WSN-based health monitoring of highway tunnel lining in seasonal cold region

*Fei Du1), Dongming Zhang2), Yan Wu3), Hongwei Huang4)

and Fushen Sun5)

1), 2), 4) Department of Geotechnical Engineering, Tongji University, China,

3) Wisen Innovation Ltd, China. 5) Jilin Provincial Transport Scientific Research Institute, China.

2) 09zhang@tongji.edu.cn

ABSTRACT

Wireless sensor networks (WSN) have attracted great attentions for structure health monitoring (SHM) in recent years. However, the application of WSN for monitoring tunnels in seasonal cold region still maintains big challenges in consideration of radio propagation and harsh environmental conditions. In this paper, a developed wireless sensor network is presented; the basic features of the hardware nodes and the WSN communication protocol are introduced briefly. A WSN system consisting of 37 nodes that can measure the inclination of the secondary lining was deployed in Laoyeling tunnel, which is a 2391m long stretch of NATM concrete-lined highway tunnel located in Jilin province, China. Over 5 months monitoring period, the network performances including the system reliability and the wireless network topology are analyzed. Furthermore, the paper illustrates the temperature and inclination monitoring results; the temperature distribution law of the surface of lining is presented according to the monitoring data, and the monitoring inclination data indicates that the lining deformation has been continuously evolving. This deployment demonstrates the feasibility and reliability of the presented WSN system for health monitoring of the long tunnels in seasonal cold region. 1. INTRODUCTION

Large numbers of highway or railway tunnels have been built in seasonal cold region around the world. The coldest monthly average temperature in some seasonal

                                                              1) Graduate Student 2) Assistant Professor 3) Engineer 4) Professor 5) Research Fellow

cold areas is lower than -20˚C, in this case, the fissure and pore water of the surrounding rock mass of tunnel become frozen, as a consequence, expansion in volume occurs. This volumetric expansion is restrained by the tunnel lining and the unfrozen surrounding rock; therefore, the frozen rock results in forces on the lining, known as the frost-heave force of the surrounding rock. The frost-heave force and other forces acted on the tunnel lining may result in deformation or cracking and lead to some kind of frost damages. Moreover, the large temperature difference over a year will result in temperature stress in tunnel lining. During their lifetime, these tunnels inevitably deteriorate because of the frost damages and temperature stress, for example, 33 railway tunnels in the seasonal cold region of China have shown different degrees of frost damage (Lai 2000), so these tunnels require timely maintenance in order to prevent possible severe disruptions or accidents.

Structure health monitoring (SHM) is progressively being seen as an effective way to minimize identified or potential risks and is increasingly applied to civil infrastructures. Traditionally, manual monitoring and wired sensors network are applied to SHM, which have the disadvantages of low efficiency, high cost, long setup time and difficulties in cabling management. Especially, it’s a huge challenge for workers to conduct monitoring work due to the very low ambient temperature in cold environment. So it is badly in need of deploying an effective, reliable, large-scale, real-time and automatic monitoring system for civil infrastructures in seasonal cold regions.

In recent years, with the rapid advancement of sensor technology and wireless communication technology, wireless sensor networks (WSN) have attracted great attentions for SHM. The use of wireless sensors and wireless communication technique, which transmits the data using radio, allows a rapid and large-scale deployment due to elimination of the cables and self-organizing capability. Meanwhile, the improvement of micro-fabrication techniques has allowed the development of Micro-Electric-Mechanical system (MEMS)-based sensors that have the advantage of small volume, light-weight, high reliability, low-power dissipation and low cost. Combining the WSN with the MEMS-based sensors, there is the possibility for significant cost saving, long-term, real-time and automatic SHM. Straser and Kiremidjian (Straser 1998) are the first to illustrate the design of a low-cost wireless sensing unit intended for SHM, and they have validated the performance of their developed wireless sensing unit prototypes by installing five of them upon a 15m span of the Alamosa Canyon Bridge. Since then, numerous researchers have developed wireless sensors prototypes (Spencer 2004, Yu 2009, Ha 2013) and also have deployed sorts of WSN systems on bridges (Lynch 2003 2005, Chung 2004, Nagayama 2009) and urban metro tunnels (Bennett 2009 2010, Yin 2015, Wang 2016). Current research investigating the application of WSN for SHM has fewer focused on the infrastructures in seasonal cold region, and very fewer groups have deployed WSN in the long mountain tunnel with the goal of long-term monitoring. The application of WSN for mountain tunnels in seasonal cold region is still a huge challenge in consideration of radio propagation inside tunnel and the very low

temperaTo

the SHMstructurathe angvertical using thmeasure(Ha 201

In the mesinclinatioinstalledinclinomdescribethe monanalyzed 2. BASI

2.1Th

the hardFig. 1.

