I NEED HELP IN SUMMARIZING THE ATTACHMENT

Modeling and Simulation of the Communication

Networks in Smart grid Yizhou Dong, Ziyuan Cai, Ming Yu, and Mischa Sturer

Dept. of Electrical & Computer Engineering

FAMU-FSU College of Engineering, FL 32310, USA.

[email protected], [email protected], [email protected], [email protected]

Abstract—A reliable and secure communication network plays

a significant role in Smart grid systems, which aims at

coordinating generation, transmission, distribution, and

consumption parts in power system. The scope of our work

ranges from utility level to end consumption level. The major

difficulties in this work can be summarized as follows: 1)

Performance requirements from the viewpoint of network have

not been clearly defined; 2) Model mapping from power system

to communication networks is not straightforward. 3) Network

performance has not been well-investigated. This paper

proposes a communication network model for a typical

program of smart gird. Moreover, application requirements,

link capacity and traffic settings have been investigated.

Simulation results validate the feasibility of this model and

provide useful network performances which can satisfy both

the non-real-time and real-time application requirements.

Keywords-Smart Grid; Communication Network; Simulation;

Performance; FREEDM; IFM

I. INTRODUCTION

Smart grid becomes to an attractive dominating topic

nowadays in both research universities and industrial

organization. The traditional power communication

infrastructure cannot meet the requirements for our future

power system which the energy will not only generated by

traditional generation facilities but also produced by

distributed facilities and new energy devices. The delivery of

both energy and information must also be end-to-end and

bidirectional. Communication network should interconnect

every device of power system from electricity generation to

end-user consumption, and even more. One view need to be

point out is that, in physical layer, geographic location for

power electricity device and communication network device

can be dissimilar.

NIST published the first definition of Smart Grid in 2009

which represent smart grid standardization in North

American. The networking parts proposed by NIST

emphasize the transformation from traditional power

communication networks to Information and Communication

Technology (ICT), which indicates that both energy and

information transmission must be bidirectional for all levels.

[1] In Europe, European Technology Platform also issued

standards to define smart grid as the target architecture which

enable all users’ connection, including generators,

transmission, and consumers. Other national organizations

and industrial companies also boost the development of

smart grid by provides the recommend standards and

proposals like The German Smart Grid Standardization

Roadmap concentrate their attentions on smart grid’s ICT

infrastructure.

Note that the research progress is developed under The

Future Renewable Electric Energy Delivery and

Management System (FREEDM system) [2], which is power

distribution system motivated by the widespread use of

information network at first. NSF engineering Research

Center established the in 2008 and this system is

headquartered by NCSU and partnering ASU, FSU, FAMU,

MST, RWTH, ETH and more than thirty seven industry

companies. The vision and framework in FREEDM includes

Intelligent Energy Management (IEM), Intelligent fault

management (IFM), Solid State Transformer (SST), Fault

Isolation Device (FID), Reliable and Secure Communication

(RSC), Distributed Gird Intelligence (DGI), i.e. Our present

research goal in this FREEDM project is to evaluate a

feasible communication system and provide corresponding

performance reference by modeling and simulation.

The motivation of our work is that the communication

network model in smart grid has not been clearly

investigated. The difficulty in our work is how to map the

architecture of power system into communication system.

Moreover, the time delay performance for real-time

applications is the major technical problem in our work.

From network view point, what are the FREEDM

communication application requirements? How to implement

it as a reliable and secure communication system? What kind

of model we should use to simulate the FREEDM scenario?

What is the network performance for a typical smart grid

system? Our paper provides some recommendations and

reference for these questions.

This paper is organized as follows. Related work is

reviewed in Section Ⅱ. Scenario formulation, including network topology model, link capacity and traffic settings, is

proposed by Section Ⅲ. The simulation results are presented

in Section Ⅳ. Section Ⅴ summarize this paper.

II. RELATED WORK

The current research progress for Smart Grid (SG)

Communication Architecture can be divided into three

categories, 1) Survey and Overview for SG communication

infrastructure, 2) Application requirements and supporting

network for SG, 3) Tentative and experimental suggestion

978-1-4577-0653-0/11/$26.00 ©2011 IEEE 2658

for communication network for SG, 4) Security issues for

SG.

