Sunteți pe pagina 1din 65

Use of Sensors for the Traceability in the Supply Chain

Brigitte Leblon, Ph.D. Faculty of Forestry and Environmental Management, U of New Brunswick, Fredericton (NB) bleblon@unb.ca
With inputs from: F. Fournier (FORINTEK), Q. Wei (UNB), J. Nader (FERIC), K. Tounis and P. Cooper (U of T), K. Groves (FORINTEK) M. Defo (FORINTEK) F. Ding (CRIQ) T. Trung (PAPRICAN) Z. Pirouz (FORINTEK)

Why Using Sensors?


Without sensors, the supply chain is like a blind person
Sensors = of the chain

Sensors = data for the supply chain optimization model

Electromagnetic spectrum

http://en.wikipedia.org/wiki/Image:EM_Spectrum_Properties_edit.svg

Relation with wood structures


Scale Unit Wood Structure Cellulosic crystal
Fibril

Submicroscopic
Microscopic nm

m mm Macroscopic
Mesoscopic Megascopic

Cell Wall Cells Annual ring Branches, Leaves, Trunk


Tree Stand

cm
m mkm

Gigascopic

km

Forests

(adapted from Bucur 2003)

Relation with wood structures


Spectral domain X-ray Visible light Optical Infrared Thermal Infrared Microwave Ultrasound Radio Frequency (NMRI) Wavelength (m) angstrm () 0.4-0.7 m 0.7-2.5 m 3 100 m cm cm-km km

Types
Need contact
Ex: Acoustic or pulsated current

No contact, but only on small wood samples


EX: NIR, some X-ray sensors

No contact on full log or board


Point measurements Ex: laser systems Images: Ex: cameras, other imaging systems

Forest
(standing tree / outdoor)
Note: only ground-based systems

Terrestrial Laser Scanning System (LiDAR)


Ex: Riegl (A) 3D Terrestrial Laser Scanning (V) Z-Series based on NIR band (900 nm) + TerraScan software for classifying scan data

Sca
Sitka spruce Plan View Measured DBH

Perspective View

(Watt and Donoghue 2005)

Scanned DBH

Stiffness: Acoustic (Sonic) sensors


Fibre-Gen (NZ) Hitman PH330 (prototype)

Influence of the moisture content R&D need

Forest / Merchandizing Yard


(log wood / outdoor)

Merchandizing yard concept


Bucking and segregation yard Block 1 Mill 1

Block 2 Mill 2

Block 3 Mill 3

(courtesy of F. Fournier, FP Innovations)

Dimensions / External Features


Length, diameter and sweep / External log features or defects

Laser beam

IR optical scanners
Microtec (I): iRED, iRAS

Rema (S): Log 3D, Log Bark Mikropuu Oy (Fin) OPMES 604/614 Microtec (I): Dishape Metso (Fin): VisiQ Prototype: Comact with FP Innovations (CDN): Mobile Scanner

Digital optical cameras


Microtec (I): ScreenLog (3D image)

Photocells
Mikropuu Oy (Fin): OPMES 211/212

Stiffness
Acoustic (Sonic) bands Stiffness = f (frequency of the hammer blow, wood density) Portable instrument:
Fibre-Gen (NZ) Hitman HM200 , Fakopp (H) Microsecond timer

In-line instrument:
Fibre-Gen (NZ) Hitman L640 Log grader

Influence of the moisture content

Rot/Moisture Content
Fakopp
Acoustic wave Rot, holes in standing timber Limitations:
Sensitive to water along the whole log Liquid water log temp > 0C No simple relationships for some hardwoods when MC > 30% Variable reg line slope between species and within the same species Contact measurements
(from Nader 2007)

Shigometer
Resistance to a pulsated current that decreases with cation cc Rot in standing timber Limitations:
Sensitive to water close to the sensor Liquid water log temp > 0C and <60C MC > fiber saturation threshold (20-30%) Not reliable if log surface is drying Contact measurements

Internal Features
x-ray path through log X-ray detector Log

X-ray equipments High energy radiation material penetration internal features

x-ray Source (courtesy of Z. Pirouz , FP Innovations)

