205
`
`The effect of process parameters on product quality
`of rotogravure printing
`
`M F J Bohan, T C Claypole and D T Gethin*
`Welsh Centre for Printing and Coating, Department of Mechanical Engineering, University of Wales Swansea, UK
`
`Abstract: This paper describes an experimental investigation into the process parameter effects on
`product quality in rotogravure printing. A production printing press was instrumented for the
`measurement of parameters including strategic temperatures, speeds and web tension. Through-the-
`run analysis was used to evaluate the fluctuation of the press parameters and the product quality.
`The trials indicated the parameters where process control is important and allowed the elimination
`of several parameters, including ink drying, from the main experimental programme. The
`parameters selected for the main experiments were those most often used for control purposes, or
`those that were varied between print jobs. Orthogonal array experiments were used, as these allow
`the simultaneous variation of several parameters and the investigation of interactions between
`parameters. The experiments were performed using a typical commercial print run once the
`production of saleable copy had been completed. The analysis of the print quality was carried out
`in the image areas using spectrophotometry. The experiments highlighted the sensitivity of the
`process to changes in ink viscosity and the non-linearity of the response. Doctor blade setting was
`also found to be important, whereas blade load and impression pressure had negligible impact. No
`interactions were found between any of the parameters investigated. For viscosity change, analysis
`of the printed image confirmed the importance of hydrodynamic mechanisms, but it was not
`possible to isolate ink dilution. For doctor blade angular setting, hydrodynamic behaviour appears
`secondary to cell transport and release mechanisms. This latter mechanism requires further
`investigation in order to establish a clear understanding.
`
`Keywords: rotogravure printing, orthogonal arrays, CIE94, ink viscosity, doctor blade angle, doctor
`blade load, impression pressure
`
`NOTATION
`
`red- green scale in the CIE L*a*b* colour space
`blue-yellow scale in the CIE L*a*b* colour
`space
`Commission Internationale de l’Eclairage
`parametric factor for chroma in the CIE94
`colour difference formula
`parametric factor for hue angle in the CIE94
`colour difference formula
`parametric factor for lightness in the CIE94
`colour difference formula
`lightness value in the CIE L*a*b* colour
`space
`
`CIE
`kc
`
`kL
`
`L*
`
`weighting function for chroma in the CIE94
`colour difference formula
`weighting function for hue angle in the CIE94
`colour difference formula
`weighting function for lightness in the CIE94
`colour difference formula
`
`chroma colour difference
`CIE94 colour difference
`CIE colour difference
`hue colour difference
`
`aH2
`
`1
`
`INTRODUCTION
`
`Tile MS Iras received on 11 December 1997 and was accepted after revi-
`sion for publication on 16 July 1999.
`*Corresponding attthor: Welsh Centre for Prhlthlg attd Coating,
`Department of Mechanical Engineering, University of IlLales Swansea,
`Singleton Park, Su,ansea SA2 8PP, UK.
`
`Knowledge of the effects of different process parameters
`on the image quality in rotogravure printing is limited.
`The purpose of this work was to address rotogravure
`printing as a manufacturing process and to establish
`
`B05199 (cid:128)~ IMechE 2000
`
`Proc lnstn Mech Engrs Vol 214 Part B
`
`FAST FELT 2009, pg. 1
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`206
`
`M F J BOHAN, T C CLAYPOLE AND D T GETH1N
`
`. ,
`
`the dominant factors to achieve a high-quality product.
`The paper describes an experimental programme on
`a rotogravure press where the influence of process
`parameters on the quality of the printed image was
`evaluated while printing on a flexible material. At present
`the process parameters for each print run are set by the
`operators, based primarily on experience. Whenever a
`new print job is run on the press, material and press
`time are wasted in obtaining the correct product quality.
`This quality then has to be maintained throughout the
`duration of the print run. The work was used to identify
`which parameters in the process have a large impact on
`the product quality during production and to put
`forward some suggestions as to why they have such an
`impact.
`
`1.1 Rotogravure printing
`
`The rotogravure process is used to print detailed colour
`images in many graphic and packaging applications.
