Data acquistion principles.pptx

8/18/2019 Data acquistion principles.pptx

1/138

Training Course on VibrationAnalysis Level-II

Centre for Vibration Analysis & Machine Condition

Monitoring (CVCM)[email protected]

h!"#$-%&'("&$

)a*!"#$-%&$+$$'

Cell!",""-+#($"("

by

mailto:[email protected]:[email protected]


2/138

VIBRATION MEASURING INSTRUMENTS VIBRATION MEASURING INSTRUMENTS

OVERVIEW


3/138

Choosing Your InstrumentationChoosing Your Instrumentation

What do you want to achieve?

What is your present and future budget for equipment & training?

Person power? Knowledge level?

Number of machines to be monitored?

Type of machines to be monitored?

Environment?


4/138

INSTRUMENT TYPES INSTRUMENT TYPES

verall !evel "eters

#uic$ %hec$ naly'ers

((T )ata %ollector* naly'ers

(ull (eature naly'ers

+eal Time ,pectrum naly'ers

-nstrument #uality Tape +ecorders

)edicated .alancing instruments


5/138

Data AcquisitionPrinciples

&

Signal Processing


6/138

Data Acquisition System

Raw Signal

Amp

AC

Output

Integrator1x, 2x

High Pass

ilter

!ow Pass

ilter

"C

Output

Amp "ete#torP$P or R%S

"ispla&

Rea'ing

A##elerometer


7/138

Data Acquisition System

/ ensortrans/ucer for measurement of 0hysical

variables.

/ ignal-Con/itionertransmission circuitry1 thatenables conversion of signal out0uts from

trans/ucers to a rea/able form for 2ataAc3uisitioninterface mo/ules.

/ The 2ata Ac3uisition 4ar/5are com0rising

6ulti0le*er1 Am0li7er1 A2 Converter1 8u9er6emory1 etc. to /igitize the analog signals for C:.

/ Com0uterC: to 0rocess the /igital /ata for /ata0rocessing1 /is0lay1 out0uts ;control


8/138

Selection of ransducers


9/138

!hat is ransducer"

/ A trans/ucer is a /evice 5hich transforms a non-

electrical 0hysical 3uantity ;i.e. tem0erature1soun/ or light< into an electrical signal ;i.e.voltage1 current1 ca0acity=<

/ In other 5or/ it is a /evice that is ca0able ofconverting the 0hysical 3uantity into a

0ro0ortional electrical 3uantity such as voltageor current.


10/138

Selection of ransducer

/ Sensiti#ity$ The trans/ucer must be sensitive enough to

0ro/uce /etectable out0ut. / %perating ange$ The trans/ucer shoul/ maintain the

range re3uirement an/ have a goo/ resolution over theentire range.

/ Accuracy$ it is /e7ne/ as ho5 close the out0ut of the

trans/ucer is to the e*0ecte/ value. 4igh accuracy isre3uire/.

/ Si'e$ 2e0en/ing on the a00lication of the trans/ucer1 thesize may be of 0rimary im0ortance

/ n#ironmental Compatibility$ the trans/ucer is

selecte/ base/ on the various environmental con/itions itcan 5ork.

/ nsensiti#ity to un*anted signals$ The trans/ucershoul/ be minimally sensitive to un5ante/ signals an/highly sensitive to /esire/ signals.

/ Cost$ The cost of a trans/ucer is an im0ortantconsi/eration1 es0ecially 5hen many sensors are nee/e/


11/138(a)le* (rans'u#er Chara#teristi#s


12/138

Vibration ransducers

Sensors+ransducers+Probes+!hat is it"

+,t basically con#erts mechanical

#ibration to an electrical signal

AccelerometerCharge Ty0e >Line 2rive

Constant Voltage >Constant Current

Velocity Probe Displacementhaft ?i/ers

ro*imity robes;//y Current robes<


13/138

Vibration ransducer ypes

/ eismic!- 8earing relative to s0ace.

0 Velocity icku0s

0 Accelerometers

0 iezoelectric velocity 0icku0s

/ ?elative!- haft relative to bearing.0 on-contact //y Current 2is0lacement robes

/ Absolute!- haft relative to s0ace.

0 haft Contact 2is0lacement robes ;inclu/ing haft

ticks an/ haft ?i/ers<


14/138

Velocity Sensors


15/138

Accelerometers

D sp acement Pro es non contact e y


16/138

D sp acement Pro es non-contact. e ycurrent probes)


17/138

Seismic ransducer

V/%C0 PC12P


18/138

Velocity Pic3ups

ADVANTAGES

/ Self 4enerating 5 no po*er supply required / Magnet inside coil generates #elocity

proportional to #ibration

/ Spring mass system

/ 67 8', to 6777 8', / Phase change 977

/ Directional mounting

/ /arge & 8ea#y

/ %utput : mV;inch;sec

/ !ide range of a#ailable outputsote !- There are t5o ty0es of velocity 0icku0s the above

a/vantages /o not a00ly to 0iezoelectric velocity

trans/ucers.


19/138

Pie'oelectric Velocity Pic3up

ADVANTAGES

/ ?emember everything that

you Bust learne/ about an

accelerometer.

/ The out0ut of the

accelerometer has been

integrate/ to velocity an/ has

a %"" 0hase change

/ $"" mVinchsec ;'

mVmmsec<

/ #"" mVinchsec ;&"mVmmsec<

SSMC A


20/138

SSMC A


21/138

Accelerometers - ad#antages– o moving 0arts1 no 5ear.

– ?ugge/.

– Very large /ynamic range.

– i/e fre3uency range.

– Com0act1 often lo5 5eight.

– 4igh stability.

– Can be mounte/ in any orientation.

