A STUDY OF THE FLOW'OF WATER
IN THE SETTLING BASINS OF THE
DETROIT WATER WORKS.

ITS EFFICIENCY AND IMPROVEMENT

Thesis for the Degree of B. 5.
Harry S. Aten
I 9 2 7

 

 

.w-

IIIIIIIII

Il.|..l\‘rIITIIA

 

 

 

I .. 4 .134 .
e « E. . . .u I I n . The
Fluffurblr r .L-Iunrhvl . .. _ . .L- rfik. . r .u r I .~ I .y

III. IL“?! l.ILIf.;.II:|I I‘thlIhlll. .... 4 1 . . nurnh E1 .1! 1 ihfihnflififlulu.

   

A STUDY OF THE FLOW OF WATER IN THE SETTLIITG BASIITS
OF THE DETROIT WATER WORKS. ITS EFFICIENCY AND
IflPROVEEENTo

A THESIS

submitted to the faculty of
the Michigan State College of
Agriculture md Applied Science.

By

Harry S. Aten
Omdidnte
for the degree of
Bachelor of science
June 1927

I wish to take this Opportunity to
express my great appreciation and to them:
the members of the mgneering Department
of the Dgtroit Water Works for their generous
assistance and hearty cooPeration given me

while collecting the data for this thesis.

I also wish to thank professor C. L. Allen
and professor H. 0. Woods for their part in

assisting me in this work.

93798

A STUDY OF THE FLOW or ‘cm'ma IN n12 gramme BASINS
OF 'IHE DETROIT WATER WON-ZS. ITS BFFICITCI‘ICY mm
IIZPRovmm-IT.

PROBLBX

It has been found from experiments on the settling
basins 0f the Detroit water works that the I'detention
tima' is a very poor measure of the actual conditions
in the tanks. It has been found, for example, that with
a theoretical detention time of a little over three hours
some of the particles of any certain charge of water find
their way through in less than fifty minutes while others
take six hours or more to get through, with the larger
portion of the charge passing through in from one to three
hours.
. Theoretically every particle of water should pass
through in the me length of time as every other particle,
that is, suppose that a certain quantity of water he
charged into the inlet of the basin in a mass, 'Jl'hat quantity
should pass through the basin and out of it in the same
mass with each particle in the same general relation to
its fellows as at the begimling. A basin with this
condition of flow would be an ideal basin. Obviously,
however, this condition cannot be fully attained but it
can be quite closely approximated.

If it were pessible to attain a complete lack of

.
0
.
‘ I
.0
h:
.\
O
. .
l
a .
V U
o g r. I
Q
r..
'h
b
:7
V ..
V s . n I u
a O. r-. II“
m .
v t
' .
. . a
o O
.1 t
v to. q
l .
O a
G
0..
0‘

7

CI

 

-2-

dis persion in a tank the ideal or perfect flow in the
tank would be accomplished. With a straight rectangular
tank open at both ends, the ideal flow tould mean that
the velocity at any point would be the the same as at
any other point and that the water entering in a vertical
plane at one and would pass out at the other as the same
vertical plane. However, with the water flowing in a curved
path. as it does at the Detroit plant, it will be necessary
that the velocities at some points be greater than at
Others to cause the water to flow in and out in the same
relation. A numerical measure of this dispersion 1n the
tanks would be a fair indication of the efficiency of the
basin. The smaller the dispersion the higher the effeciency.
the problan, then, is to find a measure for this
dispersion and then put such an arrangement of baffles,
screens, or vanes in the basin as will make this dispersion
as mall as possible.

gamne or DISPERSION
In this discussion dispersion is taken to mean

what is commonly termed short circuiting. It is believed

by some that this is a better term for it. the word

dispersion, however, would include not only short circuiting;

but also long circuiting and so would indicate the whole

action or the water in the tank and not Just a part of it.

Short circuiting is mereaa portion of the water takes a

more direct route from the in let to the outlet at a much

-3...

higher velocity than the theoretical. Long circuiting,
then, must be taken to mean Just the Opposite, or where
a portion of the water stays in an eddy for some time or

flows in a circuituous route at a low velocity.

