Annual review of.the glass industry - Laboratory tecjmiques
`
`DUALITY CONTROL AND DUALITY ASSURANCE RELATED TO
`RAW MATERIALS AND MELTING IN A MODERN CONTAINER
`GLASS PLANT - PART 1
`
`by Peter Buchmayer*
`
`Glass is one of the few industries to manufacture its own material
`- molten glass -
`for its finished products. Hence, the problems
`of quality control and assurance in a glass factory are quite
`different to those in other industries. Whereas the task of quality
`control of finished products is to provide quality assurance with
`regard to the economic value, the aim of quality control, in the
`field of raw materials and melting, serves to maintain optimum
`consistancy of that raw material called glass, with respect to its
`workability as well as to its chemical and physical properties. This
`article (the first part of which follows) reviews the subject in
`significant detail. Part Two w111 appear in the October 1986 edition.
`
`the quality criterion is not exclusively in(cid:173)
`fluenced by the content of glass forming
`oxides and the consistancy of the composi(cid:173)
`tion but, in particular, by the content of
`those constituents which are detrimental to
`the glass, such as certain metal oxides.
`In the past, the relationship between raw
`material costs and production costs, together
`with the conservative and more empiric
`opinions of glassmakers, were the reasons
`for using high quality raw materials, even for
`cheap mass-produced glass containers. For
`the future, this can only mean that the
`quality of raw materials must correspond to
`the use of the finished glass products.
`
`Sampling
`The first and most important step in the
`quality control of new materials is sampling.
`The supervising of the chemical composition,
`physical properties, grain size distribution
`and bulk density requires sample§ .re12resent(cid:173)
`ing the whole bulk of the material. The main
`criteria for sampling procedures are the
`grain sizes of the raw materials and the dis(cid:173)
`tribution of components. It is an advantage,
`
`Fig 1. Sampling with tube sampler.
`
`Due to the very different methods, facilities
`and specific knowhow required for the
`quality control of glass and raw material on
`the one hand and for containers on the other,
`these tasks, nowadays, are mainly performed
`by different departments of the glass plant.
`Also, today, normal practice is for the con(cid:173)
`trol of raw materials, cullet and the chemical
`and physical properties of the melt and the
`glass to be carried out by the plant chemical
`laboratory, an indispensible sector in a
`modern glass factory.
`
`Work of the chemical department
`The work of this department can be divided
`into three main categories:
`Testing of all incoming raw materials, auxili(cid:173)
`ary materials and particularly cullet;
`Control and assurance of the chemical and
`physical properties of the glass; and
`Identification of glass defects and their origin
`in the area of raw materials, cullet and
`melting.
`The fundamental objective of these in(cid:173)
`vestigations is to reveal deviations from
`established limits and to obtain new data for
`preventive batch corrections.
`Two important points influence the work
`of quality control departments in a modern
`container glass factory:
`Steady increase in profitability by optimising
`furnace and production output -
`accom(cid:173)
`panied, increasingly, by greatly increased
`requirements of the finished goods (new fill(cid:173)
`ing methods, increased filling speeds); and
`The minimising of production costs by using
`cheaper raw materials, ie raw materials of
`inferior quality, increasing the percentage of
`cullet and reducing glass price by a compo(cid:173)
`sition change, eg lowering the alkaline
`content, which may influence workability.
`
`DESCRIPTIONS OF PROCEDURES
`Raw materials, auxiliary agents and cullet
`Most glass raw materials are of natural origin
`and, therefore, their compositions fluctuate
`to a greater or lesser extent, depending on the
`geological situation of the sources. Often,
`
`• Peter Buchmayer, Oberland Glas AG, Bad Wurzach. Federal
`Republic of Germany.
`
`G LA S S August 1986
`
`normally, if the range of grain size distri(cid:173)
`bution of the raw materials used for glass(cid:173)
`making remains fairly constant and the
`distribution of components is homogeneous.
`The sampling procedure should be per(cid:173)
`formed by:
`Taking out individual samples from a
`delivered charge, whereby the number of
`samples depends on the material grain size;
`and dividing the collective sample -
`if
`necessary after grinding -
`to achieve suit(cid:173)
`able small amounts of material for tests.
`\Vhere acceptance agreements are made
`with suppliers, one part of the divided
`sample should be retained to have material
`for any necessary arbitration analysis. For
`humidity determinations, individual
`samples have to be taken; and because of
`water evaporation, measurements have to be
`performed quickly.
`
`Sampling procedures and devices
`It is not the intention of the author to give
`here a complete review of sampling methods
`and equipment. These descriptions are res(cid:173)
`tricted to those methods and devices which
`are normally employed in the glass industry.
