Tobacco Documents Online

J co r :3- L~ U L tJ Imperial Tobacco Limited/Limit6e ~LOATION OF "DE~IVER VERSION 3~* A COMPUTER PRO~ TO PREDICT SMON~ COMpOnENTS ! A CANADIAN PERSPECTIVE AUTHOR : C. McBride RESEARCH AND DEVELOPMENT DIVISION MONTREAL THIS REPORT MUST NOT BE COPIED OR SHOWN TO UNAUTHORIZED PERSONS. 57oSOGGIB

RESEARCH LABORATORY REPORT NO. IB1 COPY NO. IMPERIAL TOBACCO LIMITED RESEARCH & DEVELOPMENT DIVISION MONTREAL - RESTRICTED R&O-hO01-87 c.7 EVALUATION OF '~DELEVER VERSION 3" A COMPUTER PROGRAM TO PREDICT SMCKE DOMPONENIB A CANADIAN PERSPECTIVE AUTHOR : C. McBrlde ISSUED BY : DR. S.R. MASSEY DATE ISSUED : 3ANUARY 15, 1987 DISTRIBUTION Dr. P.3, 8unn Or. S.R. Massey Mr, A.L. Heard Dr. R. B1nns Mr. E.E. Kohnhorst Mr. R.G. Nlcholls Herr. E, ~Ittershaus Dr. E. Rausch Dr. C,3.P. de BIq~elra L~brary Copy NO. l 2 3 4, N 6. ? B, IO Ii ]2 13, 14, 15 570506619

• TABLE OF CONTENTS PAG~ SUMMARY ......................................................... INTRODUCTION ........................................................... 3 := 7. METN~D ............................................................. O 1.1 CIGARETTE SELECTIQ~ ...................................... 5 1,2 MEASURED DELIVERIES ...................................... D I 3 FREOICTER OELIVER~ES ..................................... 6 2. RESULTS ARD DISCUSSION ............................................. 7 2.1 STANDARD SMOKING ......................................... 8 2,I.I OWER~STIMATION OF FIRST PUFF DEtlVERIES ........... 13 2.1.2 REFERENCE DREDIE~ .......................... " .... 14 2.1.3 PURE PREDIDTION ................................... 15 2.2 HUMAN SMOKING TRIAL ...................................... ~7 S, CONCLUSIO~ ........................................................ 22 REFERENCES ......................................................... 23 APPENDIX l SUMMARY OF CANADIAR CIBARFTTE PAPER ANALYSIS 51 APPENDIX D FILTRATION EFFICIENCY AND HYDRAULIC DIAMETER MEASUREMENTS 52 5705066;~0

R&D-HO01-87 c.7 SUMMARY This report describes a study to assess the accuracy of the DELIVER {Version 3) Computer Program to predict total and puff by p~ff cigarette deliveries under both standard machine and human smoking ¢ondltlo~s, Four commercial cigarettes currently available on the Canadian market were used In the evaluation. These included a hlgh tar delivery brand (16 mg), two m¶d tar delivery brands (]l mg) and a low tar delivery brand {4 m9). The results ~an be summarlsed as follows: FOg STANDARD CONDITIONS; TOTAL AND PUFF BY PUFF OELIVERIES I. Reference Predlctlen Mode: Cigarette dellverles, excluding those of the first puffs, could be predlctEd with good agreement (within the ~easured range) regardless of the cigarette or smoke constituent examined. Pure gredlctlon Mode: a) Acceptable agreement between predicted and measured values for TPM an~ tar del~verles was found for the high and mid delivery cigarettes. b} NO agreement for TPM and tar deliveries was found for the low delivery cigarettes. c) No agreement between measured and predicted deliveries of nicotine was found for any of the ¢fgarettes examined. g. The inaccuracy wlth Which the DELIVER model predicts deliverles for the low delivery cigarette may be due to the presence of a moderate (14.5~) amount of expanded tobacco for which DELIVER does not adequately account. The relationship between averag~ tar per puff and Inc~uslon level of expanded tobacco used in DELIVER maw not be representablve of realistic Incluslon level ranges. 570~q06621

