[U1]
DAMBRK
DAMBRK Troubleshooting Hints
DAMBRK is a very complex program, and therefore getting the results you
desire will require some troubleshooting. This guide was put together to give
you some basic ideas of how to handle your DAMBRK problems, as well as
take you through some of the most common problems.
Modeling Strategies
1. What is your time step?
•
If it is too large, you will miss the peak value.
For example, let’s say you have a cross-section every 1/4 mile, but
your time step is 1 hour.
Figure 1: If time steps are too large, you may miss the actual peak.
If you “sample” every hour, then it is very easy to miss the peak
water surface elevation values.
2. What is the minimum base flow value for your inflow hydrograph?
•
DAMBRK is a “wet model,” meaning it must have some base
flow in it. It cannot start up dry (i.e., no flow in channel).
•
Note that the default error tolerances are:
Q
=
100 cfs.
WSEL
=
.01 ft
Hence, if your base flow discharge is less than this, your model can
easily blow up due to lack of flow in the model.
First, ask yourself, what is the widest channel bottom invert in
the entire model?
Figure 2: Where is the widest channel bottom in the model?
If it is very wide, then you may have to artificially insert a V-notch
bottom and/or increase the minimum base flow specified for
the upstream inflow hydrograph.
•
•
For example, when trying to get a model up an running,
experiment with increasing the base flow to 1,000 cfs.
•
Does the DAMBRK model now “spin up”?
•
If you increase the base flow that much, how much does
this really skew (or affect) your model results?
•
Try to determine this effect by increasing the base flow
another 1000 cfs. How much did your peak water
surface elevation change? Hydrograph attenuation
change?
•
If there was little to no change, run with it.
Be careful changing the error tolerance (in the Boundary
Conditions screen).
•
For example:
WSEL = .1 ft
Q = 1000 cfs.
3. What is the slope of the inflow hydrograph?
Figure 3: What does your inflow hydrograph look like?
•
Abrupt changes in inflow hydrograph can cause the model to
experience a numerical shock, causing convergence problems.
Figure 4: Abrupt change in inflow hydrograph.
•
May require that you perform some smoothing of the inflow
hydrograph.
Figure 5: Make sure your hydrograph has smooth curves.
4. Are there abrupt changes in channel top width?
Figure 6: Watch for abrupt changes in channel top width.
•
Is it reasonable to assume the entire top width as available for
flow? Or would it be wiser to count some of this increased top
width as storage?
•
“Rule of Thumb” 50% decrease, 100% increase in cross-section
to cross-section. Allow cross-section top width to increase by 100%
(i.e., 50 ft →100 ft) or decrease by 50% (i.e., 100 ft →50 ft) from
section to section. Remainder is accounted by defining as storage.
Table 1: Using the “Rule of Thumb”
Prior
XS1
Elev
100
110
120
130
140
Width
0
10
20
30
40
After
XS2
Storage
0
0
0
0
0
Elev
100
110
120
130
140
Width
0
10
70
300
500
XS2
Storage
0
0
0
0
0
Elev
100
110
120
130
140
Width
0
10
20
30
40
XS2
Storage
0
0
0
0
0
Elev
100
110
120
130
140
Width
0
10
40
60
80
5. Are there additional (internal) flow structures in your model (i.e.,
additional dams and bridges)?
Figure 7: Additional structures will complicate the calculations.
•
If so, first try removing all structures (simply undefine them in
the cross-section description screen). You will need to define
the DAMBRK model with Problem Specification 7.
•
If model continues to fail, problem lies further in the model.
Need to look at some of the other items discussed here to get
that portion of the model running.
•
Then insert the structures, one at a time — starting at the
upstream end of the model, run the model and make certain
results seem consistent. Troubleshoot each structure in turn.
6. Do you have bridges in your model?
•
Try treating the bridge as a dam with a spillway rating curve
designed to mimick flow through the bridge opening.
•
Conservative approach — assume the bridge gets washed out
totally, hence completely removed from the analysis.