Thresponsfrequencnetworkvia mob

ature. gether withM, i.e., acal deformale of rotatmembers.

he inclinates in evalu3). this paper

sh networkon of tunnd, includin

meter, 8 vibes the wirenitoring datd.

C FEATUR

1 Nodes ofhis paper pdware of th

(a)

he WSN gible for thcy modifica. Meanwhiile network

h the existccelerationation providtion in the

As such, ion, such

uating the s

r, a developk protocol el lining wg 28 WSbrating wireless netwota including

RES OF W

f WSN Sysresents a d

he system

WSN GateF

ateway (Fhe commanations, andile, it transk or local s

ing structu, displace

des informacase of han inclinothat inclin

safety of in

ped WSN are introd

was deployeSN tilt nodre (VW) inork performg the temp

WIRELESS

stem developedconsists of

eway Fig. 1 Node

Fig. 1(a)) isnd issuingd for data smits the dserver via s

ural responement andation that ihorizontal meter cannation mendividual st

is presenteduced brieed in Laoydes integrnterface nomance duriperature da

SENSOR

WSN systf WSN gat

es of the W

s used as

g, such ascollection

data and systandard R

nse indicesstrain, th

s useful inmembers evaluate asurementructural m

ed; the basfly. A WSNyeling tunnrated silicoodes and aing the moata and inc

R NETWOR

tem for struteway and

(bWSN system

s a key uns the samp

from all thystem infoRS232 con

that are che inclinatin evaluating

and the dthe deformts can beembers an

sic featureN system nel. A total on capacia WSN ga

onitoring peclination da

RK SYSTE

uctural inclWSN tilt n

) WSN tilt m

nit in the Wpling time he nodes irmation to

nnections. T

common adion changg the defledrift in themation of me used asnd entire s

es of the nofor measu of 37 noditive MEMateway. Theriod. Furthata are pre

EM

lination monodes, as s

node

WSN systinterval anvolved in

o the remotThe WSN

dopted in e during

ection via case of

members primary

tructures

odes and uring the des were

MS-based he paper hermore, eliminary

onitoring; shown in

em. It is and radio n a mesh te server gateway

is powered by four 3.6V industrial D cell batteries, thus, the deployment of a WSN gateway can be more flexible from the traditional power supply cabling. To analyze the power consumption of the WSN gateway and WSN tilt node, laboratory experiments were carried out with the hardware settings including Keysight 34401A multimeter, Atten APS3005D power supply and PC. Fig. 2(a) shows the results of the power consumption analysis for the WSN gateway. The battery life of the WSN gateway increases linearly with the increase of the sampling time interval. Note that the battery life can be further extended by a factor of 1.5 if the WSN gateway is one-sixth times less often making connections to the server.

(a) WSN gateway (b) WSN tilt node

Fig. 2 Battery life of the WSN nodes

As the sensing devices in the WSN system, the WSN tilt node is as shown in Fig. 1(b), it is capable of sensing the inclination variation of the structure as well the working temperature, processing the measurements and transmitting the data packet via its internal radio module. The WSN tilt node consists of a MEMS-based dual axis inclinometer with the sensing range from -30˚ to 30˚. Another key electronic component inside the WSN tilt node is a microcontroller with low power 2.4GHz transceiver that is compliance with IEEE 802.15.4 Standard. It is responsible for system controlling, data packets transmitting and receiving at 2.4GHz. To prevent physical damage to the inside electronic components by the harsh environment, the external component is constructed of aluminum-alloy with the size of 80mm×75mm×57mm and of IP66 protection from water and dusts. The 5dBi Omi-direction antenna with a length of approximate 200mm is used to provide longer radio communication range. A single 3.6V industrial D cell battery is used to power the WSN tilt node. The battery life of the WSN tilt node is closely related to the number of hops taking for a message to go through in the mesh networking. Fig. 2(b) shows the battery life calculated for a WSN tilt node taking no sub-mesh network and taking 9 hops of sub-mesh network of its own.

2.2 WSN Communication Protocol In this study, the WSN communication protocol operating in the ISM 2.4GHz radio

1230

89

170

321

0

50

100

150

200

250

300

350

0 10 20 30 40 50 60 70

Bat

tery

lif

e (d

ays)

Sampling time interval (mins)

242

1095

2981 4760 7310

96

434

1186 2185

4029

10

100

1000

10000

0 10 20 30 40 50 60 70

Bat

tery

lif

e (d

ays)

Sampling time interval (mins)

