From Underground Leakage to Pre-Ignition Flammable Cloud Formation in Buried Hydrogen-Blended Natural Gas Pipelines: A Review and Perspective on Urban Safety

1. Introduction

1.1. Low-Carbon Transition and HBNG

Achieving carbon neutrality demands a fundamental transformation of the energy mix, in which hydrogen has attracted growing interest owing to its zero-carbon combustion characteristics. Hydrogen has a mass energy density of approximately 120 kJ/g, roughly three times that of gasoline, and its combustion yields water as the sole product, making it inherently carbon-free [1]. Projections suggest that large-scale deployment of clean hydrogen could abate approximately 60 Gt of CO₂ by 2050, representing roughly 6% of the total emissions reduction required to meet global net-zero targets [2]. Major economies have elevated hydrogen to the level of national energy strategy. The European Union has launched the European Hydrogen Backbone initiative, targeting approximately 53,000 km of hydrogen transmission infrastructure by 2040. China is steadily pursuing its dual-carbon goals, while the United States has committed over USD 10 billion to establishing regional clean hydrogen hubs. Japan was among the first to articulate a "hydrogen society" vision, backed by a detailed national roadmap [3].

Compared with constructing dedicated pure-hydrogen pipelines, HBNG offers significant engineering advantages and has emerged as a prominent interim solution for global decarbonization. Multiple international studies and demonstration projects have shown that blending hydrogen into natural gas pipelines at 5–25 vol% enables direct reuse of existing infrastructure, eliminating the need for new pipeline construction. This approach achieves substantial emission reductions with minimal economic and social resistance, effectively advancing net-zero targets [4]. Among the infrastructure options available, pipeline transportation is widely recognized as the most efficient and economical solution for large-scale, long-distance hydrogen delivery, making it central to addressing this bottleneck [5]. However, hydrogen blending alters pipeline integrity and failure behavior, highlighting the need for end-to-end risk assessment of buried pipeline leakage scenarios [6,7].

1.2. Urban Safety Risks of Buried Pipeline Leaks

Buried pipelines form the backbone of urban gas systems, and the evolution of their leakage hazards is characterized by concealment, delayed onset, and multi-space coupling. Due to burial depth and soil properties, there is a significant lag in the subsurface migration of leaked gas and its eventual breakthrough at the ground surface. Moreover, existing inspection methods can hardly monitor this subsurface diffusion process effectively, resulting in delayed leak detection that severely compresses the early warning window. Existing studies have shown that, under identical source terms and soil conditions, HBNG is more likely than pure natural gas to shorten the time to first reach the lower flammable limit (LFL), thereby compressing the early warning window. However, the extent of this reduction depends significantly on the leak orifice size, ventilation conditions, spatial dimensions, and the definition of assessment criteria [8].

Multi-space coupling is key to the evolution of buried leaks into high-consequence urban incidents. Gas mixtures leaked from high-pressure pipelines, driven jointly by pressure and concentration gradients in heterogeneous soil, tend to escape preferentially through low-resistance pathways, such as fractures, and rapidly intrude into adjacent enclosed spaces, such as utility tunnels and basements. Owing to the low density and strong buoyancy of hydrogen, hydrogen-rich stratification is more likely to form in the upper parts of enclosed spaces [9].

1.3. Research Gap: From Fragmented Studies to Full-Chain Hazard Understanding

Current research on the leakage evolution of HBNG is mostly confined to individual stages. Underground migration studies have primarily focused on the controlling factors of seepage-diffusion and surface breakthrough [10], whereas aboveground investigations have concentrated on cloud accumulation and ventilation control in enclosed spaces [11,12,13]. Existing macroscopic safety reviews have largely addressed material compatibility or policy issues [14,15,16].

As presented in Table 1, a fundamental limitation shared by these studies is that the complete physical chain, from leak source term through soil and ground surface to enclosed space, is artificially truncated at the ground surface boundary. Because key parameters derived from underground seepage can hardly be directly converted into initial and boundary conditions for aboveground models, aboveground studies are often forced to adopt idealized, uniform source terms, thereby neglecting the decisive influence of upstream transport on the initial concentration and velocity fields. This lack of mechanistic coupling across physical fields leads to a severe disconnect in disaster evolution research, making it difficult for traditional static safety distances or fixed-threshold methods to support the dynamic risk assessment of HBNG in complex urban scenarios [17]. Therefore, there is an urgent need to enable seamless parameter transfer across the ground surface boundary and to establish a full-chain methodology for dynamic evolution analysis and risk characterization.

