Issues in ETABS results

[Badar (BAZ)]

I have observed few issues while getting the design results of reinforced concrete members from ETABS (v 9.7.4). I wanted to share this issues with you and get your input and observation on it.

• ETABS always considers the minimum eccentricity for selecting the design moment of columns irrespective of the probable behavior of the column, whether short or long column. See section 10.10.6.5 and its commentary of ACI 318-08 which deals with minimum eccentricity of long columns. You should always check the design moments that ETABS uses for columns if you want to bring down the cost of construction.
• If your model has lateral loads, ETABS will give you design moments in column irrespective of its status as braced or un-braced as per ACI 318 criteria. You should investigate if the storey under consideration is braced, or un-braced (10.10.5.2), and decide appropriate design moments of columns.
• ETABS has a tendency to select a time period of the building that is considerably less than the value obtained by the approximate method, Method A, of the section 1630.2.2  of UBC 97. To quote the FEMA 451 document: ''Because this formula is based on lower bound regression analysis of measured building response in California, it will generally result in periods that are lower (hence, more conservative for use in predicting base shear) than those computed from a more rigorous mathematical model". So, there is no need to use the value of time period that is lot less than Ta. One should always check the time period used by the software; ETABS can overestimate the seismic force by more than 2 times.

  Method A gives lower T and higher V, so FEMA 451 has advised not to use the value of time period less than this value even if rigorous analysis gives a lower value.

  I have seen the results where Etabs have use the value of time period less than Ta; in-fact as low as 0.5Ta, which can increase the base shear two times.

[Ayesha]

Great observations. 

Thank You for sharing.

[Hussain Abid]

Remarkable observations!!!!!

 

[WR1]

1

On 5/1/2016 at 2:26 PM, baz said:

• ETABS always considers the minimum eccentricity for selecting the design moment of columns irrespective of the probable behavior of the column, whether short or long column. See section 10.10.6.5 and its commentary of ACI 318-08 which deals with minimum eccentricity of long columns. You should always check the design moments that ETABS uses for columns if you want to bring down the cost of construction.

Astounding observation. Just to add my 2 cents that in exterior columns, minimum e does not govern. For interior columns it is always good idea to design columns for minimum eccentricity. Yes this is conservative and depends on the choice of designer.

 

2

On 5/1/2016 at 2:26 PM, baz said:

• ETABS always performs analysis of frame as if it is un-braced. You should investigate if the storey under consideration is braced, or un-braced (10.10.5.2), and decide appropriate design moments of column, if ETABS has included sway moments.

Even if the structure is braced, magnification factors will be so small that they will not matter much.

 

3

On 5/1/2016 at 2:26 PM, baz said:

• ETABS has a tendency to select a time period of the building that is considerably less than the value obtained by the approximate method, Method A, of the section 1630.2.2  of UBC 97. To quote the FEMA 451 document: ''Because this formula is based on lower bound regression analysis of measured building response in California, it will generally result in periods that are lower (hence, more conservative for use in predicting base shear) than those computed from a more rigorous mathematical model". So, there is no need to use the value of time period that is lot less than Ta. One should always check the time period used by the software; ETABS can overestimate the seismic force by more than 2 times.
   

See discussion below

[ILYAS]

ETABS 2015 have  option in design that whether you consider minimum eccentricity or not

 

[WR1]

ILYAS, even ETABS 9 had the similar options but that did not work.

[ILYAS]

but in etabs 2015 it works.

 

[Badar (BAZ)]

On ‎7‎/‎27‎/‎2016 at 8:42 AM, Rana said:

Astounding observation. Just to add my 2 cents that in exterior columns, minimum e does not govern. For interior columns it is always good idea to design columns for minimum eccentricity. Yes this is conservative and depends on the choice of designer.

According to the section 10.10.6.5 and its commentary (of ACI 318-08), we do not need to use minimum eccentricity if there are negligible second order effects.   

[WR1]

On 5/1/2016 at 2:26 PM, BAZ said:

I have observed few issues while getting the design results of reinforced concrete members from ETABS (v 9.7.4). I wanted to share this issues with you and get your input and observation on it.

