600281111
                  es                EPA-
              Environmental Protection
              Agency                  June 1981
4>EPA       Research  and
              Development
              ANALYSIS OF SOCMI


              VOC FUGITIVE


              EMISSIONS DATA
              Prepared for
              Office of Air Quality Planning and Standards
              Prepared by

              Industrial Environmental Research
              Laboratory
              Research Triangle Park NC 27711
                                          Mt-II

-------
                 RESEARCH REPORTING SERIES


Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination  of  traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:

    1. Environmental Health Effects Research

    2. Environmental Protection Technology

    3. Ecological Research

    4. Environmental Monitoring

    5. Socioeconomic Environmental Studies

    6. Scientific and Technical Assessment Reports (STAR)

    7. Interagency Energy-Environment Research and Development

    8. "Special" Reports

    9. Miscellaneous Reports

This report has been assigned to the  ENVIRONMENTAL PROTECTION  TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from  point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
                       EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents necessarily
reflect the views and policy of the Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.

This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                             EPA-600/2-81-111

                             June 1981
              ANALYSIS OF SOCMI

           VOC FUGITIVE EMISSIONS
                    DATA
                Final Report


                Prepared by:

                G,  J. Langley
                S.  M. Dennis
                L.  P. Provost
                 J. F. Ward

             Radian Corporation
               P. 0. Box 9948
            Austin, Texas 78766


         Contract No. 68-02-3171-28


             EPA Project Officer
             Dr. B. A. Tichenor


          Chemical  Processes Branch
                 EPA/IERL
     Research Triangle Park, N,C.   27711

                Prepared for:

Office of Air Quality Planning and Standards

-------

-------
                                  CONTENTS


Figures	  .   iii

Tables	viii


   1.  Introduction 	     1

   2.  Summaries and Conclusions	     4

            Relationship of Leak Frequency to Process Parameters
            (Section 3)	     4
            Emission Factor Development (Section 4)  	     6
            Increase in Mass Emissions Due to Occurrence and
            Recurrence (Section 5)	  ...     6
            Impact of Response Adjustments on Leak Frequency
            Estimation (Section 6)	     9

   3.  Detailed Results for the Effects of Process Parameters on
       Leak Frequency	    12

            Overview of Screening Data from 24 Chemical Units	    13
            Effect of Chemical Produced on Leak Frequencies 	    16
            Effect on Leak Frequency of Primary Chemical in the
            Process Line	    22
            Effect of Type of Valve on Leak Frequency	  .    31
            Leak Frequency for Pump Seal Classification	    43
            The Effect of Line Temperature and Line Pressure	    48
            Effects of Line Temperature and Line Pressure on Pump
            Seals, Flanges, and Open-Ended Lines	    61
            Effect of Ambient Temperature on Leak Frequency 	  ,    69
            Effect of Elevation on Leak Frequency	    71

   4.  Emission Factor Development for Three Processes	    73

            Distribution of Screening Values	    73
            Emission Factors and Cumulative Distributions of Total
            Emissions by Screening Values 	    74

   5.  Evaluation of the Effects of'Leak Occurrence, Recurrence, and
       Repair on Mass Emissions	117

            Effect of Leak Occurrence on Mass Emissions	117
            Effect of Leak Recurrence on Mass Emissions for
            Valves	119
            Further Analysis of Effect of Valve Maintenance on
            Mass Emissions	121

-------
                            CONTENTS (CONTINUED)
   6.  Impact on Leak Frequency Estimates of Applying Chemical
       Response Adjustments 	  125

            Summary of Four Leak Frequency Estimates by Primary
            Chemical	126
            Summary of Four Leak Frequency Estimates by Process
            Type	126

   7.  Statistical Considerations 	  138

            Statistical Categorical Analysis Using Funcat
            (Section 3)	138
            Chi-Square Test for Independence (Section 3)	  139
            Confidence Intervals for Percent Sources Leaking
            (Section 3)	140
            Screening Value Distributions (Section 4) 	  141
            Emission Factor Development (Section 4) 	  145
            Cumulative Emission Functions 	  151
            Increase in Mass Emissions Due to Occurrence and
            Recurrence (Section 5)	  155
            Response Model Adjustments to Screening Values
            (Section 6)	156

References	160

Appendices

   A.  Screening Data Summary	161
   B.  Detailed Information on Line Temperature and Line
       Pressure	168
   C.  Summary Statistics and Detailed Information on the Effects
       of Ambient Temperature and Elevation on Leak Frequency	190
   D.  Corrections to Screening Data	198

-------
                                   FIGURES


Number                                                                   Page^

 3-1     Effect of Process Type on Percent of Valves Leaking	    19

 3-2     Categories of Sources for Further Analysis  	    27

 3-3a    Percent Leaking with 95 Percent Confidence Intervals
           for Each Valve Type with Gas Service	    36

 3-3b    Percent Leaking with 95 Percent Confidence Intervals
           for Each Type of Valve with Light Liquid Service	    37

 3-4a    Percent Leaking for Block Versus Control Valves in
           Gas Service by Primary Material Group and Process
           Unit Type	    40

 3-5     Percent Leaking with 95 Percent Confidence Intervals
           for Each Type of Pump Seal with Light Liquid Service  ....    47

 3-6     Combined Effects of Line Temperature and Line Pressure
           on Percent Leaking for Valves in Gas Service Within
           Ethylene Process Units  	    55

 3-7     Combined Effects of Line Temperature and Line Pressure
           on Percent Leaking for Valves from Group 5	    56

 3-8     Combined Effects of Line Pressure on Percent Leaking
           for Valves in Group 3	    57

 3-9     The Effect of Line Pressure on Percent Leaking with
           95 Percent Confidence Intervals for Valves from
           Group 4 and Group 8 . »	    58

 3-10    The Effect of Line Pressure on Percent Leaking with 95
           Percent Confidence Intervals on Pump Seals in Light
           Liquid Service  	    64

 4-1     Typical Distribution of Loge (OVA Screening Value)
           Ehtylene Process, Valves in Gas Service 	    76

 4-2     Cumulative Distribution of Sources by Screening Values -
           Cumene Process, Valves in Gas Service 	    77
                                     111

-------
Number                                                                  Page

 4-3     Cumulative Distribution of Sources  by Screening Values -
           Cumene Process,  Valves in Light Liquid  Service  	    78

 4-4     Cumulative Distribution of Sources  by Screening Values -
           Cumene Process,  Pumps in Light  Liquid Service   	    79

 4-5     Cumulative Distribution of Sources  by Screening Values -
           Ethylene Process,  Valves in Gas Service  	    80

 4-6     Cumulative Distribution of Sources  by Screening Values -
           Ethylene Process,  Valves in Light Liquid  Service  	    81

 4-7     Cumulative Distribution of Sources  by Screening Values -
           Ethylene Process,  Pumps in Light  Liquid Service  	    82

 4-8     Cumulative Distribution of Sources  by Screening Values -
           Vinyl Acetate Process, Valves in  Gas Service 	    83

 4-9     Cumulative Distribution of Sources  by Screening Values -
           Vinyl Acetate Process,Valves in Light Liquid Service ....    84

 4-10    Cumulative Distribution of Sources  by Screening Values -
           Vinyl Acetate Process, Pumps in Light Liquid Service ....    85

 4-11    Emission Factors—Valves	" .    87

 4-12    Emission Factors—Pump  Seals 	    88

 4-13    Cumulative Distribution of Total  Emissions  by Screening
           Values - Cumene  Process,  Valves in Gas  Service  	    89

 4-14    Cumulative Distribution of Total  Emissions  by Screening
           Values - Cumene  Process,  Valves in Light  Liquid Service   .  .    90

 4-15    Cumulative Distribution of Total  Emissions  by Screening
           Values - Cumene  Process,  Pumps  in Light Liquid  Service  ...    91

 4-16    Cumulative Distribution of Total  Emissions  by Screening
           Values - Ethylene  Process, Valves in Gas  Service  	    92

 4-17    Cumulative Distribution of Total  Emissions  by Screening
           Values - Ethylene  Process, Valves in Light Liquid Service  .    93

 4-18    Cumulative Distribution of Total  Emissions  by Screening
           Values - Ethylene  Process, Pumps  in Light Liquid  Service  .  .    94

 4-19   . Cumulative Distribution of Total  Emissions  by Screening
           Values - Vinyl Acetate Process, Valves  with Light
           Liquid Service 	    95
                                     IV

-------
Number                                                                  l?age

 4-20    Cumulative Distribution of Total Emissions by Screening
           Values - Vinyl Acetate Process,  Valves in Light
           Liquid Service	96

 4-21    Cumulative Distribution of Total Emissions by Screening
           Values - Vinyl Acetate Process,  Pumps in Light
           Liquid Service 	    97

 4-22a   Cumulative Distribution of Sources by Screening Values -
           Cumene Process, Valves in Gas Service  	    98

 4-22b   Cumulative Distribution of Total Emissions by Screening
           Values - Cumene Process, Valves in Gas. Service	    99

 4-23a   Cumulative Distribution of Sources by Screening Values -
           Cumene Process, Valves in Light Liquid Service  	    100

 4-23b   Cumulative Distribution of Total Emissions by Screening
           Values - Cumene Process, Valves in Light Liquid  Service  .  .    101

 4-24a   Cumulative Distribution of Sources by Screening Values -
           Cumene Process, Pumps in Light Liquid Service  	    102

 4-24b   Cumulative Distribution of Total Emissions by Screening
           Values - Cumene Process, Pumps in Light Liquid  Service  .  .  .    103

 4-25-a   Cumulative Distribution of Sources by Screening Values -
           Ethylene Process, Valves in Gas Service  	    104

 4-25b   Cumulative Distribution of Total Emissions by Screening
           Values - Ethylene Process, Valves in Gas Service 	    105

 4-26a   Cumulative Distribution of Sources by Screening Values -
           Ethylene Process, Valves in Light Liquid Service 	    106

 4-26b   Cumulative Distribution of Total Emissions by Screening
           Values - Ethylene Process, Valves in Light Liquid Service  .    107

 4-27a   Cumulative Distribution of Sources by Screening Values -
           Ethylene Process, Pumps in Light Liquid Service	    108

 4-27b   Cumulative Distribution of Total Emissions by Screening
           Values - Ethylene Process, Pumps in Light Liquid Service  .  .    109

 4-28a   Cumulative Distribution of Sources by Screening Values -
         Vinyl Acetate Process,  Valves in Gas Service	    110
                                      v

-------
Number                                                                  Page

 4-28b   Cumulative Distribution of  Total Emissions by  Screening
           Values - Vinyl Acetate Process,  Valves  in Gas  Service  ....  Ill

 4-29a   Cumulative Distribution of  Sources by Screening  Values -
           Vinyl Acetate Process,  Valves  in Light  Liquid  	  112

 4-29b   Cumulative Distribution of  Total Emissions by  Screening
           Values - Vinyl Acetate Process,  Valves  in Light Liquid
           Service	  .  113

 4-30a   Cumulative Distribution of  Sources by Screening  Values -
           Vinyl Acetate Process,  Pumps in Light Liquid Service   .  .  ,  .  114

 4-30b   Cumulative Distribution of  Total Emissions by  Screening
           Values - Vinyl Acetate Process,  Pumps in Light Liquid
           Service	  115

 5-1     Before Minus After  Maintenance Leak Rate  - Valves
           Screening <10,000 ppmv After Maintenance	122

 5-2     Before Minus After  Maintenance Leak Rate  - Valves
           Screening ^10,000 ppmv After Maintenance	123

 6-1     OVA Reading vs.  Method 1 Adjustment for Cumene Process
           Valves in Gas Service	132

 6-2     OVA Reading vs.  Method 1 Adjustment for Cumene Process
           Valves in Light Liquid Service  	  133

 6-3     OVA Reading vs.  Method 1 Adjustment for Ethylene Process
           Valves in Gas Service	134

 6-4     OVA Reading vs.  Method 1 Adjustment for Ethylene Process
           Valves in Light Liquid Service  	  135

 6-5     OVA Reading vs.  Method 1 Adjustment for Vinyl  Acetate
           Process Valves in .Gas Service	136

 6-6     OVA Reading vs.  Method 1 Adjustment for Vinyl  Acetate
           Process Valves in Light Liquid Service   	  137

 B-l     Distribution of Sources Screened by Line  Pressure for
           Ethylene and High Leaking Process Units by Chemical
           Group for Valves  with Gas Service	186

 B-2     Distribution of Sources Screened by Line  Temperature for
           Ethylene and High Leaking Process Units by Chemical Group  for
           Valves with Gas Service	187
                                     VI

-------
jjumber                                                                   Page

 B-3     Distribution of Sources Screened by Line Pressure for
           Ethylene and High Leaking Process Units by Chemical
           Group for Valves with Light Liquid Service	188

 B-4     Distribution of Sources Screened by Line Temperature
           for Ethylene and High Leaking Process Units by
           Chemical Group for Valves with Light Liquid Service 	   189

-------
                                   TABLES
Number
  2-1   Estimated Emission Factors for Nonmethane Hydrocarbons
          from Valves and Pump Seals (Ibs./hr./source and kgs./
          hr. /source)	      7

  2-2   Summary of Percent of Sources Distribution Curves and Percent
          of Mass Emissions Curves at Screening Value of 10,000.
          pprav r . .  .	      8

  2-3   Comparable Estimates for Percent Leaking (Valves)	     11

  3-1   Percent of Sources Leaking1  by Source	     15

  3-2   Percent Leaking for Each Chemical Produced as a Function of
          Source Type and Stream Service	     17

  3-3   Definition of Chemical Process Groups. . .•	     20

  3-4   Leak Frequencies by Process  Unit Group, Source Type and
          Stream Service 	     21

  3-5a  Percent of Leaking Valves by Primary Material in Line. ...     23

  3-5b  Percent of Leaking Valves by Primary Material in Line. ...     24

  3-6a  Percent Leaking by Primary Material for Valves -
          Gas Service	     25

  3-6b  Percent Leaking by Primary Material for Valves -
          Light Liquid  Service	     26

  3-7a  High Versus Low Leaking Primary Chemical Groups for High
          Leaking Process Units	     29

  3-Tb  High Versus Low Leaking Primary Chemical Groups for Ethylene
          Process Units	     30

  3-8   Percent Leaking for All Types of Valves in Gas Service as
          a Function  of Process Group and Primary Material Group .  .     32

  3-9   Percent Leaking for All Types of Valves with  Light Liquid
          Stream Service by Process  Group and Primary Material
          Group	     33

  3-10  Leak Frequencies for All Types of Valves for  Gas and Light
          Liquid Stream Service	     35
                                    VI11

-------
Number                                                                 Page

  3-11  Results of Categorical Analysis on Valves1	     39

  3-12  Leak Frequencies for Pump Seals in Light Liquid Service ...     44

  3-13  Results of Analysis of the Effects of Line Temperature and
          Line Pressure on Leak Frequency for Valves	     50

  3-14  Line Temperature and Line Pressure and Their Combined Effects
          on Valves in Gas Service within Ethylene Process Units.  .  .     51

  3-15  Line Temperature and Line Pressure and Their Combined Effects
          on Valves from Group 5*	     52

  3-16  Line Temperature and Line Pressure and Their Combined Effects
          on Valves from Group 3*	     53

  3-17  Line Temperature and Line Pressure and Their Combined Effects
          on Valves from Group 7A	     54

  3-18  Effect of Line Temperature and Line Pressure on Valves from
          Group 6*	     59

  3-19  Effects of Line Temperature and Line Pressure on Valves from
          Group 4 and Group 8 by Process Unit Group1. „..,..».     60

  3-20  Effects of Line Temperature and Line Pressure on Pump Seals
          with Light Liquid Service 	     63

  3-21  Effects of Line Temperature and Line Pressure on Flanges
          in Gas Service by Process Unit Group	     65

  3-22  Effects of Line Temperature and Line Pressure on Flanges in
          Light Liquid Service by Process Unit Group	     66

•  3-23  Effects of Line Temperature and Line Pressure on Open Ended
          Lines in Gas Service by Process Unit Group	     67

  3-24  Effects of Line Temperature and Line Pressure on Open Ended
          Lines in Light Liquid Service by Process Unit Group ....     68

  3-25  Summary of the Effects of Ambient Temperature on Percent
          Leaking	     70

  3-26  Summary of the Effects of Elevation on Percent Leaking. ...     72

  4-1   Estimated Emission Factors for Nonmethane Hydrocarbons from
          Valves and Pump Seals	     86

  4-2   Summary of Percent of Sources Distribution Curves and Percent
          of Mass Emissions Curves at Screening Value of 10,000 pprav.    116

                                     ix

-------
Number                                                                 Page
  5-1   Increase in Mass Emissions by Leak Occurrence for Valves
          and Pump Seals Screening <10,000 ppmv  Initially 	   118

  5-2   Increase in Mass Emissions by Leak Recurrence for Valves
          Screening <10,000 ppmv Immediately After Maintenance   .  .  .   120

  5-3   Weighted Percent Reduction in Mass Emissions  for Valves
          Screening ^10,000 ppmv Immediately Before Maintenance  .  .  .   124

  6-1   Percent Leaking Estimates for Valves in  Light Liquid  Service   128

  6-2   Percent Leaking Estimates for Valves in  Gas Service 	   129

  6-3   Percent Leaking Estimates for Valves in  Light Liquid
          Service by Process Type	   130

  6-4   Percent Leaking Estimates for Valves in  Gas Service by
          Process Type	   131

  7-1   Comparison of Emission Factors with Quality Control
          Estimates of Mean Leak Rates for Valves  and Pump Seals   .  .   150

  A-l   Data Summary of Leak Frequencies  for Various  Sources  in
          Various Stream Services 	   162

  B-l   Summary Statistics for Line Temperature  and Line Pressure
          for Gas Service	   170

  B-2   Summary Statistics for Line Temperature  and Line Pressure
          for Light Liquid Service  	   171

  B-3   Summary Statistics for Line Temperature  and Line Pressure
          in Heavy Liquid Service Within  High and  Ethylene Process
          Units	   172

  B-4   Effects of Line Temperature and Line Pressure on Percent
          Leaking for Valves in Gas Service Within Ethylene Process
          Units	   173

  B-5   Effects of Line Temperature and Line Pressure on Percent
          Leaking for Valves in Gas Service Within High Leaking
          Process Units by Chemical Group 	   174

  B-6   Effects of Line Temperature and Line Pressure on Percent
          Leaking for Valves in Light Liquid Service  Within
          Ethylene Process Units by Chemical Group  ...  	   175
                                      x

-------
Number                                                                 j?age

  B-7   Effects of Line Temperature and Line Pressure  on Percent
          Leaking for Valves in Light Liquid Service Within High
          Leaking Process Units by Chemical Group  	   176

  B-8   Effects of Line Temperature and Line Pressure  on Percent
          Leaking for Pump Seals in Light Liquid Service  	   177

  B-9   Effects of Line Temperature and Line Pressure  for Flanges
          in Gas Service From Ethylene Process Units   	   178

  B-10  Effects of Line Temperature and Line Pressure  for Flanges
          With Gas Service From High Leaking Process Units by
          Chemical Group  	   179

  B-ll  Effects of Line Temperature and Line Pressure  for Flanges
          in Light Liquid Service Within Ethylene  Process Units
          by Chemical Group	   180

  B-12  Effects of Line Temperature and Line Pressure  on Percent
          Leaking for Flanges in Light Liquid Service  Within  High
          Leaking Process Units by Chemical Group  	   181

  B-13  Effects of Line Temperature and Line Pressure  on Percent
          Leaking for Open Ended Lines in Gas Service  Within
          Ethylene Process Units  	   182

  B-14  Effects of Line Temperature and Line Pressure  on Percent
          Leaking for Open-Ended Lines in Gas Service  Within
          High Leaking Process Units by Chemical Group  	   183

  B-15  Effects of Line Temperature and Line Pressure  on Percent
          Leakage for Open-Ended Lines in Light Liquid Service
          Within Ethylene Process Units by Chemical Group 	   184

  B-16  Effects of. Line Temperature and Line Pressure  on Percent
          Leaking for Open-Ended Lines in Light Liquid Service
          Within High Process Units by Chemical Group  	   185

  C-l   Summary of Ambient Temperature During Screening of
          Various Source Types in Gas Service	   192

  C-2   Summary of Ambient Temperature During Screening of
          Various Source Types in Light Liquid Service  	   193

  C-3   Effect of Ambient Temperature on Percent of Sources
          Leaking in Ethylene Process Units as a Function of
          the Primary Chemical Groups 	   194

-------
Number                                                                 Page

  C-4   Effects of Ambient Temperature on Percent of  Sources
          Leaking in High Leaking Process Units as a  Function
          of the Primary Chemical Groups 	    195

  C-5   Effects of Source Elevation on Percent Leaking for  Ethylene
          Process Units as a Function of Primary Chemical Groups  .  .  .    196

  C-6   Effects of Source Elevation on Percent Leaking in High
          Leaking Process Units as a Function of Primary Chemical
          Groups	    197

  D-l   Corrections Affecting Results on Previous Reports  	    200

  D-2   Corrections to Screening Data Sheets	    202
                                     xxi

-------
                                  SECTION 1
                                INTRODUCTION

     The contribution of fugitive leaks from process unit components are
being investigated as a potential source of Volatile Organic Compound (VOC)
emissions in the Synthetic Organic Chemical Manufacturing Industry (SOCMI).
The purpose of this study is to provide an in-depth analysis of data on
these emissions collected under EPA contracts 68-02-3171-1, 68-02-3173-2 and
11, 68-02-3174-5, and 68-02-3176-1 and 6 and 68-03-2776-4.  These data were
collected by Radian, PEDCO, TRW and Acurex and are summarized in References
1 and 2.  The results of this study will be available for use in evaluating
VOC fugitive emissions.

     The study design and test procedures for the data analyzed in this
report are described in References 1 and 2.  The 24 process units studied in
the data collection programs were selected to represent a cross-section of
the population of the SOCMI.  Several factors were considered during process
unit selection.  These factors included total annual production volume,
number of producers, process conditions, corrosivity, volatility, toxicity,
and value of the final chemical product.  Factors varied widely from unit
type to unit type, so that the selected process unit types represented a
reasonable sample of the variety of chemical process units encountered in
SOCMI.

     Evaluating the leak frequency in SOCMI was done by the "collection of
screening data from 24 process units, where a screening value is the maximum
repeatable concentration of total hydrocarbons detected at a source with a
portable hydrocarbon detector (Reference 1),  Evaluation of maintenance was
done by measurement of fugitive emission leak rates (Ib./hour) at selected
sources before and after maintenance at six process units representing three

-------
chemical processes (Reference 2).  The results of these two programs provide
the background information necessary for the current study:

     •  source population data
     •  screening value profiles for each source type
     •  screening-to-emission rate relationships

     The screening procedures began with the definition of the process unit
boundaries.  All feed streams, reaction/separation facilities, and product
and by-product delivery lines were identified on process flow diagrams and
in the process unit.  Process data, including stream compositions, line tem-
peratures, and line pressures, were obtained for all flow streams.

     The Century Systems Models OVA-108 and OVA-128 hydrocarbon detectors
were used for screening.  The detector probe of the instrument was placed
directly on those areas of the sources where leakage would typically occur.
For example, gate valves were screened along the circumference of the annular
area around the valve stem where the stem exits the packing gland and at
the packing gland/valve bonnet interface.  The actual leak rate measurements
were taken using a flow-through method described in Reference 8 and were
analyzed on Byron Total Hydrocarbon Analyzer.

     All accessible sources of the following source types were screened:

     •  process valves,
     •  pump seals,
        compressor seals,
        agitator seals,
     •  relief valves,
     •  process drains, and
        open-ended lines.

-------
Also, a randomly selected subset of flanges was screened.  Originally, only
five percent of all flanges were screened.  The subset was increased to 20
percent of all flanges when initial results indicated a higher frequency of
emitting flanges than had been encountered in previous programs.  The impor-
tant variables available from this study are:  screening value, source
category, stream service, source type, chemical produced, ambient temperature
elevation, line temperature and line pressure.  For the purposes of this
report, a source is defined as "leaking" if its screening value is greater
than or equal to 10,000 ppmv.

     This report is actually a presentation of four distinct data analysis
tasks.  Section 2 is a short summary of the results of all four tasks.  In
Section 3 a detailed analysis of the SOCMI screening data (from 24 process
units) is presented along with summaries of important correlating process
parameters (line pressure, etc.).  Emission factor development for three
specific chemical processes (7 units) is presented in Section 4.  The analysis
reported in Section 5, an extension of the results in Reference 2, is directed
at investigating the increase in mass emissions due to occurrence and recur-
rence of leaks.  In Section 6 the impact on leak frequency from adjusting
screening values by chemical response models is investigated.  The statistical
methods used in Sections 3 through 6 are presented in Section 7.  Appendix A
is a statistical summary of all the screening data from the 24 untis.
Appendix B contains summary statistics and information on the effect of line
pressure and line temperature on the percent leaking.  Appendix C contains
similar descriptions for ambient temperature and elevation.  Appendix D is a
summary of all corrections made to the original data.

-------

-------
                                   SECTION 2
                          SUMMARIES AND CONCLUSIONS

     This section presents the major findings from the analyses discussed
in Sections 3 through 6.

RELATIONSHIP OF LEAK FREQUENCY TO PROCESS PARAMETERS (SECTION 3)

     The process parameters that were examined for their effect on leak
frequency were:  process, service, material in the line, line pressure, line
temperature, ambient temperature and source elevation.  Data on four source
types (valves, pump seals, flanges and open ended lines) were used to examine
the effects of these parameters.  The sources were grouped into 32 categories
(see Figure 3-2) based on source type, process type, stream service and
primary chemical in the line.  These groupings were for statistical reasons
and were not based on engineering reasoning.

     Stream service was defined as either gas, light liquid or heavy liquid
(Reference 1).  Heavy liquids were not included in any analyses, since they
leaked so rarely regardless of the other conditions.  Gas stream service
generally had a higher leak frequency than light liquid service.  Proceeding
with four source types and two stream service types  the data was then cate-
gorised by process unit as either ethylene processes, high leaking processes
or low leaking processes.  The ethylene units were analyzed separately because
of the large number of sources in ethylene processes and the high leak fre-
quency.  The high leaking group consists of all other units with greater
than 1% of all source types leaking.  The low leaking group consisted of all
units with less than 1% of all source types leaking.  Since there were very
few sources leaking, the low leaking process units were not considered in
further analyses.  Within these process unit groups, the data was further

-------
subdivided by primary materials in the line.   Caution should be  used  in
these evaluations, however,  since other chemicals  in the line  may  also  have
an effect on leak frequency.

     Examination of the data within these categories resulted  in the  follow-
ing conclusions for this data setr

        Leak frequency was affected not only  by the type of chemical  process
        but also by the type of primary material in the line.

     •  Control valves had a higher leak frequency than block  valves.

        For block valves, gate valves had a higher leak frequency  than  most
        of the other types,  and plug and ball valves have lower  leak  fre-
        quencies .

        On-line pump seals had an overall leak frequency of 13.1 percent
        versus 4.9 percent for off-line pump  seals.

        These data did not show a difference  in leak frequency between  double
        mechanical pump seals and single mechanical pump seals,  although the
        type of barrier fluid was unknown and therefore unaccounted for in
        this analysis.

        Line pressure was seen to have a statistically significant effect in
        almost every case, with higher levels of pressure associated  with
        higher leak frequencies,

     •  Line temperature had no consistent effect on leak frequency.   The
        combined effect of line pressure and temperature was important  in
        some cases.

-------
        Ambient temperature had a consistent effect on leak frequency,
        however, the effect was not statistically significant for a majority
        of the cases.  Higher leak frequencies tended to be associated with
        the higher ambient temperature category.

     •  Elevation had no consistent effect on leak frequencies.  In the four
        cases where a statistically significant effect was observed, sources
        at ground level had a higher leak frequency than sources at higher
        elevations.

EMISSION FACTOR DEVELOPMENT (SECTION 4)

     The sources included in the development of the emissions factors are all
valves and pump seals screened in the seven ethylene, cumene, and vinyl
acetate process units or 51.2% (16,575) of all valves and pump seals screened
in the screening program.  Since leak rate screening value models were only
developed for these three process types, emission factor estimation was
limited to these three'processes.

     The emission factors developed in this study are reported in Table 2-1.
The emission factors for ethylene process are consistently higher than the
factors for the cumene and vinyl acetate processes.  The vinyl acetate
process tends to have the lowest emission factors of the three process
types.

     Cumulative distributions of screening values and mass emissions as a
function of screening values were also developed for each of the three
processes.  Table 2-2 gives the estimates and confidence intervals from
these curves for a 10,000 ppmv screening value.

INCREASE IN MASS EMISSIONS DUE TO OCCURRENCE AND RECURRENCE (SECTION 5)

     Further analysis of data collected during the EPA SOCMI maintenance

-------
       TABLE 2-1.  ESTIMATED EMISSION FACTORS FOR NONMETHANE HYDROCARBONS
                   FROM VALVES AND PUMP SEALS (Ibs./hr./source and kgs./hr;/
                   source)
Source Type
                                Emission Factor (95% Confidence Interval)
    (Ibs./hr.)
    (kgs./hr.)
Valves
- Gas Sen/ice
    Ethylene processes
    Cumene processes
    Vinyl Acetate processes
- Light Liquid
    Ethylene processes
    Cumene processes
    Vinyl Acetate processes
0.024(0.008, 0.07)
0.011(0.003, 0.05)
0.0046(0.001, 0.03)

0.020(0.007, 0.06)
0.0056(0.002, 0.02)
0.0003(0.0001, 0.002)
0.011(0.004, 0.03)
0.0052(0.001, 0.02)
0.0021(0.0004, 0.01)

0.010(0.003, 0.03)
0.0025(0.001, 0.01)
0.0001(0.00003, 0.001)
Pump Seals
- Light Liquid
    Ethylene processes
    Cumene processes
    Vinyl Acetate processes
0.069(0.006, 0.8)
0.052(0.001, 2.7)
0.0043(0.0001, 0.1)
0.031(0.003, 0.4)
0.023(0.0004, 1.2)
0-0020(0.00006, 0.06)

-------
                    TABLE 2-2.  SUMMARY OF PERCENT OF SOURCES DISTRIBUTION CURVES AND PERCENT
                                OF MASS EMISSIONS CURVES AT SCREENING VALUE OF 10,000 PPMV
        Source Type
                                         Percent of Sources
                                       Screening ^ 10,000 ppmv
                          Estimate
             95% Confidence
                Interval
                        Percent of Mass Emissions
                         Attributable to Sources
                   	Screening > 10,000 ppmv
                                   95% Confidence
                   Estimate           Interval
00
Valves
  Gas
      Ethylene
      Cumene
      Vinyl Acetate
  Light Liquid
      Ethylene
      Cumene
      Vinyl Acetate
15
16
3.7

26
12
0.2
(U, 16)
(13, 19)
(2,  5)

(24, 27)
(10, 13)
(0,  0.4)
94
94
90

89
80
25
(93, 95)
(90, 96)
(85, 94)

(87, 90)
(72, 86)
(9, 47)
       Pump Seals
          Light Liquid
              Ethylene
                       t
              Cumene
              Vinyl Acetate
                             30
                             14
                             1.7
                (20, 39)
                (1, 27)
                (0, 4)
                      96
                      89
                      67
                (90, 98)
                (50, 98)
                ( 5, 92)

-------
program (Reference 2) was done to estimate the effects of leak occurrence and
recurrence on mass emissions.  The following conclusions are based on these
analyses:

        The increase in emissions for valves for which a leak occurred over
        a one to six month period was estimated to be 530% (95% confidence
        interval of 200% to 900%).

     •  Not enough data was available to accurately quantify the effect on
        emissions from leak occurrence from pump seals.  However, the percent
        increase estimate was 75% with a 95% confidence interval of -100%
        to 6000%.

     •  The percent increase in emissions for valves with a leak recurrence
        within the six month period was estimated to be 510% (95% confidence
        interval of -100% to 1700%).

     •  Further analysis of the effect of valve maintenance on emissions
        showed a 98% reduction in emissions for valves which were "repaired"
        (screening valve <10,000 ppmv after maintenance) and a 63% reduction
        for sources which were "not repaired" (screening valve remained -
        ^10,000 ppmv after simple, on-line maintenance).

IMPACT OF RESPONSE ADJUSTMENTS ON LEAK FREQUENCY ESTIMATION (SECTION 6)

     Three different techniques were used to adjust the original screening
value for each source:

        the original OVA reading adjusted for the associated OVA response
        relationship of the primary chemical compound in the line,

     "  weighted logarithmic average of response of primary and secondary
        chemicals, and

-------
        weighted arithmetic average of response of primary  and  secondary
        chemicals.

     The percent of leaking valves was calculated for each  of  the three
estimates for both gas and light liquid services.  The three estimates  were
found to be similar in most cases to the leak frequency based  on the original
screening valves.  Table 2-3 presents the overall results.
                                     10

-------
        TABLE 2-3.  COMPARABLE ESTIMATES FOR PERCENT LEAKING (VALVES)
                            (24 SOCMI Process Units)



Process
Stream
Gas
Light
Liquid



Number
Screened1
9374

18,133
Percent
Leaking
Based on
OVA
Readings
11.3

6.1
Percent
Leaking
Based on
Method 1
Adjustments2
10.1

5.3
Percent
Leaking
Based on
Method 2
Adiustments3
10.2

5.6
Percent
Leaking
Based on
Method 3
Adiustments1*
10.3

5.5.
     sources with screening valves = 10,001 ppmv were excluded.
2Method 1 is the adjustment to the OVA Reading based on the response of the
 primary chemical in the line.
3Method 2 is the mixed chemical weighted logarithmic average technique.
^Method 3 is the mixed chemical weighted average technique.
                                     11

-------
                                  SECTION 3
                     DETAILED RESULTS FOR THE EFFECTS OF
                    PROCESS PARAMETERS ON LEAK FREQUENCY
     The effects of various process parameters on leak frequency are
evaluated in this section.  The process variables analyzed are source
category, stream service, source type, chemical produced, ambient temperature,
elevation, line temperature and line pressure.  Each of these variables was
examined to determine which of them is associated with high or low leak
frequencies.  Leak frequency data from four source types are analyzed in
detail in this section.  They are open-ended lines, valves, pumps, and
flanges.  Simple summary statistics for all source types are presented in
Appendix A.  The data were grouped into exclusive categories for statistical
reasons (not engineering) as outlined in the following paragraphs.

     Data for this analysis come from an EPA study in which all sources
in 24 chemical process units were screened (Reference 1).  In a data collec-
tion study such as this, it is possible to have several of the process
parameters confounded.  This means it can be difficult to separate the
effects of one parameter from that of another.  For example, if one process
source type does in fact have a high frequency of leaks, but is almost
always associated with a certain type of stream service, it may appear that
the high leak frequency is associated with the stream service.  If the data
are grouped by both source type and stream service, the effect of each of
these two variables can be seen.  To avoid this type of problem, the data
have been analyzed in smaller groups whenever a possibility of confounding
was suspected.

