2. Stephens 1
Abstract
The problemtobe solvedistominimize the weightanaerosol package willlose overtime due to
evaporationthroughmicroscopicleaksinavalve’sgasketorcrimppoints.The FairPackagingand
LabelingActenactedin1967, requiresthatall consumercommoditiesmustbe labeledtodisclose net
content,identityof commodity,andname andplace of business,etc.Thisrule wasamended in2015
withthe clause that all productsmustbe labeledinaccordance withitsdate of manufacture andnot
guaranteedthroughthe shelf lifeof the product.Therefore dataiscollectedduringmanufacturingwith
upperand lowerlimitstoensure thatcansare weighedinaccordance withthe netcontentlistedonthe
label.Thiswasgreatnewsforaerosol at the time but lateritwas discoveredthatstabilitytested needed
to be done on these materialstoensure the contentsare labeledaccuratelyand withinareasonable
limit.Toensure a productismanufacturedproperlyandwiththe rightcomponentswe lookatthe
weightlossof a finishedgood.
The designof the aerosol package isnot createdwithoutfailure.There are lossesthatoccurin this
moderndaypackaging and are to be expected.A singlesheetof steel isweldedtothe shape of a can.
Many cans are assembledas3 piecesintoone.Therefore,totestsome avoidable failureswe lookatthe
valve,inparticular,we analyze the hardnessof the gasket.Afteragood gasketmaterial isselected, a
case of the finishedgoodisplacedona twoyear stabilityperiodand monitored.The methodis
describedinmore detail throughoutthisreport.
By observingseveral factorsof the valve we are goingto analyze the weightlossof the finishedgood.
The hardness of the gasketismeasuredbya durometertest tomake a determinationof whichmaterial
ismost compatible withthe formulabeinganalyzedforthatparticularapplication. We will lookatthe
formulaharshnessasa measuredof gasketswellingtodetermine if we caninfact package the desired
formulaintoan aerosol withthe materialsavailable fromoursuppliers.If adeterminationcanbe made
we will use thatgasketmaterial basedonweightretentionthroughoutthe twoyearstabilitytestanda
lowswell indurometerresults.
Introduction
The historyof the aerosol package originate inthe late 1940s and beganmass productionshortlyafter
WorldWar II.The firstaerosol containerswere usedforinsecticidesandstill have manyof the same
characteristicsasthe aerosol packageswe use today.The bodyof the can ismass produced fromsheets
3. Stephens 2
of stainless steel andsome, recycledaluminum.The can shape variesindiameterdependingonthe
applicationandproductioncapabilities.Itisequippedwithavalve thatsitsatthe topopeningof the
can. Finally,anactuatoror buttonis placedontop of the valve tosuppressthe componentsandrelease
the contentsof the can. See Figure 1in AppendixA formore detailsonthe designof an aerosol can.
The valve containsa stem, mountingcup,a housingbodyanda diptube.Withinthe valve housingexists
a gasketand springfor actuatingthe aerosol.All these componentsare made fromdifferentmaterials
and are customizabledependingonthe manufacturer.Althoughthe mechanicsof a valve are fairly
simple,the chemical compatibilityof the formulaandpackagingiscrucial to the successof a finished
goodaerosol container.
For thisexperimentwe are observingthe effectsof the gasketmaterial onaparticularformula.A 211 x
604 stainlesssteel canwithDOTpressure ratingof 2Q will be usedto holdthe selectedgasketmaterial
and formulaforthe weightlossstudy.The selectionof the finishedpackage isatwopart process.First
we lookat the gasketmaterialsandthenthe weightloss.
The durometerreadingof the gasketisaffectedbychemical attackandtemperature.We testupto 120F
due to the DOT shippingregulations. We take readingatRT and elevatedtemperaturestoobserve a
change in hardnesswhichwill inevitablyaffectthe weightof the can.Whena gasketiscompromised,an
aerosol will undoubtedlyleakandif enoughmaterial islost,the productisnon-compliantaswell asa
dangerto the consumer.
Methodology
Part 1
In the firstpart, we analyze the swell andhardnessof the gasketatthe supplier.The gasket
compatibilitystudywillbe performed usingthe availablematerialstothatsupplier,see Table 1.The
supplierwe selectedforthis study hassix typesof gasketmaterialstotestthe swell.If agasketswells
beyondareasonable limit,thenthe gasketisdeemednotcompatibleanda differentgasketmustbe
considered. Anaverage swell of 3-8%isallowableandservesasan industrystandard.
