Emission Factor Documentation for AP-42
                                Section 7.1
            Organic Liquid Storage Tanks
                              Final Report
         For U. S. Environmental Protection Agency
       Office of Air Quality Planning and Standards
             Emission Factor and Inventory Group
                              September 2006

-------
Emission Factor Documentation for AP-42
                                 Section 7.1
             Organic Liquid Storage Tanks
                               Final Report
         For U. S. Environmental Protection Agency
        Office of Air Quality Planning and Standards
              Emission Factor and Inventory Group
               Research Triangle Park, NC 27711

                       Attn: Mr. Michael Ciolek
                               September 2006

-------
                              TABLE OF CONTENTS

                                                                              Page

1.0  INTRODUCTION	      1-1

2.0  STORAGE TANK DESCRIPTIONS	      2-1
2.1  INTRODUCTION	      2-1
2.2  TYPES OF STORAGE TANKS	      2-1
    2.2.1     Fixed-Roof Tanks	      2-1
    2.2.2     External Floating Roof Tanks	      2-2
    2.2.3     Internal Floating Roof Tanks	      2-2
    2.2.4     Domed External Floating Roof Tanks	      2-3
    2.2.5     Horizontal Tanks	      2-4
    2.2.6     Pressure Tanks	      2-4
    2.2.7     Variable Vapor Space Tanks	      2-4
2.3  TYPES OF FLOATING ROOF PERIMETER SEALS	      2-5
    2.3.1     External and Domed External Floating Roof Seals	      2-5
    2.3.2     Internal Floating Roof Seals	      2-6
2.4  TYPES OF FLOATING ROOF DECK FITTINGS	      2-7
    2.4.1     External and Domed External Floating Roof Fittings	      2-7
    2.4.2     Internal Floating Roof Fittings	      2-9
2.5  REFERENCES	    2-26

3.0  EMISSION ESTIMATION PROCEDURES	      3-1
3.1  INTRODUCTION	      3-1
    3.1.1     Total Losses From Fixed Roof Tanks	      3-1
    3.1.2     Total Losses From Floating Roof Tanks	    3-14
    3.1.3     Variable Vapor Space Tanks	    3-29
    3.1.4     Pressure Tanks	    3-30
3.2  HAZARDOUS AIR POLLUTANTS (HAPs) SPECIATION METHODOLOGY	    3-31
3.3  REFERENCES	    3-78

4.0  EMISSION ESTIMATION PROCEDURES FOR FIXED ROOF TANKS	      4-1
4.1  BREATHING LOSS EQUATIONS	      4-1
4.2  COMPARISON OF PREDICTIVE ABILITY OF TWO EQUATIONS	      4-3
    4.2.1     Predictive Ability-Actual Data	      4-4
    4.2.2     Predictive Ability-Default Values	      4-5
4.3  SENSITIVITY ANALYSIS	      4-6
4.4  CONCLUSIONS AND RECOMMENDATIONS	      4-9
4.5  REFERENCES	    4-30

5.   EMISSION ESTIMATION PROCEDURES FOR FLOATING ROOF TANKS	      5-1
5.1  STATISTICAL ANALYSES - API TANK TEST DATA	      5-1
    5.1.1     Evaluation of Rim Seal Loss Factors	      5-2
    5.1.2     Evaluation of Wind Speed Calculation	    5-24
    5.1.3     Evaluation of Diameter Function and Product Factor	    5-25
    5.1.4     Deck Fitting Loss Factors	    5-27
                                        in

-------
                          TABLE OF CONTENTS (continued)

                                                                            Page

    5.1.5      Development of the Fitting Wind Speed Correction Factor	    5-49
    5.1.6      Deck Seam Factors	    5-53
    5.1.7      Vapor Pressure Functions	    5-57
5.2  PREDICTIVE ABILITY - ACTUAL TANK TEST DATA	    5-58
    5.2.1      Standing Storage Loss	    5-58
    5.2.2      Internal Floating Roof Emissions	    5-62
5.3  SENSITIVITY ANALYSES	    5-63
    5.3.1      Standing Storage Loss	    5-63
    5.3.2      Withdrawal Loss	    5-70
5.4  CONCLUSIONS	    5-74
5.5  REFERENCES	    5-74

6.   SUMMARY OF CHANGES TO AP-42 SECTION	     6-1
6.1  CHANGES TO EMISSION ESTIMATION PROCEDURES AND FACTORS FOR FIXED
       ROOF TANKS	     6-1
6.2  CHANGES TO EMISSION ESTIMATION PROCEDURES AND FACTORS FOR
       FLOATING ROOF TANKS	     6-1
                                       IV

-------
                                       LIST OF FIGURES

                                                                                            Page

Figure 2-1.    Typical fixed rooftank	     2-11
Figure 2-2.    External floating rooftank (pontoon type)	     2-12
Figure 2-3.    External floating rooftank (double-deck type)	     2-13
Figure 2-4.    Internal floating rooftank	     2-14
Figure 2-5.    Domed external floating rooftank	     2-15
Figure 2-6.    Typical underground storage tank	     2-16
Figure 2-7.    Atypical above-ground horizontal tank	     2-17
Figure 2-8.    Vapor mounted primary seals	     2-18
Figure 2-9.    Liquid-mounted and mechanical shoe primary seals	     2-19
Figure 2-10.   Secondary rim seals	     2-20
Figure 2-11.   Deck fittings for floating roof tanks	     2-21
Figure 2-12.   Deck fittings for floating roof tanks	     2-22
Figure 2-13.   Slotted and unslotted guidepoles	     2-23
Figure 2-14.   Ladder and well	     2-24
Figure 2-15.   Bottom conditions for landing loss	     2-25
Figure 3-la.   True vapor pressure of crude oils with a Reid vapor pressure of 2 to
              15 pounds per square inch	     3-18
Figure 3-2a.   True vapor pressure of refined petroleum stocks with a Reid vapor pressure
              of 1 to 20 pounds  per square inch	     3-19
Figure 3-lb.   Equation for true vapor pressure of crude oils with a Reid vapor pressure of
              2 to 15 pounds per square inch	     3-20
Figure 3-2b.   Equation for true vapor pressure of refined petroleum stocks with a Reid
              vapor pressure of 1 to 20 pounds per square inch	     3-20
Figure 3-3.    Vapor pressure  function coefficient (A) of refined petroleum stocks with a
              Reid vapor pressure of 1 to 20psi, extrapolated to 0.1 psi	     3-21
Figure 3-4.    Vapor pressure  function coefficient (B) of refined petroleum stocks with a
              Reid vapor pressure of 1 to 20 psi, extrapolated to 0.1 psi	     3-21
Figure 3-5.    Equations to determine vapor pressure constants  A and B for refined
              petroleum stocks	     3-22
Figure 3-6.    Vapor pressure  function coefficient (A) of crude  oil stocks with a
              Reid vapor pressure of 2 to 15 psi, extrapolated to 0.1 psi	     3-23
Figure 3-7.    Vapor pressure  function coefficient (B) of crude  oil stocks with a
              Reid vapor pressure of 2 to 15 psi, extrapolated to 0.1 psi	     3-23
Figure 3-8.    Equations to determine vapor pressure Constants A and B for crude oils
              stocks	     3-24
Figure 3-9.    Equations for the daily maximum and minimum liquid surface temperatures	     3-24
Figure 3-10.   Turnover factor (KN) for fixed roof tanks	     3-25
Figure 3-11.   Vapor pressure  function	     3-26
Figure 4-1.    Sensitivity of Equation 4-1 to changes in molecular weight (Mv)	     4-10
Figure 4-2.    Sensitivity of Equation 4-2 to changes in molecular weight (Mv)	     4-11
Figure 4-3.    Sensitivity of Equation 4-1 to changes in vapor pressure (Pv)	     4-12
Figure 4-4.    Sensitivity of Equation 4-2 to changes in vapor pressure (Pv)	     4-13
Figure 4-5.    Sensitivity of Equation 4-1 to changes in the vapor space outage (HVo)	     4-14

-------
                                 LIST OF FIGURES (continued)

                                                                                          Page

Figure 4-6.     Sensitivity of Equation 4-2 to changes in the vapor space outage (HVo)	    4-15
Figure 4-7.     Sensitivity of Equation 4-1 to changes in the average ambient temperature
              range (AT)	    4-16
Figure 4-8.     Sensitivity of Equation 4-2 to changes in the average ambient temperature
              range (AT)	    4-17
Figure 4-9.     Sensitivity of Equation 4-1 to changes in the paint factor values (FP)	    4-18
Figure 4-10.   Sensitivity of Equation 4-2 to changes in the solar absorptance (<*)	    4-19
Figure 4-11.   Sensitivity of Equation 4-2 to changes in the daily solar insolation
              factors (I)	    4-20
Figure 5-1.     Emissions after a rim-mounted secondary seal as a function of primary seal
              type	    5-12
Figure 5 -2.     Efficiency of rim-mounted secondary seal as a function of primary seal type	    5-13
Figure 5-3.     Emissions after a rim-mounted secondary seal as a function of primary
              seal gap size	    5-14
Figure 5-4.     Efficiency of a rim-mounted secondary seal as a function of primary seal
              gap size	    5-15
Figure 5-5.     Effect of secondary gap on efficiency: Case 1—vapor mounted primary with
              1 inch gap	    5-16
Figure 5-6.     Effect of secondary gap on efficiency: Case 2~shoe mounted primary with
              9.4 inch gap	    5-17
Figure 5-7.     Effect of secondary gap on efficiency: Case 3~shoe mounted primary with
              39.2 inch gap	    5-18
Figure 5-8.     Effect of secondary gap on emissions: Case 4~shoe mounted primary with
              1 inch gap	    5-19
Figure 5-9.     Effect of secondary gap on emissions: Case 5—shoe mounted primary with
              13.2 inch gap	    5-20
Figure 5-10.   Calculated losses as a function of diameter exponent	    5-23
                                              VI

-------
                             LIST OF TABLES

                                                                    Page

TABLE 3-1.  LIST OF ABBREVIATIONS USED IN THE TANK EQUATIONS	   3-44

TABLE 3-2.  PROPERTIES (Mv, Wvc, PV) OF SELECTED PETROLEUM LIQUIDS	   3-46

TABLE 3-3.  PHYSICAL PROPERTIES OF SELECTED PETROCHEMICALS	   3-47

TABLE 3-4.  ASTM DISTILLATION SLOPE FOR SELECTED REFINED
          PETROLEUM STOCKS	   3-49

TABLE 3-5.  VAPOR PRESSURE EQUATION CONSTANTS FOR ORGANIC
          LIQUIDS	   3-50

TABLE 3-6.  PAINT SOLAR ABSORPTANCE FOR FIXED ROOF TANKS	   3-52

TABLE 3-7.  METEOROLOGICAL DATA (TAX, TAN, I) FOR SELECTED U.S.
          LOCATIONS	   3-53

TABLE 3-8.  RIM-SEAL LOSS FACTORS, KRa, KRb and n, FOR FLOATING ROOF
          TANKS	   3-59

TABLE 3-9.  AVERAGE ANNUAL WIND SPEED (v) FOR FOR SELECTED U.S.
          LOCATIONS	   3-60

TABLE 3-10. AVERAGE CLINGAGE FACTORS, C	   3-64

TABLE 3-11. TYPICAL NUMBER OF COLUMNS AS A FUNCTION OF TANK
          DIAMETER FOR INTERNAL FLOATING ROOF TANKS WITH
          COLUMN-SUPPORTED FIXED ROOFS	   3-64

TABLE 3-12. DECK-FITTING LOSS FACTORS, KFa, KFb, AND m, AND TYPICAL
          NUMBER OF DECK FITTINGS, NF	   3-65

TABLE 3-13. EXTERNAL FLOATING ROOF TANKS: TYPICAL NUMBER OF
          VACUUM BREAKERS, Nvb, AND ROOF DRAINS, Nd	   3-67

TABLE 3-14. EXTERNAL FLOATING ROOF TANKS: TYPICAL NUMBER OF
          DECK LEGS, NL	   3-68

TABLE 3-15. INTERNAL FLOATING ROOF TANKS: TYPICAL NUMBER OF DECK
          LEGS, Ni, AND STUB DRAINS, Nd	   3-69

TABLE 3-16. DECK SEAM LENGTH FACTORS (SD) FOR TYPICAL DECK
          CONSTRUCTIONS FOR INTERNAL FLOATING ROOF TANKS	   3-69
                                   vn

-------
                         LIST OF TABLES (continued)

TABLE 3-17. ROOF LANDING LOSSES FOR INTERNAL FLOATING ROOF TANK
          WITH A LIQUID HEEL	   3-70

TABLE 3-18. ROOF LANDING LOSSES FOR EXTERNAL FLOATING ROOF TANK
          WITH A LIQUID HEEL	   3-71

TABLE 3-19. ROOF LANDING LOSSES FOR ALL DRAIN-DRY TANKS	   3-72

TABLE 3-20. HENRY'S LAW CONSTANTS FOR SELECTED ORGANIC LIQUIDS
          (REFERENCE 15) 	   3-74

TABLE 3-21. CORRECTION OF HENRY'S LAW FACTOR FOR A TEMPERATURE
          DIFFERENT FROM STANDARD	   3-77

TABLE 4-1.  FIXED ROOF TANK BREATHING LOSS-COMPARISON OF
          ESTIMATING EQUATIONS-API DATA BASE	   4-21

TABLE 4-2.  FIXED ROOF TANK BREATHING LOSS ~ COMPARISON OF
          ESTIMATING EQUATIONS-WOGA DATA BASE	   4-21

TABLE 4-3.  FIXED ROOF TANK BREATHING LOSS-COMPARISON OF
          ESTIMATING EQUATIONS-EPA DATA BASE	   4-22

TABLE 4-4.  COMPARISON OF EMISSION ESTIMATING EQUATIONS WITH
          BREATHING LOSS AS A FUNCTION OF STOCK TYPE	   4-23

TABLE 4-5.  COMPARISON OF EMISSION ESTIMATING EQUATIONS WITH
          BREATHING LOSS AS A FUNCTION OF VAPOR PRESSURE	   4-24

TABLE 4-6.  FIXED ROOF TANK BREATHING LOSS ~ COMPARISON OF
          ESTIMATING EQUATIONS ~ WOGA DATA BASE DEFAULT
          VALUES	   4-24

TABLE 4-7.  FIXED ROOF TANK BREATHING LOSS-COMPARISON
          OF ESTIMATING EQUATIONS-EPA DATA BASE DEFAULT
          VALUES	   4-25

TABLE 4-8.  SUMMARY OF STATISTICAL ANALYSIS VALUES	   4-26

TABLE 4-9.  COMPARISON OF BREATHING LOSS ESTIMATING EQUATIONS
          (USING DEFAULT VALUES)~PREDICTIVE ABILITY AS A
          FUNCTION OF PRODUCT TYPE	   4-27
                                  Vlll

-------
                         LIST OF TABLES (continued)

TABLE 4-10.  BREATHING LOSS ESTIMATING EQUATIONS FIXED ROOF TANKS
           SENSITIVITY ANALYSIS-COMPARISON BETWEEN THE AP-42 AND
           NEW API EQUATION BASELINE CONDITIONS	   4-28

TABLE 4-11.  BREATHING LOSS ESTIMATING EQUATIONS FIXED ROOF TANKS
           SENSITIVITY ANALYSIS	   4-29

TABLE 5-1.   BASIS OF RIM-SEAL LOSS FACTORS	    5-4

TABLE 5-2.   SUMMARY OF RIM SEAL LOSS FACTORS, KRa, KRb, AND n	    5-6

TABLE 5-3.   COMPARISON OF ESTIMATING EQUATION COEFFICIENTS:
           INDIVIDUAL CASES	    5-8

TABLE 5-4.   COMPARISON OF ESTIMATING EQUATIONS FOR AVERAGE
           FITTING SEAL LOSS FACTORS	   5-11

TABLE 5-5.   COMPARISON OF SLOTTED GUIDE POLE PARAMETER
           ESTIMATES	   5-31

TABLE 5-6.   SUMMARY OF RESULTS FOR LINEAR MODEL ANALYSES OF ALL
           FITTINGS	   5-34

TABLE 5-7.   RECOMMENDED GROUPING OF SLOTTED GUIDE POLE FITTINGS
           AND PARAMETERS FOR AP-42	   5-36

TABLE 5-8.   COMPARISON OF UNSLOTTED GUIDE POLE PARAMETER
           ESTIMATES	   5-38

TABLE 5-9.   SUMMARY OF PARAMETER ESTIMATES - CB&I/API REPLICATE
           ANALYSES	   5-40

TABLE 5-10.  SUMMARY OF PARAMETER ESTIMATES - MRI ANALYSES	   5-41

TABLE 5-11.  RECOMMENDED DECK-FITTING LOSS FACTORS, KFa, KFb,
           ANDm	   5-46

TABLE 5-12.  NONCONTACT DECK SEAM LOSS FACTORS BY TEST	   5-53

TABLE 5-13.  CONTACT DECK SEAM LOSS FACTORS BY TEST	   5-53

TABLE 5-14.  TANK PARAMETERS RECORDED DURING TESTS BY THE
           WESTERN OIL AND GAS ASSOCIATION	   5-57

-------
                         LIST OF TABLES (continued)

TABLE 5-15. PREDICTED AND ACTUAL EMISSION FROM TANKS TESTED
          BY THE WESTERN OIL AND GAS ASSOCIATION	    5-58

TABLE 5-16. RELATIVE ERRORS CALCULATED FOR PREDICTED EMISSIONS
          PRESENTED IN TABLE 5-13	    5-58

TABLE 5-17. FIELD TEST TANK PARAMETERS	    5-60

TABLE 5-18. RESULTS OF SENSITIVITY ANALYSIS FOR THE STANDING
          STORAGE LOSS EQUATION FOR EXTERNAL FLOATING ROOF
          TANKS 	    5-62

TABLE 5-19. RESULTS OF SENSITIVITY ANALYSIS FOR THE STANDING
          STORAGE LOSS EQUATION FOR INTERNAL FLOATING
          ROOF TANKS	    5-66

TABLE 5-20. SENSITIVITY ANALYSIS OF WITHDRAWAL LOSSES FROM
          EXTERNAL FLOATING ROOF TANKS	    5-69

TABLE 5-21. SENSITIVITY ANALYSIS OF WITHDRAWAL LOSSES FROM
          INTERNAL FLOATING ROOF TANKS	    5-71

-------
                 EMISSION FACTOR DOCUMENTATION FOR AP-42 SECTION
                                  Organic Liquid Storage Tanks

                                     1.0 INTRODUCTION

       The document Compilation of Air Pollutant Emission Factors (AP-42) has been published by the
U. S. Environmental Protection Agency (EPA) since 1972.  Supplements to AP-42 have been routinely
published to add new emission source categories and to update existing emission factors. AP-42 is
routinely updated by EPA to respond to new emission factor needs of EPA, State and local air pollution
control programs, and industry.

       An emission factor is a representative value that attempts to relate the quantity of a pollutant
released to the atmosphere with an activity associated with the release of that pollutant. The emission
factors presented in AP-42 may be appropriate to use in a number of situations, such as making source-
specific emission estimates for areawide inventories for dispersion modeling, developing control
strategies, screening sources for compliance purposes, establishing operating permit fees, and making
permit applicability determinations. The purpose of this report is to provide background information to
support revisions to AP-42 Section 7.1, Organic Liquid Storage Tanks.

       This background report consists of six chapters.  Chapter 1 includes the introduction to the report.
 Chapter 2 gives basic descriptions of fixed roof tanks, floating roof tanks, variable vapor space tanks, and
horizontal tanks. It also includes descriptions of the different types of rim seals and deck fittings on
floating roof tanks. Chapter 3 presents the emission estimation procedures for each tank type, as well as
the methodology for hazardous air pollutant (HAP) speciation.  Chapter 4 provides an evaluation of the
equations that predict standing storage and working losses from fixed roof tanks. Chapter 5 provides an
evaluation of the equations that predict standing storage and withdrawal losses from floating roof tanks.
Chapter 6 is a summary of changes to the section since the previous edition of AP-42 (February 1996).
                                              1-1

-------
                             2.0 STORAGE TANK DESCRIPTIONS

2.1.  INTRODUCTION

       This chapter presents basic descriptions of fixed-roof tanks (vertical and horizontal);
internal, external, and domed external floating roof tanks; pressure tanks; and variable vapor
space tanks. In addition, the chapter provides descriptions of perimeter seals and fittings for
internal, external, and domed external floating roofs.

2.2.  TYPES OF  STORAGE TANKS

       Seven types of vessels are used to store volatile organic liquids (VOL):

       1. Fixed-roof tanks;
       2. External floating roof tanks;
       3. Internal floating roof tanks;
       4. Domed external floating roof tanks;
       5. Horizontal tanks;
       6. Pressure tanks; and
       7. Variable vapor space tanks.

The first four tank types are cylindrical in shape with the axis oriented perpendicular to the
foundation. These tanks are almost exclusively above ground. Horizontal tanks (i.e., with the
axis parallel to the foundation) can be used above ground and below ground.  Pressure tanks often
are horizontally oriented and "bullet" or spherically shaped to maintain structural integrity at high
pressures. They  are located above ground. Variable vapor space tanks can be cylindrical or
spherical in shape.  The discussion below contains a detailed description of each of these tank
types.

2.2.1. Fixed-Roof Tanks

       Of currently used tank designs, the fixed-roof tank is the least expensive to construct and
is generally considered the minimum acceptable equipment for storing VOL's. A typical
fixed-roof tank, which is shown in Figure 2-1, consists of a cylindrical steel shell with a cone- or
dome-shaped roof that is permanently affixed to the tank shell. Most recently built tanks are of
all-welded construction and are designed to be both liquid- and vapor-tight.  However, older tanks
may be of riveted or bolted construction and may not be vapor-tight.  A breather valve
(pressure-vacuum valve), which is commonly installed on many fixed-roof tanks, allows the tank
to operate at a slight internal pressure  or vacuum.  Breather vents are typically set at 0.19 kPa
(0.75 in. w.c.) on atmospheric pressure fixed-roof tanks.1 Because this valve prevents the release
of vapors during  only very small changes in temperature, barometric pressure, or liquid level, the
emissions from a fixed-roof tank can be  appreciable. Additionally, gauge hatches/sample wells,
float gauges, and roof manholes provide accessibility to these tanks and also serve  as potential
sources of volatile emissions. Breather vents may be called conservation vents, although hardly
any conservation of vapors occurs at such low pressure settings.  Generally, the term conservation
vent is used to describe a pressure setting of 17 kPa (67 in. w.c.) or less. Vents with settings
greater than 17 kPa (67 in. w.c.) are commonly called 'pressure' vents.
                                           2-1

-------
 2.2.2 External Floating Roof Tanks2

       A typical external floating roof tank consists of an open-topped cylindrical steel shell
equipped with a roof that floats on the surface of the stored liquid, rising and falling with the
liquid level. The floating roof is comprised of a deck, fittings, and rim seal system. Floating roof
decks are constructed of welded steel plates and are of three general types:  pan, pontoon, and
double deck.  Although numerous pan-type decks are currently in use, the present trend is toward
pontoon and double-deck type type floating roofs.  The two most common types of external
floating-roof tanks are shown in Figures 2-2 and 2-3. Manufacturers supply various versions of
these basic types of floating decks, which are tailored to emphasize particular features, such as
full liquid contact,  load-carrying capacity, roof stability, or pontoon arrangement.  The liquid
surface is covered by the floating deck, except in the small annular space between the deck and
the shell; the deck may contact the liquid or float directly above the surface on pontoons.
External floating roof tanks are equipped with a rim seal system, which is attached to the roof
perimeter and contacts the tank wall. The rim seal system slides against the tank wall as the roof
is raised and lowered.  The floating deck is also equipped with fittings that penetrate the deck and
serve operational functions. The external floating roof design is such that evaporative losses from
the stored liquid are limited to losses from the rim seal system and deck fittings (standing storage
loss) and any exposed liquid on the tank walls (withdrawal loss).

 2.2.3  Internal Floating Roof Tanks3

       An internal floating roof tank has both a permanent fixed roof and a floating roof inside.
There are two basic types of internal floating roof tanks: tanks in which the fixed roof is
supported by vertical columns within the tank; and tanks with a self-supporting fixed roof and no
internal support columns. The fixed roof is not necessarily free of openings but does span the
entire open plan area of the vessel.  Fixed roof tanks that have been retrofitted to employ an
internal floating roof are typically of the first type, while external floating roof tanks that have
been converted to an internal floating roof tank typically have a self-supporting roof. Tanks
initially constructed with both a fixed roof and an internal floating roof may be of either type. An
internal floating roof tank has both a permanently affixed roof and a roof that floats inside the
tank on the liquid surface (contact deck) or is supported on pontoons several inches above the
liquid surface (noncontact deck). The internal floating roof rises and falls with the liquid  level.
A typical internal floating roof tank is shown in Figure 2-4.

       Contact-type decks  include (1) aluminum sandwich panels with a honey combed
aluminum core floating in contact with the liquid; (2) resin-coated, fiberglass-reinforced polyester
(FRP), buoyant panels floating in contact with the liquid; and (3) pan-type steel roofs, floating in
contact with the liquid with or without the aid of pontoons. The majority of contact internal
floating decks currently in VOL service are pan-type steel or aluminum sandwich panel type.
The FRP decks are less common.

       Several variations of the pan-type contact steel roof exist.  The design may include
bulkheads or open  compartments around the perimeter of the deck so that any liquid that may
leak or spill onto the deck is contained.  Alternatively, the bulkheads may be covered to form
sealed compartments (i.e., pontoons), or the entire pan may be covered to form a sealed, double-
deck, steel floating roof. Generally, construction is of welded steel.
                                           2-2

-------
       Noncontact-type decks are the most common type of deck currently in use, and typically
consist of an aluminum deck laid on an aluminum grid framework supported above the liquid
surface by tubular aluminum pontoons. The deck skin for the noncontact-type floating decks is
typically constructed of rolled aluminum sheets (about 1.5 meters [m] [4.9 feet (ft)] wide and
0.58 millimeter [mm] [0.023 inches (in)] thick). The overlapping aluminum sheets are joined by
bolted aluminum clamping bars that run perpendicular to the pontoons to improve the rigidity of
the frame. The deck skin seams can be metal on metal or gasketed with a polymeric material.
The pontoons and clamping bars form the structural frame of the floating deck. Deck seams in
the noncontact internal floating roof design are a  source of emissions. Aluminum sandwich panel
contact-type internal floating roofs also share this design feature.  The sandwich panels are joined
with bolted mechanical fasteners that are similar in concept to the noncontact deck skin clamping
bars.  Steel pan contact internal floating roofs are constructed of welded steel sheets and therefore
have no deck seams. Similarly, the resin-coated,  reinforced fiberglass panel decks have no
apparent deck seams. The panels are butted and lapped with resin-impregnated fiberglass fabric
strips. The significance of deck seams with respect to emissions from internal floating roof tanks
is addressed in Chapter 5.

       The internal floating roof physically occupies a finite volume of space that reduces the
maximum liquid storage capacity of the tank.  When the tank is completely full, the floating roof
touches or nearly touches the fixed roof. Consequently, the effective height of the tank decreases,
thus limiting the storage capacity.  The reduction in the effective height varies from about 0.15 to
0.6 m (0.5 to 2 ft), depending on the type and design of the floating roof employed.

       All types of internal floating roofs, like external floating roofs, commonly incorporate
rim seals that slide against the tank wall as the roof moves up and down. These seals are
discussed in detail in Section 2.3.2. Circulation vents and an open vent at the top of the fixed roof
are generally provided to minimize the accumulation of hydrocarbon vapors in concentrations
approaching the flammable  range.

       Flame arresters are  an option that can be  used to protect the vessel from fire or explosion.
When these are used, circulation vents are  not provided.  Tank venting occurs through a
pressure-vacuum vent and flame arrester.

2.2.2. Domed  External Floating Roof Tanks4	
       Domed external floating roof tanks have the heavier type of deck used in external floating
roof tanks as well as a fixed roof at the top of the shell like internal floating roof tanks. Domed
external floating roof tanks usually result from retrofitting an external floating roof tank with a
fixed roof.  A typical domed external floating roof tank is shown in Figure 2-5.

       As with the internal floating roof tanks, the function of the fixed roof is not to act as a
vapor barrier, but to block the wind.  The type of fixed roof most commonly used is a self
supporting aluminum dome roof, which is of bolted construction. Like the internal floating roof
tanks, these  tanks are freely vented by circulation vents at the top of the fixed roof. The deck
fittings and rim seals, however, are basically identical to those on external floating roof tanks.
                                           2-3

-------
2.2.5.  Horizontal Tanks

       Horizontal tanks are constructed for both above-ground and underground service.
Figures 2-6 and 2-7 present schematics of typical underground and above-ground horizontal
tanks.  Horizontal tanks are usually constructed of steel, steel with a fiberglass overlay, or
fiberglass-reinforced polyester. Horizontal tanks are generally small storage tanks with capacities
of less than 75,710 L (20,000 gallons). Horizontal tanks are constructed such that the length of
the tank is not greater than six times the diameter to ensure structural integrity. Horizontal tanks
are usually equipped with pressure-vacuum vents, gauge hatches and sample wells, and manholes
to provide accessibility to these tanks. In addition, underground tanks may be cathodically
protected to prevent corrosion of the tank shell. Cathodic protection is accomplished by placing
sacrificial anodes in the tank that are connected to an impressed current system or by using
galvanic anodes in  the tank. However, internal cathodic protection is no longer widely used in
the petroleum industry, due to corrosion inhibitors that are now found in most refined petroleum
products.

       The potential emission sources for above-ground horizontal tanks are the same as those
for fixed-roof tanks. Emissions from underground storage tanks are mainly associated with
changes in the liquid level in the tank. Losses due to changes in temperature or barometric
pressure are minimal for underground tanks because the surrounding earth limits the diurnal
temperature change and changes in the barometric pressure would result in only small losses.

2.2.6.  Pressure Tanks

       Two classes of pressure tanks are in general use:  low pressure (2.5 to 15 psig) and high
pressure (higher than 15 psig). Pressure tanks generally are used for storing organic liquids and
gases with high vapor  pressures and are  found in many sizes and shapes, depending on the
operating pressure of the tank. Pressure tanks are equipped with a pressure/vacuum vent that is
set to prevent venting loss from boiling and breathing loss from daily temperature or barometric
pressure changes.

2.2.7.  Variable Vapor Space Tanks5

       Variable vapor space tanks are equipped with expandable vapor reservoirs to
accommodate vapor volume fluctuations attributable to temperature and barometric pressure
changes.  Although variable vapor space tanks  are sometimes used independently, they are
normally connected to the vapor spaces of one  or more fixed  roof tanks. The two most common
types of variable vapor space tanks are lifter roof tanks and flexible diaphragm tanks.

       Lifter roof tanks have a telescoping roof that fits loosely around the outside of the main
tank wall. The space between the roof and the  wall is closed  by either a wet seal, which is a
trough filled with liquid, or a dry seal, which uses a flexible coated fabric.

       Flexible diaphragm tanks use flexible membranes to provide expandable volume. They
may be either separate gasholder units or integral units mounted atop fixed roof tanks.

       Variable vapor space tank losses occur during tank filling when vapor is displaced by
liquid. Loss of vapor occurs only when the tank's vapor storage capacity is exceeded.
                                           2-4

-------
2.3.   TYPES OF FLOATING ROOF PERIMETER SEALS

2.3.1   External and Domed External Floating Roof Rim Seals2'6'7

       Regardless of tank design, a floating roof requires a device to seal the gap between the
tank wall and the deck perimeter. A rim seal, or in the case of a two-seal system, the lower
(primary) rim seal, can be made  from various materials suitable for organic liquid service.  The
basic designs available for external floating roof rim seals are (1) mechanical (metallic) shoe
seals, (2) liquid-filled seals, and  (3) (vapor- or liquid-mounted) resilient foam-filled seals.
Figures 2-8 and 2-9 depict these  three general types of seals.

       One major difference in seal system design is the way in which the seal is mounted with
respect to the liquid surface.  Figure 2-8 shows a vapor space between the liquid surface and rim
seal, whereas in Figure 2-9, the seals rest on the liquid surface.  These liquid-filled and resilient
foam-filled seals are classified as liquid- or vapor-mounted rim seals, depending on their location.
Mechanical shoe rim seals are different in design from liquid-filled or resilient foam-filled rim
seals and cannot be characterized as liquid- or vapor-mounted. However, because the shoe and
envelope combination precludes contact between the annular vapor space above the liquid and the
atmosphere (see Figure 2-9),  the emission rate of a mechanical shoe seal is closer to that of a
liquid-mounted rim seal than that of a vapor-mounted rim seal.

       2.3.1.1. Mechanical  Shoe Seal. A mechanical shoe seal, also known as a "metallic shoe
seal" (Figure 2-9), is characterized by a metallic sheet (the "shoe") that is held against the vertical
tank wall.  Prior to 40 CFR 60 Subpart Ka, the regulations did not specify a height for
mechanical shoe seals; however, shoe heights typically range from 75 to  130 centimeters (cm)
(30 to 51 in.). The shoe is connected by braces to the floating deck and is held tightly against the
wall by springs or weighted levers.  A flexible coated fabric (the "envelope") is suspended from
the shoe seal to the floating deck to form a vapor barrier over the annular space between the deck
and the primary seal.

       2.3.1.2. Liquid-Filled Seal.  A liquid-filled rim seal (Figure 2-9) may consist of a tough
fabric band or envelope filled with a liquid, or it may consist of a flexible polymeric tube 20 to 25
cm (8 to 10 in.) in diameter filled with  a liquid and sheathed with a tough fabric scuff band. The
liquid is commonly a petroleum  distillate or other liquid that will not contaminate the stored
product if the tube ruptures.  Liquid-filled rim seals are mounted on the liquid product surface
with no vapor space below the seal.

       2.3.1.3. Resilient Foam-Filled Seal. A resilient foam-filled rim seal is similar to a
liquid-filled seal except that a resilient  foam log is used in place of the liquid. The resiliency of
the foam log permits the seal to adapt itself to minor imperfections in tank dimensions and  in the
tank shell. The foam log may be mounted above the liquid surface (vapor-mounted) or  on the
liquid surface (liquid-mounted).  Typical vapor-mounted and liquid-mounted seals are presented
in Figures 2-8 and 2-9, respectively.

       2.3.1.4. Secondary Seals on External Floating Roofs.  A secondary seal on an external
floating roof consists of a continuous seal mounted on the rim of the  floating roof and extending
to the tank wall, covering the entire primary seal. Secondary seals are normally constructed of
flexible polymeric materials.  Figure 2-10 depicts several primary and secondary seal systems.
An alternative secondary seal design incorporates a steel leaf to bridge the gap between the roof
                                           2-5

-------
and the tank wall. The leaf acts as a compression plate to hold a polymeric wiper against the tank
wall.

       A rim-mounted secondary seal installed over a primary seal provides a barrier for volatile
organic compound (VOC) emissions that escape from the small vapor space  between the primary
seal and the wall and through any openings or tears in the seal envelope of a metallic shoe seal
(Figure 2-10). Although not shown in Figure 2-10, a secondary seal can be used in conjunction
with a weather shield as described in the following section.

       Another type of secondary seal is a shoe-mounted secondary seal. A shoe-mounted seal
extends from the top of the shoe to the tank wall (Figure 2-10). These seals do not provide
protection against VOC leakage through the envelope. Holes, gaps, tears, or other defects in the
envelope can permit direct exchange between the saturated vapor under the envelope and the
atmosphere. Wind can enter this space through envelope defects, flow around the circumference
of the tank, and exit saturated or nearly saturated with VOC vapors.

       2.3.1.5.  Weather Shield.  A weather shield (see Figure 2-9) may be installed over the
primary seal to protect it from deterioration caused by debris and exposure to the elements.
Though the NSPS's  40 CFR 60 Subparts Ka and Kb do not accept the installation of a weather
shield as equivalent  to a secondary seal, there are a large number of existing  tanks not affected by
the NSPS that have this configuration. Typically, a weather shield is an arrangement of
overlapping thin metal sheets pivoted  from the floating roof to ride against the tank wall. The
weather shield, by the nature of its design, is not an effective vapor barrier. For this reason, it
differs from the secondary seal. Although the two devices are conceptually similar in design,
they are designed for and serve different purposes.

2.3.2. Internal Floating Roof Rim Seals3'7

       Internal floating roofs typically incorporate one of two types of flexible, product-resistant
rim seals: resilient foam-filled seals or wiper seals.  Similar to those employed on external
floating roofs, each of these seals closes the annular vapor space between the edge of the floating
deck and the tank shell to reduce evaporative  losses.  They are designed to compensate for small
irregularities in the tank shell and allow the roof to freely move up and down in the tank without
binding.

       2.2.2.1. Resilient Foam-Filled Seal. A resilient foam-filled seal used on an internal
floating roof is similar in  design to that described in Section 2.3.1.3 for external floating roofs.
Resilient foam-filled seals are shown in Figures 2-8  and  2-9. These seals can be mounted either
in contact with the liquid  surface (liquid-mounted) or several centimeters above the liquid surface
(vapor-mounted).

       Resilient foam-filled seals work because of the expansion and contraction of a resilient
material to maintain contact with the tank shell while accommodating varying annular rim space
widths.  These seals consist of a core of open-cell foam encapsulated in a coated fabric. The
elasticity of the foam core pushes the fabric into contact  with the tank shell.  The seals are
attached to a mounting on the deck perimeter and are continuous around the  roof circumference.
Polyurethane-coated nylon fabric and  polyurethane foam are commonly used materials. For
emission control, it is important that the mounting and radial seal joints be vapor-tight and that
the seal be in substantial contact with the tank shell.
                                           2-6

-------
       2.2.2.2. Wiper Seals. Wiper seals are commonly used as primary rim seals for internal
floating roof tanks. This type of seal is depicted in Figure 2-8.

       Wiper seals generally consist of a continuous annular blade of flexible material fastened
to a mounting bracket on the deck perimeter that spans the annular rim space and contacts the
tank shell. The mounting is such that the blade is flexed, and its elasticity provides a sealing
pressure against the tank shell. Such seals are vapor-mounted; a vapor space exists between the
liquid stock and the bottom of the seal. For emission control, it is important that the mounting be
vapor-tight, that the seal extend around the circumference of the roof, and that the blade be in
substantial contact with the tank shell.

       Three types of materials are commonly used to make the wipers.  One type consists of a
cellular, elastomeric material tapered in cross section with the thicker portion at the mounting.
Rubber is a commonly used material.  All radial joints in the blade are joined.

       A second type of wiper seal construction uses a foam core wrapped with a coated fabric.
Polyurethane on nylon fabric and polyurethane foam are common materials. The core provides
the flexibility and support, while the fabric provides the vapor barrier and wear surface.

       A third type of wiper seal consists of overlapping segments of seal material (shingle-type
seal). Shingle-type seals differ from the wiper seals discussed previously in that they do not
provide a continuous vapor barrier.

       2.2.2.3. Secondary Seals for Internal Floating Roof Tanks. Secondary seals may be used
to provide some additional evaporative loss control over that achieved by the primary seal. The
secondary seal is mounted to an extended vertical rim plate, above the primary seal, as shown in
Figure 2-10. Secondary seals can be either a resilient foam-filled seal or an elastomeric wiper
seal, as described in Sections 2.3.2.1 and 2.3.2.2, respectively. For a given roof design, using a
secondary seal further limits the operating capacity of a tank due to the need to maintain contact
with the tank shell or keep the seal from interfering with IFRT fixed-roof rafters when the tank is
filled.  Secondary seals are not commonly used on internal floating roof tanks that are not
affected by the NSPS (40 CFR 60 Subpart Kb).

2.3.  TYPES OF FLOATING ROOF DECK FITTINGS

2.3.1. External and Domed External Floating Roof Deck
Fittings2'6'7

       Numerous fittings penetrate or are attached to an external floating roof deck.  These
fittings accommodate structural support components or allow for operational functions. These
fittings can be a source of emissions in that they must penetrate the deck.  Other accessories are
used that do not penetrate the deck and are not, therefore,  sources of evaporative loss.  The most
common fittings relevant to controlling vapor losses are described in the following sections.

       2.3.1.1. Access Hatches. An access hatch consists of an opening in the deck with a
peripheral vertical well attached to the deck and a removable cover to close the opening as shown
in Figure 2-11. An access hatch is typically sized to allow workers and materials to pass through
the deck for construction or servicing.  The cover can rest directly on the well, or a gasketed
connection can be used to reduce evaporative loss.  Bolting the cover to the well reduces losses
further.
                                           2-7

-------
       23.1.2. Gauge Float Wells. Gauge floats are used to indicate the level of stock within
the tank.  These usually consist of a float residing within a well that passes through the floating
deck, as shown in Figure 2-11. The float is connected to an indicator on the exterior of the tank
via a tape passing through a guide system. The float rests on the stock surface within the well,
which is enclosed by a sliding cover. Evaporation loss can be reduced by gasketing and/or
bolting the connection between the cover and the rim of the well. The cable passes through a
bushing located at the center of the cover.  As with similar deck penetrations, the well extends
into the liquid stock on noncontact floating decks.

       2.3.1.3. Gauge Hatch/Sample Ports. Gauge hatch/sample ports provide access for hand-
gauging the level of stock in the tank and for taking samples of the tank contents. A gauge
hatch/sample port consists of a pipe sleeve through the deck and a self-closing gasketed cover, as
shown in Figure 2-11.  Gauge hatch/sample ports are usually located under the gauger's platform,
which is mounted on the top of the tank shell.  The cover may have a cord attached so that it can
be opened from the gauger's platform.  A gasketed cover reduces evaporative losses.

       2.3.1.4. Rim Vents.  Rim vents are found on tanks equipped with a rim seal system that
creates a vapor pocket, such as a mechanical shoe seal or double wiper seal system.  The rim vent
is connected to the rim vapor space by a pipe and  releases any excess pressure or vacuum that is
present (Figure 2-12).  The rim vapor space is bounded by the floating deck rim, the primary-seal
shoe, the liquid surface, and the primary-seal fabric.  Rim vents usually consist of weighted
pallets that rest on the gasketed surface.

       2.3.1.5. Deck Drains.  Deck drains permit removal of rainwater from the surface of
floating decks. Two types of floating roof drainage systems are currently used:  closed and open.
Closed drainage systems carry rainwater from the surface of the deck to the outside of the tank
through a flexible or articulated piping  system  or through a flexible hose system located below
the deck in the product space.  Since product does not enter this closed drainage  system, there is
no associated evaporative loss. Open drainage systems, consisting of an open pipe that extends a
short distance below the bottom of the deck, permit rainwater to  drain from the surface of the
deck into the product.  Since these drainpipes are filled with product to the product level in the
tank, evaporative loss occurs from the top of the open drainpipes. Two types of roof drains are
commonly used in open drainage systems: flush drains and overflow drains.  Flush drains
(Figure 2-12) have a drain opening that is flush with the top surface of the double deck. They
permit rainwater to drain into the product.  Overflow drains (Figure 2-12) consist of a drain
opening that is elevated above the top surface of the deck, thereby limiting the maximum amount
of rainwater that can accumulate on the deck and providing emergency drainage of rainwater.
They are normally used in conjunction  with a closed drainage system. Some open deck drains are
equipped with an insert to reduce the evaporative loss.

       2.3.1.6. Deck Legs.  Deck legs prevent damage to fittings underneath the deck and allow
for tank cleaning or repair by holding the deck at a predetermined distance from the tank bottom.
These supports consist of adjustable or fixed legs attached to the floating deck as shown in
Figure 2-12.  For adjustable deck legs, the load-carrying element passes through a well or sleeve
in the deck.

       2.3.1.7. Slotted and Unslotted  Guide Poles and Wells. Antirotation devices are used to
prevent floating roofs from rotating and potentially damaging roof equipment and rim seal
systems.  A commonly used antirotation device is a guide pole that is fixed at the top and bottom
of the tank (Figure 2-13).  The guide pole passes through a well in the deck. Rollers attached to
the top of the well ride on the outside surface of the guide pole to prevent the floating roof from
                                           2-8

-------
rotating.  The guide pole well has a sliding cover to accommodate limited radial movement of the
roof.  The sliding cover can be equipped with a gasket between the guide pole and the cover to
reduce evaporative loss.  The guide pole well can also be equipped with a gasket between the
sliding cover and the top of the well to reduce evaporative loss. Openings at the top and bottom
of the guide pole provide a means of hand-gauging the tank level and of taking bottom samples.
In the slotted guide pole/sample well application, the well of the guide pole is constructed with  a
series of drilled holes or slots that allow the product to mix freely in the guide pole and thus have
the same composition and liquid level as the product in the tank. Evaporative loss from the
guidepole can be reduced by modifying the guidepole  or well, or by placing a removable float
inside the pole.  Deck fitting factors for slotted guidepoles without pole sleeves were determined
from test data on fittings where the  float top or float wiper was positioned at or above the sliding
cover elevation.  Tests were not conducted with floats where the top of the float or wiper was
below the sliding cover elevation ("short" floats); emissions from such a configuration are
expected to be somewhere between those for guidepoles with and without floats, depending upon
the float height. When a pole sleeve is used, the evaporative loss will not be affected by the
height of the float within the well, since the purpose of the pole sleeve is to restrict the flow of
vapor from the vapor space  below the deck into the slotted guidepole.

       2.3.1.8. Vacuum Breakers. The purpose of a vacuum breaker is to allow for the
exchange  of vapor and air through the floating roof during filling and emptying. Vacuum
breakers are designed to be  activated by changes in pressure or liquid level, or strictly by
mechanical means.

       Mechanical vacuum breakers are activated when the deck is either being landed on its
legs or floated off its legs to equalize the pressure of the vapor space across the deck. This is
accomplished by opening a  deck penetration that usually consists of a well formed of pipe or
framing on which rests a cover (Figure 2-12). Attached to the underside of the cover is a guide
leg long enough to contact the tank  bottom as the external floating deck approaches the tank
bottom. When in contact with the tank bottom, the guide leg mechanically opens the breaker by
lifting the cover off the well. When the leg is not contacting the bottom, the penetration is closed
by the cover resting on the well.  The closure may or may not have a gasket between the cover
and neck.  Since the purpose of the  vacuum breaker is to allow the free exchange of air and/or
vapor, the well does not extend appreciably below the deck. The gasket on the underside of the
cover, or conversely on the upper rim of the well, provides a small measure of emission control
during periods when the roof is free floating and the breaker is closed.

2.3.2. Internal Floating Roof Fittings3'7

       Numerous fittings penetrate or are attached to an internal floating deck. These fittings
serve to accommodate structural support components or to allow for operational functions.  The
fittings can be a source of evaporative loss in that they require penetrations in the deck.  Other
accessories are used that do not penetrate the deck and are not, therefore, sources of evaporative
loss. The most common fittings relevant to controlling vapor losses are described in the
following sections.

       The access hatches, deck legs, vacuum breakers, and automatic gauge float wells for
internal floating roofs are similar fittings to those described earlier for external floating roofs.
Therefore, the discussion is  not repeated.

       2.3.2.1.  Column Wells.  The most common fixed-roof designs (Figure 2-4) are normally
supported from  inside the tank by means of vertical columns, which necessarily penetrate the
                                           2-9

-------
floating deck.  (Some fixed roofs are entirely self-supporting and, therefore, have no support
columns.) Columns are made of pipe with circular cross sections or of structural shapes with
irregular cross sections (built-up). The number of columns varies with tank diameter from a
minimum of 1 to over 50 for very large tanks. A typical fixed roof support column is shown in
Figure 2-11.

        The columns pass through deck openings via peripheral vertical wells. With noncontact
decks, the well should extend down into the liquid stock. Generally, a closure device exists
between the top of the well and the column. Several proprietary designs exist for this closure,
including sliding covers and fabric sleeves, which must accommodate the movements of the deck
relative  to the column as the liquid level changes.  A sliding cover rests on the upper rim of the
column  well (which is normally fixed to the deck) and bridges the gap or space between the
column  well and the column. The cover, which has a cutout, or opening, around the column,
slides vertically relative to the column as the deck raises and lowers. At the same time, the cover
slides horizontally relative to the rim of the well.  A gasket around the rim of the well reduces
emissions from this fitting. A flexible fabric sleeve seal between the rim of the well and the
column  (with a cutout, or opening to allow vertical motion of the seal relative to the columns)
similarly accommodates limited horizontal motion of the roof relative to the column.  A third
design combines the advantages of the flexible fabric sleeve seal with a well that excludes all but
a small portion of the liquid surface from direct exchange with the vapor space above the  floating
deck.

        2.3.2.2.  Sample Pipes or Wells. A sample well may be provided to allow liquid stock
sampling. Typically, the well is funnel-shaped to allow for easy entry of a sample thief. A
closure is provided, which is typically located at the lower end of the funnel and which frequently
consists of a horizontal piece of fabric slit radially to allow thief entry. The well should extend
into the  liquid stock on noncontact decks.

        Alternately, a sample well may consist of a slotted pipe extending into the liquid stock
equipped with an ungasketed or gasketed sliding cover.

