Document 4ay9KgoD8E6JqZ8pEyw1qzazV

PRELIMINARY CORPORATION ' ' " * 1 y- " ' ' ' >- 'i" 'ir ``f : , " :'r' - '* *>> ' jSSstfV"# DETERMININam PLANT EHVIROW^^^p^^ .j .; = AND EMPLOYEE EXPO^EjfS WITH -THE* E0Cp-:ipW^Snp5TEI^^;S^^ x; Wfe's- ' -.r-:V. V%Uv:.;. ^ ; . ' ' * : - . ' ,. -J > " \ : f;\ '..*3 ,. ` . " . 'v ' p.?*#%&*-*** & ;-X; ;J v, ' .* tf *Ji '|.j,i V i^r>v ' ;..u v>*^' ;-; ->.'.*v >v f .> ^^1^; ;fr-.y -;`-?'i'-s ; August 21 1975 -' "J "~ T` ,, '"i-'ls'ii = ' - J ^ - V ,'.v >`. . Ki . ......... - ^ V'.*, 1 ", : *; *+ ' ,. ,sj4", -'"/JiV^>4H ' - '' ,.,' /.2iT^rJ:*; "`- J*V*,j"Ww V - -^T -.'. -? - ' - '*' ' -'7 ' ';. ? -v _>- >' ry- ,r `1 ` - ' &.'' ;`: y '*%m****-'.,^1`C-1''' j.jiijt'i-' \ -*^ `.'J' ' x ' 1 t ", ',J4` v/<. ?4 '--<f-f `-w,i*.wx.--ySj-.w ____ :/-......^ ::- ucc 034144 PRELIMINARY DETERMINING THE PUNT ENVIRONMENT AND EMPLOYEE EXPOSURES WITH THE EOCOM IDAAC SYSTEM August 21, 1975 EOCOM Corporation PROPRIETARY INFORMATION EOCOM CORP. Not t. . dupKct.*. ' " "pirt ,or *"y than .valua-lon. Thia -Met*. * '"* "" ______ u 1. i. .hl.lnad from inOttMT OOiea without roatliCtlOn- e ucc 034145 Section I II III IV V TABLE OF CONTENTS Description Introduction Nature of Gas in the Work Environment Sampling Algorithm Basic Reporting of theIDAAC System on the Plant Environment Employee or Job Category TWA PROPRIETARY INFORMATION EOCOM CORP. No! to bft duplicatad. usad, or difdotad in wtiola or in part for any purpoaa otltar than ava-'ua ion. Thi* roatriction does not limit tna ri*nt to naa Information containao ' ` -- uu without rottrletion. i ucc 034146 EOCOM COaFOAATiOlW Number A B C 1 2 TABLE OF FIGURES Description Small Number of Samples Per Eight (8) Hours Many Samples Per Eight (8) Hours Trend Response Log Normal and Normal Distributions With the Same Arithmetic Mean and Standard Deviation Multi-Layer Sampling Page 9 9 10 17 18 PROPRIETARY INFORMATION EOCOM CORP. Not to bo duplicated, mad, or diicloiod in whole or in part for any purpose other than evaluation. This restriction does not limit the right to use information contained herein If it Is obtained from Another source without restriction. ucc 034147 e EOCOM Determining the Plant Environment and Employee Exposures with the EOCOM IDAAC System I Introduction Over the past eighteen (18) months EOCOM has very carefully studied the concept of area monitors in conjunction with the following ob jectives: Fast Excursion Detection Accurate Employee TWA's Eliminate Requirements for Personal Sampling Management Information System to Aid in Achieving OSHA "Perfor mance" Standards After a thorough study, it was decided to embark on a measurement program to find out the in-plant variables involved in trying to achieve the objectives. In order to do this we then proceeded to: Quantify the Nature of Gases in the Work Environment 9 Determine Appropriate Sampling Sequence for an Area Monitoring System Relate the Employees to the results of the Area Monitor and de termine Their Exposure II Nature of Gas in the Work Environment In order to understand the work environment and how this relates to employee exposure, we have taken literally millions of samplespoint data from our various field installations and analyzed the data for the statistical nature of the gases in the breathing zone. The pur poses of this effort was to establish the distribution of the gas concentration levels. This obviously is effected by the air flow. UCC 034148 PROPRIETARY INFORMATION EOCOM CORP. 1- - Not to b duplicated, used, or disclosad in whole or in part for any purpose other than evaluation. This restriction docs not !.mn the right to use information contained herein if it is obtained iron; another source without restriction. EOCOM molecular weight of the gas, reactivity of the gas in the atmosphere, type of source of the gas, etc. However, from these studies we have sufficient data to say that a "log normal" distribution is sufficient to represent the distribution of the gases studied in the various plants studied. To our knowledge this is the first in-plant data to confirm the NIOSH report of April, 1975 entitled "Statistical Methods for the Determination of Non-compliance with Occupational Health Standards", by Nelson A. Lei del and Kenneth H. Busch, HEW Publication No. (NIOSH) 75-159 (Reference 1). In this report they use air pollution data to support their "log normal" distribution methods. Now with our confirmation of this data (to be published in the latter half of 1975) (Reference 2), it is possible to use