1.4. Scope and Contributions

To address the aforementioned research gaps, this study, from the perspective of urban gas infrastructure safety, presents a systematic review of the continuous evolution of HBNG from buried leakage to pre-ignition flammable cloud formation. Specifically, focusing on the core mechanisms and key controlling parameters of leak source-term dynamics, transport in porous soil media, and accumulation in enclosed spaces, this review synthesizes the main advances in experimental studies, numerical simulations, and data-driven approaches. On this basis, it critically discusses the gaps in understanding between different physical stages of the cross-medium continuous evolution and their underlying causes, and further analyzes the implications of this mechanistic understanding for pipeline integrity management, early warning, and ventilation-based hazard mitigation. The main contribution of this study lies in integrating existing research from a multi-physics coupling perspective, thereby providing a relatively complete analytical thread for understanding the continuous evolution mechanisms of HBNG leakage hazards.

2. Fundamental Mechanisms of HBNG Leakage Evolution

2.1. Key Properties

As shown in Table 2, the safety of HBNG systems is largely influenced by the differences in the physical properties of hydrogen and natural gas.

Hydrogen exhibits characteristics such as low density, low ignition energy, a high diffusion coefficient, and high burning velocity. These properties collectively affect leak and dispersion, cloud formation, and combustion dynamics through coupling effects, potentially increasing fire and explosion risks. Therefore, incorporating full-chain dynamic risks driven by physical properties into a unified management and control framework is an important prerequisite for the safe blending of hydrogen into urban gas pipeline networks [22].

Among these properties, low density and high diffusivity are key factors that affect the spatial evolution of leakage. Enhanced buoyancy drives rapid gas ascent, leading to stratification at the soil–surface interface and at the top of confined spaces, thereby inhibiting uniform dilution and promoting the localized accumulation of high-concentration gas clouds. Higher diffusivity accelerates the outward expansion of the concentration field, causing the system to reach hazardous thresholds earlier. Existing studies have shown that, under specific operating conditions, 20% hydrogen blending can significantly advance the onset of hazardous conditions and expand the hazard range. Meanwhile, a wider flammable range and lower minimum ignition energy increase the ignition sensitivity of the gas mixture, whereas an increased burning velocity may further elevate the rate of pressure rise during explosion and intensify destructive effects. In summary, hydrogen blending alters the risk evolution characteristics following conventional natural gas leaks. Accordingly, assessments based solely on static diffusion results may underestimate the actual risk boundaries, making it necessary to introduce cross-medium, full-chain dynamic parameters [23,24]

2.2. Source-Term Dynamics

In HBNG buried pipeline leakage accidents, source-term characteristics such as pressure, orifice size, and leak direction directly determine the initial gas release rate and jet momentum, thereby governing the seepage and convective diffusion behavior of the gas in porous soil media and serving as key considerations for risk assessment and control [30]. Among them, pipeline pressure and leak orifice size together constitute the initial dynamic foundation for gas migration. Higher pressure enhances the pressure-differential driving force, whereas a larger orifice size expands the effective leakage area. An increase in either parameter significantly elevates the initial jet momentum, promoting a transition toward convection-dominated transport and enabling the gas to overcome soil resistance more readily. This strong convective effect not only accelerates the expansion of the concentration field and enlarges the subsurface hazard radius, but also leads to a sharp increase in near-field gas concentration, thereby substantially shortening the hazardous time to reach the lower flammable limit [31]. Furthermore, the coupling among leak direction, gas buoyancy, and jet momentum profoundly influences the spatial evolution pathway of the gas. Upward leaks tend to induce vertical gas penetration, causing shallow subsurface zones to reach hazardous thresholds earlier. In contrast, downward leaks tend to result in gas retention in deeper soil layers, forming localized high-concentration accumulation zones with a relatively slower overall migration rate. As the pressure and leak orifice size increase, the enhanced initial kinetic energy further amplifies the spatial differentiation effects associated with the leak direction [32].

2.3. Migration Mechanisms in Porous Soil Media

As shown in Table 3, the migration and surface breakthrough processes of HBNG following a buried leakage are governed by the coupled effects of convection, diffusion-dispersion, and boundary conditions. Among these, soil heterogeneity and backfill boundary characteristics are the core factors determining the evolution pattern of the concentration field, surface breakthrough time, and subsurface hazard range [33]. In the near field of the leak, convection driven by the pressure gradient dominates, and the soil permeability directly determines the advancement rate of the convective front. A high-permeability environment significantly reduces migration resistance, accelerates gas transport, and leads to a substantially earlier breakthrough time [34]. As migration proceeds, diffusion and dispersion jointly govern gas spatial expansion. The high diffusion coefficient of hydrogen enhances pore-scale mass transfer, whereas pore structure heterogeneity and increased porosity further intensify mechanical dispersion, accelerating lateral gas expansion and enlarging the subsurface hazard range [35].