        2.3.2.3.  Ladder We 11s.  Some tanks are equipped with internal ladders that extend from a
manhole in the fixed roof to the tank bottom. The deck opening through which the ladder passes
is constructed with similar design details and considerations to those for column wells, as
discussed in Section 2.4.2.1.  A typical ladder and well are shown in Figure 2-14.
                                          2-10

-------
Breather vent (open or P/V type)
Float gauge conduit
Tank roof and shell
(not insulated)
Gauge-hatch/
sample well
Roof manhole
No floating roof

Stable (nonboiling)
stock liquid
                                         Figure 2-1 Typical Fixed Roof Tank
                                                     2-11

-------
                       Overflow drain
                   Deck leg
                   (center area)
    Rim seal
(mechanical-shoe)
Open top (no fixed roof)


       Access hatch
                                                             Gauge hatch/
                                                              sample port
                                                                  Solid guidepole
                                                                    (unstotted)
                                                                                  Tank shell
         Rim vent
         Figure 2-2.  External floating roof tank (pontoon type).
                                                                       11
                                      2-12

-------
                               Open tup :ne feod rtOTI
                                                             MMMQ
Figure 2-3.  External floating roof tank (double-deck type).11
                           2-13

-------
   MBWM
i. vapor-mounted}

     Sample port
L
                                                                                    sh*ll
                                                    Fued-rooT
                                                      support column
                                             -Decx Grain
                      Figure 2-4.  Internal floating roof tank.11
                                         2-14

-------
                                                                       Fixed-roof center vent
                                                                                           Fixed roof
                                                                                      (self-supporting
                                                                                           aluminum
                                                                                              dome)
  Peripheral venting typically
       provided at the eaves
    Rim seal
(mechanical-shoe)

         Rim vent
                                                                                           Tank shell
                                                                           Gauge float
                 Deck leg
                 (pontoon area)
                      Deck leg
                      (center area)
       Solid guidepoie
         (un slotted)
Gauge hatch/
sample port
                         Overflow drain
                                                  Access hatch
                      Figure 2-5.  Domed external floating roof tank11.
                                             2-15

-------
Figure 2-6. Typical underground storage tank.
                   2-16

-------
                                       Pressure
                                       vacuum vent











1

5







/

V








1

1











, 1
1




TT
J
/
• Hrain i/alwa











1


i

J




•
r Containment
mm


\
I
Figure 2-7. A typical above-ground horizontal tank.
                      2-17

-------
                       Roating roof deck



                               Liquid surface -
               Resilient-filled seal
       (not in contact with the liquid surface)


          (see section view below)
                         Elastomeric-coatad
                           fabric envelope
                             Resilient
                            foam core
Rim vapor
space
Liquid
surface
                                    Floating
                                        roof
                                       deck
\A r—Tank shell

   \±          /—Floating roof deck


                          Liquid surface -
                • Flexible-wiper seal
           (wiper position may vary with tne
           floating roofs direction of travel)

           (see section views below)
Tank
shell
Rim vapor
space
                                                     Liquid
                                                     surface
                            Elastomenc blade
                                                     Tank
                                                     shell
                                                     Rim vapor
                                                     space
                                                     Liquid
                                                     surface
                         EElastomaric-coated
                           fabric envelope

                            Foam core
                                     Floatino
                                        roof
                                       deck
                       Figure 2-8.  Vapor mounted primary seals.1
                                             2-18

-------
                      Floating roo! dae*



                            Liquid surface-
             R«4*int-f*»d M«|
  (bottom of «nl in contact with B» liquid fcjrfaw)

         (see aaction view below)
Tanfe-
ihri
Liqud—
•urteee
       j^^™ vw^nw
r    / (rwt «»*n «twe)

 ^^      y	R«tlBflt CO
    \    /foam of

 ^^^       ^	RMtin?
_fj	,^ f         j
i^J
                                     root
                                    dec*
                                           -Tank shall   /-FlutingKK*deck

                                                                Imwy-H
                                                                 Mote
                                            (see section view below)
                                               Motalfc
                                               epaco-


                                               Uquid.
            Figure 2-9.  Liquid-mounted and mechanical shoe primary seals11.
                                            2-19

-------
     Rim-mounted secondary seal
                 over
       resilient-filled primary seal

                       Secondary seal
                     (flexible wiper shown)
    Shoe-mounted secondary seal
                over
    mechanical-shoe primary seal
                                                 Primary seal
                                                 (mechanical
                                                 shoe)	.
                                                 Liquid —
                                                 surface
                   -Tank shell
                        Secondary-seal
                        (shoe-mounted)
     Rim-mounted secondary seal
                 over
       flexible-wiper primary seal

                       Secondary seal
                     (flexible wiper shown)
                            Rim extender
Primary seal—•
(flexible-wiper)
Liquid
surface
     Rim-mounted secondary seal
                over
     mechanical-shoe primary seal
Primary seal
(mechanical
shoe)    s.
Liquia —
surface
                   -Tank shell
                                                                         Secondary-seal
                                                                          (rim-mounted)
&
                       Figure 2-10.  Secondary rim seals.
                                        2-20

-------
Removable
cover
Floating
   roof
   deck
Well
          (see section view below)
Handle	

Removable cover

Gasket

Well

Liquid
surface
 Bolted
 closed

Floating
   roof
   deck
               Access Hatch
Cable

Removable
cover
Floating
   roof
   deck
Well
          (see section view below)
                               Removable cover
Float
Floating
roof
deck-
Pipe column
                      (see section view below)
                                         deck
                                   (noncontact
                                   type shown)
           Liquid
           surface
                   Fixed-Roof Support Column
Self-   Cord
closing
cover

Pipe
sleeve
through
the
deck
                                                 Gauge-hatch/
                                                 sample port
                                                    -Slit-
                                                   fabric
                                              sample port
                                  (internal floating roofs only)
                      (see section view below)
           Cord
           (shown pulling
            cover open)

           Gasket

           Pipe
           sleeve

           Liquid
           surface
                Gauge float                                     Sample Ports

                   Figure 2-11. Deck fittings for floating roof tanks11
                                          2-21

-------
(see section view below)

                     miBw pinhote
                                                                Flush   Healing
                                                                drain      roof
                                                                         feck
drain	f^~	        • l.^-'	Pipesiwb


          (see section view below)
                                 Flush drain
   Vacuym Breaker
               Deck Drains
                                    Tank
                                    tlnrfl^
                                                                      Rhttvent
                                    *noe*eal—'
                                             ; see sedlon view below)
                                    UquU
                                    surfac*
      Deck Leg                                     Rim Vent

       Figure 2-12.  Deck fittings for floating roof tanks11.
                               2-22

-------
                        Solid guidepole
                                  (see section views below)
Solid guidepole


Roller assembly
Sliding
cover
Well
                                                 Solid guidepole

                                                 Roller assembly

                                                          Pole
                                                        sleeve
                                 Unslotted (solid) Guidepole
                        Slotted guidepole

                        Roller assembly
                                                         Slots in guidepole
                                                         (2 staggered rows
                                                         on opposite sides)
Slotted guidepole

Roller assembly

Sliding cover
                                                                                 Slotted guidepole

                                                                                  Roller assembly
(see section views below)
                                                                                            Pole
                                                                                          sleeve
                               Slotted (perforated) Guidepole

                   Figure 2-13.  Slotted and unslotted guidepoles1:
                                            2-23

-------
Floating
roof
deck
            (see section view below)
Ladder
Liquid
surface
     Ladder

     Sliding
      cover
                                              Well
     Sliding
      cover

    Floating
       roof
     <  deck
(noncontact
type shown)
                  Figure 2-14. Ladder and well.:
                            2-24

-------
     Full Liquid Heel
      (standing liquid
  across the entire bottom)
 Partial Liquid Heel
(standing liquid only
 in or near a sump;
 clingage elsewhere)
      Drain Dry
  (no standing liquid,
only liquid is clingage)
Figure 2-15.    Bottom conditions for landing loss
                                        2-25

-------
2.5  REFERENCES

1.     Evaporative Loss from Fixed-Roof Tanks, API Draft Publication 2518, Second
      Edition, American Petroleum Institute, Washington, DC, October 1991.

2.     Evaporative Loss From External Floating Roof Tanks, API Publication 2517,
      Third Edition, American Petroleum Institute, Washington, DC, February 1989.

 3.    Evaporation Loss From Internal Floating Roof Tanks, API Publication 2519,
      Third Edition, American Petroleum Institute, Washington, DC, June 1983.

 4.    Ferry, R.L., Estimating Storage Tank Emissions-Changes are Coming, TGB
      Partnership, 1994.

 5.    Use of Variable Vapor Space Systems to Reduce Evaporation Loss, API Bulletin
      2520, American Petroleum Institute, Washington, DC, September 1964.

 6.    VOC Emissions From Volatile Organic Liquid Storage Tanks - Background
      Information for Proposed Standards, EPA-450/3-81-003a, U. S. Environmental
      Protection Agency, Research Triangle Park, NC, July 1984.

 7.    Benzene Emissions From Benzene Storage Tanks - Background Information for
      Proposed Standards, EPA-450/3-80-034a, U. S. Environmental Protection
      Agency, Research Triangle Park, NC, December 1980.

 8.    Laverman, R. J., Emission Reduction Options for Floating Roof Tanks, Chicago
      Bridge and Iron Technical Services Company,  presented at the Second
      International Symposium on Aboveground Storage Tanks, Houston, TX, January
      1992.

 9.    Recommended Practices for Underground Storage of Petroleum, New York State
      Department of Environmental Conservation, Albany, NY, May 1984.

10.    Toxic Substance Storage Tank Containment, Ecology and Environment, Inc.,
      1985.

11.    Courtesy of R. Ferry, TGB Partnership, Hillsborough, NC.
                                     2-26

-------
                          3.0  EMISSION ESTIMATION PROCEDURES

3.1 INTRODUCTION

       The following section presents the emission estimation procedures for fixed roof, external
floating roof, domed external floating roof, and internal floating roof tanks. These procedures are valid
for all petroleum liquids, pure volatile organic liquids, and chemical mixtures with similar true vapor
pressures. It is important to note that in all the emission estimation procedures the physical properties of
the vapor do not include the noncondensibles (e.g., air) in the gas but only refer to the condensible
components of the stored liquid. To aid in the emission estimation procedures, a list of variables with
their corresponding definitions was developed and is presented in Table 7.3-1.

       The factors presented in AP-42 are those that are currently available and have been reviewed and
approved by the U. S. Environmental Protection Agency. As storage tank equipment vendors design new
floating decks and equipment, new emission  factors may be developed based on that equipment. If the
new emission factors are reviewed and approved, the emission factors will be added to AP-42 during the
next update.

       The emission estimation procedures outlined in this chapter have been used as the basis for the
development of a software program to estimate emissions from storage tanks. The software program
entitled "TANKS" is available through the CHIEF website maintained by the U. S. Environmental
Protection Agency.

3.1.1  Total Losses From Fixed Roof Tanks4'8"14 -

       The following equations, provided to estimate standing storage and working loss emissions, apply
to tanks with vertical cylindrical  shells and fixed roofs. These tanks must be substantially liquid- and
vapor-tight and  must operate approximately at atmospheric pressure. The equations are not intended to be
used in estimating losses from unstable or boiling stocks or from mixtures of hydrocarbons or
petrochemicals  for which the vapor pressure is not known or cannot be readily predicted. Total losses
from fixed roof tanks are equal to the sum of the standing storage loss and working loss:

                                          LT = LS + LW                                     (3-1)
where:
         LT =  total losses, Ib/yr
         Ls =  standing storage losses, Ib/yr, see Equation 3-2
        Lw =  working losses, Ib/yr, see Equation 3-29
                                              3-1

-------
3.1.1.1  Standing Storage Loss

        The standing storage loss, Ls, refers to the loss of stock vapors as a result of tank vapor space
breathing. Fixed roof tank standing storage losses can be estimated from Equation 3-2, which comes from
the previous edition of Chapter 7 of AP-42.

                                      Ls = 365VvWvKEKs                                  (3-2)
where:
         Ls =  standing storage loss, Ib/yr
        Vv =  vapor space volume, ft3, see Equation 3-3
        Wv =  stock vapor density, lb/ft3
        KE =  vapor space expansion factor, dimensionless
        Ks =  vented vapor saturation factor, dimensionless
        365 =  constant, the number of daily events in a year, (year)"1

Tank Vapor Space Volume. Vy - The tank vapor space volume is calculated using the following equation:


                                                                                            (3-3)
where:
        Vv =  vapor space volume, ft3
          D =  tank diameter, ft, see Equation 3-13 for horizontal tanks
       HVO =  vapor space outage, ft, see Equation 3-15

The standing storage loss equation can be simplified by combining Equation 3-2 with Equation 3-3. The
result is Equation 3-4.

                                             (^  _\
                                                                                            (3-4)
where:
         Ls =  standing storage loss, Ib/yr
         KE =  vapor space expansion factor, dimensionless, see Equation 3-5, 3-6, or 3-7
          D =  diameter, ft, see Equation 3-13 for horizontal tanks
       HVO =  vapor space outage, ft, see Equation 3-15; use HE/2 from Equation 3-14 for horizontal
               tanks
         Ks =  vented vapor saturation factor, dimensionless, see Equation 3-20
        Wv =  stock vapor density, lb/ft3, see Equation 3-21
        365 =  constant, the number of daily events in a year, (year)"1
                                               3-2

-------
Vapor Space Expansion Factor. KR

       The calculation of the vapor space expansion factor, KE, depends upon the properties of the liquid
in the tank and the breather vent settings. If the liquid stock has a true vapor pressure greater than 0.1
psia, or if the breather vent settings are higher than the typical range of ±0.03 psig, see Equation 3-7. If
the liquid stored in the fixed roof tank has a true vapor pressure less than 0.1 psia and the tank breather
vent settings are ±0.03 psig, use either Equation 3-5 or Equation 3-6.
       If the tank location and tank color and condition are known, KE is calculated using the following
equation:
                   KE =0.0018AT; =0.0018 [0.72(TAX-TAN) + 0.028
(3-5)
where:
         KE =  vapor space expansion factor, dimensionless
       A TV =  daily vapor temperature range, °R
        TAX =  daily maximum ambient temperature, °R
        TAN =  daily minimum ambient temperature, °R
         a =  tank paint solar absorptance, dimensionless
           I =  daily total solar insolation on a horizontal surface, Btu/(ft2 day)
     0.0018=  constant, fR)'1
       0.72 =  constant, dimensionless
      0.028 =  constant, (°R ft2 day)/Btu

       If the tank location is unknown, a value of KE can be calculated using typical meteorological
conditions for the lower 48 states. The typical value for daily solar insolation is 1,370 Btu/(ft2 day), the
daily range of ambient temperature is 21°R, the daily minimum ambient temperature is 473.5 °R, and the
tank paint solar absorptance is 0.17 for white paint in good condition. Substituting these values into
Equation 3-5 results in a value of 0.04, as shown in Equation 3-6.
                                           KE = 0.04
(3-6)
                                               3-3

-------
       When the liquid stock has a true vapor pressure greater than 0.1 psia, a more accurate estimate of
the vapor space expansion factor, KE, is obtained by Equation 3-7. As shown in the equation, KE is greater
than zero. If KE is less than zero, standing storage losses will not occur.

                                         A7V  APF -APB
                                    K^=TL+ P  -P    >0                                (3-7)
                                         1 LA    rA  rVA
where:
       A TV =  daily vapor temperature range, °R; see Note 1
       A Pv =  daily vapor pressure range, psi; see Note 2
       A PB =  breather vent pressure setting range, psi; see Note 3
         PA =  atmospheric pressure, psia
        PVA =  vapor pressure at daily average liquid surface temperature, psia; see Notes 1 and 2 for
               Equation 3-21
        TLA =  daily average liquid surface temperature, °R; see Note 3 for Equation 3-21

Notes:

        1. The daily vapor temperature range, ATV , is calculated using the following equation:

                                   A TV =0.72 ATA + 0.028 a I                             (3-8)

where:
       A TV =  daily vapor temperature range, °R
       A TA =  daily ambient temperature range, °R; see Note 4
         a =  tank paint solar absorptance, dimensionless; see Table 7.3-6
           I =  daily total solar insolation factor, Btu/ft2 d; see Table 7.3-7

        2. The daily vapor pressure range, APV, can be calculated using the following equation:

                                         APv = Pvx-PvN                                    (3-9)

where:
       A Pv =  daily vapor pressure range, psia
        PVX =  vapor pressure at the daily maximum liquid surface temperature, psia; see Note 5
        PVN =  vapor pressure at the daily minimum liquid surface temperature, psia; see Note 5
                                               3-4

-------
       The following method can be used as an alternate means of calculating APV for petroleum
liquids:
                                                 '-LA
                                                    2 -                                 (3-10)
where:
       A Pv =  daily vapor pressure range, psia
          B =  constant in the vapor pressure equation, °R; see Note 2 to Equation 3-21
        PVA =  vapor pressure at the daily average liquid surface temperature, psia; see Notes 1 and 2 to
               Equation 3-21
        TLA =  daily average liquid surface temperature, °R; see Note 3 to Equation 3-21
       A TV =  daily vapor temperature range, °R; see Note 1

        3. The breather vent pressure setting range,  APB, is calculated using the following equation:

                                         APB = PBP-PBV                                    (3-11)

where:
       A PB =  breather vent pressure setting range, psig
        PBP =  breather vent pressure setting, psig
        PBV =  breather vent vacuum setting, psig

        If specific information on the breather vent pressure setting and vacuum setting is not available,
assume 0.03 psig for PBp and -0.03 psig for PBv as typical values. If the fixed roof tank is of bolted or
riveted construction in which the roof or shell plates are not vapor tight, assume that A PB = 0, even if a
breather vent is used.

        4. The daily ambient temperature range, ATA, is calculated using the following equation:

                                         ATA = TAX-TAN                                   (3-12)
where:
       A TA =  daily ambient temperature range, °R
        TAX =  daily maximum ambient temperature, °R
        TAN =  daily minimum ambient temperature, °R

        Table 7.3-7 gives values of TAX and TAN for selected cities in the United States.
                                               3-5

-------
        5. The vapor pressures associated with daily maximum and minimum liquid surface temperature,
PVX and PVN, respectively, are calculated by substituting the corresponding temperatures, TLX and TLN,
into the vapor pressure function discussed in Notes  1 and 2 to Equation 3-21. If TLX and TLN are
unknown, Figure 7.3-17 can be used to calculate their values.

Diameter

        For vertical tanks, the diameter is straightforward. If a user needs to estimate emissions from a
horizontal fixed roof tank, some of the tank parameters can be modified before using the vertical tank
emission estimating equations. First, by assuming that the tank is one-half filled, the surface area of the
liquid in the tank is approximately equal to the length of the tank times the diameter of the tank. Next,
assume that this area represents a circle, i.e., that the liquid is an upright cylinder. Therefore, the effective
diameter, DE, is then equal to:
                                                                                           (3-13)
where:
         DE =  effective tank diameter, ft
          L =  length of the horizontal tank, ft (for tanks with rounded ends, use the overall length)
          D =  diameter of a vertical cross-section of the horizontal tank, ft

        By assuming the volume of the tank to be approximately equal to the cross-sectional area of the
tank times the length of the tank, an effective height, HE, of an equivalent upright cylinder may be
calculated as:


                                           HE =-D                                      (3-14)
                                                 4

        DE should be used in place of D in Equation 3-4 for calculating the standing storage loss (or in
Equation 3-3, if calculating the tank vapor space volume). One-half of the effective height, HE, should be
used as the vapor space outage, HVo, in these equations. This method yields only a very approximate
value for emissions from horizontal storage tanks. For underground horizontal tanks, assume that no
breathing or standing storage losses occur (Ls = 0) because the insulating nature of the earth limits the
diurnal temperature change. No modifications to the working loss equation are necessary for either above-
ground or underground horizontal tanks.
                                               3-6

-------
               Vapor Space Outage
                      The vapor space outage, HVo is the height of a cylinder of tank diameter, D, whose volume is
               equivalent to the vapor space volume of a fixed roof tank, including the volume under the cone or dome
               roof. The vapor space outage, HVo, is estimated from:
               where:
              Notes:
               where:
                                                     HVO — HS - HL +
                      HVO =  vapor space outage, ft; use HE/2 from Equation 3-14 for horizontal tanks
                       Hs=  tank shell height, ft
                       HL =  liquid height, ft
                      HRO =  roof outage, ft; see Note 1 for a cone roof or Note  2 for a dome roof
                      1. For a cone roof, the roof outage, HRO, is calculated as follows:

                                                        HRQ = 1/3 HR
                        RO =  roof outage (or shell height equivalent to the volume contained under the roof), ft
                       HR=  tank roof height, ft
                                    (3-15)
                                    (3-16)
                                                         HR — SR RS
                                    (3-17)
                             where:
                                     SR =    tank cone roof slope, ft/ft; if unknown, a standard value of 0.0625 is used
                                     Rs =    tank shell radius, ft
                      2. For a dome roof, the roof outage, HRO, is calculated as follows:
                                                             1    1
                                                            — H—
                                                             2    6
H,
                                    (3-18)
               where:
                        RO =  roof outage, ft
                        RS =  tank shell radius, ft
                       HR =  tank roof height, ft
(
                                                                                                        (3-19)
                                    HR=   tank roof height, ft
                                    RR =   tank dome roof radius, ft
                                    Rs =   tank shell radius, ft
                                                             3-7

-------
The value of RR usually ranges from 0.8D - 1.2D, where D = 2 Rg. If RR is unknown, the tank diameter is
used in its place. If the tank diameter is used as the value for RR, Equations 3-18 and 3-19 reduce to
HRO = 0. 137 Rs and HR = 0.268 Rs.

Vented Vapor Saturation Factor. Ks

        The vented vapor saturation  factor, Ks, is calculated using the following equation:

                                                  1
                                              0.053 Pn
                                                                                           (3-20)
where:
         Ks =  vented vapor saturation factor, dimensionless
        PVA =  vapor pressure at daily average liquid surface temperature, psia; see Notes 1 and 2 to
               Equation 3-21
       HVO =  vapor space outage, ft, see Equation 3-15
      0.053 =  constant, (psia-ft)"1

Stock Vapor Density. Wy - The density of the vapor is calculated using the following equation:
                                                RTLA
where:
        Wv =  vapor density, lb/ft3
        Mv =  vapor molecular weight, Ib/lb-mole; see Note 1
          R =  the ideal gas constant, 10.73 1 psia ft3/lb-mole °R
        PVA =  vapor pressure at daily average liquid surface temperature, psia; see Notes 1 and 2
        TLA =  daily average liquid surface temperature,  °R; see Note 3

Notes:

        1. The molecular weight of the vapor, Mv, can be determined from Table 7.3-2 and 7.3-3 for
selected petroleum liquids and volatile organic liquids, respectively, or by analyzing vapor samples.
Where mixtures of organic liquids are stored in a tank, Mv can be calculated from the liquid composition.
The molecular weight of the vapor. Mv, is equal to the sum of the molecular weight, Mi; multiplied by the
vapor mole fraction, yi; for each component. The vapor mole fraction  is equal to the partial pressure of
component i divided by the total vapor pressure. The partial pressure of component i is equal to the true
vapor pressure of component i (P) multiplied by the liquid mole fraction, (xj). Therefore,
                                                                                           (3-22)

-------
where:
        PVA, total vapor pressure of the stored liquid, by Raoult's Law, is:
                                                                                            (3-23)
        For more detailed information, please refer to Section 7.1.4.

        2. True vapor pressure is the equilibrium partial pressure exerted by a volatile organic liquid, as
defined by ASTM-D 2879 or as obtained from standard reference texts. Reid vapor pressure is the
absolute vapor pressure of volatile crude oil and volatile nonviscous petroleum liquids, except liquified
petroleum gases, as determined by ASTM-D-323. True vapor pressures for organic liquids can be
determined from Table 7.3-3. True vapor pressure can be determined for crude oils using Figures 7.3-13a
and 7.3-13b. For refined stocks (gasolines and naphthas), Table 7.3-2 or Figures 7.3-14a and 7.3-14b can
be used. In order to use Figures 7.3-13a, 7.3-13b, 7.3-14a, or 7.3-14b, the stored liquid surface
temperature, TLA, must be determined in degrees Fahrenheit. See Note 3 to determine TLA.

        Alternatively, true vapor pressure for selected petroleum liquid stocks, at the stored liquid surface
temperature, can be determined using the following equation:
                                     PVA  =exP
A-\
                                                       B
                                                     JLA.
(3-24)
where:
        exp =  exponential function
          A =  constant in the vapor pressure equation, dimensionless
          B =  constant in the vapor pressure equation, °R
        TLA =  daily average liquid surface temperature, °R
        PVA =  true vapor pressure, psia

        For selected petroleum liquid stocks, physical property data are presented in Table 7.3-2. For
refined petroleum stocks, the constants A and B can be calculated from the equations presented in
Figure 7.3-15 and the distillation slopes presented in Table 7.3-4. For crude oil stocks, the constants A
and B can be calculated from the equations presented in Figure 7.3-16. Note that in Equation 3-24, TLA is
determined in degrees Rankine instead of degrees Fahrenheit.
                                                3-9

-------
        The true vapor pressure of organic liquids at the stored liquid temperature can be estimated by
Antoine's equation:
                                    \ogPVA=A-—-—                                 (3-25)
where:
          A =  constant in vapor pressure equation
          B =  constant in vapor pressure equation
          C =  constant in vapor pressure equation
        TLA =  daily average liquid surface temperature, °C
        PVA =  vapor pressure at average liquid surface temperature, mm Hg

        For organic liquids, the values for the constants A, B, and C are listed in Table 7.3-5. Note that in
Equation 3-25, TLA is determined in degrees Celsius instead of degrees Rankine. Also, in Equation 3-25,
PVA is determined in mm of Hg rather than psia (760 mm Hg = 14.7 psia).

        3. If the daily average liquid surface temperature, TLA, is unknown, it is calculated using the
following equation:

                               TLA = 0.44TAA + 0.56TB + 0.0079 a  I                         (3-26)


where:
        TLA =  daily average liquid surface temperature, °R
        TAA =  daily average ambient temperature, °R; see Note 4
         TB =  liquid bulk temperature, °R; see Note 5
         a =  tank paint solar absorptance, dimensionless; see Table 7.3-6
           I =  daily total solar insolation factor, Btu/(ft2 day); see Table 7.3-7

        If TLA is used to calculate PVA from Figures 7.3-13a, 7.3-13b, 7.3-14a, or 7.3-14b, TLA must be
converted from degrees Rankine  to degrees Fahrenheit (°F = °R - 460). If TLA is used to calculate PVA
from Equation 3-25, TLA must be converted from degrees Rankine to degrees Celsius (°C = [°R -
492J/1.8). Equation 3-26 should not be used to estimate liquid surface temperature from insulated tanks.
In the case of insulated tanks, the average liquid surface temperature should be based on liquid surface
temperature measurements from the tank.
                                               3-10

-------
       4. The daily average ambient temperature, TAA, is calculated using the following equation:


                                                                                          (3_27)
where:
        TAA =  daily average ambient temperature, °R
        TAX =  daily maximum ambient temperature, °R
        TAN =  daily minimum ambient temperature, °R

        Table 7.3-7 gives values of TAX and TAN for selected U.S. cities.

        5. The liquid bulk temperature, TB, is calculated using the following equation:

                                       TB = TAA + 6a  -1                                  (3-28)

where:
         TB =  liquid bulk temperature, °R
        TAA =  daily average ambient temperature, °R, as calculated in Note 4
         a  =  tank paint solar absorptance, dimensionless; see Table 7.3-6.

3.1.1.2  Working Loss

        The working loss, Lw, refers to the loss of stock vapors as a result of tank filling or emptying
operations. Fixed roof tank working losses can be estimated from:

                                Lw =O.OOWMVPVAQKNKP                           (3_29)

where:
        Lw =  working loss, Ib/yr
        Mv =  vapor molecular weight, Ib/lb-mole; see Note 1 to Equation 3-21
        PVA =  vapor pressure at daily average liquid surface temperature, psia; see Notes 1 and 2 to
               Equation 3-21
          Q =  annual net throughput (tank capacity [bbl] times annual turnover rate), bbl/yr
        KN =  working loss turnover (saturation) factor, dimensionless; see Figure 7.3-18
               for turnovers >36, KN = (180 + N)/6N
               for turnovers <36, KN = 1

        N = number of turnovers per year, dimensionless

                                              5.6140
                                         N =—T,	                                    (3-30)
                                                V LX
                                              3-11

-------
where:

       VLX = tank maximum liquid volume, ft3

                                      V^^tfH^                                  (3-31)
                                             4


where:
         D =  diameter, ft
       HLX =  maximum liquid height, ft
        KP =  working loss product factor, dimensionless
                      for crude oils KP = 0.75
                      for all other organic liquids, KP = 1

       Using the following  steps, Equation 3-29 can be simplified to combine all variables into one
equation.

Using Equation 3-21, the term "MvPvA" can be replaced with Equation 3-32.

                                     MVPVA=WVRTIA                                 (3-32)
Using a combination of Equation 3-30 and Equation 3-31, the term "Q" can be replaced with
Equation 3-33.
Assuming a standard value of Rto be 10.731 ft3 psia/(lb-mole °R), the result is Equation 3-34.
                       =   -             j TLANHLXD   KN KP Wv               (3-34)
                           5.614 J^      '            V4

                                             3-12

-------
       By assuming the temperature to be 60°F (520°R), and adding the vent setting correction factor,
KB, the result is Equation 3-35. The vent setting correction factor accounts for any reduction in emissions
due to the condensation of vapors prior to the opening of the vent. This correction factor will only affect
the calculation if the vent settings are greater than ±0.03 psig.
                               Lw  = N H^D  KN KP WVKB                          (3-35)
where:
        Lw =  working loss, Ib/yr
         N =  number of turnovers per year, (year)"1
        HLX =  maximum liquid height, ft
         D =  diameter, ft
        KN =  working loss turnover (saturation) factor, dimensionless; see Figure 7.3-18
                   for turnovers > 36, KN = (180 + N)/6N
                   for turnovers < 36, KN = 1
        KP =  working loss product factor, dimensionless
                   for crude oils Kp = 0.75
                   for all other organic liquids, KP = 1
        Wv =  vapor density, lb/ft3, see Equation 3-21
        KB =  vent setting correction factor, dimensionless
                   for open vents and for a vent setting range up to ± 0.03 psig, KB = 1

Vent Setting Correction Factor

        When the breather vent settings are greater than the typical values of ± 0.03 psig, and the
condition expressed in Equation 3-36 is  met, a vent setting correction factor, KB, must be determined
using Equation 3-37. This value of KB will be used in Equation 3-35 to calculate working losses.

        When:

                                             ,D +P,
                                              3-13

-------
       Then:
                                           PBP +P A
                                                                                           (3-37)
where:
         KB =  vent setting correction factor, dimensionless
         PI =  pressure of the vapor space at normal operating conditions, psig
               PI is an actual pressure reading (the gauge pressure). If the tank is held at
               atmospheric pressure (not under a vacuum  or held at a steady pressure) PI would
               beO.
         PA =  atmospheric pressure, psia
        KN =  working loss turnover (saturation) factor (dimensionless)
                      for turnovers > 36, KN = (180 + N)/6N
                      for turnovers < 36, KN = 1
        PVA =  vapor pressure at the daily average liquid surface temperature, psia; see Notes 1 and 2 to
               Equation 3-21
        PBP =  breather vent pressure setting, psig.
                                          3-5,13,15-17
3.1.2  Total Losses From Floating Roof Tanks

       Total floating roof tank emissions are the sum of rim seal, withdrawal, deck fitting, and deck
seam losses. The equations presented in this subsection apply only to floating roof tanks. The equations
are not intended to be used in the following applications:

       1. To estimate losses from unstable or boiling stocks or from mixtures of hydrocarbons or
petrochemicals for which the vapor pressure is not known or cannot readily be predicted;

       2. To estimate losses from closed internal or closed domed external floating roof tanks (tanks
vented only through a pressure/vacuum vent); or

       3. To estimate losses from tanks in which the materials used in the rim seal and/or deck fittings
are either deteriorated or significantly permeated by the stored liquid.

       This section contains equations for estimating emissions from floating roof tanks in two
situations: during normal operation, and during roof landings.
                                              3-14

-------
3.1.2.1 Normal Operation

       Total losses from floating roof tanks may be written as:

                                      LT = LR + LWD + LF + LD                               (3-38)

where:
         LT =  total loss, Ib/yr
         LR =  rim seal loss, Ib/yr; see Equation 3-39
       LWD =  withdrawal loss, Ib/yr; see Equation 3-41
         LF =  deck fitting loss, Ib/yr; see Equation 3-42
         LD =  deck seam loss (internal floating roof tanks only), Ib/yr; see Equation 3-46

       Loss factors may be estimated for deck fitting configurations that are not listed in Table 3-12, at
the zero miles-per-hour wind speed condition (IFRTs and CFRTs), from the following equation:

                                         Kfai = 0.27(,4fl)086

where:
       ATfai =  zero-wind-speed loss factor for a particular type of deck fitting, in pound-moles per year.
       Afl  =  liquid surface area within a particular type of deck fitting, in square inches. The liquid
               surface area is the area inside the deck fitting  well or leg sleeve, less any area occupied
               by an obstruction in the  deck fitting well or leg sleeve (such as a fixed-roof support
               column, unslotted guidepole, guidepole float, or deck support leg).

       The coefficient, 0.27, has units of pound-moles per (square inches)0 86-year, and the exponent,
0.86, is dimensionless.

       This equation is only applicable when the distance from the liquid surface  to the top of the deck
fitting well or leg sleeve is 12 inches or greater. Shorter deck fitting wells or leg sleeves may result in
higher loss rates. There are no similar algorithms available for estimating loss factors for shorter deck
fitting wells or leg sleeves.

       This equation is for an uncontrolled deck fitting. Effective deck fitting controls would be
expected to result in lower loss  factors than would be estimated by this equation, but there are no
algorithms available for estimating the effectiveness of deck fitting controls.

       This equation is for the zero miles-per-hour wind speed condition. There are no algorithms
available for estimating loss factors at non-zero wind speeds (EFRTs).
                                               3-15

-------
Rim Seal Loss - Rim seal loss from floating roof tanks can be estimated using the following equation:


                                  LR = (KRa + KRb vn)DP* Mv Kc                            (3-39)

where:
         LR =  rim seal loss, Ib/yr
        KRa =  zero wind speed rim seal loss factor, lb-mole/ft-yr; see Table 7.3-8
        KRb =  wind speed dependent rim seal loss factor, lb-mole/(mph)nft-yr; see Table 7.3-8
          v =  average ambient wind speed at tank site, mph; see Note 1
          n =  seal-related wind speed exponent, dimensionless; see Table 7.3-8
         P* =  vapor pressure function, dimensionless; see Note 2
                                   p* _ _ ±_A
,+(i-^r
   i    PJ
       where:

                      PVA = vapor pressure at daily average liquid surface temperature, psia;

       See Notes 1 and 2 to Equation 3-21 and Note 3 below
                      PA = atmospheric pressure, psia

         D = tank diameter, ft
        Mv = average vapor molecular weight, Ib/lb-mole; see Note 1 to Equation 3-21,
        KG = product factor;
                      KG = 0.4 for crude oils;
                      KG = 1 for all other organic liquids.
                                                                                         (3-40)
Notes:
        1. If the ambient wind speed at the tank site is not available, use wind speed data from the nearest
local weather station or values from Table 7.3-9. If the tank is an internal or domed external floating roof
tank, the value of v is zero.

        2. P can be calculated or read directly from Figure 7.3-19.
                                             3-16

-------
       3. The API recommends using the stock liquid temperature to calculate PVA for use in Equation 3-
40 in lieu of the liquid surface temperature. If the stock liquid temperature is unknown, API recommends
the following equations to estimate the stock temperature:
Tank Color
White
Aluminum
Gray
Black





Average Annual Stock
Temperature, Ts (°F)
TAA + 03
TAA + 2.5
TAA + 3.5
TAA + 5.0
       aTAA is the average annual ambient temperature in degrees Fahrenheit.

Withdrawal Loss - The withdrawal loss from floating roof storage tanks can be estimated using
Equation 3-41.
_  (0.943)QC,WL
         D
                                                     1 +
                                                         NcF
                                                          C-TC
                                                          D
(3-41)
where:
         LWD =  withdrawal loss, Ib/yr
           Q =  annual throughput (tank capacity [bbl] times annual turnover rate), bbl/yr
          Cs =  shell clingage factor, bbl/1,000 ft2; see Table 7.3-10
          WL =  average organic liquid density, Ib/gal; see Note 1
           D =  tank diameter, ft
        0.943 =  constant, 1,000 ft3-gal/bbl2
          NO =  number of fixed roof support columns, dimensionless; see Note 2
          Fc =  effective column diameter, ft (column perimeter [ft]/?i); see Note 3
Notes:
       1. A listing of the average organic liquid density for select petrochemicals is provided in
Tables 7.3-2 and 7.3-3. If WL is not known for gasoline, an average value of 6.1 Ib/gal can be assumed.

       2. For a self-supporting fixed roof or an external floating roof tank:

                     Nc = 0.

          For a column-supported fixed roof:

                     Nc = use tank-specific information or see Table 7.3-11.

       3. Use tank-specific effective column diameter or

                     FC = 1.1 for 9-inch by 7-inch built-up columns, 0.7 for 8-inch-diameter
                          pipe columns, and 1.0 if column construction details are not known
                                             3-17

-------
Deck Fitting Loss - Deck fitting losses from floating roof tanks can be estimated by the
following equation:

                                         LF = FF P*MVKC                                   (3-42)

where:
         LF =  the deck fitting loss, Ib/yr
         FF =  total deck fitting loss factor, Ib-mole/yr

                          FF = [(NFlKFl) + (NF2KF2)  + .- +(NFnfKFnf)]                     (3-43)


where:
        NF; =  number of deck fittings of a particular type (i = 0,l,2,...,nf), dimensionless
        KF =  deck fitting loss factor for a particular type fitting
               (i = 0,l,2,...,nf), Ib-mole/yr; see Equation 3-44
          nf =  total number of different types of fittings, dimensionless
               P , Mv, KG are as defined for Equation 3-39.

       The value of FF may be calculated by using actual tank-specific data for the number of each
fitting type (NF) and then multiplying by the fitting loss factor for each fitting (KF).

       The deck fitting loss factor, KF. for a particular type of fitting, can  be estimated by the following
equation:


                                    KF, = KFa, + KFb,(KvV)m'                               (3-44)
where:
        KF =  loss factor for a particular type of deck fitting, Ib-mole/yr
        KFa =  zero wind speed loss factor for a particular type of fitting, Ib-mole/yr
        KFb =  wind speed dependent loss factor for a particular type of fitting, lb-mole/(mph)m-yr
         ni; =  loss factor for a particular type of deck fitting, dimensionless
           i =  1, 2, ..., n, dimensionless
         Kv =  fitting wind speed correction factor, dimensionless; see below
          v =  average ambient wind speed, mph

        For external floating roof tanks, the fitting wind speed correction factor, Kv, is equal to 0.7. For
internal and domed external floating roof tanks, the value of v in Equation 3-44 is zero and the equation
becomes:

                                           Kn = KFai                                       (3-45)
                                               3-18

-------
Loss factors KFa, KFb, and m are provided in Table 7.3-12 for the most common deck fittings used on
floating roof tanks. These factors apply only to typical deck fitting conditions and when the average
ambient wind speed is below 15 miles per hour. Typical numbers of deck fittings for floating roof tanks
are presented in Tables 7.3-11, 7.3-12, 7.3-13, 7.3-14, and 7.3-15.

Deck Seam Loss - Neither welded deck internal floating roof tanks nor external floating roof tanks have
deck seam losses. Internal floating roof tanks with bolted decks may have deck seam losses. Deck seam
loss can be estimated by the following equation:

                                     LD = KDSDD2P*MVKC                                (3-46)


where:
        KD =  deck seam loss per unit seam length factor, Ib-mole/ft-yr
           =  0.0 for welded deck
           =  0.14 for bolted deck; see Note
         SD =  deck seam length factor,  ft/ft2
                A
               •"deck

              where:
                      LSeam = total length of deck seams, ft

                      Adeck = area of deck, ft2 =	

         D, P*, My, and Kc  are as defined for Equation 3-39.

       If the total length of the deck seam is not known, Table 7.3-16 can be used to determine SD. For a
deck constructed from continuous metal sheets with a 7-ft spacing between the seams, a value of 0.14
ft/ft2 can be used. A value of 0.33 ft/ft2 can be used for SD when a deck is constructed from rectangular
panels 5 ft by 7.5 ft. Where tank-specific data concerning width of deck sheets or size of deck panels are
unavailable, a default value for SD can be assigned. A value of 0.20 ft/ft2 can be assumed to represent the
most common bolted decks currently in use.

Note:  Recently vendors of bolted decks have been using various techniques, such  as gasketing
       the deck seams, in an effort to reduce deck seam losses. However, emission factors are
       not currently available in AP-42 that represent the emission reduction, if any, achieved
       by these techniques.  Some vendors have developed specific factors for their deck
       designs; however, use of these factors is not recommended until approval has been
       obtained from the governing regulatory agency or permitting authority.
                                             3-19

-------
3.1.2.2 Roof Landings21

       When using floating roof tanks, the roof floats on the surface of the liquid inside the tank and
reduces evaporative losses during normal operation. However, when the tank is emptied to the point that
the roof lands on deck legs, there is a period where the roof is not floating and other mechanisms must be
used to estimate emissions. These emissions continue until the tank is refilled to a sufficient level to again
float the roof. Therefore, these emission estimate calculations are applicable each time there is a landing
of the floating roof.

       This model does not address standing idle losses for partial days. It would be conservative (i.e.,
potentially overestimate emissions) to apply the model to episodes during which the floating roof remains
landed for less than a day.

       The total loss from floating roof tanks during a roof landing is the sum of the standing idle losses
and the filling losses. This relationship may be written in the form of an equation:

                                        LTL=LSL+LFL                                   (3-47)
where:
        LTL =  total losses during roof landing, Ib per landing episode
        LSL =  standing idle losses during roof landing, Ib per landing episode
        LFL =  filling losses during roof landing, Ib per landing episode

        The group of applicable equations to estimate the landing losses differs according to the type of
floating roof tank that is being used. The equations needed to estimate landing losses from internal
floating roof tanks are contained in Table 7.3-17; equations for external floating roof tanks are contained
in Table 7.3-18; and equations for drain-dry floating roof tanks are contained in Table 7.3-19. The
following sections explain these equations in more detail.

3.1.2.2.1  Standing Idle Losses

        After the floating roof is landed and the liquid level in the tank continues to drop, a vacuum is
created which could cause the floating roof to collapse. To prevent damage and to equalize the pressure, a
breather vent is actuated. Then, a vapor space is formed between the floating roof and the liquid. The
breather vent remains open until the roof is again floated, so whenever the roof is landed, vapor can be
lost through this vent. These losses are called "standing idle losses."

        The three different mechanisms that contribute to standing idle losses are (1) breathing losses
from vapor space, (2) wind losses, and (3) clingage losses. The specific  loss mechanism is dependent on
the type of floating roof tank.

        For internal floating roof tanks with nominally flat bottoms (including those built with a slight
upward cone), the breathing losses originate from a discernible level of liquid that remains in the tank at
all times due to the flatness of the tank bottom and the position of the withdrawal line (a liquid "heel").
The liquid evaporates into the vapor space and daily changes in ambient temperature cause the tank to
breathe in a manner similar to a fixed roof tank.
                                              3-20

-------
        For external floating roof tanks, which are not shielded from the surrounding atmosphere, the
wind can cause vapors to flow from beneath the floating roof. The higher the wind speeds, the more vapor
that can be expelled. These are known as wind losses.

        For tanks with a cone-down or shovel bottom, the floor of the tank is sloped to allow for more
thorough emptying of the tank contents, therefore, the amount of liquid differs significantly from tanks
with flat bottoms (see Figure 7.3-20). When the emptying operation drains the tank bottom, but leaves a
heel of liquid in or near the sump, the tank is considered to have a partial heel. A drain-dry condition is
attained only when all of the standing liquid has been removed, including  from the bottom of the sump.
However, due to sludge buildup and roughness of the inside of the tank, a small layer of liquid can remain
clinging to the sloped bottom of a drain-dry tank.  This layer of liquid will create vapor that can result in
clingage losses. The amount of vapor produced within a drain-dry tank is directly related to this clingage.
Clingage factors for various tank conditions are contained in Table 7.3-10.

Standing Idle Loss for Tanks with a Liquid Heel

        A constraint on the standing idle loss is added for floating roof tanks with a liquid heel in that the
total emissions cannot exceed the available stock liquid in the tank. This upper limit,  represented as
LsLmax, is a function of the  volume and density of the liquid inside the tank.

                     LSLmax = (area of tank) (height of liquid) (density of liquid)                 (3-48)


        Assuming that the  tank has a circular bottom and adding a volume conversion unit, the equation
can be simplified to Equation 3-49 and Equation 3-50.

                                                                                           (3-49)
                                            = 5.9 D2 hle Wt                                 (3-50)
 where:
      LsLmax =  limit on standing idle loss, Ib per landing episode
       7.48 =  volume conversion factor, gal/ft3
          D =  diameter of the tank, feet
         hie =  effective height of the stock liquid, feet
         Wi =  density of the liquid inside the tank, Ib/gal

Internal Floating Roof Tank with a Liquid Heel

       For internal floating roof tanks with liquid heels, the amount of "standing idle loss" depends on
the amount of vapor within the vapor space under the floating roof. Essentially, the mechanism is
identical to the breathing losses experienced with fixed roof tanks. The mechanism shown in Equation 3-
51 is identical to Equation 3-2.
                                              3-21

-------
                                    LSL=365VVWVKEKS                               (3-51)
 where:
        LSL =  annual breathing loss from standing storage during roof landing, Ib/yr
        365 =  number of days in a year, days/yr
        Vv =  volume of the vapor space, ft3
        Wv =  stock vapor density, lb/ft3
                                                                                           (3-52)
                                            v
                                                 RT

               Mv  =    stock vapor molecular weight, Ib/lb-mole
               P    =    true vapor pressure of the stock liquid, psia
               R    =    ideal gas constant, 10.731 (psia-ft3)/(lb-mole °R)
               T    =    temperature, °R
         KE =  vapor space expansion factor, dimensionless
         Ks =  saturation factor, dimensionless.

       Assuming that nd equals the number of days that the tank stands idle and substituting for the stock
vapor density according to Equation 3-52, the equation is further simplified to Equation 3-53.
                                  LSL = nd KE         Mv Ks                             (3-53)
        The term with the highest amount of uncertainty is the saturation of the vapor within the tank.
The factor, Ks, is estimated with the same method used to calculate the saturation factor for fixed roof
tanks in Equation 3-20. In order to establish limits on the value of Ks, the estimated factor is assumed to
be less than or equal to the saturation factor during filling (S). (For more information see Filling Losses.)
                                               3-22

-------
External Floating Roof Tank with a Liquid Heel

        For external floating roof tanks with a liquid heel, wind affects emission releases from the tanks.
As a starting point, begin with a basic equation based on rim-seal loss. The equation, shown as Equation
3-54, is equivalent to Equation 3-39.
 where:
        LRL =  annual rim seal loss during roof landing, Ib/yr
        KRa =  zero wind speed rim seal loss factor, Ib-mole/ft-yr
        KRb =  wind speed dependent rim seal loss factor, lb-mole/((mph)n-ft-yr))
          n =  seal-related wind speed loss exponent, dimensionless
               (KRa, KRb, and n are specific to a given configuration of rim seal)
          v =  average ambient wind speed, mph
          D =  tank diameter, ft
        Mv =  stock vapor molecular weight, Ib/lb-mole
         KG =  product factor, dimensionless
         P  =  a vapor pressure function, dimensionless
                                    P=-
                                                                                           (3-55)
               where:

                      PA =    atmospheric pressure, psia
                      P =     true vapor pressure of the stock liquid, psia.

       Assuming that the stock properties included in the vapor pressure function will adequately
account for differences in liquid product type, Kc is assumed to equal 1. Regardless of the type of rim seal
that is in use, it is effectively rendered a 'vapor-mounted' seal when the liquid level falls such that the rim
seal is no longer in contact with the liquid. The contribution of a secondary seal is neglected in that it is
offset by emissions through the deck fittings. The emissions are therefore based on the case of a welded
tank with an average-fitting vapor-mounted primary seal. According to Table 7.3-8, the values of Kra, Krb,
and n are 6.7, 0.2, and 3.0, respectively. The variables were substituted and the equation was converted
from annual emissions to daily emissions by dividing the equation by 365. A value of 10 mph is assigned
to the wind speed, so that estimated standing idle losses from an external floating roof tank will not be
less than for a typical internal floating roof tank. Lower values for the rim seal loss factors or the wind
speed should not be used. The equation can be simplified for daily emissions to Equation 3-56.
                                              3-23

-------
                                      LsLwind = 0.57 nd D P* Mv                              (3-56)

  where:
         LsLwind=  daily standing idle loss due to wind, Ib per day
             nd =  number of days that the tank is standing idle, days
             D =  tank diameter, ft
             P* =  a vapor pressure function, dimensionless
            Mv =  stock vapor molecular weight, Ib/lb-mole

         After the wind empties the vapor space above the remaining liquid heel, the liquid will continue
  to produce vapor.  Thus, this standing idle loss will occur every day that the tank stands idle. This
  equation is adequate at this time, but could be revised as additional testing is conducted and studied.

  Standing Idle Losses from Drain-Dry Tanks

         When a drain-dry tank has been emptied, the only stock liquid available inside the tank is a small
  amount that clings to the wetted surface of the tank interior (if a heel of liquid remains in or near a sump,
  then the tank should be evaluated as having a partial heel, and not as drain dry - see Figure 7.3-20). The
  slope prevents a significant amount of stock liquid from remaining in the tank so that evaporation is much
  lower than from tanks with liquid heels. Due to the limited amount of liquid clinging to the interior of the
  tank, as shown in Figure 7.3-20, it is assumed that vapors would not be replenished as readily as in tanks
  with a liquid heel. For this model, standing idle loss due to clingage is a one-time event rather than a daily
  event.