-2 - FQR HUMAN SMOKING CONDITIONS: Reference Prediction MoRe: a) It was possible Do match Dotal clgarette deliveries but only by altering fiItratlon effi¢lencles to very low levels. b) It was nod possible to obtain any 8greemenD between measured and predicted puff by p~ff deliveries. Based on these results. It would appear that DELIVER {version 3) is improved in terms of Reference Prediction but not in terms of Pure Predictlon. AS such It has potentlal as a descriptlve design tool for standard smoking condltions, but nQt as a tool for exploring wider hypotheses which mlght be encountered in Duman Smoking Research. Its usefulness In the laDDer area is therefore limited. ITL Project No, T-DDTg Human Smoking Behavlour Prediction Dellver Puff by Puff Computer Program Model]Inq Smoke DellvBries Clg~rette D~sign $7050GG22

INTRODUCTIOH The q~anhltetlve ~etermi~ation of smok6 constituents in ~n important actlvlhy in tobacco research. It ~s especl~lly Important in the area of Product Research $1nce accurate quant|fIcatlon of ~moke co~ponents Forms ~ ba~Is ~or the Investi~atlon of the brinclple~ underlyln~ combustlon of tobacco, the relationship helen ~obacco s~oke and prod~ct deslRn and the processes Involved in the Interactions between the smoker and ~he product. However, the measurement ~f ~oke c~mpene~t~ can be a tlme con~glng an~ COstly en~e~o~r. Any new Or a~ternatlve technique ~hIch COUld l~pr~v~ the dete~m~natlon of s~o~e constlt~nts whether through Increase~ ~curac~, improved control oE varlables or a reduction In time and cDst would be ~ ~sset. A computer program (OELIVER) designed for quan~Itatlve prediction of mainstream smoke constltuent5 has been developed by BAT (UK & E} R&O. This program has recently been made available to users with terminal access to the Soutbampton CYRER computer. If rellable, this program could plaw a role in some areas of research, In particular, Human Smoking Research. An Investigatlon of a previous version of DELZVER (version 2) assessed the ability of the program to predict under ~tandard condltlons the dellv~rles of tar &~d nlcot~ne for three C~n~dla~ Cigarettes ranging in tar dellver~ from 9 -ll mR (1). ~ased, %n part, on some ~f the recome~datlons of thls report, comm~nlcate~ Informally to BAT, the second wrsion oh DELIVER w~s refined. DELIVER version 3 Is the latest release of the DELIVER cIDarette design model. This ne~ verslon ~as developed in order to ~ccommodate recentl~ obtained experlmental and theoretical flndl~gs of ~AT (~) with input from Will~ and to lesse~ extent ~TL, as Dreviously ~eted~ These improvements include: 5'7050~G ~'3

-4 - I) Reflnements to the expressions which descrlbe product formation and Eiltration. 2) Improvements In the predlct~on of the lenRth of the tobacco rod burnt durinR the puff. 3) Improvements In the predlctlon of the length of the toPacco rod Purnt during the smoulder period between puffs. 4) The Introductlon of flow dependant filtration coefficients into toe overall filter represent~t%on. Furthermore, thls release of DELIVER offers two additional modes of operation, Reference Prediction &rid Reference Calculation as well as the orlglnal Pure Prediction mode. A description of the fundamental gas flow model and related ~quati~ns and a completa d~scrlptlon of the latest DELIVEQ program can be found In a serles of BAT reports (2-5). The specific objective of tbe present study was to investigate t~e accuracy wlth which the latest reTease of OELIVER, version 3, could predict total cigarette and puff by puff del~verles of tar and nlcotlne for Canadlan clgarettes as ganerated under bobD standard and h~man smoking condltlons. Measured and predicted deliveries ware CO~pared for four brands (dellvery range 4~16 mg) ¢~rrently available on toe market. This report describes the results of thls investigation and makes recommendatlons ~oncerning the usefulness of the DELIVER program for Human Smoking 6ehavlour studles. 570506624