`The aim of the process is to reproduce an original, be it
`photographic or computer generated, on to a substrate
`of either paper or flexible film. The rotogravure process
`is similar to other high-volume printing processes in
`that it is not possible to vary the ink film thickness
`directly. The changes in colour strength are achieved
`principally by using halftones. The halftone is a matrix
`area of dots of varying area coverage. The image to be
`printed is transformed into a series of dots varying
`between 0 and 100 per cent coverage, depending on the
`strength of colour required. These will give the impres-
`sion of varying saturation when observed under normal
`viewing conditions. The variations in colour hues are
`achieved by the use of specific inks on multiple print
`units, in addition to the overprinting of inks from differ-
`ent printing units. The combination used will depend on
`the printing press configuration and the product being
`produced. Refinement of the final printed colour is
`achieved either by mechanical adjustment of the press
`settings or by making changes to the ink, notably its
`colour or viscosity.
`The rotogravure printing process transfers ink from
`an engraved cylinder on to a substrate, and a schematic
`representation of a station of a printing press is shown
`in Fig. 1. A press can comprise up to 10 stations to
`apply inks and coatings to the substrate. Each unit is
`identical in mechanical design, comprising rollers to
`handle the web and an ink or coating transfer system.
`The ink transfer system comprises an ink duct in
`which the gravure roll rotates, thereby picking up ink
`in the cells that are engraved into its surface. The
`excess ink is scraped off the cylinder by means of a
`doctor blade arrangement. The blade is usually made
`of steel, although tapered polymeric systems are now
`finding application. In the case of a steel blade, this is
`supported by a strip placed directly against the back
`
`Impression roll
`
`~ Substrate
`
`Doctor blade ~/’~~N~ Engraved cylinder
`
`Thermocouple \~
`
`" ]
`
`~ Ink ba~
`
`Fig. 1 Schematic representation of rotogravure printing
`
`of the blade, or via a strip clamped independently
`from the doctor blade to form a Y configuration.
`The surface of the gravure cylinder is engraved with
`cells representing the colour-separated image that is to
`be printed. This includes both solid and halftone areas.
`The substrate is brought into contact with the engraved
`cylinder under pressure from an impression roll, at
`which time the ink is transferred from the cells on to
`the substrate. Multiple units are placed in sequence to
`build up the final printed image and a drying system is
`normally placed between each printing unit. The drying
`of the ink prevents contamination between ink units
`and aids in the predictability of the final colour produced
`using overprints. Additional units may also be used to
`apply a coating to the non-printed side of the film for
`sealing or to prevent contamination, for example, in
`food products.
`Many factors affect the quality of the final printed
`product (Fig. 2). These can be subdivided into five
`categories representing the pre-press, the cylinder, the
`substrate, the ink and the process. This shows the
`large and varied number of parameters that can affect
`the product quality. The majority of factors relating
`to the first four categories are not usually under the
`control of the printer and as such cannot be used for
`control purposes. The aim of the experimentation was
`to identify the process operational parameters having
`a major impact on the product quality. Those selected
`represent the variables most often altered by the press
`operators. The pressure and angle of the doctor blade
`may be altered according to many factors, including
`the image to be printed and the ink applied. The
`doctor blade was fully investigated in this study. It is
`known that the depth, size and shape of the cells in
`the cylinder influence the quantity of ink transferred
`[1]. However, in this work, the engraving type and
`control were fixed by the rolls used. The ink release
`from the cells can be altered by the ink properties,
`and the viscosity is the primary mechanism by which
`this is effected by the printers. This is carried out by
`the addition of solvent, but by altering the viscosity
`other properties such as surface tension may also be
`affected [2].
