Document QMqRazz4x0oKnn2mOZ3x4r3V5

TO: Distribution TGG: \n\ r"HT UlU- A. 1 n- Rp XF:,, Interoffice Communication FROM: DATE: SUBJ: T. G. Grumbles February 15, 1991 VENTILATION IN CONFINED SPACES VISTA The attached article reviews the principles involved and makes recommendations on the most efficient way to ventilate confined spaces where oxygen deficiency is a concern. It's a lengthy article but if you go to page 139, the recommendations are summarized. T. G. Grumbles dlj Attachment Distribution: SAFETY DIRECTORS Bruce Trego-Aber, Balt, Harry Peirce-Blane, Kathy Perez-Hmd, K. L. Fogg-LCCP, R. V. Gantz-LCLAB, G. M. Shirley-LCVCM, Brent White-Okc, R. B. Martin-Austin, J. R. Drumwright, Rick Quy VVV 0000X1847 Ventilation to Eliminate Oxygen Deficiency in a Confined Space-- Part III: Heavier-Than-Air Characteristics Richard P. Garrison and Muzaffer Erig School of Public Health, The University of Michigan, Ann Arbor, Michigan 48109 This study investigated characteristics of ventilation co elimi nate oxygen deficiency caused by heavier-than-air (HTA) contam inants in confined space (CS) models. The HTA `'contaminant'' gases were carbon dioxide, halocarbon-22, and sulfur hexaftuor- ide, having specific gravities (SG) of approximately 1.5, 3 0, and 5.0, respectively. Neutrally bouyant nitrogen (SG = 0.98) also was tested, consistent with previous studies. CS models having cubical and vertical noncubical shapes were tested. Other variable design parameters included ventilation mode (exhaust and sup ply), volume flow rate, inlet/outlet elevation, and sampling ele vation inside the CS model. Testing confirmed that HTA con taminants would stratify in the CS models. Regressions of experimental data provide a means to calculate oxygen recovery times from initial deficiencies. Ventilation time was found to increase significantly with increasing contaminant SG. The effect was most pronounced at low CS model elevations. Supply ven tilation was more effective than was exhaust. A low ventilation inlet/outlet elevation was considerably more effective than was a high elevation. Geometrically similar cubical models of different size performed in very similar manners. Results for vertical, non cubical CS models emphasized the importance of introducing ventilation near the bottom of the space. Guidelines for CS ven tilation from this and previous studies, additional issues, and recommendations are discussed. fafTiSOfl, R.P.; Erig, M.: Ventilation to Eliminate Oxvaen in . ; ,.,,fmeu Space--ran ill: Heavier- Than-Air Characteristics. Appl. Occup. Environ. Hyg. 6:131*140; 1991. Introduction Oxygen can be displaced and diluted inside confined spaces (CS) to the extent that the atmosphere inside the space is unsafe to breathe. Any air contaminant can cause this if present m sufficient quantities. Of course, toxic con taminants can also pose serious hazards in much smaller quantities, without causing oxygen deficiency'. Relatively nontoxic contaminants, for which oxygen dis placement may be the primary potential hazard, and toxic contaminants may include gases having densities quite dif ferent from that of air. Gases having greater density than air are described as being heavier-than-air." The specific gravin' (SG) of a gas or vapor is defined as the ratio of the gas/vapor density to that of air. Specific gravity can be calculated by dividing the molecular weight (MW) of a given gas/vapor by 28.9, which is the average "molecular weight" for air as a mixture of oxygen (approximately 2196, MW = 32) and nitrogen (approximately 79%, MW -- 28). If a heavier-than air (HTA) contaminant is released slowly, in sufficient quantity and without agitation, then it is pos sible for the contaminant to stratify, causing significantly higher concentrations in lower regions of confined spaces. Failure to recognize this has lead to accidents, including fatalities, during confined space entries. This study examined characteristics of HTA contaminants in laboratory CS models. The overall objective was to ob serve and evaluate parameters affecting mechanical ven tilation to eliminate oxygen deficiency' caused by HTA con taminants. A second objective was to supplement an empirical database for CS model ventilation which could be used in the development of computer-designed models to predict CS ventilation performance. This study is Part IH-of a series investigating'ventilation characteristics to eliminate oxygen deficiency in confined spaces. Part I examined characteristics of a cubical CS model, using nitrogen (neutrally buoyant. SG = 0.98) with a rel atively broad range of design parameters)l> Part II also used nitrogen and investigated characteristics of noncub ical CS models, constituting progressive expansions from the basic cubical configuration.121 These studies have been pan of a research project funded by the National Institute for Occupational Safetv and Health. ooooi'18* Experimental Method '* The experimental facilities and testing methods for this study were very similar to those of the two previous stud ies)121 A "contaminant" was used to displace oxygen (02) inside a confined space model. Oxygen concentration (%02) APPL OCCUP. ENVIRON. HYG. 6121 FEBRUARY 1991 1047-322X/91/0602-131JZ50/10 1991 AlH w.is measured at different loiations inside the C.s model, prior to and following mechanical ventilation ot the modd with fresh ( uno ilium in.ucd i .nr. The primary differences from the experimental procedures followed in the two previous studies were: GinuinunanLs oilier than