EFFECT pr marathon

Experiments conducted by Imhoff on set-rage settling
basins show that the effect of the detention time on the
clearness of the effluent diminishes progressively with
the time. that is, the first unit of time is more effective
than the second and the second is more effective than the
third and so on. Now if the dispersion in the tank is great
it is apparent that the tater that passes through in a
mart Mails has less settled parts per millon than that
which is retained a longer time but as all the water
flows out of the same outlet the water of less clearness
in mixed with that of greater clearness and the effluent
has a quality of an effluent from a tank where there is
no dispersion and a detention time somewhat less than the
theoretical period of the tank with the greater dispersion.
Therefore, a tank devoid of dispersion would give a much
clearer effluent than one having the same detention time

and great dispersion.

NEED 011A 11mm or DISPERSIOE

 

 

It would appear then that dispersion lowers the
effeciency of a settling basin. If this is true, then a

 

-4...

measure of this dispersion would be important in
indicating the action or characteristics of the basin.

In the words of Mr. horrill of the Detroit Water Works,

a measure of the dispersion in a settling basin is a
dimension, nearly, if not quite as important as the actual
films of the basin. To give the nominal time of a settling
basin without giving a measure of its dispersion is like
giving the effective size of a filter sand without giving
its unifomity coefficient.

FLOATS

Floats were at first used to determine the dispersion
in the tanks. These floats were of two types, namely,
surface floats and subsurface floats. By placing these
floats at various points and watching their respective
courses, a good general idea of the currents and eddies
in the tanks was obtained. However, floats started at the
same point but at different times behaved entirely
different from each other. This showed that the eddies
were ever changing md no representative indications
could be Obtained. some floats would pass from inlet to
outlet in a fairly direct path while others would get
into a large slow eddy and at times would one and at times

would move in the Opposite directions, still others would
get in a corner or along the wall and remain there.

Sometimes the paths would orOss each other thus indicating

that the water flowed now in one direction and then in

another.

When the floats paths were plotted up they proved
to be merely a maze of lines from which no animate
conclusions of value could be drawn. Other than showing
that there was dispersion and that the flow in the tanks

was very poor, the floats proved useless.

comm WATIili

Galoring matter was next used as an indicator.
The water was dosed with the colored luquid at the inlet
and the color front of the flow was carefully plotted at
one minute intervals. A model basin similar to the one
shown in fig. 1 was used for these and succeeding experiments.
Wires were stretched across the basin parallel to all four
sides at one foot intervals. A man standing above this
model with a chart could easily draw in the contours. A
sample of these contours is shown in fig. l. Mud was at
first used as a coloring matter but it settled out too
quickly and it also necessitated cleaning the basin after
each run. Uranine was then tried with mch better success.
This is of such a nature that it will difuse quite
thoroughly with the water. A small quantity poured in to
the water at the inlet would color it sufficiently to
enable the observer to follow the course of the water
around the talk with ease. However, where there was a

reversal or flow the colored water flowing baclmard would

Suchcf- MODEL BASIN Gd/m. ,w-Mlbm “579 £32,205, [6%
Adz/(($770, fir (1‘4"?er

 

“/37

2.3

 

 

 

 

2a?

21 \

20 ‘

19 7 \ J /

n Y/ ,
/J

 

 

 

 

 

 

 

 

I7 !

 

[6 /

lé’

 

 

 

 

 

 

 

I4-

\
,3/
/

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

~ hhl-t‘qlovcoarli}

 

C’l ‘lj/l/d / A?

54%" l
0 [I a? 3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mix with the other and no indications could be obtained.

this method gave sufficient indications of the
flow to allow experiments to be carried on using different
arrangements of screens, vanes, and baffles. Ito mmerioal
data could be obtained from this method that would in any
way give a measure of the dispersion or of the effeciency
of the basin. Quite a nunber of experiments were carried
on by this method using different arrangements in the .
tank but as a better method of indicating the dispersion
was later devised the results of these ez-zporiments will
not be given. This method did prove valublo, however, in
adjusting the flow in the model basin to make the conditions
of flow in it to that in the large basin.