• ETABS has a tendency to select a time period of the building that is considerably less than the value obtained by the approximate method, Method A, of the section 1630.2.2  of UBC 97. To quote the FEMA 451 document: ''Because this formula is based on lower bound regression analysis of measured building response in California, it will generally result in periods that are lower (hence, more conservative for use in predicting base shear) than those computed from a more rigorous mathematical model". So, there is no need to use the value of time period that is lot less than Ta. One should always check the time period used by the software; ETABS can overestimate the seismic force by more than 2 times.
   

Do you mean "There is no need to use the value of time period that is LESS than Ta" or "NOT LESS than Ta"?

 

As you quoted FEMA 451, it says

Ta is lower-bound limit (more conservative for base-shear as T-computed would be higher)

CuTa is the upper-bound limit (can be ignored for drift calculations)

 

But,

 

There is no guidance on if T-computed is less than Ta, not even in ETABS manuals. All the documents highlight the upper limit but not the lower one.

For example, consider a bearing wall building, (density of walls/floor is large), T-computed is ~ 0.10s, however Ta=0.5s; huge difference.

ETABS only checks the upper limit (lets say 1.0s) and not the lower one and selects 0.10 automatically and calculates base shear for this T.

 

However, if I am getting you right or as per; 2009 NEHRP Recommended Seismic Provisions; Chapter 4, page-41, it says;

if T-computed is less than Ta, then use Ta.

https://www.fema.gov/media-library-data/1393888487069-c071850c3050c34da135376baa478a1b/P-752_Unit4.pdf

Please shed some light on this issue.

 

[Saad Pervez]

There is no lower limit for T but there is an upper limit for Maximum V, so even if the time period calculated by ETABS is significantly lower than Ta, code provisions for maximum base shear would govern and the calculated value of T would be ignored.

Similarly, no vertical distribution of force would take place for a small time period hence your design would still be in reasonable FOS limits.

[Waqas Haider]

4 hours ago, Saad Pervez said:

Similarly, no vertical distribution of force would take place for a small time period hence your design would still be in reasonable FOS limits.

Does vertical distribution depends on time period? It must depends on weight participation and stiffness of floors.

[Saad Pervez]

12 hours ago, Engr Waqas said:

Does vertical distribution depends on time period? It must depends on weight participation and stiffness of floors.

UBC97 1630.5: Concentrated force at top, Ft, becomes zero if time period is 0.7 seconds or less. My statement was not 100% correct. Apologies.

[Badar (BAZ)]

 

On ‎12‎/‎21‎/‎2016 at 11:27 AM, Rana said:

However, if I am getting you right or as per; 2009 NEHRP Recommended Seismic Provisions; Chapter 4, page-41, it says;

if T-computed is less than Ta, then use Ta.

I meant that.

[WR1]

1 hour ago, BAZ said:

 

I meant that.

Wow. But there is no clear reference for this in ubc or asce. I got the sense and could save some structures from over-design but would be very difficult to convince authorities without a clear overt, explicit reference to building codes.

[UmarMakhzumi]

Code enforces minimum stiffness requirement through drift control and stability requirements by ensuring second order effects are not significant. I would personally be comfortable using Ta if "drift and second order effect are within limits". 

Thanks.

[WR1]

I have another point,

 

Quote

The period of formulas in the 1997 Uniform Building Code (UBC1997), the 1996 Structural Engineering Association of California (SEAOC) recommendation...are derived from those developed in 1975 as part of the ATC3-06 project [Applied Technological Council (ATC) 1978] largely based on periods of buildings measured from their motions recorded during the 1971 San Fernando earthquake.

Goel and Chopra 1997 [1] developed improved empirical formulas to estimate the fundamental vibration period of RC moment resisting frames (MRF) buildings for use in equivalent lateral force analysis specified in building codes using motions of many buildings recorded during earthquakes.

The data used in [1] have been combined from the motions of buildings recorded during the 1971 San Fernando, 1984 Morgan Hill, 1986 Mt. Lewis and Palm Spring, 1987 Whitter, 1989 Loma Prieta, 1990 Upland, 1991 Sierra Madre and 1994 Northridge earthquakes

(Source: http://www.tdmd.org.tr/TR/Genel/pdf/TDMSK010.pdf)

Lower bound and upper bound limits of regression analysis are given as in building codes.

