JOSS paper review
I have reviewed your software paper. Please find my comments below @luisaforozco
- In line 17, the paper states that the framework supports "highly flexible finite-volume spatial approximation". I don’t understand what "highly flexible" refers to in the context of discretization. Please provide more details in the Discretization section in page 2.
- In line 18, the paper states that the framework supports "efficient physics based preconditioners". Details regarding the preconditioners is unfortunately missing in the paper. Please provide more details (~ 1-2 lines) in page 2: a) does the framework use in-built preconditioners or does it use preconditioners from third-party libraries, b) which preconditioners are supported? etc.
- As per the review checklist for JOSS (see review thread), the authors need to describe how openDARTS compares with other commonly-used packages (some I could think of are TOUGH, AD-GPRS, PFLOTRAN, CSMP++, MRST etc.) in this field. This comparison has not been made in the paper.
- Motivation in the section Statement of need: a) The motivation for this software is described in lines 30-34. This section could be re-arranged such that the motivation for this software comes first; b) in the motivation part of this section, you may also write down the advantages of this software over commonly used packages in this field (as mentioned in point 3 earlier).
- Line 26: flexibility and performance in what aspects? Please add more details.
- On this webpage (https://darts.citg.tudelft.nl/capabilities/), I found additional key features of this software, such as capabilities to handle fracture networks, support for thermodynamic equations of state, support for adaptive mesh refinement, can be run on GPUs etc. If these features have already been implemented and tested, they could be added to the paper.
- Line 41-43: I don't understand the need for these lines. These lines could be instead rewritten to mention the multiphysical processes supported by the software: darcy flow, gravity, multi-component & multiphase flows, thermal flows, chemical and kinetic reactions etc.
- Lines 44-46 "Observing .... integral way" do not convey any essential information. I feel that these lines could be removed. Since finite volume method is used for the discretization, line 45-46 on "discretize ... using a uniformly integral way" is implied.
- Lines 46-48: Please combine this information with line 67 concerning discretization. Finite Volume Method -> finite volume method, Fully Implicit Method -> fully implicit method (avoid using capitalized first letters).
- Line 53: Please avoid using symbols \eta and \omega, as you do not use them in any equations.
- Lines 63-66: Please write these in a better way: a) structured grid for ...., b) radial grid for ...., c) ... etc.
- Line 72: the usage of finite volume method has already been mentioned before. Please remove that information here.
- Line 74: Is it supposed to be contact mechanics formulations? I am not aware of a word termed "friction contact mechanics". Does the framework support linear or nonlinear (large-deformation theory) poro-elastic formulations? Please mention this in the paper.
- The sections on HPC and Python can be combined into one section on programming or something like that.
- Line 91 is very specific to the framework. I feel that it can be removed.
- Please list the key features in pages 2-3 in a logical order, i.e., equations -> discretization->programming etc. From what I follow, this would be the right one, but please feel free to modify it in case I am wrong: Unified thermal-compositional PDE formulation, Geomechanics, Discretization, Operator-Based Linearization, Programming, Inverse modeling.
Please tag me once you are done with the changes so that I can update the checklist in this review thread