     Another reason the data were grouped into subsets is that the analysis
procedure used to statistically evaluate factors affecting the leak frequency
                                     12

-------
is very sensitive to frequencies of zero.  That is, if there were no sources
leaking (or very few) in a particular category (e.g., heavy liquid) the
analysis procedure is not appropriate.  To avoid this problem, the data to
be analyzed for statistical significance were first categorized to include
only groupings that displayed at least a moderate percentage of leaking
sources.  Summary statistics for the groupings not statistically analyzed
(heavy liquids and process units with less than 1 percent of the sources
leaking) are presented separately in the appendices.

OVERVIEW OF SCREENING DATA FROM 24 CHEMICAL UNITS

     Table 3-1 gives information on the number of sources screened, the
number that were leaking and the percentage that were leaking in the 24
chemical units screened.  This information is given for each source type and
each stream service within each source category.  (The stream service classi-
fications are described in Reference 1.)  It can be seen from this table
that sources in the heavy liquid service category have a fairly low leak
frequency.  There are also fewer heavy liquid service sources than gas or
light liquid in each source type.  However, even in a group such as valves,
where there were 3,632 valves in heavy liquid service, the leak rate is
very low (0.4 percent, -or 13 leaking sources). Table 3-1 shows that valves
in gas service have both a large number and high percentage of leaking
sources.  It also appears that the percent leaking varies with both source
type and stream service.

     Valves as a source type had the largest number of screening values.
Flanges and open-ended lines also had a large number (although only 5 to 20
percent of the flanges were screened).  For further analysis, these three
categories plus pump seals were investigated.  It was felt that the sample
sizes of the other categories were too small to allow meaningful subcate-
gorization of the source type.

     Since only 17 sources in heavy liquid stream service in the source types
to be further analyzed were found to be leaking, sources in this service were
                                     13

-------
not included in further anlaysis of factors affecting the leak frequency.
However, summary statistics for this stream service are included in the later
sections, where appropriate.
                                      14

-------
                              TABLE  3-1.    PERCENT OF SOURCES LEAKINGJ
                                                       (24  Process  Units)
BY  SOURCE

Source Service
Valves Gas*
Light Liquid3
Heavy Liquid
Pump Seals Light Liquid3
Heavy Liquid
Flanges Gas1
Light Liquid3
Heavy Liquid
Open Ended Lines Gas
Light Liquid5
Heavy Liquid
Process Drains Gae
Light Liquid3
Heavy Liquid
Agitator Seals Gas
Light Liquid
Heavy Liquid
Relief Valves Gas3
Light Liquid*
Heavy Liquid
Compressors Gas5
Other2 Gas
Light Liquid1
Heavy Liquid
Number
Screened
9669
18299
3632
646
97
1450
2833
607
923
3605
477
83
496
28
7
8
1
84
68
3
22
19
34
2

Number
1103
1183
13
57
2
66
36
0
54
141
2
2
19
2
1
0
0
3
2
0
2
3
2
0
Sources with Screening
Percent
U. 4
6.5
0.4
8.8
2,1
4.6
1.3
0
5.9
3.9
1.3
2.4
3.8
7.1
14.3
0
0
3.6
2.9
0
9.1
15. B
5.9
0
Values >1 0,000
95Z Confidence Interval
for Percent >10,000
(10.8, 12.0)
(6.1, 6.9)
(0.2, 0.7)
(6.4, 11.0)
(0.3, 7.3)
(3,7, 5.7)
(0,9, 1.7)
(0, 0.6)
(4.4,7.9)
(3.4, 4.7)
(0.5, 2.8)
(0.3, 8.4)
(2.3, 5,9)
(0.9, 23.5)
(0.4, 57.9)
(0, 36.9)
(0, 100)
(0.7, 10.1)
(0.4, 10.2)
(0, 70.8)
(1.1, 26.2)
(3.4, 39.6)
(0.7, 19.7)
<0, 84.2)*
1A leaking source is defined as one  with a screening value >10,000 ppmv.

2Includea filters,  vacuum breakers,  expansion Joints, rupture disks,  sight p.iass aeals, etc.

 The numbers in each column may be different from that found in Reference ; because of corrections to  the original data
 (See Appendix D).

-------
EFFECT OF CHEMICAL PRODUCED ON LEAK FREQUENCIES

     Table 3-2 describes the screening data in terms of chemical produced
by the source types and service categories outlined earlier.  Some differ-
ences between the chemical processes are apparent.  The production of ethy-
lene appears to be associated with a leak frequency that is higher than that
found with the production of any of the other chemicals.  Leak frequencies
from the Cumene and MEK units are also high.  Other processes had very low
leak frequencies for all four of the source types.  The formaldehyde unit
screened only had two leaks and the two adipic acid units had no leaks from
the four source types.  Figure 3-1 graphically presents the estimated percent
leaking along with 95 percent confidence intervals for valves in gas and
light liquid service by process type.

     It is clear from looking at Figure 3-1 and examining Table 3-2 that the
breakdown by process type, in addition to source type and stream service,
results in some subsets with few or no leaking sources.  To avoid the problem
of  analyzing such small groups, a method of grouping the chemicals produced
was devised.  Three chemical process groups based on overall leak frequency
were formed.  The groups are Low Leaking Process, High Leaking Process, and
Ethylene Process.  Each category, and the processes and unit identification
numbers that are associated with it, is given in Table 3-3.  The Low Leaking;
group contains data on chemicals whose leak frequency was less than one per-
cent for all source types and stream services.  The overall leak frequencies
for the High Leaking group range from one percent to six percent.

     Table 3-4 summarizes the data available for further analysis for the
subcategories formed by the source type, service category, and chemical
process groups.  In the analysis of the effect of other process parameters
on leak frequency, only the High Leaking and Ethylene groups were used.  The
Low Leaking group had too few leaks to adequately determine any types of
effects on leak frequency of the other variables.
                                     16

-------
TABLE 3-2.  PERCENT  LEAKING FOR EACH CHEMICAL PRODUCED  AS  A FUNCTION OF
            SOURCE TYPE AND STREAM SERVICE
Source/Chemical (units) '
Valves
Vinyl Acetate (1,3)
Kthylene (2,4,11)
Cumene (5,6)
Acttone/I'lienol (12)
lithylene bichloride (2.1,29)
Vinyl Chloride Monomer (20,28)
Formaldehyde (22)
Methyl Ethyl Ketone (31,32)
Acetaldehyde (33)
Methyl Metliacrylate (34)
Adipiu Acid (35,61)
Chlorinated Ethanes (60, 62)
Ai:rylpnitrlle (65,66)
1,1,1-Trlcliloroetlwna (&1)
Vinyl Acetate (1,3)
Ethyleme (2,1,11)'
Cumene (5,6)
Acetone/Phenol (12)
Ktliylene bichloride (21,29)
Vinyl Chloride Monomer (20,28)
Formaldehyde (22)
Methyl Ethyl Ketone (31,32)
Acetaldehyde (33)
Methyl Metliacrylate (34)
Atllpic Acid (35,64)
Chlorinated Ethanes (60,62)
Ai.'1-ylnnltrlle (65,66)
1,1 l-Triehlorocthane (filj

Number
Screened

919
6291
418
8
103
' 112
41
207
178
190
95
48
396
	
	
	
	
	
	
	
	
	
	
	
	
	
	
	
GAS
Number
Leaking

35
934
63
0
4
30
1
19
8
0
0
0
9
	
	
	
	
	
	
	
	
	
	
	
	
	
	
	
LIGHT LIQUID
Percent
Leaking

3.
14.
14.

1.
7.
2.
9.
4.



2.
•"•
	
	
	
	
	
	
	
	
	
	
	
	
	
	

7
8
1
0
0
3
4
2
5
0
0
0
3
~
_
-
-
-
-
-
-
-
-
-
~
-
-

Number
Screened

2137
4176
799
1818
2256
1209
121
671
551
1058
17
1620
1494
373
89
76
25
86
58
65
8
31
32
45
	
60
61
10
Number
Leaking

8
969
84
6
24
12
0
34
3
1
0
10
28
4
4
20
4
2
3
7
0
1
3
2
	
5
5
1
Percent
Leaking

0.
23.
10.
0.
1.
1.

5.
0.
0,

0.
0.
1.
4.
26.
16.
2.
5.
10.

3.
9.
4.
	
8.
a.
10.

4
2
5
3
1
0
0
1
5
1
0
6
9
1
5
3
0
3
2
8
0
2
4
4
-
3
2
0
Number
Screened.

124
1237
198
488
	
	
	
	
	
	
1478
12
95

5
15
3
36
	
	
	
	
	
	
30
	
B
	
HEAVY LIQUID
Number
Leaking

0
13
0
0
	
	
	
	
	
	
0
0
0

0
0
0
0
	
	
; 	
	
	
	
0
	
'I
	

Percent
Leaking

0
1.1
0
0
	

	
	
	
	
0
0
0

0
0
0
0
	
	
	

	
	
0
	
25.0
, 	
                                                                                   (Continued)

-------
               TABLE  3-2.   (continued)
CO
Source/Chemical (units) '
flanges
Vinyl Acetate (1,3)
Ethylene (2,4,11)
Cuinene (5,6)
Ace tone /Phenol (12)
Jithylene Bichloride (21,29)
Vinyl Chloride Monomer (20,28)
Formaldehyde (22)
Methyl Elliyl Ketone (31,32)
Acetaldeliyde (33)
Methyl Methacrylate (34)
Adiplc Acid (35,64)
Chlorinated Ethanes (60, 62)
Acrylonltrile (65,66)
J ,1,1-Ttrichloroethane (*1)
Open Ended Lines
Vinyl Acetate (1,3)
Ethylene (2,4,11)
Cumpne (5, 6)
Acetone/Phenol (12)
Ethylene Bichloride (21,29)
Vinyl Chloride Monomer (20,28)
Formaldehyde (22)
Methyl Ethyl Kutone (31,32)
Acetaldehyde (33)
Methyl Methacry late (34)
Ajlplc Acid (35,64)
Chlorinated Ethanes (60,62.)
Acrylonltrile (65,66)
l.Ul-Trlchloroethane (61.)

Number
Screened

107
634
367
	
25
16
2
22
32
38
49
16
142
	

145
305
6
2
100
55
14
37
34
63
19
27
116
	
GAS
Number
Leaking

3
39
19
	
1
2
0
0
0
0
0
0
2
	

8
37
0
0
0
2
0
3
3
0
0
0
1
	
LIGHT LIQUID
Percent
Leaking

2
6
5
--
4
12






1
•

5
12



3

8
8



0
—

.8
.2
.2
—
.0
.5
0
0
0
0
0
0
.4
—

.5
.1
0
0
0
.6
0
.1
.8
0
0
0
.9
—
Number
Screened

173
407
468
82
163
47
8
76
144
247
2
461
382
73

318
214
15
518
475
340
36 .
186
158
335
1
412
486
111
Number
Leaking

0
25
9
0
1
0
1
0
0
0
0
0
0
0

8
41
2
8
16
18
0
19
8
1
0
6
12
2
Percent
Leaking


6.
1.

0,

12.








2,
19,
13.
1,
3,
5,

10,
5,
0,

1,
2,
1

0
,1
6
0
.6
0
,5
0
0
0
0
0
0
0

.5
.2
.3
,5
.4
,3
0
,2
,1
.3
0
.5
,5
.8
Number
Screened

8
89
130
30
	
	
	
	
	
	
320
2
28
	

22
91
1
107
	
	
	
	
	
	
214
4
38
	
HEAVY LIQUID
Number
Leaking

0
0
0
0
	
	
	
	
	
	
0
0
0
	

2
0
0
0
	
	
	
	

	
0
0
4


Percent
Leaking

0
0
0
0
	
	

	
	
	
0
0
0


9.1
0
0
0
	
	
	
	
	
--_—
0
0
10.5


-------
r
                       Opp,r 951
                       Confidence
                       Interval
                                             VALVES-GAS SERVICE PERCENT LEAKING BY PROCESS TYPE
                                       35

                                     ^30
                                     r
                                     c 25

                                     n __


                                     I 15
                                     •
                                     ? 19
                                                                       '    I1       ]I
                                                                   I  31,32  I   34   169,   621
 2,4,11 t   12   I 20,28  t 31,32 I   34
3     5,6     21,29     22      33     35,64    65,66
             Process Unit Number1
                    ii-*-
                                          VALVES-LIQUID SERVICE PERCENT LEAKING BY PROCESS TYPE
                                      40

                                      35

                                    :3e
                                    r
                                    <= 25

                                    £ 20

                                    I 15
                                    *
                                    ? 10
                                                                       [S3  cq£3 i-4-i
                                         n—i—i—1—i—I—i—|—i—I   i    |    i   r
                                         1,3  I • 5,6   I 21,28 I   22  I   33   I 35,64 ( 65,66 I
                                            2.4,11    12    20,28   31,32    34    60,62    61
                                         Figure  3-1.   Effect of Process Type on
                                                        Percent of Valves Leaking
                     1 See Table 3-3  for definition of process  type identification numbers
                                                              19

-------
               TABLE 3-3.   DEFINITION OF CHEMICAL PROCESS GROUPS
Process Group
"Low Leaking"
<1% of all
source types
leaking



"High Leaking"
>1% of all
source types
leaking





"Ethylene"
Chemical Process
Adipic Acid
Acetone
Formaldehyde
Methyl Methacrylate
Trich.loroethylene/
Perchloroethylene
Vinyl/Ethylene Bichloride
Ace t aldehyde
Acrylonitrile
Vinyl Acetate
Vinyl Chloride Monomer
Ethylene Bichloride
1,1,1-Trichloroethane
Cumene
Methyl Ethyl Ketone (MEK)
Ethylene
Unit Numbers
35, 64
12
22
34
60
62
33
65, 66
1, 3
20, 28
21, 29
61
5, 6
31, 32
2, 4, 11
Percent
Leaking*
0.0
0.5
0.8
0.3
0.8
0.0
2.3
1.7
1.4
2.8
1.2
1.2
6.3
5.9
12.9
*For all source types and stream services
                                     20

-------
TABLE 3-4.  LEAK FREQUENCIES BY PROCESS UNIT GROUP,
            SOURCE TYPE AND STREAM SERVICE
Ethylene Process Units
Source Type
Valves

Pump Seals
Flanges

Open Ended Lines

Stream Service
gas
light liquid
light liquid
gas
light liquid
gas
light liquid
Number Number
Screened Leaking
6294 934
4176 969
76 20
634 39
407 25
305 37
214 41
Percent
Leaking
14.8
23.2
26.3
6.2
6.1
12.1
19.2
95% Confidence
Interval for
Percent Leaking
(13.8, 15.8)
(21.8, 24.6)
(16.9, 37.7)
(4.4, 8.4)
(4.0, 8.9)
(8.6, 16.3)
(14.0, 25.3)

High Leaking Process
Source Type
Valves

Pump Seals
Flanges

Open Ended Lines

Stream Service
gas
light liquid
light liquid
gas
light liquid
gas
light liquids
Number Number
Screened Leaking
2993 168
9490 197
371 28
711 27
1626 10
493 17
2089 85
Percent
Leaking
5.6
2.1
7.5
. 3.8
0.6
3.4
4.1
Units
95% Confidence
Interval for
Percent Leaking
(4.7, 6.5)
(1-8, 2.4)
(5.1, 10.5)
(2.5, 5.5)
(0.3, 1.1)
(2.0, 5.4)
(3.3, 5.0)

Source Type
Valves

Pump Seals
Flanges

Open Ended Lines

Stream Service
gas
light liquid
light liquid
gas
light liquid
gas
light liquid
Low Leaking
Number Number
Screened Leaking
382 1
4626 16
199 9
105 0
798 1
125 0
1300 15
Process Units
Percent
Leaking
0.3
0.4
4.5
0.0
0.1
0.0
1.2
95% Confidence
Interval for
Percent Leaking
(0.01, 1.5)
(0.2, 0.6)
(2.1, 8.4)
(0.0, 3.4)
(0.0, 0.7)
(0.0, 2.9)
(0.7, 1.9)
                         21

-------
EFFECT ON LEAK FREQUENCY OF PRIMARY CHEMICAL IN THE PROCESS LINE

     The effect on leak frequency of the primary chemicals in the process
lines is investigated in this section.  The definition of primary chemical
is described in Reference 1.  Only the primary chemical is investigated here;
the influence of the other chemicals in the line is not evaluated.  The
results of this section should be considered with this in mind.

     Tables 3-5a and 3-5b display the percent of leaking sources by their
primary chemical in the line for valves - gas service and liquid service,
respectively.  Large differences in leak, frequency between primary chemicals
can be seen in these tables.  Because of these differences it was decided
to further categorize the sources by the primary chemical in the line, de-
pending on the leak frequency associated with that chemical.

     To do this categorization, the percent leaking data for valves associated
with primary chemicals were analyzed for the categories previously established
(source type, stream service, and process groups).  Tables 3-6a and 3-6b dis-
play this data for valves.  It can be seen that chemicals associated with
high percent leaking in the ethylene group were also seen to be associated
with high percent leaking in the high leaking process unit grouping.  For
example, ethylene as a primary chemical in ethylene process units has a high
percent leaking and, it also was found to have a frigh percent leaking in
other process units.

     Using the data from Table 3-6, the primary chemicals were grouped into
two categories.  If the percent of leaking (from Tables 3-6a and 3-6b) was
above 5% the chemical was put into the high leaking chemical group.  Other-
wise it was put into the low leaking group.  The resulting final groupings
of the screening data for further analyses are shown in Figure 3-2.
                                     22

-------
    TABLE 3-5a.  PERCENT OF LEAKING VALVES BY PRIMARY MATERIAL IN LINE
                            (All Process Units)
Valves - Gas Service
Chemical
Ethylene
Methane
Propylene
1,2-Ethylene Bichloride
Ethane
Benzene
Acrylonitrile
Vinyl Acetate
Ac et aldehyde
Propane
Acetic Acid
Methyl Ethyl Ketone
Vinyl Chloride
Other Chemicals
Total
Number
Screened1
3134
1849
1128
525
379
332
287
272
179
145
125
116
96
851
9418
Percent of
Total Gas
Service
Valves
33.3
19.6
12.0
5.6
4.0
3.5
3.0
2.9
1.9
1.5
1.3
1.2
1.0
9.0
100%
Number
Leaking1
498
232
207
4
35
53
0
0
4
18
1
7
0
42
1101
Percent
Leaking
15.8
12.5
18.3
0.8
9.2
16.0
0.0
0.0
2.2
12.4
0.8
6.0
0.0
4.9
11.7
Numbers displayed in this table may not add up to totals in previous
sections due to missing information on primary chemicals.
                                    23

-------
   TABLE 3-5b.   PERCENT OF  LEAKING VALVES BY  PRIMARY
                  MATERIAL  IN LINE

                  (All process units)
Chemical
1,2-Ethylene Dichloride
Propylene
Ethylene
Acetic Acid
Acrylonitrile
Vinyl Acetate
1,1,2-Trichloroethane
Cumene
Vinyl Chloride
Percfaloroethylene
Phenol
Benzene
Acetaldehyde
Methyl Ethyl Ketone
Methyl Methacrylate
Methanol
Ethane
cc-Methyl Styrene
Hydrocarbons-Cs +
Tri chlo roethylene
Acetone
Methane
Sec Butyl Alcohol
Acetone Cyanohydrin
Other Chemicals
Total
Valves
Number
Screened
2809
1604
1230
1162
1126
973
914
773
611
601
594
536
456
425
393
373
328
326
323
272
209
205
202
191
1572
18208
- Light Liquid
Percent of
Total Light
Liquid
Service
Valves
15,4
8.8
6.8
6.4
6.2
3.3
5.0
4.2
3.4
3.3
3.3
2.9
2.5
2.3
2.2
2.0
1.8
1.8
1.8
1.5
1.1
1.1
1.1
1.0
8.6
100%
Service
Number1
Leaking
32
488
321
6
6
3
4
4
4
3
0
49
2
23
1
4
92
0
8
6
5
36
10
0
69
1176
Percent
Leaking
1.1
30.4
26.1
0.5
0.5
A •>
W.J
0.4
0.-5
0.6
0,5
0.0
9.1
0.4
5.4
0.2
1.1
28.0
0.0
2.5
2.2
2.4
17.6
5.0
o.o
4.4
6.5
Numbers displayed in this table may not add up to totals in previous sections
due to missing information on primary chemicals.
                               24

-------
                    TABLE 3-6a.  PERCENT  LEAKING BY  PRIMARY MATERIAL  FOR VALVES  -  GAS  SERVICE
ro
Ln
High Leaking Process Units

Ethylene
Methane
Propylene
1,2-Ethylene
Dichloride
Ethane
Benzene
Acrylonitrile
Vinyl Acetate
Acetaldehyde
Propane
Acetic Acid
Methyl Ethyl
Ketone
Vinyl Chloride
Other Chemicals
TOTAL
Number1
Screened
680
-
69
505
-
282
287
272
179
81
125
116
96
294
2986
Percent ,
of
Total
Screened
22.
-
2.
16.
-
9.
9.
9.
6.
2.
4.
3.
3.
9.
100.
8

3
9

4
6
1
0
7
2
8
2
9
0
Number*
Leaking
62
-
15
4
-
50
0
0
4
12
1
7
0
13
168
Ethylene Process Units .
Percent
of
Percent Number1 Total Number1
Leaking Screened Screened Leaking
9.
-
21.
0.
-
17.
0.
0.
2.
14.
0.
6.
0.
4^
5.
1 2454 40.6 436
1849 30.6 232
7 1059 17.5 192
8 -
379 6.3 35
7 50 0.8 3
0 - - _
0 - -
2 -
8 64 1.1 6
8 - -
0 - -
,0 - - -
4 195 3.2 28
6 6050 100.0 932
Percent
Leaking
17.8
12.6
18.1
-
9.2
6.0
-
-
-
9.4


_
14.4
15.4
                      Numbers displayed In this table may not add up to totals In previous sections due to

                      missing information on primary chemicals.

-------
           TABLE 3-6b.   PERCENT LEAKING  BY  PRIMARY MATERIAL  FOR  VALVES  -
                           LIGHT LIQUID SERVICE
High Leaking Process Units
Primary Chemical
in the Line
Number
Screened
Percent
of
Total
Screened
Number 1
Leaking
Percent
Leaking
Ethylene Process Units
Number 1
Screened
Percent
of
Total
Screened
Number
Leaking
Percent
Leaking
1,2 Ethylene         2809       29.7      32
  Dichlorlde
Propylene            253        2.7      44
Ethylene               9        0.1       0
Acetic Acid          1162       12.3       6
Acrylonitrile        1126       11.9       6
Vinyl Acetate         973       10.3       3
1,1,2 Trichlorethane   -          -
Vinyl Chloride        611        6.5       4
Perchloroethylene      -          -
Phenol                -          -
Benzene              432        4.6      48
Acetaldehyde          456        4.8       2
Methyl Ethyl Ketone   425        4.5      23

Methyl Methacrylate    -          -        -

Methanol              -          -
Ethane                           -
Hydrocarbons Cj                   -
a-Methyl Styrene       -          -
Trichloroethylene      -          -
Acetone               -          -
Methane               -          -        -
Sec Butyl Alcohol     202        2.1      10
Acetone Cyanohydrin    -          -        -

Other Chemicals       827        8.7     _12_

TOTAL               9453      100.0     193
                                                  1.1

                                                 17.4
                                                  0.0
                                                  0.5
                                                  0.5
                                                  0.3

                                                  0.7
                                                 11.1
                                                  0.4
                                                  5.4
5.0


1.4
2.0
          1351
          1221
                                                             104
                                                              68
                                                             328
32.8
29.6
                                                                      2.5
                    1.6
                    8.0
                                                             205      5.0
                                                             844      20.5
444
321
             4
            92
                                                            4121     100.0
          J>8
          966
32.9
26.3
                                                                                          1.0
           5.9
          28.1
                                                                                 36       17.6
                                                                                            .1
                                                                                          15.4
1Numbers displayed in this  table may not add  up  to totals in previous  sections due to missing
information on primary  chemicals.

-------
        SOURCE
        TYPE
        PROCESS
        UNITS
        STREAM
        SERVICE
                                   Valves
                                   Pump Seals
                                   Flanges
                                   Open Ended Lines
             Ethylene
             Unit*= 2,4,11
ho
 High
Leaking
GROUP 1
  Low
Leaking
GROUP 2
                        Light Liquid
 High
Leaking
GROUP 3
  Low
Leaking
GROUP 4
                                        Gas
 High
Leaking
GROUP 5
                                               High Leaking
                                               Unit=l,3,5,6,20,21
                                                    28,29,31,32,33,
                                                    61,65,66
                                               Light Liquid
  Low
Leaking
GROUP 6
 High
Leaking
GROUP 7
  Low
Leaking
GROUP 8
PRIMARY
CHEMICAL











Ethylene None
Methane
Propylene
' Ethane
Benzene
Propane
Other






Ethylene
Propylene
Benzene
MEK
Ethane
Methane
Butylene
C6+
Methanol
Propane
Butanes
Butadiene
Butylenes ,
C4+
C5+
Ethanol
N Butane
Benzene







mixed
Ethylene
Propylene
Benzene
Propane
MEK








1,2-Ethy- Ethylene
lene Dich- Propylene
loride Benzene
Acryloni- MEK
trile
Vinyl Ace-
tate
Acetalde-
hyde
Acetic Acid
Vinyl Chloride
Other

1,2-Ethy lene
Bichloride
Acetic Acid
Acrylonitrile
Vinyl Acetate
Cumene
Vinyl Chloride
Acetaldehyde
Sec Butyl Alcohol
Other



        1There were <4% of the sources in ethylene process units-gas streams associated with a low leaking
         primary chemical.
        2See Table 3-2 for definition of process unit type identification numbers
                              Figure 3-2.  Categories of Sources for Further Analysis

-------
      The major  reason  for  grouping  the  sources  into  these eight  categories
was  to  aggregate  the sources  into groups  that have similar  leak  frequencies.
Note that  this  categorization was done  for  the  data  analysis  and not  for
engineering  or  physical  reasons.  The final 25  groups  used  for further
analyses are internally  similar in:

        source  type
      •  stream  service
      •  leak frequency by  process type, and
      •  leak frequency by  primary chemical  in  the line.

With these groupings,  any  differences  that  the  analysis  detects  in the  other
parameters" of interest (line  pressure,  line temperature,  etc.)'will not be
confounded with these  grouping  parameters.

      The division into categories was  done  separately  for each combination
of stream  service and  process unit  category.  For this reason, a particular
chemical may be grouped  in the  high leaking group in one subset  and the low
leaking group in  another.   Also the influence  of other chemicals in the line
was  not investigated here. As  a result,  it is  difficult to quantify the
effect of  a specific chemical.

      Two additional comments  should be  made.  First, the chemical  groupings
were made  according to valve  data only, so  the  high-low  primary  chemical
breakdown  for pump seals,, flanges,  and open-ended lines  (see Tables 3-7a and
3-7b) may  not reflect  a  strict  high versus  low  leaking classification in  all
cases.  Secondly, the  numbers displayed in  the  tables  in this section may
not add up to totals  from other tables  in previous sections due  to missing
information on  primary chemicals for some sources.
                                      28

-------
  TABLE 3-7a.   HIGH VERSUS  LOW  LEAKING  PRIMARY  CHEMICAL GROUPS  FOR  HIGH LEAKING PROCESS  UNITS
Hiah Leaking Chemicals (Group 5 and Group 71)
Source Type
Valves

Pump Seals
Flanges

Open Ended
Lines
Stream
Service
Gas
Light Liquid
Light Liquid
Gas
Light Liquid
Gas
Light Liquid
Number
Screened
1228
3299
126
391
578
146
798
Number
Leaking
146
147
14
18
10
13
47
Percent
Leaking
11.9
4.5
11.1
4.6
1.7
8.9
6.0
95Z Confidence
Intervals for
Percent Leakina
(10.2
(3. 1.
(6.;:,
<2.7,
(0.7,
(4.9.
(4.J.,
, 14.1)
5.3)
18.1)
7.1)
3.6)
14.6)
7.9)
Low
Number
Screened
1758
6154
243
316
1010
347
1291
Leaking Chemicals (Group 6 an
1 Group 81)
952 Confidence
Number Percent Intervals for
Leaking Leaking Percent Leakintt
22 1.2 (0
46 0.8 (0.
14 5.8 (3.
9 2.8 (1
0 0,0 (0
4 1.2 (0.
38 2,9 (2
7, 1.7)
6, 1.0)
3, 9.4)
2, 5.3)
0.4)
4, 3.1)
0, 4.1)
'See Figure 3-2 for explanation of groups.
Note:  Chemical groupings were made according to valve data only,  so other sources may not reflect
      a strict high versus low leaking classification.

-------
                   TABLE 3-7b.   HIGH VERSUS  LOW  LEAKING  PRIMARY  CHEMICAL GROUPS  FOR  ETHYLENE  PROCESS  UNITS
w
High Leaking Chemicals (Group 1 and
Source Type
Valves
-
Pump Seals
Flanges

Open Ended
Lines

Stream Number
Service Screened
Gas 6050
Light Liquid 3514
Light Liquid 61
Gas 566
Light Liquid 32 7
Gas 284
Light Liquid 151
Number Percent
Leaking Leaking
932 15.4
957 27.2
18 29.5
39 6.9
25 7.6
37 13.0
39 25.8
Group 31
)
95% Confidence
Intervals for
Percent LeaklnK
(14,
(26,
(19,
(4.9,
(5.0,
(9.4,
(19,
16)
29)
43)
9.2)
12)
17)
34)
Low Leaking Chemicals (Group 41)
95% Confidence
Number Number Percent Intervals for
Screened Leaking Leakina Percent Leakina

607 9 1.5 (0.67, 2.8)
15 2 13.3 (1.7, 40)

70 0 0.0 (0.0, 5.1)

63 2 3.2 (.39, 11)
           'See Figure 3-2 for explanation of groups.

           Note;  Chemical groupings were made according to valve data only so other sources may not reflect
                 a strict high versus low leaking classification.

-------
EFFECT OF TYPE OF VALVE ON LEAK FREQUENCY

     The first breakdown for studying the effect of valve type is block
valves versus control valves.  Within these two categories, there are six
types of valves evaluated:  gate, globe, plug, ball and butterfly, plus one
group called "other" which includes any valve type that does not fit into the
first five categories.

     Tables 3-8 and 3-9 show the valve leak frequency data for gas and light
liquid stream service, respectively.  Confidence intervals for percent leaking
are included to help distinguish those cases with a high percent leaking but
a small sample size.

     Since some of the specific types of valves, particularly for
control valves, had very small sample sizes when the data is categorized, an
overall tabulation of valve types is given in Table 3-10.  The smaller con-
fidence limits make differences by type more easily seen.  Figures 3-3a and
3-3b show this information graphically.  For block valves, gate valves have
the highest leak frequency while plug and ball valves have the lowest leak
frequency.

     The totals for block and control valves over all individual types were
tested to evaluate the influence of both process unit and chemical in the
line.  Categorical statistical analyses were performed on these data to
determine the significance of these classifications and their combined effects
(This method of analysis is described in Section 7.)  Separate analyses were
performed for gas and light liquid stream service.  A summary of these analy-
ses are given in Table 3-11.

     The analysis of gas stream service does not include a variable to
distinguish primary material groups since no such group was defined for
ethylene process units with gas stream service.  The analysis shows a sig-
nificant effect of chemical produced, block/control and their combined
                                     31

-------
            TABLE 3-8.  PERCENT LEAKING FOR ALL TYPES OF VALVES IN GAS SERVICE
                         AS A  FUNCTION OF PROCESS GROUP AND PRIMARY MATERIAL GROUP
High Leaking Process Units
Grouo 51 Prlraarv Chemicals
Valve
Function Type
Block gate
globe
plug
ball
butterfly
other
Total
Control gate
globe
plug
ball
butterfly
other
TOTAL
Number
Screened
978
9
34
102
17
1
1141
19
39
3
5
20
1
87
Number
Leaking
107
4
0
4
1
0
116
8
11
0
1
9
0
29
Percent
Leaking
10.9
44.4
0.0
3.9
5.9
0.0
10.2
42.1
28.1
0.0
20.0
45.0
0.0
33.3
952
Confidence
Interval
(9.1, 13)
(14, 79)
( 0, 10)
(1.1, 9.6)
(0.2, 29)
(0, 100)
(8.7, 13)
(20, 66)
(15, 45)
(0, 71)
(0.5, 72)
(23, 68)
(0. IQOi
(24,44)
Cr.
Number
Screened
1124
34
245
273
21
9
1706
15
21
2
1 .
13
0
52
oup 61 Pr1
Number
Leaking
20
1
0
0
0
0
21
0
1
0
0
0
-
1

Percent
Leaking
1.8
2.9
0.0
0.0
0.0
0.0
1.2
0.0
4.8
0.0
0.0
0,0
- .
1.9
95%
Confidence
Interval
(1.1, 2.8)
(0.1. 15)
(0, 1.5)
(0, 1.3)
(0, 16)
(0, 34)
(0.8, 1.9)
(0, 22)
(0.1. 24)
(0, 84)
(0, 100)
(0, 25)
-
(0.1, 10)
Ethylene Process Units

Number
Screened
4495
73
39
834
102
261
5804
24
137
5
8
56
16
246
Number
Leaking
823
10
0
14
8
16
871
7
24
0
2
26
2
61
Percent
Leaking
18.3
13.7
0.0
1.7
7.8
6.1
15.0
29.2
17.5
0
25.0
46.4
12.5
24.8
952
Confidence
Interval
(17, 19)
(6-7, 24)
(0, 9.0)
(0.9, 2.9)
(3.3, 15)
(3.7, 10)
(14, 16)
(13, 51)
(11, 24)
(0, 52)
(3.2, 65)
(34, 62)
(1.6, 38)
(19, 30)
See Figure 3-2 for explanation of groups.

-------
TABLE 3-9.   PERCENT LEAKING FOR ALL TYPES OF VALVES WITH LIGHT LIQUID STREAM
            SERVICE BY PROCESS GROUP AND PRIMARY MATERIAL GROUP
High Leaking Process Units
Group 7 Primary Chemicals
Valve
Type
Clock
Gate
Globe
Plug
Ball
Butterfly
Other
TOTAL
Control
Gate
Globe
Plug
Ball
Butterfly
Other
TOTAL
Number
Screened
2330
37
470
255
63
5
3160

58
40
26
2
11
2
138
Number
Leaking
122
4
2
0
0
0
128

11
6
1
0
1
0
19
%
Leaking
5.2
10.8
0.4
0.0
0.0
0.0
4.1

19.0
15.0
3.8
0.0
9.1
0.0
13.8
95X
Confidence
Interval

(4.3, 6.2)
(3.0, J5)
(0.1, 1.5)
(0, 1.5)
(0, 5.7)
(0. 52)
(3.4. 4.9)

(9.9. 3D
{5.7, 30)
(0.1. 20)
(0, 85)
(0.2, 41)
(0. 85)
(8.4. 20)
Group 8 Primary Chemicals
Numb jr
Screened

332J
187
1030
1263
13
33
587 J

65
121
5)
25
12
5
281
Number
Leaking
38
1
0
0
0
0
39

2
5
0
0
0
0
7
X
Leaking
1.4
0.5
0.0
o.o
o.o
0.0
0.66

3.1
4.1
0.0
0.0
0.0
0.0
2.5
95Z
Confidence
Interval

(0.8, 1.6)
(0.01,2.9)
(0. .36)
(0, .29)
(0, 11)
(0, 9.25)
(0.5, ,91)

(0.4, 11)
(1.3, 9.5) ,
(0, 6.7)
(0. 14)
(0. 26)
(0. 52)
(1.0, 5.0)

-------
               TABLE  3-9.   (continued)
OJ
Ethylene Procese Units
Group 3 Primary Chemicals
Valve
Type
Block
Gate
Globe
Plug
Ball
Butterfly
Other
TOTAL
Control
Gate
Globe
Plug
Ball
Butterfly
Other
TOTAL
Number
Screened

3125
45
3
19
A
94
3320

20
162
2
3
7
0
194
Number
Leaking

877
2
0
4
2
15
900

9
46
0
0
2
	
55
%
Leaking

28.
. 4.
0.
18.
50.
16.
27.

