T. H. Y. TEBBUTTand D. G. CHRISTOULAS*
Abstract--Extensive pilot-scale studies on the primary settlement of sewage have been analyzed to
produce a simple mathematical expression relating sedimentation efficiency to hydraulic loading and
influent SS concentration. Comparisons are made with other published performance relationships for
primary sedimentation, lnfluent SS concentrations were shown to be more important in controlling
removal than surface overflow rate and it is concluded that conventional settlement design criteria
are very conservative.
~
down to about I #m in size. These fine particles are 300--400
200-300
nonsettleable and there is no correlation between total 0-60 100--200
-- 50--100
suspended solids and settleable solids so that for a w
S mgL ~
given SS content a tank may give quite different remo-
0.40
val efflciences on separate occasions because of the
varying content of settleable solids. These factors mean
0.20
that sedimentation is not simply a physical process
controlled by hydraulic parameters but a much more
complex process in which the variability of the feed is ] t I , [
2 a 6
~ - ~ ~
6 6 6 6 6 6 6 6
66666 6 6
0
0
~ r~ CD b~. ~,1 ~ ~"~-
6 6 6 6 6 6 6 6
4-"
0 o
,..o
[..
6 6 6 6 6 6 6 6
o
350 T.H.Y. TEBBUTTand D. G. CFIRISTOULAS
25 41 I 0'49
50 402 0.43
100 355 0.36
150 365 0-34
0.60
0.4C
= •
0"40
Ld
I 0"20
.... E=o725e-oo°°70
0.20
i I L 1
0 25 5o I00 150 50 ~00
q m ~ rn ~d q rn 3 m - 2 d-~
Fig. 3. A v e r a g e r e m o v a l of s u s p e n d e d solids in p i l o t - s c a l e
tank. Fig. 4. F i t t i n g of E = Eo c "~ for Steel's d a t a .
Perf2.rmance relationships for primary sedimentation 351
E~ritt Relationship Eo c
ILl
I
0"20
~ ;
E=O.ZTe - ° ' 0 ° 2 3 q
" ~ ~ Pilot-plant data
Smith
Steel
0.510
0.820
0.725
0.0029
0.0088
0.0047
-// Escritt 0"770 0.0023
CIRIA 0'750 0.0123
i il I
50 tO0 150
q m 3 m-2d ~1
rates. The range of data from which Escritt's formula
Fig. 5. Fitting of E = Eo e -~q for Escritt's expression.
was obtained is not known but Fig. 5 shows that the
formula gives a reasonably linear relationship over a
centration. A simple relationship for these conditions range of q of 20-130 m 3 m - 2 day- 1. The CIRIA rela-
is of the general form. tionship was derived from data with overflow rates in
the range 6-33 m 3 m - 2 day- 1 and Fig. 6 shows that
E = Eo e- ~q (4) equation (4) is a fairly good approximation to the
CIRIA expression for q values between 10 and I00
where E0 = removal efficiency for SS under quiescent m 3 m - 2 day- 1. Smith's relationship [equation ( 1)] is of
conditions; and c = constant. course already in the form of equation (4).
A least squares fitting of this form to the data in Fig. Table 3 shows the values of E0 and c obtained by fit-
3 produces the expression: ting equation (4) to the various performance data.
Clearly there are considerable differences in the values
E = 0'510e-0"0029q (5) of Eo and c but this is not unexpected because of the
widely differing sources of the data. The data from the
It should be appreciated that eq,:ation (5) is not pilot-plant indicate a relatively high proportion of non
necessarily the best fit to the experimental data but it settleable solids since the maximum efficiency permit-
was selected as a simple mathematical model for what ted by equation (5) is 0.510. The maximum efficiency
is known to be a complex process. permitted by the other relationships is in all cases
Because of the simplicity of this expression it is of in- higher than that from the pilot plant results but it must
terest to observe how it compares w!ith other perfor- be appreciated that E o does not simply measure the
mance relationships discussed earlier. Equation (4) im- settleable fraction of the influent as a direct ratio of
plies that there is a linear relationship between q and settleable to total solids. Other factors such as the SS
log E and Figs. 4-6 show Steel's data (Si = 300-400 concentration, nature of the solids and their size distri-
mg l- 1), Escritt's expression (Si = 40) mg 1-1) and the bution, flocculation and temperature all affect the settle
CIRIA relationship (S~ = 400 mg l - ! ) plotted in the ability of suspended matter. Unfortunately at the
q - log E format. Steel's data show an almost perfect present time it does not appear possible to include the
linear relationship over the whole range of overflow effects of these other factors in a general performance
relationship so that universal values of Eo and c are
not likely to be obtainable.
