Вы находитесь на странице: 1из 8

See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/239387786

Closure to Simple Equations for Aquifer


Parameters from Drawdowns in Large-Diameter
Wells by Sushil K. Singh

ARTICLE in JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING FEBRUARY 2009


Impact Factor: 1.09 DOI: 10.1061/(ASCE)0733-9437(2009)135:1(128)

CITATIONS DOWNLOADS VIEWS

5 14 35

1 AUTHOR:

Sushil Singh
National Institute of Hydrology
185 PUBLICATIONS 694 CITATIONS

SEE PROFILE

Available from: Sushil Singh


Retrieved on: 23 June 2015
DISCUSSIONS AND CLOSURES

Cooper, H. H., Jr., Bredehoeft, J. D., and Papadopulos, I. S. 1967.


Discussion of Estimating Storage Response of a finite-diameter well to an instantaneous change of
Coefficient and Transmissivity from Slug water. Water Resour. Res., 31, 263269.
Test Data by Prabhata K. Swamee
and Sushil K. Singh
September/October 2007, Vol. 133, No. 5, pp. 505507.
DOI: 10.1061/ASCE 0733-94372007133:5505 Discussion of Estimating Storage
Coefficient and Transmissivity from Slug
Robert P. Chapuis1 Test Data by Prabhata K. Swamee and
1
Dept. of Civil, Geological, and Mining Engineering, cole Polytech-
nique, PO Box 6079, Stn CV, Montreal, Quebec, Canada H3C 3A7. Sushil K. Singh
E-mail: robert.chapuis@polymtl.ca
September/October 2007, Vol. 133, No. 5, pp. 505507.
DOI: 10.1061/ASCE0733-94292007133:5505
The authors have proposed an approximation for the well function
tabulated by Cooper et al. 1967 for their slug test solution. They Zekai en1
then have applied a least-squares error method to estimate the 1
Istanbul Technical Univ., Faculty of Civil Engineering, Hydraulics and
aquifer parameters for two data sets. However, this slug test so- Water Resources Division, Maslak 34469, Istanbul, Turkey. E-mail:
lution is physically and mathematically wrong Chapuis 1998 zsen@itu.edu.tr
and thus should no longer be used by engineers. The Cooper et al.
1967 solution is in error by a factor of 5-10 for T values and
by a factor of 200-50,000 for S values Chapuis 2009. Similarly, This technical note denies one of the most significant aspects of
the solution for pulse tests Bredehoeft and Papadopulos 1980 is any modeling study: that the model has to be parsimonious, i.e.,
in error by a factor 100 for T and a factor 104 108 for S Chapuis economic in terms of the number of parameters. In this technical
and Cazaux 2002. As a result, the standard D4104 ASTM note the basic simple equations Eqs. 15 are rendered into a
2006b for slug tests was not renewed in 2002 by ASTM, and the riddle of complicated expression through Eq. 6, which involves
interpretation method of standard D4631 for pulse tests ASTM seven parameters, each of which has no physical, rational, or
2006a is likely to be modified. logical meaning. Where is the scientific objectivity in the deriva-
One may question why these authors and others continue to tion of Eq. 6? It seems to be the subjective derivation of the
use this inaccurate solution. A related discussion Chapuis 2009 authors, perhaps following much trial and error. In contrast, Eqs.
suggests that it may be the result of poor communication between 15 are based on already available tables that are very useful in
engineers and scientists. Unfortunately, the authors may have
practical applications.
been misled by this communication problem.
In the aquifer parameters estimation section the authors again
use the optimization technique with the objective of minimizing
the error as defined in Eq. 9. The application indicates that there
References are absolute relative errors of 4.5 and 57% between the Cooper et
al. 1967 transmissivity and storage coefficient estimations and
ASTM. 2006a. Standard test method for determining transmissivity
and storativity of low permeability rocks by in situ measurements authors. As far as the transmissivity is concerned, 4.5% improve-
using the pressure pulse technique D4631. ASTM International, ment if it is improvement at all is not significant practically. On
West Conshohocken, Pa. the other hand, the relative error of 57% for storativity implies
ASTM. 2006b. Standard test method for determining transmissivity of that the authors approach is not convenient. This can also be
nonleaky confined aquifers by overdamped response to instantaneous deduced from the percentage error graphs in Figs. 2 and 3, which
change in head D4104. ASTM International, West Conshohocken, imply that the errors have structural serial dependence, because
Pa.
the successive values do not fluctuate randomly around the zero
Bredehoeft, J. D., and Papadopulos, S. S. 1980. A method for deter-
mining the hydraulic properties of tight formations. Water Resour. error horizontal level. Additionally, the percent error propagation
Res., 16, 233238. with time has an increasing variance, which indicates that the
Chapuis, R. P. 1998. Overdamped slug test in monitoring wells: errors are not stationary and hence the proposed model is not
Review of interpretation methods with mathematical, physical and suitable. Thus, the final sentence in the conclusion is not valid at
numerical analysis of storativity influence. Can. Geotech. J., 355, all.
697719.
Chapuis, R. P. 2009. New methods for aquifer parameters from slug
test data: Discussion. J. Irrig. Drain. Eng., in press.
Chapuis, R. P., and Cazaux, D. 2002. Pressure-pulse test for field hy- References
draulic conductivity of soils: Is the usual interpretation method ad-
equate? Evaluation and remediation of low and dual porosity Cooper, H. H., Jr., Bredehoeft, J. D., and Papadopulos, I. S. 1967 Re-
environments, M. N. Sara and L. G. Everett, eds., ASTM Interna- sponse of a finite-diameter well to an instantaneous charge of water.
tional, West Conshohocken, Pa., 6682. Water Resour. Res., 31, 263269.

JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING ASCE / JANUARY/FEBRUARY 2009 / 125

Downloaded 21 Apr 2009 to 210.212.58.170. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
1998 for estimating the aquifer parameters is erroneous and not
Closure to Estimating Storage Coefficient justified.
and Transmissivity from Slug Test Data by In view of the scope of this closure, the writers refrain from
Prabhata K. Swamee and Sushil K. Singh further commenting on the Chapuis 1998 method and will not
discuss the issue of the solution provided by Bredehoeft and Pa-
September/October 2007, Vol. 133, No. 5, pp. 505507. padopulos 1980 as this solution was not used in the original
DOI: 10.1061/ASCE0733-94372007133:5505 technical note. However, there may be some error in the method
used by the discusser. The writers believe it is inappropriate to
Sushil K. Singh1 and Prabhata K. Swamee2 discuss ASTM procedures here. Any agency may adopt or recom-
1
Scientist, National Institute of Hydrology, Roorkee-247667, Uttarak- mend a procedure/method for a particular objective, but this
hand, India. E-mail: sukusi1@yahoo.com should not be a binding on research.
2
Emeritus Fellow, Dept. of Civil Engineering, Indian Institute of Tech- The basic solution used by the writers is correct.
nology Roorkee, Roorkee-247667, Uttarakhand, India. E-mail:
swameefce@yahoo.com