         The loss due to clingage is proportional to a clingage factor, which varies with the condition of
  the inside of the tank. A list of clingage factors are shown in Table 7.3-10. The factors are given in terms
  of barrels per thousand square feet. To convert the loss to pounds, the  density of the liquid and the area of
  the tank must be taken into account, as shown in Equation 3-57 (See NOTE).

                                       Lc  = 0.042Cs Wl (Area)                               (3-57)

  where:
             LC =  clingage loss from the drain-dry tank, Ib
         0.042 =  conversion factor, gal/bbl
             Cs=  clingage factor, bbl/1,000 ft2
             Wi =  density of the liquid, Ib/gal
          Area =  area of the tank bottom, ft2
NOTE:  Equation was corrected 8/2012
                                                                                            (3-58)
         Among the conditions shown in Table 7.3-10, the one that best approximates a sludge-lined tank
  bottom is gunite-lined. Assuming that gasoline is being stored in the tank, a clingage factor of 0.15 and
  the area term in Equation 3-58 were substituted into Equation 3-57, which simplifies to Equation 3-59.
                                                3-24

-------
                                     LSL = 0.0063 Wi    —                                (3-59)
       The clingage loss should be constrained by an upper limit equal to the filling loss for an internal
floating roof tank with a liquid heel. This is demonstrated in Equation 3-60.

                                                 (PV\
                                   LSL*- = 0.60  —£• \MV                               (3-60)
                                                 V K. 1 j
where:
        LsLmax =  maximum standing idle loss for drain-dry tanks due to clingage, Ib
          Wi =  density of the liquid inside the tank, Ib/gal
           D =  diameter of the tank, feet
            P =  true vapor pressure of the liquid inside the tank, psia
          Vv =  volume of the vapor space, ft3
           R =  ideal gas constant, 10.73 1 psia ft3 /lb-mole °R
            T =  average temperature of the vapor and liquid below the floating roof, °R  (= TAA)
          Mv =  stock vapor molecular weight, Ib/lb-mole

        Therefore, the standing idle loss for drain-dry tanks, shown in Equation 3-59, must be less than or
equal to Equation 3-393. This relationship is shown by Equation 3-361.
                                                                                           (3-61)
3.1.2.2.2 Filling Losses

       When a floating roof tank is refilled, there are additional emissions resulting from the roof being
landed. These losses are called "filling losses" and continue until the liquid reaches the level of the
floating roof.

       The first contributor to filling losses is called the "arrival" component. As liquid flows into the
tank, the vapor space between the liquid and the floating roof is decreased. The displaced vapors are
expelled through the breather vent. Once the roof is refloated on the liquid surface, the breather vent
closes.

       The second contributor to filling losses is called the "generated" component. As the incoming
liquid evaporates, additional vapors will be formed in the vapor space and will also be expelled through
the breather vent.

Internal Floating Roof Tank with a Liquid Heel

       For internal floating roof tanks with a liquid heel, the amount of vapor that is lost during filling is
directly related to the amount of vapor space and the saturation level of the vapor within the vapor space,
as shown in Equation 3-62.

                                              3-25

-------
            LFL = (yol of vapor space)(density ofvapor)(mol wt ofvapor)(satfactor)       (3-62)


       After substituting for the major terms in Equation 3-62, the equation can be simplified to
Equation 3-63.

                                            (PV^
                                      LFL=\—^\MVS                                 (3-63)
                                            V K 1 J
where:
        LFL =  filling loss during roof landing, Ib
          P =  true vapor pressure of the liquid within the tank, psia
        Vv =  volume of the vapor space, ft3
          R =  ideal gas constant, 10.731 psia-ft3/(lb-mole-°R)
          T =  average temperature of the vapor and liquid below the floating roof, °R
        Mv =  stock vapor molecular weight, Ib/lb-mole
          S =  filling saturation factor, dimension less (0.60 for a full liquid heel; 0.50 for a partial
               liquid heel).

       This equation accounts for the arrival losses and the generated losses. The main concern with this
equation is the estimation of the saturation factor. All other components are based on the ideal gas laws.
For consistency, an accepted value of 0.6, which is used elsewhere in Chapter 7, will be used for the case
of a full liquid heel. A value of 0.5 has been demonstrated for the case of a partial liquid heel.

External Floating Roof Tank with a Liquid Heel

       For external floating roof tanks with a liquid heel, the amount of vapor lost during filling will be
less than the amount for internal floating roof tanks because of wind effects. The "arrival" component will
be partially flushed out of the tank by the wind, so the preceding equation requires the addition of a
correction factor, Csf to the saturation factor as shown in Equation 3-64.


                                                                                           (3-64)
       The basic premise of the correction factor is that the vapors expelled by wind action will not be
present in the vapor space when the tank is refilled, so the amount of saturation is lowered. This is
demonstrated in Equation 3-65.

                    (one day of wind driven standing idle loss] - (one day without wind standing idle loss]
             Q=l--	—	        (3-65)
                                        one day without wind total loss
                                               3-26

-------
       The equation for the saturation factor can be simplified based on other equations contained in this
section as shown in Equation 3-66 and Equation 3-67.
                      csf=i-
( (Equation 2 -19) - (Equation 2-16)'
^ (Equation 2-16) + (Equation 2 - 26)
                                           (3-66)
                     crf=l~
                               0.57 nd DP  Mv-  nd KE
                                    d
                                                         RT
                               My K<
                                nd KE
        PVy
        RT
Mv KS\+\MV S
PVy
RT
                                                         (3-67)
where:
        Csf = filling saturation correction factor, dimensionless
         nd = number of days the tank stands idle with the floating roof landed, dimensionless
        KE = vapor space expansion factor, dimensionless
                                                 0.50 B P
                                        T  {   T(PA-P\
                                                                                        (3-68)
                A Tv  =    daily vapor temperature range, °R
                  T  =    average temperature of the vapor and liquid below the floating roof, °R
                  B  =    constant from the vapor pressure equation shown in Equation 3-24, °R
                           (If B is unknown, KE may be calculated from Equation 3-5, 3-6, or 3-7, as
                           appropriate, with the value of APB set equal to zero.)
                  P  =    true vapor pressure of the stock liquid, psia
                  PA  =    atmospheric pressure at the tank location, psia
        Vv = volume of the vapor space, ft3
                                        vv=-
                                             k, n D2
                                                         (3-69)
                 hv   =    height of the vapor space under the floating roof, ft
                 D   =    tank diameter, ft
         R = ideal gas constant, 10.731 psia ft3 / Ib-mole R
        Mv = stock vapor molecular weight, Ib/lb-mole
        Ks = standing idle saturation factor, dimensionless
          S = filling saturation factor, dimensionless
         P = vapor pressure function, dimensionless
        Wi = stock liquid density, Ib/gal
                                             3-27

-------
Drain-Dry Tanks

        The "arrival" component of filling losses for drain-dry tanks is completely covered by the
"clingage" loss. Once this initial loss occurs, there is no remaining liquid inside the tank. Therefore, there
is no vapor in the tank that could be expelled by the incoming liquid.

        However, the "generated" component remains a valid aspect of the model. Therefore, the filling
loss calculations for drain-dry tanks are identical to the filling loss calculations for internal floating roof
tanks with a liquid heel. Although the equations are the same, the  saturation factor will be lower for drain-
dry tanks due to the lack of an "arrival" component. AP-42 Chapter 5, Petroleum Industry, provides
emission factors for the loading of gasoline and crude oil into compartments according to the prior state
of the compartment. A  drain-dry tank would be most similar to a tank that was cleaned before filling
because a cleaned tank also lacks "arrival" losses. The emission factor (0.33 lb/1000 gallons) for this kind
of tank can be converted to a saturation factor by assuming a pressure of 8 psia (the same assumption
used in the formulation of the emission factor), and substituting the molecular weight of gasoline (64
Ib/lb-mole). The resulting saturation factor is 0.15. The equation is the same as Equation 3-70 with a
different assumed saturation factor.

                                             ( PV.. ^
                                                                                            (3-70)
where:
        LFL =  filling loss during roof landing, Ib
          P =  true vapor pressure of the liquid within the tank, psia
        Vv =  volume of the vapor space, ft3
          R =  ideal gas constant, 10.731 psia-ft3/(lb-mole-°R)
          T =  average temperature of the vapor and liquid below the floating roof, °R
        Mv =  stock vapor molecular weight, Ib/lb-mole
          S =  filling saturation factor, dimension less (0.15 for a drain-dry tank).
                                               3-28

-------
3.1.3  Variable Vapor Space Tanks  -

       Variable vapor space filling losses result when vapor is displaced by liquid during filling
operations. Since the variable vapor space tank has an expandable vapor storage capacity, this loss is not
as large as the filling loss associated with fixed roof tanks. Loss of vapor occurs only when the tank's
vapor storage capacity is exceeded. Equation 3-1 assumes that one-fourth of the expansion capacity is
available at the beginning of each transfer.
Variable vapor space system filling losses can be estimated from:

                                                                                   (3-71)
                                                                V2N2)]
where:
         Lv =  variable vapor space filling loss, lb/1,000 gal throughput
        Mv =  molecular weight of vapor in storage tank, Ib/lb-mole; see Note 1 to Equation 3-21
        PVA =  true vapor pressure at the daily average liquid surface temperature, psia; see Notes 1
               and 2 to Equation 3-21
         Vi =  volume of liquid pumped into system, throughput, bbl/yr
         V2 =  volume expansion capacity of system, bbl; see Note 1
         N2 =  number of transfers into system, dimensionless; see Note 2

Notes:

        1. V2 is the volume expansion capacity of the variable vapor space achieved by roof lifting or
diaphragm flexing.

        2. N2 is the number of transfers into the system during the time period that corresponds to a
throughput of Vi.

        The accuracy of Equation 3-1 is not documented. Special tank operating conditions may result in
actual losses significantly different from the estimates provided by Equation 3-1. For example, if one or
more tanks with interconnected vapor spaces are filled while others are emptied simultaneously, all or
part of the expelled vapors will be transferred to the tank, or tanks, being emptied. This is called balanced
pumping. Equation 3-1 does not account for balanced pumping, and will overestimate losses under this
condition. It should also  be noted that, although not developed for use with heavier petroleum liquids
such as kerosenes and fuel oils, the equation is recommended for use with heavier petroleum liquids in the
absence of better data.
                                              3-29

-------
3.1.4  Pressure Tanks -

       Losses occur during withdrawal and filling operations in low-pressure (2.5 to 15 psig) tanks when
atmospheric venting occurs. High-pressure tanks are considered closed systems, with virtually no
emissions. Vapor recovery systems are often found on low-pressure tanks. Fugitive losses are also
associated with pressure tanks and their equipment, but with proper system maintenance, these losses are
considered insignificant. No appropriate correlations are available to estimate vapor losses from pressure
tanks.

3.1.5  Variations Of Emission Estimation Procedures -

       All of the emission estimation procedures presented in Section 7.1.3 can be used to estimate
emissions for shorter time periods by manipulating the inputs to the equations for the time period in
question. For all of the emission estimation procedures, the daily average liquid surface temperature
should be based on the appropriate temperature and solar insolation data for the time period over which
the estimate is to be evaluated. The subsequent calculation of the vapor pressure should be based on the
corrected daily liquid surface temperature. For example, emission calculations for the month of June
would be based only on the meteorological data for June. It is important to note that a 3-month time frame
is recommended as the shortest time period for which emissions should be estimated.

       In addition to the temperature and vapor pressure corrections, the constant in the standing storage
loss equation for fixed roof tanks would need to be revised based on the actual time frame used.  The
constant, 365, is based on the number of days in a year. To change the equation for a different time
period, the constant should be changed to the appropriate number of days in the time period for which
emissions are being estimated. The only change  that would need to be  made to the working loss  equation
for fixed roof tanks would be to change the throughput per year to the throughput during the time period
for which emissions are being estimated.

       Other than changing the meteorological data and the vapor pressure data, the only changes
needed for the floating roof rim seal, deck fitting, and deck seam losses would be to modify the time
frame by dividing the individual losses by the appropriate number of days or months. The only change to
the withdrawal losses would be to change the throughput to the throughput for the time period for which
emissions are being estimated.

       Another variation that is frequently made to the emission estimation procedures is an adjustment
in the working or withdrawal loss equations if the tank is operated as a surge tank or constant level tank.
For constant level tanks or surge tanks where the throughput and turnovers are high but the liquid level in
the tank remains relatively constant, the actual throughput or turnovers should not be used in the working
loss or withdrawal loss equations. For these tanks, the turnovers should be estimated by determining the
average change in the liquid height. The average change in height should then be divided by the total
shell height. This adjusted turnover value should then be multiplied by the actual throughput to obtain the
net throughput for use in the loss equations. Alternatively, a default turnover rate of four could be used
based on data from these type tanks.
                                              3-30

-------
3.2 HAZARDOUS AIR POLLUTANTS (HAP's) SPECIATION METHODOLOGY


       In some cases it may be important to know the annual emission rate for a component (e.g., HAP)
of a stored liquid mixture. There are two basic approaches that can be used to estimate emissions for a
single component of a stored liquid mixture. One approach involves calculating the total losses based
upon the known physical properties of the mixture (i. e., gasoline) and then determining the individual
component losses by multiplying the total loss by the weight fraction of the desired component. The
second approach is similar to the first approach except that the mixture properties are unknown; therefore,
the mixture properties are first determined based on the composition of the liquid mixture.

       Case 1 — If the physical properties of the mixture are known (PVA, Mv, ML and WL), the total
losses from the tank should be estimated using the procedures described previously for the particular tank
type. The component losses are then determined from either Equation 3-72 or 3-73. For fixed roof tanks,
the emission rate for each individual component can be estimated by:

                                    LT, = (Zv,)(LT)                                       (3-72)


where:
        LT. =   emission rate of component i, Ib/yr
        Zv =   weight fraction  of component i in the vapor. Ib/lb
         LT =   total losses, Ib/yr

       For floating roof tanks, the emission rate for each individual component can be estimated by:

                         LT, = (ZV,)(LR +  LF + LD) + (ZL,)(LWD)                           (3-73)


where:
        LT. =   emission rate of component i, Ib/yr
        Zv =   weight fraction  of component i in the vapor, Ib/lb
         LR =   rim seal losses,  Ib/yr
         LF =   deck fitting losses, Ib/yr
         LD =   deck seam losses, Ib/yr
        ZL; =   weight fraction  of component i in the liquid, Ib/lb
       LWD =   withdrawal losses, Ib/yr

If Equation 3-72 is used in place of Equation  3-73 for floating roof tanks, the value obtained will be
approximately the same value as that achieved with Equation 3-73 because withdrawal losses are
typically minimal for floating roof tanks.
                                             3-31

-------
       In order to use Equations 4-1 and 4-2, the weight fraction of the desired component in the liquid
and vapor phase is needed. The liquid weight fraction of the desired component is typically known or can
be readily calculated for most mixtures. In order to calculate the weight fraction in the vapor phase,
Raoult's Law must first be used to determine the partial pressure of the component. The partial pressure of
the component can then be divided by the total vapor pressure of the mixture to determine the mole
fraction of the component in the vapor phase. Raoult's Law states that the mole fraction of the component
in the liquid (xj) multiplied by the vapor pressure of the pure component (at the daily average liquid
surface temperature) (P) is equal to the partial pressure (Pj) of that component:

                                      Pi = (P)(xi)                                         (3-74)


where:
         Pj =  partial pressure of component i, psia
          P =  vapor pressure of pure component i at the daily average liquid surface temperature, psia
         xi =  liquid mole fraction, Ib-mole/lb-mole

       The vapor pressure of each component can be calculated from Antoine's equation or found in
standard references, as shown in Section 7.1.3.1. In order to use Equation 3-74, the liquid mole fraction
must be determined  from the liquid weight fraction by:
                                                                                          (3-75)
where:
          x; =  liquid mole fraction of component i, Ib-mole/lb-mole
        ZL; =  weight fraction of component i in the liquid, Ib/lb
        ML =  molecular weight of liquid stock, Ib/lb-mole
         M; =  molecular weight of component i, Ib/lb-mole

       If the molecular weight of the liquid is not known, the liquid mole fraction can be determined by
assuming a total weight of the liquid mixture (see Example 1 in Section 7.1.5).

       The liquid mole fraction and the vapor pressure of the component at the daily average liquid
surface temperature can then be substituted into Equation 3-74 to obtain the partial pressure of the
component. The vapor mole fraction of the component can be determined from the following equation:


                                       *=:?-                                          (3-76)
                                            PvA


where:
          y; =  vapor mole fraction of component i, Ib-mole/lb-mole
          Pj =  partial pressure of component i, psia
        PVA =  total vapor pressure of liquid mixture, psia
                                              3-32

-------
       The weight fractions in the vapor phase are calculated from the mole fractions in the vapor phase.


                                      ZVi=^                                         (3-77)
                                            MV


where:
        ZV; =  vapor weight fraction of component i, Ib/lb
         y; =  vapor mole fraction of component i, Ib-mole/lb-mole
         M; =  molecular weight of component i, Ib/lb-mole
        Mv =  molecular weight of vapor stock, Ib/lb-mole

       The liquid and vapor weight fractions of each desired component and the total losses can be
substituted into either Equations 4-1 or 4-2 to estimate the individual component losses.

       Case 2 — For cases where the mixture properties are unknown but the composition of the liquid is
known (i. e., nonpetroleum organic mixtures), the equations presented above can be used to obtain a
reasonable estimate of the physical properties of the mixture. For nonaqueous organic mixtures,
Equation 3-74 can be used to determine the partial pressure of each component. If Equation 3-75 is used
to determine the liquid mole fractions, the molecular weight of the liquid stock must be known. If the
molecular weight of the liquid stock is unknown, then the liquid mole fractions can be determined by
assuming a weight basis and calculating  the number of moles (see Example  1 in Section 7.1.5). The
partial pressure of each component can then be determined from Equation 3-74.

       For special cases, such as wastewater, where the liquid mixture is a dilute aqueous solution,
Henry's Law should be used instead of Raoult's Law in calculating total losses. Henry's Law states that
the mole fraction of the component in the liquid phase multiplied by the Henry's Law constant for the
component in the mixture is equal to the partial pressure (PO for that component. For wastewater, Henry's
Law constants are typically provided in the form of atm-m3/g-mole.

Therefore, the appropriate form of Henry's Law equation is:

                                          Pi = (HA)  (CO                                    (3-78)
where:
         P; =  partial pressure of component i, atm
        HA =  Henry's Law constant for component i, atm-m3/g-mole
         Q =  concentration of component i in the wastewater, g-mole/m3; see Note

       Section 4.3 of AP-42 presents Henry's Law constants for selected organic liquids. The partial
pressure calculated from Equation 4-7 will need to be converted from atmospheres to psia
(1 atm = 14.7 psia).
Note:  Typically wastewater concentrations are given in mg/liter, which is equivalent to g/m3. To
convert the concentrations to g-mole/m3 divide the concentration by the molecular weight of the
component.
                                              3-33

-------
       The total vapor pressure of the mixture can be calculated from the sum of the partial pressures:

                                           PVA = I Pi                                     (3-79)
where:
        PVA =  vapor pressure at daily average liquid surface temperature, psia
         P; =  partial pressure of component i, psia

        This procedure can be used to determine the vapor pressure at any temperature. After computing
the total vapor pressure, the mole fractions in the vapor phase are calculated using Equation 3-76. The
vapor mole fractions are used to calculate the molecular weight of the vapor, Mv. The molecular weight
of the vapor can be calculated by:

                                          Mv = I MlYl                                     (3-80)

where:
        Mv =  molecular weight of the vapor, Ib/lb-mole
         M; =  molecular weight of component i, Ib/lb-mole
         y; =  vapor mole fraction of component i, Ib-mole/lb-mole

        Another variable that may need to be calculated before estimating the total losses, if it is not
available in a standard reference, is the density of the liquid, WL. If the density of the liquid is unknown,
it can be estimated based on the liquid weight fractions of each component (see Section 7.1.5,
Example 3).

        All of the mixture properties are now known (PVA, Mv, and WL). These values can now be used
with the emission estimation procedures outlined in Section 7.1.3 to estimate total losses. After
calculating the total losses, the component losses can be calculated by using either Equations 3-72 or 3-
73. Prior to calculating component losses, Equation 3-77 must be used to determine the vapor weight
fractions of each component.
                                              3-34

-------
I—   O.S
                                                                        1*0
                                                                       130
                                                                       120
  •  a
   10
   ti
   12
   13
   14
   15
                                  — 2
                                  — s
                               1
I—!5
                                                                       110
                                                                       100 —=
                                                                        so —=
                                                                        80
                                                                               i
                                                                               I
                                                                       50
                                      40
                                      30
                                     20 —E
   20
                                                                       to —=
                                                                        0  —3
 Figure 3-1. True vapor pressure of crude oils with a Reid vapor pressure
                    of 2 to 15 pounds per square inch.4
                                   3-35

-------
             — 0.20


             — 0.30


             — 040

                o.so

                0.60

                0.70
                0.80
                090
                1.00
             —  1.50



             —  2.00


                 250


                 3.00

                 3.50

             —  4.00



             —  5.00^


                 6.00


                 7.00


                 8.00

                 9.00

             — 10.0

             — 11.0

               -12.0
               -13.0

               - 14.0
               -15.0
               -16.0
               - 17.0
               • 18,0
               -190
               -20.0
               -21.0
               -  22.0
               - 23 0
               -240
                                                                      120 -,
                                                                      no-;
                                                                      100 —
                                                                                             90-
                                                                       »-:
                                                                      70-
                                                                       60 —
                                                                       50-
Notes:
1.5 = slope of the ASTM distillation curve at 10 percent evaporated, in degree!
       Fahrenheit per percent
     - [(°F at 15 percent) -  (T at 5 percsm)]/(10 percent).
In the absence of distillation data, the following average values of S may be used:
         Motor  gasoline—3.0.
         Aviation gasoline—2.0.
         Light naphtha (RVP of 9-14 pounds per square inch)—3.5,
         Naphtha (RVP of 2-8 pounds per square inch)—2.5.
2. The broken line illustrates a sample problem for a gasoline stock (5 = 3.0) with a
Reid vapor pressure of 10 pounds per square inch and a stock temperature of 62.ST.
                                                                       30—
                                                                       20-^
                                                                       10—
                                                                        0^1
Figure 3.-2. True vapor pressure of refined petroleum stocks with a Reid vapor pressure of
                                    1 to 20 pounds per square inch.4
                                                    3-36

-------
             P = exp
         2,799
       T +459.6
        -2.227
   loglo(RVP)-
  7,261
T +459.6
    -12.82
Where:
      P = stock true vapor pressure, in pounds per square inch absolute.
      T = stock temperature, in degrees Fahrenheit.
   RVP = Reid vapor pressure, in pounds per square inch.

Note:This equation was derived from a regression analysis of points read off Figure 7. l-13a over the full range of
         Reid vapor pressures, slopes of the ASTM distillation curve at 10 percent evaporated, and stock
         temperatures. In general, the equation yields P values that are within +0.05 pound per square inch
         absolute of the values obtained directly from the nomograph.
                      Figure 3-lb. Equation for true vapor pressure of crude oils
                    with a Reid vapor pressure of 2 to 15 pounds per square inch.4
         P = exp
0.7553-
  413.0
T +459.6
S05log10(RVP)-
1.854-
  1,042
T +459.6
                                                                                         -,0.5
                          2,416
                       T +459.6
              -2.013
Where:
            loglo(RVP)-
                8,742
              T +459.6
        + 15.64
       P = stock true vapor pressure, in pounds per square inch absolute.
       T = stock temperature, in degrees Fahrenheit.
    RVP = Reid vapor pressure, in pounds per square inch.
       S = slope of the ASTM distillation curve at 10 percent evaporated, in degrees Fahrenheit per percent.

Note:    This equation was derived from a regression analysis of points read off Figure 7.1-14a over the full range of Reid
        vapor pressures, slopes of the ASTM distillation curve at 10 percent evaporated, and stock temperatures. In
        general, the equation yields P values that are within +0.05 pound per square inch absolute of the values obtained
        directly from the nomograph.
               Figure 3-2b. Equation for true vapor pressure of refined petroleum stocks
                    with a Reid vapor pressure of 1 to 20 pounds per square inch.4
                                                 3-37

-------
      20
      18
      16
  €  14
 a
      12
      10
                                             ASTM D86 distillation slope at
                                             10 volume percent evaporated
       0.1
                 0.2
0.4    0,6 0.6 1        2        4     6   8 10

      Reid vapor pressure, RVP (psi)
Figure 3-3.  Vapor pressure function coefficient (A) of refined petroleum stocks with a
              Reid vapor pressure of 1  to 20 psi, extrapolated to 0.1 psi.1
14.UUU
12,000
QC
OQ
§ 10,000
E
| 8.000
1
5
6,000
4,000
0



""•..
_"""-•«*..
" """* —*."*****
""-- :








^-^




'



"•--
— ^.,
•»^'







•. j^^

-^
~ —








-i!
-,

t




""*(
'





t



ASTM D86 distillation slope at
10 volume percent evaporated
S (T/vol.%)

}-^
•^^. 1
$4^
5

1 0.2 0.4 0.6 0.8 1




^~^^



^
^T-
^
|



V""
§




-^
•~~^
~~-,




^^

^^








"^

2 4 6 8 10 2C
                                Retd vapor pressure, RVP (psi)


Figure 3-4.  Vapor pressure function coefficient (B) of refined petroleum stocks with a
              Reid vapor pressure of 1  to 20 psi, extrapolated to 0.1  psi.1
                                           3-38

-------
                     A= 15.64 - 1.854 S05 - (0.8742-0.3280 S05)ln(RVP)
                     B = 8,742 - 1,042 S05 - (1,049-179.4 S05)ln(RVP)
                     where:
                             RVP = stock Reid vapor pressure, in pounds per square inch
                              In = natural logarithm function
                               S = stock ASTM-D86 distillation slope at 10 volume percent
                     	evaporation (°F/vol %)	
Figure 3-5. Equations to determine vapor pressure constants A and B for refined petroleum stocks.
                                             3-39

-------
.1
   c
   3


   =5
        18
        16
        14
12
         10
          8
                        1	1	1   I 1 I  I
          b,1       0.2       0.4   0.6  0.8 1        2        4    6   8  10    15



                                Reid vapor pressure, RVP (psi)
Figure 3-6.  Vapor pressure function coefficient (A) of crude oil stocks with a

          Reid vapor pressure of 2 to 15 psi, extrapolated to 0.1  psi.1
12,000

10,000
F
Vapor pressure coefficient. B
_M *• P> ?>


*""--^










^-^




















-»-









*»










^^









•^



	 1 	






\


___









_i_







^»,

_L







\^


t 02 ' 4 0.6 0.8 1 2 4 b b IU 1!
Reid vapor pressure, ftVP (psi)
   Figure 3-7.  Vapor pressure function coefficient (B) of crude oil stocks with

          a Reid vapor pressure of 2 to 15 psi, extrapolated to 0.1 psi1
                                         3-40

-------
                           A = 12.82 - 0.9672 In (RVP)

                           B = 7,261- 1,216 In (RVP)

  where:

                      RVP = Reid vapor pressure, psi
                        In = natural logarithm function
Figure 3-8. Equations to determine vapor pressure Constants A and B for crude oil stocks.'



           Daily Maximum and Minimum Liquid Surface Temperature, (°R)

                TLX = TLA+ 0.25 ATV

                TLN = TLA-0.25 ATV
                where:
                            = daily maximum liquid surface temperature, °R
                         LA    is as defined in Note 3 to Equation 1-21
                        A TV  is as defined in Note 1 to Equation 1-7
                       TLN  = daily minimum liquid surface temperature, °R
TLA
A
T
 Figure 3-9. Equations for the daily maximum and minimum liquid surface temperatures.
                                      3-41

-------
 u.
 1
1.0

O.I

9.1

0.4

9.1

  0
                100
                  200
300     400
             PER YEAE - AMMUAL THRCMMHPOT
                          TANE CAPACITY
    Note: For 36 lu»ov<5ri p« yea* or IBM. 1C,  -1.0
Figure 3-10. Turnover Factor (KN) for fixed roof tanks.
                       3-42

-------
0.9
0.8
0.7
0.6
O.S
0.4
0.3
0.2
0.1
0.09
0.08
0.07
0.06
0.05
0.04
0.03
n QS
0.01
„
_
-
-
-
-
=
=
-
-
—

-



I/
= /
~)














/
/


1












/
/




1










/
/






1








x
1
1






,






1








/








{1
1
















I
+ (1 -
1







X
X"







'IP.
- (PIP.
\







/









))">*
1






^
/









1






X










1





xl
















/












-.1
/-

/,
/2


=
i
^
-
-
-
_
-
_
_
1
=
—
1 234 S 6 78 9 10 11 12 13 14 1!
1.0

0.7
0.6
O.S
0.4
0.3
0.2
0.1
0.09
0.08
O.O7
0.06
O.OS
0.04
0.03
0.02
0.01
i
                           Stock tnw vapor praaur*. f (pounds ptr aquam inch abMtot*)
Notes:
1. Broken line illustrates sample problem for P = 5,4 pounds per square inch absolute.
2. Curve ia for atmospheric pressure, Pv equal to 14.7 pounds per square inch absolute.
                              Figure 3-11. Vapor pressure function.4
                                                3-43

-------
                          Table 3-1. LIST OF ABBREVIATIONS USED IN THE TANK EQUATIONS
Variable Description
Variable Description
Variable Description
a      tank paint solar absorptance,
            dimensionless
n      constant, (3.14159)
A      constant in vapor pressure equation,
            dimensionless
Adeck    area of deck, ft2
Afi      liquid surface area within a
            particular type of deck fitting,
            in2
B      constant in vapor pressure equation,
            °Ror°C
C      constant in vapor pressure equation,
            °Ror°C
Cs      shell clingage factor, bbl/1,000 ft2
Csf      filling saturation factor
D      tank diameter, ft
DE      effective tank diameter, ft
Fc      effective column diameter,  ft
FF      total deck fitting loss factor,
            Ib-mole/yr
FR      rim deck loss factor, Ib-mole/ft-yr
HL      liquid height, ft
HLX    maximum liquid height, ft
HR      tank roof height, ft
HRO    roof outage, ft
Hs      tank shell height, ft
HVO    vapor space outage, ft
i        1,2,	n, dimensionless
I        daily total solar insolation factor,
            Btu/ft2-d
Kc      product factor for floating roof
            tanks, dimensionless
KD      deck seam loss per unit seam length
            factor, Ib-mole/ft-yr
KE      vapor space expansion factor,
            dimensionless
Kfai     zero wind speed loss factor
KF.     loss factor for a particular type of
            deck fitting, Ib-mole/yr
KN     turnover factor, dimensionless
KP     working loss product factor for fixed
            roof tanks, dimensionless
KRa     zero wind speed rim seal loss factor,
            Ib-mole/ft-yr
KRb     wind speed dependent rim seal loss
            factor, lb-mole/ (mph)nft-yr
Ks     vented vapor saturation factor,
            dimensionless
Kv     fitting wind speed correction factor,
            dimensionless
L       length of tank, ft
Lc     clingage factor for drain dry tanks
LD     deck seam loss, Ib/yr
LF     deck fitting loss, Ib/yr
LKL     filling loss during roof landing,
            Ib/landing event
LR     rim seal loss, Ib/yr
LRL     rim seal loss during roof landing,
            Ib/landing event
Ls     standing storage losses, Ib/yr
Lseam    total length of deck seam, ft
LSL     standing loss during roof landing,
            Ib/landing event
LT     total losses, Ib/yr
LT.     emission rate of component i, Ib/yr
LTL     total loss during roof landing,
            Ib/landing event
Lv     variable vapor space filling loss,
            lb/1,000 gal throughput
Lw     working losses, Ib/yr
LWD    withdrawal loss, Ib/yr
m1     loss factor for a particular type of
            deck fitting, dimensionless
M;      molecular weight of component i,
            Ib/lb-mole
ML     molecular weight of liquid mixture,
            Ib/lb-mole
Mv     vapor molecular weight, Ib/lb-mole
N      number of turnovers per year,
            dimensionless
n       seal-related wind speed exponent,
            dimensionless
N2      number of transfers into system,
            dimensionless
Nc      number of columns
Nc      number of columns, dimen-sionless
Nd      number of drains
nf      total number of different types of
            fittings, dimensionless
NFa.     zero wind speed loss factor for a
            particular type of deck fitting,
            Ib-mole/yr
Npb     wind speed dependent loss factor for
            a particular type of fitting,
            lb-mole/ mphm-yr
NF.     number of deck fittings of a
            particular type, dimensionless
NI      number of deck legs
NTOTAL  total number of moles in mixture, lb-
            mole
Nvb     number of vacuum breakers
P       true vapor pressure of component i,
            psia
P*      vapor pressure function,
            dimensionless
PA      atmospheric pressure, psi
A PB   breather vent pressure setting range,
            psig
PBp     breather vent pressure setting, psig
PBV     breather vent vacuum setting, psig
                                                              5-44

-------
                                   Table 3-1  (cont).
Variable Description
Variable Description
PI      gauge pressure within the vapor
            space, psig
Pi      partial pressure of component i, psia
A Pv   daily vapor pressure range, psi
PVA     vapor pressure at daily average
            liquid surface temperature, psia
PVN     vapor pressure at the daily minimum
            liquid surface temperature, psia
PVX     vapor pressure at the daily
            maximum liquid surface
            temperature, psia
Q      annual net throughput, bbl/yr
R      ideal gas constant,
            (10.731 psia-ft3/lb-mole-°R)
RR      tank dome roof radius, ft
Rs      tank shell radius, ft
SD      deck seam length factor, ft/ft2
SR      tank cone roof slope, ft/ft
A TA   daily ambient temperature range, °R
TAA     daily average ambient temperature,
            °R
TAN     daily minimum ambient
            temperature, °R
TAX     daily maximum ambient
            temperature, °R
TB      liquid bulk temperature, °R
TLA     daily average liquid surface
            temperature, °R
A TV   daily vapor temperature range, °R
v       average wind speed, mph
Vi      volume of liquid pumped into
            system, bbl/yr
V2      volume expansion capacity, bbl
VLX     tank maximum liquid volume, ft3
Vv      vapor space volume, ft3
Wi      liquid density of component i, lb/ft3
WL     average organic liquid density,
            Ib/gal
Wv     vapor density, lb/ft
x;       liquid mole fraction of component i,
            Ib-mole/lb-mole
yi       vapor mole fraction of component i,
            Ib-mole/lb-mole
ZL      liquid weight fraction of
            component i, Ib/lb
Zv      vapor weight fraction of
            component i, Ib/lb
                                         3-45

-------
    TABLE 3-2. PROPERTIES (Mv, Wvc, WL, Pv) OF SELECTED PETROLEUM LIQUIDS3
Petroleum Liquid
Crude oil RVP 5
Distillate fuel oil
No. 2
Gasoline RVP 7
Gasoline RVP 7.8
Gasoline RVP 8.3
Gasoline RVP 10
Gasoline RVP
11.5
Gasoline RVP 13
Gasoline RVP
13.5
Gasoline RVP
15.0
Jet kerosene
Jet naphtha (JP-4)
Residual oil No. 6

Molecular r-/"1"*"
-.,.,, Density
W^tat At60°F,
60 **> WL 40°F 50°F (
MV /ii ; IN
(lb/lb-mole) (lb/gal)
50 7.1 1.8 2.3
130 7.1 0.0031 0.0045
68 5.6 2.3 2.9
68 5.6 2.5929 3.2079
68 5.6 2.7888 3.444
66 5.6 3.4 4.2
65 5.6 4.087 4.9997
62 5.6 4.7 5.7
62 5.6 4.932 6.0054
60 5.6 5.5802 6.774
130 7.0 0.0041 0.0060
80 6.4 0.8 1.0
True Vapor Pressure, PVA (psi)
50°F 70°F SOT 90°F 100°F
2.8 3.4 4.0 4.8 5.7
0.0065 0.0090 0.012 0.016 0.022
3.5 4.3 5.2 6.2 7.4
3.9363 4.793 5.7937 6.9552 8.2952
4.2188 5.1284 6.1891 7.4184 8.8344
5.2 6.2 7.4 8.8 10.5
6.069 7.3132 8.7519 10.4053 12.2949
6.9 8.3 9.9 11.7 13.8
7.2573 8.7076 10.3774 12.2888 14.4646
8.1621 9.7656 11.6067 13.7085 16.0948
0.0085 0.011 0.015 0.021 0.029
1.3 1.6 1.9 2.4 2.7
190 7.9 0.00002 0.00003 0.00004 0.00006 0.00009 0.00013 0.00019
References 10 and 11
                                   3-46

-------
Table 3-3. PHYSICAL PROPERTIES OF SELECTED PETROCHEMICALS3
Name
Acetone
Acetonitrile
Acrylonitrile
Allyl alcohol
Allyl chloride
Ammonium hydroxide
(28.8% solution)
Benzene
wo-Butyl alcohol
ferf-Butyl alcoholv
w-Butyl chloride
Carbon disulfide
Carbon tetrachloride
Chloroform
Chloroprene
Cyclohexane
Cyclopentane
1 , 1 -Dichloroethane
1 ,2-Dichloroethane
c«-l,2-Dichloro-
ethylene
trans- 1 ,2-Dichloro-
ethylene
Diethylamine
Diethyl ether
Di-wo-propyl ether
1 ,4-Dioxane
Dipropyl ether
Ethyl acetate
Ethyl aery late
Ethyl alcohol
Formula
CH3COCH3
CH3CN
CH2:CHCN
CH2:CHCH2OH
CH2:CHCH2C1
NH4OH-H2O
C6H6
(CH3)2CHCH2OH
(CH3)3COH
CH3CH2CH2CH2C1
CS2
CC14
CHC13
CH2:CC1CH:CH2
C6H12
CsHio
CH3CHC12
CH2C1CH2C1
CHC1:CHC1
CHC1:CHC1
(C2H5)2NH
C2H5OC2H5
(CH3)2CHOCH(CH3)2
OCH2CH2OCH2CH2
CH3CH2CH2OCH2CH2CH3
C2H5OOCCH3
C2H5OOCCH:CH2
C2H5OH
Molecular
Weight
58.08
41.05
53.06
58.08
76.53
35.05
78.11
74.12
74.12
92.57
76.13
153.84
119.39
88.54
84.16
70.13
98.97
98.97
96.95
96.95
73.14
74.12
102.17
88.10
102.17
88.10
100.11
46.07
Boiling
Point At
1 Atmosphere
(°F)
133.0
178.9
173.5
206.6
113.2
83.0
176.2
227.1
180.5
172.0
115.3
170.2
142.7
138.9
177.3
120.7
135.1
182.5
140.2
119.1
131.9
94.3
153.5
214.7
195.8
170.9
211.8
173.1
Liquid
Density At
60°F(lb/gal)
6.628
6.558
6.758
7.125
7.864
7.481
7.365
6.712
6.595
7.430
10.588
13.366
12.488
8.046
6.522
6.248
9.861
10.500
10.763
10.524
5.906
5.988
6.075
8.659
6.260
7.551
7.750
6.610
Vapor Pressure (psia) At
40 °F
1.682
0.638
0.812
0.135
2.998
5.130
0.638
0.058
0.174
0.715
3.036
0.793
1.528
1.760
0.677
2.514
1.682
0.561
1.450
2.552
1.644
4.215
1.199
0.232
0.425
0.580
0.213
0.193
SOT
2.185
0.831
0.967
0.193
3.772
6.630
0.870
0.097
0.290
1.006
3.867
1.064
1.934
2.320
0.928
3.287
2.243
0.773
2.011
3.384
1.992
5.666
1.586
0.329
0.619
0.831
0.290
0.406
60 °F
2.862
1.083
1.373
0.261
4.797
8.480
1.160
0.135
0.425
1.320
4.834
1.412
2.475
2.901
1.218
4.177
2.901
1.025
2.668
4.351
2.862
7.019
2.127
0.425
0.831
1.102
0.425
0.619
70 °F
3.713
1.412
1.779
0.387
6.015
10.760
1.508
0.193
0.638
1.740
6.014
1.798
3.191
3.655
1.605
5.240
3.771
1.431
3.461
5.530
3.867
8.702
2.746
0.619
1.102
1.489
0.599
0.870
80 °F
4.699
1.876
2.378
0.522
7.447
13.520
1.972
0.271
0.909
2.185
7.387
2.301
4.061
4.563
2.069
6.517
4.738
1.740
4.409
6.807
4.892
10.442
3.481
0.831
1.431
1.934
0.831
1.218
90 °F
5.917
2.456
3.133
0.716
9.110
16.760
2.610
0.387
1.238
2.684
9.185
2.997
5.163
5.685
2.610
8.063
5.840
2.243
5.646
8.315
6.130
13.342
4.254
1.141
1.876
2.514
1.122
1.682
100°F
7.251
3.133
4.022
1.006
11.025
20.680
3.287
0.541
1.702
3.481
11.215
3.771
6.342
6.981
3.249
9.668
7.193
2.804
6.807
10.016
7.541
Boils
5.298
1.508
2.320
3.191
1.470
2.320
                          5-47

-------
Table 3-3. (con't).
Name
Freon 1 1
w-Heptane
w-Hexane
Hydrogen cyanide
Isopentane
Isoprene
Isopropyl alcohol
Methacry lonitrile
Methyl acetate
Methyl aery late
Methyl alcohol
Methylcyclohexane
Methylcyclopentane
Methylene chloride
Methyl ethyl ketone
Methyl methacrylate
Methyl propyl ether
Nitromethane
w-Pentane
w-Propylamine
1,1,1 -Trichloroethane
Trichloroethylene
2,2,4-trimethyl pentane
(isooctane)
Toluene
Vinyl acetate
Vinylidene chloride
Formula
CC13F
CH3(CH2)5CH3
CH3(CH2)4CH3
HCN
(CH3)2CHCH2CH3
(CH2):C(CH3)CH:CH2
(CH3)2-CHOH
CH2:CH(CH3)CN
CH3OOCCH3
CH3OOCCH:CH2
CH3OH
CHs-CsHu
CH3C5H9
CH2C12
CH3COC2H5
CH3OOC(CH3):CH2
CH3OC3H7
CH3N02
CH3(CH2)3CH3
C3H7NH2
CH3CC13
CHC1:CC12
(CH2)3CCH2CH(CH3)2
CH3-C6H5
CH2:CHOOCCH3
CH2:CC12
Molecular
Weight
137.38
100.20
86.17
27.03
72.15
68.11
60.09
67.09
74.08
86.09
32.04
98.18
84.16
84.94
72.10
100.11
74.12
61.04
72.15
59.11
133.42
131.40
114.23
92.13
86.09
96.5
Boiling
Point At
1 Atmosphere
(°F)
75.4
209.2
155.7
78.3
82.1
93.5
180.1
194.5
134.8
176.9
148.4
213.7
161.3
104.2
175.3
212.0
102.1
214.2
96.9
119.7
165.2
188.6
210.6
231.1
162.5
89.1
Liquid
Density At
60 °F (Pounds
Per Gallon)
12.480
5.727
5.527
5.772
5.199
5.707
6.573
6.738
7.831
7.996
6.630
6.441
6.274
11.122
6.747
7.909
6.166
9.538
5.253
6.030
11.216
12.272
5.76
7.261
7.817
10.383
Vapor Pressure (Pounds Per Square Inch Absolute) At
40 °F
7.032
0.290
1.102
6.284
5.878
4.757
0.213
0.483
1.489
0.599
0.735
0.309
0.909
3.094
0.715
0.116
3.674
0.213
4.293
2.456
0.909
0.503

0.174
0.735
4.990
50°F
8.804
0.406
1.450
7.831
7.889
6.130
0.329
0.657
2.011
0.773
1.006
0.425
1.160
4.254
0.928
0.213
4.738
0.251
5.454
3.191
1.218
0.677

0.213
0.986
6.344
60°F
10.900
0.541
1.876
9.514
10.005
7.677
0.483
0.870
2.746
1.025
1.412
0.541
1.644
5.434
1.199
0.348
6.091
0.348
6.828
4.157
1.586
0.889
0.596
0.309
1.296
7.930
70 °F
13.40
0.735
2.436
11.853
12.530
9.668
0.677
1.160
3.693
1.354
1.953
0.735
2.224
6.787
1.489
0.541
7.058
0.503
8.433
5.250
2.030
1.180

0.425
1.721
9.806
80 °F
16.31
0.967
3.055
15.392
15.334
11.699
0.928
1.470
4.699
1.798
2.610
0.986
2.862
8.702
2.069
0.773
9.417
0.715
10.445
6.536
2.610
1.508

0.580
2.262
11.799
90 °F
19.69
1.238
3.906
18.563
18.370
14.503
1.296
1.934
5.762
2.398
3.461
1.315
3.616
10.329
2.668
1.064
11.602
1.006
12.959
8.044
3.307
2.030

0.773
3.113
15.280
100°F
23.60
1.586
4.892
22.237
21.657
17.113
1.779
2.456
6.961
3.055
4.525
1.721
4.544
13.342
3.345
1.373
13.729
1.334
15.474
9.572
4.199
2.610

1.006
4.022
23.210
       5-48

-------
                   TABLE 3-4. ASTM DISTILLATION SLOPE FOR
                    SELECTED REFINED PETROLEUM STOCKS
Refined petroleum stock
Aviation gasoline
Naphtha
Motor gasoline
Light naphtha
Reid vapor pressure (RVP),
psia
ND
3-45
ND
9-14
ASTM distillation slope at
10 volume percent
evaporated,(°F/vol%)
2.0
2.5
3.0
3.5
Reference 1. ND = no data.
                                    3-49

-------
TABLE 3-5. VAPOR PRESSURE EQUATION CONSTANTS
            FOR ORGANIC LIQUIDS3
Name
Acetaldehyde
Acetic acid
Acetic anhydride
Acetone
Acetonitrile
Acrylamide
Acrylic acid
Acrylonitrile
Aniline
Benzene
Butanol (iso)
Butanol-(l)
Carbon disulfide
Carbon tetrachloride
Chlorobenzene
Chloroform
Chloroprene
Cresol(-M)
Cresol(-O)
Cresol(-P)
Cumene (isopropylbenzene)
Cyclohexane
Cyclohexanol
Cyclohexanone
Dichloroethane( 1,2)
Dichloroethylene(l,2)
Diethyl (N,N) aniline
Dimethyl formamide
Dimethyl hydrazine (1,1)
Dimethyl phthalate
Dinitrobenzene
Dioxane(l,4)
Vapor pressure equation constants
A
(dimensionless)
8.005
7.387
7.149
7.117
7.119
11.2932
5.652
7.038
7.32
6.905
7.4743
7.4768
6.942
6.934
6.978
6.493
6.161
7.508
6.911
7.035
6.963
6.841
6.255
7.8492
7.025
6.965
7.466
6.928
7.408
4.522
4.337
7.431
B
(°C)
1600.017
1533.313
1444.718
1210.595
1314.4
3939.877
648.629
1232.53
1731.515
1211.033
1314.19
1362.39
1169.11
1242.43
1431.05
929.44
783.45
1856.36
1435.5
1511.08
1460.793
1201.53
912.87
2137.192
1272.3
1141.9
1993.57
1400.87
1305.91
700.31
229.2
1554.68
C
(°C)
291.809
222.309
199.817
229.664
230
273.16
154.683
222.47
206.049
220.79
186.55
178.77
241.59
230
217.55
196.03
179.7
199.07
165.16
161.85
207.78
222.65
109.13
273.16
222.9
231.9
218.5
196.43
225.53
51.42
-137
240.34
                   3-50

-------
Table 3-5. (con't).
Epichlorohydrin
Ethanol
Ethanolamine(mono-)
Ethyl acetate
Ethyl acrylate
Ethyl benzene
Ethyl chloride
Ethyl ether
Formic acid
Furan
Furfural
Heptane(iso)
Hexane(-N)
Hexanol(-l)
Hydrocyanic acid
Methanol
Methyl acetate
Methyl ethyl ketone
Methyl isobutyl ketone
Methyl metharcrylate
Methyl styrene (alpha)
Methylene chloride
Morpholine
Naphthalene
Nitrobenzene
Pentachloroethane
Phenol
Picoline(-2)
Propanol (iso)
Propylene glycol
Propylene oxide
Pyridine
Resorcinol
Styrene
Tetrachloroethane( 1,1,1,2)
Tetrachloroethane( 1 , 1 ,2,2)
Tetrachloroethylene
Tetrahydrofuran
Toluene
Trichloro( 1 , 1 ,2)trifluoroethane
Trichloroethane( 1,1,1)
Trichloroethane(l,l,2)
Trichloroethylene
8.2294
8.321
7.456
7.101
7.9645
6.975
6.986
6.92
7.581
6.975
6.575
6.8994
6.876
7.86
7.528
7.897
7.065
6.9742
6.672
8.409
6.923
7.409
7.7181
7.01
7.115
6.74
7.133
7.032
8.117
8.2082
8.2768
7.041
6.9243
7.14
6.898
6.631
6.98
6.995
6.954
6.88
8.643
6.951
6.518
2086.816
1718.21
1577.67
1244.95
1897.011
1424.255
1030.01
1064.07
1699.2
1060.87
1198.7
1331.53
1171.17
1761.26
1329.5
1474.08
1157.63
1209.6
1168.4
2050.5
1486.88
1325.9
1745.8
1733.71
1746.6
1378
1516.79
1415.73
1580.92
2085.9
1656.884
1373.8
1884.547
1574.51
1365.88
1228.1
1386.92
1202.29
1344.8
1099.9
2136.6
1314.41
1018.6
273.16
237.52
173.37
217.88
273.16
213.21
238.61
228.8
260.7
227.74
162.8
212.41
224.41
196.66
260.4
229.13
219.73
216
191.9
274.4
202.4
252.6
235
201.86
201.8
197
174.95
211.63
219.61
203.5396
273.16
214.98
186.0596
224.09
209.74
179.9
217.53
226.25
219.48
227.5
302.8
209.2
192.7
       3-51