 

More exact solutions (such as in analytical models) give higher T than Ta due to absence of partition and infill walls etc. But my concern is for buildings having large shear walls areas such as tunnel construction, very very dynamically stiff (e.g. 0.1s). In such cases (T-computed < Ta) and as mentioned by following paper, we should not use Ta (as per discussion in posts above) and should use T-computed.

Paper can be found here: http://erolkalkan.com/Pubs/12.pdf

Quote

As is observed from those figures, the obtained finite-element analysis results significantly differ from the code-referred values...The significantcant deviation between current code given formulas and finite-element analysis leads to intolerable errors for the dynamic parameters and corresponding design loads.

Also see this: http://www.sciencedirect.com/science/article/pii/S1687404815000565

[Syed Umair Haider]

I think both ETABS and UBC 97 provisions are correct.

As ETABS completely follows UBC97 therefore only UBC's provisions are elaborated below.

UBC seismic design philosophy limits the minimum time period (in the form of maximum base shear ) to Ts (UBC response spectrum) which is equivalent to Cv/2.5Ca.

For elaboration ,

V=CV.I.W/R.T

Substitutue T=Ts=Cv/2.5Ca in above eqn

therefore 

V=(CV.I.W/R)X(2.5Ca/Cv)

i.e V=2.5Ca.I/w (the maximum base shear that ETABS and UBC97 uses)

Therefore time period regardless of its shorter value below Ta cannot be lesser than Ts which is in contrast with UBC97 provisions that are inherent in ETABS.

However, analysis indicating time period lesser than Ta indicates a over stiff/over design structure, therefore it is recommended to maintain a natural period closer to 0.1x (no of floors).

 

 

 

 

[WR1]

1 hour ago, Syed Umair Haider said:

I think both ETABS and UBC 97 provisions are correct.

As ETABS completely follows UBC97 therefore only UBC's provisions are elaborated below.

UBC seismic design philosophy limits the minimum time period (in the form of maximum base shear ) to Ts (UBC response spectrum) which is equivalent to Cv/2.5Ca.

For elaboration ,

V=CV.I.W/R.T

Substitutue T=Ts=Cv/2.5Ca in above eqn

therefore 

V=(CV.I.W/R)X(2.5Ca/Cv)

i.e V=2.5Ca.I/w (the maximum base shear that ETABS and UBC97 uses)

Therefore time period regardless of its shorter value below Ta cannot be lesser than Ts which is in contrast with UBC97 provisions that are inherent in ETABS.

Dear Umair, I understand what you said, but, that is only one of the condition I asked about (see below). Ok let me put this in following way;

lets call T-computed simply as T, there are following 4 cases;

a. T < Ts < Ta

b. T > Ts but < Ta

c. T > Ta but < CuTa {not my concern}

d. T > CuTa {not my concern}

Now lets focus on condition a & b; Let's say; Ts=0.3s and Ta=0.7s

Condition a

Assume T=0.1s which is less than both Ts and Ta

In order to jump to V, we need to first set the time period.

a1. If we use Ts, then obviously maximum base shear would govern. This is what you are referring to.

a2. if we use Ta (which is higher than Ts), base shear would be less than Max. V so we get some saving.

Now the question is what T you'd use? Ts or Ta?

As per FEMA (moment frame example) use Ta.

As per research papers I have mentioned for tunnel construction you should use actual T (or in other words Ts as Max V would govern, as you rightly pointed out).

 

Condition b

Assume T=0.5s which is more than Ts but less than Ta.

Still, if we go with FEMA approach, we should use Ta which will give us lesser base shear. And use of actual T would give us higher base shear.

 

So the question remains, what value of T to be used?

Quote

However, analysis indicating time period lesser than Ta indicates a over stiff/over design structure, therefore it is recommended to maintain a natural period closer to 0.1x (no of floors).

0.1 x (no of floors) is for certain moment frame buildings as per ASCE 7-10 seismic design examples book (example 17), that might not be correct for buildings with large no/area of shear walls. Anyway, the intention, as i see, is to use approximate method (Ta) if analysis T is less than Ta, which is in-line with what Baz said. However, after going through the above research papers, I am again inclined towards using T from analysis instead of Ta for such buildings. For moment frames, yes we could use Ta.

[WR1]

[Badar (BAZ)]

My initial post was to point to fellow engineers that one must look at the value of the time period used by the ETABS for calculating the base shear.

