This content is not included in your SAE MOBILUS subscription, or you are not logged in.
Fuel Permeation Performance of Polymeric Materials
ISSN: 0148-7191, e-ISSN: 2688-3627
Published May 07, 2001 by SAE International in United States
Annotation ability available
This paper presents an extensive set of permeation data on automotive fuel system materials. It adds significantly to the information provided by the same authors in SAE paper 983160 . The permeation measurements refer to three different test fuels: fuel C, CE10 and CM15 at 40, 50 and 60°C. The materials examined include poly-ethylenes, nylons, polyketons, ethylene-vinyl alcohol copolymers, acetals, fluoropolymers and fluoroelastomers. These data are important in the design of automotive fuel system components capable of meeting LEVII or PZEV requirements. In particular, data of this kind are crucial in optimizing the permeation performance of multilayer structures for fuel system applications.
CitationNulman, M., Olejnik, A., Samus, M., Fead, E. et al., "Fuel Permeation Performance of Polymeric Materials," SAE Technical Paper 2001-01-1999, 2001, https://doi.org/10.4271/2001-01-1999.
- Fuel Permeation performance of polymeric materials analyzed by gas chromatography and sorption techniques. Nulman M., Olejnik A., Samus M., Fead E., and RossiG., SAE Technical Paper 981360 (1998)
- Code of Federal Regulations. 40 CFR, Ch. I, Subpart B - §86.101 to §86.157.
- California Code of Regulations Title 13. See in particular: Evaporative Emission Standards and Test Procedures for 1978 and Subsequent Model Motor Vehicles. Recent information on new and proposed CARB regulations can be accessed through the CAR webpage at: http://www.arb.ca.gov
- Both the EPA  and the CARB test  rely on propane calibrated Flame Ionization Detectors (FID) to measure hydrocarbon emissions. In addition to hydrocarbons (e.g., species containing only carbon and hydrogen) such FIDs detect other carbon containing species (alcohols, ethers, fluorocarbons,…), so that these species also contribute to evaporative emissions total.
- See: Real-Time Non-Fuel Background Emissions. Haskew H.M., Cadman W.R., Liberty T.F. and Mitsopoulos C.G.; SAE Technical Paper 912373 (1991). Powertrain fluids, such as engine and transmission lubricants, are known to have quite low volatilities (as measured by tests such as ASTM D5800 and ASTM D5480): however, during service, fuel hydrocarbons can become dissolved in the engine oil due to combustion gases that make their way into the crankcase by leakage past the piston rings (a process known as “blow by”) and eventually contribute to the evaporative emissions.
- See: Activated Carbon Canister Performance During Diurnal Cycles: An Experimental and Modeling Evaluation. Johnson P.J., Jamrog J.R. and Lavoie G.; SAE Technical Paper 971651 (1997).
- Speciation of evaporative emission from Plastic Fuel Tanks. Fead E., Vengadam R., Rossi G., Olejnik A. and Thorn J.; SAE Technical Paper 983176.
- Plots documenting the decrease in evaporative emissions since 1971 can be found in: Diurnal Emissions from In-Use Vehicles. Haskew H.M. and Liberty T.F.; SAE Technical Paper 1999-01-1463 (1999).
- Such a tool has been recently developed and it is described in “Predicting multicomponent fuel permeation across multiplayer polymeric walls.”, RossiG. and SamusM.. (To be published).
- These definitions agree with those found in SAE Recommended Practice J1681 SEP93, p12.76. However, part of the permeation data presented in Tables I, II and III were obtained using mixtures prepared on a weight rather than on a volume basis (see footnotes to these tables). The relevant solvent densities are 0.867 for toluene, 0.692 for isooctane, 0.791 for methanol, and 0.785 for ethanol. Therefore, a 50/50 mixture by weight of toluene and isooctane (fuel Cw) corresponds to a 44.4/55.6 mixture, by volume, of toluene and isooctane. Similarly, a 42.5/42.5/15 mixture by weight of toluene, isoctane and methanol (fuel CwM15) corresponds to a 37.7/47.6/14.7 mixture, by volume, of toluene, isooctane and methanol. Finally, a 45/45/10 mixture by weight of toluene, isooctane and ethanol (fuel CwE10) correspond to a 39.9/50.3/9.8 mixture of toluene, isooctane and ethanol. In other words these mixtures contain a somewhat (about 10%) smaller amount of toluene, and essentially the same amount of alcohols as the corresponding SAE J1681 fuels C, CM15 and CE10.