THE SALT TEST

The salt test is the same in principle as the
color test with this exception, it lends itself quite
readily to chemical and mathematical analysis. To make
this test a mixing chamber is required in which to mix up
the salt caution and some means is necessary whereby the
solution may be dosed into the water in as short a. time
as possible. Accordingly a tank was constructed large
among) to hold a quantity of water sufficient to disolve
about 4% pound of salt for use with the model basin. An
outlet was provided that would empty the tank in approximately
one minute. By thking qualitative analyses of the effluent

 

.. 7 ..
at one minute intervals from the time of dosing to the time
when no more excess salt appears and plotting these results
on ordinary gaph paper, some will be obtained that
closely resmbles a normal distribution curve. The data
thus obtained will be in parts per million per minute of
salt.

{the method used to make the quantative analysis of
the effluent was to measure the resistances at the outlet
by means of an Ohmeter. 'Ihis consisted essentially of a
wheatstone bridge and a cell containing two platinized
electrodes. 'lhis instrument measures the resistance of the
liquid in ohms. To use this data it must be changed into
p.p.m. of salt mich necessitated calibrating the instrummt
to read in p.p.m. In calibrating the instrument it was
found that temperature change effected the readings materially.
A change of one tenth of a degree centegrade being sufficient
to change the resistance of the solution about '7 ohms.
With the temperature canstant, a change of one ohm indicates
a change of about .2 p.p.m. of salt. Accordingly it is
necessary from the point of accuracy to read temperatures
to the hundredth of a degree. The cell should be placed in
such a place in the outlet as to secure the most representative
readings.

If each p.p.m. of salt be considered as an individual

phmcmn with a certain number occurring or passing out
in each unit of time, it will be seen that they truly
represent a frequency distribution . When plotted as a

 

frequency series the curve will confom more or less closely
to the curve of normal distribution, depending on the
effeciency of the basin. If the data be considered as a
grouped frequmcy series, it will be seen that a certain
number of molecules of salt will appear a certain number
of mimtes after the dosing, a certain rumber in the next
minute, and a certain number in each minute thereafter
until the completion of the run when no more excess salt
appears. the high point of the curve, which is the point
where the molecules of salt appear most frequently, should
be somewhere near the nominal time of the basin. It should
be evident that the sieser the high point of the curve is
to the nominal time, the more effecient the basin would be.
Fig. ,2a shows a curve plotted from data obtained from a
salt run on the original basin. Fig. 2b shows a curve
obtained from the model basin with the curved vanes in
place. giving a much better flow. This indicates the same
thing that the color did, namely, that there is dispersion
and eddying in the basin. Withgreat dispersion the curve
is irregular, rather flat and drawn out, and the center
of gravity does not coincide very closely with the nominal
time of the basin. At the time of writing this thesis the
curved vanes had not yet been installed in the original
basin so that no check on the results could be obtained.
If the basin had a complete lack of dispersion, the entire

.. 9 ..
quantity of salt would come out of the outlet at or very
close to the nominal time of the basin md the curve
would rise from zero a minute or two before this time to
a considerable height at the time and return to zero I
immediately after it. 'the center of gravity of this curve
would coincide with the nominal time of the basin.

gomnpmrpr DISPERSIOIL

Now if a numerical value could be obtained to
indicate the difference between the center of gravity of
the curve obtained frOm plotting the data collected from
a salt run and the center of gravity of the ideal curve,
or, in other words, the line through the theoretical time
of the basin, it would be a fair measure of the dispersion
in the basin. If dispersion be defined as that preperty
of a series by which the several variates tend to differ
in value from the average, it will be found that two series
with identical means may have entirely different dispersions.
There are several methods used in statistical work to
measure this dispersion. Of these the average deviation
is probably the best adapted to the present needs.