Taking no hop

Taking 9 hops

band is designed based on the IEEE 802.15.4 wireless protocol standard. The radio frequency is between 2.405GHz and 2.480GHz with 16 channels (channel 11 to channel 26) of 5 MHz bandwidth at each channel. The transmit power is less than 1.4mW (i.e., 1.5dBm) typically with the minimum transmission rate of 250Kbit/s. The WSN communication protocol supports 10 hops network structure (e.g., the radio link from WSN gateway to the 1st layer node is called the 1st hop). The presented WSN possesses self-organizing capability to organize various network forms, such as ring network, mesh network, star network and line network etc. Single node environmental coverage has been measured and predicted using Modified 2D Finite-Difference Time-Domain (FDTD) method (Wu 2009); the results indicate that the communication distance is 100m at least while the antenna is placed at approximate 100mm away from the segments inside metro tunnel. 3. DEPLOYMENT OF WIRELESS SENOR NETWORK IN HIGHWAY TUNNEL

3.1 Field Deployment Site Jilin province is located in the northeast of China where the monthly average

temperature is lower than 0˚C from November to March of the following year; January is the coldest month when monthly average temperature is approximately -25˚C ~ -20˚C (China meteorological data service center, CMDC). The frozen earth type is seasonal frozen earth; the maximum frozen depth is approximately 1m ~ 3.5m over the years; in one year, the ground surface will be frozen in the latter of October, and the freezing period is about 112 days ~ 171 days (Zhou 2000).

Laoyeling Tunnel is a one-way and double lane tunnel with disjunctive up and down lines, located below Laoyeling Mountain in Jilin province. The tunnels were excavated by New Austrian Tunneling Method (NATM). Up and down line tunnel is 2360m and 2291m long respectively. Fig. 3 shows a geological longitudinal section alone Laoyeling down line tunnel. The maximum overburden depth of the tunnel is approximately 250m. There are three rock formations in Laoyeling Mountain passed through by the down line tunnel, the two main rock formations are alteration tuff and granite porphyry, the surface of the mountain is a relatively thick layer of clay containing reduced stones. There are four faults in the district alone the tunnel, as shown F1~F4 in Fig. 3, the tunnel gets through the fault F1, F2 and F4. The tunnel direction intersects these three faults at an angle of 60˚~90˚. Underground water consists of Quaternary pore water and bedrock fissure water, the overburden of the underground water is between 7.42m and 18.45m.

Figsection ithe tunnfabric shThe PVseconda

3.2

Fig

g. 4 showsis a circuit

nel section,hotcrete anC waterprary lining.

2 Wireless

g. 3 Geolog

s a typicaarch havin, compositnd rock booof board

Fig. 4

Sensor N

gical condit

l cross-seng clear hee lining walts; and theand geote

4 Typical c

etwork Ge

tions and o

ction of Laeight of 7.1as adoptede secondaextile was

cross-sectio

eometry

overburden

aoyeling dm and clea

d. The primary lining w

lay betwe

on of the tu

n in the tun

down line tar width of

mary suppowas C30 reeen the pr

unnel

nnel

tunnel. Thf 10.6m. Toort consistseinforcing crimary sup

he tunnel o support s of steel concrete. pport and

Thshown in(VW) intsame Wsystem installedlong aw(MS1~MlongitudinstalleddiagramWSN tiltbetweenhopping

Wh

seconda1 and Fi

he networkn Fig. 5. Aterface nod

WSN commwas deplo

d in the cenway from eMS5 in Fig. inal directi

d at the arcm is shown

t nodes ann the two a.

hen installary concretig. 6 show

k geometryA total of 37des and a

munication oyed on Dntral of theeast tunne

5) were inon of the tuch springinin Fig. 5.

nd two VWadjacent M

Fig. 5 WSN

led the note lining; ththe node p

y design o7 nodes wWSN gateprotocol w

December tunnel, 11l portal. Astrumenteunnel. For ng and spaThe MS s

W interfaceMS section

N system d

odes, plug hen the nodprototypes

of WSN sywere installeway. Notewere deplor 14th and 150m longA total of fied by four W

each MS candrel on tsections are nodes spns, allowing

deploymen

bolts werdes were f

s after insta

ystem adoed; 28 WS

e that the Voyed as re

15th, 201away fromive cross WSN tilt nocross sectthe both sire spaced paced by g the wirel

nt in Laoye

re used tofixed to theallation in s

pted for LSN tilt nodeVW interfacelay nodes6. The W

m west tunnsections o

ode at four ion, four Wdes of tunabout 560110m or 1ess netwo

ling Tunne

o fix the me metal holsite.

Laoyeling Tes, 8 vibrace nodes us. The whoWSN gatewnel portal aof secondalocations a

WSN tilt nodnnel, the sc0m and 5720m was

ork using e

el

metal holdelder using

Tunnel is ating wire using the ole WSN way was and 1140 ary lining alone the des were chematic 0m. Two installed

extensive

er to the bolt. Fig.