`The simplest tool, especially for the cross(cid:173)
`division method, is a shovel which should be
`of stainless steel.
`If the grain size of a material is very fine,
`samples are taken with a tube-sampler, the
`simplest type of which is a stainless steel
`tube of about 50mm diameter and a length
`of 1000mm-1500mm. It may be extended by
`a metal rod (fig 1). A more comfortable
`solution is a closable sample-taker (fig 2),
`especially for fine and dry materials which
`tend to flow easily. But, if the composition
`of raw materials varies considerably over a
`short period of time, the frequency of samp(cid:173)
`ling must be increased. In such cases, it is
`necessary to use an automatic sample-taker.
`Page292 ...
`
`Fig 2. Closable sample taker.
`
`291
`
`O-I Glass, Inc.
`Exhibit 1029
`Page 001
`
`

`

`:--
`,4nnual review of the glass industry - Laboratory techniques
`
`•
`Fig 3. Turning spoon sampler connected to an
`automatic sample divider (sand}.
`Fig 5. 'Ricards' sampling diagram.
`
`· Fig 4. Turning spoon sampler, connected to· an
`automatic sample divider ( cul/et}.
`
`such a time-regulated device may be a ch~te,
`which takes samples from the free-falling
`material on the dumping point of the con(cid:173)
`veyor belt transfer by a back-and-forward·
`movement, or may be a turning-spoon
`5ampler, which takes samples frofu the
`conveyor belt. These devices are best
`connected up with an automatic sample
`divider (figs 3 and 4). The application of
`such equipment gives an increase in samp(cid:173)
`ling accuracy.
`
`Sampling specifications
`At present, there are no special sampling
`specifications for glass raw materials and
`cullet. Using the available and suitable
`German standards (DIN) and corresponding
`publications as a basis, sampling may be
`performed as follows:
`- Minimum quantity of individual samples
`to be taken from a charge:
`For bulk densities ;;i, 1, which is normally
`the case for glass raw materials, Table I
`shows the mini.mum weights ofindividual
`samples to be taken·, -as a function of
`maximum grain size_
`- Minimum number of_individual samples
`,
`to be taken from a charge:
`Table II shows the minimum number of
`individual samples as a function of the
`weight of a delivered charge, the standard
`deviation range of the attributive com(cid:173)
`ponent and sampling accuracy.
`Explanations:
`6 = standard deviation or the attributive oomponent of an
`individual raw material sample.
`26
`'
`N = ( ~ )
`
`n = number of individual samples:
`
`\ = standard deviation of sampling:
`
`81 g sampling accuracy
`
`26
`O A = 26 A = -
`.Jn
`01 is the maximum deviation .between the
`mean-value of the attributive component of
`a collective sample (sum of inclividual
`samples) and the whole of the particular raw
`material bulk supply, which will not be ex(cid:173)
`ceeded, the probability being about 950/o
`(26" -range).
`Depending on the required sampling ac(cid:173)
`c~racy (13 1) and the range of standard devia(cid:173)
`tion (o), the necessary number of individual
`samples can be taken from Table II.
`.
`Page295,..
`Table I. Minimum weight of individual samples.
`
`-
`
`-
`maxlrr.um grain size
`1n 111111
`
`above
`I--.__
`
`up to
`
`minimum veight or
`individual .sample
`in kg
`
`,,,,,
`s
`4
`
`J
`
`1
`
`1--
`
`,
`
`0.8
`
`6
`
`-
`
`.....
`
`I; L,,
`
`~
`
`~
`
`...
`
`-· '-- ~+--- p bo'
`J • ....-
`...
`-
`
`~ I;
`
`V
`
`I ....
`
`""
`
`I-
`
`Q
`
`0.
`
`0.1
`
`0.1
`
`I . /
`~I/
`.....
`V'
`V
`I--• LL'j...... t..-;
`, ..... ~
`.
`-- +-··
`" v! w..- ........
`~ i
`'
`l
`O.J 1!4 o.s 0.6 0.6 1.0
`1
`J
`4 S678910
`10
`COLLECTIVE SAMPLE f kg J ,·,
`
`v
`1,1/
`.....~
`r I /
`1./
`i..., ....
`v""
`l,..L
`-
`i..,"' v1,
`i...,
`
`I;
`
`I,,
`
`I,,
`
`~
`
`V
`
`I • L.,,
`
`'-
`
`I..,
`
`I; ...
`
`..... .....
`...
`I..., ....... v,...