45.
28.
0.
0.
28.
__
28.

1
4
0
2
0
0
1

0
4
0
0
6
-
4


. 95X
Confidence
Interval

(26,
(0.5
(0,
(2.3
(6.8
(9.2
(25,

(23,
(22,
(0,
(0,
(3.7
—
(23,

30)
, 15)
71)
, 20)
, 93)
, 25)
29)

68)
3f>
84)
71)
, 71)
-
35)
Group 4 Primary Chemicals
HI mibc r
Screened

529
15
2
8
1
• JO
575

7
20
5
0
0
0
32
Number
Leaking

7
0
0
0
0
0
7

0
2
0
0
	
	
2
2
Leaking

1.3
0.0
0.0
0.0
0.0
0.0
1.2

0.0
10.0
0.0
	
	
	
6.2
95%
Confidence
Interval

(0.
(0,
(0,
(0,
(0,

-------
                          TABLE 3-10.  LEAK FREQUENCIES FOR ALL TYPES OF VALVES FOR GAS AND

                                       LIGHT LIQUID  STREAM SERVICE
UJ
Ln
Gaa Service
Type
Block
Gate
Globe
Plug
Ball
Butterfly
Other
TOTAL
Control
Gate
Globe
Plug
Ball
Butterfly
Other
TOTAL
Number
Screened
6976
145
440
1272
160
275
9268
61
207
10
15
91
17
401
Number
Leaking
952
15
0
18
9
16
1010
15
36
0
4
35
3
93
Percent
Leaking
13.7
10.3
0.0
1.4
5.6
5.8
10.9
24.6
17.4
0.0
26.7
38.5
17.6
23.2
95X
Confidence
Limits
(13. 15)
(5.9, 17)
(0, 0.8)
(0.7, 2.5)
(2.6, 10)
(3.5, 9.2)
(10, 11)
(14, 37)
(13, 24)
(0, 31)
(7.8, 55)
(28, 49)
(3.8, 43)
(19, 28)
Nuiaber
Screened
11017
755
:'.-'. 7 9
2732
157
378
17518
182
417
91
33
34
25
782
Liaht Liquid Service
Number
Leaking
1059
8
2
4
2
17
1092
22
61
3
1
3
1
91
Percent .
Leaking
9.6
1.1
0.1
0.2
1.3
4.5
6.2
12.1
14.6
3.3
3.0
8.8
4.0
11.6
95X
Confidence
Limits
(8.6, 9.8)
(0.5, 2.1)
(0.01, 0.3)
(0.04, 0.4)
(0.15. 4.4)
(2,7, 7.8)
(5.9, 6.6)
(7.8, 18)
(12, 19)
(0.7, 9.3)
(0.1, 16)
(0.7, 20)
(0, 20)
(10, 15)

-------
                                VALVES IN GAS SERVICE
I_1KJ
p
e
r
c 40
e
n
t
30
1

e
a 20
k
i

n
g 10


-
-
—
_
-

_
-



_
-
r —













i
— i
" LJ3-J
-
-
-

i
L_













i


i
_i
















*~
.
















— i ^.
. J




i — i




i








i _


t

1 — • i_

i
L '
, rf_, rm ' '
GA
TE
	 	 •*!
PLUG
B-F
GLOBE BALL
i __

	 RI nru'

"LY
GA
TE
r—






i —


— i

i

_i



* r
PL
— i
t








— ,



i




i — i

i



i — i
i



t




i— — i
i —









i




_j
i »
UG
B-F
"LY

OTHER GLOBE BALL OTHER


	 rnMTPni 	
Figure 3-3a.   Percent Leaking with 95 Percent  Confidence  Intervals  for Each Valve Type with
             Gas Service

-------
                               VALVES  IN  LIQUID  SERVICE
-
-
:S
—
-
-
: 3

i
GATE






c
• 'Hi I"T
PLUG



i
l

i
7-1 i —
B-F
GLOBE BALL

	 RI nr.K



r—
— , t

t_
i
_i
'LY
i—
i
-i "-
i —
~i
i
i t_
— i
i —
i
_j
r~
—i
r


L_
— i


1 i
_i
i -
GA
TE
PL
UG
r~
— i

i


i
L_
— > i —

i

t

— i
, — » '
B-F
"LY
— i



i


i

OTHER GLOBE BALL OTHER

:_ - ^- 	 — rnMToni 	 	
    p
    e
    r
    c
    e
    n
    t
    a
    k
    i
    n
15
Figure 3-3b.  Percent Leaking with 95 Percent Confidence Intervals for Each Type of Valve with
            Light Liquid Service

-------
effect on the leak frequency.  For valves in gas stream service, Table 3-8
shows control valves with a higher percent leaking than block valves for each
group.  The significant combined effect indicates that the difference between
block and control valves is significantly greater in the high leaking process
units than in the ethylene process units.

     The second analysis summarized in Table 3-11 is for light liquid
service.  The variables used here are the same as for gas service with the
addition of a category by primary material in the line and all of the two-way
combined effects.  All of the main effects and two of the combined effects
are highly significant.  The combined effect of primary material in the line
and block/control is also statistically significant.  Table 3-9 shows the
direction of these differences.  It can be seen from this table that the per-
centage of sources leaking for control valves from high leaking process units
with high leaking primary materials is about three times that of block valves
in the same group.  For the low leaking primary material group, it was about
four times.  Ethylene process units with high leaking primary materials in
the line had similar leak frequencies between block and control groups.

     Figure 3-4a and Figure 3-4b provide a graphical display of the differ-
ences .in leak frequency between the block and control valves for each process
and primary material category.

     Since the analysis found a significant different in leak frequency
between block and control valves, the comparison by specific type of valve
was done for each of these valve classifications.  Figures 3-3a and 3-3b
show leak frequencies for each valve type with 95 percent confidence intervals
for the leak frequencies.  The larger number of block valves tested makes
this group the easier one to examine for differences by valve type.  For
both gas and light liquid service, gate valves have the highest leak fre-
quency, and plug and ball valves have the lowest leak frequency.  Globe and
butterfly valves in light liquid service also have low leak frequencies,
The^e last two types of valves have comparatively wide confidence intervals
for gas service because of the small number of valves of those types found.
                                      38

-------
          TABLE 3-11. RESULTS OF CATEGORICAL ANALYSIS ON VALVES1
Source
Gas Stream Service"
Chemical Process
Block/Control
Combined Effects
Light Liquid Stream Service
Chemical Process
Primary Material
Block/Control
Combined Effects
Process by Material
Process by Block/Control
Material by Block/Control
Chi-Square
Statistic
655.40
19.44
13.81
207.1
60.1
25.4
64.9
15.7
4.4
Probability of
No Effect
<0.01
<0.01
<0.01
<0.01

-------
                                       VALVES IN GAS  SERVICE
O
             45

             40
          P
          e  35
          r

          •  30
          n
          t  25

          1  20
i
n  10
9
    5
i
                        [i
BLOCK     CONTROL

       Group  5
   P rimary Ch emic als
                                            BLOCK
                                            CONTROL
                                                    Group 6
                                                Primary  Chemicals
                                              BLOCK    CONTROL
                                                                  Group 1
                                                              Primary  Chemicals
      Figure 3-4a.
                    'High  Leaking Processes  '  —	'    '    '    Ethylene Processes
         Percent Leaking for Block Versus Control Valves in Gas Service by Primary
         Material Group and Process Unit Type
                 Note:  See Figure 3-2 for explanation of  primary material groups.

-------
r
                                               VALVES  IN  LIQUID  SERVICE
-TO
40
P
e 35
r
c 30
e
n ^
t
I 20
e
a 15
k
i
n '
9
5
-
-
-
-
-
-
_.
"
_,
i —
-
__
- i
-
t_
-
f nn



i —

T1
T
ul_i '

-. «-


I

_j

- I


r~
— i



1
(
_t

j 	 ,
t *""
i_i_i





— i





I



— j




                           BLOCK CONTROL  BLOCK CONTROL  BLOCK CONTROL  BLOCK  CONTROL
                               Group 7            Group 8
                           Primary Chemicals   Primary Chemicals
    Group 3
Primary Chemicals
    Group 4
Primary Chemicals
                                 High Leaking Processes
          Ethylene Processes
               Figure 3-4b.   Percent Leaking  for Block Versus  Control Valves in Light Liquid Service
                            by Primary Material Group and Process Unit Type

                          Note: See Figure 3-2 for explanation of primary material groups.

-------
For gas service, they appear to fall in the middle range between the high
leaking gate valves and low leaking plug and ball valves.  A comparison of
types of control valves is more difficult because of the small sample sizes.
The actual percent leaking for gate valves is higher than that of plug and
ball valves, but there are overlapping confidence intervals.  In the group
of control valves in light liquid service, globe valves show a significantly
higher leak rate than plug valves from the same group.
                                     42

-------
LEAK FREQUENCY FOR PUMP SEAL CLASSIFICATIONS

     Pump seals comprise a much smaller group of sources than valves.  For
this reason, the groupings by Process uti.it type and primary materials in the
line are not reported in this section.  When these subcategorizations were
examined, the small sample size in these categories resulted in such large
confidence limits that no statistical differences could be seen.  Combining
the categories did not effect any of the trends observed.

     The primary classifications for pump seals are on-line versus off-line,
single versus double seals, mechanical versus packed seals, and location of
the emission point.  Table 3-12 gives the number of pump seals screened, the
number leaking, the percentage leaking and appropriate 95% confidence inter-
vals for these classifications of pump seals.  On-line and off-line single
mechanical seals with emission point at seal are the two largest groups.

     A chi-square test (see Section 7 for details)  was performed to determine
if there was a statistically significant difference in the leak frequencies
between on-line and off-line pump seals when the emission point was at the
seal.  The outcome is below:
                      NO LEAK
                   Number     %
     LEAK
Number     %
       On-line
       Off-line
        Total
271 86.9
232 95.1
41 13.1
12 4.9
    503      90.5
Chi-Square  _ -,Q .
Statistic
 53
9.5
        TOTAL
                    312
                    244
556

     This test indicates that there is a significant difference between
on-line and off-line pump seals, with the leak frequency for off-line pumps
                                     43

-------
TABLE 3-12.  LEAK FREQUENCIES FOR PUMP SEALS IN LIGHT LIQUID SERVICE
On Line/
Off Line
On-Line








Off-Line








Mechanical/ Single/
Packed Double
Mechanical Single


Mechanical Double


Packed Single


Mechanical Single


Mechanical Double


Packed Single


Emission
Point
Seal
Vent
Other
Seal
Vent
Other
Seal
Vent
Other
TOTAL AT
THE SEAL
Seal
Vent
Other
Seal
Vent
Other
Seal
Vent
Other
TOTAL AT
THE SEAL
Numb e f
Screened
215
24
30
92
3
3
5
0
1 '
312
139
9
17
86
0
1
19
2
0
244
Number
Leaking
28
0
2
13
1
1
0
-
0
41
9
0
0
3
-
0
0
0
_
12
Percent
Leaking
13.0
0.0
6.7
14.1
33.3
33.3
o.o
_
0.0
13.1
6.5
0.0
0.0
3.5
-
0.0
0.0
0.0
-
4.9
95Z Confidence
Interval
(9, 20)
(0, 14)
(0.8, 22)
(7.7, 23)
(0.8, 91)
, (0.8, 91)
(0, 52)
-
(0. 100)
(9.0, 17)
(3.0, 12)
(0, 34)
(0, 20)
(0.7, 9.9)
-
(0, 100)
(0, 18)
(0, 84)
_
(2.6, 8.6)

-------
about one-third that of on-line pumps.
     A Chi-square test was also used to compare the leak frequency for the
single mechanical pump seals to double mechanical pump seals.  Separate tests
were performed for on-line and off-line seals.  Only data with the emission
point at the seal was considered.  Single packed pump seals had no leaks in
either case and so could not be included in the test.  The following table
describes this test:
     On-Line Pump
        Seals
   NOT LEAKING
Numb er      %
                        266
            86.6
     LEAK
Number      %
Single
Mechanical
Double
Mechanical
187 87.0
79 85.9
28 13.0
13 14.1
           Chi-Square Statistic = 0.07
                        p >0.10
                                                               215
                                                                 98
  41     13.4     307
     Off-Line Pump
        Seals
   NOT LEAKING
Number      %
   LEAKING
Number      %
Single
Mechanical
Double
Mechanical
130 93.5
83 96.5
9 6.5
3 3.5
                                                               139
                                                                87
                        240.        94.7       12       5.3     252
           Chi-Square Statistic = 0.94        p >0.10

-------
     The leak frequency for single mechanical versus double mechanical was
not significantly different for either the on-line or the off-line data.
Figure 3-5 shows this same information graphically.  Note that the presence
or type of barrier fluids was generally not recorded for this data.  This
may have been a factor in the lack of a significant difference between single
mechanical and double mechanical pump seals.
                                     46

-------
                                PUMP SEALS
  45
p
e
  35
r
c
e 30
n
t 25

1 20
e
a
k '5

n 10
9
   5
                                A-J-
                                     A-L
      SING-MECH
SING-PACK
                 DOUB-MECH

                 ON-LINE  -
           DOUB-MECH
SING-MECH           SING-PACK
                       OFF-LINE
    Figure 3-5.  Percent Leaking with 95 Percent Confidence Intervals for Each Type
              of Pump Seal with Light Liquid Service

-------
THE EFFECT OF LINE TEMPERATURE AND LINE PRESSURE

     The effects of line temperature and line pressure are examined in paral-
lel in this section.  The effects- of these two variables are evaluated for
the four major source types:  flanges, open-ended lines, valves,  and pump
seals.  It was found, that different levels of temperature and pressure are
present in the 24 units studied depending both on the type of chemical produced
and also on the primary material in the line.  The data are grouped by these
variables as they have been defined earlier in this report.  Appendix B
contains summary statistics in tabular form for line pressure and line temper-
ature for each of the groups.

     Since valves constituted the largest group by source type, the effects
of line temperature and line temperature on leak frequency for valves could
be studied in the greatest detail.  Categorical statistical analysis,
described in Section 7, was used to determine the significance on leak fre-
quency of line temperature and line pressure and their combined effects
(interaction).  This method of analysis is biased by empty cells (any tempera-
ture and pressure categories with no leaks).  As a result, the only groups
to be studied for possible combined effects were valves with high leaking
primary chemical groups (see Figure 3-2 for an explanation of groups).  The
results of this analysis are given in Table 3-13.

     The results of the categorical analysis show that for ethylene process
units with valves in gas service, both line temperature and line pressure,
and also their combined effect, were significant.  For the valves in light
liquid service, pressure and the combined effect of temperature and pressure
were significant.  Both of the groups from the high leaking process units
showed only pressure to have a significant effect on leak frequency.  Tables
3-14 to 3-17 give the data used in this analysis.  Figures 3-6 to 3-9 graph-
ically show the results of these analyses.  Figure 3-6 provides a good
example of significant combined effects (interaction) of line temperature
and line pressure.  It shows that the effects of increased pressure on the
percent leaking is not the same for all temperature groups.  If there was no
                                      48

-------
significant combined effect, the lines would be parallel.

     Tables 3-18 and 3-19 show the effects of line temperature and line
pressure on the valve for primary material groups 4, 6 and 8.  The categories
of temperature and pressure were chosen to agree with those in Appendix B
It appears that pressure may .have an effect on each of these three groups
(group 4, group 6 and group 8).  Temperature appears to have an effect on
valves from group 4.

     In summary, higher levels of pressure appear to result in higher leak
frequency in almost every instance.  For example, valves from Primary
Material Group 1 have a 4.1 percent leaking in the "less than 25 psig"
pressure group and 25.8 percent leaking in the "greater than 200 psig"
pressure group.  In those cases where this is not seen, it may be due to
the smaller sample sizes.  Temperature appears to be significant in only a
few cases.  In those cases, it was the middle range of line temperature
rather than the extremes that was associated with higher leak frequency.
Valves in gas service from ethylene process units had the greatest percent
leaking (16.2%) at temperatures between 0°F and 49°F.  The combined effects
of line temperature and line pressure could only be studied for valves.  It
had a significant effect for the ethylene process units only.  Higher leak
frequencies for high pressure and middle level temperature were found.
Figures 3-6 to 3-8 graphically show the effect of the interaction.  The
significant combined effect is apparent in the fact that the lines for
levels of line temperature are not parallel.
                                     49

-------
        TABLE 3-13.   RESULTS OF ANALYSIS OF THE EFFECTS OF LINE
                     TEMPERATURE AND LINE PRESSURE ON LEAK FREQUENCY

                     FOR VALVES
Process
- Unit
Group
Ethylene
Processes




High
Leaking
Processes



Primary
Material
Group 1
Group 1


Group 3


Group 5


Group 7



Stream
Service
Gas


Light
Liquid

Gas


Light
Liquid.



Source
Temperature
Pressure
Combined effects
Temperature
Pressure
Combined effects
Temperature
Pressure
Combined effects
Temperature
Pressure
Combined effects
Degrees
of
Freedom2
3
3
9
2
2
4
2
2
4
2
2
4

Chi-
Square
9.1
268.6
42.4
3.4
67.2
25.7
3.2
7.6
3.1
3.4
13.9
3.1

Signifi-
cance
A
AA
AA
n.s.
AA
AA
n.s .
A
n.s.
n.s.
A
n.s.
1See Figure 3-2 for explanation of groups.

2The total degrees of freedom for gas service in the ethylene process units
 is higher than the total for the other groups because, in this group, four
 levels of both temperature and pressure could be used without producing any
 empty cells in the analysis.
•s
 *probability of no significant effect is less than 0.05
 **probability of no significant effect is less than 0.01
 n.s.-no significant effect
                                     50

-------
TABLE 3-14.   LINE TEMPERATURE AND LINE PRESSURE AND THEIR COMBINED EFFECTS
             ON VALVES IN GAS SERVICE WITHIN ETHYLENE PROCESS UNITS
Temper atu
Pressure
-15-25
25-99
100-199
200-1050
TOTAL
re <°F>
Number
Screened
348
180
55
415
998
-267-0
Number
Leaking
1-8
30
12
80
140

Percent
Leaking
5.2
16.7
21.8
19.3
14.0

Hunter
Screened
506
449
104
392
1451
0-49
Number
Leaking
29
54
8
144
235

Percent
Leaking
5.7
12.0
7.7
36.7
16.2

Number
Screened
542
881
332
491
2246
50-99
Number
Leaking
14
96
71
146
327

Percent
Leaking
2.6
10,9
21.4
29.7
14.6

Number
Screened
244
276
350
714
1584
100-1570
Number
Leaking
7
IB
55
149
229

Percent
Leaking
2.9
6.5
15.7
20.9
14.5

Number
Screened
1640
1786
841
2012
6279
TOTAL
Number
Leakina
68
198
146
519
931

Percent
Leaking
4.1
11.1
17.4
25.8
14.8

-------
             TABLE 3-15.   LINE  TEMPERATURE AND  LINE  PRESSURE AND THEIR COMBINED EFFECTS
                             ON VALVES  FROM  GROUP  5*
Temper at ur
Pressure
(pslg)
-15-99
100-199
200-1050
TOTAL
e <°F)
Number
Screened
120
16
73
209
-267-99
Number
Leaking
8
4
10
22

Percent
Leaking
6.7
25.0
13.7
10.5

Number
Screened
141 •
83
91
315
100-149
Number
Leaking
8
7
14
29

Percent
Leaking
5.7
8.4
15.4
9.2

Number
Screened
187
118
236
541
150-1570
Number
Leaking
22
15
45
82

Percent
Leaking
11.8
12.7
19.1
15.2

Number
Screened
448
217
400
1065
TOTAL
Number
Leaking
38
26
69
133

Percent
Leaking
8.5
12.0
17.2
12.5
*See Figure 3-2 for explanation of groups.  Group 5 Is the high leaking primary chemical group from high leaking processes in gas service.

-------
                               TABLE 3-16.  LINE TEMPERATURE  AND  LINE  PRESSURE AND THEIR COMBINED
                                              EFFECTS ON  VALVES FROM GROUP 3*
Temperatur
Pressure
(psig)
-15-99
100-199
200-1050
TOTAL
e {op)
Number
Screened
683
65
1325
2073
-267-49
Number
Leaking
106
6
360
647

Percent
Leaking
15.5
9.2
2?. 2
31.2

Number
Screened
56
282
490
828
50-99
Number
Leaking
1
90
196
287

Percent
Leaking
1.8
31.9
40.0
34.7

Number
Screened
104
108
396
608
100-1570
Number
Leaking
16
14
163
198

Percent
Leaking
15.4
13.0
42.4
32.6

Number
Screened
843
455
2211
3509
TOTAL
Number
Leaking
123
110
724
957

Percent
Leaking
14.6
24.2
32.7
27.3
w
             *See Figure 3-2 for explanation of groups.  Group 3 is the high leaking chemical group from light liquid service from ethylene processes.

-------
                     TABLE  3-17.   LINE TEMPERATURE AND LINE PRESSURE AND  THEIR COMBINED



                                     EFFECTS  ON VALVES  FROM  GROUP 7*
Temper a tur
Pressure
(palg)
-15-99
100-199
200-1050
TOTAL
» CF)
Number
Screened
795
216
143
1154
-267-99
Hunter
Leaking
14
12
19
45

Percent
Leaking
1.8
5.6
13.3
3.9

Number
Screened
322
234
263
8.9
100-149
Number
Leaking
4
12
20
36

Percent
Leaking
1.2
5.1
7.6
4.4

Number
Screened
663
245
390
1298
150-1510
Number
Leaking
24
12
27
63

Percent
Leaking
3.6
4.9
6.9
4.8

Number
Screened
1780
695
796
3271
TOTAL
Number
Leaking
42
36
66
144
•
Percent
Leaking
2.4
5.2
8.3
4.4
*See Figure 3-2  for explanation of groups.  Group 7 Is the high leaking primary chemical group for high leaking processes in light liquid service.

-------
p
E
R
C
E
N
T

L
E
A
K
I
N
G
         40
32-
24-
16-
          8
          0
                  COMBINED EFFECTS OF TEMPERATURE AND PRESSURE

                  ETHYLENE PROCESS UNITS —  GAS SERVICE
                -15  to 25
                        26 to  99
100  to 199
200+
          ~  TEMPC<0:>
          ••' TEMPC0-49D
          ~'  TEMPC50-995
          — TEMPO 100)
                            PRESSURE
Figure 3-6.  Combined  Effects of Line Temperature and Line Pressure on Percent
            Leaking for Valves in Gas Service Within Ethylene Process Units.

-------
Ui
Os
R
C
E
N
T

L
E
A
K
I
N
G
                        30
                        24-
                         8"
                        12-
                         6-
                         0
                                     COMBINED EFFECTS OF TEMPERATURE AND

                                     PRESSURE ON  VALVES IN GROUP  5*
                                                            /  x
                                                                    S

                                                                            V    •
                                                                                 . *N
                                  -15  to 90
             TEMPO 00-149)
             TEMP01495
                                               1
                                         100 to  199
                                                    PRESSURE:
  1
200+
              Figure 3-7.   Combined Effects  of Line Temperature and Line Pressure on
                           Percent Leaking for Valves from Group 5*.
                *See Figure 3-2  for explanation of groups.

-------
                              COMBINED EFFECTS OF LINE  TEMPERATURE AND LINE PRESSURE

                              ON VALVES IN GROUP  3*
Ui
                              ss
                              11
                                                            100-199
>199
                                  TEMP
                                  TEMP CG0-99D
                                  TEMP 099)
             Figure  3-8.  Combined Effects of Line Pressure on Percent  Leaking for Valves in Group 3.*

              *See Figure 3-2  for explanation of groups.

-------
00
                     30
r
c
e
n
t

I
e
o
k
I
n
9
                     20
                      10
                                                         1
                                           CED
                                                                      nri
                        •15  to 49      100  to 249      -15  to 49      100  to 249
                                 50  to 99          250 «•         50 to 99          250 •*-

                         i	ETHYLENE PROCESS UNITS	"	HIGH LEAKING PROCESS UNITS	[
                             GROUP 4 Primary  Chemicals           GROUP 8 Primary Chemicals
                     Figure 3-9.  The Effect of  Line Pressure on Percent Leaking With 95 Percent
                                 Confidence Intervals for Valves from Group 4 and Group 8.*
                     *See  figure 3-2 for explanation of groups

-------
       TABLE 3-18.  EFFECT OF LINE TEMPERATURE AND LINE PRESSURE ON
                    VALVES FROM GROUP 6*
Pressure
-15-49
50-99
100-249
250-1050
Total
Temperature
-267-49
50-99
100-199
200-1570
Total
Number
Screened
1267
141
188
162
1758

45
335
823
555
1758
Number
Leaking
7
2
13
0
22

0
4
2
16
22
Percent
Leaking
0.5
1.4
6.9
0.0
1.2

0.0
1.2
0.2
2.9
1.2
95%
Confidence
Limits
(0.2, 1.1)
(0.2, 5.1)
(3.7, 12)
(0, 2.3)
(0.78, 1.9)

(0, 7.9)
(0.3. .3. 2)
(0.03, 0.9)
(1.7, 4;?>
(0.8, 1.9>
*See Figure 3-2 for explanation of groups
                                    59

-------
         TABLE 3-19.  EFFECTS  OF LINE TEMPERATURE  AND LINE PRESSURE  ON VALVES



                       FROM GROUP 4 AND GROUP 8 BY  PROCESS UNIT GROUP1
Group 41 PRIMARY CHEMICALS
Number Number
Pressure (psig) Screened Leaking
-15-49
50-99
100-249
250-1050
O Total
Temperature (°F)
-267-49
50-99
100-199
200-1570
Total
173 0
215 4
181 4
38 1
607 9
•
29 0
127 2
341 '6
110 1
607 9
95%
Number Confidence
Leaking Intervalu
0.0 (0, 2.1) ;.
1.9 (0.5, 4.8)
2.2 (0.8, 5,6)
' 2.6 (0.1, 14)
1.5 (0.7, 2.8)

0.0 (0, 12)
1.6 (0.2, 5.5)
1.8 (0.6, 3.8.1
0.9 (0.0, 5.0)
1<5 (0.7, 2.8)
Group 81 PRIMARY CHEMICALS
Number
Screened
2432
1567
1916
205
6120

96
1912
2583
1563
6154
Number
Leaking
16
5
18
7
46

2
11
16
17
46
Number
Leaking
0.7
0.3
0.9
3.4
0,8

2.1
0.6
0.6
1.1
0.8
95%
Confidence
Intervals
(0.4,
(0.1,
(0.6,
(1.4,
(0.5,

(0.3,
(0.3,
(0.4,
(0.6,
(0.5,
1.1)
0.7)
1.5)
7.1)
1.0)

7.3)
1.0)
1.0)
1.7)
1.0)
*See Figure 3-2 for explanation of groups

-------
EFFECTS OF LINE TEMPERATURE AND LINE PRESSURE ON PUMP SEALS, FLANGES, AND
OPEN-ENDED LINES

     There was not enough data available to study the possible combined
effect of line temperature and line pressure for the remaining source types,
and so the effects of these two variables were examined separately.   Catego-
ries were used that would conform to earlier tables and at the same  time
provide an approximately even distribution of sources screened.

     Table 3-20 shows this information for pump seals for both groups of
process units and all primary materials.  The 95 percent confidence  intervals
indicate that no significant effects of temperature or pressure can  be seen.
If the overall number screened was increased, the size of the confidence
intervals would be decreased and it is possible that some significant dif-
ferences might then be seen.  Figure 3-10 shows the percent leaking  with
95 percent confidence intervals as a function of pressure for this source
type.

     Tables 3-21 and 3-22 give the leak frequencies by line temperature and
line pressure for flanges in gas and light liquid services, respectively.
Increasing levels of line pressure result in increased leak frequency.  The
effect is most clear for gas service streams.  These data are presented by
process unit group since for the light liquid service, there appears to be
some differences between these two groups.

     Tables 3-23 and 3-24 show the leak frequency of open-ended lines by
line temperature and line pressure.  Ethylene process units services in gas
service show a higher leak frequency at the highest level of pressure.
Otherwise the gas service show overlapping confidence intervals.  Open-ended
lines in light liquid service within ethylene process units show an  increased
leak frequency at higher pressure levels and also a higher frequency at the
upper two pressure levels when compared to the high leaking process  units.
The leak frequency from the high leaking process units does not appear to
                                     61

-------
be affected by line temperature or line pressure.   Line temperature does
not appear to have an effect on open-ended lines within ethylene process
units either.
                                     62

-------
     TABLE 3-20.  EFFECTS OF LINE TEMPERATURE AND  LINE  PRESSURE ON PUMP
                  SEALS WITH LIGHT LIQUID SERVICE
Pressure
(psig)
-15 - 49
50 - 99
100 - 249
250 - 1050
Number
Screened
146
115
116
65
Number
Leaking
10
19
11
12
Percent
-Leaking
6.8
16.5
9.5
18.5
95% Confidence
Intervals
(3.3, 12)
CIO, 25)
(4.9, 16)
(9.9, 30)
     Total        442

Temperature  (°F)

-267 - 49          53
  50 - 99         148
 100 - 199        146
 200 - 1570       100

     Total        447
52
12
14
13
_i
48
11.8
22.6
 9.5
 8.9
 0.9

10.7
(9.0,  15)
(12, 36)
(5.2, 14)
(4.8, 14)
(4.2, 16)

(8.0, 14)
                                       63

-------
                               PUMP SEALS —  LIGHT  LIQUID
           35
        P 30
        e
        r
        c 25
        n
        t 20
e  ' ">
o
k
i  '
n
9  5
                      ii
                                      u
                  15  to 49       50 to  99      100  to  249

                                       PRESSURE
                                                            250 •*•
Figure 3-10.  The Effect of Line Pressure on Percent Leaking with 95 Percent Confidence Intervals
             on Pump Seals in Light Liquid Service

-------
             TABLE 3-21.  EFFECTS OF LINE TEMPERATURE AND LINE  PRESSURE ON FLANGES
                          IN  GAS SERVICE BY PROCESS UNIT GROUP
1 Ethylene Process Units
Number
Pressure (pain) Screened
-15-49
50-99
100-249
250-1050
o\
Ln
TOTAL
Temperature (°F)
-267-49
50-99
100-199
200-1570
210
102
136
182
630

129
335
155
15
Number
Leaking
3
4
8
24
39

17
15
5
2
Percent
Leaking
1.4
3.9
5,9
13.2
6.2

13.2
4.5
3.2
13.3
95%
Confidence
Intervals
(0.3,
(1.1,
(2.5,
(8.7,
(4.4,

(7.8,
(2.6,
(1-0,
(1.7,
4.3)
9.6)
11)
19)
8.0)
•
20)
7.8)
7.4)
41)
Number
Screened
301
76
117
217
711

16
99
268
321
High Leaking Process Units
Number
Leaking
6
2
2
17
27

0
1
4
22
Percent
Leaking
2.0
2.6
1.7
7.8
3,8

o.o
1,0
1.5
6.8
95*
Confidence
Intervals
(0.
to.
(0.
(4,
(2.

(0,
(0.
(0.
(4.
7, 4.3)
3, 9.2)
2, 6.1)
8, 13)
5, 5.5)

21)
5.5)
4, 3.7)
4, 10)
TOTAL
634
39
                                   6.2
(4.4, 8.0)
704
                                                                  27
3.8
(2.5,  5.5)

-------
TABLE 3-22.  EFFECTS OF LINE TEMPERATURE AND LINE PRESSURE ON FLANGES
             IN LIGHT LIQUID SERVICE BY PROCESS UNIT GROUP


Number
Pressure (psig) Screened
-15-49
50-99
100-249
250-1050
TOTAL
Temperature (°'F)
-267-49
50-99
100-199
200-1570
TOTAL
70
52
74
200
396

133
141
121
9
404
Ethylene
Number
Leaking
1
0
5
19
25.

13
10
2
0
25
Process Units
Percent
Leaking
1.4
0,0
6.8
9.5
6.31

9.8
7.1
1.7
o.o-
6.2
95%
Confidence Huriber
Intervals Screened
(0,
, (0,
(2.
(5.
(4.

(5.
(3.
CO.
(0,
(4.
7.7) !i83
6.8) 364
2, 15) 372
8, 14) 27<<
1, 9.1) 1596

3, 16) 2£
5, 13) 493
2, 5.8) 638
34) 452
1, 9.1) 1607
High Leaking Process
Number
Leaking
0
0
2
8
10

' 0
2
6
2
10
Percent
Leaking
0.0
0.0
0.5
2.9
0.6

0.0
0.4
0.9
0.4
0.6
Units

95%
Confidence
Intervals
(0,
CO,
CO.
(1-
(0.

CO,
Co.
(0.
(o.
(0.
0.7)
1.1)
1, 2.0)
3, 5.7)
3, 1.2)

13)
1, 1.4)
3, 2.1)
1, 1.6)
3, l.D

-------
TABLE 3-23.  EFFECTS OF LINE TEMPERATURE AND LINE PRESSURE ON OPEN ENDED
             LINES IN GAS SERVICE BY PROCESS UNIT GROUP


Number
Pressure (psig) Screened
-15-49
50-99
100-249
250-1050
TOTAL
Temperature < °F)
-267-49
50-99
100-199
200-1570
TOTAL
160
53
35
55
303

173
96
31
5
305
Ethylene
Number
Leaking
6
7
3
20
36

16
17
4
0
37
Process Units
Percent
Leaking
3.8
.13.2
8.6
36.4
11.9

9.2
17.7
12.9
0.0
12.1
95%
Confidence
Intervals
(1.
(5.
(1.
(24
(8.