The writers are thankful to the discussers for raising certain Points by Sen
points, which are clarified below:
The parameters of a model and the constants of an equation
should be interpreted differently, yet the discussion does not dif-
ferentiate between the two. In Eq. 6, there are only two param-
Points by Chapuis eters of the model, i.e., and . The values k1-k7 are seven
constants, which were obtained by minimizing the error between
The Cooper et al. 1967 solution pertaining to slug tests is physi- the tabulated values of W and Eq. 6, resulting in Eq. 7. The
cally and mathematically correct and applicable to the field tests writers Eq. 7 is computationally much simpler than Eq. 2,
that satisfy the assumptions involved in the derivation of the so- which involves numerical integration of a complex function con-
lution. The reasoning given by Chapuis 1998 for the invalidity sisting of Bessel functions of different types and orders. Eq. 7
of this solution is not justified. This solution is based on the can be used for calculating W for any set of values of the argu-
solution of an analogous problem a cylindrical metallic perfect ments, while it may require double interpolation if tabulated val-
conductor in an infinite medium of lower thermal conductivity of ues of W are used. This restricts the practical application of the
heat conduction given by Carslaw and Jaeger 1959, p. 341-342. tabulated values.
The analogy gives rise to the following assumptions: 1 there is The Cooper et al. 1967 solution Eqs. 15 and curve
no skin effect between the well and aquifer; 2 the water moves matching method for estimating aquifer parameters should be
in the well with no resistance, which denotes a unity storage treated differently as far as accuracy is concerned. Eqs. 15
coefficient for the well space; and 3 the water transfer between are analytically derived, and Cooper et al. 1967 stated that the
the well and aquifer is analogous to the heat transfer between the storage coefficient cannot be estimated reliably using their curve
metallic conductor and medium. The writers have used rw = rc matching method. The estimates obtained using the Cooper et al.
giving = S, which does not violate the assumption considered by 1967 curve matching method should not be used as a bench-
Carslaw and Jaeger 1959, i.e., is twice the ratio of the heat mark. The writers method yields more reliable estimates of the
capacity of an equivalent volume of the medium to that of the parameters than those obtained using the Cooper et al. 1967
perfect conductor. Based on the similar reasoning, Eq. 3 of the method, which has been substantiated in Table 1 of the original
writers technical note, given by Cooper et al. 1967, seems cor- technical note along with the standard error of estimate SEE as
rect. a reliability criterion.
The Cooper et al. 1967 analytical solution and curve match-
ing method should not be treated as equally accurate. More reli-
able methods a diagnostic curve method and a simple objective
method for estimating aquifer parameters from slug test data References
have been proposed by Singh 2007. Both sets of data used by
Chapuis 1998 for assessing the reliability of the Cooper et al. Bredehoeft, J. D., and Papadopulos, S. S. 1980. A method for deter-
1967 curve matching method pertain to partially penetrating mining the hydraulic properties of tight formations. Water Resour.
wells, while the Cooper et al. 1967 method is applicable for Res., 16, 233238.
fully penetrating wells. Therefore, the comparison and conclu- Carslaw, H. S., and Jaeger, J. C. 1959. Conduction of heat in solids,
sions as given by Chapuis 1998 are not justified. The large Oxford University Press, New Delhi, India.
errors reported in the discussion may be the result of partial pen- Chapuis, R. P. 1998. Overdamped slug test in monitoring wells: Re-
view of interpretation methods with mathematical, physical and nu-
etration and the empirical nature of the method used by the dis-
merical analysis of storativity influence. Can. Geotech. J., 355,
cusser. A large error may be involved in numerically comparing
697719.
the derivative as suggested by Chapuis 1998 and would be still Cooper, H. H., Jr., Bredehoeft, J. D., and Papadopulos, I. S. 1967.
more if measurements are made at longer time intervals. The Response of a finite-diameter well to an instantaneous charge of
straight-line velocity graph of Chapuis 1998 is based on inap- water. Water Resour. Res., 31, 263269.
propriate data, i.e., on a partially penetrating well. Also, the semi- Singh, S. K. 2007. New methods for aquifer parameters from slug test
logarithmic curve matching procedure suggested by Chapuis data. J. Irrig. Drain. Eng., 1333, 272275.

126 / JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING ASCE / JANUARY/FEBRUARY 2009

