Structural damage to Dutch terraced houses due to flood actions

Abstract

’De Watersnood van 1953’, the largest Dutch flood in recent history, caused the death of 1795 people in the Netherlands directly from the flood conditions, while in the UK, 315 were recorded. Most of them were among those whose residence collapsed due to high water depth, quick rise rate of the water or strong flow velocity. Based on historical data of floods with similar flood characteristics and comparable buildings, mortality functions were developed to estimate the number of fatalities. These functions are still used, but the correlation between the flood characteristics and the damage observation is not clear according to multiple studies. The current study contributes to improving these functions by investigating which flood conditions may lead to collapse of the residences in the current Dutch building stock. From the BAG-registration (in Dutch: Basisregistratie Adressen en Gebouwen) it is found that 50% of the Dutch live in terraced houses (in Dutch: rijtjeshuizen), which is similar in the areas which are most likely to be affected by flooding. Most of these residences are built in the period of the housing shortage between 1965 and 1975 and the energy crisis between 1975 and 1994, which are considered as ’the typical Dutch residence’. This residence type consists of cavity walls with a load-bearing leaf of concrete or unreinforced masonry (URM), which can be clay or calcium-silicate. This inner leaf is tied to the outer leaf of URM consisting of perforated clay units, wood-based materials, or concrete. Stability is provided by piers in the façades in case of the URM walls or rigid connections between the concrete floor and walls. To define the properties of the building materials, existing experimental research on the masonry is used.
Experiments with a physical model were conducted herein to measure the quasi-steady load in the form of pressures acting on different elements of the residence. This enables the comparison of the quasi-steady flood load and the lateral load due to wind on different elements of a building. Similar to FEMA (2011), it was found that the pressure coefficient decreases when the width-to-water depth ratio decreases. However, higher coefficients are found from the experiments than those provided by FEMA, resulting in higher hydrodynamic loads. Furthermore, the orientation of the residence compared to the flow direction changes the angle of attack. When the flow is perpendicular to the wall, the pressure coefficient is the largest. Decreasing the angle of attack causes a decrease of the pressure due to equal flood conditions. The pressure coefficients obtained from the experiments are used to define the hydrodynamic load due to flooding. The resistance of the load-bearing cavity walls, windows and piers were compared to the acting moment due to different depth-flow velocity combinations. The resistance of out-of-plane bending of the load-bearing wall is the critical failure mechanism for typical Dutch residences. Residences with calcium-silicate masonry walls and system floors have a higher resistance than residences with clay masonry walls and timber floors. Cracks start to develop at a small lateral load resulting in zero tension strength after cracking and an eccentricity of the normal force. This makes the influence of the dead weight carried by the wall, in combination with the compression strength and the thickness, more important than the flexural bending strength.
All types of residences, using design values, already collapse before the hv-product (water depth times flow velocity) of 7 m2/s is reached according to Clausen (1989). A water depth of ±1.2 meters for the older residences (1965-1975) and ±1.8 meters for the newer residences (1975-1994), already cause the design moment resistance of the wall without taking the velocity or wave action into account. If the flood water has a flow velocity of 2 m/s or waves are generated by a wind speed of 29.5 m/s over a fetch of 100 m, the critical water depth reduces to respectively ±0.9 and 1.5 meters.