– 6easures casing vibration

– 6easures absolute vibration

– Integrate to Velocity

– asy to mount

– Large range of fre3uency res0onse

– Available in many con7gurations


22/138

Accelerometer ypes

The three most common are :-

– Com0ression Ty0e

– Inverte/ Com0ression Ty0e

– hear Ty0e

C i t l t


23/138

Compression type accelerometer

lectric connector

Seismic Mass

Preload Stud

Acoustic Shield

Pie'oelectric Material

CP Ampli=erMounting Studeceptacle

>ase

C i A l t


24/138

Compression ype Accelerometers

Advantages

0 ?elatively lo5 cost

Disadvantages

0 ensitive to base strain

0 ensitive to Thermal transients

0 Can cause over-saturation an/

trans/ucer settling 0roblems

Wide! "sed


25/138

n#erted compression type

Pie'oelectric MaterialCP Circuit

Mounting stud receptacle

Seismic Mass

Preload Slee#e

Sh A l t


26/138

Shear ype Accelerometer

lectric connector

Seismic Mass

Post

Acoustic ShieldPie'oelectric Material

CP CircuitMounting Studeceptacle

>ase

Ad#antages Shear ype


27/138

Ad#antages - Shear ype

Advantages

0 Lo5er sensitivity to base strain

0 Large /ynamic range

0 6uch less sensitive to tem0erature transients

0 tabilizes 3uickly 5hen taking measurements at lo5

fre3uencies.Disadvantage: -

0 Denerally higher cost /ue to a//e/ com0onents


28/138

Parameters;Speci=cations

y u y


29/138

y u yesponse

requency esponse oun ng


30/138

requency esponse oun ngechnique-6


31/138

S e n s i t i #

i t y

?req,

StudMount

8andProbe

Dual ail

Magnet

?latMagnet

MountingPad

6,@18' 6718' B18'

echnique (B)

u y u


32/138

u y uechnique ()


33/138

What is the #re$"enc! range o# !o"r%

&nstr"ment%Ca'es

%Sensor %Sensor Co"(ing

!hat is the fault frequency you are loo3ingfor "

Sensor freq, :6B 18' nstrument

freq, : 718'

Cablelength "

Sensorcoupling "

.Application

?requency ange


34/138

+'-

2H.

2

1

$1

1/H.

R e l a t i 0 e S e n s i t i 0 i t &

seul re3uen#& Range14 limit 5+ +'- limit 56

$2

$+

$7

re3uen#& 8x 9

re3uen#& Response o Sensor

?requency ange


35/138


36/138

PC12PC%/

MA4


37/138

a*ial shaft 0osition.

– 2i9erential e*0ansion bet5een case an/ rotor.

specially eecti#e on machinery *ith high mass rigid casingsand relati#ely lo* mass

rotors supported in Eournal type bearings,


38/138


39/138

a on ac sp acemen ro es(Absolute)

– haft ticks

– 4ar/5oo/1 7sh-tail1 7*e/ to

accelerometer or velocity 0icku0

– 6easures vibration am0litu/e >

0hase– haft ?i/ers

Sh ft id


40/138

SHA( SRACE

:O:$%E(A!!IC (IP

%ACHI:E HOSI:;

SHA( RI"ER ASSE%-!<

PIC/P %O:(I:; S("

Shaft ider

Direct Contact $ Absolute Measurements


41/138

Direct Contact $ Absolute Measurements

– haft ?i/ers ;0ermanently installe/<

– haft ticks or )ishtails

0 safety issue

0 very useful belo5 cou0ling of vertical 0um0s

ypical 2ses of Vibration ransducers


42/138

?a/ial CasingVibration

?a/ial haftVibration > osition

A*ial haft

Vibration > osition

ypical 2ses of Vibration ransducers

Measurement Parameter


43/138

Measurement Parameter– )in/ the FGattestH

s0ectrum ormally

velocity is use/.

– )or very slo5 running

machine ; ("" ?6<

/is0lacement is0referre/.

–

)or 4igh fre3uency/iagnostics use

acceleration

– Al5ays use

acceleration for

A##eleration

Velo#it&

"ispla#ement

Monitoring echniques


44/138

Monitoring echniques

A##eleration

Velo#it&

"ispla#ement

Displacementaccentuates LE fre3uencies1an/ attenuates 4ID4 fre3uencies.

VelocityFGatH treats all fre3uencies e3ually.

Accelerationaccentuates 4ID4 fre3uencies1an/ attenuates LE fre3uencies.

re35

Vi)5

Comparison of ransducers


45/138

– asy to install

– 4ood for detecting high

frequency faults

–


46/138

;ear

-la'es

Rolling Element

-earings

Shat

Rotating

Spee'

2x

+x

=ournal

-earings

insta)ilit&

1 /H. +/H. 26/H.

:on Conta#t "ispla#ement

Velo#it& Pro)e

A##elerometer

Vibration Pic3ups


47/138

(&pes o -earings

=ournal -earings

• Stationar& Signals• Relati0e !ow re3uen#&

=ournal -earings

• Stationar& Signals• Relati0e !ow re3uen#&

Rolling Element -earings

• %o'ulate' Ran'om :oise

• Pulsating signals• High re3uen#&

Rolling Element -earings

• %o'ulate' Ran'om :oise

• Pulsating signals• High re3uen#&

se Proximit& pro)es

se A##elerometers

Monitoring TechniquesMonitoring Techniques


48/138

Transucers an Mounting TechniquesTransucers an Mounting Techniques

lthough there are many different types of transducers available1 the most common

type used for day to day data collection are ccelerometers2 These transducers provide an electrical charge proportional to acceleration by

stressing pie'oelectric crystals typically 344m5*g sensors are used2


49/138

!ata "ua#it$ !ata "ua#it$

Whether it is your 6ob to collect the data and*or analyse the data it is important to

understand that the technologies will not give you the answer to a machinesproblem unless you have collected meaningful1 quality data

