Nondestructive Impact and Acoustic Testing For Quality Assessment of Apples by Itzhak Shmulevich, Naftali Galili and M. Scott Howarth A G E NG 2002 Budapest,

Nondestructive Impact and Nondestructive Impact and Acoustic Testing ForAcoustic Testing For Quality Quality

Assessment of ApplesAssessment of Applesby by

Itzhak Shmulevich, Naftali Itzhak Shmulevich, Naftali GaliliGalili

andand

M. Scott Howarth M. Scott Howarth

AAGGEENGNG 2002 2002

Budapest, Hungary June 30-July 4, 2002Budapest, Hungary June 30-July 4, 2002

The Department of The Department of Agricultural Agricultural EngineeringEngineeringTechnion-Israel Institute of TechnologyTechnion-Israel Institute of Technology

Presentation outlinePresentation outline

Introduction - firmness quality nondestructive Introduction - firmness quality nondestructive measurements;measurements;

Impact technique vs. acoustic technique;Impact technique vs. acoustic technique; Experimental report on various fruits;Experimental report on various fruits; Results; Results;

Discussion; Discussion;

Conclusions.Conclusions.

QUALITY QUALITY ASSESSMENTASSESSMENT

Quality Factors of Quality Factors of Agricultural ProductsAgricultural Products

Appearance - visual Texture - feel Flavor - taste and smell Safety Nutritive Value

TextureTexture

Texture can be defined by subjective Texture can be defined by subjective terms such as:terms such as:

Firmness Firmness Mealiness, Hardness, Softness, Mealiness, Hardness, Softness, Brittleness, Ripeness, Toughness, Brittleness, Ripeness, Toughness, Chewiness, Smoothness, Crispness, Chewiness, Smoothness, Crispness, Oiliness, Springiness, Toughness, Oiliness, Springiness, Toughness, Fibrousness, or Juiciness etc.Fibrousness, or Juiciness etc.

Quality Sensing in Quality Sensing in Commercial SettingsCommercial Settings

RequirementsRequirements

NondestructiveNondestructive

External and internal propertiesExternal and internal properties

AccuracyAccuracy

Speed (5-15 fruits/sec)Speed (5-15 fruits/sec)

Recognize inherent product variabilityRecognize inherent product variability

NONDESTRUCTIVE NONDESTRUCTIVE

SENSOR SENSOR TECHNOLOGYTECHNOLOGY

NondestructiveNondestructive Firmness Firmness

Measurement Measurement TechniquesTechniques Fruit Response to ForceFruit Response to Force

Detection by Impact ForceDetection by Impact Force Forced VibrationsForced Vibrations Mechanical or Sonic ImpulseMechanical or Sonic Impulse Ultrasonic TechniquesUltrasonic Techniques Indirect Firmness MeasurementIndirect Firmness Measurement

Research Research ObjectiveObjective

The motivation of the present work is to develop

a fast nondestructive method for quality firmness

testing of apples.

The general objective of the research is to

compare sensing the fruitfirmness using low

mass impulse excitation to the acoustic response

for quality assessment of apples.

TextureTexture

Acoustic vs. ImpactAcoustic vs. Impact

Relationship between turgor Relationship between turgor pressure and tissue rigiditypressure and tissue rigidity

E=3.6 p +2.5 x10E=3.6 p +2.5 x1077 [ [ dynes/cmdynes/cm22]]

Modulus of ElasticityModulus of Elasticity

NondestructiveNondestructive Firmness Firmness MeasurementMeasurement

Impact Force TechniqueImpact Force Technique

Quality Detection by Quality Detection by Impact ForceImpact Force

Time [msec]

Force[N]

Low-Mass Impact (LMI) Low-Mass Impact (LMI) FirmnessFirmness

IQ FirmnessIQ Firmness

IQ IQ TM TM Firmness Tester Firmness Tester

Sinclair International Sinclair International LTDLTD

NondestructiveNondestructive Firmness Firmness MeasurementMeasurement

Acoustic TechniqueAcoustic Technique

Firmalon PrototypeFirmalon Prototype

FirmalonFirmalon

Typical Acoustic Fruit Typical Acoustic Fruit ResponseResponse

Frequency DomainFrequency DomainTime DomainTime Domain

(b)-5

-3

-1

1

3

5

0 5 10 15 20 25TIME (ms)

AM

PL

ITU

DE

(V

)

