THE COMPUTATIONAL HYDRAULICS & HYDROLOGY

International Academic Hub for River Basin Research

2010-12-01T03:10:00.000-08:00

Poster Session

The best poster award was won by Ratih Indri Hapsari, Ph.D (Candidate). Congratulations!

details see http://www.g coe.yamanashi.ac.jp/e/event/20 10/20101129-30symp.htm

Interested in PhD. Scholarship @GCOE-UY, 2011?
see .. http://www.gcoe.yamanashi.ac.jp/e/event/2010/interview2011.html

IWRM- Training

2010-02-21T16:24:00.001-08:00

Phase-1, January 2010

Phase-2, April 2010

Flood Fighting Drill in Japan

2009-09-09T05:36:00.001-07:00

Flood Fighting (the fight against flooding).

Suikou deployment team and Opening ceremony.

Compared with among other types of disaster, flood often occurs anywhere but timely forecasted. Thus, it is available for giving an early flood warning or making flood fighting arrangement.

In Japan, by the law it's required to conduct the flood drills annually in order to provide understanding of the type of flood fighting activities by involving local residents, private companies, local disaster managers and students.

The benefits from this activity are raising public awareness on flood fighting (i.e. enhancement of public preparedness) and implementing school based flood disaster education (i.e. capacity development). Public awareness is the basis for flood fighting and it is essential for public to be proactive and look for the problems and causes before floods happens. Meanwhile, within the education sector it urges to create curriculum on disaster management to educate school children and students in the flood fighting techniques.

Flood drills (banks protection in Fujikawa).

Rescue and evacuation demonstration.

Experiencing of making sandbags.

山梨発　水と緑を考える大学・地域・産業連携国際シンポジウム

2009-09-05T00:24:00.001-07:00

STOP PRESS !!

2009-08-25T18:54:00.000-07:00

Day of Defense - Completing a Doctoral Dissertation

2009-08-16T20:50:00.000-07:00

Two weeks ago, She turned to sweat although she could handle everything elegantly (well done, Congrats !! ).

Absorbing (non-reflecting) boundary condition

2009-07-31T19:23:00.001-07:00

An absorbing BCs has been formulated and applied to SUIKOU2D in conjunction with MacCormack and TVD schemes. The inclusion of the boundaries is very important in the successful application of any numerical techniques. Hyperbolic equations are particularly sensitive because errors introduced at the boundaries are propagated and reflected throughout the grids. This, in many cases, may results in instability or incorrect flow values.

Numerical modeling of inundation flow often involves computational domains that are very large. Discretizing the entire domain and solving the flow equations turn out to be a tedious jobs in term of times and memory. Therefore, the physical domain has been truncated by introducing a pseudo boundary. Here, the use of pseudo boundary as open boundary aids in reducing the size of computational domain and and emphasizes the interest area only.

There are two possible flow conditions may occur in the pseudo boundary; sub-critical or super-critical flows accompanied with shocks where it needs special treatment on each. An ideal pseudo boundary condition should meet the following criteria:
(i) it should be compatible with physical conditions of the nature.
(ii) it should not degrade/defect the numerical solutions.
(iii) it should absorb the outgoing waves to avoid any reflections.

Several established methods are available for pseudo boundary. The simplest approach is by implementing two ways, including
1. extrapolation of flow variables from the interior node (Froude >=1);
2. solved by MoC (Froude <1).

Test Case

A 10m height of water in reservoir in upstream part of computational domain. It is assumed some buildings in the downstream part as represented by random blocks. Initially, it is dry in the downstream. Suddenly the reservoir partially bursts and inundates downstream part.

The simulations are carried out on the basis of;
a. The first case, all of outer BCs of the computational domain is considered as wall.
There is no incoming/outcoming flux through walls. The shocks will be reflected as they hit the walls.
b. Second, the downstream part is truncated by pseudo BC.
All incoming/outcoming shocks will be absorbed as represented flows pass through the downstream wall without any reflections.

The animations of water surface during 1200 seconds after breach are as shown below.
(click on the image to run animation)

a. case no.1 (walls BC).

b. case no 2. (non reflecting BC).

c. Water depth along x-axis at center of breach.

Obviously seen, the lowering water depth in reservoir occurs in case no. 2 since there is out flowing flux through pseudo BC, whilst no.1 is reflected back.

to be continued....

