Rob S 0 Report post Posted May 15, 2009 I am setting up a large regional scale TUFLOW model which has a dam spill way at the the downstream boundary. I have modelled the spillway with a HQ type 2d_bc line along the entire dam embankment. For some reason this boundary is very unstable and is actually pushing water back into the dam at the begining of the simulation. Stability is only improved by reducing the time steps right down to 0.5 seconds and even then it isn't great. The grid size is 10m so a half second time step seems smaller than it should need to be and it is causing model run times to be excessive. I know that HQ boundaries are not the most stable boundary condition but does anyone have an idea on how this could be improved. Note I have tried applying constant bathymetry and 'n' values along the boundary and immediatly upstream but this didn't help. Anyone got any ideas on how this could be improved? Cheers, Rob Quote Share this post Link to post Share on other sites
cardno 0 Report post Posted May 15, 2009 What slope have you used in the model for your HQ boundary? Our experience has been that very low slopes produce the effect you are describing. I suggest you use a steeper slope, to get a stable boundary. Quote Share this post Link to post Share on other sites
Rob S 0 Report post Posted May 15, 2009 What slope have you used in the model for your HQ boundary? Our experience has been that very low slopes produce the effect you are describing. I suggest you use a steeper slope, to get a stable boundary. No slope adopted. I input the QH relationship directly based on the dams weir and operating rules. good idea though... Cheers, Rob Quote Share this post Link to post Share on other sites
RhysHJ 0 Report post Posted May 15, 2009 Hi Rob, I've had the same experience as cardno for the automatic slope HQ boundary that steepening the slope will improve the stability (although check your boundary is far enough away from you area of interest by doing some sensitivity checks on this). Given that it's a spillway and is likely to have critical flow over the crest, a steep slope boundary is probably a reasonable assumption. If you want to specify your HQ relationship, make sure that the values at low flow make sense with your topography (ie the first entry in the table, with Q=0, should have an H value similar or even slightly lower than your spillway crest zpts, to avoid the flowback behaviour you described). Also, you can get a similar effect to steepening the automatic slope by reducing the H values in your rating curve by a constant amount. Hope that helps, Rhys Quote Share this post Link to post Share on other sites
paul_ollett 0 Report post Posted May 18, 2009 What about trying to remove the dam&spillway from the 2D domain (DEM) altogether? Then extend the HQ relationship downwards, eg, RL H Q 0.0 -100.0 0 100.0 0.0 0 100.01 0.01 1 100.1 0.1 20 100.2 0.2 60 etc Just wondering whether the checks/calcs for sub/super critical flow in the 2D domain may be at conflict with the HQ rating curve? If you lower the grid cell values so that flow regimes remain constant at the dam, then the HQ relation may function without interruption. Just a thought... Perhaps even shift your boundary downstream of the spillway, and let TUFLOW calculate the natural HQ at the dam. I thought TUFLOW uses a weir equation for cells under critical flow, and that you can locally adjust the weir coefficient in the 2d domain. Therefore you may be able to calibrate the TUFLOW 2D domain to first-principle estimates. There may be too much stuff going on at your d/s boundary, ie, the spillway cell dynamics + the HQ relationship. Trying to separate these may help. Cheers Paul. Quote Share this post Link to post Share on other sites
