Delivered by Ingenta toRagnar KullenbergIP 812247934

Tue 14 Dec 2010 031112

RESEARCH

ARTIC

LE

Copyright copy 2010 American Scientific PublishersAll rights reservedPrinted in the United States of America

Journal ofBiobased Materials and Bioenergy

Vol 4 363ndash366 2010

Dual-Energy X-Ray Absorptiometry Analysis for theDetermination of Moisture Content in Biomass

R Kullenberg1lowast M Hultnaumls2 V Fernandez2 M Nylinder2 S Toft3 and F Danielsson31Department of Radiation Physics University of Gothenburg SE41345 Gothenburg Sweden

2Department of Forest Products Swedish University of Agricultural Sciences SE75651 Uppsala Sweden3Mantex AB SE16440 Kista Sweden

The aim of the present study was to describe and validate a novel method able to fast accuratelyand with a high precision determine the moisture content of biological materials ranging 10 to70 A calibrated Dual X-ray photon absorptiometry technique is validated against moisture contentin wood chips Conventional criteria such as the Pearson correlation coefficient the root meansquare errors of calibration and coefficient of variation were used for the evaluation of the methodThe method is independent of the thickness and density of the object but dependent of the effectiveatomic number of the material The results shows that this photon absorptiometric method givesaccurate (lt2) and precise (lt1) results for moisture content determination in wood chips

Keywords Moisture Content Photon Absorptiometry Gravimetric Method Dual Energy X-RayAbsorptiometry Spruce Chips Pine Chips

1 INTRODUCTION

Previously known methods to estimate the amount ofmoisture content in wood material with reasonable goodaccuracy (lt1ndash2) involve drying of the material1 Thesegravimetric methods using hot air are often used as thestandard method for moisture content determination23

However such methods are cumbersome and tedious andit would normally take a day or more until a correct mea-sured value could be obtained which delays the overallprocessing It is therefore a need for a fast and reliablemethod to estimate the moisture contentSimilar needs exist in other industries handling biolog-

ical material For example it would be advantageous tohave a fast and reliable method for estimating the mois-ture content of biomass in the bio energy field in order tocontrol the burning process more precisely and improveits efficiencyWater can be found in most biological and non-

biological material in different forms Due to their hygro-scopic nature biological materials can also gain or losemoisture depending on the material-air interaction that takeplace during storage In addition to the gravimetric methodto measure the moisture content there is near infrared andmicrowave technology45 Also capacitance gauges may be

lowastAuthor to whom correspondence should be addressedEmail ragnarkullenbergteliacom

used as well as other electrical devices like conductivitymeasurement6 The limitation of above-mentioned mois-ture measurement technologies is often the dominant influ-ence of surface moisture content to the end result Alsoambient conditions for example snow and ice or tempera-ture below zero may decisively disturb the determinationThe use of photon absorptiometry has the potential to over-come the difficulties with present day methods since theradiation has penetrating power and snow and ice will notaffect the moisture measurementPhoton absorptiometry methods have been used in medi-

cal applications mainly for the determination of bone min-eral content in the skeleton Single X-ray absorptiometryhas been used for measurements on the peripheral skeletonand dual X-ray absorptiometry for measurements on theaxial skeleton7 Photon absorptiometry is sensitive to theeffective atomic number and density of the material beinganalyzed8 The absorption and scattering of photons in theenergy range that is most convenient for material analysisis a combination of two photonndashmatter interactions Thesetwo are called the photoelectric effect and Compton scat-tering and are energy dependentIn this work a new method using two different photon

X-ray energies from an X-ray tube will be presented Withthis method it is possible to determine the moisture con-tent in most biological materials for instance crops cottonpulp and wood In this paper the application of the methodfor moisture content in wood chips is described

J Biobased Mater Bioenergy 2010 Vol 4 No 4 1556-656020104363004 doi101166jbmb20101101 363

Delivered by Ingenta toRagnar KullenbergIP 812247934

Tue 14 Dec 2010 031112

RESEARCH

ARTIC

LE

Dual-Energy X-Ray Absorptiometry Analysis for the Determination of Moisture Content in Biomass Kullenberg et al

2 MATERIALS AND METHOD

A Mantex desktop scanner (Mantex AB Kista Sweden)has been used for the absorptiometric method measure-ments in this study This device consists of an X-ray tubeoperated at 90 kVp 2 mA and 40 kVp 5 mA This givesthat the effective energies was about 58 and 30 keV Thedetector consisted of a semiconductor linear array detectorThe samples were contained in plastic boxes about

