Revision addresses new approaches for storage tank loads

Dec. 8, 2003
The American Petroleum Institute (API) Standard 650, Welded Steel Tanks for Oil Storage, provides design rules for petroleum storage tanks.

The American Petroleum Institute (API) Standard 650, Welded Steel Tanks for Oil Storage, provides design rules for petroleum storage tanks. These rules are intended to provide reasonable margins against failure of these tanks due to loads imposed on them. The loads include such environmental loads as snow, wind, and earthquakes, and those due to use of the tanks, such as roof live loads, internal and external pressure, and liquid in the tanks.

Through its 10th Edition (1998), API 650 addressed loads simplistically. For example, API 650 prescribed a 25-psf live load for all fixed roofs (unconservative for some northern locations) and provided no method to determine snow loads.

Wind loads were based on a 100-mph fastest wind speed, although the US National Weather Service phased out this measurement method more than a decade ago in favor of 3-sec gust wind speeds. (3-sec gust: the maximum speed associated with an averaging time of 3 sec.)

API 650 designers usually did not design tanks for wind uplift on their roofs, even though building codes historically required this approach.

Furthermore, API 650 combined loads unscientifically. Some load combinations were very conservative: Tanks were deemed to be full of liquid when the design earthquake strikes, but empty when the design wind blows.

In other cases, API 650 provided no guidance on combining loads: For example, what portion of the snow load is to be combined with seismic loads?

Lastly, API 650 addressed loads and load combinations in various places in the standard. This made it difficult to add new loads (such as external pressure) to the standard or to ensure uniform safety margins for different load combinations and safety margins appropriate for the consequences of failure.

These approaches became increasingly troublesome as storage tanks came under closer scrutiny from building officials and as engineering practices advanced. Consequently, the API Pressure Vessel and Tank committee recently concluded a 5-year effort to address these issues and in 2005 will publish the results in the 11th Edition of API 650.

This two-part series describes the revisions to API 650 shown in that edition. This first part addresses the new loads; the second part discusses the load combinations.

Background

API 650 requires that a tank be designed so that its strength (S), usually divided by a safety factor (Fs), equals or exceeds the effect of loads (L), which are sometimes multiplied by load factors (γ).

Expressed as an equation in the case of a single load results in the following:

S /Fs ≥ γ L

For example, the tank shell's tensile strength, divided by a safety factor of 2.5, must equal or exceed the tension imposed by stored liquid pressure (Section 3.6.2.1). Another example is that an anchored tank's counterbalancing weight (Section F7.5a) must be greater than 1.5 times the uplift from the design internal pressure.

Strengths are computed with the minimum strengths given in material specifications for the tank components and the design equations provided in API 650 (for example, the equation for the required section modulus of wind girders in Section 3.9.6.1). As for other permanent structures, most loads are those that are expected to occur, on average, once every 50 years.

Dome roof and cone roof tanks, such as these at Colonial Pipeline Co.'s Greensboro, NC, pipeline terminal, are subject to API 650, which specifies loads for tanks. (Photograph by Robert Ferry, TGB Partnership, Hillsborough, NC; used with permission of Colonial Pipeline Co.)
Click here to enlarge image

null

While it is inevitable that some material will be weaker than the minimum specified strength or that the 100-year snow will fall, these events are sufficiently infrequent that tanks are not explicitly designed for their occurrence.

Factors on strength and factors on loads serve the same purpose: They provide a margin against failure, but they can be used to address different aspects of this margin. Load factors distinguish between the accuracy with which loads are known and the likelihood of different loads occurring at the same time.

In API 650, as with other structural design standards, factors on strength (also called safety factors) vary depending on the consequence of failure.

Where the consequence is severe, such as rupture of a tank full of petroleum product, the safety factor is relatively large. Conversely, when the consequence of failure is not as potentially life-threatening, such as roof columns buckling under snow load or wind damage to the tank roof, the safety factor is smaller.

