[CIG-SHORT] Fault direction in 3D fault

[Brad Aagaard]

Matt,

In 3-D the long-used convention for defining the two shear directions 
(along-strike, and up-dip) is based on the fault normal and the
"up" direction. The convention breaks down for a horizontal fault 
(vertical normal direction). The underlying assumption is that faults in 
nature are not horizontal, and this holds up quite well. I have never 
seen a convention for horizontal faults, so we don't have an easy fall back.

We can try a tolerance of 1.0e-6 and see how well it works.

Brad


On 05/05/2014 01:15 PM, Matthew Knepley wrote:
> On Mon, May 5, 2014 at 2:59 PM, Brad Aagaard <baagaard at usgs.gov> wrote:
>
>> Charles,
>>
>> We use the same orientation function for Neumann BC and fault BC. The
>> issue is whether we use a strict mag == 0.0 check for parallel directions
>> or a mag <= tolerance check. Currently, we use mag == 0.0. If we use mag <=
>> tolerance, then we need a robust value for tolerance. That is, we want it
>> to catch cases that are problematic without triggering errors in cases that
>> are okay.
>
>
> Since faults can now be pretty twisty, does it make sense to have just one
> coordinate system? Could we have
> one for each element if we wanted?
>
> I think its fair to warn the user about a very ill-conditioned coordinate
> transform. I think there is no problem with
> a fairly large tolerance, 1e-6, since then you really should not be doing
> that transform.
>
>     Matt
>
>
>>
>> Brad
>>
>>
>>
>> On 05/05/2014 12:34 PM, Charles Williams wrote:
>>
>>> If I’m not mistaken, when we’re applying Neumann BC, we actually exit
>>> with an error when up_dir and normal_dir are aligned.  This obviously
>>> doesn’t happen on faults, though.
>>>
>>> Charles
>>>
>>>
>>> On 6/05/2014, at 3:22 am, Brad Aagaard <baagaard at usgs.gov> wrote:
>>>
>>>   On 05/05/2014 05:30 AM, Matthew Knepley wrote:
>>>>
>>>>> On Sun, May 4, 2014 at 8:58 PM, Charles Williams <willic3 at gmail.com>
>>>>> wrote:
>>>>>
>>>>>   It appears that the problem is that your fault at the bottom of the
>>>>>> slab
>>>>>> is completely horizontal over much of its extent.  This causes
>>>>>> problems for
>>>>>> PyLith when the default up_dir [0,0,1] is used.  The way to diagnose
>>>>>> this
>>>>>> problem is to have PyLith output the orientation information for the
>>>>>> faults, as Brad suggested previously.  When you plot the strike_dir as
>>>>>> vectors, you will see it varies quite a bit for the horizontal portion
>>>>>> of
>>>>>> the fault, because the direction is undefined.  The solution is to
>>>>>> choose a
>>>>>> vector that does not align with the fault normal vector.  I chose a
>>>>>> vector
>>>>>> that points slightly to the west.  When this is done, the various fault
>>>>>> slip directions are well-defined.
>>>>>>
>>>>>>
>>>>> Great find, Charles. Is there a way we can automate this check?
>>>>>
>>>>
>>>> The difficulty is defining a good tolerance. Right now, we compute the
>>>> cross product of the up direction and the fault normal to differentiate
>>>> between along-strike shear and up-dip shear. If the magnitude of the cross
>>>> product is strictly zero, we give an error. We could make it epsilon, but
>>>> we may have to play around with the order of magnitude so that it is not
>>>> overly aggressive.
>>>>
>>>> Brad
>>>>
>>>> _______________________________________________
>>>> CIG-SHORT mailing list
>>>> CIG-SHORT at geodynamics.org
>>>> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
>>>>
>>>
>>> Charles A. Williams
>>> Scientist
>>> GNS Science
>>> 1 Fairway Drive, Avalon
>>> PO Box 30368
>>> Lower Hutt  5040
>>> New Zealand
>>> ph (office): 0064-4570-4566
>>> fax (office): 0064-4570-4600
>>> C.Williams at gns.cri.nz
>>>
>>> _______________________________________________
>>> CIG-SHORT mailing list
>>> CIG-SHORT at geodynamics.org
>>> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
>>>
>>>
>>
>
>