The 9/11 Forum

Dr. G wrote:Dave Rogers:

Am I "accusing NIST of deliberately and fraudulently misrepresenting their own results?" NOT AT ALL! I am simply asking NIST to explain what I perceive to be a discrepancy in their Draft Report, because I have no idea how such contradictory results could be obtained.

metamars wrote:

Dictator Cheney wrote:
http://wtc.nist.gov/media/NIST_NCSTAR_1 ... omment.pdf

Page 538 - 543 it isnt that much to read.

and if you want i can get you more details about the LS-Dyna standard materials they used, end of next week, when i have acces to LS-Dyna.

On p. 539 (201 in the pdf reader), we read that

For the regions of WTC 7 subjected to heating by fires (between Floors 7 and 14), termperature-dependent material models were used for the framing. The material model used for the steel in the fire-affected floors was the Elastic-Viscoplastic Thermal (Type 106) model in LS-DYNA, which included thermal expansion and thermal degradation in material stiffness and strength. This model used the same parameters to define the nonlinear material behavior of steel at room temperature as the Type 24 model, but included additional parameters to define temperature dependence. The yield strength, elastic modulus, Poisson's ratio, and thermal expansion coefficient were all identified as a function of temperature. The temperature-dependent models used the same failure criterion as that applied at room temperature. The temperature-dependent material properties and constitutive model parameters for the steels and bolts used in WTC 7 are presented in Appendix E. Example of the stress-strain curves for the 50 ksi steel at various temperatures is shown in Figure 12-2.

However, when we look at E.3.1, p. 708, we read

The WTC investigation developed a methodology for estimating the creep properties of untested steels based on creep models of existing steels.....Although tensile strength scaling produced the best agreement, in many cases the agreement was not very good. Frequently the predicted strain differed from the actual by more than a factor of ten.

Well, that doesn't sound too good. However, the very next sentence says

Differences of this magnitude correspond to temperature offsets of around 35 deg C or stress offsets of 17 MPa.

Frankly, while I have no intuition on the 17 MPa offset figure, 35 deg C doesn't sound bad, at all. So, color me confused.

However, the last sentence is also mysterious to me, and also raises doubts about how accurate their modeling was:

Because no steel relevant to the creep modeling was recovered from WTC7, it is impossible to create more accurate models.

The problems I have with this are as follows:

1) If they have accurate knowledge about exact steels used, and good models for those same steels, then if their fire models were accurate and adequate (and if they had enough computer horsepower), steel specimens would not be essential, other than as sanity checks.
2) If they are missing some essentials I list in 1), and physical specimens could have been used to provide data needed to tweak the models to make them accurate, the fact that these were not recovered means that they can't even do that! So, while NIST is implying that their models are good enough, I don't see how they can have any confidence that this is so.

Getting back to my quote, even miniscule amounts of Boron added to a steel can have a big effect on it's behavior at high temperatures. So, I don't have have a warm fuzzy feeling about using models for "similar" steels, as, e.g., is mentioned on p. 716, where they used behavior based on "Austrailian AS149 steel, which is similar to ASTM A 36"

As if that's not enough, on p. 718, we read

The models developed were technically for the steels from which the bolts were made, rather than for the bolts. Bolt failure is complex at both room- and elevated-temperature, and no methodology exists for modeling the failure of bolts, as distinct from the steels form which they are made, at elevated temperature.

(emphasis mine)

While it may not even be possible to make an accurate model of such a complex, large system as WTC 7 + fires, should NIST be claiming explanations based on what amounts, apparently, to a massive guess, just because they made an effort? Did they give themselves an 'A', just for effort?

Dictator Cheney wrote:

metamars wrote:

Dictator Cheney wrote:
http://wtc.nist.gov/media/NIST_NCSTAR_1 ... omment.pdf

Page 538 - 543 it isnt that much to read.

and if you want i can get you more details about the LS-Dyna standard materials they used, end of next week, when i have acces to LS-Dyna.

On p. 539 (201 in the pdf reader), we read that

For the regions of WTC 7 subjected to heating by fires (between Floors 7 and 14), termperature-dependent material models were used for the framing. The material model used for the steel in the fire-affected floors was the Elastic-Viscoplastic Thermal (Type 106) model in LS-DYNA, which included thermal expansion and thermal degradation in material stiffness and strength. This model used the same parameters to define the nonlinear material behavior of steel at room temperature as the Type 24 model, but included additional parameters to define temperature dependence. The yield strength, elastic modulus, Poisson's ratio, and thermal expansion coefficient were all identified as a function of temperature. The temperature-dependent models used the same failure criterion as that applied at room temperature. The temperature-dependent material properties and constitutive model parameters for the steels and bolts used in WTC 7 are presented in Appendix E. Example of the stress-strain curves for the 50 ksi steel at various temperatures is shown in Figure 12-2.

However, when we look at E.3.1, p. 708, we read

The WTC investigation developed a methodology for estimating the creep properties of untested steels based on creep models of existing steels.....Although tensile strength scaling produced the best agreement, in many cases the agreement was not very good. Frequently the predicted strain differed from the actual by more than a factor of ten.