Downloaded 21 Apr 2009 to 210.212.58.170. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
increases the tediousness of the aquifer parameter estimation pro-
Discussion of Simple Equations for cedure application in practice. The classical type curve matching
Aquifer Parameters from Drawdowns in procedures show the local deviations from the relevant type
Large-Diameter Wells by Sushil K. Singh curve, and an aquifers heterogeneity can be interpreted accord-
May/June 2007, Vol. 133, No. 3, pp. 279281. ingly en 1994.
DOI: 10.1061/ASCE0733-94372007133:3279
References
Zekai en1
1
Istanbul Technical Univ., Faculty of Civil Engineering, Hydraulics and
Papadopulos, I. S., and Cooper, H. H. 1967. Drawdown in a well of
Water Resources Division, Maslak 34469, Istanbul, Turkey. E-mail:
large diameter. Water Resour. Res., 31, 241244.
zsen@itu.edu.tr
en, Z. 1994. Hydrogeophysical concepts in aquifer test analysis.
Nord. Hydrol., 253, 183192.
Aquifer test analysis and parameter estimations should be based en, Z. 1995. Applied hydrogeology for scientists and engineers, CRC
on scientific, philosophical, logical, physical, and then mathemati- Lewis Publishers, Boca Raton, Fla.
cal procedures by considering the hydrogeological setup of the Wikramaratna, R. S. 1985. A new type curve method for the analysis
well location and its vicinity for the best benefit, reliable interpre- of pumping tests in large diameter wells. Water Resour. Res., 212,
tations, and parameter estimations en 1995. It is not possible to 261264.
rely on any mathematical procedure if the physical foundations Zdankus, N. T. 1974. Analysis of shallow hard rock well pumping and
are missing, yet the derivations of the author lack such founda- recovery test data. Ground Water, 125, 310317.
tions. Although the author uses the phrase Simple Equations in
the title of his technical paper, in fact his equations are compli-
cated and physically meaningless. See, for example, Eqs. 5
13, which are the convenient versions of the basic and
physically plausible Eqs. 13. In addition, the simplifications
Discussion of Simple Equations for
of the aquifer test mathematics Eqs. 13 are presented either Aquifer Parameters from Drawdowns in
in the form of convenient tables or as type curves for practical Large-Diameter Wells by Sushil K. Singh
uses. The author tries to cover this simple procedure with a te- May/June 2007, Vol. 133, No. 3, pp. 279281.
dious approach that includes curves similar to type curves as in DOI: 10.1061/ASCE0733-94372007133:3279
Fig. 1 and physically implausible equations, which yield aquifer
parameter estimations without visual inspection of genuine time-
Mesut imen1
drawdown data from the field. 1
Faculty of Engineering-Architecture, Dept. of Civil Engineering,
In the application section, the author generates synthetically Sleyman Demirel Univ., 32260, Isparta, Turkey. E-mail: mesutcim@
time-drawdown data from available tables and hence adopts an mmf.sdu.edu.tr
ideal curve from already available type curves with whole exten-
sions for small and large times. Then he tries to calculate the
The author suggested a method for aquifer parameter estimates.
aquifer parameters from his proposed curves. The 50 min data
He states in the technical note, Due to the subjectivity of the
requirement is an implied result of such a generation, and it is
existing curve methods namely, the Papadopulos and Cooper
well-known from field applications that 50 min is a very short
PC 1967 and Wikramaratna 1985 methods, the estimated
duration for aquifer response, especially in large-diameter wells,
parameters are not very accurate and different people may arrive
which are used in low transmissivity aquifers. It is already stated
at different estimates. Hence the author determined the peak and
in the technical note that It is appropriate to mention here that
these data up to 50 min nearly fall on the straight line portion of time of the peak for each value of by using the tabulated values
the Papadopulos and Cooper 1967 or Wikramaratna 1985 given by Papadopulos and Cooper 1967. He wrote, If such a
curve. This shows that the aquifer parameters cannot be estimated peak is determined and even the data seem to fall on the straight
from these data using the curve matching methods of Papadopu- line portion of the PC curve, the aquifer parameters can be accu-
los and Cooper 1967 or Wikramaratna 1985. The author ad- rately estimated using the proposed method. However, the pro-
mits that the data up to 50 min nearly falls on the straight line posed method still contains subjectivity.
portion of the Papadopulos and Cooper curve, which means that 1. The peak value of the dimensionless time factor u p be-
the pump discharge comes from well storage. Hence the aquifer comes 0.0010517 for = 0.01. The values of the dimension-
parameter estimations cannot be regarded as reliable because less well function for this u p in the PC and Theis 1935
there is not yet aquifer response. methods are 4.43112 and 6.28116, respectively. According to
On the other hand, the author alleges, without any scientific these values of the well function, a fit does not seem possible
evidence, that for data from Zdankus 1974 the Papadopulos and because the straight-line portion of the PC curve matched
Cooper 1967 curve method cannot yield a reliable estimate of with the Theis curve begins after u = = 2.105, for which the
the aquifer parameters. In fact this method yields in many cases values of the well function in both the PC and Theis methods
reliable estimates of the transmissivity but not storativity. It are 10.201 and 10.2426, respectively.
would have been more illuminating if the author had presented 2. The proposed method depends on the observed values of t p
the unreliable Papadopulos and Cooper 1967 parameter estima- and s2wp / t p and a very short duration of the pumping. The
tions and compared them with the results in his technical note. pumping measurements are generally made in certain steps
In conclusion, the procedure presented by the author is math- of time. Therefore the determination of the peak values from
ematical fantasy, not a physical interpretation or a reliable param- the observations is possibly arbitrary and results in subjec-
eter estimation of an aquifer test. The definition of the unimodal tivity. Besides, a very short duration of pumping does not
curve has no physical basis, and it is a sort of type curve. This give sufficient information about the aquifer response, which

JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING ASCE / JANUARY/FEBRUARY 2009 / 127