Furthermore, boundary effects induced by stratigraphic heterogeneity and moisture content macroscopically constrain gas migration trajectories and breakthrough patterns. Compared with homogeneous soil, layered structures and heterogeneous backfill induce significant preferential flow along high-permeability pathways, making the evolution of the hazard range highly sensitive to the stratigraphic structure [33]. Meanwhile, an increase in moisture content, or saturation, not only weakens gas-phase seepage and dispersion capacity by occupying pore space, but also increases gas-liquid two-phase resistance, resulting in restricted vertical migration and passively enhanced horizontal diffusion. This water-gas coupling effect dynamically alters gas migration paths, leading to markedly nonlinear characteristics in the surface breakthrough time and evolution of the hazardous area [36].

2.4. Cross-Medium Breakthrough and Enclosed Space Accumulation

The breakthrough of HBNG leakage through the ground surface and its subsequent intrusion and accumulation in enclosed spaces represent a critical spatial turning point in the full-chain risk evolution, capable of rapidly converting concealed subsurface seepage into high-consequence fire and explosion accidents. Once inside such spaces, the gas mixture, driven by buoyancy, tends to form pronounced vertical stratification, while the high molecular diffusivity of hydrogen significantly alters the evolution of the concentration field and affects the gas cloud volume. Therefore, the accurate characterization of gas cloud dynamics at this stage is a core issue in the safety assessment of urban pipeline networks [38].

The concentration field distribution within an enclosed space is primarily governed by the coupled effects of leak source-term intensity and ventilation boundary conditions. Regarding the leak source term, an increase in leak orifice size and hydrogen blending ratio accelerates convective gas injection and species diffusion, leading to higher local peak concentrations and shorter alarm and response times. Existing studies have indicated that under high-flow-rate leakage or high hydrogen blending conditions, the gas ingress rate may exceed the ventilation capacity of conventional systems, thereby limiting the effectiveness of existing ventilation-control measures. Consequently, it is necessary to develop more targeted and enhanced ventilation strategies [39].

Furthermore, the interaction between gas cloud stratification characteristics and the spatial flow field further influences the fire and explosion hazard potential. As the hydrogen blending ratio increases, the buoyancy-dominated accumulation of high-concentration flammable gas at the top becomes more pronounced, which is an important mechanism for risk amplification in sensitive urban enclosed spaces [40]. Although optimized ventilation layouts can reduce hazardous gas cloud volumes to some extent, uncontrolled zones may still arise in high-hydrogen-blend systems under unfavorable conditions such as complex flow field disturbances, ventilation dead zones, or disadvantageous leak locations. This highly non-uniform spatial concentration distribution, combined with uncertainties in the leak location and ignition delay time, further affects the transient response characteristics of the explosion overpressure within the space [41].

3. Research Progress on the Full Hazard Chain

3.1. Research Progress on Underground Leakage and Soil Migration

3.1.1. Experimental Studies

Experimental studies on underground leakage and soil migration of HBNG from buried pipelines can be categorized by scale into small-scale soil box experiments and large-scale field simulation experiments. Small-scale experiments use controlled soil tanks to measure near-field concentration distributions and temperature field evolution, providing reasonably accurate data for mechanism validation. Large-scale experiments simulate complex geological conditions, such as actual burial depth and soil heterogeneity, to quantify gas migration behavior and surface breakthrough time. Current research has expanded from single-parameter testing to multi-factor coupled validation frameworks, offering experimental references for assessing leakage risks in pipeline networks [42].

Small-scale soil box tests primarily focus on local migration characteristics, investigating the influence of source-term conditions on subsurface seepage through adjustment of parameters such as soil particle size and moisture content. Under upward leakage conditions, the vertical gas migration rate is relatively high, and the time required to form the concentration front is shortened. An increased hydrogen blending ratio promotes molecular diffusion, whereas soil heterogeneity induces irregular diffusion paths, thereby expanding the horizontal influence range [17]. Large-scale experiments are used to evaluate the aforementioned physical mechanisms under complex conditions. In specific high-pressure leakage scenarios, the combined effect of increased pressure and larger orifice size enhances jet momentum, leading to a 20%–50% reduction in surface breakthrough time. Anisotropic backfill media make the hazard radius highly sensitive to layered structures, and fluctuations in moisture content exert complex nonlinear effects on the first hazardous time. Recent experiments have combined undisturbed cutting-ring soil sampling with field temperature monitoring and introduced porous media thermo-hydraulic coupling analysis to quantitatively assess the constraining effects of backfill physical boundaries on subsurface seepage. Based on mechanistic analysis, the Joule-Thomson throttling effect during high-pressure gas expansion generates a corresponding temperature gradient, and related pore parameter tests confirm that high-pressure gas exhibits preferential flow and fingering in heterogeneous soil. This seepage trajectory influenced by spatial heterogeneity provides a physical basis for validating full-chain dynamic evolution models [43].