`....,
`-
`
`t....,,
`
`V
`
`V
`
`v V
`
`~
`
`~
`
`c!t
`~ ...
`--
`... --
`·-
`"
`V
`
`~
`
`.....
`
`1 ......
`
`I
`
`I
`
`I , - - . --1t
`
`.
`'
`
`-
`
`I
`
`30 40 5464 ao m1r9
`
`Table II. Minimum number (n} of individual samples.
`
`weight of a
`delivered
`charge int
`
`above up to
`
`1
`
`standard deviation range
`
`t1 .. 1
`
`1 < ds2
`
`2 < d s: .3
`
`3<0.S5
`
`n
`
`2
`
`81
`
`::s 1 ,41
`
`n
`
`4
`
`81
`
`.S 2,00
`
`81
`
`n
`
`81
`
`:!!!: 2, 12
`
`16
`
`:<; 2,50
`
`n
`
`B
`
`12
`
`100
`
`--
`
`50
`........._
`
`io
`r--__
`
`10
`1--._
`
`3
`t---.._
`
`1
`t---...._
`
`--
`
`100
`
`50
`
`20
`
`10
`
`3
`
`1
`
`30
`
`15
`
`5
`
`2
`
`0,5
`
`0,2
`
`o,o:;
`
`I
`
`1
`
`5
`
`10
`
`50
`
`5
`
`10
`
`50
`
`100
`
`100
`
`500
`
`500
`
`1000
`
`3
`
`4
`
`6
`
`8
`
`12
`
`16
`
`1, 15
`
`1,00
`
`0,82
`
`0,71
`
`0,58
`
`0,50
`
`1000
`
`2000
`
`20
`
`,s 0,45
`
`6
`
`B
`
`12
`
`16
`
`24
`
`32
`
`40
`
`1,63
`
`1, 73
`
`1,41
`
`~6
`
`1,50
`
`24
`
`32
`
`2,04
`
`1, 77
`
`1,15
`
`.24
`
`1,22
`
`48
`
`1,44
`
`1,00
`
`0,82
`
`0,71
`
`32
`
`48
`
`64
`
`1,06
`
`0,87
`
`64
`
`96
`
`1,25
`
`1,02
`
`0,75
`
`128
`
`0,88
`
`"'0,63
`
`80
`
`.s o,67
`
`160
`
`~ 0,79
`
`G L A S S August 1986
`
`O-I Glass, Inc.
`Exhibit 1029
`Page 002
`
`

`

`Annual review of the glass industry - Laboratory techniques
`
`The following example may serve to
`provide a better understanding:
`The result of a chemical analysis of dolo(cid:173)
`mite, over a longer delivery period, for the
`main component MgO is 18.50/o ± 1.750/o.
`The maximum grain size is < 2.0 and
`> 1.6mm. Therefore, the minimum weight
`of individual samples has to be 50g (Table I).
`Assuming that the delivered charges are
`about 15 tons, ie in the range 10 tons to 50
`tons, the minimum number of individual
`samples to be taken has to be 12 (Table II).
`In this case the sample accuracy (B 1) would
`be 1.15.
`Another authorOl proposes the appli(cid:173)
`cation of a diagram originally developed by
`RichardsC2l for ores and minerals. In this
`diagram, the maximum grain size is plotted
`against the quantity of a collective sample
`(fig 5). The five curves represent different
`content ranges for the main components.
`Thus, curve I is valid for low contents of
`about 1 OJo, eg Fe20 3 in sand, curve III for
`mean contents of about 100/o, eg Na20 in
`feldspar and curve V for high percentage
`contents, eg Cao in limestone. For the
`dependence of the number of individual
`samples on the delivery charge, the author
`quotes the following:
`
`Less than I ton
`
`I ton
`IOtons
`llllltons
`
`Minimum of six individual
`samples
`10 individual samples
`20 individual samples
`30 individual samples
`
`Thus, for a delivered charge of sand of
`about 18 tons with an Si02 content of about
`98070 and a maximum grain size of 0.8mm,
`using curve V, the weight of the collective
`sample has to be 7kg. Since the number of
`individual samples for 18 tons is a minimum
`of 20, the weight of the individual samples
`has to be.about 350g.
`It must be remembered that this method
`does not take into account variations in the
`main components of a raw material. If varia(cid:173)
`tions (standard deviations) are large, it is
`recommended that the number of individual
`samples should be increased.
`
`Sample preparation
`From the bulk of the collected individual
`samples, the collective sample must be divi(cid:173)
`ded to obtain, first, a 'laboratory sample' (a
`few kg) and, finally, the small amounts of a
`few grams which are necessary for the per(cid:173)
`formance of chemical and physical tests.