`
`Proc Instn Mech Engrs Vol 214 Part B
`
`B05199 ~, IMechE 2000
`
`FAST FELT 2009, pg. 2
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`THE EFFECT OF PROCESS PARAMETERS ON PRODUCT QUALITY OF ROTOGRAVURE PRINTING
`
`207
`
`INK
`
`Viscosity
`¯ Solvent, recireulation, colour
`Elasticity
`Surface tension
`Overprinting
`Canfer properties
`Colourimetric properties
`¯ Pigment, opacity
`
`PRE-PRESS
`Scanning & scanner
`Origination
`RIPs
`Transfer algorithms
`Gamut compression
`Colours used
`
`SUBSTRATE
`Surface treatment
`Suhstrate type
`Surface finish properties
`¯ Porosity, chemistry, surface energy
`Opacity
`Colour
`Thickness
`Tension
`
`CYLINDER
`Engraving type
`Cell geometry
`Cylinder type
`Cylinder geometry
`Chemistry properties
`Image composition
`Coating
`
`PROCESS
`Doctor blade
`¯ type
`¯ load
`¯ angle
`Colour sequence
`Print speed
`Impression roll
`¯ pressure
`¯ rubber properties
`¯ geometry
`Environmental
`conditions
`¯ temperature
`¯ humidity
`¯ solvents
`Web properties
`Web manipulation
`Machine
`¯ design
`¯ accuracy
`Curing
`¯ type
`¯ total
`¯ cross unit
`Cooling systems
`
`Fig. 2
`
`Factors affecting print quality in rotogravure printing
`
`2 EXPERIMENTAL INVESTIGATION
`
`2.1 Strategic approach to the experimental programme
`
`High-quality colour printing is a complex manufacturing
`process with a high operating cost and the experimental
`time for the investigation of parameter effects needs to be
`minimized while the results obtained are maximized.
`Orthogonal array techniques, which are subsets of a
`full factorial experiment, were utilized to achieve this
`[3, 4]. The subset is balanced, and each variable setting
`occurs the same number of times with no two experi-
`ments being repeated or mirror imaged. This allows
`multiple parameters to be investigated simultaneously,
`at multiple levels, and the effect of interactions to be
`quantified. The technique, by its statistical design, is
`advantageous in reducing the number of experiments
`while obtaining the maximum information from a
`limited set of trials. There are other designs of fractional
`factorial experiments, but in most cases these are less
`efficient or do not cover the full range of combinations
`of factors. In addition, with certain elements of the
`arrays it is possible to ignore the interactions, as their
`effects can be distributed equally across all the results.
`These points aid in minimizing the time and cost,
`which was of importance, as the trials were carried out
`using a commercial press.
`
`A series of monitoring exercises were carried out to eval-
`uate the process variability and the controls of the roto-
`gravure press prior to the orthogonal array experimental
`programme. These were used to identify the important
`process parameters that would be investigated using the
`orthogonal array trials along with their normal operating
`ranges. These monitoring trials were performed during
`production with no loss of quality or product.
`
`2.2 Assessment of product quality
`
`The applicability and effectiveness of the monitoring and
`orthogonal array technique are dependent on the quality
`characteristic used to evaluate the effects of changes
`made on the press. The quality of the finished product
`has been traditionally assessed by observation by the
`operator. However, this is a subjective assessment and
`is not sufficiently accurate or repeatable for use with
`any experimental programme.
`Various techniques are available to assess colour
`objectively, including colour atlases, densitometry and
`spectrophotometry. The colour atlases use a visual inter-
`pretation of the colour, with a colour match being made
`between the sample and atlas. The colour specified is
`dependent on the visual assessment, with the ability of
`the eye to discriminate and the conditions under which
`
`B05199 ~ IMechE 2000
`
`Proc Instn Mech Engrs Vol 214 Part B
`
`FAST FELT 2009, pg. 3
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`208
`
`M F J BOHAN, T C CLAYPOLE AND D T GETHIN
`
`the match was made limiting the colour specification.
`Again, the errors introduced using the colour atlas are
`too great for detailed experimental evaluation.
`Densitometers and spectrophotometers both illumi-
`nate the sample with a controlled light source and use
`the reflected light spectrum to calculate a numerical
`description for the colour. Densitometers use filter sets
`to investigate only a portion of the reflected light to
`calculate a density value. The instrument is designed
`for use with the four process colours of cyan, magenta,
`yellow and black. The filter sets are optimized so that,
`as the ink film thickness increases, the density value
`increases in an approximately linear manner. These
`instruments are limited to the measurement of the
`process colours.