3.1.2. Numerical Simulation Studies

Numerical simulations of the buried leakage of HBNG have gradually formed a technical pathway ranging from mechanistic analysis to intelligent prediction. This framework, which is centered on computational fluid dynamics and porous media seepage models, has been further extended to empirical correlations and machine learning methods, providing important methodological support for dynamic risk quantification under diverse operating conditions [44].

In the mechanistic analysis stage, computational fluid dynamics models serve as fundamental tools for investigating gas flow and diffusion characteristics. Although early studies often relied on simplified assumptions regarding the soil medium, this approach revealed the fundamental evolution patterns of the flow field. Relevant simulation results indicate that under specific conditions, when the hydrogen blending ratio increases to 20%, the subsurface hazard range expands and the first hazardous time is correspondingly shortened owing to changes in gas diffusion kinetics [45]. To bring simulation conditions closer to actual geological environments, some models have further incorporated nonlinear seepage and mechanical dispersion effects in porous media. The results show that after accounting for the variations in flow resistance caused by heterogeneous backfill boundaries, the horizontal gas diffusion range further increases, which to some extent compensates for the inability of simplified models to describe irregular fingering fronts [46].

With the increasing demand for assessment timeliness in engineering applications, related research has gradually evolved toward rapid prediction. To overcome the time-consuming nature of three-dimensional high-fidelity simulations, researchers have converted extensive simulation results into empirically based predictive models suitable for engineering practice. These multivariable correlation-based computational tools can relatively quickly estimate hazard boundaries under complex conditions such as high pressure and large orifice sizes, providing references for on-site emergency decision-making [47].

In recent years, the application of data-driven techniques has further advanced risk simulation toward dynamic prediction. Machine learning methods, represented by artificial neural networks, can reasonably well fit complex nonlinear relationships among multi-source hazard-inducing parameters through training on large datasets, thus serving as important complements to traditional mechanistic models. This technology to a certain extent enables the rapid prediction of hazard ranges in complex urban scenarios and establishes a new connection between fundamental physical mechanism analysis and dynamic monitoring of pipeline networks [48].

3.1.3. Key Driving Factors and Engineering Implications

The underground migration of buried HBNG leakage is a physical process involving the synergistic coupling of multiple parameters, including the hydrogen blending ratio, leakage pressure, orifice size, burial depth, release direction, soil pore characteristics, and heterogeneous boundaries. Among these, the physical properties of the gas mixture are primarily influenced by the hydrogen blending ratio. An increase in this ratio accelerates the flow field evolution rate through enhanced diffusion characteristics, thereby advancing the first hazardous time and expanding the subsurface hazard boundary [49]. In terms of leak source-term dynamics, pressure and orifice size jointly determine the initial jet momentum, and the combination of high pressure and large-orifice leakage typically further shortens the migration period of hazardous gases. Additionally, gas released upward, coupled with buoyancy, tends to break through the ground surface more rapidly, whereas an increased pipeline burial depth can delay this process to some extent [50]. Soil environmental characteristics also profoundly influence macroscopic migration paths. High porosity generally accelerates convective processes, whereas heterogeneous backfill structures reshape the gas diffusion front morphology by altering the medium resistance distribution, thereby expanding the overall potential impact zone.

These physical mechanisms provide important references for the integrity assessment and risk management of HBNG pipeline networks. The shortening of breakthrough time caused by an increased hydrogen blending ratio, combined with the nonuniform plume characteristics in the leak near field, renders conventional periodic inspection methods inadequate for timely response, thereby driving the transition toward real-time dynamic early warning systems. The expansion of hazard boundaries under high pressure further elevates the overall system risk. Accordingly, pipeline network operators can establish dynamic, tiered risk control zones by integrating the evolution patterns of hazard boundaries in buried leakage scenarios and the subsequent gas cloud characteristics [51]. Furthermore, considering the non-uniform distribution characteristics of the gas mixture in the soil-surface transition zone, the rapid buoyant rise and escape of hydrogen may weaken the effective coverage of conventional single-component sensors. Based on this, pipeline integrity management systems may consider introducing a multi-domain feature fusion monitoring framework, identifying adverse conditions such as high-pressure operation, upward leakage, and high-permeability soil as key control scenarios, thereby advancing pipeline risk management toward more proactive predictive maintenance [52,53].

3.2. Surface Escape and Cloud Evolution

3.2.1. Surface Plumes and Escape

The quantitative assessment of the surface escape behavior of HBNG currently relies primarily on three-dimensional numerical simulations [54,55]. This process, a critical transitional step in the conversion of leaked gas into a hazardous pre-ignition gas cloud, is influenced by multi-physics coupling effects. The flow field evolution exhibits a certain sensitivity to the hydrogen blending ratio. Existing simulation results indicate that under specific conditions, when the hydrogen blending ratio increases from 0% to 20%, both the alarm response time and the time to reach the lower flammable limit are reduced. This evolutionary trend reflects the role of the higher molecular diffusivity of hydrogen in promoting the gas cloud accumulation rate.

