Research on the Technical Factors of Risk Prevention and Control in

Smart Ports and Supply Chain Resilience

Yuxuan Huang

School of Mathematics and Physics, Department of Applied Mathematics, Xi’an Jiaotong-Liverpool University,

Qianqiao Sub-district, Wuxi, China

Keywords: Smart Terminal, Supply Chain Resilience, Digital Twin Technology.

Abstract: With the increasing complexity of global trade and logistics, ports face growing pressure to improve

operational efficiency, reduce delays, and enhance safety. Digital twin technology emerges as a transformative

solution to address these challenges by enabling real-time simulation, predictive analytics, and dynamic

decision-making. In the process of digitalization and intelligentization of supply chains, smart ports have

emerged as a typical outcome of this evolution.The rise of Industry 4.0 and the Internet of Things (IoT) has

accelerated the digital transformation of port operations. As key nodes in global supply chains, smart ports

leverage advanced technologies like digital twins to optimize resource allocation, automate workflows, and

strengthen resilience against disruptions. This study investigates the effectiveness of digital twin technology

in enhancing port visibility, flexibility, and agility. By examining practical applications in Rotterdam Port,

Qingdao Port, and Shenzhen Yantian Port respectively, the research summarizes its advantages and

limitations. The findings demonstrate its application value in mitigating port accident risks. Over the next

decade, digital twin technology is expected to transcend its current role as a "technical tool" and evolve into

the intelligent hub of complex port systems.

1 INTRODUCTION

Supply chains constitute the core network connecting

production and consumption, whose significance far

exceeds simple logistics and transportation. In recent

years, with the advancement of globalization, supply

chains have increasingly played a pivotal role in

global economic operations. However, under volatile

international situations, their technical architectures

and operational models have exposed significant

systemic vulnerabilities, posing substantial

challenges to supply chain resilience. Concurrently,

smart ports, as products of the digitalization and

intelligentization phase in supply chains, demonstrate

concrete applications. For instance, Yangshan Phase

IV Port interconnects 128 intelligent devices through

a 5G private network, achieving an annual throughput

of 6.3 million TEU (Li and Zhang, 2022). The CTN

system at Hamburg Port utilizes machine learning to

predict vessel arrival times with over 95% accuracy,

resulting in an 18% improvement in berth utilization

and annual fuel savings exceeding 8 million euros

https://orcid.org/0009-0009-4811-5230

(Schulz and Bergmann, 2023). This paper focuses on

the technical factors of risk prevention and control in

smart ports, selecting digital twin technology as a

representative case for in-depth analysis. Through

systematic investigation of its functional

implementation mechanisms across three dimensions

- visibility, flexibility, and agility - this research

reveals how digital twins empower smart ports with

dynamic risk mitigation capabilities in complex

operational environments.

2 CURRENT TECHNICAL RISK

LANDSCAPE IN SMART

PORTS

Singapore Port pioneered the global implementation

of the first container terminal management system in

1990, reducing vessel port stay time by 30% (Lee and

Tongzon, 1994). Since then, smart port development

has evolved through four phases: Mechanization and

Huang, Y.

Research on the Technical Factors of Risk Prevention and Control in Smart Ports and Supply Chain Resilience.

DOI: 10.5220/0014351500004718

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 2nd International Conference on Engineering Management, Information Technology and Intelligence (EMITI 2025), pages 289-294

ISBN: 978-989-758-792-4

289

Informatization Foundation (1980s-2000s),

Automation Transformation (2000s-2010s), Digital-

Intelligent Integration (2010s-2020s), and Smart

Resilience Enhancement (2020s-present).

Contemporary smart ports face three critical technical

risks.

2.1 Data Interface Fragmentation

Rotterdam Port's initial digital transformation (2016-

2020) encountered 27 isolated systems with

incompatible interface protocols, resulting in a 6.3%

container status misreporting rate due to data format

conflicts (Van and Langen, 2021).

2.2 Algorithmic Rigidity

Busan Port's fourth-generation automated terminal

(2022) revealed inflexible AGV path planning during

typhoon events. Overreliance on static digital maps

without real-time integration of water level sensors

caused 12 AGVs to stall in flooded areas (Korea

Maritime Institute, 2023).

2.3 Decision-Making Latency

Antwerp Port's 2022 chemical spill incident exposed

bureaucratic inefficiencies: 127-minute approval

process across three hierarchical levels (safety

inspector → operations manager → port director). 18-

minute delay from sensor alarm to emergency team

activation. 43-minute fireboat deployment lag due to

paper-based chart coordination, expanding

contamination area by 37% (European Maritime

Safety Agency, 2023).