`Spectrophotometers use the whole of the spectrum to
`produce a set of tristimulus values to describe the
`colour. The instruments measure colour by illuminating
`the sample with a controlled light source. The reflected
`light is analysed across the visible spectrum, nominally
`380- 730 nm, to produce a reflectance curve. Using the
`reflectance curve, the illuminant and a set of standard
`colour matching functions, it is possible to calculate a
`set of.CIE tristimulus colour space values [5, 6]. These
`can be used for unique numerical description of the
`colour of the sample being assessed. The uniformity of
`these systems to perceptible colour differences varies
`between the many different systems. The CIE L*a*b*
`colour space has been used for the analysis as this
`provides the most uniform colour space and is generally
`used in the graphic arts industry. This defines the colour
`with a lightness L*, red-green scale a* and blue yellow
`scale b*, shown schematically in Fig. 3. Spectrophot-
`ometers were used for the measurement of the colour
`as they could most accurately identify colour and
`measure the changes in the printed copy since they use
`the whole of the reflected light spectrum.
`The colour difference between two samples, AEa.~., is
`calculated as the distance between two points in the
`three-dimensional colour space and is defined as
`
`AEa.b. = v/(AL*)2 + (Aa*)2 + (Ab*)2
`
`(l)
`
`+L*
`white
`/~ +b*
`~.yellow
`"~ ~ +a*
`-a* ~
`green ~_~_~/ r red
`
`-b /
`blue
`+L*
`black
`
`Fig. 3
`
`Representation of CIE 1976 L*a*b* colour space
`system
`
`Recent developments have led to the specification of a
`new colour difference equation, CIE94, to compensate
`further for the non-uniformity within the CIE L*a*b*
`colour space [7]. This takes into account the location in
`the colour space and applies weighting functions to the
`calculation:
`
`(A 8 2
`
`(2)
`
`The results were analysed and presented using the CIE94
`colour difference equation. The weighting functions
`kc and kn were all set to unity.
`Extensive analysis of printed images using spectro-
`photometry has indicated that the colour fluctuates
`between consecutive copies [8]. The colour variations
`between copies were analysed with Fourier and moving
`average analysis, using sample sizes of up to 130
`copies. Fourier analysis of the samples indicated that
`the colour fluctuations were random, with no cyclic fre-
`quency occurring. Moving average techniques were
`used to optimize the number of consecutive measure-
`ments required for statistically repeatable measurements
`to be attained. The analysis indicated that 30 consecutive
`cylinder revolutions should be measured, with the CIE
`colour space values averaged. In performing these
`measurements on consecutive copies, the variance in
`AE was typically 0.09.
`
`2.3 Experimental instrumentation
`
`The purpose of fitting instrumentation on to the press
`was to provide information relating to the major process
`parameters. This enabled a quantitative evaluation of the
`impact of various process changes. The printing press on
`which the experiments were performed consists of nine
`individual print units with the web running through
`each. The instrumentation was installed to provide
`general information from each of the individual units,
`which included the ink and burner temperatures. Addi-
`tional instrumentation was then added to the units
`used in the orthogonal array programme to measure
`the doctor blade temperature and surface strain. The
`instrumentation incorporated into the press had to
`comply with strict electrical regulations owing to the
`quantities of solvent used in the process. Therefore,
`wherever possible, existing process instrumentation was
`used. The instrumentation installation is summarized in
`the following paragraphs.
`The temperatures of the ink in the trays were measured
`on the gear and operator side of the press for each of the
`units, using sheathed K-type thermocouples of 0.8 mm
`diameter. These were used to detect temperature varia-
`tions at different stations along the press and across the
`width of the web. Changes in ink temperature will have
`an effect on the properties of the ink, such as its viscosity
`
`Proc Instn Mech Engrs Vol 214 Part B
`
`B05t99 (’: IMechE 2000
`
`FAST FELT 2009, pg. 4
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`THE EFFECT OF PROCESS PARAMETERS ON PRODUCT QUALITY OF ROTOGRAVURE PRINTING
`
`209
`
`Table 1 Summary of parameter level settings
`
`Parameter
`
`Blade load
`
`Blade angle
`
`Viscosity (Pa s)
`
`Impression pressure
`
`Level
`Setting
`
`1
`1.0b
`
`2
`1 Ab
`
`3
`l.Sb
`
`l
`Steep
`
`2
`Mid
`
`3
`Shallow
`
`l
`0.032
`
`2
`0.038
`
`3
`0.044
`
`1
`1.0b
`
`2
`1.3b
`
`3
`1.6b
`
`and surface tension, and their change principally owing
`to solvent evaporation. This is likely to have an impact
`on the transfer characteristics in the process, both from
`the engraved cell and on to the substrate [2]. The tem-
`peratures in the four-burner systems and hood dryers
`above each of the units were monitored using existing
`press thermocouple instrumentation. The press line
`speed was monitored using the existing press instru-
`mentation. Finally, web tension was measured using
`load cells at the unwind, in-feed, intermediate and out-
`feed of the press. The load cells were calibrated prior
`to the experimental programme. The data capture of
`these variables was effected using a data logger, allowing
`40 channels of data to be recorded on a personal
`computer.