15 cm in width and 21 cm in length and 17 cm in heightThe volume was about 3 liters These boxes were scannedonce for each energy with a total measurement time of60 seconds The measurement setup is shown in Figure 1The method has been calibrated for measurements of

spruce pine spruce and pine mixture and pine bark Thespruce and pine mixture consisted of about 50 sprucewood chips and 50 pine wood chips Samples of about05 m3 were collected and divided into subsamples ofabout 3 liters The different moisture levels were achievedby air drying The subsamples were for this purpose spreadon metal plates and allowed for drying several hours beforemeasurement Before measuring the subsamples they werehomogenized in a closed bucket in order to achieve ahomogenous sampleFigure 2 shows the measurement equipment used for

this studyThe Desktop scanner used in this study was lead

shielded in such a way that the dose equivalent rate doesnot exceed 02 Svh at a distance of 10 centimetres at anypoint around the box This could be compared to the doserate from natural background radiation in a normal envi-ronment in Sweden which is about 04 Svh A switch ismounted at the lid of the measurement space This turnsoff the X-ray equipment if the lid is opened during oper-ation The dose rate from the Mantex desktop device andthe design of the box shows that the device is an enclosedinstallation

3 THEORYCALCULATION

We assume that both photon energies are attenuated expo-nentially and thus the condition narrow beam geometry isvalid

Fig 1 Schematic drawing of the measurement setup

Fig 2 A Mantex desktop scanner was used for the measurements inthis study

The following relations are then valid

N1 = N01 expminus1X (1)

N2 = N02 expminus2X (2)

Where Ni is the observed transmitted countrate for energyi N01 is the countrate for energy I with no attenuator i

is the linear attenuation coefficients for energy I and X isthe thickness of the attenuatorEquations (1) and (2) can also be written

lnN01N1= 1X (3)

lnN02N2= 2X (4)

The countrates without attenuator could be determinedbesides the object or attenuatorFrom Eqs (3) and (4) an expression called k can be

derivedk = R1R2 (5)

where R1 = lnN01N1 and R2 = lnN02N2 andtherefore

k = 12 (6)

The expression (6) is independent of the thickness anddensity of the object It is only dependent of the linearattenuation coefficient The linear attenuation coefficientis dependent upon the effective atomic number of the tar-get material and the energy of the radiation8 This comesfrom the fact that the attenuation of different materialsdepends on the atomic number of the constituents andnot the chemical composition of these constituents In theenergy region used in this paper the photoelectric absorp-tion is important and the absorption per electron dependson about the third power of Z where Z is the atomicnumberThe effective atomic number of a compound can be

defined from the following expression9

Zeff =294 radicf 1times Z1294+f 2times Z2

294+f 3times Z3294

364 J Biobased Mater Bioenergy 4 363ndash366 2010

Delivered by Ingenta toRagnar KullenbergIP 812247934

Tue 14 Dec 2010 031112

RESEARCH

ARTIC

LE

Kullenberg et al Dual-Energy X-Ray Absorptiometry Analysis for the Determination of Moisture Content in Biomass

where f 1 f 2 and f 3 are the fractional contents of elec-trons belonging to elements Z1 Z2 and Z3 respectively

Since the effective atomic number of a mixture of dif-ferent elements will be dependent on the amount of waterit is possible to use the expression (5) to determine theamount of moisture in different materials where the onlyvariable component is the water content When the mois-ture content varies the effective atomic number will varyand thus the linear attenuation coefficient will varyBy plotting the change in k as a function of moisture

content linear regression could be used to fit the modelCalibration measurements have been carried out on sam-

ples of different biomasses For these calibration measure-ments the correct moisture contents have been determinedby the gravimetric method2 For each material 4 or 5 dif-ferent samples have been measured at least 3 times anddried in several steps Final drying of all samples havebeen in an oven at 105plusmn 2 degrees Celsius for at least24 hours and weighed The samples are completely driedwhen the weight is constant within 1 hourMoisture content has been calculated according to the

formula

MC= Wet weightminusDry weight)Wet weightlowast100The reproducibility was measured using the coefficient

of variation (CV= stdmeanlowast100) for 20 measurementson a sample of spruce chips after allowing the device towarm upIn order to evaluate if the measurements are affected if

the samples are frozen measurements were carried out onboth frozen and unfrozen samples

4 RESULTS

The parameter k has been determined for 4 different sam-ples of spruce wood chips with different amounts of watersee Table I The 4 samples have been dried in five stepsand measured three times at every step The moisture

Table I Moisture measurements on four samples of spruce wood chips

Sample 1 () k Sample 2 () k Sample 3 () k Sample 4 () k

5867 174687 5875 174657 5861 174665 5872 1746255867 174697 5875 174666 5861 174539 5872 174675867 174721 5875 17465 5861 174707 5872 1746585181 17331 5177 173436 5163 173409 5138 1734665181 17321 5177 173396 5163 173423 5138 173365181 173373 5177 17344 5163 173424 5138 1733044576 172402 4604 172552 4670 172718 4553 1721984576 172267 4604 172707 4670 172583 4553 1723234576 172428 4604 172618 4670 172755 4553 1722173855 171981 3893 171283 4188 171903 3939 1716763855 171686 3893 171366 4188 171965 3939 1716183855 17194 3893 171316 4188 172042 3939 1717012492 168982 2531 168776 2853 169962 2533 1689092492 169039 2531 168887 2853 170021 2533 1689312492 169027 2531 168776 2853 169993 2533 169095