While IoT, AI, and automation technologies

enable port intelligence, escalating system

complexity amplifies associated technical risks.

3 APPLICATION AND

DEVELOPMENT OF DIGITAL

TWIN TECHNOLOGY

Digital twin refers to a virtual dynamic mapping

model of physical entities (equipment, systems, or

processes), enabling real-time data interaction

through IoT, big data, and AI technologies to support

simulation, prediction, and optimization.

The evolutionary trajectory spans Conceptual

Emergence Period (1960s-2000s): NASA's Apollo

program developed the "Mirroring System" for

spacecraft state simulation, recognized as the

prototype of digital twin concepts, Technological

Incubation Period (2010-2015): General Electric

pioneered commercial applications in aviation engine

health management, achieving predictive fault

diagnosis (Banerjee, 2016), Application Explosion

Period (2016-2020): Cross-industry proliferation

across manufacturing, energy, and urban

infrastructure sectors, Intelligent Convergence Period

(2021-Present): Deep integration with edge

computing, blockchain, and autonomous decision-

making systems.

In 2023, the Maritime and Port Authority of

Singapore (MPA), in collaboration with IBM,

developed a digital twin platform for Singapore Port.

This system integrates real-time data from over

100,000 IoT sensors with a refresh rate under 500ms.

Its 3D port modeling achieves Level of Detail 500

precision, enabling multi-dimensional simulations of

wind speed, tidal patterns, and cargo flows.

Leveraging LSTM neural networks, the platform

elevates quay crane gearbox fault prediction accuracy

to 89%, reducing maintenance costs by 22%. For

emergency scenarios like chemical spills, evacuation

path optimization algorithms decrease response time

by 37% through digital rehearsals.

The technology has been demonstrated to possess

the capability of collecting physical entity data in real

time through sensors, cameras, and other such

devices. It is then able to construct high-fidelity

virtual models based on Building Information

Modelling (BIM), Computer Aided Design (CAD),

and Finite Element Analysis (FEA), to name but a

few. Furthermore, it is capable of supporting dynamic

decision-making through the utilisation of machine

learning and physical simulation. In addition, it is

able to control commands to reverse physical entities

in the Augmented Reality (AR) and Virtual Reality

(VR) interface to carry out algorithmic analyses and

interactive feedback. It is therefore evident that the

technology can serve as a key optimisation and

improvement solution for dock visibility, flexibility,

and agility (Tao,2018).

As Europe's largest port and a global smart port

benchmark with annual throughput exceeding 15

million TEUs, Rotterdam historically faced

challenges including system fragmentation, delayed

dynamic responses, and inadequate environmental

adaptability. The 2021-launched PortXcel digital

twin project, integrating IoT, blockchain, and AI

technologies, established the world's first full-

logistics-chain digital twin platform, achieving triple

breakthroughs in supply chain visibility, efficiency,

and resilience.

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence

290

The PortXcel platform has been deployed with

over 85,000 industrial-grade sensors at the equipment

layer, including shorebridge vibration monitoring,

AGV battery health diagnosis and temperature and

humidity tracking of refrigerated containers. At the

environmental layer, the platform integrates

meteorological satellites, tides and laser radars to

provide real-time monitoring of wind speed, tide

height and visibility, as well as 26 other

environmental parameters. At the business layer, the

platform connects to the dispatch systems of 72

shipping companies and over 200 logistics service

providers via EDI and API, processing over 4,000

logistics events per second. These three layers of data

sensing form the platform's core.

The data was processed by a neural network

computing node (NVIDIA Jetson) prior to being

uploaded to a cloud-based digital twin engine, which

generated a port operations heat map. For instance,

during the 2023 typhoon season, the platform utilised

a combination of meteorological forecasts and

automated steering for the real-time adjustment of the

container stacking area, thereby reducing the average

waiting time for vessels from 18.7 hours to 14.9

hours. This resulted in a reduction of approximately

€1.2 million in the cost of fuel for each vessel.

Moreover, the Rotterdam Port Authority's

financial report for 2024 indicates a decline in the

turnover rate of the port yard by 18%, accompanied

by a decrease in ship demurrage of 23 million euros

per annum. The implementation of this technology

has enabled cargo owners to ascertain the location of

their cargoes in real time, with an ETA (error <15

minutes), thereby enhancing the customer satisfaction

index from 7. The International Association of Ports

and Harbours (IAPH) has awarded the Port of

Rotterdam a rating of 2 to 8.9 (out of 10) in the

category of 'digital twin practice'. This prestigious

accolade has attracted significant investment from

major shipping entities such as Maersk and Duffy,

with an additional €750 million pledged to the port.

