Energy and Power Engineering, 2013, 5, 540-545
doi:10.4236/epe.2013.54B103 Published Online July 2013 (http://www.scirp.org/journal/epe)
Modeling of IEEE1588 on OPNET and Analysis of
Asymmetric Synchronizing Error in Smart Substation
Jiongcong Chen1, Haizhu Wang2, Chunchao Hu1, Ka i Ma1, Zexiang Cai2
1Electric Power Research Institute, Guangdong Power Grid Corporation, Guangzhou, China
2South China University of Technology, Guangzhou, China
Received March, 2013
ABSTRACT
The IEEE1588 network time synchronization, matched with smart substation information network transmission, is be-
coming the next generation advanced data synchronization of the smart substation. It is known that the inherent asym-
metry error of the network synchronization approach in the smart substation is highlighted, which is concerned particu-
larly. This paper models the synchronization process of the IEEE1588 based on the communication simulation software
of OPNET Modeler. Firstly, it builds the models of master-slave clock, IEEE1588 protocol and network synchroniza-
tion model, and analyzes the composition and influencing factors of the asymmetry error. Secondly, it quantitatively
analyzes the influence of the synchronous asymmetric error of the IEEE1588 affected by the network status differences
and the clock synchronization signal transmission path differences. Then its correction method is analyzed, in order to
improve the IEEE1588 synchronization reliability and gives the solutions to its application in smart substation.
Keywords: Smart Substation; IEEE1588; Synchronization; Asymmetric Error
1. Introduction
As the important node of the smart grid, the smart sub-
station based on network communication provides grid
application oriented integration data platform, its biggest
feature and advantage is information sharing [1-3]. The
synchronization of high precision data is the important
premise and basic properties to realize information shar-
ing. Application system such as relay protection, SCA-
DA, PMU and so on has a high requirement on the data
synchronization which is paid much attention as the core
problem of smart substation technology [4-6].
At present, the smart substation data synchronization
usually adopts IRIG-B, and uses point-to-point link with
exclusive fiber optic, sending the unidirectional synchro-
nization signal. This type of flow reliability conflicts
with the network information transmission mode of smart
substation [7-8]. IEEE1588 as precision clock synchroni-
zation protocol matches with Intelligent substation net-
work transmission mode based on IEC61850 [9], can
share network resources with other packet, and elimi-
nates the timestamp mark error of NTP/SNTP protocol
mode and its synchronization is accurate to sub-micro-
second, which is becoming the next generation synchro-
nization mode of the smart substation, with good devel-
opment prospects.
At the same time, with the inherent defect of uncertain
network synchronization, synchronized asymmetry error
may occur in the synchronization of IEEE1588 because
of differences running state and transmission path in the
network.
Under the high data synchronization requirements of
smart substation, the issue becomes more prominent
[10-12]. At present, due to the lack of quantitative analy
-sis tools and methods, the study of IEEE1588 is limited
to the field test, the study on the forming mechanism,
influence degree and countermeasure of the dissymmet-
ric error is nearly in a gap, which limits its popularization
and application [13-14].
This paper develops a IEEE1588 synchronization proc-
ess modeling of smart substation with a communication
simulation software named OPNET Modeler, and analy-
ses the differences of network status and the master-slave
clock synchronization signal transmission path quantita-
tively [15], then studies the influence on IEEE1588
synchronous asymmetry error[16], as also as its correct-
ing method.
2. Asymmetric Synchronizing Error
IEEE1588 precision clock synchronization protocol was
promulgated by the IEEE standard committee in 2002,
which is to meet the high-precision requirement of the
measurement and control applications in a distributed

