9/11 WTC 2 Aircraft Impact Orientation (Draft)


The intention of this document is to determine the fuselage impact orientation and nose-to-tail impact trajectory of the aircraft that crashed into WTC 2 on 9/11, and to document the methods used to derive the orientation and trajectory values.


Throughout all steps of this method, the corresponding values used by NIST within their computer simulations and impact damage assessment studies will be included, and any differences highlighted.


This method intimately combines accurate 3D models of the aircraft and buildings with photographs and video of the actual event to produce self-validating visualisations of all orientation values.


The rotoscoping of virtual 3D models over actual photographic and video evidence of the real event is an inherent part of this method, and provides an additional level of validation not yet presented within other methods, such as those used by NIST.


Digital resources utilised will be made available via links in the appendix to allow full personal validation of all enclosed details, measurements and values.



The Aircraft


The aircraft which impacted WTC 2 was identified as a long-range version of the Boeing 767-200, specifically a Boeing 767-222/ER.


Basic Scale Specification


Length            48.5m             159ft 2in

Wingspan       47.6m             156ft 1in




The 3D Aircraft Model


The 3D model used is a public-domain rendition of the Boeing 767-200, sourced from the Google SketchUp directory, which has been converted into 3DS format and is specifically scaled in world coordinates. The world coordinate system used is 1ft=1unit. Primary dimension scaling, such as aircraft wing-span, is accurate to within 1 inch. Sub-feature scaling is accurate to much greater precision, relative to the scaling of primary dimensions. Accuracy of the model is more than reasonable for the purposes of this method.






Comparison to the NIST 3D Aircraft Model



The identified aircraft (Flight 175 – UAL175)




The Buildings


World Trade Centre Complex


A correctly scaled and very detailed public-domain model of the WTC Complex is used for all three dimensional rotoscoping tasks.


The following images show the accuracy and detail of the model in an artistic sense, followed by specific comparison between the model and technical diagrams used by NIST.



(Add more images)

Note the accuracy in alignment between the tower external columns, spandrel plates and corner construction elements.



Alignment Methods


The first method of alignment used is the definition of the specific extremities of the aircraft impact damage outline, which can be determined by inspection of photographic and video evidence.


Three ‘control points’ are defined:


1)     The port (left) wing-tip façade penetration point.

2)     The starboard (right) wing-tip façade penetration point.

3)     The nose-cone impact point.


Any derived orientation and trajectory must result in the nose-cone making contact at the nose-cone impact point, the port wing-tip penetrating the façade at the port wing-tip penetration point and the starboard wing-tip penetrating the façade at the starboard wing-tip penetration point.


These three simple defined points strictly limit the orientation and trajectory possibilities.

Any orientation and trajectory that does not adhere to these control points, within reasonable errors of margin, cannot be correct.


The second method of alignment is the rotoscoping of a virtual three-dimensional environment upon photographic and video evidence of the physical event.

Placement of the footage camera is determined and all appropriate viewport perspective parameters are adjusted to ensure maximum match quality between virtual and real views. Due to the limitations and implications of perspective and aperture adjustments available in the version of 3D studio I have available, the virtual camera placement cannot be exactly ‘on’ the defined real-world camera position, but is behind and below it. It is however ‘in-line’. This will be shown to provide accurate angular referencing between virtual object placement and real-world imagery, with angular resolution accuracy increasing as the distance between the real-world camera location and the ‘target’ (WTC2 in this case) increases.



Control Points


Port Wing-Tip






The NIST point placement is mid-way across column 442, at the bottom of the 78th floor spandrel plate.


This study will define the point as being at the following location:





Starboard Wing-Tip


Due to the lack of high-resolution photographs of the starboard wing-tip penetration point that are not obscured by smoke, the accuracy of the penetration point position is lower than for the Port wing-tip. The reduction in positional quality is indicated by an increase in Control Point marker size in the virtual model.








NIST Point placement is on the right hand edge of column 404, mid-way between the inter-spandrel plate space separating floors 85 and 86.


This study will define the point as being:






Sources to determine nose-cone impact position are clearly limited to video footage.




NIST Nose Impact Point Placement




NIST point placement is at the top of the 81st floor spandrel plate, and in the middle of column 423.


Here is that point placed on the 3D building model:



Accurate placement of the nose-cone impact point will be undertaken in conjunction with building model rotoscoping.

Definition will be conducted a little later in this document, after camera location of the video footage has been documented, and viewpoint orientation and perspective has been determined. This will allow for more accurate definition of the control point.



Virtual Building Positioning


WTC Complex


Purely to increase rendering performance, buildings in the full WTC complex base model other than WTC1 and WTC2 have been removed.

WTC 1 and 2 relative positioning is fully retained, positioned such that the coordinate of the centre-point of WTC 2 footprint is at (0,0).

The grid shown is in 10ft units.




Additional Buildings


Google Earth/Sketchup models of additional buildings present in the video footage being rotoscoped have been converted to 3DS format and imported into the environment.

As 3D building placement and scaling within Google Earth is not exact, placement of each building has been performed with reference to the underlying satellite photography available within Google Earth, and each model individually scaled according to published building heights. Further refinement of these models is required to fully finalise this method, however every care has been made to ensure the placement is as accurate as possible.



