venti}}$, the air absorbs heat, causing a temperature rise $\Delta T=T_{\mathrm{out}}-T_{\mathrm{mref}}$. The total heat removal rate $Q$ (in W) for the segmment is given by the connective energy balance:

$$Q=\stackrel{°}{m}\cdot c_{\mathrm{p},\mathrm{air}}\cdot \Delta T$$ 13

Here, $\Delta T$ is approximated as $\left(T-T_{\mathrm{mref}}\right)$, where $T$ represents the local average temperature in the air domain, and $T_{\mathrm{mref}}$ is the inlet reference temperature, providing a linear approximation for heat transfer.

3) Heat Removal per Unit Length: To distribute the heat removal along the tunnel, the rate per unit length $q_{\mathrm{l}}$ (in W/m) is:

$$q_{\mathrm{l}}=\frac{Q}{L_{\mathrm{venti}}}=\frac{\stackrel{°}{m}\cdot c_{\mathrm{p},\mathrm{air}}\cdot \left(T-T_{\mathrm{mref}}\right)}{L_{\mathrm{venti}}}$$ 14

For the representative unit length $L_{\mathrm{rep}}$ (set to 1 m for simplicity in the 2D cross-section), the adjusted negative heat source $q=-q_{\mathrm{l}}\cdot L_{\mathrm{rep}}$.

This negative heat source, derived from air mass flow rate, heat removal over a ventilation segment, and distribution per unit length, reduced simulation time from 5,283 s to 47 s (Fig. 4a) while capturing cooling effects. Parametric simulations varied shaft spacing $L_{\mathrm{venti}}$, air velocity $v_{\mathrm{wind}}$, and current, with conductor (90°C for cables, 105°C for GIL) and air (40 °C) temperature constraints. Capacity, losses, and LCC per unit transmission task were calculated using cost assumptions from Methods, yielding optimal cost surfaces (Supplementary Fig. 7). The cost-optimal $L_{\mathrm{venti}}$ and $v_{\mathrm{wind}}$ balanced capacity gains with ventilation energy costs, optimizing economic performance.

In summary, this integrated techno-economic framework unifies electromagnetic, thermal, and fluid dynamics analyses with economic cost optimization to design efficient underground transmission systems for renewable energy. Through baseline modeling, electromagnetic safety analysis, spatial layout thermal assessment, and 2D ventilation optimization, it provides a comprehensive methodology for simulating and optimizing tunnel systems. By incorporating multi-physics simulations and cost evaluations, the framework guides the planning, design, and operation of sustainable energy infrastructure, maximizing capacity, ensuring safety, and minimizing LCC per unit transmission task.

Author contributions

Conceptualization, Q.H. and H.F.; Methodology, H.F., Z.Z., and X.T.; Validation, Q.H., Z.Z., A.H.K., and M.S.; Formal Analysis, H.F. and A.H.K.; Investigation, H.F. and Q.H.; Resources, Q.H. and K.L.; Writing—Original Draft, H.F.; Writing—Review & Editing, Q.H., A.H.K., Z.Z., and M.S.; Visualization, H.F., A.H.K., and K.L.; Funding Acquisition, Q.H.

Peer review

Peer review information

Nature Communications thanks Anders Bergman, Illia Diahovchenko and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The datasets generated and analyzed during the current study are available in figshare. 10.6084/m9.figshare.30226579⁷¹. The repository contains the numerical outputs of the finite-element simulations (electromagnetic, thermal, and ventilation scenarios) and derived results that underpin the figures and tables reported in the main text and Supplementary Information. Source data are provided with this paper.

Code availability

The custom MATLAB and Python codes used for data processing, figure generation, current-carrying capacity calculations, and ventilation optimization are available in figshare. 10.6084/m9.figshare.30226567⁷². The repository includes the complete scripts referenced in the paper together with the input data needed to run them.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-71221-5.