many_ of their analytical, techniques and statistics. These can be used to analyze data from an area monitor to determine compliance (at various confidence levels), to predict plant performance based on the "log normal" statistical model and to easily analyze current plant perfor mance against past plant performance for detection of significant trends. More specifically the attached Figure 1 shows the difference between the ''log normal" distribution and the more comnonly used normal dis tribution. To define the "log normal" distribution it is only neces sary to determine the Geometric Mean (GM) and the Geometric Standard Deviation (GSD) (similar to the Arithmetic Mean (y) and Standard Deviation (ej for the normal distribution). GSD * e [e |^+ 1] PROPRIETARY INFORMATION EOCOM CORP. _2_ Not to be duplicated, used, or disclosed in whole or In part for any purpose other than evaluation. This restnctien dees net limit therifht to use information contained herein it it is obtained from another source without restriction. ucc 034149 e ECOOACOPO*TIMQN EOCOM CORP. Not to be duplicated, wed. or disclosed In whole or In pert tor any purpose other than evaluation. This restriction doaa not lim it th e rlfh t to use Information contained {ir^ln If It Is obtained from another source without restriction. The 6M and GSD (for any time period long with respect to plant varia tions), completely defines, from a statistical viewpoint, the "log normal" curve for the frequency of concentration of various levels of the gas being monitored. That is to say that you now have a definition-of your plant environ ment that allows projection of the probabilities of various occur rences in the plant as well as to determine compliance with various OSHA "Performance" Standards. Thus, the EOCOM data system, as appro priate, calculated these parameters as well as TWA's. Ill Sampling Algorithm Given that the EOCOM data system keeps track of the GM and GSD of each breathing zone sensor probe* (of the area monitor), it is then possible to evolve a sampling algorithm that will have: Higher Probability of Excursion Detection Turn Warning Systems Off in the Shortest Time to Minimize the Use of Masks Isolate the Actual Probe Nearest the Cause for the Excursion A. Probe Data First using just the sample probe data (employee time/motion studies will be considered later), management selects a parameter(s) NOTE: For this discussion, no consideration is given to the number of samples from a probe that yield a useful GM and GSD. Also the effect of high instrument noise (not "log normal" distributed) relative to the variations in plant levels is not considered in this discussion. Both these parameters are a function of the area monitor used with the data system and should be evaluated on an indivi dual case basis and could effect the distribution or time inter vals used with the statistics. -3- UCC 034150 (P [M3) of the gas that is key to their plant operation under the OSHA Standard. For instance in the case of the VCM Standard, management might select individual probe readings of greater than 25ppm until April 1, 1976, and 5ppm after that time with the objective of having the correct mask on at all the required times. On the otherhand, depending on the general levels of VCM in the plant, management might select eight (8) hour averages of the probes that are'higher than 0.5ppm and l.Oppm as two (2) parameters to determine the sampling algorithm. Another para meter that could be selected is the GSD of a probe. In any case the selection of the desired parameter (one or more) is made by management as a function of plant conditions relative to the standard as well as any other requirements they may have. For instance, let's take the case of selecting the probability of the_ sample probe having a reading of 5ppm or greater P [> 5ppm/reading]. Using the GM and GSD for each probe, the axis plural for the "log normal" distribution of Figure 1 is determined and a set of probabilities, one for each probe, is determined. P [M], P2 [M]...........Pn [M] These probabilities of probe readings at or above 5ppm are then used to determine those probes that should be sampled more often (i.e. higher P [Mj) and from these probabilities is developed the sampling pattern where the sample probe selection is under the control of the EOCOM IDAAC System. The result being that those points with most likelihood of a 5ppm excursion are monitored more often. B. Employee Data A second level of effort is to also include as