1 Introduction

The ocean accounts for about 71% of the Earth's surface area. The diversity and abundance of its species and resources make marine ecosystems an important part of the Earth's ecosystems (Liu and Su, 2014). However, with economic and social development, a large number of pollutants are discharged into the ocean. Oil spills are one of the main sources of marine pollution. Major oil spills will bring huge losses to China's marine fisheries, mariculture, coastal tourism, and other industries, as well as pollute the ocean and coastal environment. The marine ecosystem is already irreversibly damaged. In the Bohai Sea, the explosion of the oil pipeline in the Dalian Xingang crude oil tank farm in July 2010 and the oil spill on the seabed near the Penglai 19-3 oilfield platform in June 2011 caused a large amount of oil-based mud to sink to the seabed. Evidently, oil spills have caused serious pollution in the Bohai Sea.

With the rapid development of China's marine economy, offshore oil exploration and development, coastal oil ship transportation, the construction of large oil storage bases, and petrochemical projects are increasing, which has also exacerbated the risk of marine oil spills. Especially in the Bohai Sea, hidden dangers of oil spill risk in offshore pipelines and other facilities are present. At this stage, comprehensive risk assessment and risk management experience in response to marine pipeline oil spills is lacking, aggravating the harm of marine oil spill disasters to marine ecosystems. To deal with potential oil spill risks of offshore pipelines, a comprehensive oil spill risk assessment model and regional oil spill risk assessment are imperative.

In recent years, research on marine oil spill risks has received widespread attention from researchers worldwide due to the frequent occurrence of marine oil spills. Experts have conducted more in-depth research into the analysis of oil spill risk sources and the assessment of oil spill risk probability. Such work mainly includes analyzing the risk sources of the leakage accidents, discussing the probability of oil spill accidents, and estimating the possible oil spill size.

Among domestic researchers, Lin and Ge (2010) used the discrete binomial probability distribution model to calculate the probability of ship oil spills in the waters of Luoyuan Bay and Baima Port. Ma (2012) used random distribution and analytic hierarchy processes to calculate the probability of an oil spill. Han et al. (2013) simulated and analyzed the migration of oil spills over time and their spatial distribution characteristics based on the oil particle model. Liu et al. (2013) developed environmental pollution accident risk mapping for risk assessment in the Minhang District, Shanghai. Wu and Zhao (1992) analyzed the expansion, dispersion, and migration movement patterns of oil spills and established oil spill models to analyze oil spill pollution in Lingdingyang. Liu et al. (2015) assessed the oil spill risk in the entire China Bohai Sea by using historical data from drilling platforms and ships. Huang et al. (2015) established a model based on the environmental fluid dynamics code model for the impact of oil spill accidents at the Nanjing section of the Yangtze River. Song et al. (2019) constructed a risk management model for hazardous material process facilities based on Bayesian influence diagrams and discussed management and control strategies that can effectively reduce system risks. Tong et al. (2020) combined an oil spill model with the single-shot detector curve method to assess the impact of the oil spill on March 19, 2011, in Bohai Bay on the ecological environment. Ji et al. (2021) used the overall system risk analysis model to simulate large-scale oil spill accidents as an example of oil spill prevention and response.

