INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN FLUIDS
Int. J. Numer. Meth. Fluids (2015)
Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/fld.4202
Local maximum principle satisfying high-order
non-oscillatory schemes
Ritesh Kumar Dubey
1,
*
,†
, Biswarup Biswas
1
and Vikas Gupta
2
1
Research Institute, SRM University, Tamilnadu, India
2
LNMIIT, Jaipur, Rajasthan, India
SUMMARY
The main contribution of this work is to classify the solution region including data extrema for which high-
order non-oscillatory approximation can be achieved. It is performed in the framework of local maximum
principle (LMP) and non-conservative formulation. The representative uniformly second-order accurate
schemes are converted in to their non-conservative form using the ratio of consecutive gradients. Using
the local maximum principle, these non-conservative schemes are analyzed for their non-linear LMP/total
variation diminishing stability bounds which classify the solution region where high-order accuracy can be
achieved. Based on the bounds, second-order accurate hybrid numerical schemes are constructed using a
shock detector. The presented numerical results show that these hybrid schemes preserve high accuracy at
non-sonic extrema without exhibiting any induced local oscillations or clipping error. Copyright © 2015
John Wiley & Sons, Ltd.
Received 22 July 2015; Revised 29 September 2015; Accepted 7 November 2015
KEY WORDS: hyperbolic conservation laws; smoothness parameter; non-sonic critical point; total variation
stability; finite difference schemes
1. INTRODUCTION
We consider the 1D scalar conservation law associated with the conserved variable u.x;t/,
@
@t
u.x;t/ C
@
@x
f .u.x; t // D 0; .x; t/ 2 R R
C
u.x; 0/ D u
0
.x/
(1)
where f .u/ is a nonlinear flux function. The numerical approximation for the solution of (1) is
carried out by the discretization of the spatial and temporal spaces into N equispaced cells I
i
D
Œx
i
1
2
;x
i C
1
2
; i D 0; 1; : : : N 1 of length x and M equispaced intervals
t
n
;t
nC1
; n D
0; 1; : : : ; M 1 of length t , respectively. Let x
i
D ix and t
n
D nt denote the cell center of
cell I
i
and the n
th
time level, respectively, then a conservative numerical approximation for (1) can
be defined by
u
nC1
i
D u
n
i
F
n
i C
1
2
F
n
i
1
2
; D
t
x
: (2)
where u
n
i
D u.x
i
;t
n
/ and F
n
i ˙
1
2
is the numerical flux function defined at the cell interface x
i ˙
1
2
at
time level n. The characteristics speed a.u/ D
@f .u/
@u
associated with (1) can be approximated as
*Correspondence to: Ritesh Kumar Dubey, Research Institute, SRM University, Tamilnadu, India.
†
E-mail: riteshkd@gmail.com
‡
The carried work is financially supported by SERB-DST, India project SR/FTP/MS-015/2011.
Copyright © 2015 John Wiley & Sons, Ltd.