Professional Documents
Culture Documents
Illudas 1 - 81
Illudas 1 - 81
M. D. Watson
The mathematica~ modelling approach has proved very suitable for urban
drainage studies and challenging research has been commenced on these
lines. This particular report documents the adaptation and use of an
isochronal-type model for simulating the flow in storm-water drains.
ii
ABSTRACT
TABLE OF CONTENTS
Page
Preface i
Abstract ii
Chapter 1 : Introduction 1
Page
References 84
CHAPTER 1 INTRODUCTION
The major objectives of this report are to assess how well the
- 5 -
model will perform under local conditions and how the parameters
may be estimated for design purposes. The limited availability of
local urban rainfall and runoff data make this assessment
difficult and further data collection and subsequent assessment
are recommended. A modified metric version of the model is
presented and tentative recommendations for parameter estimation
are made. The successful performance of the model makes
it a promising design tool and it is therefore recommended for
local design use.
- 6 -
2.1 GENERAL
In the tests by Watkins (1962) runoff from pervious areas was found
to be negligible and therefore ignored in the calculations. The
percentage of impervious area ranged from 2% to 100% and runoff
from such areas was determined using a percentage loss rate or in
some cases a uniform loss rate based on observations.
Storage:discharge relationships used in the calculations were
determined from the recession portions of observed hydrographs. The
mean ratio of computed to observed peak discharge for all areas was
0,98 with a standard deviation of 0,15.
(iv) The model should not be promoted for general use (in the
USA) because as indicated in (ii) and (iii) under certain
circumstances the results are not satisfactory
Papadakis and Preul (1973) tested the original TRRL method for two
storms (three peaks) on Oakdale Avenue catchment in Chicago, USA.
The catchment has an area of 5,2 ha of which 45% is paved. The mean
ratio of computed to observed peak discharge was 1,15 with a
standard deviation of 0,31. Data from this area had been previously
used by Terstriep and Stall (1969) who noted that some observed
peaks were truncated due to problems with the recording device.
Further errors can partly be accounted for
in neglect of pervious runoff and initial losses.
(1971), and by Heeps and Mein (1973) are presented. The mean value
of the ratio of computed to observed peaks was 0,95 with a high
standard deviation of 0,45. Aitken concluded that the TRRL method
did not accurately simulate the runoff process. Examination of the
paper by Heeps and Mein (1974), however, shows that apart from the
sampling errors many of the errors can be ascribed to neglect of
pervious area runoff.
The results of all these tests are summarised in Table 2.1. For
further information on performance of the TRRL method the reader
is referred to the original publications as well as an excellent
summary by Colyer and Pethick (1976).
- 12 -
- 13 -
2.3 ILLUDAS
Data from two rural catchments were used to test the grassed
area runoff computations. The catchment sizes were 3 ha and
7 ha. Parameters were estimated in the same fashion as for
urban catchments. In all 12 events were tested. ILLUDAS predicted
zero runoff for four of these, three of which had an average
runoff of 0,4 mm and one of 15 mm. The mean ratio of computed to
observed discharge for the remaining G events was 0,65 with a
standard deviation of 0,20. The prediction of runoff volume was
slightly better with a ratio of computed to ob-
- 14 -
- 15 -
The presence of data errors for both urban rural catchments was
recognised but no attempt was made to eliminate them. This and
the use of estimated parameters prevents direct comparison of the
results with those from other verification studies.
Some of the data used to assess the performance of the TRRL method
were used in the assessment of ILLUDAS. A comparison of results
shows a decrease in the mean absolute error for the prediction of
peak discharge on 8 catchments from 42% to 37%. On another
catchment the error increased (20% to 49%). This was probably due
to inadequate consideration of antecedent moisture conditions.
ILLUDAS has had fairly widespread use in the USA and Canada since
its inception. Its use in Canada has led to the following re-
commendations (Wisner et al, 1979) regarding practical application
to design:
Applicability of
Use
ILLUDAS
Rural areas Fair
Preliminary analysis Very good to excellent
Detailed analysis of free surface peak flows Good
Detailed analysis of valley storage Fair
Surcharged flow Poor
2.4 DISCUSSION
ILLUDAS is a logical extension of the TRRL method to account for
pervious area runoff. Its use for design seems reasonable because
of its logical accounting for losses during severe events. The
test results presented in this chapter do not, however, show any
real improvement over the TRRL method in its capability of
reproducing observed events. This could be partly due to the fact
that most observed events are not sufficiently intense to produce
significant runoff from pervious areas. It does seem, however,
that the major cause could be the presence of errors
in the data and the different methods of assessment adopted in
the various studies.
Data errors can be due ,to many causes including spatial non-
uniformity of rainfall, discharge rating errors, drowned
weir, leakage or blockage of pipes and gauge malfunctions. Some
of these difficulties are reported both by Watkins (1962) and
Terstriep and Stall (1974) in the original assessments of the
two models. The different approaches adopted by these re-
- 17 -
3.1 INTRODUCTION
The rainfall input is the same as that for the paved area and
is shown in Fig. 3.4c. Runoff from the paved area draining onto
the grassed area is shown in Fig. 3.4d and is termed
supplementary paved area runoff (SPARO). This runoff is assumed
to be instantly and uniformly distributed over the grassed area
and is added to the rainfall input.