With regard to recent foreign research, Thomas and Thomas (1986) introduced three basic methods of oil spill risk analysis and assessment. At present, using modeling tools has become a common practice for analyzing oil trajectories for the planning or assessment of oil spills (Olita et al., 2012). BP Exploration LTD (2002) applied the oil spill information system model to assess the oil spill risk in the Caspian Sea and calculated the oil spill size, risk probability, and landing time under different oil spill scenarios in summer and winter. Mcallister (2009) calculated the probability of riser accidents of different sizes based on the statistical results of 120 offshore pipeline accidents in the United Kingdom. Det Norske Veritas (DNV) (2011) estimated the risk of pollution from marine oil spills in Australian ports and waters to support a review by the Australian Maritime Safety Authority of the National Plan to Combat Pollution of the Sea by Oil. The Australian Maritime Safety Administration (DNV, 2011) developed an oil spill risk database (OSRD) model based on a comprehensive analysis and calculation of the various types of risk probability. Nelson et al. (2015) focused on deep water oil spills and utilized an oil spill simulation model called the blowout and spill occurrence model in the Gulf of Mexico. Canu et al. (2015) and Cucco et al. (2012) assessed oil spill risks by using the Bonifacio oil spill operational model. Arzaghi et al. (2018), considering the uncertainty of the input variables, proposed the fugacity model to describe the migration and fate of crude oil leakage from a pipeline. Lavine et al. (2020) studied the impact of sea level rise on the drift of the TELEMAC-2D oil spill model. Sajid et al. (2020) used Bayesian networks to evaluate the cost and cycle of marine ecological restoration caused by oil spills. Barreto et al. (2021) constructed an oil spill coupling model and an oil spill emergency response model and conducted a case study during the trial production period of the deep spill oilfield.

Studies on oil spills from pipeline drift diffusion simulation and risk evaluation have been extensively conducted around the world. The above-mentioned research results, such as the hydrodynamic model, oil particle model, and oil spill phased diffusion theory, provide a good reference for this article. However, a search through the existing research shows that a systematic and large-scale oil spill risk assessment method in China is still lacking. In particular, in the coastal area of China, no systematic risk assessment research combines risk probability and dynamic modeling. Using the Bohai offshore pipeline as an example, this paper constructs a method of oil spill risk assessment on offshore pipelines and a calculation of oil spill risk to provide a scientific basis for comprehensive risk assessment and risk management in the sea area.

2 Study Areas

The research area is shown in Fig. 1. Currently, 27 offshore oil platforms, groups of offshore oil and gas fields, and more than 1600 km of oil pipelines are found in the Bohai Sea. After more than 30 years of development and operation, the oil spill risk of pipelines from offshore oil exploration, development, and transportation facilities has become tremendous.

3 Model and Methodology

The oil spill risk assessment process of the Bohai offshore pipeline constructed in this paper is shown in Fig. 2.

The research content of this paper is mainly divided into three parts: Bohai Sea offshore pipeline oil spill risk source assessment, oil spill risk diffusion simulation, and regional risk assessment in the Bohai Sea.

The first step is to assess the source of oil spill risk in offshore pipelines. This article refers to Lu et al. (2014) with regard to the pipelines' oil spill probability. The oil spill risk of the Bohai Sea pipeline is evaluated from five criterion levels: corrosion, fatigue, natural hazard, third party, and operational. The second step is the oil spill risk diffusion model. The hydrodynamic conditions, such as tidal currents and residual currents, and sea surface meteorological conditions, such as wind-sea currents and wind-induced oil film drift, are considered in the Bohai Sea to simulate the drift and diffusion of oil spills. The sea surface, weather conditions, water temperature, and oil product parameters, among others, are considered comprehensively to construct an oil spill weathering model. The simulation results of oil spill drift and diffusion are multiplied by the risk probability, and the simulation results of different scales of oil spill size are superimposed. Finally, the calculation results of the 12 months are superimposed to obtain the calculation results of the oil spill risk of the Bohai Sea pipeline.

3.1 Offshore Pipeline's Oil Spill Frequency Model

To estimate the oil spill frequency of risk sources, we adopted the OSRD model established by DNV. The OSRD model includes the frequency of oil spills of more than 1 ton due to pipelines within the safety zone of 6.3 × 10⁻⁴ × 7.3^−0.46 = 2.5 × 10⁻⁴ per pipeline-km year and the frequency of oil spills of more than 1 ton due to pipelines in the open sea (i.e., between the platform safety zone and the pipeline landfall) of 4.9 × 10⁻⁵ × 7.3^−0.46 = 2.0 × 10⁻⁵ per pipeline-km year.

On the basis of the OSRD model, the probability assessment combined with the Bohai pipeline oil spill (Lu et al., 2014), a risk source probability model for Bohai Sea oil pipelines is established.