part of the sampling probability model, the probability of any employee being at any one PROPRIETARY INFORMATION EOCOM CORP. herein It it is attuned train inathtr source without restriction. UCC 034151 EOCOM breathing zone probe. This data is derived from time/motion studies of the involved employees (or job categories). From the time/motion study, one obtains the probability of each employee (category) with respect to probes. P [TA^. P [Ta2]......... P [TAn] P TZl], P CTz23......... P CTZn] By combining the probabilities of all the employees (categories) with the P [M] for each point, you then have the probability for that probe that any employee may be exposed at that probe. P [E-, J > P [E2J.........P [En] These probabilities are then used to generate a sampling distri bution that is used in the sampling algorithm of the EOCOM IDAAC System. t. Search Sampling In addition to the above statistically controlled sampling, the * IDAAC System also goes into a search routine when an initial alarm or excursion occurs. This search is performed.to be sure that the highest point of concentration is located so that effective corrective action can be taken as soon as possible. This search routine takes into account the wind direction (normally known because of forced ventilation) as well as the probabilities of the points around the initial excursion. After establishing all of the alarm conditions in the area, the IDAAC System then proceeds PROPRIETARY INFORMATION EOCOM CORP. _5_ Not to be duplicated, used, or disclosed in whole or in part for any purpose other than evaluaiien. This restriction dees net limit the right to use information contained herein If it is obtained from another sourco without restriction. e EOCOM COMOAATlON ucc 034152 to monitor the rest of the system while establishing a higher priority for reading the alarmed area more often until the condition is corrected. Again all of the TWA (P)'s are calcu lated correctly by the trapezoidal integration. D. Multi-Level Sampling The IDAAC System also has the capability of multi-level samp ling. The multi-level sampling concept is to monitor the mixture of several points simultaneously. That is to take for instance, 3 points and mix them in the gas manifold. Thus the measurement of the three (3) points would provide the average of the three (3) points (1, 2, and 3). In order to make a decision on a 5ppm-' alarm, the IDAAC analyzes the data as shown in Figure 2. Multi-level sampling has the advantage of "sweeping" the plant at up to three (3) times faster than single point sampling. It has certain requirements to be effective and still not deterior ate the TWA and other calculations being performed by the IDAAC System. 1, In order to almost achieve the factor of 3 speed improvement the average levels in the plant (for the case of Figure 2) would have to be below 5/3 or 1.66ppm. In no case is the approach slower than single station sampling. The worst case is when all levels in the plant are 5ppm, then the system will measure R, R-j and Rg, or three (3) measurements. Most cases we have observed are such that almost the full fac tor of three (3) can be achieved, 2. One has to take care in selecting the points that are to be averaged so as not to effect the other calculations.- We PROPRIETARY INFORMATION EOCOM CORP. haioin If it i obtain** from another *ourc* without restriction. EOCOM ucc 034153 have shown that this is accomplished by selecting points with roughly equal P [E] and/or P [M] 3. Although the IDAAC System has the multi-level sampling capa bility, it may require different gas manifolding than present ly in use. In summary, the multi-level sampling capability does improve: * Probability of Measuring Excursions Average Time To Survey the Plant E, Summary of Sampling In summary, the sampling algorithm of the IDAAC System allows the user to; Have a system that responds to its environment. Each day the GM and GSD are updated and the sampling algorithm cor rected for plant trends and variations Sample the area monitor in an intelligent fashion based on the "living" statistical trends of the plant environment Select the sampling decision parameters that are deemed important to management Enter new or changed time/motion studies Get to excursions significantly faster than a sequential sample system Correct excursions fast PROPRIETARY INFORMATION EOCOM CORP. Not to be duplicated, used, or disclosed in whole or in part for any purpose other t*ten evaluation. This restriction does not limit