Making Space for Water

2009-05-19T13:48:00.000-07:00

Afflux and Backwater

2009-04-21T04:30:00.001-07:00

As water engineers we often know terms of 'backwater' rather than 'afflux'. Why?. Don't be confused!
In hydrology, afflux is defined as a rise in the water level immediately upstream of and due to a natural or artificial obstruction. Backwater is a consequence of afflux, in other words the afflux brings 'backwater effect'.

TVD Scheme for SUIKOU-2D v2.0

2009-04-10T11:07:00.001-07:00

TVD stands for Total Variation Diminishing, a scheme to eliminate the numerical oscillations or commonly occurred in the explicit numerical methods when they are approaching the high gradient regions. The mathematical expressions of TVD can be reviewed here. I am not going to explain detail about TVD but systematically outlining its implementation to the current version of SUIKOU-2D.
As in previous post, SUIKOU-2D v1.0 was developed to solve full set of Shallow Water Equations (SWEs) using the explicit MacCorm ack Method . Artificial viscosity (AV) scheme, a control of numerical dissipations to handle shock waves was added. AV is acting as a fashioned approach where the viscosity/turbulence parameters are given as fixed values to all computational steps. It takes series of iterations to get a convergence result. Several improvement efforts has been sought and now it's being altered by TVD scheme in the latest codes of SUIKO-2D. The convergence is quite faster compared to previous one. The general computation blocks are drawn as below figure.

In SUIKOU v2.0 a five-points symmetric TVD term has been used to remove the numerical oscillations and calculated after the corrector step.

where r( ) is scalar product of two vector Ui+1/2 and Ui-1/2 within the point brackets below:

in which U=(h,u,v)

these scalar products can be rewritten in detail as

The function G( ) is defined as

while the flux limiter is:

and C variable is a function of courant number, Cr

The imaginary grid points of above variables are described as below:

Back to AV, it's obviously seen that it is likely a simple model of TVD. The corrections are only made on the basis of water depth, h, whilst TVD deploys all flow variables (u,v,h) to furnish final results.

Numerical treatment of the source terms in SUIKOU-2D v2.0

2009-04-08T21:19:00.000-07:00

The history of the numerical treatment of hydrodynamic model is extremely long and has been provided a lot of variance in the numerical approaches. Commonly, the flooding models are simplified by elimination the source terms. In consequences they are limited to simulate bore propagation or flash flood occurred in dambreak events.

Using TVD-MacCORMACK, a two-stage scheme gives a benefit where source terms can be easily treated and preserves that whole scheme has second order accuracy in both time and space. However, if the bottom topography shows strong variation it is required a special treatment of the source terms to eliminate or reduce the artificial numerical error caused by adding TVD corrections to the MacCORMACK. The treatment of source terms related to strong bed slope is adopted here.

to be continued ....

Visualize results with Gnuplot

2009-04-04T07:54:00.001-07:00

This article is briefly showing a hands-on guide to turn the output data of SUIKOU-2D into professional-looking graphs with Gnuplot, a freely distributed plotting tool.

According to FAQ 1.7, Gnuplot is freeware in the sense that you don't have to pay for it. However, it is not freeware in the sense that you would be allowed to distribute a modified version of your gnuplot freely.

The program could be downloaded here. Please read and accept the Copyrights to use it. Basically, Gnuplot is a command line driven plotting tool with extensive supports for many types of plots in either 2D and 3D with capabilities to draw using lines, points, boxes, contours, vector fields, surfaces, and various associated text, but it would be very nice to be able to have the FORTRAN codes to continue execution without fires it up in the command-line mode.

By inserting the below fortran code in SUIKOU-2D.for, it automatically displays interactive plots at the end of running program. Control plots could be managed by user in a separate text file of gnuport.txt, a compilation of selected shell scripts. It only requires "wgnuplot.exe" (1,906KB) that must be copied into folder where SUIKOU-2D.for is otherwise the plots won't be displayed.