$ {F_s} = p \times 2.5 \times {10^{ - 4}}{Q^{ - 0.46}}, $

(1)

$ {F_o} = p \times 2.0 \times {10^{ - 5}}{Q^{ - 0.46}}L, $

(2)

where F_s is the annual probability of oil spills occurring in pipelines within the safety zone, F_o is the annual probability of oil spills in pipelines in the open sea, p is the probability evaluation score of an oil pipeline in the Bohai Sea divided by 45, Q is spill size in ton, and L is pipeline length, the unit is km.

3.2 Hydrodynamic Model

The marine hydrodynamic model used in this study is FVCOM, which is a prognostic, unstructured grid, finite-volume, free-surface, three-dimensional (3D) primitive equation coastal ocean and estuarine model (Chen et al., 2003). With its accurate geometric representation of irregular coastlines and islands and sufficiently high horizontal resolution in narrow channels, FVCOM accurately simulates the tidal current in the bays and also resolves the strong tidal flushing processes in narrow channels (Zhao et al., 2006).

The simulated sea area extends from the North Yellow Sea in the east to the entire Bohai Sea in the west. The 3D computing grid plane is a triangular grid (Wu et al., 2013). The highest resolution reaches 100 – 200 m. The number of triangular grids is 38718, as shown in Fig. 3.

The main tidal current ellipse elements S₂, M₂, K₁, and O₁ in the study area are shown in Fig. 4.

This study uses the 12-month climatic wind field, as shown in Fig. 5.

3.3 Oil Particle Model 3.3.1 Movement of seawater on particle transport velocity V₁

Tidal ellipse elements are used to quickly calculate the tidal currents. Circulation and wind-ocean current are superimposed to obtain the velocity of the particle location, where u₁, u₂, u₃, and u₄ represent the tidal current, residual current, wind-ocean current, and wind drift, respectively.

$ {V_1} = f({u_1}, {u_2}, {u_3}, {u_4}) . $

(3)

3.3.2 Random movement of oil particle velocity V₂

The Monte Carlo method is used to simulate random turbulent motion, where ε and k_h represent the distributed random number and the turbulent eddy viscosity coefficient on the [−1, 1] area, respectively, and dt is the calculated time step.

$ {V_2} = \varepsilon \sqrt {6{k_h}/{\text{d}}t} . $

(4)

3.3.3 Vertical movement velocity V₃

V₃ is caused by the density difference between the oil spill and sea water.

$ {V_3} = kd_i^2g(1 - {\rho _c}/{\rho _w})/{v_w}, $

(5)

where V₃ is the vertical velocity of the difference between the density of oil particles and sea water affected by buoyancy, k is the buoyancy coefficient (recommended value: 0.056), d_i is the particle diameter, g is the acceleration of gravity, ρ_w is the density of sea water, ρ_c is the density of hazardous chemicals, and v_w is the seawater dynamic viscosity coefficient.

3.3.4 Spilled oil particle diffusion velocity V

The calculation method of the oil spill particle diffusion velocity is shown below, where V_f is the wind-induced drift velocity.

$ V = V({V_1}, {V_2}, {V_3}, {V_f}) . $

(6)

3.4 Oil Spill Risk Assessment Methods

According to the established oil spill frequency model, the hydrodynamic model, and the oil particle model of the Bohai offshore pipeline's oil spill, an integrated risk diffusion model of the Bohai oil pipeline is established. This model can be used to simulate and calculate the oil spill diffusion field of a single-risk point source and a certain oil spill size in a certain month and multiply the oil spill probability of the risk point source to calculate the oil spill risk diffusion field.

To reduce the number of simulation calculation cases, the pipeline risk line source is converted into a risk point source with a resolution of 3 km. The number of point sources after conversion is 455. The spill sizes of the risk source are divided into three simple rankings in this study by performing many numerical experiments. These rankings are 1 t to 10 t (A, typical spill size is 5 t), 10 t to 100 t (B, typical spill size is 50 t), and more than 100 t (C, typical spill size is 1000 t). In this study, simulation calculation analysis is conducted on a monthly basis. The oil film distribution field is output every hour. The monthly oil spill wind direction diffusion field is obtained by superimposing within the month. The total number of oil spill risk diffusion simulation calculation cases is 16380 (455 × 3 × 12 = 16380).