The interaction between leak source-term parameters and local environmental conditions further reshapes the evolution trajectory of the flow field. At the dynamic level, a higher initial leak rate, by enhancing turbulent mixing, typically accelerates the concentration field expansion and shortens the alarm time. Compared with horizontal release, vertically upward release, under the coupled effect of buoyancy and jet momentum, more readily breaks through the ground surface and forms a high-concentration front above the ground Furthermore, local physical obstacles in open environments hinder free plume diffusion and induce local gas retention, thereby further shortening the time to reach hazardous thresholds in specific scenarios [56].

The interaction between external wind fields and building structures has a macroscopic influence on the ultimate diffusion boundary of the gas cloud. An increase in wind speed typically suppresses the vertical development of the plume and expands the horizontal hazard range in the downwind direction. Notably, when the leak source is directly facing a building façade, the gas may rise along the exterior wall and infiltrate the interior owing to the wall-adherence effect, thereby altering the distribution of flammable components inside the building. Relevant modeling studies indicate that installing physical baffles along critical diffusion paths can to some extent reduce indoor leak concentrations. This intervention mechanism, based on flow obstruction effects, can provide a reference for the safety protection design of urban surface pipeline networks and adjacent sensitive spaces [57].

3.2.2. Accumulation in Semi-Enclosed Spaces

After HBNG escapes into semi-enclosed infrastructures such as valve chambers, trench ducts, and utility tunnels, the constrained geometric boundaries further amplify the accumulation risk of flammable gas clouds. An increased hydrogen blending ratio exacerbates the vertical stratification of the gas within a semi-enclosed space. Inside facilities such as valve chambers, hydrogen, driven by buoyancy, tends to reach the top first and form localized hydrogen-rich accumulation zones. This upper-enrichment characteristic enhances the fire and explosion hazard following ignition, and accordingly multi-component detection devices should be considered at the top of the space [58]. Similar fluid coupling effects also exist in local dead zones within utility tunnels or semi-enclosed transition areas with overhead covers. Such locally obstructed structures impede the free convective diffusion of the gas, causing highly diffusive gas components to more readily form a high-concentration retention layer beneath the roof slabs. Strategically placed flow-guiding barriers within these structural blind zones can help limit hazardous gas cloud migration toward adjacent underground spaces.

In spaces with channel-like characteristics, the flow field evolution exhibits different dynamic patterns. Narrow, elongated trench ducts may develop flow features that resemble convective circulation. The combination of bottom obstructions and upward jet release intensifies local accumulation and shortens the early warning window [59]. The macroscopic flow field evolution within utility tunnels is directly influenced by leak source-term parameters. In general, an increase in leak orifice size and hydrogen blending ratio jointly accelerates the concentration field expansion, shortening the alarm response time while raising the equilibrium concentration of the gas mixture in the space. Under high hydrogen blending scenarios, conventional mechanical ventilation rates may fail to meet safety control requirements, necessitating more targeted adaptive ventilation strategies. Relevant simulations indicate that a higher hydrogen blending ratio expands the equivalent explosive volume of the flammable gas cloud. On the other hand, optimizing the spacing of ventilation ports and enhancing local exhaust efficiency can to some extent suppress hazardous gas accumulation. Under adverse conditions, high-pressure pipelines often require higher emergency ventilation capacities. For extreme release scenarios, relying solely on ventilation control may be insufficient to contain gas cloud expansion, making it necessary to engage safety interlock systems to cut off the leak at the source in a timely manner [60].

3.2.3. Pre-Ignition Flammable Cloud Dynamics

The evolution characteristics of flammable gas clouds formed after HBNG escapes into semi-enclosed spaces serve as an important basis for assessing the full-chain dynamic risks. AS a direct indicator of the fire and explosion potential of the system, the hazardous gas cloud volume typically exhibits an expanding trend with an increasing hydrogen blending ratio. Driven by buoyancy, upward-released gas mixtures can fill a semi-enclosed space relatively quickly, and the volume evolution exhibits a staged process characterized by initial rapid growth, followed by dynamic equilibrium and gradual decay. During this dynamic evolution, the absence of forced ventilation leads to persistent hazardous gas retention and enlargement of the gas cloud volume, whereas optimizing the layout of air inlets and outlets to enhance ventilation efficiency can serve as an important measure for controlling the fire and explosion potential [61].