-------
                                        Table 3-5. (con't).
Trichlorofluoromethane
Trichloropropane(l,2,3)
Vinyl acetate
Vinylidene chloride
Xylene(-M)
Xylene(-O)
6.884
6.903
7.21
6.972
7.009
6.998
1043.004
788.2
1296.13
1099.4
1426.266
1474.679
236.88
243.23
226.66
237.2
215.11
213.69
aReference 11.
             TABLE 3-6.  PAINT SOLAR ABSORPTANCE FOR FIXED ROOF TANKS3
Paint Color
Aluminum
Aluminum
Aluminum
Beige/Cream
Brown
Gray
Gray
Green
Red
Rust
Tan
White
Paint Shade or Type
Specular
Diffuse
Mill finish, unpainted


Light
Medium
Dark
Primer
Red iron oxide

NA
Paint Factors (a)
Paint Condition
Good
0.39
0.60
0.10
0.35
0.58
0.54
0.68
0.89
0.89
0.38
0.43
0.17
Poor
0.49
0.68
0.15
0.49
0.67
0.63
0.74
0.91
0.91
0.50
0.55
0.34
Notes:
a Reference 8. If specific information is not available, a white shell and roof, with the paint in good condition, can be
  assumed to represent the most common or typical tank surface in use. If the tank roof and shell are painted a
  different color, a is determined from a  = (a R + a s)/2; where a R is the tank roof paint solar absorptance and
  a s is the tank shell paint solar absorptance.
bThis refers to aluminum as the base metal, rather than aluminum-colored paint.
NA = not applicable.
                                               3-52

-------
Table 3-7. METEOROLOGICAL DATA (TAX, TAN, I) FOR SELECTED U.S. LOCATIONS3
Location
Birmingham, AL


Montgomery, AL


Homer, AK


Phoenix, AZ


Tucson, AZ


Fort Smith, AR


Little Rock, AR


Bakersfield, CA


Long Beach, CA


Los Angeles AP, CA


Sacramento, CA


San Francisco AP, CA


Pro
Symbol
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
perty
Units
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
Monthly Averages
Jan.
52.7
33.0
707
57.0
36.4
752
27.0
14.4
122
65.2
39.4
1021
64.1
38.1
1099
48.4
26.6
744
49.8
29.9
731
57.4
38.9
766
66.0
44.3
928
64.6
47.3
926
52.6
37.9
597
55.5
41.5
708
Feb.
57.3
35.2
967
60.9
38.8
1013
31.2
17.4
334
69.7
42.5
1374
67.4
40.0
1432
53.8
30.9
999
54.5
33.6
1003
63.7
42.6
1102
67.3
45.9
1215
65.5
48.6
1214
59.4
41.2
939
59.0
44.1
1009
Mar.
65.2
42.1
1296
68.1
45.5
1341
34.4
19.3
759
74.5
46.7
1814
71.8
43.8
1864
62.5
38.5
1312
63.2
41.2
1313
68.6
45.5
1595
68.0
47.7
1610
65.1
49.7
1619
64.1
42.4
1458
60.6
44.9
1455
Apr.
75.2
50.4
1674
77.0
53.3
1729
42.1
28.1
1248
83.1
53.0
2355
80.1
49.7
2363
73.7
49.1
1616
73.8
50.9
1611
75.1
50.1
2095
70.9
50.8
1938
66.7
52.2
1951
71.0
45.3
2004
63.0
46.6
1920
May
81.6
58.3
1857
83.6
61.1
1897
49.8
34.6
1583
92.4
61.5
2677
88.8
57.5
2671
81.0
58.2
1912
81.7
59.2
1929
83.9
57.2
2509
73.4
55.2
2065
69.1
55.7
2060
79.7
50.1
2435
66.3
49.3
2226
June
87.9
65.9
1919
89.8
68.4
1972
56.3
41.2
1751
102.3
70.6
2739
98.5
67.4
2730
88.5
66.3
2089
89.5
67.5
2107
92.2
64.3
2749
77.4
58.9
2140
72.0
59.1
2119
87.4
55.1
2684
69.6
52.0
2377
July
90.3
69.8
1810
91.5
71.8
1841
60.5
45.1
1598
105.0
79.5
2487
98.5
73.8
2341
93.6
70.5
2065
92.7
71.4
2032
98.8
70.1
2684
83.0
62.6
2300
75.3
62.6
2308
93.3
57.9
2688
71.0
53.3
2392
Aug.
89.7
69.1
1724
91.2
71.1
1746
60.3
45.2
1189
102.3
77.5
2293
95.9
72.0
2183
92.9
68.9
1877
92.3
69.6
1861
96.4
68.5
2421
83.8
64.0
2100
76.5
64.0
2080
91.7
57.6
2368
71.8
54.2
2117
Sept.
84.6
63.6
1455
86.9
66.4
1468
54.8
39.7
791
98.2
70.9
2015
93.5
67.3
1979
85.7
62.1
1502
85.6
63.0
1518
90.8
63.8
1992
82.5
61.6
1701
76.4
62.5
1681
87.6
55.8
1907
73.4
54.3
1742
Oct.
74.8
50.4
1211
77.5
53.1
1262
44.0
30.6
437
87.7
59.1
1577
84.1
56.7
1602
75.9
49.0
1201
75.8
50.4
1228
81.0
54.9
1458
78.4
56.6
1326
74.0
58.5
1317
77.7
50.0
1315
70.0
51.2
1226
Nov.
63.7
40.5
858
67.0
43.0
915
34.9
22.8
175
74.3
46.9
1151
72.2
45.2
1208
61.9
37.7
851
62.4
40.0
847
67.4
44.9
942
72.7
49.6
1004
70.3
52.1
1004
63.2
42.8
782
62.7
46.3
821
Dec.
35.2
55.9
661
59.8
37.9
719
27.7
15.8
64
66.4
40.2
932
65.0
39.0
996
52.1
30.2
682
53.2
33.2
674
57.6
38.7
677
67.4
44.7
847
66.1
47.8
849
53.2
37.9
538
56.3
42.2
642
Annual
Average
73.2
51.1
1345
75.9
53.9
1388
43.6
29.5
838
85.1
57.3
1869
81.7
54.2
1872
72.5
49.0
1404
72.9
50.8
1404
77.7
53.3
1749
74.2
53.5
1598
70.1
55.0
1594
73.4
47.8
1643
64.9
48.3
1608
                               3-53

-------
Table 3-7 (cont).
Location
Santa Maria, CA


Denver, CO


Grand Junction, CO


Wilmington, DE


Atlanta, GA


Savannah, GA


Honolulu, HI


Chicago, IL


Springfield, IL


Indianapolis, IN


Wichita, KS


Property
Symbol
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
Units
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
Monthly Averages
Jan.
62.8
38.8
854
43.1
15.9
840
35.7
15.2
791
39.2
23.2
571
51.2
32.6
718
60.3
37.9
795
79.9
65.3
1180
29.2
13.6
507
32.8
16.3
585
34.2
17.8
496
39.8
19.4
784
Feb.
64.2
40.3
1141
46.9
20.2
1127
44.5
22.4
1119
41.8
24.6
827
55.3
34.5
969
63.1
40.0
1044
80.4
65.3
1396
33.9
18.1
760
38.0
20.9
861
38.5
21.1
747
46.1
24.1
1058
Mar.
63.9
40.9
1582
51.2
24.7
1530
54.1
29.7
1554
50.9
32.6
1149
63.2
41.7
1304
69.9
46.8
1399
81.4
67.3
1622
44.3
27.6
1107
48.9
30.3
1143
49.3
30.7
1037
55.8
32.4
1406
Apr.
65.6
42.7
1921
61.0
33.7
1879
65.2
38.2
1986
63.0
41.8
1480
73.2
50.4
1686
77.8
54.1
1761
82.7
68.7
1796
58.8
38.8
1459
64.0
42.6
1515
63.1
41.7
1398
68.1
44.5
1783
May
67.3
46.2
2141
70.7
43.6
2135
76.2
48.0
2380
72.7
51.7
1710
79.8
58.7
1854
84.2
62.3
1852
84.8
70.2
1949
70.048.1
1789

74.6
52.5
1866
73.4
51.5
1638
77.1
54.6
2036
June
69.9
49.6
2349
81.6
52.4
2351
87.9
56.6
2599
81.2
61.2
1883
85.6
65.9
1914
88.6
68.5
1844
86.2
71.9
2004
79.4
57.7
2007
84.1
62.0
2097
82.3
60.9
1868
87.4
64.7
2264
July
72.1
52.4
2341
88.0
58.7
2273
94.0
63.8
2465
85.6
66.3
1823
87.9
69.2
1812
90.8
71.5
1784
87.1
73.1
2002
83.3
62.7
1944
87.1
65.9
2058
85.2
64.9
1806
92.9
69.8
2239
Aug.
72.8
53.2
2106
85.8
57.0
2044
90.3
61.5
2182
84.1
65.4
1615
87.6
68.7
1709
90.1
71.4
1621
88.3
73.6
1967
82.1
61.7
1719
84.7
63.7
1806
83.7
62.7
1644
91.5
67.9
2032
Sept.
74.2
51.8
1730
77.5
47.7
1727
81.9
52.2
1834
77.8
58.0
1318
82.3
63.6
1422
85.6
67.6
1364
88.2
72.9
1810
75.5
53.9
1354
79.3
55.8
1454
77.9
55.3
1324
82.0
59.2
1616
Oct.
73.3
47.6
1353
66.8
36.9
1301
68.7
41.1
1345
66.7
45.9
984
72.9
51.4
1200
77.8
55.9
1217
86.7
72.2
1540
64.1
42.9
969
67.5
44.4
1068
66.1
43.4
977
71.2
46.9
1250
Nov.
68.9
42.1
974
52.4
25.1
884
51.0
28.2
918
54.8
36.4
645
62.6
41.3
883
69.5
45.5
941
83.9
69.2
1266
48.2
31.4
566
51.2
32.9
677
50.8
32.8
579
55.1
33.5
871
Dec.
64.6
38.3
804
46.1
18.9
732
38.7
17.9
731
43.6
27.3
489
54.1
34.8
674
62.5
39.4
754
81.4
66.5
1133
35.0
20.3
402
38.4
23.0
490
39.2
23.7
417
44.6
24.2
690
Annual
Average
68.3
45.3
1608
64.3
36.2
1568
65.7
39.6
1659
63.5
44.5
1208
71.3
51.1
1345
76.7
55.1
1365
84.2
69.7
1639
58.7
39.7
1215
62.6
42.5
1302
62.0
42.2
1165
67.6
45.1
1502
      3-54

-------
Table 3-7 (cont).
Location
Louisville, KY


Baton Rouge, LA


Lake Charles, LA


New Orleans, LA


Detroit, MI


Grand Rapids, MI


Minneapolis-
St. Paul, MN

Jackson, MS


Billings, MT


Las Vegas, NV


Newark, NJ


Property
Symbol
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
Units
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
Monthly Averages
Jan.
40.8
24.1
546
61.1
40.5
785
60.8
42.2
728
61.8
43.0
835
30.6
16.1
417
29.0
14.9
370
19.9
2.4
464
56.5
34.9
754
29.9
11.8
486
56.0
33.0
978
38.2
24.2
552
Feb.
45.0
26.8
789
64.5
42.7
1054
64.0
44.5
1010
64.6
44.8
1112
33.5
18.0
680
31.7
15.6
648
26.4
8.5
764
60.9
37.2
1026
37.9
18.8
763
62.4
37.7
1340
40.3
25.3
793
Mar.
54.9
35.2
1102
71.6
49.4
1379
70.5
50.8
1313
71.2
51.6
1415
43.4
26.5
1000
41.6
24.5
1014
37.5
20.8
1104
68.4
44.2
1369
44.0
23.6
1190
68.3
42.3
1824
49.1
33.3
1109
Apr.
67.5
45.6
1467
79.2
57.5
1681
77.8
58.9
1570
78.6
58.8
1780
57.7
36.9
1399
56.9
35.6
1412
56.0
36.0
1442
77.3
52.9
1708
55.9
33.2
1526
77.2
49.8
2319
61.3
42.9
1449
May
76.2
54.6
1720
85.2
64.3
1871
84.1
65.6
1849
84.5
65.3
1968
69.4
46.7
1716
69.4
45.5
1755
69.4
47.6
1737
84.1
60.8
1941
66.4
43.3
1913
87.4
59.0
2646
71.6
53.0
1687
June
84.0
63.3
1904
90.6
70.0
1926
89.4
71.4
1970
89.5
70.9
2004
79.0
56.3
1866
78.9
55.3
1957
78.5
57.7
1928
90.5
67.9
2024
76.3
51.6
2174
98.6
68.6
2778
80.6
62.4
1795
July
87.6
67.5
1838
91.4
72.8
1746
91.0
73.5
1788
90.7
73.5
1814
83.1
60.7
1835
83.0
59.8
1914
83.4
62.7
1970
92.5
71.3
1909
86.6
58.0
2384
104.5
75.9
2588
85.6
67.9
1760
Aug.
86.7
66.1
1680
90.8
72.0
1677
90.8
72.8
1657
90.2
73.1
1717
81.5
59.4
1576
81.1
58.1
1676
80.9
60.3
1687
92.1
70.2
1781
84.3
56.2
2022
101.9
73.9
2355
84.0
67.0
1565
Sept.
80.6
59.1
1361
87.4
68.3
1464
87.5
68.9
1485
86.8
70.1
1514
74.4
52.2
1253
73.4
50.8
1262
71.0
50.2
1255
87.6
65.1
1509
72.3
46.5
1470
94.7
65.6
2037
76.9
59.4
1273
Oct.
69.2
46.2
1042
80.1
56.3
1301
80.8
57.7
1381
79.4
59.0
1335
62.5
41.2
876
61.4
40.4
858
59.7
39.4
860
78.6
51.4
1271
61.0
37.5
987
81.5
53.5
1540
66.0
48.3
951
Nov.
55.5
36.6
653
70.1
47.2
920
70.5
48.9
917
70.1
49.9
973
47.6
31.4
478
46.0
30.9
446
41.1
25.3
480
67.5
42.3
902
44.4
25.5
561
66.0
41.2
1086
54.0
39.0
596
Dec.
45.4
28.9
488
63.8
42.3
737
64.0
43.8
706
64.4
44.8
779
35.4
21.6
344
33.8
20.7
311
26.7
11.7
353
60.0
37.1
709
36.0
18.2
421
57.1
33.6
881
42.3
28.6
454
Annual
Average
66.1
46.2
1216
78.0
57.0
1379
77.6
58.3
1365
77.7
58.7
1437
58.2
38.9
1120
57.2
37.7
1135
54.2
35.2
1170
76.3
52.9
1409
57.9
35.4
1325
79.6
52.8
1864
62.5
45.9
1165
      3-55

-------
Table 3-7 (cont).
Location
Roswell, NM


Buffalo, NY


New York, NY
(LaGuardia
Airport)
Cleveland, OH


Columbus, OH


Toledo, OH


Oklahoma City, OK


Tulsa, OK


Astoria, OR


Portland, OR


Philadelphia, PA


Property
Symbol
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
Units
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
Monthly Averages
Jan.
55.4
27.4
1047
30.0
17.0
349
37.4
26.1
548
32.5
18.5
388
34.7
19.4
459
30.7
15.5
435
46.6
25.2
801
45.6
24.8
732
46.8
35.4
315
44.3
33.5
310
38.6
23.8
555
Feb.
60.4
31.4
1373
31.4
17.5
546
39.2
27.3
795
34.8
19.9
601
38.1
21.5
677
34.0
17.5
680
52.2
29.4
1055
51.9
29.5
978
50.6
37.1
545
50.4
36.0
554
41.1
25.0
795
Mar.
67.7
37.9
1807
40.4
25.6
889
47.3
34.6
1118
44.8
28.4
922
49.3
30.6
980
44.6
26.1
997
61.0
37.1
1400
60.8
37.7
1306
51.9
36.9
866
54.5
37.4
895
50.5
33.1
1108
Apr.
76.9
46.8
2218
54.4
36.3
1315
59.6
44.2
1457
57.9
38.3
1350
62.3
40.5
1353
59.1
36.5
1384
71.7
48.6
1725
72.4
49.5
1603
55.5
39.7
1253
60.2
40.6
1308
63.2
42.6
1434
May
85.0
55.6
2459
65.9
46.3
1597
69.7
53.7
1690
68.5
47.9
1681
72.6
50.2
1647
70.5
46.6
1717
79.0
57.7
1918
79.7
58.5
1822
60.2
44.1
1608
66.9
46.4
1663
73.0
52.5
1660
June
93.1
64.8
2610
75.6
56.4
1804
78.7
63.2
1802
78.0
57.2
1843
81.3
59.0
1813
79.9
56.0
1878
87.6
66.3
2144
87.9
67.5
2021
63.9
49.2
1626
72.7
52.2
1773
81.7
61.5
1811
July
93.7
69.0
2441
80.2
61.2
1776
83.9
68.9
1784
81.7
61.4
1828
84.4
63.2
1755
83.4
60.2
1849
93.5
70.6
2128
93.9
72.4
2031
67.9
52.2
1746
79.5
55.8
2037
86.1
66.8
1758
Aug.
91.3
67.0
2242
78.2
59.6
1513
82.3
68.2
1583
80.3
60.5
1583
83.0
61.7
1641
81.8
58.4
1616
92.8
69.4
1950
93.0
70.3
1865
68.6
52.6
1499
78.6
55.8
1674
84.6
66.0
1575
Sept.
84.9
59.6
1913
71.4
52.7
1152
75.2
61.2
1280
74.2
54.0
1240
76.9
54.6
1282
75.1
51.2
1276
84.7
61.9
1554
85.0
62.5
1473
67.8
49.2
1183
74.2
51.1
1217
77.8
58.6
1281
Oct.
75.8
47.5
1527
60.2
42.7
784
64.5
50.5
951
62.7
43.6
867
65.0
42.8
945
63.3
40.1
911
74.3
50.2
1233
74.9
50.3
1164
61.4
44.3
713
63.9
44.6
724
66.5
46.5
959
Nov.
63.1
35.0
1131
47.0
33.6
403
52.9
41.2
593
49.3
34.3
466
50.7
33.5
538
47.9
30.6
498
59.9
37.6
901
60.2
38.1
827
53.5
39.7
387
52.3
38.6
388
54.5
37.1
619
Dec.
56.7
28.2
952
35.0
22.5
283
41.5
30.8
457
37.5
24.6
318
39.4
24.7
387
35.5
20.6
355
50.7
29.1
725
50.3
29.3
659
48.8
37.3
261
46.4
35.4
260
43.0
28.0
470
Annual
Average
75.3
47.5
1810
55.8
39.3
1034
61.0
47.5
1171
58.5
40.7
1091
61.5
41.8
1123
58.8
38.3
1133
71.2
48.6
1461
71.3
49.2
1373
58.1
43.1
1000
62.0
44.0
1067
63.4
45.1
1169
      3-56

-------
Table 3-7 (cont).
Location
Pittsburgh, PA


Providence, RI


Columbia, SC


Sioux Falls, SD


Memphis, TN


Amarillo, TX


Corpus Christi, TX


Dallas, TX


Houston, TX


Midland-Odessa,
TX

Salt Lake City, UT


Property
Symbol
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
Units
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
°F
°F
Btu/ft2-d
Monthly Averages
Jan.
34.1
19.2
424
36.4
20.0
506
56.2
33.2
762
22.9
1.9
533
48.3
30.9
683
49.1
21.7
960
66.5
46.1
898
54.0
33.9
822
61.9
40.8
772
57.6
29.7
1081
37.4
19.7
639
Feb.
36.8
20.7
625
37.7
20.9
739
59.5
34.6
1021
29.3
8.9
802
53.0
34.1
945
53.1
26.1
1244
69.9
48.7
1147
59.1
37.8
1071
65.7
43.2
1034
62.1
33.3
1383
43.7
24.4
989
Mar.
47.6
29.4
943
45.5
29.2
1032
67.1
41.9
1355
40.1
20.6
1152
61.4
41.9
1278
60.8
32.0
1631
76.1
55.7
1430
67.2
44.9
1422
72.1
49.8
1297
69.8
40.2
1839
51.5
29.9
1454
Apr.
60.7
39.4
1317
57.5
38.3
1374
77.0
50.5
1747
58.1
34.6
1543
72.9
52.2
1639
71.0
42.0
2019
82.1
63.9
1642
76.8
55.0
1627
79.0
58.3
1522
78.8
49.4
2192
61.1
37.2
1894
May
70.8
48.5
1602
67.6
47.6
1655
83.8
59.1
1895
70.5
45.7
1894
81.0
60.9
1885
79.1
51.9
2212
86.7
69.5
1866
84.4
62.9
1889
85.1
64.7
1775
86.0
58.2
2430
72.4
45.2
2362
June
79.1
57.1
1762
76.6
57.0
1776
89.2
66.1
1947
80.3
56.3
2100
88.4
68.9
2045
88.2
61.5
2393
91.2
74.1
2094
93.2
70.8
2135
90.9 70.2
1898

93.0
66.6
2562
83.3
53.3
2561
July
82.7
61.3
1689
81.7
63.3
1695
91.9
70.1
1842
86.2
61.8
2150
91.5
72.6
1972
91.4
66.2
2281
94.2
75.6
2186
97.8
74.7
2122
93.6
72.5
1828
94.2
69.2
2389
93.2
61.8
2590
Aug.
81.1
60.1
1510
80.3
61.9
1499
91.0
69.4
1703
83.9
59.7
1845
90.3
70.8
1824
89.6
64.5
2103
94.1
75.8
1991
97.3
73.7
1950
93.1
72.1
1686
93.1
68.0
2210
90.0
59.7
2254
Sept.
74.8
53.3
1209
73.1
53.8
1209
85.5
63.9
1439
73.5
48.5
1410
84.3
64.1
1471
82.4
56.9
1761
90.1
72.8
1687
89.7
67.5
1587
88.7
68.1
1471
86.4
61.9
1844
80.0
50.0
1843
Oct.
62.9
42.1
895
63.2
43.1
907
76.5
50.3
1211
62.1
36.7
1005
74.5
51.3
1205
72.7
45.5
1404
83.9
64.1
1416
79.5
56.3
1276
81.9
57.5
1276
77.7
51.1
1522
66.7
39.3
1293
Nov.
49.8
33.3
505
51.9
34.8
538
67.1
40.6
921
43.7
22.3
608
61.4
41.1
817
58.7
32.1
1033
75.1
54.9
1043
66.2
44.9
936
71.6
48.6
924
65.5
39.0
1176
50.2
29.2
788
Dec.
38.4
24.3
347
40.5
24.1
419
58.8
34.7
722
29.3
10.1
441
52.3
34.3
629
51.8
24.8
872
69.3
48.8
845
58.1
37.4
780
65.2
42.7
730
59.7
32.2
1000
38.9
21.6
570
Annual
Average
59.9
40.7
1069
59.3
41.2
1112
75.3
51.2
1380
56.7
33.9
1290
71.6
51.9
1366
70.7
43.8
1659
81.6
62.5
1521
76.9
55.0
1468
79.1
57.4
1351
77.0
49.9
1802
64.0
39.3
1603
      3-57

-------
                                                                    Table 3-7 (cont).
Location
Richmond, VA


Seattle, WA
(Sea-Tac Airport)

Charleston, WV


Huntington, WV


Cheyenne, WY


Property
Symbol
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
TAX
TAN
I
Units
°F
°F
Btu/ft2 day
°F
°F
Btu/ft2 day
°F
°F
Btu/ft2 day
°F
°F
Btu/ft2 day
°F
°F
Btu/ft2 day
Monthly Averages
Jan.
46.7
26.5
632
43.9
34.3
262
41.8
23.9
498
41.1
24.5
526
37.3
14.8
766
Feb.
49.6
28.1
877
48.8
36.8
495
45.4
25.8
707
45.0
26.6
757
40.7
17.9
1068
Mar.
58.5
35.8
1210
51.1
37.2
849
55.4
34.1
1010
55.2
35.0
1067
43.6
20.6
1433
Apr.
70.6
45.1
1566
56.8
40.5
1294
67.3
43.3
1356
67.2
44.4
1448
54.0
29.6
1771
May
77.9
54.2
1762
64.0
46.0
1714
76.0
51.8
1639
75.7
52.8
1710
64.6
39.7
1995
June
84.8
62.2
1872
69.2
51.1
1802
82.5
59.4
1776
82.6
60.7
1844
75.4
48.5
2258
July
88.4
67.2
1774
75.2
54.3
2248
85.2
63.8
1683
85.6
65.1
1769
83.1
54.6
2230
Aug.
87.1
66.4
1601
73.9
54.3
1616
84.2
63.1
1514
84.4
64.0
1580
80.8
52.8
1966
Sept.
81.0
59.3
1348
68.7
51.2
1148
78.7
56.4
1272
78.7
57.2
1306
72.1
43.7
1667
Oct.
70.5
46.7
1033
59.5
45.3
656
67.7
44.0
972
67.6
44.9
1004
61.0
34.0
1242
Nov.
60.5
37.3
733
50.3
39.3
337
55.6
35.0
613
55.2
35.9
638
46.5
23.1
823
Dec.
50.2
29.6
567
45.6
36.3
211
45.9
27.8
440
45.2
28.5
467
40.4
18.2
671
Annual
Average
68.8
46.5
1248
58.9
43.9
1053
65.5
44.0
1123
65.3
45.0
1176
58.3
33.1
1491
oo
a References 13 and 14
TAx = daily maximum ambient temperature
TAN = daily minimum ambient temperature
I = daily total solar insolation factor
                                                                          3-58

-------
                     TABLE 3-8.  RIM-SEAL LOSS FACTORS, KRa, KRb and n,
                                 FOR FLOATING ROOF TANKS3
Tank Construction And
Rim-Seal System
Welded Tanks
Mechanical-shoe seal
Primary onlyb
Shoe-mounted secondary
Rim-mounted secondary
Liquid-mounted seal
Primary only
Weather shield
Rim-mounted secondary
Vapor-mounted seal
Primary only
Weather shield
Rim-mounted secondary

(Ib-mole/ft-yr)


5.8
1.6
0.6

1.6
0.7
0.3

6.7C
3.3
2.2
Average-Fitting Seals
[lb-mole/(mph)n-ft-yr]


0.3
0.3
0.4

0.3
0.3
0.6

0.2
0.1
0.003

n
(dimensionless)


2.1
1.6
1.0

1.5
1.2
0.3

3.0
3.0
4.3
Riveted Tanks
Mechanical-shoe seal
Primary only
Shoe-mounted secondary
Rim-mounted secondary

10.8
9.2
1.1

0.4
0.2
0.3

2.0
1.9
1.5
"Reference 11.
blf no specific information is available, a welded tank with an average-fitting mechanical-shoe primary seal can be
 used to represent the most common or typical construction and rim-seal system in use for external and domed
 external floating roof tanks.
°If no specific information is available, this value can be assumed to represent the most common or typical rim-seal
 system currently in use for internal floating roof tanks.
                                               3-59

-------
TABLE 3-9. AVERAGE ANNUAL WIND SPEED (v) FOR SELECTED U. S. LOCATIONS3
Location
Alabama
Birmingham
Huntsville
Mobile
Montgomery

Alaska
Anchorage
Annette
Barrow
Barter Island
Bethel
Bettles
Big Delta
Cold Bay
Fairbanks
Gulkana
Homer
Juneau
King Salmon
Kodiak
Kotzebue
McGrath
Nome
St. Paul Island
Talkeetna
Valdez
Yakutat

Arizona
Flagstaff
Phoenix
Tucson
Wind
Speed
(mph)

7.2
8.2
9.0
6.6


6.9
10.6
11.8
13.2
12.8
6.7
8.2
17.0
5.4
6.8
7.6
8.3
10.8
10.8
13.0
5.1
10.7
17.7
4.8
6.0
7.4


6.8
6.3
8.3
Wind
Location Speed
(mph)
Arizona (continued)
Winslow 8.9
Yuma 7.8

Arkansas
Fort Smith 7.6
Little Rock 7.8

California
Bakersfield 6.4
Blue Canyon 6.8
Eureka 6.8
Fresno 6.3
Long Beach 6.4
Los Angeles (City) 6.2
Los Angeles Int'l Airport 7.5
Mount Shasta 5.1
Sacramento 7.9
San Diego 6.9
San Francisco (City) 8.7
San Francisco Airport 10.6
Santa Maria 7.0
Stockton 7.5

Colorado
Colorado Springs 10.1
Denver 8.7
Grand Junction 8.1
Pueblo 8.7

Connecticut
Bridgeport 12.0
Hartford 8.5
Location
Delaware
Wilmington
District of Columbia
Dulles Airport
National Airport

Florida
Apalachicola
Daytona Beach
Fort Meyers
Jacksonville
Key West
Miami
Orlando
Pensacola
Tallahassee
Tampa
West Palm Beach

Georgia
Athens
Atlanta
Augusta
Columbus
Macon
Savannah

Hawaii
Hilo
Honolulu
Kahului
Lihue

Wind
Speed
(mph)

9.1

7.4
9.4


7.8
8.7
8.1
8.0
11.2
9.3
8.5
8.4
6.3
8.4
9.6


7.4
9.1
6.5
6.7
7.6
7.9


7.2
11.4
12.8
12.2

                              3-60

-------
Table 3-9. (cont.)
Location
Idaho
Boise
Pocatello

Illinois
Cairo
Chicago
Moline
Peoria
Rockford
Springfield

Indiana
Evansville
Fort Wayne
Indianapolis
South Bend
Iowa
Des Moines
Sioux City
Waterloo

Kansas
Concordia
Dodge City
Goodland
Topeka
Wichita

Kentucky
Cincinnati Airport
Jackson
Lexington
Louisville
Wind
Speed
(mph)

8.8
10.2


8.5
10.3
10.0
10.0
10.0
11.2


8.1
10.0
9.6
10.3

10.9
11.0
10.7


12.3
14.0
12.6
10.0
12.3


9.1
7.2
9.3
8.4
Location
Louisiana
Baton Rouge
Lake Charles
New Orleans
Shreveport

Maine
Caribou
Portland

Maryland
Baltimore

Massachusetts
Blue Hill Observatory
Boston
Worcester
Michigan
Alpena
Detroit
Flint
Grand Rapids
Houghton Lake
Lansing
Muskegon
Sault Sainte Marie

Minnesota
Duluth
International Falls
Minneapolis-Saint Paul
Rochester
Saint Cloud

Wind
Speed
(mph)

7.6
8.7
8.2
8.4


11.2
8.8


9.2


15.4
12.5
10.1

8.1
10.4
10.2
9.8
8.9
10.0
10.7
9.3


11.1
8.9
10.6
13.1
8.0

Location
Mississippi
Jackson
Meridian

Missouri
Columbia
Kansas City
Saint Louis
Springfield

Montana
Billings
Glasgow
Great Falls
Helena
Kalispell
Missoula
Nebraska
Grand Island
Lincoln
Norfolk
North Platte
Omaha
Scottsbuff
Valentine

Nevada
Elko
Ely
Las Vegas
Reno
Winnemucca


Wind
Speed
(mph)

7.4
6.1


9.9
10.8
9.7
10.7


11.2
10.8
12.8
7.8
6.6
6.2

11.9
10.4
11.7
10.2
10.6
10.6
9.7


6.0
10.3
9.3
6.6
8.0


      3-61

-------
Table 3-9. (cont.)
Location
New Hampshire
Concord
Mount Washington

New Jersey
Atlantic City
Newark

New Mexico
Albuquerque
Roswell

New York
Albany
Birmingham
Buffalo

New York (Central Park)
New York (JFK Airport)
New York (La Guardia
Airport)
Rochester
Syracuse
North Carolina
Asheville
Cape Hatteras
Charlotte
Greensboro-High Point
Raleigh
Wilmington


North Dakota
Bismark
Fargo
Williston
Wind
Speed
(mph)

6.7
35.3


10.1
10.2


9.1
8.6


8.9
10.3
12.0

9.4
12.0
12.2
9.7
9.5

7.6
11.1
7.5
7.5
7.8
8.8



10.2
12.3
10.1
Location
Ohio
Akron
Cleveland
Columbus
Dayton
Mansfield
Toledo
Youngstown

Oklahoma
Oklahoma City
Tulsa

Oregon
Astoria
Eugene

Medford
Pendleton
Portland
Salem
Sexton Summit
Pennsylvania
Allentown
Avoca
Erie
Harrisburg
Philadelphia
Pittsburgh Int'l
Airport
Williamsport

Puerto Rico
San Juan

Wind
Speed
(mph)

9.8
10.6
8.5
9.9
11.0
9.4
9.9


12.4
10.3


8.6
7.6

4.8
8.7
7.9
7.1
11.8

9.2
8.3
11.3
7.6
9.5
9.1

7.8


8.4

Location
Rhode Island
Providence

South Carolina
Charleston
Columbia
Greenville-
Spartanburg
South Dakota
Aberdeen
Huron
Rapid City
Sioux Falls

Tennessee
Bristol-Johnson
City
Chattanooga
Knoxville
Memphis
Nashville
Oak Ridge
Texas
Abilene
Amarillo
Austin
Brownsville
Corpus Christi
Dallas-Fort Worth

Del Rio
El Paso
Galveston
Houston
Lubbock
Wind
Speed
(mph)

10.6


8.6
6.9
6.9


11.2
11.5
11.3
11.1


5.5

6.1
7.0
8.9
8.0
4.4

12.0
13.6
9.2
11.5
12.0
10.8

9.9
8.9
11.0
7.9
12.4
      3-62

-------
                      Table 3-9. (cont.)
Location
Texas (continued)
Midland-Odessa
Port Arthur
San Angelo
San Antonio
Victoria
Waco
Wichita Falls

Utah
Salt Lake City
Vermont
Burlington
Virginia
Lynchburg
Norfolk
Richmond
Roanoke
Washington
Olympia
Quillayute
Seattle Int'l. Airport
Spokane
Walla Walla
Yakima
West Virginia
Belkley
Charleston
Elkins
Huntington
Wind
Speed
(mph)

11.1
9.8
10.4
9.3
10.1
11.3
11.7


8.9

8.9

7.7
10.7
7.7
8.1

6.7
6.1
9.0
8.9
5.3
7.1

9.1
6.3
6.2
6.6
Location
Wisconsin
Green Bay
La Crosse
Madison
Milwaukee

Wyoming
Casper
Cheyenne
Lander
Sheridan



















Wind
Speed
(mph)

10.0
8.8
9.9
11.6


12.9
13.0
6.8
8.0



















'Reference 13.
                           3-63

-------
                            TABLE 3-10. AVERAGE CLINGAGE FACTORS, C
                                       (Barrels per 1,000 square feet)a
Product stored
Gasoline
Single -component stocks
Crude oil
Shell condition
Light
rust
0.0015
0.0015
0.0060
Dense
rust
0.0075
0.0075
0.030
Gunite
lining
0.15
0.15
0.60
"Reference 10.

Note:  If no specific information is available, the values in this table can be assumed to represent the
                    most common or typical condition of tanks currently in use.
                TABLE 3-11. TYPICAL NUMBER OF COLUMNS AS A FUNCTION OF TANK
                  DIAMETER FOR INTERNAL FLOATING ROOF TANKS WITH COLUMN-
                                      SUPPORTED FIXED ROOFS3
Tank diameter range D, (ft)
0
-------
  Table 3.1-12. DECK-FITTING LOSS FACTORS, KFa, KFb,
AND m, AND TYPICAL NUMBER OF DECK FITTINGS, NFa
Fitting Type And Construction Details
Access hatch (24-inch diameter well)
Bolted cover, gasketedb
Unbolted cover, ungasketed
Unbolted cover, gasketed
Fixed roof support column welld
Round pipe, ungasketed sliding cover
Round pipe, gasketed sliding cover
Round pipe, flexible fabric sleeve seal
Built-up column, ungasketed sliding cover0
Built-up column, gasketed sliding cover
Unslotted guide-pole and well (8-inch
diameter unslotted pole, 21 -inch
diameter well)
Ungasketed sliding coverb
Ungasketed sliding cover w/pole
sleeveGasketed sliding cover
Gasketed sliding cover w/pole wiper
Gasketed sliding cover w/pole sleeve
Slotted guide-pole/sample well (8-inch
diameter slotted pole, 21 -inch
diameter well)6
Ungasketed or gasketed sliding cover
Ungasketed or gasketed sliding cover,
with float8
Gasketed sliding cover, with pole wiper
Gasketed sliding cover, with pole sleeve
Gasketed sliding cover, with pole sleeve
and pole wiper
Gasketed sliding cover, with float and
pole wiper8
Gasketed sliding cover, with float, pole
sleeve, and pole wiper11
Gauge-float well (automatic gauge)
Unbolted cover, ungasketed
Unbolted cover, gasketed
Bolted cover, gasketed
Gauge-hatch/sample port
Weighted mechanical actuation,
gasketedb
Weighted mechanical actuation,
ungasketed
Slit fabric seal, 10% open area0
Vacuum breaker
Weighted mechanical actuation,
ungasketed
Weighted mechanical actuation, gasketedb







Loss Factors
TS TS
(lb-mole/yr) (lb-mole/(mph)m-yr)

1.6 0
36° 5.9
31 5.2

31
25
10
51
33



31 150
25 2.2
25 13
14 3.7
8.6 12



43 270

31 36
41 48
11 46

8.3 4.4

21 7.9
11 9.9

14C 5.4
4.3 17
2.8 0


0.47 0.02

2.3 0
12


7.8 0.01
6.2C 1.2








Typical Number Of
m Fittings, NF
(dimensionless)
1
0
1.2
1.3
Nc
(Table 7.1-11)






1
1.4
2.1
2.2
0.78
0.81


f
1.4

2.0
1.4
1.4

1.6

1.8
0.89
1
1.1
0.38
0
1

0.97

0

Nyb (Table 7.1 -
13yDeckdrain(3-
4.0 inch diameter)
0.94 Openb
90% closed
1.5
1.8
0.21
0.14
1.7
l.lNd (Table 7.1-13)
                      3-65

-------
Fitting Type And Construction Details
Stub drain (1-inch diameter )k
Deck leg (3 -inch diameter)
Adjustable, internal floating deck0
Adjustable, pontoon area - ungasketedb
Adjustable, pontoon area - gasketed
Adjustable, pontoon area - sock
Adjustable, center area - ungasketedb
Adjustable, center area - gasketed111
Adjustable, center area - sock™
Adjustable, double-deck roofs
Fixed
Rim vent11
Weighted mechanical actuation, ungasketed
Weighted mechanical actuation, gasketedb
Ladder well
Sliding cover, ungasketed0
Sliding cover, gasketed
Loss Factors
KFa Kpb m
(Ib-mole/yr) (lb-mole/(mph)m-yr) (dimensionless)
1.2

7.9
2.0 0.37 0.91
1.3 0.08 0.65
1.2 0.14 0.65
0.82 0.53 0.14
0.53 0.11 0.13
0.49 0.16 0.14
0.82 0.53 0.14
0 00

0.68 1.8 1.0
0.71 0.10 1.0

98
56
Typical Number Of
Fittings, NF
Nd (Table 7. 1-1 5)
NI (Table 7.1-15),
(Table 7. 1-14)








1


ld


Note:  The deck-fitting loss factors, KFa, KFb, and m, may only be used for wind speeds below  15 miles
        per hour.

a  Reference 5, unless otherwise indicated.
b  If no specific information is available, this value can be assumed to represent the most common or typical deck fitting
   currently in use for external and domed external floating roof tanks.
0  If no specific information is available, this value can be assumed to represent the most common or typical deck fitting
   currently in use for internal floating roof tanks.
d  Column wells and ladder wells are not typically used with self supported fixed roofs.
e  References 16,19.
f  A slotted guide-pole/sample well is an optional fitting and is not typically used.
B  Tests were conducted with floats positioned with the float wiper at and 1 inch above the sliding cover. The user is cautioned
   against applying these factors to floats that are positioned with the wiper or top of the float below the sliding cover ("short
   floats"). The emission factor for such a float is expected to be between the factors for a guidepole without a float and with a
   float, depending upon the position of the float top and/or wiper within the guidepole.
h  Tests were conducted with floats positioned with the float wiper at varying heights with respect to the sliding cover.  This
   fitting configuration also includes a pole sleeve which restricts the airflow from the well vapor space into the slotted
   guidepole. Consequently, the float position  within the guidepole (at, above, or below the sliding cover) is not expected to
   significantly affect emission levels for this fitting configuration, since the function of the pole sleeve is to restrict the flow of
   vapor from the vapor space below the deck into the guidepole.
j   Nvb = 1 for internal floating roof tanks.
k  Stub drains are not used on welded contact internal floating decks.
m  These loss factors were derived using the results from pontoon-area deck legs with gaskets and socks.
n  Rim vents are used only with mechanical-shoe primary seals.
                                                        3-66

-------
                 TABLE 3-13. EXTERNAL FLOATING ROOF TANKS: TYPICAL NUMBER OF
                              VACUUM BREAKERS, Nvb, AND DECK DRAINS, Nda
Tank Diameter
D (feet)b
50
100
150
200
250
300
350
400
Number Of Vacuum Breakers, Nvb
Pontoon Roof Double-Deck Roof
1
1
2
3
4
5
6
7
1
1
2
2
3
3
4
4
Number Of Deck drains, Nd
1
1
2
3
5
7
ND
ND
Note:   This table was derived from a survey of users and manufacturers.  The actual number of vacuum breakers
       may vary greatly depending on throughput and manufacturing prerogatives. The actual number of deck
       drains may also vary greatly depending on the design rainfall and manufacturing prerogatives.  For tanks
       more  than 300 feet in diameter, actual tank data or the manufacturer's recommendations may be needed for
       the number of deck drains. This table should not supersede information based on actual tank data.

       "Reference 10. ND = no data.
       blf the actual diameter is between the diameters listed, the closest diameter listed should be
         used. If the actual diameter is midway between the diameters listed, the next larger
         diameter should be used.
                                               3-67

-------
        TABLE 3-14.  EXTERNAL FLOATING ROOF TANKS: TYPICAL NUMBER OF
                                                 DECK LEGS, NLa
Tank
diameter, D
(feet)b
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
210
220
230
240
250
260
270
280
290
300
310
320
330
340
350
360
370
380
390
400
Pontoon roof
Number of
pontoon
legs
4
4
6
9
13
15
16
17
18
19
20
21
23
26
27
28
29
30
31
32
33
34
35
36
36
37
38
38
39
39
40
41
42
44
45
46
47
48
Number of
center legs
2
4
6
7
9
10
12
16
20
24
28
33
38
42
49
56
62
69
77
83
92
101
109
118
128
138
148
156
168
179
190
202
213
226
238
252
266
281
Number of
legs on
double-
deck roof
6
7
8
10
13
16
20
25
29
34
40
46
52
58
66
74
82
90
98
107
115
127
138
149
162
173
186
200
213
226
240
255
270
285
300
315
330
345
Note:   This table was derived from a survey of users and manufacturers. The actual number of roof legs may vary
       greatly depending on age, style of floating roof, loading specifications, and manufacturing prerogatives.
       This table should not supersede information based on actual tank data.
""Reference 10.
blf the actual diameter is between the diameters listed, the closest diameter listed should be used. If the actual
diameter is midway between the diameters listed, the next larger diameter should be used.
                                               3-68

-------
          TABLE 3-15. INTERNAL FLOATING ROOF TANKS:  TYPICAL NUMBER
         	OF DECK LEGS, Nh AND STUB DRAINS, Nda	
                Deck fitting type
Typical number of
   fittings, NF
 Deck leg or hanger wellb
                                                               10  600
  Stub drain (3-inch diameter)0
                                                                125
aReference 5.
bD = tank diameter, ft
°Not used on welded contact internal floating roof decks.
   TABLE 3-16. DECK SEAM LENGTH FACTORS (SD) FOR TYPICAL DECK CONSTRUCTIONS
                              FOR INTERNAL FLOATING ROOF TANKS3
Deck construction
Continuous sheet construction13
5 ft wide
6 ft wide
7 ft wide
Panel constructiond
5 x 7.5 ft rectangular
5 x 12 ft rectangular
Typical deck seam length factor,
SD (ft/ft2)

0.20C
0.17
0.14

0.33
0.28
Reference 5.  Deck seam loss applies to bolted internal floating decks only.
bSD = 1/W, where W = sheet width (ft).
°If no specific information is available, this value can be assumed to represent the most common bolted
   decks currently in use.
dSD = (L+W)/LW, where W = panel width (ft) and L = panel length (ft).
                                        3-69

-------
Table 3-17. ROOF LANDING LOSSES FOR INTERNAL FLOATING ROOF TANK WITH A LIQUID
                                  HEEL
Standing Idle Loss
Standing Idle Saturation Factor
Filling Loss Equation
Filling Saturation Factor (S)
Equation 2-16
T V -m V \A V
^ rj,
T 
-------
   Table 3-18. ROOF LANDING LOSSES FOR EXTERNAL FLOATING ROOF TANK WITH A
                                      LIQUID HEEL
Standing Idle Loss
                  L,r = 0.57 n, D P* Mv
                   SL        a         V
                      <
                      -
                                                        Equation 2- 19
                                                                          Equation 2- 13
Standing Idle
Saturation Factor
Not applicable
Filling Loss
Equation
                  Lm =
        PVy
        RT.
V (c, s)
                                                        Equation 2-27
Q=1-
                                                           v
                                 A\
                                                                          Equation 2-30
Filling Saturation
Factor (S)
S = 0.6 for a full liquid heel
S = 0.5 for a partial liquid heel

Csf S> 0.15
                                           3-71

-------
              Table 3-19. ROOF LANDING LOSSES FOR ALL DRAIN-DRY TANKS
Standing Idle Loss


Standing Idle Saturation
Factor
Filling Loss Equation

Filling Saturation Factor (S)
/ 2\ Equation 2-22
T O OOfi'? W
LSL 0.006JI1^ 4 J
f pi/- \ Equation 2-23
T -- 060 \ v Vf

Not applicable
( p y A Equation 2-26
T -\ v \ \f S*
I /? T )
S = 0.15
where:
         Ls =  standing idle loss per landing episode (Ib)
         nd =  number of days the tank stands idle with the floating roof landed (dimensionless)
        KE =  vapor space expansion factor (dimensionless)
                                       kTv I     0.50BP
                                       ~r( +T(PA-P)^
       A Tv =  daily vapor temperature range (°R)
         T =  average temperature of the vapor and liquid below the floating roof (°R)
         B =  constant from the vapor pressure equation shown in Equation 1-24 (°R)
         P =  true vapor pressure of the stock liquid (psia)
         PA =  atmospheric pressure at the tank location (psia)
        Vv =  volume of the vapor space (ft3)
                                       vv=-
                                            k, n D2
         hv =  height of the vapor space under the floating roof (ft)
         D =  tank diameter (ft)
         R =  ideal gas constant (psia ft3 / Ib-mole R) = 10.731
        Mv =  stock vapor molecular weight (Ib/lb-mole)
        Ks =  standing idle saturation factor (dimensionless)
         S =  filling saturation factor (dimensionless)
                                            3-72

-------
 P =  vapor pressure function (dimensionless)
                             P=-
Wi =  stock liquid density (Ib/gal)
hie =  effective height of the stock liquid (ft)
LF =  filling loss per landing episode (Ib)
Csf =  filling saturation correction factor (dimensionless)
                                       3-73

-------
TABLE 3-20. HENRY'S LAW CONSTANTS FOR SELECTED ORGANIC LIQUIDS

MO FORMULA
1 C2H40
2 C2M50N
3 C2H3N
C8H80
C3H40
C3NSNO
C3N402
C3H3N
C3H5CL
10 C6N7N
11 C7H9NO
12 C6M6
13 C7HSCL3
14 C7H7CL
15 C12N10
Uf*'tta± fn ^**
C2H4CL20
17 CNM3
18 C4H6
19 C6H110N
20 CU
21 CCL4
22 C2N3C102
23 C8N7CLO
24 C6MSCL
25 CNCL3
26 C4H5CL
27 C7H80
29 C7H80
30 C7H80
31 C9H12
32 C6H40.2
33 C4H8CL20
34 C3H4CL2
35 C4N11N02
36 C8N11H
37 C4N1004S
38 C14N16N2
39 C3N7NO
40 C2N8N2
41 C10N100*
42 C2H6S04
43 C6N3N2O4
44 C7M6N204
45 C4H8Q2
4* C12N12Ht
47 C3H5CLO
48 C3H802
49 C8N10
SO C2N5CL

MAM
ACETALOEHYOI
ACETAMIOE
ACETONITRILI
ACETOPHENOMf
ACROLEli
ACRYLAHIOt
ACRYLIC ACIO
ACRYLONITRILE
ALLYL CHLORIDE
ANIL1N8
0-ANISI01N8
BENZEMC
8ENZOTR I CHLORIDE
UNZYL CHLORIDE
•IPHEKYl
At £tf*tAt J^M^^KYUVI %MTyXft
•ISC CHLORMTMYL jETHf •
•ROMOFORN
1,3-KJTAOIEMI
CAPROUCTAM
CARMN OISOLFI08
CARSON TETMCHLORIDE
CHLOROACEtIC ACIO
2- CHLOROACETOPHENONf
CHLOROKNZElNl
CHLOROFORM
CHLOROPRfNI
M-CRESOl.
f*tttf Cfv tf y^ABAVI ff* Af*ffftffC0W
CKESOLS/UltSTLIC AClUdSOT
0-CRESOL
P-OESOL
CUNEN8
1,4-01CHLOROMN2EN8
OICHLOROETHYL ETHER
1.3-OICHLOROFROWIRI
DIETHAMOLAMINi
N,N<0!NETYUNIL!lBl
01 ETHYL SulFATI
OIMETHYL1M2IDHNI
DIMETHYL FORHAMIOi
1,1*0 1 METHYLHYORAZINi
DIMETHYL PHTHALATE
DIMETHYL SULFATI
2,4-OIHITROMtEMOl
214-OIHITMTOLUEJRI
1,4-OIOXAlM
1.2-OIPHEIIYLHYORA2IIRI
EFtCHLOROHYORIII
ETHYL ACRYLATE
rrimmoMi
ETHYL OHORIOi
Honry'* Law Constant.
H 8 25 C
4.8730000
0.0000986
1.1076388
0.5089400
4.5711400
0.0000145
0.0223962
5.4484900
513.4180500
0.0977600
0.0092393
308.3400000
54.5177107
17.7286753
22.6700000
29.5600000
3961.1453000
0.0001639
1064.0713500
1677.7900000
0.0036272
1.5713000
209.4500000
221.3300000
51.6355560
0.0394800
0.0911500
0.0396800
727.7800000
176.1100000
1.1390000
197.2200000
0.0000001
0.7701322
0.3405000
0.1780100
0.0098341
0.091075*
0.0548542
0.2226700
0.4756000
0.3996900
0.3079797
0.0135700
1.8590400
14.1169500
437.8100008
672.2300008
H • ata/aal fraction
•ASIS
Expariaantal
UNIFAC
VLE Oata
Solubility Oata
Solubility Oata
UNIFAC
VLE Data
Solubility Oata
Solubility Oata
Solubility Oata
UNIFAC
Exp*ria»ntal
UNIFAC
UNIFAC
ExpariHantal
Raactlon witti wata?
E jipar 1 avnta I
Solubility Oata
UNIFAC
Solubility Bata
Expariaantal
UNIFAC
Solubility • EttiHt**1
CxpaHaantal
ExpaHaantal
UNIFAC
Solubility Oata
Solubility Oata
Solubility Oata
Expariaantal
Expariawital
Solubility Oata
ExpaHaamai
UNIFAC
UNIFAC
Solubility Oat*
Solubflity - Eati**ta4
VLE Oata
VLE Oata
UNIFAC
Solubility Oat*
Solubility Oata
Solubility Oat*
VIE Oata
Solubility * EattMt**V
Solubility Oat*
Solubility Oata
Expariawttal
E"**riMM*t
                                     3-74

-------
TABLE 3-20. (cont.)