API 650's load provisions may be divided into three categories: those dealing with overall loads, local loads, and load combinations.

Overall loads act on the entire surface or projected area of the roof or shell (for example, snow on the entire roof or wind on the entire projected area of the tank). The revision establishes overall loads, such as wind and snow, consistent with ASCE 7-02 (American Society of Civil Engineers' Minimum Design Loads for Buildings and Other Structures) because ASCE 7 is widely recognized as a rational basis for loads and is the basis for many building codes. The overall loads are consolidated into a new Section 3.2.1 in API 650.

Local loads act over a limited portion of the roof or shell of the tank, and include:

1. 36-psf wind pressure for wind girder requirements (3.9.6, 3.9.7).

2. External loads on shell openings (Appendix P).

3. Concentrated loads on platforms, walkways, stairs (3.8.10), rolling ladders (C.3.7), internal floating roofs (H.4.2.2), or aluminum dome roof panels (G.4.2.6.2).

4. Live loads applied to supports of floating roofs (C.3.10.2, H.4.2.5).

5. Uniform panel loads on aluminum dome roofs (G.4.2.6.1).

6. Unbalanced live loads on aluminum dome roofs (G.4.2.2.2).

The revisions leave these local loads as they currently appear in API 650 (10th Edition) and do not include them in load combinations. They are still considered, therefore, to act separately from overall loads. Overall loads can be thought of as loads averaged over the entire projected area of the tank, while local loads can be considered peak loads that act over a small part of the tank.

Loads

Before load combinations can be rationalized, the tank loads must be established. The revisions to API 650 consolidate and define overall loads in Section 3.2.1 (see accompanying box on previous page).

Although expressed differently, most of the loads are the same as in API 650 10th Edition. API 650, however, included little documentation for how its loads were determined, casting doubt on compliance with building codes and good engineering practice.

The following addresses the derivation of these loads.

Minimum roof live load

API 650 Section 3.10.2.1 specified a 25-psf minimum roof live load. This consisted of a 1-in. water column (w.c.; 5.2 psf) external pressure and a 20-psf roof live load because Section 3.2.4 states that tanks designed to API 650 can withstand a 1-in. w.c. external pressure and other API 650 provisions (such as the maximum height of unstiffened shells in 3.9.7.1) include a 5-psf external pressure.

ASCE 7's commentary notes that minimum roof live loads "provide for occasional loading due to the presence of workers and materials during repair operations" and sets minimum roof live loads (Lr) based on the roof slope and the tributary area of a structural member. The minimum roof live load, as determined by ASCE 7 Section 4.9.1, varies from 12 to 20 psf.

For low slope (less than 4:12) roofs with members with tributary areas not exceeding 200 sq ft (many supported cone roofs meet these criteria), Lr is 20 psf. For the steepest steel and aluminum dome roofs API 650 allows, the ASCE minimum roof live load based on slope considerations only is 18.4 psf, and for common profiles the minimum roof live load is 20 psf.

A 20-psf minimum live load was adopted for the revised loads even though it is slightly conservative for some dome roofs because it matches what API 650 previously used. The total downward load changes slightly, however, because the roof live load and external pressure are combined differently under the new rules for load combinations, as will be seen later.

Snow

Although API 650's 10th Edition required a 25-psf roof live load, it did not explicitly address snow loads. ASCE 7's approach is to determine the 50-year mean recurrence interval ground snow load (pg) from a map (ASCE 7 Fig. 7-1 or Table 7-1 for Alaskan sites). Then factors are applied for slope (Cs), exposure (Ce), thermal (Ct), and importance (I) effects to determine the roof snow load (ps):

ps = 0.7 Cs Ce Ct I pg

In the new API 650 Section 3.2.1(e), the slope factor (Cs) is conservatively taken as 1.0. The exposure factor (Ce) is set at 1.0, appropriate for most types of terrain. The thermal factor (Ct) is taken as 1.2, the factor for unheated structures. (For heated tanks, used for asphalt and some other services, the thermal factor is 1.0). The importance factor (I) is taken as 1.0, appropriate for ASCE 7 Category II structures.