`The most important requirement is that the
`composition of these small quantities corres(cid:173)
`pond to that of the collective sample and
`even to the whole charge. The dividing stages
`depend on the grain size of the material. For
`normal raw materials for glass melting, ie
`materials of medium homogeneity, the quan(cid:173)
`tity of divided sample as a function of grain
`size has to be calculated as follows:
`m = 0.1 X d'
`
`where:
`m = quantity of divided material in kg
`d = maximum grain size in mm
`Hence, the dividing procedure normally
`consists of a combination of reducing and
`
`GLAS S August 1986
`
`grinding steps. Obviousl:9, materials have to
`be dried beforehand. Sample dividing may
`be performed either manually or mechanic(cid:173)
`ally. Applying the manual cross-division
`method, the dried and, if necessary, crushed
`and ground collective sample, is spread out
`on a clean flat surface. The material is mixed
`with a shovel and piled up into a cone. Then
`the cone is flattened. This pile is reduced by
`quartering as shown in fig 6. The removed
`two quarters are mixed again before the next
`quartering occurs. Thus, step by step, the
`material is reduced to obtain the required
`partial quantity.
`For the mechanical dividing and reduction
`of glass raw materials, two different types of
`sample dividers with different dimensions
`are suitable. Widely applied is the turning
`
`Fi'g 6. Manual sampling division.
`
`COllECTIVf OR
`RAND0H SAHPlE
`
`1 :
`
`(G. lOKG
`
`10KG
`
`'"j
`'''j
`'")$
`
`Z•J
`
`1,4
`
`!ABOR.1,TQRY
`SANPL(
`
`I
`DRYING
`
`CRUS/1/NG AND GRINDING A
`
`l•J
`7.4
`SIEVE TEST
`[H[Hl(Al
`·-
`~ A.La.
`
`Fig 7. Turning tube divider with feeding device.
`
`tube divider, which is used for the periodical
`or continuous dividing of bulk material (figs
`3, 4 and 7). A turning tube distributes the
`continuously-fed material (feeding channel)
`inside a vessel. A small part is able to pass
`through an outlet slot, which can be con(cid:173)
`tinuously adjusted by moving a shutter.
`Normally, the dividing ratio can be varied
`from 1: 10 to about 1: 100.
`Another rotary divider is the laboratory
`sample divider for representative portions.
`A type with eight flow-out exits with attach(cid:173)
`ments for small bottles or bags, which divides
`material into eight equal portions. Since, for
`different tests to be performed, several equal
`samples are required (sieve test, bulk density,
`chemical analyses) this type is preferred in
`laboratories (fig 8). To achieve constant
`feeding of the sample divider, particularly
`with fine heavy-flowing materials, it is neces(cid:173)
`sary to use a feeding device incorporates a
`Page296 ....
`
`Fig 8. Rotary divider with eight outflows for
`laboratory samples.
`
`Fig 9. Sampling plan for sand.
`
`~ 1 s a a d
`1"0,11~,~11 S10,•IJ.J%:¢.5
`,o-,,ia ••~1,,~a, :P'.11 • I.S, ma, 9•e,~ lrff d, 1 ~.~
`
`IIJ
`
`I01ndu1avol1oa,lt,
`.t 5~~ ra, II« ull«t.·l•i
`/;poc1n,amp1,1ah1J
`
`rolluf,-, somp1,1aay
`
`"'~":::: .... , ... ,,.
`~ ,, ... 1 .. ,
`
`lllfflllll lvL•r ~t,,ftr
`(f IOI
`
`hltntfll '"~' a,,,,,.,
`(t ISi
`
`'"'""~~
`
`' " '~ '
`
`th~miral
`anoly111
`
`fo r11ert
`
`l•1Mtato,1 lal"~lf
`
`,,.,.,,.,
`
`(a II
`
`295
`
`O-I Glass, Inc.
`Exhibit 1029
`Page 003
`
`

`

`Annual review of the glass industry - Laboratory techniques
`
`-:--..._ _________
`there is a change in the viscosity curve.
`Calculation of viscosity data have shown
`that deviations of _the main oxides of a given
`glas~ •fOnJ.po~itiori witHin the r~ng~ of
`± 0.1 wt.% result in teml,efature changes of·
`the viscosity curve within the range of ::1: 5°C
`(Table III). This is within the dimensions of
`accuracy of a good temperature control in a
`forelrearth.