The average deviation is the mean of the absolute
(without regard to plus and minus signs) deviations of the
several variates from the median. In the following mrk it
will be assmned sufficiently accurate to measure the dev-
iations from the mean rather than from the median because
of the somewhat greater facility of measuring the deviations
from the mean. If the frequency be denoted by(f) and the

 

 

time by x , thanfo ’21?“ g H, the mean time. Now let r 2:
percent excess of salt, then fr will be the total percent
of excess salt. Let t a the theoretical time and ”o 3 time
in minutes from the beginning of the run. then (1: - t6)-
the deviation from the mean for that minute. The average
deviation, then, will be the mation of the product of
each percentage and its deviation divided by the total

percentage, or, emrossed as a formula, A.D. Mt - tAj.
SP a"

 

the coefficient of dispersion may now be defined as 100 - K,
where K is taken as the average deviation divided by the
average mean time and expressed in percent. On the following

page will be found an example illustrating this principle.

mICRII'IHITS

Following this discussion is a mammary of the
ezmeriments that has been conducted in the atta'npt to
improve the conditions of flow in the sedimentation basins
at Detroit. The poor arrangements are given as well as the
good with the idea of showing the effect of the various
devices used. Many experiments were run with floats and
others with coloring but the results thus obtained did

not yeald much information and they will not be discussed.

Design of Vanes

 

'ihe arrangement shown in fig. 21 is the one that

- 11 n

was chosen to be used in the large basin. The vanes as

designed are to be made of two sheets of iron bolted to

channel irons thus forming a double well. At the upstream

and of the vanes are placed moveabls wings so that adjust-

ment can be easily made for varying rates of flow.

The

exact position of these vanes in the large basin will be

determined largely be experiment. It is unfrotmzate that

time does not permit of the actual results being given in

118 thesis.

Kenna

 

ills following is an example of the method of computing

the coefficient of dispersion.

x f r .2Fr t - t r t - t fx
Time in p.p.m. Excess ( (9 ( 0)

min. NaCl

1 70.8 0.0
3 71.8 1.0 1.10 1.10 5.8 6.18 3.0
4 73.5 2.7 3.00 4.10 4.8 14.40 10.8
5 76.0 5.2 5.77 9.87 3.8 21. 2 26.0
6 79.5 8.7 9.66 19.53 2.8 27.06 52.2
7 82.8 18.0 13.34 32.87 1.8 24.00 84.0
8 80.6 15.8 17.57 50.44 0.8 14.06 126.4
9 83.4 2.6 14.00 64.44 0.2 2.80 113.4
10 80.8 10.0 11.00 75.44 1.2 13.20 100.0
11 78.5 7.7 8.56 84.00 2.2 18.83 84.7
12 75.9 5.1 5.66 89.66 3.2 18.10 61.2
15 7405 505 3089 93055 4.3 16038 4505
14 73.1 2.3 2.56 96.11 5.2 13.34 32.2
15 78.7 1.9 2.11 98.82 6.2 155.06 28.5
16 72.0 1.2 1.33 99.55 7.2 9.58 19.2
17 71.1 0.3 0.33 99.88 8.2 2.46 5.1
18 70.9 0.1 0.11 99.99 9.2 0.92 1.8
19 70.8 0.0 --~- --—- 19.2 ---- ---~
20 70.8 0.0 ---- ----- 11.2 ---- ---—-
9 0 . 0 100 . 00?! T7673 '79 II. 0

 

-12..

1.168.113fo : 794.0 4- 8.8 minutes.
ET. W

A.D. .zzrgt - 3Q]. 216.23 . 2.1623

1'

K 3 _ A.D.*4_ a 3.162; é—‘f .246
“Mean 8.8

 

coefficient of diaspersion . 100 - (0.246 x 100) g 75.4%

a...
.09u'qu;
b-vos‘».O4—0

a

.,’ .

>9- .*o+‘6
.‘ oil.

MKHflGAN

-o-.“~‘4o

. .-.-'

..¢+L-o O 6 ‘

-.b¢—.L. n
9.-

STATE COLLEGE

. .6...oo
.45.’¢<.é—-
..-o.+¢—.-¢~

.L.-&......
. . -.

-n--.

“—j \m‘ 'v

|—.-

4-s—o >4;- .

  
 

USPAH I Huh!