Sensor-1

Sensor-2

Sensor-3

(d)0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0 400 800 1200 1600FREQUENCY (Hz)

AM

PL

ITU

DE

(V

)

Sensor-1

Sensor-2

Sensor-3

Microphone Based Microphone Based System for Acoustic System for Acoustic

Firmness TestingFirmness Testinghp 35665A

Impact

Microphone Signal conditioner

Source: J. De BaerdemaekerSource: J. De Baerdemaeker

Comparison Between Comparison Between Two Acoustic Test Two Acoustic Test

MethodsMethods

35302520151055

10

15

20

25

30

35

FIRMNES INDEX - METHOD-A

FIR

MN

ES

S I

ND

EX

- B y = 0.00328 + 0.9954x

R^2 = 0.977

BRAEBURN

3025201510500

5

10

15

20

25

30

FIRMNESS INDEX - METHOD-AFI

RM

NE

SS

IN

DE

X -

B

y = -0.30995 + 1.0080x R^2 = 0.989

JONAGOLD

Method-A:Method-A: Microphone Microphone Method-B:Method-B: Piezoelectric-Film Sensor Piezoelectric-Film Sensor

Source: N. Galili & J. De BaerdemaekerSource: N. Galili & J. De Baerdemaeker

Acoustic Firmness Acoustic Firmness SensorSensor

A F SA F S TMTM

Source: AWETASource: AWETA

DestructiveDestructive Firmness Firmness

MeasurementMeasurement

22

p

p

T

FC

p

p

T

FC 1

1.5 2 2.5 3 3.5

x 10-3

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Time [sec]

Am

plit

ude [

volt]

d

sec

N

Fp

Tp

td

Quality Detection by Impact Quality Detection by Impact ForceForceChen. P (1996), Farabee (1991)

Delwiche (1989 ,1991), Nahir et al. (1986 )

Quality Detection by Quality Detection by ImpactImpact

0 0.5 1 1.5 2 2.5 3 3.5 4

x 10-3

0

10

20

30

40

50

60

Time [sec]

F [N

]

day 1day 2day 3day 4

0 100 200 300 400 500 600 700 800 900

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Frequency [Hz]

Norm

alize

Am

plitu

de

day 1day 2day 3day 4

Source: Shmulevich et. al. Source: Shmulevich et. al. ( 2000 )( 2000 )

The Acoustic The Acoustic Parameters Parameters

of a Fruitof a Fruit Natural frequencies and firmness index - Natural frequencies and firmness index - FIFI

FI = f FI = f 22 m m 2/32/3 {10{104 4 kgkg2/32/3 s s-2-2} }

where: where: f f - first spherical resonant frequency - first spherical resonant frequency

mm - fruit’s mass. - fruit’s mass. Damping ratio - Damping ratio - The centeroid of the frequency response - The centeroid of the frequency response - ffcc

Parameters Parameters extracted from the extracted from the

measurementsmeasurements Low-Mass Impulse parameters: Low-Mass Impulse parameters:

C1 = Fp/Tp; C2 = Fp/Tp2 ; (-20); and fc(in).

IQ , IT

Acoustic parameters:

f1 ; FI ; and fc;

Destructive parameters: E ; MT.

Method and Method and MaterialsMaterials

• Three apples cultivers;Three apples cultivers;

• Shelf life conditions: 20 Shelf life conditions: 20 00C C 50%RH;50%RH;

• 25 fruits were tested daily 25 fruits were tested daily both both nondestructively and nondestructively and destructively;destructively;

• Destructive test - (MT, E’);Destructive test - (MT, E’);

.

ResultsResults

Results - Results - Rubber Spheres Rubber Spheres

Results - Results - Rubber Spheres Rubber Spheres

0

20

40

60

80

100

120

140

160

0 1 2 3 4 5 6 7 8

Displacement (mm)

Fo

rce

(N

)

White

Black

Yellow

Red

Typical input signal in time domain of theLow-Mass Impact system for different

Rubber Spheres

2000

2100

2200

2300

2400

2500

2600

2700

2800

50 100 150 200 250 300 350 400

Time steps

Am

plit

ud

e

White

Black

Yellow

Red

Typical input signal in time domain of theLow-Mass Impact system by Sinclair for

different Rubber Spheres

1800

2000

2200

2400

2600

2800

3000

3200

3400

3600

3800

50 100 150 200 250 300 350 400

Time steps

Am

plit

ud

e

White

Black

Yellow

Red

Results - Results - Rubber Spheres Rubber Spheres

IQ = 3.0396x + 0.2495

R2 = 0.9984

IT = 2.936x - 0.6056

R2 = 0.9975

FI = 1.5694x - 0.8186

R2 = 0.99650

5

10

15

20

25

30

35

40

45

50

0 2 4 6 8 10 12 14 16

E' (MPa)