N_>--sjiealicallv. the i IT\ gases carbon dioxide (COG halocarbon-22 (chlorodifluoromethane. HC22), and sultur hexalluoride (SF;>) were tested. The test gases were selected on the basis ot specific gravity and ot having relatively low hazard po tential. They were introduced slowlv at the bottom of the CS model to minimize dispersion prior to venti lation testing. Different sampling locations--with HTA contaminants, it was necessary to place greater emphasis on effects at different elevations inside the CS model, For this study, samples were taken at four elevations on the vertical central axis. In previous studies, the sampling locations were in different vertical quadrants ol the CS models. Limited CS model configurations--in order to have a manageable number of tests in the final experimental protocol, it was necessarv to reduce the number of CS model shapes and sizes and the variations of ventilation design parameters. Figure l illustrates the CS model shapes, ventilation con figurations. and sampling locations tested in this study. Three CS models were used: 1) a basic cubical shape <CS Model AV. 2) a vertical, noncubical shape I (IS Model H2) three times higher and greater (volume) than CS Mode) A; and 3) a double-sized cubical shape tCS Model b) geo metrically similar to CS Model A and having eight times greater volume. The CS model designations A. E2. and b are the same as used in the previous study.1 J' Additional designations (lower case letters) were also used to name specihe test configurations for each model shape. Table 1 summarizes the six CS modeLventilatioa sam pling configurations investigated in this study. These con figurations were selected to provide tour types ol comparist >n.v CS Models A;t versus Ah versus Ac--ventilation flow rate variation for the basic cubical model, with flow rate measured in terms of air changes per hour (ACH), i.e.. the 1-hour (oO-mmute) volume flow divided by the volume of the O model. CS Models Aa versus ba--doubling all CS Model ven tilation and sampling dimensions (i.e., maintaining geometric similarity) and using the same nondimen sional ventilation volume flow rate (ACH) for two cu bical models of different size. CS Models Aa versus E2a--expanding die cubical model to a vertical, noncubical shape, with ventilation inlet/ outlet (LO) and sampling elevations moving propor tionately to maintain the same nondimensional heights C\>H. percent of CS model height, H) and the same nondimensional flow rate (ACH); CS Models Ah versus E2b--adding cubical volumes on top of the original basic cube to form a vertical, non cubical shape, keeping the actual dimensional flow rate (cfm > and the I/O and sampling elevations (inches) the same. The final test protocol involved 96 test cases (6 x 2 x 2 x -4) consisting of the following: 6 CS model/ventilation configurations; 2 ventilation modes (exhaust and supply); 2 iniet/outiet elevations (25 and "5 %H); and -i contami nants of different specific gravities (N>, CO?, HC22. and SF(,, having SG = 0.98, 1.50. 3 00, and 5.05, respectively). Four sampling elevations were monitored for each test case: at 10. 25. *0, and ~0 oH for most cases. Figure 2 shows the experimental facilities A test run for VVV 000011849 + denotes sampling point a0 Aa.b.c Basic cube E2a `Expanded" E2b "Add-on" Vertical Noncubes Double-size cube geometrically similar to basic cube FIGURE 1. CS model configurations, with ventilation inlet/outtet and sampling elevations, for testing HTA contaminants. 132 ARPL OCCUR. ENVIRON. HYG fU2l . FFOOIMPV ,otu TABLE t. Characteristics of CS Models Tested with Heavier-Than-Air Contaminants Config uration Aa Description Basic cube and flow rate CS Volume Ccf) 8 Ventilation Flow Rate ACH cfm 20 2? Inlet/Qutlet Elevations %H inches 25 6 75 18 A& Basic cube, high Dow rate 8 60 8 25 6 75 18 AC Basic cube, higher flow rate E2a Vertical noncube, nondimensional expansion, sampling and I/O elevations moved upward E2b Vertical noncube, add-on voiume expansion, sampling and I/O elevations in bottom cube Ba Double-sized cube, geometrically similar to basic cube 8 24 24 64 120 16 20 8 same as Aa 18 54 20 8 8.3 same 25 as Ab 20 64 same as Aa 12 26 Sampling Elevations %H inches 10 25 40 70 10 25 40 70 same as Aa & Ab same as Aa 33 8.3 13.3 23.3 same as Aa 2.4 60 96 168 2.4 6.0 96 16.8 7.2 18.0 28.8 50.4 same as Ab 48 12.0 19 2 Contaminant Volume (cf) 4 4 12 same 3S Ab 32 involved checking and adjusting all four sampling locations to read a nominal ambient concentration of 21.0 %Ov introducing the desired volume of a particular contaminant gas into the CS model: turning ventilation ON; and recording %0> versus time (minutest using a com puter with analog/digital data conversion from the fourchannel O2 monitor. Data were collected continuously, with one measurement approximately every second for each sampling location until oxvgen recovery was very nearly complete (within 0.1 %02 of ambient) The CS models were constructed of wood (four sides) and clear plextglas (two sides). The overall dimensions (width x depth x height) of the CS models were as follows: Model A wax 0.61 x 0.61 x 0.61 m (2 x 2 x 2 ft); Model E2 was 0.61 x 0.61 x 1.83 m (2 x 2 x 6 ft); and Model B was 1.22 x 1 22 x 1.22 m (-4 x 4 x 4 ft). The circular CS top opening, Itxiated in the center of a quadrant, had a diameter of 15.2 cm (6 in.) for CS Models A and E2. and 30.4 cm {12 in.) for CS Model B. The inside diameter of the ventilation pipe was 5.1 cm (2 in.) for CS Models A and E2, and 10.2cm (4 in.) for CS Model