4 Results and Discussion 4.1 Distribution of Oil Spill Risk Value

On the basis of the established marine oil spill risk model, the simulation results of the oil spill diffusion field of each offshore pipeline risk point source are obtained, each featuring a typical oil spill size. In this research, the oil spill diffusion field of each month and hour is calculated. The hourly oil spill diffusion field is multiplied by the probability (corresponding to the probability of the pipeline risk point source and oil spill size) to obtain the hourly oil spill risk diffusion field. This study superimposes the oil spill risk diffusion fields of all point sources and oil spill size to obtain the hourly oil spill risk diffusion fields in the study area. Further temporal superposition can be performed to obtain monthly, seasonal, and annual oil spill risk fields.

The seasonal results of the oil spill risk diffusion field of the Bohai Sea Pipeline are shown in Figs. 6 and 7. Figs. 6(a – d) show the distribution of oil spill risks in four seasons. Fig. 7 shows the annual distribution of oil spill risk. The median line in the figure is set to 1, 2, 5, 10, and 20, and the unit is t km⁻².

In the spring (March to May, Fig. 6(a)), the overall risk of oil spills from pipelines is lower than that in other months. They are mainly distributed near dense areas of oil pipelines and near the northern coast. The western Bohai Bay, the northern coastal waters, the northern waters of the mouth of the Yellow River, and the northern waters of Liaodong Bay have a higher oil spill risk. The highest value obtained is 6.90, which is found in the coastal area of Tangshan in the northern part of Bohai Bay.

Areas that have a high oil spill risk in spring are distributed in dense pipelines as a result of the action of the northerly monsoon. The flow moves from the west coast of the Bohai Sea to the north, turns to the western waters of the top of Liaodong Bay, and flows south along the east coast of Liaodong Bay. Meanwhile, the inflow velocity of Liaodong Bay is faster than the outflow velocity, causing possible oil spill pollutants to accumulate on the top of the bay and not spread easily, resulting in the above-mentioned high-risk area.

In summer (June to August, Fig. 6(b)), the overall risk of oil spills from pipelines is relatively higher than that in other months. They are mainly distributed in the coastal waters of the northwestern part of the dense offshore pipelines. The areas with a higher oil spill risk are Liaodong Bay, northern Bohai Bay, and western coastal waters. The highest value is 10.75, which is found in the sea area west of the Liaohe Estuary at the bottom of Liaodong Bay.

Although the total area of risky sea areas in summer has been reduced, the average risk value has increased significantly. Under the influence of the southerly monsoon, it flows northward from the east coast of the Liaodong Bay to southward on the west bank. Pollution may be transported from the east bank to the west bank along with the flow. Therefore, the pollution source may drift westward. According to previous research results, in the coastal waters of Qinhuangdao and Huludao, the two currents converge, and their circulation structure is different in summer from that in spring and autumn and all the way northward into Liaodong Bay. This condition results in a high-value area of pipeline oil spill risk in the coastal waters of Qinhuangdao in summer.

In autumn (September to November, Fig. 6(c)), the overall risk of oil spills from pipelines is relatively lower than that in other months. They are mainly distributed near dense areas of pipelines and in the southeastern coastal waters. The areas with higher oil spill risk are the eastern part of Liaodong Bay, the southeast of Bohai Bay, near the mouth of the Yellow River, and the eastern waters of Laizhou Bay. The highest value obtained is 6.34, which is found in the sea area north of Dalian in the east of Liaodong Bay.