The resulting load is 0.7(1.0)(1.0) (1.2)(1.0) pg = 0.84 pg for unheated tanks and 0.7 pg for heated tanks. For unheated tanks (the common case), a ground snow load (pg) of 23.8 psf corresponds to the 20-psf roof live load previously in API 650.

The revised API 650 allows two options for determining the ground snow load:

1. The ground snow load may be determined from ASCE 7.

2. The 50-year ground snow load may be provided by the purchaser. The purchaser-specified snow load option addresses sites outside the US because they are not covered by ASCE 7.

The new provisions also allow two options for determining the roof design snow load:

1. The design snow load may be simply taken as 0.84 pg .

2. The design snow load may determined from the ground snow load using ASCE 7 (which can give slightly lower design loads than No. 1 if the tank is heated or has a high profile roof).

Wind

API 650 (10th Edition) Section 3.11.1 specified overall wind loads of 18 psf on projected areas of cylindrical surfaces and 15 psf on projected areas of conical and double-curved surfaces, and Footnote a of 3.9.7.1 noted that the local wind load considered for wind girder design was 31 psf. These pressures were based on a 100-mph fastest mile speed.

Although API 650 specified that the 15-psf pressure be applied to "projected areas" of conical surfaces, API 650 designers traditionally applied the 15-psf pressure to the vertical projected area of cone roofs and not the horizontal projected area. They did so because applying uplift to the roof with the corresponding API 650 requirement that the tank be considered empty for the overturning stability check would require anchoring many tanks.

The practice of not applying uplift, however, exactly opposed building codes. For overturning, these codes require uplift but do not require roof horizontal pressure because the suction on the windward side exceeds suction on the leeward side, counteracting overturning.

API 650 tanks do, however, have a satisfactory history of performance against overturning, so that the challenge was to revise API 650 to follow standard engineering practice without substantially changing design outcomes. As will be seen, this was accomplished by addressing the load combinations for wind overturning.

ASCE 7 design wind pressures are determined by the basic wind speed (V), given in ASCE 7 Fig. 6-1, which is based on 3-sec gust speeds (discussed further presently). The ASCE 7 design wind pressure is:

p = qz G = 0.00256Kz Kzt Kd V2 I G

Virtually all tanks are located in exposure Category B (urban and suburban areas, wooded areas, or other terrain with numerous closely spaced obstructions having the size of single-family dwellings or larger) or exposure Category C (open terrain with scattered obstructions with heights generally less than 30 ft, including flat open country, grasslands, and all water surfaces in hurricane prone regions).

The wind load in Exposure B is about 70% of that in Exposure C, so that using Exposure C for all cases is not extremely conservative and, as will be seen presently, gives design wind pressures consistent with those previously used in API 650.

The velocity pressure exposure coefficient Kz for Exposure C at a height of 40 ft is 1.04, and less at lower heights (ASCE 7 Table 6-3). A height of 40 ft is used because the mid-height of most tanks is less than 40 ft.

The topographic factor Kzt is 1.0 for all structures except those on isolated hills or escarpments (6.5.7). The directionality factor Kd is 0.95 for round tanks (Table 6-4). The importance factor I is 1.0 for ASCE 7 Category II structures (Table 6-1). The gust factor G is 0.85 (6.5.8.1).

Click here to enlarge image

By the 1990s, the US National Weather Service changed how wind speeds are measured and recorded from fastest mile to 3-sec gust. For the default 100-mph fastest mile wind speed previously used by API 650, the wind speed was averaged over a time period equal to (1 mile)/(100 miles/hr) = 0.01 hr = 36 sec.

This being a much longer time period than 3 sec, the fastest-mile wind speeds are considerably less because peak gusts are faster over short time periods.