B. A contaminant volume equal to 50 percent of the CS model volume was used for most test cases, consistent with previous studies using nitrogen. Each sampling lo cation was connected to one sensor of a four-channel, electrochemical oxygen monitor (ENMET Corp). The same fiow rate (approximately' 0.25 L'min) was drawn through each sample line, connected to a manifold at the inlet of small diaphragm pump. Ventilation flow rate was mea sured with a calibrated orifice plate and manometer. Previous .studies11-1 have shown that oxygen recoverv could be described reasonably well with a simple expo nential relationship, specifically. %0? = 21 -(21 - B)e-Cl in this model, tile coefficients (21 - B) and C represent the initial oxygen concentration and an oxygen recovery' time constant, respectively. The time constant (C) de scribes the rate of oxygen recovery from an initial defi ciency. This parameter can be used to calculate the ven tilation time required for the oxygen level to change from an initial value to some final value. Ventilation time (mm) _ In ((21 - %02 initiai)/(21 - %Qg final)] for oxygen recovery " q Using this mcxiel, regressions were performed on the experimental data to obtain values for C. For the 96 test cases, and four sampling locations for each case, the av- VVV 000011350 FIGURE 2. Experimental facilities showing CS models, with contaminant de livery, ventilation, sampling, and oxygen monitoring systems. APPt 0CCUP. ENVIRON. HYG. 6(2) FEBRUARY 1991 folt rj TABLE II. Regression Values of Oxygen Reccvery Time Constant (C) lor Exhaust * Ventilation with Different Specific Gravity Contaminants and CS Model Configurations Oxygen Recovery Time Constant C CS Model Parameters Contaminants (Specific Gravity) Configuration Volume Flow Rate (ACH) Inlet Elevation f%H) Sampling Elevation (%H) N, (0.98) CO, (1.50) HC22 (3.00) SF* (5.05) Aa 20 25 10 027 0.10 007 005 25 050 041 041 030 40 062 061 061 044 70 100 120 123 100 75 10 032 002 0013 0.012 25 030 0021 0017 0016 40 0 32 0 043 0036 0035 70 0 32 too 082 130 Ab 60 25 10 0.90 025 013 009 25 0.80 1.10 1 00 0 73 40 0 82 1 50 1.30 100 70 0 82 2.40 2 20 1.70 75 10 0.70 0 04 0 021 001 25 061 0045 0027 0011 40 067 0.10 0 05 0.021 70 0 67 1 20 080 050 Ac 120 25 10 1 20 050 020 0.14 25 too 1 50 1 70 200 40 1.00 210 230 2.50 70 1.10 2 40 300 250 75 10 095 0 055 004 0014 25 088 010 0 05 0.015 40 080 0 24 0 22 004 70 0 85 150 020 1 40 E2a 20 25 10 035 003 0 026 002 21 0 33 100 060 0.50 4(1 0 33 1 30 080 0 70 70 0 33 150 1 40 1.10 < < o Q O E2b o I-4 CO U1 t-- 75 10 0.32 0017 0016 0 012 25 032 002 0045 0.035 40 0 31 005 011 010 70 033 200 2.00 2 40 20 B.3 3.3 090 017 0.11 009 3.3 080 080 1.00 079 13.3 080 1.00 1.50 1.00 23 3 080 150 240 230 25 3.3 0 70 0022 0.018 0014 8.3 060 0029 0026 0015 133 0 70 006 006 0022 23 3 0 70 008 0.85 060 TABLE III. Regression Values of Oxygen Recovery Time Constant (C) for Supply Ventilation with Different Soecitlc Gravity Contaminants and CS Model Configurations Oxygen Recovery Time Constants. C Config uration Aa Ab Ac fc2a E2b CS Model Parameters Volume Flow Rate (ACH) Outlet Elevation (%H) 20 25 75 60 25 75 120 25 75 20 25 75 20 83 25 Sampling Elevation (%H) 10 25 40 70 10 25 40 70 10 25 40 70 10 25 40 70 10 25 40 70 10 25 40 70 10 25 40 70 10 25 40 70 3.3 83 133 23 3 33 83 133 23.3 Contaminants (Specific Gravity) N, (0.98) 084 083 080 051 041 0 38 0.41 0.40 0 89 086 092 092 0 92 086 0 94 092 1.40 1.40 1 40 1.48 150 f 60 155 160 061 061 061 0 59 041 042 042 042 0.90 086 092 0.92 090 0.90 092 0.92 CO, 0.50) 0 30 030 031 031 0 10 011 020 0 44 083 084 088 0 89 0 75 0 74 002 083 1.50 150 1 00 1.40 1 50 140 140 130 0 30 0 30 030 031 014 0.20 039 0.43 0 70 050 0.53 054 045 0.42 044 0 43 HC22 (3.00) 016 030 0 30 030 004 005 011 060 081 081 082 081 048 049 069 0 70 1.20 130 1.30 120 070 080 0 70 050 028 0 37 037 037 014 0.20 0 43 0 42 100 0.87 077 080 040 0.55 067 0 70 Sf. (5.05) 011 020 0 30 040 0 027 0 032 0 087 050 069 0 69 069 068 0 20 0 30 0 53 060 0 65 080 080 065 060 065 0 70 0 70 020 030 0 35 0 30 007 012 030 025 068 069 069 060 020 0.33 0 55 063 A P P L OCCUP. ENVIRON. HYG. 6(2) FEBRUARY 19 erage coefficient of determination <R2) was 92 percent, ranging from 80 to 99 percent. The standard errors of the regression estimates of C were within 1-5 percent. These results were considered acceptable, although previous studies involving only N2 were found to have somew'hat higher R2 and lower standard errors. Results Tables II and III provide the regression values of C for all of the test cases except those involving CS Model Ba. which are adequately represented by the data for CS Model Aa (discussed subsequently). Values of C are given for each sampling elevation, for the test configurations, and for ex haust (Table II) and supply (Table HI) ventilation. These tables supplement the database of C values established in the previous rwo studies.1121 The regression database can be used to compare ventilation effectiveness for different CS model situations. A one hundredfold difference in C values represents a one hundredfold difference in venti lation time, e.g., 3-9 minutes when C = 1.2 and 390 minutes when C = 0.012 for oxygen recovery from 10 to 20.9 %02. Figures 3 and 4 present oxvgen recovery characteristics (%C>2 versus minutes) of HTA contaminants for the basic cubical CS Model Aa at a flow rate of 20 ACH. Figure 3 compares results for the lightest (N2, SG = 0.98) and heaviest (SF^, SG = 5 05) gases at three sampling eleva tions (10, 25, and 40 %H) for exhaust and supply ventilation RGURE 3. Oxygen recovery for the cubical CS Model Aa for N2 and SF6 at three sampling elevations (10. 