Integration and development of energy and information network in the Pan-Arctic region

1 Introduction

The construction of the Global Energy Interconnection scheme has been proposed by China in 2010 based on the successful breakthrough of ultrahigh voltage (UHV) technology.It provides a scientific and reasonable plan for promoting world energy transformation and sustainable development.To deal with the problems of economic adjustment,insufficient growth momentum,and the weak economy,the global energy interconnection scheme constitutes a new approach used for the promotion of the transformation of the world economy based on the integration of energy,information,and transportation.The new pattern of network integration will promote the transformation and the upgrade of the world economy and achieve the development goals of electrification,intelligence,globalization,and humanization.

The important goal of the global energy interconnection involves the construction of a backbone network formed by UHV grids around the world that will mainly consume clean energy [1].The Global Energy Interconnection Development Cooperation Organization proposed in 2018 “nine horizontal and nine vertical backbone networks” for global energy inteconnection [2].These networks connect the main grids in countries around the world,thus forming a new pattern of global power allocation across time zones and seasons.

Among these backbone networks,the Pan-Arctic region network is very important for global energy interconnection.First,the Pan-Arctic region is rich in wind energy resources,and the circulation is almost constant.The equivalent utilization time of wind energy ranges between 4000-4500 h per year.The theoretical and technical annual power generation capacities in this region respectively account for 18.5% and 17.2% of the global wind energy resources.The Arctic area is an ideal power base for the global energy interconnection scheme [3].Secondly,the Arctic energy interconnection channel has the largest time span and the longest route.It starts from Norway in Northern Europe,traverses Russia,and crosses the Bering Strait to connect to Alaska.Its length is approximately 12,000 km,and spans 19 time zones.Once successfully built,the Arctic Energy backbone network will connect 80% of the power systems in the Northern hemisphere.Finally,the low temperatures in the Arctic region make it an ideal location for building green energy-saving data centers [4-6] that operate on the basis of integrating energy and information networks.

Given the unique geographical location and climatic conditions of the Pan-Arctic region,this study introduces and evaluates a possible plan of coordinated development of fiber-optic and power networks in this region.We analyze the advantages of this plan based on the traffic flow prediction and on the analysis of global data centers and networks.The results presented in this study can support the decision making of developing the global energy interconnection scheme.

The rest of this study is organized as follows:Section 1 introduces the development of global data centers.Section 2 presents the analysis and prediction of global communication traffics.Section 3 describes the framework and advantages of building integrated power and information networks in the Pan-Arctic region,and Section 4 outlines the conclusions of this study.

2 Developmental analyses of global data centers

2.1 Increasing energy consumptions of data centers

A data center gathers a large number of information technology (IT),power distribution,and refrigeration and air-conditioning equipment.It is responsible for performing data storage,management,information processing and exchanging.To ensure that the IT equipment in a good status,data centers require cooling systems.Cooling systems run all year round,and have costs which amount to a large proportion of the total electricity cost of the data center [7,8].In a typical data center,IT equipment converts more than 99% of its electrical energy into heat.To protect the data center’s operating environment,70% of the heat needs to be removed from the data center.As the number of server devices grows at a rate of more than 10% per year,data storage requirements are growing at an average annual rate of more than 50%,which ultimately leads to an overall increase of the energy consumptions of the data centers.

The energy consumption of data centers can be classified in two parts:IT energy consumption and infrastructure energy consumption.The IT equipment system includes servers,storage devices,and network equipment; the infrastructure includes uninterrupted power supplies (UPS),air conditioning and refrigeration,and auxiliary lighting systems.The energy consumptions of all these parts are plotted in Fig.1.Generally,the energy consumptions of the IT equipment and infrastructure are equivalent.Airconditioning systems are energy-intensive,and account for approximately 37%.Of these,air-conditioning and refrigeration systems account for approximately 25%,while the air supply systems account for the remaining 12%.