Converting a 100-mph fastest mile wind speed to a 3-sec gust wind speed per ASCE 7-95 Fig. C6-1 gives a 117 mph non-hurricane wind speed and a 121-mph hurricane wind speed. A 3-sec gust wind speed of 120 mph, therefore, is used here to compare to previous practice in API 650.

The conversion factor of 1.2 given to adjust fastest-mile wind speed to 3-sec gust wind speed has an error typically less than 2% for the range of wind speeds likely to be considered. Wind speeds of 3-sec gusts range from 85 to 150 mph in the US, depending on location.

The design wind pressure is shown in an accompanying box below.

This wind pressure matches the 31 psf in API 650 Section 3.9.7.1 Footnote (a) before a 5-psf internal vacuum is added. As explained in Footnote (b) of 3.9.7.1, the 36-psf pressure is assumed to act uniformly over a local area of the tank shell (equal to the theoretical buckling mode), so that a shape factor is not applied.

Click here to enlarge image

This is consistent with the distribution of wind pressure shown in Fig. 1, which shows the maximum inward pressure to occur on the windward side of the tank at a pressure equal to 1p.1 Thus a 120 mph 3-sec-gust design wind speed exerts the same wind pressure (31 psf) as that previously used for wind girder design and derived more than 40 years ago.2

Shell pressure

The average wind pressure on a cylindrical tank given in ASCE 7 is the design wind force (F) averaged over the projected area normal to the wind (Af):

F/Af = p Cf = pavg

ASCE 7 Fig. 6-19 gives a force coefficient (Cf ) of 0.5 for API tanks (round cross-section tanks with moderately smooth surfaces, a height-to-diameter [h/D] ratio of 1 and for which D.œqzw > 2.5).

Applied to the previously determined design wind pressure of 31 psf, this yields an average pressure on the tank shell of 15.5 psf. For h/D equal to 2 (rare for API tanks), Cf is 0.52 and the average wind pressure is 16.1 psf. Rather than change the current API 650 requirement of 18 psf, however, the committee conservatively left the shell pressure at 18 psf for the revised loads.

Roof pressure

For cone roofs, ASCE 7 Fig. 6-6 gives external pressure coefficients (Cp), and Fig. 6-5 gives internal pressure coefficients (GCpi) to be used to determine the design wind pressure p in the equation:

p = qGCp – qi (GCpi)

ASCE 7 classifies a tank as an enclosed structure because the area of openings in the windward shell does not exceed the sum of the area of openings in the leeward shell and roof. For roofs of enclosed structures, the design wind pressure p is (6.5.12.2.1):

p = qhGCp – qh (GCpi)

= qh (GCp - GCpi)

ASCE 7 Fig. 6-5 gives the internal pressure coefficient (GCpi) for an enclosed structure as ±0.18.

Supported cone roofs typically have a roof slope of 3/4 on 12, or 3.6° (see API 650 3.10.4.1). This is a roof-height-to-tank-diameter ratio of (3/4)/12/2 = 0.031. ASCE 7 does not specifically address cone-roof wind loads, but they could be determined by approximating the cone roof as a dome roof or as a hip roof. ASCE 7 Fig. 6-7 for dome roofs, however, indicates that hip roof pressures should be used for dome roofs if the roof-height-to-tank-diameter ratio is less than 0.05.

Therefore, wind loads for cone roofs must be determined as for hip roofs by Fig. 6-6. For both domes and cones, however, the ASCE 7 wind-pressure coefficients are a function not only of the roof-height-to-tank-diameter ratio (f/D), but also the tank-height-to-tank-diameter ratio (h/D). API tank-height-to-tank-diameter ratios vary from about 0.2 (for a 40 ft tall tank 200 ft in diameter) to about 1.33 (for a 40 ft tall tank 30 ft in diameter) and average about 0.6.

Click here to enlarge image

Considering a cone as a hip roof and using ASCE 7 Fig. 6-6 for a roof slope of 3.6° (< 10°) and a tank-height-to-tank-diameter ratio of 0.5 gives the values for Cp and pressure as shown in Table 1.

For example:

p = qh (GCp - GCpi)

p = 36.4 (-0.85(0.9) – 0.18)

= 36.4 (-0.945)

= 34.4 psf(uplift)

ASCE 7 provides wind pressures for dome roofs in its Fig. 6-7. As for cone roofs, dome pressures are a function of the tank-height-to-diameter ratio, distance from the windward edge, and roof profile.

For typical profiles permitted by API 650 (for steel domes [3.10.6.1], Rd can range from 0.8D to 1.2D, and for aluminum domes (G.6.2), Rd can range from 0.7D to 1.2D) on an 80 ft diameter, 48-ft tall tank, ASCE 7 Fig. 6-7 gives an approximate average Cp = -0.97, so that the design wind pressure is

p = qh (GCp - GCpi)

p = 36.4 (-0.97(0.85) – 0.18)

= 36.4 (-1.005)

= 36.6 psf(uplift)

Click here to enlarge image

Table 2 shows ASCE 7 wind pressure on domes for three tanks, using a dome radius equal to the tank diameter (D), a typical radius for API 650 tanks, resulting in a dome-height-to-tank-diameter ratio of 0.13.

In each case, the uplift on the windward side is about three times the uplift on the leeward side, producing a net horizontal force in a direction opposite to the wind direction. API 650 G.4.2.3.1 similarly specifies a higher outward pressure on the windward side than the leeward side. Therefore, the horizontal effect of the wind counteracts overturning and can be conservatively neglected.

The wind pressures given above for dome roofs are considerably greater than those given in API 650 for aluminum domes in G.4.2.3.1, which average 0.7(31 psf) = 22 psf. The G.4.2.3.1 pressures are less accurate because they were developed before ASCE 7 addressed dome wind loads and were based on arched roofs and do not include internal pressure.

In light of this, API selected a 30-psf roof uplift pressure as a reasonable average for all roofs for the revised loads. The new API 650 provisions, however, properly limit the wind uplift on the roof to the strength of the compression ring at the top of the tank (given in F.4.1). For self-supported cone roofs, this compression ring is often deliberately designed to be weaker than the shell-to-bottom joint so that in an overpressure, the top joint fails before the bottom joint, avoiding loss of the tank's contents.

The revision allows three options for wind speed:

1. A 120 mph 3-sec gust wind speed.

2. The site's 3-sec gust wind speed from ASCE 7.

3. Or, the site's 3-sec gust, 50-year wind speed provided by the purchaser.

The purchaser-specified wind speed is intended to address sites outside the US and its territories because ASCE 7 does not cover them. The 120-mph option provides a wind speed if ASCE 7 does not cover the site and the speed is not specified by the purchaser.

External pressure

Because even in storage tanks with open vents some external pressure acts (to empty them and when the air space in the tank cools), API 650 currently includes a 1-in. w.c. (1/2 oz/sq in., 0.036 psig, or 5.2 psf) external-pressure requirement.

Because API is developing design rules for higher external pressures, 3.2.1(i) was worded so that when such rules are added it can be easily revised.

Seismic load

API 650's Appendix E addresses seismic loads, which include the dead weight of the tank and the effect of the full amount of stored liquid for the two checks given there: vertical compression in the shell and overturning stability.

Because API is updating seismic loads to meet new building code requirements, they were addressed separately.

References

1. Boardman, H.C., "Analysis of Bending Stresses in Shell Plates Due to Wind," internal paper; Chicago Bridge and Iron Co., 1929.

2. McGrath, Raymund V., "Stability of API Standard 650 Tank Shells," Proceedings of the American Petroleum Institute, Section III-Refining, Vol. 43, pp. 458-469: New York, 1963.

