Emissions from Fixed-Roof Storage Tanks

EMISSIONS FROM FIXED-ROOF STORAGE TANKS

ABSTRACT

Benzene emission from D-14401, the Styrene Unit benzene shore tank in Block 180, has been identified by Environmental Services (ES) as one of the top priority items for abatement. Previous calculations showed benzene vent losses could range from 175 to 337 tons per year. Maximum ground level concentrations (MGLC) of benzene, due to boat unloading at 1,250 USGPM and thermal breathing, could reach as high as 7.2 ppmv using the MOE program STACK1. This compares to a currently regulated value of 3 ppmv in Reg 308 (Concentration at point of impingement - 1/2 hour average). ES recommended that D-14401 be modified to contain the losses and to limit the MGLC to 1 ppmv. It should be noted, there has been a recent decision to locate the Kiln Project next to the D-14401 dyked area.

In order to clarify design procedures, an engineering design manual, Emissions From Fixed-Roof Storage Tanks, was prepared to assist in the training of junior engineers working in this area for the first time. To expedite preparation of the manual, the routines were developed on the Macintosh computer using a spreadsheet. D-14401 was used as an example.

D-14401 SUMMARY

Calculations now show the total venting losses from D-14401 are no more than 90 tons per year. From an environmental regulation point of view, the MGLCs (5 mph wind, "D" stability, 200m) due to boat unloading at 1,300 USGPM and thermal breathing (9 AM Peak) are:

BenzenePump InBreath'gTotal Month ppmv lb/hr lb/hr lb/hr

June 3.92 205.27112.21317.48

Aug3.77 217.7797.58315.35

May 2.81160.0667.55227.61

Mar1.28 83.4520.31103.76

90% of the time during the year the losses are due to thermal breathing alone. During August and June at 9 AM the peak MGLC, due to thermal breathing alone, reaches approximately 1.2 ppmv, or very nearly the ES recommended limit.

Because D-14401 has a low design pressure (2 inches of water) and very narrow working pressure range (0.67 inches of water), only very low pressure drop options can be considered to reduce the losses.

Thermal vent losses can be reduced by 90 to 95% by installing an internal floating roof. However, during boat unloading at 1,300 USGPM, seal leakage benzene will be displaced and the MGLC will still reach 2 ppmv. Technical problems associated with an internal floating roof must still be resolved.

In order to reach the recommended MGLC level of 1 ppmv during high boat unloading rates, vapor return lines, flaring or burning at the Power Plant or in the future Kiln (Kiln Project) should be considered.

CONCLUSIONS AND RECOMMENDATIONS

1.0 It is recommended that this instruction be used as a training manual for junior engineers working on storage tank venting for the first time.

2.0Some of the DOWCAD programs on the Vax, namely VENT2 and BLOSS, need to be corrected and enhanced. If resources can be made available, the new programs, described herein, should be added to DOWCAD.

3.0This instruction gives us some insight to the environmental problems associated with boat unloading. 90% of the time during normal breathing only, the maximum ground level concentration (MGLC) of benzene should be less than the ES recommended value of 1 ppmv.

4.0During boat unloading at 800 USGPM, with an internal floating roof, and even with benzene refrigerated to 60°F, the MGLC of benzene will reach a 1.3 ppmv. Hence, the only way to solve the environmental problem is to eliminate boat unloading or to burn the vents during boat unloading.

5.0 It is recommended that the following options be examined in further detail:

(a) An internal floating roof,

(b) A vapor return line to the boat,

(c) A vapor line to burn the vapors in the Power Plant or in the Kiln Project,

(d) A vapor line to a small local ground level flare.

DESIGN MANUAL INTRODUCTION

Environmental Services (ES) recently made a study of emissions of volatile organic chemicals (VOC's) from a number of fixed-roof, API type storage tanks located throughout the Division. Benzene losses from D-14401, at 337 tons/yr, were found to be one of the Division's top 5 priorities for abatement. Under worst-case boat unloading and out breathing conditions, the vents from D-14401 gave a benzene ground level concentration of 7.2 ppmv, which is well over the current Reg 308 limit of 3 ppmv. ES recommended we achieve 1 ppmv and that benzene be stored in a tank, which meets the "Best Available Technology Economically Achievable" criteria.

Process Engineering was asked to evaluate a number of alternatives and to recommend the most viable method to reduce the yearly benzene losses and to meet their recommended air quality requirement.

On reviewing all the process engineering work to-date there appeared to be some conceptualizing and calculation problems with the alternatives that had been considered.

This report presents a number of design routines on the Macintosh computer that will assist in the evaluation of alternatives to reduce emissions from low-pressure API storage tanks. The programs and routines are:

* BLOSS, which more reasonably predicts the yearly breathing evaporation losses.

* API-2000, for PVRV and ERV design and for specification of the maximum allowable working pressure range of a low pressure API storage tank.

* API-2518, the unabridged and modified version for calculating the yearly breathing evaporation losses from fixed-roof API tanks.