25, 40 %H), with high (75 %H) 1/0 elevation at 20 ACH. at high i" %\{ \ I/O elevation. Figure 4 shows 02 recovery for the same CS Model Aa at 20 ACH. with high ("5 %H) and low 125 %H) I/O elevations for all four contaminants (N>, CO>, HC22. and SF,,) at the lowest (10 %H) sampling elevation. Figure 5 provides a general comparison of the two cu bical models of different size iCS Models Aa and Ba) which w-ere geometrically similar and were ventilated at the same nondimensional volume flow rate of 20 ACH. Data wrere taken from 02 recovery curves of all 32 test cases for these models. The purpose of this comparison was to evaluate whether findings for small-scale laboratory models could be applied reasonably to larger models, perhaps ultimately to actual confined spaces. Estimating the length of time needed to ventilate a con fined space is a fundamental design objective. For 02 de ficiency. it is obviously a function of the rate of recovery, which can be represented empirically by an 02 recover.' time constant (C). Previous studies*1-2' have compared dif ferent CS models using ventilation times calculated using regression values of C; specifically, "delta T." the time for nearly (99%) complete recovers- from a 10 %02 initial level to 209 %02. For consistency, this same lime parameter w as used for comparisons in this study. However, it should be kept in mind that recovery often required more time, sometimes considerably more, for HTA conditions involv ing initial oxygen levels below 10 %02 and, therefore, requiring additional time to reach 10 %02. Figure 6 illustrates effects of changing ventilation vol ume flow rate on ventilation time, delta T, for oxygen recovery from 10 to 20.9 %02, calculated from time con stants in Tables II and III. These data are for the basic cubical shape (CS Models Aa, Ab, and Ac) for all four contaminants under exhaust and supply ventilation with low (25 %H) I/O elevation. Figure 7 provides comparisons of ventilation time char acteristics for cubical (CS Models Aa, Ab) and larger, ver tical, noncubical (CS Models E2a. E2b) configurations. These comparisons present data for the lowest sampling eleva tion, for which oxygen recover.' times were greatest. Con sequently, the HTA characteristics illustrated are more pro nounced than they were at higher sampling elevations. The comparisons of Figure 7 include significant differ ences in ventilation design. CS Models Aa and E2a utilized the same nonditnensiotial design parameters: ventilation flow rate of 20 ACH (2.7 cfm for Aa; 8.0 cfm for E2a); I/O elevations of 25 and -,5%H; and sampling elevation at 10 %H. CS Models Ab and E2b had the same dimensional design parameters: ventilation flow rate of 8 cfm (60 ACH for Ab; 20 ACH for E2b); I/O elevations of 6 and 18 inches; and a .sampling elevation of 2.4 inches. Figure 7 can be described as involving comparisons between the basic cu bical CS model and a proportional, threefold, vertical "ex pansion" of the cube and the ventilation design parameters (Aa versus E2a) and comparisons between the basic cube and a vertical noncube made by "adding on" two cubic volumes on top of the basic cube, leaving the ventilation design in the bottom cube unchanged (Ab versus E2b). APPL 0CCUP. ENVIRON. HYG. 6(2} . FEBRUARY 1991 WV 000011S52 135 Oxygen concentration (%Oj) EXHAUST Ventilation time (minutes) FIGURE 4. Oxygen recovery tor the cubical CS Model Aa at the lowest sampling elevation (10 %H) for exhaust and supply ventilation at 20 ACH. FIGURE 5. Comparison of oxygen recovery data for two geometrically similar cubical CS models (Ba vs. Aa) of different size for all test cases, with ventilation at 20 ACH. vvv 000011353 136 APPL OCCUP. ENVIRON. HYG. 612) . FEBRUARY 1991 Discussion In ail of the test eases, mechanical ventilation caused more rapid air mixing and contaminant dilution than would have occurred without it. Figures 3-~ show that recovery' from HTA contaminant stratification and oxygen deficiency occurred within periods of time ranging from minutes with effective ventilation design to hours for heavy contami nants with poor ventilation design. Preliminary testing in dicated that O: recovery by diffusion alone would take much longer--roughly 2-4-36 hours for the basic cubical CS model containing 5Fft. Contaminant density had a significant effect upon ven tilation effectiveness. Figures 3. 4, 6. and ^ show that ven tilation time generally increased with increasing contam inant density-. This effect was most pronounced for the lowest < 10 elevation in the CS models, as shown in Figure 3 Differences between contaminants were least ev ident for effective air mixing, such as caused by low (25 %H) outlet supply ventilation in Figures 4. 6, and?. Supply ventilation was generally- more effective than ex haust ventilation. Figures 3, -t, 6. and ' show supply' ven tilation causing oxygen recovery- in less time than for ex haust under otherwise identical conditions for ail of the CS models. This advantage tended to improve with in creasing contaminant density (Figures 3. 