J. C. CHEN ET AL. 541
network of timing synchronization. It’s a brand-new at-
tempt to introduce IEEE1588 into power domain that is
consistent with smart substation network transmission,
and is beneficial to take advantage of information sharing.
But as a network synchronization mode, the problem of
uncertainty is also prominent which needs to be studied
in depth.
2.1. IEEE 1588 Protocol and Synchronization
Process
1) IEEE 1588 protocol feature
IEEE1588 protocol is a distributed network synchro-
nization protocol, and it doesn’t need networking singly
[17]. It can share hardware resources on network, and
chooses multiple transmission paths flexibly, also can get
the real-time status of other nodes on network. It ensures
the reliability of the synchronization system.
Traditional network synchronization time stamp was
performed by the software, and the position of time
stamp is on application layer, so the message parsing and
packaging processing delay constitute a part of synchro-
nization error. As IEEE1588 Protocol time stamps on
physical layer exit by hardware using the technology of
accessing MAC by Ethernet media, which eliminates part
error of time delay, as shown in Figure 1.
2) Synchronization process
IEEE1588 synchronization process is shown in Figure 2.
Figure 1. Time stamp of network synchronization.
Figure 2. Synchronization process of IEEE1588.
Assuming that the master clock and the slave clock
deviation is offset, that is offset = tslave - tmaster , the slave
clock at the local time t1 sends synchronization request
packets to the master clock, the master clock receives
synchronization request packets at the local time t2. As-
suming transmission delay from the slave clock to the
master clock is 1

, so
21
tt offset1


 (1)
The master clock sends synchronization packets to the
slave clock at t3, and the synchronization packet includes
two time scales as t2 and t3. The slave clock receives the
synchronization packet at t4. If transmission delay from
the master clock to the slave clock is 2

, so
43
tt offset2


 (2)
All t1, t2, t3 and t4 are known quantities in formula (1)
and (2), but offset, 1

and 2

are unknown quantities.
Assumes 1

= 2

, so the slave clock can solve and
correct the time error offset based on the formula (1), so


21 43
(
2
tt tt
offset )


 (3)
2.2. The Influences of IEEE1588 Asymmetric
Error
The importance assumption of IEEE1588 synchroniza-
tion is round-trip transmission path symmetry, transmis-
sion delay 1

= 2

, since the differences in both net-
work status and transmission path dynamics, transmis-
sion delay 1

and 2

generally are not equal, which
cannot be deleted directly, so formula (3) should be cor-
rected as:

 
21432 1
2
tt tt
offset




 (4)
21
2
t


  (5)
t

is asymmetric error, the influences of IEEE1588
asymmetric error mainly includes:
1) The difference of transmission path
Transmission delay

is composed of link transmis-
sion delay and switch transmission delay
link
D
s
witch
D,
so
link switch
DD


 (6)
Synchronous packet is stored and forwarded by the
switch address analytical table dynamically, as round-trip
transmission path may be different when the synchronous
packet passes switch and link, link and D
s
witch are not
symmetrical generally. The transfer rate of optical fiber
link is 2/3 of velocity of light, that is to say the
transmission delay of one 1000-meter link is 5μs. A
switch packet processing delay is microsecond level, but
D

J. C. CHEN ET AL.
542
the synchronous precision requirement of IEEE1588
protocol is sub-microsecond level, so the unequal t

from transmission path cannot be ignored.
2) The difference of network status
The communication running state is dynamic, and
running state i is closely related to load conditions of
network, so the switch processing delay
S
s
witch is dif-
ferent under different running state i. As a result, even
round-trip transmission path is symmetry, as the network
status is different, then transmissio n delay
D
1
S

and 2

are not equal.
All the people hold the same attitude that the asym-
metric error makes IEEE1588 synchronization precision
jitter, but the study of asymmetric error stays on qualita-
tive analysis level because of limited to the research
methods. Once the synchronous system of smart substa-
tion out of step, the protection device may lock, wrong
operator miss operate, even more severe is that the sec-
ondary system becomes breakdown. So it is necessary to
build a synchronous process simulation model of
IEEE1588 and quantitative analysis IEEE1588 asymmet-
ric error and then formulate countermeasures.
3. Synchronous Process Modeling and
Asymmetric Error Analysis
IEEE1588 synchronous process model includes master-
slave clock model and synchronous network model, and
the model reflects several features: IEEE1588 protocol
packet processing mechanism, the technique of time
stamp on the physical layer exit by hardware, synchroni-
zation network made by switch and optical fiber and
asymmetric error [18].
3.1. IEEE Master-Slave Clock Model
Figure 3 is based on IEEE1588 master-slave clock model,
it mainly completes following several functions:
Figure 3. Clock model based on IEEE1588.
1) Synchronous packet formation
Master-slave clock sending model produces four main
synchronous packets according to different trigger condi-
tions, master clock produces a Sync packet per second,
then put Sync packet sending time T1 on Follow up, after
receiving the request Delay_req from slave clock, put
receiving time T4 on response Delay_resp packet.
2) IEEE1588 protocol package and analysis
The four main synchronization packets of IEEE1588
are sync, follow up, delay_req and delay_resp, and they
have the same format and the same 64Byte long frame
with padding bytes, but their time stamps have different
information. Use the following up packet model as an
example, origin Timestampas master clock sends time T1,
and sync interval is synchronous period(usual is 1 sec-
ond), grandmaster Clock Variance is the change rate of
master clock. The data link layer package and analysis
packet MAC address, and interface module is general
module, completing the protocol analysis and the data
packet filling from the data link layer to application lay-
erin IEEE1588.
3) Time stamp marker and read
Different from NTP/SNTP traditional synchronization
protocol which is marking and reading time stamp on
application layer, time node model of IEEE1588 puts
time stamp marker function op_pk_stamp(pkptr) and read-
ing function op_pk_stamp_time_getpkptr) on the exit of
the data link layer, to realize time stamp marker and read
function of clock, and reduce protocol analysis time.
4) Calculation and correction of Master-slave clock
offset and asymmetric error
The slave clock uses formula (7) to calculate and cor-
rect master-slave clock offset and asymmetric error t