Fig.1 Energy consumption components of data centers

In 2010,the energy consumption of data centers accounted for 1.1-1.5% of the total global electricity consumption.Energy consumption of data centers in the United States accounts for 1.7-2.2% of the total national electricity consumption [9,10].In the future,data centers will become the world’s largest consumer of energy,and the energy consumption—as a percentage of total global electricity consumption—will increase by 4.5% in 2025 [11].The global data center used 325 billion kWh of electricity in 2010.With the rapid development of the internet,many data centers have been built.In 2017,global data centers used 560 billion kWh of electricity.It is estimated that by 2020 and 2030,electricity consumption of data centers will reach 760 billion kWh and 1.7 trillion kWh,respectively.Air-conditioning systems in data centers will consume more than one-third of the total energy [12].

2.2 Additional data center requirements for social intelligence

With the development of the next-generation information technologies,such as the Internet of Things,big data,and artificial intelligence,human society generates increasingly more data than ever.At present,the global data volume is growing at an average rate of 40% per year.Cloud computing reduces the cost of information applications.Meanwhile,the demand for mining big data has accelerated the development of machine intelligence technology.Big data and machine intelligence have spawned an intelligent era that will lead to the rapid growth of data in global communications.The data volume of global data communication and computing was 2.8 ZB in 2011,but by 2017 it reached 10.7 ZB.It is expected that the data volume will respectively reach 19.1 ZB and 61.8 ZB by 2020 and 2030.Meanwhile,the data stored in data centers will reach almost 1 TB in 2020,and more data will be moved to data centers over time.

2.3 Growth of global network traffic

Tremendous increases have been noted in the development of new generation information technologies and global network communication traffic,especially in data center traffic.According to Cisco’s visual network index (VNI) 2016,global digital transformation will continue to have a major impact on the demand of the internet protocol (IP) network in the next five years.Internet users will grow from 3.3 billion to 4.6 billion (58% of the global population).Personal equipment and machine-to-machine (M2M) connections will be more adopted extensively,and the number is anticipated to increase from 17.1 billion in 2016 to 27.1 billion in 2021.The average broadband speed will increase significantly by 53.0 Mbps from 27.5 Mbps.Video viewing volume will also increase,while the proportion of total IP traffic will increase from 73% to 82%.Global IP traffic is expected to increase almost threefold from 1.2 ZB in 2016 to 3.3 ZB in 2021 [13].

According to the reports released by Cisco [14,15],the global data center IP traffic is expected to grow from 4.7 ZB per year (390 EB per month) to 15.3 ZB per year (1.3 ZB per month) by the end of 2020,with a compound annual growth rate (CAGR) of 27%.The network traffic generated by data centers will account for 97% of the global network traffic,including the network traffic between the different data centers and different users (as shown in Fig.2).With the rapid growth of data center traffic,data centers are increasingly demanding communication networks.As a result,the access to the backbone fiber-optic network is one of the important factors in the construction of large-scale data centers.

Fig.2 Global data traffic distribution

3 Near-Arctic region:a hot spot in data center construction

Air-conditioning refrigeration systems of data centers consume more than one-third of the energy consumption of the entire data center.With the rapid development of global intelligence,a large number of data centers have been built.It is estimated that by 2020 and 2030,the energy consumption of air conditioning systems in data centers will reach 280 billion kWh and 650 billion kWh,respectively.In 2030,the energy consumption of air-conditioning systems in data centers will be more than one-tenth of China’s electricity consumption in 2017 [12].Extensive research work has been conducted to reduce the energy consumption of data centers [16-19].

World renowned Internet companies,such as Facebook and Google,have gradually noticed the advantages of building data centers in the Arctic region.Inexhaustible clean energy,such as wind power,is abundant in the Arctic region.The cold air and ideal humidity can reduce the cost of cooling systems in data centers.If new data centers are built in the Arctic,the overall pattern of global data centers will change.Most of the existing data centers were built in well-developed regions,such as Europe and the United States.Although the response of data communication is rapid,these data centers consume a huge amount of energy.A reasonable plan involves the construction of data centers for “cold information” in the near-Arctic region and “hot information" data centers for near users.This will not only reduce the energy consumption and carbon emissions,but it will also reduce the impact on global climate change.Building data centers in the Arctic area can also change the geographical restrictions of global data communication,improve the efficiency of data routing and exchange,and enhance the security of the intelligent grid.

