[cig-commits] r15323 - short/3D/PyLith/trunk/libsrc/faults
brad at geodynamics.org
brad at geodynamics.org
Wed Jun 17 12:47:45 PDT 2009
Author: brad
Date: 2009-06-17 12:47:44 -0700 (Wed, 17 Jun 2009)
New Revision: 15323
Modified:
short/3D/PyLith/trunk/libsrc/faults/FaultCohesiveKin.hh
Log:
Added notes on fault implementation.
Modified: short/3D/PyLith/trunk/libsrc/faults/FaultCohesiveKin.hh
===================================================================
--- short/3D/PyLith/trunk/libsrc/faults/FaultCohesiveKin.hh 2009-06-17 19:11:30 UTC (rev 15322)
+++ short/3D/PyLith/trunk/libsrc/faults/FaultCohesiveKin.hh 2009-06-17 19:47:44 UTC (rev 15323)
@@ -21,16 +21,38 @@
* fault.
*
* The ordering of vertices in a cohesive cell is the vertices on the
- * one side of the fault, the corresponding entries on the other side
- * of the fault, and then the corresponding constraint vertices.
+ * "negative" side of the fault, the corresponding entries on the
+ * "positive" side of the fault, and then the corresponding constraint
+ * vertices.
*
- * [ K C^T ] [ U ] = [ Fe ]
- * [ C 0 ] [ Fi ] = [ D ]
+ * The system without Lagrange multipliers is
*
- * where K is the stiffness matrix, C is the matrix of Lagrange
- * constraints, U is the displacement field, Fe is the vector of
- * external forces, Fi is the vector of Lagrange multipers (forces), D
- * is the fault slip.
+ * [A(t+dt)]{u(t+dt)} = {b(t+dt)}
+ *
+ * With Lagrange multipliers this system becomes
+ *
+ * [A(t+dt) C^T ]{ u(t+dt) } = {b(t+dt)}
+ * [ C 0 ]{ L(t+dt) } {D(t+dt)}
+ *
+ * where C is the matrix of Lagrange constraints, L is the vector of
+ * Lagrange multiplies (internal forces in this case), and D is the
+ * fault slip.
+ *
+ * We solve for the increment in the displacement field, so we rewrite
+ * the system as
+ *
+ * [A(t+dt) C^T ]{ du(t) } = {b(t+dt)} - [A(t+dt) C^T ]{ u(t) }
+ * [ C 0 ]{ dL(t) } {D(t+dt)} [ C 0 ]{ L(t) }
+ *
+ * We form the residual as
+ *
+ * {r(t+dt)} = {b(t+dt)} - [A(t+dt) C^T ]{ u(t)+du(t) }
+ * {D(t+dt)} [ C 0 ]{ L(t)+dL(t) }
+ *
+ * The term D does not involve integration over cohesive cells. We
+ * integrate the Lagrange multiplier terms over the cohesive cells
+ * because this introduces weighting of the orientation of the fault
+ * for the direction of slip at the vertices of the cohesive cells.
*/
#if !defined(pylith_faults_faultcohesivekin_hh)
More information about the CIG-COMMITS
mailing list