NO FORMULA
51 C2H4BR2
52 C2H4CL2
53 C2H602
54 C2H40
55 C2H4a2
54 CH20
57 C4H1002
58 C4H1002
59 C8H1604
60 C6H1203
61 C6H1403
62 CSH1003
63 C8H1803
64 C6H1403
65 C3H802
66 C6M1202
67 CBH1 002
68 CSH1202
69 can 1803
70 C6M1402
71 C8H1804
72 C8H 1503
73 C6CL6
74 C4CL6
75 C2CL6
76 C6H14
77 C8H4oa
78 C9H140
79 C4H203
80 CH40
81 CH38R
82 CH3CL
83 C2H3CL3
84 C4H80
85 CH4M2
84 C6H120
mf (•OUtat/1
of dnjMJ
88 C5H802
89 CSH120
90 CH2CL2
91 C15H10M202
92 C13H14N2
93 C10H8
94 C6H5N03
95 C6H5N03
94 C3H7N02
97 C6H40
98 C6M8N2
99 COCL2
100 C8M40S
Harry
MAME
ETHYLEME 01 BROMIDE
ETHYLEME 01 CHLORIDE
ETHYLEME GLYCOL
ETHYLENE OXIDE
ETHYLIOEME BICHLORIDE
FORMALDEHYDE
ETHYLENE GLYCOL DIMETHYL ETHER
ETHYLENE GLYCOL MOMOETHYL ETHER
01 ETHYLENE GLYCOL NONOETNYL ETHER ACETATE
ETHYLENE GLYCOL MONOETHYL ETHER ACETATE
01 ETHYLEME GLYCOL MONOETHYL ETHER
ETHYLENE GLYCOL MOMOMETHYL ETHER ACETATE*
0 1 ETHYLEME GLYCOL MON08UTYL ETHER
OIETHYLENE GLYCOL DIMETHYL ETHER
ETHYLENE GLYCOL MOMOMETHYL ETHER
ETHYLENE GLYCOL MONOPROPYL ETHER
ETHYLEME GLYCOL MONOPHENYL ETHER
OIETHYLENE GLYCOL NONOMETHVL ETHER
OIETHYLEMf GLYCOL 01 ETHYL ETHER
ETHYLEME GLYCOL MON08UTYL ETHER
T«I ETHYLEME GLYCOL DIMETHYL ETHER
ETHYLENE GLYCOL NON08UTYL ETHER ACETATE
HEXACMLOROKHZENf
HEXACNLOR08UT AO I EME
HEXACHLOROETHANf
MEXAM
HYOROaUINONf
ISOPHORONE
KALE 1C ANHYDRIDE
METHANOL
METHYL MOMIDf
METHYL CHLORIDE
METHYL CHLOROFORM
METHYL ETHYL KETON8
METHYL HYORAZINf
METHYL ISORUTYL KETOmf
mtJtTUVl ICfWAHAW
WC T NT L I SQCTMMTB
METHYL NETNACRYLATf
METHYL TERT-aum ETHER
METHYLENE CHLORIDE
NfTNYLENi OIPMENYL OIISOCYANATE**
4,4-METHYLENEOIANILIMf
MAPHTHALEMB
NITROMNZM
4-NITROPttCNQL
2-NITROPROPAMI
PHENOL
P-PHEMYLEKEOIAMINf
PHOSGENE**
PMTHALIC AHNYMIM
'« Law Conetant,
H 3 25 C
34.1100000
65.3800000
0.0001051
13.2280793
312.2300000
0.0187000
1.9471244
0.0409170
0.0358404
0.0986300
0.0026793
0.1218685
0.0012481
0.0837494
0.0405801
0.0474169
0.0037600
0.0022577
0.1189224
0.0292288
0.0025951
0.2744400
94.4500000
572.2300000
443.8900000
42447.0100000
0.0000800
0.3682100
0.0121451
0.2885032
381.0578800
490.0000000
944.4700000
7.2200000
0.0248008
21.6700000


7.8317700
30.8401800
164.4500000
0.0026600**
0.0284900
24.8300000
1.3300000
0.0044400
6.6111800
0.0722000
0.0007700
780.0225300**
0.0441508
N • ataVmel fraction
BASIS
Experimental
Experimental
VLE Data
VLE Data
Experimental
Experimental
VLE Data
VLE Data
UN IF AC
Solubility Data
UN IF AC
UN IF AC
UN IF AC
UN IF AC
Cor relation
UN IF AC
Solubility Data
UN IF AC
UN IF AC
VLE Data
UN IF AC
Solubility Data
Experimental
Experimental
Experimental
Experimental '
Solubility Data
Solubility Data
UN IF AC
VLI Data
solubility Data
Experievntal
Experimental
Experimental
UN IF AC
Expertawntal

H VVC * 1 Ofi Mi %It KaUwt.
Solubility • Estiamtaal
Solubility Data
Experimental
Solubility • Estiamtaa)
Solubility Data
Experimental
Experimental
Solubility Data
Solubility Data
Experimental
Solubility Data
Solubility Data
Solubility Data
       3-75

-------
                                     TABLE 3-20. (cont.)
NO  FOUMULA
HAW
tianryU Law Constant,   H • ata/aal fraction

       N » 23 C        lASIS
101 C3H402
102 C3N60
103 C3M6CL2
104 C3H60
109 C6M402
106 C8M8
107 C2H2CL4
108 C2CL4
109 C7H8
110 C7H10M2
111 C9N6N202
112 C7H9M
113 C6K3C13
114 C2H303
115 C2HCU
116 C6M3O30
117 C6N15N
118 C8H18
119 C4M602
120 C2M30.
121 C2H2CU
155
14*
123 C8N10
124 C8M1Q
123 C8H10
IETA**«0*IOUCTONf
WWIONAlDSHYOt
WWmEIH OICNUXIDE
PtgaYLEHf OX IDE
QUINONf
STYRENf
1,1.2,2-TETIACMLOtOETHAJH
TETMCNLOROETNYLElll
TOLUEMt
2,4-TOLUCNf OtAMiNf
2,4-TOLUENf OIISOCTAHATI"
O-TOLUIDIM
1 ,2,4-TRICNLOKOMNZEIM
1,1,2-T»ICMLOtOfTHAJ«
rmCNLOROtNYlEIK
2,4,5-TIICHLOWPHfKX
TRUTNYUMINf
2,2,4-T«!MITHYLKMTANC
VIMYl ACETATE
VINYL CHLiWlOl
VINYLIOENf CMLMIOI
Wf aTilatfl f t4l JL H f VTt •atC«l ^
XTkCNK* (iSQMCK* * nlXlUREvl
M-XYLENC
0-XYLEMf
P-XYlEHf
0.0063801
3.3224900
138.7100000
19.7742986
0.0376800
144.7133400
13.8900000
983.3400000
336.6700000
0.0000742
0.0091900**
0.1344600
106.6700000
43,7700000
566.6700000
0.4841100
6.9428000
183431.3318600
28.2111800
1472.2300000
1438.9000000


413.3400000
270.5600000
413.3400000
UNIFAC
Solubility Oata
Expariawttal
VIE Oata
Solubility Oata
Solubility Oata
Expariaantal
Expariaantai
Expariaantal
UNIFAC
Solubility • Eatia*tad
Solubility Oata
Expariawital
Expariaantal
Enpariaantal
Solubility Oata
Solubility Oata
Solubility Oata
Solubility Oata
Expariaantal
Exparia»ntal


Expariaantal
Expariaantal
Expariaantal
Notaa:

1. * • E*t1natad valuta for eo*ffiei*nta in vapor praaaura actuation.

2. ** * Raacta Mitd Matar.
3. For baaia of UN IF AC,  tha aatiaation of tha activity eoafficiant at  infinita dilution nakaa uta of
tha group contribution contribution aathod uaing tha UK IF AC aquationa (CMdUnf, J., *. Raaauuan and
A. Fradanalund, Ind. En«.  Cham, •roeaaa Oaa. Oav.,  2L 11« (1982»>

4. For baaia of Solubility - Estiaatad, tha aatiMtion of natar aolubility aakaa uaa of axpariaantal
oata which  ia  availaMa on  rafaranca  coMpounda  that  ara vary cloaa  In a*lacular •truetura  ta tha
compound of intaraat. Tha addition of  a avlacular  iroup (or araupa) ta tha lafaianca caacvund than
providaa a nalaoular atructura that ia (oantical ta tha avlaauiar atructura af tha eaapaund of intaraat
(lot S • lot 8^, * AOraup).
                                              3-76

-------
       TABLE 3-21. CORRECTION OF HENRY'S LAW FACTOR FOR A TEMPERATURE
                              DIFFERENT FROM STANDARD

                                     6
                                   106     T2    P\  T2

where:

       HI = Henry's Law Constant at standard temperature, atm-m3/mol
       H2 = the Henry's Law Constant at the actual temperature, atm-m3/mol
       P i = the compound vapor pressure at standard temperature, atm
       P*2 = the compound vapor pressure at actual temperature, atm
       TI = the standard temperature, 298 °K
       T2 = the actual temperature, °K
     MW = the average molecular weight of the liquid, g/mole
       106 = the density of water, g/m3

Source:  Lyman, Warren J., William Reahl, and David Rosenblatt. Handbook of Chemical Property
                      Estimation Methods. McGraw-Hill Book Company, New York, New York,
                      1982.  Section 14, pp. 3-25.

To convert from H in atm/vol fraction to:

H in atm/ (mol/m3), divide by 55,556
H in mmHg/mol fraction, multiply by 760
H in psia/mol fraction, multiply by 19.7
H in kPa/mol fraction, multiply by 101.325
H in kPa/mol/m3), multiply by 101.325/55,556
                                           5-77

-------
3.3  REFERENCES

 1.  Evaporative Loss From Fixed-Roof Storage Tanks, Bulletin No. 2518, Second Edition, American
       Petroleum Institute, Washington, DC, October 1991.

 2.  Estimating Air Toxics Emissions From Organic Liquid Storage Tanks, EPA-450/4-88-004, U. S.
       Environmental Protection Agency, Research Triangle Park, NC, October 1988.

 3.  Barnett, H.C., et al., Properties of Aircraft Fuels, NACA-TN 3276, Lewis Flight Propulsion Laboratory,
       Cleveland, OH, August 1956.

 4.  Petrochemical Evaporation Loss From Storage Tanks, Bulletin No. 2523, First Edition, American
       Petroleum Institute, Washington, DC, 1969.

 5.  Evaporation Loss From Internal Floating Roof Tanks, Bulletin No. 2519, Third Edition, American
       Petroleum Institute, Washington, DC, June 1983.

 6.  Manual of Petroleum Measurement Standards: Chapter 19-Evaporative Loss Management, Section 2—
       Evaporative  Loss from Floating Roof Tanks, Draft Publication, American Petroleum Institute,
       Washington, DC, December 1994.

 7.  SIMMS Data Base Management System, U. S. Environmental Protection Agency, Research Triangle
       Park, NC.

 8.  Comparative Climate Data Through 1990, National Oceanic and Atmospheric Administration,
       Asheville, NC, 1990.

 9.  Input for Solar Systems, National Oceanic and Atmospheric Administration, Asheville, NC, November
       1987.

10.  Evaporation Loss From External Floating Roof Tanks, Bulletin No. 2517, Third Edition, American
       Petroleum Institute, Washington, DC, February 1989.

11.  Ferry, R. L., Documentation of Rim Seal Loss Factors for the Manual of Petroleum Measurement
       Standards, American Petroleum Institute, Washington, DC, April  1995.

12.  Memorandum from A. Parker, R. Neulicht, Midwest Research Institute, to D. Beauregard, U. S.
       Environmental Protection Agency, Fitting Wind Speed Correction Factor for External Floating Roof
       Tanks,  September 22, 1995.

13.  Memorandum from R. Jones, et al., Midwest Research Institute, to D. Beauregard, U. S. Environmental
       Protection Agency, Final Fitting Loss Factors for Internal and External Floating Roof Tanks, May
       24,1995.

14.  Memorandum from A. Parker, Midwest Research Institute, to D. Beauregard, U. S. Environmental
       Protection Agency, Final Deck Fitting Loss Factors for AP-42 Section 7.7, February 15, 1996.
                                              3-78

-------
15.  Memorandum from A. Marshall, Midwest Research Institute, to D. Beauregard, U. S. Environmental
       Protection Agency, Review of New Loss Factor Data, December 20, 1996.

16.  Use of Variable Vapor Space Systems to Reduce Evaporation Loss, Bulletin No. 2520, American
       Petroleum Institute, New York, NY, 1964.

17.  Yaws, C.L., Henry's Law Constants for HAPs, U. S. Environmental Protection Agency, September 30,
       1992.
                                              5-79

-------
            4.0 EMISSION ESTIMATION PROCEDURES FOR FIXED ROOF TANKS
       Two emission estimation procedures were examined for estimating standing storage or breathing
loss emissions from fixed roof tanks. The first equation is a version of that developed by the American
Petroleum Institute (API) in 1962.1  This breathing loss equation was used in both the industry and
regulatory communities for 30 years. More recently, API proposed a new equation for predicting
breathing losses from fixed roof tanks.2 This chapter presents the results of a comparative evaluation of
the two equations to determine the most accurate method for predicting breathing losses from fixed roof
tanks.  In order to evaluate the effectiveness of the equations in predicting actual breathing losses, the
equations were used to estimate breathing losses in two scenarios.  In the first scenario, actual parameters
recorded during previous emissions  testing are the variables used in the estimating equations. In the
second scenario, default values that  are likely to be used in the regulated community are the  variables
used in the estimating equations. In addition to evaluating the predictive abilities of each of the
equations, the sensitivity of each equation to various parameters was analyzed using default  values and a
typical range of data points for each parameter.

4.1 BREATHING LOSS EQUATIONS

       The standing storage or breathing loss equation that has been used historically is based on the
assumption that the breathing loss is a function of the vapor pressure of the stored liquid, tank diameter,
vapor space outage, ambient temperature, and the tank paint color and condition. The old breathing loss
equation*  is as follows:
                                                                                          (4-1)
where:
         Lb =  breathing loss, Ib/yr
        Mv =  stock vapor molecular weight, Ib/lb-mole
         P  =  vapor pressure of stored liquid at bulk liquid conditions, psia
         PA =  average atmospheric pressure at tank location, psia
         D  =  tank diameter, ft
         H  =  average vapor space height, including roof volume correction, ft
         AT =  average ambient diurnal temperature change, °F
         Fp =  paint factor, dimensionless
         C  =  adjustment factor for small diameter tanks, D<30 ft, dimensionless
         KG =  product factor, dimensionless
*Note: This equation was updated as described on the following page in Supplement E of the 4th edition
of AP42, October 1992.
                                             4-1

-------
The new breathing loss equation proposed by API was developed based on theoretical equations and is a
function of the following parameters:

        1. KE = the vapor space expansion factor;
        2. Ks = the vapor space saturation factor;
        3. Vv = the tank vapor space volume; and
        4. Wv = the stock vapor density.

The expressions describing each of the above parameters were derived from theoretical equations and are
themselves functions of tank diameter, vapor space outage, stock molecular weight, vapor pressure, and
environmental conditions. The expression developed for Ks contains a mass transfer coefficient for
which no value is available. Therefore, a correlation based on EPA, Western Oil and Gas Association
(WOGA), and API data was developed to describe Ks.2 The theoretical derivation was used as a guide in
developing the data-based correlation.  Thus, KE, Vv, and Wv are based on theoretical derivations and Ks
is based on actual test data.  The following is the new breathing loss (standing storage loss) equation that
has been developed by API:
                                     Ls = 365VvWvKEKs                                 (4-2)
where:

         Ls =  standing storage loss, Ib/yr

        Vv=
W               My PyA
          v —  	
               ATV
                TLA      PA-PVA
         KS =
               l + 0.053P^/fra
        365 =  constant, the number of daily events in a year, (year)"1
          7t=  constant, 3.1459
          D =  tank diameter, ft
       HVO =  vapor space outage, ft
        Mv =  stock vapor molecular weight, Ib/lb-mole
        PVA =  vapor pressure at daily average liquid surface temperature, psia
          R =  ideal gas constant, 10.731 psi-ft3/lb-mole-°R
        TLA =  daily average liquid surface temperature, °R
       A TV =  daily vapor temperature range, °R
       APV =  daily vapor pressure range, psia
       APB =  breather vent pressure setting range, psia
        PA =  atmospheric pressure, psia

       Of the above parameters, D, Hvo, Mv, PV, APB, and R are readily available or can be determined
using basic assumptions, tables, or figures.  The remaining parameters, however, are not readily available
and are themselves functions of the following variables:

        TAA = daily average ambient temperature, °R

                                              4-2

-------
       Of the above parameters, D, HVo, MV) PV, APB, and R are readily available or can be determined
using basic assumptions, tables, or figures. The remaining parameters, however, are not readily available
and are themselves functions of the following variables:

        TAA = daily average ambient temperature, °R
          TB = liquid bulk temperature, °R
          a = tank paint solar absorptance, dimensionless
           I = daily total solar insolation, Btu/ft2-d
The above values may not be readily available, but default values are provided by API.

4.2 COMPARISON OF PREDICTIVE ABILITY OF TWO EQUATIONS

       The  predictive ability of the two breathing loss equations was evaluated under two sets of
circumstances.  In the first situation, data collected for the WOGA, EPA, and API studies were used in
each equation to generate breathing loss  emission estimates for comparison against actual measured
values. In the second situation, typical default values that are  likely to be used by the regulated
community were used in each breathing loss equation.  The breathing loss calculated by each equation,
based on default values, was then compared against actual measured values.

       The  emissions estimated by the two breathing loss equations were compared in Section H of the
Documentation File for API 2518, second edition.3  The WOGA, EPA, and API test data that were used
in each equation are identified in Section H. The API test data used in this analysis are the same as the
test data presented in Section H of the API 2518 documentation file and are for tanks storing fuel oil
No. 2. The WOGA test data used in this analysis are taken from the document entitled "Hydrocarbon
Emissions From Fixed Roof Petroleum Tanks," prepared by Engineering Science for WOGA in
July 1977.4  This report is the original report containing the data and field data sheets from the emission
testing.  The data in the Engineering Science report  are, in some instances, different from those presented
in Section H of the API 2518 documentation file. However, the data in the Engineering Science  report
were used because this report is assumed to be the most accurate source.  These data are for tanks storing
crude oil.  The EPA data used in the report are also slightly different from those presented in Section H  of
the API 2518 documentation file.  The EPA data used in this analysis were taken from the document
entitled "Breathing Loss Emissions From Fixed-Roof Petrochemical Storage Tanks" (Third Draft),
prepared by  Engineering Science for EPA in December 1978.5 As with the WOGA data, this document
contains the  original field data sheets and is therefore assumed to be the most accurate source of
information. These data describe tanks storing several different petrochemicals.
                                             4-3

-------
4.2.1  Predictive Ability-Actual Data

        A summary of breathing loss estimates calculated by the two equations is presented in Table 4-1
for fuel oil (API data base), Table 4-2 for crude oil (WOGA data base), and Table 4-3 for petrochemicals
(EPA data base). The breathing loss values presented were calculated using actual measured parameters,
not default values. Also presented in these tables are the stock type contained in test tanks, the measured
vapor pressure of the stock, and the actual breathing loss emissions measured.  The measured breathing
losses were compared to those calculated using the two emission estimation equations by establishing the
bias, standard deviation, and root mean squared error of the models, which are the emission estimation
equations.  The bias is used to describe the systematic error in a certain model, for example, whether the
model consistently overpredicts or underpredicts the actual measured emission loss. The standard
deviation describes the precision or reproducibility of the model.  If the standard deviation is large, this
indicates that there is a lot of scatter in the data base, or in the case of the breathing loss equations, that
the equation has poor reproducibility.  The root mean squared error is an expression that is used to
incorporate both types of error in an equation, the bias and the standard deviation. The root mean squared
error of each model is expressed in one number. The bias, standard deviation,  and root mean squared
error are all expressed in pounds per day (Ib/d) for each emission estimation procedure.

        From the first "row" of data presented in Table 4-4, which corresponds to the aggregate of data
from all three data bases, the revised equation (Equation 4-2) has less bias and a smaller variability than
Equation 4-1. Based on all tanks in the data base, Equation 4-1 underestimates actual measured emissions
by approximately 49 Ib/d and has a standard deviation  of 116 Ib/d. The root mean squared error is
126 Ib/d.  In comparing measured breathing losses to those calculated using the revised breathing loss
estimating equation, and considering all tanks in the data base, Equation 4-2 underestimates by only
2.5 Ib/d and has a standard deviation of 52.9 Ib/d. The root mean squared error of Equation 4-2 is also
52.9 Ib/d.

        It  is imperative that, when reviewing the data presented in Table 4-4, the reader note how the two
equations  predict breathing losses for individual stock types, and consider the number of data points
associated with each stock type. For example, the fuel oil No. 2 (API data base) and crude oil (WOGA
data base) comparisons are based on 10 and 8 data points, respectively.  The other comparisons-
isopropanol, ethanol, acetic acid, ethyl benzene, and cyclohexane (all from the EPA data base)-only
comprise 2 or 3 data points each.  If one equation happens to show better predictability in a stock type
that has a  large number of data points, as is the case with Equation 4-2 for crude oil, the overall
(aggregate) results will be biased.  Therefore, even though the overall (aggregate) results indicate that
Equation 4-2 is a better predictor of breathing losses, the two equations are for the most part comparable,
except in the case of crude oil, in which Equation 4-2 clearly is a better predictor (both equations
underpredict, with Equation 4-1 being a substantially worse predictor than Equation 4-2).

        A comparison of the calculated breathing losses as a function of vapor pressure is shown in Table
4-5.  From the table, Equation 4-1 appears to be a better predictor for liquids having a vapor pressure of
less than 0.5 psia and Equation 4-2 appears to be a better predictor for liquids with vapor pressures above
2 psia. In the vapor pressure range between 0.5 and 2.0 psia, the two equations are comparable
predictors. However, based on the product type analysis, the difference between the two equations may
be more a result of product type than vapor pressure. For example, no data points for crude oil fall within
the vapor pressure range less than 0.5 psia, which happens to coincide with the vapor pressure range
where Equation 4-1 appears to be a better predictor.  On the other hand,  four data points from crude oil
fall within the vapor pressure range greater than 2.0 psia where Equation 4-2 is a better predictor.
                                               4-4

-------
4.2.2  Predictive Ability-Default Values

        It is assumed that many users of the breathing loss equation have only basic information about a
particular tank. This information would include the physical characteristics of the tank (size, color), the
location of the tank, and the tank stock. Therefore, many of the parameters needed to use either breathing
loss estimation procedure are default values provided in AP-42 or API 2518.  These default values may
be given specifically or presented in tables and figures. In the case of Equation 4-1, values of Mv,  PV,
TAA, AT, and Fp are obtained from tables and figures. The value of H is obtained by assuming H equals
one-half of the actual tank shell height.  In the case of Equation 4-2, the same assumptions as for
Equation 4-1 are made.  The vapor space height, H,  is denoted as the vapor space outage, HVo, in
Equation 4-2.  However, additional parameters of a  and I are based on tables provided in API 2518; these
parameters would not typically be measured.

        A summary of the estimates calculated by the two breathing loss estimating equations (using
default values) is presented for the WOGA and EPA data bases in Tables 4-6 and 4-7. In the API testing,
the test tank was located indoors. Therefore, no default meteorological data could be assumed and no
comparison with the WOGA and EPA default value data bases is possible.  Also presented in Tables 4-6
and 4-7 are the stock type contained in the test tanks and the actual breathing  loss emissions measured.
Overall, as compared to measured data, Equation 4-1 (using default values) underpredicts by
approximately 67.9 Ib/d and has a standard deviation of 127.9 Ib/d. The root mean squared error is
144.8 Ib/d.  Equation 4-2 (using default values) underpredicts by approximately 16.2 Ib/d and has a
standard deviation of 85 Ib/d.  The root mean squared error is 86.5 Ib/d.

        A summary of the bias, standard deviation, and root mean squared error is presented in Table 4-8
for three cases. The first case compares the two equations using all actual data. The second compares the
two equations  using actual and default values, when the fuel oil data base is excluded (this is done
because default values were not entered for the API-fuel oil testing).  In the third case, the two equations
with actual and default data are compared, excluding the fuel oil and crude oil data. This was done
because the revised equation, Equation 4-2, is obviously a better predictor in the case of crude oil.

        A summary of the comparison between the two equations by product type is provided in
Table 4-9. There are no important differences between the equations' performance for any product except
crude oil.  Using the default values for the parameters in the equations improved the performance of
Equation 4-1 and resulted in a slightly worse performance for the revised equation, Equation 4-2.
However, for crude oil, Equation 4-2 is still better than Equation 4-1.  The major difference is in the bias.
The precision of the equations is comparable.

        In summary, with the likely use of the default values for the parameters in the equations, the
Equation 4-1 is a slightly better predictor for the products in the data base with the  exception of crude oil.
For crude oil,  the revised equation, Equation 4-2, is better.

4.3 SENSITIVITY ANALYSIS

        As indicated by the above discussion,  both breathing loss equations are basically functions  of the
same variables; they differ in the extent to which they depend on those variables. In order to assess the
extent to which each of the equations depends on its independent variables, a sensitivity analysis was
conducted.  The results of that analysis are discussed below.

        The sensitivity of Equation 4-1 was evaluated with respect to molecular weight (Mv), vapor

                                              4-5

-------
pressure (Pv), vapor space height (H), average ambient temperature change (AT), and the paint factor (Fp).
 A sensitivity analysis had previously been done for the new breathing loss equation developed by API, as
documented in Section G of the documentation file for API 2518, second edition.2  The parameters that
were evaluated included Mv, PV, H, and AT as well as the following parameters:

       1. Tank diameter (D);
       2. Solar insolation  (I);
       3. Solar absorptance (a);
       4. Ambient temperature;
       5. Breather vent pressure and vacuum settings; and
       6. Reid vapor pressure.

       The sensitivity of the two breathing loss equations to the various parameters was evaluated by
maintaining all parameters constant except for the one being evaluated. The baseline case that was used
for the sensitivity analysis was the WOGA test conducted on May 18,  1977, involving a crude oil storage
tank.4  Parameters of diameter, Reid vapor pressure, stock type, tank location, and tank color were
obtained from field data sheets. These parameters are the parameters that are assumed to be available to
all end users. For all other parameters (molecular weight, vapor pressure, vapor space height, temperature
data, and solar insolation), it was assumed that default values would be used by the end users. A
summary of the baseline values for the sensitivity equation is provided in Table 4-10.

       For each equation, the sensitivity  evaluation of each parameter was conducted using the
following procedure. First,  for each parameter, the default value was identified. Then, based on actual
data (EPA, WOGA, and API data bases), the amount by which the default value is likely to vary was
determined.  For example, the default value for the molecular weight of crude oil is 50 pounds/pound-
mole (Ib/lb-mole). Actual data indicated that the measured molecular weight may be as high as 70 Ib/lb-
mole.  Therefore, the default value range was assumed to be 20 Ib/lb-mole. Once the default range was
determined,  the corresponding change in the calculated breathing loss was calculated using each equation.
 For example, in a 20 Ib/lb-mole range of molecular weight, the breathing loss calculated by Equation 4-1
differs by approximately 64 pounds. The  same analysis was conducted for each parameter using both
breathing loss estimating equations. The results are summarized in Table 4-11 and Figures 4-1 to 4-11.
Note that because one baseline case is being used in the sensitivity analysis, the actual measured
breathing loss is the same in all cases, 574 Ib/d, as shown in Figures 4-1 to 4-11. Also note that all figures
have different scales.

       As previously stated, the  actual molecular weight of crude oil may vary by 20 Ib/lb-mole. The
sensitivity of Equation 4-1 to changes in the molecular weight (all other parameters held constant) is
provided in Figure 4-1. As  indicated in Figure 4-1, the relationship between molecular weight and
breathing loss is linear. Using the default  value of 50 Ib/lb-mole, a breathing loss of 159.3 Ib/d is
estimated. Based on Figure 4-1, a difference of 20 Ib/lb-mole in the molecular weight would result in
breathing loss emissions ranging between  159.3 and 223.1, a difference of approximately 64 Ib/d in
breathing loss emissions. Figure  4-2 shows the sensitivity of the revised equation to variance in
molecular weight. Again, the relationship is linear.  Using the default value of 50 Ib/lb-mole, a breathing
loss of 377.9 Ib/d is estimated. Based on a difference of 20 Ib/lb-mole in the molecular weight value used
in the equation, the breathing loss would range between 377.9 and 529.1 Ib/d, a difference of
approximately 150 Ib/d.

       The vapor pressure  of the stored material is  also a parameter for which a default value from a
reference table is used. (The default values are a function of temperature and therefore vary with the
stored  liquid temperature.) The WOGA data base indicates that the Reid vapor pressure of the same type

                                              4-6

-------
of stock, stored in the same tank, can vary considerably.  Thus, the vapor pressure, which is a function of
the Reid vapor pressure, will also vary even at a constant stored temperature. For example, in WOGA
tests 7b and 8a, breathing losses from the same tank storing crude oil were measured.  The Reid vapor
pressure during the first test was measured as 1.2 psi. During the second test, 3 months later,  the Reid
vapor pressure was measured as 3.4 psi. Assuming a constant default value for ambient temperature, the
true vapor pressure varies from 0.5 psi to 2.0 psi, a difference of 1.5 psi.  The sensitivity of Equation 4-1
to changes in the stock vapor pressure is linear, as shown in Figure 4-3. Based on a 1.5 psi difference in
the vapor pressure value used in the equation, the difference in the breathing loss in the 0.5 to 2.0 psi
range varies between 37.5 and 103.9 Ib/d, a difference of approximately 64 Ib/d. Performing the same
analysis with the revised equation, the difference in the calculated breathing loss ranges between 79.2 and
236.6 Ib/d, a difference of approximately 158 Ib/d. The sensitivity of the revised equation to vapor
pressure changes is also linear, as depicted graphically in Figure 4-4.

        A value for the average vapor space height, or outage (HVo), is required as an input for both
equations  (it is required as H in Equation 4-1). If this value is not available, it is recommended that HVo
be assumed  to equal one-half the tank height. The API 2518  bulletin does not direct the user as to what
value to use if the actual HVo is unavailable.  Based on the available data, a typical tank height is 40  feet;
thus, a typical value for HVo would be 20 feet. In the baseline case, the value of HVo is 20.75 feet, one-
half of the actual tank height of 41.5 feet. Based on the HVo value measured at tanks with a height of
approximately 40 feet,  the actual HVo varied from approximately 5 to 35 feet.  This reflects a  difference in
+ 15 feet from the default value. The sensitivity of the two  equations to various HVo values, all other
parameters held constant, is depicted graphically in Figures 4-5 and 4-6, respectively. As indicated in
both figures, the relationship between HVo and breathing loss is nonlinear. Thus, in using  either equation
to calculate breathing loss, the difference between measured and calculated emissions depends on whether
the value used in the equation is less than or greater than the default value.  Using Equation 4-1, if a value
of 5 feet (-15 feet) is assumed for HVo, a breathing loss of 77.1 Ib/d is obtained. Using the default value
of 20 feet, a breathing loss of 156.4  Ib/d is obtained; the difference in breathing loss emission estimates is
approximately 80 Ib/d. If a value of 35 feet (+15 feet) is assumed for HVo, a breathing loss of 208 Ib/d is
estimated. As compared to the estimate of 156.4 Ib/d using the default value, the difference in breathing
loss is approximately 50 Ib/d. Performing the same analysis using Equation 4-2, the difference in
breathing loss emissions is again found to depend on whether the value for HVo is less than or greater than
the default value.  If the value used for HVo is 5  feet, the breathing loss is estimated as 226 Ib/d. Using a
default value of 20 feet, the breathing loss estimated by the revised equation is 374.9 Ib/d,  a difference of
approximately 150 Ib/d. If  the value used for HVo in the revised equation is 35 feet, a breathing loss of
413.9 Ib/d is estimated. The difference as compared to the  default value is approximately 40 Ib/d.

        Both of the estimating equations also incorporate a value for average daily ambient temperature
(AT) into the emission  estimating equation. In Equation 4-1,  the breathing loss is  directly  related to  the
value of AT raised to the 0.5 power; the relationship is approximately linear.  In the revised equation, the
value of AT is not used directly in the emission estimating equation.  It is, however, used to calculate
values for the daily vapor temperature change and the liquid surface temperature; these parameters are
used directly in calculating  the breathing loss emissions. Thus, the relationship between AT and breathing
loss is different for each equation. The tank in the baseline case, which is considered in this analysis, is
located in California.  The default AT value of 15°F was obtained by using meteorological data for the
areas that are contained in both AP-42 and API 2518.  Based on actual measured  temperature changes of
20° to 40°F in the same area, the value of ambient temperature change may vary by 25°F. The
sensitivity of the two equations to changes in the value of AT is shown graphically in Figures 4-7 and 4-8,
respectively. As indicated by these  figures, the relationship between AT and breathing loss is
approximately linear in both cases.  Using Equation 4-1 and a default value of 15°F, the breathing loss is
estimated as 159.3 Ib/d. Based on an increase of 25 °F in the  value of AT, the emission rate predicted by

                                               4-7

-------
the old equations is 259.3, a difference of 100 Ib/d. Using Equation 4-2 and a default value of 15 °F, a
breathing loss of 377.9 is estimated. With an increase in AT of 25 °F, the calculated breathing loss is
642.8 lb/d, a difference of approximately 265 Ib/d.

        The effect of tank color on the breathing loss emission rate is accounted for in both equations. In
Equation 4-1, the breathing loss is directly related to the value for the paint factor (Fp) that is used in the
equation.  The value for Fp depends on tank shell and roof paint color and paint condition. The values for
Fp were developed based on testing that involved measurement of solar reflectance and an evaluation of
the relationship between solar reflectance and emission rates.  In the revised equation, the effect of tank
shell and roof paint color and paint condition is described by the solar absorptance (a) of the paint.  A
linear relationship  exists between the values of a and Fp. The values of a provided in API 2518, second
edition, were calculated using this relationship.  However, Equation 4-2 incorporates the a value much
differently than Equation 4-1 uses the  paint  factor, Fp. A value of a is not used directly in calculating the
breathing loss emission rate. It is used to calculate the liquid bulk temperature, the average liquid surface
temperature, and the vapor temperature range.  These parameters are all used in the emission estimating
equation; thus, the relationship between a and the emission rate is complicated, and a and Fp are not
directly comparable.

        For the baseline case used in this analysis, the tank shell was white and the tank roof was light
gray.  The paint condition was reported as good/poor (good condition was assumed). Based on this
information and tabular listings of Fp and a values in AP-42 and API 2518, a value of 1.3 was used for Fp
and a value of 0.355 was used for a. Both the AP-42 and API 2518 documents state that if information is
not known, assume a white shell and roof, with the paint in good condition.  If these conditions are
assumed, a value of 1.0 is used for  Fp and a value of 0.17 for a. The sensitivity of the breathing loss
emission rate calculated by Equation 4-1 to  changes in the value of Fp is linear, as shown in Figure 4-9.
Using the actual value of 1.3 for  Fp, a breathing loss of  159.3 lb/d is estimated. Using the default value of
1 for Fp (a change of 0.3), the breathing loss is  estimated as 122.6 lb/d. Using the default value, the
breathing loss emission rate varies by approximately 37 lb/d.

        The sensitivity of the breathing loss emission rate calculated by Equation 4-2 to changes in the a
value is also linear, as shown in Figure 4-10. Using the actual value of 0.355 for a, a breathing loss of
377.9 lb/d is estimated. Using the default value of 0.17 for alpha, the breathing loss is estimated as
259.8 lb/d, a difference of 188  lb/d.

        One parameter that is used in Equation 4-2 but is not used in any way in Equation 4-1 is the daily
solar insolation (I). The value of I is not used directly in the revised equation, Equation 4-2, to estimate
breathing loss emissions but is used to calculate bulk storage temperature, average liquid surface
temperature, and the vapor temperature range.  The value of I depends on the tank location. Default
values for I are provided in API 2518, based on the city and State in which the tank is located.  It is
assumed that the default values of I will be used by the regulated community. In the baseline case, a
value of 1,594 Btu/ft2d was used due to location of the tank in California.  Based on actual values for I
that were measured at the same location, the values of I may vary from 933 to 2,050 Btu/ft2d.  The
sensitivity of the revised equation to changes in the value of I is linear, as shown in Figure 4-11. Based
on a variance of 600 Btu/ft2d as the value of I, the breathing loss estimated would range between
336.2 lb/d at I = 1,300 and 421 lb/d at I = 1,900. The difference in the emission rate calculated by the
revised equation varies by approximately 85 lb/d.

        Based on the results of the  sensitivity analysis, it is concluded that for both equations, the
predicted emission rates decrease when a lower value than the default value is used and increase when a
higher value than the default value  is used.  In general, the revised equation, Equation 4-2, for estimating

                                               4-8

-------
emissions from fixed roof storage tanks, is more sensitive to variations in the values of molecular weight,
vapor pressure, temperature change, and in some instances, vapor space height, than Equation 4-1.  Both
equations account for the effect of tank shell and roof paint color and condition on the breathing loss
emission rate.  In Equation 4-2, the effect  is described by alpha. In Equation 4-1, the effect is described
by Fp.  Variations in the value of a will affect the resultant calculated emission rate much more than
variations in the value of Fp used in Equation 4-1 to describe the same effect.  However, Fp and  a are not
directly comparable.  Finally, a new parameter, I, is introduced into the Equation 4-2.  The breathing loss
emission rate calculated by Equation 4-2 is sensitive to changes in the value of I that is used.

        Based on the sensitivity analysis,  it is concluded that use of Equation 4-2 would result in more
uncertainty or variability in the predicted value if any of the variables are subject to uncertainty or error.

4.4 CONCLUSIONS AND RECOMMENDATIONS

        For all stock types other than crude oil, Equation 4-1 is an adequate predictor of breathing loss.
For crude oil stocks, the revised equation  is a clearly superior predictive equation.  In the case of crude
oil, the Equation 4-2  underpredicts by 55.6 Ib/d, as compared to the underprediction of 179.5 Ib/d of
Equation 4-1.  Equation 4-2 is very sensitive to the value of a that is used.  In a situation in which the
default value of a is incorrect, an underprediction of 118 Ib/d may occur. Adding this factor to the
underprediction of 55.6 Ib/d in the case of crude oil, Equation 4-2 underpredicts by 173.6 Ib/d,
comparable to Equation 4-1. However, it  is recommended that the revised  equation, Equation 4-2, be
incorporated in the AP-42 for estimating emissions from fixed roof tanks as it is a better predictor for
crude oil and is comparable to Equation 4-1 for other stored materials.
                                              4-9

-------
X
o
"D

v~
4)
a
T3
C
D
O
a
en
c
o
1)
m
600



550



500



450



400



350



300



250



200



150



100
                   40
                                                                     L-jneasuned = 574 Ib/day
                                  60
80
                                    Molecular Weight, pound/pound mol

                           D   Lb, AP-42 #/day      +   Lb, measured  If/day
100
Figure 4-1. Sensitivity of Equation 4-1 to changes in molecular weight (Mv).

-------
          800

-------
           700
X
o
/c3ay
01
0


O»
C
o
0)
L
m
           300
           200
           100
             0
          0
                                            1                       2


                                            Vapor Pressure (Pv), psia
                   D   Lb AP-42,
                                                         +    Lb measured, #/day
                               Figure 4-3. Sensitivity of Equation 4-1 to changes in vapor pressure (Pv).

-------
00
                   X
                   o
                   13

                   v_
                   
-------
 o
 TO
 a
 C
 3
 O
 a
01
c
o
0>
t.
DQ
           600
          500
          400
          300
          200
          100 -
                                                                             I* measured = 574  Ib/day
                                             20
40
                                         Vapor Space  Height (H), feet

                            a   Lb AP-42, #/day      +   Lb measured,  #/day
                                                                                                60
                          Figure 4-5. Sensitivity of Equation 4-1 to changes in the vapor space outage (HVo).

-------
          600
          550
                                                                           I^measured = 574 Ib/day
O
t)
a
T3
C
3
O
a
in

O
_i
o>
c
r
•*-<
o
i_
          500
          450
          400
          350
          300
          250  -
          200
                                            20
                                                                      40
                                        Vapor Space Height (H), feet

                             a   Lb API, I/day       +    Lb measured,  #/day
                                                                                               60
                        Figure 4-6. Sensitivity of Equation 4-2 to changes in the vapor space outage (HVo).

-------
o
TD

-------
 0)
 Q.
C
3
O
a
in
W)
O
o>
c
o

-------
oo
                  o
                  "O
                  ^
                  Q)
                  a

                  (fl
                  "O
                  c
                  3
                  O
                  a
                  w
                  w
                  o
Qi
c
'_c
*-*
o
 600



 550



 500



 450



 400



 350



300



250



200



150



100
                                                                                             L. measured = 574 Ib/day
                                                      1.2
                                                                        1.4
                                                             1.6
                                                         Paint Factor (Fp),  dimensionless

                                              a   Lb AP-42, #/day      +   Lb  measured, #/doy"
                                                                                                          1.8
                                          Figure 4-9. Sensitivity of Equation 4-1 to changes in the paint factor values (FP).

-------
CD


D
U
111
O
tfl
-a
.5
0
Ct
_J
in
O'
c
-C
"b
in




tlLJLf
-56O
J54O
S20
SGO
480
46O
44O
42O
3SO
- 36O
34 O
-320



1 1 f 1 h 1 Jl 1 i - --{-
11 measured -574 llj/Oay

-
-
^*r~~^*
^ar-^**"
J^^^^^'
^^^^^3'^'
: ^-«-"~~

1 1.2 1.4 1.6 },S
(Thousands)
DaHy Soltir l^moltjlion, (BTU/aq.ft.aay)
1 1 i-ti Apt, -ft /aay i I h measured, jj',/d
-------
J^
I

o
                    X
                    o
                    0)
                    a
                    c
                    3
                    O
                    a
                    o
                    E
                    '£
                    +-•
                    o

                    L
600


580  -


560  -


540  -


520


500


480


460


440


420


400


380


360


340


320


300


280
                                                                                              L, measured = 574 Ib/day
                                                  D
                           1.2               1-4
                                            (Thousands)

                           Daily  Solar insolation,  (BTU/sq,ft.day)

                        Lb API, #/day       -t-   Lb measured, #/day
                                          Figure 4-11.  Sensitivity of Equation 4-2 to changes in the daily solar insolation factors (I).