4, 6, and 7), was greatest at low elevations (Figure 3), and was not affected significantly by CS shape variations (Figure 7). The mag nitudes of the differences between supply and exhaust i ventilation can be dramatic, e g., from several minutes for complete recovery under supply ventilation to an hour or more for HTA contaminants under exhaust ventilation. The reason for more rapid contaminant dilution under supply ventilation is the increased dynamic mixing caused by the airflow jet discharged from the ventilation outlet. I/O elevation was found to have very significant effects upon ventilation characteristics. Figures 4 and 7 show that ventilation time was less for low (25 nH) I/O elevations compared to higher C75 %H) elevations. This advantage of tow I/O over high I/O was most pronounced for high contaminant densities and low CS mode! sampling ele vations (Figure 4). Ventilation volume flow rate can significantly affect ven tilation time for oxygen recovery. Figure 5 shows venti lation time, delta T. dropping substantially for increasing volume flow rate (ACH) for both supply and exhaust ven tilation. This increase was not in proportion to change in flow rate. A three-times change from 20 to 60 ACH reduced ventilation time for HTA contaminants by less than 50 per cent for exhaust ventilation and by variable amounts with different contaminants for supply ventilation at the lowest (10 %H) sampling elevation, with different characteristics at higher elevations. Reductions in ventilation time were observed over the full range (20-120 ACH) tested for ex haust, but there was little improvement above 60 ACH for supply ventilation for the cubical CS Model A Geometric similarity is necessary for two CS models of different size to experience similar ventilation perfor- Vantllotlon volume flow rote (ACH) FIGURE 6. Oxygen recovery times, delta T(10--20.9 %0j), for different ven tilation volume flow rates (cubical CS Models Aa, b. c) lor different contam inants. with low (25 %H) 1/0 elevation. ...... _ -^ VVV 000011854 mance. Similarity' must apply for both the CS configuration and the ventilation design. Figure 6 shows that geometric similarity' was sufficient for the two cubical CS models (Aa and Ba) of this study to exhibit nearly the same oxygen recovery. Empirical results such as found by this study (Tables II and III) are limited to the CS model and ven tilation design parameters which were tested. Geometric similarity may make it possible to extend application of the data to larger field CS situations. However, it may be unlikely that any field spaces will conform closely enough io die icsi uaia ui aiiow- nigniv accurate predictions of ventilation time. This study allows limited consideration of CS shape ef fects; .specifically, Figure " provides comparisons of cubical and vertical, noncubical CS models. The effects of differ ences between cubical CS Model Aa and the "expanded," noncubical Model E2a include a three times greater ven tilation rate and amount of contaminant. Higher flow rates (cfm) for CS Model E2a resulted in increased vertical air flow velocities through the unchanged horizontal crosssectional areas of the CS model and the ventilation pipe. Higher velocities caused increased air mixing within CS Model E2a. Low (25 %H) elevation supply ventilation was most effective for eliminating O2 deficiency in both the cubical and noncubical CS models. Findings for the "ex panded," vertical, noncubical CS model indicate a signif- APPL 0CCUP. ENVIRON- HYG 6(21 FEBRUARY 1991 137 .'iiii-J * C( jJlfcv g 500 500 400 300 EXHAUST Ventilation Inlet elevation 6 In. 18 In. 200 100 + f Contaminants : d~l N 0 Ab 17/2X CO^ EZ8 HC22 fa Hi E2b SF< 500 SUPPLY Ventilation ,,t 400 Outlet elevation 300- 6 In. 18 In. VVV 0 0 0 0 1 1 8 5 5 200 100 + 0 Ab E2b Ab E2b Ventilation rate : 8 cfm (Ab=60 ACH, E2b=20 ACH) Cubical CS Model (Aa) vs. "Expanded" Vertical-Noncubicaf CS Model (E2a) Cubical CS Model (Ab) vs. "Add-on" Verticai-Noncubical CS Model (E2b) FIGURE 7. Oxygen recovery times, delta T (10 to 20.9%02), for the cubical and the vertical, noncubical CS models. Aa and E2a had the same nondimensiona! ventilation design parameters (ACH,l/0-%H). Ab and E2b had the same dimensional parameters (cfm, 1/0-inches) S icaru advantage in reducing ventilation time when using low I/O elevation for HT\ contaminants. Findings for the "add-on," vertical, noncubical CS model also show the importance of providing ventilation near the bottom of a CS containing HTA contaminants. Ventilation characteristics for the cube and vertical noncube were similar, suggesting that ventilation design for the "cubical'' bottom of a vertical noncube may be predicted reasonably well using cubical CS model data, with relatively little shape (depth) effects. Data in Figures 3, 4, 6, and 7 show that N2 was not a good predictor for HTA contaminant effects. Nitrogen was the only contaminant used in the previous studies of cu bical and noncubical CS models.0 Findings for nitrogen may be more useful for dilution ventilation to control toxic contaminants in low concentrations for which the mixture density is nearly the same as that of air. The relationships between CS ventilation design param eters (e.g., CS shape, contaminant SG, ventilation I/O, ACH, and iocation/elevation in CS) are very complex. Combining these with other relevant parameters (many not yet stud ied) would result in an empirical database so complicated as to defy almost any effort to establish a rational, accurate, and comprehensive computer design model based solely on empirical findings. A truly comprehensive design model for confined space ventilation probably must come from an analytical basis. This requires developing user-friendly computer programs to solve difficult mathematical models. Research should continue towards development of prac tical guidelines for effective CS ventilation. More