therifht to use Information contained herein if it is obtained from another source without restriction. -7- e EOCOM CSn.QM.TlON ucc 034154 IV Basic Reporting of the IDAAC System on the Plant Environment In order to provide management with significant information concerning the plant environment, its changes and the degree of compliance with a standard; the IDAAC system reports several parameters regarding the plant data measured by the probes. First, individual probe readings are available through the printer or by optional mass storage device. However, except for investigating specific events, it is felt that the following summary parameters are sufficient to fully define the environment for each probe *> S "8 is, hi TWA (P) Trend TWA: TWA (TP) 0 in 1ttQ. sf 5 i 2PitUSo f! ? --*!io "S S5 S s! S 8I & 3ko ^e5 TSe, CoL I'll jf= f *f I-s F4l1 *i * GM (P) Trend GM: GM (TP) GSD (P) Trend GSD: GSD (TP) The total plant (floor or area) environment is defined by the sum of the probe data for the plant (floor or area). A) TWA (P): TWA (P) is calculated from the basic data points measured by the sensor probes. It is important to note that the non-uniform sampling (with time caused by the sampling theorem in section III above) requires careful attention to the means of calculating the TWA. The IDAAC system calculates the TWA with a trapezoidal integration of. the individual readings, yielding the correct value for the area under the curve, which then for any given time period (15 minutes, 1 hour, 4 hours, 8 hours, etc.) provides the correct TWA (P). The area under the curve is shown in Figure A. It is this true area divided by the time that yields TWA values. It is interesting to note that most management selection criteria for.P [M] will lead to an Note: The IDAAC System calculates many other parameters as noted in its specific specification sheet. The parameters discussed here are those.'; used to describe the plant environment and employee exposure. -8- UCC 034155 optimum sampling of the eight (8) hour period. That is to say that most of the selection criteria used by management will lead the sampling algorithm to sample active exposure areas more often than less active exposure areas. This is shown by the following conditions: A) Small number of samples per eight (8) hours: * Low P (M) Low P (E) B) Many samples per eight (8) hours: High P (M) Not to be duplicated, used, or disclosed in whole or In part for any purpoaa ottiar Sen evaluation. Thii restriction does not limit the right to use information contained -9~ herein If it i* obtained from another source without restriction. e EOCOM ucc 034156 In both cases the system samples the data correctly to provide the correct time weighted TWA (P). C) Trend Responsive TWA (TP): In order to compare today's performance with past performance the IDAAC System outputs an exponential TWA which is: TWA (TP)o - O.X TWA (P)Q + ( 1 - O.X) TWA (TP)y Where the subscript 0 indicates today's data and the subscript Y indi cates yesterday's data. The factor X determines the number of days that are used to determine the trend TWA (TP). The factor X is selectable by the user and can be varied to give very long term trends or shorter term (such as a week) trends. The effect of setting X to a time constant of seven (7) to ten (10) days is to smooth out any abnormalities of * a given day, while at the same time giving data on the most recent performance of the plant. This is depicted in the following figure: PROPRIETARY INFORMATION EOCOM CORP. _10- Not to t* duplicated, used, or diactoced in whole or in part for any purpoaa othar ttan cvaioaiion. This leiirictton doea net limit therifht to u*e information contained nerein i* it ii obtained from another eource without restriction. e EOCOM CdMPOAATIQM UCC 034157 By this method you can readily compare today's performance TWA (TP) with the average of a previous (selectable) set of days and decide if today's performance is significantly different. An alternate path is to simply calculate the TWA for extended periods of time (rather than exponentially). This is possible with the system, however, it is our opinion that the most recent data contains more management significance, however, either is available from the IDAAC System. 