The valve materialsare analyzedusingthe complete formulathatwouldgointothe package.The
complete formulaisthe bulkliquidwithpropellant.Forthisexperimentwe are lookingata harsh
formulawitha ratingunitof 10 anda slightlylessharshformulaatratingof 7. The formulais96% of the
4. Stephens 3
finishedgoodpackage andCO2is our propellantat4%. The harshnessisdeterminedbyanumberof
factors includingplasticcrackingandcrazing,a Kb Value,andparticle size.
The report fromthe supplierissenttothe internal labanda final decisionismade onthe valve type
basedon the durometerdata.The valve ismanufacturedatthe supplierandsenttothe lab for
assemblingthe finishedgood.The durometerisobservedattwoweeks, twomonthsandtwodifferent
temperatures.
A testrun or sample casesare createdusingone or two valveswithacceptable gasketswell materials.If
there isnot one goodchoice,thenthe formulaharshnessmustbe reconsidered andadditional testing
will be performedafteralowerharshnessratingisassignedtoa new formula.
Table 1: Materials by Supplier
Summit Precision
GASKET DIP TUBE GASKET DIP TUBE
36101.042 Buna 38301 Series 05-5110-51 BUNA
B175
09-2010 PE
36202.042 Neoprene 05-5110-20 BUNA KA-
6712
09-2050 MDPE
36155.042 Butyl 05-5130-19
NEOPRENE N-200
09-35XX PP
36251.042 Viton 05-5150-00 BUTYL
U133
05-5160-00 VITON V-
500
05-5110-27 B218
For thisexperimentwe are consideringvalvesthatcontain similarcomponents.The onlydifference is
the gasketmaterial.All the valvesare standardS90 valveswithstainlesssteel springand have identical
diptubes.Thisiscalledidentical form,fitandfunctionwithalternative gasketmaterials. We will analyze
at two temperaturesbutwe are notconsideringalternativeformulas.The applicationforthisformula
has strictrequirementsforVolatile OrganicCompoundlimitationsandflammabilityrequirements.We
alsohave to meeta highKb Value and show minimal attackonplasticsandvariouspolymers.There is
verylittle roomformuchmanipulationof the chemical formulasowe will focusonthe components.
Part 2
The gaskethardnessisdeterminedandthe appropriate material isselectedbasedonthose results.We
nowhave to setup an in house stabilityof the formula,valve andcanforfurthertesting.Appendix D
5. Stephens 4
showsan excerptof the stabilitydatawe will collectfromthree differentgasketmaterials.Table 2
showsthe means of weightloss collectedovertwo yearsof stability.The dataisTable 2 isthe averages
of 12-24 cans overtwo years.Inthisexperimentswe lookata sample size of 24. We selectedthree
gaskettypes,based onindustryknowledgeandbasedonthe durometerdatafromPart 1 of thisstudy.
Since there isno significantchemical effectwe couldsee fromthe SPSSdata on any material,thanwe
have a highprobabilityof achievinggoodresultsforweightloss. Anaverage istakenforeachmonthand
compiledintoTable 2.
Results
Part 1
We testedall possible gasketmaterialsavailable inthisvalve setup fromthe valve supplier.Table 3
belowisanexcerptof the data receivedfrom the valve manufactureronhardnessof the gasketsat
room temperature andelevatedtemperature.The full reportcanbe seeninAppendix B. The initial
readingisat two weeksof exposureandthe final readingisattwomonths. One setof gasketsissetup
withthe final formulaatroom temperature.Anotherexperimentisrunat 120°F probablyinan ovenand
a verydry atmosphere.We are givendatafor the swell of these gasketsbutwe are onlylookingatthe
effectof temperature andchemical attackonthe durometer.