Here is the Fortran code with an example of main program to call.
$freeform
program gnuport
call porting_gnuplot ( 'gnuport.txt' )
stop ''
end

subroutine porting_gnuplot(console)
character (len=100) command
character (len=*) console
integer status
integer system
write (command, *) 'wgnuplot ' // trim (console)
status=system(trim(command))
if (status.eq.0) then
write (*,*)
write ( *, '(a)' ) ' Normally end.'
else
write(*,*)
write ( *, '(a)' ) ' Fatal error!'
stop ''
end if
return
end

and a simple example of gnuport.txt
unset key
set title "Visualization with Gnuplot" font "Times-Roman,14"
set size ratio 0.35
set ticslevel 0
set ztics 5
set noxtics
set noytics
set zrange [0:10]
set xlabel "x-grid"
set ylabel "y-grid"
set cbtics 5
set cblabel "water depth (m)" offset character -2,0
set cbrange [0:10]
set palette rgbformulae 3,0,13
splot "y0004.gnu" matrix with pm3d
pause -1
set size ratio 1.35
set view map
splot "y0004.gnu" matrix with pm3d
pause -1
set view map
set contour base
set palette rgbformulae 3,0,13
set cbtics 5
set cbrange [0:10]
unset surface
splot "y0004.gnu" matrix with lines palette
pause -1
a=0.01
plot "u0004.gnu" using 1:2:(a*$3):(a*$4) with vector
pause -1

Screenshots:

It's a simple work to make fruitful life, isn't? :)

Link:
http://gnuplot.sourceforge.net/demo/

Dam burst disaster

2009-03-29T16:11:00.000-07:00

It's a classical word that institution/government reacts aftermaths rather than preventive countermeasures. See the recent disaster of the dam burst, Situ Gintung. The water authority has been immediately issuing responses to take actions for similar cases may occurred.
http://www.malangpostnews.co.id/index.php?option=com_content&view=article&id=690:waspadai-bendungan-karangkates&catid=47:agropolitan&Itemid=75

Surely there is an emergency action plan (Rencana Tindak Darurat-RTD ?) for such hazard due to flash flood caused by dam break for any dams and water structures. Definitely, this plan is including the worst scenarios and the mitigation plan. It is an obligation for most of water projects. Then how it's being implemented. Are there any early warning systems? or just pile up the documents in the office.

Moreover, it's still questionable and debatable on the accuracy of past RTD studies. The rising question is when the study was carried out and what kind approaches they used. If it was done several years or even decades ago, no doubt to revise it.

Take a simple fact. Was RTD being studied with 1D model? If the answer is yes, definitely it has to be updated. But if the answer is no (used advanced approaches) another question is still coming.

What kind of topography data was taken? How was the land use data?
Doing study of flash flood, it's absolutely required high accuracy DEM. Using derived DEM from1:25.000 map is not sufficient enough since more details (roads, housing etc) will be uncovered. As consequences the results are inaccurate. The landuse during RTD studies and current situation is different. It has been altered. That sounds better if it was properly projected within these studies.

Then what shall we do?
Well. As I understand so far, this DEM problem is mostly concerns not only by developing countries but also developed countries. It is very costly to have it and needs more resources to analyze (eg. LiDAR data). Perhaps the aerial photo of the prone areas to inundation is a mid solution. If it's required the ground leveling has to be done at specific area needs to be detailed. Combined with the existing DEM, more accurate topography map could be obtained and a flash flood modeling could be set up. Alternatively, a physical model test with detail villages is also worth to be done.

Despite how much money will be spent on this study and who will do it, those are just about my opinions as water practitioner dealing with such flood risks. Indonesia is still my lovely country.
May God bless us, all victims and their family in this disaster, Amien.

Old articles but still worth readings....

2009-03-29T12:07:00.000-07:00

Almost a year I didn't update this blog since the last writing. I don't know whether I am lazy or busy with my recent jobs but that's the facts!. At the moment, I've been involved in a project with new challenges slightly different with this game. According to the regulations it's strictly prohibited to write results here before being published publicly. I promise then I'll share it here. It's so exciting for hydraulics ...LoL..

By the way, this weekend I tried to check my old backups and do a regular maintenance. Oh lucky me ! I found the old stuffs written almost ten years ago.

Those were class notes used to lecture undergraduate level. It seemed to be so ambitious by pushing the students to master it since it's usually given at graduate level or higher.

Again, whether they did like and enjoy this mathematical game or were really hate to learn it, I believe someday they will recognize how its worth. Particularly when they found some jobs required those skills. Probably some of them are pursuing higher education and got stuck on their research works. Surely! it's not bad to recite it. Those who are in undergraduate level and want to learn this game, it's right time to start it now.