0) GM and GSO: The significance of GM and GSD is that they allow one to completely describe the plant in terms of statistical numbers. This uniquely de fines the log normal curve in Figure 1 from HEW Document No. (NIOSH) 75-159, (Reference 1). Again' it is significant to compare the present to the most recent plant data. By using this data, you can easily assess the plant performance for the day both in.actual levels and in variations. Since it has been shown that the data is log-normal distributed it is also possible to use the GSD and GM from a series of days to project the plant environment into the next shift or day with a confidence level attached to projected performance on a future shift or day. Again, the continuing GSD and GM are reported as trends also. Thus the system reports: GM (P), GM (TP), GSD (Pj and GSD (TP) in the same fashion as the TWA (P) and TWA'(TP) values. Further, using the GM and GSD allows one to assess whether any given shift or period is within the previous experience and whether a com bination of shifts represents a new trend. Depending on the process and results from the system it may be more appropriate to keep separate records on the shifts in each day rather than running the average from shift to shift through the day into the next. In certain industrial cases we have noted a substantial difference in the value of GM and GSD from shift to shift. In one case, the night shift runs significantly higher than the other two shifts. PROPRIETARY INFORMATION EOCOM CORP. -11- Not to be duplicated, used, or disclosed in whole or In port for ony purpose other tpen evaluation. This restriction does net limit the right to use information contained herein if it is oOUined from another source without restriction. EOCOM ucc 034158 It is worthwhile to comment on the effect of the sampling algorithm on the calculation of GM and GSD for the eight (8) hour periods. The actual log normal distributions are based on random samples. The samples taken (at least for a short period) cannot be considered random, however, our data to date does show a substantially log normal dis tribution for periods as short as one hour. PROPRIETARY INFORMATION EOCOM CORP. Not to be duplicated, used, or disclosed In whole or In port far any pufpaaa other than evaluation. Thia reetrtctlon doaa not hndt the rifht to use Information contained herein If It la obtained from another aource without restriction. 12 EOCOM ucc 034159 V Employee or Job Category TWA The IDAAC System calculates employee (or job category) exposure by using the measured TWA (P) for the time period of interest for each point and the time/motion study data (P [T^]...........P [T^]) This approach leads to extremely good long term TWA of employees. Certainly it is based on more information than a 1 or 2 time/month personal sample. The time that the IDAAC System takes to converge on accurate TWA (when compared to charcoal tubes) is a function of many variables, such as: Conditional exposures Validity of time/motion study and its variability Placing of the sensors in the breathing zone Number of samples per measurement period provided by the area monitor Cyclic (or non-cyclic) nature of the process etc. It has been our experience to date with an installed VCM monitor system that with the inclusion of the conditional exposure (man in reactor) that calculated TWA values converge to the measured TWA values in less than eight (8) hours. For those situations where the employee may perform several job categories within a shift or from shift to shift, the job category concept should be used and the employee should log in on a per shift basis the time spent in each job category that is a regulated job category. PROPRIETARY INFORMATION EOCOM CORP. Not to bo duplicated, used, or disclosed in whole or In part tor nny purpose other tben evaluation. This restriction does not limit the right to use information contained herein if it is obtained trom another source without restriction. -13- EOCOM ucc 034160 EPA Requirements With the present proposed standards for EPA monitoring of plant effluents for substances such as VCM (EPA 40 CFR Part 61: National Emission Standards for Hazardous Air Pollutants Proposed Standard for Vinyl Chloride) E0C0M intends to offer an optional accessory to IDAAC System to perform either a mass calculation of VCM in the air in the plant and that VCM that escapes the plant into the environment or to accept data from an exhaust sensor (one or more) and provide the appropriate data analysis, such as lOppm per hour limit. In view of the status of this proposed EPA VCM standard we have not chosen the final approach.. PROPRIETARY INFORMATION EOCOM CORP. Not to bo duplicated, uiod, or dlieloood In who!