my comments are in the attached pdf file

This article uses a highly non-linear, elastic plastic, soil model, which allows for initial porosity, compaction and plastic flow under deformation, potentially in concentrated "shear zones" in response to imposed boundary stresses. It presents a large set of numerical experiments on wedge like geometries (thrust belts) and their response to differing material and decollement properties and decollement angles.

By varying experimental conditions in a semi-systematic way, a range of deformation geometries and bulk material behaviours are generated. Specifically, the study sets out to examine under what conditions "back thrusts" are generated. The final conclusions of the paper are based on a statistical analysis of the significance of the individual parameters on the propensity for back thrusting. 

Regarding this final result, it has to be said that given the experiments concern a complex system (thrust wedge) there is almost certain to be a degree of cross-correlation and covariance of parameters and their influence on back thrusting. A valid statistic ought to explore this and I am unsure if that is the case. The details of the correlation method are not given. This is a significant issue for the paper, since it attempts to take a straightforward experimental approach to the question and thus, the correlation results are the main conclusion presented. I am not an expert in multivariate regression methods, but I am aware that this is a very large and complicated field and requires serious consideration to get right.  

Regarding the problem and the method itself, thrust wedges are, as already stated, complex systems. The plastic soild model used is a reasonable analogue for such a system, but this also ought to be discussed explicitly in the paper. The phenomenon under investigation is backthrusting, an emergent property of a complex system. Any attempt to study this problem needs to consider whether the model used is able to produce such emergent properties. There is little doubt that the critical state soil model is able to do so, but still, this should be a starting point for the paper. In this way, the initial discussion of critical wedges could be far better incorporated into the rest of the paper. 

The experiments themselves seem perfectly reasonable, but their presentation leaves a lot to be desired. Specifically, the experimental conditions are all placed in a single table, and then experimental results are presented by referring to the experiment's number. This is impossible to follow, since it requires endless switching back and forth between text, figures and the table to actually know what paramters are being used to get a particular result. Hence, the values of the the parameters should be visible on every figure showing model results, and ought to be incorporated into the text and discussion. 

The scope of the experiments is a slightly different question. One thing I noticed after a while is the nature of the basal decollement angle. It is assumed to be constant (as far as I can tell). There is very good reason (due to elastic flexure) to assume any decollement at the scale of thrust belt will have a degree of elastic curvature dependent on the internal friction of the thrust wedge material. This in turn will modify the thrust wedge / critical taper response of the entire system. This was quite well dealt with by Willett and Schlunegger (2010) in a paper on the Swiss Molasse Basin/Western Alpine thrust system. You may argue that at a small enough scale, the curvature no longer has an effect. The results of these experiments are also perfectly valid as they stand. But you cannot dismiss the potential significance of the elastic bending of the lithosphere in many cases, and you should discuss this effect somewhere in the paper and how it may modify things. That is not to say you need to run more experiments (although you could if you wanted) but if you don't, then you must acknowledge the limitations of the results in the broader context of thrust wedges. 

I found quite a lot of issues with the use of English in the paper. Some of these are highlighted in an annotated pdf.