Downloaded 21 Apr 2009 to 210.212.58.170. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
shows fundamental properties of the confined, unconfined, compared to the writers Eqs. 813. Eq. 1 contains the well
and leaky aquifers. function. The numerical evaluation of this function for given val-
3. The author states that for Zdankus 1974 data the PC curve ues of the arguments is tedious as it involves Bessel functions of
matching method cannot yield a reliable estimate of the aqui- different types and orders. If one intends to use the tabulated
fer parameters. However, is found as 0.151 whereas Eqs. values, double interpolation is required, which makes the process
811 are developed using the data for 101 105. cumbersome. Singh 2007 has proposed a computationally
The value obtained is not within this range. This shows that simple function for approximating the well function for large-
the proposed method involves subjectivity. diameter wells.
As consequence the method presented by the author is still The difficulties in using the Papadopulos and Cooper 1967
handicapped by subjectivity. The reliability of the method should type curves for estimating the aquifer parameters are 1 unique
be tested by the PC curve matching method on a set of field data. estimates of the aquifer parameters cannot be obtained; 2 reli-
able estimates of the storage coefficient cannot be obtained; 3 a
long duration pumping test is needed, which may not be feasible
References with constant discharge; and 4 apart from the subjectivity in-
volved in matching the curves, interpolation for the value of is
Papadopulos, I. S., and Cooper, H. H. 1967. Drawdown in a well of required. The advantages of the writers method over the Papa-
large diameter. Water Resour. Res., 31, 241244. dopulos and Cooper 1967 curve matching method are 1 unique
Theis, C. V. 1935. The relation between the lowering of the piezomet- and reliable estimates of the aquifer parameters are obtained; 2
ric surface and the rate and duration of discharge of a well using
only a short duration pumping test is needed, which can save time
ground water storage. Trans., Am. Geophys. Union, 16, 519524.
Wikramaratna, R. S. 1985. A new curve method for the analysis of
and money; and 3 no subjectivity is involved and the aquifer
pumping test in large-diameter wells. Water Resour. Res., 212, parameters can be estimated using computationally simple equa-
261264. tions. However, the peak and time to peak of a unimodal curve
Zdankus, N. T. 1974. Analysis of shallow hard rock well pumping and must be determined.
recovery test data. Ground Water, 125, 310317. In the writers method, the estimates of the aquifer parameters
are based effectively on one point, i.e., the peak of the unimodal
curve. Hence, the reliability of the estimated parameters can be
checked in an ideal sense. All the drawdowns can be simulated
using the estimated parameters and compared to the respective
Closure to Simple Equations for Aquifer observed drawdowns with a visual or numerical inspection. If the
Parameters from Drawdowns in simulated drawdowns depart systematically and substantially
Large-Diameter Wells by Sushil K. Singh from those observed, a nonideal well, aquifer, or pumping condi-
tion is confirmed.
May/June 2007, Vol. 133, No. 3, pp. 279281.
Although an alternative set of curves has been proposed by the
DOI: 10.1061/ASCE0733-94372007133:3279
writer Fig. 1, it has not been used for the estimation of aquifer
parameters for the application example. However, one can use it
Sushil K. Singh1 for estimating the aquifer parameters by curve matching. It has
1
Scientist, National Institute of Hydrology, Roorkee-247 667,
the advantage of analyzing short duration pumping-test data.
Uttarakhand, India. E-mail: sukusi1@yahoo.com
However, a set of semilogarithmic diagnostic curves for estimat-
ing the aquifer parameters from shorter duration pumping-test
The writer is thankful to the discussers for raising certain points, data has been elsewhere proposed by the writer see Singh
which are clarified below: 2008a. Using the Singh 2008a approach, it can be verified that
the synthetic data used by Wikramaratna 1985 for the verifica-
tion of his method do not pertain to the ideal conditions for which
Points by Sen his curves were developed.
It is the capability of a particular method to ascertain how long
When analyzing the pumping-test data for aquifer parameters, a pumping duration should be in order to accurately estimate the
types of well, aquifer, and pumping should be considered in using aquifer parameters. Most field applications are generally based on
an appropriate analytical model that practically satisfies the field the Papadopulos and Cooper 1967 curve matching method. With
conditions. It is not appropriate to use an analytical model, which the availability of the writers method, a comparatively shorter
does not practically satisfy the types of well, aquifer, and pump- duration pumping-test data can be used to accurately estimate the
ing. The writer has proposed computationally simple equations aquifer parameters. The 50 min data used in the first application
for estimating the aquifer parameters. The types of well, aquifer, example show a substantial contribution from the aquifer storage
and pumping that have been considered are a large-diameter to the pumped discharge as the hypothetical drawdowns, for no
well, a confined aquifer, and a constant discharge pumping, re- contribution from the aquifer storage is greater than the respective
spectively. The phrase Simple Equations in the title and else- observed drawdowns. This is clear in Table 1 of this closure. The
where is meant to convey handy and computationally simple. aquifer contribution to the pumped discharge increases with time,
Eqs. 47 are derived from Eqs. 13; hence, both sets of and at the end of 50 min, 30% of the pumped volume of water is
equations bear the same physical meanings for the parameters and from the aquifer storage. To be clear, nearly fall on should be
variables. Eqs. 813 have been proposed for the explicit esti- appropriately read as seem to nearly fall on.
mation of the aquifer parameters from drawdowns in a large- The curve matching method of Papadopulos and Cooper
diameter well, which are computationally simple calculations 1967 cannot yield the estimates of the aquifer parameters from
can be performed even on a handheld calculator. Eqs. 13, only 4 h data as considered in the second application example,
proposed by Papadopulos and Cooper 1967, do not explicitly hence no comparison was made. The reason the aquifer param-
yield the estimates for the aquifer parameters and hence cannot be eters were not estimated using the Papadopulos and Cooper