.
(1)()()()
_
iii
offsetoffsetTdelay offset
i

 (7)
T is master-slave clock’s correction value through the
calculation of t1, t2, t3 and t4 according to formula (3), and
delay_offset is asymmetric error . t
Some codes of master clock model areas follows:
/* Setpacket formal,send information and so on,and
put sending time t1 on the */
/* packge of follow_up */
pkptr =op_pk_create (pksize);
t1=(op_pk_creation_time_get(pkptr))*10000000000;
pkptr_Sync=op_pk_create_fmt("Sync");
pkptr_Follow_up=op_pk_create_fmt("Follow_up");
op_pk_nfd_s et(pkptr_Fo llow_up,"o riginTimesta mp",t
1);
….
op_pk_fd_set(p kptr,1,OPC_FI ELD_TYPE_INTEGER
,1,16);
op_pk_fd_set(p kptr,2,OPC_FIELD_ TYPE_PACKET,
pkptr_Sync,464);
/* Put receiving time t4 of Delay_req packet on

J. C. CHEN ET AL. 543
response packetDelay_resp. */
pkptr = op_pk_get (op_intrpt_strm ());
op_pk_stamp(pkptr);
t4=(op_pk_stamp_time_get(pkptr))*10000000000;
op_pk_nfd_set( pkptr_Delay_resp,"or iginTimestamp",t
4);
Some codes of slave clock model areas follows:
/* Calculate and correct master-slave clock offset and
asymmetric errordelay_offset. */
T=((t2-t1)- (t4-t3))*0.5;
delay_offset=((t2-t1)+(t4-t3)) *0.5;
offset=offset-T- delay_offset;
3.2. Network Synchronization Model
Network synchronization model includes running status
network model, transmission path network model, the
switch with 500000 p/s processing capacity and the100M
optical fiber link, shown as Figure 4. In Figure 4(a), to
observe the transmission delay with different network
running status, inject background traffic into switch. In
Figure 4(b), to observe the transmission delay with
different transmission path, increasing the number of
switch from 0 to 3 on master-slave synchronous packet
transmission path.
Different network running status
3.3. Asymmetric Error Simulation
Table 1 is the performance of transmission delay, injects
different background traffic to the switch. When back-
ground traffic is 25%, transmission delay is 262.993
s

,
which costs 1
s

more compared to transmission delay
without background traffic. Transmission delay becomes
2.667
s

when background traffic is 40%. When back-
ground traffic increases to 95%, transmission delay is
337.993
s