contents were determined by weighing of the samples ateach drying step and also when the samples had been com-pletely dried From these measurements a calibration couldbe made by plotting the k-values against water or moisturecontent see Figure 3The reproducibility measurements on 20 spruce chips

samples gave a precision of 09In Table II are a few different materials and the lin-

ear relationship between the moisture content and k-valuepresentedIn order to validate the regression analysis measure-

ments were done on separate samples consisting of aspruce (70) and pine (30) mixture The results are plot-ted in Figure 4The materials in Table III were measured frozen and

then again after defrosting No significant differences werefound between frozen and not frozen samples at a confi-dence level of 95 The mean difference was 026 witha standard error of estimate (SEE) of 07

5 DISCUSSION

This report describes a novel method for moisture contentdetermination in biological samplesThe results show that a linear calibration equation

albeit with slightly differing coefficients for different mate-rials provides excellent fit (r2 values over 085) Themethod advantage is usability under many circumstancesThe results are for example not affected if the materialsare frozen it is a fast method and measures through thematerial The transmission measurement covers the entirematerial cross-section in the path of the photons and there-fore representative results will be achieved even when themoisture is not homogeneously distributed This measure-ment technology is non destructive and also contactlessand therefore does not influence the material or wear ofthe measuring systemThe results for moisture measurements of pine bark

were somewhat worse compared to spruce and pine This

J Biobased Mater Bioenergy 4 363ndash366 2010 365

Delivered by Ingenta toRagnar KullenbergIP 812247934

Tue 14 Dec 2010 031112

RESEARCH

ARTIC

LE

Dual-Energy X-Ray Absorptiometry Analysis for the Determination of Moisture Content in Biomass Kullenberg et al

Fig 3 Moisture content in spruce wood chips measured by the gravi-metric method and plotted as a function of k-values The correlationcoefficient R2 became 099 and the regression equation MC= 601lowastk-valueminus9906

Table II Linear relationship between moisture content (MC) and k-values correlation coefficients and standard error of estimate (SEE) forspruce pine pine bark and mixture of spruce and pine chips

Material Linear equation R2 SEE ()

Spruce MC= 601lowastk+9906 099 12Pine MC= 629lowastk+10390 098 18Pine bark MC= 125lowastk+21712 085 39Spruce and pine mixture MC= 610lowastk+10067 099 13

could be an effect of the more inhomogeneous samplesfrom bark The bark samples contained also soil and smallamounts of gravelThe precision shows that the method has a good repro-

ducibility and is well suited for repeated measurements ofthe same objectSince this method is independent of the density and

thickness of the material it could also be adapted to mea-surements on belt drives and in tubes Further studies areneeded in order to evaluate these applications Other appli-cations of this technology may be moisture determinationin paper pulp and in different biomasses in agricultureThe reference method to which the photon absorptio-

metric method has been compared was the gravimetricmethod2 This method has an accuracy of better than2 and a precision of better than 1 Both the preci-sion and accuracy of the Mantex absorptiometric devicewere similar to these figures However the oven dryingmethod is not an absolute method since biomass frequentlycontains varying amounts of volatile compounds whichmay evaporate while oven drying10 In order to quantifythe errors inherent in the oven drying method a studywhere resinous wood species are compared to non-resinousspeciessamples may be undertaken Since the accuracyof the absorptiometric method described in this paper isbased on the results of the gravimetric method the errorsinherent in the latter are superimposed the absorptiometricmethod The accuracy of the absorptiometric method willtherefore be better than the present results shows

Fig 4 Validation of regressions analysis on samples of a mixture ofspruce and pine chips The Moisture content measured by the desktopscanner was compared to the moisture content by the gravimetric methodThe correlation coefficient R2 became 099 and the standard error ofestimate (SEE) 19

Table III Comparison between measurements of frozen and not frozensamples

Moisture content

Frozen Not frozen Delta ()

Pine 129 134 minus05Pine 128 135 minus07Pine 96 89 07Pine 107 112 minus05Spruce 488 488 00Spruce 471 471 minus01Spruce 311 312 minus01Spruce 148 146 02Pine bark 659 679 minus20Pine bark 667 663 04Mean 320 323 03

References

1 R Samuelsson J Burvall and R Jirjis Biomass Bioenergy 30 929(2006)

2 SS-En 14774-2 Swedish standards institute Stockholm Sweden(2009)

3 ASTM E 871-82 American Society for Testing and Materials(ASTM) (1996)

4 P Daugbjerg Jensen H Hartman T Boumlhm M TemmermanF Rabier and M Morsing Biomass Bioenergy 30 935 (2006)

5 R King W S James and Y Y Yen Proc 1st Int Conf ScanningTechnol Sawmilling Miller Freeman Publications San FranciscoCA (1985)