(Defo 2008)

><: Confusion wood moisture and wood density NIR

One X-ray View Scanner


Density projection in 2D Good for: determining knottiness locating knot clusters detecting metal (WWII) measuring bark thickness Some species identification Not good for pinpointing or separating defects:
Cannot determine the feature (e.g. knot) location along x-ray path (case 1) Effects of a knot (higher density) and a check (density gap) may cancel out (case 2) Multiple dense features (knots) may appear as one knot (case 3) x-ray Detector

x-ray Source

(1)

(2)

(3)

(courtesy of Z.Pirouz, FP Innovations)

Multiple X-ray Views


Each view adds more information to the scan Helps to localize & distinguish defects About 250K ($380K) per additional view Currently 1-4 views are available (4 European vendors) Most European mills use 1-2 views Ex: Rema (S) Log X-Ray, Mikropuu (Fin) OPMES AX1

(Courtesy of Z. Pirouz, FP Innovations)

Computerized Tomography CT X-Ray images


CT images (or multi X-ray views) for: Detecting check, resin pocket, etc. Measuring defect size & location Breakdown optimization Need of good image quality Need of appropriate image processing algorithms
Microtec (I) Tomolog

Bintec (Fin) Wood X

Classified sugar maple CT images using the MLC classifier

a) raw CT image

b) MLC classified image

c) MLC classified image filtered using a 5 5-pixel median filter


(Wei, Leblon et al 2008)

Black Spruce

a) raw CT image

b) MLC classified image

c) MLC classified image filtered using a 5 5-pixel median filter

(Wei, Leblon et al 2008)

BP ANN Classifiers topology selection

The number of layers and nodes in each layer define the BP ANN classifiers topology. One input layer (9 input nodes), one hidden layer, one output layer (4 output nodes). The hidden node number is empirically selected as function of the classification accuracy , mean square error (MSE), the number of training iterations
and training time.
(Wei, Leblon et al 2008)

Classification by the BP ANN classifier

Overall accuracy: 98.5% (training log), 82.0%(validation log) for sugar maple; 97.6% (training log), 67.6%(validation log) for black spruce.
(Wei, Leblon et al 2008)

3D Reconstruction of Internal Log Characteristics: sugar maple

(Wei, Leblon et al 2008)

3D Reconstruction of Internal Log Characteristics: black spruce

(Wei, Leblon et al 2008)

Austria: Bentec Wood X


X-ray based log sorting See details for the UPM Sawmill at Steyrermhl, Austria at the following URL
http://www.suomenlinkki.fi/english/uutinen1.html

Sweden: Microtec Tomolog


Swedish installation of TINA in1980s One claim of 10-30% productivity increase Sorting based on internal defects (mostly knots) Some mills have more than 80 sorting bins Scaling (under bark diameter) Metal detection (debris from WWII) Customized and calibrated for each mill and their species / defects

(courtesy of Z. Pirouz, FP Innovations)

Other imaging systems (experimental)


l

Gamma Ray

X-Ray

Infra-red

a) Neutron imaging

b) CT imaging

c) Thermal imaging

Microwave

Radio

Radio
(Wei, Leblon et al 2008)

d) Microwave imaging

e) Ultrasound imaging f) MR or NMR imaging

Microwave Imaging (Experimental)


Long wave penetration Microwave = f(roughness, dielectric constant e)
Material Water (20oC) Dry material Wet material

e
0.36 0.94 0.59 Knot

Influence of the moisture content

(Nuclear) Magnetic Resonance Imaging (NMRI) (experimental)


Internal wood structure Rot detection Absorption of preservatives Free water and Bound water (hydrogen)

8.9%

8.0%

7.0%

4.0%

Pinus densiflora CW=compression wood CX= conducting xylem (water) CZ= cambial zone (water) Cor= Cortex LW = latewood P = pith

In the Plant
Sawn Wood / Controlled Environment Sawing, Trimming, Edging Optimization

Grading System

Laser Beam (profilometer) + (color or greyscale digital camera)


