27 April 2018
Shaking table tests of the Newmark displacement method for earthquake induced landslides
Posted by Dave Petley
Shaking table tests of the Newmark displacement method for earthquake induced landslides
Earthquake induced landslides are probably the most difficult mass movement phenomena to anticipate, and thus management of these hazards is proving to be very challenging. The issue is urgent – landslides triggered by the large earthquakes in Pakistan in 2005 and China in 2008 killed over 20,000 people in each case, and many more people live in earthquake-prone high mountain areas. The next tragedy could come at any time.
Assessment of the potential for earthquake induced landslides is often based upon the Newmark displacement method, often known as the Newmark sliding block model. In this conceptual model, the landslide is considered to be a rigid block on an inclined slope:-
Conceptual model of the Newmark displacement method, via the USGS
.
This set up allows the stresses operating on the system to be calculated in both a static (i.e. non-earthquake) and dynamic (i.e. during an earthquake) state. The fundamental idea is that there is a critical acceleration at which siding will initiate, so an analysis is run to determine this value. The model allows the displacement to be calculated for the parts of the earthquake for which this acceleration is exceeded. An assumption is made that when the total displacement exceeds a critical value the slope will fail. There are some problems with this – it is very hard to know what this critical displacement might be – but that is a separate issue.
The Newmark displacement method can be applied to any slope. In recent years digital datasets have allowed it to be applied across an entire landscape, which creates the opportunities for a spatial hazard analysis. This approach has been developed by the USGS, and others, and has been taken up quite widely. But the question has always remained as to how good the Newmark Displacement method might actually be.
Recently, I was a part of a team that examined one aspect of this – wave phasing. We hypothesised that the phasing of horizontal and vertical accelerations might critically change the displacements. We ran a series of experiments using a highly specialised piece of lab equipment that was developed for us by GDS Instruments – the Dynamic Back Pressured Shear Box (DBPSB). These results were published a while ago (Brain et al. 2015) – they ask some quite key questions about the Newmark displacement method.
In a paper just published in the journal Landslides, Li et al. (2018) have looked at a different aspect of the Newmark approach. In their work, they have examined the behaviour of a simple physical slope model when exposed to earthquake accelerations delivered by a shaking table. This is their model set up:
The physical model of the Newmark displacement method used by Li et al. (2018).
.
They undertook several analyses of the behaviour of the model, and in some cases they compared the response to shaking with that predicted by the Newmark displacement method. They conclude the following:-
By comparing critical acceleration between the experimental and theoretical results, the results indicated that Newmark’s method overestimates critical acceleration during seismic-induced dip slope failure. In reality, slip planes along existing discontinuities develop much more easily and can lead to even greater disasters.
In other words, for reasons that are not clear the Newmark displacement method suggests that the critical acceleration value is too high. This means that the analysis will indicate that, in some cases, slopes are stable when in reality they may be unstable during a modeled earthquake. This does not mean that the approach should not be used, and indeed it remains a key tool, but it does indicate that great care is needed. And of course it suggests that further research is required to improve the technique, or to develop alternatives.
References
Brain, M.J., Rosser, N.J., Sutton, J., Snelling, K., Tunstall, N. & Petley, D.N. 2015. The effects of normal and shear stress wave phasing on coseismic landslide displacement. Journal of Geophysical Research: Earth Surface. 120, 1009–1022
Li, HH., Lin, CH., Zu, W. et al. 2018. Dynamic response of a dip slope with multi-slip planes revealed by shaking table tests. Landslides . https://doi.org/10.1007/s10346-018-0992-2
Dave Petley is the Vice-Chancellor of the University of Hull in the United Kingdom. His blog provides commentary and analysis of landslide events occurring worldwide, including the landslides themselves, latest research, and conferences and meetings.
This is not the proper way to analyze the problem. There are two main types of landslides that are triggered by earthquakes. The first is the shallow slips of sediments on slopes that are usually triggered by a change in the ‘composition’ of a weak layer at the base of the slide. Often it involves some process similar to liquefaction that weakens a layer (or a previous rupture surface) and allows the unit to slip along that surface. The Newmark analysis is not really relevant to this unless the change in character of the rupture surface itself is included. The second is just as damaging, and that is the generation of rockfalls at the face of steep slopes, which are produced by the expansion and contraction, through a few cycles, of the slope face. That generates the extension of expansion cracks behind the face itself, which leads to an extension failure of the face. Those failures have nothing to do with the sliding geometry of the Newmark analysis.
[I agree that Newmark does not capture either of these two mechanisms well. But it is not really correct to say that these are the only two mechanisms in earthquake induced landslides. There are frequent “conventional” translational landslides for which Newmark is appropriate. D.]