Storage
0
0
30
240
420
7. Simplify, Simplify, Simplify
•
Try:
•
No hydrograph, simply a constant Q
•
1 Structure
•
Shorter time step
•
Less cross-sections
•
Undefine your structure failure— For example, WSEL at
time of failure = 20,000 ft
8. Develop a procedure (or system) for troubleshooting your model.
•
Get a notebook and document what you have tried.
•
Save your files and note what you have done to change it from
the last run.
•
Saving stages of your development efforts allows you to
immediately return to the prior stage that was working when
you hit a dead-end.
9. Try different problem options.
1 Structure: Problem Specification Option 1, 11 and 13
2 Structures: Problem Specification Option 12 and 14)
•
Option 1
Figure 8: Cross-section downstream of structure.
•
This will not work when there is a concern about
tailwater submergence, since DAMBRK computes
outflow from the structure first, without any tailwater
effect and then routes it (2 step process).
•
Note that option 1 is very similar to option 13, simply
respecify problem option in Project Description input
screen
•
Options 11 & 13, and 12 &14
•
Fully dynamic (does dam failure analysis simultaneously
with routing), and hence can account for tailwater
submergence affects on dam failure.
•
Option 11 assumes a storage/elevation relationship to
define reservoir volume.
•
Option 13 requires additional upstream cross-sections to
define reservoirs, due to a negative wave traveling
upstream.
Figure 9: Placement of additional cross-sections due to existence of a negative wave.
DAMBRK Additional Tips
1. There must be a positive slope at the last two downstream
cross-sections, otherwise initial conditions can’t be computed
2. If trouble is with the loop rating curve downstream boundary
condition, then try the standard rating curve.
3. May want to move downstream section farther downstream to get
accuracy at downstream cross-section of interest.
4. Stay simple and systematic and then add complexity.
5. Effective storage top width for wide expansion
Figure 10: Effective top widths.
6. Contraction/Expansion coefficients
•
Use 4:1 expansion, 1:1 contractions to define effective flow
width.
•
Use lower range of contraction/expansion coefficients since
there is so much going on.
•
Later, check these coefficients to see how adjusting them affects
your answer. Should have a minimal effect.
7. Cross-section placement
Figure 11: Correct cross-section placement.
8. Low Flow Filter
Figure 12: Re cognizing the pre-wave effect due to compuation.
•
Dip can drop below channel bottom.
9. Downstream Boundary
If flow dumps into a lake and elevations stays constant for any flow
— do not specify it as a rating curve, but do as a tide dependent.
The constant value rating curve will confuse DAMBRK.
10. Composite n
•
V=
149
.
2
1
R 3 S f2
n
Figure 13: Locating the variables.
•
Horton & Einstein Method
a) Proportioning Scheme where R is replaced by A/P.
2
b) V =
1
. A 3 S f2
149
2
nP 3
[
c) nP 3 = ∑ Pn1.5
2
]
2
3
[∑ Pn ] = [Pn + P n + P n ]
=
1.5
d) ncomposite
2
P3
2
3
1.5
1 1
1.5
2 2
2
P3
1.5
3 3
2
3
[U2]
DAMBRK
DAMBRK Typical Problems
DAMBRK Typical Problem 1
Typical Problem 1: Computation Stops During Computation of Initial Conditions
Possible Causes and Solutions
1. Adverse slopes or scour holes in channel profile invert can cause
problems.
•
In these situations smooth the channel profile invert.
2. Review input data reported by the program in the output file.
•
Look for errors.
3. Review river reach for likely trouble spots.
•
Bridges - Take them out.
•
Steep reaches - Try raising “n.” Also try sub/super critical flow
options in boundary conditions.
•
Sudden changes between cross-section geometry - Make the
distance between interpolated cross-sections smaller.
4. Change downstream boundary condition.
•
Beware of channel control (loop rating) for small slopes. This
can cause problems, especially for flat slopes.
•
Add another (artificial cross-section downstream) to give a
steeper slope, but be aware that this can affect the results.
•
Temporarily, try a different boundary condition.
5. Check initial flow.
•
Is headwater level at dam(s) giving a reasonable outflow?
•
Must have flow through the dam. Cannot start-up model
“dry.”
•
Error tolerance in DAMBRK (default) is 100 cfs.