There are certain considerations that must be ta$en prior to any data being

collected1 these are7

good understanding of the internal ma$e up of the machine1 in order to

understand the best transmission path for data collection 8 bearing locations1

load 'ones etc2

Ensure data is collected in a repeatable manner so we can compare two or more

readings to each other 8 trending purposes

5ariable speed machines 8 it is very important to collect data with the correct

running speed enter into the analyser


50/138

Transmission PathTransmission Path

)amaged caused to a machine component will cause a certain amount of

vibration*sound or heat to propagate away from the initial impact2 -t is the effect of the impact*force that we are trying to detect

-n many cases the further you are away from the initial event the wea$er the signal

will become1 resulting in the data appearing to be lower in value2

-n more e9treme cases the impact can be lost amongst other machine noise by the

time it has reached your transducer1 resulting in no detection of a machine problem2

:sually the best place to acquire data from a machine1 is at the bearings2

/ This is because the bearings are the only part of the machine that connect

the internal rotating components to the stationary components ;%asing<


51/138

Re%eatae !ataRe%eatae !ata

%ollect data in the same manner each time2

This consistency will allow you to trend the machinery condition and properly 6udge the

progression of faults

-n order to aid with repeatable data the analyser requests for data to be collected in

certain locations on the machine2

These are called ="easurement Points>

A measurement point is 'etermine' )& three #hara#ters an' a 'es#ription5 Ea#h #hara#ter reers to a parti#ular pla#e on the ma#hine )eing monitore'

–E5g5 %1H is a t&pi#al measurement point


52/138

Measurement PointsMeasurement Points

measurement point is defined as three alpha numeric digits along with their

respective definition – Orientation and location on each component

The image on the right is ta$en

from the screen of the 3@4

analyser during a collection

=route>

The measurement =point

identifier> can be seen in the top

right while the =point description>

is shown 6ust below


53/138


The first letter of the =Point -dentifier> refers to the type of machine being monitored

– " A "otorP A Pump( A (an

The second character represented by a number indicates the location on the machine

– -nboard ;)rive End< or utboard ;Non )rive End<

The third letter refers to the orientation of the sensor or the type of processing being

done by the analyser – B A Bori'ontal 5 A 5ertical P A Pea$vue %hange in ),P of nalyser


54/138


The following e9ample shows how the numbering system changes as you cross from

one component to the ne9t

Notice how the =3> is not always the =utboard>

This changes when the ne9t component is required for data collection

The numbering system starts from 3 again

12

1

2

"3B / "otor utboard Bori'ontal

"3P / "otor utboard Bori'ontal Pea$vue

P3B / Pump -nboard Bori'ontal

P3P / Pump -nboard Bori'ontal Pea$vue


55/138

'ocating Turning S%ee 'ocating Turning S%ee


56/138

T i S


57/138

BFI - Example 6

Ex6 -P2V P UMP OUTBOARD VERTICAL

Route Spectrum

!-"a#-$6 %&'%('&%

OVRALL) %*&2 V-D+

RMS ) %*(6LOAD ) %!!*!

RPM ) 2$,*

RPS ) ($*,

! 2! (! 6! ,! %!! %2! %(! %6!

!

2

(

6

,

%!

%2

Fre.ue#c/ 0# 1CPM

R M S V e l o c 0 t / 0 # m m 2 S e c

BFI - Example 6

Ex6 -P2V P UMP OUTBOARD VERTICAL

Route Spectrum

!-"a#-$6 %&'%('&%

OVRALL) %*&2 V-D+

RMS ) %*(6LOAD ) %!!*!

RPM ) 2$,*

RPS ) ($*,

! 2! (! 6! ,! %!! %2! %(! %6!

!

2

(

6

,

%!

%2

Fre.ue#c/ 0# 1CPM

R M S V e l o c 0 t / 0 # m m 2 S e c

Fre.'

Or3r'

Spec'

2*$,

%*!!!

*%%&

Turning S%ee Turning S%ee

The spectrum is showing numerousimpacts appearing at different

frequencies2

.y locating the turning speed1 it is veryclear that the impacts are Non8

synchronous


58/138

Ana#$sis Techniques Test Ana#$sis Techniques Test

Bave a loo$ at the spectrum below2

– Where was the data ta$en?

When the turning speed has been

located

– What type of energy is

present?

Lo4 - Example

E5 -P2V Pump Outoar3 Vert0cal

A#al/7e Spectrum

%&-8o9-$& %!'!!'%6

RMS ) %*2LOAD ) %!!*!

RPM ) *

RPS ) %2*2,

! 6!!! %2!!! %,!!! 2(!!! !!!!

!

!*2

!*(

!*6

!*,

%*!

Fre.ue#c/ 0# CPM

R M S V e l o c 0 t / 0 # m m 2 S e c

Lo4 - Example

E5 -P2V Pump Outoar3 Vert0cal

A#al/7e Spectrum

%&-8o9-$& %!'!!'%6

RMS ) %*2LOAD ) %!!*!

RPM ) *

RPS ) %2*2,

! 6!!! %2!!! %,!!! 2(!!! !!!!

!

!*2

!*(

!*6

!*,

%*!

Fre.ue#c/ 0# CPM

R M S V e l o c 0 t / 0 # m m 2 S e c

Fre.'

Or3r'

Spec'

6*,6

%*!!!