Table 3: Gasket Swell and Durometer Results
Material Type
Temperature Temperature
Room Temp
Swell %
120°F
Swell %
Trail 1 Trial 2 Trial 1 Trial 2
(1) BUNA B175 77 75 2.67 78 74 2.04
(2) BUNA KA-6712 75 69 4.88 75 66 3.94
(3) N-200 80 75 2.04 81 75 2.84
(4) BUTYL U133 77 68 23.11 78 62 23.98
(5) VITON V-500 79 63 7.99 80 66 7.69
(6) B218 80 75 4.72 80 78 3.77
We will thenanalyze the respective durometerreadingsfortwo replicatesatdifferenttemperaturesfor
each differentmaterialtype. The durometerreadingisameasure of hardnessonthe Shore A scale.The
Table 2: Three Types of Gsket Materials Weight Loss Over 2Years
Project Gasket Valve Formula 1 Mo 2 Mo 3 Mo 4 Mo 5 Mo 6 Mo 9 Mo 1 Yr 1.5 Yr 2 Yr
2887 B175 1210 697 0.63 1.13 1.70 2.06 2.24 3.03 4.63 7.21 10.60 12.59
2888 BUNA KA 52899 VS64-13A 1.72 3.55 5.23 7.52 7.75 8.93 13.76 20.79 21.75 22.32
2890 N-200 40253 VS64-13A 1.30 2.54 3.66 4.58 5.12 6.05 9.36 14.48 22.99 27.96
6. Stephens 5
numbersignifiedthe relativehardnessof the material andthe range ingeneral isfrom40-90, with40
beingverysoftlike agummybearand 90 beingveryhard suchas a hard hat.We are inthe range of
rubberbandsand pencil eraserstogive areference point.
Alphaisthe Material Type and we have types1, 2, 3, 4, 5 and 6. The data usedforBeta isa
measurements atRT(RoomTemperature or70 F) and 120 F so we have two factorsfor temperature.
We have tworeplicatesof eachmaterial. Ourdependentvariable isthe durometerresultsowe have an
overall of 24 observations.
All hardness have atolerance of +/- 5 points. Thismeansa 70A rubbercan measure 65A, and a 60A
rubbercan alsomeasure 65A and both wouldbe acceptable. We alsouse a5% significance. A 5%
significance level islogical forthisdatabecause inourharshestformulawe have lessthan25g weight
lossso we are withinthatthreshold.Below is the formulaforourtwofactor anova.
yij = ɥ + αi + ßj + ɛij
The 2 factor ANOVA testingof the gaskethardnessresultsinthe followinghypothesisforourmaterial
types(alpha) where are Ho statesall material meansare the same for the effectof the material.Forthe
alternative Hypothesis, Ha isthat at leastone αi is notequal to 0. Our alphais the gasketmaterial sowe
have 6 differenttypesof materials.We are makingthe hypothesisthatall the material meansare equal
to each otherandtherefore will notbe significant.The alternative isthatatleastone meanwill notbe
equal.
Ho: α1=α2=α3=α4=α5=α6=0
Ha= at least one α ≠ 0
Hypothesisforbetaisthe effectdue totemperature,it isthatthe two levelsatroomtemperature and
elevatedtemperatureof 120°F will have equal meansandthe alternativeisthatat leastone will have an
effectonthe variable andnotequal 0.
Ho: ßRT = ß120 =0
Ha: at least one ß ≠ 0
The hypothesisforall interactionsare thatthe meansformaterial andtemperature are the same under
all conditions.The alternative isthatatleastone material andtemperature meansare notequal.
The resultsof the Univariate ANOVA testfromSPSSoutputshowsthat alphaand betameansare both
not significantunderthese conditionsandtherefore we fail torejectournull hypothesisthatthese
7. Stephens 6
effectsare equal tozero. The factor of gasketmaterial hasa significance of 0.444 whichismuch higher
than our significance at0.05, therefore ourgasketmaterialsdonoteffectdurometerresults. Ourbeta
factor of temperature is1 therefore itisnotsignificantwhenanalyzingthe durometerhardnesseither.
The interactionbetweenthe factorsisnotsignificantbecause we donothave enoughcontinuousdata
inthe analysis. Althoughthisdoesnotdirectthe decisionof whichgasketmaterial isbest,we can
eliminatethe harshnessof the formulaonthe gasket. The material andtemperatureare notsignificant.