* API-2517, for calculating yearly breathing evaporation losses from external floating roof API tanks.

* D.M. Gammell's method for determining the maximum vapor flow to a vapor recovery or to a vapor collection system.

* Typical Loading/Unloading logistics to demonstrate how normal volume change affects losses, how monthly liquid level affects losses and how inventory and loading factors can be manipulated to reduce yearly losses.

* Zanker's hourly out breathing routine to determine maximum flow rates and concentrations for dispersion studies.

Benzene losses from D-14401 will be used as an example.

INSTRUCTIONAL SUMMARY

The following is a summary of the design routines that should be used when determining:

(a) Yearly venting losses.

(b) Flow rates to a vapor recovery system.

(c) Flow rates for pressure / vacuum ERV and PVRVs.

(d) Worst case flow rates and concentrations for dispersion modeling.

Venting from D-14401, Plant 52's benzene shore tank in Block 180, is used as an example. The reader is reminded that the primary purpose of this report is to be instructional for first time junior process engineers. Although D-14401 is used as an example, the results described herein are only Phase I quality. Accordingly, any recommendations that are made should be considered as direction for further detailed studies.

(1)One of the first things the designer must do is determine the yearly loading and unloading logistics for the storage tank. Factors that will affect the yearly breathing losses are the:

(a) Monthly liquid levels,

(b) Maximum or minimum inventory policy,

(b) Pump in and pump out flow rates,

The more accurate the loading logistics, the more accurate will be the prediction of breathing losses. Examples for D-14401 are included to demonstrate the role loading logistics plays in storage tank venting. Future work should examine the logistics and decide the operating parameters that will minimize the impact on the environment. Options that can be examined are:

(i) It has already been recognized that unloading a boat at 1/2 rates helps meet current environmental criteria.

(ii) Keeping D-14401 inventory levels as high as economically and safely possible during warm summer months also helps.

(iii) Filling D-14401 only between the hours of 2 PM and 2 AM so that inbreathing compensates somewhat for filling losses.

2.0Prior to becoming involved with storage tank venting, the designer needs to read some reference literature. The designer should be familiar with API-2518 for predicting evaporation loss from fixed-roof tanks. The designer should also become familiar with the study of Beckman and Gilmer who developed a model for predicting emissions from fixed-roof storage tanks. Beckman showed the API-2518 method was high by a factor of two. Beckman also gives an excellent account of vapor movement vs time, which teaches us how to establish maximum flows and concentrations for dispersion studies.

The Macintosh computer program contains many Beckman features. The unabridged API-2518 version and a modified version of API-2518, which predicts lower answers more in line with the findings of Beckman, are also available in the Mac program.

3.0 One of the next things the designer should do is determine the magnitude of the vent losses. The Zanker method on the Macintosh computer should be used to predict yearly breathing losses. When coupled to loading logistics, described above, when average daily temperature profiles are used, and when vapor pressures are set to the proper liquid or ambient temperatures, this method predicts a more reasonable, though still conservative, yearly breathing loss.

Pump in losses should be calculated by the simple volumetric displacement of vapors at particular times during the day. This program assumes filling losses equivalent to the worst-case composition and therefore these losses are conservative on a yearly basis. Previously, the pump-in losses were determined from a program that should only be used to size relief valves, which have a safety factor of two built into them. See numerous examples in the Zanker file.

Previous calculations on D-14401 showed benzene losses ranged between 175 and 337 tons/year. The new, more rigorous calculations now show; if the company continues to follow a minimum inventory policy and if liquid levels are allowed to be as high as possible under this policy during the warm summer months, the filling and breathing benzene losses should not exceed 80 tons per year. In dollar value this is a loss of $25,600 per year.

This low dollar value suggests that any viable vapor recovery system will have to be fairly inexpensive to give a positive ROI. If a positive ROI is not possible, then the most economic system to reduce the environmental impact will be the one with the lowest capital and operating costs.

4.0Pressure and vacuum protection of a very fragile API tank is of primary importance from a loss of containment point of view. There are many examples where API tanks have failed because the system was not designed correctly or the tank operated outside narrow pressure limits. Prior to considering vapor recovery alternatives, the designer needs to know the mechanical strength of the vessel and the ERV and PRV settings which will dictate the working pressure range available for normal movement of vapor in and out of the tank.

API-2000 is generally followed to establish ERV and PVRV flows. PVRV settings are established after detailed study of the working pressure ranges required for the vapor recovery system, if any. Unless you have knowledge of the vapor recovery system or the padding system, you can get caught in a time consuming cycle trying to get the whole system designed properly.

By design, the API tank provides economical storage if the design and vacuum setting can be kept as low as possible. The economic advantage can be easily lost if the tank is designed for a high design pressure in order to allow a comfortable working pressure range. Considering all the above complexities, these details need to be examined very closely by senior design personnel to get the system right. For existing tanks retrofitted into new services, you work with given pressures and there may be limited opportunity for vapor recovery systems.