4

 

O

U" ”A! h..flll“

 

. "—~’.~ _‘

:30

700

.— -9-.. —-—.—. .-

mcmGAN CTATI COLLEGK

 

00-...
--m+«+
90—fHQO

Ho—o
~A~o~o¢¢o 9+” “0
0.0.0.0.. po—oo 4o-

>5H0Q6- 9-0-9.-.
-oo—Ocoo... “-4---
. 0‘7. 0’ , o

H‘o—OO-

5.--- .9

- ....
—--‘H~ -
~~9oo~v~
9“A-A-.

o-->.oo---
o—-.‘o4..—.

p... 09“ .

roononq‘

H b. oo-owo

H

~¢O*¢ o

-$ o—o
‘4‘
co-o ‘

 

-MW

 

ULPAN '4'.“ ur no.

0

 

naMAlug
N

60

J0

20

10

O

 

 

In figure 1’- is shown the /
‘asin with no iupefliment
except the corner baffles. I
The arrows indicate the fi
direction of flow and the
eddies.

The accompanying graph |

shows the result of the

salt test. It is realily

 

seen from this graph that
ml

 

 

 

 

4U?

 

 

 

J

0 TI me I'D Minute: I 5 0

this arrangement of the basin is a very poor one. The yeal
of the curve occurs much too soon and the lonj time that
it takes the curve t0 come back to normal indicat€s

that there is considerable flspersion in the Latin. The

coefficient of efficiency was not figured for th

srwl

H'

because the graph alone is sufficient to show thr‘ t

arran‘ement is undesirable.

 

 

O'K‘r? f, . g
.1 .Ir,‘.rgra
,0 +- K.

‘1" to :u V

‘ ” S l
‘ " not
'4?! n.3,: 3- i . V V.“ or:

.
I 'Q

.1“ i‘
,4. 0‘:

‘.o t a"

 

1.;

s 1
J. L-.
«1
J.

 

-1n_

 

1 ‘ vault *1 a? ,
(e 019...: 1‘1":_,(_‘:' x f
‘ C" wr' 'o- \th '

17 --.';‘,- .L
- J L

n 4 ‘- '
(-1. 11; “2.: ' 3

 

 

 

 

 

A 1

77076 in M'n-c. "fps

‘ A n . _

- (0.50"

 

 

 

20

 

-17-

fiig. F is a sketch of the nofi/’

model basin showing screens
and a curved vane in place.
below is the curve plotted
from a salt run with this
arrangement. Corner baffles
and the original baffle were
also in place.

This curve shows that
this arrangement is somewhat

bett~r ‘han the previous

 

20 4b 77179.9 fr; )7 Ifidftkf

 

 

 

 

 

 

’3‘

ones but is quite spread out and very irregular.
The total flow was 122 gal. at the rate of 58 gal.

per minute divi ed flow.

\./

 

 

 

In fig. 6 is shown nether / ”MN/1
\s
1'1 f‘.‘.‘., 7" Voile
am ..n.,,ment 0.. screens, baffle 3 Jo 6,,
and vane. r
:14MeJ/7

The curve lotted from this flats

is ler‘s irregular than the
previous ones but the peak
cones rather early and it is

Ion;- dravm out on the end

 

 

 

 

 

' K-JOMPJ’!
indicating that there is still
Or; [ha/H
quite a lot of dispersion Jam/e Z...
26$ PT I,
30.
15»
10 »
5 .
O : -— - f 4 :

 

 

 

I0 :0 .30 -;0 3‘0 60 7'0 \50 so zoo
Time I"? ”mun-4
and that "here is a strong direct current that carries

most of the water through in a short while.

 

he arrangement shown in

 

 

 

 

 

Thnowk7whnufe:

14Mealr
51;. V gives but little better vane
{W JOMOJ’?

r ~sultr than that in fig. 6.

The curve is a. little smoother

ant comes to a narrower peak.

Other than this it has the

sens chartcteristies as the

preceding one.

0’49,ng
Jaw"
L’; .9.
I:
1"ng 7
lo
16‘
Q. I0.
1‘
J.
" IO 40 J0 4a a 60 70 so so ILO

 

 

- 20 -

, Corner-841?“: 3

 

 

 

 

 

 

 

/-‘J .
The arrrmgenent shown in P1,".
shwwash
. is not much better than the
original baffle alone. The fined.
czzrv: is fairly smooth but
t returns to normal Very
slowly indicating very great
dispersion.
the original baffle was
left: out and 14 and m mesh
45.14"er
screens were placed around Ii“ '44-
l‘fir l‘JOfiféU/z
Fig. 8
10.
I!»
19..
Q
J»
0 :3 i0 i0 4:0 '3‘? Io 7? 8‘0 9'0 ’1"

Tu'ne in Ml'nufes

-t.