4 Development of fiber-optic and power networks in the Pan-Arctic region

4.1 Data center traffic constitutes the main body of future network traffic growth

To predict network traffic in the future,global data center traffic is analyzed via the regression model.The linear regression model is a commonly used predictive model with an important assumption,whereby a linear relationship exists between the response variable y and the explanatory variable x [20,21].However,the global data center traffic has increased tremendously owing to the widespread expansion of the Internet of Things,cloud computing,big data,and artificial intelligence,and has led to a weak linear relationship (or to the lack of a linear relationship altogether) between time variable x and the data center traffic y.For this reason,the regression analysis of the data center traffic in this study is extended from the linear model to polynomial regression of nonlinear models [22].Polynomial regression is a linear combination polynomial that converts a first-order feature to a higherorder feature according to

The loss function is defined as

To improve the generalization ability of the model,a regularization term was added,and the revised expression is

where η represents the regularization coefficient.The gradient descent was employed to determine the coefficients a0 ,a1 ,a2 ，… , am,and the gradient value can be solved as follows,

The gradient update rate is λ,and the gradient update formula is,

input in the above model for training,and the values of the parameters were updated to fit the samples.In this manner,the regression model for the network traffic prediction was obtained.

In the above regression model,addition of a higher-order feature (such as a square or a cubic term) was equivalent to an increase of the degrees-of-freedom of the model to capture nonlinear data changes.However,when higher-order terms are added,the complexity of the model is also increased.As the complexity of the model increases,the capacity of the model and the ability to fit the data increase,which can further reduce the training loss.However,the overfitting risk of the model is also increased.To this end,the optimal value is determined by cross-checking,as shown in Fig.3.

The training samples {( x 0 ,y0 ),( x1 ,y1 ),...,( x n ,y n)} were

Fig.3 Polynomial regression model complexity and prediction error relationship

Based on Cisco’s research reports [18,19],all the components of the global data center traffic during the period of 2012-2020 are listed in Table 1.Using the regression model of Eq.(1),the different traffic components of the global data centers were modeled and predicted from 2021 and before 2030 based on Cisco’s data in Table 1,as shown in Fig.4 and 5.

Data show that the global network traffic will reach 10 ZB or more in 2030,and it will definitely introduce great challenges to the global backbone optic fiber network.It is worth noting in Figs.4 and 5 that the network traffic will exhibit a downward trend with an onset on 2027.This is attributed to the fact that with the development of the new generation of information technology,such as artificial intelligence,big data,and the 5G network,edge computing will become more mature,and will result in a large number of calculations in the terminal,which will reduce the transmission of data flow.

Table1 Cisco global data center traffic forecast

YEAR 2012 2013 2014 2015 2016 2017 2018 2019 2020 Types (EB/Year)Data center to user 427 560 711 744 933 1164 1438 1772 2183 Between data centers 167 221 281 346 515 713 924 1141 1381 Inside the data centers 1971 2560 3123 3587 5074 6728 8391 10016 11770 Total (EB/Year)IP traffic 2565 3341 4115 4677 6522 8605 10753 12929 15334

Fig.4 Prediction of global network component traffic during the period of 2010-2030

Fig.5 Prediction of the total global network traffic during the period of 2010-2030

4.2 Advantages of developing fiber-optic networks in the Pan-Arctic region

Building fiber-optic networks in the Arctic region can provide shorter routes for the connections of Asia,Europe,and America,and will thu s result in fewer network communication delays.To take this advantage,a number of trans-Arctic fiber optic network projects have been proposed in the last few years.These projects include the Arctic Fiber project of Canada,the ROTACS project of Russia,and the Cinia project of Finland [23,24].Canada’s Arctic Fiber project will involve the construction of a fiberoptic network in the Northwest Passage of Canada,to connect Japan,Alaska,and the United Kingdom,spanning a total length of 15,000 km.The line is expected to reduce the data transmission delays between Japan and London by 24 ms.Russia’s ROTACS project plans to build a trans-Arctic submarine cable from Japan through Northern Russia to the United Kingdom,and is expected to reduce the transmission delay between Japan and the United Kingdom by 60 ms.Finland’s Cinia project proposes to establish submarine cables which will connect Europe and Asia along the Northern coast of Russia aiming to the reduction of network transmission delays.The aforementioned projects are under construction and have not been completed yet,but they have highlighted the advantages of building fiber-optic networks in the Arctic region.