* or In port for any purpoaa tpan ovalua.ion. Thl* i**trtetien doo* not limit thartftit to wo Information *nt*Jn*0 haraln If It h obtained from another aoum* without roatrlctlon. -14- e cEonOpoCBAOftOMM UCC 034161 Effect on Employees A true interactive intelligent monitor, as installed in a VCM monitoring situation, has been observed to have the following effects: 1. Excursions due to work practice have significantly decreased through to the speed of the monitor and its ability to capture 94% of the excursions. This has, without intervention of manage ment, brought a force to bear on the employees which had the effect of dramatically reducing excursions due to work practices. 2. A previous monitor system was circumvented by the employees knowing the sequential nature of the sampling device. Knowing this they were able to perform critical operations that might cause excursions while the monitor is not sampling in the area. With installation of the new system and its sampling algorithm this was no longer possible and a significant change in work practices occur, which also improved product!vety. PROPRIETARY INFORMATION EOCOM CORP. Net to be duplicated, used. or disclosed in whole or In pert for any purpose other than evaluation. This restriction does not limit the right to use information contained herein if it is obtained from another source without restriction. -15- e EOCOM ucc 034162 Nomenclature GM : Geometric Mean GSD : Geometric Standard Deviation P [M] : Management Selected parameter for determining sampling algorithm (one or more can be selected) P CTAl3 : Probability of man A being at point 1, etc. P [E] : Probability of exposing any employee at a sensor probe TWA : Time weighted average as defined in 29 CFR 1910.93 (d) (1) TWA (C) : Measured eight (8) hour time weighted average of an employee or job category TWA (P) : Eight (8) hour time weighted average of a sensor probe TWA (TP) : Exponential average of eight (8) hour time weighted average of a sensor probe GM (P) % : Eight (8) hour geometric mean of a sensor probe GM (TP) : Exponential average of eight (8) hour geometric mean of a of a sensor probe GSD (P) : Eight (8) hour GSD of a.sensor probe GSD (TP) : Exponential average of an eight (8) hour GSD of a sensor probe .A Subscripts 1 , 2, 3, n. . . : Probe numbers PROPRIETARY INFORMATION EOCOM CORP. Not to be duplicated, used, or disctowd in whole or to pan tor any puiposo other than evaluation. This restriction deaa net limit the right to uao Information contained harain If It is Obtained from another source without restriction. A, B, C...........: Employee or Job Category -16- EOCOM CQAFOH4tlON ucc 034163 x = ARITHMETIC MEAN 53M x 10 15 CONCENTRATION ppm Figure 1 - Log normal and normal distributions with the same arithmetic mean and standard deviation. . . FROM: HEW Document No. (NIOSH) 75-159 "Statistical Methods for the Determination of Noncompliance Occupational Health Standards" -17- UCC 034164 Situation 1. P [M] 5ppm 2. Simplified version of program. Note real program takes into account recent measurements of points and probability of most likely point of excursion. Reading = R (Average of Points 1, 2 and 3) Is R X 3 < 5 Yes Proceed with sampling is No Measure Pt 1 ' Is R] < 5 * <5 Is 3|-~ <5 Proceed With Sampling Measure R2 I Now have enough data.to determine R"l> R2 R3 Proceed With Sampling Measure Rg 1 Now have enough data to determine Rl r2* R3 Multi-Layer Sampling Figure 2 PROPRIETARY INFORMATION EOCOM CORP. Not to be duplicated, used, or di*elo**d in whole or in part for any pnrpee taan evaluation. Thi* restriction doe* not limit the right to u* information harein If it i* aktained from another source without roatriction. EOCOM UCC 034165 References Reference One HEW Publication No. (NIOSH) 75-159 "Statistical Methods for the Determination of Noncompliance with Occupational Health Standards" Reference Two "The Use of Area Monitors for Determination of Compliance with OSHA "Performance" Standards" by Joe Coppola, David Mattson, Ken Lindelin and Dr. S. Thomas Dunn (To be published) PROPRIETARY INFORMATION EOCOM CORP. Net to be duplicated* used, or disclosed in whole or In part for any purpose other than eva:ua.<on. This restriction does not limit the right to use into##nation contained herein H it is obtained from another source without restriction. -19- e edcom C.oe*Qn*TiON ucc 034166 O- - ,> \ 5--"` .. ': ir* ` ' y [ '*? oi .I Of O- ! ' , -a , .--j- .r' ;, EQCOM CORPORATtON/lS7Ba JAMBOREE BLVO-, IRVINE, CALIFORNIA BSBB4/TELEPHONS (714) B33~87B1- ' ucc 034167