<![CDATA[Hardy Geoscience Substack]]>https://hardygeoscience.substack.comhttps://substackcdn.com/image/fetch/$s_!c9qM!,w_256,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fbce87a8c-eb77-4c54-9b1b-666760e73592_224x224.pngHardy Geoscience Substackhttps://hardygeoscience.substack.comSubstackTue, 01 Sep 2026 17:37:52 GMT<![CDATA[Digressions]]>https://hardygeoscience.substack.com/p/digressionshttps://hardygeoscience.substack.com/p/digressionsTue, 22 Jul 2025 20:42:12 GMTMore digressions in visualisation. Frictional-cohesive cover, homogenous, 200 m displacement accruing on basement fault. 4 movies of the evolution of this 2D discrete-element simulation all on identical "time" loops: geometry, a zoom of a high-level portion of the section, incremental displacement magnitude over previous 2m of slip, and incremental max. shear strain over the same interval. When does information overload occur? Are 2 panels optimum, 3 okay and 4 too much? I am genuinely interested to know……

download movie here

]]><![CDATA[It's all very superficial I know...]]>https://hardygeoscience.substack.com/p/its-all-very-superficial-i-knowhttps://hardygeoscience.substack.com/p/its-all-very-superficial-i-knowSat, 12 Jul 2025 20:32:37 GMTFirst post after an intentional, and much needed, break. Not my usual scale/domain, but here is an example of a "high-resolution" (average element radius 0.7m) 2D discrete element model of high-level (almost surficial) deformation. 100m thick cover, frictional-cohesive - think basalt or limestone - overlying a 60 degree basement fault at depth. I was testing here not only high-level, high resolution deformation but also looking at the geometric and (capturing) the shear strain evolution during quite small increments of fault displacement. Here, I show the total model evolution up to 15m displacement at increments of 20cm in terms of simple geometry and incremental max. shear strain. (model starts with a period of equilibration, nicely illustrated by the inc. shear strain)

full high-res video

Read more

]]><![CDATA[Too hot....time to dig out old code blocks]]>https://hardygeoscience.substack.com/p/too-hottime-to-dig-out-old-code-blockshttps://hardygeoscience.substack.com/p/too-hottime-to-dig-out-old-code-blocksSun, 29 Jun 2025 20:48:39 GMTIt's almost a bit too hot to do much at all... so I am going through a series of bespoke, maybe oddball, boundary conditions and set-ups in my 2D discrete element codes and, well, just seeing what comes "out in the wash". Here is an example of a code block that contains this functionality: some kind of indenting body - cover has a lower, frictionless unit overlain by a layered, frictional-cohesive unit. This movie shows the geometric evolution (top panel) and the total max. shear strain during the simulation...

download movie here

]]><![CDATA[Nestor does his magic....]]>https://hardygeoscience.substack.com/p/nestor-does-his-magichttps://hardygeoscience.substack.com/p/nestor-does-his-magicMon, 23 Jun 2025 20:02:16 GMTI have always wondered about the “best” way to view my (or anyone else’s) 3D discrete element modelling results. Personally, I think that ParaView is a bit “clunky” but very powerful, whilst ImageTank on the Mac is incredibly powerful and slick but not everyone has a Mac. However, in the end, you need to invest quite a bit of time … almost becoming an expert user - with them both to get the most out of them. For most mere mortals this is not an option. After a frenzied day of experimentation (and WhatsApp) around cdem3D output and Python visualisation, Nestor Cardozo has done his thing again producing a Python cdem3D viewer, with instructions, examples etc etc. He is an incredibly fast worker. Contact me for more details about cdem3D etc. Here is the github resource (or whatever you are meant to call it)

link to viewer

Hardy Geoscience Substack is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

]]><![CDATA[That sinking feeling]]>https://hardygeoscience.substack.com/p/that-sinking-feelinghttps://hardygeoscience.substack.com/p/that-sinking-feelingSat, 21 Jun 2025 15:33:01 GMTThat sinking feeling. A quick movie of a high-resolution 2D discrete element simulation, the average element radius is 0.75m. Here we have a 100 m thick, homogeneous, frictional-cohesive cover subject to a total of 30m block subsidence at its base. This simulation is part of a series where I have been looking at different strengths of such high-level covers. The simulation (like all) is subject to an initial "equilibration" phase, which is shown here for completeness. In the top half of the movie the simple geometric "collapse" is shown, whilst in the bottom half the incremental max. shear strain over the previous 1m subsidence is shown. High-resolution models like this are computationally quite intensive but worth the wait I think.