field test ing is needed. This will enhance working experience and improve the predictive capability of laboratory experi ments, particularly when done in conjunction with scale model testing. The development of analytical models should be encouraged. Such models could be tested and evaluated against the empirical database obtained in these studies for oxygen deficiency. There are many additional and un tested design parameters, such as different CS model shapes (e.g., cylindrical), internal CS surfaces, more than one CS opening, and CS opening size variation, to list a few. Guidelines and Recommendations The conclusions from this study, and those from the two previous studies/1'21 help to establish a framework of de sign guidelines for CS ventilation which can be summa rized as follows: Air movement caused by mechanical ventilation will greatly enhance contaminant dilution inside a CS. Con taminant dilution by diffusion and/or natural ventila tion can be much slower than with mechanical ventilation. Heavier-than-air contaminants can stratify' inside con fined spaces. Rates of contaminant dilution tend to decrease as contaminant density increases, especially near the bottom of a confined space. Supply ventilation is generally more effective than ex haust in reducing the length of time required for di APPL OCCUP. ENVIRON. HY6. 8(2) . FEBRUARY 1991 lution of atmospheric contaminants because air mixing is enhanced by the supply jet. Decisions to ventilate by either supply or exhaust methods should also in clude considerations of other advantages and disad vantages (discussed subsequently). A directed supply jet of fresh air can provide very effective air mixing and dilution, such as rapid oxygen recovery, at locations aligned with the jet. It is highly advisable in many CS ventilation situations to direct fresh air to the breathing zones of workers. Low ventilation I/O elevations (e.g., less than 25 %H) are generally more effective than higher I/O elevations (e.g., above 75 %H). This characteristic can be pro nounced for HTA contaminants, increasing with con taminant density. Ventilation volume flow rate can have significant effects upon ventilation time. Reductions in ventilation time typically are not proportional to increases in volume flow rate. Benefits from exceeding 60 ACH may be minimal for supply ventilation. Flow rates below 20 ACH are not recommended for most CS situations. If geometric and air flow similarity are maintained be tween two confined spaces of different size (i.e., same CS and ventilation geometry and same ACH), then they will probably have similar ventilation characteristics. If geometric similarity is not maintained, then venti lation characteristics may be very different. The shape of a CS can have significant effects upon ventilation characteristics. HTA contaminants may cause substantial differences between characteristics of cu bical versus noncubical shapes in vertical and hori zontal orientations, A cubical CS model may be able to predict ventilation effects in the "cubical end" of a noncubical CS. Variations in ventilation effects with CS shape (e.g., cubical versus noncubical) are less pronounced when contaminant density' is not significantly different from air, such as for neutrally buoyant nitrogen and for toxic contaminants in relatively low concentrations. Findings from these studies do not address all important issues of ventilation control for confined spaces. Consid eration should be given to other important engineering and administrative control measures, to advantages and disadvantages of different ventilation alternatives, and to ventilation equipment, testing, and training. Mechanical ventilation of a confined space does not pre clude the importance of stria administrative control mea sures for safe CS entry. Both engineering and administra tive controls are needed unless either can be confirmed to be unnecessary (and this can be difficult to do). Entries should be avoided whenever possible. Confined spaces should always be opened as much as possible prior to entry. Advantages of supplyventilation for a CS can include: rapid and moveable localized dilution directly to worker breathing zones; it can cause more effective air mixing throughout the CS; and it can accomplish mancooling. Dis advantages include: enhanced dispersion/evaporation of VVV 00G011B56 139 dust.v liquids; contaminant discharge from CS openings; contaminant buildup if supply air is not fresh: and possiblv hazardous mancooling with cold ambient temperatures. Advantages of exhaust ventilation include: the tendency to be nondispersive of contaminants; contaminants can be discharged away from CS openings; localized control is possible if inlets are positioned close to localized sources; and lower general air flow velocities can reduce man cooling effects. Disadvantages of exhaust ventilation in clude: less effective general air mixing and dilution; the possibility of drawing contaminants through breathing zones; and the possibility of high localized contaminant concentrations. Local exhaust ventilation (LEV) can be used effectively in confined spaces provided the inlets (hoods) can be kept close enough to localized contaminant sources. LEV offers the possibility of air cleaning to remove contaminants. LEV may not have sufficient volume flow rate to satisfy needs for dilution ventilation if contaminants are not captured at the hood(s). Natural ventilation, typically caused by wind and/or ther mal convection, can have significant effects