For the paved area the program uses Manning's equation to compute
the travel time in the gutter assuming a hydraulic radius of 60
rom and a retardance coefficient, n, of 0,02. The n value of 0,02
is reasonable for street gutters while the assumed hydraulic
radius implies an average flow depth of about 70 rom for a 1 in
10 cross-slope gutter - a reasonable value for minor road drains.
The overland flow time for runoff to reach the gutter is assumed
equal to 2 minutes. This is added to the gutter travel time to
obtain the entry time.
3.6 INFILTRATION
Corrections are made for the rainfall being less than the in-
filtration capacity and the equation is solved by the Newton-
Raphson technique.
The paved and grassed area hydrographs determined for each sub-
catchment are combined to become an inlet hydrograph to the
drainage system. If the sub-catchment is at the uppermost end of
a series of pipes or open channels, the inlet hydrograph is
entered into the system by routing it downstream to the next
input point. If the sub-catchment occurs somewhere below the
upper end, its inlet hydrograph is combined with the upstream
hydrograph and the resulting combined hydrograph is routed
downstream to the next input point. If the sub-catchment is
located at the confluence of two or more pipes, the inlet
hydrograph is combined with the converging hydrographs before
routing downstream.
The first is the more economical of the two from the point of
view of computing time and is adequate in many instances. The
second is based on a sounder theoretical approach. Neither
method considers non-uniform flow, dynamic effects or the effect
of increased head due to surcharge.
where 11 and 01 are the initial inflow and outflow; 12 and 02 are
the final inflow and outflow; CAREA is the cross-sectional flow-
area function in which the kinematic wave assumption that Sf=So
and Manning's equation are used, where Sf is the friction slope
and S is the reach slope; and L is the reach length. Eqs. (3.3)
and (3.4) can be combined to yield:
∆s = I - 0 . . . . . . . . . . . . . . . .(3.9)
I = inflow
O = outflow
- 32 -
4.1 GENERAL
Travel
Length Slope Diameter Height Width Lateral Discharge
Reach Section* time
(m) (%) (mm) (m) (m) slope capacity (m3/s)
(min)
* C = circular; T = trapezoidal
- 39 -
- 40 -
- 41 -
The runoff records were screened for large events and ten were
selected for further analysis. Three consecutive events had
hydrographs truncated at a discharge well below the maximum
observed. This was probably due to a temporary pipe blockage and
the records of these events had to be discarded. The remaining
seven events are tabulated in Appendix E.
- 44 -
4.4 DISCUSSION
5.1 GENERAL
This synthetic distribution has had fairly widespread usage and has
become known as the Chicago design storm. It can read-
- 61 -
t = tb + ta
Substituting for t from eqs (5.4) and (5.5) in eq. (5.3) gives the
following relationships for intensities before and after the peak:
(iv) Compute the points before and after the peak by inte-
grating the design curve and calculating the discrete
- 64 -
a = 3930
γ = 0,12
b = 21
c = 1,1
and with an assumed value of c one can solve for the constants
(a1/c) and (-b) by linear regression using (I-l/c) as the abscissa
and t as the ordinate. This is repeated using a different value
of c until an acceptable fit has been achieved. Alternatively eq.
(5.1) can be written as
Region b c γ60
Inland 14,4 0,883 44,9
Coastal 12,6 0,737 23,5
For MAP greater than 1 000 mm, eq. (5.14) gives results differing by
up to 20% from those derived from the curves in Fig. 4 of
HRU 2/78. The maximum difference for MAP less than 1 000 mm
is 6%.
From the above data it seems that values for r will generally be
less than 0,5. Higher values will probably be more appropriate for
coastal regions and lower values for inland regions. However, due
to the limited data it is suggested that a value of 0,4 be used
for all localities until further data become available.
CHAPTER 7 CONCLUSIONS
REFERENCES:
14. M.M. DILLION LTD. Storm drainage criteria manual for the
city of Burlington. Ontario, Canada, April 1977.
15. FORD, W.G. The adaptation of the RRL hydrograph method for
tropical conditions. Proc. Nairobi flood hydrology
symposium, Oct. 1975, Transport and Road Research La-
boratory SR259, pp 409-455.
23. KEIFER, C.J. and CHU, H.H. Synthetic storm pattern for
drainage-design. J. Hyd. Div., Amer. Soc. Civ. Engrs.,
Vol. 83, No. HY4, Aug. 1957, paper no. 1332.
34. RAGAN, R.M. and DURU, J.O. Kinematic wave nomograph for
times of concentration. J. Hyd. Div., Amer. Soc. Civ.
Engrs., Vol. 98, No. HY10, Oct. 1972, pp 1765-1771.
50. WILLIAMS, D.W., CAMERON, R.J. and EVANS, G.P. TRRL and
Unit hydro graph simulations compared with measurements in
an urban catchment. J. Hydrol., Vol. 48, No. 1/2, Aug.
1980, pp 63-70.