(5.
(11
(3-
CO,
(8.
4, 8.0)
5, 25)
8, 23)
, 50)
6, 16)

5, 15)
,0, 27)
6, 30)
52)
6, 16)
Number
Screened
272
59
74
51
456

13
83
225
134
455
High
Number
Leaking
5
4
1
4
14

0
8
5
1
14
Leaking Process Units
Percent
Leaking
1.8
6.8
1.3
7.8
3.1

0.0
9.6
2.2
0.8
3.5
95%
Confidence
Intervals
CO
(i
(0
(2
(1

(0
(4
(0
(0
(1
.6, 4.3)
.9, 16)
, 7.3)
.2, 19)
.7, 5.8)

, 25)
.2, 18)
.7, 5.1)
, 4.1)
.7, 5.8)

-------
                    TABLE 3-24.  EFFECTS OF LINE TEMPERATURE AND LINE PRESSURE ON OPEN ENDED
                                 LINES IN LIGHT LIQUID SERVICE BY PROCESS UNIT GROUP
CO
Ethylene Process Units
Pressure (psig)
-15-49
50-99
100-249
250-1050
TOTAL
Temperature ( F)
-267-49
50-99
100-199
200-1570
TOTAL
Number
Screened
30
48
38
98
214

75
56
62
21
214
Number
Leaking
0
2
7
32
41

15
13
11
2
41
Percent
Leaking
0.0
4.2
18.4
32.6
19.2

20.0
23.2
17.7
9.5
19.2
95%
Confidence
Intervals
(0,
(0.5
(7.7
(24,
(15,

(12,
(13,
(9.2
(1.2
(15,
12)
, I*)
, 34)
«)
26)

31)
36)
, 30)
, 30)
26)
Number
Screened
813
479
474
125
1891

56
668
685
^88
1897
High Leaking Process Units
Number
Leaking
27
22
14
3
66

0
28
24
14
66
Percent
Leaking
3
4
2
2
3

0
4
3
2
3
.3
.6
.9
.4
.5

.0
.2
.5
.9
.5
9^2
J Jn
Confidence
Intervals
(2.2,
(2.9,
(1.7,
(0.5,
(2.7.

(0, 6.
(2.8,
(2.4,
(1.6,
(2.7,
4.7)
6.8)
4.9)
6.9)
4.4)

4)
5.9)
5.2)
4.8)
4.4)

-------
EFFECT OF AMBIENT TEMPERATURE ON LEAK FREQUENCY

     This section evaluates the effects on leak frequency of the ambient
temperature.  The ambient temperature was measured at the same time that the
source was screened.  Ambient temperature was measured as a continuous
variable, but to evaluate its effect on leak frequency, it was grouped as
less than 70 F or greater than or equal to 70°F.  Appendix C contains summary
statistics for this variable.

     Statistical tests were performed for each primary material group to
determine if there was a significant difference in leak frequencies between
the two classifications of ambient temperature.

     Table 3-25 gives a summary of the effects of ambient temperature on
leak frequencies of sources.  In those cases where the percent leaking was
not affected by the primary material in the line or the type of process unit
or both, groups were combined.  Those groups that did show a significant
effect of ambient temperature are noted with asterisks.  In one case (open-
ended lines in gas service from high leaking process units) a significant
difference in leak frequencies was seen when the primary material groups were
combined (but the differences were not significant when they were separated) .

     Overall, ten of the 25 groupings of sources showed a statistically
significant effect of ambient temperature on the leak frequency.  Four of
the seven comparisons for valves were significant.  Generally higher leak
frequencies were associated with the high ambient temperature classification.
Differences in leak frequencies between the two ambient temperature categories
range from three percent leaking to 14 percent leaking.
                                     69

-------
           TABLE  3-25.    SUMMARY OF  THE  EFFECTS OF AMBIENT TEMPERATURE ON PERCENT LEAKING
Source
Type
Valves




Pump
Seals
Flanges

Open Ended
Lines



Stream
Service
Gas

Light
Liquid


Light
Liquid
Gas
Light
Liquid
Gas


Light
Liquid
Process
Group
E thy I en e
High
Leaking
Ethylene
High
Leaking

Both
Both
Both
Ethylene
High
Leaking

Both
Primary
Material Ambient
Group Temperature
Group
1
Group 5
and Group 6
Group 3
and Group 4
Group
-Group
Group
Group
Group
Group
Group
Group
Group
7
a
3,4,
7,8
1,5,6
3,4,
7,8
1
5
6
3,4,
7,8
<70°
70"+
<70"
70°+
<70°
70e+
<70*
70°+
<70"
70°+
<70"
70°+
<70"
70°+
<70°
70"+
<70°
70"+
<70"
70°+
<70°
70e+
<70°
70°+
Number
Screened
3760
2534
1591
1402
1906
2215
2435
803
2861
3293
245
202
288
1057
457
1576
223
82
71
75
204
143
1288
1015
Number
Leaking
474
460
67
J01
448 .
518
52
95
17
29
21
27
14
52
7
28
19
18
4
9
1
3
84
42
Significant
Percent Effect of
Leaking Temperature
12.
18.
4.
7.
23.
23.
2.
11.
0.
0.
9.
13.
4.
4.
1.
1.
8.
22.
5.
12.
0.
2.
6.
4.
6 **
2
2 ,
2 **
5
4
1 **
8
6
9
0
0
7
9
5
6
5
0 **
6
0 2
5
1
5
1
952 Confidence
Interval
(11.
(17,
(3.3
(5.8
(21,
(21,
(1.6
(9.6
(0.4
(0.6
(5.3
(9.1
(2.6
(3,7
(0,6
(1.2
(5.2
(14,
(1.6
(5.6
(0,
(0,5
(5.2
(3.0
14)
20)
, 5.4)
, 8.7)
26)
26)
, 2.8)
, 14)
, 1.0)
, 1.2)
, 13)
, 19)
, 7.8)
, 4.3)
, 3.2)
, 2.6)
. 13)
32)
, 14)
, 22)
2.8)
, 6.1)
, 8.1)
, 5.6)
1-See Figure 3-2 for explanation of groups.
2-Showed significance when groups 5 and 6 were combined.
**Probabillty of no difference in leak frequency between ambient temperature categories is less than one percent.

-------
EFFECT OF ELEVATION ON LEAK FREQUENCY

     This section evaluates the effect of source elevation on leak frequency.
The elevation of each screened source was recorded at the time of screening.
This elevation was expressed as the process unit landing level closest to
the screened source.  For analysis in this report, the source elevation was
categorized as either ground level or above ground.  The data is presented
in the same format as that for ambient temperature.  Appendix C contains
summary statistics for elevation and the results of all statistical tests
on the elevation categories.

     Table 3-26 gives the same type of summary for the effects of elevation
that was given for ambient temperature.  Groups of process units and primary
materials were combined wherever the effect of elevation was consistent.

     Only five of the 25 source type/primary chemical groups evaluated
indicated a significant effect of elevation on leak frequency.  In all of
these cases, the sources at ground level had a higher leak frequency than
the sources above ground.  The differences between the elevation categories
for those groups ranged from 1.7 to 6.0 percent leaking.
                                    71

-------
                                   TABLE 3-26.   SUMMARY  OF  THE  EFFECTS  OF  ELEVATION ON  PERCENT  LEAKING
ro
Source Stream .
Type Service
V«lvea Gaa


Light
Liquid


pump Light
Seals Liquid
Flanges Gas
Light
Liquid
Open Ended
Lines Gas


Light
Liquid


Prln*ry
Proceaa Material
Group Group
Ethylene Group 1
High Group S
Group 6
Ethylene Group 3,4
High Group 7
Group 8
Both Group 3,4
7.8
Both Group 1,5.6
Both Croup 3,4,
7,8
Ethylene Croup 1
High Group 5
Group 6
Ethylene Group 3,4
Htgh Group 7
Group 8
Elevation
Ground .
Abova
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Ground
Above
Number
Screened
3298
3977
479
• 749
423
1333
3123
1041
2494
795
4394
1743
437
10
481
863
1414
610
235
69
59
87
72
274
163
51
623
172
949
340
Number
Leaking
475
453
54
92
12
10
727
238
121
25
35
10
48
0
22
44
25
9
25
12
6
7
2
2
31
10
45
2
29
9
Significant
Percent Effect of
Leaking Elevation
14.4
15.2
11.3
12.3
2.8 **
0.7
23.3
22.9
4.8 *
3.1
0.8
0.6
11.0
0.0
4.6
5.1
1.8
1.5
10.6
17.4
10.2
8.0 2
2.8
0.7
19.0
19.6
7.2 »*
1.2
3.1
2.6
95 J Confidence
Interval
(13, 16)
(14, ">
(8.5, 14)
(10, 14)
(1.4, 4.9)
(0,4, 1.4)
(22. 25)
(20, 26)
(4.0, 5.8)
(2.0, 4.6)
(0.6, 1.1)
(0.3, 1.0)
(B.3, 14)
(0, 3D
(3.0, 7.8)
(3.7. 7.0)
(1.1. 2.6)
(0.7, 2.8)
(7.1. 16)
(9.3, 28)
(3.8, 21)
(3.3. 16)
(0.3, 9.7)
(0.1, 2.6)
(13, 2«)
(9.8, 33)
(5.3, 9.4)
(0.1, 4.1)
{2.0. 4.4)
(1.2, 4.9)
                                1-See Figure 3-2 for explanation of groups.
                                2-Therc wao a  Qigniflcant difference between elevation categories when chemical groups 5 and 6 were combined.

                                 *Probability  of no difference in leak frequency between elevation categories Is less.than five percent.
                                **Probablllty  of no difference in leak frequency between elevation categories is less than one percent.

-------

-------
                                  SECTION 4
               EMISSION FACTOR DEVELOPMENT FOR THREE PROCESSES

     This section presents detailed results of the investigations in the
following three areas:

     *  Distribution of screening values,
        Estimation of emission factors, and
        Mass emission distribution over the range of screening
        values.

DISTRIBUTION OF SCREENING VALUES

     Distributions of OVA screening values were examined for each process,
source type (valve or pump seal) and service.  From past experience with
the refining industry, it was expected that the distributions of the nonzero
screening values could be modeled with a lognormal distribution.  It was
anticipated that censoring above 100,000 ppmv would occur due to the inconsis-
tent use of a secondary OVA dilution probe.  Figure 4-1 shows a typical
histogram of the logarithms of the nonzero screening values with a pattern
that frequently occurred:  a large number of observations nominally at 100,000
ppmv with positive skewness (more large values occurring than expected from
a normal distribution) and negative kurtosis (flatter peak and shorter tails
than a normal distribution).  Further examination of the distributions by
primary material classification showed similar departures from the lognormal
distribution.  Two approaches were subsequently taken in modeling the distri-
bution of screening values:  fitting an empirical cumulative distribution,
which reflects detailed features of the data, and fitting a cumulative log-
normal distribution to the nonzero screening values, with adjustment for
censoring of the data.  Section 7 contains a more detailed discussion of these
                                     73

-------
distribution models.  Figures 4-2 through 4-10 compare the lognormal models
with the empirical distributions.  The departure from lognormality of the
screening data does not appear large in magnitude.  The lognormal model was
therefore used in the development of both screening value distributions and
distributions of mass emissions.

EMISSION FACTORS AND CUMULATIVE DISTRIBUTIONS OF TOTAL EMISSIONS BY
SCREENING VALUES

     This section briefly describes the estimated emission factors and mass
emission functions.  A more detailed discussion of the statistical methods
and assumptions employed is found in Section 7.

     Table 4-1 presents estimated emission factors for nonmethane hydrocarbon
fugitive emissions from valves and pump seals.  Figures 4-11 and 4-12 show
graphically how these emission factors compare between the three processes
considered in this report.

     Comparison of emission factors among the three processes by means of
their related confidence intervals shows only one difference that can be
considered to be statistically significant:  ethylene has a significantly
larger emission factor than vinyl acetate for valves with light liquid service
Note, however, that ethylene consistently shows the largest emission factor,
followed by cumene and vinyl acetate.  For pump seals, ethylene and cumene
have about the same emission factor.  With the exception of vinyl acetate,
pump seals have larger emission factors than do valves.  Finally, for com-
parable sources, gas service has higher emission factors than light liquid
service.

     Fugitive emissions may also be compared by means of the cumulative
distribution of total mass emission by screening value.  These curves relate
the OVA screening value to the percentage of the total mass emission which
can be expected from all sources with screening values greater than any given
                                     74

-------
value.  These cumulative functions have been estimated for each process,
source type, and service.  Figures 4-13 through 4-21 display the cumulative
mass emission estimated by an empirical function, with the lognormal model
superimposed for comparison.  Both the lognormal model and the empirical
function are described more fully in Section 7.

     Confidence bounds are given to indicate how well the cumulative mass
function has been estimated from the data collected in both the screening
and maintenance programs.  The development of these intervals is discussed
in Section 7.  In using  these estimated functions and confidence intervals,
it should be kept in mind that the relationship between screening values and
mass emissions is imperfect.  Also, the true distribution of screening values
is not known precisely:  it is estimated from the observed screening value
distribution.  These two sources of variation contribute to the width of the
confidence bands shown in the figures.

     Figures 4-22 through 4-30 show the cumulative distribution functions for
screening values (Part a) and mass emissions (Part b) based on the lognormal
model for the screening  values.

     Application of Figures 4-22 through 4-30 may be illustrated through the
use of Table 4-2, which  exhibits point estimated and 95% confidence intervals
for both the percentages of sources screening ^ 10,000 ppmv and the percentage
of total mass emissions  attributable to sources screening ^ 10,000 ppmv.  For
example, approximately 15% of ethylene process valves in gas service can be
expected to have screening values above 10,000 ppmv (Figure 4-25a).  However,
these 15% of the valves  are responsible for an estimated 94% of the mass
emissions (Figure 4-25b).  In the same manner, other specific screening
values (or percentage of sources) could be chosen and the corresponding
percentage of mass emissions found.
                                      75

-------
bno
2 Oil
inr
              NOTE:
Log (10) = 2,30  '
Log (100) = 4.61
Log (1000) = 6.91
Log (10,000) = 9.21
Log (100,000) = 11.51
Log (1,000,000) = 13.82
I
1
1
t-
1
1
1
1
1
I
1
t
1
1
I
1
+
1
1
1
1
•t
1
1
1
1


*****
*****
•1 t * * *•
*****
-t **+•*
M 1 t * -t t +- 1 *
t * t * * t****
*+M* *****
***** •«*•***
**.!** J t * * *
** * **
*****
*****
*****
*****
***» * •
***.* +
*****
*****
*****
******
** *** '
*****
** * **
*****
*** **
*****
*** t *
*****
*****
*****
**•** *
*****
*****
**•»-* -I
*****
t** **
* *Jt **
*.*» * t
*****
*****
»,*-, *• *
*****
*** * *•
*****
*****
*****
**»* *
*****
** * 4 *
*****
* 4 ^ t *
*****
**** *
*****


*****
*****
*****
*****
*****
*****
* ****
* * * ^* *
*****
*****
*****




*****
*****
*****
*****
** ***
*****
* * * * #
*****
*****
*****


*****
*****
*****
*****
** ***
*****
*****
*****
*****
*****
*****
*****






**-***
*****
** ***
**•«*:*:
*****
*****
*****







*****
*****
*****
*****
*****
*****
* ****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*•**.**
*****
' *****
*****
*****
*****
*****










*****
***** *****
                                                                 10
                                                                       11
                                                                              12
                                LOGP  (OVA  SCREENING VALUE)
               Figure  4-1.   Typical Distribution of Log  (OVA Screening Value)

                            Ethylene Process,  Valves in Gas Service

-------
R    i
C 35  +
El
N    I
T    I
  30  +
Si
C
     I
R    t
E 25 +
El
Ml
II
W 20 +
C    I
     I
G    1
R 15 +
El
A    I
T    I
E 10 +
     I
R
              * *
              x x ***
                  xxx

                         xxx
                         **x  x
                               x '
                               * x
                               * *
                                   X x , ,
                                     XX..  .
                                 *    XXX
                                  *****    X  ....
                                      +*     X
                                       *  *  XXX ., .
                                 ..           XX ...
                                   . . . ,      ****XXX  . ,
                                      ..        * XXX.
                                       .  .  ,    **  XX
                                            ,     ***
                                            ...     *
                                                                   x = lognormal  model
                                                                   * = empirical  function
                                                                   * = 95% confidence limits
                                                                       for empirical function
                                                       XX  . *
                                                       * XX  . . .
                                                       **  X
                                                        *** XX   ....
                                                          * *  XX     . .
                                                             **  X
                                                       ...      **  X X
                                                                  * *XX
                                                                      XX *
                                                                         X
1.0
                 1.5
2,0
2.b
i.O
   3.5      4.0

SCREENING VALUE)
                                                                  H,5
                                                                           5.0
                                                                                   5.0
                                                                                           t>.U
           Figure 4-2.  Cumulative Distribution of  Sources by Screening Values -
                        Cumene Process, Valves in Gas  Service

-------
  35
P
E 30
R
C
E
M
T 25

S
C
ft
E 20
E
N
I
N
& 15

G
H
E
A 10
T
E
R

   5
   0 +•
*      . •
X X*
  xxx***  ...
     xxxx
. .      XXX ..
   ....   *xx ..
      ..     XX ..
       ...  ** XX .
         .i  **  XX.
           .   **  XV,
           . .      XXX
             ..   **   XX. . .
                   *   xx..
               .<   **   X *.
                     *+** xx,.
                        *  xx..
                   ...    *  XX. ,
                          ***  xx..
                      ...   ** X  ...
                         .    ** XX  .,
                          ...   ** xx
                                  + *xx
                                     *xxx...
                                       *xx .  ..
                                        * XX
                                     *   ***  X
                                      ..   *  *x
                                        .    *
                                           X = lognormal  model
                                           * = empirical  function
                                           • = 95%  confidence limits
                                               for  empirical function
                                                x** **
                                                 XX   ****
                                                    xxxx
                                                       XXX
l.o
i.a
a.o
                      i.o
                                                  i.b
t.b
6.U
              Figure 4-3.  Cumulative Distribution of  Sources by Screening Values  -
                           Cumene  Process, Valves in Light Liquid Service

-------
64
60
p
E
R
C 56
E
M 52
36
  44
R
E 40
E
N
I
N 32
G
  28
G
R 24
E
A 20
T
E 16
K
  12  +
     I
   8  -f
     I
                                                                      lognormal model
                                                                      empirical function
                                                                      95% confidence limits
                                                                      for empirical function
                                              x *
                                                X
 X

 *

5.0
        1.0
                        2,U
                                         3.0
                                               3.b     4.0

                                            SCKELNING  VALUE)
4.5
5.3      b.U
         Figure 4-4.  Cumulative Distribution  of  Sources by Screening Values
                      Cumene Process, Pumps in Light  Liquid Service

-------

45

40
P
t
R
C 35
E
N
T
JO
S
c
R
E 25
E
N
I
Co N 20
0 G
U
R 15
(,
A
T
E 10
K

5



0


XX. •*••..
, XXXXK+**..
jXxxx ^ = l°gnormal model
..xxx * = empirical function
•**x*x • = 95% confidence limits
± .*. « *t y y v
xxx for empirical function
«... xxx
.**** . xxxx
..*** AX
	 xxx
****.. XX
«.«***.. XXX
* w Y • T 4 M «t rt A
.***.. XX
•..**... XX
.»***,, XX
...***.. XX
..***.. xxx
*.***•*. *xx
.,****, ..XX
*'"!!!!I**xx.
. ,***tx. .
...**XKK.*.*
. ..*xxx*. ....
....•OK***. 	
* t*XXt******
XXA...4
xxx.
xxx
XX
	 XXX AX
**** * * 	
	 ** *** «»
l.C 1 . iJ «!,U 2,t> 4.0 3.5 t+.O 4.5 5.0 5.3 fa.U
Figure 4-5.  Cumulative Distribution of Sources by Screening Values  -
             Ethylene Process4  Valves in Gas Service

-------
00
H1
               70
P
L 60
R
C
E
N
T 50

S
C
R
E 40
E
N

N
G 30

G
K

A 20
T
E
f<

  10
                      * ** ***  ,4,,

                      X XX XXX  X***...
                     1.0
                 l,b
                                          ,.****,XXX*
                                            ...,**..xxxx
                                                        ...**#. xxx
                                                                                X =  lognormal model
                                                                                * -  empirical function
                                                                                • =  95%  confidence limits
                                                                                     for  empirical function
                                                                ,.****xx
                                                                  ...****xx
                                                                     ...*«xxxx
                                                                        .****.xx.
                                                                         ...»***!**.,.
                                                                            	»**«X	
                                                                                              X X
                                                                                                  X  X
                                                                                                      xxx
                                                                                          .  ..••   •*.*
                                                                                          t  * * * *   **#*
._•!_.
 2.0
2.i>     3.0

     L.OG1UIOVA
                                                               3,5
  1.0

VALUE)
                                                                   H,5
                                                                            5.0
                                                           5.3      6.U
                         Figure 4-6.  Cumulative Distribution of  Sources by Screening  Values -
                                      Ethylene Process, Valves  in Light Liquid Service

-------
CO.
              70
            E 60
            R
            C
            E
            N
            T 50
S
C
il
E HO
E
N
I
i-J
G 30

6
R
E
A 20
r
E
              10
                     *
                   X X*
                      X
                                      X
                                      *  *
                                          *   XX
                                                  *    X
                                                   *  *  X
                                                         *x
                                                         *  *     »
                                                            *xx
                                                             *
                                                                X
                                                              *  XX
                                                                      X
                                                                    *   x
                                                                 X = lognormal  model
                                                                 * = empirical  function
                                                                 • = 95%  confidence limits
                                                                     for  empirical function
                    i.o
                1.3
2.0
3.0
                        3.b
                                                         4.0
                                                                                      5.0
                                                                                          fa.U
                           Figure 4-7.  Cumulative Distribution  of  Sources by Screening Values -
                                        Ethylene Process, Pumps  in  Light Liquid Service

-------
17.5
E  ib.o
R
C
E
N
T  12.5 -f
S
C
R
E  10,0
E
N
I
N
G   7.5

G
H
E
A   5,0
T
E
R

    2,5 *
       XX**
            X  X *
       .  .     x
 0.0
                 XX,.
                   *XX
                    * XXX.
                 .   *   X
                      **
                              . .
                             XXX
                                 If
                             *     ».
                              **  .  xxx
                               *     . .x
                                 *    .XX
                                   **
                              . *     *.
                                *,   **
                                        **      X      .  *
                                           * *     XXX
                                             *   * *   x
                                                   ** *x
                                        ..             *  X
                                                                    = lognormal model
                                                                    = empirical function
                                                                    = 95% confidence limits
                                                                      for empirical function
                                                           XKX
                                                                   ***   *
                                                                   xxx
                                                                        X
                                                                                        • - + -.
                                                                                         b.U
                                       3.0

                                    LOGlO(UV/\  y
          Figure  4-8.   Cumulative Distribution of  Sources  by Screening Values -
                        Vinyl Acetate Process, Valves in Gas  Service

-------
00
p
E 6
R
C
E
N
T 5

S
C
R
E 1
E
N
I
W
ti 3

G
R
E
A 2
T
E
R

  1
                     **

                     XX
                    1.0
                         X M
                              X*
                                 X.
                                 X!
 X.
  X
[*   X i
 *   X  .
  *      .
       X
.   **    X
 *     *
  I      *

    ..    *
                                            XX
              *+x
                 X   ,,
                * X     .
                 **  XX   .,
              •.    * X   ..
                     ** XX
                         X
                       ***x
                  .       **
                                            X = lognormal  model
                                            * = empirical  function
                                            • = 95%  confidence limits
                                                     empirical function
                                                                      ..
                                                                *  xx  *
                                                                      X    *   *
                l.b
                                    2.0
                    4.0      3.5      4.0

                 IU&1UIOVA sCKEENlNf, VALUE)
                                                                  . 5
.- + _.
5.0
•- + -•
b.5
.- + -.
d.U
                        Figure 4-9.  Cumulative Distribution of Sources by Screening Values  -
                                     Vinyl Acetate Process,  Valves in Light Liquid Service

-------
00
27.
25.
24.
23.
22.
21.
20.
19.
Id.
16.
15.
1H.
13.
12.
11.
10.
 9.
 7.
 6.
 5.
 4.
 3.
 2.
 1.
 0.
oon
875
750
625
500
375
250
125
000
875
750
625
500
375
25n
125
000
875
750
625
500
375
250
125
000
                                                                              X = lognormal  model
                                                                              * = empirical  function
                                                                              • = 95% confidence limits
                                                                                  for empirical function
                       K  X
                          * X
                       i.o
*     X
  *     X
      *   X
       *    X
         *
            *
                                             X  X
                                               * X
                                                     XX
                                                     *
                                                      *   X
                                                          *   X
                                                             *
                1.5
   3.U      3.5

UUdlOtOVA
 4.0

VALUt)
                                                               H.5
                                                             5.0
                                                                                                          6.U
                       Figure 4-10.
                      Cumulative Distribution o£  Sources by Screening Values
                      Vinyl Acetate Process, Pumps  in  Light Liquid Service

-------
                        TABLE 4-1.  ESTIMATED EMISSION FACTORS FOR NONMETHANE HYDROCARBONS
                                    FROM VALVES AND PUMP SEALS
       Source Type
                                                             Emission Factor  (Confidence^
                         Number Screened
                        (Ibs./hr./source)
                                 (kg,/hr./source)
00
Valves
  Gas
    Ethylene
    Cumene
    Vinyl Acetate
  Light Liquid
    Ethylene
    Cumene
    Vinyl Acetate
6,294
  448
  949
 i
4,176
  799
2,137
0.024(0.008, 0.07)
0.011(0.003, 0.05)
0.0046(0.001, 0.03)

0.020(0.007, 0.06)
0.0056(0.002, 0.02)
0.0003(0.0001, 0.002)
0.011(0.004, 0.03)
0.0052(0.001, 0.02)
0.0021(0.0004, 0.01)

0.010(0.003, 0.03)
0.0025(0.001, 0.01)
0.0001(0.00003, 0.001)
       Pump  Seals
         Light Liquid
           Ethylene
           Cumene
           Vinyl Acetate
                                76
                                25
                                89
                   0.069(0.006, 0.8)
                   0.052(0.001, 2.7)
                   0.0043(0.0001, 0.1)
                            0.031(0.003, 0.4)
                            0.023(0.0004, 1.2)
                            0.0020(0.00006, 0.06)

-------
00
           GJ
           H
           id
           o
           WJ
           o
          •a
          a
          D
          P.
                   0. t
                  0.01
                 0.001
                0.000
                         CUMENE
V. ACETATE
                                   ETHYLENE
                        J-,	Qas  Source
         ETHYLENE
CUMENE            V. ACETATE
	Light Liquid Service	1
                                   Figure 4-11.  Emission Factors—Valves

-------
g-
p-
en
en
H-
O
3
                                        88
                         pounds/hour/source (log  scale)
                     ©
                              G3
                               i    ~r
OS

C

i-l

ffi
o
C
2
rn
Z
rn
               r
               m
               z
               rn
CO
ro
                 H
                 o
                 R
                 -I
                 >
                 -I
                 rn

-------
00
100
 90
 60
             70
          T  60
          0
          T
          A  50
          L
             30
 20
             10
                               X**
                                       .....
                            lognormal model
                            empirical function
                            95% confidence limits
                            for empirical function
                                                                ***** XX
                                                                    * **XXX
                                                                       *** * X
                                                                         * * *»X
                                                                .             ***X
                                                                                 *  X
                                                                                   *
 *•
1.0
1.5      2.0     2.5      3.0      3.5

                      LOGIOIOVA
H.O
4.5
                                                                        b.O
                                                                                                     6.0
                    Figure  4-13.   Cumulative Distribution of Total Emissions by Screening Values
                                  Cumene Process,  Valves in Gas Service

-------
      \
      I
      I
      I
  100  *   *    X   X  «  XXXXXXXXXXm*	  •
      i                	.*****XKXXXXXXXX                    	
      I	    ****XXKXXXX                         ...
   90  *                               ...         »*XXXX
P|                                 	       **KXXX
E     I                                      ..          ***XX
R  tiO  +                                        ...  '        *XxX
C     I                                          .*•          *XXX
El                                            ..          ***  X
ij  70  +                                             . .          *  »X
T     I                               ....         #*  **
      I                                                 ...           X* *»
T  £.0  t                                                   •.           X   *
01                                                                  X   **
T     I                                                      ,             XX   *
ft  50  +                                           '                          X
LI                                                       ..XX
      I                                                         •
E  HO  +                                                           .
Ml
II                                                      '        -.
s  30  +           x  *  lognormal model
[i           *  «  empirical function                               *\
a  20  +            .  -  95% confidence limits
(J     [                for empirical  function                                *

   10  +
      i
      i
    o  +                                                                       .
      i
      I
      I
      i

         1.0      1.5     2.0      2.5      3.0     3.5      4.0      1.5     5.0      5.5      6.0

                                       LOGlbfOVA SCREENING VA|_UE)


          Figure  4-14.   Cumulative Distribution of  Total Emissions by Screening Values
                         Cumene Process, Valves in Light Liquid Service

-------
  100
   90

E
K  (10
•C
E
N  70
 T  60
 0
 T
 A  50
 L
   50
 0  20
                                             *   X X
                                                  *
lognormal model
empirical function
95% confidence  limits
for empirical function
         1.0      1.5      2.0     2.5      3.C      3.5      4.0      4.5     5."     5,5      6.0

                                       LOGlOtOVA SCREENING
           Figure 4-15.  Cumulative Distribution of Total Emissions by Screening Values
                         Cumene Process, Pumps in Light Liquid Service

-------
      I
      I
      I
      t
  100 *    x***ju*»***m»****x*******x******if *******
      i                                               	XXKXXM**
   90 *                                                    ...   XXX«**»*
P     I                                                       ....  AXXX ****
E     I                                                           ...   XXX  »***
H  ao +                                       '                      ...   xx   *
C.I                                                                ..XX
El                                                                ..XX
N  70 +                                                                  ..XX
T     I                                                                   ..XX
      I                                                                     ..   X
T  60 +                                                                      ..X
0     I                                                                            X
T     I                                                                       ...    X
A  50 +                                             '                             , .  X
LI                                                                            .  . X
      I
C  40 +                                                                                X   ..
HI                                                                                X    ...
\     '         X =  lognormal model
si         * =  empirical function         '                             »**» *       x
I'         . =  95% confidence limits                                        *  * **   x
"  20 \              for empirical function                                              *** *
!1     I                                                                                     X  «
      I
   10 +
      I                                                                                       X
      I                                                                                        X
    0 +  .                                                                     	
      I
      t

      I

                                          3.0      3.5      1.0      4.5      &,0     5.b      6.1)

                                       tOGtU
            Figure 4-16*  Cumulative Distribution of Total Emissions by Screening Values
                          Ethylene Process,  Valves in Gas Service

-------
  loo  +
p
E
K  60  +
C
e
N  70
T  60  +
0
T
A  50
L

E  40
   30
I
S
S
I
0  20 +
ft

   10 +
           *K
                 lognormal model
                 empirical function
                 95% confidence limits
                 for empirical function
                                                ,**xx*xxx	
                                                  	»XXXXM..»,
                                                     ....   *XXX
                                                                    X**
                                                                     xxx***
                                                                   ,. XX ***
                                                                     ..  XX
                                                                       .. XX
                                                                        .. x
                                                                                      XX
                                                                                       ...X
                                                                           *  * * * »   ***
                                                                                         *
1.0      1.3     2.0      2.5      3.0      3.5

                              LOG10(OtfA
                                                          4.0
                                                                  4.5
b.O
          Figure 4-17.  Cumulative Distribution  of  Total Emissions by Screening Values
                        Ethylene Process, Valves in Light Liquid^Service

-------
  100
   90
P
E
R  60
C
E
M  70
T

T  60
0
T
A  50
L

E  tO
N
I
S  30
S
I
0  20
N

   10
    0
                                                    Kilt *
                                    •  •  *
X * lognormal model
* = empirical function
. = 95%  confidence limits
    for  empirical function
                                               **   X
                                                   *
                1.5     ?,0      2.5      3.0      3.5     4,0

                                      UOG1CHDVA SCREENING
                                                              5.0
S.5
6.0
          Figure 4-18.   Cumulative Distribution of^Total Emissions by Screening Values
                        Ethylene Process, Pumps in Light Liquid Service

-------
100
 90
 flO
 70
 50
 30
 20
 1.0
                             *H X K   X «** *
                                        XX XH *  **
                                              x    *   *
                                                XX      * +
                             ..                   x      *   *
                                . .                   X
                                                        XX
                                                         X
X = lognormal model
* = empirical functipn
. = 95% confidence limits
    for empirical function
                                                                        xx
      i.o
2.0      2.5     3.0      3.5      1,0

              UOGlOfOUA SCREENING VA| Ui
                                                             5.0
                                                                                5.5
       Figure  4-19.   Cumulative Distribution of Total Emissions by Screening Values
                     Vinyl  Acetate  Process,  Valves with Light Liquid Service

-------
      I

      I
      I
  100  +   XX » X t  	
      |            XXM VXX
      I   .. . .      ***X XX
   go  +                     * x  *
PI            .             x x*«
El             .              XX **                        '      .
R  80  t              .                XX*
C     I               ..               X**
El                 .               XX *
ftl  70  +                  .                  **
T     I                    .                XX  *                           .
      I                     •               X  **
T  60  +                       .                X **
01                         .               XX     *
T     I                                         XX       *
A  50  +                         .                    '     *  *
U     I                          ..                              *
      I
E  40  +                              .                 X               «
HI
II                                                  X                 *
S  30  +                                .                 X
S     I                                                     X.
I     I    X =  lognormal model
^   °  J    * =  empirical function       '
      I    . "  95% confidence limits      ..                         x
   10  *         for empirical function
      i                                                                   x
      I                                                                      x

      I
      I
      I
      I

        1.0      1.5      £.0     2.5      3.0      3.5     4.0      4.5      5.0      5.5      6.0

                                       LOt](J(OVA SCREENING VA(.UE)



         Figure 4-20.   Cumulative Distribution  of  Total  Emissions by Screening Values
                        Vinyl Acetate Process, Valves  in  Light  Liquid Service

-------







p
E
R
C
F.
iJ
T

T
0
T
A
L

E
\o ,
-J I
S
S
I
0
•'








1
1
1
1
1 X * XX
1 * X
90 +. X
1
1
00 + .
1 *
1 X
70 + .
1 . X
I
60 +
1 •
t
50 + "X
1
1 *
HO +
1
1
30 + ,
|
1 .
20 +
1
*
10 +
1
(I •*• *.*• .(•
1
1
1
1
1.0 1.5 2.0 2.5 3,0 3.5 4.0 4.5 5.0 5.5 6.0
                             LOGJOIOVA SCREENING VALUE)
Figure 4-21.  Cumulative Distribution of Total Emissions by  Screening Values
              Vinyl Acetate Process, Pumps  in Light  Liquid Service

-------
CO
p
E
R
C
E
N
T

S
C
R
E
E
N
I
N
G

G
R
E
A
T
E
R
                        40
                          0
                                                          •7—  Estimated Percent of Sources
                                                               Screening Greater Than the Selected Source