upon contam inant concentrations in confined spaces. Natural ventilation has the advantage of not being subject to mechanical fail ure. However, it can also be changeable without warning and should be used only with strict administrative controls. Natural ventilation should be utilized to whatever extent it may exist. However, mechanical ventilation should be required whenever highly toxic contaminants are present. Equipment limitations can be significant factors in ven tilation design for a particular CS entry. Ventilation lime is a function of the fan volume flow rate. Planning and designing for CS ventilation should be made well before an entry is needed in order to be certain that proper equipment will be available. Ventilation equipment can cause problems for egress, visibility, communication, undesired mancooling, and other aspects of CS entry safety. Testing before and during CS entries should always em phasize monitoring for atmospheric contaminants, even when effective tentilation is in place. Testing during CS entries should also include observations and measure ments of ventilation parameters to help evaluate ventilation effectiveness. Training for CS entries should include setup and op eration of all ventilation equipment for all persons in volved in routine entries and emergency rescues. Training should address ventilation testing and the calibration and operation of gas/vapor/oxygen monitoring instruments. It should emphasize "hands-on" experience. Training drills at actual confined spaces, with potentially hazardous con taminants present, are highly recommended in addition to classroom instruction. There will be some CS entries for which ventilation is impossible, such as when air must be excluded from the CS (eg., pyroforic catalysts) or when space size and ge ometry prevent effective ventilation (e.g., long, complex utility tunnels). It is advisable to provide ventilation as close as possible to work areas that cannot be ventilated directly. It is necessary in such situations to place special emphasis on administrative controls to offset the loss of safety resulting from poor or nonexistent ventilation. Poor ventilation is a defining characteristic of a confined space. There is great need to improve the awareness of potential atmospheric hazards in CS and of ventilation as a primary means of control. More specific information is needed on metfiods for implementing effective CS ventilation. Mechanical ventilation, even lacking solid comprehen sive quantitative design criteria, should be used for most confined space entires. Mechanical ventilation does what no administrative control can do (e.g., entry attendant, permit, atmospheric testing, or personal protective equip ment)--it will directly reduce concentrations of potentially hazardous air contaminants in confined spaces. References 1. Garrison, R.P.; Nabar, N.; Erig, M.: Ventilation to Eliminate Oxygen Deficiency in a Confined Space--Pan I. A Cubical Model. Appl. Ind. Hyg. 4:1-11 <1989). 2. Garrison, R.P.; Erig, M.: Ventilation to Eliminate Oxygen Deficiency in a Confined Space--Part 11: Noncubical Models. Appl. Ind. Hyg. 4:260-268 0989) Received 12/26/89: review decision 2/23/90: revision 8/3/90; accepted 9/4/90 VVl/ 0ooii857 140 APPL 0CCUP. ENVIRON. HYG. 6(21 FEBRUARY 1991 TO: Distribution Interoffice Communication FROM: DATE: SUBJ: T. G. Grumbles February 15, 1991 CMA CAER WORKSHOP vc Attached is information regarding regional CAER implementation workshops. It appears this workshop is focusing on employee and community outreach. T. G. Grumbles dlj .265 Attachment Distribution: SAFETY DIRECTORS Bruce Trego-Aber, Balt, Harry Peirce-Blane, K. L. Fogg-LCCP, R. V. Gantz-LCLAB, G. M. Shirley-LCVCM, Brent White-Okc, R. B. MartinAustin, J. R. Drumwright, Rick Quy cc: PLANT MANAGERS R. W. Seymour-Aber, L. R. Eauer-Balt, G. D. Williams-Blane, J. Pavao-Hrad, J. Friend-LCCP, J. W. Ware-LCLAB, R, A. Conrad-LCVCM, H. D. Garrison-Okc, P. L. Foote-Prem, V. W. Weiss-Austin ENVIRONMENTAL COORDINATORS F. G. Jeanson-Aber, D. L. Mahler-Balt, G. D. Williams-Blane, Matt Tonkovich-Hmd, M. G. Hayes, G. L. Foshee, J. L. Johnson, Julie Bozich, Diane Johnson-LCCP, Tony Salah-LCLAB, D. R. Booth-LCVCM, Joyce Callen-Okc, G. C. Lipps-Prem, R. B. Martin, J. R. RoheimAustin, R. B. Quy M. S. Reynolds, C. Bozman, M. A. Stephenson WV 000011858 Registration Form Plg^e register the following person(s) for the jCode.of Management Practices Workshop. A check in the amount of per person) Is enclosed. Mall registration form and check or money order, payable to the Chemical Manufacturers Association, to: Heather Feltmate, Chemical Manufacturers Association, 2501 M Street, N.W., Washington, DC 20037. Refunds will be made if cancellations are received at least two weeks before the date of the workshop. Workshop Location: Register by: .4/10/91 New Orleans, LA .4/24/91 Chicago, IL .5/14/91 Sacramento, CA 4/1/91 4/9/91 5/6/91 So we can better identify you on your meeting badge!, please register using your name as you wish It to appear on the badge. Name Title. Name___ Title____ Company Address _ City________________________ Stat* Zip Phon_J______ L Workshop Agenda 8:00-8:30 a.m., 8:30-6:45 i 8:45-9:00 9:00-10:15 10:15-10:30 10:30-12:00 noon SESSION 1: SESSION 2: SESSION 3: 12:00-1:00 Registration and Continental Breakfast Welcome and Opening Remarks James Hinton, Chairman CAER Task Group Dow Chemical U.S.A--Texas Operations Welcome from the Chemical Industry Council Community Outreach Overview Members of the Responsible