Dimension, Defects (knots, rot/decay, stain, wane, holes, cracks / splits, pith, resin), Wood cell structure, Color, Board number tracking
Examples:
Comact (CDN): EdgExpert (Stain detection, Knot Detection) , GradExpert, Trimmer System Rot Detector, Blue Stain Detector Autolog (CDN): Cant optmizer, Carriage optimizer, Edger, Trimmer, Planner Bar Mullin Inc. (USA): CompuRip, CellScan, Discan USNR (USA): LASAR 3D, Smart TriCam LuxScan (L): Easy-Scan, Combi-Scan C200/C400, MatchScan, Rip Scanner R200, Shape-Scan L Microtec (I): Boardscan (Boardwide, Boardthick), IDiScan (Tracking), Wanescan Innovativ Vision AB (S): WoodEye CRIQ (CDN): Floor Wood Classification System, Multisensor grader/optimizer (BorealScan)

Video of the Comact Stain detection system

X-ray scanner
Dimension Examples: Microtec (I): DenSCAN Defects (knots, rot/decay, stain, wane, holes, Cracks / Splits, pith, resin) Density Luxscan (L): X-Scan Fibre cross-sectional dimensions SilviScan (CSIRO, AUS) Microfibril angle Coarseness

SilviScan
On wood sample

(Defo, 2008)

SilviScan: Cell Scanner

(Defo, 2008)

SilviScan: X-ray Densitometer

(Defo, 2008)

SilviScan: X-ray Densitometer + Cell Scanner

(Defo, 2008)

X-Ray Diffractometry

(Defo, 2008)

X-Ray Diffractometry
Principle

Typical diffraction pattern & main cellulose-I planes

Relationship to MFA

Stiffness (MOE)
(Defo, 2008)

SilviScan measurements
Direct Measurements Derived Measurements

(Defo, 2008)

X-ray + Laser
Dimension, defects (knots, rot/decay, stain, wane, holes, Cracks / Splits, pith, resin) Ex: Microtec(I) Goldeneye

Stiffness (MOE)
Acoustic (Sonic) bands Stiffness = f (frequency of the sound, wood density) Portable Instrument
Hammer blow: Brookhuis MicroElectronics BV (NL) Timber Grader MTG

In-line instrument:
Vibration: Microtec (I) ViSCAN

Influence of the moisture content

Moisture Content

Microtec (I): M3SCAN Correction for density Wavelength??

Brookhuis Micro-Electronics BV (NL) FMI-L or FMI-X Correction for density (mass/vol) Dielectric constant

Dscher & Dscher (D): Timberscan or Venscan Microwave Density independent

SCS Forest Products (USA) (NMI from BC): MC Pro 1500 Transverse Moisture Density Sorter (MC PRO TRAC bar code integration system)
Measures MC and density (Microtec technology) Sorts the lumber based on MC and density Capacitance measurements

Kiln Drying

SCS Forest Products (USA): MC Pro 2000 Kiln Moisture Measurement System Capacitance measurements Moisture content of zones Average moisture content Schedule data, drying time, date stamp

NIR spectroscopy (experimental)


Surficial measurement Related to chemical composition (water, lignin, cellulose) Density influence removed
Specific absorption bands

Water, Cellulose and/or Lignin

Cellulose

Lignin

Species Discrimination
2 Spruce Jack pine Fir WSpruce Lodgepole Hemlock 1.5 1 0.5

-0.5

-1 -1.5

-2 -2 -1.8 -1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

(Cooper, Leblon et al. 2008)

Moisture content
60 Pr edicte d Y Slope 0. 980 02 7 0. 955 60 0 Offse t 0. 285 65 6 0. 629 39 9 RMSE 2. 091 68 6 3. 031 81 2 R- Squar e 0. 980 02 7 0. 960 27 6

Southern Pine

40

20

-2 0 -1 0 0 10 pls_ CS_EW, ( Y- var , PC): (MC,4 ) (MC,4 ) 20 30 40

Meas ure d Y 50

(Cooper, Leblon et al. 2008)