                                                          — —  95X Confidence Limits
                                                     345
                                             LOG10COVA  SCREENING VALUED
                     Figure 4-22a.  Cumulative Distribution of Sources by  Screening Values
                                    Cumene Process, Valves in Gas Service

-------
   P
   E
   R
   C
   E
   N
   T

   T
   0
   T
   A
   L

   E
   M
   I
   S
   S
   I
   0
   N
100
 40
          0
~^Estimated Percent of Total
  Mass Emissions Attributable
  to Sources With Screening
  Values Greater Than the
  Selected Value

— 95% Confidence Limits
           0
                          2345
                       LOG10COVA SCREENING  VALUE)
Figure 4-22b.   Cumulative Distribution of Total Emissions by  Screening Values
                Cumene Process,  Valves in Gas  Service

-------
o
o
p
E
R
C
E
N
T

S
C
R
E
E
N
I
N
G

G
R
E
A
T
E
R
                       80
                       40
                       20
                                                           —— Ectiiutad Percent of Sour CM
                                                               Screening Greater Than the Selected Source
                                                           — — 95X Confidence Unite
                                                   345
                                            LOGI0COVA  SCREENING  VALUE)
                                                      6
                   Figure 4-23a,
Cumulative Distribution of Sources by Screening Values
Cumene Process,  Valves in Light Liquid Service

-------
E
R
C
E
N
T

T
0
T
A
L

E
M
I
S
S
I
0
N
       40
         0
          0
                    — Estimated Percent of Total
                      Mass Emissions Attributable
                      to Sources With Screening
                      Values Greater Than the
                      Selected Value

                    — 95% Confidence Limits
                              2345
                          LOG10COVA  SCREENING VALUE?
6
Figure 4-23b.   Cumulative Distribution of  Total Emissions by  Screening Values
                Cumene Process,  Valves in Light Liquid Service

-------
P
E
R  100
c
E
N
T    80
S
C
R
E
E
N
I
N
G

G
R
E
A
T
E
R
60
40
  0
                                 • '  Eetlnated Percent of Sourcee
                                    Screening Greater Than Che Selected Source
                               	  95X Confidence Limits
                2           34            5
                     LOG10COVA  SCREENING  VALUE)
6
 Figure 4-24a.
            Cumulative Distribution of Sources by  Screening Values
            Cumene Process, Pumps in Light Liquid  Service

-------
  P
  E
  R
  C
  E
  N
  T

  T
  0
  T
  A
  L

  E
  M
  I
  S
  s
  I
  0
  N
00
60
40
             — Estimated Percent of Total
               Mass Emissions Attributable
               to Sources With Screening
               Values Greater Than the
               Selected Value

             — 95% Confidence Limits
                     12345
                             LOG10COVA SCREENING  VALUE)
Figure 4-24b.   Cumulative  Distribution of Total Emissions by  Screening Values
                Curaene Process,  Pumps in Light-Liquid Service

-------
O
•P-
P
E
R
C
E
N
T

S
C
R
E
E
N
J
N
G

G
R
E
A
T
E
R
                       100
                         60
40
                         20
                           0
                                                            —  Estimated Patcant of Sources
                                                               Sct««nlng Gteatar Than the Selectad Source

                                                            —  95X Confidence Limits
                                                     345
                                              LOG10COVA SCREENING  VALUE}
                   Figure  4-25a.  Cumulative Distribution of Sources by  Screening Values
                                   Ethylene Process,  Valves in Gas  Service

-------
o
Ul
p
E
R
C
E
N
T

T
0
T
A
L

E
M
I
S
S
I
0
N
                       103
                        40
— Estimated Percent of Total
  Mass Emissions Attributable
  to Sources With Screening
  Values Greater Than the
  Selected Value

-- 95% Confidence Limits
                          0
                           0
                              2345
                          LOG10COVA  SCREENING VALUE)
                                                        6
                  Figure 4-25b,
              Cumulative Distribution of Total Emissions by  Screening Values
              Ethylene Process,  Valves in Gas  Service

-------
P
E
R
C
E
N
T

S
C
R
E
E
N
I
N
G

G
R
E
A
T
E
R
100
                                    — Estimated Percent of Soureaa
                                       Screening Greater Than th* Selected Source

                                    — 951 Confidence Limits
 60
 40
 20
   0
                              345
                      LOG10COVA  SCREENING  VALUE)
Figure  4-26a.
            Cumulative Distribution of Sources  by Screening Values
            Ethylene  Process, Valves in Light Liquid Service

-------
   p
   E
   R
   C
   E:
   N
   T

   T
   0
   T
   A
   L

   E
   M
   I
   S
   S
   I
   0
   N
108
 60
 40
         0
— Estimated Percent of Total
  Mass Emissions Attributable
  to Sources With Screening
  Values Greater Than the
  Selected Value

— 95% Confidence Limits
          0
                          2345
                       LOG10COVA SCREENING VALUE)
                                                         6
Figure 4-26b,
          Cumulative Distribution of Total Emissions by Screening Values
          Ethylene Process, Valves in Light Liquid Service

-------
o
O3
p
E
R
C
E
N
T

S
C
R
E
E
N
I
N
G

6
R
E
A
T
E
R
                        80
                        60
                        20
                         0
                                                             — Estimated Percent of Sources
                                                               Screening Greater Than ehe Selected Source

                                                               95Z Confidence Limit*
                                       234           5
                                             LOGI0COVA SCREENING VALUE)
                    Figure 4-27a.   Cumulative Distribution of Sources by  Screening Values
                                    Ethylene Process, Pumps in Light Liquid  Service

-------
      100
R
C
E
N
T

T
0
T
A
U

E
M
I
S
S
I
0
N
                        Estimated Percent of Total
                        Mass Emissions Attributabl*
                        to Sources With Screening
                        Values Greater Than the
                        Selected Value

                        95% Confidence limits
                                 2345
                             LOG10COVA  SCREENING VALUE)
                                                                         6
Figure 4-27b.   Cumulative Distribution of Total  Emissions by Screening Values
                Ethylene Process,  Pumps in Light  Liquid Service

-------
p
E
R  100
C
E
N
T    80
S
C
R
E
E
N
I
N
G

6
R
E
A
T
E
R
60
40
20
  0
                                    —  Estimated Percent of Sources
                                       Screening Greater Than the Selected Source

                                    —  95% Confidence Limits
                            345
                     LOG10COVA SCREENING  VALUED
   Figure 4-28a.
              Cumulative Distribution of Sources by Screening Values
              Vinyl Acetate Process,  Valves in Gas Service

-------
E
R
C
E
N
T

T
0
T
A
L

E
H
I
S
S
I
0
N
      813
       40
        0
                   —- Estimated Percent of Total
                      Mass Emissions Attributable
                      to Sources With Screening
                      Values Greater Than the
                      Selected Value

                    — 95% Confidence Limits
                     12345
                            LOG10COVA  SCREENING VALUE)
                                                                         6
Figure 4-28b,   Cumulative  Distribution of  Total Emissions  by Screening Values
                Vinyl Acetate Process, Valves in Gas Service

-------
p
E
R
C
E
N
T

S
C

E
E
N
I
N
G

G
R
E
A
T

R
 108
   80
   60
                                    	 Estimated Percent of Sources
                                        Screening Greater Than the Selected Source

                                    — — 951 Confidence Limits
   20
    0
       345
LOG10COVA SCREENING VALUE)
                                                                     6
Figure 4-29 a.
                  Cumulative Distribution  of Sources by Screening Values
                  Vinyl Acetate Process, Valves in Light Liquid

-------
                  Estimated Percent of Total
                  Mass Emissions Attributable
                  to Sources With Screening
                  Values Greater Than the
                  Selected Value
                  95X Confidence Limits
                      12345
                             LOG10COVA SCREENING  VALUE)
Figure 4~29b.   Cumulative Distribution of Total  Emissions by  Screening Values
                Vinyl Acetate  Process, Valves  in  Light liquid  Service

-------
  p
  E
  R
  C
  E
  N
  T

  S
  C
  R
  E
  E
  N
  I
  N
  G

  G
  R
  E
  A
  T
  E
  R
100
                                    —— Estimated Percent of Sources
                                        Screening Greater Than Che Selected Source

                                    	 95X Confidence Limits
 40
 20
   0
                              3           4.5
                      LOG10COVA  SCREENING VALUE)
Figure 4-30a.
           Cumulative Distribution  of Sources by Screening Values
           Vinyl Acetate Process, Pumps in Light Liquid  Service

-------
H
K
Ui
E
R
C
E
N
T

T
0
T
A
L

E
M
I
S
S
I
0
N
                       100
                        80
                        60
                        40
                          0
                                —- Estimated Percent of Total
                                   Mass Emissions Attributable
                                   to Sources With Screening
                                   Values Greater Than the
                                   Selected Value

                                — 95X Confidence Limits
                                                 2345
                                              LOG10COVA SCREENING  VALUE}
                 Figure  4-30b.   Cumulative Distribution of  Total Emissions by  Screening Values
                                 Vinyl Acetate Process, Pumps  in  Light Liquid Service

-------
TABLE 4-2.  SUMMARY OF PERCENT OF SOURCES DISTRIBUTION CURVES AND PERCENT
            OF MASS EMISSIONS CURVES AT SCREENING VALUE OF 10,000 PPMV
Percent of Sources
Screening > 10,000 ppmv
Source Type
Valves
Gas
Ethylene
Cumene
Vinyl Acetate
Light Liquid
Ethylene
Cumene
Vinyl Acetate
Pump Seals
Light Liquid
Ethylene
Cumene
Vinyl Acetate
Estimate


15
16
3.7

26
12
0.2


30
14
1.7
95% Confidence
Interval


(14
(13
(2,

(24
(10
(0,


(20
(1,
(0,


» 16)
> 19)
5)

, 27)
, 13)
0.4)


, 39)
27)
4)
Percent of Mass Emissions
Attributable to Sources
Screening ^ 10,000 ppmv
Estimate


94
94
90

89
80
25


96
89
67
95% Confidence
Interval


(93,
(90,
(85,

(87,
(72,
( 9,


(90,
(50,
( 5,


95)
96)
94)

90)
86)
47)


98)
98)
92)

-------
                                 SECTION 5
               EVALUATION OF THE EFFECTS OF LEAK OCCURRENCE,
                 RECURRENCE, AND REPAIR ON MASS EMISSIONS
     This section presents the results of investigations of leak occurrence,
recurrence, and maintenance effects on VOC mass emissions.   The analysis is an
extension of previous work on these aspects of fugitive emissions control
presented in Reference 2 (maintenance study).

EFFECT OF LEAK OCCURRENCE ON MASS EMISSIONS

     Leak occurrence was defined in Reference 2 for sources initially
screening < 10,000 ppmv as the first occurrence of a leak (screening ^ 10,000
ppmv) at any time after the initial screening.  In the maintenance study,
described in Reference 2, there were 651 valves and 89 pumps which screened
below 10,000 ppmv initially, and were subsequently rescreened two to six times
over a six month period.  Estimated leak occurrence rates were developed in
Reference 2 for both valves and pump seals.  This section presents estimates
of the effect on mass emissions from those leak occurrences.  The statistical
procedures used to develop these estimates are discussed in Section 7.

     Table 5-1 snow estimates of the weighted percent increase (WPI) and the
increase in mass emissions for the sources for which leaks did and did not
occur.  The WPI is applicable as an estimate of the effect of leak occurrence
on mass emissions.  The mean emission estimates (Ib/hr/source) are applicable
only to the data from the maintenance study since they represent the combined
data from three specific chemical processes.
                                     117

-------
                     TABLE 5-1.   INCREASE IN MASS  EMISSIONS BY LEAK OCCURRENCE1'  FOR VALVES

                                 AND PUMP SEALS SCREENING < 10,000 ppmv INITIALLY
H
H*
Oo
Source Category
(number of sources
in category)
Sources with leak
occurrences
Valves (30)
Pump Seals (15)
Sources without
leak occurrence
Valves (621)
Pump Seals (74)
Weighted Percent
Increase (%)


530
(200, 900)
75
(-100, 6000)

-37
(-56, -18)
-47
(-100, 11)
Mean Emissions
At Initial Screening


0.0052
(0.001, 0.03)
0.013
(0.001, 0.1)

0.00065
(0.0002, .002)
0.0014
(0.0002, 0.01)
(Ibs/hr/source) Mean Emissions
At First Leak Occurrence
or Last Screening


0.033
(0.006, 0.1)
0.99
(0.005, 10)

0.00041
(0.0001, 0.002) (-0
0.00075
(0.00004, 0.005)
Increase
(Ibs/hr/source)


0.028
(0.005, 0.2)
0.98
(0.02, 20)

-0.00024
.001, 0.00002)
-0.00066
(-0.02, 0)
        1)  Screening ^ 10,000 ppmv the first time following initial screening


        Note:   leak rates estimated at initial screening and measured (or estimated)  at either (1)  time of
               first occurrence of (2) time of last screening


        Note:   estimates are reported with an approximate 95% confidence interval

-------
     The valves with a leak occurrence had a WPI in emissions of 530% while
the valves without leak occurrence showed a slight decrease in emissions
(WPI = -37%).  These estimates can be combined with the occurrence rates
estimates in Reference 2 to estimate the total..impact of leak occurrence on
mass emissions-  The confidence intervals for these estimates should be con-
sidered in analyses of this type.  The confidence intervals for the WPI
estimates for pumps are quite large and include zero (no increase).

EFFECT OF LEAK RECURRENCE ON MASS EMISSIONS FOR VALVES

     Leak recurrence was defined in Reference 2 for maintained valves
(screening value < 10,000 ppmv immediately after maintenance) as a leak
(screening value ^ 10,000 ppmv) at any time after maintenance.  In the
maintenance study (Reference 2) there were 28 valves with the potential for
leak recurrence (i.e., with screening value ^ 10,000 ppmv before maintenance
and < 10,000 ppmv immediately after maintenance).  Eight valves exhibited a
leak recurrence during the six month period after maintenance.  Leak recur-
rence rates for valves were estimated in Reference 2 using these data.  This
section presents estimates of the effect on mass emissions from these leak
recurrences.

     Table 5-2 shows estimates of the weighted percent increase (WPI) and
estimates of the mean emissions before maintenance, after maintenance, and
after recurrence or at time of last screening.  As with the occurrence esti-
mates, the mean emission estimates are applicable only to the data from the
maintenance study.  The confidence intervals for the WPI estimate include
zero in both cases due to the small number of sources studied for recurrence,
The estimates can be combined with recurrence rate estimates in Reference 2
to evaluate the impact of recurrence on emissions from valves, but the con-
fidence intervals should be considered in these evaluations.
                                     119

-------
                 TABLE 5-2.  INCREASE IN MASS EMISSIONS BY LEAK RECURRENCE   FOR VALVES

                             SCREENING < 10,000 ppmv IMMEDIATELY AFTER MAINTENANCE

Source Category
Valves with leak
recurrence
(8 valves)
Valves without
leak recurrence
(20 valves)

Weighted Percent
Increase (%)
510
(-100, 1700)
-50
(-96, -5)

Mean Emissions (Ib/hr/source)
Before After At first Recurrence
Maintenance Maintenance or Last Screening
0.26 0.0033 0.02
(0, 0.6) (0, 0.02) (0, 0.08)
0.024 0.0016 0.0008
(0.01, 0.04) (0.0001, 0.01) (0.0001, 0.002)
Mean Emissions
Increase at
Recurrence or
Last Screening
(Ib/hr/source)
0.017
(-0.04, 0.2)
-0.0008
(0.008, 0.002)
1)  Screening > 10,000 ppmv the first time following after maintenance screening

Note:  leak rates measured (or estimated) after maintenance and at either (1) time of first
       recurrence or (2) time of last measurement

Note:  estimates are reported with an approximate 95% confidence interval

-------
FURTHER ANALYSIS OF EFFECT OF VALVE MAINTENANCE ON MASS EMISSIONS

     A statistical analysis was done to expand on the analysis of the
immediate effect of valve maintenance in Reference 2.  Reference 2 reported
a weighted percent reduction (WPR) of 71% (95% confidence interval of 54%
to 88%) for 155 valves for which maintenance was performed.  The WPR for the
97 valves with a before maintenance screening valve of > 10,000 ppmv was 70%
(95% confidence interval of 46% to 95%).  Reference 2 also reported that
only 29% of the 97 valves were "repaired" by simple on-line maintenance,
where a "repair" is defined as screening below 10,000 ppmv immediately after
maintenance.  This analysis compares the reduction for the 29% of the sources
repaired with the 71% not repaired.

     Table 5-3 summarizes this comparison and Figures 5-1 and 5-2 show the
before minus after maintenance leak rates plotted against the before mainten-
ance leak rates for  the "repaired" and "non-repaired" valves.  The weighted
percent reduction for repaired valves was 97.7% (95%, 100%) compared with
62.6% (41%, 85%) for non-repaired valves.  This significant difference in
emissions reduction between the two groups of valves can be seen by comparing
the data plots in Figure 5-1 and 5-2.

     Table 5-3 also contains estimates of the mean emissions from the valves
before and after maintenance.  These estimates are only applicable to the
sources in the data base since they represent the combined data from valves
from three specific chemical processes.
                                      121

-------
ro
to
             lO.OOOo
          B
          E
          F   1*0000
          0
          R
          E   0*1000

          H
          I   0*0100
          N
          U
          s   o.noio

          A
          F   O.OOOi
          T
          E
          R   Q.QOOo
          E  -O.OOOl
          A
          K
             -0*0010
          H
          A
          T  -0.010Q
             -O.lOOfl
                                                     "Repaired"  Valves

                                                     LEGENDI A s 1 DBS,  B  =  2 QbSt ETC.
 perfect repair
(100% reduction)
                     --t	+	+—„ ——-+	+	+-	--+.._—„—+—^—--+	--^—+-
                    0,0001     0.0nl)3    0.0010    0.0030    0.0100    0.0300     0*1000    0.30UO    1.0000    3.0000

                                                    BEFORt MAINTENANCE LEAK  HATE
                                Figure 5-1.  Before Minus After Maintenance Leak Rate -
                                              Valves Screening < 10,000  ppmv After Maintenance

-------
NJ
Lo
             10.0000
              1.0000 +
                    I
              0,0001  +
                     1
                     I
R   O.OOOQ +•
          I
L         I
E  -O.OQOi +
A         I
K         I
   -0.001Q +
K         !
A         I
          T  -O.Olflo  +
          E          I
                     I
             -O.lOOQ  +

                     I
                    0.0001
                                                     "Non-Repaired" Valves

                                                     LtGENOI A = 1 DBS* B = 2  0HS, ETC.
                                                                                             perfect  repair
                                                                                            (100% reduction)
                                                    A
                                                    AA
0.0003     0.0010    0.0030     0.0100    0.0300    0*1000

                      BEFORE  MAINTENANCE LEAK RATE
                                                                            0.5000    1.0000   3.0000
                                Figure 5-2.  Before Minus After Maintenance Leak Rate -
                                             Valves Screening ^ 10,000  ppmv After Maintenance

-------
                 TABLE 5-3.  WEIGHTED PERCENT REDUCTION IN MASS EMISSIONS FOR VALVES

                             SCREENING > 10,000 ppmv IMMEDIATELY BEFORE MAINTENANCE
  Source Category
                  Mean Emissions  (Ib/hr/source)
                                                                                    Mean Emissions
                                                                                      Reduction
Weighted Percent
 Reduction (%)     Before Maintenance   After Maintenance   (Ib/hr/source)
Sources repaired'
   (28 valves)
                              97.7
                            (95, 100)
                      0.09
                    (0,  0.2)
                                             0.0002
                                            (0, 0.02)
                        0.088
                   (-0.007,  0.2)
Sources not repaired
     (69 valves)
  62.6
(41,  85)
                          0.10
                      (0.04, 0.2)
    0.038
(0.02,  0.05)
                                                                                         0.062
                                                                                    (0.006, 0.12)
       Total
   (97 valves)
  70.1
(46,  95)
1)  Screening < 10,000 ppmv Immediately after maintenance

Note:   leak rates measured before and after maintenance

Note:   estimates are reported with approximate 95% confidence interval

-------
                                  SECTION 6
                    IMPACT ON LEAK FREQUENCY ESTIMATES OF
                   APPLYING CHEMICAL RESPONSE ADJUSTMENTS
     The goal of the analysis in this section was to investigate the effect of
applying chemical(s) specific response adjustments to the OVA readings to
estimate the frequency of leaks from SOCMI process units.  This was accom-
plished by calculating adjusted screening values based on the original screen-
ing value and chemical response factor corrections.  For the purposes of this
study a source is said to be leaking if its screening value is ^ 10,000 pprov.
Three different techniques were used to adjust the original OVA screening
value:

     1)   the original OVA reading adjusted for the associated OVA
          response relationship of the primary chemical compound in
          the line (see Section 7 for more detail),
     2)   weighted logarithmic average of response of primary and
          secondary chemicals (see Section 7 for more detail),
     3)   weighted arithmetic average of response of primary and
          secondary chemicals (see Section 7 for more detail).
     The percent of valves leaking was calculated for each of the three
estimates for both gas and light liquid services.  The three estimates were
found to be similar to the leak frequency estimate based on the original
screening value.
          *
     It should be noted that the total number of valves used in this analysis
may not match totals from previous sections of this report.  The reason is
that in certain process units a dilution probe was not used.  This resulted
in 119 sources having a recorded OVA reading of 10,001 ppnrv (indicating a
concentration above 10,000 ppmv).  Many of the adjustments of these data
                                     125

-------
resulted in estimates just below 10,000 (i.e., -9,997) ppmv).  Therefore,
all sources with OVA readings equal to 10,001 ppmv were excluded from the
analysis.  These 119 observations came from the following process types:

         Acrylonitrile - 37 observations,
         Chlorinated Ethanes - 11 observations,
         Ethylene Dichloride - 28 observations,
     •   Formaldehyde - 1 observation, and
         Vinyl Chloride Monomer - 42 observations.

Because of these deletions, the percent leaking estimates will have a small
negative bias.  However, the comparison of the four estimates is still valid
since the relative sizes of the estimates is the important aspect to be
evaluated.

SUMMARY OF FOUR LEAK FREQUENCY ESTIMATES BY PRIMARY CHEMICAL

     The percent leaking estimates resulting from the three adjustment
methods are presented in Tables 6-1 and 6-2.  Also included is the percent
leaking estimate from unadjusted OVA readings for comparison purposes.  As
seen in the tables, the three leak frequency estimates based on adjusted
screening values are similar to the unadjusted estimates.

SUMMARY OF FOUR LEAK FREQUENCY ESTIMATES BY PROCESS TYPE

     The main question to be answered by this investigation is, "If the OVA
readings for a given process unit are adjusted for chemical response, will
significantly different estimates of the percent of leaking sources result?"
From the summarizations shown in Tables 6-3 and 6-4, it is evident that there
are no drastic changes in the estimates of percent leaking.  However, there
is a general trend for a small reduction in the estimated frequencies.

     To show the relationship between OVA readings and Method 1 estimates,
                                    126

-------
plots of these two variables are shown for specific process types in Figures
6-1 through 6-6.  The effects of specific chemicals with a process type can
be seen as straight lines.  This is especially apparent in Figures 6-3 and
6-4.  The northwest quadrant of these plots indicate valves where the original
screening value was below 10,000 ppmv and the Method 1 estimates are ^
10,000 ppmv.  The southeast quadrant represents the opposite situation.  The
other two quadrants indicate no change in the leak designation for those
valves.

     For the high leaking processes the adjustments to the gas service valves
result in consistently lower percent leaking estimates.  These estimates
are approximately 3 percentage points lower.  The estimates in all other cases
are almost indistinguishable from the unadjusted estimate.
                                      127

-------
                           TABLE  6-1.   PERCENT  LEAKING ESTIMATES  FOR VALVES  IN LIGHT LIQUID  SERVICE
N>
oo
Chemical
Propylene
Ethane
Ethylene
Methane
Benzene
Methyl pthyl Ketone
Sec Butyl Alcohol
Hydrocarbons-Cs
Acetone
Methanol
Acetic Acid
Cumene
Acetaldehyde
Tr ichloroethylene
Vinyl Acetate
Methyl Methacrylate
Perchloroethylene
1,1,2 Tr Ichloroethane
1,2 Ethylene Dlchloride
Acrylonltrlle
Vinyl Chloride
Phenol
ct-Methyl Styrene
Acetone Cyanohydrln
Other Chemicals
TOTAL
OVA
Response
Factor @
10,000 ppmv
Responae
0^80
0.65
0.70
1.00
0.29
0.60
0.76
0.52
0.80
4.39
1.60
1.87
1.14
0.95
1.30
0.99
2.97
1.25
0,95
0.97
0.80
AAA4
113.9
3.51

Number
Screened
1583
328
1230
205
536
425
202
323
209
373
1162
773
456
267
973
393
599
911
2777
1120
607
594
326
191
15 JO
18,133s
Percent Leaking
Based on
OVA. Readings
Number
Leaking
467
92
321
36
49
23
10
a
5
4
6
4
2
1
3
1
1
1
0
0
0
0
0
0
67
1101
Percent
Leaking
29.50
28.05
26.10
17.56
9.14
5.41
4.95
2.48
2.39
1.07
0.52
0.52
0.44
0.37
0.31
0.25
0.17
0.11
0
0
0
0
0
0
4.27
6.07
Percent Leaking
Baaed on Method 1
Adjustments1
Number
Leaking
417
70
271
36
26
14
8
7
4
11
7
9
2
0
3
0
6
1
0
0
0
1
3
0
69
965
Percent
Leaking
26.34
21.34
22,03
17.56
4.85
3.29
3.96
2.17
1.91
2.95
0.60
1.16
0.44
0
0.31
0
1.00
0.11
0
0
0
0.17
0.92
0
4.39
5.32
Percent Leaking
Baaed on Method 2
Adjustments2
Number
Leaking
446
75
273
47
30
10
7
7
5
11
8
11
4
1
4
1
6
1
5
1
0
2
0
0
68
1023
Percent
Leaking
28.17
22.87
22.20
22.93
5.60
2.35
3.47
2.17
2.39
2.95
0.69
1.42
0.88
0.37
0.41
0.25
1.00
0.11
0.18
0.09
0
0.34
0
0
4.33
5.64
Percent Leaking
Baaed on Method 3
Adjustments3
Number
Leaking
431
76
2^3
38
28
13
7
6
4
11
6
9
2
0
3
1
5
1
0
0
0
0
1
0
65
990
Percent
Leaking
27.23
23,17
23.01
18.54
5,22
3,06
3,47
1,86
1.91
2,95
0,52
1,16
0,44
0
0.31
0.25
0.83
0,11
0
0
0
0
0.31
0
4.14
5.46
                ?S°3 i I" ^ a?Ju't[n*nt to1the °VA read1"8 baBed °n the «sponee of the primary chemical in the line.
                Method 2 la the mixed chemical weighted logarithmic average  technique.
                Method 3 ia the mixed chemical weighted average technique.
                A response of 10,000 ppmv for Phenol was experimentally unattainable.
                74 aourcea with OVA Reading - 10,001 ppmv were excluded.

-------
                               TABLE 6-2.   PERCENT  LEAKING  ESTIMATES FOR  VALVES IN GAS  SERVICE
ho




Chemical
Propylene
Benzene
Ethyl ene
Methane
Propane
Ethane
Methyl Ethyl Ketone
Acetaldehyde
Acetic Acid
1,2-Ethylene Dichlorlde
Acrylonitrile
Vinyl Acetate
Vinyl Chloride
Other Chemicals
TOTAL (
OVA
Response
Factor @
10,000 ppmv
Response
0.80
0.29
0.70
1.00
0.60
0.65
0.60
1.14 .
1.60
0.95
0.97
1.30
0.80
—



Percent
Baaed
Leaking
on
OVA. Read inns
Number
Screened
1119
33Z
3104
1849
145
379
116
179
125
521
287
272
96
850
93741*
Number
Leaking
198
53
468
232
18
• 35
7
' 4
1
0
0
0
0
41
1057
Percent
Leaking
17.69
15.96
15.08
12.55
12.41
9.23
6.03
2.23
0.80
0
0
0
0
4.82
11.28
Percent
Based on
Leaking
Method 1
Adjustments'
Number
Leaking
168
31
422
232
19
25
4
4
1
0
0
0
0
40
946
Percent
Leaking
15.01
9.34
13.60
12.55
13.10
6.60
3.45
2.23
0.80
0
0
0
0
5.29
10.09
Percent
Baaed on
Leaking
Method 2
Adjustments2
Number
Leaking
189
36
425
213
18
31
3
4
1
0
0
0
1
40
961
Percent
Leaking
16.89
10.84
13.69
11.52
12.41
8.18
2.59
2.23
0.80
0
0
0
1.04
4.71
10.25
Percent
Baaed an
Leaking
Method 3
Adjustments3
Number
Leaking
173
32
437
234
18
29
3
4
1
0
0
0
0
37
968
Percent
Leaking
15.46
9.64
14.08
12.66
12.41
7.65
2.59
2.23
0.80
0
0
0
0
4.35
10.33
               1  Method 1 is the adjustment to the OVA reading based on the response of [he primary chemical in the line.
               2  Method 2 is the mixed chemical weighted logarithmic average technique.
               1  Method 3 la the mixed chemical weighted average technique.
               *  45 sources with OVA Readings - 10,001 ppmv were excluded.

-------
              TABLE  6-3.   PERCENT LEAKING  ESTIMATES FOR  VALVES  IN LIGHT LIQUID SERVICE BY PROCESS TYPE
CO
o
Percent Leaking
Based on
OVJL Read Inns
Process (unit #'B)
Ethylene (2,4,11)
Cumene (5,6)
Methyl Ethyl Ketone (31,32)
Acetaldehyde (33)
Vinyl Acetate (1,3)
Acetone/Phenol (12)
Chlorinated Ethanes (60,61,62)
Methyl Methacrylate (34)
1,2-Ethylene Dichlorlde (21,29)
Acrylonitrile (65,66)
Vinyl Chloride Monomer (20,28)
Formaldehyde (22)
Adiplc Acid (35,64)
TOTAL
Number
Screened
4121
762
671
551
2137
1818
1982
1058
2232
1466
1197
121
17
18,133
Number
Leaking
966
80
34
3
8
6
3
1
0
0
0
0
0
1101
Percent
Leaking
23.44
10.50
5.07
0.54
0.37
0.33
0.15
0.09
0
0
0
0
0
6.07
Percent Leaking
Baaed on Method 1
Adjustments
Number
Leaking
852
62
23
3
9
9
7
0
0
0
0
0
0
965
Percent
Leaking
20.67
8.14
3.43
0.54
0.42
0.50
0.35
0
0
0
0
0
0
5.32
Percent Leaking
Based on Method 2
Adjustments
Nunber
Leaking
895
69
18
4
12
9
8
1
4
2
1
0
0
1023
Percent
Leaking
21.72
9.06
2.6B
0.73
0.56
0.50
0.40
0.09
0.18
0.14
0.08
0
0
5.64
Percent Leaking
Based on Method 3
Adjustments
Number
Leaking
882
63
,21
2
9
6
6
1
0
0
0
0
0
990
Percent
Leaking
21.40
8.27
3.13
0.36
0.42
0.33
0.30
0.09
0
0
0
0
0
5.46
                Method
                      1 la the adjustment to the OVA reading baaed on the response of the primary chemical in the line.
             2  Method 2 is the mixed chemical weighted logarithmic average technique.


             1  Method 3 is the mixed chemical weighted average technique,

-------
    TABLE  6-4.   PERCENT LEAKING ESTIMATES FOR VALVES  IN GAS  SERVICE BY  PROCESS TYPE




Percent
Baaed
Leaking
on
OVA. Readings

Process (unit #'s)
Ethylene (2,4,11)
Cumene (5,6)
Methyl Ethyl Ketone (31,32)
Acetaldehyde (33)
Vinyl Acetate (1,3)
1,2-Ethylene Dlchloride (21,29)
Acrylonitrile (65,66)
Vinyl Chloride Monomer (20,28)
Methyl Mathacrylate (34)
Adiplc Acid (35,64)
Chlorinated Ethanes (60,61,62)
Formaldehyde (22)
Ace tone/ Phenol (12)
TOTAL
Mumber
Screened
6050
1443
207
178
949
397
387
382
190
95
48
40
8
9374
Number
Leaking
932
63
19
8
35
0
0
0
0
0
0
0
0
1057
Percent
Leaking
15.40
14.22
9.18
4.49
3.69
0
0
0
0
0
0
0
0
11.28
Percent
Based on
Leaking
Method 1
Adjustments'
Number
Leaking
849
45
13
8
31
0
0
0
0
0
0
0
0
946
Percent
Leaking
14,03
10.16
6.28
4.49
3.27
0
0
0
0
0
0
0
0
10.09
Percent
Based on
Leaking
Method 2
Adjustments2
Number
Leaking
856
49
12
8
33
1
1
0
0
0
0
1
0
961
Percent
Leaking
14,15
11.06
5.80
4.49
3.48
0.25
0.26
0
0
0
0
2.50
0
10.25
Percent
Based on
Leaking
Method 3
Adjustments1
Number
Leaking
873
44
11
8
32
0
0
0
0
0
0
0
0
968
Percent
Leaking
14.43
9.93
5.31
4.49
3.37
0
0
0
0
0
0
0
0
10.33
Method 1 id the adjustment to  the OVA reading based on the response of the primary chemical in the line.
Method 2 is the mixed chemical weighted logarithmic average technique.
Method 3 is the mixed chemical weighted average technique.

