Etikettieren und Kennzeichnen
Roll-fed wrap-around labelling of cylindrical containers with hot-melt glue
Die wissenschaftlichen Inhalte dieser Seite werden auf Englisch und Spanisch veröffentlicht; hier werden sie auf Englisch angezeigt.
Kurz gefasst
Etikettieren und Kennzeichnen kann höchstens 100 von je 100 geplanten Stunden nutzen. Der Rest entfällt auf Grenzen, die kein Werk vermeiden kann.
- 100höchstens nutzbar
Die drei Grenzen, die das festlegen
Laufzeit: bis 100 %
Der Rest sind Stillstände, die Physik oder Vorschrift erzwingen: Reinigung, Umrüsten, Abtauen.
Geschwindigkeit: bis 100 %
Wie nah an ihrer Höchstgeschwindigkeit sie laufen kann, ohne Produkt oder Anlage zu schädigen.
Gutprodukt: bis 100 %
Ein Teil des Produkts liegt auch bei gut beherrschtem Prozess außerhalb der Spezifikation.
Technisches Detail
Unsicherheitsintervalle, physikalische Modelle je Station und Quellen, für alle, die jede Zahl prüfen möchten.
1Szenario
Filled and capped 0.5 L PET bottles of a carbonated soft drink are labelled downstream of the filler on a rotary roll-fed labeller. The label web is drawn from a reel, each label is cut to length at the print mark, receives hot-melt glue on its leading and trailing edges and is wrapped once around the cylindrical label panel of the rotating container; the trailing edge is glued over the leading edge. The same geometry describes wrap-around paper labels on cans of vegetables. The physics-limited quantities are the overlap of the glue seam (geometry and cutting registration) and the web speed at which the registration still keeps the seam capable. The labeller runs faster than the filler (the V-graph of a bottling line); its documented set performance and failure behaviour come from measurements in an industrial bottling plant.
2Ergebnisse
- Cut length of the label (panel circumference plus nominal overlap)212 [203–221] mm
- Standard deviation of the overlap at standstill registration (cutting and container diameter only)0,434 [0,295–0,580] mm
- Largest web speed at which the overlap stays z standard deviations above its minimum3,00 [2,04–3,93] m/s
- Standard deviation of the overlap at the capable web speed0,741 [0,419–1,332] mm
- Share of labels whose overlap is shorter than the minimum glue seam6,20 × 10⁻¹⁰ [6,25 × 10⁻³⁰–4,39 × 10⁻⁴]
- Share of labels placed outside the vertical tolerance of the label panel5,73 × 10⁻⁷ [1,44 × 10⁻¹⁶–1,79 × 10⁻²]
- TheoretischMaximum OEE quality factor (overlap seam and placement) at the capable web speed1,000 [0,982–1,000]
- Quality factor if the labeller ran at its maximum web speed regardless of registration capability1,000 [0,982–1,000]
- Containers per hour at the maximum web speed of the declared machine50.895 [34.701–67.675] 1/h
- Containers per hour at the capable web speed50.895 [34.561–67.168] 1/h
- Best dokumentiertSet performance of a labelling machine measured in an industrial bottling plant33.197 1/h
- TheoretischThermodynamic minimum energy per label (the glue returns to ambient)0 J
- Heat to bring the glue of one label to the application temperature, without recovery5,70 [3,38–14,39] J
- Glue heating duty per kg of labelled product0,0110 [0,0065–0,0277] kJ/kg
- Run time of one label reel at the capable rate22,2 [10,6–41,8] min
- TheoretischMaximum label-material yield (splice and start-up waste)0,9994 [0,9980–0,9997]
- TheoretischMaximum availability (no stop is forced by physics or regulation: reel changes are removable by an automatic splicer)1,00
- Best dokumentiertAvailability of a labelling machine from its measured MTBF and MTTR0,973
- Availability with declared manual reel splices0,881 [0,705–0,960]
- TheoretischPerformance factor: capable web speed over maximum web speed1,00 [0,78–1,00]
- TheoretischPhysics-bounded OEE1,000 [0,775–1,000]
- OEE with declared manual reel splices0,881 [0,643–0,958]
3Obergrenzen je Dimension
- Theoretisch
- Beste verfügbare Technik
- Best dokumentiert
- Typisch
- TheoretischMaximum OEE quality factor (overlap seam and placement) at the capable web speed1,000 [0,982–1,000]
- Best dokumentiertSet performance of a labelling machine measured in an industrial bottling plant33.197 1/h
- TheoretischThermodynamic minimum energy per label (the glue returns to ambient)0 J