376 inlet at a radius 03 approximately one uni end half

I4)

’Z‘Ot o

v-0 .1" ‘Wj'j ; 'r'"‘;-P—~——-' rl-fi-g ~

 
  
  
  
  
   
  
  
   
  

-21..

 

njthis arrangement seven K
gmqrfianes were used at the inlet
fli© deflect the water over 30 M,”
{sheenfiire Width of the pass.
A mass}! screen was also used at the
m1::..T-hese gave the most
WEEK? flow of any combination
refitrted as is evidenced by
mam-en? uniform curve.

.' 7 Olga:
Van's

 

 

 

 

go Jo 40 JD ‘ ca 70' .90 8.0 , (aq-

 

T1)": in Mimic?! '_

 

-22~

 

 

 

 

 

 

 

 

f’ere (3 tin vanes were used \
.. 6-1 f". . _
ith tl-e OI‘i_.lI‘81 baffle in rlade. "new!
whose vanes were curved and so placed / 30mm.
as to distribute the flow as
uniformly as possible over
the entire passage way. A 14
mas}; and a 39 mesh screen
were used at the turn as shown.
This arrangement gave much. the ,
W's!
0
same curve as the glass vanes. I;
25' ‘andnes
Fig. 10
20 +
1:.
2‘ .
0:10
o;
.S' r
O "5 20 3o 40 a 4:0 50 .50 8‘0 7;.

771:": In Mina f0: .

'i'hr‘, tin Van-vs have the advantage over Lhe glass of bein;

flexible and easier to adjust.

-23..

  
  
  
  
   
  
  
  
   
   
  
  
   

 

 

 

 

 

 

"This is the same arrangement Joan/c
as that on the previous page nh'o'f/fcfpr
with the exception that there ’{lj‘fifl’l’
{is a curved tin vane for a I
deflector at the and of the
sorter baffle. The deflector
makes very little difference
in the appearance of the curve.
'L‘his same arrangement was .. /
tried without the 14 mesh Zia,"
4.5T Egg/e4
I Fig. 11
1 - - h
‘0 n so so mo

 

7701c In M'nu're:
screen and with the 30 mesh run all the way across. The
only effect tha' this had was to spread out +he curve

a. 11 ttho

 

 

 

I[his set up is mantle.“ 3‘ ”a”
that shown in fig. 11 with the 17}. m. >/
addition of 27 - g inch V743.“
couplings on the bottom at
the center line of each pass
transversly. ...... :3”...
These give practically ‘lhe {7.}a¢6%/
same results as the two

previousX arrangements.

 

 

 

 

 

 

orélh.’
This was also tried with 32 """'
4/4.
134* .3}: 77;: Maine:
10 "
I"
‘C
m [OI
Q
J»
O ’3 L 14 J0 ~70 Jo so 70 0:0 .6. ’70

Time in Minutes
1 inch couplings on each side with very little change
in results. The distribution effeciency was computed by

the method perviously discussed and was found to be 81.95"

 

-25-

 

This combination of screens
curved vane at the end of the

center baffle, couplings, and

 

25 small tin vanes at the inlet
gave very poor results. It
seems that the vanes at the ............ .
argyzdhmqy.,7'
inlet were too small and
difficult to adjust. The curve
indicates that there was a

strong direct current from the

 

 

 

 

 

 

 

.1: V411?!
inlet to the outlet and that .u4% #_
I! r
Fig. 15

to r
t.IJ’ r
gm
fr

9 lb .10 a; 4:0 fa £0 73 50 30 ’00

Time in lfinutes

eddi es were numerous.

-26-

- '

   
 
  
  
 
  
  
   

. .

