You are just repeating questions that has already been answered.
Hi Carlos!
Actually, many of the questions that I raise go unanswered, so I sympathize completely. But this question has been underanswered.
The sticky point is how do these buckling beams, that have lost their vertical support (fig 11-35) push off? The buckling can't happen if they are allowed to push the 79 to 44 girder, and NIST describes exactly that. This is describing a pin-roller. A pin-roller won't allow the force development necessary to buckle. If so this is new to me, and I'd like an explaination from you. Just cite it in your own words,
how can beams that are restrained by the stiff exterior with moment connections at the columns (although they say at one point that the exterior columns buckling while this buckling-push is occuring - amazingly enough) can push off the 79 connection and then buckle? This defies the accumulated practice of solid mechanics. But with NIST, a lot of things are new.
"The public should really recognize the science is really behind what we (NIST) have said," adding, "The obvious stares you in the face." - Dr. Sunder - 2008
Please explain that clearly. Any other twoofie-eating JREFer is welcome to try this as well. Also, there are a number of other problems with NIST's reports, but this is what Tom offered me in explaining this amazing pushing/buckling floor beam from thermal expansion: 2 NIST quotes that are in and of themselves baseless and unsubstantiated, that is, unless you believe in garbage in – garbage out computer models.
Tom does, I don't. I am too "skeptical".
No sir, this is no answer Carlos, this is called delusions.
Tom did not answer the question. Nor has anybody. And yes, we will get to the columns, sooner or later. But keep in mind, none of this buckle-push thermally expanding beam phenomena (
or is it push-buckle?) was not substantiated by witnesses. Also keep in mind, none of these buckling-pushing differential thermally expanding woo beams were verified, witnessed that we know of or substantiated by testing the evidence.
Stages of General Practice in Failure Analysis - ASM Vol. 10, p.10 - Failure Analysis and Prevention - 8th Edition
1. Collect background data and selection of samples.
2. Preliminary examination of failed part
3. Nondestructive testing
4. Mechanical testing - hardness, toughness and tensile strength
5. Selection, identification, preservation and cleaning of specimens
6. Macroscopic/microscopic examination and analysis
7. Selection, preparation and analysis of metalographic sections
8. Determination of failure mechanism
9. Chemical analysis
10. Fracture mechanics analysis
11. Testing under simulated service
12. Analysis of all evidence, formulation of conclusions, and reporting
And how many of these stages of "General Practice in Failure Analysis" were followed, Carlos? How many?
"Originally Posted by NIST
Criteria to Determine Lateral-Torsional Buckling of Beams and Girders
When lateral support of the top (compression) flange was lost, floor beams and girders could laterally displace and buckle in a lateral-torsional mode. ... The criterion used for removal of buckled beams or girders was based on the destabilizing effect of gravity loads on laterally displaced beams or girders. If a beam or girder twisted half of its flange width laterally, it would not be able to support its gravity loads and would be removed from the analysis."
But this lateral movement was not quantified; they only said it was monitored. Monitored by what? Where is this data? And how are we (JREFers excluded, they will always believe NIST, that's a given) to buy off on the WTC 7 column 79 “walk off” break(s) on a heavily constrained member when a lightly constrained Cardington member did not break? The problem is that this is not forensic based. This is not witness based. This is NIST woo-based, and we can't see the backup for the woo:
"We are, however,
withholding 3,370 files.
The NIST Director determined that the release of these data might jeopardize public safety. This withheld data include the remaining input and all results files of the ANSYS 16-story Case B collapse initiation model, break element source code. ANSYS scripts files for the break elements, custom executable ANSYS file, and all spreadsheets and other supporting calculations used to develop floor connection failure modes and capacities.
Sincerely, Catherine S. Fletcher, Freedom of Information Act Officer.”
We are speculating on NIST's
woo models and amazing reports. So let's examine Tom's other twoof-crushing nugget:
"Buckling in the floor beams was due to the combined effects of (1) loss of lateral restraint, (2) increased axial loads due to thermal expansion effects, and (3) gravity loads from the floor slab. Floor beams lost lateral restraint when the majority of their shear stud connections failed, either by differential thermal expansion between the steel beams and the concrete slab, or by local concrete failure due to fires on the floor slab.
As has been previously raised (but just gets ignored by the twoofie-devouring JREFers), (1) the lateral direction of the NIST break connection for the seat supports in the model was not quantitized. NIST states that this was "monitored". Again, monitored by what? Does anyone know? Does NIST? The model COMBIN37 control element could only account for displacement in one direction, the axial direction.
(2) This is truly amazing, they point to Cardington Test #3 as their "ya see ya see?!". But what happened at Cardington Test #3? That's right, more heat and longer duration yet no buckling of the top flange, nor did the shear angle clip break.
So how exactly does NIST derive at their break threshold? Where the break strength data, remaining input and all results are files of the ANSYS (FEA) 16-story Case B collapse initiation model, break element source code, ANSYS (FEA) scripts files for the break elements, custom executable ANSYS (FEA) file, all spreadsheets and other supporting calculations used to develop floor connection failure modes and capacities, connection models?
Oh yeah, Dr. Sunder doesn't want to jeopardize public safety. Ryan the twoofie-slayer said his reason for withholding was ITAR. Want to revisit that one Rayn?
Here is Dr. Sunder's reasoning for this refusal of release:
"The decision to withhold the data was based on the fact that the capabilities of the WTC 7 collapse initiation and global collapse models are unprecedented, in that they provide validated models that can predict collapse of typical tall buildings. If released, these models would provide a powerful tool to groups and individuals interested in simulating building collapses and devising ways to destroy buildings."
NIST likes to use terms like "good agreement" and "salient features" and other non-quantitized hollow language. It is amusing to me that the Cardington test is a NIST premise that their woo model has validity.
Just to quickly review NIST's high standards, please remember, this is the same NIST that stated: "The interior walls [including insulated steel columns] were assumed to have the properties of gypsum board [0.5 W/m/K]." NCSTAR 1-5F, p 52
and
"The steel was assumed in the FDS model to be thermally-thin, thus, no thermal conductivity was used." NCSTAR 1-5F, p 20
and
“Although the floor slab actually consisted of a metal deck topped with a concrete slab...the thermal properties of the entire floor slab were assumed to be that of concrete [1.0 W/m/K]." NCSTAR 1-5F, p 52
and loaded these values into their 16-story ANSYS model:
Item Actual Density lb/ft3 NIST woo Density lb/ft3
Steel 500 0.5
Concrete 150 0.15
Why did NIST assign densities of the beam and shell material 1/1000th that of generally accepted engineering values in the 16 story ANSYS model? Guesses?
Or their “omniscience” with language such as:
“All the floor connections to these columns, as well as to the exterior columns failed, and the floors fell on the east side of the building.” NCSTAR 1A p. 22
Who saw “all these connections” fail? Was testing done to determine this? Is “all” a bit of a stretch Tom or do you believe everything NIST tells you?
Never mind that for now. Let's discuss 3.
(3) gravity loads from the floor slab.
This is a 5.5" slab (most floors), mesh reinforced, dowelled into the perimeter bent plate. Is this a dominant factor in failure modes? Is not already redundant in the structural design? Please explain.
"Floor beams lost lateral restraint when the majority of their shear stud connections failed, either by differential thermal expansion between the steel beams and the concrete slab, or by local concrete failure due to fires on the floor slab."
NIST states that ''even though steel and concrete have similar coefficients of thermal expansion, differential thermal expansion occurred between the steel floor beams and concrete slab when the composite floor was subjected to fire."
Explain this woo differential thermal expansion "''even though steel and concrete have similar coefficients of thermal expansion (we went on and ignored science anyway)" Tom. Don't forget your top hat and cane.
or even better:
"No thermal expansion or material degradation was considered for the slab, as the slab was not heated in this analysis." NCSTAR 1-9, p. 352.
The NIST partial-floor model did not allow the slab to expand thermally with
the steel beams, and neglecting thermal expansion of the slab has the effect of imposing additional relative displacement on the shear studs connecting the concrete to the steel. This subsystem analysis formed the basis for special connection elements used in the global analyses as described in the following passages.
"The failure modes in this model [the partial floor] were incorporated into the 16 story ANSYS and 47 story LS-DYNA analyses." NCSTAR 1-9, p. 353.
This is the same 16 story ANSYS model that had concrete and steel densities lowered by two orders of magnitude from reality. Credit to Gery Warner PE for this notices and making me and others aware of this woo find.
"These results helped to guide the development of special connection elements…that captured the salient features and failure modes of the various types of connections used in the floor system of WTC 7." NCSTAR 1-9, p. 359.
Is salient a quantifiable term appropriate for this context? This relative displacement occurred in the ANSYS model, and no physical testing was done to verify its magnitude in the steel-and concrete structure. Did NIST take steps to maximize the destructive effects of any relative displacement due to thermal movement? Is this why they created "differential thermal expansion" woo right after admitting that "though steel and concrete have similar coefficients of thermal expansion"? Is this why the ANSYS model has 6 degrees of freedom on its connections? Is this why NIST manipulated thermal and density values, to make a more breakable model? Explain this to me please Tom.
NCSTAR 1-9 Chapter 11 discusses structural analysis of the initial failure event based on the 16-story ANSYS model. Although this model was capable of including thermal conductivity, NIST does not mention this important material property. Why?
"The [ANSYS] model accounted for nonlinear geometric effects, temperature dependent behavior of members and connections (including thermal expansion and stiffness and strength degradation), the sequential failure of structural framing and connections under fire conditions, and removal of failed elements (with user intervention)." NCSTAR 1-9, p. 457.
Heat transfer within structural elements and between structural elements was considerable in the steel framing, and it dissipated heat energy from the hottest parts of the steel. Did the analysts consider heat transfer, or was this property simply ignored to enhance computational performance? ANSYS results were input to the LS-DYNA model to cover up predict woo failure modes.
"The purpose of the ANSYS model was to simulate the accumulation of local damages and failures up to the initiation of overall global collapse due to fire." NCSTAR 1-9, p. 484.
"The fire-induced damage from the ANSYS model were [sic] input into the LS-DYNA model as initial conditions." NCSTAR 1-9, p. 457.
"…it was not necessary to input more than one solution to the global analysis of the collapse. The fire induced damage produced by Case B temperatures at 4.0 h was carried forward as the initial condition for the LS-DYNA analysis." NCSTAR 1-9, p. 535.
"Column splices were also not modeled for interior columns, as the purpose of the ANSYS model was to accumulate local failures up to the point of buckling in a column. When column buckling appeared to be imminent, the analyses were continued in the LS-DYNA 47 story model." NCSTAR 1-9, p. 476.
The preceding statements imply that the 47-story LS-DYNA model was initially
damaged due to preexisting fire effects, and NIST controlled the initial conditions by using the 16-story ANSYS model to predict an initial failure state for the 47-story model. The LS-DYNA model was loaded with gravity dead loads plus 25 percent of the original design live loads in addition to the high-temperature thermal loading Case B. The initial damage state for the LS-DYNA model included debris impact damage from WTC 1 plus the accumulated fire-induced damage predicted by the ANSYS analysis. Was the LS-DYNA model capable of predicting the initial failure resulting from the Case B temperature distribution without preexisting damage imposed?
NIST supplied the initiating event data to ARA even though the contract states that ARA would perform analyses to determine the location and cause of collapse initiation. ARA only looked at failure modes of floors eight through 46 even though previous engineering studies by FEMA engineers stated clearly that ''the most likely [structural failure] event would have been the collapse of Truss 1 and/or Truss 2 located in the east end of the 5th and 6th floors."
http://wtc.nist.gov/solicitations/wtc_awardQ0186.htm
According to the contractual language ARA did not look for possible failure modes on floors one through seven, and the analysis documented by ARA was required to support the initiating-event hypothesis as determined by NIST.
The Introduction to NCSTAR 1-9A clearly states the purpose of the LS-DYN analysis.
"The purpose of this work was to analyze the global response of WTC 7 to an initial failure event due to fire and to analyze the resulting component and subsystem failures to determine the events that led to the global collapse." NCSTAR 1-9A, p. 1.
The initial failure event was predetermined by NIST. ARA was not responsible for analysis of the structural response to the fires and varying temperature distribution from the start, although LS-DYNA is capable of analyzing thermal softening and thermal expansion of structural materials. NCSTAR 1-9A also states the LS-DYNA model of WTC 7 ''was focused on capturing the entire collapse initiation and collapse propagation process of the building…'' NCSTAR 1-9A, p. 1.
This is false; the LS-DYNA model of WTC 7 was initialized with data representing fire-induced damage that NIST estimated had occurred leading to collapse initiation. A two-floor subassembly model was constructed by ARA to ''assess the model behavior for failure events during the model development and to assess the global model performance…''NCSTAR 1-9A, p. 64. Two temperature profiles were considered during the two-floor model analyses. These are described as Case A and Case B at five hours85, but NCSTAR 1A and NCSTAR 1-9 discuss only temperature profiles with 3.5-hour and four-hour duration. The final reports are inconsistent with respect to this important detail.
ARA analyzed their two-floor model with several specific load cases in conjunction with the Case A and Case B temperatures at five-hour duration. Load Case 1 had no imposed (preexisting) connection or support failures. NCSTAR 1-9A, p. 70. The Case A temperature distribution did not lead to instability of the floor structure. The Case B temperature distribution predicted a partial collapse of the framing, but this did not occur at the east end of the building as predicted by the ANSYS analysis. Only Load Cases 2 and 3 exhibited a partial collapse at the east end of WTC 7, and these load cases imposed preexisting failures of connections at columns 79 and 81. Not one of the three load cases predicted a collapse of floor framing at the northeast corner as predicted by the ANSYS model—the event described by NIST as causing collapse initiation.
"The loads applied to the LS-DYNA global model included gravity, debris impact damage, Case B temperatures (applied smoothly in two seconds), and fire-induced damage from the ANSYS analysis." NCSTAR 1-9, p. 563.
"In the model, the debris damage was instantaneously applied to approximate the actual dynamic event." NCSTAR 1-9A, p. 83.
"The final step in the initialization process was to apply fire-induced damage from the 16 story ANSYS analysis." NCSTAR 1-9A, p. 118.
"…the fire-induced damage obtained from the 16-story ANSYS analysis, including damage to floor beams, girders, and connections, was applied instantaneously." NCSTAR 1-9A, p. 51.
"Any imposed structural damage was applied instantaneously immediately following temperature initialization." NCSTAR 1-9A, p. 65.
The elevated temperatures and fire-induced damage to structural elements occurred over a period of several hours, and sudden removal of damaged structural elements does not account for a gradual redistribution of static loads. Thermal conductivity and heat flux affect the temperature distribution
as a function of time. What effect does the rate of application of heat and fire-induced damage have on the global analysis? This is one more question the report does not address.
Damage to framing and connections was taking place in the LS-DYNA analysis prior to the application of the ANSYS estimated damage.
"During the temperature application cycle in the LS-DYNA analysis, combined thermal expansion and thermally degraded material properties resulted in beam and girder connection damage throughout the heated floor structures. The connection damage and buckled beam data transferred from the 16 story ANSYS analysis were then applied." NCSTAR 1-9A, p. 79.
If the application of elevated temperatures were sufficient to cause framing and connection damage throughout the floor structures, and the LS-DYNA analysis considered thermal expansion and thermally-degraded material properties, then why was it necessary to impose additional fire-induced damage determined by the NIST ANSYS analysis?
Models of framing connections used in the LS-DYNA analysis were compared to the ANSYS connection models.
"A comparison was performed between the LS-DYNA and ANSYS FHK [fin, header, and knife] shear connection models. The comparison showed good agreement for selected connections, which increased confidence in both of the separately developed modeling approaches. NCSTAR 1-9, p. 555.
What is considered ''good agreement'', and what about connections other than the ''selected connections''? NIST does not show any documentation of this comparison. NCSTAR 1-9A Figure E-2 shows the elements of a seated connection model. NCSTAR 1-9A, p. xxxvii. This connection model appears to have the necessary components for prediction of connection performance and any failure due to thermal stresses. So why does the LSDYNA global analysis depend on the 16-story ANSYS analysis performed by NIST to predict the fire-induced damage to framing members and connections? NIST attempts to explain this procedure.
"The ANSYS analysis estimated the damage that occurred as the fires grew and spread on Floors 7, 8, and 9 and Floors 11, 12, and 13. The LSDYNA analysis, by comparison, considered only a temperature profile at the time when thermally-induced damage was transferred from the ANSYS analysis." NCSTAR 1-9A, p. xxxix.
This does not explain why the LS-DYNA analysis was not started cold and allowed to develop the thermally-induced damage from data provided by the NIST fire simulation. Not only does the LS-DYNA temperature profile go from zero to nearly 500 degrees Centigrade in two seconds, but the thermal damage estimated by NIST occurred gradually over several hours, and it was applied to the structural model instantaneously. This is not credible for a structural model used to predict the response and interaction of structural materials with time and temperature-dependent properties. But don't let facts trip you up.
NIST compared visual observation times and analytical prediction times of various events leading up to and including the global collapse. The first entry in Table 3-1 of NCSTAR 1A indicates an observation time of minus six seconds for the cascading floor failures that preceded the buckling failure of column 79. This ''event'' was not observed by NIST or anyone else, so the table is erroneous to imply that it was observed before column buckling or the start of global collapse. The buckling of columns 79 through 81 and the horizontal progression of core column buckling were also not observed events as clearly shown in the table.
A significant discrepancy is obvious in the last two observations listed in Table 4-2 of NCSTAR 1-9A. These include the vertical motions of the roof-mounted screen wall (between the east and west penthouses) and the west penthouse. Visual observations clearly show the screen wall falling prior to the west penthouse. The global LS-DYNA model (including debris impact damage) indicates the west penthouse falling out of sequence prior to the screen wall, and NIST falsely claims ''the simulation closely matched the observed behavior." NCSTAR 1-9A, p. 120.
This is related to the column failures in the western core that occurred out of sequence in the global model. How do ARA and NIST explain this woo discrepancy? Additionally, NIST makes contradiction with regards to the alleged core damage:
"In the analysis with debris impact damage, the core framing damage on the west side resulted in a more rapid failure of the west interior columns in the last stages of the horizontal progression." NCSTAR 1-9, p. 599.
What core framing damage on the west side? There was no core framing damage on the west side according to NCSTAR 1-9, page 182.
The horizontal progression of failure was sensitive to the extent of the estimated initial structural damage in WTC 7 due to the collapse of WTC 1. NCSTAR 1-9, p. 612.
"The initial westward progression and the overall speed of the collapse was [sic] not sensitive to the extent of the estimated structural damage to WTC7 due to the debris from the collapse of WTC 1." NCSTAR 1-9, p. 625.
Which is it? What changed NIST's opinion of this "sensitivity" between NCSTAR 1-9, p. 612 and p. 625?
Figures 4-13 and 4-14 of NCSTAR 1-9A illustrate the 47-story model during collapse progression. These figures are viewed from the northeast rather than the northwest as labeled, and they indicate significant distortion in the upper stories that were not apparent in any of the photographs or videos taken during the event on 9/11.
"This behavior created numerical difficulties in the analysis, which were not likely to occur in the structure." NCSTAR 1-9, p. 489.
The ''behavior'' referred to above is the torque applied to spandrel beams from ''softened'' slab elements that carried floor live loads but had reduced stiffness. In some cases the supporting beam elements had failed and had been removed from the analysis. How many other numerical difficulties were encountered in the complex finite-element models that were not likely to
occur in the steel and concrete structure?
"Computer simulations…can be used to predict a complex degradation and collapse of a building." NCSTAR 1-9, p. 625.
Professor E.L. Wilson points out with regard to computer simulations: ''Remember the result obtained from a computer model is an estimation of the behavior of the real structure. The behavior of the structure is dictated by the fundamental laws of physics and is not required to satisfy the building code or the computer program's user manual."
Figure 8-21 of NCSTAR 1-9 shows the connection at column 79 supporting the W33x130 girder that spanned between columns 44 and 79. This column had three girders framing into it, but NIST says:
"The details of the connections of the other two girders are not shown." NCSTAR 1-9, p. 348.
Why not? The other two girders also provided lateral bracing for column 79, and the connection details are important. This is the connection that doomed the entire building.
Damage to framing connections from the ANSYS analysis was applied to the LSDYNA model as shown in NCSTAR 1-9 Figure 12-36 (and in NCSTAR 1-9A Figure 3-58.) A 100 percent failure state was assumed to occur for any calculated damage over 75 percent. The report says this assumption was made due to ''the coarseness of the shell element modeling of the fin, knife, and header connections in the LSDYNA model…'' NCSTAR 1-9, p. 566. Residual connection strength of 25 percent of the original strength, however, is substantial considering the safety factor used to ensure adequate design. The area under the strain stain curve is energy required for ultimate failure. If a 100 percent failure state was assumed to occur for any calculated damage over 75 percent, close examination of the stress-strain curve will reveal that the energy for failure is reduced by approximately 50% based on NIST's assumption for "the coarseness" excuse. Is that scientific or roughshod woo for the purpose of created a doomsday model?
"The buckling failure of the east floor beams and exterior columns was caused by restrained thermal expansion and failure of the shear studs along the beam length." NCSTAR 1-9, p. 537.
It is not clear what buckling failure of exterior columns is referred to in the preceding statement, and NIST previously stated …the beam displaced the girder at the interior end of the floor beam but did not displace the exterior frame at the other end of the floor beam." NCSTAR 1-9, p. 526.
If thermal expansion of the floor beams did not displace the exterior frame, then how would buckling of exterior columns occur?