Document 3Jdbnv92DadzOOKBE71vY7wvy

I u <-> "T r* A Model for the Diffusion of Vinyl Chloride Monomer from PVC Under Various Conditions of Storage -sri 30 Co CO UNWOOD B. CRIDER, MICHAEL M. O'MARA, and ROBERT L. BOWLES BFCoodrich Chemical Division Avon Lake Technical Center 1 Avon Lake, Ohio 44012 Otn oo 4* O This example of employee safety activity in the vinyl resin industry is illustrative of the efforts put forth by many to achieve acceptable VCM levels in work areas during all conditions ofhandling. Since use conditions differ gTeatly at temperatures above T, temperatures and are nonstatic,, it was deemed necessary to develop a model embracing many variables to allow prediction of atmospheric vinyl chloride monomer levels for any combination of the variables. . INTRODUCTION Trx7ithin thff past several years significant advanceV V mentshave been made within the PVC industry to minimize enrployee exposure to vinyl chloride monomer (VCM). Improvements in PVC manufacturing processes to reduce the residual vinyl chloride monomer (RV'CM) content of finishedf resins has been a key factor in maintaining acceptable VCM levels in work areas during packaging, processing, shipment and stor age. The diffusion of vinyl chloride monomer out of finished PVC resins has been adequately studied (1) at temperatures above-'the glass transition temperature (Tp). The results of.these studies have been particularly useful in reducing RVCM during manufacturing and to predicting RV.GM escape to the environment during processing. This new knowledge has also led to a rapid, simple gas chromatographic method (2)for the determi nation of RVCM' in PVC from the analysis of the vapor phase (head space) over PVC powders in a closed con tainer. This somewhat ideal behavior does not exist at tem peratures below Tp, however, and consequently one cannot use existing equilibrium data to predict the con centration ofVCM in the airspace above PVC resins and in particular under non-static conditions such as exists during the storage and transportation of bagged or bulk resin. In addition to the relative slow diffusion rate at or near ambient temperatures, other factors that affect VCM release include variable ventilation rates, massto-volume ratios, RVCM content of the resin and resi dence time in the storage or shipment compartment. The primary objective of the research reported in this paper was to develop a physical model for the diffusion of VCM from PVC resin under various conditions of expo sure during storage and shipment. The developed model is based on a statistically significant number of large scale diffusion experiments to show the interaction of all the above variables and to allow the prediction of atmospheric vinyl chloride monomer (AVCM) levels for any combination of the variables. EXPERIMENTAL It can be postulated that some reasonable understand ing of these storage variables and their interactions can be developed from a physical modeling of a storage or shipment compartment. It can also be rationalized that the accuracy of such a model will be much improved if the experimental design is on a large scale that can physically simulate all of the storage variables but under highly controlled conditions. Any experimental design of this type; however, must be thoroughly tested to establish the validity of the test measurements and to assure the absence of any errors as may occur through poor test controls, adsorption of VCM by the system, and possible interfering components that could lead to spurious results. Using a Computer Optimized Experimental Design (COED), 24 large scale experiments were selected to study the variables listed below to describe their effects and interaction on AVCM in a storage compartment or warehouse. The variables and ranges included: 1. Temperature--74-115F 2. Ventilation rate--0.5-3.0 turnovers/h 3. Loading (PVC volume/storage volume)--11-34 percent 4. Resin RVCM--0.01-123 ppm 5. Storage time--1-170 h In order to meet the demands of the statistically de signed (COED) diffusion study, the experimental phase of this program had to be capable of controlling the following experimental parameters: temperature, venti lation rate, and mass/volume ratio. The basic physical model that was chosen to meet these requirements was one based on sealed metal storage bins (55-gal drums' housed in an environmental chamber where the tem perature could be varied and controlled. A schematic of the entire experiment is shown in Fig. 1. The schematic * 168 JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1979, VOL. 1, NO- 3 * f 'crjiO-4! w*Gi Ci, i. . O" If: C,, -- concentration of incoming air stream Ct = concentration in drum at any time (f) V = volume of drum M -- flow into and out of drum t ~ min ~~~~ then: VCM into drum = VCM out of drum -1- accumulation MC0dt = MCtdt + Vdc Vdjdt - MC,, - MCt djdt = (C,, - Ct) R&S 005841 d< ce - ct = M/V d, (1) is divided intafourkey areas: (A) the ventilation system,. (B) the environmental chamber, "(G) the sampling sys tem and (D) the gas chromatograph with data handling system. Clean, dry nitrogen (A) enters the environmen tal chamber (B)'through a split manifold. The ten drums were piped into the two manifolds through ten flow regulators. La this way, ventilatk>n.rates could be inde pendently set.in each drnm-. The exhausted atmosphere from the drums was.directed ta a sampling system (C) that provided a split stream; one for vent and one for sampling. The sampling stream was interfaced to a gas sampling valve/gas chromatograph/data handling sys tem (D). Sequential sampling ofeach of the drums was possible with this-system. The environmental chamber is a Conrad Model WD^24 temperature humidity chamber with an internal vommeof938 cu ft. Tempera ture control inside the chamber was 2C; a maximum exposure temperature of I00C is possible within the chamber. . Calibration of the system was accomplished through the use of commercially-available standards (Precision Gas Sampling-Corporation) and dynamic dilution of those standards. A regression analysis of the calibration data indicates good linearity over the range from 1282000 ppb ofVCM in the atmosphere. The slope from the linear regression