In autumn, the circulation pattern at the top of Liaodong Bay is roughly the same as that in spring, but the average velocity of the entire Bohai Sea, the velocity on the west coast of the Bohai Sea, and the velocity of the northern Liaodong Bay (Jinzhou sea area) are larger than those in spring and summer, and the coastal waters of the northern Liaodong Bay (Panjin, Yingkou). The scope and intensity of pollution are smaller than in summer. The rainy season of the Yellow River is from June to September, and the runoff is large, especially in autumn. When the runoff of the Yellow River increases, the flow in the waters near its mouth also increases. After reaching the middle of Laizhou Bay, the flow turns westward, forming a counterclockwise flow near the mouth of the Yellow River. The ring causes pollutants to accumulate here. As a result, in summer and autumn, pollutants accumulate in the west of Laizhou Bay, resulting in a high inorganic nitrogen content concentration in the coastal waters of the west of Laizhou Bay.

In winter (December to February of the following year, Fig. 6(d)), pipeline oil spills have the highest overall risk value, which is mainly distributed in the coastal waters southeast of dense pipelines. The areas with higher oil spill risk are near the mouth of the Yellow River, the eastern part of Liaodong Bay, the north-central part, the southern part of Bohai Bay, and the waters southeast of Laizhou Bay. The highest value obtained is 15.43, which is found in the waters north of the mouth of the Yellow River.

At this time, the remnant of the Yellow Sea warm current enters the Bohai Sea through the northern part of the Bohai Strait and travels west to the west bank of the Bohai Sea. It is divided into two branches – the south and the north – by the coast. Then it goes south along Liaodong Bay, forming a clockwise circulation in Liaodong Bay. The south branch enters Bohai Bay and goes south along the coast. After passing through Laizhou Bay, it flows out of the Bohai Sea through the Bohai Strait, forming a counterclockwise large current loop. In addition, the decrease in runoff at the mouth of the Yellow River in winter, coupled with the influence of topography, reduces the area of oil spill risk. However, the high-risk areas are concentrated in the northern waters of the mouth of the Yellow River.

The oil spill risk of pipelines in different waters of the Bohai Sea basically depends on the circulation pattern. Moreover, the temporal and spatial evolution trends of the four seasons in spring, summer, autumn, and winter are basically consistent with the previous research results on the Bohai circulation. The findings of many experts and scholars, such as Wei et al. (2001), on the circulation of the Bohai Sea confirm the above analysis. The east coast of Liaodong Bay and the east coast of Laizhou Bay can quickly exchange with the open sea, and pollutants do not accumulate easily. In Bohai Bay, the southerly wind blows in summer, and the seawater gathers toward the northwest corner of Bohai Bay. After the autumn wind turns, Bohai Bay overflows. The oil risk is reduced as a whole, and the water exchange capacity in the western part of Liaodong Bay and the north side of the mouth of the Yellow River is weak, causing pollutants to bypass and accumulate again.

The four-quarter oil spill risk diffusion field is superimposed to obtain the annual oil spill risk distribution of the Bohai oil pipeline (Fig. 7). The high-risk areas are mainly located near the mouth of the Yellow River and the north, the bottom of Liaodong Bay, and the northern part of Bohai Bay near Tangshan and Tianjin.

4.2 Pipeline Oil Spill Risk Level

The oil spill risk rating table of the Bohai offshore pipeline is shown in Table 1.

The calculated oil spill risk scores of the Bohai Sea in three grades of spill sizes are shown in Fig. 6. Fig. 8 shows that, in the Bohai Sea (about 78000 km²), the first-level risk area (the lowest risk area) for oil spills from offshore pipelines is 52702 km², accounting for approximately 67.57%. The second-level risk area is 19254 km², accounting for approximately 24.68%. The third-level risk area is 5421 km², accounting for about 6.95%. The fourth-level risk area is 611 km², accounting for about 0.78%. The fifth-level risk area is 12 km², accounting for only 0.015%. However, once oil spills in this area, the hazard is extremely serious.