In the first category, the common case is to give a big

picture and official background by investigate national and

organizational documents. Such as the NIST report on Smart

Gird [3], SMB smart grid interoperability standards[4], IEC

61850-1 Communication networks and systems in

substations [5] [6].

In the second category, propose a communication

infrastructure in conceptual level and evaluate both basic

applications and advanced applications are their major works

[7] [8]. A bunch of features and characteristics of smart gird

network have been raised like data digitalization,

expandability and adaptability, Intelligence, Sustainability

and Customization [9]. Requirements for different levels in

smart grid system have been distinguished. Various

communication configuration depends on application have

been proposed such as phasor measurement units (PMUs)

[10], advanced metering infrastructure (AMI). Related

applications includes some basic application like smart meter,

monitoring control and also some advanced application like

security video surveillance, automatic distributed control

[11].

In the third category, few tentative evaluations mentioned

smart grid traffic profiles and most of them are estimated by

standards or collected from residential power user. [12]

Security issue is fairly significant for smart grid and an

increasing amount of papers focus on this field. For this

paper, security issue is not our key concern.

All in all, most of the work is still in exploratory level and

there is no verified authoritative communication

infrastructure at present. In addition, only a few works have

step into modeling and simulation part for the intergral Smart

Gird communication system.

III. SENARIO FORMULATION

In the view point of power system, smart grid can be classified into four levels: Generation, Transmission, Distribution and Consumption. Consider the range of FREEDM project. We are focusing on the power delivery from distribution substation and local utility to end consumers. In order to implement the communication network in FREEDM system, some specific features should be included in the model such like Intelligent energy management (IEM), Intelligent fault management (IFM), Distributed Renewable Energy Resource (DRER), Distributed Energy Storage Device (DESD).

It is very important to provide enough evidence to verify the feasibility and credibility of communication network topology. So the follows illustrate our investigation and research for Network Model, Link Capacity and Traffic Settings.

A. Network Topology Model

As mentioned before, our network model mainly

considerate in the scope of our project, range from Control

Center level to home network level, which is from WAN

level to LAN level in terms of network perspective. In the

background of FREEDM project, IEM and IFM are typical

future communication network devices. Due to its

importance, we promoted the node model in our topology to

illustrate their impact for communication network

performance.

Traditional CCs

(SCADA)

IFM/IEDs IFM IEM/MDMS.. ...

WAN

NAN

Substation level

Control Center level Smart CCs

(DGI)

Neighbor level IFM/RTU IEM/Relay

Home level IEM/SM IEM/SM...

Fig. 1. Network hierarchical structure

As shown in the Fig.1, we propose totally four levels of

network. Firstly, Control Center level represents the current

power grid control center like SCADA system and also

future smart grid center control. The Substation level refers

to current equipment in substation or future equipment in

FREEDM for wide area control, for instance, the IEM and

IFM are exactly in this level. The Neighbor level describes

relay devices or region equipment in a certain zone and the

Home level refers to the equipment in resident area.

The IFMs can be deployed together with RTU and

substations, or standalone in the Substation level. The IEM

can be deployed together with smart meters, the relay nodes

(i.e., the roadside data forward equipment) and the meter

data management system (MDMS) in Substation level.

From the physical distance view point, the Control

Center level is usually far away from Substation level. Thus

we can implement the transmission part between these two

levels as a WAN. However, the range for Neighbor level and

Home level is restricted in a small region. Therefore, we use

NAN to denote the Neighbor level and use LAN to

implement the simulation topology.

2659

Fig. 2. Simulation scenario topology

We provide our network topology model for OPNET

simulation in Fig.2.

In the Control Center level, Billing center represent the

center to collect and analysis the metering data from MDMS.

Control center act as a commander and he take the

responsibility to order the specified control information for

dedicated equipment. WAN1 and WAN2 refers to the routers

in the WAN between Control Center level and Substation

level. In the Substation level, we use MDMS model to

collect all the data information from Home level, which not

merely provide the accumulation function but also can

process the data from sub-layer. In addition, IFM in

Substation level is also considered. In Neighbor level, we use

a LAN model, a server model and a corresponding router to

indicate the relay for a specific region. The server model can

be regarded as IFM or IEM. Inside of the LAN model, each

workstation denotes a device in Home level.