- 5 - METHDD l.I CIGARETTE SELECTION Three brands were chosen to re~resent hlgh, mid and low delivery clgarettes currently avali~bie on the m~rket. These brands were du gautier Klng Size (16 mg declared tar del~wry}, Playerls Extra Light Kin~ Slze (11 mg declared tar delivery) and Matlnde Extra Mild Kinq Slze (4 mg declared tar delivery). A fourth brand, du Maarier Light K~ng S~ze (]~ mg declared tar) was chosen for the Humen Smoking comparison. After cond~tlonlng at 60% R.H. and q2°~ for 48 hQurs, the cl@arettes were selected for weight, pressure ~rop a~d ventilation levels ~ccocdln~ to a mean v~Tue (n=~O). ~etai~ of the cI~ret~e~ ~e ~Ive~ in Table I. A sufflclent number of clgarette$ w~re seIected foc both smoklng a~d for the determlnat~on of the various Input parameters required by the gELIVER program. Thls was to en~ure that the measured and p~edlc~ed deliveries could be attrlbuted to the san~ cigarettes and wo~id therefore be co.parable, 1,2 MEASURED DELIVER{ES A 2~ port Borg~aldt rotary smoking engine was used to collect the malnstr~m smoke for both tota~ cigarette and puff by puff delIverle~. Statldard ~mok~g ~ond~tlons were ~$ed In ~be flr~t part of this st~dy. Te~ c~g~rettes per ru~ were ~oked, i.e, the flr~t p~F5 Of te~ cigarettes were ¢oilect~d on a s~ngle Cimbrldge Filter pad. ~e second p~f~s of the same ten clgarett~s o~ a ~eeond Cambridge Fllt~r pad ~nd $o o~ ~t~] the specified butt length w~s r~ched~ I~ the cas~ ~ Matinee E. M. ~Ing SIZe, 20 c~r~tes were SIBoked per run. E~eh ru,~ was ~e~e~ted six t~me$. In ~r~er to minimize an~ potential 7o~ses~ all Ca~brld~ filter a~se~b]le~ wer~ sealed upon completion Of s~okln~ i~edlately ,weighed a~d the p~ds p~aced in ~ apdro~rlate extraction solutions. 570506625

The cigarettes were smoked to the follow%no butt lengths: du Maurter K.S. PTayer'$ E. I.K.S. 30 t tel du Maurler L. K.S, Matln~e E. M, K.S, - 33 mm Human mouth deliveries for du Maurter Light K.S. were detgrmlned using the ITL Smoke Dup71catar (8). Dup]Icatlon was also repeated six t~me~. Total partQculate matter (TPM), tar, nicotine Bnd water deliveries, both t~tal cigarette and puff by puff, were determined for each brand following standard ITL'procedures (?}. 1.3 PREDICTED DELIVERIES The cigarette design parameters required as input by the DELIVER progra~ are ~i~t~d In Table 2. The Values In th~s table represen~ the mean values for I00 clgarattas ~e~ected at the same time and to the s~me speclflcatIons as those selected for smoking. The exceptlons to thls are the values for: I) The clq~rette paper diffusion coefficients {D2T) which w~re measured by BAT (OK & E) R~D. 2) The average perforation hole diameter (HD) for Which the mean Df 24 measurements was used. 3) The flltrat~on efF~cle~cles (FET, FEN and FEW) which were determined from measur~ment~ o~ 25 c~arettes. 5705066;~6

_y _ The procedures for the determination of these values and the Individual results are appended IAppendlx l, Append!x 2), Deliveries for the three brands smoked under standard conditions were predicted using the three operating modes available wlth the DELIVER program; Pure Prediction, Reference Calcu]ahlon and Reference Prediction. The deliveries for the brand smoked under human conditions were Dredlcheg uslng the Reference Prediction ~ede. A descrlpt%on of the operatlng modes can be found in two BAT reports (2 and 5). RESULTS AND DISCUSSION The overall results are lll~strated In Table g. Since the Reference Predlct~on Mode use~ the results from the Reference CaIcu~atlon Mode to predict dellverle~, %+e. ~eference Ealculahlon 15 an Intermedla~e ~bep. only the R@ference Prediction Mode results ~re presented here. The Reference Prediction Mode an~ its relabionsdlp with the Reference Calculation Mode is explalne~ in greater detail later ~n this report. Throughout this study, good agreement between measured and predlcted deliveries was assumed when all the Individual puffs' predicted values fell between the two extreme measured values, and acceptable agreement when the majority of these vlIues fell within the measurement range. It was declded to use this criterion for comparison for tbe folIwing reasons: I) It offers the widest possible realistic range in which DELIVER may be applied. ~) It can be concluded that If predicted values fall within a measured range, then the model Is IndIcBtlve of experimental values regardless of the acceptabilit~ of that measured range. 570506627