- TheoretischMaximum label-material yield (splice and start-up waste)0,9994 [0,9980–0,9997]
- TheoretischMaximum availability (no stop is forced by physics or regulation: reel changes are removable by an automatic splicer)1,00
- Best dokumentiertAvailability of a labelling machine from its measured MTBF and MTTR0,973
- TheoretischPerformance factor: capable web speed over maximum web speed1,00 [0,78–1,00]
- TheoretischPhysics-bounded OEE1,000 [0,775–1,000]
4Verlustkaskade
- TheoretischBest dokumentiertVerfügbarkeit1,00 → 0,973 (Abstand 0,0273)
No stop of a roll-fed labeller is forced by physics or by regulation: the reel change is removable by an automatic splicer and there is no hygienic cleaning duty on the label path, so the availability ceiling is 1. The documented labeller of an industrial bottling plant stopped on average every 1 423 s for 40 s, an availability of 97.3 %. Such short, frequent stops are typical of container handling: containers that jam in the infeed star wheel or the centring bells, labels that fail to transfer from the vacuum or glue pallet, glue that strings or chars, and sensor faults at the print mark. Each of these is a mechanical or control loss, not a physical limit; in a line, an accumulation conveyor between filler and labeller absorbs most of them (see the conveying model).
- TheoretischTypischEnergie je Produkteinheit
The glue is heated and cools back to ambient on the container, so with ideal heat recovery no net energy would be needed: the thermodynamic minimum is zero. Without recovery the least heat is that of bringing the few tens of milligrams of glue per label to the application temperature, a few joules per label and a few thousandths of a kilojoule per kilogram of product. A real glue station keeps a melt tank, hoses and a glue roller at temperature all the time and loses heat continuously to the room, and the drives of the turret, the vacuum drum and the conveyors dominate the electricity of the machine; the energy of labelling is set by machinery, not by the physics of the label.
5Physikalisch begrenzte OEE
- Belegungszeit (ideal)1,000
- VerfügbarkeitFaktor 1,001,000 (−0,000)
- LeistungFaktor 1,00 [0,78–1,00]1,000 (−0,000)
- QualitätFaktor 1,000 [0,982–1,000]1,000 (−0,000)
- Physikalisch begrenzte OEE1,000 [0,775–1,000]
6Vollständige Herleitung
Scope. Filled and capped 0.5 L PET bottles of a carbonated soft drink are labelled downstream of the filler on a rotary roll-fed labeller. Each label is cut from the web at its print mark, receives hot-melt glue on its leading and trailing edges and is wrapped once around the cylindrical label panel of the rotating bottle; the trailing edge is glued over the leading edge. The same geometry describes wrap-around paper labels on cans of vegetables. The labeller runs faster than the filler (the V-graph of a bottling line). Symbols and their values, units, uncertainty distributions and sources are listed in the parameter table generated from config/physics/labeling.yaml. Numbers quoted here are point values; the output table gives the Monte Carlo intervals.
1. Mass balance — maximum label-material yield
Labelling consumes no product. Label web is lost at each reel splice (a length around the splice) and at each changeover while the cut registration settles ( labels). During a campaign of duration at the capable rate the web used for good labels is , the number of reels is , and
where is the cut length of one label. A container rejected for a bad label is counted as a quality loss (section 4), not as a yield loss: the product and the container are not consumed by the labeller.
2. Energy balance — minimum energy per label
The glue is melted, applied and cools back to ambient on the container; the label and the container end in their initial thermal state. In the reversible limit all the heat could be recovered, so the thermodynamic minimum is zero. Without recovery the least heat is that of bringing the glue mass of one label from ambient to the application temperature :
0.011 kJ per kg of product. Glue mass, specific heat and temperature are declared scenario values. A real glue station keeps a melt tank, hoses and a glue roller hot all the time and loses heat continuously; the turret, vacuum drum and conveyor drives dominate the electricity of the machine.