128 / JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING ASCE / JANUARY/FEBRUARY 2009

Downloaded 21 Apr 2009 to 210.212.58.170. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Table 1. Comparison of Hypothetical Drawdowns for No Aquifer
Contribution with Observed Drawdowns k1 = ln
t1
t2
ln
t2
t3
ln
t3
t1
5
Hypothetical drawdowns Observed
Time with no aquifer drawdowns Eqs. 15 were obtained by fitting a parabola through the
min contribution m m points of the triplet. Two triplets containing the peak are
0.50 0.032 0.030 possible. Using Eqs. 15, one set of coordinates of the
1.25 0.080 0.072 peak is obtained for each triplet. If the coordinates of the
2.50 0.159 0.136
peak in the case of both triplets differ moderately, the aver-
age of the two should be accepted or a smooth curve should
5.00 0.318 0.245
be sketched and the smoothed coordinates of the triplet
12.50 0.796 0.484
points should be considered or a new triplet should be con-
25.00 1.529 0.723
sidered after deleting a outlier point. The above proposed
50.00 3.183 0.960 objective method of determining the peak of a unimodal
curve can also be used for the method proposed by Singh
1967 method is the same as in the case of the first application 2000.
example. 3. Although Eqs. 811 are developed using the data for
101 105 the sign of inequality in the original note
should be read with correction, these are also applicable for
Points by Cimen 101 and 105. Possible care to this effect has been
taken while developing the equations.
1. Using Eq. 8, u p = 0.001052 is obtained for = 0.01. The
value of well function for a large-diameter well is less than
References
that of the small-diameter well for this value of u p. This
indicates that only a part of the pumped discharge is from
Papadopulos, I. S., and Cooper, H. H. 1967. Drawdown in a well of
well storage and the remaining part is from aquifer storage.
large diameter. Water Resour. Res., 31, 241244.
Therefore, there is no logical problem in applying the writ- Singh, S. K. 2000. Simple method for confined-aquifer parameter es-
ers method for estimating the aquifer parameters. Applica- timation. J. Irrig. Drain. Eng., 1266, 404407.
tion of the writers method has nothing to do with the time at Singh, S. K. 2001. Confined aquifer parameters from temporal deriva-
which the Papadopulos and Cooper 1967 curves merge tive of drawdowns. J. Hydraul. Eng., 1276, 466470.
with the Theis 1935 curve; the writers method can analyze Singh, S. K. 2003. Storage coefficient and transmissivity from residual
much shorter duration data. drawdowns. J. Hydraul. Eng., 1298, 637644; Errata. 12912,
2. A short duration pumping-test data cannot yield accurate 1024.
estimates of the aquifer parameters is a conception devel- Singh, S. K. 2007. Approximation of well function for large diameter
oped because the appropriate methods were not available to wells. J. Irrig. Drain. Eng., 1334, 414416.
Singh, S. K. 2008a. Estimating aquifer parameters from early
analyze the short duration pumping-test data. Such appropri-
drawdowns in large-diameter wells. J. Irrig. Drain. Eng., 1343,
ate methods for small-diameter wells have been elsewhere 409413.
proposed by the writer see Singh 2000, 2001, 2003, 2008b. Singh, S. K. 2008b. Diagnostic curve for confined aquifer parameters
Determination of the peak and time to peak may involve a from early drawdowns. J. Irrig. Drain. Eng., 1344, 515520.
little subjectivity if one needs to sketch a smooth curve Theis, C. V. 1935. The relation between the lowering of the piezomet-
through few but sufficient number of selected points contain- ric surface and the rate and duration of discharge of a well using
ing the peak. Graphic software can be used to fit a polyno- groundwater storage. Trans., Am. Geophys. Union, 16, 519524.
mial of adequate degree through these points to reduce the Wikramaratna, R. S. 1985. A new type curve method for the analysis
subjectivity. Then the coordinates of the peak are determined. of pumping tests in large diameter wells. Water Resour. Res., 212,
A logarithmic time-axis is preferable. How to accurately read 261264.
the coordinates for a logarithmic scale is given by Singh
2008b. However, to approximately determine the peak and
time to peak of s2w / t on natural scale versus t on logarith-
mic scale curve in a completely objective manner, a triplet
Discussion of Flooding Probability
with the coordinates of points as x1 , t1, x1 , t1, and x1 , t1
containing the peak is selected and the following equations Constrained Optimal Design of Trapezoidal
are used: Channels by Amlan Das
January/February 2007, Vol. 133, No. 1, pp. 5360.
k1
t p = t1 + 1 DOI: 10.1061/ASCE0733-94372007133:153
2k2
Rajib Kumar Bhattacharjya, Aff.M.ASCE1; and
s2wp k21
= x1 + 2 M. G. Satish2
1
tp 4k1k2 Asst. Prof., Dept. of Civil Engineering, National Inst. of Technology,