-------
TABLE 4-1. FIXED ROOF TANK BREATHING LOSS-COMPARISON OF
          ESTIMATING EQUATIONS-API DATA BASE
Test
description
API-1
API-2
API-3
API-4
API-5
API-6
API-7
API-8
API-9
API- 10
Stock type
Fuel Oil No. 2
Fuel Oil No. 2
Fuel Oil No. 2
Fuel Oil No. 2
Fuel Oil No. 2
Fuel Oil No. 2
Fuel Oil No. 2
Fuel Oil No. 2
Fuel Oil No. 2
Fuel Oil No. 2
Vapor pressure,
psia
0.0049
0.0049
0.0050
0.0051
0.0051
0.0051
0.0075
0.0101
0.0101
0.0101
Breathing loss, Ib/d
Eqn. 4-1
0.0523
0.0575
0.0715
0.0708
0.0701
0.0425
0.0874
0.1302
0.1112
0.0890
Eqn. 4-2
0.011
0.0145
0.0172
0.0171
0.0155
0.0053
0.0374
0.053
0.0397
0.0382
Actual
0.0152
0.017
0.0216
0.0219
0.0179
0.0079
0.0334
0.0502
0.0478
0.0441
TABLE 4-2. FIXED ROOF TANK BREATHING LOSS - COMPARISON OF
        ESTIMATING EQUATIONS - WOGA DATA BASE
Test
description
WOGA-7B
WOGA-8A
WOGA-13A
WOGA-13B
WOGA-16A
WOGA-16B
WOGA-17A
WOGA-17B
Stock type
Crude oil
Crude oil
Crude oil
Crude oil
Crude oil
Crude oil
Crude oil
Crude oil
Vapor pressure,
psia
1.5
0.6
0.8
0.8
2.7
2.7
3.4
3.4
Breathing loss, Ib/d
Eqn. 4-1
136.3
46.5
17.7
14.0
99.8
98.7
122.9
129.2
Eqn. 4-2
254.6
145.2
22.0
19.5
282.7
278.7
448
446
Actual
196
80
129
146
177
256
574
576
                           4-21

-------
TABLE 4-3. FIXED ROOF TANK BREATHING LOSS-COMPARISON OF
         ESTIMATING EQUATIONS-EPA DATA BASE
Test
description
EPA-1A
EPA- IB
EPA-2A
EPA-2B
EPA-2C
EPA-3A
EPA-3B
EPA-5A
EPA-5B
EPA-6A
EPA-6B
EPA-6C
Stock type
Isopropanol
Isopropanol
Ethanol
Ethanol
Ethanol
Acetic acid
Acetic acid
Ethyl benzene
Ethyl benzene
Cyclohexane
Cyclohexane
Cyclohexane
Vapor pressure,
psia
0.65
0.715
0.895
0.895
0.895
0.23
0.23
0.2
0.2
1.97
1.97
1.97
Breathing loss, Ib/d
Eqn. 4-1
9.71
9.04
11.03
12.65
10.35
22.12
27.44
9.67
10.64
38.28
31.08
34.04
Eqn. 4-2
13.35
12.15
12.99
16.73
10.42
45.64
70.5
13.56
15.59
48.64
48.64
48.41
Actual
15
17
6
3.4
5.7
24
45
11
15
20
17
14
                           4-22

-------
TABLE 4-4. COMPARISON OF EMISSION ESTIMATING EQUATIONS
   WITH BREATHING LOSS AS A FUNCTION OF STOCK TYPE
ARRreRate of data
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
Fuel oil No. 2
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
No. of data points
Isopropanol
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
No. of data points
Ethanol
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
No. of data points
Acetic Acid
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
No. of data points
Ethyl benzene
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
No. of data points
Cyclohexane
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
No. of data points
Crude oil
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
No. of data points
Equation 4-1

-48.7
115.9
125.7

0.051
0.013
0.052
10.0

-6.6
1.3
6.8
2.0

6.3
2.1
6.6
3.0

-9.7
7.8
12.5
2.0

-2.8
1.5
3.2
2.0

17.5
2.5
17.6
3.0

-186.9
155.3
243.0
8.0
Equation 4-2

-2.5
52.9
52.9

-0.003
0.004
0.005
10.0

-3.3
1.6
3.6
2.0

8.3
3.6
9.1
3.0

23.6
1.9
23.6
2.0

1.6
1.0
1.9
2.0

31.6
2.4
31.7
3.0

-29.7
95.2
99.7
8.0
                           4-23

-------
 TABLE 4-5. COMPARISON OF EMISSION ESTIMATING EQUATIONS
  WITH BREATHING LOSS AS A FUNCTION OF VAPOR PRESSURE
Vapor pressure range, psia
VP < 0.5
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
No. of data points
0.5 2.0
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
No. of data points
Equation 4-1

-1.8
4.5
4.9
14

-34.5
52.4
62.8
8

-4.5
38.2
38.5
4

-285.8
165.5
330.2
4
Equation 4-2

3.6
8.2
9.0
14

-18.7
60.4
63.2
8

38.3
11.9
40.1
4

-31.9
100.5
105.4
4
TABLE 4-6. FIXED ROOF TANK BREATHING LOSS - COMPARISON OF
 ESTIMATING EQUATIONS - WOGA DATA BASE DEFAULT VALUES
Test
description
WOGA-7B
WOGA-8A
WOGA-13A
WOGA-13B
WOGA-16A
WOGA-16B
WOGA-17A
WOGA-17B
Stock type
Crude oil
Crude oil
Crude oil
Crude oil
Crude oil
Crude oil
Crude oil
Crude oil
Vapor pressure,
psia
1.5
0.6
0.8
0.8
2.7
2.7
3.4
3.4
Breathing loss, Ib/d
Eqn. 4-1
54.7
19.9
7.7
7.7
144.8
144.8
159.3
159.3
Eqn. 4-2
168.9
56.8
11.8
11.8
341.9
341.9
377.9
377.9
Actual
196
80
129
146
177
256
574
576
                           4-24

-------
TABLE 4-7. FIXED ROOF TANK BREATHING LOSS-COMPARISON OF
  ESTIMATING EQUATIONS-EPA DATA BASE DEFAULT VALUES
Test
description
EPA-1A
EPA- IB
EPA-2A
EPA-2B
EPA-2C
EPA-3A
EPA-3B
EPA-5A
EPA-5B
EPA-6A
EPA-6B
EPA-6C
Stock type
Isopropanol
Isopropanol
Ethanol
Ethanol
Ethanol
Acetic acid
Acetic acid
Ethyl benzene
Ethyl benzene
Cyclohexane
Cyclohexane
Cyclohexane
Vapor pressure,
psia
0.65
0.715
0.895
0.895
0.895
0.23
0.23
0.2
0.2
1.97
1.97
1.97
Breathing loss, Ib/d
Eqn. 4-1
9.9
9.9
16.1
16.1
16.1
25.7
25.7
10.2
10.2
43.5
43.5
43.5
Eqn. 4-2
10.1
10.1
20.1
20.1
20.1
33.7
33.7
7.9
7.9
50.4
50.4
50.4
Actual
15
17
6
3.4
5.7
24
45
11
15
20
17
14
                           4-25

-------
TABLE 4-8. SUMMARY OF STATISTICAL ANALYSIS VALUES
Statistical analysis, Ib/d
Bias
Standard deviation
Root mean squared error
All data
Actual data
EQN4-1
-48.7
115.9
125.7
EQN4-
2
-2.5
52.9
52.9
Excluding fuel oil data
Actual data
EQN4-1
-73.1
135.5
153.9
EQN 4-
2
-3.7
64.7
64.8
Default data
EQN 4-1
-67.9
127.9
144.8
EQN 4-
2
-16.2
85
86.5
Excluding fuel oil data and crude oil data
Actual data
EQN 4-1
2.7
10.8
11.1
EQN 4-2
13.6
13.5
19.2
Default data
EQN 4-1
6.4
14.4
15.7
EQN 4-2
10.2
16.2
19.2

-------
TABLE 4-9. COMPARISON OF BREATHING LOSS ESTIMATING EQUATIONS
   (USING DEFAULT VALUES)-PREDICTIVE ABILITY AS A FUNCTION
                     OF PRODUCT TYPE
Product
Isopropanol
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
Data points
Ethanol
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
Data points
Acetic Acid
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
Data points
Ethyl benzene
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
Data points
Cyclohexane
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
Data points
Crude oil
Bias, Ib/d
Standard deviation, Ib/d
Root mean squared error, Ib/d
Data points
WOGA and EPA data bases, Ib/d
Equation 4-1
-6.1
1.0
6.2
2

11.1
1.2
11.1
3

-8.8
10.5
13.7
2
-2.8
2.0
3.4
2
26.5
2.4
26.6
3

-179.5
140.9
228.2
8
Equation 4-2
-5.9
1.0
6.0
2

15.1
1.2
15.1
3

-0.8
10.5
10.5
2
-5.1
2.0
5.5
2
33.4
2.4
33.5
3

-55.6
122.7
134.8
8
                             4-27

-------
     TABLE 4-10. BREATHING LOSS ESTIMATING EQUATIONS FIXED ROOF TANKS
      SENSITIVITY ANALYSIS-COMPARISON BETWEEN THE AP-42 AND NEW API
    	EQUATION BASELINE CONDITIONS	
Parameter
Default value
Molecular weight (Mv)
Reid vapor pressure (RVP)
Vapor pressure (Pv)
Tank diameter (D)
Vapor space height (H or HVo)
Average ambient temperature (TAA)
Average ambient temperature change (AT)
alpha
Paint factor (Fp)
Solar isolation (I)
Breather vent pressure/vacuum difference (APB)
Small diam. adjustment (C)
Product factor (Kc)
50 Ib/lb mole
5.5 psia
3.35 psia
175.8 ft
20.75 ft
62.6°F
15.1°F
0.355
1.3
1,594 Btu/ft2 d
0.06 psig
1
0.65
                                          4-28

-------
          TABLE 4-11. BREATHING LOSS ESTIMATING EQUATIONS FIXED ROOF TANKS SENSITIVITY ANALYSIS
Parameter
Molecular weight
Vapor pressure
Vapor space
height
Ambient
temperature
change
Paint color/
condition Fp
(AP-42 equation)
alpha (API
equation)
Solar insolation
Baseline
(default) value
50
3.35
20.75
15.1
1.3
0.355
1,594
Units
Ib/lb-mole
psia
ft
°F
NA
NA
Btu/ft2-d
Difference in calculated breathing loss,
Ib/d
Typical
variance
of default
value
20
1.5
-15
+15
25
0.3
0.18
600
Current
equation
64
64
80
50
100
37
NA
NA
Revised
equation
150
158
150
36
265
NA
118
85
Change in emissions/change in
default variable
Current
equation
3.2 Ib/lb-mole
42.7 Ib/psia
-80 lb/- 15 ft
+501b/+15ft
1001b/10°F
+37 lb/0.3
NA
NA
Revised equation
7.5 Ib/lb-mole
105.3 Ib/psia
-1501b/-15ft
+361b/+15ft
265 lb/10°F
NA
1181b/0.18
85 lb/600 Btu/ft2-d
J^
I


CD

-------
4.5  REFERENCES

1.  Evaporation Loss from Fixed-Roof Tanks, Bulletin 2518, First Edition, American Petroleum Institute,
   Washington, DC, June 1962.

2.  Evaporative Loss from Fixed-Roof Tanks, Bulletin 2518, Second Edition, American Petroleum
   Institute, Washington, DC, October 1991.

3.  Evaporative Loss from Fixed-Roof Tanks; Documentation File for API Publication 2518, Second
   Edition, American Petroleum Institute, Washington, DC, September 1990.

4.  Hydrocarbon Emissions from Fixed-Roof Petroleum Tanks, Engineering Science, Inc., prepared for
   the Western Oil and Gas Association, July 1977.

5.  Breathing Loss Emissions from Fixed-Roof Petrochemical Storage Tanks, Third Draft, Engineering
   Science, Inc., prepared for the U. S. Environmental Protection Agency, December 1978.

6.  Compilation of Air Pollution Emission Factors, AP-42, Supplement E, Chapter 12.  September 1992.
                                             4-30

-------
           5. EMISSION ESTIMATION PROCEDURES FOR FLOATING ROOF TANKS
       The purpose of this chapter is to present the statistical analyses that were performed on the
floating roof tank emission estimation procedures.  In the first section, a discussion of the results of the
statistical analyses on the test tank data performed by API in developing the emission estimation
equations is presented. The second section documents the results of a statistical analysis to determine
how well the procedures predict emissions based on actual tank test data.  The third section documents the
results of sensitivity analyses performed on selected variables in the emission estimating procedures. The
fourth section presents the conclusions from the overall analyses.

5.1 STATISTICAL ANALYSES - API TANK TEST DATA

       The first step in the analysis of the floating roof tank  emission estimating factors and equations
was a comprehensive review and evaluation of the procedures that the American Petroleum Institute
(API) and their contractors, Chicago Bridge & Iron Company (CBI); Cermak, Peterka, Peterson, Inc.
(CPP); and the TGB Partnership used to develop them.  The primary objectives of the analysis were to
assure that valid statistical procedures were used to evaluate the data and to develop better information on
the precision of the coefficients that were developed for the estimating equations. Information on the
precision of the coefficients can provide some indication of the uncertainty of the emission estimates
generated by the equations. The primary sources of information used in the analysis were API
Publications 2517 and 2519 and their associated background documentation files.1"4 Additional sources
of information were two draft documents API submitted to EPA which summarized the procedures API
used for revising the evaporative loss estimating procedures and factors.  These documents are: Manual
of Petroleum Measurement Standards Chapter 19-Evaporative Loss Measurements. Section
2-Evaporative Loss From Floating-Roof Tanks: and Documentation of Rim-seal Loss Factors for the
Manual of Petroleum Measurement Standards. Chapter 19—Evaporative Loss from Floating-Roof
Tanks.5"6

       Personnel at API were extremely helpful in providing the background information needed for the
analysis. However, documentation was lacking on the basis of the development of clingage factors. The
analyses focused on seven components of the floating roof tank estimating equations—the development of
the rim-seal loss factors, the diameter function, the product factors, the deck fitting loss factors, the fitting
wind speed correction factor, the IFRT deck seam loss factors, and the vapor pressure function. The
evaluations of the analyses for each of these seven components are discussed in the following sections.
                                              5-1

-------
5.1.1  Evaluation of Rim-seal Loss Factors

       5.1.1.1  Development Methodology. The rim-seal loss equation and factors introduced in the
Second Edition of API Publication 2517 and used in the previous version of AP-42 Section 7.1 (July
1995) did not allow for a total rim-seal loss factor other than 0 at a wind speed of 0 miles per hour (mph)
and were not considered to be valid at wind speeds below 2 mph.  The new floating roof tank rim-seal
loss estimating equation  developed by API involves three rim-seal loss factors: (1) the zero wind speed
rim-seal loss factor, KRa; (2) the wind speed dependent rim-seal loss factor, KRb; and (3) the rim-seal wind
speed exponent, n. In the new equation, the KRa term extends the applicability of the equation and allows
for a nonzero value of the 0 mph loss factor:
                                           = KRa+KRbVn
where:

         FR =  total rim-seal loss factor, Ib-mole/yr
        KRa =  zero wind speed rim-seal loss factor, Ib-mole/yr
        KRb =  wind speed dependent rim-seal loss factor, lb-mole/(mph)n-yr
          v =  site ambient wind speed, mph
          n =  seal wind speed exponent, dimensionless

        Sets of these new factors were developed by API for 12 distinct tank construction/rim-seal
configurations for average fitting rim-seals and 9 distinct tank construction/rim-seal configurations for
tight fitting rim-seals.  The data used to develop estimates for the rim-seal loss factors were obtained from
the documentation file for API Publication 2517.2 This file contains data from a test program conducted
for API by CBI on a 20 foot diameter test tank. In that test program, emission estimates were generated
based on hydrocarbon measurements downstream from the test tank under steady state conditions and a
constant wind speed. Test runs were conducted for a number of primary and secondary rim-seal
configurations and gap speeds at wind speeds that ranged from  2.2 to 13.1  mph.

        For each rim-seal configuration, information for tanks with two to four sets of gap sizes were
averaged to obtain the final  loss factors for average fitting rim-seals and the 0 gap size data were used to
obtain loss factors for tight fitting rim-seals.  Two distinct computational procedures were used,
depending on the availability of information. In the first case, data were available for the specific
combination of primary and secondary rim-seal of interest for all gap sizes included in the analysis. In
the second case, data were unavailable for the specific combination of primary and secondary rim-seal of
interest, so secondary rim-seal emission reductions were estimated from analogous configurations.
Procedures used for each case are outlined below.

        For cases with data available for the combination of primary and secondary rim-seal of interest,
estimating equation coefficients (KRa, KRb, and n) were obtained using a three step process. The first two
steps generated coefficients for specific gap sizes, while the third step averaged across the gap sizes of
interest.  However, prior to the first step, the raw data from the documentation file were modified by
replacing emission rates for replicate tests at the same wind speed with  the average emission rate for all
replicates at that wind speed. The three steps used to obtain the final coefficient estimates are outlined
below.

                                              5-2

-------
        In the first step, the values of SRa (a coefficient analogous to the zero wind speed loss, KRa, but in
units of Ib/d) were determined using an iterative process. First, an exponential curve fit routine was used
to estimate a starting value for SRa.  Using that estimate of SRa, net emission rates were calculated for each
test by subtracting SRa from the measured value of the emission rate. A standard least squares regression
routine was then used to fit a linear equation with the log transform of the net emission rate as the
dependent variable and the net wind speed as the independent variable. The estimate of SRa was then
changed iteratively and the process was repeated. The estimate which yielded the best fit linear equation
was assumed to be the best estimate of SRa.

        In the second step, the estimate of SRa obtained from Step 1 was subtracted from the measured
emission rate for each test to generate a net emission rate (Enet) for the test. Both the net emission rate and
the measured wind speed for the test were log transformed and an equation of the following form was fit,
where Enet and SRb have units of Ib/d rather than Ib-mole/ft-yr.

                                    log(Enet) = log(SRb) + n* log(v)
        If the value of Enet obtained was less than 0, that test was eliminated from the analyses.  Least
squares regression was used to obtain estimates of n and log(SRb), which were then exponentiated to
obtain an estimate of SRb (analogous to KRb, the wind dependent loss factor).

        In the third step, estimates of the percentage of tanks represented by each gap size used in Steps 1
and 2 were used to generate a weighted average estimating equation for each rim-seal configuration. For
each gap size considered, the equations generated in Steps 1  and 2 were used to generate estimated
emission rates in Ib/d at wind speeds of 0, 4, and 10 mph. The percentage weights were then applied to
obtain average emission rates in Ib/d at each of these wind speeds.  The value obtained for a wind speed
of 0 mph was used as the  estimate of SRa.  To obtain the estimates of n and SRb, net emission rates were
calculated by subtracting the estimate of SRa from the average emission rates  at 4 and 10 mph.  The
resulting net emission rates at two wind speeds were log transformed and a linear equation was fit to the
two points to obtain estimates of n and log(SRb), which was exponentiated to  obtain an estimate of SRb.
Finally, SRa and SRb were  converted to KRa and KRb.

        For six of the primary/secondary combinations of interest, no test data were available.  However,
test data were generally available for all primary rim-seals of interest with no secondary rim-seal and all
secondary rim-seals of interest applied in combination with at least one of the primary rim-seals.
Consequently, loss factors for the primary/secondary combinations without test data were developed by
applying the reduction, or "control efficiency," achieved by the secondary seal of interest applied in
combination with a different type primary rim-seal to the "uncontrolled" emissions from the primary rim-
seal of interest.

        Table 5-1 provides an overview of the data that were used to develop the rim-seal loss factors for
floating roof tanks; the comment column indicates the factors that were calculated indirectly.  The actual
factors are presented in Table 5-2, and those factors that were calculated indirectly, by applying a
secondary rim-seal control efficiency, are denoted by an asterisk.
                                               5-3

-------
TABLE 5-1. BASIS OF RIM-SEAL LOSS FACTORS
Tank
construction
Welded
Rim seal system
Basic
Mechanical shoe
primary
LMRF seal"
VMRF sealb
Extension
Primary only
Shoe-mounted
secondary
Rim-mounted
secondary
Primary only
Weather- shield
Rim-mounted
secondary
Primary only
Weather shield
Rim-mounted
secondary
Gap area (in.)
Prim.
0
1.7,3.8
9.4
0
0
0,1.7,3.8
1.7,3.8
0
1.3,2.6
0,1.3
1.3,2.6
0,1.3
1.3,2.6
0
1
0
1
0,1
1
Sec.
NA
NA
NA
0
0.4
0
0
NA
NA
>9<27
>9<27
0
0
NA
NA
>9<27
>9<27
0
1
Gap
percent
10
80
10
75
25
75
25
65
35
65
35
75
25
65
35
A
NA
75
25
Tests/
points
10/47
3/10
1/5
1/6
1/6
NA
NA
1/7
3/22
NA
NA
NA
NA
10/57
1/13


4/20
3/16
Case
lal
Ia2
Ia3
Ibl
Ib2
Icl
Ic2
2al
2a2
2bl
2b2
2cl
2c2
3al
3a2
3bl
3b2
3d
3c2
Comments





Calculated from a set of rim mounted secondary
data on a LRMF applied to reduce emissions for
mechanical shoe primary seal only data.


One set only of weather-shield data on primary
with 2.6 gap. Used these data to calculate
reduction and applied to primary tight-fitting and
average fitting data.
One set of data for rim mount secondary on
primary with 2.6 gap. Used these data tight fitting
and average fitting to calculate reduction and
applied to primary data.


Efficiencies from weathershield on LMRF primary
applied to VMRF primary seal.


                  5-4

-------
                                                         TABLE 5-1. (continued)
Tank
construction
Riveted
Rim seal system
Basic
Mechanical
shoe primary
Extension
Primary only
Shoe-mounted
secondary
Rim-mounted
secondary
Gap area (in.)
Prim.
0
1.7,3.8
9.4
13.3
0-13.3
0-3.8
1.7-3.8
9.4
Sec.
NA
NA
NA
NA
0.4
0
0
0.76-2.66
Gap
percent
5
55
35
5

20
70
10
Tests/
points
10/47
3/10
1/5
1/5
NA
NA
NA
NA
Case
4al
4a2
4a3
4a4
4bl
4cl
4c2
4c3
Comments




Calculated control efficiency using primary seal only
(three tests) and one test with a secondary seal;
applied this control efficiency to the average primary
only factor

Efficiency of rim-mounted secondary seal with LMRF
primary applied to selected mechanical shoe primary
seal only data

""Liquid-mounted, resilient foam seals.
bVapor-mounted, resilient foam seals.
                                                                    5-5

-------
                                   TABLE 5-2.  SUMMARY OF RIM-SEAL LOSS FACTORS, KRa, KRb, AND n
Tank construction and rim-seal
system

Average-fitting seals

KRa KRb n



lb-mole/mphnft-yr Ib-mole/ft-yr (dimensionless)

Mechanical-shoe seal
Primary only
Shoe-mounted secondary
Rim-mounted secondary
Liquid-mounted resilient-filled seal
Primary only
Weather shield
Rim-mounted secondary
Vapor-mounted resilient-filled seal
Primary only
Weather shield
Rim-mounted secondary

Mechanical-shoe seal
Primary only
Shoe-mounted secondary
Rim-mounted secondary
Welded tanks

5.8a 0.3a
1.6 0.3
0.6* 0.4*

1.6 0.3
0.7* 0.3*
0.3* 0.6*

6.7b 0.2
3.3* 0.1*
2.2 0.003
Riveted tanks

10.8 0.4
9.2* 0.2*
1.1* 0.3*


2.1a
1.6
1.0*

1.5
1.2*
0.3*

3.0
3.0*
4.3


2.0
1.9*
1.5*
Tight-fitting seals

Kna KRb
lb-mole/mphnft-yr Ib-mole/ft-yr


1.5 0.4
1.0 0.4
0.4* 0.4*

1.0 0.08
0.4* 0.2*
0.2* 0.4*

5.6 0.2
2.8* 0.1*
2.2 0.02


c c
c c
c c


n
(dimensionless)


1.9
1.5
1.0*

1.8
1.3*
0.4*

2.4
2.3*
2.6


c
c
c
 Note:  The rim-seal loss factors KRa, KRb, and n may only be used for wind speeds below 15 miles per hour. Factors calculated indirectly are denoted by an asterisk.
alf no specific information is available, a welded tank with an average-fitting mechanical-shoe primary seal only can be assumed to represent the most common or typical
 construction and rim-seal system in use for external and domed external floating roof tanks.
blf no specific information is available, this value can be assumed to represent the most common or typical rim-seal system currently in use for internal floating roof tanks.
°No evaporative-loss information is available for riveted tanks with consistently tight fitting rim-seal systems.
                                                                            5-6

-------
        5.1.1.2 Evaluation. The complete results of the evaluation of rim-seal loss factors are
documented in a report prepared for MRI entitled:  Evaluation of Rim-seal Loss Factors for AP-42 Use.
prepared by Dr. Dennis Wallace and dated September 1995 7

        The verification of the API estimating equations addressed three questions: (1) whether the
computations performed by API could be replicated; (2) whether the replacement of the raw test data with
an average emission rate at a particular wind speed had a significant effect on the final estimates; and (3)
whether the linearity assumed in the averaging process, particularly when ratio estimates were involved,
had a significant effect on estimates. Also, analyses were conducted to assess the effect of primary seal
type, primary seal gap size, and secondary seal gap size on secondary seal performance.  Finally,
uncertainty estimates were developed for two estimation scenarios (individual tanks and population
means).

        5.1.1.2.1 Verification of API computations. The computations performed by API were  replicated
using the raw data from the documentation file; no substantive problems were identified.

        5.1.1.2.2 Replacement of the raw test data with an average emission rate. To address the second
question, three alternative analyses were performed. First, nonlinear regression was used to estimate all
three parameters. Second, the procedures used by API to compute estimates were replicated using the
measured emission rates for each test, rather than the average emission rate for a particular wind speed.
Third, the procedures used by API to compute estimates were replicated using average emission rates for
a given wind speed, but with only a single observation at each wind speed rather than replication of the
averages.

        The results for the individual test conditions are presented in Table 5-3.  Note that the coefficients
generated using the nonlinear  models are quite different than both the API coefficients and the
coefficients generated by MRI using linear model analyses on log transformed data. Examination of the
residuals from the different analyses suggested that the linear model approaches provided better  fits and
were more consistent with model assumptions than were the nonlinear model  results.  Consequently, the
linear model approach used by API is considered appropriate.
                                               5-7

-------
TABLE 5-3. COMPARISON OF ESTIMATING EQUATION COEFFICIENTS: INDIVIDUAL CASES
Case
lal
Ia2
Ia3
Ibl
Ib2
Icl
Ic2
2al
2a2
2bl
2b2
2cl
2c2
Sal
3a2
3bl
3b2
3cl
3c2
Number of
tests3
47 (39)
10(9)
5
6
6
57 (45)b
7
22
16
c
d
c
57*(43)
13
e
f
20 (16)
16
API coefficients
SRS
0.39
1.03
5.8
0.24
0.88
0.37
0.241
0.68
0.18

1.41
2.14

0.55
0.52
Sub
0.0956
0.0491
0.1522
0.0902
0.0380
0.1049
0.0197
0.1935
0.1001

0.0484
0.0821

0.0060
0.0007
n
1.976
2.130
1.972
1.458
1.994
1.922
1.848
1.283
1.340

2.390
3.209

2.597
4.922
Nonlinear model coefficients
SRS
2.383
1.103
4.011
0.212
0.186
2.168
-0.062
0.473
0.475

0
0

0.733
-10
Sub
0.00108
0.0448
0.4109
0.0981
0.2644
0.00239
0.1343
0.3179
0.0639

0.599
4.110

0.000357
0.329
n
3.780
2.167
1.598
1.428
1.228
3.438
1.100
1.121
1.493

1.326
1.478

3.859
2.418
Linear model coefficients
Test specific
SRS
0.39
1.03
5.8
0.24
0.88
0.37
0.241
0.68
0.18

1.41
2.14

0.55
0.52
$Rb
0.0947
0.0456
0.1522
0.0902
0.0380
0.1033
0.0197
0.1935
0.1001

0.0739
0.0821

0.00566
0.00067
n
1.974
2.162
1.972
1.458
1.994
1.923
1.848
1.283
1.340

2.200
3.209

2.622
4.922
Wind speed average
$Ra
0.39
1.03
~
0.37
..
~

1.41

0.55
$Rb
0.0993
0.0525
~
0.1085
..
~

0.0542

0.0058
n
1.928
2.101
~
1.876
..
~

2.342

2.631

-------
                                                      TABLE 5-3. (continued)
Case



4al
4a2
4a3
4a4
4bl
4cl
4c2
4c3
C31
C36,C38
C35
W24,W25
Number of
tests3


g
b
h

i
j
a
h
7
12
8
16
API coefficients


SRS



1.46
2.689



0.43
1.53
0.76
0.12
Sub



0.1612
0.922



0.0372
0.0179
0.0170
0.0579
n



1.694
2.012



1.113
2.457
2.191
1.979
Nonlinear model coefficients


SRS



2.155
—



—
-0.58
0.557
-4.2
Sub



0.0653
—



—
0.925
0.0664
1.478
n



2.046
—



—
0.845
1.602
0.8575
Linear model coefficients

Test specific
SRS



1.46
2.689



0.43
1.53
0.76
0.12
$Rb



0.1612
0.0631



0.0372
0.0179
0.0170
0.0579
n



1.694
1.9041



1.113
2.457
2.191
1.979
Wind speed average
$Ra



~
—



—
—
	

$Rb



~
—



—
—
	

n



~
—



—
—
—

aNumbers in parentheses represent average wind speed.
bSame as Case Ia2
°Same as Case 2a2
dSameasCase2bl
eSame as Case Sal
fSame as Case 3a2
8Same as Case lal
hSame as Case Ia3
'Weighted average from 4a
JSame as Case Icl
                                                                5-9

-------
       There were some differences in the specific coefficients for the different linear model approaches,
but they were relatively minor. Furthermore, the values of SRb and n tend to adjust in opposite directions.
 Consequently, although using the actual data rather than the average data would have been more
appropriate, the results differ so little that modification of the estimating equations is not warranted.

       5.1.1.2.3 Linearity assumption. To address the third question, two alternative average estimates
were computed. The first replicated the API procedures, but measured, rather than average, emission
rates were used. The second procedure was fundamentally different in that emission rates were computed
at 0.5 mph increments from 0.5 mph to 14 mph, weighted average emission rates were computed at each
wind speed (using the appropriate ratio for secondary seal performance at that wind speed), and a linear
regression equation was fit to log transformed net emissions and log transformed wind speeds to estimate
coefficients. The results of the analyses are described below.

       Estimating equation coefficients and estimated emission rates were calculated at wind speeds of
4, 8, and 12 mph in units of Ib-mole/ft-yr for three averaging techniques.  The first MRI technique was
comparable to the API technique  described above, but actual, rather than average, emission data were
used in the  computations. The second MRI technique involved computation of average emissions at wind
speeds over the range of 0 to 14 mph and fitting a regression model to the computed values.  The second
MRI procedure provides better estimates if the ratio procedures used to estimate emissions result in
extreme nonlinearities. Table 5-4 summarizes the results of using the alternative averaging techniques.
The coefficients varied somewhat, but the effects on emissions were minimal, particularly in light of the
uncertainties in the estimates.  Average emissions for the 12 scenarios of interest varied little as a function
of averaging method.  Again, these analyses indicate that the equations generated by API are acceptable
and provide no reason to modify the equations.
                                              5-10

-------
TABLE 5-4. COMPARISON OF ESTIMATING EQUATIONS FOR AVERAGE FITTING RIM-SEAL LOSS FACTORS
Case
1A
Ib
1C
2A
2B
2C
3A
3B
3C
4A
4B
4C
Description
Mechanical shoe— primary only
Mechanical shoe— shoe-mounted
secondary
Mechanical shoe— rim-mounted
secondary
Liquid mounted seal— primary only
Liquid mounted seal — weathershield
Liquid mounted seal— rim-mounted
secondary
Vapor mounted seal— primary only
Vapor mounted seal — weathershield
Vapor mounted seal— rim-mounted
secondary
Riveted: mechanical shoe secondary
Riveted: mechanical shoe shoe-mounted
secondary
Riveted: mechanical shoe rim-mounted
secondary
Method8
API
MRI1
MRI2
API
MRI1
MRI2
API
MRI1
MRI2
API
MRI1
MRI2
API
MRI1
MRI2
API
MRI1
MRI2
API
MRI1
MRI2
API
MRI1
MRI2
API
MRI1
MRI2
API
MRI1
MRI2
API
MRI1
MRI2
API
MRI1
MRI2
Coefficients
KRJ
5.8
5.8
5.8
1.6
1.6
1.6
0.60
0.60
0.60
1.6
1.6
1.6
0.71
0.71
0.71
0.34
0.34
0.34
6.7
6.7
6.7
3.3
3.3
3.3
2.2
2.2
2.2
11
11
11
9.2
9.2
9.2
1.1
1.1
1.1
KRb
0.25
0.24
0.24
0.29
0.29
0.30
0.38
0.37
0.19
0.29
0.29
0.57
0.33
0.33
0.29
0.57
0.57
0.23
0.19
0.22
0.29
0.12
0.14
0.18
0.0035
0.0034
0.012
0.37
0.36
0.36
0.23
0.25
0.29
0.27
0.26
0.21
N
2.1
2.1
2.1
1.6
1.6
1.6
1.0
1.0
1.3
1.5
1.5
1.2
1.2
1.2
1.3
0.33
0.33
0.75
3.0
3.0
2.9
3.0
2.9
2.8
4.3
4.3
3.7
2.0
2.0
2.0
1.9
1.9
1.8
1.5
1.5
1.6
Emissions (Ib-mole/ft-yr)
4mph
10.4
10.2
10.2
4.3
4.3
4.4
2.1
2.1
1.8
3.9
3.9
4.6
2.5
2.5
2.5
1.2
1.2
1.0
18.9
20.8
22.9
11.0
11.1
12.0
3.6
3.5
4.2
16.9
16.8
16.8
12.4
12.7
12.7
3.3
3.2
3.0
8mph
25.5
24.7
24.7
9.7
9.7
10.0
3.6
3.6
3.4
8.2
8.2
8.5
4.7
4.7
5.0
1.5
1.5
1.4
104
119
127
64.7
61.5
64.1
29.0
28.2
28.5
34.7
34.0
34.0
21.2
22.2
21.4
7.2
6.9
7.0
12mph
52.0
50.1
50.1
17.1
17.1
17.6
5.2
5.0
5.4
13.7
13.7
12.8
7.2
7.2
8.0
1.6
1.6
1.8
335
387
398
211
192
193
155
151
120
64.3
62.8
62.8
35.0
37.3
34.6
12.3
11.9
12.3
                         5-11

-------
       5.1.1.2.4 Effect of rim-seal type and gap size on rim-seal performance.  As noted previously, no
specific test tank data were available for six of the primary/secondary rim-seal configurations.  For each
of those configurations, a comparison case was used to estimate the reduction in emissions associated
with applying a secondary rim-seal. An assumption embedded in these computations is that the emission
reduction,  or the incremental control efficiency, achieved by a secondary rim-seal is not affected greatly
by the type of primary rim-seal or rim-seal gap size.  To evaluate this assumption, analyses were
conducted to assess the effect of primary rim-seal type on secondary  rim-seal performance, the effect of
primary seal gap size on secondary rim-seal performance, and the effect of secondary seal gap size on
secondary rim-seal performance.

       To evaluate the effect of primary rim-seal type on secondary rim-seal performance, three cases
were compared. The results are graphically presented in Figures 5-1  and 5-2, respectively.  For both
emission rates and reduction percentages, the rim-mounted secondary seal appears to perform comparably
for mechanical shoe and liquid-mounted primary rim-seals (API estimate), but quite differently for vapor-
mounted primary rim-seals.  Because the comparison results were not used for vapor-mounted rim-seals
in the API analyses, these results provide no basis for modifying any of those emission factors.  However,
they do suggest that care should be taken in evaluating secondary rim-seal performance on future tests
conducted under the new protocol being developed by API for certifying loss factors for new rim-seal
configurations.

       To examine the effect of primary rim-seal gap size on secondary rim-seal performance, five cases
were examined. Each case involved a rim-mounted secondary seal with no gap. The results are
graphically presented in Figures 5-3 and 5-4. With the exception of the vapor-mounted seal with no
primary gap, the data indicated some difference in efficiency at low wind speeds. However, for wind
speeds above 6 to 7 mph, the data indicated that the primary seal gap size has little effect on efficiency.
These results suggest that the primary seal gap will have little effect on the efficiency for cases generated
by API, so no changes in the loss factors are recommended. However, the analyses again point to the
anomalous results for vapor-mounted primary seals, raising some concerns about the reliability of those
factors.
                                            5-12

-------
    70
    60
    50
	API Estimate





	Mechanical Shoe





	Vapor—Mounted
            -^-r..rT----T-T-T-r---..rT..r.T..,^	r j.^.y...^-^.-.^,-^-^	t	,	t	|	j	,	t	t	p., -T	f. T.|. ,,,,„,	T. T ^, rr.,r ., .^ r-r-f|	|	;' I j I I" f-pij-




       0123456789    10   1112   13   14




                                  Wind Speed, mph




Figure 5-1. Emissions after a rim-mounted secondary seal as a function of primary seal type.
                                      5-13

-------

    1.00



    0.95-




    0.90-




    0.85:



    0.80-
 S3

£


 §  0.75

'•g



    0.70
 rt
    0.65-
    0.60 H
 o
 o>
OJ
    0.55-




    0.50^




    0.45 ^




    0.40 :
 i
                 7	
	  API Estimate



  	Mechanical Shoe




  	 Vapor—Mounted
    0.35


        0123456789   10   111213   14


                                 Wind Speed, mph



Figure 5-2. Efficiency of rim-mounted secondary seal as a function of primary seal type.
                                   5-14

-------
      75
      70:
      65:
      60
      55-
  -  50-
  •f  45 i
  g
  -Q
   _.  40-
   I
      351
   §  30 i
  I
  s  25:
      20-
      15 :
      10 :
      5
SM Seal Gap=39.2
SM Seal Gap=9.2
SM Seal Gap=0
VM Seal Gap=l
VM Seal Gap=0
      0
        0123456789   10   U1213   14
                                Wind Speed, mph
Figure 5-3.  Emissions after a rim-mounted secondary seal as a function of primary seal gap size.
                                5-15

-------
  LOO
  0.95
             -SMSeal Gap=39.2
            •-SM Seal Gap=9.2
             -SM Seal Gap =
              VM Seal Gap =
              VM Seal Gap =
      01    23456789    10   1112   13  14
                             Wind Speed, mph
Figure 5-4.  Efficiency of a rim-mounted secondary seal as a function of primary seal gap size.
                                5-16

-------
Finally, the effect of secondary rim-seal gap size on secondary rim-seal performance was examined. A
series of five cases were examined. The results are graphically presented in Figures 5-5, 5-6, 5-7, 5-8,
and 5-9. The results suggest that even a small difference in secondary seal gap size can affect seal
performance. Generally, the results show that performance improves with increasing wind speed for gap
sizes 2 in2/ft or less, but tends to deteriorate with increasing wind speed for larger gap sizes.
                                             5-17

-------
    LO
    0.9-
 s   '
 o
 c"  0.7
 o
13  0.6 i
 8  0-5:
 
-------
1.00
0.90 '' • <	
    0123456789   10   U1213   14

                              Wind Speed, mph


             Figure 5-6. Effect of secondary gap on efficiency:
              Case 2--shoe mounted primary with 9.4 inch gap
                                5-19

-------
 e
 o
    LO
    0.9-
    0.8-
    0.7-
 c

•B  0.6
 u

T3
•a  o-s
 o
t/2
 fi  O-4
 o
OQ
    0.3-
    0.2-
    0.1-
          \
            \
              \
	Sec Gap=4.9




	  Sec Gap=L8




                 t=0
    0.0 T"'"1	'	I	'	'	'	'	1
       0123456789    10   1112   13   14



                                 Wind Speed, mph




                   Figure 5-7.  Effect of secondary gap on efficiency:

                   Case 3--shoe mounted primary with 39.2 inch gap.
                                     5-20

-------
	No Secondary

     Sec Gap = 0.43

     Sec Gap=0
56789
 Wind Speed, mph
                                        10   1112   13   14
    Figure 5-8. Effect of secondary gap on emissions:
     Case 4--shoe mounted primary with 1 inch gap.
                      5-21

-------
so
70
60
0
	No Secondary

	Sec Gap=2.56

	Sec Gap=0.76
      	I'1 '' I | I ' I I |
  0123456789   10   111213   14

                           Wind Speed, mph

            Figure 5-9.  Effect of secondary gap on emissions:
            Case 5--shoe mounted primary with 13.2 inch gap.
                              5-22

-------
       The analyses suggested that secondary rim-seal performance is strongly affected by gap size.
Given the limitations in the current data base, no modifications to the loss factors based on these findings
are recommended at this time.  However, the strong effect of gap size on performance for vapor-mounted
rim-seals suggests that further analysis of these rim-seals is warranted.  Also, the effect of gap size on
performance has implications for the design of test programs to demonstrate the performance of new rim-
seal configurations.

       5.1.1.2.5 Uncertainty estimates. Uncertainty estimates were developed at two levels (unique test
cases and average estimating equations) for two estimation scenarios (individual tanks and a tank
population  mean).  When the equations are applied to populations of tanks, the uncertainties are generally
quite reasonable. However, the uncertainties for the equations when applied to individual tanks are quite
large. Also, the uncertainties for vapor-mounted rim-seals are quite large in comparison to those for shoe-
mounted and liquid-mounted rim-seals. With the exception of the vapor-mounted rim-seals, the upper
bound for the population mean was generally within a factor of two of the point estimate.

       In conclusion, the results of the analyses performed to verify and evaluate API's calculations
indicate that the new rim-seal loss  factors generated by API are acceptable for use in Section 7.1 of AP-
42. Although the analyses do indicate that secondary seal gap size does affect rim-seal performance,
limitations  in the data base preclude any recommendation for modifications to the factors. However, due
to the strong effect of wind speed and gap size on seal performance for vapor-mounted seals, further
analysis of these seals may be warranted.  The results of these analyses should also be considered in API's
development of test methods to evaluate performance of new rim-seal designs.

5.1.2 Evaluation of Wind Speed Calculation

       The relationship between emissions and wind speed in the floating roof tank estimating equations
was developed by regressing daily emissions against fixed wind speeds under equilibrium conditions.
The equation was then modified to generate annual emissions by embedding a multiple of 365.25  in the
equation's constant coefficient and by assuming that as wind speed varied over the year, the average
annual emissions could be obtained by evaluating the function at the average wind speed. Because
emissions and wind speed are related nonlinearly, this latter assumption is not strictly correct.  A
sensitivity analysis was conducted to determine how using the estimating equations at average wind
speeds rather than averaging the emissions obtained by applying the estimating equations over a typical
wind speed distribution might affect the emission estimate.

       Information presented by Justus, et al., and Corotis et al., indicates that wind speed distributions
can be represented reasonably well by a two parameter Weibull distribution.1'8  An analysis of the general
form of the distribution indicated that the error in the emission estimate introduced by applying the
estimating equation to the average wind speed rather than averaging emissions  over the wind speed
distribution depends only on the coefficient of variation of wind speed (i.e., the ratio of the standard
deviation to the mean of wind speed) and the wind speed exponent in the estimating equation. The error
in the estimate increases as the  coefficient of variation increases and as the exponent moves away from 1.
 (Both methods produce the same result if the exponent equals 1).  The  Justus et al., and Corotis et al.,
papers also provided information on typical values for those parameters that could be used to evaluate
possible errors. For an exponent of 1.5, the estimate obtained using the average wind speed was 7 to
11 percent lower than the estimate obtained by averaging over the wind speed distribution when the
coefficient  of variation was in the range of 0.45 to 0.6. For an exponent of 0.4 using the same range of
coefficient  of variation, the estimates obtained using the average were 2 to 5 percent higher than estimates
obtained by averaging over wind speeds. However for an exponent of 2.6, the estimates obtained using
                                             5-23

-------
average wind speed were 30 to 45 percent lower than those obtained by averaging over wind speeds.
These results suggest that if the estimating equations are used with exponents in the range of 0.4 to 1.5,
using average wind speeds produces reasonable results.  However, if the application involves larger
exponents, the use of wind speed distribution data should be considered to improve the accuracy of the
estimates.

5.1.3 Evaluation of Diameter Function and Product Factor

        Because the data base that could be used to evaluate alternative diameter exponents for the
floating roof tank estimating equations is quite sparse, the mathematical techniques used by API in the
analysis are  straightforward. As such neither the data nor the analytical procedures lend themselves to
complicated statistical procedures. The  procedures used by API are reasonable and the results of the
analyses are supported. No change in the  diameter exponent of 1 is recommended. However, the
paragraphs below briefly describe the API analyses and summarize the available information on the
uncertainty associated with the diameter exponent.

        Field test emission data were collected for three tanks (denoted as C, T, and P) to evaluate the
effect of tank diameter on emissions.  To conduct the evaluation, API identified appropriate CBI test tank
conditions that matched the field test conditions. For tanks T and P, CBI test conditions were found that
closely matched the field tank. However, for tank C no directly applicable conditions were found, so
multiple conditions were used to simulate the tank C conditions; consequently, the results for tank C are
less reliable  than those for tanks P and T.  Estimating equations were then developed based on those field
tank-specific test conditions, and emissions from the field tanks were estimated using those equations
with different diameter exponents.

        The results of the analyses, which are shown graphically in Figure 5-10, suggest that 1 is a
reasonable exponent, particularly for tanks that are higher emitters. The error bands for the field tank
emission estimates suggest that the exponent can reasonably be expected to be within the range of 0.7  to
1.2.  If this range of exponents is applied to a 100 foot diameter tank, the emission estimate ranges from
25 percent to 250 percent of the estimate obtained for an exponent of 1. Hence, even though the exponent
uncertainty is relatively small, it does introduce substantial uncertainty into the emission estimates.
                                             5-24

-------
*
O
IUUU
900
800
700
600
400
200
100
90
80
70
60
50
40
30
20
10
9
a
7
6
5
4
3
2
1
C
—
—
_
—
—
_
—
z:
_.
_
—
r_
_
>-
-
_
=
~
—
—
_
-
-
-
-
E
-
-






IV
(F


























easure<
icunds f







i loss
jer day)

173 —







2J











1
.3 0.4








































35% cor

















• ~^








X
s
ifidence
\
\
\



^










~*~~*







4
'A
interval



\
\T
1
M -
^










. 	 "







t
/
Ta
»







Ta
















X
nkC






^
nkT









TarvkP
-— —
— -*







K»







•-*










M"






















































—
—
—
_
-
-
_
=
—
=
—
—
_
_
—
-
_
^
—
—
_
—
-
—
-
_
-
=
I
—
0.6 0.8 1.0 1.2 1.4 1.
                                         Diameter exponent



              Figure 5-10. Calculated losses as a function of diameter exponent.
                                          5-25

-------
        The procedures used by API to develop the product factors for crude oil and gasoline products are
outlined in the documentation file of Appendix E of the 1983 Edition of API 2517.2 The procedure relies
on a straightforward comparison of paired test conditions that had the same seal configurations but
different stored products. For the crude oil analysis, three test pairs that had different seal configurations
and a wide variability in wind speeds were used. For each of the six test conditions, a regression model
was fit for log emissions versus log wind speed.  Resultant emission estimates at 5,  10, and 15 mph were
then adjusted for vapor pressure and molecular weight and the ratio was compared.  The results for the
three scenarios produced product factor estimates of 0.28 and 0.32 for configurations with a primary seal
only and 0.51 for a configuration with a secondary seal for an overall product factor of 0.4 for crude oil.

        Replication of the regression analyses conducted by API verified these results. Furthermore,
these analyses indicated that the uncertainty associated with the three individual product factors and the
average factor was quite small.  Based  on the variances of the regression coefficiencies, the 95  percent
confidence intervals for both the individual pairs and the average are estimated to be on the order of
+ 10 percent. Hence, the product factor is likely to provide precise estimates over a large tank  population.
However, the large difference in the factor for different scenarios suggests that the factor could introduce
substantial error for a single tank. Hence, the use of Kc = 0.4 for crude  oil is recommended for AP-42.

        During the IFRT, 20-ft diameter pilot scale studies conducted by CBI, several tests were
conducted during each phase for which the only difference was the vapor pressure of the octane/propane
mixture. Tests were available under the same conditions for the single-component stocks with  n-hexane
and n-octane stored in the tank.  The emission data for each test condition were regressed against true
vapor pressure. These regression lines were demonstrated graphically to provide a good fit to the
octane/propane data.  Furthermore, with one exception, the regression lines generated estimates for the
single-component tests that were greater than or equal to the measured values. The  single exception was
an n-octane test during Phase I, under tight seal conditions, with a very low true vapor pressure (0.3 psia).
Given these results, API determined that a product factor of 1.0 for single-component stocks provides a
conservative emission estimate.

        As a part of this review, MRI corroborated the API results. Given the limited  quantity  of data on
single-component stocks, the procedures used by API are reasonable. The single problem point appears
to be an anomalous data point. Hence, the use of Kc =1.0 for single-component mixtures is
recommended for AP-42.
                                             5-26

-------
                             ,9-13
5.1.4  Deck Fitting Loss Factors

        5.1.4.1  Development Methodology.  The floating roof tank fitting loss estimating equations
involve three loss factors: the zero wind speed loss factor, or equation intercept, (Ka), wind speed
dependent loss factor (Kb), and wind speed exponent (m). Factors were developed for both slotted and
unslotted guide poles under a variety of control scenarios that involved the application of gaskets, floats,
float wipers, pole sleeves, and pole wipers. Other fittings used to develop the estimating equations were
deck legs assumed to have a 3 in. diameter opening, a gauge hatch/sample port assumed to have a 8 in.
diameter opening, a vacuum breaker assumed to have a 10 in. diameter opening, an access hatch assumed
to have a 24 in. diameter opening, a gauge-float well assumed to have a 20 in.  diameter opening, and a
deck drain assumed to have a 3 in. diameter opening. With  the exception of the deck leg and deck drain
fittings, all fittings were tested with and without gasketing.

        The tests were performed in a wind tunnel, at wind  speeds of 0, 4.3, 8.5, and 11.9 miles per hour.
 The deck fitting was mounted on a product reservoir that rested on a digital platform scale.  The top of
the deck fitting extended into the wind tunnel. Precise measurements of the test facility weight were used
to obtain data on cumulative evaporative loss over time; the duration of each test depended on the loss
rate for a particular fitting.  For slotted guide  poles, tests were conducted with the slots oriented at
different angles to the primary wind direction through the tunnel. Details of the guide pole test
procedures and the raw data from the test are  contained in the May 1994 addendum to API Publication
No. 2517 and its background documentation.9 Details of the other fitting test procedures can be found in
Volume 1 of the Final Report, Testing  Program to Measure  Evaporative Losses from Floating Deck
fittings, dated October 1, 1993.

        The deck fitting loss factor estimating equations were developed by CBI/API using a three stage
procedure, outlined below.

        1. The raw data from each fitting test for cumulative loss versus time were fit to either a linear or
quadratic polynomial in time using a least squares analysis,  and the initial loss rate for each test was
determined by evaluating the derivative of the polynomial at time zero.

        2. For each combination of fitting  and wind speed,  an average initial loss rate was determined by
taking the arithmetic average of all initial loss rates obtained for that combination of fitting and wind
speed. This process generated four data points for each fitting, one point for each wind speed.  Note,
however, that in some cases, only one test was conducted, and this single test value was used as the
average initial loss rate for that combination of fitting and wind speed.

        3. For each fitting, a software  package, "Table Curve for Windows™," was used to fit the four
data points to a curve of the form:
                                         KF = KFa + KFbvm


where:

               KF =   Fitting loss factor, Ib-mole/yr
               KF  =  Zero wind speed fitting loss factor, Ib-mole/yr
               KFfe =  Wind speed dependent fitting loss factor, lb-mole/(mph)nyr

                                             5-27

-------
               v =    Ambient wind speed, mph
               m =   Wind-dependent loss exponent, dimensionless

        The software used by API utilized a Levenburg-Marquardt algorithm, a standard technique for
nonlinear curve fitting, to fit this curve. To force the intercept of the function (KFa) to fit the data exactly,
the algorithm was modified to give a large weight to the zero wind speed data point.

        The results of MRI's analyses of the deck fitting factors recommended by API are presented in the
following sections.  The analyses of guide pole fitting and other deck fittings are discussed separately.
        5.1.4.3 Evaluation of the Guide Pole Deck Fitting Loss Factors. There were four primary
concerns about the statistical methods used by API to evaluate the data.  First, the method used to force
the equation through the measured zero wind speed data point prevented obtaining meaningful
information about the goodness of fit of the equation to the data. Because KF is simply a linear
                                                                       a
transformation of the equation from the intercept, a much better method for forcing the equation through
the zero wind speed data point is simply to subtract that intercept from the other data points and fit a two-
parameter equation to the remaining three data points.

        The next two concerns involved the method used to average the data. If the relationship between
evaporative loss and wind speed is the hypothesized power function, then evaporative loss and wind
speed are linearly related on a log-log scale. If that assumption holds, then averaging emissions on the
log scale rather than the linear scale appears to be more appropriate. Also, the data indicate that loss rate
is strongly related to slot orientation relative to wind direction. There  were sometimes different numbers
of tests for  different slot orientations, so equal weighting of orientations rather than equal weighting of
individual tests appears to be more appropriate.