In the context of the increasing fluctuations that

characterise contemporary globalised supply chains,

it is imperative that ports implement a closed-loop

intelligence system that encompasses the three phases

of 'perception-decision-execution'. The advent of

closed-loop 'simulation-optimisation' technology,

underpinned by digital twins and dynamic strategy

tuning driven by real-time data, has emerged as a

pivotal factor in overcoming the rigidity inherent in

conventional port operations.

The present study employs the Qingdao Port as a

case study to demonstrate the enhancement of multi-

scenario adaptability and anti-disturbance capability

through the implementation of the proposed

technology.

closed-loop system is comprised of three core

modules. Firstly, the real-time sensing module

constructs the physical entities of the terminal by

integrating data from AGV, LIDAR, ship AIS, yard

camera, etc. Secondly, the simulation derivation

module constructs the discrete event simulation

model based on Anylogic or FlexSim. Thirdly, the

dynamic optimisation module generates the Pareto-

optimal solution set by adopting deep reinforcement

learning and genetic algorithms (Wang,2023).

The berthing sequence is dynamically adjusted by

a fuzzy comprehensive evaluation model,

incorporating factors such as schedule urgency, cargo

value, and tidal window. During the 2022 typhoon in

Qingdao port, the model reduced the average waiting

time of 100,000-tonne container ships by 42%. The

specific operation entails the input of the typhoon's

trajectory, wave height, and gust number, followed by

modelling. Subsequently, the twins will evaluate the

three alternative loading and unloading schemes:

early berthing, bollard reinforcement, and transfer to

inland waterways. The twins will then recommend a

combination strategy of 'phased berthing + locking on

the quay bridge'. Consequently, the actual operation

recovery time was reduced from the estimated 36

hours to 11 hours, thereby reducing the economic loss

by more than 230 million RMB.

The integration of digital twin technology with

edge computing, 5G communication and artificial

intelligence is effecting a transformation in the real-

time responsiveness and decision-making efficiency

of port operations. The port's hybrid architecture,

integrating localised simulation and cloud-based co-

optimisation, enables a millisecond-level closed loop

of "perception-decision-execution" in complex

dynamic scenarios. In terms of technical architecture,

the system integrates the NVIDIA Jetson Xavier and

other edge computing units into AGVs, shore bridges

and other equipment. These units possess an

arithmetic power of 32 TOPS, facilitating TensorRT-

accelerated real-time reasoning and enabling the

deployment of edge nodes. The system employs a

URLLC private network, characterised by an end-to-

end latency of less than 10 milliseconds and jitter

control of ±0.The system operates with a processing

speed of 5 milliseconds and utilises local processing

for high-frequency data, such as LIDAR point clouds,

and low-frequency data, including business logs. The

hierarchical processing of data is completed by

uploading low-frequency data, such as business logs,

to the cloud.

Research on the Technical Factors of Risk Prevention and Control in Smart Ports and Supply Chain Resilience

291

The present paper adopts the automatic guided

trolley in Shenzhen Yantian Port as a case study to

analyse the utility of digital twin technology in

agility. The automated guided trolley has been

equipped with the Velodyne VLS-128 LiDAR, with

the edge nodes running the YOLOv7 model for the

identification of obstacles. The digital twin model

generates avoidance paths based on reinforcement

learning, and the delay in instruction issuance has

been reduced to 50 milliseconds from 1.2 seconds in

the traditional cloud. The collision rate has been

reduced to 0.03 collisions per 10,000 TEUs in 2023,

and the time consumed for accident response has been

shortened by 91%.

Furthermore, the capacity for intelligent decision-

making to undergo continuous evolution is facilitated

by the AI training closed loop. The system is capable

of simulating 12 different types of disturbance

scenarios, including typhoons and equipment

failures, and is capable of generating 18,000 sets of

training samples per second. The system utilises a

multi-intelligence body depth deterministic strategy

gradient to train automatic guided trolley cluster

cooperative obstacle avoidance and path planning

strategies. In 2024, Yantian Port successfully

completed 200 AGV path replanning in 30 seconds,

exhibiting a mere 11% increase in operation delay

during sudden heavy rainfall.