In addition,the construction of fiber-optic networks in the Pan-Arctic region will cause the move of global data centers toward the North.This will further optimize the global communication network and enhance information security.At present,network resources are shifting to data centers.The traffic and architecture of the network are also changing owing to the development of data centers.The traffic of the Internet is no longer completely determined by the routing policy,but the traffic is highly influenced by the business demand mode.Moreover,the traffic between data centers has increased significantly,and the demand for network bandwidth is increasingly large.The traffic model of the Internet is changing from “data center-user” to “data center-data center.” It is expected that the annual growth rate of global communication traffic will reach 25% from 2018 to 2020.More than 97% of network traffic will be related to data centers in 2020.

Building large-scale data centers in the Arctic region will directly lead to changes in global communication routes.The architecture of global network will be flattened.This is beneficial for breaking geopolitical constraints of data communication,increasing global network bandwidth,and improving the routing and switching efficiency.Conversely,building data centers in the Arctic region is beneficial to avoid the data center monopoly by a few specific countries.Accordingly,this will improve the security and reliability of information,and will ensure the safe operation of the smart grid.

4.3 Advantages of coordinated development of fiber-optic and power networks

According to the current mature technical indicators,the cost of terrestrial cable construction mainly includes:optical cable materials (approximately 0.15 million USD/km),relay stations (45,000 US dollars/month,one per 200 km),and construction costs (approximately 45,000 USD/km,including labor costs,material costs,etc.),terminal stations (approximately 2 million USD/station,with two or more stations).The total cost for constructing a line spanning 10,000 km of terrestrial cable is approximately 68.5 million US dollars.

Submarine cable construction costs mainly include fiber optic cable materials (approximately 15,000 USD/km),repeaters (1.5 million USD/month,one installed per 100 km),installation costs (approximately 30,000 USD/km,including special equipment,labor costs,etc.),fiber optic cable terminal stations (approximately 10 million USD/station,two or more).The total cost for the construction of a submarine cable line spanning 10,000 km is approximately 485 million US dollars.

Considering the climatic conditions of the Pan-Arctic region,and based on the existing construction technology,the ambient temperature of the terrestrial optical cable should not be lower than -50 °C,and the working environment of the equipment,such as the optical cable line connectors and relays must not be lower than -10 °C.In addition,considering the labor cost,construction progress,material transportation,and other factors in cold regions,the cost of constructing terrestrial optical cables in extremely cold regions will increase significantly compared to other regions.Additionally,the assumptions for the cost of each part of the terrestrial cable in the Arctic region are as follows:

·Optical cable material cost:It is assumed that in the future,the optical cable can be used in the Pan-Arctic region at -50 °C,so the cost of this part will not increase

·Relay station cost:The construction of the relay station,including the addition of necessary insulation,will result in a significant increase in cost.Based on the analysis of the data center infrastructure costs [25],it is assumed that this part of the cost will be equal to 160% of the original

·Facility cost:Owing to poor and difficult construction conditions in cold regions,the cost of the facility will increase significantly.However,under the framework of global energy Internet,the concurrent construction of communication optical cables and other cables,and the sharing of underground pipe corridors,the cost of laying facilities can also be effectively controlled.Accordingly,the cost of laying facilities will be 140% of the original

·Terminal station cost:End stations and relay stations face the same problems.Correspondingly,it is assumed that the cost will increase to 160%

According to the above analysis,the cost of construction of 10,000 km of land optical cable in the Arctic region is approximately 88 million US dollars.