-------
u>
N>
                    loouooon
                     1U 0 0 0 0 It
                      10000(1
                       AOOOO
                        loon
                         100
                           1
         1        10



Legend:  A * 1 OBS, B - 2 OBS, etc.
                                                  "•*•-•
                                                  ion
1000



 OVtt
lUOOo
                                                                             100000
1UUOOUO  10000000
                                 Figure 6-1.   OVA Reading vs. Method 1 Adjustment for

                                               Cumene Process Valves in Gas  Service

-------
I-1
                     10001)00!,
                      1UUOOOC
                        10001)
                         100P
                          100
                           10
       *-••— 4_-*-'I<»te-*l»*' — .^**— •— • — ^l-*-.f.«—_»»_» — <-4->f«*

         \        10        100      1000


Legend:  A *  1 DBS, B = 2 DBS, etc.          f'VA
                                                                      1UOOO
• --_ -f ••>.
100000
                                                                                        10000UO  100000UO
                                Figure 6-2.   OVA Reading  vs. Method 1 Adjustment for
                                               Cumene Process Valves  in Light Liquid Service

-------
1000 f) 000
 jUUOOOu
  10000(1
   11) 0 0 0
    1000
      11)
       1
           1        10        100       1000

Legund:  A " 1 OBS, B - 2  OBS,  etc.           OVA
                                                        100000
10000UU   1000QOUO
           Figure  6-3.   OVA Reading vs. Method 1 Adjustment  for
                         Ethylene  Process Valves in Gas Service

-------
LO
                      lOOUOOOf) •*•
                       1 0 U 0 0 0 0  +•
                        1UOOOO
                         ioono +
                          1001)
                           100
                            10  t
                             1
                A  Slhli
              AARLKF  VflZM
             fJ?trj  tiftM
          PZG1    Dj^tVI
       HIM      rvQifi
    IAL       nt^iist:
AF A       KfJFCVtDl
         UntlAflAHA
       PL
     AH  K> B
               1         10


Legend: A = 1 DBS,  B = 2 OBS, etc.
100
                   1000
                                                                        lUOOf)
10 () 0 0 0
                                      10UOOUO  10000DUO
                                Figure 6-4.   OVA Reading vs.  Method 1 Adjustment  for
                                               Ethylene Process Valves in Light Liquid Service

-------
r
                                    1 U I) 0 0 0 0
                                     1U 0 0 0II
                                      iuoon
                                       1000
                                        100
                     A
                     AD
                 3 C A
               U&C  /A
            fCF, A A
         HEAHAA
            A
   Hl*C     A A
A H  AllA R
    AA   A
       A

c      '  „
                                               1        10


                                  Legend:  A = 1 DBS, -B - 2 DBS, etc.
         100
1POO

 UVA
1UOOQ
                                                                                            100000
                                             1QUUOUU  1QOOOUUO
                                          Figure  6-5.   OVA Reading vs. Method  1 Adjustment  for
                                                        Vinyl Acetate Process Valves in Gas  Service

-------
                        luouoooo •*
OJ
--J
1
1
h
ft
r
t

u
K

p
I*
M
 V
                                  "'I'	H	"Jo	ioou	iuooo""""ioDOOo"    iouooou  loooouuo
                       Legend:  A = 1 DBS, 6 = 2 DBS, etc.
                                                                 OVA  Kt.«Ult|l»
                               Figure  6-6.  OVA Reading vs. Method 1 Adjustment for
                                             Vinyl  Acetate Process Valves  in Light Liquid Service

-------

-------
                                   SECTION 7
                          STATISTICAL CONSIDERATIONS

      This section discusses the assumptions and technical details of the
 statistical methods employed in the analysis of data within this study.

 STATISTICAL CATEGORICAL ANALYSIS USING FUNCAT (SECTION 3)

      The Funcat procedure in SAS (a computer software system) was used to
 test for significance in leak frequency between categories.  This procedure
 is. used in Section 3 to consider leak frequency as a function of line temper-
 ature and pressure for valves in gas stream service.

      The analysis is based on fitting a log-linear model to the cell frequen-
 cies.  The model is:

                      ln(F; :, ) = 9 + a. + Y; + ay, -             where:
                          •Cj hi         -C     j     A^j

     F. -, = expected cell frequency of leaking or not leaking at each
            level of temperature and pressure
        0 = intercept term
       a- = main effect of factor a at level  - (in this case, temperature)
        ••C                                    A^.
       Y- = main effect of factor y at level  . (in this case, pressure)
     o^Y;; = interaction (combined effects)  of temperature and pressure
       *-j

     The program tests the significance of the main effects and interaction
via x2 tests.  The resulting analysis tables and interpretations are similar
to analysis of variance tables and their interpretations.
                                      138

-------
     The following table is typical of the form of the output from Funcat
Analysis.

           Source          elf      CHI-SQUARE         P_
           Temperature      3          54.35       0.0001
           Pressure         3         252.31       0.0001
           Interaction      9          39.69       0.0001

     In this example the main effects, temperature and pressure, are
significant as is the interaction of these two variables.  The P_ statistics
is the probability of making an incorrect significance statement.  The
interaction, or combined effect, can be seen graphically when either pressure
or temperature is plotted against percent leaking,  with a separate line
drawn for each level of the other variable.   Where  there is significant
interaction, the lines will be non-parallel.

CHI-SQUARE TEST FOR INDEPENDENCE (SECTION 3)

     The two-way chi-square test is a technique for testing that two
characteristics are independent.  Here the term "independent" means the
distribution of one characteristic should be the same regardless of the
level of the other characteristic.  This test is used in Section 3.

     When there are two levels of both variables in the two-way classifi-
cation, the computational formula for testing the hypothesis of independence.
is:
          X2 =
               N AD-BC -j)
               (A+B) (C+D) (A+C) (B+D)
where the letters A through D refer to the cell frequencies, N is the total
number of observations and the data is tabulated in a 2 x 2 table as shown:
                                      139

-------
                      VARIABLE I
          VARIABLE II
A
C
B
D
The degrees of freedom for the x2 calculated from this formula is one.   A x
value which exceeds the tabulated value (the specified probability (P)  point
of a chi-squre distribution) indicates a dependence of one variable on the
other.

CONFIDENCE INTERVALS FOR PERCENT SOURCES LEAKING (SECTION 3)

     Confidence intervals for the percent of leaking sources were computed
using the Binomal Distribution.  The Binomial is used to model data when a
random sample is selected and each item is classified into one of two cate-
gories (leaking or non-leaking here).  Exact confidence limits (level 1-a)
for the estimate of percent leaking can be obtained by iteration, solving for
PT in
          n
          X
         i-k
(1 - P,)     = —r for the lower limit and for P   in
      L       • 2                             u
          k
          I
         i=o
(1 - P )     = —r for the upper limit,
where n = number of sources screened and k = number of leaking sources.
Tables of these solutions (Reference 6), available for most cases, were used
to develop 95% confidence intervals reported in Section 3.
                                      140

-------
SCREENING VALUE DISTRIBUTIONS (SECTION 4)

     In order to utilize the results of previous work on the estimation of
mass emissions over a range of screening values (Reference 8) » it was neces-
sary to confirm that the screening values followed a distribution close to
lognormal in form.  Summary statistics for Log  (OVA screening value) were
                                              e
generated, including coefficients of skewness and kurtosis, for cumene units,
ethylene units, vinyl acetate units, and for each source type and service.
From an earlier study (Reference 2) , it was decided that since the detection
limit of the OVA is approximately 10 ppmv, that this number would be used to
define an emitting (not leaking) source.  Separate statistics for screening
values below 10 ppmv and for screening values between 10 and 100,000 ppmv were
generated to evaluate the effect of a larger-than-expected number of observa-
tions at 100,000 ppmv.  The patterns of skewness and kurtosis were similar in
both cases.  An empirical approach was taken in the development of the screen-
ing value distributions and their confidence intervals for comparison with the
lognormal models (described later) .

     Chi-square tests were performed to compare the percentage of each
screening value category (processes and units within process by source type
and service) > 10 ppmv.  The statistic computed was
                           X2(r-l,d.f.) =
                                          All
                                      Categories
where
           3, comparing 3 processes
           2 or 3 » comparing 2 or 3 units  ,
             within a process
     0 =   Observed number of sources < 10 ppmv.
           or ^ 10 ppmv., for each process or unit,
                                     141

-------
     E =  Expected number of sources < 10 ppmv.
          or ^ 10 ppmv., for each process or unit, and
     d.f. = the degrees of freedom.

     The empirical cumulative distribution function  (CDF) of  screening
values, defined by

                  T?f   \   Number of  sources screening ^ xn
                          Total number of sources

for  each category  (process by source type by service) was computed.   The
curves displayed in Figures 4-2 through  4-10 are of  the reverse  cumulative
distribution  functions (RCDF):
                          s\
                        (1-F  (x0)) x  100  vs. Log10(x0)

showing the percentage of sources screening greater  than given value  XQ.

     Confidence limits for the RCDF  were constructed using  a  Kolmogorov
2-sided critical value, w (using tables  from Reference 4).  Upper  and lower
'approximate 95 percent confidence limits for F(x)  (UCL and  LCL,  respectively)
were obtained using the two  following equations:
                                 F(x) + w, if F(x) + w < 1
                                              ^
                                ^.0      » if F(X) + w > 1

                   and
                          ( , _ J F(x) - w, if F(x) - w > 0
                        * W
                                          , if F(x) - w < 0
                                     142

-------
where w is the tabulated critical value and n is the number of sources
screened in the given category.  The resulting limits for the RCDF are
      UCL(x)  =
                                             x 100
                     and
                          LCL-(x) =  (l-Fu(x)) x 100,
     A lognormal distribution was used to model the distribution of screening
values greater than 10 ppmv.  This distribution has the property that when
the original data are transformed by taking natural logarithms, the trans-
formed data will follow a normal distribution.  The lognormal distribution is
often appropriate when the standard error of an individual value is propor-
tional to the magnitude of the value.  The form of the lognormal distribution
is as follows:
      f(x) =
[-
                          x -
                        X
 2a2
G/2?
                                                 for 0 > x > °°
                                                 for x < 0  .
                                     143

-------
     In order to develop cumulative screening value distribution curves,
the "non-emitting" sources (with screening values less than 10 ppmv) also
had to be modeled.  A mixed distribution, specifically a lognormal distribu-
tion with a discrete probability mass at 0, was used for this purpose.  Let-
ting p equal the fraction of non-emitting sources in the population, this
mixed-lognormal distribution has the following form:
      f(x) =
                 fi   ^
                 (1-p)  exp  -
     Mean =  (1-p)  exp
+
                                                    for 10 < x < «>
                                                    for  0 < x < 10
                                                     for x <  0
      Another  set  of  curves  (4-22a  through  4-30a)  contains  the estimated
 cumulative distribution of  log  screening values.   The  curves show 100 percent
 minus the cumulative percent, or the  estimated  percent of  sources which would
 have  screening values greater than any  particular screening value.  These
 cumulative distribution functions  were  estimated  by  fitting a lognormal
 distribution, as  described  above,  to  the screening data and then generating
 the cumulative distribution.

      There was some  difficulty  in  fitting  the lognormal distribution to the
screening values.  Figure 4-1 shows a typical histogram of log screening
 values for valves in gas  service.  The  histogram  appears to approximate a
 normal distribution  adequately  up  to  100,000 ppmv (5.0 on  Iog10 scale).  The
 spike at 100,000  ppmv was due to the  inability  of the  screening device to
 measure beyond 100,000 without  a modification to  the dilution probe.  The
 modified dilution probe was used in only a few  cases in the screening process
 during this program.
                                     144

-------
     To overcome the bias caused by this spike, only log screening values
less than 5.0 were used to estimate the parameters of this distribution.
Formulas from "censored" normal distribution theory (discussed in Reference
3) were then used to arrive at unbiased estimates of the entire distribution.
These estimates were used to generate the cumulative distribution function
for each source type/process stream grouping.

     Confidence intervals for these cumulative-functions were obtained using
the Binomial Distribution.  The 95 percent confidence interval for individual
probabilities were approximated using
       ± 1.96 [p(l - p)/n]
                          I/ 2
where p is the estimated cumulative percent and n is the number of screening
values for each particular source type and stream group.

     The estimated lognormal cumulative distribution functions were compared
with the empirical distribution function and appeared to fit the data reason-
ably well.  Figures 4-2 through 4-10 show the lognormal and empirical dis-
tributions for the source type and service classifications.  Discrepancies
were found at the 100,000 ppmv screening value (5.0 log screening value)
in almost all cases, but this was to be expected since the sample function
had a big jump at this point.

EMISSION FACTOR DEVELOPMENT (SECTION 4)

     Predicted log leak rates were generated for all sources with screening
values greater than 10 ppmv, using the prediction equations (Reference 2)
developed from modeling the available measured leak rates with asso-
ciated OVA screening values.  Emission factors were estimated from the
predicted leak rates using
              (leak rate) = a + 6 [Log10 (OVA Value)] + Z (standard error)
where a and 3 are model parameters developed in the maintenance study
                                    145

-------
(Reference 2)  Z is a standard normal random number and the standard error
is associated with the prediction equation.  Because the true leak rate/
screening relationship is unknown, there is a potential bias introduced when
these predicted leak rates are used in developing emission factors.  This
potential bias was taken into account in developing confidence intervals
discussed below.

     As described in the previous subsection, a lognormal distribution was
used to model the distribution of leak rates for emitting sources (i.e.,
sources with a screening value > ppmv).   Coefficients of skewness and kurtosis
for Log  (leak rates) were computed and histograms examined for normality.
This assumption is adequate for the generated emissions data.  To account for
the non-leaking sources, a mixed distribution with a discrete probability mass
at zero was fit to the data.  The precise form of this distribution was given
earlier in this section.  The best, unbiased estimator of the population
mean emission rate from this distribution is:

                                       -        *
                        m = C(l-p) exp(y)]  g
                                               2  / *

where

     g  ——  - bias correction factor (discussed in detail in Reference 8).

     Confidence intervals for the percent of sources screening > 10 ppmv
were computed using the Binomial Distribution.  Binomial Confidence Interval
tables, available for most cases, were used for computing 97.5 percent con-
fidence intervals which were then used in developing confidence intervals for
emission factors.  The 97.5 percent was selected so that approximate 95
percent confidence intervals for emission factors would result when the
estimated percent leaking was combined with the estimated mean leak rate
(0.975 x 0.975 = 0.95).
                                    146

-------
     The 97.5 percent confidence intervals were computed for the average, y,
of the Log  leak rate estimates using:
                                    2          i/2
         = lower limit = y - 2.24 [s2/(n - r)]
and
                                    2         1/2
      G  = upper limit = y + 2.24 [s2/(n - r)]
where
          f}
         s  = the variance of the log  leak rate estimates and
                                     e
      n - r)= the number of leaking sources.
Then confidence intervals for the mean leak rate  (emission factor estimate)
was computed using

      C' = lower limit = exp[C»] g(s2/2)
       X*                      )fj

and
       " = upper limit = exp[C ] g(s2/2)
where
        (-^—-) is the bias correction factor.
      To obtain 95 percent confidence limits for the emission factors, the
confidence limits for the percent leaking and for the mean leak rate were
combined as follows:
      lower 95% limit for emission factor = p   (Cj)
      upper 95% limit for emission factor = P   (C')
       rr                                  •  u   u
                                     147

-------
These confidence intervals are conservative in the sense that 95 percent
is a lower bound for the confidence coefficient for the intervals.  The
intervals consider random sampling variation and random test error, with
no adjustments for potential bias in the estimation of the log  rates.

      An adjustment was applied to the emission factor confidence intervals
to account for the potential bias due to estimating leak rates.  The stan-
dard error of the predicted average Iog10 leak rate (SEP)  was calculated from
where n = number of leaking sources,
      k = number of data pairs used to estimate
          the prediction equation, and
     x. = ith screening value (Log10 scale) used to estimate the prediction
          equation.

The reported confidence intervals for the emission factors were widened by a
factor of
      10
        2(SEP)
 A similar procedure was used to adjust the confidence intervals  for  the mean
 emission estimate in Section 5.

     As a Quality Control measure on the emission factor estimation,  an
alternative approach to estimate emission factors was also explored.   The
alternative model was
                                       n
             E.F. alternative =  -i_  Y"*  C-10logl° (leak rate)
                                  n   •—•.                    '
                                     148

-------
where n = number of sources and C = bias correction factor.  This approach
employs an estimator based on the arithmetic mean computed in arithmetic
scale of the leak rates.  (The bias correction factor is actually part of
the predicted leak rate in the arithmetic scale as discussed in Reference 2) .
This alternative estimator is unbiased regardless of the distribution of leak
rates and avoids using the generated error term Z (standard error of estimate)
for the predicted leak rates.  Comparison of the results of these approaches
is shown in Table 7-1.

     Confidence limits for the alternative estimates of mean leak rates
were based on computing the mean leak rates for each of the two limiting
distributions of screening values given by the confidence bounds for the
empirical CDF as described earlier in this section.  These bounds were
further adjusted to account for the potential bias in using predicted leak
rates.  For the alternative estimates, the standard error of the mean (SEM)
was calculated from
       SEM -                     CIO   CLOS10 (leak rate)]'   1/2
where cr = residual error for the fitted prediction equation.  The confidence
limits were adjusted by adding or subtracting 2.24 SEM to the upper or lower
confidence limits, respectively.  The results (Table 7-1) are an attempt to
approximate 95 percent confidence limits for these alternative estimates.
Note that the lower confidence is zero in most cases for the alternative
estimate indicating that the ± SEM limits do not adequately reflect the
skewness of the distribution of the alternate estimates.

     The emission factor estimates are not consistently higher or lower than
the quality control estimate.
                                      149

-------
                        TABLE  7-1.   COMPARISON OF EMISSION FACTORS WITH QUALITY CONTROL
                                     ESTIMATES OF MEAN LEAK RATES FOR VALVES AND PUMP SEALS
        Process
Source Type
Service
 Emission Factor
(Ibs./hr./source)
Quality Control Estimate'
    (Ibs./hr./source)
        Cumene
        Ethylene
        Vinyl Acetate
Ui
o
Valves

Pump Seals
Valves

Pump Seals
Valves

Pump Seals
Gas -
Light Liquid
Light Liquid
Gas
Light Liquid
Light Liquid
Gas
Light Liquid
Light Liquid
 0.011(0.003, 0.05)
 0.0056(0.002, 0.02)
 0.052(0.001', 2.7)
 0.024(0.008, 0.07)
 0.020(0.007, 0.06)
 0.069(0.006, 0.8)
 0.0046(0.001, 0.03)
 0.0003(0.0001, 0.002)
 0.0043(0.0001, 0.1)
    0.0079(0,  0.02)
    0.0061(0,  0.02)
    0.030(0, 0.3)
    0.010(0.004,  0.02)
    0.013(0,008,  0.02)
    0.085(0,0.7)
    0.0027(0,  0.03)
    0.0003 (0,0.005)
    0.0051(0,0.06)
         Emission factor reported with 95% confidence interval
         r\
         Quality Control estimate reported with approximate 95% confidence interval
         based on estimate ±2 (standard error of the estimate)

-------
CUMULATIVE EMISSION FUNCTIONS

     A cumulative function for the percentage of total mass emissions for
all sources screening greater than a given value was estimated by integrating
the leak/screening regression relationship over a lognormal distribution of
screening values.  This function has the following form:

       CF m J° C(10)Nx)Bl
where
       So = selected upper screening value for integration,
       C  = log/arithmetic scale bias correction factor,
       BO = log10 regression intercept term,
       BI = Iog10 regression slope term,
       u  = mean of the log  (screening values),
       a2 - variance of the log  (screening values),
       x  = screening values over which the integration is being done, and
       CF = cumulative function described above in Ibs/hr
       D  = numerator of CF evaluated at S0 = 1,000,000

     The form of the cumulative function can be simplified by algebraic
reduction and change of variables to obtain:
          , ,.^.,-J - u - BiCT | /^ | Jin(1,000,000) - u - Bier'
     CF — 


-------
in each case.  The log/log least-squares regression estimates were used for
the scale bias correction factor and for BO and BI.  Division by the numer-
ator of the function evaluated at one million ppmv forced the function to
1.0 at one million ppmv.  These function values were then subtracted from
1.0 and multiplied by 100.0 to obtain the functions shown in Figures 4-22£>
through 4-30b.

     The estimated lognormal cumulative emissions functions were compared
with the empirical functions (discussed below) and found to adequately
approximate the data.  Figures 4-13 through 4-21 show the lognormal and
empirical functions for the source type and service classifications.
The biggest discrepancies were near the 100,000 ppmv screening value where
the sample function has a big jump.  This area is more critical for this
function than the cumulative distribution function since most of the emis-
sions are attributable to sources with screening values greater than 100,000
ppmv.  It is important to note that very little screening data are available
with screening values greater than 100,000 ppmv.  Thus, this portion of the
curve is based on extrapolations using models developed from screening values
less than 100,000 ppmv.

     This cumulative function is a very complex nonlinear function of three
sample statistics.  Due to the complexity of this function, it was not
possible to derive a closed-form analytical expression for the confidence
intervals.  Thus, a Monte-Carlo computer method was used to generate the
confidence intervals.

     This method involved regenerating the cumulative function 200 times.
Each time, the data collected in the project (the number of sources with
screening values greater than 10 ppmv) were regenerated, except with an
independent set of random variations.  The distributional properties of the
leak rate and screening data were used in computing the required random
numbers.
                                     152

-------
     For each of the 200 trials, sample estimates of the three parameters
required to compute the cumulative function were computed.   Then these
estimates were used to generate a new cumulative function.   The one percent
lower result and the 99 percent upper result from the 200 trials for any
given screening value were then selected as approximate 95  percent confidence
limits for the population cumulative function.

     Since these confidence limits address the uncertainty  in the cumulative
function for the entire sampled population of a particular  source type, they
are not necessarily applicable to a finite sample of sources in a particular
situation.  The variation of this function depends on the number of sources
in a complex manner, so it is not possible to draw a general conclusion for
the effect of sample size.

     Empirical functions computing the percentage of total  mass emissions
for all sources screening greater than a given screening value were devel-
oped using the estimator
£
- x < x0
G (-xuJ
Z-^
all x
c • 10 Lo§1° (leak rate)
c . 10 Log10 (leak rate)
in addition to the approach based on the lognormal distribution discussed
                                       ^N
earlier.  Note that the denominator of G(XO) is an expression for the total
mass emission used in the quality control check for estimating emission
factors.

                           A                                 ^
     Confidence bounds for G(XO) were obtained by evaluating G(x0) for the
screening value distributions corresponding to the confidence bounds shown
in Figures 4-13 through 4-21.  Applying a standard approximation to calculate
                                      153

-------
the variance of a ratio (Reference 5), the following  expression for the stan-
             /\
dard error  of G(XQ) was derived:
     SE = CT(Loge(10))
                       x
                     \A11
x
where A = 10
             [Log10  (leak rate)]
                                     LJ
                                    x
\
                         A-
                    x> x     x
     A     _^  A
x 
-------
INCREASE IN MASS EMISSIONS DUE TO OCCURRENCE AND RECURRENCE  (SECTION 5)

     Increase (or reduction) of mass emissions reported in this study has
been expressed in two ways:

     •  Weighted percent increase (reduction)
     •  Mean increase (reduction)

The first measure of change in mass emissions has been discussed in detail in
Reference 2 as Weighted Percent Reduction (WPR), defined as the percent of
total emissions reduced due to maintenance:
            n                         n
    -__. _  (£ mass emissions before - Z mass emissions after)
    WrK. —                                                       X 100%
                       n
                       I  mass emissions before

(See Reference 2 also for a discussion of the development of confidence
intervals for this estimate.)  Weighted percent increase (WPI) is defined as
WPI = -WPR, where "before" and "after" refer to before and after leak occur-
rence, recurrence, or maintenance, depending upon the application.
     The mean increase is defined as the difference in mass of the
average before and after maintenance emissions from individual sources.
Confidence intervals for this measure of increase in mass emissions are
given by:
                                 *

                         Mean increase ± £0.975 x SE,
   SE = Standard deviation of increase// n , where n = number of sources.

     Reference 2 provides details of the estimation of nonmeasured leak rates
from screening values via prediction equations appropriate to source type and
service type.  The equations, as expressed in arithmetic scale, were applied
as discussed in Reference 2, except for sources in unit 1.  In unit 1 there
was no information on service type for valves in the occurrence analysis.
                                      155

-------
The following equation was used for these valves:

 Predicted Leak Rate = 5.08  •  10 C-5'22 + °"67 Log"
-------
The other two adjustment methods used to estimate actual concentration take
into account the mixed nature of the chemical composition.  The OVA response
to a mixture of compounds is intermediate to the individual responses to
each compound at the same concentration.  Using this concept, one of the
mixed chemical adjustment methods (Method 3) used a weighted average of the
responses to estimate actual concentration.  An estimate of the weighted
response is
where, R  = the estimated weighted average response,
        £~L
       p. = the fraction of the mixture total concentration accounted for
        i
            by compound i (P. = C./C ),
       a. = exp (A) with "A" from Brown, et al (1980) for component i,
        i                                 -    '
       b. = coefficient "B" from Brown, et al (1980) for component i,
        i                                  --
       s, = parameters "SE" from Brown, et al (1980) for component i,
        i                                 " —
       C_ = EC., the total concentration, and
       C: = the concentration in the mixture of compound i.

The coefficients A, B, and SE can be found in Tables 5-169 and 5-170 of
Brown, ot_ al (1980) (Reference 7) for selected compounds.

    The above discussion involves the prediction of an instrument response
when the actual concentration of mixture components are known.  For this
study, the reverse is the case:  the response is observed and it is desired
to estimate the total concentration of the constituents.  Basically, this
cannot be done without some additional information.  The compound identifi-
cation of the constituents must be known.  If the constituent proportions
are also know, the total concentration can be computed assuming the above
model is correct.  The total concentration (C ) is estimated by solving
equation 1.
                                     157

-------
     Equation 1 cannot be solved explicitly for total concentration.  An
iterative solution is required.  This can be done using the Newton-Raphson
method.   Let
                         f«v>= E p±ai cibi e%s x- R
where R is the observed instrument response, and
                      f'(CT) -
                                            .-l
Then the iteration formula is
                         C    -C  -f(C)/f(C),
A reasonable starting value C  is R, the observed instrument response.
                             o

     The other mixed chemical adjustment method (Method 2) used for estimat-
ing actual concentration was a weighted logarithmic average.  In this case
(2)
                  log(RL) = Yl P± [log a± + %a| + b± log
where R^ is the estimated instrument response using a weighted logarithmic
average
     In contrast to the previously given weighted arithmetic average model
(equation 1) , this weighted logarithmic average model (equation 2) has an
explicit solution for actual total concentration:
                   = exp
                           log R -    ?.. (log a
                                      158

-------
     Both of the chemical mixture methods used the information on primary
and secondary chemicals and their percentage of the total concentration.
If their percentages did not total 100 percent, (i.e., there were other
chemical compounds in the line) the rest of the percentage was assigned a
response factor of 1.
                                      159

-------
                                 REFERENCES
1.  Blacksmith, J.  R.,  G. E,  Harris,  and G.  J.  Langley.   Frequency of  Leak
    Occurrence for  Fittings in Synthetic Organic Chemical Plant Process Units:
    Final Report.   EPA-600/2-81-003 (NTIS Ho.  PB81-14566), U.  S.  Environmental
    Protection Agency,  Research Triangle Park,  N. C.,  1980.

2.  Langley, G. J.  and  R. G.  Wetherold.   Evaluation of Maintenance for
    Fugitive VOC Emissions Control:  Final Report.   EPA-600/S2-81-080, U.  S.
    Environmental Protection Agency,  Cincinnati, Ohio, 1981.

3.  Wetherold, R, G, and L. P. Provost.   Emission Factors and  Frequency of
    Leak Occurrence for Fittings in Refinery Process Units.  EPA-600/2-79-044,
    U. S. Environmental Protection Agency, Washington, D. C.,  1979.

4.  Conover, "W. J.   Practical Nonparametric Statistics.   John  "Wiley and  Sons
    Inc., New York, 1971.

5.  Mood, A. M., F. A.  Graybill, and D.  C. Boes.  Introduction to the  Theory
    of Statistics,  Third Edition.  McGraw-Hill Book Company, New York, 1974.

6.  Beyer, W. H. (ed.).  CRC Handbook of Tables for Probability and Statistics,
    Second Edition.  The Chemical Rubber Co.,  Cleveland, Ohio, 1968.

7.  Brown, G. E., D. A. DuBose, W. R. Phillips, and G. E. Harris.  Response
    Factors of VOC  Analyzers Calibrated with Methane for Selected Organic
    Chemicals.  EPA-600/2-81-002 (NTIS No. PB81-136194), U. S. Environmental
    Protection Agency,  Research Triangle Park,  N.C., 1980.

8.  Wetherold, R, G, and D. D, Rosebrook,  Environmental Assessment of Atmo-
    spheric Emissions from Petroleum Refinery.   EPA-600/2-80-075a (NTI'S  No.
    PB80-225-253),  U. S. Environmental Protection Agency, Research Triangle
    Park, N. C., 1980.
                                      160

-------

-------
                                 APPENDIX A
                            SCREENING DATA SUMMARY
     Section 3 of this report contains an analysis of screening data collected
on an earlier EPA project (Reference 1).  This appendix gives detailed source
type groupings of this data.   Table A-l gives the number of sources screened,
the number leaking and the percent leaking for each possible source type and
for each type of stream service, including heavy liquids.
                                     161

-------
  TABLE A-l.  DATA SUMMARY  OF LEAK FREQUENCIES FOR VARIOUS
               SOURCES IN VARIOUS STREAM SERVICES
Source
Flange


Process Drain


Open-End Line


Agitator Seal


Relief Valve


Block Valve-
Gate Type

Block Valve-
Globe Type
•
Block Valve-
Plug Type

Block Valve-
Ball Type

Service
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
C-as
T.ighf Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Number
Screened
1450
2833
607
83
496
28
923
3605
477
7
8
1
84
68
3
6976
11017
2034
1 /. C
-ICC
129
440
2479
1031
1272
2732
251
Number
Leaking1
66
36
0
2
19
2
54
141
6
1
0
0
3
2
0
952
1059
9
L3
8
0
0
2'
0
18
4
4
Percent
Leaking1
4.5
1.3
0.0
2.4
3.8
7.1
5.8
3.9
1.3
14.3
0.0
0.0
3.6
2.9
o.o
13.6
9.6
0.4
10.3
1. 1
0.0
0.0 .
0.1
0.0
1.4
0.1
1.6
leaking defined as OVA reading >10,000 ppmv.
(continued)
                               162

-------
   TABLE A-l.  DATA SUMMARY OF  LEAK FREQUENCIES FOR VARIOUS
                SOURCES IN VARIOUS STREAM  SERVICES  (CONTINUED)
Source
Block Valve -
Butterfly Type
Block Valve -
Other Types
Control Valve -
Gate Type
Control Valve -
Globe Type
Control Valve -
Plug Type
Control Valve -
Ball Type
Control Valve - •
Butterfly Type
Control Valve -
Other Tir**es
On-Line Pump Seals
Single Mechanical-
Emission
Point at Seal
Service
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid
Gas
Light Liquid
Heavy Liquid

Light Liquid
Heavy Liquid
Number
Screened
160
157
8
275
378
35
61
182
27
207
417
107
10
91
0
15
33
3
91
34
6
•17
25
1

215
60
Number
Leaking
9
2
0
16
17
0
15
22
0
36
61
0
0
3
4
1
0
35
3
0
3
1
0

28
2
Percent
Leaking
5.6
1.3
0.0
5.8
4.5
0.0
24.6
12.1
0.0
17.4
14.6
0.0
0.0
3.0
26.7
3.0
0.0
38.5
8.9
0.0
17. 6
i.n
0.0

13.0
3.3
leaking defined as OVA reading ^10,000 ppmv.
                                                    (continued)
                                163

-------
   TABLE A-l.   DATA SUMMARY  OF LEAK FREQUENCIES FOR VARIOUS
                  SOURCES IK VARIOUS  STREAM  SERVICES  (CONTINUED)
Source
                      Service
                                           Number
                                          Screened
                                     Number
                                     Leaking,1
                           Percent
                           Leaking1
On-Line Pump  Seals
Single Mechanical -
Emission
Point at Vent

On-Line Pump  Seals
Single Mechanical -
Other
Emission Point

On-Line Pump  Seals
Double Mechanical -
Emission
Point at Seal

On—Line Pump  Seals
Double Mechanical -
Emission
Point at Vent

On-Line Pump  Seals
Double Mechanical -
Other
Emission Point

On-Line Pump  Seals
Single, Packed,
Emission
Point at Seal

On—Line Pump  Seals
Single, Packed,
Emission
Point at Vent

On-Line Pump  Seals
Single, Packed,
Other
Emission Point

On-Line Pump  Seals
Sealess Pumps
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
Light-Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
24
 0
30
 0
92
 2
13
 0
               0.0
               6.7
14.1
 0.0
                              33.3
                               0.0
                               0.0
                               0.0
1 Leaking defined as OVA reading 3*10,000 ppmv.
                                                              (continued)
                                       164

-------
    TABLE A-l.   DATA  SUMMARY  OF  LEAK FREQUENCIES FOR VARIOUS
                   SOURCES  IN VARIOUS  STREAM SERVICES  (CONTINUED)
Source
                      Service
                     Number
                    Screened
            Number
            Leakingl
Percent
Leaking1
Off-Line Pump Seals
Single Mechanical -
Emission
Point at Seal

Off-Line Pump Seals
Single Mechanical -
Emission
Point at Vent

Of f-Lin& Pump Seals
Single Mechanical -
Other
Emission Point

Off-Line Pump Seals
Double Mechanical -
Emission
Point at Seal

Off-Line Pump Seals
Double Mechanical -
Other
Off-Line Pump Seal
Single, Packed
Emiss ion
Point at Seal

Off-Line Pump Seals,
Single, Packed
Emission
Point at Vent

Off-Line Pump Seals,
Single, Packed
Other
Emission Point
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
                      Light Liquid
                      Heavy Liquid
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
Light Liquid
Heavy Liquid
139
 24
 17
  1
 86
  1
                         19
    6.5
    0.0
                               0.0
    0.0
    0.0
    3.5
    0.0
                                                      0.0
                               0.0
                               0.0
                                                       0.0
 1Leaking defined as OVA reading  10,000 ppmv.
                                             (continued)
                                       165

-------
     TABLE  A-l.   DATA  SUMMARY OF  LEAK FREQUENCIES FOR VARIOUS
                    SOURCES  IN VARIOUS  STREAM  SERVICES  (CONTINUED)
Source
                      Service
                                           Number
                                          Screened
                                     Number
                                     Leaking1
Percent
Leaking!
Off-Line Pump Selas,
Sealess Pumps
On-Line Compressor
Seals, Single,
Mechanical,  Emission
Point at Seal
Light Liquid
Heavy Liquid
Gas
On-Line Compressor
Seals, Single,
Mechanical,  Emission
Point at Vent          Gas
On-Line Compressor
Seals, Single,
Mechanical,  Other
Emission Point
                      Gas
                                                                          33.3
On-Line Compressor
Seals, Double,
Mechanical,  Emission
Point at Seal         Gas
                                                                          0.0
On—Line Compressor
Seals, Double,

Point it Vent
                                                                          0.0
On—Line Compressor
Seals, Double,
Mechanical,  Other
Emission Point
                      Gas
                                                                          0.0
On—Line Compressor
Seals, Single,
Packed, Emission
Point at Seal
                      Gas
                                                                          0.0
On-Line Compressor
Seals, Single,
Packed, Emission
Point at Seal
                      Gas
                                                                        100.0
'Leaking defined as OVA reading ^10,000 ppmv.
                                        (continued)
                                      166

-------
TABLE A-l.   DATA SUMMARY OF LEAK FREQUENCIES  FOR VARIOUS
              SOURCES  IN VARIOUS  STREAM SERVICES (CONTINUED)
                                     Hunfoer       Humber       Percent
 jaource               Service        Screened	Leaking *      Leaking

 On-Line Compressor
 Seals, Single,
 Packed, Other
 Emission Point        Gas                1            0      '     0.0

 Other Source Types     Gas               19            3          15.8
                      Light Liquid       33            2           6.1
                      Heavy Liquid        2            0           0.0
  1Leaking defined as OVA reading ^10,000 ppmv.
                                167

-------

-------
                                 APPENDIX B
                           DETAILED INFORMATION ON
                     LINE TEMPERATURE AND LINE PRESSURE
     Section 3 contains an analysis of the effect of temperature and
pressure on leak frequency.  This appendix contains statistical information
on temperature and pressure.