One issue need to be explained in this topology is that we

use two-layers in Neighbor level. Due to the consideration of

the different distance and layers in Neighbor level, some of

the relay region can be positioned as upper-layer relays, and

also the others can be regarded as lower-layer relays.

Table I Node model used in simulation Name Node model Description

BillingCenter,

Controlcenter,

MDMS

ppp_wkst Workstation for

PPP link

WAN slip8_gtwy_adv Gateway model for

SLIP

Substation,

Relay

ethernet2_slip8_gtw

y_adv

Gateway model for

interconnect

Ethernet and SLIP

IFM, IEM ethernet_server_adv Ethernet server

model

LAN Eth_switched_lan_a

dv

Ethernet LAN mode

B. Link Capacity

Link capacity is a critical factor in simulation scenario

which needs to meet the requirements for both application

requirements and also the practical using.

As illustrated in network topology model, the communication networks between the Control Center level

and Substation level are WANs, such as the connections of

dedicated fibers or leased wired lines. For the practical

implementation in companies, VPN or other kinds of private

wired links is being used. Nevertheless, in order to simplify

our network topology, we only regard it as point-to-point

link model to denote this sort of lines. Similarly, we use

point-to-point link to implement the links in backbone

network for Substation level and Neighbor level. But for

Home level, we can regard it as the access part and we

usually use Ethernet link to formularize the links in LAN.

The following provide the link model and link capacity

we use in our simulation for different levels.

Table II link model used in simulation level Link model Link capacity

(Mbps)

Control Center

level

DS3 44.736

Substation level T1 1.544

Neighbor level

and Home level

10BaseT 10

C. Traffic Settings

Traffic settings for simulation is largely depend on

application requirements and project background. We

consider three kinds of main existing power grid application,

i.e., advanced metering infrastructure (AMI), substation

automation and fault information management.

Advanced metering infrastructure (AMI) is the main

application for power grid no matter it’s a traditional power

system or a future smart gird. In our scenario, we generate

traffic model from all the meters in Home level and transmit

to MDMS in Substation level. After aggregated and

processed in substation, then forward it to BillingCenter in

Control Center level. Data for metering is a large amount of

information transmitted in the network and usually this

application can be regarded as background traffic which is

mostly the basic prerequisite for network performance

problem.

Substation automation (SA), which is an application

requirement can rapidly response to real time events with

appropriate actions to avoid great damage cause by

equipment failure, power disturbance and natural accidents.

We develop traffic demands in our scenario to simulate the

control commands from control center in Control Center

level to each IFM device. Comparing to AMI, substation

automation traffics basically are small amount control

information, but the time delay requirement is more

demanding.

Fault information management application for FREEDM

request the IFM device monitor the work condition for

crucial power device and report it to upper-layer’s control

center. For every half circle of one waveform, IFM need to

sample it sixteen times and transmit these fault management

2660

information to upper control center.

Table III Traffic settings Traffic type Traffic rate

(kbps)

Uplink/d

ownlink

Proto

col

AMI(part1) 20 Uplink UDP

AMI(part2) 100 Uplink UDP

SA 5 Downlin

k

TCP

IFM 10 Uplink TCP

IV. SIMULATION RESULTS

Our main concern and purpose for this simulation is to

give primary data-type results rather than theoretical analysis

for smart grid communication network performance, thereby

provide a big picture with more detail information for

reference.

In the simulation section, we use OPNET Modeler 16.0A

PL4 to investigate the performance of FREEDM

communication network under various traffic settings and

different packet size. The parameters and values are given in

Table 1, 2 and 3for topology set up and traffic configuration.

The topology dimension for this simulation is and the network topology shown in Fig. 2.

The performance metrics we mainly concern are defined

as follows:

1) Maximum top-bottom delay for control data: The maximum time latency since a packet is transmitted

from control center to smart meter in Home level.

2) Maximum delay for metering: The maximum time latency since a packet is transmitted from smart

meter to MDMS in Substation level.

3) WAN link maximum utilization: The maximum ratio of the data rate of a certain link to the link capacity in

WAN network.

4) WAN bandwidth efficiency: The ratio of the total consumed bandwidth to the whole bandwidth in

WAN network.

5) Packet loss for control data: The ratio of the number of control packets unsuccessfully delivered to the

total number of packets send out by a source node.

6) Packet loss for metering data: The ratio of the number of metering data packets unsuccessfully

delivered to the total number of packets send out by

a smart meter.

7) End-to-End packet delay: The packet latency for a dedicate end-to-end flow traffic.

We develop three main scenarios for different

performance evaluation. In our first scenario, simulations are

conducted to examine the performance impact by adding

smart meter nodes.

Fig. 3. Maximum top-bottom delay for control data and Maximum delay

for metering.

Fig. 3 shows that the maximum delay for metering hold at zero until the numbers of nodes increase to 200 and with

that metering sharply goes up. However, the time latency for

control information sends from Control Center level to Home

level remains as zero no matter how many meter nodes we

configure. We need to make it clear that for the point with

200 meter nodes, the traffic setting for each node is 20kbps

as mentioned before, thus the data rate for uplink from

Neighbor level to Substation level roughly reach to 1Mbps

which is close to the maximum link capacity of T1 link. As a

result, delay for metering data occurs due to the increase of

queuing delay for routers. Moreover, owing to the different

priority requirement of diverse applications, control

information for substation automation undoubtedly maintains

a higher priority. Thus it is reasonable to configure UDP

protocol for metering data and TCP for control signal. After

all, even when the numbers of meter node is relatively high,

our simulation result shows that we can still keep the delay

for control information at a very small level which means our

work result can meet the requirements for power system.

Fig. 4. WAN link maximum utilization and WAN bandwidth efficiency

Fig. 4 illustrate the maximum link utilization and bandwidth efficiency for WAN by the increment of node

number. Note that the link maximum utilization here is

obtained by choose the highest utilization ratio from all the

links. The reason is that the links between Substation level

and Neighbor level undertake heaviest load in this network.

However, for the most parts of this network, bandwidth

efficiency remains to a very low level because of the

unemployment of most link resource especially for

2661

downlink.

Fig. 5. Packet loss ratio for control data and meter data

Fig.5 indicates the packet loss ratio for two main types of traffic flow in our network by increase the node number. It

can be seen that packet ratio for control information is

always stay as zero which is desirable for our application

requirement for substation automation. But for data

information transmits from smart meters, packet loss appears

when node number keeps increase.

In our second scenario, we investigate the impact of

offered load for control traffic on the corresponding time

delay. Fig.6 shows the corresponding result for this part. The

packet size in this scenario all keeps at 1500Byte and also fix

the node numbers. Two traffic classes are categorized by

different hop counts. When we increased the data rate of the

control traffics, the flow delay in the figure appears a slow

linear ascending before 500kbps. After keep increasing the

offered load, the results show a sharp nonlinear rise in

control traffic delay because the increasing load leads to the

queuing delay. The delay performance for these two traffics

is similar. This results show us the queuing is also apply to

this model and 500kbps in this scenario is the turning point.

Figure 6 queuing influence the control traffic delay

In our third scenario, we fix our meter node numbers as 50 and also the traffic data rate for uplink metering data. We

mainly focus on the relationship between delay performance

for control information and control information packet size.

Due to the security concern in smart grid, the usual way to

implement security issue in simulation is to configure the

corresponding security protocol and add relevant overhead in

packet.

Due to the hierarchy structure of network, IFM/IEM

can be deployed at Substation level and Neighbor level. We

can divide our control information from control center to

each IFM/IEM into three categories: control center to

substation, control center to relay level-1, and control center

to relay level-2. The partition here is mainly base on the hop

number, for example, the hop number for CC-sub is 3 or 4,

CC-relay1 is 4 or 5, and CC-relay2 is 5 or 6.

Fig. 7. End-to-End delay for three traffics

Fig.7 shows end-to-end delay for these three traffic

categories by varying the packet size of downlink control

packets. With the increasing packet size, delay linearly

increase and growth rate is primarily depends on the hop

number. One significant drop at the point which packet size

is 1500Byte is due to the fact that TCP protocol validates the

flow control function by adaptively change the mechanism.