Thisshowsthat we are startingPart 2 of the analysisunderthe assumptionsthatthe gaskethardness
will notaffectthe weightlossobservedinPart2. It isokay to assume thatwe have normalityinthisdata
and that the observationsof the durometerreadingsare independentfromone another. Weightloss
will notbe an effectof chemical attackon the gasketmaterial andwill be due tosome otherfailure or
withinanormal acceptance level topackage the material. We will checkfornormalityinthe weightloss
data.
Part 2
Stabilityof anaerosol isalwaysa twopart processbecause a gasketmaterial will be selectedandthena
case of the potential finishedgoodswillbe placedonanin house stability.A case of product is12 cans
of the same formulawiththe selectedgasketmostsuitableforthe formula.Since we evaluatedthatthe
temperature andmaterial wasnotsignificant,we canfurtherevaluate byselectingataminimumtwo
differentmaterialsandatmost,all the materialsforthe inhouse stability.
In Part 2 of the stabilitytestingwe will lookatthe qualityof the case setup onstability.Initially,12cans
of the selectedtreatmentwill be weighed.All the datawill be collectedonacontrolledstability
document.The cans will be observed every30daysfor the first6 months,thenonly1 time in 6 month
incrementsuntil the 2yeardeadline. Alongwiththe cansbeingweighed,ageneral observationof the
appearance of the can is made during thistime. Appendix Dshowsanexcerptfromthe stabilityfiles.
It isimportantto make a connection between the durometerreadingof the gasketmaterial andthe
weightsof the product.Althoughgasketswell anddurometerreadingwill be determinedinthe
beginning,the shelf life of the finishedpackagedcontainermustbe testedpriortoproduction.If the
gasketisa badselectionforthe formulatreatment,thenwe will seehighweightlossesinthe can.
Typicallythistype of failure happensearlyinPart2 of the stabilitytest suchasin the first3 months.
Howeverwe have datafromtwo materialsandwe can lookto checkfor normalityof thisdataand
compare the meansof the weightlosstoselectafinal package. Below isahistogramon the average
8. Stephens 7
weightlossfor3 gasketmaterialsover2 years.There seemstobe a trend inincreasingweightlossas
time passes.
Chart 1: Histogram of Average Can Weight Loss for 3 Gaskets
In Part 1, the gasketsare visuallyidentical otherthanusingchemical analysis testingtodetermine which
material isbeingobserved.Majorissuesinthisprojectiskeeporganizedand all the componentsin the
rightorder.In part 2 of the stabilitytestingwe canrunintomajor issuesif eachcan isnot labeledclearly
because eachcan must be documentedcorrectlyandinthe rightorder.Otherwise variationinthe
weightswill notbe accurate.Toweighthe cans,we use a Balance Linksoftware systemsotranscribing
any data will notbe an issue. Ourreplicationswillcome inPart2 whenwe weigh all 12 cans eachmonth
to monitorthe weightlossfromthe selectedgasketmaterial.
The data from each12 cans,then11 cans, then10 cans is observedabout10-12 timesovera 2 year
period. We can lookat the can afterweighingmore thoroughlybydispellingthe contentsandtiming
the expelledproduct.We canfinddata onrate and see if itchanges.We will alsocutopena can to
inspectitfor flashrusting.We will alsoensure the diptube elongationiswithinareasonable limit. If we
lookat the meanvalue of weightlossof the seriesof cansbeingweighedatone particulartime, we can
see that thisdata isnormallydistributedonthe normalityplotsinChart2.
The weightloss,whenagoodgasketis selectedhasnormality. The Rsquaredvalue forthislinearfitis
89% forgasketB175. The gasketB175 has the lowestswell percentage aswell whichmayindicate
componentcompatibility.The onlytime we wouldsee askewingof thisdataisif we selecta poor
9. Stephens 8
gasket.Inthe eventof a poorgasketselectionwe will see dramaticdifferencesinthe meansof weight
loss.We will see veryhighnumbersinthe beginningof the stabilityoratthe endof the stability.This
symbolizesvast,quickswellinginthe beginningoraslow trickle of leakage overtime (andhighloss
meansat the endof 2 yearperiod). These highlossescanalsobe consideredoutlierswhenwe lookat
the average of the losses. The dataforChart 2 can be foundinAppendixE.