Well, that doesn't sound too good. However, the very next sentence says

Differences of this magnitude correspond to temperature offsets of around 35 deg C or stress offsets of 17 MPa.

Frankly, while I have no intuition on the 17 MPa offset figure, 35 deg C doesn't sound bad, at all. So, color me confused.

However, the last sentence is also mysterious to me, and also raises doubts about how accurate their modeling was:

Because no steel relevant to the creep modeling was recovered from WTC7, it is impossible to create more accurate models.

The problems I have with this are as follows:

1) If they have accurate knowledge about exact steels used, and good models for those same steels, then if their fire models were accurate and adequate (and if they had enough computer horsepower), steel specimens would not be essential, other than as sanity checks.
2) If they are missing some essentials I list in 1), and physical specimens could have been used to provide data needed to tweak the models to make them accurate, the fact that these were not recovered means that they can't even do that! So, while NIST is implying that their models are good enough, I don't see how they can have any confidence that this is so.

Getting back to my quote, even miniscule amounts of Boron added to a steel can have a big effect on it's behavior at high temperatures. So, I don't have have a warm fuzzy feeling about using models for "similar" steels, as, e.g., is mentioned on p. 716, where they used behavior based on "Austrailian AS149 steel, which is similar to ASTM A 36"

As if that's not enough, on p. 718, we read

The models developed were technically for the steels from which the bolts were made, rather than for the bolts. Bolt failure is complex at both room- and elevated-temperature, and no methodology exists for modeling the failure of bolts, as distinct from the steels form which they are made, at elevated temperature.

(emphasis mine)

While it may not even be possible to make an accurate model of such a complex, large system as WTC 7 + fires, should NIST be claiming explanations based on what amounts, apparently, to a massive guess, just because they made an effort? Did they give themselves an 'A', just for effort?

the whole problem is indeed they have no steel sample to do real life tests. to compare it to the FE sims.

but they know what kind of steel was used, like you pointed out, A36. there are pretty detailed specifications of that steel.
im not so sure about the ASTM norms as i dont work with them. But i guess they also normed the Chemical ingridents, and normally such standards are a good starting point.
but then the diffrent connection types under diffrent high temperatures can not be tested from samples of the building.
sure it is somehow guessing and testing.

the collapse model and Fire model was not one simulation.
they first simulated the fires and then the data and conclusions from that sim was brought to the collapse sim. no other way atm.

with more effort and more computer power they could have worked out a maybe more accurate Sim, but how unaccurate is it now? does it need more? i dont know.(i am amazed about the lack of the Computing power, almost can compete with theyr computing power) But i think it is a good start.

Dictator Cheney wrote:for a simulation of a total collapse there is no reason to use deflection scales other than 1:1

metamars wrote:Frankly, while I have no intuition on the 17 MPa offset figure, 35 deg C doesn't sound bad, at all. So, color me confused.

Chainsaw wrote:

Dictator Cheney wrote:

metamars wrote:

Dictator Cheney wrote:
http://wtc.nist.gov/media/NIST_NCSTAR_1 ... omment.pdf

Page 538 - 543 it isnt that much to read.

and if you want i can get you more details about the LS-Dyna standard materials they used, end of next week, when i have acces to LS-Dyna.

On p. 539 (201 in the pdf reader), we read that

For the regions of WTC 7 subjected to heating by fires (between Floors 7 and 14), termperature-dependent material models were used for the framing. The material model used for the steel in the fire-affected floors was the Elastic-Viscoplastic Thermal (Type 106) model in LS-DYNA, which included thermal expansion and thermal degradation in material stiffness and strength. This model used the same parameters to define the nonlinear material behavior of steel at room temperature as the Type 24 model, but included additional parameters to define temperature dependence. The yield strength, elastic modulus, Poisson's ratio, and thermal expansion coefficient were all identified as a function of temperature. The temperature-dependent models used the same failure criterion as that applied at room temperature. The temperature-dependent material properties and constitutive model parameters for the steels and bolts used in WTC 7 are presented in Appendix E. Example of the stress-strain curves for the 50 ksi steel at various temperatures is shown in Figure 12-2.

However, when we look at E.3.1, p. 708, we read

The WTC investigation developed a methodology for estimating the creep properties of untested steels based on creep models of existing steels.....Although tensile strength scaling produced the best agreement, in many cases the agreement was not very good. Frequently the predicted strain differed from the actual by more than a factor of ten.

Well, that doesn't sound too good. However, the very next sentence says

Differences of this magnitude correspond to temperature offsets of around 35 deg C or stress offsets of 17 MPa.

Frankly, while I have no intuition on the 17 MPa offset figure, 35 deg C doesn't sound bad, at all. So, color me confused.

However, the last sentence is also mysterious to me, and also raises doubts about how accurate their modeling was:

Because no steel relevant to the creep modeling was recovered from WTC7, it is impossible to create more accurate models.

The problems I have with this are as follows:

1) If they have accurate knowledge about exact steels used, and good models for those same steels, then if their fire models were accurate and adequate (and if they had enough computer horsepower), steel specimens would not be essential, other than as sanity checks.
2) If they are missing some essentials I list in 1), and physical specimens could have been used to provide data needed to tweak the models to make them accurate, the fact that these were not recovered means that they can't even do that! So, while NIST is implying that their models are good enough, I don't see how they can have any confidence that this is so.

Getting back to my quote, even miniscule amounts of Boron added to a steel can have a big effect on it's behavior at high temperatures. So, I don't have have a warm fuzzy feeling about using models for "similar" steels, as, e.g., is mentioned on p. 716, where they used behavior based on "Austrailian AS149 steel, which is similar to ASTM A 36"

As if that's not enough, on p. 718, we read

The models developed were technically for the steels from which the bolts were made, rather than for the bolts. Bolt failure is complex at both room- and elevated-temperature, and no methodology exists for modeling the failure of bolts, as distinct from the steels form which they are made, at elevated temperature.

(emphasis mine)

While it may not even be possible to make an accurate model of such a complex, large system as WTC 7 + fires, should NIST be claiming explanations based on what amounts, apparently, to a massive guess, just because they made an effort? Did they give themselves an 'A', just for effort?

the whole problem is indeed they have no steel sample to do real life tests. to compare it to the FE sims.

but they know what kind of steel was used, like you pointed out, A36. there are pretty detailed specifications of that steel.
im not so sure about the ASTM norms as i dont work with them. But i guess they also normed the Chemical ingridents, and normally such standards are a good starting point.
but then the diffrent connection types under diffrent high temperatures can not be tested from samples of the building.
sure it is somehow guessing and testing.

the collapse model and Fire model was not one simulation.
they first simulated the fires and then the data and conclusions from that sim was brought to the collapse sim. no other way atm.

with more effort and more computer power they could have worked out a maybe more accurate Sim, but how unaccurate is it now? does it need more? i dont know.(i am amazed about the lack of the Computing power, almost can compete with theyr computing power) But i think it is a good start.

This might help, in understanding the steel, http://www.key-to-steel.com/default.aspx?ID=CheckArticle&NM=177.

einsteen wrote:

Dictator Cheney wrote:for a simulation of a total collapse there is no reason to use deflection scales other than 1:1

I would also say that. It is useful for things that are hard or not to see directly, but the whole world knows the building kept its shape while falling and then 1:1 would be the perfect choice for comparisons.

Dr. G wrote:Thanks Einsteen!

Thats great, ...... you da man!

The most striking feature of these smearograms is that the collapse initiation is so fast.

One question though:

What would you guess the vertical scale to be in your figure 3?

12.3.1. Controlling Displaced Shape Displays
You can control displaced shape displays in two ways:

By superimposing undisplaced and displaced shapes. A display of a structure's displaced shape will often be more meaningful if you can compare the displaced configuration against the original configuration. You can do this by using the KUND argument on the PLDISP command.

By multiplying displacements for distortion displays. In most small-deformation structural analyses, the displaced shape is hard to distinguish from the undisplaced shape. The program automatically multiplies the displacements in your results display, so that their effect will be more readily apparent. You can adjust this multiplication factor, using the /DSCALE command (Utility Menu> PlotCtrls> Style> Displacement Scaling). The program interprets exactly zero values of this multiplier (DMULT = 0) as the default setting, which causes the displacements to be scaled automatically to a readily discernible value. Thus, to obtain "zero" displacements (that is, an undistorted display) you must set DMULT = OFF.

/DSCALE, WN, DMULT

Sets the displacement multiplier for displacement displays.

GRAPHICS: Scaling

MP ME ST PR PRN <> <> FL EM <> <> PP <>

WN
Window number (or ALL) to which command applies (defaults to 1).

DMULT

AUTO or 0 — Scale displacements automatically so that maximum displacement (vector amplitude) displays as 5 percent of the maximum model length, as measured in the global Cartesian X, Y, or Z directions. This is the default setting when NLGEOM is OFF.

1 — Do not scale displacements (i.e., scale displacements by 1.0, true to geometry). Often used with large deflection results. This is the default setting when NLGEOM is ON.

FACTOR — Scale displacements by numerical value input for FACTOR.

OFF — Remove displacement scaling (i.e., scale displacements by 0.0, no distortion).

USER — Set DMULT to that used for last display (useful when last DMULT value was automatically calculated).



Command Default
The default value is 1.0 when NLGEOM is ON, and AUTO when NLGEOM is OFF.

NLGEOM, Key

Includes large-deflection effects in a static or full transient analysis.

SOLUTION: Nonlinear Options

MP ME ST PR PRN <> <> <> <> <> <> PP <>

Product Restrictions

Key
Large-deflection key:

OFF — Ignores large-deflection effects (that is, a small-deflection analysis is specified). This option is the default.

ON — Includes large-deflection (large rotation) effects or large strain effects, according to the element type.

Dr. G wrote:Thanks Einsteen!

Thats great, ...... you da man!

The most striking feature of these smearograms is that the collapse initiation is so fast.

One question though:

What would you guess the vertical scale to be in your figure 3?

einsteen wrote:DC, this is a video measurement and no simulation...