`The viscosity readings were monitored continuously
`using a video facility, with the data being extracted later
`at set time intervals. The displays that are used as indica-
`tors for viscosity control were located above the main
`control console. The video gave additional indication of
`changes in the viscosity setting that were carried out by
`the print crew as part of their normal printing practice.
`Two doctor blade designs were used in the orthogonal
`array experimental programme. Three K-type thermo-
`couples were mounted on to each of the doctor blades,
`located at each end and the middle. To obtain an estima-
`tion of the blade surface strain, gauges were applied to
`the doctor blade/backing plate configuration. Three
`strain gauges were placed on both the doctor blade and
`backing plate for the straight doctor blade, with only
`three located on the backing plate for the Y-shaped
`configuration. All strain gauges were located in line
`with the thermocouples. The strain gauge data were
`collected using a high-speed data acquisition system,
`sampling at up to 100000 data points per second, to
`investigate possible changes occurring around the
`circumference of the engraved cylinder.
`
`2.4 Orthogonal array experimental design
`
`The invasive experimental programme was designed
`using orthogonal arrays. This minimized the experi-
`mental time and allowed the evaluation of interactions
`between the parameters. L9 orthogonal arrays were
`used for the experiments [9]. This allowed the linearity
`of the response to be evaluated with the parameters
`investigated being set at three levels. The analysis also
`provides information relating to any possible inter-
`actions between the parameters.
`
`The effects of doctor blade angle and doctor blade load
`were assessed in experiment 1 and the ink viscosity and
`impression roll pressure in experiment 2. The parameters
`were paired according to where interactions were most
`likely. It was not possible to investigate all four param-
`eters using one array owing to the possible significant
`interactions between all the parameters resulting in a
`much larger set oftrials (64 tests for a full factorial experi-
`ment). The experiments were repeated on two units to
`assess the effect of different doctor blade configurations,
`with the straight blade used on unit 7 printing blue ink
`and the Y-shaped blade on unit 3 printing red ink.
`The parameter levels were selected to represent condi-
`tions found during normal press operation and were
`obtained from the through-the-run analysis. These
`levels are shown in Table 1. The changes in blade load
`and impression pressure were effected by adjusting the
`pneumatic pressure to their loading systems. The blade
`angle was adjusted manually in its holder, and solvent
`was added to the ink to reduce its viscosity. Once the
`adjustments were completed on the press, a 5 min stabi-
`lization period was allowed and a marker was inserted
`into the reel to indicate the collection point. The reels
`were divided after the completion of the trials into
`those representing the individual runs within each experi-
`ment.
`The print image for the orthogonal array experiments
`was selected on the basis of it containing large areas of
`single ink colours. This allowed accurate, repeatable
`and consistent measurements to be carried out on the
`film using spectrophotometry.
`
`3 EXPERIMENTAL RESULTS
`
`The results from the two experimental programmes are
`subdivided into those obtained from the assessment of
`the physical press parameters and those from the
`colour assessment of the print. The through-the-run
`results will be considered initially, followed by the ortho-
`gonal array investigations.