•
Some initial flows are unfortunate choices given channel, bridge
or dam characteristics. For example, if flow is just about to
overtop a bridge, dam, overbank, etc., DAMBRK will have
difficulty.
DAMBRK Typical Problem 2: Non-Convergence During Routing
Typical Problem 2: Non-Convergence During Routing
Possible Causes and Solutions
1. Sources of information
•
Non-convergence message - when and where does the problem
first appear?
•
Re-run for a detailed printout at apparent trouble spots - Froude
number, velocity, top width, etc. Check for these often going to
critical depth. Froude number near 1, velocity suddenly
changing, top width suddenly changing.
•
Look for inconsistencies in initial conditions.
•
Review graphics if available–run up just prior to time step with
convergence problem and look at stage hydrographs.
•
Try taking out possible trouble-makers one by one (i.e., bridges,
downstream boundary, etc).
2. Look for sudden changes with respect to time or space where the
problem first appears
•
•
•
Vertical change in flow width at a cross-section
•
Substitute storage for active flow
•
Use flood plain and convergence
Longitudinal change in flow width between cross-sections
•
Substitute storage
•
Decrease distance between interpolated cross-sections
Sudden change in discharge (e.g., dam failure)
•
Adjust time step
•
•
If acceptable, increase failure time (you may have to
perform a substitute study to see what effect this has on
results)
Sudden change in rating curve at a boundary (e.g., bridge
overtopping)
•
Storage Adjustments
Vertical change
Figure 14: Storage adjustments with a vertical change.
Longitudinal change
Figure 15: Storage adjustments with a longitudinal change.
3. Other Problem Sources
•
Flow depth becomes too small
•
Prevented by “safety-net” feature
Figure 16: Where the flow depth can become too small.
•
Supercritical flow (or just barely critical flow)
•
Use internal boundary at location of critical depth (as
with dam and rating curve)
•
Use sub/super critical flow options
•
Adjust Manning’s n to get rid of supercritical flow for
Froude number near 1 (i.e., critical depth)
•
Break model into separate models.
•
Downstream Boundary, (esp. loop rating curve)
•
Inconsistent cross-section geometries
DAMBRK Typical Problem 3
Typical Problem 3: Strange Results (Program runs, but gives suspicious results)
Possible Causes and Solutions
1. Check summary tables for unreasonable depths, velocities, time to
peak, etc.
2. Did the hydrograph have enough time to pass through the entire
reach?
•
May need to lengthen inflow hydrograph description to route
flood peak all the way downstream
•
Check the stage hydrograph at key cross-sections along the
model
3. Same troubleshooting strategies as previously described.
Strange results example
Figure 17: Results that are wrong for mysterious reasons.
Fixed Results
Take unreasonable channel top width increase and place in storage.
Increase in friction caused additional head to drive flow.
Figure 18: Results after a minor fix in the data.
Dambrk Output Results
Location
(Mi)
Elevation
(ft MSL)
(cfs)
Stage
(hr)
Velocity
(ft/sec)
(ft MSL)
Elevation
(hr)
7.683
1118.02
84204
20.362
3.43
.00
.00
7.885
1117.44
99314
20.562
3.52
.00
.00
8.088
1117.16
93405
20.362
3.76
.00
.00
8.209
1116.57
100664
20.537
4.06
.00
.00
8.509
1116.44
94069
20.362
4.11
.00
.00
8.727
1115.96
101670
20.487
4.07
.00
.00
8.946
1115.90
95309
20.362
4.00
.00
.00
9.164
1115.44
102448
20.437
4.31
.00
.00
9.393
1115.44
100858
20.362
4.17
.00
.00
9.601
1115.25
105937
20.437
4.31
.00
.00
9.820
1115.48
111944
20.412
4.70
.00
.00
10.124
1115.28
122006
20.387
5.44
.00
.00
10.428
1115.13
135448
20.362
6.55
.00
.00
10.732
1117.28
167601
16.712
9.42
.00
.00
11.036
1115.98
158499
20.362
7.88
.00
.00
11.340
1112.34
279101
16.762
39.46
.00
.00