A m

from all mobile ngatewayAccordinthe battein the wWSN tiltmonitor 4 WSN S

4.1Th

hardwarrating ofhardwarharsh.

Thtransmitwirelessevaluatetime intehour (hanode sh2017. Apacket ttransmit

mobile netthe nodes

network. Thy is one-sng to the laery life of W

worst case t node witthe lining s

SYSTEM P

1 System Rhere are twre and radf all the nore failures

he radio cotted from Ws sensor ned. As meerval takenalf an hourould send

An indicatothat is drotted. In this

Fig.

twork is avs can be the samplinsixth timesaboratory pWSN tilt norespectivethout meastructure fo

PERFORM

Reliability wo main aredio connecdes is moreven thou

onnectivityWSN gatenetwork contioned be

n for data pr in the beg

to the WSr for the ra

opped (i.e.s paper, the

6 WSN tilt

vailable intransmittedng time intes often mapower conodes and Wely. The basurement or one yea

MANCE

eas of conctivity (Benre than IP6ugh the e

y has expeeway to reonnectivityefore, the ackets from

ginning). IdSN gatewayadio conne not transe data loss

nodes dep

Laoyelingd to the reerval of theaking consumption a

WSN gatewattery life o

function. Tar at least.

ncern whennnett 201066, the WSnvironmen

erienced cemote servy, so that sampling tm all nodedeally, the y is 2952 fectivity quasmitted) tos ratio is de

ployment in

tunnel, memote serve WSN tilt nnections tanalysis, away is at leof the VW Thus, the

n it comes ). Because

SN deploymntal conditi

connectivitver contaithe radio time intervs to reach total numbfrom Deceality is the the total efined by t

n site

eaning thaver by the nodes is 1 to the seras shown ineast 4029 dinterface ndeployed

to reliabilite the ingrements haveons in La

ty issues. n the infoconnectiv

val of WSNthe WSN g

ber of datamber 14th,

e ratio of thexpected he Eq. (1):

at the dataWSN gatehour and t

rver, i.e., n Fig. 2 andays and 4

node is lonWSN sys

ty of WSNess protece not suffe

aoyeling tu

The data ormation avity qualityN tilt nodegateway a

a packets t, 2016 to Ahe numbenumber o

:

a packets eway via the WSN 6 hours.

nd Fig. 3, 482 days nger than stem can

N system: ction (IP) ered from nnel are

packets bout the

y can be and the

re both 1 hat each

April 19th, r of data

of packet

% 1 100 (1)r

i i G i G G iR T T T where iR is the data loss ratio for node i, G iT   is the counts of commands from WSN

gateway to node i, i.e., the expected value of the number of data packet transmitted from node i to WSN gateway, i GT   is the actual number of data packet from node i and

r

i GT   is the repeated number of data packet from node i. The number of data packet

from the nodes is counted at the WSN gateway, so the loss by traveling across several nodes to reach to the WSN gateway is also included in this ratio.

Fig. 7 Data loss ratio of nodes versus spatial distance to the WSN gateway

Fig. 7 shows the data loss ratio of all WSN tilt nodes and VW interface nodes with

different spatial distance to the WSN gateway. The minimum data loss ratio is 0%, i.e., all the data packets of the node were transmitted to the WSN gateway, there are total 4 nodes with the data loss percentage of 0%, the maximum distance to WSN gateway of these 4 nodes is 350m. The maximum data loss ratio is 54.98%, i.e., 1612 data packets were dropped, this WSN tilt node was installed on the spandrel close to the east portal of the tunnel which is about 1130m apart away from the WSN gateway. In general, the data loss ratio increases with the spatial distance to the WSN gateway, there are 6 nodes whose data loss ratio is more than 15%, and the distance to the WSN gateway of these 5 nodes is more than 1000m. Shortening the distance between the WSN gateway and the nodes is an effective way to decrease the data loss ratio.

Fig. 8 shows the percentage of the nodes with different data loss ratio, the number in bracket is the data loss ratio range. For the deployed WSN system with only one WSN gateway, there are 63.9% of all the nodes, i.e., 23 nodes, whose data loss ratio are less than 5%, whereas, 16.7% of all the nodes (6 nodes) whose data loss ratio are more than 15%.

Beyond doubt, for the long tunnel WSN monitoring, two effective solutions to reduce the amount of the nodes with large data loss ratio would install more WSN gateways or install more relay nodes so that the distance between adjacent nodes can be shortened, but this increased reliability comes with a subsequent increase in cost use.

0

10

20

30

40

50

60

Da

ta l

oss

rat

io (

%)

Distance to gateway (m)

0 200200 400400 600600 800800 10001000 12001200

West East

4.2Ba

that inditopologydifferentpassing momentthe VW beside tthe archas show

Ttransmitthe greeradio linand the from thethere we23:19 Aphops, tha certainpacket wuntil thetime is mstored w

Fig.