*2(&

P2V

Synchronous Energy


59/138

Things to Remem&er a&out a RouteThings to Remem&er a&out a Route

route includes information from one area only

route does not have to include all the equipment defined in that area The order of the equipment in the route can differ from that of the database Equipment can appear in more than one route .:T can not appear in the same route

twice +oute measurement points may not include all the points configured on the

equipment +oute measurement points do not have to be in the same order as they appear in the

database

)ata is not stored at the route level but in the database with the measurement points1there for routes can be deleted but will not loose data from the database ma9imum of C4 routes can be stored to each area Each equipment has a ma9imum of 3DD points nd one route can only contain 34DD measurement points

-mportant7 route file contains the equipment and measurementpoint -)>s and definitions * speeds2 (or this reason the route doesnot recognise points if their -)>s are altered in the database


60/138

(requenc$ Bans(requenc$ Bans

• "i0i'e spe#trum in re3uen#& )an's )ase' on thet&pes o me#hani#al aults that might appear on thema#hine

%5

25

5- 65

BEARI8+ BA8D %BEARI8+ BA8D 2

$-!5 RPM

!-&!5 RPM


61/138



%5

25

5- 65


$-!5 RPM

!-&!5 RPM

Im)alan#e


62/138



%5

25

5- 65


$-!5 RPM

!-&!5 RPM

Im)alan#e

%isalignment


63/138



%5

25

5- 65


$-!5 RPM

!-&!5 RPM

Im)alan#e

%isalignment

!ooseness

-earing -an' 1-earing -an' 2


64/138



%5

25

5- 65


$-!5 RPM

!-&!5 RPM

Im)alan#e

%isalignment

!ooseness

-earing -an' 1


65/138



%5

25

5- 65


$-!5 RPM

!-&!5 RPM

Im)alan#e

%isalignment

!ooseness

-earing -an' 1-earing -an' 2


66/138

Digita Signa+rocessing

Analog Signals


67/138

Anaog signas:

/ 2irectly measurable 3uantities in terms ofsome other 3uantity.

E,am(es:

/ Thermometer – mercury height rises as temperaturerises.

/Car Speedometer – Needle moves farther right as youaccelerate.

/ Stereo – Volume increases as you turn the knob.

Digital Signals


68/138

Digita Signas:

/ 4ave only t5o states.

/ )or /igital com0uters1 5e refer to binary states1" an/ $. F$H can be on1 F"H can be o9.

E,am(es:

/ Light s5itch can be either on or o9.

/ 2oor to a room is either o0en or close/.

Famples of Analog to Digital


69/138

Applications / Micro(hones - take your voice varying

0ressure 5aves in the air an/ convert them into

varying electrical signals.

/ Strain Gages - /etermines the amount ofstrain ;change in /imensions< 5hen a stress is

a00lie/. / / Thermoco"(e J tem0erature measuring

/evice converts thermal energy to electricenergy.

/ Votmeters

/ Digita M"ti-meters


70/138

(ime Signal

A)solute Vi)rationwith ree$Spa#e

Machine Vi&ration Signa# Machine Vi&ration Signa#


71/138

AC Signal

"C

Signal

Relati0e Vi)ration with

mounting position o Prox5 Pro)e

Machine Vi&ration Signa# Machine Vi&ration Signa#


72/138

Pea>

Pea>

to

Pea>

R%S

A0g

Alwa&s as>5555

Are &ou measuring R%S or Pea> , et# ?? What is the re3uen#& range ??

How mu#h a0eraging?

re35 @ 1(ime

re35 @ H.

@ re05 per se#on'

%a#hine re35 are un#tion o RP%

i5e5 re05 per minute

Ban%ass Measurement Ban%ass Measurement

! t t


73/138

R%S

(rue pea> $ pea>

AT

a t dt RMS

T

= ∫ !

"

# $

A A peak RMS =

! %

or Sine wa0es onl&*

a

( @ a0eraging perio'

R%S

A peak

A peak peak −

A peak

A peak peak −

!etector !etector

M hi Si # TM hi Si # T


74/138

Stationar& Signals

:on $ Stationar& Signals

– V0rat0o# :rom rotat0#; mac


75/138

Converts anaog signas into 'inar! ords

/ Guanti'ing$ breaking /o5n analog value is aset of 7nite states.

/ ncoding$ assigning a /igital 5or/ or numberto each state an/ matching it to the in0utsignal.

Guanti'ing-6


76/138

E,am(e:

/ Kou have "-$"Vsignals.

/ e0arate them

into a set of/iscrete states

5ith $.V

increments. / ;4o5 /i/ 5e get

$.V

/ ee ne*t sli/e=

Eut0uttates

2iscrete Voltage?anges ;V<

" ".""-$.

$ $.-&.#"

& &.#"-,.M#

, ,.M#-#.""

' #.""-(.

# (.-M.#"

( M.#"-+.M#

M +.M#-$"."

Guanti'ing-B


77/138

The n"m'er o# (ossi'e states that theconverter can o"t("t is:


78/138

/ 4ere 5e assign

the /igital value

;binary number< to

each state for the

com0uter to rea/.

utput,tates

utput .inary Equivalent

4 444

3 443

434

@ 433

D 344

C 343

334F 333

Accuracy of A;D Con#ersion


79/138

/ There are t5o 5ays to best im0rove

accuracy of A2 conversion!

0 increasing the resolution 5hich im0roves the

accuracy in measuring the am0litu/e of the

analog signal.

0 increasing the sam0ling rate 5hich increases

the ma*imum fre3uency that can be

measure/.

esolution


80/138

/ ?esolution ;number of /iscrete values the converter can

0ro/uce< N Analog Ouantization size ;O


81/138

Sampling Rate: The sampling rate #S&$ is the rate at 'hich amplitude values are

digiti(ed from the original 'aveform or )re*uency at 'hich +,C evaluates

analog signal.

