After being able to access wind tunnel data of the Citicorp building, Dat Duthinh writes that he now understands the basis for the decision to perform emergency repairs on this famous skyscraper.
Following structural engineering practice of the 1970s, engineers designed the Citicorp building for the action of wind in each of the structure’s principal axes.
Combining simultaneous wind loads
One problem they faced was how to determine design values by combining simultaneous wind loads from the means and root mean squares of these loads.
Given the technology available at the time, the simplifying assumption was made of decomposing the effects of a corner wind into the sum of the effects on two adjacent faces, as if these wind loads were static and perfectly correlated, which they are not.
As a result of this assumption, the connections in the diagonal braces were deemed overstressed, and emergency repairs had to be performed on the just completed building.
Nowadays, the Database-Assisted Design method and modern computer technology are capable of accounting for simultaneous dynamic loads properly, performing structural analyses for all time steps where measurements are available, and extrapolating to longer duration windstorms using extreme value distributions.
Modern analysis thus determines design loads on a more rational basis and shows that the combinations of wind loads that caused such concern in 1978 do not need to be considered for mean recurrence intervals of practical interest.
'Wind engineering an emerging technology'
“Wind engineering is an emerging technology and there is no consensus on certain aspects of current practice. Unfortunately, the use of ASCE 7 with wind tunnel-produced loadings is not straightforward.”
This statement was written in 2004 (Ref. 1) by a leading architectural and engineering firm of tall buildings, Skidmore, Owings and Merrill (SOM), in reference, not to the Citicorp building, but to another New York City landmark, the World Trade Center towers (destroyed in 2001), and the American Society of Civil Engineers Standards on Minimum Design Loads for Buildings and Other Structures.
This assessment was certainly true in 1978, when the urgent decision was made to strengthen the Citicorp building.
Recently, using data of wind tunnel tests on a model of similar proportions and a modern computer method called database-assisted design, researchers concluded that face winds and not corner winds govern the design of the Citicorp building (References 2 and 3).
Demands on eight-storey chevron braces
The demands on the eight-storey chevron braces caused by corner winds are significantly less than those caused by face winds. The researchers cautioned that these conclusions do not account for the effects of neighbouring buildings or the exact geometry of the Citicorp building, with its iconic wedge-shaped top and midside base columns.
References 2 and 3 called for a re-examination of the reasons for the emergency repairs of the Citicorp building shortly after its completion, in light of the progress in wind engineering and structural engineering of the last four decades, which now allows a structure to be analysed at every time step using dynamic loads measured simultaneously, thus taking the guesswork out of the determination of the governing load cases.
This note, based on a lecture given by Le Messurier at MIT in 1995 (Reference 4) and wind tunnel practice in 1975 (Reference 5), is intended to shed further light onto this famous case.
In his lecture, Le Messurier mentioned that the wind tunnel tests performed at the University of Western Ontario (UWO) included buildings within a 2,000ft (610m) radius of the Citicorp building, and showed that face winds caused higher overturning moments than corner winds.
Doubling of stresses
This responds to the concerns of References 2 and 3. Le Messurier also showed a sketch, where the effects of corner winds can result in the doubling of stresses in the chevron braces relative to the effects of face winds.
This doubling of stresses was not accounted for in the original design and construction and was the major (but not the only) reason for the emergency strengthening of the connections in the chevron braces.
Let us examine the overturning moments caused by face winds and corner winds on the Citicorp building, as calculated by UWO (Reference 5). From wind tunnel measurements for all wind directions α, every 10°, UWO calculated overturning moments about the building axes x and y, which are normal to the South (S) and West (W) faces of the building (Table 1 and Figures. 1-5).
M_x and M_y refer to moments caused by wind forces F_x and F_y respectively. Wind angle α is measured from Reference North (N), which is 18° from the North face.
The positive (negative) peak moments are obtained by adding (subtracting) 3.5 times the root mean squares (rms) to the mean values, as recommended in Ref. 5, under the assumption that random vibrations follow a Gaussian distribution. The overbar indicates the mean value.
Note: Wind speed 100 mph (44.7 m/s, 161 km/h), damping 0.005 of critical, actual building configuration with roof sloping South. 1 ft·lbf = 1.356 N·m.
Peak± = mean ± 3.5 × root mean square (rms) for random vibrations.