2 Network ased on thvidual datay can demot moment,

of vehiclets. The filleinterface

the nodes h springingwn in Fig. 5These two t the data pen solid arrk from 3rd hblack soli

e nodes to ere 5 hopspril 18th, 2e rest of thn momentwould be se node wasmore than 2will be elim

8 Percent

Topology e layout oa packet constrate ththe netwo

es. Fig. 9ed circle renodes, anis the nod, and the o

5 and Fig. 6figures botpackets. Trows are thhop nodesd arrows athe WSN g

s at this co017, 44%

he nodes wt doesn’t mstored in ths re-online24 hours, tinated, tha

tage of the

of the wholould take f

he travelingork topolog shows twepresents nd the filledde ID. The other two r6. th illustrate

The cyan sohe radio lin, the blue sare the radgateway. Fommunicatof all the n

were offlinemean that he data mee. Howeverthe reboot

at is, the da

nodes wit

le WSN syfrom the ng route of tgy will diffwo differenthe WSN d pentagraoutside tw

rows of the

e that the olid arrowsk from 4th solid arrowdio link froFig. 9(a) shtion mome

nodes, i.e., e. It is nece

the data emory of tr, the nodeof a node

ata packets

th different

ystem, theode to thethe data pafer due to nt networktilt nodes;

am is the wo rows ofe nodes we

WSN systs are the rhop nodes

ws are the rm 1st hop

hows that aent, where16 nodes

essary to npacket wahis node ae will autocomes wit

s are dropp

data loss

re are ma WSN gateackets for the extern

k topologie the filled WSN gate

f the nodesere installe

em uses madio link fr

s, the red sradio link frnodes, tha

all the nodeas, as showere onlinote that a

as droppedand would matically rth a subseped.

ratio

any possibeway. Thea specific nal condites at two square re

eway. Thes were insed on the s

multi-hop rrom 5th hosolid arrowrom 2nd hoat is, the res were onown in Figne and thernode was

d; this unsnot be tra

reboot if thquent that

le routes e network

node. At ions and different presents

e number talled on spandrel,

outing to p nodes, s are the

op nodes, radio link nline and . 9(b), at re were 4 offline at

sent data ansmitted he offline the data

Thinformatprobabil

where P

i jn is

number Fig

the linkslinks wicommundifferenton the snode A4vehiclesare locacommunpossiblebe a relapackets to the W

he data paction, it is feity of the li

i jP is the

the numbe

of the datag. 10 shows with a prith a probnication bet possibilitiesame side 46C, node s will influeated on tnication pre, consequay node, fare hoppe

WSN gatew

(a) Netw

(b) NetwFig. 9 Net

cket from aeasible to nk from no

e possibility

er of data

a packets wws the mosrobability obability beetween noes for routiwith WSN 22FF andnce the dirhe opposotocol makently, clos

for instanceed to the W

way, thus, th

work topolo

work topoltwork topo

a node hascompute

ode i to nod

iP

y of the lin

packets tr

which are tst likely comof above 5etween 10des is quiing. As cangateway a node A01rect commite side wkes nodeser to the We, the nod

WSN gatewhe nodes lo

gy at 21:48

ogy at 23:ologies at tw

s both a pathe probabde j is intro

i j i jP n N

nk from no

ransmitted

transmittedmmunicati

50%, where0% and 5ite busy an be seen fare more li15 etc., this

munication bwith the Ws to commuWSN gatewde 222C coway when tocated aw

8 January

19 April 18wo differen

arent nodebility of eaoduced usi

iN

ode i to no

from nod

d from nodon links. Teas the bla50%. Thisnd complicfrom Fig. 1ikely to rels appearanbetween WWSN gateunicate dirway doesnompares wthe nodes ay from the

7th, 2017

8th, 2017 nt moments

e ID and a ach link beng the Eq.

de j (includ

e i to node

de i. The red thicack thin sos figure ilcated; all 0, the noday the datance occurs

WSN gateweway. In grectly to th’t mean th

with the nodcannot coe WSN gat

s

node ID. Fetween nod. (2):

ding the g

e j. iN is

ck solid arolid arrowsllustrates the nodes

des that area packets,s due to paway and nogeneral, thhe WSN gahe more pode 2329. Tmmunicateteway tend

From this des. The

(2)

gateway).