6 A Fuchs M Moser H Zangl and T Bretterklieber Int J SmartSensing and Intell Systems l2 2 (2009)

7 G M Blake and I Fogelman Seminars in Nuclear Medicine 27 210(1997)

8 I Akkurt B Mavi B Akkurt C Basyigit S Kilincarslan and H AYali Journal of Quantitative Spectroscopy and Radiative Transfer94 3 (2005)

9 W V Mayenord The significance of the roumlntgen Acta of the Inter-national Union Against Cancer 2 (1937)

10 J Reeb and M Milota 50th Meeting Western Dry Kiln AssociationPortland (1999) pp 66ndash74

Received 12 February 2010 RevisedAccepted 5 June 2010

366 J Biobased Mater Bioenergy 4 363ndash366 2010

Delivered by Ingenta toRagnar KullenbergIP 812247934

Tue 14 Dec 2010 031112

RESEARCH

ARTIC

LE

Dual-Energy X-Ray Absorptiometry Analysis for the Determination of Moisture Content in Biomass Kullenberg et al

2 MATERIALS AND METHOD

A Mantex desktop scanner (Mantex AB Kista Sweden)has been used for the absorptiometric method measure-ments in this study This device consists of an X-ray tubeoperated at 90 kVp 2 mA and 40 kVp 5 mA This givesthat the effective energies was about 58 and 30 keV Thedetector consisted of a semiconductor linear array detectorThe samples were contained in plastic boxes about

15 cm in width and 21 cm in length and 17 cm in heightThe volume was about 3 liters These boxes were scannedonce for each energy with a total measurement time of60 seconds The measurement setup is shown in Figure 1The method has been calibrated for measurements of

spruce pine spruce and pine mixture and pine bark Thespruce and pine mixture consisted of about 50 sprucewood chips and 50 pine wood chips Samples of about05 m3 were collected and divided into subsamples ofabout 3 liters The different moisture levels were achievedby air drying The subsamples were for this purpose spreadon metal plates and allowed for drying several hours beforemeasurement Before measuring the subsamples they werehomogenized in a closed bucket in order to achieve ahomogenous sampleFigure 2 shows the measurement equipment used for

this studyThe Desktop scanner used in this study was lead

shielded in such a way that the dose equivalent rate doesnot exceed 02 Svh at a distance of 10 centimetres at anypoint around the box This could be compared to the doserate from natural background radiation in a normal envi-ronment in Sweden which is about 04 Svh A switch ismounted at the lid of the measurement space This turnsoff the X-ray equipment if the lid is opened during oper-ation The dose rate from the Mantex desktop device andthe design of the box shows that the device is an enclosedinstallation

3 THEORYCALCULATION

We assume that both photon energies are attenuated expo-nentially and thus the condition narrow beam geometry isvalid

Fig 1 Schematic drawing of the measurement setup

Fig 2 A Mantex desktop scanner was used for the measurements inthis study

The following relations are then valid

N1 = N01 expminus1X (1)

N2 = N02 expminus2X (2)

Where Ni is the observed transmitted countrate for energyi N01 is the countrate for energy I with no attenuator i

is the linear attenuation coefficients for energy I and X isthe thickness of the attenuatorEquations (1) and (2) can also be written

lnN01N1= 1X (3)

lnN02N2= 2X (4)

The countrates without attenuator could be determinedbesides the object or attenuatorFrom Eqs (3) and (4) an expression called k can be

derivedk = R1R2 (5)

where R1 = lnN01N1 and R2 = lnN02N2 andtherefore

k = 12 (6)

The expression (6) is independent of the thickness anddensity of the object It is only dependent of the linearattenuation coefficient The linear attenuation coefficientis dependent upon the effective atomic number of the tar-get material and the energy of the radiation8 This comesfrom the fact that the attenuation of different materialsdepends on the atomic number of the constituents andnot the chemical composition of these constituents In theenergy region used in this paper the photoelectric absorp-tion is important and the absorption per electron dependson about the third power of Z where Z is the atomicnumberThe effective atomic number of a compound can be

defined from the following expression9

Zeff =294 radicf 1times Z1294+f 2times Z2

294+f 3times Z3294

364 J Biobased Mater Bioenergy 4 363ndash366 2010

Delivered by Ingenta toRagnar KullenbergIP 812247934

Tue 14 Dec 2010 031112

RESEARCH

ARTIC

LE

Kullenberg et al Dual-Energy X-Ray Absorptiometry Analysis for the Determination of Moisture Content in Biomass

where f 1 f 2 and f 3 are the fractional contents of elec-trons belonging to elements Z1 Z2 and Z3 respectively