- The most common method to measure fuel permeation until now has been an adaptation of procedure ASTM-E96 (the “cup test”) for determination of water vapor transmission rates. Fuel (rather than water) is used as the permeant. The overall weight loss of a cup containing fuel and sealed with the material being tested is measured as a function of time. See for example: Fuel-Alcohol Permeation Rates of Fluoroelastomers, Fluoroplastics and other Fuel Resistant Materials. Stahl W.M. and Stevens R.D.; SAE paper 920163 (1992). Other methods (SAE test procedures J30, J527 and J1737) devised to determine fuel permeation from finished fuel system components (such as tubes, hoses and fuel line assemblies) are also based on measuring an overall weight change.
- Fuel Permeation Method Analysis Correlation. Stevens M. and Demorest R.. SAE Technical Paper 1999-01-0376 (1999).
- See for example: Permeability and Diffusion Data. Pauly S.; in “Polymer Handbook” edited by Brandrup J.; Wiley, New York (1999), pag. VI/543. In fact, the amount of solvent going through the slab will in general depend on the solvent partial pressures p and p-Δ on both sides of the slab. In other words, the steady state flux, Fss, across a slab of thickness L is a function, Fss(p, p-Δp), of both p and p-Δp. In order for P to be defined unambiguously, that is independently of the units used to measure the pressure, the partial pressure difference Δp between the two sides must be infinitesimal, i.e., P=P(p)= Fss(p, p-Δp)L/Δp, is defined in the limit of Δp going to zero.
- The amount of solvent present within the slab depends on the partial pressures p and p-Δp on both sides of the slab. Again (see ) one has to assume that there is an infinitesimal partial pressure difference between the two sides: then the local solvent concentration ϕ of solvent within the slab is nearly constant, i.e., it is not a function of the location x within the slab, and depends only on p.
- In terms of P(p) (see ) the vapor transmission saturated pressure of the solvent.
- Multicomponent transport in a model elastomeric system has recently been examined for simple binary mixtures. See Swelling and transport in polyisoprene networks exposed to benzene-cyclohexane mixtures: a case study in multicomponents diffusion. Schlick S., Gao Z., Matsukawa S., Ando I., Fead E. and Rossi G.; Macromolecules 31, 8124-8133 (1998).
- The main goal of the preconditioning treatment is to desorb water that may have been taken up by the films (especially nylons and EVOH) because of exposure to ambient humidity. In the sorption tests this is a significant concern because having water present in a film sample prior to the beginning of the test results in an erroneous reading of the initial film weight. In the permeation tests, the problem is not as severe, especially if one is interested only in the steady state vapor transmission rate. The initial presence of water in the film may accelerate solvent diffusion in the early part of the test, but since one side of the film is continuously swept with dry nitrogen, a significant amount of the water present will eventually be desorbed from the film: this desorption process should be accelerated when the permeant contains alcohols, as in the case of CE10 and CM15 fuels.
- At steady state the flux no longer changes with time. One may (arbitrarily) define tss, the time required to reach steady state, as the time at which the flux reaches, say 95%, of its steady state value.
- The two (liquid and vapor contact) data points for ethanol at t=14.5 h correspond respectively to fluxes of 1.63 and 0.35 gm/m2 day: this large difference cannot be attributed to experimental uncertainties in temperature or slab thickness. In view of the fact that no such difference is seen in any other pair of data and in view of the fact that we have obtained good agreement between liquid and vapor contact permeation results in many other instances (see Ref. ), we are led to view this discrepancy as the result of some handling error.