`Thus, after fixing the required or desired
`maximum deviation range (~0.1%) of-the
`glass oxides, the maximum variation of raw
`material constituents, particularly ·the main
`oxides which do not cause violation of this
`range, are to be calculated for each raw
`mu~~-
`,
`This can be demonstrated as follows:
`Considering a limestone of rel~tive · po~
`quality, used for an amber glass batch which'-'..
`contains 40% cullet, the result of chemical
`analysis is for the main component Cao is:
`0x = 53.S'lo CaO, s • ,i, 0.62'1',, 2s • +/- l.24f/o, 3s • 1.86'1',
`Under the condition that the maximum
`deviation in the Cao content of this glass is
`fixed at ± 0.1 oio, the batch calculation shows
`that the deviation limit of CaO which will
`should be made. As is well known, the
`not cause a violation of this range is: Ax
`standard deviation, (s), does not indicate the
`magnitude of the error of an individual Limit = :1: 3.0%, which means: nX Limit
`measured value but is a criterion of the > 3s.
`·
`reproducibility of a measuring method.· Its
`Assuming the supplier is reliable, ie that
`numerical value indicates that the error of · this quality level can also be expected for
`measurement fluctuates within the limits + s
`future deliveries, control of the limestone
`and-sonly with a probability of 68.3%. If· can be restricted to ~andom tests.
`.
`.
`If the same glass 1~ t? be me!te~ w1~h ~nly_ .
`a higher statistical certainty is desired, the
`,
`scattering value, i.\x, of the expected result
`5% c~llet, the perm!ss1ble d;v1.at1on hm1t of_
`~aO m our ~x8!11-ple 1s: AX ~1m1t = :I: _1.S~o,
`increases. For a statistical certainty of
`te _3s > x L1m1! > 2s. Strictly speaking, m
`99.50/o, Ax is about 2s and for a statistical
`certainty of 99. 711/o, Ax is about 3s.
`this case all delivered lots have to be tested.
`In fact, 2s means tha_t ~5% o~ aJI future
`In contrast to the standard deviation, s,
`the scattering value, !1x, is a parameter of meas~re~ results are w1thm the hm1ts :1: 2s.
`Cons1dermg furthermore,
`that normally
`-the error of the individual measured value.
`Second the influence of variations in the
`several deliveries are stored together in a silo,
`constitue~ts of the raw materials on the glass which will. c~use a ~ixing when material _is
`composition have to be ascertained.
`extracted, 1t 1s ~uffic1ent _to tes_t 0;11ly a certam
`number of delivery lots 1~ x Limit ;> 2s. The
`The permissible variations of the main
`oxides in a given glass composition depend
`number d~pends on the_ silo capactty and on
`raw ?1-atenal ~onsumpt1?n. If, for example,
`directly on the maximum variations in work-
`ing properties which will not cause working
`the sil_o con tam~ five dchvery lots, each fifth
`problems, ie which are negligible. These may
`supplied lot should be tested.
`be variations in the working range and/or
`variations in the working temperatures, ie
`
`,,_,, ...
`~~ -~ c.-...-.,.,, ti..-,., .• ~ ... .,.~ .J- t., J
`,.. .... ,,,.
`
`tloJ~fda,
`
`.~,, .. ,& ,J••
`• ,1, ,.,,.,
`,.,..,. .... ,,.,..,,,
`·,,, ,.111,
`
`0
`
`ffl'l6,~fll,,o/ IO~H
`
`=-_;,-
`
`,_,,, ,. ····•.r
`.,_,, ,-,1,,,_
`
`l11:1MOtf'f '"'""'' .....
`
`11•-,al
`_.,,.,.
`
`,,,.,.,,
`
`Fig 10. Sampling plan for cu/let.
`
`Page298~
`
`channel and an electromagnetic drive (fig 7).
`For the sampling of the various raw materials
`it is advantageous to establish a sampling
`plan ,(fig 9).
`
`Sampling of cullet
`The sampling of cleaned and processed
`foreign cullet, especially when be;ing used in
`the batch in high percentages, requires
`special considera~ion. On the one hand,
`there is no practical experience available; on
`the other hand, a consequent application of
`the above-mentioned specifications is neither
`possible nor necessary. Thus, for a cullet
`charge of 20 tons with a maximum grain size
`of 50mm to 100mm and a standard deviation
`of the main componentin the range 1 < 6 :,.-;; 2,
`12 individual samples of l5kg each ( = 180kg)
`would have to be taken and prepared.
`For an- estimation of the amount of con-
`' tamination in the cul1C4 such as, ceramics,
`metals, stones, etc, which are more inhomo(cid:173)
`geneously distributed-(wJnimum 3 < 6:;;; 5), a
`minimum of 48 individual samples of 15kg
`each ( = 720k'g!) would.have to betaken.
