Bazant, Le, Greening and Benson give an interpretation of the Seismic data as indicating 12.58 +/- 0.5 and 10.09 +/- 0.5 seconds for the North and South towers respectively.
NIST says (in their FAQ):
So NIST says essentially, "we don't know".
Now, if the acceleration was constant, it would be possible to extrapolate from the video portions where the top is visible. I think Benson (mentioned above) and Einsteen have worked somewhat on this.
There are many theoretical calculations of the expected collapse times usually measuring the time from collapse initiation (visible movement) to impact of the collapse front with the bedrock, or sometimes ground level. Most have focused on the North tower because it is a less severe case in terms of the potential survival of the structure. The results of these calculations usually range from 12-25 seconds. Some have the collapse arresting. All the calculations I have seen are too simplified to give reliable results.
Many people arguing against controlled demolition like to argue that since portions of the core were standing the building wasn't collapsed yet. Technically this is correct, but this has no bearing on the point in time at which the collapse front hits the bedrock which could be compared to a calculated expected collapse time.
In order to do more realistic calculations of the collapse times the following must be established:
- failure mode - floor by floor column buckling is usually used, but there is little evidence supporting it
- column elastic strain energy - the spring(s) stiffness C is not generally agreed upon
- floor elastic strain energy - some energy is absorbed by flexion of the floor diaphrams
- energy due to inelastic collision - this is pretty well known and accepted
- floor plastic strain energy - often referred to as comminution of concrete. Newtons Bit denies this is a factor due to the ineastic collision, but the collision is not purely inelastic. In fact the upper floor of the lower part is restrained for some period of time at each impact.
- column plastic strain energy - we don't have a generally agreed upon value for this at the first impact
- scaling of column plastic strain energy - most models don't use this but it should be added
- column plastic energy during the initial collapse - most models do not include this
- elastic and plastic strain energies for floors and horizontal members - most models do not include this
- sheading of mass - there is no real agreement on the amount of mass expelled. It probably ramps up to a maximum at the first mechanical floors reached by the collapse front.
There are more issues, but this would be a start.