the total

rrows are s are the that the

s provide e located such as assing of odes that he WSN ateway if ossible to The data e directly d to send

the dataselected

Fig. 10with a

a 5. MONI

A chave befield in tWSN tiltsecondaexperimmodel ttempera

5.1Th

for the tutunnel cthe WSNabout 5cregarded

Fothe temtemperadirectiontemperalongitudthe two is not syeast tun

a packets d node tran

0 Network probability

above 10%

ITORING R

certain numeen measutunnel cant nodes of ary lining. ents haveto elimina

ature fluctu

1 Temperahe temperaunnel desi

can be obtN tilt nodescm away fd as the te

or each MSmperature ature distribn of the tunature of thinal directisides of th

ymmetricalnnel portal

to one of nsmits the

topology sy of above 5%; : WSN

RESULTS

mber of dared by the

n be obtaithe MS seBefore the

e been alrate the peations.

ature Monitature field ogn in seasained thros of the fivfrom the suemperatureS section, tvalue meabutions of tnnel. In Dehe surfaceon, that is

he portal to, the distanis approxi

the WSN data packe

showing the50%; blacktilt node;

OF WSN

ta packets deployed ned accor

ections (she installatioready conderformance

toring Resuof the liningsonal coldugh the tee MS secturface of lie of the surthe temperasured bythe surfaceecember, Je of lining, the parabo inside of nce of the mate three

tilt nodesets to the W

e most likek thin solid

: VW inte

SYSTEM

s including WSN syst

rding to thown in Fig

on of the Wducted fore degrada

ults g and surrregion. Th

emperatureions werening. In thrface of linrature of thy all four e of lining iJanuary ang follows tbola distribthe tunnelcross-sect

e quarters

s or VW inWSN gatew

ely routes (d arrows: perface node

temperatutem. The ve tempera

g. 5) do meWSN tilt nor build theation of t

rounding rohe variatione monitorininstalled ois paper, ting.

he surface WSN tilt

in differentnd Februarthe same bution. Thel, howevertion with thof the tun

nterface noway or ano

(red thick srobability oe; : WSN

re and inclvariation ruature monieasure the odes in sitee temperatthe WSN

ock of tunnn rules of teng data. Son the lininghe monitor

of lining isnodes. F

t month alory, that is, i

distributioe temperatr, the temphe highest nnel length

odes and other node

solid arrowof below 50N gateway

lination infules of temitoring dat

e movemente, the temture comp

tilt node

nel is the ktemperatur

Shown as Fg with the ring tempe

s the meanFig. 11 shong the lonin cold seaonal trendture increaperature dis

temperatu. What’s m

then the .

ws: links 0% and

y )

formation mperature

a. Some nts of the

mperature pensation

due to

key factor re field in Fig. 1(b), distance

erature is

value of hows the ngitudinal ason, the s in the ses from stribution ure to the more, the

temperature closed to the west tunnel portal is high than the temperature closed to the east tunnel portal. The temperature is the coldest in January, when the coldest monthly average temperature of the surface of the lining is approximately -16°C, and the largest temperature difference along the longitudinal direction is approximately 6°C. In March and April, temperature distributions are not parabolic distribution, the temperature difference along the longitudinal direction of the tunnel becomes smaller and smaller, the temperature of the surface of lining almost linearly increases from the east portal to the west portal with the gradient of 0.0012°C/m in March, and in April, the temperature along the tunnel length is almost equal.

Fig. 11 Lining temperature along the tunnel longitudinal direction

5.2 Inclination Monitoring Results In the case of rocks surrounding ground, there are gaps and spaces remained

between the lining and surrounding rock, such gaps or spaces are caused by uneven excavated face and primary support which consists of shotcrete, rock bolts and lattice girders. For the rock tunnel in seasonal cold region, the water filling in the gaps and spaces will be frozen up in cold seasons; the frost-heave pressure will come into being due to the frost swelling confined by rock and lining (Wang 2004). Moreover, the temperature difference over a year and at different cross-sections will result in temperature stress in tunnel lining (Luo 2010). The lining of the tunnel in seasonal cold region will deteriorate under the frost-heave pressure, freeze-thaw cycles effects (Tan 2013, Feng 2015) and temperature stress, and thus, the lining is more likely to be deformed. The inclination of the lining can reflect the entire deformation and evaluate the safety of the tunnel to some extent.

The results for 5 months from MS1 section are shown in Fig. 12; the inclination data have been compensated. The data shows the change in the readings from the WSN tilt nodes since they were installed. The location of the WSN tilt nodes is shown in Fig. 13. As the Fig. 12 shows, three installed WSN tilt nodes do measure the inclination change of the lining, but the inclination change values are no more than 0.1˚. Fig. 13 shows the movement direction (the dashed arrow: rotation direction at previous stage,

-20

-15

-10

-5

0

5

10

0 0.5 1 1.5 2 2.5 3

Tem

per

atu

re (˚C

)

Distance to west tunnel portal (km)

Dce.Jan.Feb.Mar.Apr.

the solidnumbersmeasuremeasuremeasure

F

Thmovemedeterminwhich ex

-0-0-0-0-0-0-0-0000000

∆In

cli

na

tio

n (˚)

d arrow: ros beside ted by noded by noded by node

F

ig. 13 Mov

he movement trends ne the exaxactly is th

0.160.140.120.100.080.060.040.020.000.020.040.060.080.10

otation direthe nodes)e A18C ade A3BE e A38B hav

Fig. 12 Inc

vement me

ments for aof differe

act mechane future wo

ection at la) from thend node Ahave rota

ve not rota

clination mo

easured of

all other Ment MS senism of tunork of this

ater stage)e installatioA3BB haveated in twated.

onitoring re

MS1 sectio

MS sectioections arennel movestudy.