 

'79" .Qt'illin: baffle shown '
_ v . . ,4 [‘1er

    

14 was drilled with
we. ' 7- aem far
5'" 11?],(38. COI‘ICS were fitted ”' aw‘
"4 'f" M- 93/:
mgtgipse holes 30 that any
11:66. number of them could
)9 .{Elosed With all of them
s 5.21 _

min ‘the teat gave a fair

“gorv‘e but not as good as some

 

 

 

 

 

 

 

 

. oer arrangements previously
Bil-35% g :hwm! Burk
- {'9 hung}
Fig. .14
’5
Vic .59 J0 4'9 3'0 0'9 A 79 9'9. .9: ”.7 . -

Time in Iflnutes

 

/4 ”(Jkfl

 

Here the stilling baffle
was used with 16? holes and

tin deflectors were placed as

 

51} :ovm .

This gave a very irregular
curve indicating that eddies
were prevalent and that currentd
were ever changing.

The arrangement is not good.

 

 

 

 

 

 

 

 

 

”II/’5, 84”/?
I 6.9-3 ' f)
at
F1 5 o 1 5

2 0 r

1.9"
Q/o

Q'

4" »

J
‘ I 0 1; 4'0 140’ 4:» 60 7‘} JO :0 1:;

Time in itinutes

This set up was tried with the deflector marked 1

removed and a 14 mesh screen placed as shown by the dotted

line. This gave practically no change except to smooth out

\1

+he curve a little.

 

 

-23-

The arrangement shown in fig
It: seems to give the best
results of any yet tried.
The curve is fairly smooth
showing that eddies are few
and small and that the larger
portion of the dose passes
through in abuut 30 minutes.
With an overflow weir at
the outlet as shown by the

(Ir
20*

15‘-

 

‘ A
v v v v

JO

 

«1‘0

 

. '1‘ ‘ 4
TWLDeflhdUw;/5f"
/ ,4 Me..-
I f :4 ~11

SfingEhfiW'

 

I“- g'lule:

 

 

 

 

‘0

40
Time in Lttnutes

Fig. 18

dotted line in the figure the curve was spread out much

more at the base indicating that the weir did more harm

than good.

The arrangement shown in fig. 16 was also tried without

the tin vanes but this too was unsatisfactory.

 

In this set up the 14 mesh / “"'”"’” \
screens were doubled and tripled
3-14»!er
at the points where the velocity I-3 "a"
was greatest and the 50 mesh
placed between as shown. It did
not seem to improve conditions
much except to move the center
of gravity of the curve a little

farther over.

An arrangement similar to ma.” g7”!

 

 

 

 

 

:0.

GP.”
4— .

 

a» 23 i0 W 4'0 «in 30 in do do to.

Tina in Minutes

 

 

this but with vertical slots in the stilling baffle fitted

with small swinging gates was tried. This gave a better

looking curve than that shown above but the coefficient

of dispersion was only 77.7% . The gates when placed on an

angle did not distribute the flow as was expected but rather

let the greater flow through where the velocity was the greatest.

 

 

 

gsfloes the coefficient. ‘\.

£0-

I?FLA£

[3&-

In this set up a larger
amount of water was cut off in
corner by a wall and an A
baffle was substttuted for the
screens at the turn. A detail
of the A baffle is shown below.
The coefficient of dispersion
for this arrangement is 74...
The curve shows that there is

still too much dispersion as

[Or-

 

,/5- 64/. Car an“

 

p

fl 0

Well

A

Shi/Il'ry Ba ”/0

A Baff/P

 

kry/IV; hale:

 

 

ere v

 

 

 

lb 20 .30 4'0 .110 {a

 

in

Time in.hinutes

Fig. 18

.90 1.00

This was also tried with %" holes at A with little change

in results.

 

.— _, _..-

— ._ —-’

ZBofiorv—r of Tank

‘—

Wa‘hr ngce [Inc/Laval Baff/PJ

Dora/7 07" A Baffle

-31..

The A baffle was also tried with cut off walls at
both 1?. and C, fig. 18, and the stilling baffle abd tle
-"‘" holes as shown. The coefficient for this run was 77-30.