Since the effective atomic number of a mixture of dif-ferent elements will be dependent on the amount of waterit is possible to use the expression (5) to determine theamount of moisture in different materials where the onlyvariable component is the water content When the mois-ture content varies the effective atomic number will varyand thus the linear attenuation coefficient will varyBy plotting the change in k as a function of moisture

content linear regression could be used to fit the modelCalibration measurements have been carried out on sam-

ples of different biomasses For these calibration measure-ments the correct moisture contents have been determinedby the gravimetric method2 For each material 4 or 5 dif-ferent samples have been measured at least 3 times anddried in several steps Final drying of all samples havebeen in an oven at 105plusmn 2 degrees Celsius for at least24 hours and weighed The samples are completely driedwhen the weight is constant within 1 hourMoisture content has been calculated according to the

formula

MC= Wet weightminusDry weight)Wet weightlowast100The reproducibility was measured using the coefficient

of variation (CV= stdmeanlowast100) for 20 measurementson a sample of spruce chips after allowing the device towarm upIn order to evaluate if the measurements are affected if

the samples are frozen measurements were carried out onboth frozen and unfrozen samples

4 RESULTS

The parameter k has been determined for 4 different sam-ples of spruce wood chips with different amounts of watersee Table I The 4 samples have been dried in five stepsand measured three times at every step The moisture

Table I Moisture measurements on four samples of spruce wood chips

Sample 1 () k Sample 2 () k Sample 3 () k Sample 4 () k

5867 174687 5875 174657 5861 174665 5872 1746255867 174697 5875 174666 5861 174539 5872 174675867 174721 5875 17465 5861 174707 5872 1746585181 17331 5177 173436 5163 173409 5138 1734665181 17321 5177 173396 5163 173423 5138 173365181 173373 5177 17344 5163 173424 5138 1733044576 172402 4604 172552 4670 172718 4553 1721984576 172267 4604 172707 4670 172583 4553 1723234576 172428 4604 172618 4670 172755 4553 1722173855 171981 3893 171283 4188 171903 3939 1716763855 171686 3893 171366 4188 171965 3939 1716183855 17194 3893 171316 4188 172042 3939 1717012492 168982 2531 168776 2853 169962 2533 1689092492 169039 2531 168887 2853 170021 2533 1689312492 169027 2531 168776 2853 169993 2533 169095

contents were determined by weighing of the samples ateach drying step and also when the samples had been com-pletely dried From these measurements a calibration couldbe made by plotting the k-values against water or moisturecontent see Figure 3The reproducibility measurements on 20 spruce chips

samples gave a precision of 09In Table II are a few different materials and the lin-

ear relationship between the moisture content and k-valuepresentedIn order to validate the regression analysis measure-

ments were done on separate samples consisting of aspruce (70) and pine (30) mixture The results are plot-ted in Figure 4The materials in Table III were measured frozen and

then again after defrosting No significant differences werefound between frozen and not frozen samples at a confi-dence level of 95 The mean difference was 026 witha standard error of estimate (SEE) of 07

5 DISCUSSION

This report describes a novel method for moisture contentdetermination in biological samplesThe results show that a linear calibration equation

albeit with slightly differing coefficients for different mate-rials provides excellent fit (r2 values over 085) Themethod advantage is usability under many circumstancesThe results are for example not affected if the materialsare frozen it is a fast method and measures through thematerial The transmission measurement covers the entirematerial cross-section in the path of the photons and there-fore representative results will be achieved even when themoisture is not homogeneously distributed This measure-ment technology is non destructive and also contactlessand therefore does not influence the material or wear ofthe measuring systemThe results for moisture measurements of pine bark

were somewhat worse compared to spruce and pine This

J Biobased Mater Bioenergy 4 363ndash366 2010 365

Delivered by Ingenta toRagnar KullenbergIP 812247934

Tue 14 Dec 2010 031112

RESEARCH

ARTIC

LE

Dual-Energy X-Ray Absorptiometry Analysis for the Determination of Moisture Content in Biomass Kullenberg et al

Fig 3 Moisture content in spruce wood chips measured by the gravi-metric method and plotted as a function of k-values The correlationcoefficient R2 became 099 and the regression equation MC= 601lowastk-valueminus9906

Table II Linear relationship between moisture content (MC) and k-values correlation coefficients and standard error of estimate (SEE) forspruce pine pine bark and mixture of spruce and pine chips

Material Linear equation R2 SEE ()

Spruce MC= 601lowastk+9906 099 12Pine MC= 629lowastk+10390 098 18Pine bark MC= 125lowastk+21712 085 39Spruce and pine mixture MC= 610lowastk+10067 099 13

could be an effect of the more inhomogeneous samplesfrom bark The bark samples contained also soil and smallamounts of gravelThe precision shows that the method has a good repro-

ducibility and is well suited for repeated measurements ofthe same objectSince this method is independent of the density and

thickness of the material it could also be adapted to mea-surements on belt drives and in tubes Further studies areneeded in order to evaluate these applications Other appli-cations of this technology may be moisture determinationin paper pulp and in different biomasses in agricultureThe reference method to which the photon absorptio-

metric method has been compared was the gravimetricmethod2 This method has an accuracy of better than2 and a precision of better than 1 Both the preci-sion and accuracy of the Mantex absorptiometric devicewere similar to these figures However the oven dryingmethod is not an absolute method since biomass frequentlycontains varying amounts of volatile compounds whichmay evaporate while oven drying10 In order to quantifythe errors inherent in the oven drying method a studywhere resinous wood species are compared to non-resinousspeciessamples may be undertaken Since the accuracyof the absorptiometric method described in this paper isbased on the results of the gravimetric method the errorsinherent in the latter are superimposed the absorptiometricmethod The accuracy of the absorptiometric method willtherefore be better than the present results shows

