12.07.2015 Views

Comparison between 1D and 2D models to analyze the dam break

Comparison between 1D and 2D models to analyze the dam break

Comparison between 1D and 2D models to analyze the dam break

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Comparison</strong> <strong>between</strong> <strong>1D</strong> <strong>and</strong> <strong>2D</strong> <strong>models</strong> <strong>to</strong> <strong>analyze</strong> <strong>the</strong> <strong>dam</strong> <strong>break</strong> waveFigure 1. Representative vertical section through <strong>the</strong> fluiddomain showing <strong>the</strong> bed of a lake <strong>and</strong> <strong>the</strong> water surface.2.3 Two-dimensional Shallow Water equations.Depth averaging of <strong>the</strong> free surface flow equationsunder <strong>the</strong> shallow water hypo<strong>the</strong>sis leads <strong>to</strong> a commonversion of <strong>the</strong> two dimensional SWE, wich inconservative form is:∂z∂(zu)∂(zv)+ + = 0∂t∂x∂y∂(zu)∂(zu+∂t∂x2) ∂(zuv)+∂y(5)∂z∂zf (6)− f ⋅ zv + g ⋅ z ⋅ + g ⋅ z ⋅∂x∂x22∂ ( zu)∂ ( zu)+ g ⋅ z ⋅ Sfx− E ⋅ − E ⋅ = 022∂x∂y∂(zv)∂(zuv)∂(zv+ +∂t∂x∂y+ g ⋅ z ⋅ Sf2)∂z∂zf (7)+ f ⋅ zu + g ⋅ z ⋅ + g ⋅ z ⋅∂y∂yy22∂ ( zv)∂ ( zv)− E ⋅ − E ⋅ = 022∂x∂yThe equations above are valid under <strong>the</strong> followingassumptions:1. The fluid is well-mixed vertically with a hydrostaticpressure gradient.2. The density of <strong>the</strong> fluid is constant.3. We study water waves of long wave lengths.4. The viscosity term is negligible.Where z is <strong>the</strong> depth of water, zu <strong>and</strong> zv are unitdischarges along <strong>the</strong> co-ordinate directions. Inaddition, u <strong>and</strong> v are <strong>the</strong> velocities in <strong>the</strong> x <strong>and</strong> ydirections respectively, g is <strong>the</strong> acceleration due <strong>to</strong>gravity <strong>and</strong> Sf x <strong>and</strong> Sf y are <strong>the</strong> friction terms in <strong>the</strong> x<strong>and</strong> y directions respectively. The turbulent dissipationterms (E) <strong>and</strong> Coriolis effect (f) are present in <strong>the</strong>equations, but not <strong>the</strong> wind effect, which is notsignificant in <strong>the</strong> usual valleys where <strong>the</strong> <strong>dam</strong>-<strong>break</strong>occurs.2.3.1 The friction term.Many two-dimensional depth-averaged <strong>models</strong> includeonly friction at <strong>the</strong> bot<strong>to</strong>m. Specifically, <strong>models</strong> thatassume vertical channel side-walls <strong>and</strong> use free-slipboundary conditions do not account for <strong>the</strong> friction at<strong>the</strong> walls. Neglecting this effect, open channel flowwould likely show a marked variation in water depthfrom <strong>the</strong> one measured experimentally.Evaluation of <strong>the</strong> friction slope that quantifies <strong>the</strong>energy loss at <strong>the</strong> side-walls of <strong>the</strong> channel is required.It is common <strong>to</strong> use a uniform flow law, Manning orChezy formulae, <strong>to</strong> calculate this term. However, <strong>the</strong>selaws were developed for one-dimensional flow <strong>and</strong>must be extended <strong>and</strong> properly incorporated <strong>to</strong> twodimensionalequations.Manning’s formula used in one-dimensional shallowwater<strong>models</strong> is expressed in <strong>the</strong> formSfn2A⋅ Q ⋅ Q= (8)2⋅ R4 / 3Where n is <strong>the</strong> Manning coefficient <strong>and</strong> R = A/P is <strong>the</strong>hydraulic radius, which depends on <strong>the</strong> wettedperimeter P. For a rectangular channel P = b + 2z<strong>and</strong> for an irregular basin P = b.A distinction must be made <strong>between</strong> an arbitrarychannel cross-section <strong>and</strong> a cross-section with verticalwalls (fig. 2). Most two-dimensional <strong>models</strong> assumevertical walls <strong>and</strong> free-slip wall boundary conditionsfor rectangular channels, while <strong>the</strong> equations used <strong>to</strong>compute <strong>the</strong> friction term are based on irregular crosssections.To correct this inconsistency, <strong>the</strong> frictionslope equation has been modified here <strong>to</strong> reflect <strong>the</strong>vertical side-wall assumption. Basically, <strong>the</strong>modification ensures that <strong>the</strong> entire wetted perimeter(bot<strong>to</strong>m width <strong>and</strong> side-walls) is accounted for.For an arbitrary cross-section, <strong>the</strong> wetted perimeter ina cell is equal <strong>to</strong> <strong>the</strong> bot<strong>to</strong>m width b (fig. 2), so that <strong>the</strong>hydraulic radius adopts <strong>the</strong> form R = z, <strong>the</strong> waterdepth. In this case, <strong>the</strong> two-dimensional friction termsare written in <strong>the</strong> form (Brufau <strong>and</strong> García-Navarro2000):22n ⋅ u ⋅ u +Sf x 4/ 3zv2= (9)22n ⋅ v ⋅ u +Sf y 4 / 3zv2= (10)5

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!