`This amount of eleborate work cannot be
`applied in practice.
`In fact, the bulk of processed cullet is-very
`homogeneous with regard to its chemical
`comppsition, as a result of collection, depo-
`1 sition and processing, which includes crush-
`I ing and mixing stages. Since the chemical
`I composition of a single bottle is-veryhomo-
`1 geneous, it can be assumed that aftercrush-
`i ing, the chemical composition of. broken(cid:173)
`. glass in all grain size ranges is identical.
`This is confirmed by comparative chemical
`analyses of different sieve fractions of
`crushed cullet. Hence it seems to be permis(cid:173)
`sible to use the average grain size, or the
`grain size of the main part of processed cull et
`respectively to dete~ine the minimum
`weight of the individual samples. The
`average grain size of proc~sed cullet is
`normally in the range of 10mm to 20mm.
`Therefore, the minimum weight of ·the
`individual samples to be taken is 2kg. This
`quantity of material permits the application
`of an automatic spoon sampler (fig 4) .. ~which
`enables the taking of a large number of
`individual samples per unit time. Fig 10
`shows a practical sampling plan for cullet.
`
`Testing raw materials ·
`It is obvious that the investigation of all
`delivered raw materials is not possible, as a
`plant laboratory must work economically.
`Therefore, it is necessary to use a schedule
`for the required tests and analyses and their
`frequency. The number of raw' material
`analyses to _be performed should be deter(cid:173)
`mined on a basis of an evaluation which
`takes into· account the periodic variations in
`composition on the one hand and their'
`effects on the established glass composition
`on the other.
`Hence, first of all the degree of variation,
`particularly of the main compon~ts, ie the
`scattering value, ~x. of each raw material,
`has to be ascertained as well as the consist(cid:173)
`ancy over a longer period of time.
`Although a knowledge of statistical prin(cid:173)
`ciples is taken for granted, some explanations
`
`296
`
`Table Ill. Dependence of viscosity data on glass composition (Lakatos).
`oxides basis glass with
`(%)
`deviations
`72,0
`2,0
`
`1197
`1031
`
`1193
`1028
`
`- 4
`- 3
`
`log Vise. basis glass with difference
`(OC)
`(pasc)
`deviations
`(OC)
`
`-. Si02
`Al2o~
`CaO
`
`MgO
`Na2o
`K2o
`
`11,0
`
`1,0
`
`13,0
`1,0
`
`71,5
`2,1
`11,1
`
`1, 1
`
`13, 1
`1,1
`
`ii) Emhart
`
`+) log 2 -
`
`log 6,65
`
`2
`
`3
`6,65
`
`8
`
`9
`12
`
`13,~
`WRI*
`RMS*
`WR+
`
`'
`
`732
`
`674
`640
`
`565
`
`538
`167
`
`137
`465
`
`731
`
`673
`639
`
`564
`
`538
`167
`
`137
`462
`
`- 1
`- 1
`
`- 1
`
`- 1
`0
`
`0
`
`0
`
`- 3
`
`G L A S S August 1986'
`
`O-I Glass, Inc.
`Exhibit 1029
`Page 004
`
`

`

`Annual review of the glass industry - Laboratory techniques
`
`Assuming, also, that raw materials of very
`inferior quality are to be accepted (preven(cid:173)
`tative glass correction), it can occur that x
`Limit < s. In these cases the investigation
`of all incoming raw material lots is in(cid:173)
`dispensable for carrying out necessary batch
`corrections.
`Finally, it has to be mentioned that an
`oxide can originate in different raw materials,
`eg CaO in both limestone and dolomite. It is
`self-evident that both have to be taken into
`account, if the value ofx Limit is to be deter(cid:173)
`mined.
`
`Chemical analysis of raw materials
`It would be going too far to describe all
`available methods of analysis in detail. But
`consideration should be given to the method
`of carrying out the necessary analyses both
`quickly and economically. Comparing the
`essential preparation and measuring time on
`the one hand and the costs of equipment on
`the other, of the two eligible modern
`methods, X-ray fluorescence analysis and
`atomic absorption-spectrophotometry
`(AAS), the time-ratio for the preparation of
`samples and measuring is about 1 :3, whereas
`the ratio of equipment costs is about 10: I.