Silchar 788010, Assam, India. E-mail: rajibkbc@gmail.com
2 2 2
t2 t3 Prof., Dept. of Civil and Resources Engineering, Dalhousie Univ., Hali-
k1 = x2 x1 ln x3 x1 ln 3 fax, NS, Canada B3J1Z1. E-mail: Mysore.Satish@dal.ca
t1 t1

k2 = x2 x1ln t2
t1
x3 x1ln
t3
t1
4 The author of the paper developed an optimization-based model
for optimal design of trapezoidal channel section incorporating

JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING ASCE / JANUARY/FEBRUARY 2009 / 129

Downloaded 21 Apr 2009 to 210.212.58.170. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
nated solutions and will not find any position on the Pareto-
optimal front.
Eqs. 2527 of the technical paper constitute a multiple ob-
jective formulation as a tradeoff that exists between objective one
Eq. 25, i.e. minimization of channel cost, with objective two
Eq. 26, i.e., minimization of flooding probability. The author
used a classical optimization technique to generate the Pareto-
optimal solutions. Most of the classical methods transform the
multiple-objectives optimization problem into a single-objective
optimization problem, and then apply single-objective classical
optimization techniques to solve the transformed problem. This
type of classical optimization technique gives only one Pareto-
optimal solution at a time. The process is then repeated to gener-
ate other points on the Pareto-optimal front. There are several
methods available to convert a multiple-objectives problem into
single-objective problems. The author did not discuss in details
Fig. 1. Feasible boundary and Pareto-optimal front for composite the method employed in his study for conversion of the multiple-
roughness case objectives problem into a single-objective problem. The
-constrained method can be used to convert the multiple-
objectives problem into a single-objective problem Deb 2001.
flooding probability as an additional constraint. The discussers The transformed single-objective formulation in this case may be
would like to comment on some related issues on the technical written as:
paper that would help the readers to better understand the authors
Minimize J1 = c1A f + c2 P1f + c3 P2f + c4 P3f 1
work.
Figs. 2 and 3 of the technical paper show the tradeoff between
the total construction cost of a trapezoidal open channel and Subject to Py y + f 2
flooding probability of the channel for composite and uniform
roughness scenario respectively. It can be observed that for both Q A5/3
=0 3
scenarios, the total cost of the channel decreases to a minimum So n3/2
i P i
2/3
value and then increases along with the increase in flooding prob-
ability value. Intuitively, the total cost of the channel should de- For different values of , the single-objective optimization prob-
crease along with the increases in flooding probability values and lem may be solved using classical optimization techniques. The
should reach the minimum value at a deterministic optimum discussers solved the formulated optimization problem in Eqs.
channel section. And then it should maintain total cost of the 13 for composite roughness and uniform roughness sce-
deterministic optimal channel section value with further increase narios. The sequential quadratic programming routine available in
in flooding probability. The total cost of the channel reaches a MATLAB is used to solve the single-objective optimization prob-
minimum at flooding probability equal to 0.3693 for the compos- lem. The feasible boundary obtained by solving the optimization
ite roughness case and at 0.3827 for the uniform roughness case. problem is shown in Fig. 1 for composite roughness case and in
The corresponding costs of the channel are 22.935 and 24.963 in Fig. 2 for uniform roughness case. It can be observed that the
