29 i f u p1 pu f u f u 10210b 10210c split

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ng 26]. Van Leer 27] found an improvement that provided better performance near sonic and stagnation points of the ow. The split uxes are evaluated by upwind techniques. Thus, at an interface x = xj , f + is evaluated using Uj (xj t) and f ; is evaluated using Uj+1(xj t). Calculating uxes based on the solution of Riemann problems is another popular way of specifying numerical uxes for vector systems. To this end, let w(x=t uL uR ) be the solution of a Riemann problem for (10.1.1a) with the peicewise-constant initial data (10.1.25). The solution of a Riemann problem \breaking" at (xj tn) would be w((x ; xj )=(t ; tn) Uj (xj tn) Uj (xj+1 tn)). Using this, we would calculate the numerical ux at (xj t), t > tn , as F(Uj (xj tn) Uj+1(xj tn)) = f (w(0 Uj (xj tn) Uj+1(xj tn)): (10.2.11) 32 Hyperbolic Problems Example 10.2.4. Let us calculate the numerical ux based on the solution of a Riemann problem for Burgers' equation (10.1.16). Using the results of Example 10.1.8) we know that the solution of the appropriate Riemann problem is 8 if Uj Uj+1 > 0 > Uj > >U > j +1 if Uj Uj +1 < 0 < if Uj < 0 Uj+1 > 0 w(0 Uj Uj+1) = > 0 : > Uj if Uj > 0 Uj+1 < 0 (Uj + Uj+1)=2 > 0 > > :U j +1 if Uj > 0 Uj +1 < 0 (Uj + Uj +1 )=2 < 0 (The arguments of Uj and Uj+1 are all (xj tn). These have been omitted for clarity.) With f (u) = u2=2 for Burgers' equation, we nd the numerical ux 82 > Uj =2 if Uj Uj +1 > 0 >2 > U =2 if U U < 0 > j +1 < j j +1 if Uj < 0 Uj+1 > 0 F (Uj Uj+1) = > 0 : > Uj2 =2 if Uj > 0 Uj +1 < 0 (Uj + Uj +1 )=2 > 0 > >2 : U =2 if U > 0 U < 0 (U + U )=2 < 0 j j +1 j j +1 j +1 Letting u+ = max(u 0) u; = min(u 0) we can write the numerical ux more concisely as F (Uj Uj+1) = max (Uj+)2 =2 (Uj; )2=2]: +1 When used with a piecewise-constant basis and forward Euler time integration, the resulting discontinuous Galerkin scheme is identical to Godunov's nite di erence scheme 18]. This was the rst di erence sche...
View Full Document

This document was uploaded on 03/16/2014 for the course CSCI 6860 at Rensselaer Polytechnic Institute.

Ask a homework question - tutors are online