Fig 4 Validation of regressions analysis on samples of a mixture ofspruce and pine chips The Moisture content measured by the desktopscanner was compared to the moisture content by the gravimetric methodThe correlation coefficient R2 became 099 and the standard error ofestimate (SEE) 19

Table III Comparison between measurements of frozen and not frozensamples

Moisture content

Frozen Not frozen Delta ()

Pine 129 134 minus05Pine 128 135 minus07Pine 96 89 07Pine 107 112 minus05Spruce 488 488 00Spruce 471 471 minus01Spruce 311 312 minus01Spruce 148 146 02Pine bark 659 679 minus20Pine bark 667 663 04Mean 320 323 03

References

1 R Samuelsson J Burvall and R Jirjis Biomass Bioenergy 30 929(2006)

2 SS-En 14774-2 Swedish standards institute Stockholm Sweden(2009)

3 ASTM E 871-82 American Society for Testing and Materials(ASTM) (1996)

4 P Daugbjerg Jensen H Hartman T Boumlhm M TemmermanF Rabier and M Morsing Biomass Bioenergy 30 935 (2006)

5 R King W S James and Y Y Yen Proc 1st Int Conf ScanningTechnol Sawmilling Miller Freeman Publications San FranciscoCA (1985)

6 A Fuchs M Moser H Zangl and T Bretterklieber Int J SmartSensing and Intell Systems l2 2 (2009)

7 G M Blake and I Fogelman Seminars in Nuclear Medicine 27 210(1997)

8 I Akkurt B Mavi B Akkurt C Basyigit S Kilincarslan and H AYali Journal of Quantitative Spectroscopy and Radiative Transfer94 3 (2005)

9 W V Mayenord The significance of the roumlntgen Acta of the Inter-national Union Against Cancer 2 (1937)

10 J Reeb and M Milota 50th Meeting Western Dry Kiln AssociationPortland (1999) pp 66ndash74

Received 12 February 2010 RevisedAccepted 5 June 2010

366 J Biobased Mater Bioenergy 4 363ndash366 2010

Delivered by Ingenta toRagnar KullenbergIP 812247934

Tue 14 Dec 2010 031112

RESEARCH

ARTIC

LE

Kullenberg et al Dual-Energy X-Ray Absorptiometry Analysis for the Determination of Moisture Content in Biomass

where f 1 f 2 and f 3 are the fractional contents of elec-trons belonging to elements Z1 Z2 and Z3 respectively

Since the effective atomic number of a mixture of dif-ferent elements will be dependent on the amount of waterit is possible to use the expression (5) to determine theamount of moisture in different materials where the onlyvariable component is the water content When the mois-ture content varies the effective atomic number will varyand thus the linear attenuation coefficient will varyBy plotting the change in k as a function of moisture

content linear regression could be used to fit the modelCalibration measurements have been carried out on sam-

ples of different biomasses For these calibration measure-ments the correct moisture contents have been determinedby the gravimetric method2 For each material 4 or 5 dif-ferent samples have been measured at least 3 times anddried in several steps Final drying of all samples havebeen in an oven at 105plusmn 2 degrees Celsius for at least24 hours and weighed The samples are completely driedwhen the weight is constant within 1 hourMoisture content has been calculated according to the

formula

MC= Wet weightminusDry weight)Wet weightlowast100The reproducibility was measured using the coefficient

of variation (CV= stdmeanlowast100) for 20 measurementson a sample of spruce chips after allowing the device towarm upIn order to evaluate if the measurements are affected if

the samples are frozen measurements were carried out onboth frozen and unfrozen samples

4 RESULTS

The parameter k has been determined for 4 different sam-ples of spruce wood chips with different amounts of watersee Table I The 4 samples have been dried in five stepsand measured three times at every step The moisture