Despite the potential of digital twin technology to

enhance operational efficiency in ports, its

implementation is hindered by numerous obstacles,

particularly for small and medium-sized ports that

lack sufficient capital and technological

accumulation, thereby establishing a substantial

barrier to adoption. It is imperative that ports integrate

IoT sensors, traditional management systems and

external data. However, the differences in interfaces

between heterogeneous systems result in data

cleansing accounting for more than 40% of the

project cycle. It is an acknowledged problem that

small and medium-sized ports often face the issue of

information silos becoming fragmented due to a

paucity of standardised data architecture. The

establishment of a digital twin platform necessitates

considerable investment in hardware, software and

human capital. For a medium-sized European port,

the initial investment is approximately 12 million

euros, and the annual maintenance cost in the

subsequent stage accounts for 15-20% of the total

investment. For ports with an annual throughput of

less than 1 million TEU, the payback period may

exceed five years. The construction of models

demands a multifaceted, interdisciplinary expertise,

yet a mere 12% of global ports have a dedicated

digital twin team. It is evident that smaller and

medium-sized ports frequently depend on external

suppliers, which consequently engenders diminished

technological autonomy and delayed upgrades

(Deloitte, 2023).

The following recommendations are thus proposed by

the present study.

The optimisation of technical architecture is

achieved by the establishment of a twin system

characterised by layered synergy.

The strategic deployment of edge computing

nodes at critical infrastructure points, such as

automatic guided trolleys and shore bridges, is

paramount for the effective execution of real-time

operating systems. These systems must be configured

to handle high-frequency, low-latency localisation

tasks, including LIDAR and vibration sensor

monitoring, to ensure seamless operational

efficiency. In order to facilitate instantaneous

obstacle-avoidance decision-making and fault early

warning, and to achieve edge-layer deployment, non-

real-time data is uploaded to the cloud via 5G URLLC

slicing network. The training of the global

optimisation model and the generation of a dock-level

resource scheduling strategy to enable cloud

collaboration is performed by GPU clusters.

IThe following paper will set out a cost control

strategy. The promotion of inclusive technology is of

paramount importance.

In order to overcome the financial and technical

challenges faced by small and medium-sized ports, it

is essential to promote the adoption of 'on-demand

subscription, lightweight deployment' of Software as

a Service (SaaS) solutions. Vendors such as

Microsoft Azure and Aliyun are encouraged to launch

modular digital twin platforms, providing plug-and-

play function packages such as AGV scheduling and

yard simulation. Ports have the option to subscribe to

it at a graded level according to throughput, thereby

eliminating the procurement and operation and

maintenance costs of servers that can easily amount

to millions of euros (Gartner, 2023).

The training of talent and the implementation of

organisational change.

The establishment of a 'Digital Transformation

Office' within the port was a strategic initiative aimed

at facilitating the coordination of several key

processes. These processes encompassed the

planning of technical routes, the promotion of

interdepartmental collaboration, and the

implementation of a KPI assessment system. The

department has established a data governance group,

a model development group and a change

management group with the objective of innovating

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence

292

the organisational structure. Furthermore, it has

collaborated with universities to establish a micro-

profession of 'Port Digital Twin Engineer'. The scope

of this micro-profession encompasses the

development of IoT sensors, a Python data science

toolchain and optimisation algorithms of operations

research. The department has also introduced a

virtual-realistic and real-life practical training

platform to enhance the practical capabilities of

students. This initiative is designed to promote the

integration of industry and education. It is imperative

to promote the integration of industry and education.

IV. The present paper sets out the argument for policy

and ecological synergy.

The government has allocated funds to support the

digital transformation of ports. This financial

assistance encompasses the procurement of edge

computing equipment and the training of models.

Furthermore, the government has stipulated that

funded ports must open source a portion of their non-

core code, and has granted tax reductions to pilot

ports. Additionally, a fund has been established to

facilitate infrastructure development. The

government has also initiated a collaborative

laboratory between ports, technology enterprises, and

universities, with a focus on technological

advancements in digital threading and causal

reasoning. Finally, the government has mandated the

implementation of standardised data interfaces for

devices such as quay bridges, AGVs, and other

equipment. Finally, it is imperative that shore bridges,

AGVs and associated equipment are equipped with

standardised data interfaces.