3. Rate-limiting step — web speed and registration
Label length. A label wraps the panel of diameter once and overlaps itself by the nominal overlap :
Rate. The web feeds one label length per container, so the rate at web speed is : 50 900 containers per hour at the maximum web speed of the declared machine, = 3 m/s. The web speed is a machine design value without an open source, so the rates carry no theoretical label (assumption a3).
Registration error grows with speed. The cut is triggered at the print mark. A timing error of mark detection and cut actuation becomes a length error , so the standard deviation of the cut length is
where is the error of the cutting mechanism at standstill registration and the timing jitter. The faster the web, the wider the scatter of the overlap; this is the physical trade-off between speed and quality of the operation.
Capable web speed. The realised overlap is (section 4); with independent cut length and container diameter its standard deviation is
A capable process keeps the mean overlap standard deviations above the minimum glue seam , i.e. . The largest capable web speed is
With the declared values the registration limit is 6.0 m/s, above the machine's 3 m/s, so ; when the jitter is large (upper part of its range) the capable speed falls below the maximum and the performance factor drops below 1.
4. Process capability — quality ceiling
Overlap seam. The trailing-edge glue bead must land on the label itself; if the overlap is shorter than the seam carries only part of the bead and the label opens or flags. With the share of short seams is
with = 0.74 mm and the standard normal distribution function.
Vertical placement. The label must stay within of its nominal height on the panel; with a vertical position error the misplaced share is .
Running at the maximum web speed regardless of registration capability gives quality_at_max_speed, which equals here because the declared machine is capable at full speed. Wrinkles, flags caused by the glue, print defects and code legibility are not modelled, so is an upper bound. No open source quotes the minimum glue overlap, the cutting accuracy, the timing jitter or the placement tolerance of roll-fed labellers; they are declared scenario values with ranges, and the Monte Carlo interval of reflects them.
5. Unavoidable stops — maximum availability
A reel holds a finite web length and runs out after = 22 minutes at the capable rate. A manual splice stops the labeller, but an automatic splicer removes the stop (not the splice waste); no cleaning-in-place applies to the label path, so the cleaning model is not used. No stop is forced by physics or regulation:
With manual splices of duration ,
which enters only oee_scenario.
Documented availability. In an industrial bottling plant (mineral water and soft drinks, returnable glass bottles), the labelling machine was measured at a set performance of 33 197 bottles per hour with a mean time between failures of 1 423 s and a mean time to repair of 40 s (Bär et al., 2021, Table A1). With the availability defined in the same source as ,
the 97.3 % of the source table. These short, frequent stops (jams in the star wheels, failed label transfers, glue and sensor faults) are mechanical and control losses, not physical limits; the conveying model shows how much of them an accumulation buffer keeps away from the filler.
6. Physics-bounded OEE
Scope of the bound. The theoretical label of means physics-bounded under the declared scenario: the ceiling follows from the geometry and registration physics of the operation for the declared container, label and machine, and it is a bound only conditional on those declarations. It follows the common definition of all models: counts only stops required by regulation or forced by a physical mechanism; is the ratio of the physics-limited ideal cycle to the cycle of the capable process; is the share of conforming units, and mass losses of conforming material are reported only as yield.
The labeller does not bound the line by physics: its ceiling is set by the registration capability, which the declared machine meets at full speed. With the declared manual splices oee_scenario = 0.881; the documented failure behaviour of a real labeller gives an availability of 0.973.
Consistency with the other models. The labeller runs faster than the filler, so in a line it never sets the rate; its documented set performance (33 197 per hour against 28 998 per hour of the filler of the same plant, a table cell the filler shares with the empty bottle inspector) is the V-graph used in the conveying model. Filling, closing and the product are outside this model.
7. Assumptions and validity
See the assumption and validity lists generated from the configuration file. The most influential inputs (sensitivity table) are the timing jitter, the maximum web speed and the overlap margin for performance, the vertical placement dispersion for quality, and the reel length and splice time for the scenario availability.
References
- Bär, R. M., Zeilmann, M., Nophut, C., Kleinert, J., Beyer, K., Voigt, T. (2021). Simulation of energy and media demand of beverage bottling plants by automatic model generation. Sustainability, 13, 10089. doi:10.3390/su131810089
7Parameter und Herkunft
- 65 mm
Diameter of the cylindrical label panel of the container
Szenario · Gleichverteilt 62 bis 68 mm
Deklarierte Annahme0.5 L PET bottle; no open bottle drawing was used. Declared assumption.