Fir

mn

ess

FI (mean)

IQ (mean)

IT (mean)

Typical Acoustic Fruit Typical Acoustic Fruit ResponseResponse

Frequency DomainFrequency DomainTime DomainTime Domain

(b)-5

-3

-1

1

3

5

0 5 10 15 20 25TIME (ms)

AM

PL

ITU

DE

(V

)

Sensor-1

Sensor-2

Sensor-3

(d)0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0 400 800 1200 1600FREQUENCY (Hz)

AM

PL

ITU

DE

(V

)

Sensor-1

Sensor-2

Sensor-3

Golden Delicious Golden Delicious n=270n=270

Pearson linear correlation between the Pearson linear correlation between the nondestructive and destructive tests,nondestructive and destructive tests,

n=270 Golden Delicious -Apple, n=270 Golden Delicious -Apple, Correlation is significant at the 0.05 levelCorrelation is significant at the 0.05 level

E' MT FI IQ IT

E' 1 0.542 0.913 0.911 0.923

MT 1 0.520 0.517 0.547

FI 1 0.923 0.927

IQ 1 0.990

IT 1

Results -Golden Results -Golden DeliciousDeliciousGolden Apples (N = 270)

MT = 0.2925x + 2.9894

R2 = 0.2942

FI = 4.0346x - 2.9391

R2 = 0.8339

IQ = 2.1593x + 6.3759

R2 = 0.8292

IT = 4.0464x + 8.1449

R2 = 0.85260

5

10

15

20

25

30

35

40

45

50

0 2 4 6 8 10

E' (MPa)

Fir

mn

es

s

MT (mean)

FI (mean)

IQ (mean)

IT (mean)

Results-Golden Results-Golden DeliciousDeliciousGolden Delicious

IT = 0.0456x2 - 1.9376x + 39.995

R2 = 0.9646

IQ = 0.023x2 - 1.0098x + 23.336

R2 = 0.9679

FI = 0.0515x2 - 2.1226x + 29.909

R2 = 0.957

E' = 0.0074x2 - 0.372x + 7.3076

R2 = 0.9592

MT = 0.0009x2 - 0.0956x + 5.224

R2 = 0.93770

5

10

15

20

25

30

35

40

45

0 2 4 6 8 10 12 14 16 18 20 22

Day

Fir

mn

ess

E' (average)

MT(average)

FI (average)

IQ (average)

IT (average)

Starking Apples Starking Apples n=270n=270

Pearson linear correlation between the Pearson linear correlation between the nondestructive and destructive tests,nondestructive and destructive tests,

n=270 Starking -Apple, n=270 Starking -Apple, Correlation is significant at the 0.05 levelCorrelation is significant at the 0.05 level

E' MT FI IQ IT

E' 1 0.597 0.947 0.837 0.850

MT 1 0.598 0.532 0.542

FI 1 0.829 0.842

IQ 1 0.990

IT 1

Results-Starking Results-Starking ApplesApples

IT = 2.8958x + 13.914

R2 = 0.7226

IQ = 1.6916x + 8.6502

R2 = 0.7006

FI = 3.003x - 2.5367

R2 = 0.8966

MT = 0.3071x + 3.9389

R2 = 0.3565

0

5

10

15

20

25

30

35

40

45

50

0 2 4 6 8 10

E' (MPa)

Fir

mn

ess

MT (mean)

FI (mean)

IQ (mean)

IT (mean)

Results-Starking Results-Starking ApplesApples

IT= 0.0097x2 - 0.8215x + 37.34R2 = 0.9627

IQ = 0.0065x2 - 0.5113x + 22.477R2 = 0.9302

FI = 0.0113x2 - 0.8938x + 21.854R2 = 0.9966

E' = 0.0028x2 - 0.2475x + 7.694R2 = 0.9888

MT = 0.0007x2 - 0.0751x + 6.3496R2 = 0.9206

0

5

10

15

20

25

30

35

40

0 10 20 30 40 50Day

Fir

mn

ess

E'-average

MT-average

FI-average

IQ-average

IT-average

Granny SmithGranny Smith n=270n=270

Pearson linear correlation between the Pearson linear correlation between the nondestructive and destructive tests,nondestructive and destructive tests,

n=270 Granny Smith -Apple, n=270 Granny Smith -Apple, Correlation is significant at the 0.05 levelCorrelation is significant at the 0.05 level