The early warning window of the system is jointly governed by the persistence time of the gas cloud and the rate at which the lower flammable limit is reached. The enhanced diffusivity associated with a higher hydrogen blending ratio exerts a dual effect on the early warning window. On the one hand, it accelerates the overall dissipation of the hazardous gas field. On the other hand, it also promotes the faster propagation of the flammable mixture, thereby shortening the available warning time. Within semi-enclosed spaces, the buoyancy-dominated gas concentration distribution typically exhibits pronounced vertical stratification, making the top region more likely to reach the lower flammable limit first and form a hazardous gas cloud. This concentration evolution pattern, governed by spatial non-uniformity, further challenges the monitoring effectiveness of conventional single-component sensors [62].

The underlying mechanisms of the above evolution characteristics lie primarily in the differential diffusion and non-uniform accumulation of multi-component gases. Upon leakage, hydrogen, driven by stronger buoyancy owing to its lower density, tends to accumulate upward and form a high-concentration zone in the upper region of the semi-enclosed space. In contrast, methane tends to remain distributed across the middle and lower regions, exhibiting more pronounced lateral spreading characteristics. This tendency toward spatial separation of components increases the overall concentration gradient, leading to actual peak concentrations at the top that are often higher than those predicted by conventional homogeneous mixing assumptions. The introduction of mechanical ventilation under these conditions generates more complex flow disturbances. Although forced convection can weaken the component stratification at the top and dilute peak concentrations, it may also carry hazardous gases into a larger area during flow field restructuring, thereby altering the dissipation dynamics of the entire flow field to some extent [63].

3.2.4. Key Driving Factors and Engineering Implications

The pre-ignition risk management of HBNG in semi-enclosed spaces is highly dependent on the identification and quantitative assessment of core hazard parameters. Existing studies indicate that key source-term driving factors, such as the hydrogen blending ratio, operating pressure, and leak orifice size, significantly influence the scale of hazardous gas cloud evolution. A quantitative indicator system constructed based on these driving factors provides an important basis for defining the safe operating boundaries of pipeline networks and offers theoretical references for engineering prevention measures, including high-consequence area delineation, ventilation system optimization, and intelligent monitoring and early warning [64,65].

High-consequence area delineation is fundamental for developing evacuation and risk prevention strategies. The combined effect of an increased hydrogen blending ratio and confined geometric boundaries is an important mechanism that leads to the expansion of the potential high-consequence area. The resulting increase in hazardous gas cloud volume can be mitigated to some extent by optimizing the spacing between air inlets and outlets. Given that traditional static assessment methods may underestimate the early-stage evolution risk under high-hydrogen-blend systems, it is necessary in engineering management to incorporate the dynamic evolution characteristics dominated by the aforementioned core driving factors into risk prediction models, thereby improving the alignment between risk zoning and emergency response time windows [66].

Local ventilation design is a key engineering measure for enhancing the safety of underground facilities. Under adverse conditions such as a high hydrogen blending ratio or large-orifice leakage, conventional fixed ventilation rates often fail to maintain ambient concentrations below safety thresholds. Introducing variable frequency adaptive ventilation strategies based on spatial concentration feedback can serve as an important means to mitigate this issue. As the hydrogen blending ratio increases, correspondingly raising the local ventilation frequency and optimizing the layout of the top exhaust ports helps reduce the area of indoor hazardous zones and the equivalent explosive volume. During the engineering design phase, sensitivity analyses incorporating source-term parameters such as the hydrogen blending ratio and leak orifice size should be conducted to promote the transition of pipeline network ventilation protection from a static fixed mode to a dynamic control mode [67].

Real-time monitoring of key nodes, such as valve chambers and utility tunnels, is an important safeguard for achieving early accident warnings. An increased hydrogen blending ratio not only compresses the effective warning time of the system but also induces stratification and dispersion characteristics that may weaken the coverage of conventional single-component sensors. To address this issue, the integration of deep learning algorithms with multiparameter fluid models offers a new technical pathway for the rapid inversion and localization of leak sources. At the engineering implementation level, multi-component array sensors can be deployed inside semi-enclosed facilities, especially in top areas prone to accumulation, to enhance the early risk perception capability of the full pipeline network chain.

In summary, pre-ignition dynamic warnings are highly dependent on the systematic assessment and comprehensive prediction of various hazard-driving factors. Pipeline network operators can incorporate quantitative indicators, such as the hydrogen blending ratio, leak orifice size, and their derived surface breakthrough time and hazardous gas cloud volume, into the pipeline system-level integrity management framework. By introducing intelligent algorithms to construct multidimensional dynamic warning models, pipeline integrity management is expected to gradually evolve from passive accident response to more proactive predictive maintenance, thereby reducing the probability of high-consequence fire and explosion accidents over the entire lifecycle [68].