Table I Moisture measurements on four samples of spruce wood chips

Sample 1 () k Sample 2 () k Sample 3 () k Sample 4 () k

5867 174687 5875 174657 5861 174665 5872 1746255867 174697 5875 174666 5861 174539 5872 174675867 174721 5875 17465 5861 174707 5872 1746585181 17331 5177 173436 5163 173409 5138 1734665181 17321 5177 173396 5163 173423 5138 173365181 173373 5177 17344 5163 173424 5138 1733044576 172402 4604 172552 4670 172718 4553 1721984576 172267 4604 172707 4670 172583 4553 1723234576 172428 4604 172618 4670 172755 4553 1722173855 171981 3893 171283 4188 171903 3939 1716763855 171686 3893 171366 4188 171965 3939 1716183855 17194 3893 171316 4188 172042 3939 1717012492 168982 2531 168776 2853 169962 2533 1689092492 169039 2531 168887 2853 170021 2533 1689312492 169027 2531 168776 2853 169993 2533 169095

contents were determined by weighing of the samples ateach drying step and also when the samples had been com-pletely dried From these measurements a calibration couldbe made by plotting the k-values against water or moisturecontent see Figure 3The reproducibility measurements on 20 spruce chips

samples gave a precision of 09In Table II are a few different materials and the lin-

ear relationship between the moisture content and k-valuepresentedIn order to validate the regression analysis measure-

ments were done on separate samples consisting of aspruce (70) and pine (30) mixture The results are plot-ted in Figure 4The materials in Table III were measured frozen and

then again after defrosting No significant differences werefound between frozen and not frozen samples at a confi-dence level of 95 The mean difference was 026 witha standard error of estimate (SEE) of 07

5 DISCUSSION

This report describes a novel method for moisture contentdetermination in biological samplesThe results show that a linear calibration equation

albeit with slightly differing coefficients for different mate-rials provides excellent fit (r2 values over 085) Themethod advantage is usability under many circumstancesThe results are for example not affected if the materialsare frozen it is a fast method and measures through thematerial The transmission measurement covers the entirematerial cross-section in the path of the photons and there-fore representative results will be achieved even when themoisture is not homogeneously distributed This measure-ment technology is non destructive and also contactlessand therefore does not influence the material or wear ofthe measuring systemThe results for moisture measurements of pine bark

were somewhat worse compared to spruce and pine This

J Biobased Mater Bioenergy 4 363ndash366 2010 365

Delivered by Ingenta toRagnar KullenbergIP 812247934

Tue 14 Dec 2010 031112

RESEARCH

ARTIC

LE

Dual-Energy X-Ray Absorptiometry Analysis for the Determination of Moisture Content in Biomass Kullenberg et al

Fig 3 Moisture content in spruce wood chips measured by the gravi-metric method and plotted as a function of k-values The correlationcoefficient R2 became 099 and the regression equation MC= 601lowastk-valueminus9906

Table II Linear relationship between moisture content (MC) and k-values correlation coefficients and standard error of estimate (SEE) forspruce pine pine bark and mixture of spruce and pine chips

Material Linear equation R2 SEE ()

Spruce MC= 601lowastk+9906 099 12Pine MC= 629lowastk+10390 098 18Pine bark MC= 125lowastk+21712 085 39Spruce and pine mixture MC= 610lowastk+10067 099 13

could be an effect of the more inhomogeneous samplesfrom bark The bark samples contained also soil and smallamounts of gravelThe precision shows that the method has a good repro-

ducibility and is well suited for repeated measurements ofthe same objectSince this method is independent of the density and

thickness of the material it could also be adapted to mea-surements on belt drives and in tubes Further studies areneeded in order to evaluate these applications Other appli-cations of this technology may be moisture determinationin paper pulp and in different biomasses in agricultureThe reference method to which the photon absorptio-

metric method has been compared was the gravimetricmethod2 This method has an accuracy of better than2 and a precision of better than 1 Both the preci-sion and accuracy of the Mantex absorptiometric devicewere similar to these figures However the oven dryingmethod is not an absolute method since biomass frequentlycontains varying amounts of volatile compounds whichmay evaporate while oven drying10 In order to quantifythe errors inherent in the oven drying method a studywhere resinous wood species are compared to non-resinousspeciessamples may be undertaken Since the accuracyof the absorptiometric method described in this paper isbased on the results of the gravimetric method the errorsinherent in the latter are superimposed the absorptiometricmethod The accuracy of the absorptiometric method willtherefore be better than the present results shows

Fig 4 Validation of regressions analysis on samples of a mixture ofspruce and pine chips The Moisture content measured by the desktopscanner was compared to the moisture content by the gravimetric methodThe correlation coefficient R2 became 099 and the standard error ofestimate (SEE) 19

Table III Comparison between measurements of frozen and not frozensamples

Moisture content

Frozen Not frozen Delta ()

Pine 129 134 minus05Pine 128 135 minus07Pine 96 89 07Pine 107 112 minus05Spruce 488 488 00Spruce 471 471 minus01Spruce 311 312 minus01Spruce 148 146 02Pine bark 659 679 minus20Pine bark 667 663 04Mean 320 323 03

References

1 R Samuelsson J Burvall and R Jirjis Biomass Bioenergy 30 929(2006)

2 SS-En 14774-2 Swedish standards institute Stockholm Sweden(2009)

3 ASTM E 871-82 American Society for Testing and Materials(ASTM) (1996)

4 P Daugbjerg Jensen H Hartman T Boumlhm M TemmermanF Rabier and M Morsing Biomass Bioenergy 30 935 (2006)

5 R King W S James and Y Y Yen Proc 1st Int Conf ScanningTechnol Sawmilling Miller Freeman Publications San FranciscoCA (1985)

