Capacity for Rail
Monitoring Technologies & Sensors identificationMigration of innovative technologies to existing structures
Madrid – 21/09/2017
Mani EntezamiResearch Fellow
Overview
2
• Sensing technologies and energy harvesting
• Technology identification and evaluation
• Sensors assessment
• System design
• Field tests
• Demonstrator
• Conclusions
Identified Technologies
Technology to be Developed
Technology Evaluation
Context
Capacity for Rail
Sensing Technologies
Sensing technologies
4
Technology Key Factors
Vision Frame rate, Resolution, Colour depth / sensitivity, Field of vision (lens angle)
VibrationRange of operation, Sensitivity, Drift / stability, Sampling rate, Power consumption,
Shock tolerance
Sensor resonance frequency
Mechanical
(Strain)
Bridge configuration, Robustness (operating environment), Target geometry,
Installation method
Environmental
(Temperature, Humidity,
Wind)
Range of operation, Robustness, Accuracy vs. cost, Size, Installation method
AcousticRange of operation (sensitivity), Frequency range, Directionality, Physical limitations
(size)
Robustness / operating principle
ElectricalRange of operation, Sensitivity, AC or DC, Sampling rate, Physical limitations (size)
Installation method, Isolation
Specialist / Multifunction
(Fibre)
Operational benefits arising from multi-modal operation (sensing and
communications)
Sensing technologies
5
Vision
Displacement of the drive rod
Camera results vs displacement sensor
Vision
High speed camera - FASTCAMMovement of the overhead lines
Eurostar @ 300 km/h
Energy harvesting and storage
6
Solar panelThermogeneratoer• Energy Harvesting• Solar • Thermal• Wind• Electromagnetic• Vibration
• Storage• Batteries • Flywheels • Supercapacitors• Hydrogen • Thermal
Wind TurbineGel battery
Supercapacitor
Example: Solar panel power depends on the location
Monthly average generated power
Capacity for Rail
Technology identification and evaluation
Technology identification framework
8
Drivers
Technology Market Place
Technology
Capability
Barriers Applicability
Enabler
Required Competency
Environmental
Technological
Social
Economic
Political
Which technologies could be developed given the Market Place?What is missing in the Market Place to progress the technologies?
What is the technology in question?
Technology identification framework
9
Drivers
Technology Market Place
Which technologies could be developed given the Market Place?What is missing in the Market Place to progress the technologies?
Technology
Capability
Barriers Applicability
What is the technology in question?
Where could the Technology be applied?
Where are the drivers for introducing the Technology?
Where are the barriers to introducing the Technology?
What capability is required to realise the Technology?
Technology identification framework
10
Applicabilityapplication, stakeholders, business case and business model
Wireless Vibration Monitoring
Driverspolitical, economic, social, technological, environmental
Barrierspolitical, economic, social, technological, environmental
Capabilityenablers, competencies and relevant stakeholders
No cablesNo trip hazardEasy to install and retrofitTrack movements
Accelerometers and gyroscopes
Signal processing
Communication
Power
Track access and possession
Power consumption
Energy harvesting environmental limitations
Switches and crossings
Track geometry monitoring
Wheelflat detection
• Testing has been undertaken at the Long Marston facility, Uk
• A variety of different grade (cost) accelerometers have been evaluated
• Cross-comparison of sensors and evaluation against geophones
• High quality (cost) sensor displays reasonable correlation
• Lower quality (cost) sensor displays significant drift – can be reduce by post processing
11
Sensor evaluation / comparison
7 7.2 7.4 7.6 7.8 8-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
Time into experiment [s]
Vert
ical speed [m
s-1]
geo2
vz2
Accelerometers
12
KS76a (Piezo) ADXL001 (MEMS)
Interface IEPE Voltage
Power ~ 132 mW < 1 mW
Range ±120 g ± 250 g
Resonant frequency > 34 kHz 22 kHz
Sensitivity 50 mV/g 4.4 mV/g
Noise 80 µg (20 – 50000 Hz) 95 mg (100 – 400 Hz)