The stilling baffle was then taken out and the
original baffle replaced at the inlet. Coefficient '7-3"..'=.

rI'hree tin vanes were then placed at the end of the
original baffle and adjusted to give what appeared to be
the best distribution of the water. The coeffidient was
than 79.3, which is a decided improvement over the same
arrangement without the vanes.

Four tin vanes were then tried in place of the three
at the inlet. This gave a coefficient of 3.3.

It would appear that three vanes could be adjusted
to give a better flow than a greater number.

The original baffle was again replaced by the stilling
baffle but. this time the stilling baffle contained vertical
slots and sliding vanes. The highest coefficient that
could be obtained by moving the sliding vanes was 7".f .

The A baffle and corner walls were neat tried using
seven glass vanes at the inlet. This arrangement gave a
coefficient of 78.0.

Ser eral different devices were tried on the A baffle
such as perferated tin, 14‘; mesh screens,and glass vrnef,
“out they did not improve it a noticeable amount. It would
appear from these experiments that the A baffle would not

be of any value.

 

 

-52-

 

Six tin vanes were next #"n" curd" V\

placed as shown in fig. 19 J TI.” Van" \\
L...—-'
with corner boards and the "\

original baffle. Several
tests were made on this
arrangements with the vanes
in various positions. ri‘he
highest coefficient that was
obtained was 81.9.

 

 

 

 

  
 

a e ,
From these runs it would Z) , rm my.”
app ear the this arrangement Z) a;

14‘ ’

‘0 I-

[ti
E,o s Average nearehc'a/ 75514- Zfiél‘fl"?
Q

4- .

 

 

 

I0 2'0 4:0 1'0 «in (a 72 9‘0 Jo loo
Time in Minutes
that tin vanes give a more even flow than any thing that
has yet been tried.
The six vanes were also tried with the corner baffles
curved and in addition a curved corner baffle at A. The
coefficient then was only 75.6 but this might have been

battered by shifting the curved vanes to various positions.

 

 

 

‘l. Isl-J _

It was thought that by
putting the curved vanes in
the dead corners that fine
eddies might be elliminatcd
but it was found that large
slow eddies occured in bath
passes. The coefficient was

only 46. The curve also shows a

poor distribution.

 

 

#77.” vex/76’s am

 

 

 

 

 

 

Org/0a,
; Bgf/F ‘fin Vane
14"
F1 3. 20
’00
M?
g
0:
‘5' b
.2 A ? v t - v 1 m
0 10 .10 J0 40 we so )0 80 so ’00

Timein minutes

 

1",

 

This ii? the arran gement that Brvexner\
was finally adopted to be used
in the large basin. It consists W\\

of nine curved vanes, one in east

“-.

of the three dead corners,
three at the inlet and three
at the turn. The highest
coefficient obtained form

thia Bet 111) was 8.1090
curved Vane:

/)/J; ear/64 Vane

Fig. v.1

 

 

 

 

10‘?

 

 

.fo 20 7:0 9'0 9'0 10°
Time in Minutes
The exact location of these vanes and other details
have been previously given. "he exact ss ting of these
vanes is quite materially effected by the quantity of flow.
Theref one the ends of the vr'nes have been so designed that

trey can be moved to acconodate varying conditions of flow.

 

 

 

).
., .
. .
. .2

u. I. as
I.‘ .
o-INII

{rank

‘

.‘v‘ i!!!“ \‘l

(I Et’

I ‘ll.\

‘ nflaHVNMh mm.”

V ‘

  

   

 

 

 

 

 

 

 

  

‘ll . . .. t I... . .. L a v . v :. v .. t , . . n. v
.I...‘.«..\... A. a. . l I ..3.... A . A? , I . . l . ... n vara‘..l.u... .. I ....., . . «3.
~. \—..§. ... t . In . 7. - .phflmul $1.}; n.«.fi A4 ... v. r.‘ “.1. I. L , t .

or. \s..l....:. 3.1.1:;

 

. ., .- I. a...
‘ aura-lbw... Pier“...

 

ft .
.u n -

 

. (he .» try
.1 ,

  

..

 

A i 1....-.:......i«n.!..:1 .51.
5.4.1.251! .. .2. :L y .. 43.33 .it .

 

 

2 473

30

”8
“0
|m3
may
“:2
“1
m3