Date

) and the ion time to e rotated

wo differen

esults of M

on (View fr

ons are sue different.ement from

nclination 30th April

in one dirent direction

MS1 section

rom Chang

ummarized At prese

m the readi

change val, 2017. Tection, whns, and th

n

g Chun sid

d in Fig. ent, it is dings taken

A18C

A3B

A38B

A3B

alue (the he lining

hile lining he lining

de)

14. The ifficult to to date,

C

E

B

B

Fig 6. CONC Ththe hardnode, thwirelessfeaturesadvanta

In monitorisystem nodes aregion ohardwarconnectthe radiogateway

g. 14 Summ

CLUSION

his paper pdware nodhe WSN cs protocol s of this Wge of ultraorder to

ng for a higin seasona

and 8 VW of China. Ure failure. ivity betweo connectivy is an issu

mary of the

resents a des of the communicastandard

WSN system-low powe

realize tghway tunal cold reginterface

Up to now, The data

een the novity betwee

ue even tho

e movemen

developedsystem mation protoand operam are intror consumpthe real-tinel and deion. A WSnodes wasthe WSN loss ratio

odes and Wen the nodough there

nts measu

WSN systainly consocol is deates in theoduced brption and aime, autoemonstrateSN system s deployedsystem hao is propoWSN gatewes far awaare many

ured (View

tem for strusist of the esigned bae ISM 2.4iefly. The W

artificial intematic, lon

e the applicincluding a

d in Laoyeas been oposed to eway. By ca

ay from therelay node

from Chan

uctural inclWSN gate

ased on thGHz radioWSN systelligence seng-term ancability of tha WSN gaeling tunneperating novaluate thalculating t WSN gate

es with the

ng Chun si

lination moeway and he IEEE o band. Ttem posseelf-organiznd remotehe present

ateway, 28 el in seasoormally withe quality the data loeway and tinterval of

de)

onitoring, WSN tilt 802.15.4 he basic sses the

zing. e health ted WSN WSN tilt

onal cold hout any of radio

oss ratio, the WSN f 110m or

120m. The maximum data loss ratio is 54.98%, and the data loss ratio of 36.1% of all nodes (13 nodes) is more than 5%. For the long tunnel monitoring, more than one WSN gateway should be installed in site besides the installation of relay nodes to reduce the quantities of the dropped data packets. The WSN system uses multi-hop routing to transmit the data packets; the communication between nodes is quite busy and complicated. Sometimes, all the nodes were online, which can insure that all the data packets can be transmitted to the WSN gateway, while some nodes were offline due to the influence of some external conditions such as passing of vehicles. In general, the WSN communication protocol makes nodes to communicate directly to the WSN gateway if possible, consequently, some expected relay nodes closed to the WSN gateway can’t relay the data packets frequently. Thus, it is not necessary to deploy the relay nodes near the WSN gateway.

A certain quantity of data has been measured by the deployed WSN system for almost 5 months. According to the monitoring results, the temperature distribution of the surface of lining follows the parabola distribution along the longitudinal direction of the tunnel in cold seasons, the temperature increases from the two sides of the portal to inside of the tunnel. The temperature monitoring results can be used for the mechanical analysis as the boundary conditions.

The compensated inclination monitoring results do indicate movements of the tunnel lining even though the magnitudes of the inclination change value are no more than 0.1˚. The magnitudes and the rotation direction of all five monitoring sections are different as the movement of the lining is complex. In the future, the mechanism of tunnel movement will be studied; evaluating the health condition through the inclination of the lining is also a valuable work.

ACKNOWLEDGEMENTS

This work was supported by the Transportation Science and Technology Program

Research Project of Jilin Province (2015/1/18) and Natural Science Foundation Committee program (51278381). The support is gratefully acknowledged. The authors want to thank Tao Chen and Xiaoyan Huang for their support installing the devices in site and thank Mr. Yongguang Wang for the field data collecting.

REFERENCES Bennett, P.J., Kobayashi, Y., Soga, K. and Wright. P. (2009), “Wireless sensor network

for monitoring of underground tunnel”, Proceedings of the ICE - Geotechnical Engineering.