4. Research Gaps and Future Perspectives

4.1. Fragmentation in Full-Chain Physical Evolution Research

In the existing body of research on the leakage risks of buried HBNG pipelines, there remains a pronounced fragmentation of the physical stages. Most current studies focus on a single stage, such as subsurface diffusion or accumulation in semi-enclosed spaces. Subsurface diffusion studies have predominantly examined how soil properties influence the subsurface pressure field and hazard extent, but their scope typically terminates at the moment of surface breakthrough, leaving the non-uniform surface escape flux unincorporated as a dynamic input condition for subsequent aboveground evolution. On the other hand, studies on semi-enclosed spaces focus on the volume expansion and concentration stratification of flammable clouds within semi-enclosed geometries, yet they generally simplify the surface release as an idealized uniform source, thereby neglecting the actual influence of soil transport on the spatial distribution of the plume. Furthermore, existing macroscopic safety reviews mostly emphasize system-level qualitative assessments or single-scenario analyses, lacking systematic integration of the continuous evolution mechanisms from subsurface migration to aboveground dispersion. This disconnect in the physical boundary conditions makes it difficult for early warning indicators obtained from the subsurface stage to directly support predictions of subsequent accumulation and hazard evolution. Future research needs to further bridge the continuous evolution chain connecting subsurface migration, surface escape, and spatial accumulation to address the knowledge gaps in the key physical transition stages [69,70].

4.2. Lack of Full-Scale Multi-Physics Coupled Experimental Data

In research on the leakage risk assessment of HBNG, the lack of full-scale, multi-physics coupled experimental data significantly limits the engineering generalization capability of relevant predictive models. Existing physical experiments are mostly confined to small-scale laboratory setups or pure natural gas conditions, making it difficult to realistically reproduce the full-chain leakage evolution process under actual burial depths and complex soil characteristics. Although numerical simulations are widely adopted in related studies, their validation benchmarks are often derived from limited-scale, low-pressure test data. This scale discrepancy may further amplify prediction deviations when numerical models are applied to the real-world conditions of high-pressure urban pipeline networks. Currently, the industry still lacks a reasonably reproducible, standardized test dataset covering the entire process, resulting in insufficient data support for model calibration and early warning assessment. In the future, prioritizing the construction of full-scale, multi-physics integrated experimental platforms and establishing high-quality standardized test databases will be an important foundation for advancing the practical engineering application of dynamic risk models [71].

4.3. Limitations in Soil Simulation

Current numerical simulations have certain limitations in the physical characterization of soil environments. Most subsurface diffusion studies rely heavily on seepage models based on the assumption of homogeneous media, often neglecting the stratigraphic differences, compaction effects, and spatial non-uniformity of moisture content distribution in actual trench backfill operations. Field tests and comparative experiments indicate that heterogeneity in soil pore structure and permeability alters the preferential seepage paths of gas and promotes the formation of irregular diffusion fronts, leading to varying degrees of expansion of the actual horizontal diffusion range. Simplifying such locally compacted structures often causes the simulated concentration field distribution to deviate from the actual evolution pattern, thereby introducing systematic bias into risk assessments. Although some models have attempted to incorporate relevant parameters for correction, standardized data from real field geological conditions currently remain lacking. Future research should focus on obtaining the macro and micro-scale physical parameters of actual trench backfill and developing seepage models calibrated against real heterogeneous soil conditions to reduce prediction errors between theoretical simulations and engineering practice [72].

4.4. Bottlenecks in High-Resolution Measurement of Gas Clouds

The accurate characterization of the dynamic diffusion process of pre-ignition flammable gas clouds remains constrained by the temporal and spatial resolution limitations of current measurement techniques. Existing experiments mostly rely on fixed-point concentration sensors for low-frequency sampling, and this conventional discrete measurement approach struggles to capture the local transient non-uniform flow field evolution induced by turbulence-buoyancy coupling. Although relevant experiments can macroscopically observe the phenomenon of hydrogen rising first and forming stratification, the lack of fine-scale monitoring data on the temporal and spatial evolution of local regions with steep concentration gradients introduces considerable uncertainty into the assessment of effective warning system coverage and limits the validation accuracy of numerical models. While some studies have attempted to introduce high-frequency optical velocimetry and full-field visualization techniques, these methods are currently mostly applied to low-pressure, small-scale conditions owing to constraints in the equipment field of view and testing conditions, making it difficult to fully cover the entire dynamic process from surface breakthrough to gas cloud formation in enclosed spaces. To improve the prediction accuracy of non-uniform cloud evolution, future efforts should focus on developing synchronized multi-parameter high-resolution non-contact measurement techniques to enable the quantitative capture of transient flow field evolution characteristics under complex leakage scenarios [73].