Care Public Advisory Panel History and overview of the code The need for community outreach The benefits of community outreach Break Three Concurrent Break-Out Sessions: Community Advisory panels Who is involved Function of the panel How ft works Types of panels Employee Participation and Communications Stimulating interest Methods of communication Employees' role in outreach activities Outreach to Target Audiences Educators and students Local, state and federal officials Medical community Lunch 1:00-2:30 2:30-2:45 2:45-4:15 o o o o > > > Repeat Three Concurrent Workshops Break Repeat Three Workshops H( \t&! A block of rooms has been reserved at the following hotels for the CMA workshops. This block of rooms will be held until two weeks prior to the workshop. After that date, room reservations will be subject to availability. Please refer to the CAER Code Workshop when making reservations.; New Orleans, LA April 10,1991 Royal Sonesta Hotel 300 Bourbon Street New Orleans, LA 70140 (504) 5866300 Room Rate: $115/Single Chicago, IL April 24, 1991 The Westin Chicago 909 N. Michigan Ave. Chicago, IL 60611 (312) 943-7200 Room Rate: $130/lngie Sacramento, CA May 14,1991 Sacramento Hilton Inn 2200 Harvard Street Sacramento, CA 95815 (916)922-4700 Room Rate: $78/Single $85/Double V CMA IS NOT RESPONSIBLE FOR MAKING HOTEL RESERVATIONS. Responsible ore CAER Code fy^lanagement ctices Workshop {re$ch programs are becoming an fngiylmportant part of the chemical ifhe ability to deal with the public and fjc segments is becoming a more ilOSfilewill' for plant managers to have in the (9903: :J*' pie^f^jcal industry bears greater scrutiny fonmental groups and their growing the media and a more skeptical jbUc. Tljle ill and its reporting requirements iav'palspd the profiles of chemical facilities in Se community. its member companies have long community outreach. The public )rpe aware of industry operations telndu^try has to be sensitive to commuIpcerns. Under Responsible Care, the H'fty Awareness and Emergency iespgfjSe.'Or CAER, process is the vehicle for )mf|iuntcatlpg the industry's commitment to Its operations. CAER participation is Ijgajion for CMA members. They have to >tjn 6 the community. But they also should arryp for the benefits it brings. tfjg jhree CAER workshops planned this jring, participants will learn how to meet the somrriunjty awareness elements of the CAER Management Practices. Speakers and s will discuss their experiences in com ing with a number of specific audiach speaker will discuss strategy and jsjworkshop;Wlll bffnjustr Sara1 members of the public.' *M * will be beneficial for plant R coordinators and others irj communication with the )mnlun|ty. Registrafie.. Registration for the CAER Code Workshop is $175 per person, which includes breakfast, lunch and beverage breaks. Telephone registration and cancellations will be considered binding. To register, please notify CMA by close of business April 1 (New Orleans), April 9 (Chicago), May 6 (Sacramento). 3 If you have registered and are unable to " attend, please call (202) 887-1273 by dates . listed. CMA cannot make a refund for cancel lations received after those dates. To register, detach and mail the registration form along with a check or money order payable to the ,, Chemical Manufacturers Association jo: Heather Feltmate Chemical Manufacturers Association 2501 M Street, N.W. Washington, D.C. 20037 (202) 887-1273 < < < o o o o *" 'B < CD o T . ,V o M&.-r -4i'-; ' :fr recycled paper / vx 'ifc The Responsible Care CAER Cod of Management Practices Worksh ,; - i I (S CICI CICC . i_hYi . jvii-s.- yI CHEMGAtj" V4 . . rMANUR\CTUBEBS^ mUtf * i f VMfrnsmt ; ASSOCIATION . - ^ J. A Public \ TO: R. D. Gamblin Interoffice Communication FROM: DATE: SUBJECT: T. G. Grumbles February 15, 1991 PROGRESS REPORT VISTA 1. The draft Responsible Care Employee Safety and Health Code is out for review by CMA Member Companies, It has been distributed internally and comments are due back prior to April 1. 2. The initial Responsible Care Distribution Code is out for self-evaluation. This code is difficult to review because multiple departments and locations are involved in the code element areas. S&T, Manufacturing and Environmental all have ownership of the code elements. 3. A visit was made to Sun Polymers in Baton Rouge to assess their capability to process Aberdeen No. 2 Pond Resin. Sun Polymers already handles off-grade PVC materials for Vista. Sun also processes PVC from BFG, Oxy, Certainteed, GeorgiaGulf and Formosa. The plant is not run the way Vista would operate,/ but considering the low hazard of the PVC, is a reasonable risk. 4. JCL and Aberdeen Plant personnel met with MS DEQ to discuss renewal of the Aberdeen Air Emission Operating Permit. The MS Air Toxics Policy was discussed in detail and our consultant, Peter Voytek of Clement International, presented information on the appropriate vinyl chloride unit risk factor. DEQ will contact EPA regarding the unit risk factor and get back to us. 5. JCL and Dave Penney of R&D met with EPA Office of Toxic Substances regarding a potential SARA 313 delisting petition for di-n-octylphthalate. Our alcohol customer, Aristech, also attended the meeting. The intent of the meeting was to get an indication from EPA of our chances of successfully delisting DNOP. At this time, chances seem slim, due to potential objections from the EPA Office of Water. A meeting with OW is being arranged by our outside counsel. \\ cv-------------- T. G. Grumbles /Ijw / wv 000011861