. Value of transmission delay varies from
0.211
s

to 76
s

on different background traffic, and
it enlarges constant. Asymmetric error fromdifferent
network running status couldnot be ignored.
t
Figure 4. Netw or k synchronization model.
2) Different transmission path
From Table 2, when the number of switch increases
one more, transmission delay will increase 87. So when
configuration of communication network switch change
and master-slave clock round-trip transmission path differ
one or one more switch, correction of asymmetric error
∆t must exist.
So assumption of master-slave clock round-trip trans-
mission path is equal exists defect, asymmetric error from
different network status and transmission path should be
a necessary link of synchronous correction.
4. Asymmetric Error Correction
There are some differences between smart substation
communication network and general LAN: 1) the struc-
ture of network is simple. Configuration of smart substa-
tion secondary system is normative according to voltage
level, the number of devices are limited, switches are no
more than four in general; 2) Traffic flow is specific. All
automatic business such as measurement, protection, test
and control and etc. have their corresponding SV/GO-
OSE/MMS/Synchronization information flow, and the
feature of information flow is specific, little uncertain
flow. Specific feature of smart substation communication
network makes asymmetric error correction feasible.
4.1. Asymmetric Error Correction Method
Assuming at time t, master-slave clock transmission de-
lay is measured by network monitoring device.
Table 1. Impact of background traffic on transmission de-
lay.
Background traffic Delay(μs)
0% 261.993
5% 262.204
10% 262.438
20% 262.993
40% 264.660
80% 277.993
95% 337.993
Table 2. Impact of path on transmission delay.
The number
of switch Delay(μs) The increase of transmission delay(μs)
0 88.553 /
1 176.207 87.654
2 261.993 85.786
3 349.145 87.152

J. C. CHEN ET AL.
544
11tlinktswitch t
DD
1

 (8)
22tlinktswitch t
DD
2

 (9)
According to formula (5), asymmetric error t t

is:
21
2
t
t
tt

 
 (10)
Put t from formula (10) into master-slave clock
correction, that is
t
_
()delayoffset i in formula (7).
Since 1t

and 2t

are got from measurement in op-
eration, correction error exists because of measurement.
4.2. Asymmetric Error Correction Simulation
Analysis
1) Simulation scenario setting
Asymmetric error correction simulation model is shown
as Figure 5. Master clock transmission path is from
switchA to slave clock, while return path is composed of
EDCB.
Setting parameters of asymmetric error simulation as
Table 3:
2) Analysis of asymmetric error simulation results
Results of asymmetric error simulation are shown in
Figure 6. Curve1 is uncorrected asymmetric error t

,
from 0 to 34 s, range of is 0-0.987 t
s

, synchronous
error less than 1
s

, from 35s to 100s, with the increasing
Figure 5. Synchronization network model in asymmetric
error simulation.
Table 3. Parameters of asymmetric error simulation.
Path of master-slave
clock
Path of slave-master
clock
Simulation
time(s) switch Background
traffic(%) switch Background traffic(%)
0-100 B 0-100
101-115 CB
116-130 DCB
131-145
A 0
EDCB
0-14
background traffic, t

enlarges to 1.143-41.937
s

.
Obviously t

has completely over the minimum
requirement that IEEE1588 synchronous precision
should less than 1
s

. At the time of 101s, 116s and 131s,
the number of slave clock path switch turn into 2, 3 and 4,
t

enlarges to 43.5010-136.169
s

further.
As correction error is 1%, curve2 is asymmetric error
t

which uses the corrected method of this paper to
correct. From 0 to 100s, and when master-slave round-
trip switches have the same number, the range of back-
ground traffic is 0-100%, is 0-0.419 t
s

after
correction. From 101 to 130s, the differ number of
round-trip switch is 2, and the range of background
traffic is 0-14%， t

increases to 0.435-0.909
s

. At
the time of 131s, the number of switch turn into 3, t

is1.305
s

, over 1
s

, and it cannot meet the
requirement of IEEE1588 synchronous precision.
Curve3 is the change condition of after correction
when assume correction error is 0.5%. The range of
tt

is 0-0.681
s

after correction, even the differ number of
master-slave round-trip is 3, it still can meet the precision
requirement of 1
s

.
Figure 6. Results of asymmetric error simulation
To sum up, asymmetric error of master-slave
clock link impacts IEEE1588synchronous precision se-
riously, while the correction method of this paper will
reduce the impact of asymmetric error greatly. Relative
certainty of smart substation communication network
makes the measurement of 1t
t

and 2t

feasible, the
problem of asymmetric error of network synchronization
can be solved effectively by the error correction which is
according to this.
5. Conclusions
This paper develops IEEE1588 synchronous process
model, quantitatively analyses asymmetric error from
network running status and the difference of master-slave
clock transmission path, researches that the correction
can supply a new approach and method expecting for actual