- 0,1 mm
Standard deviation of the label-panel diameter from container to container
Szenario · Gleichverteilt 0,05 bis 0,15 mm
Deklarierte Annahme - 8 mm
Nominal overlap of the trailing edge over the leading edge (label design)
Szenario · Gleichverteilt 7 bis 9 mm
Deklarierte Annahme - 3,5 mm
Least overlap that still carries the full glue bead of the seam
Szenario · Gleichverteilt 3 bis 4 mm
Deklarierte AnnahmeNo open source states the minimum glue-seam overlap; declared assumption.
- 0,3 mm
Standard deviation of the cut length from the cutting mechanism alone (standstill registration)
Szenario · Gleichverteilt 0,2 bis 0,4 mm
Deklarierte Annahme - 0,2 ms
Timing jitter of print-mark detection and cut actuation
Szenario · Gleichverteilt 0,05 bis 0,4 ms
Deklarierte AnnahmeA timing error becomes a length error v_web sigma_t; declared assumption.
- 3,5
Capability margin of the overlap in standard deviations
Szenario · Gleichverteilt 3 bis 4
Deklarierte AnnahmeOperating choice, point value at the centre of the declared range.
- 3 m/s
Maximum label-web speed of the declared labeller
Szenario · Gleichverteilt 2 bis 4 m/s
Deklarierte AnnahmeMachine design value; no open source was used.
- 1,5 mm
Allowed vertical displacement of the label on the panel
Szenario · Gleichverteilt 1 bis 2 mm
Deklarierte Annahme - 0,3 mm
Standard deviation of the vertical label position
Szenario · Gleichverteilt 0,2 bis 0,5 mm
Deklarierte Annahme - 120 °C
Application temperature of the hot-melt glue
Szenario · Gleichverteilt 110 bis 140 °C
Deklarierte Annahme - 25 °C
Ambient temperature
Szenario · Fest
Deklarierte Annahme - 30 mg
Mass of hot-melt glue per label (pick-up and overlap beads)
Szenario · Gleichverteilt 15 bis 60 mg
Deklarierte Annahme - 2 kJ/kg/K
Specific heat of the hot-melt glue
Szenario · Gleichverteilt 1,8 bis 2,6 kJ/kg/K
Deklarierte AnnahmeTypical of polymer melts; no open source was used. Declared assumption.
- 0,52 kg
Mass of product per container
Szenario · Fest
Deklarierte Annahme0.5 L of a sweetened soft drink of density about 1040 kg/m3.
- 4.000 m
Label-web length on one reel
Szenario · Gleichverteilt 2.000 bis 6.000 m
Deklarierte Annahme - 2 m
Label web wasted at each reel splice
Szenario · Gleichverteilt 1 bis 5 m
Deklarierte Annahme - 30
Labels wasted while registration settles after a changeover
Szenario · Gleichverteilt 10 bis 60
Deklarierte Annahme - 8 h
Production time between two label changeovers
Szenario · Gleichverteilt 4 bis 24 h
Deklarierte Annahme - 3 min
Machine stop per manual reel change and splice (zero with an automatic splicer)
Szenario · Gleichverteilt 1 bis 6 min
Deklarierte Annahme - 33.197 1/h
Set performance of the labelling machine of an industrial bottling plant (glass bottles, wet-glue labels; measured)
Vergleichswert · Fest
Verifiziertes ZitatWörtliches Zitat anzeigen
Labelling machine 33,197 30 1423 40 97.3Simulation of Energy and Media Demand of Beverage Bottling Plants by Automatic Model Generation (Bär, R. M.; Zeilmann, M.; Nophut, C.; Kleinert, J.; Beyer, K.; Voigt, T.), Sustainability 13, 10089 (2021), CC BY 4.0 (öffnet in neuem Tab). MDPI (copy distributed by a co-author's institution), 2021. S. 18. doi:10.3390/su131810089
- 1.423 s
Measured mean time between failures of the labelling machine
Vergleichswert · Fest
Verifiziertes ZitatWörtliches Zitat anzeigen
Labelling machine 33,197 30 1423 40 97.3Simulation of Energy and Media Demand of Beverage Bottling Plants by Automatic Model Generation (Bär, R. M.; Zeilmann, M.; Nophut, C.; Kleinert, J.; Beyer, K.; Voigt, T.), Sustainability 13, 10089 (2021), CC BY 4.0 (öffnet in neuem Tab). MDPI (copy distributed by a co-author's institution), 2021. S. 18. doi:10.3390/su131810089
- 40 s
Measured mean time to repair of the labelling machine