Wish you all success and enjoy your time !
(articles are available upon email request)

Updated ... SUIKOU-2D Ver. 1.1 (April 2008)

2008-04-01T22:35:00.000-07:00

What's new with the updated version??

1. diffusion-dispersion terms, turbulent and dispersion stresses are added in the new version of SUIKOU-2D. The dispersion terms are produced because of the non-uniformities of the fluid flow velocity in the vertical direction during the depth-averaging. The determination of these terms requires knowledge of the information in the secondary flows across the depth which is, however, usually not known a priori. These terms are frequently found to be negligible in cases such as straight channel flows. However, the dispersion terms can be very important, particularly when there are strong secondary flows, such as fluid flows through a complex domain. The diffusion terms are supposed to overcome the elevation depressions or obstacles.

2. a wetting-drying algorithms.

3. integrating TVD (total variation diminishing) and Artificial Viscosity schemes to get better numerical dissipation control. It has been proved to increase the model robustness.

4. better formulation for numerical treatment of Bed Slope Source Terms

The worked example (click image for animation):
1. Dambreak with rectangular barrier
(Currently there is problem on blogger.com to support animation)

2. Multiple blocks

Sediment Transport in Reservoir

2008-02-27T21:23:00.000-08:00

Environmental concerns in lakes and reservoirs due to anthropogenic changes has prompted a need for mathematical modeling tools that can be used as a decision support system for watershed management issues. In order to get a better understanding of flow circulations, the interactions between flow and bottom and the sediment distribution, the hydrodynamic modeling is absolutely required. Here is a summary of an example numerical simulation of reservoir hydrodynamics, the sediment-transport process and its distribution.

The upper Brantas River
The 2002 and 2003 survey provided 17 cross-sections along the upper part of Brantas River and 32 cross-sections along Lesti River. The reaches are 3,645 m 7,746 m from dam axis for Brantas River and Lesti river respectively.

Sediment Transport
By employing GIS-techniques, changes of reservoir volume and bathymetry are determined. The portion of results is given in the below figures.

Compared to the observed data, they seem to be good enough.
Year 2002, the difference = 4.925.454,618 m3 – 4.527.583.411 m3 (observed)
= 397.871,207m3 ~ 8.8% of error
and
year 2003, the difference = 3.698.383,96 m3 – 3,663,658 m3 (observed)
= 34.725,64 m3 ~ 0,95% of error.

NUMERICAL MODELING
1. Grid generation - DEM ( derived from 1:25.000 map)

2. Hydrology Data

Example run #1.
Assuming the reservoir is almost empty in this stage, only 0.1m in depth of water playing as initial condition. The spillway on Dam maintains the water depth of the reservoir at 0.65m. The boundaries are set as a constant water depth in Brantas River at 0.10m and a flood in Lesti River. The flood is rising up to 1.65 m within 1200s and slowly decreasing to 0.25 at 7200s. Total computational time is 10800s.

(click here for an animation with 360s in interval).

be continued with sediment transport modules ...

INUNDATION MODEL FOR HIGHLY URBANIZED AREA

2008-01-24T22:02:00.000-08:00

The flood inundation model for highly urbanized area is different from general flood models due to the increase of the complexity of nature and the internal boundaries. As consequences, the flow is more complex and requires a stable numerical model. Here is a flood inundation model for highly urbanized area (SUIKOU-2D) that meets all those challenges.

Suikou has meaning the flow of water in Japanese. Sui is mizu or water and Kou is Iku or go. Although Japanese don't use this word "Suikou" instead of "Kyuuryuu" or "gekiryuu", this word was especially adopted as a name of this numerical code. SUIKOU-2D was written in FORTRAN-77 for Finite Difference Method, MACCORMACK SCHEME combined with ARTIFICIAL VISCOSITY SCHEME. These schemes allow to smoothly perform a numerical simulation of two-dimensional depth-averaged flow.

This model was tested to simulate and to evaluate the past flood inundation in 2004, Aioi area in Kofu, Yamanashi. It was a heavy storm with 78mm/hr of intensity resulted overflows from drainage channels and yielded a severe inundation in Aioi area.
Compared to the observed data, the numerical result was good. Then, it is supposed to be a useful tool for the water authority and the local government to design the countermeasures, to assess the inundation processes and to provide the potential inundation hazard map.