6 A Fuchs M Moser H Zangl and T Bretterklieber Int J SmartSensing and Intell Systems l2 2 (2009)

7 G M Blake and I Fogelman Seminars in Nuclear Medicine 27 210(1997)

8 I Akkurt B Mavi B Akkurt C Basyigit S Kilincarslan and H AYali Journal of Quantitative Spectroscopy and Radiative Transfer94 3 (2005)

9 W V Mayenord The significance of the roumlntgen Acta of the Inter-national Union Against Cancer 2 (1937)

10 J Reeb and M Milota 50th Meeting Western Dry Kiln AssociationPortland (1999) pp 66ndash74

Received 12 February 2010 RevisedAccepted 5 June 2010

366 J Biobased Mater Bioenergy 4 363ndash366 2010

Delivered by Ingenta toRagnar KullenbergIP 812247934

Tue 14 Dec 2010 031112

RESEARCH

ARTIC

LE

Dual-Energy X-Ray Absorptiometry Analysis for the Determination of Moisture Content in Biomass Kullenberg et al

Fig 3 Moisture content in spruce wood chips measured by the gravi-metric method and plotted as a function of k-values The correlationcoefficient R2 became 099 and the regression equation MC= 601lowastk-valueminus9906

Table II Linear relationship between moisture content (MC) and k-values correlation coefficients and standard error of estimate (SEE) forspruce pine pine bark and mixture of spruce and pine chips

Material Linear equation R2 SEE ()

Spruce MC= 601lowastk+9906 099 12Pine MC= 629lowastk+10390 098 18Pine bark MC= 125lowastk+21712 085 39Spruce and pine mixture MC= 610lowastk+10067 099 13

could be an effect of the more inhomogeneous samplesfrom bark The bark samples contained also soil and smallamounts of gravelThe precision shows that the method has a good repro-

ducibility and is well suited for repeated measurements ofthe same objectSince this method is independent of the density and

thickness of the material it could also be adapted to mea-surements on belt drives and in tubes Further studies areneeded in order to evaluate these applications Other appli-cations of this technology may be moisture determinationin paper pulp and in different biomasses in agricultureThe reference method to which the photon absorptio-

metric method has been compared was the gravimetricmethod2 This method has an accuracy of better than2 and a precision of better than 1 Both the preci-sion and accuracy of the Mantex absorptiometric devicewere similar to these figures However the oven dryingmethod is not an absolute method since biomass frequentlycontains varying amounts of volatile compounds whichmay evaporate while oven drying10 In order to quantifythe errors inherent in the oven drying method a studywhere resinous wood species are compared to non-resinousspeciessamples may be undertaken Since the accuracyof the absorptiometric method described in this paper isbased on the results of the gravimetric method the errorsinherent in the latter are superimposed the absorptiometricmethod The accuracy of the absorptiometric method willtherefore be better than the present results shows

Fig 4 Validation of regressions analysis on samples of a mixture ofspruce and pine chips The Moisture content measured by the desktopscanner was compared to the moisture content by the gravimetric methodThe correlation coefficient R2 became 099 and the standard error ofestimate (SEE) 19

Table III Comparison between measurements of frozen and not frozensamples

Moisture content

Frozen Not frozen Delta ()

Pine 129 134 minus05Pine 128 135 minus07Pine 96 89 07Pine 107 112 minus05Spruce 488 488 00Spruce 471 471 minus01Spruce 311 312 minus01Spruce 148 146 02Pine bark 659 679 minus20Pine bark 667 663 04Mean 320 323 03

References

1 R Samuelsson J Burvall and R Jirjis Biomass Bioenergy 30 929(2006)

2 SS-En 14774-2 Swedish standards institute Stockholm Sweden(2009)

3 ASTM E 871-82 American Society for Testing and Materials(ASTM) (1996)

4 P Daugbjerg Jensen H Hartman T Boumlhm M TemmermanF Rabier and M Morsing Biomass Bioenergy 30 935 (2006)

5 R King W S James and Y Y Yen Proc 1st Int Conf ScanningTechnol Sawmilling Miller Freeman Publications San FranciscoCA (1985)

6 A Fuchs M Moser H Zangl and T Bretterklieber Int J SmartSensing and Intell Systems l2 2 (2009)

7 G M Blake and I Fogelman Seminars in Nuclear Medicine 27 210(1997)

8 I Akkurt B Mavi B Akkurt C Basyigit S Kilincarslan and H AYali Journal of Quantitative Spectroscopy and Radiative Transfer94 3 (2005)

9 W V Mayenord The significance of the roumlntgen Acta of the Inter-national Union Against Cancer 2 (1937)

10 J Reeb and M Milota 50th Meeting Western Dry Kiln AssociationPortland (1999) pp 66ndash74

Received 12 February 2010 RevisedAccepted 5 June 2010

366 J Biobased Mater Bioenergy 4 363ndash366 2010