download movie below

high-res movie

Read more

]]><![CDATA[With fresh eyes]]>https://hardygeoscience.substack.com/p/with-fresh-eyeshttps://hardygeoscience.substack.com/p/with-fresh-eyesThu, 12 Jun 2025 10:08:13 GMTAs ever, revisiting something you did a while ago with fresh eyes/tools/ideas/capabilities is a good thing. Here is a case in point: a 2D discrete-element simulation: a simple 10 km wide section, maximum thickness 1.25 km - total shortening 3.1 km solely by movement of right end-wall. The "footwall" is static and carved out of the section, the base, fault segments have no movement imposed on them at all, but the are very low friction and cohesion-less - so they (or segments of them) are activated when the mechanics "says so".... (no mylar sheet being pulled or anything like that), in addition there are 4 weak, thin, layers in the cover section. Anyway, I ran this for some fun, and hopefully some insight as a result of testing with fresh eyes. The movie shows the geometric evolution in the top panel, and the incremental shear strain (over previous 2m shortening) in the bottom panel.

dowload the movie here

Read more

]]><![CDATA[The 32 Threads]]>https://hardygeoscience.substack.com/p/the-32-threadshttps://hardygeoscience.substack.com/p/the-32-threadsSat, 07 Jun 2025 13:54:01 GMTWhen modelling deformation, high resolution and high displacement combined can be a tall order. Computationally it is quite demanding but if you have a machine that you can "hijack" for a while the results are worth waiting for I think. Here is a (2 1/2 mins or so) movie of 6 km shortening being accrued/accommodated in a thrust wedge. Initial geometry 1.25 x 20.0 km, so 30% shortening of this frictional cohesive section, very weak base. Resolution is approx. 10m, c. 80,000 elements. Right end-wall is displaced horizontally 0.0005m per time step to accrue the 6000m shortening (that is 12 million time-steps). But the 2D discrete element code handles this with ease and is stable. 32 threads OpenMP running under Windows. First half of the movie is just the "simple" geometry as it develops showing lots of nice features, second half of movie is the incremental max. shear strain (calculated over previous 2m of shortening) which is a great way of visualizing localization, activation of faults etc etc.

download the movie here

]]><![CDATA[Back on that relay-ramp trail....]]>https://hardygeoscience.substack.com/p/back-on-that-relay-ramp-trailhttps://hardygeoscience.substack.com/p/back-on-that-relay-ramp-trailTue, 03 Jun 2025 12:36:18 GMTOn my 3D discrete element testing/investigation journey I have ended back up at relay-ramps... this time looking at a ramp at surface, developing in a 300 m thick frictional-cohesive cover above 2 overlapping, 65 degree, normal faults at depth. Fault separation 450 m and total displacement here is 230 m. Just experimenting with different views, perspectives etc. This movie stops just as the ramp is about to be breached.... "a breach too far" movie will follow, just need to find time/inclination.

download movie here

]]><![CDATA[More of the (interesting) same]]>https://hardygeoscience.substack.com/p/more-of-the-interesting-samehttps://hardygeoscience.substack.com/p/more-of-the-interesting-sameThu, 29 May 2025 11:22:41 GMTSometimes a rather mundane, seemingly simple, numerical model can tell you quite a lot if you just stop to really examine it in some detail (Part II). I present my case in point: a 2D discrete element model of simple horizontal contraction, section initially 20 km x 1.25 km, frictional-cohesive cover, base v. low friction. Not a lot more to say really, total of 2000 m contraction in this movie, so only 10% shortening. Deliberately so, I wanted to see maybe just a simple forward propagation of some thrust faults. This (rather speedy) movie shows the (longer than 1st post) evolution at 2 m intervals (after initial equilibration) firstly expressed as rather ordinary geometry and secondly expressed in terms of incremental max. shear strain calculated at each element over the previous 2 m of displacement. (slow it down to speed x0.5 if it seems too fast)

download movie here

]]><![CDATA[Boring can be surprisingly interesting]]>https://hardygeoscience.substack.com/p/boring-can-be-surprisingly-interestinghttps://hardygeoscience.substack.com/p/boring-can-be-surprisingly-interestingWed, 28 May 2025 20:40:38 GMTSometimes a rather mundane, seemingly simple, numerical model can tell you quite a lot if you just stop to really examine it in some detail. I present my case in point: a 2D discrete element model of simple horizontal contraction, section initially 5 km x 1.25 km, frictional-cohesive cover, base v. low friction. Not a lot more to say really, total of 500 m contraction in this movie, so only 10% shortening. Deliberately so, I wanted to see maybe just a simple forward propagation of 1 or 2 faults. The movie shows the evolution at 2 m intervals (after initial equilibration) firstly expressed as rather dull geometry and secondly expressed in terms of incremental max. shear strain calculated at each element over the previous 2 m of displacement. I rest my case: sometimes trawling through those incremental results and doing something with them pays off...enjoy!