Result of 1 hour flooding (flood.avi).

An example of inundation process is as follow:

Link:
Sisinggih D., S. Oishi and K. Sunada: Flood Inundation Model for Highly Urbanized Area and Its Application to Simulate the Flood Inundation in 2004, Kofu-City, Japan, Annual Journal of Hydroscinece and Hydraulic Engineering, Vol.52, 2008.2

Partial dambreak

2008-01-24T18:40:00.000-08:00

To run a simple test of two-dimensional problem with SUIKOU-2D.FOR, there are seven input files required. Say, an example of simple dambreak problem. The computational domain is 200m x 200m and 5m x 5m mesh. The initial water level is maintained as 10m at the upstream dam and tailwater was 5m.

The input files to run the codes are:
geo.txt --> contains the geometry information on each node.
mesh.txt --> describe the mesh size and its numbers (dx,dy, nx,ny)
ctrl.txt --> setting up parameters, and hydraulic constants.
elv.txt --> contains the information of bed elevation data.
ho.txt --> initial water depth.
uo.txt --> initial velocity in x-direction.
vo.txt --> initial velocity in y-direction.

Screenshot of Running program :

The output file is ready formatted for GNUPLOT visualization.

Assumed it is in the directory of output files, then simply run the GNUPlot.
Type a command shown in the figure below (for advanced plotting, please refer GNUPLOT manual).

and it gave :

If it's need, the animations could be made.

The Water Surface animation

The vector velocity animation

(Reference: Open Channel-Flow, Chaudhry M.H, 1993. pp. 371-374)

Flow around rectangular block

2008-01-24T05:19:00.001-08:00

The water surface animation

The vector velocity

Aioi's flood in 2004

2008-01-22T01:00:00.001-08:00

This animation was created with SUIKO-2D. (approximately 5MB).

SUIKO-2D ver. 1.0

2007-10-24T20:34:00.000-07:00

Coming soon !!!!..... GIS BASED free software for general two-dimensional unsteady flow ...
SUIKOU-2D ver. 1.0 (2007)

Capabilities :
- Interfacing with GIS based data for pre and post processing
- Excellent mass balance conservation
- Shock capture availability

Contact : suiko2d@gmail.com

The example project : Flood in high urbanized area

SUIKOU for One-Dimensional Problems

2007-06-19T05:00:00.000-07:00

For the particular case study of one-dimensional flow, SUIKOU-1D.FOR was developed. There are two input files need to run. First is GEO.TXT which contains the grid of computational area. Another one is a control file, CNTL.TXT which contains parameters and program controls.

Let’s take a simple example of 1D flow over rough-variable topography. There is a single straight channel consists with 4 reaches and different bed slopes (So1=0.03, Lo1=40m; So2=0.008, Lo2=40m; So3=0.03, Lo3=40m; So4=0.011, Lo4=40m). The unit discharge is 2.36m2/s at the upstream and the water depth at downstream was set to 0.8m. Manning roughness coefficient was assumed as 0.03. The grid size is 1m. The downstream end elevation is set to be 0m.

The input files is described as below .

GEO.TXT is containing the information of channel properties. They are 4-coloumns separated by tab. First column is a node number. Then it's followed by the channel width (B) at the second column. Third column is representing the datum elevation on each node and the fourth-column is representing the initial water level to start the simulation.

CNTL.TXT is containing simulation control and hydraulic constants. First row is representing the upstream and downstream boundary conditions respectively. The second row is n manning roughness, bank slope and the gravity constant which used in the simulation. Third row is describing the momentum coefficient (), optional switch of artificial viscosity and-coefficient . Lasty, the fourth column is containing time step (dt) and total simulation time in seconds.

In the 1D code, it writes only one output file namely as OUT.TXT. This is a text file containing six columns and separated by tab. First column is indicating the node numbers and the second one is datum elevation of nodes. The third column is the elevation of water surface. The fourth column is the discharge unit. Finally, the fifth column represents the flow velocity whilts the sixth column is the froude number. All units are described on the metric system (SI).

Here is the detail format of OUT.TXT.

To visualize the water level along the channel by GNUPLOT, the following simple commands can be used.

Assume it's the directory contains the output files and GNU-Plot is running.
Type a command as shown in the figure below (for advanced plotting, please refer GNUPLOT manual):

For the discharge along the channel :

Animations: