Tank Overturning Stability Criteria

[Muhammad@ZEL]

Hi All,
I hope this message finds you well.
I am currently reviewing the anchorage and overturning stability requirements for vertical aboveground storage tanks designed in accordance with API 650, particularly under the combined influence of wind and seismic loads.

Based on my understanding so far, the need for anchorage arises primarily from two major load scenarios:

Wind Case

1. For unanchored tanks, API 650 outlines the following uplift stability criteria under wind loads and internal pressure:​

Unanchored tanks shall satisfy all of the following conditions:​

1.  0.6Mw + MPi < MDL / 1.5 + MDLR
2.  Mw + Fp(MPi) < (MDL + MF)/2 + MDLR
3.  Mws + Fp(MPi) < MDL / 1.5 + MDLR

For unanchored tanks with supported cone roofs, satisfying the following condition applies:
  Mws + Fp(MPi) < MDL / 1.5 + MDLR

It is clear that:

• The tank's ability to resist wind uplift strongly depends on the actual liquid level at the time of the wind event.
• An empty or near-empty tank becomes particularly vulnerable to uplift.

I would appreciate further clarification on how to treat uplift stability in scenarios where the tank may be only partially filled (e.g., dead stock level only) — which is a likely operational scenario in industry. In cases where a tank operates at minimum dead stock level (e.g., 5%–10% fill), how is the resisting moment treated in wind uplift checks? Is the liquid weight ignored entirely?
And Also

• if I run calculations using full liquid height (up to shell height), the tank passes easily.
• However, if I use the actual design liquid level (DLL) instead of shell height — uplift may fail, requiring anchors

So, what’s the proper industry practice here? Should wind uplift checks be done using:

• Design Liquid Level (DLL)?
• Full shell height?
• Or even dead stock level (worst case)? (do we separately consider it ?)

Seismic Case
Also, in API's seismic Examples, the product height is taken up to full shell height — not the design liquid level.

1. Is it acceptable to use design liquid level (DLL) instead of full shell height when calculating seismic overturning moment and anchor bolt sizing?
2. While testing different design scenarios, I observed that:

When I use the full shell height as the product height in seismic overturning calculations, the required anchor bolt size increases noticeably. However, if I reduce the product height to the design liquid level the required bolt size decreases.

Freeboard
API 650 Annex E (Seismic Design) considers sloshing and requires freeboard per Table E.7 for SUG II & III, but it's optional for SUG I.
"Sloshing of the liquid within the tank or vessel shall be considered in determining the freeboard required above the top capacity liquid level."
“Purchaser shall specify whether freeboard is desired for SUG I tanks.”
So does the freeboard affect anchor design or overturning checks, or is it only for sloshing margin?


Tank Design Data (For Reference):
Location: Pakistan
Seismic Zone: 2A
Peak Ground Acceleration (Sp): 0.16g
Seismic Use Group (SUG): I
Site Class: D
Tank Diameter: 8.0 m
Shell Height: 13.0 m
Product Specific Gravity: 0.76
Roof Type: Self-supporting Cone
Bottom Type: Cone Down
Shell Thickness (1st course, corroded): 6 mm
Bottom Plate Thickness: 8 mm
Wind Speed: 160 km/h (3-sec gust)

Best regards,

 

[sdz]

Muhammad@ZEL,

For the depth of liquid refer to;

5.11.2.3 The liquid weight (wL) is the weight of a band of liquid at the shell using a specific gravity of 0.7 and a height of one-half the design liquid height H. wL shall be the lesser of 140.8 HD for SI Units (0.90 HD for USC units) or the following:

i.e you use 1/2 the design liquid depth AND only consider an strip around the edge of the tank. Most of the liquid weight is ground supported and does not resist overturning.

I assume this is based on the low probability that the design wind will occur when the tank is empty.

I am a little perplexed by the overturning criteria myself

API650-12ed 5.11.2.1 Unanchored tanks shall satisfy all of the following uplift criteria:

1) 0.6Mw + MPi < MDL /1.5 + MDLR
2) Mw + Fp(MPi) < (MDL + MF)/2 + MDLR
3) Mws + Fp (MPi) < MDL /1.5 + MDLR

where

FP is the pressure combination factor, see 5.2.2; defined as the ratio of normal operating pressure to design pressure, with a minimum value of 0.4.
Mpi is the moment about the shell-to-bottom joint from design internal pressure;
Mw is the overturning moment about the shell-to-bottom joint from horizontal plus vertical wind pressure;
MDL is the moment about the shell-to-bottom joint from the nominal weight of the shell;
MF is the moment about the shell-to-bottom joint from liquid weight; (see 5.11.2.3, tank half full)
MDLR is the moment about the shell-to-bottom joint from the nominal weight of the roof plate plus any attached structural;
Mws is the overturning moment about the shell-to-bottom joint from horizontal wind pressure.

I really don't understand;
1) why use 2/3 of the wall weight but all the roof weight?
2) why use 1/2 of the wall weight but all the roof weight?
3). In what circumstances will you have horizontal wind pressure without vertical wind pressure?

 

[Muhammad@ZEL]

Muhammad@ZEL,

For the depth of liquid refer to;


i.e you use 1/2 the design liquid depth AND only consider an strip around the edge of the tank. Most of the liquid weight is ground supported and does not resist overturning.

I assume this is based on the low probability that the design wind will occur when the tank is empty.

I am a little perplexed by the overturning criteria myself


I really don't understand;
1) why use 2/3 of the wall weight but all the roof weight?
2) why use 1/2 of the wall weight but all the roof weight?
3). In what circumstances will you have horizontal wind pressure without vertical wind pressure?

Thanks for your response.
Honestly, I’m also a bit confused and still looking into it.

 

[JStephen]

I think one of the principles being used here is that if you have two independent effects, you don't necessarily assume that both happen at the same time. So, for example, for vertical and lateral seismic acceleration, both are considered, but one is reduced when combined with the other.

With the tanks and wind, you have two varying conditions: Tank "emptiness" and wind speed. So both COULD hit maximum at the same time, but it's less likely. So I think API is trying to factor that in, where you figure full wind plus some emptiness or some wind plus complete emptiness but not full wind plus complete emptiness.

In the older codes, only lateral wind loading was considered, no uplift, but then no product was considered, either. I think the results were reasonable, but calculated wind loads were lower than reality, but then overturn resistance was higher than reality, so overall results were generally satisfactory.

Another issue that plays in here, a great many of the tanks subject to higher winds are in hurricane zones (IE, Gulf coast) and tank operators have some advance notice when a hurricane is blowing in. So, if desired, they can make sure tanks are not completely empty. Net effect is that experience with winds is probably better than it would be if winds just randomly hit 140 mph one day.

Summary on that: If the tank ONLY needs to comply with API-650, you check the three conditions there, and if it meets them, it need not be anchored. Per the standard, you don't need to worry about "but what if it's empty" cases specifically.

If you want to be extra conservative, or you think your tank is likely to be completely empty 2/3 the time or something, there's nothing preventing you from factoring that in.

If the tank needs to comply with area building codes, ASCE 7-16 and newer have wind uplift provisions that may require anchorage and do not have any provisions for product. So a tank may require anchorage per the building code even when not required by API-650. Note especially that groups of 3 or more tanks will jump the wind loads up, whereas being able to use Exposure B in some cases will reduce the loads. I think ASCE 7 also needs to be revised allow for this "emptiness" factor, but I don't know if any effort is underway to rewrite that.

For the seismic loading, H is defined as "Maximum design product level", so yes, it can be less than shell height.

On the 1/2 or 2/3 of weights used, I have interpreted that as being different safety factors applied to those items, not necessarily just using a portion of the weight. But I don't know the reasoning behind the difference. I think part of the reasoning here is still trying to bring calculated overturning results in line with actual overturning results. IE, if tanks were overturning right and left all over the country, they'd be jacking those safety factors up. But if thousands of tanks have set there for decades with no problem, then you come along and calculate that they're all in danger of overturn, it's tempting to adjust your calculation methods to match reality rather than the other way around. So they've added increased overturning forces in the code, but also added increased overturning resistance to try to compensate.

 

[Muhammad@ZEL]

I think one of the principles being used here is that if you have two independent effects, you don't necessarily assume that both happen at the same time. So, for example, for vertical and lateral seismic acceleration, both are considered, but one is reduced when combined with the other.

With the tanks and wind, you have two varying conditions: Tank "emptiness" and wind speed. So both COULD hit maximum at the same time, but it's less likely. So I think API is trying to factor that in, where you figure full wind plus some emptiness or some wind plus complete emptiness but not full wind plus complete emptiness.

In the older codes, only lateral wind loading was considered, no uplift, but then no product was considered, either. I think the results were reasonable, but calculated wind loads were lower than reality, but then overturn resistance was higher than reality, so overall results were generally satisfactory.

Another issue that plays in here, a great many of the tanks subject to higher winds are in hurricane zones (IE, Gulf coast) and tank operators have some advance notice when a hurricane is blowing in. So, if desired, they can make sure tanks are not completely empty. Net effect is that experience with winds is probably better than it would be if winds just randomly hit 140 mph one day.

Summary on that: If the tank ONLY needs to comply with API-650, you check the three conditions there, and if it meets them, it need not be anchored. Per the standard, you don't need to worry about "but what if it's empty" cases specifically.

If you want to be extra conservative, or you think your tank is likely to be completely empty 2/3 the time or something, there's nothing preventing you from factoring that in.

If the tank needs to comply with area building codes, ASCE 7-16 and newer have wind uplift provisions that may require anchorage and do not have any provisions for product. So a tank may require anchorage per the building code even when not required by API-650. Note especially that groups of 3 or more tanks will jump the wind loads up, whereas being able to use Exposure B in some cases will reduce the loads. I think ASCE 7 also needs to be revised allow for this "emptiness" factor, but I don't know if any effort is underway to rewrite that.

For the seismic loading, H is defined as "Maximum design product level", so yes, it can be less than shell height.

On the 1/2 or 2/3 of weights used, I have interpreted that as being different safety factors applied to those items, not necessarily just using a portion of the weight. But I don't know the reasoning behind the difference. I think part of the reasoning here is still trying to bring calculated overturning results in line with actual overturning results. IE, if tanks were overturning right and left all over the country, they'd be jacking those safety factors up. But if thousands of tanks have set there for decades with no problem, then you come along and calculate that they're all in danger of overturn, it's tempting to adjust your calculation methods to match reality rather than the other way around. So they've added increased overturning forces in the code, but also added increased overturning resistance to try to compensate.

Thanks a lot for your detailed explanation — really appreciate your time and clarity.

 

[sdz]

On the 1/2 or 2/3 of weights used, I have interpreted that as being different safety factors applied to those items, not necessarily just using a portion of the weight. But I don't know the reasoning behind the difference. I think part of the reasoning here is still trying to bring calculated overturning results in line with actual overturning results. IE, if tanks were overturning right and left all over the country, they'd be jacking those safety factors up. But if thousands of tanks have set there for decades with no problem, then you come along and calculate that they're all in danger of overturn, it's tempting to adjust your calculation methods to match reality rather than the other way around. So they've added increased overturning forces in the code, but also added increased overturning resistance to try to compensate.

Thanks for your comments.