`
`3.1 Through-the-run analysis
`
`3.1.1 Physical parameters
`
`The aim of the through-the-run analysis was to evaluate
`during production the natural parameter variations
`occurring in a rotogravure printing press and those in
`
`B05199 :( IMechE 2000
`
`Proc lnstn Mech Engrs Vol 214 Part B
`
`FAST FELT 2009, pg. 5
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`210
`
`M F J BOHAN, T C CLAYPOLE AND D T GETHIN
`
`25
`
`,_, 20
`
`~ 15
`
`1o
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`16
`
`Time (hours)
`
`--average temperature unit 5 --average temperature unit 7 --speed * samples collected
`
`Fig. 4 Ink temperatures in the inking trays
`
`the printed product. The data indicate the current
`control limits on the process parameters and provide
`ranges for the orthogonal array investigation. The dis-
`ruptions caused by the monitoring had to be minimized,
`and therefore samples of printed film were collected at
`the end of each reel only. This approximated to 50 min
`intervals. The results presented are for one day in
`which a single, long-duration job was printed. The
`parameters investigated included the press temperatures,
`line speed, ink properties and web tension.
`Temperatures in the burners and hood were examined
`initially. The temperature in the hoods and burners was
`very consistent with typical values of 120 4- 1 "C. The
`temperatures of the burners or hoods were not affected
`by process changes, including alterations in the press
`speed. Owing to the consistency of these temperatures,
`they were not used as a control parameter in the ortho-
`gonal array experimental programme. Their consistency
`is a requirement for in-register printing of flexible film
`owing to its increased extension at higher temperatures.
`The temperatures of the ink in the ink trays on units 5
`and 7, the sample collection intervals and the press speed
`are shown in Fig. 4. The ink temperatures are not
`significantly affected by the changes in production
`speed. The results indicate a maximum ink temperature
`variation of 2 C throughout the complete production
`run..The ink is stored in a tank below the unit and
`then circulated to the inking tray where it is resident
`for only a short time. The stability of the ink temperature
`is in contrast to other printing processes, such as web
`offset [10] where there is a gradual and much larger
`increase in the ink temperature during production. In
`
`addition, there is no cross-web variation in ink tempera-
`ture, a feature also found in other printing processes. The
`consistent temperatures found during this analysis are
`the result of" the limited shearing action occurring in
`the ink. This is due to the press not having a long
`roller train for ink transfer, the short ink dwell time on
`the cylinder and the low ink viscosity in comparison
`with that used in other printing processes.
`While the temperature of the ink remains consistent,
`its viscosity varies considerably through the print run
`(Fig. 5). The variation in viscosity differed between the
`units as a function of the ink and the quantity used,
`with ink viscosity changes of up to 40 per cent occurring.
`The ink viscosity is controlled by the addition of solvent,
`which evaporates with time and thereby causes a gradual
`increase in the ink viscosity. The large sudden changes in
`the viscosity were a direct result of adjustments made by
`the press crew by the addition of"ink to the tank reservoir
`or the introduction of solvent.
`The web tension was nominally constant throughout
`the print run. At the in-feed, web tension was nominally
`90 N and at the out-feed it was nominally 80N. These
`levels ensure the successful flow of film through the
`press and a high-quality rewind of the film for further
`processing, such as slitting. Small variations through
`each reel of film were detected at the unwind nip, and
`these were consistent between reels. More extreme fluc-
`tuations were noted during reel changeover. During the
`monitoring stage, it was possible to collect samples
`only at the end of each reel and it was not possible to
`assess fully the effects of tension fluctuations. However,
`they are expected to be minimal since the unwind to
`
`Proe Instn Mech Engrs Vol 214 Part B
`
`B05199 ( IMechE 2000
`
`FAST FELT 2009, pg. 6
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`THE EFFECT OF PROCESS PARAMETERS ON PRODUCT QUALITY OF ROTOGRAVURE PRINTING
`
`211
`
`0.06
`
`0.05
`
`~ 0.04
`
`..~ 0.03
`
`0.02
`
`0.01
`
`8
`
`9
`
`10
`
`11
`
`12
`
`13
`
`14
`
`15
`
`Time (hours)
`
`--*--unit 2 --~-unit 5 --n-unit 7 ---~-unit 8 o Samples collected
`
`Fig. 5 Viscosity variations for the print units
`
`out-feed stations are electronically geared and controlled
`to maintain a constant tension. These measurements
`confirmed that the tension control within the printing
`section of the press was precise.
`
`3.1.2 Colour measurement
`
`The printed image was a food wrapper that consisted of a
`series of solid and halftone areas repeated four times
`across the width of the web and four times around the
`circumference of the cylinder. A series of test patches
`were printed along the edge of the web, where, tradition-
`ally, any colour measurements would be taken. How-
`ever, results obtained from these are prone to edge
`effects and therefore are not appropriate. Hence, the
`measurements were made on areas located within the
`repeated image. At these positions, only one ink colour
`was printed.