Continuous analog signal from a sensor Analog signal sampled at discrete intervals

Sampling ate (contHd)


82/138

/ Sampling Rate: the sampling rate #S&$ is the rate at 'hich amplitude

values are digiti(ed from the original 'aveform or )re3uency at

5hich A2C evaluates analog signal.

/ As 5e see in the secon/ 0icture1 evaluating the signal more

often more accurately /e0icts the A2C signal.

Aliasing


83/138

/ Aliasing Eccurs 5hen the in0ut signal is

changing much faster than the sam0le rate.

/ )or e*am0le1 a & k4z sine 5ave being sam0le/

at $.# k4z 5oul/ be reconstructe/ as a #"" 4z

;the aliase/ signal< sine 5ave.N!$"ist R"e:

/ :se a sam0ling fre3uency at least t5ice as high

as the ma*imum fre3uency in the signal toavoi/ aliasing.

?requency ?iltering


84/138

*iters

/ A 7lter is a function that in the fre3uency /omain has avalue close to $ in the range of fre3uencies that the analyst

5ishes to retain an/ close to zero in the range of

fre3uencies that the analyst 5ishes to eliminate.

/ The 7lter can be a00lie/ in the time /omain or in the

fre3uency /omain but their function is best un/erstoo/ in

the fre3uency /omain

/ If the 7lter is a mechanical or electrical /evice 5hich

o0erates on the continuous time 0hysical signal it is calle/

an analogue 7lter.

/ If the 7lter is a numerical algorithm or mechanical /evice

5hich o0erates on sam0le/ /ata it is calle/ a /igital 7lter.

?requency ?iltering


85/138

8igh-Pass =lters As the name im0ly1 a high 0ass 7lter allo5s highfre3uencies to 0ass. ;lo5er fre3uency limit<

/o*-Pass =lters

Allo5 lo5 fre3uencies to 0ass through;u00er limit<

>andpass =lters Allo5s only fre3uencies 5ithin the ban/

Anti-aliasing =ltersLo5 0ass 7lter at half the sam0ling fre3uencies

(es o# /ters:

f

f

?requency ?iltering (contHd)


86/138

ypical spectrum of a machine *ithoutany =lter applied

8igh-Pass =lters

/o*-Pass =lters

>andpass =lters

Discrete ?ourier ransform


87/138

/ The signal that comes to the analyzer is analog signal. It

must be /igitally sam0le/ by the analyzer. This 0rocess is a

variation of ))T an/ is kno5n as 2)T.

/ )or 2)T the 5aveform is re-create/ in the analyzer by

/igitally sam0ling an/ then transforme/ into the fre3uency

/omain.

/ To un/er stan/ the ))T /igital sam0ling 0rocess 15e must

have the un/er stan/ing of!

/ L.E.?

/ )ma*

/ Length of aveform

/ 2igital am0le ize

?? (D?) - Pitfalls


88/138

/ 2iscrete )ourier Transform ;2)T< - itfalls

/ ))T - )ast )ourier Transform is an ePcient means of

calculating a 2)T ;2iscrete / )ourier Transform


89/138

Sampling rate too slo*

8igh frequency analysis results in false lo* frequencysignal

Solution$

/ :se Anti-aliasing 7lter.

/ Ty0ically a $Q ;$"&' 0oint< transform1 #$& fre3uency com0onents are

calculate/.

/ an/ '"" lines /is0laye/. imilarly a &Q transform +"" lines are

/is0laye/.

?? pitfalls - /ea3age


90/138

6st Sample

Bnd Sample

$0e

$0e

B0e

B0e

+,,gi#e discontinuities*hen ends Eoined

2se 8anning !indo*

?? - Pic3et ?ence ect


91/138

ActualSpectrum

MeasuredSpectrum

/ines of esolution


92/138

/ RLines of resolutionS /etermine the clarity of the s0ectral

/ata.

/ The better the resolution the more accurate the

fre3uency /is0laye/

/ The number of lines of resolution selecte/ are /ivi/e/

into the ma*imum fre3uency scale ;)ma*< to arrive at thebin 5i/th ;8! : ?maF;/%

/ The lines are actually centre fre3uencies of bins of energy

/ All the energy 5ithin the bin is summe/ u0 to give a

single am0litu/e fre3uency

esolution


93/138

/ 8 N )ma* LE?

/ nergy is summe/ u0

5ithin a 8in an/ 0lotte/

at the centre fre3uency. C e n t r e r e 3

u e n # &

-an'wi'th

/ines of esolution


94/138

/ines of esolution (contHd)


95/138

L2 - TA %6

TA%6 -M%@ Motor Outoar3 @or07o#tal

A#al/7e Spectrum

%-Mar-!% !$'%'&

P ) *!,

LOAD ) %!!*!

RPM ) %($6*

RPS ) 2(*$(

! (!! ,!! %2!! %6!!

!

!*%

!*2

!*

!*(

!*&

Fre.ue#c/ 0# @7

P A c c e l e r a t 0 o # 0 # + - 4

L2 - TA %6

TA%6 -M%@ Motor Outoar3 @or07o#tal

A#al/7e Spectrum

%-Mar-!% !$'%('%6

P ) *,&2

LOAD ) %!!*!

RPM ) %($*

RPS ) 2(*$&

! (!! ,!! %2!! %6!!

!

!*!(

!*!,

!*%2

!*%6

!*2!

Fre.ue#c/ 0# @7

P A c c e l e

r a t 0 o # 0 # + - 4

0 The s0ectrum sho5n /is0lays/ata at +"" L.E.? 5ith an )ma*of $("" 4z

he second spectrumdisplays the same data but*ith B77 /,%, o#er thesame ?maF

!indo*ing and /ea3age


96/138

/ ))T base/ measurements are subBect to errors from an

e9ect kno5n as leakage.