Bennett, P. J., Soga, K., Wassell, I., Fidler, P., Abe, K., Kobayashi, Y., and Vanicek, M. (2010). “Wireless sensor networks for underground railway applications: case studies in prague and london.” Smart Struct. Syst., Vol. 6(5-6), 619-639.

China meteorological data service center, CMDC. “China’s surface climate map (1971-2000). Available online: http://data.cma.cn/en/?r=data/weatherBk.

Chung, H-C., Enomoto, T., Shinozuka, M., Chou, P., Park, C.,Yokoi, I. and Morishita, S. (2004). “Real-time visualization of structural response with wireless MEMS sensors.” Proceedings of 13th World Conference on Earthquake Engineering.

Feng, Q., Wang, G., and Jiang, B.S. (2015). “Analytical method for thawing analysis of surrounding rock in seasonal cold region tunnels.” (In Chinese) Chin. J. Geotech. Eng., Vol. 37(10), 1835-1843.

Ha, D.W., Park, H.S., Choi, S.W., and Kim, Y. (2013). “A wireless mems-based inclinometer sensor node for structural health monitoring.” Sensors, Vol. 13(12), 16090-16104.

Joo, J.W., and Choa, S.H. (2007). “Deformation behavior of MEMS gyroscope sensor package subjected to temperature change.” IEEE Trans. Compon. Packag. Technol., Vol. 30(2), 346-354.

Lai, Y.M., Wu, Z.W., Zhu, Y.L., He, C.X., and Zhu, L.N. (1999). “Nonlinear analysis for the coupled problem of temperature and seepage fields in cold regions tunnels.” Cold Reg. Sci. Technol., Vol. 29(1), 89-96.

Luo, Y.B., Chen, J.X., Qiao, X., and Wang, M.S. (2010). “Mechanics state analysis of secondary lining concrete structure on tunnel based on temperature effect.” (In Chinese) China. J. Highw. Transp., Vol. 23(2), 64-70.

Lynch, J.P., Kiremidjian, A.S., Sohn, H., and Farrar, C.R. (2003). “Field validation of a wireless structural monitoring system on the alamosa canyon bridge”, Proceedings of SPIE - The International Society for Optical Engineering, 5057(215), 267-278.

Lynch, J.P., Wang, Y., Law, K.H., Yi, J.H., Chang, G.L., and Yun, C.B. (2005). “Validation of a large-scale wireless structural monitoring system on the geumdang bridge.” Proc.of Int.conf.on Struct.safety & Reliability.

Nagayama, T., Spencer Jr., B.F., and Rice, J.A. (2009). “Autonomous decentralized structural health monitoring using smart sensors.” Struct. Control. Health. Monit., Vol, 16(7-8), 842–859.

Spencer Jr., B.F., Ruiz-Sandoval, M. and Kurata, N. (2004). “Smart sensing technology for structural health monitoring.” Proceedings of the 13th World Conference on Earthquake Engineering.

Straser, E. G. & Kiremidjian, A. S. (1998). A modular wireless damage monitoring system for structures. Ph.D. Thesis, Dept. of Civil and Environmental Engineering, Stanford University, Stanford, CA.

Tan, X.J., Chen, W.Z., Wu G.J., and Zheng, P.Q. (2013). “Study of thermo-hydro-mechanical-damage (THMD) coupled model in the condition of freeze-thaw cycles and its application to cold region tunnels.” (In Chinese) Chin. J. Rock. Mech. Eng., Vol. 32(2), 239-250.

Wang, F., Huang, H.W., He, B., Wu, Y., Shao, H., and Wu, H.M. (2016). “Wireless sensing on shield tunnels in Shanghai.” International Conference on Smart Infrastructure and Construction.

Wang, J.Y., Hu, Y.F. (2004). “A discussion on frost heave force acting on tunnel lning.” (In Chinese) J. Railw. Eng. Soc., Vol. 21(1), 87-93.

Wu, Y., Lin M. and Wassell, I.J. (2009), “Modified 2D finite-difference time-domain based tunnel path loss prediction for wireless sensor network applications”, J. Commun., 4(4), 214-223.

Yin, J.G., and Huang, H.W. (2015). “Real time monitoring method for the longitudinal settlement of shield tunnel using wireless inclinometer.” 2015 information technology and mechatronics engineering conference.

Yu, Y., Ou, J.P., Zhang, J., Zhang, C.W. and Li, L.Y. (2009), “Development of wireless MEMS inclination sensor system for swing monitoring of large-scale hook structures”, IEEE T. Ind. Electron., 56(4), 1072-1078.

Zhou, Y.W., Guo, D.X., Qiu, G.Q., Cheng G.D., and Li, S.D. (2000), Geocryology in China (In Chinese), Science Press, Beijing.