The Macintosh computer program includes API-2000 routines for establishing flows to the ERV and the PVRV and for establishing the working pressure available for a vapor recovery system. See example.

Since D-14401 can only withstand a maximum pressure of 2 inches of water, pressure relief protection (PRV) is set at 1 inch of water. Since the maximum working pressure should be below the PRV set pressure, say 0.67 inches of water, there is insufficient working pressure differential (0 to 0.67 inches of water) to consider vapor recovery options. Preliminary discussions with our mechanical design department have indicated that strengthening the tank may be a possibility, but that it would be expensive. Based on this information, vapor recovery options were not considered any further.

5.0If a vapor recovery system is a viable option, the method of Gammell on the Macintosh shows us how to determine the maximum flow rate to the system. Gammell uses a modified API-2518 method, which uses mean of monthly temperature data. Although Gammell tells us the maximum flow rate for pressure drop and compressor sizing, it does not tell us the size of carbon beds or regeneration cycles within the vapor recovery system. These factors are usually determined by consultation with vendors. Examples are included.

6.0One of the main concerns on venting hydrocarbons is the flammability aspect and whether or not a flame can flash back into the tank. The lower flammable limit (LFL) of benzene is 1.35% by volume and the upper flammable limit (UFL) is 6.75%. The vapor composition can be as high as 12.5% when the liquid is 77°F. Hence, the vents pass through the flammable range as they disperse in the atmosphere. The PRV for D-14401 was designed to breath through a flame arrestor. However, since benzene freezes at 42°F, there is the risk of plugging the flame arrestor and it was suggested by operations that the PRV on D-14401 no longer contains the flame arrestor. For D-14401, the flammability problem needs to be looked at as soon as possible. Generally the tank is padded with an inert gas, such as nitrogen, if a flame arrestor is not suitable. D-14401 is not padded with nitrogen. The vapor space contains free breathing air.

7.0Converting D-14401 to an internal floating roof tank ($162,000 to + $350,000) appears to be one of the remaining alternatives to reduce the yearly losses. Horton CBI, Limited claim 95 to 97% reduction of yearly losses, based on limited pilot plant data. If CBI is correct, this would reduce the yearly breathing losses from 63 to 3 tons/year. However, since the vapor space above the floating roof will still contain benzene, the filling losses during boat unloading will still be an environmental problem. In addition, since benzene freezes at 42°F, the vapor space must be kept warm to prevent condensation on the walls and the roof. Keeping these factors in mind, there is still a fair bit of technology to learn about internal floating roofs.

8.0 Previous ES dispersion calculations were based on pump in flows from programs that are normally used to size PRVs. Consequently the previous flow rates were conservative by factors of 2. The work of Beckman and Gilmer showed us that the maximum rate of venting occurs at 9 AM and in a month when there is a large temperature swing between night and day. The work of Zanker gives us a method to calculate daily tank breathing losses based on rigorous calculus. However, in the integrated form it is not suitable to predict the maximum hourly rate.

In order to perform an incremental analysis, the two fundamental equations developed by Zanker were solved using the classical trapezoidal rule (calculate the area under the curve for small increments). This allows the designer to study the rate of change of vapor pressure with time and from this you can see how the vapor flows change incrementally between the hours of 2 AM and 2 PM. By using the trapezoidal rule, called the Zanker Hourly Method, the findings of Beckman were confirmed, namely that the maximum flow rate occurs at 9 AM. In addition, the results from the Zanker Hourly Method confirmed the integrated solution for a particular month. This showed us Zanker integrated the two fundamental equations correctly when an error in the original article was corrected.

The Zanker Hourly Method on the Macintosh computer allows the designer to include pump in filling losses with the breathing losses. Hence, the program determines flows and compositions for any combination of pump in and breathing flows at any time of the day.

Based on the Zanker Hourly Method and STACK1, the current MOE dispersion program, a number of dispersion studies were performed to test for compliance. Table 1 summarizes the D-14401 dispersion studies.

9.0Case 2 shows a maximum ground level concentration (MGLC) of 1.17 ppmv for "D" stability at 5 mph wind and for breathing only at maximum rates at 9 AM in August. This shows that most of the time (≈ 300 days per year) we should be venting at less than the ES recommended level of 1 ppmv. Only during June, July and August does normal breathing exceed 1 ppmv, and not by much.

10.0Case 7 on Table 1 shows a maximum out breathing and boat unloading pump in loss of 317.48 lb/hr of benzene at 9AM if the boat is unloaded during June. During August (Case 1), the rate is 315.35 lb/hr. During June, this study shows the MGLC to be 3.92 ppmv for "D" stability at 5 mph wind. The ES study used 603 lb/hr for boat unloading and breathing and MGLC was 7.2 ppmv. Hence, boat unloading and breathing during the summer months puts us out of compliance. Production planning should look at inventory logistics to see if boat unloading can be restricted between June and October.