Instrument Cos. (USS). Operational Costs. (US $). S/ Fruit* @ various grading speeds. (seconds) .... T.-M. Jin. L. Liu. X.-W. Tang. T.M. Jin, L. Liu and X.W. Tang.
09.2-4
Challenges and opportunities in the use of near infrared for the analysis of
intact, high moisture plant products Stanley J. Kays, Gerald G. Dull and Richard G. Leffler
Depanment of Horticulture. The University of Georgia.Athens. OA30602-7273. USA. Introduction
Near infrared (NIR) spectroscopy has a wide range of advantages for the quality assessment of in
tact fruits, vegetables and other high-moisture products. It is non-destructive, fast, highly cost^eclive in appropriate applications and requires no sample preparation or chemical
contrasted wtih other uses for NIR.n i stam l ent deveo l pment for hg i h mositwe crops has Pfog^^^e^ slowly over the past decade due. in part, to a unique set of problems presemed by fruits
Constraints include: 1) a high product moisture content which absorbs NIR
2) substanta i l vara i to i nn i product szie and shape whcih resutlsn i ann i conssitem ^P^
and 3) variation in the composition of target attributes throughout individual prc^uct units. Recent
advances have begun to circumvent these constraints, seting the stage for the rapid commercialisadon of economically viable applications. The folowing article reviews current applications and economic considerations relative to instrument development and use.
Fruti and vegetabe l s thate l nd themsevles to quatiyl assessment usn i g NIR tend to haven i dvid i ual
product units that are of reasonably high value and a quality atribute(s) that: sumers' percepto i n of the vau l e of the product; 2) can be measured accurateyl Md 3) can ascertained by the customer prior to purchase. Critcal quality trarts tend to be senso^ tn NIR analysis involves the correlation of spectral data with these atnbutes. ImportMt traits mcMe in teroal components modua l tn i g taste (sweet and sour), dry mater and n i ternal defects; traB i tat cm not be accurately determined by consumers prior to purchase via external Msessment b"'
tremeyl m i portant n i the consumers eventual satsi facto i n wtih the product. Thus, gradn i g frati M
vegetabe l susn i gNIRprovd i eswhoe l sae l rsandretae lirswth i theabtyil toguaranteethequatyil ofthe product their customers purchase.
Applications and instruments
Transmission NIR assessment techniques have been established in ^
cross-section of quality attributes and crops (Table 1). Quality traits range from acidity to moisture
content wtih sou l be l sod il s and dry mater the most frequentyl measured. In most mst^ces. ^^atsi de
scribed as sugar content is a measure of soluble solids (Brix).Applications in which cominercial in struments are reportedly available are indicated in Table 1. The rapid increase in NIR aoDcilato i ns over the past severalyears has been due. predomn i ateyl, to four compane i s (Tabe l 2).
Ji^^Ja?SarsorcpreL„,crn icaL l„cem.andNIRhasbeenshown»beefccuvcm.denofymg defects in several products, facilitating the removal of infenor matenal (Table 3).
Figure 1 ilustrates the performance of NIR for three non-destructive quality evaluMon applica to i ns. We rea l to i nshp i between actual sou l be l sod il s n i peaches v. that predci ted usn i g NIR si gvi en n i
NIR Analysis on Intact, High Moisture Plant Products
Table 1. Potential and current transmission NIR applications and instruments for the non-destructive quality evaluation of high moisture crops. Applications for which instruments are report^ to be avail
able are listed in bold (abbreviations: A = acidity, DM = dry matter, F = firmness, M = moisture,
OA = organic acids, SS = soluble solids, S = starch, SC = sugar content*). ^
Apricot" (OA):Apple" (A.DM.F.M.SS.SC); Cantaloupe' (SS.SC): Chinese cabbage" (SS.SC); Cucumber" (DM)- Date' (M); Honeydew melon' (SS.SC); Japanese pear (SS.SC); Japanese per^mmon (SS.SC); Kiwifmit"^ (DM.SS); Mango" (DM.SS); Mushroom" (DM); Nectarine" (SS); Ooion m M.SSy. Orange^
(A SS.SC); Papaya' (SS); Peach"(A.SS.SC); Pear'" (A.SS.SC); Peppermint" (SS.S); Pjneapple (SS); Plum" (SS); Pomelos (A.SS.SC): Potato""" (DM.S); Strawberry"' (SC); Tangenne (A.SS.SC); Tomato"" " (A.OA); Watermelon (SS.SC).
Table 2. Companies currently marketing transmis
Table 3. The use of transmittance NIR for
sion NIR instruments for the non-destructive qual
identifyng internal defects of intact fruit and
ity evaluation of fruit and vegetables.
vegetables.
Company
Location
Product
Disorder
Agricultural Innovations
United Sutes
Apple
Water core
]Moud l
Fantec
Japan
Peanut
Mitsui Mining and Smelting
Japan
Pouto
Hollow heart
Sumitomo Metal Mining
Japan
Tangerine
Section drying
Citation 5 13 2 27
Figure 1(a)'. The data represents four cultivars (Redhaven. Windblo. Blake and Encore) over three
years.The e l vel of precsio i n si suffice i nt for peaches to Ij readyli sorted n i to three sweetness ca l sses based upon their soluble solids content (i.e. sweetness)." . j u j • ,
Amajor portion of the tomatoes grown worldwide are reduced in volume through dehydrauon to a
paste or sauce. Increasn i g the fruti sod il s content from 5% to 6% woud l represent an n i cre^m vau le of US$70-80 milion a year for the processing tomato industry in the United States alone. The poten
tial to rapidly measure soluble solids would be highly advantageous as a selecuon tool in tomato breeding programmes focusing on high solids and in processing plants where the pnce paid for raw
products is tied to solids content. In the NIR application ilustrated in Figure 1(b). Je soluble solids
content of the cooked product is predicted from fresh, unprocessed fruit using three diferent methods of analysis [multiple linear regression, partial least squares regression and neural network]. A third example of an NIR appUcation is for an internal disorder in citrus fruits. A significant prob
lem in citrus fruit is a physiological disorder caled dry vesicle. The cels in the vesicles appear to^
firm ("granulated") and when squeezed release virtually no juice. Since the disorder is internal, the fruit can not be separated using external symptoms nor using density diferences. Dry vesicle occurs in virtually ail citrus crops and increases with the length of storage. Ininal work, using the tangenne^ a model, has indicated that high quality and defective fruit can be readily separated [Figure l(c)J. Economic considerations
The ability to accurately determine internal quality attributes in the laboratory or under simulated packing house conditons does not guarantee the commercialisation of an application.Adoption of an application is largely an economic decision involving the cost to benefit rati^or boti the insdument manufacturer and the user. Due to questions about the econonuc viability of NIR grading of high mois-
Stanley J. Kays, Gerald G. Dull and Richard G. Leffler lure crops, the following analysis is presented. Cantaloupe production in the United States is
z g
used as the model and estimates of grading
costs/fruit are then applied to a cross-section of other fruit crops.
For a manufacturer exploring the possibility
of developing an instrument for a specific appli
cation. two critical considerations are; I) how
many instruments can be sold and 2) at what
price? To address these questions, projections of
the cost/fruit for grading cantaloupes for sweet ness (soluble solids) in the United States were calculated (Table 4). The annual production of
cantaloupes in the United States (1998) was 1.9
million metric tons" for which the following
criteria were imposed: 1) only one half of the
crop is graded; 2) the average weight of an indi
vidual fruit is approximately 1.8 kg; 3) the aver
age length of the cantaloupe season for a packing house is 70 days; 4) the packing house lines run 12 hours a day; 5) an NIR instrument
has a life expectancy of five years and 6) the in
strument maintenance costs are US$5,000 a
year. Using these criteria, the number of instru ments required to grade one half of the United
States cantaloupe crop can be calculated (Table
4). The number of instruments required ranged
from 34 to 853. depending upon the rate at
which individual product units could be graded. If an instrument has an estimated life expec
tancy of five years, then the total number of fruit
that can be graded/instrument during this time
period ranged from 75.8 x 10' for the fastest
grading rate (0.2 s per fruit) to 3.0 x 10' for the S«con
Figure 1. Examples of NIR applications for the non-destructive quality evaluation of high mois
ture crops, (a) The relationship between labora
tory determined peach soluble solids content and that predicted using NIR." (b) Cumulative per
centage of tomato fruit plotted against the abso
lute residual in percent soluble solids as classified
by multiple linear regression (MLR), partial least squares (PLS) regression, and neural network (NN)
calibrations." (c) The relationship between sec ond derivative optical density values at 768 nm
and 960 nm for normal and defective sections of tangerine."
slowest (5.0 s per fruit).
The second consideration from a manufac
turing stand-point is the retail price of an instru ment. If the purchase priceyinstrument is varied
from US$50,000 to $500,000 (depreciated over
five years), the maintenance costs are $5.0(W a year and the rate of grading an individual fruit is
varied from 0.2 to 5 seconds, the cost per fruit
ranged from $0.00099 for the fastest and least
expensive instrument (Table 5) to $0.17301 for the slowest and most expensive instrument.
What are realistic grading rates for on-line NIR instruments? Losses in precision with increas
ing rates vary with the type of measurement
NIR Analysis on Intact, High Moisture Plant Products
Table 4. The number of instruments required to grade 50% of the U.S. cantaloupe crop. I I Trtwi r.rnHino Tinw?' I InslTumeiits Required Total fruit graded in
5 years/instrument (x 10')
• Grading 70 days. 12 hr a day
Tbble 5. Grading cost per fruit at various instrument purchase prices and grading speeds K
j
I
W
I
V
Instrument Cos
Operational Costs
(USS)
(US $)
S/ Fruit* @ various grading speeds (seconds) .2
.5
1,0
5.0
25,000
.00099
,00247
,00494
,02471
50.000
25,000
,00165
,00412
,00824
,04119
100,000
25,000
,00297
,00741
,01483
,07414
200.000
,01730
,03460
.17301
500.000
JW.WV
25,000
I
,00692 I
,
•Instrument + operational costs / (# fruit in 5 years / instrument) for each grading speed
made (for exampe l , spectral anayl ssi v.sn i ge l wavee l ngth readn i gs), type of °fSrcond PromoUonal material for existing instntments give gradmg rates
thoueh theoretical rates as high as 20 fhit/second have been proposed. The relauonship between
speedandaccuracy,however,hasyettobeadequateyl expo l red.Forcomparsionpurposes,however, rates from 0.2 to 5.0 seconds/fruit were factored against various instrument costs (purchase + mainte nance) to calculate the grading cost/individual fruit (data not presented). The increase in price/kg of fruit, therefore, is a function of the cost of grading
and the size of the individual fruit. The cost increase
ranged from $0.00009 kg"' for watermelons at the lowest grading cost/fruit ($0.001) to $39.70 kg for
blueberries at the highest grading cost/fruit ($0.05). Thus, as the value of an individual fruit declines,
grading has a progressively greater impact on the re tail price, as ilustrated for a cross-section of crops in Figure 2. The percentage increase in retail price per kg ranged from 0.03% for watermelons at the low est grading price/fruit to 755% for blueberries at the highest grading price per fruit. Since grading costs are passed along to the con sumer as part of the retail price, what is an accept-
Figure 2. The relationship between the value
of individual product units for a cross-section of fruits and the percentage increase in retail
price at three grading costs due to NIR non-destructive quality grading.
Stanley J. Kays, Gerald G. Dull and Richard G. Leffler
able additional cost and at what point does it become unacceptable? A wide range of factors mo