        The final concern was whether or not the use of an overall strategy that analyzed averaged data
on a fitting-specific basis to obtain parameter estimates was more appropriate than the use of a more
general modeling strategy that used individual test data modeled concurrently to obtain both parameter
estimates and their standard errors.  Use of averages masks the inherent variability among the tests and
makes any assessment of the precision of the estimates unreliable, while a more general modeling strategy
provides better insight into the relationship of the performance of the different control strategies.
Outlined below are the approaches that were used to analyze the data and address the above concerns.

        5.1.4.3.1 Slotted guide  poles.

        Fitting specific analysis. For slotted guide poles, three alternative fitting-specific modeling
strategies were examined. For the first (MRI-1), the equation was forced through the measured data point
by subtracting the zero wind speed value for a particular fitting from the nonzero wind speed values for
that fitting. The resultant values were then averaged arithmetically,  and the parameter estimates were
obtained by fitting a linear equation to log transforms of loss rate and wind speed rather than fitting a
nonlinear equation to the raw values. This first approach provides an assessment of modifying the
approach to forcing the curve through a particular intercept and of using a linearized approach to fitting
the equation rather than a nonlinear curve fit approach. The second  and third approaches (MRI-2 and
MRI-3) both used weighted regression approaches that employed test-specific loss rates rather than
averages. This weighted approach resulted in "averaging" on the log transformed scale rather that on the
linear scale. The MRI-2 approach gave equal weight to each test, while MRI-3 gave equal weight to each
wind orientation.
                                             5-28

-------
        Eleven different control scenarios that involved the application of gaskets, floats, float wipers,
pole sleeves, and pole wipers to slotted guide poles were tested.  Two additional tests, fitting Nos. 31 and
C01, were conducted after these analyses were performed.  Data from these tests were also used in
developing slotted guide pole loss factors. Table 5-5 identifies each of the slotted guide pole fitting
configurations and compares the parameter estimates for each of the analysis methods. For seven of the
configurations, an equal number of tests was conducted at each slot orientation at any given wind speed,
so the results for the last two methods are equal.  From Table 5-5, it can be seen that parameter results
produced by the different methods vary considerably. However, graphs of the estimating equations
indicate that substantially different loss rates are estimated for most configurations only at low wind
speeds.  Only for fitting Nos. 20 and 30 are the loss rates at high wind speeds substantially different.
Further, when the differences are substantial, the two methods that use "averaging" on the linear scale
(API and MRI-1) consistently predict higher emissions than do the methods based on log scale analyses.
This difference is a result of the greater relative weight given to the large values on a linear scale.
                                              5-29

-------
TABLE 5-5. COMPARISON OF SLOTTED GUIDE POLE PARAMETER ESTIMATES

1
25
3
26
20
2
Control measure
Gasket"
0
1
0
1
1
1
Floatb
0
0
1
1
0
0
Pole sleevea
0
0
0
0
0
1
Pole wiper0
0
0
0
0
1
0
Estimation
technique"1
API
MRI-1
MPJ-2
MPJ-3
API
MRI-1
MRI-2
MRI-3
API
MRI-1
MRI-2
MRI-3
API
MRI-1
MRI-2
MRI-3
API
MRI-1
MRI-2
MRI-3
API
MRI-1
MRI-2
MRI-3
Parameter estimate
Ka
45.4
45.5
45.5
45.5
40.7
40.6
40.6
40.6
35.7
35.9
35.9
35.9
25.8
25.6
25.6
25.6
41.2
41.2
41.2
41.2
16.3
16.4
16.4
16.4
Kb
698
461
186
186
311
364
386
386
102
37.5
26.4
38.2
9.06
35.9
33.1
33.1
130
90.1
47.9
47.9
132
39.4
4.92
21.4
M
0.974
1.16
1.54
1.54
1.29
1.22
1.17
1.17
1.71
2.15
2.29
2.12
2.54
1.94
1.94
1.94
1.23
1.40
1.45
1.45
1.19
1.73
2.45
1.82
                            5-30

-------
                                       TABLE 5-5. (continued)
Fitting
30

23

29
Control measure
1

1

1
3

2

1
0

0

1
1

1

2
Estimation
technique"1
API
MRI-1
MRI-2
MRI-3

API
MRI-1
MRI-2
MRI-3

API
MRI-1
MRI-2
MRI-3
Parameter estimate
13.8
13.8
13.8
13.8
17.9
17.9
17.9
17.9
19.2
19.2
9.09
9.09
9.09
9.09
13.7
9.47
10.2
10.4
54.2
16.9
13.5
13.5
15.0
6.99
13.4
15.0
15.0
15.0
1.94
2.10
1.70
1.67
1.10
1.62
1.58
1.58
0.935
1.19
0.512
0.457
0.456
0.456
   aO = control measure not implemented; and 1 = control measure implemented.
   bO = no float used; 1 = float with 0.25 inch gap and wiper 1 inch above cover; 2 = float with 0.25 inch gap;
   and wiper at cover elevation; and 3 = float with 0.125 inch gap and no wiper.
   °0 = no pole wiper; 1 = pole wiper at sliding cover elevation; and 2 = pole wiper 6 inches above sliding cover.
   dAPI = API estimates published in addendum to API Publication No. 2517; MRI-1 = results of MRI analyses
   using a linear regression technique with different wind orientations averaged on a linear scale; MRI-2 =
   results of MRI analyses using a linear regression technique with different wind orientations averaged on a log
   scale; and MRI-3 = results of MRI analyses using a linear regression technique with different wind
   orientations averaged on a log scale weighted so that each direction received equal weight.
   eAPI excluded two tests (4BNF and 8AWG) from their analysis for fitting No. 4. MRI included these tests in
   the analysis. Subsequent conversations with API indicated that there were valid reasons for eliminating these
   tests.

                The results of this  analysis indicate that the estimating equations generated by API will
provide conservative estimates of VOC loss rates. At wind speeds of 12 miles per hour or less, the
positive bias will be relatively small for most fittings and will be less than a factor of 2 for all fittings
except No. 30.  This fitting had a float that fit tighter to the guide pole than is typical, but did not have a
float wiper. This is not a design in actual use. The apparent rationale  for this test was to determine
whether this design could be considered equivalent to a standard float. The measured losses were slightly
lower at  0 mph, but about twice as  high at typical wind speed levels, as compared to a standard float.
                                               5-31

-------
        General model analyses. The use of a more general modeling strategy that used individual test
data modeled concurrently to obtain both parameter estimates and their standard errors was investigated.
A general model analysis was performed to provide more information about the relative  performance of
the different control strategies and evaluate the precision of the parameter estimates.  There were two
stages in the analysis. The first was an analysis of the zero wind speed data to determine equation
intercepts (KFa). The second used the nonzero wind speed data to determine the wind speed coefficient
(KFb) and exponent (m). This approach was selected over a single modeling effort for two reasons. First,
it assures that the intercept term will be nonnegative, which is a physical constraint of the system being
modeled. Second, because measured emissions at zero wind speed were quite low for all fittings, any
errors introduced by using only the zero wind speed data to generate the intercept are expected to be small
relative to the overall errors in the data collection and analysis procedures.

               Because only a single data point was obtained for each fitting at zero wind speed, the
quantity of data was insufficient to model statistically.  The data were examined visually, and engineering
judgment based on the relative magnitude of the loss factor and the fitting configuration was used to
determine which groups of data to average.  An average zero wind speed loss was determined for fitting
Nos. 2,  23, and 24; fitting Nos. 4 and 26; and fitting Nos. 25 and 20. For all other fittings the measured
loss rate at zero wind speed was used for KF .

        Since KF was fixed in stage 1, a linear model approach based on log transforms of loss rate and
wind speed was used in stage 2 to determine KFfe and m. The linear model approach has the advantage
over a nonlinear modeling approach by having an exact analytical solution to the least squares normal
equations. To develop the linear models, the appropriate value for KF  was subtracted from each initial
                                                               a
loss rate and logs of the residual loss rate and wind speed were determined.  Then a sequence of linear
models  was fit starting with a saturated model with individual estimates of KFfe and m for each of the
fittings.  The saturated model was subsequently reduced by removing parameters that had no significant
effect on the ability of the model to fit the measured data. A weighted  regression  approach was used,
with each slot orientation equally weighted.

        The saturated model generated the same parameter estimates as those generated  by the weighted
regression method MRI-1. The final reduced model contained six parameters: four to estimate KF  and
two to estimate wind speed exponents. For all fittings except No. 29, which had a pole wiper 6 inches
above the sliding cover instead of at the cover level, a single wind speed exponent of 1.6 fit the data well.
 Fitting  No. 29 had a wind speed exponent of 0.45. The four parameters to estimate log  KFfe were found to
be linear combinations of four types of fitting controls (gasket, float [with float location  being
unimportant], pole sleeve, and pole wiper at the sliding cover level). The resultant parameter estimates
and their standard errors are presented in Table 5-6. For the uncontrolled fitting, the estimate of log (KFfe)
is 2.3. Adding  a gasket reduces log (KF )  by 0.13, a float by 0.36, a pole sleeve by 0.63, and a wiper at
the sliding cover by 0.72. The use of this method allows direct comparison of the effects of particular
control  options and reduces the standard errors of the parameter estimates, thereby providing more precise
and stable estimates.
                                             5-32

-------
                               TABLE 5-6. SUMMARY OF RESULTS FOR LINEAR MODEL ANALYSES OF ALL FITTINGS
Fitting
No.
1
25
o
3
26
20
2
30
23
4a
24
29
31
32
C01
Comparison of parameter estimates
Ka
45.5
40.9
35.9
25.9
40.9
17.8
13.8
17.8
25.9
(2.42)
17.8
9.1
4.89
8.3
4.86
Individual regression
Log Kb
Est.
2.27
2.58
.59
.52
.68
.32
.02
.13
0.940
(0.669)
0.874
1.20
1.85
0.64
0.97
s.e.
0.36
0.28
0.28
0.32
1.01
0.77
0.93
0.69
0.47
0.12
0.071
0.51
0.50
0.06
m
Est.
1.54
1.17
2.12
1.94
1.45
1.83
1.67
1.58
1.66
(1.93)
1.17
0.430
1.09
1.6
1.03
s.e.
0.40
0.31
0.31
0.36
1.12
0.86
1.03
0.77
0.53
0.14
0.089
0.47
0.65
0.11
Saturated model
Log Kb
Est.
2.27
2.58
.58
.52
.68
.32
.02
.13
0.940
(0.669)
0.874
1.20
b
b
b
s.e.
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.58
0.39
b
b
b
m
Est.
1.54
1.17
2.12
1.94
1.45
1.83
1.67
1.58
1.66
(1.93)
1.17
0.430
b
b
b
s.e.
0.64
0.64
0.64
0.64
0.64
0.64
0.64
0.64
0.64
0.64
0.19
b
b
b
Reduced model
Log Kb
Est.
2.30
2.17
1.95
1.81
1.44
1.54
1.09
1.09
1.09
0.46
1.18
b
b
b
s.e.
0.18
0.17
0.18
0.17
0.18
0.17
0.17
0.17
0.17
0.18
0.18
b
b
b
m
Est.
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
0.451
B
b
b
s.e.
0.18
0.18
0.18
0.18
0.18
0.18
0.18
0.18
0.18
0.18
0.24
B
B
B
OJ
OJ
      aNote that MRI originally included the data from test runs 4BNF and 8AWG for fitting No. 4 in the analyses (Reference 16).  Subsequently, it was determined
       that these were valid reasons for eliminating these tests from the analyses. The analysis results reported here are the original results including these two test
       runs; the values in parentheses for fitting four are the revised parameter values, excluding these two test runs.
      bTests 31,32, and C01 were conducted after these analyses were performed so only the individual regression results are included here.
                                                                        5-33

-------
        The results of the analyses showed that some of the control configurations tested were not
statistically different (i.e., represented the same level of control). The control configurations tested and
the corresponding control groups are shown in Table 5-7.  As shown in Table 5-7, 13 guide pole
configurations were tested that represented 7 distinct control levels.  The results of the analysis showed
that (1) the addition of a gasket to the sliding cover does not result in any appreciable reduction in
emissions for the two configurations tested in both the gasketed and ungasketed condition, (2) the
presence of a float in the guide pole  tends to reduce emissions, however, for the two configurations tested
(height of the float top or wiper at or 1 inch above the sliding cover) the height of the float wiper in the
guide pole is not a critical factor, and (3) when a pole sleeve is employed, the height of the float or float
wiper (at the sliding cover, 1-inch above the sliding cover, or 5-inches below the sliding cover) does not
have a significant impact on emissions.
                                             5-34

-------
               TABLE 5-7. RECOMMENDED GROUPING OF SLOTTED GUIDE POLE FITTINGS AND PARAMETERS FOR AP-42
Fitting
No.


1

25
3

26
20
2
31
23

4
24

29
C01
30

32
Control measure

Gasketa

0

1
0

1
1
1
1
1

1
1

1
1
1

1
Floatb

0

0
1

1
0
0
0
2

1
2

1
4
3

0
Pole
sleevea
0

0
0

0
0
1
1
0

0
1

1
1
0

1
Pole
wiper0
0

0
0

0
1
0
0
1

1
1

2
1
1

1
Comment



Presence of gasket had relatively little effect on emissions from
configuration, compared to effect of other control measures.

Presence of gasket had relatively little effect on emissions from
configuration, compared to effects of other control measures.




Position of float wiper (at or 1 inch above cover) did not
significantly affect emissions

Positions of float and pole wipers did not significantly affect
emissions when a pole sleeve was used


Configuration not included in analyses because configuration was
atypical

Parameter estimates

KF

43


31


41
11

21


11



14

8.3
KF

270


36


48
46

7.9


9.9



10

4.4
M

1.4


2.0


1.4
1.4

1.8


0.89



1.7

1.6
IJ1
OJ
         aO = control measure not implemented
          1 = control measure implemented
         bO = no float used
          1 = float with 0.25 inch gap and wiper 1 inch above cover
          2 = float with 0.25 inch gap and wiper at cover elevation
          3 = float with 0.125 inch gap and no wiper
          4 = float 5 inches below sliding cover
         C0 = no pole wiper
          1 = pole wiper at sliding cover elevation
          2 = pole wiper 6 inches above sliding cover
                                                                             5-35

-------
        Conclusions.  The general modeling approach does provide valuable information regarding the
contribution of the different controls and was useful in providing information for deciding how to group
the different fittings for establishing the AP-42 factors.  However, in some cases, factor-specific linear
regressions provide better estimates for the specific factor than the general model.
        Furthermore, in the near future API plans on establishing test protocols that will allow
manufacturers/users of tank components to evaluate evaporative losses from new deck fitting
configurations. Consequently, establishing  fitting-specific evaporative loss factors makes sense.
Therefore, the factor-specific individual regression parameter estimates (Method MRI-3) are
recommended for use  in AP-42.

        Since the analysis showed that the presence of a sliding cover gasket has relatively little effect on
emissions, separate factors for gasketed and ungasketed covers are not warranted. Consequently, the
parameter estimates from fitting Nos. 1 and 25, and 3 and 26 were combined to provide recommended AP-42
factors for slotted guide poles with a gasketed or ungasketed sliding cover and slotted guide poles with a
gasketed or ungasketed sliding cover and  a float, respectively. Similarly, because the analysis indicated that
for the configuration tested float height had little effect on emissions, the parameter estimates for fitting
Nos. 23 and 4 and for fitting Nos. 24, C01, and 29 were combined to provide the recommended AP-42 loss
factors for sliding cover, with float and pole wiper and sliding cover with float, pole sleeve, and pole wiper,
respectively. Data for fitting Nos. 2 and 31  were combined, since fitting No. 31 is identical to fitting No. 2,
but was tested after the initial test program was completed.   Since fitting No. 30 is not typical of fittings in
use, a fitting factor for this fitting configuration is not being recommended for AP-42.

        In cases where parameter estimates  for fittings were combined, the calculation was as follows:
        1. For (KFa), average of the KFa values estimated for the individual fittings;
        2. For (KFfe), average of the log(KFb) values estimated for the individual fittings; and
        3. For (m), average of the m values estimated for the individual fittings.

Table 5-7 presents the recommended fitting factors for slotted guide poles.

        5.1.4.3.2  Unslotted guide poles.

        Data analysis.  A single alternative modeling strategy was used to analyze the data gathered from
the five unslotted guide pole configurations. First, for each fitting the zero wind speed loss estimate was
subtracted from each data point.  For the three resulting data points, a linear model was fit for the log of
the loss factor versus the log of wind speed. This equation is equivalent to the power function defined
earlier with the intercept removed.  The parameter estimates obtained for unslotted guide pole fittings by
API and by this alternative method are presented in Table 5-8. The parameter estimates obtained by the
two methods differ somewhat but the resulting equations have very similar predicted emissions over the
applicable wind speed range. The equations that do differ display the  largest differences in absolute
magnitude at wind speeds of 15 miles per hour.  Since all the data used to generated the estimating
equations were obtained at wind speeds less than 12 miles per hour, the reader is cautioned against
applying the equations at wind speeds above 12 miles per hour. With this caution, the API equations for
unslotted guide poles were found to provide acceptable estimates of VOC evaporative loss rates.
                                             5-36

-------
     TABLE 5-8.  COMPARISON OF UNSLOTTED GUIDE POLE PARAMETER ESTIMATES
Fitting
18
28
27
17
19
Control measure3
Gasket
0
0
1
1
1
Float
0
0
0
0
0
Pole
sleeve
0
1
0
0
1
Pole
wiper
0
0
0
1
0
Estimation
technique
API
MRI
API
MRI
API
MRI
API
MRI
API
MRI
Parameter estimate
Kpa
31.1
31.2
25.0
24.9
25.0
24.9
13.7
13.7
8.63
8.63
Kpb
372
148
0.0267
2.21
1.05
12.8
5.78
3.73
13.8
12.3
m
1.03
1.44
4.02
2.12
3.26
2.18
0.587
0.781
0.755
0.808
   aO = control measure not implemented; 1 = control measure implemented.

       Conclusions. The unslotted guide pole parameters calculated by API using a curve fitting
procedure provide reasonable predictions of VOC evaporative losses.  Midwest Research Institute
analyzed the data by an alternative approach using a simple linear regression.  In order to force the
intercept of the fitting equation through the measured loss rate at zero wind speed (KF ), the measured loss
                                                                             a
rate at zero wind speed was subtracted from the loss rates of other wind speeds. Then, the remaining
nonzero wind speed data were plotted on a log-log scale and a linear regression performed to obtain the
other loss factors (KF  m). The emission estimating parameters calculated by MRI's procedure (and
presented in Table 5-8) are recommended for AP-42.

       5.1.4.4 Evaluation of Other External Deck Fitting Loss Factors. Fittings considered in these
analyses were deck legs assumed to have a 3 in. diameter opening, a gauge hatch/sample port assumed to
have a 8 in. diameter opening, a vacuum breaker assumed to have a  10 in. diameter opening, an access
hatch assumed to have a 24 in. diameter opening, a gauge-float well assumed to have a 20 in. diameter
opening, and a deck drain assumed to have a 3 in. diameter opening. With the exception of the deck leg
and deck drain fittings, all fittings were tested under two conditions, one without gasketing (uncontrolled)
and one with gasketing (controlled).  Four deck leg tests were conducted. One set of tests was for an
ungaskseted deck leg associated with the  center of the deck on a double deck roof.  The other three  sets of
tests were for the pontoon area under three different conditions—no control, gasket, and sock.  The deck
drain was tested under two conditions, open (uncontrolled) and 90 percent closed (controlled).

       Three distinct sets of analyses were conducted.  The first essentially replicated the procedures
used by CBI to develop the coefficients for the EFRT estimating equations proposed by API. The second
also replicated those results, but modified the procedure used to force the equation through the measured
zero wind speed value for KF . For the third set of analyses, a nonlinear model was fit to the data for all
fittings to examine  CBI's decision to model some of the fittings with a linear rather than a power function.
 Each of the analyses is described below.
                                             5-37

-------
       As described earlier (Section 5.1.4.1), the fitting loss equations were developed by CBI/API using
a three stage procedure.  Because the raw data were unavailable at the time of the review, the first step of
the procedure was not replicated, and all analyses began with the initial loss rates in units of Ib-mole/yr
generated by CBI. Note that for all fittings (except one) considered in these analyses only a single test
was conducted at each wind speed, resulting in four data points for each fitting; one data point each at
wind speeds of 0, 4.3, 8.5, and 11 mph. The MRI curve fitting analyses were performed using SAS
procedures that employ the same curve fitting techniques as those used in the CBI software package.

       For the first set of analyses, the zero wind speed data were  given a weight of 3,000 and the other
data points were given a weight of 1. The CBI procedure and MRI replication produce valid parameter
estimates, but they produce incorrect measures of the precision of those estimates. First, by weighting a
single point heavily to force the intercept (KFa) of the models to fit the measured zero wind speed
emission rate, the majority of the "data" fit the curve exactly, giving an incorrect picture of the fit.
Second, the use of the average emission rate rather than the actual calculated emission rate for each test
artificially reduces the variability of the estimates (this only relates to fitting No. 5, 3 in. roof leg--
pontoon area/no  control; for all other fittings only one test was conducted at each wind speed).

       The second set of analyses incorporated two modifications to overcome these problems.  First, to
force the models through the measured intercept, the calculated emissions at zero wind speed were
subtracted from the calculated emissions at the other wind speeds.  Then, simplified models that excluded
the term KF were fit using only the nonzero wind speed data. Second, the  actual calculated values for
           a
each test were used in the analyses, rather than the average values at each wind speed.  (Note that this
second modification only affected fitting 5 because all other fittings had only one test at each wind
speed.)  This second set of analyses produced essentially the same parameter estimates as the first set, but
it provided a more reliable measure of the precision of the estimates.

       The final set of analyses was conducted using basically the same methods as the second set, but a
power curve was fit for all fittings. The purpose of this last set was to examine the basis of the decisions
by CBI to use a linear predictor rather than a power curve for some of the fittings.

       The results of the first two sets of analyses are shown in Tables 5-9 and 5-10, respectively. Both
analyses produced equivalent parameter estimates and replicated those reported by CBI. However, the
confidence intervals for the second method, which are a much better reflection of the actual precision of
the estimates, are much wider than those for the first method. Also, many of the intervals are quite wide
relative to the parameter estimates.
                                             5-38

-------
TABLE 5-9.  SUMMARY OF PARAMETER ESTIMATES - CB&I/API REPLICATE ANALYSES
Fitting No.
5
6
7
8
9
10
11
12
13
14
15
16
21
22
Fitting description
Deck leg (3" dia.) -Pontoon
Area/No control
Deck leg (3" dia.) - Center
deck, Double Deck roof
Deck leg (3" dia.) - Pontoon
Area/gasket
Deck leg (3" dia.) - Pontoon
Area/sock
Gauge Hatch/Sample Port (8"
dia.) - ungasketed
Vacuum breaker (10" dia.) -
ungasketed
Gauge-Float well (20" dia.) -
unbolted, ungasketed
Access Hatch (24" dia.) -
unbolted/no gasket
Gauge hatch/sample well (8"
dia.) - gasketed
Vacuum breaker (10" dia.) -
gasketed
Gauge float well (20" dia.) -
unbolted/gasket
Access Hatch (24" dia.) -
unbolted/gasket
Deck drain (3" dia.) - open
Deck drain (3" dia.) - 90%
closed
Parameter estimates
A = Ka
Estimate
2.012
0.825
1.349
1.190
2.340
7.786
14.420
36.440
0.471
6.174
4.254
31.440
1.462
1.841
B=Kb
Estimate
0.3067
0.0873
0.0365
0.0614
-0.0226
0.0287
6.548
8.545
0.0205
0.9942
16.77
10.65
0.1376
0.1951
95% confidence interval
(0.3056, 0.3078)
(0.0857, 0.0089)
(0.0361, 0.0369)
(0.0610, 0.0618)
(-0.0239, -0.0213)
(0.0282, 0.0291)
(6.528, 6.568)
(8.416, 8.583)
(0.0201, 0.0209)
(0.9902, 0.9982)
(10.73, 16.80)
(10.52, 10.78)
(0.1349,0.1402)
(0.1926,0.1976)
C = m
Estimate
1
1
1
1
1
3.398
1
1
1
1
0.3891
1
1.927
1
95% confidence interval
~
~
~
~
~
(3.392, 3.404)
~
~
~
~
(0.3881,0.3990)
~
(1.919, 1.935)
~
                                5-39

-------
TABLE 5-10. SUMMARY OF PARAMETER ESTIMATES - MRI ANALYSES
Fitting No.


5
6
7
8
9
10
11
12
13
14
15
16
21
22
Fitting description


Deck leg (3" dia.) -Pontoon
Area/ungasketed
Deck leg (3" dia.) - Center deck,
Double Deck roof/ungasketed
Deck leg (3" dia.) - Pontoon
Area/gasket
Deck leg (3" dia.) - Pontoon
Area/sock
Gauge hatch/sample port (8" dia.)
- ungasketed
Vacuum breaker (10" dia.) -
ungasketed
Gauge-float well (20" dia.) -
unbolted, ungasketed
Access hatch (24" dia.) - unbolted/
ungasketed
Gauge hatch/sample port (8" dia.)
- gasketed
Vacuum breaker (10" dia.) -
gasketed
Gauge-float well (20" dia.) -
unbolted/gasketed
Access hatch (24" dia.) -
unbolted/gasketed
Deck drain (3" dia.) - open
Deck drain (3" dia.) - 90% closed
Parameter estimates
A = Ka
Estimate
2.012
0.825
1.349
1.190
2.340
7.786
14.420
36.440
0.471
6.174
4.254
31.440
1.462
1.841
B = Kb
Estimate
0.3067
[0.37]
0.0873
[0.53]
0.0365
[0.08]
0.0614
[0.14]
-0.0226
[0.0]
0.0287
[0.01]
6.548
[5.4]
8.544
[5.9]
0.0206
[0.02]
0.9943
[1.2]
16.77
[17]
10.65
[5.2]
0.1376
[0.21]
0.1951
[0.14]
95% confidence interval
(0.2276, 0.3858)
(-0.0229,0.1975)
(0.0101, 0.0629)
(0.0362, 0.0866)
(-0.110, -0.0648)
(0.0952,0.1525)
(5.145,7.951)
(5.892, 11.20)
(-0.0075, 0.0487)
(0.7195, 1.269)
(6.326, 27.20)
(1.46, 19.84)
(-0.6256, 0.900)
(0.0235, 0.3667)
C = m
Estimate
1
[0.91]
1
[0.14]
1
[0.65]
1
[0.65]
1
[0.0]
3.398
[4.0]
1
[1.1]
1
[1.2]
1
[0.97]
1
[0.94]
0.3891
[0.38]
1
[1.3]
1.927
[1.7]
1
[1.1]
95% confidence interval
~
~
~
~
~
(1.632,5.165)
~
~
~
~
(0.1042,0.6739)
~
(-0.387, 4.242)
~
                          5-40

-------
        To develop a better understanding of the effect of the uncertainty of these parameter estimates on
the actual estimated emissions generated from the equations, MRI developed graphs of the estimating
equations and 95 percent confidence intervals (95 percent CI's) for those estimates.  The graphical results
indicate that for the four roof leg conditions, there is virtually no difference among fitting Nos. 6, 7, and
8; subsequent statistical analysis showed that they can be modeled by a single equation. Generally, for
each of the 6 types of fittings, at lower wind speeds particularly, the difference in the estimates for
controlled and uncontrolled fittings is of the same order of magnitude as the uncertainty in emissions.
This uncertainty could be the result of a small sample size or of the inherent variability in emissions.

        Further examination of the data for  access hatches, vacuum breakers, and gauge-float wells raised
other concerns about the reliability of the  emissions estimates produced by these equations. First,
although the gauge-float well (20 in. diameter) is intermediate in size between the 10 in. diameter vacuum
breaker and the 24 in. diameter access hatch, the emissions at high wind speed are lower than those from
the other fittings.  Also, the pattern in reduction from adding a gasket to the access hatch is much different
than the patterns in reduction from adding gaskets to the vacuum breaker or the gauge-float well. The
vacuum breaker and gauge-float well show substantial reductions at high wind speeds when a gasket is
applied, while the access hatch shows essentially no difference.

        The results presented in Table 5-10  indicate an estimated wind speed exponent (m) equal to 1 for
all fittings except Nos.  10, 15, and 21.  In other words, a linear correlation between evaporative loss and
wind speed was assumed.  This assumption  is embedded in the approach followed by API in the data
analysis submitted to MRI. Subsequently, as with the procedure used for other analyses (e.g., guide poles
and rim-seals), both the net emission rate  (at each wind speed) and the measured wind speed were log
transformed and a least squares regression was performed in order to obtain estimates of m and log KF
These estimates were then exponentiated to  obtain an estimate of KFfe.  The fitting loss factors based on
the linear regression of the log-transformed  data are the factors recommended for AP-42. Table 5-10
presents these factors in brackets.

        It should be noted that the results for the ungasketed gauge hatch/sample port were interpreted
differently than the other fittings. If the raw test data for this fitting  are used, the model would predict
that emissions decrease as the wind speed increases. Since this result is contrary to good engineering
judgement, the emission losses at varying wind speeds were averaged to determine a single zero-wind
speed loss factor (KF = 2.2) for this fitting.  Using this approach, it is inherently assumed that there is no
wind effect on this fitting configuration.  Note that API recommended a factor based upon averaging the
emission losses at zero wind speed only (KF = 2.3).

        The deck fitting evaporative loss data evaluated as described in the preceding paragraphs were
obtained from the most recent test program conducted by API. However, this test program did not
include evaluation of several external floating deck fitting configurations. These configurations include:
        1.  Rim vent, gasketed;
        2.  Rim vent, ungasketed;
        3.  Access hatch, bolted cover, gasketed;
        4.  Gauge-float well, bolted cover, gasketed;
        5.  Deck leg, adjustable, double deck roof;
        6.  Deck leg, fixed;
        7.  Deck leg, center area, gasketed; and
        8.  Deck leg, center area, sock.
For rim vents, gasketed and ungasketed, the factors from the previous edition of AP-42 are recommended:
                                             5-41

-------
        (KFa = 0.71, Kpb = 0.10, and m = 1.0 for gasketed rim vents; Kpa = 0.68, KFfe = 1.8, and m = 10 for
ungasketed rim vents).

        For an access hatch with a bolted, gasketed cover, fitting factors based on the original IFR test
data for an access hatch with a bolted, gasketed cover are recommended (KF = 1.6, KF,  = 0, m = 0).
                                                                     a         b
Given the fact that the zero wind speed estimates for common fittings should be equivalent for both types
of floating roofs, it is consistent to use  the zero wind speed factor for an IFR tank for an EFRtank also.
In addition, it is not anticipated that there would be any wind effects on an access hatch with a bolted,
gasketed cover. In the case of the gauge-float well with a bolted, gasketed cover, the IFR factor from the
previous AP-42 edition for the same fitting was also examined. The original zero wind speed (IFR) factor
for the gauge-float well was 5.1. However, an examination of the basis of the original IFR factor revealed
that the  zero wind speed factor was based on a sum of the zero wind speed factor for an access hatch with
a bolted, gasketed cover and three times the zero wind speed factor for a 1-in. stub drain. The rationale
behind the development of this factor was that the gauge-float well was approximately the same size  as
the access hatch and the fact that there  are typically three small penetrations in the top of the float well
cover in an IFR tank. One of the small penetrations was for the cable that connected to  the float in the
well. Mr. Rob Ferry of TGB indicated that the other penetrations were sealed off for the most part.  In
reexamining the development of the zero wind speed factor for the gauge-float well, it was determined
that the  factor should be based on the sum of the zero wind speed factors for the access hatch with a
bolted, gasketed cover and the 1-in. stub drain. Thus, the 1-in stub drain was added in only once and not
three times.  This revision is recommended because on EFR tanks there is typically only one small
penetration in the gauge float well to allow the cable to pass through for the float; also, on IFR tanks  to
the remaining two penetrations are for the most part sealed off.  The revised factors are  KFa = 2.8, KFfe = 0,
m = 0.

        The same factors  for the adjustable deck leg, center area, are recommended for the adjustable
deck leg for double deck roofs (KF  =0.82, KF =0.53,m = 0.14).  The fixed deck leg factors (0) from the
previous edition remain unchanged.
                                             5-42

-------
       Factors for center-area deck legs with gaskets and socks were developed using the factors for
pontoon-area deck legs.13 Loss factors were derived using a method similar to that used to develop rim
seal loss factors when no test data were available for certain configurations.  The percent reductions
achieved by gaskets and socks on pontoon-area deck legs were applied to the ungasketed center-area leg
at wind speeds of 0, vl5 and Vj  mph (where v; = 10~100 mph and Vj = 4 mph) and the form of the loss
equation was assumed to be:
                                         E = KFa + KFbvn
where:
          E =  deck leg loss, Ib-mole/yr
        KF  =  zero wind speed loss factor, Ib-mole/yr
          a
        KFfe =  wind-dependent loss factor, lb-mole/mphm-yr
          v =  ambient wind speed, mph
         m =  wind speed exponent, dimensionless

By assigning the 0 mph loss value to KF , KF and m are determined using the derived loss values at v; and
                                    V   rb
Vj.
       Note that 21/2 in. deck legs were not tested during the most recent tests by API. This size roof leg
is atypical; consequently, factors for these fittings are not recommended in this edition of AP-42.

       Conclusions. The deck fitting parameters calculated by API using a simple linear regression
provide reasonable predictions of evaporative losses. However, to be consistent with procedures used for
determining the loss factors for other fittings, data analysis procedures based upon a linear regression of
log-transformed wind speed and evaporative losses are recommended for developing factors for deck
fittings. However, for the ungasketed gauge hatch/sample port fitting test data, this approach results in
factors that predict decreasing emissions with increasing wind speed. Since this result is contrary to good
engineering judgment, a single zero wind speed loss factor based upon the average losses measured at
zero wind speed is recommended (KF = 2.3).
                                             5-43

-------
        5.1.4.5  Evaluation of Other Internal Floating Deck Fitting Factors. Internal floating roof tanks
are equipped with a fixed roof; therefore, the wind effect on evaporative losses is mitigated. The
individual fitting loss factors for internal floating roof tanks are represented by emission measurements at
zero wind speed. Many of the fittings used on external roof tanks are also used on internal floating roof
tanks (i.e., access hatches) or the fittings are of a similar construction between tank types (i.e., column
wells on IFR's compared to guide pole wells on EFR's). Although the most recent API test program did
not include internal floating deck fittings, the zero wind speed measurement data for similar fittings were
used to revise the IFR factors, where applicable. Otherwise, the IFR fitting factors from the previous
edition of AP-42 remain unchanged. Specifically:

        1. Fixed roof support column well factor.

        a. The round pipe fixed roof suport wells with and without a gasketed sliding cover are similar in
design to the unslotted guide-pole wells with and without a gasketed sliding cover. Therefore, the new
fitting factors developed for the unslotted guide pole wells are used for the round pipe fixed roof support
column wells (KFa = 31 for ungasketed;  KFa = 25 for gasketed).

        b. The factor for  a round pipe fixed roof support column well with a flexible fabric sleeve seal
remains unchanged (KFa =10).

        c. The factors for built-up column fixed roof support wells with a gasketed sliding cover (KF
 = 33) and without a gasketed sliding cover (KF = 47) remain unchanged.

        2. Gauge hatch/sample port with slit fabric seal. This factor remains unchanged (KFa =12).

        3. Stub drain (1 in.). This factor remains unchanged (KF =  1.2).
                                                           a

        4. Ladder well with sliding cover, with and without a gasket. These factors remain unchanged
(KF = 76 for ungasketed; KF = 56 for gasketed).

        5. Deck leg for internal floating deck. This factor remains unchanged (KF =7.9).

        5.1.4.6  Recommended AP-42 Deck Fitting Loss Factors. The recommended deck fitting loss
factors for AP-42 are summarized in Table 5-11.
                                             5-44

-------
TABLE 5-11. RECOMMENDED DECK-FITTING LOSS FACTORS, KFa, KFb, AND ma
Fitting Type And Construction Details
Access hatch (24-inch diameter well)
Bolted cover, gasketedb
Unbolted cover, ungasketed
Unbolted cover, gasketed
Fixed roof support column welld
Round pipe, ungasketed sliding cover
Round pipe, gasketed sliding cover
Round pipe, flexible fabric sleeve seal
Built-up column, ungasketed sliding cover0
Built-up column, gasketed sliding cover
Unslotted guide-pole and well (8-inch
diameter unslotted pole, 21 -inch
diameter well)
Ungasketed sliding coverb
Ungasketed sliding cover w/pole
sleeveGasketed sliding cover
Gasketed sliding cover w/pole wiper
Gasketed sliding cover w/pole sleeve
Slotted guide-pole/sample well (8-inch
diameter slotted pole, 21 -inch
diameter well)6
Ungasketed or gasketed sliding cover
Ungasketed or gasketed sliding cover,
with float8
Gasketed sliding cover, with pole wiper
Gasketed sliding cover, with pole sleeve
Gasketed sliding cover, with pole sleeve
and pole wiper
Gasketed sliding cover, with float and
pole wiper8
Gasketed sliding cover, with float, pole
sleeve, and pole wiper11
Gauge-float well (automatic gauge)
Unbolted cover, ungasketed
Unbolted cover, gasketed
Bolted cover, gasketed
Gauge-hatch/sample port
Weighted mechanical actuation,
gasketedb
Weighted mechanical actuation,
ungasketed
Slit fabric seal, 10% open area0
Vacuum breaker
Weighted mechanical actuation,
ungasketed
Weighted mechanical actuation, gasketedb
Loss Factors
TV- TV-
(lb-mole/yr) (lb-mole/(mph)m-yr)

1.6 0
36° 5.9
31 5.2

31
25
10
51
33



31 150
25 2.2
25 13
14 3.7
8.6 12



43 270

31 36
41 48
11 46

8.3 4.4

21 7.9

11 9.9

14C 5.4
4.3 17
2.8 0


0.47 0.02

2.3 0
12


7.8 0.01
6.2C 1.2

Typical Number Of
m Fittings, NF
(dimensionless)
1
0
1.2
1.3
Nc
(Table 7.1-11)






1
1.4
2.1
2.2
0.78
0.81


f
1.4

2.0
1.4
1.4

1.6

1.8

0.89
1
1.1
0.38
0
1

0.97

0

Nvb (Table 7.1-13)'

4.0
0.94
                              5-45

-------
                                        Table 5-11 (cont.)
Fitting Type And Construction Details
Deck drain (3-inch diameter)
Openb
90% closed
Stub drain (1-inch diameter )k
Deck leg (3 -inch diameter)
Adjustable, internal floating deck0
Adjustable, pontoon area - ungasketedb
Adjustable, pontoon area - gasketed
Adjustable, pontoon area - sock
Adjustable, center area - ungasketedb
Adjustable, center area - gasketed™
Adjustable, center area - sock™
Adjustable, double-deck roofs
Fixed

Rim venf
Weighted mechanical actuation, ungasketed
Weighted mechanical actuation, gasketedb
Ladder well
Sliding cover, ungasketed0
Sliding cover, gasketed
Loss Factors
TV- TV-
(Ib-mole/yr) (lb-mole/(mph)m-yr)

1.5 0.21
1.8 0.14
1.2


7.9 0.37
2.0 0.08
1.3 0.14
1.2 0.53
0.82 0.11
0.53 0.16
0.49 0.53
0.82 0
0

0.68 1.8
0.71 0.10

98
56

Typical Number Of
m Fittings, NF
(dimensionless)
Nd (Table 7. 1-1 3)
1.7
1.1
Nd (Table 7. 1-1 5)
N[ (Table 7.1-15),
(Table 7. 1-14)
0.91
0.65
0.65
0.14
0.13
0.14
0.14
0

1
1.0
1.0
ld


                                                          14-16
5.1.5  Development of the Fitting Wind Speed Correction Factor

       Evaporative loss from external floating roof tanks has been shown to be wind dependent.  The old
EFRT fitting loss estimating equation included the assumption that the wind speed across the deck is
equivalent to the local ambient wind speed. It is known from field experience, however, that the shell of
the tank partially shields the floating roof from the wind. Therefore, the fitting wind speed correction
factor has been added to the deck fitting loss equation to account for the reduction in wind speed across
the floating roof as compared to the ambient wind speed. This addition results in the following form of
the fitting loss estimating equation:

                                      KF = KFa+KFb(Kvv)m
where:
               KF =loss factor for a given deck fitting, Ib-mole/yr
               KF =zero wind speed loss factor, Ib-mole/yr
               KFfe =wind speed dependent loss  factor, lb-mole/mphm-yr
               Kv =fitting wind speed correction factor, dimensionless
               v =average ambient wind speed at tank location, mph
               m =deck fitting loss exponent, dimensionless
                                             5-46

-------
       While this shielding effect was previously a known phenomenon, no data were available to
quantify the reduction in wind speed across the deck.  Therefore, wind tunnel testing, a roof height
survey, and an evaluation of field data were conducted to determine the actual reduction.

       The CPP wind tunnel test program modeled EFRT's of 48,  100, and 200 feet in diameter, with the
floating roof positioned at three different heights in each tank.14 The roof heights chosen were grouped to
result in three ranges of roof height as follows:

                                        0.35 < R/H < 0.75
                                        0.80 < R/H < 0.90
                                            R/H =1.0

(The ratio R/H is the ratio of the floating roof height to the tank shell height.)

       Average horizontal wind speeds were calculated for each roof height range at 28 locations across
each floating roof. The test program concluded that a single factor could reasonably be used to account
for the reduction in wind speeds for all areas of the floating roof, at all roof heights and tank diameters.
This factor was determined by calculating separate correction factors for each of the roof height ranges
and then calculating a weighted average of these three factors based on an assumed distribution of time
that the floating roof would spend in each height range. The distribution was based on a complete cycle
of a floating roof, where the tank begins empty, rises through  each height range, and then empties back
through each range.  This assumption results in the following  distribution:

                                        0.35 < R/H < 0.75         40 percent
                                        0.80 < R/H < 0.90         40 percent
                                            R/H =1.0             20 percent

       This test program determined that the wind speed on the floating roof is about 0.4 times the
ambient site wind speed in the first two height ranges, but increases to about 0.7 times the ambient at the
third roof height. Although the third roof height (R/H = 1.0) is not a position that occurs in the normal
operation of storage tanks, it was conservatively included in the calculation of the weighted average
correction factor. A value of 0.52 was calculated by CPP for the single fitting wind speed correction
factor.

       To investigate the validity of the assumption regarding roof height distribution, a survey of roof
heights was conducted. Forty tanks were evaluated based on  12 consecutive monthly records of liquid
level. These liquid levels were then compared to the height of the tank shell, and the ratio of liquid level
to shell height determined.  Each ratio was then assigned to one of the height ranges from the wind tunnel
study, with any readings not falling within a range being assigned to the next higher range.  This approach
resulted in tanks over 0.9 being assigned to the R/H =1.0 range.  The resulting distribution was as
follows:
                                             5-47

-------
Assigned range
0.35 < R/H < 0.75
0.80 < R/H < 0.90
R/H =1.0
Assumed distribution, percent
40
40
20
Actual distribution, percent
77.7
15.6
6.7
       While the weighted average single factor had assumed the floating roof to be at the top of the
tank shell 20 percent of the time, in the survey it was found to be in the top 10 percent of the shell height
only 6.7 percent of the time.  The distribution assumed in the wind tunnel test study was therefore
conservative compared to the distribution determined from this survey.

       Field wind speed data were also evaluated in the analysis of the fitting wind speed correction
factor. Measurements of wind speed were taken at two external floating roof tanks at a petroleum
refinery over an eleven month period. Site wind speed was measured at a platform located at the top of
the shell of one  of the tanks.  Wind speed across the floating roof of each tank was measured at two
locations on the deck, one near the perimeter and one near the center of the deck. Both horizontal and
vertical wind speed were measured. Approximately 30 readings were taken per day.

       Five months worth of data from one of these tanks were evaluated.  Daily average wind speeds
were determined at each of the two locations on the floating roof, as well as at the platform. However, in
this analysis, only the horizontal wind speed measurements were used. Due to some interruptions in the
data, only 142 days were included in the evaluation.  The wind speeds were summed for each
measurement location and the ratio of floating roof to ambient wind speed was calculated for the two
deck locations.  The resulting ratios were 0.45 for the outer area of the deck and 0.53 for the inner area.
The resulting average, 0.49, corresponds well with the value of 0.52 calculated for the single factor in the
wind tunnel test program. A correction factor of 0.5 was adopted by API and published in their draft
Manual of Petroleum Measurement Standards. Chapter 19. Section 2.5 A summary  of the analysis and
conclusions is presented in the report entitled "Documentation of the Fitting Wind-Speed Correction
Factor."15

       The single wind speed correction factor developed by API in the wind tunnel test program was
determined using generally conservative procedures.  As stated, the correction factor for the highest roof
position was weighted at 20 percent,  while the roof height survey indicated that the  roof would be at this
height less than 10 percent of the time. Further, the factor was developed based on wind speed at an
isolated tank. For a  tank farm scenario, the wind speed at the tank is expected to be only 80 percent of the
site ambient wind speed.

       The field data were evaluated and used by API to compare to the wind speed correction factor
developed from the wind tunnel study. The field data were also evaluated by MRI;  during MRI's review,
several questions were raised. Those questions and MRI's evaluation of the field data are discussed in the
following paragraphs.

       During the wind tunnel study performed to determine evaporative loss rates from deck fittings,
only a horizontal wind speed component was measured; however, the total wind speed was equal to the
horizontal wind speed (i.e., there was no vertical component). The wind tunnel turbulence was much less
than that expected on an actual tank (7 percent in the tunnel versus 20 to 100 percent expected).

                                             5-48

-------
Although the effect of turbulence on evaporative loss from fittings is unknown, the CPP report indicates
that an increase in turbulence may cause an increase in evaporative loss. During the testing, the wind
speed in the tunnel was held as constant as possible to reduce variability in loss rates and provide a stable
estimate of the rate at each test condition. Also, only the horizontal wind speed vector was measured
during the wind tunnel study performed to determine the fitting wind speed correction factor, although a
vertical wind speed component may have been present. Development of the wind speed correction factor
by API was based on the wind tunnel data and, consequently, only a horizontal wind speed vector.

       The factor API developed from their analysis of the field data agrees well with the factor they
developed from the wind tunnel study. However, both horizontal and vertical wind speed vectors were
measured in the field study. Questions regarding what effect the vertical vector of the wind speed has on
emissions and whether the vertical component should be incorporated into the calculation of the wind
speed correction factor were raised during MRI's analysis.  Since no data exists on the effects of the
vertical component of wind speed on evaporative loss, it is difficult to assess whether that component
should be included in the calculation of the wind speed correction factor.

       Another issue concerns the accuracy of the field data.  For example, there are some measurements
of deck wind speed that are more than 10 times the ambient wind speed. In addition, no information was
provided concerning the accuracy of the measurement devices. The field data were used as a validation
of the wind tunnel study, but it is unclear whether or not the data are accurate.

       Finally, the fitting loss factors in AP-42 Section 7.1 are applicable only for wind speeds up to
15 miles per hour.  Average ambient wind speeds during the field test were generally low, with a
maximum wind speed of about 6 m/s (13 mph).  However, CPP's wind tunnel tests were conducted at
wind speeds of greater than 6 m/s (13 mph).

       Analyses of the field data were conducted by MRI to replicate API's analysis  and to determine the
value of the wind speed correction factor when the vertical wind speed component is included. The first
analysis replicated API's analysis, using only the horizontal wind speed measurements. A correction
factor of 0.5 resulted. The second analysis used both the horizontal and vertical vectors of the wind speed
measured on the floating deck.  Daily average wind speeds were calculated for the 5 month period used in
the API analysis. A vector addition (Horiz2 + Vert2 = Total2) was then performed to determine a total
deck wind speed vector for each daily average at both inner and outer locations.  The ratio of deck to
ambient wind speed was calculated for each data point and measurement location. An average ratio was
then determined for the inner and the outer locations. These two ratios were then averaged and an
average fitting wind speed correction factor of 0.69 was calculated.

       The field data indicate that a vertical wind speed component is present at the deck surface on an
EFRT. However, no data are available to evaluate the effect of a vertical wind speed component on
evaporative loss. Further, no wind tunnel data are available that quantify a vertical wind speed
component for different tank configurations.

       Previously, API recommended a wind speed correction factor of 0.5 based on their analyses of
the wind tunnel and field data.  Due to the fact that no data are available on the effects of the vertical wind
speed component on evaporative loss, the most conservative approach is to use the factor calculated by
MRI from the field data, even though the accuracy of the field data is uncertain, and the field data
represent only one test condition. Therefore, the fitting wind speed correction factor recommended for
use in AP-42 is 0.7.

                                             5-49

-------
5.1.6  Deck Seam Loss Factor4-17"19

       In developing the IFRT deck seam loss factor, API assumed that deck seam losses from welded
decks are zero.  The deck seam loss factor for bolted decks was calculated based on the difference in the
average emissions from two sets of pilot tank tests with all loss sources sealed and all loss sources but
deck seams sealed. The IFRT deck seam loss factor previously in AP-42 was KD = 0.34 Ib-mole/ft-yr; it
applied to both contact and noncontact internal floating decks with bolted deck seams. Additional testing
has been conducted by CBI in their weight loss test facility on both contact and noncontact floating roof
bolted deck seams of various sizes. The purpose of the additional testing was to develop and validate
API's draft weight loss test method for the measurement of deck seam loss factors for internal floating
roof tanks.  The test data may be found in References 17-19.

