Predicted and actual performance of masonry parapets in the 2007 Gisborne earthquake

Technical Abstract

The author previously developed a methodology that can be used to predict the earthquake performance of the walls and parapets of masonry buildings. This procedure has been incorporated, in modified form, into the New Zealand Society of Earthquake Engineering procedures for assessing and improving the structural performance of earthquake risk buildings.

On the 20th of December 2007 a Mi 6.9 earthquake occurred 47 km southeast of Gisborne. This earthquake caused damage to unreinforced masonry buildings in the Gisborne CBD that included the collapse of 22 parapets. Fortunately a record of the ground motion in the Gisborne CBD as obtained. This record could be used to evaluate whether the risk of parapet collapse in the earthquake was consistent with that predicted by the procedure developed by the Author and as modified by the NZSEE.

To enable the assessment procedures to be used to compare the predicted and actual behaviour of the parapets in the Gisborne CBD details were obtained for 19 of the 22 parapets that collapsed in the 2007 Gisborne earthquake. A further 82 parapets were also surveyed so that they could be evaluated using the assessment procedures.

It was found that the best prediction of the actual parapet behaviour was given by a modified version of the Author’s procedure. This modified procedure is actually an amalgam of the Author’s original procedure and the NZSEE procedure.

The parapets in the Gisborne CBD had been exposed to an earlier 1993 Ormond earthquake. A comparison between the earthquake records of the 1993 and 2007 earthquakes indicated that the 1993 earthquake was significantly weaker, particularly in the direction parallel to the main street. Almost all the parapets that collapsed in the 2007 earthquake were subjected to this component of the earthquake. This relatively greater damage was consistent with what would be expected from using the assessment procedures.

In the other direction, where the parapets are parallel to the main street, the spectral displacement intensities of the two earthquakes were similar and only one parapet collapsed in the 2007 earthquake and none collapsed in the 1993 earthquake. The relative amount of damage was, therefore, consistent with the relative spectral intensities of the two earthquakes as would be expected from the assessment procedures.

However in this direction, more collapses would have been predicted in both earthquakes using the assessment procedures. It is quite possible that these parapets had been subjected to significantly higher earthquake motions during the even earlier 1966 Gisborne earthquake which would have “culled” the weaker parapets. This “culling” may help to explain why fewer of the parapets orientated parallel to the main street collapsed in the 2007 than would have been predicted by the Author’s modified assessment procedure.

Only approximately 6% of the total number of parapets exposed to the earthquake actually collapsed during the 2007 Gisborne earthquake. This study indicated that the Author’s modified procedure would have predicted a greater probability of parapet collapse. It is suggested that a significant proportion of this difference in performance can be explained by the manner in which the external walls of the masonry buildings responded to the earthquake. It is believed that many of the external walls of the Gisborne CBD buildings responding, at least in part, as free standing vertical cantilever walls rocking about a horizontal “crack” opening at foundation level. This type of behaviour is to be expected when the floor and roof diaphragm of a building is not strong or stiff enough to significantly affect the response of the walls. Vertical cracks at the corners of some of the Gisborne masonry buildings where the external walls had separated, indicates that the external walls may have responded, at least in part, as vertical cantilevers.

When the external walls respond to an earthquake as vertical cantilevers the parapets are not subjected to as highly amplified ground motion and are not as likely to collapse. Paradoxically this suggests that if roof diaphragms are stiffened and strengthened the parapets will be subjected to amplified ground motions and be more likely to collapse in a moderate earthquake. The Author’s modified procedure may then more accurately predict the actual risk of collapse.