Chart 2: Normality Plots per 3 Selected Gaskets
Summary
About48% of thisformulacontainsa flammable component.Therefore todeemthisproduct
nonflammable,anazeotrope wasformedwithabout52% of a nonflammable component.Due tothis
narrow threshold,weightlossof eithercomponentiscrucial tothe label claimsof the finishedproduct.
In additionthe nonflammable componentisnoteconomical sothe narrow thresholdisalso crucial to
the marginof the product.
We have foundinthisexperimentthatusingthe durometerresultstodetermine gasketselectionwas
not a feasible task.The durometerhardnesswasnotasignificantfactorof temperature orswell and
therefore we couldnotnarrowour selection.Inthe future we shouldconsiderotherfactorsandmaybe
y = 0.1128x - 0.9685
R² = 0.8829
y = 0.1322x - 1.3605
R² = 0.8839
y = 0.2475x - 0.9963
R² = 0.8903
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
0.00 5.00 10.00 15.00 20.00 25.00 30.00
ExpectedZvalues
730 days
Normality Plot Avg Wt Loss Data
N200
BUNA KA
B175
Linear (N200)
Linear (BUNA KA)
Linear (B175)
10. Stephens 9
performmultifactoranalysisonthese results.More dataneedstobe collectedandmanymore
replicatesmustbe observedtoidentifythe significanceinmore interactionsusingANOVA.
Otherthan increasingthe sample size,infuture experimentswe shouldconsiderchange inthe formula
to a harsherchemical to analyze the attackon Durometerhardness.Thiswillhelpdetermine isthisisa
goodmeasure of stabilityof the aerosol. Ourdatafor the weightlossthatwascollecteddoesfall under
our assumptionsfornormalityandwe testthatusing ourR squaredinthe linearmodels.Althoughthe
histogramof the meansshoweda positive trend,whenlookingatthe linearfitwe see thatthe BUNA
B175 hasthe bestfit to a linearmodel withanR squaredvalue of 89%.Other modelswere usedforfit
testingbutthe model wasbestfittedtolinear.Tocheckthisassumptionwaslookat the residual versus
fitplotdata to testthe model. InAppendix Fthe graphsshow that the linearmodel isagoodfitand the
residual plotisrandomlydistributedandhasconstantvariance.
Althoughthe ANOVA of the gasketmaterial showednosignificance onthe durometerreading,our
independentvariable,we didlearnthatthe weightloss dataisnormallydistributedandhasa good
linearfit.This information isvaluabletothisexperimentbecause the B175 withthe bestfitwasa good
choice and the weightlosswasminimizedinthe finishedgoodpackage.
11. Stephens 10
Appendix A
Aerosol Diagrams
Figure 1 depictsthe componentsmentionedinthe Introductionof the aerosol package.
The aerosol isa simple designbuthighlytechnical compositiondue tohighpressuresand
chemical compatibility. (1) isthe buttonthat actuatesthe stem.(2) is the stemof the housingwhich
containsthe gasketmaterial.(3) isthe external orificeof the buttonwhichexpelsthe contentsof the
vale whenthe actuatorengagesthe stem.
Figure 1
AerosolCan Functionality
Figure 2 is a zoomedinside slice of the aerosol valve andhousing.
The orifice existsinside the actuatorandissuppliedbythe manufacturer.Inhouse,thisis
selectedbythe marketingteamandisa representationof the product.Below the stemisthe gasket
12. Stephens 11
where the swell andshrinkage will be describedindetail viathisreport.The valve cupandhousingare
standardto the valve and the diptube isattachedto the bottomof it.
Figure 2
AerosolValveBreakdown
13. Stephens 12
Appendix B
Gasket Swell Testing
Table 2 isan organizedviewof the dataprovidedbythe valve manufacturer.
The valve supplierandmanufacturerdoesaninitial screeningof all six gasketmaterials.A initial
durometerreadingistakenatRoom Temperature and120°F. Once the gasketisexposedtothe formula
a final readingistakenafter2 months.The swell isobservedforeachmaterial andrecorded.The Dip
Tube elongationisalsoobservedandnotedinthe supplierreport.
The secondtable isprovidedbyDiversifiedCPC. Whenthe SPSSanalysisshowsourfactorsare
insignificantwe use thisguide tomake abesteducationguessasto whichelastomerwill leadtothe
lowestweightlossonstability.