analysis was used as input to the integrator/calculator in order to calculate all data. It is I important to note that as the sampling proceeded to a range outside of the calibration range, new calibrations were carried out to reflect these different ranges. At no time was a calibration extrapolated beyond the experi mentally determined range and used as a basis for analysis. It was necessary to check out the performance of the system in terms of theoretical diffusion. The primary reason for doing this was to provide a final and indepen dent check of the model, the sampling system and the analytical system interacting together. It was further felt that this approach would lead to a definition ofany VCM system absorption (or leakage) problems. The ab initio derivation of the mathematical model which describes the build-up of a gas in a dynamic system is described below. integrating this over the limits from 0 --C( and 0 --* t yields: -- n(C0 -- C,) --,M/V(t) + A or * Ct = C0- ATm'in at t = o, C = o A - C0 C( = C.[l - e-*"TM] ' In essence, Eq 1 describes the build-up of VCM in a 55-gal drum when VCM at a concentration of C,, is flowed into the drum at a rate equal to M. At any time = t, the concentration ofVCM in the drum is Ct. To test this equation; a`l ppm VCM standard was flowed into a drum (216 liters) at a rate of 1.82 liters/min. The data from this experiment is presented in Fig. 2 and shows a satisfactory agreement between the theoretical and ex perimentally determined concentration in the modeling system. The analytical methodology used to determine the VCM in the exit vents from the drum experiments is as follows: . . ' Gas Chromatograph System --Hewlett-Packard Model 5711 Flame Ionization Gas Chromatograph 1000 x---* -theoretical X - EXPERIMENTAL r 10011 - I I I I I I I I 1 I 0 10 SO 30 40 50 60 70 60 90 100 110 TIME (Min.) Fig. 2. Theoretical vs experimental build-up ofVCM in 55 gal drum after experimental modification. r- r I t: r l r JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1979, VOL. 1, NO. 3 169 R&S 005842 "Method--10 ft x Vi in. Porapak Q, mesh 80/100; , * isothermal at 150C; nitrogen carrier gas / at 30 cc/min; detector--250C; injection port--150C I Gas Sampling Valve--10 cc sample loop jl Data Handling System System --Hewlett-Packard Model 3380 Integrator-Calculator RESULTS AND CONCLUSIONS The results obtained from the 24 COED designed experiments and a multiple correlation analysis of the results permitted: l)-the effects and interactions of the variables-on AVCM to be established and 2) the deriva tions of a model which permits the AVCM levels to be predicted for any combination of the variables studied. The resulting model for PVC resins with a porosity of 0.2 is: ' AVCM = Ro`-4rLO3V-'-i6[r0V<24-1 + 00001 m _j_ 2e-0<><>0,1Sr,] where: .AVCM Ro L V T t atmospheric concentration of VCM in parts per billion initial RVCM of the resin in parts per million percent loading; (resin volume/ warehouse volume) X 100 ventilation rate in turnovers per hour temperature in F time in h As expected. AVCM decreases as: RVCM of the resin decreases, Volume of resin being stored decreases. Ventilation rate increases. Storage temperature decreases (for time periods of 1-9 days); however, if the resin has been stored for longer periods of time (10+ days), then the AVCM is lower at higher storage temperatures. (This is the cumulative result ofa higher temperature causing more rapid diffusion at the beginning of aging; therefore, more rapid depleting of RVCM and, thus, lower AVCM at longer agings.) Storage time increases. j Using the relationships described in the model, -- 120^, ventilation = 0.5 turnovers/h, 80 percent load ! maximum RVCM resin levels can be determined which ing of a warehouse volume with VCM containing mate { will assure that AVCM levels will be less than 500 ppb rial), then the resin RVCM must be no higher than 8,-5 Ii OSHA action level for a variety of storagoe conditions. ppm. (See Figs. 3-7 which show the effects of temperature, Figure 8 shows how well the model predicts the time, ventilation rate, loading and RVCM on AVCM.) AVCM levels in BFGoodrich warehouse samples ob When PVC is stored under the least severe conditions tained in April, 1975 at Avon Lake, Ohio, Louisville. (temperature = 80F, ventilation = 0.5 tumovers/h, 30 Kentucky and Pedricktovvn, New Jersey. As the storage percent loading of a warehouse volume with VCM con taining material) the RVCM of PVC resin must be no conditions varied in temperature, ventilation rate, load ing and RVCM (all were approximately at one month's greater than IS ppm to assure that the action level is not aging = 720 h), the model predicts the different AVCM exceeded at 5 days. results quite well. The somewhat low predictions can If the 500 ppb (OSHA action level) must be met even probably be explained by the fact that the warehouse under the most adverse storage conditions (temperature samples were bagged (slowing the VCM diffusion 170 JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1979, VOL. 1, NO. 3 Fig. 7. AVCM os RVCM and warehouse loading. R&S 005843 slightly) whereas the model was based on material being stored in bulk form. The accuracy of tbe-model is 2:100 percent in being able to predict the "true" AVCM from a set of storage conditions and resin possessing a certain RVCM. ACKNOWLEDGMENTS The authors gratefully acknowledge the considerable contributions of H. T. Kim for his assistance in development of the experimental design and the development of a mathematical model evaluating effective dif- Fig. 8. Prediction of AVCM in warehouse samples. fusivities of VCM in PVC resins; to A. R. Berens for consultation during the course of this work; to C. J. Tomanek for assistance in most of the experimental work; to M. R. Ritland for assistance in the statistical analysis of the data and to a large number of people in BFG Manufacturing who cooperated in providing large quantities of the specific resins required for this study. REFERENCES h A. R. Berens, Pohjm. Prepr. 15,203(1974). 2. A. R. Berens, L. B. Crider, C.J. Tomanek, and J.M. Whitney, J. Appl. Polym. Sci., 19, 3169 (1975). JOURNAL OF VINYL TECHNOLOGY, SEPTEMBER 1979, VOL. 1, NO. 3 171