     Tables B—1 to B-3 contain summary statistics for line pressure and line
temperature for gas, light liquid, and heavy liquid stream services.  Separate
values are given for each of the major source types.  Differences between the
types of processes, ethylene versus high leaking, and also between the groups
of primary chemicals in the line can be seen at this stage.  For example, the
average line temperature for the high leaking process units appears to be much
higher than that for the ethylene units.  The minimum temperature for the
ethylenes is also much lower.  Line pressure seems to differ more by type of
chemical in the line.  The heavy liquids are not broken down by primary
material groups in the line since they had a low leak frequency.

     Although line temperature and pressure were recorded as continuous
variables, they are grouped for evaluating leak frequency.  Tables B-4 to B-16
                                #
show the number screened, percent screened, number leaking, and percent
leaking at different levels of temperature and pressure.  This information is
given for ethylene process units and for high leaking process units and also
for primary "-material groups for all source types but pump seals.  Possible
reasons for some of the differences in leak frequencies for the different
categories can be seen from these tables.  None of the high leaking group
sources are at very low temperatures.  This group also has some screening
values for each source type at the higher temperatures.  The ethylene group
exhibits a different distribution of temperatures.  There are some values in
                                     168

-------
the very low temperature group; and on the average, the temperatures found in
the ethylene unit sources are lower.  If the data were not separated into
these groups, differences that were actually attributable to the type of
process unit might appear to be due to line temperature.

     Figures B-l to B-4 show the distributions of the sources screened as a
function of line temperature and line pressure for valves in gas and light
liquid service.
                                      169

-------
                                      TABLE  B-l.   SUMMARY  STATISTICS FOR LINE TEMPERATURE

                                                   AND LINE PRESSURE FOR GAS SERVICE
H
^J
O
High Leaking

Variable
Line Temperature
(°F)





,.




Line Pressure
(psig)










PRIMARY CHEMICAL Group * :
Type Statistics
Valves Average
Standard Deviation
Minimum
Maximum
Flanges Average
Standard Deviation
Minimum
Maximum
Open Ended Average
Lines Standard Deviation
Minimum
Maximum
Valves Average
Standard Deviation
Minimum
Max imum
Flanges Average
t Standard Deviation
Minimum
Maximum
Open Ended Average
Lines Standard Deviation
Minimum
Maximum
Group 5

210.8
153.2
30
825
271.3
146.8
30
800
128.2
82.1
30
392
184.7
167.4
-10
600
273.6
184.9
-9
600
132.0
142.5
0
600
Process Units
Group 6

217.0
195.4
20
1000
235.3
228.5
20
1000
218.1
190.6
20
1000
56.4
99.5
-15
650
37.7
85.0
-15
590
52.1
92.9
-15
450
Ethvlene Process Units
Group 1 Group 2

60.7
101.8
-267
1570
73.3
94.0 v •-
-267
750
46.1
74.7
-267 - .
720
166.7
178.7
0
1050
184.6
160.0
0
805
120.6
160.3
1
805
            lSee Figure 3-2 for explanation of groups.

-------
                 TABLE B-2.   SUMMARY  STATISTICS FOR LINE TEMPERATURE
                              AND LINE PRESSURE FOR LIGHT LIQUID  SERVICE


Variable
Line Temperature
<8F)














Line Pressure
(psig)













'

PRIMARY CHEMICAL Group1:
Type Statistics
Valves Average
Standard Deviation
Minimum
Maximum
Pump Seals Average
Standard Deviation
Minimum
Maximum
Flanges Average
Standard Deviation
Minimum
Maximum
Open Ended Average
Lines Standard Deviation
Minimum
Maximum
Valves Average
Standard Deviation
Minimum
Maximum
Pump Seals Average
Standard Deviation
Minimum
Maximum
Flanges Average
Standard Deviation
Minimum
Maximum
Open Ended Average
Lines Standard Deviation
Minimum
Maximum
High Leaking
Group 7

147.2
94.1
20
500
129.5
82.4
32
540
165
93
20
500
142.8
100.6
30
500
161.0
190.5
-10
740
116.2
152.8
0
720
247,3
210.4
-9
740
123.7
171.1
0
740
Processing Units
Group 8

145.5
90.8
15
1000
133.8
68.2
32
345
148.3
102.8
30
1000
137.2
83
20
1000
80.2
78.5
-20
700
79.3
77.5
0
700
70,6
77.0
-20
700
66,9
67.4
-20
700
Ethylene Process Units
Group 3

20.3
86.4
-267
190
8.0
73.6
-145
118
44.7
71.9
-212
190
44.4
84.5
-267
190
372.0
368.1
0
2270
512.2
427.6
80
1960
380.7
396.3
0
2270
379.4
383,9
0
2270
Group 4

129-4
57.2
40
235
112.6
70.8
40
235
128.4
49.6
40
235
154,6
74.7
40
235
101.8
111.6
2
500
65.1
57.4
2
165
79.2
79.8
2
500
75.5
96.6
0
500
*See Figure 3-2 for explanation of groups.

-------
TABLE B-3.  SUMMARY STATISTICS FOR LINE TEMPERATURE AND LINE PRESSURE IN HEAVY
            LIQUID SERVICE WITHIN HIGH AND ETHYLENE PROCESS UNITS

Variable Type Statistics
Line Temperature Valves Average
(°F) Standard Deviation
Minimum
Maximum
Pump Seals Average
Standard Deviation
Minimum
Maximum
Flanges Average
Standard Deviation
Minimum
Maximum
Open Ended Average
Lines Standard Deviation
Minimum
Maximum
Line Pressure Valves Average
(psig) Standard Deviation
Minimum
Maximum
Pump Seals Average
Standard Deviation
Minimum
Maximum
Flanges Average
Standard Deviation
Minimum
Maximum
Open Ended Average
Lines Standard Deviation
Minimum
Maximum
High Leaking
Process Units
228.8
177.7
60
600
153.6
146.8
72
460
219.9
63.4
60
500
93.9
41.1
60
260
58.5
55.4
1
230
78.6
10.2
62
92
63.9
67.3
1
230
50.0
40.4
1
120
Ethylene Leaking
Process Units
128.1
70.0
25
370
168.1
60.5
90
300
124.8
63.8
25
300
156.6
90.1
60
30
97.0
^ 110.4
0
540
48.4
57.7
0
170
89.5
112.9
0
320
68.8
95.9
0
480

-------
         TABLE  B-4.  EFFECTS OF LINE TEMPERATURE AND LINE PRESSURE
                     ON PERCENT LEAKING FOR VALVES IN GAS SERVICE
                     WITHIN ETHYLENE PROCESS UNITS




Pressure (psie)
-15 - -1
0 - 49
50 - 99
100 - 149
150 - 199
200 - 249
250 - 299
300 - 349
350 - 399
400 - 449
450 - 499
500 - 549
550 - 999
1000 - 1050



Number
Screened
0
2123
1072
681
157
321
502
196
144
94
267
316
167
4

Group
Percent
of
Total
Screened
0
35.1
17.8
11.3
2.6
5.3
8.3
3.2
2.4
1.6
4.4
5.2
2.8
0.1

1s PRIMARY CHEMICALS


Number
Leaking
_
94
171
110
36
73
136
51
54
20
54
91
37
2



Percent
Leaking
_
4.4
16.0
16.2
22.9
22.7
27.1
26.0
37.5
21.3
20.2
28.9
22.2
50.0

TOTAL
6043
929
15.4
Temperature p
-267 -
0 -
50 °F -
100 °F -
150 °F -
200°F -
250°F -
300 °F -
350°F -
400 °F -
TOTAL
-1
49
99
149
199
249
299
349
399
1570

998
1452
2035
1011
373
78
32
4
19
48
6050
16.5
24.0
33.6
16.7
6.2
1.3
0.5
0.1
0.3
0.8

140
236
327
111
72
27
8
0
7
4
932
14.0
16.2
16.1
11.0
19.3
34.6
25.0
0.0
36.8
8.3
15.4
JSee Figure 3-2 for explanation of groups.
                                    173

-------
           TABLE B-5.  EFFECTS OF LINE TEMPERATURE AND LINE PRESSURE
                        ON PERCENT LEAKING FOR  VALVES IN  GAS SERVICE
                        WITHIN HIGH LEAKING PROCESS UNITS BY CHEMICAL
                        GROUP
Group1:






Pressure (pslg)
-15 -
0 -
50 -
100 -
150 -
200 -
250 -
300 -
350 -
400 -
450 -
500 -
550 -
1000 -
-1
49
99
149
199
249
299
349
399
449
499
549
999
1050


Hunter
Screened
5
203
282
267
53
87
136
10
15
0
80
37
53
0
'.Group 31 PRIMARY CHEMICALS
Percent
of
Total
Screened
0.4
16.5
23.0
21.7
4.3
7.1
11.1
0.8
1.2
0.0
6.5
3.0
4.3
0.0


Number
Leaking
0
9
31
30
6
9
13
0
4
-
21
9
14
-


Percent
Leaking
O.D
4.4
11.0
11.2
11.3
10.3
9.5
0.3
26.7
-
26.2
24.3
26.4
~


Number
Screened
66
1201
141
136
34
18
20
84
13
13
6
18
8
0
Group -41 PRIMARY CHEMICALS
Percent
of
Total
Screened
3.8
68.3
8.0
7.7
1.9
1.0
1.1
4.8
0.7
0.7
0.3
1.0
0.5
0.0


Number
Leaking
0
7
2
12
1
0
0
0
0
0
0
0
0
—


Percent
Leaking
0.0
0.6
1.4
8.8
2.9
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
~
  TOTAL
             1228
                                146
                                        11.9
                                                 1758
                                                                     22
                                                                             1.2
Temperature (Pp>
-267 -
0 -
50 °F -
100 "F -
150 °F -
200 8F -
250°F -
300 °F -
350 "F -
400°F -
TOTAL
-1
49
99
149
199
249
299
349
399
1570

0
12
243
355
127
43
77
109
113
146
1225
0.0
1.0
19.8
29,0
10.4
3.5
6.3
8.9
9.2
11.9

-
1
27
31
9
2
8
14
22
31
145
-
8.3
11.1
8.7
7.1
4.6
10.4
12.8
19.5
21.2
11.8
0
45
335
454
369
156
113
63
7
216
1758
0.0
2.6
19.1
25.8
21.0
8.9
6.4
3.6
0.4
12.3

-
0
4
2
0
2
1
1
0
12
22
-
0.0
1.2
0.4
0.0
1.3
0.9
1.6
0.0
5.6
1.2
   See Figure 3-2 for explanation of groups.

-------
                                    TABLE B-6.   EFFECTS  OF LINE TEMPERATURE  AND LINE
                                                  PRESSURE ON PERCENT LEAKING  FOR VALVES
                                                  IN LIGHT LIQUID SERVICE WITHIN ETHYLENE
                                                  PROCESS  UNITS BY  CHEMICAL  GROUP
                                           GROUP 31 PRIMARY CHEMICALS
                                                                              GROUP 4 PRIMARY CHEMICALS
Ui






Pressure (pale)
-15 -
50 -
50 -
100 -
150 -
200 -
250 .-
300 -
350 -
400 -
450
500 -
550 -
1000 -

-1
49
99
149
199
249
299
349
399
449
. 499
549
999
1050
TOTAL


Number
Screened
0
480
363
226
229
138
477
141
310
109
273
242
312
211
3511
Percent
of
Total
Screened
0.0
13,7
10.3
6.4
6.5
3.9
13.6
4.0
8.8
3.1
7.8
6.9
8.9
6.0



Number
Leaking
_
57
66
54
56
55
156
43
94
28
59
93
97
99
957


Per :ent
Le iking

111.9
18. 2
11,9
>4.4
VI.9
C1.7
Jtl.5
311.3
. 25.7
.21.6
38.4
31.1
46.9
2/..3


Number
Screened
0
173
215
107
54
20
0
0
0
0
0
38
0
_JL
607
Percent
of
Total
Screened
0.0
28.5
. 35.4
17.6
8.9 ,
3.3
0.0
0.0
0.0
0.0
0.0
6.3
0.0
0.0



Number
Leaking

0
4
1
0
3
_
_
-
_
_
1
-
-
9
•

Percent
Leaking

0.0
1.9
0.9
0.0
15.0
_
_
_
_
_
2.6
_
-
1.5
Temperature (°F)
-267 -
0 -
50 -
100 -
150 -
200 -
250 -
300 -
350 -
400 -

-1
49
99
149
199
249
299
349
399
1570
TOTAL
1349
724
829
500
108
0
0
0
0
0
3510
38.4
20.6
23.6
14.2
3.1
0.0
0.0
0.0
0.0
0.0

265
207
287
177
21
-
-
-
-
-
957
l'J.6
28.6
34.6
33.4
1').4


_
~
-
27.3
0
29
127
252
89
110
0
0
0
0
607
0.0
4.8
20.9
41.5
14.7
18.1
0.0
0.0
0.0
0.0

_
0
2
3
3
1
_
_
_
-
9
_
0.0
1.6
1.2
3.4
0.9
_
_
_
_
1.5
                            'See Figure 3-2 fur explanation of groups.

-------
             TABLE B-7.   EFFECTS OF  LINE TEMPERATURE AND LINE
                          PRESSURE ON PERCENT LEAKING FOR VALVES
                          IN LIGHT LIQUID SERVICE WITHIN HIGH
                          LEAKING PROCESS UNITS BY  CHEMICAL GROUP








Pressure (pela)
-15
0
50
100
150
200
250
300
350
400
450
500
550
1000
- -1
- 49
- 99
- 149
- 199
- 249
- 299
- 349
- 399
- 449
- 499
- 549
- 999
- 1050
TOTAL



Number
Screened
6
1075
702
383
317
100
45
168
29
33
36
174
224
3292
Group
Percent
of
Total
Screened
0.2
32.6
21.3
11.6
9.6
3.0
1.4
5.1
0.9
1.0
.1.1
5.3
6.8

7 '* PRIMARY CHEMICALS


Number
Leaking
1
20
21
20
16
9
6
10
4
5
3
15
17
147


Percent
Leaking
16.7
1.9
3.0
5.2
5.0
9.0
13.3
6.0
13.8
15.2 '
8.3
8.6
7.6
A. 5


Number
Screened
21
2411
1567
1034
621
261
21
140
24
A
0
0
16
6120
Group 81 PRIMARY CHEMICALS
Percent
of
Total
Screened
0.3
39.4
25.6
16.9
10.2
4.3
0.3
2.3
0.4
0.1
0.0
0.0
0.3



Number
Leak Inn
0
16
5
13
3
2
0
5
0
0
-
-
2
46


Percent
Leaking
0.0
0.7
0.3
1.3
0.5
0.8 i
0.0
3.6
0.0
. 0.0
-
-
12.5
0.8
Temperature (°F)
-267 -
0 -
50"F -
100 °F -
150 °F -
200"F -
250°F -
300 °F -
350°F -
400°F -
-1
49
99
149
199
2A9
299
349
399
1570
TOTAL
0
122
1039
819
287
634
89
161
15
112
3278
0.0
3.7
31.7
25.0
8.8
19.3
2.7
4.9
0.5
3. A

-
7
38
36
22
2A
3
11
1
2
144
-
5.7
3.7
A. A
7.7
3.8
3. A •
6.8
6.7
1.8
A. A
0
96
1912
1922
661
743
512
158
68
82
6154
0,0
1.6
31.1
31.2
10.7
12.1
8.3
2.6
1.1
1.3

-
2
11
10
6
3
9
0
0
5
A6
-
2.1
0.6
0.5
0.9
0.4
1.8
0.0
0.0
6.1
0.8
    See Figure 3-2 for explanation of groups.

-------
TABLE B-8.  EFFECTS OF LINE TEMPERATURE AND LINE PRESSURE ON
            PERCENT LEAKING FOR PUMP SEALS IN LIGHT LIQUID
            SERVICE
Pres:
-15
0
50
100
150
200
250
300
350
400
450
500
550
1000


sure (psig)
-1
- 49
99
- 149 •
- 199
- 249
- 299
- 349
- 399
- 449
- 499
- 549
- 999
- 1050
TOTAL
Number
Screened

0
146
115
66
44
6
8
15
9
6
8
6
13
0
442
Percent of
Total
Screened

0.0
33.0
26.0
14.9
9.9
1.4
1.8
3.4
2.0
1.4
1.8
1.4
2.9
0.0

Number
Leaking


10
19
3
8
0
3
0
1
3
1
2
2
-
52
Percent
Leaking


6.8
16.5
4.5
18.2
0.0
37.5
0.0
11.1
50.0
12.5
33.3
15.4
-
11.8
Temperature (°F)
-267
0
50
100
150
200
250
300
350
400

-1
- 49
99
- 149
- 199
- 249
- 299
- 349
- 399
- 1570
TOTAL
26
27
148
112
34
73
21
4
0
2

447
5.8
6.0
33.1
25.1
7.6
16.3
4.7
0.9
0.0
0.4

8
4
14
11
2
8
1
0
—
0

48
30.8
14.8
9.5
9.8
5.9
11.0
4.8
0.0
—
0.0
10.7
                             177

-------
                 TABLE B-9.  EFFECTS OF LINE TEMPERATURE AND LINE PRESSURE
                             FOR FLANGES IN GAS SERVICE FROM ETHYLENE PRO-
                             CESS UNITS

GROUP I1 PRIMARY MATERIALS




Pressure (psig)
-15
0
50
100
150
200
250
300
350
400
450
500
550
1000
-1
- 49
- 99
- 149
- 199
- 249
- 299
- 349
- 399
- 449
- 499
- 549
- 999
- 1050

Number
Screened
0
145
102
36
34
65
66
30
22
6
18
29
11
0
Percent
of
Total
Screened
0.0
25.7
18.1
6.4
6.0 .
11.5
11.7
5.3
3.9
1.1
3;2
5.1
2.0
0.0

Number
Leaking

3
4
2
4
2
13
4
3
1
1
2 '
0 '
-

Percent
Leaking
_
2.1
3.9
5,6
11.8
3.1
19.7
13.3
13.6
16.7
5.6
6.9
0.0
-
TOTAL
564
39
6.9
Temperature  (°F)
•267' -
0' -
50 -
100 -
150
200
250
300
350
400 -
TOTAL
-1
49
99
149
199
249
299
349
399
1570

71
58
270
117
35
8
0
1
1
5
566
12.5
10.2
47.7
20.7
6.2
1.4
0.0
0.2
0.2
0.9

7
10
15
4
1
1
-
0
0
1
39
9.9
17.2
5.6
3.4
2.9
12.5
-
0.0
0.0
20.0
6.9
 See Figure 3-2 for explanation of groups
                                    178

-------
                 TABLE B-10.   EFFECTS OF LINE TEMPERATURE AND LINE
                               PRESSURE FOR FLANGES  WITH GAS  SERVICE
                               FROM HIGH LEAKING PROCESS UNITS BY
                               CHEMICAL GROUP
GROUP 51 PRIMARY CHEMICALS






Pressure (psig)
-15
0
50
100
150
200
250
300
350
400
450
500
530
1000
-1
- 49
- 99
- 149
- 199
- 249
- 299
- 349
- 399
- 449
- 499
- 549
- 999
- 1050


Number
Screened
1
34
53
46
5
52
72
10
2
0
52
33
31
0
Percent
of
Total
Screened
0.3
8.7
13.6
11.8
1.3
13.3
18.4
2.6
0.5
0.0
13.3
8.4
7.9
0.0


Number
Leaking
0
1
1
1
0
1
2
0
0
-
6
5
1
-


Percent
Leaking
0.0
2.9
1.9
2.2
0.0
1.9
2.8
0.0
0.0
-
11.5
15.2
3.2
-
GROUP 61 PRIMARY CHEMICALS


Suiter
Screened
17
245
23
7
6
1
2
11
0
0
0
0
4
0
Percent
of
Total
Screened
5.4
77.5
7.3
2.2
1.9
0.3
0.6
3.5
0.0
0.0
0.0
0.0
1.3
0.0


Number
Leaking
~0
5
1
0
0
0
0
1
—
-
-
-
2
-


Percent
Leaking
0.0
2.0
4.4
0.0
0.0
0.0
0.0
9.1
-
-
-
-
0.0
-
  TOTAL
              391
                                 18
                                         4.6
                                                  316
                                                                              2.8
Temperature (CF)
•267
0
5"
100 -
150 -
200 -
250 -
30C -
350 ' -
400 -
TOTAL
-1
49
99
149
199
249
299
349
399
1570

0
1
44
75
36
11
22
68
51
83
391
0.0
0.3
11.2
19.2
9.2
2.8
5.6
17.4
13.0
21.2

0
0
1
2
0
0
0
3
5
7
18
0.0
0.0
2.3
2.7
0.0
0.0
0.0
4.4
9.8
8.4
4.6
0
15
51
95
62
22
5
15
3
48
316
0.0
4.8
16.1
30.1
19.6
7.0
1.6
4.8
1.0
15.2

-
0
0
2
0
3
0
2
0
2
9
-
0.0
0.0
2.1
0.0
13.6
0.0
13.3
0.0
4.2
2.8
  'See Figure 3-2 for explanation of groups.
                                        179

-------
            TABLE B-ll.  EFFECTS  OF LINE TEMPERATURE AND LINE
                           PRESSURE FOR FLANGES IN  LIGHT  LIQUID
                           SERVICE  WITHIN ETHYLENE  PROCESS UNITS
                           BY CHEMICAL  GROUP
                    GROUP 31 PRIMARY CHEMICALS
                                                          GROUP 4l PRIMARY CHEMICALS






Pressure (psig)
-15 -
0 -
50 -
100 -
150 -
200 -
250 -
300 -
350 -
400 -
450 -
500 -
550 -
1000 -
-1
49
99
149
199
249
299
349
399
449
499
549
999
1050


Number
Screened
0
39
32
24
16
15
42
19
32
4
32
21
23
25
Percent
of
Total
Screened
0.0
12.0
9.9
7.4
4.9
4.6
13.0
5.9
9.9
1.2
9.9
6.5
7.1
7.7


Number
Leaking
_
1
0
2
3
0
1
2
2
0
6
4
0
4


Percent
Leaking
_
2.6
0.0
8.3
18.7
0.0
2.4
10.5
6.2
0.0
18.7
19.0
0.0
16.0


Number
Screened
0
31
19
10
5
4
0
0
0
0
0
1
0
0
Percent
of
Total
Screened
0.0
44.3
27.1
14.3
7.1
5.7
0.0
0.0
0.0
0.0
0.0
1.4
0.0
0.0


Number
Leaking
_
0
0
0
0
0
-
-
-
-
-
0
-
-


Percent
Leaking
_
0.0
0.0
0.0
0.0
0.0
-
-
-
_
-
0.0
_
-
    TOTAL
Temperature C°F)
              324
    TOTAL
              324
                                  25
                                           7.7
                                                     70
                                  25
                                           7.7
                                                     70
                                                                                 0.0
•267 -
0 -
50 -
100 -
150 -
200 -
250 -
300 -
350 -
400 -
-1
49
99
149
199
249
299
349
399
1570
66
63
125
60
10
0
0
31
36
44
20.4
19.4
38.6
18.5
3.1
0,0
0.0
20.5
23.8
29.1
7
6
10
2
0
-
-
7
8
13
2.2
9.5
8.0
3.3
0.0
-
-
22.6
22.2
29.5
0
2
8
40
11
9
0
0
0
0
0.0
2.9
11.4
57.1
15.7
12.9
0.0
0.0
0.0
0.0
-
0
0
0
0
0
-
-
-
-
-
0.0
0.0
0.0
0.0
0.0
_
-
-
-
                                                                                 0.0
1See Figure 3-2 for explanation of groups.
                                          180

-------
           TABLE B-12.  EFFECTS  OF LINE  TEMPERATURE AND LINE
                         PRESSURE ON PERCENT LEAKING FOR FLANGES
                         IN LIGHT LIQUID  SERVICE  WITHIN  HIGH
                         LEAKING  PROCESS  UNITS BY CHEMICAL GROUP
                   GROUP 71 PRIMARY CHEMICALS
GROUP 81 PRIMARY CHEMICALS






Pressure (psig)
-15
0
50
100
150
200
250
300
350
WO
450
500
550
1000
- -I
- 49
- 99
- 149
- 199
- 249
- 299
- 349
- 399
- 449
- 499
- 549
- 999
- 1050


Hunter
Screened
1
116
82
63
30
35
36
30
16
13
22
70
54
0
Percent
of
Total
Screened
0.2
20.4
14.4
11.1
5.3
6.2
6.3
5.3
2.8
2.3
3.9
12.3
9.5
0.0


Hunter
Leak! OR
0
0
0
2
0
0
1
3
0
0
0
2
2
-


Percent
Leakirg
0.0
0.0
0.0
3.2
0.0
0.0
2.8
10.0
0.0.
0.0
0.0
2.9
3.7
—


Humber
Screened
6
446
282
117
61
40
10
23
4
0
0
0
1
0
Percent
of
Total
Screened
0.6
45.0
28.5
11.8
6.2
4.0
1.0
2.3
0.4
0.0
0.0
0.0
0.1
0.0


Number
Leaking
0
0
0
0
0
0
0
0
0
-
-
-
0
-


Percent
Leaking
0.0
0.0
0.0
0.0
0.0
o.o
0.0
0.0
0.0
_
-
-
0.0
_'
  TOTAL
              568
                                  10
                                          1.8
                                                   990
                                                                              0.0
Temperature (°F)
-267
0
50
100
150
200
250
300
350
400

-1
- 49
- 99
- 149
- 199
- 249
- 299
- 349
- 399
- 1570
TOTAL
0
14
143
125
80
125
21
32
14
18
572
0.0
2.4
25.0
21.8
14.0
21.8
3.7
5.6
2.4
3.2

-
0
2
0
6
1
1
0
0
0
10
-
0^0
1.4
0.0
7.5
0.8
4.8
0.0
0.0
0.0
1.8
0
12
310
309
124
134
69
12
7
33
1010
0.0
1.2
30.7
30.6
12.3
13.3
6.8
1.2
0.7
3.3

-
0
0
0
0
0
0
0
0
0
0
_
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
o.o
0.0
     See Figure 3-2 for explanation of groups.
                                        181

-------
     TABLE  B-13.  EFFECTS OF LINE  TEMPERATURE AND LINE PRESSURE ON
                  PERCENT LEAKING  FOR OPEN-ENDED LINES IN GAS
                  SERVICE WITHIN ETHYLENE PROCESS  UNITS
GROUP I1 PRIMARY CHEMICALS
Pressure (psig)
-15 - -1
0 - 49
50 - 99
100 - 149
150 - 199
200 - 249
250 - 299
300 - 349
350 - 399
400 - 449
450 - 499
500 - 549
550 - 999
1000 - 1050
TOTAL
Temperature (°F)
•267 - -1
0-49
50 ' - 99
100 - 149
150 - 199
200 - 249
250 - 299
300 - 349
350 - 399
400 - 1570
TOTAL
Number
Screened

0
139
53
12
3
20
20
5
2
2
12
2
12
0
282

42
131
77
17
12
3
0
1
0
	 1
284
Percent of
'Total
Screened

0..0
49.3
18.8
4.3
1.1
7.1
7.1
1.8
0.7
0.7
4.3
0.7
4.3
0.0


14.8
- 46.1
27.1
6.0
4.2
1.1
0.0
0.4
0.0
0.4

Number
Leaking

—
6
7
3
0
0
5
3
0
1
5
2
4
-
36

8
8
17
3
1
0
-
0
-
0
37
Percent
Leaking

—
4.3
13.2
25.0
0.0
0.0
25.0
60.0
0.0
50.0
41.7
100.0
33.3
-
12.8

19.0
6.1
22.1
17.6
8.3
0.0
-
0.0
-
0.0
13.0
1See Figure 3-2 for explanation of groups.
                                    182

-------
TABLE  B-14.  EFFECTS OF  LINE TEMPERATURE  AND LINE  PRESSURE  ON PERCENT LEAKING
              FOR OPEN-ENDED LINES  IN GAS  SERVICE WITHIN HIGH LEAKING  PROCESS
              UNITS BY CHEMICAL GROUP

                   GROUP 51 PRIMARY CHEMICALS                      GROUP 61 PRIMARY CHEMICALS


Pressure (psig)
-15 - -1
0 - 49
50 - 99
100 - 149
150 - 199
200 - 249
250 - 299
300 - 349
350 - 399
400 - 449
450 - 499
500 - 549
550 - 999
1000 - 1050
TOTAL
Temperature (°F)
-267 - _1
0-49
50 - 99
100 - 149
150 - 199
200 - 249
250 - 299
300 - 349
350 - 399
400 - 1570
TOTAL

Number
Screened

0
29
28
42
5
3
7
0
4
0
0
. 0
4
0
122

0
2
37
49
13
6
4
1
9
0
121

Percent of
- Total
Screened

0.0
23.8
23.0
34.4
4.0
2.5
5.0
0.0
3.3
0.0
0.0
0.0
3.3
0.0


0.0
1.6
30.6
40.5
10.7
5.0
3.3
0.8
7.4
0.0


Number
Leaklnit .

-
3
3
1
0
0
2
-
0
-
-
-
2
-
11

-
0
8
2
0
1
0
0
0
_
11

Percent
LeakinR

-
D.3
L0.7
1. 4
0.0
0.0
ifl.6
-
0
. -
-
-
5D. 0

9.0

-
0,0
n.,6
/..i
0.0
lb.7
II .0
n.o
0.0
-
!'.09

Number
Screened

16
227
31
5
12
7
2
29
1
3
1
0
0
0
334

0
11
46
84
79
36
21
17
0
40
334

Percent of
Total
Screened

4.8
68.0
9.3
1.5
3.6
2.1
0.6
8.7
0.3
0.9
0.3
0.0
0.0
0.0


0.0
3.3
13.8
25.2
23.6
10.8
6.3
5.1
0.0
12.0


Number
LeakinR

0
2
1
0
0
0
0
0
0
0
0
-
-
-
3

-
0
0
2
1
0
0
0
-
0
3

Percent
Leaklna

0.0
0.9
3.2
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
-
-
-
0.9

-
0.0
0.0
2.4
1.3
0.0
0.0
0.0
-
0.0
0.9
  'See Figure 3-2 for explanation of groups.

-------
                 TABLE  B-15.  EFFECTS OF  LINE TEMPERATURE AND LINE PRESSURE ON  PERCENT LEAKAGE
                               FOR OPEN-ENDED LINES IN LIGHT LIQUID SERVICE WITHIN ETHYLENE
                               PROCESS UNITS BY  CHEMICAL  GROUP
                                        GROUP 31 PRIMARY CHEMICALS
GROUP 41 PRIMARY CHEMICALS
00


Number
Screened
Percent of
Total
Screened
Number
Leaking
Percei
Leak!
Pressure (pale)
-15
0
50
100
150
200
250
300
350
400
450
500
550
1000

-1
- 49
- 99
- 149
- 199
- 249
- 299
- 319
- 399
- 449
- 499
- 549
- 999
- 1050
TOTAL
0
18
9
16
7
6
14
12
23
1
12
12
7
14
151
0.0
11.9
6.0
10.6
4.6
4.0
9.3
8.0
15.2
0.7
8.0
8.0
4.6
9.3

_
0
0
3
2
2
6
6
5
0
1
2
5
7
39
_
0.0
0.0
18.7
28.6
33.3
42.9
50.0
21.7
0.0
8.3
16.7
71.4
50.0
25.8
Temperature (°F)
-267
0
50
100
150
200
250
300
350
400
- -1
- 49
- 99
- 149
- 199
- 249
- 299
- 349
- 399
- 1570
31
36
44
36
4
0
0
0
0
0
20.5
23.8
29.1
23.8
2.6
0.0
0.0
fi.O
0.0
0.0
7
8
13
11
0
_
-
,
-
-
22.6
22.2
29.5
30.6
0.0
-
_
-
-
-
Number
Screened
0
12
39
6
3
0
0
0
0
0
0
3
0
0
63
0
8
12
12
10
21
0
0
0
0
Percent of
Total
Screened
0.0
19.0
61.9
9.5
4.8
0.0
0.0
0.0
0.0
0.0
0.0
4.8
0.0
0.0

0.0
12.7
19.0
19.0
15.9
33.3
0.0
0.0
0.0
0.0
Number
Leaking

0
2
0
0
-
-
_
-
-
-
0
-
-
2
_
0
0
0
0
2
-
-
-
-
Percent
Leakitifi

0.0
5.1
0.0
0.0
-
-
-
-
-
-
0.0
-
-
3.2
_
0.0
0.0
o.o
0.0
9.5
-
-
-
-
                      TOTAL
                                151
                                                     39
                                                              25.8
                                                                           63
                                                                                                         3.2
                   See Figure 3-2 for explanation of groups.

-------
TABLE  B-16.  EFFECTS OF  LINE TEMPERATURE AND LINE PRESSURE ON PERCENT  LEAKING
              FOR OPEN-ENDED LINES  IN LIGHT LIQUID SERVICE WITHIN HIGH  PROCESS
              UNITS BY  CHEMICAL  GROUP
                      GROUP 71 PRIMARY CHEMICALS
GROUP 81 PRIMARY CHEMICALS
Number
Screened
Percent of •
Total
Screened
Pressure (palg)
-15
0
50
100
150
200
250
300
350
400
450
500
550
1000
- -1
- 49
- 99
- 149
- 199
- 249
- 299
- 349
- 399
- 449
- 499
- 549
- 999
- 1050
TOTAL
0
289
140
54
86
12
2
28
1
5
5
19
36
0
677
0.0
42.7
20,7
8,0
12,7
1.8
0.3
4.1
2.0
0.7
0.7
2,8
5.3
0.0
Number
Leaking

12
10
2
3
1
0
1
0
0
0
1
0
_
31
Percent
Leaking
.
4.2
7.1
3.7
3.5
8.3
0,0
3,6
• 0.0
20.0
0.0
5.3
0,0
_
4.6
Number
Screened
5
519
339
174
138
10
0
23
4
1
0
0
1
0
1214
Percent of
Total
Screened
0.4
42.8
27.9
14.3
11.4
0.8
0.0
1.9
0.3
0.1
0.0
0.0
0.1
0.0
Number
Leaking
0
15
12
7
1
0
-
0
0
0
-
-
0
-
35
Percent
Leaking
0.0
2.9
3.5
4.0
0.7
0.0
-
0.0
0.0
0.0
-
-
0.0
-
2.9
Temperature ("F)
-267
0
50
100
150
200
250
300
350
400
-1
- 49
- 99
- 149
- 199
- 249
- 299
- 349
- 399
- 1570
TOTAL
0
41
228
•154
40
133
15
40
0
21
672
0.0
6.1
33.9
22.9
6.0
19.8
2.2
6.0
0.0
3.1
-
0
13
6
2
7
1
2
. -
0
31
-
0.0
5.7
3.9
5.0
5.3
6.7
5.0
-
0,0
4.6
0
15
440
330
161
178
74
15
3
9
1225
0.0
1.2
35.9
26.9
13.1
14.5
6.0
1.2
0.2
0.7
-
0
15
12
4
2
1
0
0
1
35
-
0.0
3.4
3.6
2.5
1.1
1.4
0.0
0.0
11.1
2.9
       'see Figure 3-2 for explanation of groups.