Vergleichswert · Fest
Verifiziertes ZitatWörtliches Zitat anzeigen
Labelling machine 33,197 30 1423 40 97.3Simulation of Energy and Media Demand of Beverage Bottling Plants by Automatic Model Generation (Bär, R. M.; Zeilmann, M.; Nophut, C.; Kleinert, J.; Beyer, K.; Voigt, T.), Sustainability 13, 10089 (2021), CC BY 4.0 (öffnet in neuem Tab). MDPI (copy distributed by a co-author's institution), 2021. S. 18. doi:10.3390/su131810089
8Vergleichswerte
| Vergleichswert | Obergrenze | Wert | Herkunft |
|---|---|---|---|
Availability of the labelling machine of an industrial bottling plant, measured (MTBF / (MTBF + MTTR))Wörtliches Zitat anzeigenLabelling machine 33,197 30 1423 40 97.3Simulation of Energy and Media Demand of Beverage Bottling Plants by Automatic Model Generation (Bär, R. M.; Zeilmann, M.; Nophut, C.; Kleinert, J.; Beyer, K.; Voigt, T.), Sustainability 13, 10089 (2021), CC BY 4.0 (öffnet in neuem Tab). MDPI (copy distributed by a co-author's institution), 2021. S. 18. doi:10.3390/su131810089 | Best dokumentiert | 97,3 % | Verifiziertes Zitat |
9Unvermeidbare Stillstände
- Label reel change and splice (scenario stop with a manual splicer: availability_scenario only)Deklarierte Annahme
A reel holds a finite web length; at the capable rate it runs out after L_r / (rate L). A manual splice stops the labeller; an automatic splicer removes the stop but not the splice waste.
- Random machine failures (documented availability only)Verifiziertes Zitat
Measured mean time between failures and mean time to repair of a labelling machine in an industrial bottling plant; failures are not forced by physics, so they enter availability_documented, not availability_max.
Wörtliches Zitat anzeigen
Labelling machine 33,197 30 1423 40 97.3Simulation of Energy and Media Demand of Beverage Bottling Plants by Automatic Model Generation (Bär, R. M.; Zeilmann, M.; Nophut, C.; Kleinert, J.; Beyer, K.; Voigt, T.), Sustainability 13, 10089 (2021), CC BY 4.0 (öffnet in neuem Tab). MDPI (copy distributed by a co-author's institution), 2021. S. 18. doi:10.3390/su131810089
10Annahmen
- Label cut length, container diameter and vertical position are independent normal variables; the registration error of the cut grows with the web speed as v_web times the timing jitter of mark detection and cut.
- A label is nonconforming when its overlap is shorter than the minimum glue seam or when it lies outside the vertical tolerance; wrinkles, flags, glue defects and print defects are not modelled, so the quality ceiling is an upper bound. A nonconforming container is rejected (quality loss); the container and its product are not counted as yield loss.
- Performance: the ideal cycle is that of the maximum web speed of the declared machine; the capable process runs at the largest web speed at which the overlap keeps the declared capability margin. Machine design values (web speed, glue mass) are declared, so the rates are reported without a theoretical label.
- Availability: no stop is forced by physics or regulation, so availability_max = 1; manual reel splices enter only availability_scenario and oee_scenario, and the documented random failures of a labelling machine only availability_documented. No cleaning-in-place applies to a labeller, so the cleaning model is not used.
- The documented machine is a wet-glue labeller of returnable glass bottles in a mineral-water and soft-drink plant; its set performance and failure behaviour are used as the documented level of a labelling machine in a bottling line, not as a property of the roll-fed scenario.
11Gültigkeitsbereich
- Wrap-around labels on cylindrical label panels (bottles or cans) with a glued overlap seam; pressure-sensitive, sleeve and wet-glue patch labels have other failure modes.