• MEMS vs Piezo• MEMS average draw of
0.75 mW compared to Piezo of 132 mW
• MEMS Peak draw of 5 mA (1.5 mW)
Wide temp range (-40 to 85 ˚C)
Sensor Evaluation / Comparison
13
Vibration calibrator
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2-2
-1
0
1
2MMF Sensor
Accele
ration [g]
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2-2
-1
0
1
2Analog Device Sensor
Time [s]
Accele
ration [g]
Piezo
MEMS
• Lower SNR in MEMS• Good match in Freq-
domain and in displacement calculation
0 100 200 300 400 5000
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Frequency [Hz]
Am
plit
ude(f
)
100 150 200 250 300 350 400 450 5000
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Frequency [Hz]
Am
plit
ude(f
)159.2 Hz calibration signal
Piezo MEMS
• Example of sensor evaluation using the framework
SENSORS
Ref Description Au
tom
ate
d d
ata
co
lle
cti
on
De
tec
tio
n o
f in
cip
ien
t fa
ult
s
Ev
en
t lo
ca
liza
tio
n
Wa
ke
up
un
de
r e
ve
nt
Dif
fere
nt
tim
e f
or
sen
sin
g/s
en
din
g
Sc
ala
bil
ity
En
vir
on
me
nta
l c
om
pa
tib
liit
y
Da
ta c
oll
ec
tio
n a
t li
ne
sp
ee
d
Dif
fere
nt
me
asu
rem
en
t m
od
es
Cu
sto
m r
ep
ort
ing
of
pa
ram
ete
rs
Cu
sto
m f
au
lt d
ete
cti
on
ru
les
Cu
sto
m s
ub
mis
ion
ra
te o
f m
ea
sure
me
nts
Se
lf-d
iag
no
stic
Lo
ng
te
rm s
tab
ilit
y
Lo
ng
te
rm r
ob
ust
ne
ss a
nd
re
lia
bil
ity
Ca
lib
rati
on
Ge
om
etr
ica
l c
om
pa
tib
ilit
y
Co
mp
ati
bil
ity
wit
h t
rac
k m
ain
ten
an
ce
Hig
h a
va
ila
bil
ity
on
co
mp
on
en
t le
ve
l
Hig
h a
va
ila
bil
ity
on
se
nso
r n
od
e
Re
sist
an
ce
to
ele
ctr
om
ag
ne
tic
fie
lds
Mo
un
tin
g s
imp
lic
ity
SE
NS
OR
SC
OR
E
WEIGHT 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% 5%
S1 MEMS Accelerometer ADXL345 10 10 10 10 10 5 10 10 0 0 10 10 0 10 10 5 10 10 10 10 10 5 8.0
Capacity for Rail
System Design
System Design
15
Vibration based track movement motoring system using accelerometer on sleepers
➢ Low power ➢ Low cost➢ Easy installation
➢ No wires➢ Use of an energy harvesting system➢ Remote access
System design
16
Wireless Vibration Nodes
Wayside central unitWith remote connection
Short range and low power wireless technology
Use of EH for track side system
UoB wireless node system overview
17
Accelerometer
MCU + Memory
ISM Wireless Module
Temperature
Battery Voltage
ISM Wireless Module
ARM device
SD Memory
Card
WiFi / 3G module
Low-powerLow-frequencyISM band
Sleeper Node
Master Node
System components
18
➢ Data acquisition and processing• Microchip PIC24• Low cost, low power• Easy to integrate with the sensor (SPI bus)• Capability for communication modules• Suitable processing power and • Data storage• Wide temperature range (-45 to 85 ˚C)
➢ Short range wireless system• Microchip MRF89XAM8A• Low frequency ISM band, 868 MHz• Ultra low-power• SPI Interface with Interrupts• Small size: 18 mm x 28 mm• Wide temperature range (-45 to 85 ˚C)
Cost about £3
Cost about £5
COMMUNICATIONS
Ref Description Fast
dat
a tr
ansm
issi
on
Wir
ele
ss c
om
mu
nic
atio
n
Stan
dar
d i
nte
rfac
e f
or
wir
ele
ss
Ind
ust
rial
Eth
ern
et
com
mu
nic
atio
n
Cu
sto
m m
ess
age
fo
rmat
Tim
e s
yncr
on
izat
ion
CO
MM
UN
ICA
TIO
N S
CO
RE
17% 17% 17% 17% 17% 17%
C1 MRF89XAM8A 5 10 10 0 10 10 7.5
Wireless module evaluation
Central unit and remote connection
19
• Low power and low cost• Use of energy harvesting• Remote connection• Storage system• Weatherproof• Easy integration into final product: minimise product
development, quicker time to market• Capable of using the selected wireless technology
• Raspberry PI 3A➢ Quad core ARM8 processor➢ Serial communication bus➢ Storage – SD card➢ Network connections➢ Linux OS - Open source➢ Small and easy to mount➢ Low power (<3W)
• GPS module – Time sync• Short range wireless • IP67 Enclosure• Operation range
– (-40 to 85 ˚C) without LAN– (0 to 60 ˚C) with LAN, used only for
remote access
£26
£5
£10
All together with enclosure ~ £50
Energy systems
20
ENERGY HARVESTING
Ref Description Suit
abil
ity
for
inst
alla
tio
n a
t d
iffe
ren
t si
tes
Mo
nit
ori
ng
and
re
po
rtin
g o
f b
atte
ry s
tatu
s
Self
-dia
gno
stic
Envi
ron
me
nta
l co
mp
atib
lity
Re
sist
ance
to
ele
ctro
mag
ne
tic