For normal construction in Pakistan, the construction in which the area of shear walls seldom exceeds 1% of the floor area, the time period of the building should not be significantly less than the value obtained by empirical formulae of UBC 97. If ETABS is using the value which is, say 60%, of that value, then there should be some mistake in modeling, and the design engineer should be aware of that.

[WR1]

10 hours ago, BAZ said:

My initial post was to point to fellow engineers that one must look at the value of the time period used by the ETABS for calculating the base shear.

For normal construction in Pakistan, the construction in which the area of shear walls seldom exceeds 1% of the floor area, the time period of the building should not be significantly less than the value obtained by empirical formulae of UBC 97. If ETABS is using the value which is, say 60%, of that value, then there should be some mistake in modeling, and the design engineer should be aware of that.

100% agreed. Usually shearwall/floor area is ~ 1%

[Syed Umair Haider]

 

Dear Waseem

Your concern is valid but the lower limit given in FEMA example doesn't make sense to me as in a general the usage of time period in seismic analysis is not more than the evaluation of ground acceleration that will be imparted in structure in accordance with specific structural characteristics.

Similarly if the specific structural characteristics of a building reveals a time period lesser than generalized approximate time period , then it doesn't make sense to skip the accurate time period and to use approximate one when code itself defines approximate time period as a basis to start analysis for actual T.

More interesting is to note that indicated FEMA example is based on 2009 NEHRP seismic provisions , whereas in a separate document issued by NEHRP afterwards (named expanded seismic commentary to ACE-10) it is recommended to use "Tcomp" if "Tcomp < Ta".

Moreover, i didn't find any lower bound on time period in UBC97 and even in ASCE 7 which is based on same ELF procedure as given in FEMA example.

Therefore, it seems that indicated provisions in FEMA example are overlooked that they have fixed later and usage of Tcomputed if "Tcomputed < Tapproximate" seems valid.

 

 

 

[UmarMakhzumi]

ETABS can provide fairly accurate time periods if modelling assumption, soil structure interaction is accounted well. See attachment. T computed should be used if a person knows what he is doing. That is a big "IF.

Thanks.

 

ETABS vs Field Time Periods.pdf

[WR1]

13 hours ago, Syed Umair Haider said:

 

Dear Waseem

Your concern is valid but the lower limit given in FEMA example doesn't make sense to me as in a general the usage of time period in seismic analysis is not more than the evaluation of ground acceleration that will be imparted in structure in accordance with specific structural characteristics.

Similarly if the specific structural characteristics of a building reveals a time period lesser than generalized approximate time period , then it doesn't make sense to skip the accurate time period and to use approximate one when code itself defines approximate time period as a basis to start analysis for actual T.

More interesting is to note that indicated FEMA example is based on 2009 NEHRP seismic provisions , whereas in a separate document issued by NEHRP afterwards (named expanded seismic commentary to ACE-10) it is recommended to use "Tcomp" if "Tcomp < Ta".

Moreover, i didn't find any lower bound on time period in UBC97 and even in ASCE 7 which is based on same ELF procedure as given in FEMA example.

Therefore, it seems that indicated provisions in FEMA example are overlooked that they have fixed later and usage of Tcomputed if "Tcomputed < Tapproximate" seems valid.

 

 

 

Ok, what about these;

 

#1

2015 NEHRP Recommended Seismic Provisions: Design Examples FEMA P-1051/July 2016

https://www.fema.gov/media-library-data/1474320077368-125c7a1d1a3b864648554198526d671f/FEMA_P-1051.pdf

"Calculation of a period based on an analytical model of the structure is encouraged, but limits are placed on the results of such calculations. These limits prevent the
use of a very flexible model in order to obtain a large period and correspondingly low acceleration."

Page: 2-17

and

"A lower limit to the base shear determined from the modal analysis procedure is specified based on the static procedure, and the approximate periods specified in the static procedure."

Page: 2-18

Also consider the example in section 13.2.6 and I have attached the sketch explaining that example for easy understanding here; (Ta was used instead of Tc).

 

#2

See ASCE publication seismic design guidelines based on ASCE 7-10, Example 17 on page; 117 & 118;

and

 

[Hussain Abid]

interesting discussion!!! but how can we check T i.e the fundamental time period in ETABS?

 

 

Edited January 12, 2017 by Abid Qasmi