/ This e9ect occurs 5hen the ))T is com0ute/ from of a block

of /ata 5hich is not 0erio/ic.

/ To correct this 0roblem a00ro0riate 5in/o5ing functions

must be a00lie/.

/ The user must choose the a00ro0riate 5in/o5 function for

the s0eci7c a00lication.

/ hen 5in/o5ing is not a00lie/ correctly1 then errors may beintro/uce/ in the ))T am0litu/e1 fre3uency or overall sha0e

of the s0ectrum.



97/138

/ Leakage! Leakage is cause/ 5hen the time 5aveform signal

/oes ET begin an/ en/ at the same 0oint1 intro/ucing

s0urious fre3uencies.

/ The in/o5 or 5eighting function attenuates the signal

to5ar/s the e/ge of the 5in/o5 J minimizing leakage.

)re3uency )re3uency



98/138

?e3uire/ to solve FLeakageH

everal Ty0es!

/ 4anning J 6ost Commonly use/

/ :niform

/ 4amming

/ 8lackman-4arris4anning in/o5! best com0romise bet5een fre3uency

resolution an/ am0litu/e accuracy for stea/y-statemachinery analysis

/ :niform or )lat-To0! is the best choice for transientmachinery analysis.



99/138

/ The most common 5in/o5s an/ their features are given

belo5.

/ This table can be use/ to choose the best 5in/o5ing

function for each a00lication.

A#eraging / Linear Averaging! In linear averaging1 each instantaneous s0ectrum is


100/138

g g g g 0

a//e/ to the ne*t an/ the sum is /ivi/e/ by the total number of

s0ectra. This metho/ is useful in obtaining re0eatable /ata for fault

tren/ing1 as use/ in most 0re/ictive maintenance 0rograms.

/ eak-hol/ Averaging! eak hol/ is not a true averaging metho/. 2uring

sam0ling time1 the 0eak value registere/ in each analysis cell is

ca0ture/ an/ /is0laye/. This metho/ is very useful in vie5ing

transients or for stress analysis calculations.

/ *0onential Averaging! This techni3ue takes the most recent s0ectrum

collecte/ an/ 5eighs it more heavily than the 0ast /ata. It is useful in

observing con/itions that are changing slo5ly 5ith res0ect to sam0ling

time i.e.1 a stea/y-state 0rocess.

/ ynchronous Averaging! This metho/ utilizes a synchronizing signal

from the machine being analyze/. The synchronizing signal is usually

/erive/ from a 0hotocell1 electromagnetic 0icku01 or some other form

of tachometer in0ut The vibration in0ut is sam0le/ at 0recisely the

%#erlap A#eraging


101/138

/ hen more than $ average is use/ to calculate the ))T1 it is

0ossible to use overla00ing sam0les1 as sho5n in )igure

belo5!

/ This 5orks 5ell since the 7rst 0art an/ last 0art of the

sam0le have their am0litu/es re/uce/ in normal averaging1

5hile the overla00ing sam0le takes full rea/ings at these

0ositions.

/ The re/uction in accuracy is very small1 an/ for ))Ts 5ith a

lo5 )ma* an/ a lot of averages1 collection times can be

re/uce/ consi/erably.

/ )or e*am0le1 an ))T 5ith '"" lines1 an )ma* of (""" C61

an/ + averages 5ithout overla00ing takes ,& secon/s to

gather the sam0les. ith #" overla0 averaging1 sam0ling

re uires onl $+ secon/s.

Display;Storage (6)


102/138

/ The /ata collectoranalyzer can collect an/ store only a

limite/ amount of /ata. / Therefore1 the /ata must be /o5nloa/e/ to the com0uter

to form a history an/ long-term machinery information

/atabase for com0arison an/ tren/ing.

/ To 0erform the above tasks1 management an/ analysis of

machinery /ata1 /atabase management soft5are are

re3uire/.

/ These /atabase management 0rogramme for machinerymaintenance store vibration /ata an/ make com0arison

bet5een current measurements1 0ast measurements an/

0re/e7ne/ alarm limits.

6easurements transferre/ to the vibration analysis

Display;Storage (B)


103/138

/ 6easurements transferre/ to the vibration analysis

soft5are are ra0i/ly investigate/ for /eviations from

normal con/itions. / Everall vibration levels1 ))T1 time 5aveforms.

/ re0orts can be generate/ sho5ing machines 5hose

vibration level cross alarm threshol/s.

/ Current /ata are com0are/ to baseline /ata for analysis

an/ also tren/e/ to sho5 vibration changes over a 0erio/

of time.

/ Tren/ 0lots give early 5arnings of 0ossible /efects an/ are

use/ to sche/ule the best time to re0air.

?ormulae)


104/138


105/138

)a*e+orms)a*e+orms

!a#eforms (6)The time 5aveform is the electrical signal from the sensor


106/138

/ The time 5aveform is the electrical signal from the sensor.

/ It is a trace of the voltage changes as the instantaneous

vibration changes from moment to moment.

/ This voltage is gra0he/ 5ith time. Thus the name Time

5aveform. The 5aveform 0rovi/es a vie5 into e*actly ho5

that 0oint is moving or vibrating over time.

!a#eforms (B)


107/138

/ Uust like the s0ectral there are certain 0atterns an/

characteristics to look for 5hen con/ucting 5aveform

analysis.

/ Ence the characteristics have been i/enti7e/1 the

analyst can rule out certain faults

e.g! if the 5aveform is 0erio/ic faults likeLooseness1 8earing /efects1 Cracks coul/be rule/ out.

/ 2ata from the time 0lot 5ill in/icate 5hat ty0e of

vibration is 0resent. The 7ve ty0es of vibration are

harmonic1 0erio/ic1 beating1 im0ulsive1 or ran/om

)a*e+orms ,-.)a*e+orms ,-.