Effect of Surface Orientation


0.0015 PC2 5 0.0010 Scores

0.0005

86 2

-0.0005

-0.0010

3 5 7 24 3 6 5 7 3 12 1 7 4 4 18 7 16 20 29 2 9 1 9 3 25 5 0 19 2 15 24 13 7 2 9 26 0 23 5 22 1 6 11 8 10 1

28 5 3 28 25 17 29 20 19 14 7 1 23 8 22 13 2 31 24 1511 26 6 27 4 10 12 18 16

-0.0015 -0.003 RESULT22, X-expl: 79%,3% -0.002 -0.001 0 0.001

tangential radial radial-tangential cross-section


PC1 0.002 0.003 0.004

Lodgepole Pine

(Cooper, Leblon et al. 2008)

Wood Chemistry

(So et al. 2004)

Wood Quality
Density Microfibril Angle Stiffness

Lodgepole pine

(Schimleck et al. 2003)

Sawmill Trial (Grading based on stiffness) Correctly accepted cants Incorrectly accepted cants Correctly rejected cants Incorrectly rejected cants 18% 19% 53% 10% R&D needs: Sample measurements whole lumber Single spectra imaging systems

Pulp and Paper / BioEnergy


Chips, wood pellets, wood briquettes Volume Moisture content Chemical composition Color

Laser Beam
Chip volume:

Metso (Fin) VisIQ

Color camera + NIR


Chip Management System (CMS):
Digital camera chip brightness chip freshness NIR chip moisture content Bark and Plastic Detection

Chip Weighting System (CWS):


Wet/dry weight Volume sensor Mass and volume flow rates Bulk and basic Density

CRIQ Patents

Chip Sizing System (CSS):


3D Measures (width, length, thickness and area) Chip size distribution and index
(adapted from F. Ding)

Microwave
Long wave penetration Microwave = f(e)
Moisture Content

Dscher & Dscher (D): MoistureScan Microwave Density independent

NIR spectroscopy
Solid organic product chemical analysis Moisture content Paper sheet quality monitoring Ex: FOSS NIRSystems (USA)

Fourier-transform(FT)-IR
Chemical composition of liquors Ex: Effective alkali (EA) of the black liquor FP Innovation patent
FTNIR Auto-Titrator

(Kestner, Trung et al. 2004)

Wood Composites

NIR Spectroscopy
Stiffness assessment of Laminated Veneer Lumber (LVL) panels or plywoods

Density, MOE, MOR and internal bonds for medium density fiber boards (MDF) Ex: PanelPro of Metso Panelboard (Fin)

Oriented Strand Boards


Drier wood (Mountain Pine Beetle) fines NIR camera to estimate the % of fines FP Innovation Patent

(Groves 2007)

Oriented Strand Boards

(Groves 2007)

Thermal Imaging System


Termite activity in wood walls Monitoring kiln-drying temperatures

Cracks in lumbers

Thermascope SLK

Conclusions (I)
Variable Dimensions External Features Stiffness Forest/Yard (logwood) Laser Beam, Digital camera, IR sensor Acoustic sensor Plant (sawn wood) Laser Beam, Digital camera Acoustic sensor X-ray scanner

Dimensions Internal Features (defects)


Rot, decay Moisture Content

X-ray scanner or X-ray CT image

Acoustic sensor Pulsated current Acoustic sensor Pulsated current R&D need Microwave Capacitance

Conclusions (II)
Pulp and Paper / Biorefinery
Chip volume: Laser Beam Chip freshness: Digital camera Chip moisture content: NIR spectroscopy, Microwave Chemical composition: NIR spectroscopy, FT-IR spectroscopy Paper quality: NIR spectroscopy

Emerging technologies (R&D needs)


Moisture content for log wood (NIR spectroscopy, NMRI, microwave) NIR spectroscopy / hyperspectral imaging for log and sawn wood NMRI for solid wood and chips Neutron imaging for solid wood and chips

Acknowledgements
F. Fournier (FORINTEK), T. Trung (PAPRICAN), J. Nader (FERIC), K. Groves (FORINTEK) M. Defo (FORINTEK) Q. Wei (UNB), K. Tounis and P. Cooper (U of T),

F. Ding (CRIQ)

Thank you!

S-ar putea să vă placă și