`Without doubt, therefore, the latter method
`is the most economic method for the majority
`of glass plants. Accurately speaking, im(cid:173)
`proved preparation and measuring methods
`as well as AA-spectrophotometers with high
`precision render it possible to accomplish an
`analysis (or several glass analyses) within
`three hours. Because of the very accurate
`results, it is possible to calculate the Si02
`content by difference, for high contents of
`Si02 cannot be measured by AAS with the
`necessary accuracy - one of the small dis(cid:173)
`advantages of this method.
`Up to now, the main problem has been the
`dissolution of glass and silicates. As for
`AAS, dissolution with a mixture of hydro(cid:173)
`fluoric and perchloric acid is indispensable
`(sulphuric acid cannot be used and digestion
`by melting with suitable agents causes high
`salt concentrations in measuring solutions);
`also expensive platinum dishes and a special
`fume cupboard, suitable for perchloric acid,
`have been necessary. In addition, digestion
`in platinum basins takes up time and con(cid:173)
`tinuous control is essential.
`In the meantime, new equipment has been
`developed (fig 11)(3) which makes it possible
`to dissolve normal quantities of glass or
`silicate samples, ie 200mg to 500mg of
`ground material, in special carbon beakers
`
`Fig 11. Disintegration equipment.
`
`298
`
`within one to two hours. A thermostat(cid:173)
`controlled heating block, provided with
`holes for eight carbon (or glass) beakers,
`with the temperature continuously variable
`up to 300°C should be used. The advantages
`of this equipment are that acid vapours are
`extracted immediately using a special
`suction pump and neutralised in an alkaline
`solution, so that normally a fume cupboard
`is not necessary. The process is reproducible,
`which allows optimisation of the dissolution
`parameters. With the help of a timer the
`digestion process may be performed ~ver(cid:173)
`night. This equipment has been installed and
`tested in numerous glass plants in different
`countries.
`For the actual analyses, a procedure has
`been established and tested by different
`laboratories which, with the smallest
`expenditure, permits the analysis of all kinds
`of soda lime glasses and almost all raw
`material components using only three
`standards solutions. The working instruc(cid:173)
`tions will be subject of a further publication.
`In addition to the above-mentioned Si02
`which in most cases can be calculated afte;
`determining all other oxides, high contents
`of CaO and MgO (~ 15%), eg in limestone
`and dolomite cannot be determined as ac(cid:173)
`curately as required. In these cases, the quick
`and accurate method of complexometric tit(cid:173)
`ration should be applied. For the deter(cid:173)
`mination of sulphur compounds in raw
`materials and glass, the pyrolytic decompo(cid:173)
`sition, oxidation and following titration of
`generated and dissolved S03, is a very quick
`and convenient method.
`A plant laboratory, equipped in such a
`manner should be able to perform a con(cid:173)
`siderable number of raw material and glass
`analyses with little personnel cost being
`involved.
`
`Physical investigations of raw materials
`The essential investigations to be performed
`are the determination of the grain size distri(cid:173)
`bution, the moisture content and in certain
`cases the bulk density of raw materials.
`Increased melting rates and the problems of
`dust, particularly with regard to damage to
`refractories and environmental pollution,
`require a strict observance of grain size dis(cid:173)
`tribution. Investigations of refractory
`damage as a result of dust have demonstrated
`that grain size fractions < 0.09mm should
`be minimised as much as possible ( ~ lOOJo ).
`The upper limit normally should be 1.0mm
`maximum.
`For sieve testing, several types of labora(cid:173)
`tory sieve shakers are available. In accord(cid:173)
`ance with the standardised performing of
`sieve tests, sieving machines with a rotary
`plane movement should be preferred. It is
`obvious that sieving parameters should be
`optimised. Sample preparation must be
`performed according to the above-mentioned
`specifications. If the sieve test results of raw
`material suppliers and those of the plant
`laboratory are to be compared, it is prefer(cid:173)
`able to use the same type of shaker. In any
`case sieves should always be obtained from
`the same sieve manufacturer. For moisture
`content control modern devices - combina(cid:173)
`tions of electronic weighing scales with
`
`Fig 12. Drying hot plate for cul/et.
`
`Fig 13. Sorting out different cul/et contaminations.
`
`are available. The
`infrared drying hoods -
`determination of bulk density, eg for testing
`soda ash, should be performed with an auto(cid:173)
`matic stamp-volumenometer, in order to
`achieve comparable results.
`
`Control of contamination cullet
`The determination of contaminants in cullet
`is very important. Existing test methods(4J,
`eg the investigation of random samples of
`0.2% of a supplied cullet charge, are unreli(cid:173)
`able and laborious.
`Much more reliable is the investigation of
`samples which have been taken from a con(cid:173)
`veyor belt by a time-regulated sampler/
`divider combination (fig 4). Such sampling
`equipment should be installed preferably in
`the cullet plant itself. Thus, a permanent
`control of sorting efficiency is possible. The
`divided sample, dried beforehand (fig 12), is
`spread out on to a sorting table and the
`contaminants sorted out manually (fig 13).
`The percentage of different contaminants
`has to be ascertained by weighing. This
`method allows a quantitative estimation of
`the main contamination: stones, ceramics
`and porcelain, aluminium, lead and also the
`amount of coloured glass in flint cullet. For
`the time being, the following values can be
`considered as a realistic base for cullet quality
`examination:
`Ceramics/porcelain 50g/ton - l OOg/ton
`Aluminium
`50g/ton - IOOg/ton
`Lead
`30g/ton - 50g/ton
`
`The actual acceptance level, of course,
`depends on local experiences, the percentage
`of added cullet, the pull and, finally, on
`the number of inclusions ('stones') in
`the finished containers caused by cullet
`contamination.
`(To be concluded)
`
`G L A S S August 1986
`
`O-I Glass, Inc.
`Exhibit 1029
`Page 005
`
`

`

`DUALITY CONTROL AND DUALITY ASSURANCE RELATED TO
`RAW MATERIALS AND MEL TING IN A MODERN CONTAINER
`GLASS PLANT - PART II
`
`by Peter Buchmayer*
`
`Laboratory techniques
`
`B ATCH D E MO N STRAT?.O N RECXPE
`
`• • • BA T C H CHA N GE rs N ECE SS A RY • • •
`
`IIU 1t oJ c,, ..... •
`
`, 11. 11.B~
`
`'IC9 I"'".,.
`Ft.I" .. t.
`'
`'IC9 '"'".,.
`- ' - - ' - " -~ ' - ' ~ - ' - ' -"M...:.9• ..... ,a
`,,..1.in
`'Su.Cl
`::!081.11
`!.J7.~
`!,ll'.l7
`•
`•Soga
`200.•1
`'U-.tonlt '
`lff,47
`z::.,
`• 11a1 ... 1t1t
`:,2.~
`'
`• •• ..... ••t • u.1•
`~.14
`:a:.::
`'°'111" , .... t •
`. . . . , ..... t ' 0
`• c,,., .. .. u
`z:.~
`•
`• S.decol.
`•
`O
`'S..lh,U
`10
`1.:
`,c,,,.i,.,,,
`
`:lM, . 16 . n.r.. . "
`
`1(9
`ION!
`l
`'C..Lht I
`r...o
`Y.q
`•
`• c..11u::
`•p\..,.tCUI' ~ Y-lj
`
`::0
`1:;
`:
`
`GI:~ ~I~:::~ 1n t:9
`
`outofn.iil•t
`
`'"""'
`
`eou,o..- ton .. ,u,o.-1cu ,,...., t1.11.1os::
`
`ba1cnory
`qllulroa1u1cr.
`&Hcl\ . ,er, cull•t
`
`, .-q,~
`, 1:0.~
`, 1::.,1.b
`
`,::O<IQ ·-
`
`l.l>u•., dlty
`
`!
`'
`
`:
`
`• R& .. -..t. c on
`' I t Qi u •
`1::.ao
`40 . 84
`
`' '·" 1,:6
`
`.., . 1.76 . -~ ...
`
`,.
`"
`
`lle ltl11<Jl<1••1n1
`11,..,.
`, 11.~·
`_,.,.
`,::o.:
`
`so::.-ou ... 1r,tq1 t qi. u
`froa....., t o
`bU<:l'I
`bat~4nCICull • l
`
`I ... 1';:
`,.,1!,
`
`Concluding this d~tailed survey of '.his important subject, the author
`considers prever:ttve glass correct/On, 'the material glass',
`chemical analysis of g_lass, control of glass colour, glass defects
`caused by raw matenals, cul/et and melting, and their
`identification. In addition to a summary, a useful list of references
`is included. (Continued from page 298, August 1986 issue) .
`
`Preventive glass correction
`In order to describe approximately the mix(cid:173)
`ing behaviour of a modern glass melting
`furnace, the model of a reactor combination
`of a so-called ideal stirrer vessel connected
`with a tube reactor (unidirectional flow) can
`be considered. Changes in glass composition
`(or colour) caused by varying raw material or
`cullet compositions mainly take place in the
`melting area of the furnace (stirrer vessel),
`while the composition of molten glass in the
`area of forehearth and feeders (tube reactor)
`will be little influenced. The necessary time
`to pull out this part of the molten glass, the
`so