J. C. CHEN ET AL.
545
measurement for the quantitative analysis ofIEEE1588
synchronization process, and develops research thought
of IEEE1588 synchronization uncertainly problem.
Although IEEE1588 has necessitated solved questions,
IEEE1588 adapts to the development trend of smart sub-
station network transmission. To search and improve
reliable method of IEEE 1588 synchronization, overcome
network uncertainly shortcomings, make full use of net-
work synchronization is the inevitable trend for the syn-
chronization technique of smart substation.
REFERENCES
[1] S. L. Zhao, M. Q. Hu, X. B. Dou and Y. S. Du, “Research
of Time Synchronization in Digital Substation Based on
IEEE 1588,” Power System Technology, 2008, Vol. 32,
No. 21, pp. 97-102.
[2] L. Zhou and Y. F. Zhao, “Analysis and Experiment of
Transparent Clock in Time Synchronization System of
Smart substation,” Automation of Electric Power Systems,
Vol. 36, No. 6, 2012, pp. 106-111.
[3] Y. Z. Chen, “Researcher on Time Synchronization Tech-
nology of Power Systems,” Guangdong Electric Power,
Vol. 20, No. 6, 2007, pp. 41-44.
[4] H. S. Hua, D. H You and H. W. Yu, “Study on Method of
Time Synchronization in Digital Substations,” Guang-
dong Electric PoeR, Vol. 21, No. 5, 2008, pp. 52-55.
[5] W. B. Xia, “Construction of Synchronous Clock System
of Guangdong Power Digital Network,” Guangdong
Electric Power, Vol. 13, No. 2, 2012, pp. 11-14.
[6] J. Jasperneite, K. Shehab and K. Weber, “Enhancements
to the Time Synchronization Standard IEEE-1588 for a
System of Cascaded Bridges,” Factory Communication
System, Proceedings,2004 IEEE International Workshop
onVolume, No. 22-24 , 2004 , pp. 239-244.
[7] R. David, “IEEE 1588 Boundary Clock and Transparent
Clock Implementation Using the DP83640,” National
Semiconductor Application Note 1838, April28, 2008.
[8] S. Martins, H. Clemens, S. Joachim, “Get on the Digital
Bus to Substation Automation,” IEEE Power & Energy
Magazine, Vol. 5, No. 3, 2007, pp. 51-56.
doi:10.1109/MPAE.2007.365820
[9] Y. C. Hu, X. S. Zhang, W. G. Xu, et al., “Error Analysis
and Detection Method for IEEE1588 Clock Synchroniza-
tion System,” Automation of Electric Power Systems, Vol.
34, No. 21, 2010, pp. 107-111.
[10] D. Annex, “Transport of PTP Over User Datagram Pro-
tocol Over Internet Protocol Version 4,” IEEE
Std1588-2008, 2008.
[11] Y. J. Wei, “Effect of IEC 61850 Standard on Substation
Automation System in China,” Guangdong Electric
Power, Vol. 23, No. 6, 2010, pp. 31-33.
[12] L. P. Jiang, X. J. Zeng, Z. Li, A. Penh, “Realization of
High Accuracy Synchronous Clock for Power System
Based on GPS,” Power System Technology, Vol. 35, No.
2, 2011, pp. 201-206.
[13] Z. L. Yin, W. S. Liu, Q. X. Yang, et al., “A New IEEE
1588 Based Technology for Realizing the Sampled Val-
ues Synchronization on the Substation Process Bus,” Au-
tomation of Electric Power Systems, Vol. 29, No. 13,
2005, pp. 60-63.
[14] X. Y. Huang, “Discussion on the Application of NTP in
the time Synchronization of Power Automation Devices,”
Automation of Electric Power Systems, Vol. 29, No. 15,
2005, pp. 93-95.
[15] P. F. Yu, Q. Yu, H. Deng, et al., “The Research of Preci-
sion Time Protocol IEEE 1588,” Automation of Electric
Power Systems, Vol. 33, No. 13, 2009, pp. 99-103.
[16] T. Bauer, H. P. Lips, G. Thiele, et al., “Operational Tests
on HVDC Thyristor Modules in a Synthetic Test Circuit
for the Sylmar East Restoration Project,” IEEE Transac-
tions on Powe r Delivery, 1997.doi:10.1109/61.636927
[17] S. B. Zhou, Z. G. Tian, Y. B. Zhao, et al., “Substation
Clock Server Meeting IEC61850,” Power System Protec-
tion and Control, Vol. 38, No. 7, 2010, pp. 56-58.
[18] Y. Ju, “Modes of Web Communication of Digital Substa-
tion,” Power System Protection and Control, Vol. 38, No.
1, 2010, pp. 92-95.