- A recent SAE paper: A Comparison of Vapor and Liquid Fuel Permeation of Fuel System Polymers. Brahmi A. and Wolf R.. SAE Technical Paper 1999-01-0380 (1999), describes experimental results that seem to imply that permeation through polymers in contact with a liquid fuel is larger than permeation through the same polymers in contact with the saturated vapor of that fuel. The results presented in this paper may in fact be correct without violating the thermodynamic principle that diffusion is driven by chemical potential differences: this principle implies that permeation through a plastic film does not depend on whether the film is in contact with liquid or with the vapor in equilibrium with the liquid. Brahmi and Wolf show that plasticizers present in the polymer are extracted more easily in the case of liquid contact, a result that is reasonable, since in this case the bulky (low volatility) plasticizer molecules can be more easily convected away from the surface of the polymer in contact with the liquid. This in turn modifies the property of the polymer film and in particular may make it more brittle. In these conditions leakage would be favored. Note that fuel loss from the apparatus used by Brahmi and Wolf is measured gravimetrically, so that the experiments cannot distinguish between losses due to permeation and losses due to leakage .
- From a practical experimental standpoint one should be aware of the fact that elastic stresses due to the combined effect of swelling and of the constraints imposed by the test geometry (specifically in the locations where the film is fastened between the cups) are different for films of different thicknesses. Furthermore, very thin films may undergo different deformation depending on whether the film is in contact with the liquid or with its saturated vapor, since in the first case the film has to support the weight of the liquid solvent. Finally, it should be kept in mind that the thickness of commercial plastic films is not uniform: a standard deviation of, say, 2 μm in the thickness is much more significant for films that are 20 μm thick than for films that are 200 μm thick. For all of these reasons one should not be surprised to find in practice some deviations from the simple relation of inverse proportionality between steady state flux and film thickness.
- Of these data, only part of those contained in Table III were reported in Ref. . A few data relevant to fuel systems can be found in Ref.  as well as in: Permeability and other film properties of plastics and elastomers. Plastic Design Library, Norwich, NY (1995): however, there are questions as to the level of accuracy (and self-consistency) of the data reported in these two latter sources.
- A study of solvent induced changes in the glass transition temperature of EVOH using FTIR and dynamic mechanical spectroscopy. Samus M. A. and Rossi G.; in Multi-Dimensional Spectroscopy of Polymers, ACS Symposium Series 598 (1995) edited by Urban M. W. and Provder T., pg. 535 - 550.
- Methanol absorption in Ethylene-Vinyl Alcohol copolymers: relation between solvent diffusion and changes in Tg in glassy polymeric materials. Samus M. A. and Rossi G.; Macromolecules 29, 2275 - 2288 (1996).
- Macroscopic description of solvent diffusion in polymeric materials. Rossi G.; Polym. Trends 4, 337-343 (1996).
- Estimating Real Time Diurnal permeation from Constant Temperature Measurements, Lockhart M., Nulman M. and Rossi G.; SAE Technical Paper, 2001-01-0730.
- See for example: “Barrier Polymers” by DeLassus P. in Kirk-Othmer Encyclopedia of Chemical Technology, Fourth Edition, Wiley, 1992; Vol.3, p.946-948.
- In fact, suppose that PA and PB are the vapor transmission rate results determined from two separate measurements at temperatures TA and TB: in other
- Eq. (3) assumes that the speciated vapor transmission rate at time t during the cycle equals the speciated state vapor transmission rate Pi* at the temperature T(t) corresponding to time t. Typically, before a SHED test the vehicle or component being tested is stored at temperatures lower than those that it experiences during the cycle: therefore, the actual vapor transmission rate of fuel constituent i at time t is smaller than Pi*(T(t)). As a result the value of Fi*(cycle) computed from Eq. (3) produces an upper limit for the expected SHED test result.
- In fact, for a wide range of dependencies of D on solvent concentrations, Eq. (5) overestimates tss by about
- A Method to measure Air Conditioning Refrigerant contributions to Vehicle Evaporative Emissions (SHED Test). SieglW.O.; GuentherM.T., SAE Paper 1999-01-1539.