Downtown Athletic Club


Height – 162.8m / 534ft (Source)


Whitehall Building


Height – 129.24m / 424ft (Source)


Geometry placement


The following overview shows the relative placement of ‘scene’ elements, including WTC1, WTC2,  Downtown Athletic Club, Whitehall Building and Camera.

The grid shown is in 100ft units.

The blue cube near to WTC2 SW corner represents to camera ‘target’.




A Quick Look at the NIST Orientation


Before continuing, it is prudent to highlight the most fundamental issue with the NIST impact orientation values.


The following table contains the trajectory and orientation data that NIST based their impact simulations upon:



The Lateral Approach Angle and Lateral Fuselage Orientation values are a significant focus of this study.


The simplest way to begin is to look at the relative portion of the starboard (right) wing which has not penetrated the façade at the point in time that the port (left) wing tip penetrates the façade. Several video clips include enough detail to highlight this metric, and two will be referenced here.



This still frame is taken as near to the point in time that the port wing tip penetrates the façade as footage allows:



Note the portion of the starboard wing that has not yet penetrated the façade, and the length/orientation of the remaining fuselage segment.


Compare it with this virtual render, which has the aircraft set to the base NIST orientation (NCSTAR2 E-23):


(We have not yet fully defined the video rotoscoping, and it is not suggested that the virtual camera perspective illustrated is spot-on. In the context here that is irrelevant, as we are looking at relative feature proportions.)




If the problem is not obvious enough (viewing angle really does make a big difference) cross-referencing with another well-known clip may help…



It does of course require careful inspection to see the problem, which can be summarised as being that, at the point in time that the port wing-tip penetrates the façade, in the actual footage a maximum of one half of the starboard wing width is visible, with the other half having already entered. Whereas in the virtual render (in which viewing angle is totally irrelevant) the full width of the wing has not yet penetrated the façade and even the wing root is still visible.


Take some time to compare the actual footage with a rendered view of the impact with the NIST base orientation:




Whilst you may think that the difference is not extreme, it will be shown later that the resultant effect of orientation and trajectory changes upon potential damage inside the building is highly significant.

When the video rotoscoping is performed, the orientation differences are further highlighted.



Aircraft Dihedral Wing Angle


Even in normal flight, forces upon the aircraft airframe result in ‘bending’ and flexing of the wings. When performing increased ‘G’ manoeuvres, the amount of wing flexing increases.

The resulting change in dihedral wing angle must be taken into account when determining the impact orientation, as change in this angle affects the position of the wing-tips relative to the nose-cone, and also results in very slight narrowing of the ‘plan-view’ wing-span as the angle increases.


NIST Representation of Wing Deflection



Aircraft Model Representation of Wing Deflection


A similar, but very simple, process is also applied to the aircraft model in this study, namely a simple ‘bend’ modifier applied to the aircraft:



The Importance of treating Trajectory separately from Orientation


NIST specify trajectory and orientation separately, but appear to have combined the two axis values together within their simulations.


“The aircraft and exterior wall models were used to visualise the impact scenario in the figures and the view shown was aligned with the aircraft trajectory.” NCSTAR1-2 E.6



The simplest way to understand the need to treat them separately is to visualise the behaviour of an aircraft whilst landing, in which the orientation is inclined at a significant angle, whilst the direction of vertical travel is in entirely the opposite direction:




Another look at the base NIST Orientation


We now have the wing-tip penetration points defined, and the NIST nose-cone impact point.

This provides enough information to test the NIST base impact orientation values (Which they align to trajectory).


Do the base NIST values result in the wing-tips passing through the control points ?


Port Wing Without Deflection:


Port Wing with 10 degree Deflection:


Port wing-tip penetration with the NIST base values is significantly off, both with and without deflection.


Starboard Wing with No Deflection:


Starboard Wing with 10 degree Deflection:


Starboard wing-tip penetration point accuracy is pretty much spot-on with deflection, and also fairly reasonable without deflection.


HOWEVER, it has already been shown, by looking at relative amounts of the wing that have penetrated the façade, that the horizontal angle is incorrect.

This point will be emphasised and made clearer when the rotoscoping is performed.


Sample rotoscope, using a non-final camera location (Orientation: NIST 6,38,13)


Anything seem to be amiss here ? The rendered aircraft looks enourmous.

However, please note that if both the vertical and horizontal angles were different, then this apparent ‘bizarre’ view is then clarified…


Sample rotoscope, using non-final camera location (Orientation: TEST 3,37,3.5)


Much better. Not perfect yet, but a lot closer to matching the footage than the NIST orientation.

Please remember that there is no change to the size of the aircraft at any point in time.

Even though the camera viewpoint is non-final, it is clear, even at this point, that the NIST orientation is significantly incorrect.


The difference between the orientation angles above is quite marked, with a difference of 3 degrees vertical and a massive 9.5 degrees horizontal.


As you can see, using the video rotoscoping method provides an additional mechanism to validate orientation.


The base NIST orientation seems to be a good match for the starboard wing penetration point.

The base NIST orientation seems to be a dubious match for the port wing penetration point.


The base NIST orientation viewed from a rotoscoping perspective is atrocious, and has serious and far-ranging implications for NIST’s subsequent impact damage analyses.


The rest of this study will focus upon increasing the accuracy of the nose-cone impact location, the correct aircraft orientation and trajectory, and the rotoscoping camera viewpoint.


I will upload “part 2” as soon as possible. This version is for comment within the911forum only, and does not constitute an ‘official’ release.