Referring to the operation and maintenance costs of the existing optical cable system,the annual operation and maintenance cost of submarine optical cable or land optical cable is approximately 5% of the total construction cost.

Of the operating costs of submarine cables,submarine line maintenance costs account for approximately 45%,land site maintenance costs account for 15%,personnel costs account for 17%,and other costs account for 23%.

Among the operating costs of terrestrial optical cables,line maintenance costs account for approximately 30%,site maintenance costs account for approximately 20%,personnel costs account for approximately 10%,and other costs account for 40%.

According to the above cost analysis,the items required for the construction of a 10,000 km optical cable in different modes are listed in Table 2 (among which the labor cost of operation and maintenance,line detection and maintenance,station detection and maintenance,and other maintenance costs are listed,and refer to annual average costs).It can be observed from Table 2 that from the perspective of construction cost,for the construction of an optical cable in the Arctic region,the costs for a length of 10000 km are 485 million US dollars and 88 million US dollars,respectively,while the construction cost of the land optical cable is only approximately 18% of the cost of the submarine optical cable.However,in practice,land use is an important factor that cannot be ignored.Therefore,most of the international communication optical cables still adopt submarine cable solutions.The main reasons are as follows:first,the construction of terrestrial optical cables involves bilateral or multilateral negotiations between countries that increase the difficulty of implementation.Secondly,building terrestrial optical cables have to take the land use costs into account.This will also greatly increase construction costs.In the end,terrestrial cables are more vulnerable to human activities than submarine cables.

Considering the framework of the development of the global energy and information networks,the problem of land use should be skillfully avoided by relying on the power network to build land communication optical cables,while reducing at the same time the construction cost.In addition,based on the framework of energy information bus,and from the perspective of operation and maintenance,the fiber-optic network shares the same corridor with the power network.The operation and maintenance cost is less than 18% of the submarine optical cable cost.This can also ensure the safety of the optical fiber network without being damaged by humans.Therefore,the combination of the fiber-optic and power networks is a feasible solution.

5 Conclusion

The integration of energy and information networks is an important direction for the development of the energy industry.The rapid development of a new generation of information technology represented by the Internet of Things,big data,and artificial intelligence,can provide effective support for the efficiency,economy,and security of the entire energy chain.In the context of the global energy interconnection,there will be a significant impact on the global data center layout.

Relying on the Arctic energy interconnection channel,the “Pan-Arctic Energy Information Bus” can be constructed to meet the needs of multiple network integrations.It can achieve power sharing across time zones,clean energy development and delivery,and data information transmission.Combining power interconnection with information interconnection can help us make full use of the channels and projects of power interconnection to benefit all countries in the world.Moreover,building global data centers in the Pan-Arctic region can effectively reduce energy consumption and carbon emissions.The construction of communication networks in the Arctic region will also greatly improve the global communication landscape and optimize data transmission routing.

Table2 Construction,operation,and maintenance cost of a 10 km optical cable system at different modes (unit:$)

Construction mode Global decentralization mode Pan-Arctic region concentration mode Submarine cable - separate construction mode Cost comparison(unit:USD 10000) Fiber-optic network and power network - combined construction Construction cost 230 million 68.50 million plus land cost 485 million 88 plus land cost 88 million Land cable - separate construction mode Submarine cable - separate construction mode Land cable - separate construction mode Annual operation and maintenance cost 11.5 million ＞ 34 2430 ＞ 44 ＜ 44 Annual operation and maintenance cost and its composition Labor cost of operation and maintenance 17% 10% 17% 10% ＜ 10%Cost of line inspection and maintenance 45% 30% 45% 30% ＜ 30%Cost of site inspection and maintenance 15% 20% 15% 20% ＜ 20%Other costs 23% 40% 23% 40% ＞ 40%

Acknowledgements

This work was supported by the Corporation Science and Technology Program of Global Energy Interconnection Group Ltd.(GEIGC-D-[2018]024),and by the National Natural Science Foundation of China (61472042,61772079).