`Measurements were carried out on yellow (print unit
`2), green (print unit 5) and blue (print unit 7) in solid
`ink patches within the repeated image, for thirty conse-
`cutive cylinder revolutions. The CIE L’a*b* values
`showed a relatively large colour difference across the
`width of the web for nominally identical colour patches
`under the same print conditions. Analysis indicated an
`average AE~4 of 0.7 for both the blue and yellow patches
`and a maximum of 1.8 for the blue and 1.65 for the
`yellow. In these measurements the variance was 0.36.
`The human eye can detect colour changes to an AE~a
`value of approximately 0.5 or greater, and therefore
`the changes noted above are just discernible. The
`measurement on the green patch was carried out to
`
`confirm the findings for the blue and yellow, and this
`will be discussed further below.
`The reasons for these differences are not immediately
`apparent and there was no opportunity to investigate
`the gravure roll surface for image consistency, Indeed,
`topographical measurement equipment that is only
`now becoming available is not suited to performing
`measurement of gravure cell volume either on a cell or
`local area basis, and so a quantitative comparison is
`not possible at this time.
`The variation in colour through the run between suc-
`cessive reels is shown in Fig. 6. The CIE AE~4 values
`shown have been calculated with respect to the final
`sample and therefore the figure only displays change.
`The results are presented for each of the blue patches
`and an average of all the yellow patches. The results
`show a large variation in the blue patches through the
`run. However, the trends are similar on all patches, indi-
`cating the cause of the variation to be acting equally
`across the web width. The analysis of the yellow areas
`shows a much better consistency of colour with AE~4
`variations of ~< 1.0 from the second sample set onwards.
`As shown in Fig. 6, the blue does not stabilize until
`well into the print run. In an attempt to establish a
`reason for this, colour and viscosity changes were inves-
`tigated and correlation between the two carried out. Cor-
`relation coefficients of 0.8 were obtained, confirming that
`there is a reasonable correlation which indicates the
`dominant effect of ink viscosity on the press. To confirm
`this, additional measurements were carried out on the
`green patches, and these showed colour shifts where
`there were significant changes in ink viscosity.
`
`B05199 J" IMechE 2000
`
`Proc Instn Mech Engrs Vol 214 Part B
`
`FAST FELT 2009, pg. 7
`Owens Corning v. Fast Felt
`IPR2015-00650
`
`

`
`M F J BOHAN, T C CLAYPOLE AND D T GETHIN
`
`12
`
`11
`
`to
`
`9
`
`8
`
`4
`
`3
`
`2
`
`1
`
`Measurement sample
`
`9 10
`
`¯ blue 1
`
`blue2 ~ blue3 --*--blue4 ~ yellow average
`
`Fig. 6 Variation in AE~4 through the duration of a run for blue and yellow colour patches
`
`Because the findings from the analysis indicated the
`viscosity to be a significant factor in the ink transfer
`characteristic, the orthogonal array trials focused on
`the ink viscosity and the press component on which it
`has a direct impact. This includes the doctor blade con-
`figuration (load and angle) and impression roll pressure.
`
`3.2 Orthogonal array experimental trials
`
`3.2.1 Physical parameters
`
`The programme was subdivided into two orthogonal
`array experiments, with the doctor blade angle and
`load investigated in the first and the ink viscosity and
`impression pressure in the second. The parameters were
`placed in these groupings as interactions were thought
`to be least likely between those from the different
`groups. The press operational parameters were captured
`using the data acquisition system and the results are
`presented for both experiments.
`There was very little temperature variation across the
`width of the press either in the duct or across the
`doctor blade, as had been indicated during the through-
`the-run exercise. The average temperatures of the ink on
`the Y-shaped doctor blade and of the ink in the tray are
`shown in Fig. 7. Also indicated are the times at which
`samples were collected from the reels. During the first
`experiment (up to 13 h) when the ink viscosity was kept
`constant, the temperature shows little variation. How-
`ever, during experiment 2, each time solvent was added
`to the i