Click here to enlarge image

The authors
J. Randolph Kissell [tgbjrk@ mindspring.com] is a senior partner with the TGB Partnership. He formerly was engineering manager of Conservatek Industries. He is the secretary of the engineering advisory committee of the Aluminum Association, chairman of the ASME B96 committee for welded aluminum alloy storage tanks, secretary of the American Welding Society's subcommittee on aluminum structures, and a member of the ASTM light metal alloys committee, the Canadian Standards Association's committee on strength design in aluminum, and the ASCE's load standards committee. Kissell holds a BSCE (1976) from Cornell University and is a registered professional engineer in North Carolina.

Click here to enlarge image

Phil Myers (Phil.Myers@ Chevrontexaco.com) is a codes and standards specialist at ChevronTexaco Corp. He is a tank, pressure vessel and piping specialist and a risk-management consultant for the company. He holds a chemical engineering degree from the University of California and is a registered professional engineer in California. Among his many industry activities, he chairs the API subcommittee for tanks and pressure vessels and serves on the API storage tank task force and the task force's leak-detection work group for aboveground storage tanks. He is currently chairing the API seismic task committee and is updating API 2350 for overfill protection of storage tanks.

Overall loads per revisions to API 650*

3.2.1 Loads Loads are defined as follows:

(a) Dead Load (DL): the weight of the tank or tank component, including any corrosion allowance unless otherwise noted.

(b) Stored Liquid (F): the load due to filling the tank to the design liquid level (see 3.6.3.2) with liquid with the design specific gravity specified by the purchaser.

(c) Hydrostatic Test (Ht): the load due to filling the tank with water to the design liquid level.

(d) Minimum Roof Live Load (Lr): 1.0 kPa (20 lb/ft2) on the horizontal projected area of the roof.

(e) Snow (S): The ground snow load shall be determined from ASCE 7 Fig. 7-1 or Table 7-1 unless the ground snow load that equals or exceeds the value based on a 2% annual probability of being exceeded (50 yr mean recurrence interval) is specified by the purchaser. The design snow load shall be 0.84 times the ground snow load. Alternately, the design snow load shall be determined from the ground snow load in accordance with ASCE 7. The design snow load shall be reported to the purchaser.

(f) Wind (W): The design wind speed (V) shall be 190 km/hr (120 mph), the 3 sec gust design wind speed determined from ASCE 7 Fig. 6-1, or the 3 sec gust design wind speed specified by the purchaser (this specified wind speed shall be for a 3 sec gust based on a 2% annual probability of being exceeded [50 yr mean recurrence interval]). The design wind pressure shall be 0.86 kPa [V/190]2, [(18 lbf/ft2)(V/120)2] on vertical projected areas of cylindrical surfaces and 1.44 kPa(V/190)2, [(30 lbf/ft2)(V/120)2] uplift (2) on horizontal projected areas of conical or doubly curved surfaces, where V is the 3 sec gust wind speed. The 3 sec gust wind speed used shall be reported to the purchaser.

1) These design wind pressures are in accordance with ASCE 7 for wind exposure Category C. As an alternative, pressures may be determined in accordance with ASCE 7 (exposure category and importance factor provided by purchaser) or a national standard for the specific conditions for the tank being designed.

2) The design uplift pressure on the roof (wind plus internal pressure) need not exceed 1.6 times the design pressure P determined in F.4.1.3) Windward and leeward horizontal wind loads on the roof are conservatively equal and opposite and therefore they are not included in the above pressures.

3) Windward and leeward horizontal wind loads on the roof are conservatively equal and opposite and therefore they are not included in the above pressures.

4) Fastest mile wind speed times 1.2 is approximately equal to 3 sec gust wind speed.

(g) Design Internal Pressure (Pi): shall not exceed 18 kPa (2.5 lbf/in.2).

(h) Test Pressure (Pt): as required by F.4.4 or F.7.6.

(i) Design External Pressure (Pe): shall not be less than 0.25 kPa (1 in. of water). This standard does not contain provisions for external pressures greater than 0.25 kPa.

(j) Seismic (E): seismic loads determined in accordance with Sections E.1 through E.3.

*To be part of API 650, 11th Edition (2005).