       5.1.5.1  Noncontact Deck Seam Data.  The noncontact deck seam loss factor data include the pilot
scale tank test data used in developing the old deck seam loss factor and data from laboratory scale tests
conducted using different sizes of pan-type assemblies. The laboratory scale deck seam test assemblies
consist of panels bolted together with deck seams and sealed  over a rectangular reservoir which contains
the test liquid. Two test assemblies are used: a blank test assembly with no deck seams and a deck seam
test assembly. The test assemblies are filled with a volatile test liquid of known properties (normal
hexane) and are suspended from load cells. The rate of weight loss from the test assembly is then
measured over time and compared to the weight loss rate from a blank assembly without deck seams to
determine the deck seam loss factor. It was observed that for deck seams, the weight loss versus time data
could be correlated using  a linear relationship. Data are adjusted so a correlation is obtained at the
average test temperature.  Table 5-12 describes each noncontact deck seam assembly tested.
       The deck  seam loss factor, KD, for each test is  determined as follows.  First, the vapor pressure
function is calculated for the deck seam test assembly and the blank using the mean stock vapor pressure
and mean test room atmospheric pressure during  the test:
                                    P--	Z.
where:
               P = vapor pressure function, dimensionless
               P = mean vapor pressure of test liquid, psia
               Pa = mean atmospheric pressure in test room, psia
                                             5-50

-------
The loss factor for the test deck seam assembly is then determined using the loss rate, vapor pressure
function, and stock vapor molecular weight, as follows:

                                          F = E/p*MvKc

where:

               F =loss factor of the test assembly, Ib-mole/hr
               E =loss rate of the test assembly,  Ib/hr
               P* =vapor pressure function, dimensionless
               Mv =vapor  molecular weight of the test liquid, Ib/lb-mole
               Kc =product factor, dimensionless (for organics, Kc = 1)

The loss factor for the blank assembly is determined in the same manner.

        The deck seam loss factor, KD, for the test is then determined using the loss factors for the deck
seam and blank test assemblies:

                                   _ (365.25 d/yr)(24 hr/d)Qr)(FD - FB)
                                KD              IT
                                                 4Lo

where:

        KD = deck seam loss factor, Ib-mole/ft-yr
        FD = deck seam test assembly loss factor, Ib-mole/yr
        FB = blank test assembly loss factor, Ib-mole/yr
        LD = total length of the test deck seams, ft

        Deck seam loss factors for each noncontact deck seam assembly tested are given in Table 5-12.
The deck seam loss factors appear to vary with the size of the test assembly used. In addition, the
perimeter seal loss rates for blank assembly tests 9 and  13 were an order of magnitude higher than
expected,  and CBI notes in the test report (Reference 18, Interim Report No. 7, October 7, 1996) that it
was expected that the perimeter seal loss factors for these tests would be closer to those measured in tests
4, 4R, and 9R. It was discovered that the assembly used for blank test 13 had a substantial leak.
However, the repeat test, 13R, showed an even higher perimeter seal loss rate, so CBI used the data from
test 13 to calculate the deck seam loss factor for test 12 in the test report. In addition, the temperature
coefficients calculated for tests 13 and 13R are negative, indicating that as temperature increases, weight
loss decreases. Therefore, MRI discarded blank tests 13 and 13R. MRI recalculated the deck seam loss
factor for test  12 using test 9R as the blank, resulting in a deck seam loss factor of 0.030 Ib-mole/ft-yr,
ratherthan 0.015 Ib-mole/ft-yr.
                                             5-51

-------
          TABLE 5-12. NONCONTACT DECK SEAM LOSS FACTORS BY TEST
Deck seam loss factor test
Pilot scale tanka
Large test cells
Small test cells
Test assembly
description
18' 10" diameter
7'xl 8' panels
2 deck seams
6'xl6'8" panels
2 deck seams
TestNos.
Deck seam
76R
8R
8
Blank
77
9R
9
Average
I'xl6'8" panels
6 deck seams
4"x4' panels
5 deck seams
12
1
1R
2R
3
3R
9R
4
4R
4R
4
4R
Average
Deck seam loss
factor, KD
(Ib-mole/ft-yr)
0.12
0.089
0.016
0.053
0.030
0.017
0.015
0.010
0.0051
0.0064
0.011
aUsed to develop old factor.
            TABLE 5-13. CONTACT DECK SEAM LOSS FACTORS BY TEST
Deck seam loss factor test



Pilot scale tanka


Large test cell

Small test cell

Petrex pan test

Test assembly
description


18' 10" diameter
7'xl 8' panels
2 deck seams
2'4"x7' panels
6 deck seams
4"x4' panels
5 deck seams
12"x20" panel
1 deck seam
TestNos.


Deck seam
56


10

5R

2

Blank
55


NA

7R

1

Deck seam loss
factor, KD
(Ib-mole/ft-yr)

0.57


0.13

0.11

0.12

 Used to develop old factor.
                                   5-52

-------
        5.1.5.2 Contact Deck Seam Data.  The deck seam test data for internal floating decks that contact
the liquid surface include: (1) the pilot scale tank contact deck seam data used in the development of the
current deck seam loss factor; (2) a "small" contact deck seam test assembly of the same dimensions as
those in the noncontact deck seam tests; (3) a "large" contact deck seam test assembly; and (4) a small test
assembly with one deck seam, tested by Petrex, Inc. in 1984. The test assemblies are described in
Table 5-13.

        A blank test assembly was not used for the "large" test assembly (test 10) because the perimeter
seals were welded.  Otherwise, the deck seam loss factors for each test assembly were calculated in the
same manner as those for the noncontact deck seams. The loss factors for each contact deck seam
assembly tested are presented in Table 5-13.  With the exception of the pilot tank test, the loss factors are
very similar for the contact deck seam tests.

        5.1.5.3 Revised Deck Seam Loss Factor.  It can be observed from Table 5-12 that the noncontact
deck seam loss factors for the individual test assemblies vary with the size of the deck seam assembly
tested. Sufficient data are not available to determine whether the variation is a direct result of the size of
panel/assembly used, or due to some error or variation in test procedure or test assembly construction.
However, the data available do suggest that as the size of the test assembly increases, the deck seam loss
factor increases. The noncontact test data range in value  from 0.0064 to 0.12 Ib-mole/ft-yr.  In contrast,
the laboratory scale contact deck seam loss factors for the individual test assemblies are consistent with
each other, ranging from 0.11 to 0.13 Ib-mole/ft-yr. The  pilot test tank, however, exhibited a much higher
loss factor of 0.57 Ib-mole/ft-yr.

        There are a few possible explanations for the higher loss factors  from the pilot scale tank. For the
pilot tank tests, tests were  conducted (a) with all loss sources sealed, and (b) with all loss sources  sealed
except deck  seams.  The deck seam loss rate was determined as the difference in loss rate for these two
tests. However, the loss rate experienced when all loss sources were sealed was still considerable in
comparison to the loss rate attributed to the deck seams (for the case of the noncontact bolted deck, the
loss rate attributed to the deck seams was only about 4 percent of the total loss rate).  On the other hand,
in the laboratory scale testing, the  blank assembly loss rates were  an order of magnitude lower than the
deck seam assembly loss rates.

        There is a wide range in the data for the various test configurations. The data available are
inconclusive in supporting the use of any particular size panel for loss factor development.  In addition, a
t-test showed that the means of the contact and noncontact data are not statistically different at the
95 percent confidence level. Therefore, an average deck  seam loss factor was calculated for each test
assembly construction and then those data were averaged to obtain a revised deck seam loss factor.  The
resulting average  deck seam loss factor is 0.14 Ib-mole/ft-yr.  This value is approximately 40 percent of
the old deck seam loss factor of 0.34 Ib-mole/ft-yr.
                                             5-53

-------
5.1.6  Vapor Pressure Functions

       Three alternative vapor pressure functions were considered for the vapor pressure scaling
component of the floating roof tank equations. Those functions included two functions considered in
different versions of API Publication 2519 and an alternative recommended by TRW.  These three
functions are:
                                       Pi =
                                              Pv/14.7
                                           1+1-
 Pv
14.7
                                                      0.5
                                         P2 =
                                               Pv/14.7
                                                        0.7
                                              l-Pv/14.7
where:
               Pv=    true vapor pressure (psia)
               N=    a scaling exponent (optimal value to be determined from experimental data)

       Data from the 20-ft diameter pilot tank studies conducted by CBI were used to evaluate the
relative merits of the three functions.17 These studies evaluated emissions for three types of IFRT decks
a bolted noncontact deck with vapor mounted flexible wiper seals (Phases 1 and 1R), a welded contact
deck with a liquid-mounted, resilient-filled seal (Phases 2 and 2R), and a bolted, contact deck with a
vapor-mounted, resilient-filled seal  (Phases 3 and 3R). Tests were conducted primarily for
octane/propane mixtures, but limited single -component testing was conducted using n-hexane and
n-octane. The analyses of the vapor pressure function used only the octane/propane mixture data.

       Evaluation of the vapor pressure functions was conducted in two stages.  First, the data were
evaluated to determine the optimal value of N for P3*.  Then, using that value of N, linear regression
analyses were performed for each of the three test phases (i) and each vapor pressure function (P/) to
estimate the slope by for the equation:
These values, by, were then used to predict emissions for each test data point, and the correlation between
predicted and actual emissions was used to evaluate the different vapor pressure functions.
                                             5-54

-------
        To determine the optimal value of N, API first used the average emissions and vapor pressure for
each test sequence within a test phase to estimate bj and N; for the three different phases by fitting the
equation:

                                        E = bi (Pv /14.7)Ni

Then the three values of N; were averaged to obtain the optimal value of N.  The estimates of N; for the
three phases ranged from 0.62 to 0.76 with an average of 0.68 (which API appropriately rounded to 0.7).

        A better statistical approach to this problem is to fit a single model, which permits different b;'s
but only a single N, to the complete data set.  Using the data presented in documentation file B.I, MRI
conducted such an analysis and obtained an optimal value forN of 0.71, which is consistent with the API
results.  Hence, the use of N = 0.7 for P3* appears to be reasonable.

        In the second stage of the analyses, API fit separate regression models for each test phase and
vapor pressure function and used these regression models to predict emissions for each test used in the
analysis. Because the number of data points was quite small (4 to 8 per test phase) and the predictions
were developed for the same data sets that were used to develop the models, the correlation coefficients
were predictably quite high.  Furthermore,  over the range of vapor pressures used in the analysis (0.7 to
6 psia), the values of the three functions differ very little.  Because the three functions provided
comparable results,  API chose to recommend PI* because it was being used in API Publications 2517 and
2519 at the time of the analysis.

        As a part  of this study MRI reviewed the API calculations and found them to be correct.
However, the vapor pressure function has not been validated on an independent data set, so neither its
validity nor its reliability  is well established.  Consequently, the performance of the function is uncertain.
 Furthermore, no information is available on its validity for vapor pressures greater than 6 psia. A
footnote in AP-42 tables that cautions the reader about the lack of information on the performance of the
vapor pressure function at true vapor pressures greater than 6 psia is recommended.

5.2 PREDICTIVE ABILITY - ACTUAL TANK TEST DATA

5.2.1 Standing Storage Loss

        The standing storage loss equations (rim-seal and deck fitting losses) were used to predict
emissions based on actual storage tank parameters recorded from external floating roof tanks tested by the
Western Oil and Gas Association (WOGA) in 1977.8  Table 5-14 presents the parameters recorded for
each tank tested.  As shown in Table 5-14,  all of the tanks stored petroleum distillates.  Therefore,
possible biases in the emission estimation equation could be determined only for petroleum distillates.
Analyses of the performance of this equation for different stored liquids were not possible.
                                             5-55

-------
           TABLE 5-14.  TANK PARAMETERS RECORDED DURING TESTS BY THE
                           WESTERN OIL AND GAS ASSOCIATION
Tank
WOGA 134
WOGA 1757
WOGA 428
WOGA 400 x 20
WOGA 47
WOGA 48
WOGA 192
WOGA 200 10
WOGA 80210
WOGA 330
WOGA 100514
Product type
Gasoline (RVP 9)
Gasoline (RVP 9)
IP. 4 (RVP 6)
Gasoline (RVP 9)
Gasoline (RVP 10)
Gasoline (RVP 10)
Gasoline (RVP 11)
Gasoline (RVP 9)
Light Naphtha (RVP 12)
Gasoline (RVP 9)
Gasoline (RVP 9)
Tank diameter, ft
153
67
78
85
90
90
115
55
120
120
135
Type of deck
Double deck
Pontoon
Pontoon
Double deck
Pontoon
Pontoon
Double deck
Double deck
Pontoon
Double deck
Pontoon
Tank construction
Riveted
Welded
Riveted
Welded
Riveted
Riveted
Riveted
Welded
Welded
Welded
Riveted
Seal
configuration
Shoe-mounted
"Tube"
Shoe-mounted
Shoe-mounted
Shoe-mounted
Shoe-mounted
Shoe-mounted
Shoe-mounted
"Tube"
"Tube"
Shoe-mounted
       In the WOGA test report, both vapor-mounted and liquid-mounted rim-seals were characterized
as "tube" seals.  For the tanks that were reported to be equipped with "tube" seals, the emission estimation
equations were used to predict emissions for two different assumed configurations-liquid- and vapor-
mounted rim-seals. No specific information was available on the type and status (gasketed, covered, etc.)
of deck fittings on each external floating roof tank tested. However, the test report did state that the
fittings on all the tanks met the following conditions: (1) deck leg openings  were sealed; (2) emergency
deck drains were at least 90 percent covered; (3) slotted gauging devices were equipped with a floating
type plug; (4) roof guide openings were closed; and (5) all tank gauging or sampling devices were
covered,  except at the time of sampling.  Due to the limited information available on fittings and other
tank parameters, some assumptions were made. For example, tanks with pontoon decks are typically not
equipped with deck drains.  Also, most storage tanks are equipped with unslotted guide poles rather than
slotted guide poles. However to account for possible deviations from this practice, the emission
estimation equations were used to develop separate emission predictions for both slotted and unslotted
guide poles.

       Table 5-15 presents the predicted and actual emissions from 11 tanks tested by WOGA.
Column 1 presents the range in predicted emissions for liquid-mounted and  vapor-mounted rim-seals for
those tanks equipped with "tube" seals assuming the guide pole was slotted  with an ungasketed cover and
float.  Column 2 presents the predicted emissions based on the assumption that all tanks are equipped with
slotted guide poles with an ungasketed cover and float and Column 3 presents the predicted emissions
assuming the guide pole is unslotted with a gasketed cover.  For tanks equipped with "tube" seals in
Columns 2 and 3, it was assumed that the tank was equipped with the type seal that more  closely
approximated actual emissions from the tank. This assumption will minimize the "estimation error", and
if it is incorrect the error estimates will be biased low. For each tank, Column 4 presents the predicted
emissions selected from either Column 2 or 3 that most closely approximated actual emissions from the
tanks tested.
                                            5-56

-------
         TABLE 5-15.  PREDICTED AND ACTUAL EMISSION FROM TANKS TESTED
                       BY THE WESTERN OIL AND GAS ASSOCIATION
Tank
WOGA 134
WOGA 1757
WOGA 428
WOGA 400 x 20
WOGA 47
WOGA 48
WOGA 192
WOGA 200 10
WOGA 80210
WOGA 330
WOGA 100514
Predicted
emissions, lb/da
Column 1
NA
54 - 103
NA
NA
NA
NA
NA
NA
54 - 293
41 - 220
NA
Predicted
emissions, lb/db
Column 2
86
54
58
72
73
73
100
63
66
41
63
Predicted
emissions, lb/d°
Column 3
97
109
63
78
86
86
115
69
83
52
73
Predicted
emissions, Ib/d
Column 4
86
54
63
72
73
73
115
63
66
52
73
Default
emissions, Ib/d
Column 5
79
60
57
79
74
74
96
58
83
51
74
Actual
emissions, Ib/d
Column 6
33
49
108
35
32
33
265
20
25
52
167
aRange of emissions assuming vapor-mounted and liquid-mounted seals. NA = not applicable.
bAssumes guide-pole was slotted with float.
GAssumes guide-pole was unslotted with ungasketed cover.
        TABLE 5-16. RELATIVE ERRORS CALCULATED
                                 PRESENTED IN TABLE
FOR PREDICTED EMISSIONS
5-15
Data point
WOGA 134
WOGA 1757
WOGA 428
WOGA 400x20
WOGA 47
WOGA 48
WOGA 192
WOGA 200 10
WOGA 80210
WOGA 330
WOGA 100514
Relative errors
Column T
1.61
0.10
-0.46
1.06
1.28
1.21
-0.62
2.15
1.64
-0.21
-0.62
Relative errors
Column 3b
1.94
1.22
-0.42
1.23
1.69
1.61
-0.57
2.45
2.32
0.00
-0.56
Relative errors
Column 4°
1.61
0.10
-0.42
1.06
1.28
1.21
-0.57
2.15
1.64
0.00
-0.56
Relative errors
Column 5d
1.39
0.22
-0.47
1.26
1.31
1.24
-0.64
1.90
2.32
-0.02
-0.56
aMean value = 0.65; standard deviation = 1.03; and variance = 1.05.
bMean value = 0.99; standard deviation = 1.16; and variance = 1.36.
"Mean value = 0.68; standard deviation =0.99; and variance = 0.98.
dMean value = 0.72; standard deviation = 1.05; and variance = 1.10.
                                           5-57

-------
       In order to determine the predictive ability of the emission estimation equations, the relative
errors ([predicted emissions - actual emissions]/actual emissions) were calculated for Columns 2 through
4. The values for the relative errors are shown in Table 5-16.  After computing the relative errors, the
mean, standard  deviation, and variance of the relative errors were calculated for each column of data.
Table 5-16 also presents these values.

       A comparison of the means and standard deviations of the relative errors for Columns 2 and 3
shows that the predicted emissions calculated assuming that the tanks were equipped with a slotted guide
pole (Column 2) more closely approximate actual emissions than the predicted emissions calculated
assuming that the tanks were equipped with an unslotted guide pole (Column 3). The predicted emissions
based on slotted guide poles (Column 2) overestimated actual emissions by 65 percent on average;
whereas, predicted emissions based on unslotted guide poles (Column 3) overestimated actual emissions
by 99 percent on average.  It should be noted that tanks are typically equipped with unslotted guide poles
instead of slotted guide poles. Based on the mean relative error for Column 4, the best-estimate predicted
emissions overestimated actual emissions by 68 percent on average.

       In addition to using actual parameters recorded during emission testing, the emission estimation
equations were  used to predict emissions using default values  for the tanks with the exception of the type
of seal specified. Default parameters included the paint color (white), the fittings (typical), and the
meteorological  data (temperature and wind speed) based on the tank's location.  The predicted emissions
assuming default values are presented in Column 5 of Table 5-14. The predictive ability of the emission
estimation equations assuming default values was determined from the relative errors presented in
Column 5 of Table 5-16. In addition, Table 5-16 presents the mean, standard deviation, and variance of
the relative errors in Column 5.  On average the predicted emissions assuming default values
overestimated emissions by 72 percent, compared to 68 percent using actual tank parameters.

5.2.2 Internal Floating Roof Emissions

       In Appendix B of the API documentation file on internal floating roof tanks, information
regarding the predictive ability of the internal floating roof tank emission estimation procedures is
presented.  The  basis of this study was an emission test on an internal floating roof tank performed by
Radian Corporation in May 1979.20 Table 5-17 presents the tank configuration and stored liquid
properties of the internal floating roof tank tested.  The measured emissions were 77.6 Ib/d.
                                             5-58

-------
                        TABLE 5-17.  FIELD TEST TANK PARAMETERS
 Tank description
 Type
 Diameter
 Shell height
 Product volume
 Product type
 Distillation slope 10%
 Product temperature, °F
 Product vapor pressure, psia
 Product molecular weight, Ib/lb-mole
Internal floating roof
100ft
30 ft, 8 in.
926,310 gallons
Regular, unleaded gasoline (RVP 10.4)
2.5
77.5
7.4
62.4
 Internal floating roof description:
 Type
 Rim-seal type
 Deck fittings
Welded, contact
Vapor-mounted, resilient foam-filled
1 - Automatic gauge float well, unbolted cover, ungasketed
13-Column well, built-up column, sliding cover, gasketed
1 - Ladder well, sliding cover, gasketed
28 - Adjustable deck legs
1 - Sample Well, weighted mechanically actuated, ungasketed
1 - Vacuum breaker, weighted mechanically actuated, ungaskested
1 - Syphon drain
        In most cases, detailed information on the deck fittings is not available to the extent shown in
Table 5-17.  Therefore, the emission estimation procedures were used to predict emissions assuming
different levels of knowledge regarding the status of the fittings that more closely approximates what a
user of the internal floating roof tank emission estimation procedures would do to model emissions.

        In the first scenario, the emission estimation procedures were used to predict emissions by using
the fitting factors presented in Chapter 3 for all fittings except the sample well and  syphon drains. For
these fittings, a vacuum breaker was used to model emissions for the sample well and 36 1-in. stub drains
were used to model emissions from the syphon drain.  The estimated emissions under this scenario are
48 Ib/d. In the second scenario, the tank was assumed to be equipped with typical fittings and the default
quantities of those fittings. The estimated emission calculated under the second scenario are 32 Ib/d.
Under the third scenario, the tank was assumed to be equipped with typical fittings but the number of the
fittings was based on the actual tank data.  The estimated emissions under this scenario are 42 Ib/d.  For
all three scenarios, the estimated emissions underpredict the actual tank emissions of77.6 Ib/d and the
amount of the underprediction  ranged from 38 to 59 percent. Using these scenarios, it appears that the
internal floating roof tank emission estimation procedures have a tendency to underpredict emissions if
detailed information is not available for the tank. However, due to the limited availability of actual tank
test data, no conclusive findings can be presented on the predictive ability of the internal floating roof
tank procedures.
                                             5-59

-------
5.3 SENSITIVITY ANALYSES

       A sensitivity analysis was performed on both internal and external floating roof tanks to
determine the independent variables that have the most influence on the emission estimating procedures.
The results of these analyses can be used to determine the relative importance of defining a particular
parameter for a tank. If the variable has a strong influence on emissions then it is important to measure
that variable explicitly for the tank for which the emissions are estimated. Alternatively, if the variable
has only a minor influence on emissions, then the default value for the parameter can be used without
substantially biasing the emission estimate. In order to determine the influence of the independent
variables on EFRT and IFRT standing storage losses (fitting, rim-seal, and deck seam losses) and
withdrawal losses,  default tanks were used as baseline cases.

5.3.1  Standing Storage Loss

       5.3.1.1 External Floating Roof Tank. The default external floating roof tank is 153 feet in
diameter and has a pontoon floating roof with typical deck fittings. This default tank stores gasoline
(RVP 10) with a vapor molecular weight of 66 Ib/lb-mole.  The shell and roof of the tank are painted
white and the tank  is located in Long Beach, California (wind speed = 6.4 mph). Table 5-18 presents the
results of the sensitivity analyses on EFRT standing storage losses (deck fitting and rim-seal losses). The
variables that have the strongest influence on standing storage losses from EFRT's are the rim-seal factor,
the wind speed, and the guidepole fitting factor.
                                             5-60

-------
TABLE 5-18. RESULTS OF SENSITIVITY ANALYSIS FOR THE STANDING STORAGE
         LOSS EQUATION FOR EXTERNAL FLOATING ROOF TANKS
Independent variables
Default tanke
Tank characteristics
1. Paint color
a.Red primer
b . Aluminum/specular
2. Shell construction
a.Welded
b.Riveted
3. Deck type
a.Pontoon
b.Double deck
4. Fittings
a.Typical
b. Controlled
c. Guide-pole
(l)usugf
(2)usg8
(3)sug or sg w/o float11
(4)sug or sg w/float1
Meteorological conditions
1. Wind speed
a.2 mph
b.6.4 mph
c.10 mph
2.Temperature
a.66°+20°F
b.66°F
c.66° -20°F
Liquid properties
1. Vapor pressure
a.vp = 3.84 psia
b.vp = 5.74 psia
c.vp = 8.32 psia
Standing storage losses, Ib/d
Vapor-mounted primar
rim-seal
None3
234


312
255

234
NA

234
232

234
05

234
214
255
222


38
234
760

390
234
143


143
234
390
rmb wsc
f Liquid-mounted primary
rim-seal
None3
70 133 54




93 178 73
76 145 59


70 133 54
NA NA NA


70 133 54
69 132 53


70 133 54
42 105 26


70 133 54
50 113 35
91 155 76
58 122 43




17 23 18
70 133 54
268 409 95


117 222 91
70 133 54
43 81 33




43 81 33
70 133 54
117 222 91
rmb
37


49
40

37
NA

37
36

37
9

37
17
59
26


13
37
62

62
37
23


23
37
62
wsc
44


59
48

44
NA

44
43

44
16

44
25
66
33


14
44
76

74
44
27


27
44
74
Shoe-mounted primary
rim-seal
None3 rr
103


137
112

103
125

103
101

103
74

103
83
124
91


33
103
205

171
103
63


63
103
171
nb smd
43 58


58 77
47 63

43 58
53 87

43 58
42 57

43 58
15 30

43 58
24 38
65 79
32 46


14 18
43 58
73 103

72 97
43 58
26 35


26 35
43 58
72 97
                             5-61

-------
                                       TABLE 5-18.  (continued)
Independent variables
2.Vapor molecular weigh
a.MW = 46 Ib/lb-mole
b.MW = 66 Ib/lb-mole
c.MW = 86 Ib/lb-mole
Standing storage losses, Ib/d
Vapor-mounted primary
rim-seal
None3

163
234
304
rmb

49
70
93
wsc

93
133
173
Liquid-mounted primary
rim-seal
None3

38
54
71
rmb

26
37
48
wsc

31
44
58
Shoe-mounted primary
rim-seal
None3 rr

71
103
134
nb smd

30 40
43 58
56 75
3No secondary rim-seal.
bRim-mounted secondary rim-seal.
°Weather shield.
dShoe-mounted secondary rim-seal.
eThe default tank is 153 feet in diameter and is located in Long Beach, California. The tank has a pontoon floating
 roof with typical deck fittings.  The shell and roof of the tank are painted white.  The stored liquid is gasoline
 (RVP 10) [VP = 5.74 at ambient conditions]  with a vapor molecular weight of 66 Ib/lb-mole.  The wind speed at
 the tank site is 6.4 miles per hour.  The mean liquid surface temperature is 66°F.
fUnslotted, ungasketed.
8Unslotted, gasketed.
hSlotted, ungasketed or gasketed without float.
'Slotted, ungasketed or gasketed with float.
                                                5-62

-------
       Emission estimates were generated for the default EFR tank for each type of seal configuration.
As shown in Table 5-15, the emission estimates ranged from 37 Ib/d to 234 Ib/d. Based on the emission
estimates for the different types of primary rim-seals, it is most important to determine whether the rim-
seal is a vapor-mounted seal versus a liquid- or shoe-mounted seal.  In the case of vapor-mounted seals, it
is also important to know whether a secondary rim-seal is used on the tank because of the wide variability
in the emission estimates (70 to 234 Ib/d) for a vapor mounted primary only seal and a vapor mounted
primary with secondary rim-seals. For liquid-mounted seals, the variability with and without secondary
rim-seals is not as significant as that for vapor-mounted rim-seals. The emission estimates for liquid-
mounted rim-seals varied from 37 to 54 Ib/d.  For shoe-mounted rim-seals, the variability with and
without secondary rim-seals is slightly more pronounced than that for liquid-mounted  rim-seals ranging
from 43 to 103 Ib/d. Given the wide variability in the emission estimates among primary rim-seal
configurations, the reliability of the emission estimates depends strongly on identifying the primary rim-
seal type for the tank.  The relative importance of the secondary rim-seal configuration depends on the
type of primary rim-seal used.

       The color of the shell is a variable in the emission estimation procedures. The color of the shell
of the tank is used indirectly to determine the  surface temperature of the stored liquid.  For each color
combination, a solar absorptivity factor is developed that accounts for the ability of the color to absorb or
reflect heat.  The lighter the color, the less heat absorbed and the lower the liquid surface temperature.
The overall effect of shell color on the standing storage loss is fairly small for tanks that are painted with
colors that have a high reflectivity such as white, aluminum, etc. The effect becomes more pronounced if
the tank color is a dark color  such as red primer. For example, the emissions estimate  for the default EFR
tank (painted white) with vapor-mounted primary seals was 234 Ib/d. However, if the actual paint color
had been red primer, the emissions would have been underestimated by 25 percent.  On the other hand, if
the actual  paint color had been aluminum, the emissions would have been underestimated by only
8 percent.

       The type of deck used, double-deck or pontoon, has no appreciable effect on the EFR emission
estimates.  The emission estimates assuming double-deck floating roofs are slightly lower (less than
0.1 percent) than those generated for pontoon decks.

       A comparison of the  emission estimates when the tank is assumed to be equipped with typical
deck fittings versus being equipped with controlled fittings indicates that the emission estimates using
controlled fittings are 12 to 75 percent less than the estimates using typical fittings.  However, the one
fitting that has a significant effect on the emission estimate is the guide pole.  The use  of a slotted guide
pole rather than an unslotted guide pole can result in total emission increases from 10 to over 200 percent
depending upon the type of rim-seal used on the tank. Emission estimates for tanks equipped with the
more efficient rim-seal systems are more effected by inaccuracies associated with incorrect specification
of the configuration of the guide pole than are those with inefficient rim-seal systems.

       As expected, the wind speed at the tank's location is a dominant factor in the emission estimates
from the storage tank.  The wind speed of the  default tank was 6.4 miles per hour (mph). The emission
estimates for the default tank were recalculated assuming wind speeds of 2 and 10 mph. A wind  speed of
10 mph is  the default value recommended by API for use if no other data are available. The lower wind
speed of 2 mph resulted in a 50 to 85 percent decrease in emissions from those estimated at a wind speed
of 6.4 mph.  The higher wind speed of 10 mph resulted in a 70 to 280 percent increase in emissions from
these estimated at a wind speed of 6.4 miles per hour. These results highlight the importance of accurate
wind speed measurements for estimating emissions.

                                            5-63

-------
       The sensitivity of both the temperature and vapor pressure were examined concurrently since
vapor pressure is dependent upon the liquid surface temperature.  The base case temperature and vapor
pressure conditions are 66°F and 5.74 psia. A temperature change of +20°F degrees results in vapor
pressures of 3.84 psia and 8.32 psia. At the lower temperature and vapor pressure, the emission estimates
are reduced by about 40 percent from the base case.  At the higher temperature and vapor pressure, the
emission estimates are 65 to 69 percent higher than those estimated at the base case conditions.

       A change in the vapor molecular weight of the mixture also results in changes in the emission
estimates.  Vapor molecular weights of crude oils and gasolines can vary by as much as 40 Ib/lb-mole.
The base case molecular weight is 66 Ib/lb-mole. At the lower molecular weight of 46 Ib/lb-mole, the
emission estimates resulted in a 30 percent reduction from those calculated at 66 Ib/lb-mole. At the
higher molecular weight of 86 Ib/lb-mole, the emission estimates  increased by 30 percent from those
generated  at 66 Ib/lb-mole.

 5.3.1.3  Internal Floating Roof Tank. The default IFR tank is 50
feet in diameter and has a bolted deck with typical fittings. The default tank stores acetone, which has a
molecular weight of 58 Ib/lb-mole.  The shell and roof of the tank are painted white and the tank is
located in  Greensboro, North Carolina.  Table 5-19 presents the results of the sensitivity analyses on IFR
standing storage losses (deck fitting, rim-seal, and deck seam losses). The variables that have  the
strongest influence on estimating IFR standing storage losses are the physical properties of the liquid
(liquid type, vapor pressure, and molecular weight). For IFRT's, the configuration of the tank  does not
have as much influence on the estimated standing storage loss as for external floating roof tanks.
                                             5-64

-------
              TABLE 5-19. RESULTS OF SENSITIVITY ANALYSIS FOR THE
    STANDING STORAGE LOSS EQUATION FOR INTERNAL FLOATING ROOF TANKS
Independent variables
Default tankc
Tank characteristics
1 .Paint color
a.Red primer
b. Aluminum
2. Roof type
a.Column-supported
b. Self-supported
3. Deck type
a.Bolted
b.Welded
4. Deck construction
a.Cont. sheet 5 ft
b.Cont. sheet 6 ft
c.Cont. sheet 7 ft
d.Panel 5 x 7.5
e.Panel 5 x 12
5. Fittings
a.Typical
b.Controlled
Meteorological conditions
1 .Temperature
a.+20°F
b.-20°F
Liquid properties
1 .Vapor pressure
a.vp = 2.91 psia
b.vp = 1.69 psia
c.vp = 4.76 psia
2. Vapor molecular weight
a.MW = 38 Ib/lb-mole
b.MW = 58 Ib/lb-mole
c.MW = 78 Ib/lb-mole
Liquid type
a.Organic liquid (2. 91 psia)
b.Gasoline(RVPlO)
c.Crude oil (RVP 5)
Standing storage losses, Ib/d
Vapor-mounted primary rim-seal Li
None" ssb
6.3 4.3
8.4 5.8
6.9 4.7
6.3 4.3
5.2 3.3
6.3 4.3
4.6 2.6
6.3 4.3
6.1 4.1
5.9 3.9
7.3 5.3
6.9 4.9
6.3 4.3
5.6 3.6
luid-mounted primary rim-seal
None" ssb
4.1 3.5
5.4 4.7
4.4 3.8
4.1 3.5
3.0 2.4
4.1 3.5
2.4 1.8
4.1 3.5
3.8 3.3
3.6 3.1
5.1 4.5
4.7 4.1
4.1 3.5
3.4 2.8

11.2 7.7
3.5 2.4
7.2 6.2
2.3 1.9

6.3 4.3
3.5 2.4
11.2 7.7
4.1 2.8
6.3 4.3
8.5 5.8
4.1 3.5
2.3 1.9
7.2 6.2
2.7 2.3
4.1 3.5
5.5 4.7

6.3 4.3
13.7 9.4
2.1 1.5
4.1 3.5
8.9 7.6
1.4 1.2
aNo secondary rim-seal.
bRim-mounted secondary rim-seal.
The default tank is 50 feet in diameter and is located in Greensboro, North Carolina. The tank has a
bolted deck with typical deck fittings. The shell and roof of the tank are painted white. The stored
liquid is acetone [vp = 2.91 at ambient conditions] with a vapor molecular weight of 58 Ib/lb-mole.
                                          5-65

-------
       Emission estimates were generated for the default IFR tank for each type of rim-seal
configuration. As shown in Table 5-19, the emission estimates ranged from 3.5 Ib/d to 6.3 Ib/d. With
regard to rim-seal type, the estimated emissions for the rim-seal configurations presented in Table 5-16
indicate that it is most important to identify the use of a vapor-mounted primary seal without a secondary
seal versus the use of all other seal configurations. For the vapor-mounted seal without a secondary seal,
the emission estimate is 6.3 Ib/d.  For the vapor-mounted rim-seal with a secondary rim-seal and for the
liquid-mounted rim-seal with and without a secondary rim-seal, the emission estimates are 4.3, 3.5, and
4.1 Ib/d, respectively.

       The color of the shell is a variable in the IFR emission estimation procedures. The emissions
estimate for the default tank (white) with a vapor-mounted primary seal was 6.3 Ib/d.  However, if the
actual paint color had been red primer, the emissions would have been underestimated by 25 percent; if
the actual paint color had been aluminum, the emissions would have been underestimated by 9 percent.

       The type of fixed roof used has a minor influence on the IFR standing storage loss emissions.
Column-supported fixed roofs resulted in higher emission estimates than self-supported fixed roofs
because of the deck penetration required to accommodate the columns in the tank.  The emission
estimates for column-supported fixed roofs are 23 to 27 percent higher than those for self-supported fixed
roofs.

       The type of deck used, bolted or welded, has a similar effect on the IFR emission estimates as the
rim-seal configuration. The emission estimates assuming bolted decks are higher than those from welded
decks because of the deck seam losses for bolted IFR decks. The emission estimates for welded decks are
25 to 40 percent lower than those for bolted decks. In addition, the construction of the bolted deck has a
minor influence on the emission estimates. Five basic deck construction parameters were examined to
determine the effect of the variation of the deck construction on IFR standing storage losses. The
emission estimates can vary from an underestimate of 3 to 10 percent to an overestimate from 10 to
20 percent.

       The sensitivity of the estimating equations to whether controlled or uncontrolled fittings are
specified is small compared to other factors.  The emission estimates are decreased by 11 to 20 percent
when controlled fitting factors are used versus typical fitting factors.

       The sensitivity of both the temperature and vapor pressure were examined concurrently because
vapor pressure is dependent upon the liquid surface temperature. The base case temperature and vapor
pressure are 66 °F and 2.41 psia. A temperature change of+20 °F resulted in vapor pressures of 1.69 psia
and 4.76 psia.  At the lower temperature and vapor pressure, the emission estimates were reduced by
approximately 45 percent. At the higher temperature and vapor pressure, the emission estimates were 75
to 80 percent higher than those estimated at the base  case conditions.

       A change in the vapor molecular weight also results in changes to the emission estimates. At the
lower vapor molecular weight selected for the sensitivity analysis, 38 Ib/lb-mole, the estimated emissions
are 35 percent less than the estimated emissions for the base case,  58 Ib/lb-mole. At the higher molecular
weight selected for the sensitivity analysis, 78 Ib/lb-mole, the emission estimates increased by
approximately 35 percent over those calculated at the base case molecular weight, 58 Ib/lb-mole.
                                             5-66

-------
       As expected, the type of liquid stored in the tank has the strongest influence on the IFR emission
estimates. For example, if the tank had been storing crude oil (RVP 5) rather than acetone, the emission
estimates would have been overestimated by 200 to 300 percent.  If the tank had been storing gasoline
(RVP 10) instead of acetone, the emission estimates would have been underestimated by approximately
46 percent. Therefore, it is very important to determine the type(s) of liquid(s) stored in the tank during
the period over which the emission estimates are calculated.

5.3.2  Withdrawal Loss

       In terms of overall emissions from floating roof tanks, the independent variables in the
withdrawal loss equation have a fairly insignificant effect because the majority of the emissions occur
from standing storage. Typically, withdrawal loss emissions account for less than 5 percent of the overall
emissions from floating roof tanks. However, it should be noted that as the tank diameter increases, the
sensitivity of the withdrawal loss equation to certain variables increases.

       5.3.2.1  External Floating Roof Tanks.  The independent variables evaluated during the
sensitivity analysis of the withdrawal loss equation for external floating roof tanks consisted of the shell
condition, the turnover rate, and the liquid type.  The results of the analysis are presented in Table 5-20.
All three factors have an effect on emissions, but shell condition has the greatest effect.

       Table 5-20 shows that the condition of the tank shell (light rust, dense rust, or gunite lined) has a
greater effect on estimated emissions from crude oils than from gasolines.  The default value for the shell
condition is light rust. However, if the tank is gunite lined, the withdrawal loss is dramatically
underestimated compared to the base case, especially for crude oils (40 Ib/yr versus 4,000 Ib/yr).

       The influence of the turnover rate on EFR emissions is fairly insignificant compared to the effect
of the shell condition or the effect of turnovers on fixed roof tanks. At 10 turnovers per year, EFRT
withdrawal losses were estimated to  be 14 Ib/yr. At 100 turnovers per year,  withdrawal losses were
estimated at 140 Ib/yr.

       The type of liquid stored in the tank has a small impact on estimating EFRT withdrawal losses.
Estimated withdrawal losses for gasoline are 26 Ib/yr less than those estimated for crude oil. The higher
withdrawal loss estimate for crude oil is a result of the higher clingage factor for heavy crudes than for
lighter gasolines.  Heavy crude oils tend to cling to the tank shell more than the lighter gasolines.
                                             5-67

-------
   TABLE 5-20. SENSITIVITY ANALYSIS OF WITHDRAWAL
     LOSSES FROM EXTERNAL FLOATING ROOF TANKS
Independent variables
Shell condition3
Gasoline (RVP 8.5)
a.Light rust
b.Dense rust
c.Gunite lined
Crude Oil (RVP 5)
a. light rust
b. dense rust
c. gunite lined
Turnover rateb
a. 10 turnovers per year
b. 50 turnovers per year
c. 100 turnovers per year
Liquid type3
a. Crude oil (RVP 5)
b. Gasoline (RVP 8.5)
Withdrawal loss, Ib/yr

14
67
1,300
40
200
4,000

14
67
140

40
14
3Based on 10 turnovers per year.
'"Based on gasoline (RVP 8.5).
                        5-68

-------
       5.3.2.2  Internal Floating Roof Tanks.  The independent variables evaluated during the sensitivity
analysis of the withdrawal loss equation for IFRT's consisted of the column diameter, roof type, shell
condition, and the turnover rate. The results of the analysis are presented in Table 5-21.  Shell condition
has the greatest effect on estimated withdrawal losses. Turnover rate also has an effect on estimated
withdrawal loss. However, roof type essentially has no effect on estimated withdrawal loss.

       Table 5-21 shows that the  condition of the tank shell (light rust, dense rust, or gunite lined) has a
greater effect on estimated withdrawal loss when storing crude oil than when storing other volatile
organic liquids. The default value for the shell condition is light rust. However, if the tank is gunite
lined, the withdrawal losses are underestimated dramatically compared to the base case, especially for
crude oils (310 Ib/yr versus 31,000 Ib/yr).
       The type of liquid stored in the tank has an impact on the estimated withdrawal losses. Estimated
withdrawal losses for acetone were 2.7 times lower than the estimated losses for crude oils (110 Ib/yr
versus 310 Ib/yr). The higher estimated losses for crude oil is a result of the higher clingage factor for
heavy crudes than for lighter volatile organic liquids. Heavy crude oils tend to cling to the tank shell
more than the lighter volatile organic liquids.

       In the case of the turnover rate, the influence of the turnover rate on emissions is fairly
insignificant compared to the effect of the  shell condition or the effect of turnovers on fixed roof tanks.
At 10 turnovers per year, estimated emissions are  23 Ib/yr. At 100 turnovers per year, estimated
emissions are 230 Ib/yr.

       The type of fixed roof has an insignificant effect on the withdrawal loss emissions. A column-
supported fixed roof, in theory, has a higher withdrawal loss than that of a self-supported fixed roof
because of the clingage of the liquid on the columns in the column-supported fixed roofs. However, the
diameter of the columns are fairly  small, therefore, the difference between the estimated withdrawal
losses between the different roof types is insignificant (within two  significant figures).

       The column diameter also is a variable in  IFR withdrawal loss for the same reason as that stated
above for a column-supported internal floating roof tank. As with  the column-supported fixed roofs, the
column diameter has essentially no effect on the working loss emissions. Column diameters typically
range from 0.8 to  1.1 ft. The use of the various column diameters did not produce any effect on the
overall withdrawal loss estimate of 110 Ib/d.
                                             5-69

-------
TABLE 5-21. SENSITIVITY ANALYSIS OF WITHDRAWAL
  LOSSES FROM INTERNAL FLOATING ROOF TANKS
Independent variables
Shell condition
Volatile organic liquids
a.Light rust
b.Dense rust
c.Gunite lined
Crude oils
a.Light rust
b.Dense rust
c.Gunite lined
Turnover rate
a. 10 turnovers per year
b.50 turnovers per year
c.100 turnovers per year
Roof type
a. Column-supported
b. Self-supported
Column diameter
a.Built-up
b.Pipe
c. Unknown
Withdrawal loss, Ib/yr

110
570
11,000
310
1,500
31,000

23
110
230

110
110

110
110
110
                   5-70

-------
5.4 CONCLUSIONS

       The results of these analyses support the continued use of the API floating roof tank estimating
equations in AP-42. The estimating equations for tanks with mechanical shoe rim-seals appear to be very
reliable for estimating emissions for large tank populations.  The equations for the liquid- and vapor-
mounted resilient-filled rim-seals are less reliable than those for mechanical shoe seals, but they provide
reasonable estimates for large tank populations. While the equations do provide good estimates of
emissions for tank populations, their ability to present reliable estimates for single tanks is limited.
Assuming all input parameters for a single tank are correct, the inherent uncertainty in the coefficients
used makes the emission estimate imprecise.  The 95 percent confidence interval for annual emissions
from a single tank typically spans an order of magnitude or more.

5.5 REFERENCES

 1.  Evaporative Loss From External Floating Roof Tanks, Bulletin No. 2517, Third Edition, American
    Petroleum Institute, Washington, DC, 1989.

 2.  Evaporative Loss From External Floating Roof Tanks:  Documentation File for API
    Publication 2517, American Petroleum Institute, Washington, DC.

 3.  Evaporative Loss from Internal Floating Roof Tanks, API Publication 2519, American Petroleum
    Institute, Washington, DC,  March 1990.

 4.  Evaporative Loss from Internal Floating Roof Tanks: Documentation File for API Publication 2519,
    American Petroleum Institute, Washington, DC.

 5.  Manual of Petroleum Measurement Standards Chapter 19—Evaporative Loss Measurements; Section
    2—Evaporative Loss Floating-Roof Tanks, Preliminary Draft, American Petroleum Institute,
    December 31,  1994.

 6.  Documentation of Rim-seal Loss Factors for the Manual of Petroleum Measurement Standards,
    Chapter 19—Evaporative Loss from Floating-Roof Tanks, Revised draft, The TGB Partnership,
    Hillsborough, Hillsborough, NC, April 5, 1995.

 7.  Wallace, D., Evaluation of Rim-Seal Loss Factors for AP-42 Use, University of Alabama,
    Birmingham, AL, September  1995.

 8.  Hydrocarbon Emissions Floating Roof Petroleum Tanks, Engineering Science, Inc., prepared for
    Western Oil and Gas Association, Los Angeles, CA, January 1977.

 9.  Addendum to Publication 2517—Evaporative Loss From External Floating Roof Tanks, American
    Petroleum Institute, Washington, DC, May 1994.

10. Memorandum from R.  Jones,  D. Wallace, Midwest Research Institute, to D. Beauregard, U. S.
    Environmental Protection Agency, Review  of Fitting Analyses Conducted in Support of an Addendum
    to API Publication 2517 for External Floating Roof Tanks, December 5, 1994.

11. Memorandum from R.  Jones,  D. Wallace, Midwest Research Institute, to A. Pope, U. S.
    Environmental Protection Agency Review of Guide Pole Fittings Analyses Conducted in Support of
                                            5-71

-------
    an Addendum to API Publication 2517 for External Floating Roof Tanks, May 25, 1994.

12. Memorandum from A. Parker, Midwest Research Institute, to file, Final Guidepole Fitting Factors:
    Revised Analysis, November 8, 1995.

13. Memorandum from A. Parker, Midwest Research Institute, to D. Beauregard, U. S. Environmental
    Protection Agency, Final Deck Fitting Loss Factors for Use in AP-42 Section 7.1, February 23,
    1996.

14. Wind Tunnel Testing of External Floating Roof Storage Tanks, Cermak, Peterka, Peterson, Inc.,
    prepared for the American Petroleum Institute, CPP Reports 92-0869, 93-0934, and 93-1024.

15. Documentation of the  Fitting Wind-Speed Correction Factor, The TGB Partnership, Hillsborough,
    NC, March 25, 1995.

16. Memorandum from A. Parker, Midwest Research Institute, to D. Beauregard, EPA, Fitting Wind
    Speed Correction Factor for External Floating Roof Tanks, September 22, 1995.

17. Laverman, R.J., Haynie, T.J., and Newbury, J.F., Testing Program to Measure Hydrocarbon
    Emissions from a Controlled Internal Floating Roof Tank, Chicago Bridge and Iron Company,
    March 1982.

18. Chicago Bridge & Iron Technical Services Company, Loss Factor Measurements of Internal
    Floating Roof Deck Seams, Interim Reports 1-7, Prepared for the American Petroleum Institute
    Committee on Evaporation Loss Measurement, September 1994-October 1996.

19. Petrex, Inc., Test for Vapor Loss Through Clamp Bars Used on Petrex Internal Floating Roof
    System, Prepared by Petrex, Inc., March 5, 1984.

20. Field Testing Program to Determine Hydrocarbon Emissions from Floating Roof Tanks, Final
    Report, Volumes I and II, Radian Corporation, May 1979.
                                           5-72

-------
                     6.  SUMMARY OF CHANGES TO AP-42 SECTION

       The following sections summarize the major changes made since the previous version of
Section 7. I—Organic Liquid Storage Tanks (September 1997) of AP-42.
6.1  CHANGES TO EMISSION ESTIMATION PROCEDURES AND FACTORS FOR FIXED ROOF
TANKS

       An update of Section 7.1.3.1, mainly concerning low pressure tanks, was added. Table 7.1-6 was
updated, giving new paint solar absorptance factors for additional tank colors. All changes are based on
research performed by The American Petroleum Institute (API) and The TGB Partnership.  For more
information, the API Manual of Petroleum Measurement Standards, Chapter 19.1 (APIMPMS19.1)
should be consulted.
6.2  CHANGES TO EMISSION ESTIMATION PROCEDURES AND FACTORS FOR FLOATING
ROOF TANKS

       A new section was added to address emissions that originate during the landing of a floating roof.
All changes are based on research performed by The American Petroleum Institute (API) and The TGB
Partnership. For more information, the API Manual of Petroleum Measurement Standards, Chapter 19.2
(APIMPMS 19.2) should be consulted.
                                          6-1

-------