Download the movie here

Read more

]]><![CDATA[Geostorage Insights]]>https://hardygeoscience.substack.com/p/geostorage-insightshttps://hardygeoscience.substack.com/p/geostorage-insightsWed, 28 May 2025 09:26:17 GMTIt can be lonely world out there as a Geoscientist on Substack…… so why not follow Conor for some expert, practitioner, insights into storage in geological materials, a hot topic these days, and one where I hope that numerical modelling (of various flavours) can play a key role!

Geostorage Insights
Thoughts on using the ground under our feet to store carbon dioxide and heal our planet.
By Conor O'Sullivan PhD
]]><![CDATA[Localisation, shear, faults and visualisation]]>https://hardygeoscience.substack.com/p/localisation-shear-faults-and-visualisationhttps://hardygeoscience.substack.com/p/localisation-shear-faults-and-visualisationMon, 26 May 2025 18:43:36 GMTI am a firm believer in calculating as much as I possibly can (and displaying it) as I am running my 2D (and to an extent 3D) discrete element codes. This allows so much more “computational steering” than the traditional post-processing approach. Obviously, for simple attributes such as displacement magnitude and components this was pretty trivial. However, for stress and strain it required quite a bit of blood, sweat and tears to get it firstly correct and secondly fast (enough) when doing it “on the fly” along with normal force, velocity calculations etc. But the result is rather pleasing. I take a bit of a 2-pronged approach - display as you go along using my code’s built in graphics and for slicker movies/animations/visualisations use something that is designed to do just that. I almost exclusively (these days) run my codes on Windows 10/11 for many different reasons. However, I really couldn’t find an easy-to-use piece of visualisation software on Windows - yes I know about ParaView and pyvista but they didn’t feel right to me (a very personal thing I know). Anyway as a result I “finish off” my more polished results/animations using ImageTank on the Mac, it is a wonderful, elegant piece of software. Look it up! Here is a movie of localisation in part of an evolving frictional-cohesive thrust wedge (original thickness 2.5km), showing some of the detail, that with a bit of work in your code, you can display to the full. Here I am just showing the total Max. Shear strain as it accrues in this part of the 2D model.

]]><![CDATA[Get your faults in order]]>https://hardygeoscience.substack.com/p/get-your-faults-in-orderhttps://hardygeoscience.substack.com/p/get-your-faults-in-orderMon, 26 May 2025 09:58:35 GMTBack to my 2D discrete element code and testing different styles of boundary conditions, bigger, longer and more shortening.... here we have a frictional-cohesive thrust wedge formed by pulling the base and right-wall towards a static left-wall buttress (in contrast to pushing the left-wall towards the right). Internal coefficient of friction is higher than basal, resolution about 20m, cover initially 2.5 km thick, final shortening 8.5 km. First part of movie is just the geometric evolution and second part shows how the total max. shear strain accumulates. I run these tests/examples when any machine I have is idle as I think that no test/experiment is a waste of time, and those machines might be used to their max. Here I think this time was certainly not wasted, as the movie actually shows up some rather interesting shear strain/faulting patterns and evolution through time, with all sorts of relationships preserved in the final state...the "window" shows approx. the left-hand part of the model in order to highlight the interesting detail

download movie here

]]><![CDATA[Change your reference frame...]]>https://hardygeoscience.substack.com/p/change-your-reference-framehttps://hardygeoscience.substack.com/p/change-your-reference-frameSun, 25 May 2025 13:42:12 GMTBack to my 2D discrete element code and testing different styles of boundary conditions.... here we have a thrust wedge formed by pulling the base and right-wall towards a static left-wall buttress (in contrast to pushing the left-wall towards the right). Internal and basal coefficients of friction the same, resolution about 20m, cover initially 2.5 km thick, final shortening 6.5 km. First part of movie is just the geometric evolution and second part shows how the total max. shear strain accumulates. (ps. the right wall really is fixed to the base, the base to the right of it is made up of a series of "trailing" elements...)

download the movie

Read more

]]><![CDATA[I called this movie "punchy"]]>https://hardygeoscience.substack.com/p/i-called-this-movie-punchyhttps://hardygeoscience.substack.com/p/i-called-this-movie-punchyThu, 22 May 2025 20:22:03 GMTdownload the full movie here

This is one of my early attempts at animating a 2D discrete element simulation of a kind of indenter deforming an overlying, layered, cover. The lower part of the cover is frictionless, and basically “flows”, whilst the top part is frictional-cohesive. This was a kind of “proof of concept” test, but I feel that it can actually be quite useful for teaching or training. The other aspect that I feel should be brought home is displaying the manner in which the Max. shear strain accrues through time, I calculate all strains etc as my simulations are running and they can be displayed that way from the start; the calculations are not trivial but are easily incorporated and calculated as the code is running rather than post-processing…. much more opportunity to “computationally steer”. Anyway, in the archive it goes.