The results of the oil spill risk assessment indicate that the high-value areas of the oil spill risk diffusion field of the Bohai Bay offshore pipeline are mainly distributed in the coastal areas of the estuary. Specifically, four major areas of medium and high risk exist, which are the bottom of Liaodong Bay, the bottom of Bohai Bay, near the Tangshan area, and the northern part of the mouth of the Yellow River. The northern waters of the Yellow River Estuary, in particular, have a high-value area where the oil spill risk spreads and the oil spill risk level is relatively high, thus requiring key protection.

4.3 Discussions

The historical oil spill situation is combined in this paper. After the qualitative score is calculated, it is combined with DNV's OSRD model to calculate the oil spill volume and oil spill probability combination of the risk source. The wind field data are obtained by using the QuickSCAT satellite wind field to process the monthly average climate. The boundary forces of the model are the tide level data of the open boundary. The results from the large-area (the Bohai Sea, Yellow Sea, and East China Sea) model are then output. FVCOM calculates the tidal current and tidal residual current. The wind force is not added to the calculation of the hydrodynamic model. Therefore, in Eq. (3), u₁, u₂, u₃, and u₄ represent the tidal current, residual current, wind-ocean current, and wind drift, respectively.

5 Conclusions

Wind is the major driver for the ocean current and the oil distribution, yet tidal action, including tidal currents and residual currents, also has a great influence on oil film distribution. An actual oil spill accident can be simulated based on the actual occurrence time, the actual amount of oil spilled, and the actual wind field. Several sets of simulation conditions can be set when some elements are uncertain. The total amount of computation is limited.

Calculating risk is a different process. Theoretically, each element condition is uncertain; thus, simulation calculation cannot be performed. To ensure the feasibility of the calculation, a certain degree of simplification is required; for example, we divided the period into 12 months for simulation. The monthly average wind field in the climatic state is used for calculation; we applied this method to reduce the amount of calculation. This study is also simplified in other aspects. The risk sources of oil spills are consolidated into 455 point sources, and the oil spill volume is divided into three grades. The total simulation times reached 16380. Through these simplifications, considering the oil spill risk calculation of physical oceanography is feasible.

This study has produced an oil spill risk map for pipelines in the Bohai Sea, which shows the distribution of different risks in the Bohai Sea. Through the combination of the oil & gas producers' data and the OSRD model with the historical oil spill accidents in China's coastal waters, the oil spill risk model of oil pipelines is established. Then, the whole area is divided into grids, and the frequency of oil spill risk for each risk source grid is calculated. Afterward, the calculation methods of oil spill risk assessment are studied and probed, and a relevant model and formula are established. Three risk score contour maps are produced for each typical spill size ranking by the presented model. The offshore pipeline oil spill risk map of the Bohai Sea is thus completed. This risk map indicates that the high-risk oil spill areas of the Bohai Sea oil pipeline are mainly distributed in the northern waters of the mouth of the Yellow River, the waters around Caofeidian, and the central and northern waters of Liaodong Bay. The produced risk maps are of practical and direct benefit to operators of the oil pipeline, marine administrators, and environmental agencies. This study aims to inform policy regarding resource allocation for oil spill prevention, preparedness, and research response activities. At present, the marine oil spill risk model is still in its preliminary stage, and many aspects can be further improved. For example, additional risk indicators can be introduced in the risk source analysis for the calculation, and more in-depth calculations of the spread of risk sources based on the dynamic marine environment can be performed. This research direction is our next goal. On the basis of this risk model, we will conduct oil spill risk research on different sources of oil spill risk in different regions.

Acknowledgements

The study is supported by the Special Funds for Fundamental Scientific Research Operation of Central Universities (No. 202113011), the Guangxi Key Laboratory of Marine Environmental Science, Guangxi Academy of Sciences (No. GXKLHY21-04), the Shandong Provincial Social Science Planning Research Youth Project (No. 21DSHJ2), the General Project of National Social Science Fund for Research on the Ideological and Political Courses in Colleges and Universities (No. 21VSZ102), and the Ministry of Natural Resources Departmental Budget Project 'Research on the Policy and Operation System of the Control System for Land and Space Use' (No. 121107000000190014).