4.5. Disconnection Between Mechanism Studies and Risk Assessment

The disconnect between fundamental physical mechanism research and macroscopic risk assessment systems is an important weakness that currently constrains the safe management of hydrogen-blended pipeline networks. Although existing fluid dynamics simulations can quantify the effects of the hydrogen blending ratio and heterogeneous boundaries on surface breakthrough time and the scale of hazardous gas cloud evolution, most of these quantitative indicators remain at the level of physical phenomenon description and have not yet been effectively incorporated as dynamic input parameters into quantitative risk assessment (QRA) models. When calculating individual and societal risks, current assessment frameworks largely rely on historical static failure probabilities and fixed safety distance benchmarks derived from conventional pure natural gas systems. This practice fails to fully capture the actual influence of hydrogen's high diffusivity and vertical stratification characteristics on transient hazard evolution in space, potentially leading to an underestimation of the fire and explosion risk of actual pipeline networks. Meanwhile, the alarm threshold settings in existing pipeline condition monitoring schemes are mostly based on ideal homogeneous mixing assumptions without adequately incorporating actual fluid dynamics principles, such as hydrogen buoyancy-driven stratification. This disconnect between the underlying physical mechanisms and upper-level assessment tools hinders the transformation of conventional passive inspection modes into proactive predictive maintenance. Future research should focus on bridging the gap between microscale diffusion mechanisms and macroscale risk assessment models, as well as establishing an integrated management framework driven by flow field dynamic parameters [74,75,76].

4.6. Lag in Full-Lifecycle Integrity Management of Hydrogen-Blended Pipeline Networks

For HBNG pipeline networks aimed at low-carbon transition, the development of a full-lifecycle integrity management system still exhibits a certain degree of lag. Current pipeline integrity management standards are largely established based on operational experience from conventional pure natural gas networks, and they gradually reveal significant applicability limitations when supporting the safety justification of pipeline networks with medium-to-high hydrogen blending ratios. Although some pioneering efforts have attempted to introduce information technologies, such as digital twins, to enhance pipeline health assessment capabilities, the underlying drivers of relevant predictive models remain highly dependent on static historical failure data and have not yet achieved deep coupling with the transient physical evolution mechanisms of hydrogen-blended leakage flow fields. The information barrier between the physical evolution and macro-management layers makes existing control platforms prone to systematic prediction biases when responding to non-uniform hazardous gas cloud evolution. Moreover, current monitoring systems, when optimizing sensor array topology, generally rely on ideal diffusion assumptions and fail to effectively integrate prior fluid dynamics knowledge, such as the non-uniform dispersion dominated by hydrogen buoyancy, resulting in significant sensing blind zones in early warning networks. To meet the macro-level requirements of low-carbon transition in urban gas systems, future efforts should urgently focus on developing a full-lifecycle intelligent control platform that deeply integrates physics-informed and data-driven approaches, thereby advancing the integrity management system of hydrogen-blended pipeline networks toward a dynamic, proactive predictive maintenance paradigm [77,78,79].

5. Conclusions

(1) The buried leakage of HBNG is a continuous physical evolution process. High-pressure gas seeps through the soil and escapes to the ground surface, and the state of this surface escape directly constitutes the initial condition for subsequent spatial accumulation. Gas stratification in enclosed spaces is profoundly influenced by the prior subsurface seepage behavior. An objective physical correlation exists between underground and aboveground fluid evolution. Therefore, pipeline network risk assessment must be conducted based on the complete evolutionary process.

(2) Current research is mostly confined to isolated single stages, such as subsurface seepage or aboveground space ventilation, and lacks sufficient analysis of the transition phases of gas evolution. Most numerical models adopt homogeneous soil assumptions, making it difficult to accurately reflect non-uniform flow field variations under realistic backfill conditions. The scarcity of large-scale experimental data and high-resolution measurement techniques limits the validation of diffusion mechanisms, leading to a tendency for single-stage assessment results to underestimate the overall risk.

(3) Future risk control of pipeline networks should be based on realistic fluid physical mechanisms. Relevant assessments need to convert parameters such as the hydrogen blending ratio and leakage parameters into direct quantitative indicators, including the surface breakthrough time and hazardous volume. The principles of fluid stratification and concentration diffusion can be directly applied to guide the rational layout of detection equipment and the formulation of ventilation strategies. Preventive measures based on the underlying physical laws can substantially enhance the safety protection capability of pipeline networks.

(4) Incorporating of fluid dynamic parameters into management systems is an engineering requirement for enabling pipeline networks to adapt to the low-carbon transition. An assessment framework based on realistic evolution mechanisms can compensate for the errors inherent in traditional static historical models. Combining leakage physics with intelligent early warning algorithms enables a more accurate definition of safety boundaries under high-blend-ratio conditions. This approach can provide practical technical support for the safe utilization and predictive maintenance of urban gas infrastructure.