4 CONCLUSIONS

The findings of the research indicate that digital twin

technology has demonstrated its fundamental value in

high complexity industrial scenarios through the use

of virtual-reality mapping, real-time interaction and

dynamic optimisation. In the port sector, the

implementation of a digital twin facilitates

millisecond-scale dynamic scheduling of AGV path

planning and yard space allocation through the

integration of edge computing and 5G

communication technologies. The Port of

Rotterdam's implementation of this technology has

yielded notable results, as evidenced by a 20%

reduction in ship demurrage time, a 18% decrease in

yard turnover, and a more than 23 million euro

reduction in annual operational costs. These

observations signify a substantial enhancement in

operational efficiency and enhanced visibility. In

addition, the twin system is capable of simulating

disturbance scenarios, such as typhoons and

equipment failures, by means of a multi-source fusion

of meteorological, equipment and operational data.

Furthermore, it is able to generate emergency

response strategies in advance. For instance, the Port

of Qingdao succeeded in reducing the operation

recovery time from 36 hours to 11 hours during the

2022 typhoon. This was achieved through simulation

and deduction, resulting in a 76% reduction in the risk

of equipment damage. The port's enhanced risk

resilience and flexibility are evident in these findings.

The integration of AI-driven predictive maintenance

and natural language interaction has been

demonstrated to significantly reduce the necessity for

human intervention, thereby facilitating a paradigm

shift from an 'experience-driven' to a 'data-driven'

model. This transition is evident in the enhanced

decision-making intelligence and optimised supply

chain agility. However, the implementation of the

technology still faces challenges such as the difficulty

of breaking down data silos, the cost pressure on

small and medium-sized ports, and the shortage of

interdisciplinary talents. The realisation of its value

requires a four-dimensional synergy of technological

innovation, cost reconstruction, organisational

adaptation and policy escort. The modular

architecture serves to reduce the threshold,

standardisation promotes the scale effect, and

ecological cooperation accelerates innovation. The

findings of this study demonstrate that the

implementation of digital twin technology has

precipitated significant advancements in academic

theories and the reconstruction of industry standards.

This, in turn, has resulted in the convergence of

traditional port engineering with computer science

and operations research, giving rise to novel research

directions such as 'port information physical systems'.

Consequently, this has led to the realisation of dual

innovation in both academic and industrial research.

In the next decade, digital twins will evolve beyond

the confines of 'technical tools' and transform into the

intelligent nexus of port complex systems. It is

imperative that the research underpinning this field

integrates the insights of multiple disciplines,

including but not limited to quantum physics,

neuroscience and sociology. Furthermore, the

establishment of a comprehensive 'technology-ethics-

policy' trinity governance framework is essential.

Subsequent research can be conducted with a focus

on the aforementioned aspects, ensuring a thorough

and comprehensive exploration of the subject.

Research on the Technical Factors of Risk Prevention and Control in Smart Ports and Supply Chain Resilience

293

REFERENCES

Banerjee, A., et al. (2016). A digital twin approach to

predictive maintenance of gas turbine engines. ASME

Turbo Expo 2016: Turbomachinery Technical

Conference and Exposition.

Deloitte. (2023). Digital twin talent shortage in global ports

(Industry Survey Report).

European Maritime Safety Agency. (2023). Chemical spill

response analysis at Port of Antwerp: Lessons for

automated decision systems (Report No.

EMSA/2023/ANT-006). Brussels: EMSA Publications.

Gartner. (2023). Market guide for digital twin as a service

in transportation.

Korea Maritime Institute. (2023). Post-disaster analysis of

Busan automated terminal during Typhoon Hinnamnor

(Report No. 2023-AMI-005). Busan: KMI Press.

Lee, S. Y., & Tongzon, J. L. (1994). Automation and

productivity in ports: A case study of Singapore.

Maritime Policy & Management, 21(2), 141–154.

Li, W., & Zhang, Q. (2022). 5G-enabled automation in

global container terminals: A case study of Yangshan

Phase IV. Journal of Ports & Shipping, 15(4), 45-60.

Schulz, A., & Bergmann, T. (2023). Machine learning for

berth allocation optimization: The case of Hamburg

Port's CTN system. Maritime Economics & Logistics,

25(2), 287-305.

Tao, F., et al. (2018). Digital twin-driven product design,

manufacturing and service with big data. International

Journal of Advanced Manufacturing Technology, 94(9-

12), 3563-3576.

Van, den, B., R., & Langen, P. W. (2021). Port community

system integration: Lessons from Rotterdam's Pronto

platform. Transportation Research Part E: Logistics and

Transportation Review, 153, 102456.

Wang, Y., et al. (2023). Digital twin-driven closed-loop

optimization for port operations: A reinforcement

learning approach. IEEE Transactions on Intelligent

Transportation Systems, 24(6), 6015-6029.

EMITI 2025 - International Conference on Engineering Management, Information Technology and Intelligence

294