]]><![CDATA[Different points of view...]]>https://hardygeoscience.substack.com/p/different-points-of-viewhttps://hardygeoscience.substack.com/p/different-points-of-viewWed, 21 May 2025 13:34:13 GMTThis animation/movie is not for those of you who get dizzy easily Part II, variant on a theme... anyway, in my kind of back to front mindset, I got to thinking wouldn't it be interesting if, instead of viewing a complete 3D discrete element model in all its glory, we looked at a subset of sections through the data, and their evolution in time. Sort of join the structures in 3D. So here are 2 transport parallel and one depth sections through my complete 3D volume through time, with a bit of rotation of viewpoint thrown in. I guess for completeness you should look at my original (-2) post first. "enjoy", I think that the depth section evolution shown at the same time as the "traditional" sections works rather well...

download the movie here

Read more

]]><![CDATA[Sometimes you just have to look at things differently]]>https://hardygeoscience.substack.com/p/sometimes-you-just-have-to-look-athttps://hardygeoscience.substack.com/p/sometimes-you-just-have-to-look-atTue, 20 May 2025 20:39:19 GMTThis animation/movie is not for those of you who get dizzy easily... anyway, in my kind of back to front mindset, I got to thinking wouldn't it be interesting if, instead of viewing a complete 3D discrete element model in all its glory, we looked at a subset of sections through the data, and their evolution in time? Sort of join the structures in 3D. So here are 2 transport parallel and one orthogonal sections through my complete 3D volume through time, with a bit of rotation of viewpoint thrown in. I guess for completeness you should look at my penultimate post first. "enjoy"

you can download the movie from here

Read more

]]><![CDATA[Thrust Wedge/Lateral Ramp - full movie]]>https://hardygeoscience.substack.com/p/lateral-ramp-full-moviehttps://hardygeoscience.substack.com/p/lateral-ramp-full-movieTue, 20 May 2025 15:19:49 GMTAs promised I will, from time to time, add movies here to build up a teaching/training resource. This movie is of a 3D discrete-element simulation of a thrust wedge developing via lateral contraction of the end-wall, but with a step-jump in displacement. Original model size is 2.9km x 2.9km x 400m, so this is all very high-level stuff, but it illustrates several features nicely. Cover is frictional-cohesive, resolution c. 10m, final endwall displacements are 500 and 1000m. Movie can be downloaded via this OneDrive link

https://icreacat-my.sharepoint.com/:v:/g/personal/stuart_hardy_icrea_cat/EZgHqKez2tROm42MjsBRI_YBRnYuiyIr669nbyTN3z1eiw?e=8qey8a

Read more

]]><![CDATA[Oblique Kinematic Waves]]>https://hardygeoscience.substack.com/p/oblique-kinematic-waveshttps://hardygeoscience.substack.com/p/oblique-kinematic-wavesTue, 20 May 2025 11:36:19 GMTJust a post of a rather nice image from a contractional experiment with a lateral ramp, here the model is shown at approx. 950m displacement and all elements are coloured (red-blue) by their total displacement orthogonal to “transport direction”, makes you think about 2D restoration and/or forward modelling in such a situation

…

Hardy Geoscience Substack is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

]]><![CDATA[Visualization of numerical modelling results should be easy but it often isn't in my experience...what's your take?]]>https://hardygeoscience.substack.com/p/visualization-of-numerical-modellinghttps://hardygeoscience.substack.com/p/visualization-of-numerical-modellingMon, 19 May 2025 19:53:47 GMTI suppose that this is a general kind of question… we spend a lot of time getting algorithms correct, stable, optimising codes so that runtimes are tractable etc.. but often (and this is partly from reviewing journal papers) I think that the presentation/visualization of our results can let us down. I wonder why this is, it might be that some of our data is inherently difficult to convey, it might be that easily available/free software is often not very good at displaying geological-type data/problems, or that some commercial softwares just don’t deliver. I don’t know the answer, but I’d like opinions. In my case, with my codes, writing my own front-end for the 2D stuff was within my abilities but 3D is beyond me - so I ended up spending a lot of time (on various platforms) trying to “fit” my data to various products. In the end I am using ImageTank on a Mac to visualize results from 2D and 3D discrete element model results that are either generated on Linux or Windows… it works very well… what is your experience?

]]>