E' MT FI IQ ITE' 1 0.450 0.915 0.820 0.791

MT 1 0.461 0.426 0.406FI 1 0.883 0.885 IQ 1 0.978 IT 1

Results- Granny Results- Granny SmithSmith

MT = 0.155x + 4.9758R2 = 0.2497

FI = 2.5615x + 3.1173R2 = 0.8374

IT = 1.5315x + 18.428R2 = 0.6264

IQ = 0.9198x + 11.384R2 = 0.672

0

5

10

15

20

25

30

35

40

45

0 2 4 6 8 10 12 14

E' (MPa)

Fir

mn

ess

MT (mean)

FI (mean)

IQ (mean)

IT (mean)

Results- Granny SmithResults- Granny Smith

IT = 5.1747x + 19.314R2 = 0.7387

IQ = 4.5916x + 18.456R2 = 0.7055

FI = 2.4549x + 6.924R2 = 0.992

MT = 0.1472x + 5.2212R2 = 0.7301

E' = 0.834x + 2.1082R2 = 0.9198

0

5

10

15

20

25

30

35

40

0 10 20 30 40 50Day

Firm

ness

E' (mean)

MT (mean)

FI (mean)

IQ (mean)

IT (mean)

SummarySummary

•The impact firmness parameter IQ and IT of The impact firmness parameter IQ and IT of

the calibration balls, obtained by the the calibration balls, obtained by the

instrumented instrumented

hammer and the Sinclair sensor, were very closehammer and the Sinclair sensor, were very close

(R-average = 0.992), while the Sinclair sensor (R-average = 0.992), while the Sinclair sensor

predicted slightly better the elastic modulus predicted slightly better the elastic modulus

of the balls (R = 0.9992).of the balls (R = 0.9992).

Summary (Cont,)Summary (Cont,)

•The elastic modulus E’, which is the The elastic modulus E’, which is the physical measurement of firmness, was physical measurement of firmness, was predicted well by the IQ and IT impact predicted well by the IQ and IT impact parameters in Golden Delicious apples (R-parameters in Golden Delicious apples (R-average = 0.917). average = 0.917).

•The acoustic firmness index FI was The acoustic firmness index FI was equivalent to IQ in Golden Delicious, but equivalent to IQ in Golden Delicious, but improved the prediction of E’ in Starking and improved the prediction of E’ in Starking and Granny Smith apples (R-average = 0.931).Granny Smith apples (R-average = 0.931).

Summary (Cont.)Summary (Cont.)

•High correlation was found between the High correlation was found between the

individual readings of IT and IQ in all individual readings of IT and IQ in all

apples (R = 0.9867). Hence, the simple IT apples (R = 0.9867). Hence, the simple IT

algorithm may replace the IQ algorithm, if algorithm may replace the IQ algorithm, if

quick firmness calculation is needed to quick firmness calculation is needed to

increase the present operation speed of the increase the present operation speed of the

low-impact sorter (600 fruit per minute per low-impact sorter (600 fruit per minute per

lane).lane).

Summary (Cont.)Summary (Cont.)

•The changes in the penetration force MT The changes in the penetration force MT

(the yield strength of fruit tissue) during the (the yield strength of fruit tissue) during the

test period were very low, and their test period were very low, and their

correlation with the elastic modulus and correlation with the elastic modulus and

firmness parameters of all apples’ varieties firmness parameters of all apples’ varieties

was poor (R < 0.60).was poor (R < 0.60).

Summary (Cont.)Summary (Cont.)

•These findings in the tested apple varieties These findings in the tested apple varieties

indicate that the acoustic firmness index FI, if indicate that the acoustic firmness index FI, if

successfully applied in a sorting line, may successfully applied in a sorting line, may

improve the sorting capacity of a multi-sensor improve the sorting capacity of a multi-sensor

(FI and IQ) automatic machine.(FI and IQ) automatic machine.

Thanks Thanks For Your AttentionFor Your Attention