jgostick
commented
8 years ago Hello Edgar Your issues with importing are coming at a very opportune time. We have JUST embarked on an extensive pull request to add more IO methods to the program (importing using the MatFile class in the Network module is not very convenient to be sure). I have built up a basic set of rules for file formatting and an outline of how the code generally works, so we're in a good position to add more IO options.
Now I would LOVE to hear your use case and learn about these STATOIL files. I could definitely cook-up an IO class for these for you. Perhaps you could attach these 4 files to this issue in a zip archive? I"ll then take a look at them, and see what I come up with. Basically, the work flow would be something like:
pn = op.Network.Import() pn.from_statoil(Node1='filename1', Node2='filename2', Link1='filename3', Link4='filename4')
Alternatively, perhaps you'd like to write the IO class and contribute it to our project? I'd be happy to work with you the whole way.
On Sun, Jan 24, 2016 at 11:03 PM, edgargmartinez notifications@github.com wrote:
Dear all,
I am not writing regarding an OpenPNM issue, the main reason is to ask some ways to import image-based networks, and how to modify the original data in order to be supported by the MatFile class Because of this, I decided open this issue to get this conversation going
First of all, let me introduce myself My name is Edgar G Matinez, and I’m currently working on my Petroleum Engineering bachelor dissertation, which is advised by PhD Martin Diaz-Viera The purpose of my work is to present pore network models as a methodology to compute effective properties caused by flow through porous media The OpenPNM accessibility, versatility and programming language influenced me to use it, and perform my simulations on it So, we've been following your work almost by a year, and I dare say that we have enough basis on PNM: the essential components, the phenomena, and mathematical models that PNM can support When I started working on OpenPNM, I must to admit, I knew nothing of Python This situation demanded me to learn on my own; however, I still have no knowledge of a high-performance programmer Whereby, this knowledge status has allowed me to understand only a 50% of your implementations
After handling some basic exercises and corroborate the obtained results against pencil calculations, my research took the next step: to consider more complex network topologies, such as image-based networks
Even though I couldn’t get rock images in the place where I’m doing my work, my assessor and I opted to manipulate open data from the Imperial College (Martin Blunt) The main drawback of this option is to convert the original data to a mat file The original information is given in a format called Statoil, which provides four dat files, a pair are for pores, and the other one for throats The properties that each file contains are presented next Node1 Node2 Link1 Link2 Pore Index Pore Index Throat index Throat index x-coord Volumen Pore 1 index Pore 1 index y-coord Radius Pore 2 index Pore 2 index z-coord Shape factor radius Lenght of pore 1 Coordination number Clay volume shape factor Lenght of pore 2 connecting pores indices Throat total lenght Lenght of throat Pore Inlet/outlet status volume connecting throats indices clay volume
Until now, I’ve been able to take the necessary properties for the mat file, but I had to edit part of the information to create ptype and ttype values The original data don’t specify whether pores belong to a surface; merely, it implicitly points out which pores are connected to artificial inlet/outlet pores So that, I generated these artificial inlet/outlet pores through deducting the coords and the pore connections, in order to obtain pores labeled as front/back, and two boundary surfaces With a mat file proposal, I tried it on OpenPNM for a StokesFlow simulation, but the internal throats' physics couldn’t be assigned Instead of numerical values for throatdiffusive_conductance, throatthermal_conductance, and throathydraulic_conductance, not-a-number and infinity are returned This situation also occurs when I'm using the test_pn mat file
— Reply to this email directly or view it on GitHub https://github.com/PMEAL/OpenPNM/issues/489.
edgargmartinez
007craig
Mandzhi
Dear all,
I am not writing regarding an OpenPNM issue, the main reason is to ask some ways to import image-based networks, and how to modify the original data in order to be supported by the MatFile class. Because of this, I decided open this issue to get this conversation going.
First of all, let me introduce myself. My name is Edgar G. Matinez, and I’m currently working on my Petroleum Engineering bachelor dissertation, which is advised by PhD Martin Diaz-Viera. The purpose of my work is to present pore network models as a methodology to compute effective properties caused by flow through porous media. The OpenPNM accessibility, versatility and programming language influenced me to use it, and perform my simulations on it. So, we've been following your work almost by a year, and I dare say that we have enough basis on PNM: the essential components, the phenomena, and mathematical models that PNM can support. When I started working on OpenPNM, I must to admit, I knew nothing of Python. This situation demanded me to learn on my own; however, I still have no knowledge of a high-performance programmer. Whereby, this knowledge status has allowed me to understand only a 50% of your implementations.
After handling some basic exercises and corroborate the obtained results against pencil calculations, my research took the next step: to consider more complex network topologies, such as image-based networks.
Even though I couldn’t get rock images in the place where I’m doing my work, my assessor and I opted to manipulate open data from the Imperial College (Martin Blunt). The main drawback of this option is to convert the original data to a mat file. The original information is given in a format called Statoil, which provides four dat files, a pair are for pores, and the other one for throats. The properties that each file contains are presented next.
Until now, I’ve been able to take the necessary properties for the mat file, but I had to edit part of the information to create ptype and ttype values. The original data don’t specify whether pores belong to a surface; merely, it implicitly points out which pores are connected to artificial inlet/outlet pores. So that, I generated these artificial inlet/outlet pores through deducting the coords and the pore connections, in order to obtain pores labeled as front/back, and two boundary surfaces. With a mat file proposal, I tried it on OpenPNM for a StokesFlow simulation, but the internal throats' physics couldn’t be assigned. Instead of numerical values for throat.diffusive_conductance, throat.thermal_conductance, and throat.hydraulic_conductance, not-a-number and infinity are returned. This situation also occurs when I'm using the test_pn mat file.