currency units respectively. For further increase in flooding prob- total cost of the channel decreases along with the increase in
ability value, the total cost of the channel should remain the same flooding probability value. The total cost reaches the minimum
at 22.935 for the composite roughness case and at 24.936 for the value at flooding probability equal to 0.3693 for composite rough-
uniform roughness case. However, these solutions will be domi- ness case and at 0.3827 for uniform roughness case. With a fur-
ther increase in flooding probability value, the total construction
costs of the channel remain unchanged. However, the solutions
for a flooding probability value greater than 0.3693 for the com-
posite roughness case and 0.3827 for the uniform roughness case
are not the member of the Pareto-optimal front. The Pareto-
optimal solutions can be obtained using the non-dominating sort-
ing algorithm Deb 2001. The Pareto-optimal fronts are also
shown in Fig. 1 for a composite roughness case and in Fig. 2 for
a uniform roughness case.
Eqs. 14 and 15 of the technical paper are the simplified
form of Eqs.2 and 12 when n1 = n2 = n3. In this scenario, as per
the notation used in the technical paper, Eqs. 14 and 15 should
have been
SoA5/3 SoA5/3
Q= or Q= 4
n1P1 + P2 + P3 2/3
n1P2/3
or

Fig. 2. Feasible boundary and Pareto-optimal front for uniform


roughness case
5 dA

2 d
A dy P dy
P=
+
2

+
2
3 SQ2 Sn1 Sso
Sy Q2 n21 4S2o
1/2

130 / JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING ASCE / JANUARY/FEBRUARY 2009

Downloaded 21 Apr 2009 to 210.212.58.170. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
References deterministic optimal-channel section value with further increase
in flooding probability. This statement of the discussers opposes
Deb, K. 2001. Multi-objective optimization using evolutionary algo- the existence of the flooding probability constraint in that region,
rithms, Wiley, New York. which is unacceptable. The results from the writer show that there
is no conflict between the two objectives, although flooding prob-
ability constrained optimal design is viable in the region.
Moreover, the discussers solution results contain incomplete
information that make it redundant and improper on the part of
Closure to Flooding Probability
the writer to give a closing comment on the particular issue of the
Constrained Optimal Design of Trapezoidal discussers.
Channels by Amlan Das Regarding methodology of conversion of the multiobjective
January/February 2007, Vol. 133, No. 1, pp. 5360. problem to a single-objective optimization problem, an average
DOI: 10.1061/ASCE0733-94372007133:153 reader of ASCE journal paper must access the references cited in
the paper to learn the methodology. Also, the methodology has
Amlan Das1 become a standard practice because of its wide use in research.
1
Professor, Dept. of Civil Engineering, National Institute of Technology, Therefore, the short description given in the writers original
Durgapur, Mahatma Grandhi Ave., West Bengal 713209, India.
paper suffices the need.
E-mail: adas_wrpm@yahoo.com
The discussers -constrained method and the writers con-
strained methods of conversion of the multiobjective problem to a
The discussers deserve an acknowledgment for taking interest in single-objective optimization problem are identical. They have
the writers work. The discussers interpretations of Pareto- identical drawbacks and advantages.
optimal solutions of Figs. 2 and 3 may mislead a common reader. Regarding Eqs. 3 and 4 from the discussers, the explana-
They stated that the total cost should maintain total cost of the tions given in the writers original work are self-explanatory.

JOURNAL OF IRRIGATION AND DRAINAGE ENGINEERING ASCE / JANUARY/FEBRUARY 2009 / 131

Downloaded 21 Apr 2009 to 210.212.58.170. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

Вам также может понравиться