Even though the designer had access to the time histories of wind tunnel measurements (local pressures from dozens of taps on walls, and global shear forces and overturning moments from base balance) recorded on magnetic tapes at UWO, the computers at the time did not have the capacity to calculate the time histories of member stresses for all these simultaneous loads.
The problem was then to determine design values by combining simultaneous loads from their means and root mean squares evaluated over time. Figure 3, for example, shows that the moments M_x and M_y due to winds from the North (N) face, West (W) face or Northwest (NW) corner are of comparable range.
Decomposed effects of a corner wind
Following the practice at the time, the structure was designed for the action of wind in each of the building principal axes. The designer appears to have decomposed the effects of a corner wind, NW, into the sum of the effects on two adjacent faces, N and W, as if these wind loads were static and perfectly correlated (Figure 6, adapted from Reference 4).
These components would be approximately equal to the effects of winds on the N and W faces acting separately. Compared to face winds, corner winds would thus lead to a doubling of axial forces in some chevron braces and their nullification in some others.
θ_w=0° θ_w=45°
This point is crucial: given the measurements and the computer technology available at the time, the designer appears to have made some simplifying assumptions and decomposed the effects of a corner wind into the sum of the effects on two adjacent faces, as if these wind loads were static and perfectly correlated, which they are not, as shown vividly in Figure 7 ( Reference 6).
As indicated in his MIT lecture (Ref. 4), he decided as a result to perform emergency repairs on the just completed building. This is surprising since, before making his decision, Le Messurier consulted Davenport of UWO, who had developed methods to account for randomness in calculating maximum wind loads (Reference 7).
Along- and across-wind effects
Figure 6 could conceivably apply also to the simultaneous along- and across-wind effects caused by wind coming from one direction. References 2 and 3 show that the most demanding resultant overturning moment is caused by face winds and has across-wind component M_y equal to the peak M_y, but along-wind component M_x much less than the peak M_x. Thus, the doubling of stresses shown in Figure 6 would not occur.
Thanks to modern computer technology and database-assisted design, which account for simultaneous dynamic loads properly, perform structural analyses for all time steps where measurements are available, and extrapolate to longer duration windstorms using extreme value distributions, it is now possible to show that the combinations of wind loads that concerned the designer do not need to be considered for mean recurrence intervals of practical interest (References 2 and 3).
In fact, the current ASCE 7-16 (2017, Reference 8), Figure 27.3-8, specifies that wind loads acting simultaneously should be considered at 75% of their specified values when acting separately.
Author: Dat Duthinh, National Institute of Standards and Technology, Gaithersburg, Maryland 20899-8611; Tel: 301 975 4357; email: dduthinh@nist.gov
References
1. SOM (Skidmore Owings and Merrill LLP) (2004) 'Report on estimation of wind effects on the World Trade Center towers' http://wtc.nist.gov /NCSTAR1/NCSTAR1-2intex.htm
2. Duthinh, D (2019) 'Blown Away: Citicorp Center Tower Repairs Revisited', Engineers Journal.ie, Dublin, Ireland, July 1, http://www.engineersjournal.ie/2019/07/01/blown-away-citicorp-center-tower-repairs-revisited/ [cited 12.4. 2019]
4. Le Messurier , W (1995) 'The Fifty-Nine Story Crisis: a Lesson in Professional Behavior', MIT Mech. Engg. Colloq. Nov. 17, https://www.youtube.com/watch?v=um-7IlAdAtg [cited 12.4. 2019]
5. Isyumov, N, Holmes, JD, Surry, D and Davenport, AG (1975) 'A study of wind effects for the First National City Corporation project – New York', USA, University of Western Ontario Research Report BLWT-SS1-75, London, Ont. Canada, April
6. Dutta, S, Panigrahi, PK , and Muralidhar, K (2008) 'Experimental Investigation of Flow Past a Square Cylinder at an Angle of Incidence', ASCE J Engineering Mechanics 134, 9, 788-803 DOI: 10.1061/(ASCE)0733-9399(2008134:9(788)
7. Davenport, AG (1964) 'Note on the distribution of the largest value of a random function with application to gust loading', J Inst. Civ. Eng. 24, 187-196
8. ASCE Standard 7-16 (2017) 'Minimum Design Loads and Associated Criteria for Buildings and Other Structures', American Soc. of Civil Engineers, Reston, Va
Disclaimer: The policy of the National Institute of Standards and Technology is to use the International System of Units (SI) in its technical communications. This paper, however, discusses measurements performed 45 years ago in the same customary units used then (and now) in US construction industry.