108/138

The analysis of time waveform data is not a new technique2 -n the early days of vibration

analysis time waveform data was viewed on oscilloscopes and frequency components

calculated by hand2 The relationship between frequency and time is as follows7

f A 3*p where7

0 f is the frequency in B'

0 p is the period in seconds ;the amount of time required to complete 3 cycle<

/nowle'ge o this

relationship permits the

'etermination o

re3uen#& #omponents

rom the raw wa0eorm

'ata5

)a*e+orms,/.)a*e+orms,/.


109/138

(or e9ample7

, ., .

(his wa0eorm was a#3uire'

rom a 1D6 RP% pump5 (he

time spa#ing )etween the impa#ts

is 5++ se#on's5 rom thisinormation the re3uen#& #an )e

'etermine'5

@ 1p @ 1 5++ @ 25F

H. @ 1D CP%

(his in'i#ates that the impa#t is

o##urring at a re3uen#& o 1 x

RP%5

)a*e+orms,0.)a*e+orms,0.


110/138

-n most situations the time waveform pattern is very comple9 as illustrated below and

therefore the determination of frequency components is e9tremely difficult using this

method and is not recommended2

-n most situations time waveform data is best utili'ed by applying the principles of

pattern recognition and if necessary calculating the frequency components of the

ma6or events in the waveform pattern2

)hen to Use Time )a*e+orm)hen to Use Time )a*e+orm


111/138

T0me ?a9e:orm ca# e u4e3 e::ect09el/ to e#


112/138

T0me ?a9e:orm ca# e appl0e3 to a#/ 90rat0o# prolem* I# 4ome 40tuat0o#4

#ormal 4pectral a#3 p


113/138

The $ey to successful analysis of time waveform data is in the set up of the

instrument2 The following items have to be considered when setting up theinstrument2

0 U#0t o: mea4ureme#t

0 T0me per0o3 4ample3

0 Re4olut0o#

0 A9era;0#;

0 0#3o?4

Units o+ Measurement,1.Units o+ Measurement,1.


114/138

mplitude measurement units should be generally selected based upon the

frequencies of interest2 The plots below illustrate how measurement unit selection

affects the data displayed2 Each plot contains @ separate frequency components of

4B'1 @44B'1 and JC4 B'2

(his 'ata was ta>en using 'ispla#ement

note how the lower re3uen#& at F H. is

a##entuate'5

Units o+ Measurement,2.Units o+ Measurement,2.


115/138

(he same 'ata is now 'ispla&e' using 0elo#it& note

how the +H. #omponent is more apparent5

(he same 'ata is now 'ispla&e' using a##eleration

note how the large lower re3uen#& #omponent is

'iminishe' an' the higher re3uen#& #omponent

a##entuate'5

(he unit o measurement 'ispla&e' in time

wa0eorm 'ata shoul' )e the natural unit o the

trans'u#er use'5 or example i a 'ispla#ement

rea'ing is re3uire', then a 'ispla#ement trans'u#ershoul' )e use'5 In most #ases where mo'ern 'ata

#olle#tors are emplo&e' this means that a##eleration

will )e the unit o #hoi#e5 I 'ata is gathere' rom

non$#onta#t pro)es on slee0e )earing ma#hines

'ispla#ement is usuall& use'5

Time Perio Sam%#e,1.Time Perio Sam%#e,1.


116/138

The appropriate ("H and Total ,ample Period can be calculated

by the following formula7

("H %P"L A !ines of +esolution 9 +P"

M f +evolutions desired

Total sample period secondsL A 4 9 M of revolutions desired

+P"

Time Perio Sam%#e,2.Time Perio Sam%#e,2.


117/138

Re4olut0o#' (or time waveform analysis it is recommended that 344 lines

;D4J samples are used


118/138

The classic sine wave illustrated below is rarely seen in acceleration time waveform2

This is because acceleration emphasi'es the higher frequency components that are

almost always present in the vibration signal2 This de8emphasi'es the underlying lower

frequency signal2

U#ala#ce

Inter%retation o+ )a*e+orm !ata,2.Inter%retation o+ )a*e+orm !ata,2.


119/138

The waveform below is more representative of sinusoidal vibration when viewed in

acceleration2 Note the high frequency components superimposed on the lower

frequency2

Inter%retation o+ )a*e+orm !ata,-.Inter%retation o+ )a*e+orm !ata,-.


120/138

M04al0;#me#t

lthough the classic symptoms of misalignment are " and W shapes in the time

waveform1 these symptoms cannot be relied upon2 The relative phase angle between

the 3 9 +P" and 9 +P" components determines the shape or pattern of the plot2

The pattern above illustrates the classic pattern of misalignment2

Inter%retation o+ )a*e+orm !ata,/.Inter%retation o+ )a*e+orm !ata,/.


121/138

-n the pattern below the relative phase between 3 9 +P" and 9 +P" was changed J4

degrees resulting in a very different pattern2



122/138

The pattern below originates when the 3 91 & 91 vibrations are 4 degrees apart2

A l0t 3 S t



123/138

Ampl0tu3e S/mmetr/

When observing time waveform data symmetry above and below the centerline

a9is is important2 ,ymmetrical data indicates that the machine motion is even on

each side of the center position2 Non8symmetrical time waveform data indicates

the motion is constrained possibly by misalignment1 or rubs2

This waveform pattern is symmetrical above and below the 'ero line2



124/138

The following waveform pattern is non8symmetrical above and below the 'ero line2 The

amplitudes below the line are significantly higher than those above the line2 -n this case

a misalignment condition was the source2 The mar$ers on the plot indicate 3 9 +P"2