Thus 1500Byte packet size for TCP in our scenario is the

maximum one. The maximum end-to-end delay show in fig.

7 is 18.275ms which is produced by a top-bottom control

traffic pass through six routers. Generally, for a 60Hz power

system, the requirement for teleprotection is close to 8ms

which is a requirement we cannot reach for an arbitrary

topology and traffic. However, this result also give us the

hint that we can limit the hop number and configure a certain

packet size, thus we can achieve the most demanding

requirement in smart gird application.

V. CONCLUSION

This paper has proposed the communication network

model for smart grid project – FREEDM, especially from

utility level to consumption level. The model provide a

validated pursuable topology compared to other theoretical

models. Moreover, this paper investigate and summary the

application requirements and corresponding traffic demands.

To our knowledge, this is the first proposal for FREEDM

communication network architecture.

The simulation results for first scenario evaluate the

performance metrics by varying the node number of smart

meters, i.e. WAN link utilization, time delay, and packet loss

ratio. It reasonably illustrate that control data send by Control

center level has minor time delay which can meet the

requirements for Substation Automation. However, delay for

2662

upload metering data is not independent of node number. In

addition, various packet sizes for security concern indeed

have influence on the delay for control information. The

second scenario shows us the queuing when we alter the

offered load for control information. The result in the third

scenario provides a useful data for researchers to achieve the

application requirements especially for high-priority control

command by configure the packet size and hop number.

REFERENCE

[1] V. Li, F. F. Wu, and Z. Jin, “Communication requirements for risk-limiting dispatch in Smart grid,” 2010 IEEE Int. Conf. on

Communication (ICC), 23-27 May. 2010, pp. 1-5. [2] A. Q. Huang, M. L. Crow, G. T. Heydt, J. P. Zheng, and S. J. Dale,

“The future renewable electric energy delivery and management

(FREEDM) system: the energy internet,” Proceedings of the IEEE,

vol. 99, no. 1, Jan. 2011, pp. 133-148. [3] NIST, “NIST special publication 1108 NIST framework and

roadmap”, 2010. [4] SMB smart grid strategic group (SG3), “IEC smart grid

standardization roadmap,” Jun. 2010 [5] S. Mohagheghi, J. Stoupis, and Z. Wang, “Communication protocols

and networks for power systems-current status and future trends,”

IEEE Power Systems Conference and Exposition (PSCE '09), 15-18

Mar. 2009, Seattle, Washington. [6] S. Rohjans, M. Uslar, R. Bleiker, J. Gonzalez, M. Specht, T. Suding,

and T. Weidelt, “Survey of smart grid standardization studies and

recommendations,” 2010 First IEEE International Conference on

Smart grid Comms., 4-6 Oct. 2010, pp. 583 – 588. [7] W. Luan, D. Sharp, and S. Lancashire, “Smart grid communication

network capacity planning for power utilities,” IEEE Transmission

and distribution conference and exposition, 19-22 April, 2010, pp.1-4. New Orleans, LA.

[8] T. Sauter and M. Lobashov, “End-to-end communication architecture for smart grid,” IEEE Trans. on Industrial Electronics, vol. 58, no. 4,

Apr. 2011, pp. 1218-1228. [9] F. li, W. Qiao, H. Sun, H. Wan, J. Wang, Y. Xia, Z. Xu, and P. Zhang,

“Smart transmission grid: vision and framework”, IEEE Trans. on

Smart Grid, vol. 1,no. 2, pp. 168-177, Sep. 2010. [10] A. Bose, “Smart transmission grid applications and their supporting

infrastructure,” IEEE Trans. On Smart Grid, vol. 1, no. 1, Jun. 2010,

pp. 11-19. [11] M. Souryal, C. Gentile, D. Griffith, D. Cypher, and N. Golmie, “A

methodology to evaluate wireless technologies for the smart grid,” 2010 First IEEE International Conference on Smart Grid

Communications, 4-6 Oct. 2010, pp. 356 – 361. [12] P.G. Alvaro, O. Juan, and G. David, “An intelligent agent-based

distributed architecture for smart-grid integrated network management,” 2010 IEEE 35th Conference on Local Computer Networks (LCN), 10-14 Oct. 2010, pp. 1013 – 1018.

2663