12Sensitivität
- Allowed vertical displacement of the label on the panelElastizität 0,000Spearman-ρ +0,44
- Standard deviation of the vertical label positionElastizität 0,000Spearman-ρ -0,58
- Nominal overlap of the trailing edge over the leading edge (label design)Elastizität 0,000Spearman-ρ +0,17
- Least overlap that still carries the full glue bead of the seamElastizität 0,000Spearman-ρ -0,08
- Timing jitter of print-mark detection and cut actuationElastizität 0,000Spearman-ρ -0,32
- Maximum label-web speed of the declared labellerElastizität 0,000Spearman-ρ -0,20
- Standard deviation of the label-panel diameter from container to containerElastizität 0,000Spearman-ρ -0,05
- Standard deviation of the cut length from the cutting mechanism alone (standstill registration)Elastizität 0,000Spearman-ρ -0,03
- Capability margin of the overlap in standard deviationsElastizität 0,000Spearman-ρ -0,04
- Größe steigt mit dem Parameter
- Größe sinkt mit dem Parameter
13Validierungsprüfungen
- Dimensionsanalyse22 outputs converted to their declared unitsBestanden
- Jeder zitierte Wert an seiner Quelle verifiziert5 verified, 0 failedBestanden
- Kein dokumentierter Wert übertrifft die theoretische GrenzeBestanden
- Obergrenzen in physikalischer ReihenfolgeBestanden
- Anteile zwischen 0 und 1Bestanden
14Verlustursachen in dieser Operation
- Labeling machine changeover and setupRüsten und Einstellen2,3 % [0,8–5,1 %]B85 Studien
- Labeling machine equipment failureAnlagenausfall (Störungen)2,0 % [0,6–4,7 %]B85 Studien
- Inkjet printer failureAnlagenausfall (Störungen)0,9 % [0,2–2,3 %]C85 Studien
- Laser Marking Machine Speed LossVerringerte Geschwindigkeit0,8 % [0,2–2,3 %]D85 Studien
- Incorrect label applicationProzessfehler (Ausschuss und Nacharbeit)0,7 % [0,1–1,9 %]D85 Studien
- Incorrect or Illegible Product CodingProzessfehler (Ausschuss und Nacharbeit)0,6 % [0,1–1,7 %]D85 Studien
- Star wheel and guide adjustment during format changeoverRüsten und Einstellen0,6 % [0,1–1,6 %]D85 Studien
- Preparation of tools and PPE before changeoverRüsten und Einstellen0,6 % [0,1–1,6 %]D85 Studien
- Conveyor belt and guide rail adjustmentRüsten und Einstellen0,6 % [0,1–1,7 %]D85 Studien
- Changeover parts placementRüsten und Einstellen0,6 % [0,1–1,7 %]D85 Studien
- Glue roller and scraper adjustmentRüsten und Einstellen0,6 % [0,1–1,7 %]D85 Studien
- Label magazine cleaningRüsten und Einstellen0,6 % [0,1–1,7 %]D85 Studien
- Centering disc replacement (format changeover)Rüsten und Einstellen0,6 % [0,1–1,6 %]D85 Studien
- Infeed timing (worm) screw changeover adjustmentRüsten und Einstellen0,6 % [0,1–1,6 %]D85 Studien
- Reject/discard system adjustmentRüsten und Einstellen0,5 % [0,1–1,5 %]D85 Studien
- Label applicator jam or misfeedAnlagenausfall (Störungen)0,4 % [0,1–1,4 %]D85 Studien
- Label too short / undersized labelProzessfehler (Ausschuss und Nacharbeit)0,4 % [0,1–1,3 %]D85 Studien
- Missing product code/date stampProzessfehler (Ausschuss und Nacharbeit)0,4 % [0,1–1,3 %]D85 Studien
- Inkjet printer/coder adjustmentRüsten und Einstellen0,4 % [0,1–1,2 %]D85 Studien
15Literatur
- Simulation of Energy and Media Demand of Beverage Bottling Plants by Automatic Model Generation (Bär, R. M.; Zeilmann, M.; Nophut, C.; Kleinert, J.; Beyer, K.; Voigt, T.), Sustainability 13, 10089 (2021), CC BY 4.0 (öffnet in neuem Tab). MDPI (copy distributed by a co-author's institution), 2021. S. 18. doi:10.3390/su131810089
- doi:10.3390/su131810089 — Bottling-plant simulation with measured set performances, MTBF and MTTR of every machine (Table A1).