125/138

S/mmetr/ o: t


126/138

This e9ample1 is of two frequency sources that are not harmonically related2 ;C B' and

34 B'< This is the $ind of signal that could be created when a 8pole motor has an

electrical hum problem2

-t can be seen that the higher frequency wave does not always start at the same part of

the lower frequency cycle and therefore appears to Oride on the other wave causing

symmetry to be lost2 %are must be e9ercised when determining symmetry of the time

a9is2 3 9 +P" mar$ers are available in most software programs and should be used to

avoid confusion2



127/138

t first glance this waveform appears to have large impacts occurring with somewhat

similar spacing2 The hori'ontal a9is is scaled in time units2

.y using +P" as the hori'ontal a9is and applying 3 9 +P" mar$ers the ma6or impacts

can be seen to be occurring at appro9imately the same part of the revolution2 Bowever

closer inspection reveals that the spacing is not e9actly synchronous -n this case the

problem was a single large defect on the inner race of a bearing2 The change in

amplitude of the defect was due to the defect coming in and out of the load 'one2



128/138

This is the ((T ta$en from the above machine note the highest amplitude at .P(- is

Q 424C ipsR ;3Cg p$ S .P(- A 23@ ips


129/138

Beat4 Mo3ulat0o# e::ect4

nother e9cellent application for time waveform is the observation of beat frequencies

and modulation effects2 ften these phenomena are audible2 The time span for data

collection should be set to capture D8C cycles of the beat2

The time period between the beats on the above waveform is 42C s2 (rom this informationthe frequency of the beat is calculated at 34 %P"2 This represents the frequency

difference between the two source frequencies -n this case the beat was caused by

interaction between a H +P" vibration source and a 9 f! vibration source on an

induction motor2

Inter%retation o+ )a*e+orm !ata,1-.Inter%retation o+ )a*e+orm !ata,1-.


130/138

Impact4

When the ((T process is applied to a signal that contains impacts the true

amplitude of the vibration is often greatly diminished2 The following time waveform

was ta$en from an 344 +P" machine2 -t shows several random impacts with

magnitudes over g p$2 The cause of this signal was a failed rolling element

bearing2 The shape of the waveform often appears to be a large spi$e followed by a

Oring down2

Inter%retation o+ )a*e+orm !ata,1/.Inter%retation o+ )a*e+orm !ata,1/.


131/138

The plot below was a velocity spectrum ta$en from the same bearing note the

amplitude of vibration is less than 424D ips2



132/138

%are must be e9ercised when

assessing the amplitude severity of

3 9 +P" impact vibration using

spectrum2 This is the spectrum of a

machine with where the $ey is

impacting the coupling guard2 The

amplitude scale indicates

amplitudes of less than 424 ips2

This is the time waveform from

the above machine the

amplitude of the impacts

e9ceeds 423C ips2

-n this case severe damage had

occurred to the $ey and shaft of

the machine in question2

)a*e+orm Ana#$sis,1.)a*e+orm Ana#$sis,1.


133/138

-n conclusion1 how is this information practically applied in a condition based

maintenance program2

Time waveform analysis is an analysis tool2 -t would not recommend that it be ta$en on

all measurement locations on a regular basis2 This would add significantly to the time

required and data storage requirements2

:se Time waveform for the following selected analysis situations to enhance ((T

information2

0 Lo? 4pee3 appl0cat0o#4 =le44 t*

0 I#30cat0o# o: true ampl0tu3e 0# 40tuat0o#4 ?


134/138

:se an appropriate measurement unit

0 Roll0#; eleme#t ear0#;4 ;ear4 loo4e#e44 ru4 0mpact4 accelerat0o#

0 Slee9e ear0#; mac

0 Beat4 Mo3ulat0o#

0 Impact4 =4

Rememer u4e t0me ?a9e:orm to E8@A8CE #ot REPLACE 4pectral 3ata*

Crest (actor Crest (actor


135/138

I#tro3uct0o#

The %rest (actor is equal to the pea$ amplitude of a waveform divided by the +", value2 The

purpose of the crest factor calculation is to give an analyst a quic$ idea of how much impacting

is occurring in a waveform2 -mpacting is often associated with roller bearing wear1 %avitation and

gear tooth wear2

-n a perfect sine wave1 with an amplitude of O31 the +", value is equal to 2F4F1 and the crest factor

is then equal to 32D32 perfect sine wave contains no impacting and therefore crest factors with a

value higher than 32D3 imply that there is some degree of impacting2



136/138

Tt wor$ so well when one has a signal

that consists of non8periodic events1 impacts or random noise 2 .oth impacts and random

noise appear the same in the spectrum although they mean different things in the conte9t of

machinery vibration analysis2 The crest factor is therefore useful in giving the analyst a quic$

idea of what is occurring in the time waveform2



137/138

Compar04o# o: 2 a9e:orm4

-n below figures we can see an e9ample of the use of the %rest (actor2 The waveform in figure

on left has a crest factor of @2432 The waveform in figure on right has a crest factor of 3232

The data in figure on left represents a machine with serious rolling element bearing wear1 and

the crest factor is relatively high due to the amount of impacting occurring within the bearing2

The data in figure on right represents a machine with an unbalance1 but no impacting related to

bearing wear2



138/138

Co#clu40o#

The %rest (actor is a quic$ and useful calculation that gives the analyst an idea of how

much impacting is occurring in a time waveform2 This is useful information that is lost if

one is only viewing a spectrum as the ((T cannot differentiate between impacting and

random noise2 -mpacting in a time waveform may indicate rolling element bearing wear1

gear tooth wear or %avitation2 #uite often1 the %rest (actor is trended over time in order

to see if the amount of impacting is increasing or not2

Data acquistion principles.pptx

Documents

Transcript of Data acquistion principles.pptx