fie
lds
Mo
un
tin
g si
mp
lici
ty
ENER
GY
HA
RV
ESTI
NG
SC
OR
E
Weight 17% 17% 17% 17% 17% 17%
E1 Solare panel BP SX20U 5 5 0 5 10 10 5.8
• 50 cm automotive solar panel (used in traffic lights)
• Up to 20 W power• Easy to install on the track-
side
• Wide operating temperature range
• Resilient unit, does not require further housing / protection
UoB sleeper node
21
• Easily deployable networks of sensors
• Internal accelerometer• ‘Sleeps’ until a train is detected• Samples at 1600 Ss-1
• Downsamples to 800 Ss-1
• Stored in local memory• Transmitted to master node
after train has passed• Battery powered
• ~5 years (3A Lithium Iron) • EH for local master node
• Includes a temperature sensor • Wide temperature range to
operate (-20 to 60 ˚C)Around £20 for a prototype
Capacity for Rail
Field Testing / Demonstration Activities
UoB - Live trial initial tests
23
• Monitoring sleepers on the UK High Speed 1 line using low power accelerometers and embedded microcontrollers – Eurostars, 300 km/h
– Javelins, 220 km/h
– Freight, 100 km/h
• Around 1400 train passages were recorded over a 2 week period
Data analysis - Accelerometers
24
• Displacement curves for the three accelerometers
• One is significantly larger than the other two
Less-well supported sleeper
Sensor evaluation
25
10 kS/s
0.8 kS/s
UoB & UPorto - Live trial demonstration testing
26
• Transition zone onto a bridge
• Use of EH at trackside
• Battery powered nodes
• Short range wireless system
• LTE for remote access
Alcácer do Sal
27
4x Battery power Wireless nodes
Vibration node with wireless power, Solar and battery (UPORTO)
Energy harvesting~120W
Central Units
Alcácer do Sal
28
Main control panel
4G Router
UPorto PU
Gel Batteries and Regulating system
UPorto Vibration Node UoB Vibration Node
UoB Central PU
29
Alcácer do Sal
UoB - Recording on April 20 at 10:54am
Results
30
• Vibration data in m/s2
• Different vibration behaviour
• Different vibration level
• Interesting peaks on
Node 2
Data available live on UoB webserver
• Displacement in mm
• Using a bandpass filter and double integration for displacement
• Consistency in number of wheels passed
• Consistency in the displacement level
Data Analysis
31
• Freight train – Peak analysis
• Crest factor spectrom
• Load monitoring and wheel impact
System performance
32
• Waking up every 200ms
• Log maximum 11 seconds if there is a train
• Estimated to work for two months at 25˚C
• Replaced battery after 40 days
• Difference between industrial and commercial batteries
• Triggering to be improved
System performance
33
• Temperature effect on the batteries of the nodes
• No significant change in battery voltage within a month
• Temperature monitoring
• Max temp about 60˚C– Lost network connection
• Data stored locally
• Connection retrieved after a couple of weeks
Instrumentation plan – NR HS, UK
34
➢ 16 wireless nodes➢ 14 battery powered➢ 2 solar panels and battery
➢ Vibration and temperature monitoring➢ Lateral acceleration - 4 nodes➢ Vertical acceleration – 12 nodes
35
Network Rail High Speed - Kent
• Location identified by NR HS• Ballast issues• Nodes installed • Solar panel • Access for UoB staff visit to
be granted
Picture provided by NR inspection Helicopter
Wireless nodes
Solar panel
DB Demonstrator
36
• Mesh network• ~250 sensors – battery
powered• Three gateways • Solar panel for the gateways• Sensor grid within the rail
expansion • Tilt-sensors on the cess and
four-foot sides of the sleepers
Conclusions
37
• Sensing technologies introduced including their key parameters for evaluation
• Field test demonstrations of a range of technologies– Extends laboratory testing– Demonstrates integration of technologies
• Monitoring system design comprising of:– Low cost system– Energy harvesting – Low power electronics and sensing technologies– Short range wireless methods and LTE networks– Distributed data collection and processing network– Data processing and condition monitoring techniques
• Demonstrates interactions / trade-offs between technologies in whole system development
Thank you for your kind attention
Mani Entezami Research Fellow
The University of Birmingham