-------
                                                VALVES—GAS
00
p
E
R
C
E
N
T

S
C
R
E
E
N
E
D
                               0-49
                           Group 1

                           Group 6
                           Group 7
                          100-149  200-249 300-349  400-449 500-549

                                     PRESSURE
10001-
           Figure B-l.  Distribution of Sources Screened by Line Pressure for Ethylene  and High
                       Leaking Process Units by Chemical Group for Valves with Gas  Service
           *See Figure 3-2 for explanation  of groups.

-------
                                    VALVES—GAS
P
E
R
C
E
N
T

S
C
R
E
E
N
E
P
                     0-49
                 Group 1

                 Group 5
                 Group 6
100-149     208-249
 TEMPERATURE CO
300-349
400+
Figure  B-2.  Distribution of Sources  Screened by Line Temperature for Ethylene  and
            High Leaking Process Units by Chemical Group for Valves with Gas Service

*See Figure 3-2 for explanation of groups.

-------
                                               VALVES—LIGHT  LIQUID
M
OO
oo
50-
P
E 40-
R
C
E
N
T 30-
L
E
A
K '20-
I
N
G

10~


0«— .




. — . _ . _ . _ .
-

i
' ,
/ \ •'
/ Y* '•
/ "^. * \
//.\ *
>:\\\
'/; \ \ .
/'" ' •
//• v. V.
//•' \ ^'
y /- 'x- x
/;' / ^ *** • A. / \
•* / ^ "-\\ / \ -^

h ^ — ^ <• i x ,' Nv '""*""-."]-• x"-
*i -V N^ / •'•" !. N
f v"x^^^-'- -^ ''-. x\
IiiiiiiiiTii.il
< -1 50-99 150-199 250-299 350-399 450-499 550-999
r-Dnift-49 100-149 200-249 300-349 400-449 500-549 1000+
bKuUP o
GROUP 4
GROUP 7 PRESSURE
y^»r*i^*il 11^* x-.
             Figure  B-3.  Distribution of Sources  Screened by Line Pressure for Ethylene and High

                         Leaking Process Units by Chemical Group for Valves with Light Liquid Service


             *See  Figure 3-2 for explanation of groups.

-------
                                            VALVES—LIGHT  LIQUID
O3
50 ~
P
E 40 ~
R
C
E
N
T 30 ~
L
E
A
K - 20 "
I
N
G
10"


ra —

t
•"•
\ .' *.
\
\
\ t_ — • — l •
x ; x \ '•
/ :\ \ -.
1 J-; N-\-
\ ^ -j A \ \ .

// •' x ^ \ '. /'
' S \ '• - • " '^
f •" "\ ' \ " ' • '\
/(•' \ .v-C—- 
-------
                                 APPENDIX C
         SUMMARY STATISTICS AND DETAILED INFORMATION ON THE EFFECTS
           OF AMBIENT TEMPERATURE AND ELEVATION ON LEAK FREQUENCY
     This appendix contains detailed information on the effects of ambient
temperature and source elevation on leak frequency.  Tables C-l and C-2 con-
tain summary statistics for ambient temperature for the groupings of sources
described in Section 3.

     Ambient temperature was measured as a continuous variable, but to
evaluate its effect on leak frequency, it was grouped as less than 70 °F or
greater than or equal to 70°F.  Tables C-3 and C-4 give the number of sources
screened, number leaking and percent leaking for both of the groups of ambient
temperatures.  The statistics are categorized by source type, stream service,
the type of process unit, and the primary material group.  Table C-3 contains
the data for ethylene process units and Table C-4 contains the data for high
leaking process units,

     Chi-square tests were performed on each group to determine if there was
a significant difference in leak frequencies between the two categories of
ambient temperatures.  The results are given in Tables C-3 and C-4.  For the
ehtylene process units (Table C-3), the leak frequencies of valves are signif-
icantly different in all categories.  For both gas and light liquid service
in the high leaking primary material group, higher leak frequencies were
found at the higher levels of ambient temperature.  For the low leaking pri-
mary material group, the higher leak frequencies occurred at the lower ambient
temperature level.  The only other group in the ethylene process units that
showed a significant effect of the ambient temperature is open-ended lines in
gas stream service in the high leaking primary material group.
                                     190

-------
     Table C-4 contains the same information for the high leaking process
units.  Valves in group 7 showed the only significant effect on leak frequen-
cies of ambient temperature.  The higher level of ambient temperature was
associated with the higher leak frequency.

     Tables C-5 and C-6 contain the data on the effects of elevation on
leak frequency for ethylene and high leaking process units, respectively.
Chi-square tests were performed to determine differences in percent leaking
for the two levels.  There were no significant differences in leak frequencies
for any source types in the ethylene process units.  The high leaking process
units showed a few significant effects of elevation on leak frequencies.
These effects were seen for valves and open-ended lines irx light liquid stream
service and with high leaking primary materials in the line.  For the even
numbered groups only valves in gas stream service were significantly affected.
In each of these cases the higher leak frequency occurred at the ground level.
                                     191

-------
TABLE C-l.  SUMMARY OF AMBIENT TEMPERATURE DURING SCREENING
            OF VARIOUS SOURCE TYPES IN GAS SERVICE
HIGH LEAKING PROCESS
SOURCE
VARIABLE TYPE
VALVES



FLANGES



OPEN-ENDED
LINES



SUMMARY STATISTICS FOR
AMBIENT AIR TEMPERATURE
Average ("F)
Standard Deviation
Minimum
Maximum
Average (OF)
Standard Deviacion
Minimum
Maximum
Average (°F)
Standard Deviation
Minimum
Maximum.
HIGH LEAKING
PRIMARY MATERIALS
73.1
18.5
33
104
86.2
13.0
33
102
69.5
17.5
33
100
LOW LEAKING
PRIMARY MATERIALS
63.6
19-5
30
100
79-8
16.4
33
100
66.0
20.4
30
100
ETHYL ENE
HIGH LEAKING
PRIMARY MATERIALS
58.
20.
11
187
73.
14.
20
120
48.
20.
20
90
.3
,6


.2
.5


.2
,1


                             192

-------
TABLE C-2.  SUMMARY OF AMBIENT TEMPERATURE DURING SCREENING OF VARIOUS
            SOURCE TYPES IN LIGHT LIQUID SERVICE
HIGH LEAKING PROCESS
SOURCE SUMMARY STATISTICS FOR
TYPE AMBIENT AIR TEMPERATURE
VALVES



PUMP SEALS



FLANGES



OPEN-ENDED
LINES


Average (°F)
Standard Deviation
Minimum
Maximum
Average (°F)
Standard Deviation
Minimum
Maximum
Average (°F)
Standard Deviation
Minimum
Maximum
Average (°F)
Standard Deviation
Minimum
Maximum
HIGH LEAKING
PRIMARY MATERIALS
57.8
19.0
29
100
54.5
15.1
32
98
77.6
19.0
29
100
52.2
13.5
29
98
LOW LEAKING
PRIMARY MATERIALS
72.4
17.7
29
101
74.2
17.6
32
100
82.0
15.4
29
102
77.6
17.4
29
10ft
ETHYLENE
HIGH LEAKING
PRIMARY MATERIALS
62.8
19.9
21
91
56.6
21.2
22
85
74.1
13.6
24
91
51.3
21.1
24
91
LOW LEAKING
PRIMARY MATERIALS
65.
20,
22
91
62,
21,
36
88
78,
14,
30
91
48.
19.
22
90
,5
,7


.9
.2


.1
.7


,2
,4



-------
     TABLE C-3.   EFFECT  OF AMBIENT  TEMPERATURE ON  PERCENT OF SOURCES LEAKING IN
                   ETHYLENE PROCESS UNITS  AS A  FUNCTION OF THE PRIMARY CHEMICAL
                   GROUPS
PRIMARY CHEMICAL
SOURCE
TYPE
VALVES



PUMP SEALS

FLANGES



OPEN-ENDED
LINES



STREAM
SERVICE
Gas

Light Liquid

Light Liquid

Gas

Light Liquid


Gas

Light Liquid

AMBIENT
TEMPERATURE,
"F
<70°
70°+
<70°
70°+
<70°
70"+
<70"
70°+
<70"
70°+

<70°
70°+
<70°
70"+
NUMBER
SCREENED
3760
2534
1666
1848
29
32
165
469
68
259

223
82
110
41
NUMBER
LEAKING
474
460
446
511
7
11
10
29
6
19

19
18
30
9
Group 1
PERCENT
LEAKING
12.6
18.2
26.8
27.6
24.1
34.4
6.1
6.2
8.8
7.3

8.5
22.0
27.3
21.9
and Group 3
CHI-SQUARE p2
36.8 <0.001

0.34 >0.05

0.4 >0.05

0.01 >0.05

0.16 >0.05


10.1 <0.01

0.04 >0.05

NUMBER
SCREENED


240
367
7
8


9
61



50
13
PRIMARY
NUMBER
LEAKING


2
7
0
2


0
0



2
0
CHEMICAL Group 41
PERCENT
LEAKING CHI-SQUARE


0.8 1.1
1.9
0.0 2.1
25.0


0.0 *
0.0



4.0 0.54
0.0

P2


>0.05

>0.05








>0.05

'See Figure 3-2 for explanation of groups.
 Probability of no significant difference in leak frequency due to ambient temperature.
*Expected values were too low for Chi-square test.

-------
   TABLE C-4.   EFFECTS OF AMBIENT TEMPERATURE  ON PERCENT  OF SOURCES  LEAKING IN HIGH

                 LEAKING PROCESS UNITS AS  A FUNCTION  OF THE PRIMARY CHEMICAL GROUPS
PRIMARY CHEMICAL Gronp 5 and' Group 71 PRIMARY CHEMICAL Grouo 6
SOURCE STREAM
TYPE SERVICE
VALVES Gaa

Light Liquid

PUMP SEALS Light Liquid

FLANGES Gas

Light Liquid

OPEN-ENDED
LINES Gas

Light Liquid

AMBIENT
TBMPERATURE,
°F
<70°
70°+'
<70°
70°+
<70°
70°+
<70°
70°+
<70e
70"+

<70°
70°+
<70"
70"+
NUMBER
SCREENED
499
729
2435
803
108
18
46
345
161
417

71
75
713
85
HUMBER
LEAKING
50
96
52
95
11
3
1
17
1
9

4
9
45
2
PEFCENT
LEAKING
10.0
13.2
2.1
11.8
10.2
16.7
2.2
4.9
0.6
2.2

5.6
12.0
6,3
2.4
NUMBER
CHI-SQUARE P1 SCREENED
2.8 >0.05 1090
668
135.6 <0.001 2861
3293
0.7 >0.05 101
142
0.7 >0.05 77
239
0.6 >0.05 219
791

1.8 >0,05 204
143
2.1 >0.05 415
876
NUMBER
LEAKING
17
5
17
29
3
11
3
6
0
0

1
3
7
31
PERCENT
LEAKING
1.6
0.8
0.6
0.9
3.0
7.8
3.9
2.5
0.0
0.0

.5
2.1
1.7
3.5
and Group S1
CHI-SQUARE P1
2.2 >0.05

1.7 >0.05

2.5 >0.05

0.4 >0.05

*


1.9 >0.05

3.3 >0.05

'See Figure 3-2 for explanation of groups.

^Probability oE no significant difference in leak frequency due to ambient temperature.

*
Expected values were too low for Chi-square test.

-------
    TABLE C-5.   EFFECTS OF SOURCE ELEVATION ON PERCENT LEAKING FOR  ETHYLENE PROCESS
                  UNITS AS  A FUNCTION  OF PRIMARY CHEMICAL  GROUPS
PRIMARY CHEMICAL Group 1 and Group 31 PRIMARY CHEMICAL Group 21
SOURCE
TYPE
VALVES



PUMP SEALS

FLANGES



OPEN-ENDED
LINES



SERVICE ELEVATION
Gas • Ground
Above
Light Liquid Ground
Above
Light Liquid Ground
Above
Gas Ground
Above
Light Liquid Ground
Above
Ga^ Ground
Above
Light Liquid Ground
Above
NUMBER NUMBER
SCREENED LEAKING
3298
2844
2578
926
61
0
246
387
234
91
235
69
109
42
475
452
716
237
' 18
-
13
26
16
a
25
12
29
10
PERCENT
LEAKING CHI-SQUARE pa
14.9 1.2 >0.05
15.9
27.8 1.6 >0.05
25.6
29.5 *
-
5.3 0.53 >0.05
6.7
6.8 0.37 >0.05
8.8
10.6 2.3 >0.05
17.4 2.0
26.6 0.12 >0.05
23.8
NUMBER NUMBER PERCENT
SCREENED LEAKING LEAKING CHI-SQUARE p2


494 8 1.6 0.34 >0.05
113 1 0.9
15 2 13.3 *
0 — —


55 0 0.0 *
15 0 0.0


54 2 3.7 0.3 >0.05
9 0 0.0
 See Figure 3-2 for explanation of groups.

 Probability of no significant difference in leak frequency due to source elevation.

Insufficient data for Chi-square test

-------
     TABLE  C-6.   EFFECTS  OF SOURCE ELEVATION ON PERCENT  LEAKING IN  HIGH  LEAKING PROCESS
                   UNITS AS A FUNCTION OF PRIMARY CHEMICAL GROUPS
PRIMARY CHEMICAL Group 5 and Group 7 PRIMARY CHEMICAL Group 6
B
SOURCE
TYPE SERVICE ELEVATION
VALVES Gas Ground
Above
Light Liquid Ground
Above
PUMP SEALS Light Liquid Ground
Above
FLANGE Gaa Ground
t-1
^D Above
-J
Light Liquid Ground
Above
OPEN-ENDED
LINE Gas Ground
Above
Light Liquid Ground
Above
NUMBER
SCREENED
479
749
2494
795
122
4
155
236
417
160
59
87
623
172
NUMBER
LEAKING
54
92
121
25
14
0
5
13
9
1
6
7
45
2
PERCENT
LEAKING CHI-SQUARE p 2
11.3 0.3 >0.05
12.3
4.fl 4.1 <0.05
3.1
11.5 *

3.2' 1.0 >0.05
5.!.
2.1! 1.6 >0.05
O.li
10.!! 0.2 >0.05
8.0
7.;; 8.9 0.05

*

2.1 >0.05
*

2.1 >0.05

0.1 >0.05

'See Figure 3-2 for explanation of groups.
^Probability of no significant difference in leak frequency due to source elevation.
*Insufficlent data for Chi-squares test

-------
                                 APPENDIX D
                         CORRECTIONS TO SCREENING DATA

     During the SOCMI fugitive emission screening project (Reference 1),
occasional corrections were required on the original data sheets.   These cor-
rections were subsequently documented along with an explanation of why they
were necessary.

     One clarification that affected almost all of the units screened was
dua to the decision to exclude water when calculating the primary chemical
concentration.  To make this adjustment all primary material concentrations
that were below thirty percent (30%) were changed to reflect ninety to one-
hundred percent (90-100%), if no secondary material other than water existed.
That is, if the primary chemical was twenty percent (20%) of the stream and
water made up the other eighty percent (80%), then the primary chemical con-
centration was adjusted to one hundred percent (100%).  So, the concentration
number was adjusted to reflect the percent of total VOC's.

     Table D-l and D-2 summarize all adjustments and corrections made to the
original data sheets.  Table D-l is a summary of the detailed corrections
listed in Table D-2 which affected the overall number screened, the number
not screened and the number screened greater than or equal to 10,000, as
reported earlier (Reference 1).  After the corrections described in this
appendix were made, the data reported in this reference were used for the
analyses in Section 3 of this report.

     Other clarifications, mostly due to miscoding by the recorder, are
listed in Table D-2, by unit, then source identification sequence.  Coding
corrections covered a wide range of source identification codes and few were
changed from the same code number.  Therefore, it was not feasible to list
                                      198

-------
all the old codes along with their corrections
                                     199

-------
                                            TABLE D-l.   CORRECTIONS  AFFECTING RESULTS  ON PREVIOUS REPORTS
                                    SOURCE/SERVICE
 NUMBER
 SCREENED
OLD    NEH
                                                       1443   1450
                                      Light Liquid     2897   2833
  NUMBER
HOT SCREENED
OLD     NEW


 No Change




 76     142
     HU10JER
SCREENED >°10000
OLD        NEW


    NCI Change




    Nc  Chinge
O
O
                                   Open-Ended Lines

                                      Light Liquid    3603   3605
                                                                          417     415
                                          No Change'
                                   Relief Valves

                                      Gas
                                      Light Liquid
                                                        85
                                                               84
                                                        69     68
                                                                          226     227
                                                                           47      48
                                          No Change



                                          No Ch.-mge
                                                                                                                     EXPLANATION
                                                           Unit 2  had 4 sources
                                                           and Unit 4 had 3 sources
                                                           which were reclassified
                                                           from compressors.
                                                           Unit 60 had 64 sources
                                                           in which the values were
                                                           changed from 0 to miss-
                                                           ing. Unit 4 had 2 sources
                                                           which were reclassl£led
                                                           from pumps and their
                                                           values  changed from 0 to
                                                           missing.  Unit 60 had 2
                                                           sources which should have
                                                           been recorded with Unit 61.
                                                                                                                  Unit 29 had 2  sources in
                                                                                                                  which the values were
                                                                                                                  changed from missing to 0.
                                                                                                                  Unit 60 had 2  sources which
                                                                                                                  should have been recorded
                                                                                                                  with Unit 61.
                                        Unit 20 had 1 source  in which
                                        the value was changed from 0
                                        to missing.

                                        Unit 20 had 1 source  in which
                                        the value was changed from 0
                                        to missing.

-------
               TABLE  D-l  (continued)
                                                          NUMBER
                                                          SCREENED
                    NUMBER
                  HOT SCREENED
                                    SOURCE/SERVICE      OLD    HEW         OLD     NEW
     H'JMT.FR
SCREENSD :-10000
OLD        NEW
                                                                                                                         EXPLANATION
                                                        9668   9669
                                       Light Liquid     18294   18300
                                                                          2047    2046
                                                                          2553    2548
                                                                                                  Ho Ch.inge
                                                                                              1174       U83
to
O
                                    PUIDDB

                                      Light Liquid
                                    Compressors
                                      Gaa
647   646
                                                           29     22
                  29
                           28
                                        No Change
                                        Ho Change
                                    Other

                                      Light Liquid
 33    34
                                                                             No Change
                                        No Change
                                                            Unit 29 had 1 source in
                                                            which the valve was changed
                                                            from missing to (J.

                                                            Unit 29 had 6 sources in
                                                            which the values were changed
                                                            from missing to (J.   Unit 20
                                                            had 1 source in which the value
                                                            was changed from 0  Co missing.
                                                            Unit 3 had 1 source in which the
                                                            value was changed from missing to
                                                            0,  Unit 60 had 9 sources which
                                                            had values of 10,000 that were
                                                            not included in the number screen-
                                                            ed over 10,000.  Unit 60 had 8
                                                            sources which should have been
                                                            recorded with Unit  61, 1 of which
                                                            had a value of 10,000.
                       Unit  29 had  1  source in which the
                       value was  changed  from missing to
                       0.  Unit 4 had 2 sources which
                       were  reclasslfied  as flanges.
                       Unit 2  had  4  sources and Unit 4
                       had 3 sources which were reclassi-
                       fied to flanges.  Unit 64 had 1
                       source  in which  the value was
                       changed from  missing to (t.  Unit
                       4 had 1 source which was reclassl-
                       fled to other.
                       Unit 4 had  1  source which was
                       reclassified  from compressors.

-------
       TABLE D-2.   CORRECTIONS  TO SCREENING DATA  SHEETS
Unit
                    Source  ID
               Change
            33


            133-140

            534,  1052,  1355-1356, 1393

            365

            1552


            1580-1581
Elevation to 1
Comment to missing

Secondary material concentration to 4

Comment to 1

Deleted

Source type to 54
Line temperature to 140

Service to 2
            918
            1574-1576,  1594-1596,
            1606-1608,  1624,  1633,
            1646,  1723

            1756
            1757

            1758, 1822



            1873, 1895



            2215-2219


            2440, 2487

            3004


            3005-3020

            3021-3024




            3025
Screening value to missing
Comment to 1 (inaccessible)

Source type to 2
Source type to 1
Line temperature to 143
Line pressure to 525

Source type to 1

Source type to 52
Line temperature to 143
Line pressure to 525

Source type to 1
Line temperature to 188
Line pressure to 155

Screening value to missing
Comment-to 1 (inaccessible)

Source type to 2

Line temperature to 300
Line pressure to 6

Line pressure to 6

Primary material to 3
Primary material concentration to 3
Secondary material to 1
Secondary material concentration  to  2

Screening value to missing
Comment to 1 (inaccessible)
                                                                 (continued)
                                    202

-------
   TABLE  D-2.   CORRECTIONS  TO  SCREENING  DATA SHEETS  (CONTINUED)
Unit
                  Source  ID
                                                          Change
            3028-3030
            3031-3033

            3034-3038, 3121




            3236-3298

            3239


            3240

            3241-3243

            3289
            3300,  3309-3317,
            3332-3347, 3350-3373

            3378-3388
            3447


            3463-3469, 3473-3477

            3478-3498

            3552-3578

            3604


            3605-3607

            3608-3630


            3631-3634

            3635-3660
Primary material to  3
Primary material concentration to 3
Secondary material to  1
Secondary material concentration to 2
Line temperature to  260

Line temperature to  260

Primary material to  3
Primary material concentration to 3
Secondary material to  1
Secondary material concentration to 2

Line temperature to  5

Source type to 40
Line temperature to  260

Line temperature to  5

Line temperature to  5

Screening value to missing
Comment to 1 (inaccessible)

Service to 1
Service to 1
Line pressure to 2

Screening value to  missing
Comment to 1 (inaccessible)

Primary material to 1

Primary material concentration to 7

Line temperature to 25

Service to 1
Line pressure to 500

Service to 2

Service to 1
Line pressure to 2

Service to 2

Service to 1
Line pressure to 2


                  (continued)
                                   203

-------
  TABLE  D-2.    CORRECTIONS  TO SCREENING  DATA SHEETS  (CONTINUED)
Unit
               Source ID
                                                         Change
            3683-3684
            3700-3714
            3716, 3718
            3719-3720
            3721-3722
            3734-3740
Service to  1
Primary material concentration to  5
Primary material concentration to  5
Secondary material to 2
Secondary material concentration to  2

Primary material concentration to  4
Secondary material to- 2
Secondary material concentration to  2

Primary material to 3
Primary material concentration to  5
Secondary material to 1
Secondary material concentration to  3

Primary material concentration to  5
Secondary material concentration to  3
3751-3754
3755-3758
3762-3783
3801-3809
3882
3901
3913-3917
3968-3975, 3987
4668, 5048
5339
5690
3 1215-1216
1349
1638
1639
Service to 2
Service to 1 t
Secondary material concentration to 1
Service to 1
Source type to 42
Source type to 30
Line temperature to 62
Line pressure to 480
Service to 1
Source Type to 35
Screening value to 10,000
Source type to 52
Service to 1
Primary material to 5
Line temperature to 195
Line pressure to 225
Comment to 1
Secondary material concentration to 0
Source type to 32
Source type to 42
                                                             (continued)
                                     204

-------
TABLE  D-2.   CORRECTIONS TO SCREENING DATA  SHEETS  (CONTINUED)
 Unit
                  Source ID
                                                         Change
  11
   12
            97-98, 849, 956.
            956-959

            1120-1139
2161,  2173

3375

5192,  5837

6760



If Service=missing


21-40

607 (2nd)

608 (2nd)

613 (2nd)

735 (2nd)

885

1745

1953

2358-2359

2426-2429

2566-2585,  2626-2641


2731-2734

2889

3156

3409,  3411,  3418, 3422


If screening team=13 or 15

If source  type=*
31-100
Comment to 1  (inaccessible)


Deleted


Source type to  1

Source type to  32

Source type to  1

Service to 1
Source type to  1


Service to 2


Day to 21

Source id to  4105

Source id to  4106

Source id to  4107

Source id to  4108

Source type to  10

Unit to 11

Source type to  10

Service to 10

Comment to 3

Secondary material  concentration to
missing

Comment to 3

Deleted

Service to 2

Comment" to 3


Source id to:  id plus 1959

Source type to  1
Month to 3


                  (continued)
                                  205

-------
 TABLE D-2.   CORRECTIONS  TO  SCREENING DATA  SHEETS  (CONTINUED)
Unit
                 Source  ID
                                                          Change
 12
            87

            221-240

            310,  338,  339,  362

            561-562
            565-580

            2730-2739
            2788-2791
            3416
Source type to 40

Service to 2

Source type to 40

Service to 2
Primary material to 40
Primary material concentration to 6
Secondary material to 45
Secondary material concentration to 3
Line temperature to 170
Line pressure to 100
Ambient air temperature to  60

Service to 2

Primary material to 42
Primary material concentration to 9
Secondary material to missing
Secondary material concentration to
missing

Primary material to 41
Primary material concentration to 9
Secondary material to missing
Secondary material concentration to
missing

Source type to 3
 20*        If screening value=
            missing

            1049,  1081

            1239,  1251, 1314

            3156

            3201
Comment to 1


Source typ'e to 33

Screening value to missing

Service to 2

Screening value to missing
 21*
            947
                                           Service to 2
 22
            154

            221-240
Comment to 1 (inaccessible)

Ambient air temperature
 *Pedco used duplicate unit #*s, so the VMC unit  screened between 2-14 and 2-20 was
  changed to unit  28 and the EDC unit screened between 2-12 and 2-15 was changed  to
  unit 29.
                                                                  (continued)
                                     206

-------
   TABLE  D-2.   CORRECTIONS TO SCREENING DATA SHEETS  (CONTINUED)
   Unit
                        Source ID
                                                              Change
 28  (20)*      If primary material=19





               2710-2720



               2777

               2793

               2855

               3257-3276


 29  (21)*      If screening value is blank

               19

               359-360

               432-440



               626-640



               308-809

               '865-880



               1017, 1025

               2103-2120

               2135-2140



               3116

               3181-3200

               3239-3240
Primary material to 14
Primary material concentration  to 9
Secondary material to missing
Secondary material concentration to
missing

Secondary material to missing
Secondary material concentration to
missing

Screening value to missing

Source type to 45

Screening value to 30

Line temperature Co 355


Screening value to 0

Source type to 35

Comment to 1

Secondary material to missing
Secondary material concentration to
missing

Secondary material to missing
Secondary material concentration to
missing

Comment to 1 (inaccessible)

Secondary material to missing
Secondary material concentration to
missing

Source type to 35

Comment to missing

Secondary material to missing
Secondary material concentration to
missing

Comment to 1

Instrument to 2

Secondary material to missing
Secondary material concentration to
missing
*Pedco used duplicate unit :.-'s, so the VMC unit screened between
 2-14 and 2-20 was  changed to unit 28 and the EEC unit  screened
 between 2-12 and 2-15 was changed to unit 29.
                                      207

-------
  TABLE  D-2.   CORRECTIONS TO SCREENING  DATA SHEETS  (CONTINUED)
Unit
                 Source  ID
                                              Change
 32
435, 463

646-647
Comment to 2

Secondary material  concentration to 2
 33
331, 337-338
                                           Comment to 1
 34
            741-760

            1521
                               Elevation to 2

                               Service to 2
 60
            1-505
            (excluding 361-364j
            398-401,  425-428)

            39

            506-696
            897-913

            1044-1057


            1058-1068

            1269-1278

            1301-1720

            1810 (2nd)

            1829-1831



            1941-1960

            2005-2080


            2081-2123
            2354-2406,  2421-2449.
            2436-2500

            2591-2600
                               Primary material to 6
                               Source type to 30

                               Primary material to 6
                               Primary material concentration  to 6
                               Secondary material to 14
                               Secondary material concentration to 2

                               Service to 2

                               Secondary material to 9
                               Elevation" to 3

                               Secondary material to 9

                               Secondary material to 8

                               Primary material to 6

                               Source id to 1811

                               Secondary material to missing
                               Secondary material concentration to
                               missing

                               Primary material to 6

                               Primary material to 4
                               Secondary material to 15

                               Primary material to 4
                               Secondary material to 15
                               Secondary material concentration to 0

                               Primary material to 5
                               Secondary material to missing
                               Secondary material concentration to
                               missing

                                                       (continued)
                                    208

-------
TABLE D-2.   CORRECTIONS  TO  SCREENING  DATA SHEETS  (CONTINUED)
Unit
                  Source ID
                                                         Change
 60
 61
2628

2743-2747,  2787-2800



2890-2900



4113-4130

6970-6974




3021-3027,  3044-3060

3053


3121-3127

3188

3233,  3235-3237

3246-3260

3261-3280

3281-3297

3298-3305


3306-3328

3329-3335


3521-3716

3742-3760



3781-3830


3805-3806
Source type to 1

Secondary material to missing
Secondary material concentration to
missing

Secondary material to missing
Secondary material concentration to
missing

Primary material to 5

Secondary material to missing
Secondary material concentration to
missing
Secondary material to 3

Source type to 1
Screening value to 3600

Primary material to 6

Screening value to 200

Screening value to 0

Secondary material to 6

Secondary material to 3

Primary material to 3

Primary material to 3
Secondary material to 6

Primary material to 6

Primary material to 3
Secondary material to 6

Primary material to 3

Secondary material missing
Secondary material concentration to
missing

Primary material to 3
Secondary material to 6
                                           Comment  to  1
                                                             (continued)
                                    209

-------
                               TABLE  D-2.   CORRECTIONS TO  SCREENING  DATA SHEETS  (CONTINUED)
                             Unit
                                              Source ID
                                              Change
                              62
                              64
•
                              65
'
                                        3841-3860
If service=2 and
primary material  is
2, 4, or 5

3933
3941-395Q


4357, 4362

4604-4609, 4611-4612

4631-4640


4801-4820

4900 (2nd)

5241-2

5243


5244-5280

5286-5290, 5299-5300

5340

5354-5356, 536*4-5367

5393

5404, 5413-5414,
5464-5466

5468

5511

5521-5540

5979-5986, 5988-6000

6011-6013

6021-6039
Unit to 62


Service to 3



Service to 1
Primary material concentration to 1
Secondary material concentration to 0

Secondary material to 3
Secondary material concentration to 2

Service to 3

Comment to 1 (inaccessible)

Source id=id + 46


Ambient air temperature  to 8

Source id to 6751

Service to 2

Service to "2
Elevation to 2

Service to 2

Comment to 1 (inaccessible)

Source type to 1

Screening value to missing

Source type to 10

Screening value to missing


Service to 1

Comment to 2

Line pressure to 45

Line pressure to -5

Comment to 1 (inaccessible)

Ambient air temperature  to 84


                  (continued)
                                                                    210

-------
TABLE  D-2.   CORRECTIONS TO SCREENING  DATA SHEETS (CONTINUED)
Unit
                  Source ID
                                                       Change
 65
            6040
            6113-6114

            6221-6222, 6224-6225,
            6243-6249

            6601-6620
                              .Ambient  air temperature to 84
                              Comment  Co 2

                              Screening value to missing

                              Commenc  Co I {Inaccessible)


                              Line pressure to -5
 66
7083

7141-7160

7173,  7249, 7264, 7336,
7354

8574,  8808-8810
            3850, 3362
Source type  to 30

Unit to 66

Source type  co 30
Secondary material to missing
Secondary material concentration to
missing

Source type  to 30

                                  211

-------
                                TECHNICAL REPORT DATA
                         (Please read httrucrions on the reverse before completing}
. REPORT NO.
EPA-600/2-81-111
            3. RECIPIENT'S ACCESSION- NO.
 TITLE ANDSUSTITLE
Analysis of SOCMI VOC Fugitive Emissions Data
            5. REPORT DATE
            June 1981
                                                      S. PERFORMING ORGANIZATION CODE
. AUTHORlS)
            B. PERFORMING ORGANIZATION REPORT NO.
G. J.Langley, S.M.Dennis, J.F.Ward, and
 L. P. Provost
. PERFORMING ORGANIZATION NAME AND ADDRESS
Radian Corporation
P.O. Box 9948
Austin, Texas  78766
            10. PROGRAM ELEMENT NO.

            C9HA1A
            11. CONTRACT/GRANT NO.

             68-02-3171, Task 28
12. SPONSORING AGENCY NAME AND ADDRESS
 EPA, Office of Research and Development
 Industrial Environmental Research Laboratory
 Research Triangle Park, NC  27711
            13. TYPE OF REPORT AND PERIOOCOVER6D
            Task Final; 12/80 - 5/81
            14. SPONSORING AGENCY CODE
              EPA/600/13
is. SUPPLEMENTARY NOTES TERL_RTP prOject officer is Bruce
 919/541-2547.
            A. Tichenor, Mail Drop 63,
16. ABSTRACT ,pne rep0rt gives results of an examination of fugitive emission data from
 Synthetic Organic Chemical Manufacturing Industry (SOCMI) processing units (col-
 lected under earlier EPA studies) for  correlations between process variables and
 leak frequency. Although line temperature did not have a consistent relationship with
 leak frequency, the data showed that leak frequency increased with increasing line
 pressure. Also, emission factors for  three process types (vinyl acetate, cumene,
 and ethylene) were developed and presented. Increases in mass emissions due to
 occurrence and recurrence  of leaks for these three process types are also estima-
 ted. Finally, the effect of adjusting portable hydrocarbon readings by chemical
 response factor curves on leakage frequency estimates is investigated. Despite the
 wide range of response factors encountered, the adjusted leak frequencies were
 essentially the same as the  unadjusted frequencies.
17.
                             KEY WORDS AND DOCUMENT ANALYSIS
                DESCRIPTORS
                                          b.lOENTIFIERS/OPEN ENDED TERMS
                        c.  COSATI Field/Group
 Pollution            Cumene
 Volatility            Ethylene
 Organic Compounds
 Processing          Hydrocarbons
 Leakage
 Vinyl Acetate
Pollution Control
Stationary Sources
Volatile  Organic Com-
  pounds
Fugitive Emissions
13B
20M
07C
13H
14G
13. DISTRIBUTION STATEMENT
 Release to Public
                                          19. SECURITY CLASS (This Report)
                                          Unclassified
                         21. NO. OF PAGES
                             225
20. SECURITY CLASS {Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (»-73)
                                          212

-------

-------

-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
    Office of Research and Development
  Center for Environmental Research Information
           Cincinnati, Ohio 45268

           OFFICIAL BUSINESS
   PENALTY FOR PRIVATE USE  S3OO
  AN EQUAL OPPORTUNITY EMPLOYER
        POSTAGE AND FEES PAID

US ENVIRONMENTAL PROTECTION AGENCY

               EPA-335
                                       If your address is incorrect, please change on the above label
                                       tear off; and return to the above address.
                                       If you do not desire to continue receiving these technical
                                       reports. CHECK HffffD; tear off label, and return it to the
                                       above address.
                                     Publication No. EPA-600/2-8i-m

-------