0,60 ~ / /
/ Effect of influent suspended solids
E=O.75e-Oo~23q
It appears likely that the total SS concentration will
/ ~ CIRIA have an effect on the maximum removal efficiency (Eo)
0"60 since flocculation will be more significant at higher SS
levels. It is also possible that higher SS concentrations
hi contain a greater proportion of settleable solids and
there is some evidence that for sewages in the U.K.
I 0'40 / ~ ¢ there is a fairly constant non-settleable SS con-
centration of around 150-200 mg 1-1.
In such a situation the SS remaining in the effluent
from a sedimentation tank after a given time can be
0.20 expressed in the form
S, mg t -r
and for Escritt's expression equation (9) gives a satisfac-
tory description in the range S~ = 150-800 mg 1- 1 as
follows
0-60 /
E = 0 ' 7 6 e -[q46,'S3+°'il°23ql. (11)
The pilot plant data has been investigated for the effect
of suspended solids variations in two ways, both by 0.40
double regression analysis. The first analysis was made E /j/(/ I00 m 3 m-2cl -'
S i mg L- t
The second analysis was performed on the original
data grouped into the S~ranges 100-300, 200-400, 300-
600, 400-800 and 600-.1200 mg 1- ~ with the following
result 0.60
E = l-138 e [(358/S,l+O.OO16q]. ( I 3) /
/
The grouped data and the curves derived from equa- 0.40
tions (12) and (13) are shown for each rate of flow in
Fig. 7. As would be expected, equation (13) provides a E 150 m 3 m-2d -I
better fit to the data since it was derived directly from o.zo
the points plotted and is somewhat influenced by over-
lapping of the data. The fit provided by equation (12) ~7'/~ , J I , , , ,
is however reasonably satisfactory and this equation o 500 ~000
must be considered more acceptable since it does not Si mg L- t
give an ultimate maximum efficiency in excess of unity
as does equation [13) which is mathematically a some-
- - - Equation 12
what closed representation of the plotted data. In addi-
. . . . Equation t3
tion it is believed that the SS range of 200-600 mg 1
is likely to be more representative of normal sewage.
Figure 8 gives plots of equation (12) for a range of in- Fig. 7. Effect ot influent suspended solids on removal
fluent SS concentrations and illustrates the consider- efficiency.
Performance relationships for primary sedimentation 353
O'BO
0-60
0.4O tion 12
o-2o ~ -CIRIA
, I , J
o I J J Retention, h 2 I
150 tO0 50 25
q m3 m-2d-~ (2.500m deep)
Fig. 9. Comparison of various suspended solids removal relationships with curve derived from equation
II 21 over flow range used experimentally.
354 T.H.Y. TEBBt:T'rand D. G. CHRISTOULAS
Li - La,,. + On + Ki)Si
more effect on the behaviour of the biological oxi-
dation process which follows so that it is pertinent to Le = Ld(O~ + m S i + K , , S e 117)
also consider primary settlement as a process which
In general, K,, = ~b(E) and S,, = (I - E) Si and equa-
controls the feed quality to the biological treatment
tion (17) may be rewritten as
units. In this context the volatile suspended solids and
the oxygen demand present in the tank effluent are the L~ - L,,
. . . . K, - (1 - E)O(E). (18)
important characteristics, Examinations of the exper- Si
imental data showed a linear relationship between
Since the experimental data show that Ks does not
volatile and total suspended solids. For design pur-
vary with S~ the term (L~ - L,,)/(S3 is simply a function
poses the oxygen demand going forward from the pri-
of the SS removal efficiency E. Using the grouped ex-
mary settlement units to the biological oxidation stage
perimental results in the range S~ = 200-800 mg 1-1
is probably the most important single parameter.
where there is adequate data, (Ls - L~)/(Si) has been
The chemical oxygen demand (COD) in a sample is
plotted against E in Fig. 11 which indicates K~ ( = ~bE)
partly in solution and partly associated with the sus-
is a constant at least in the range of E values 0.26-0.63
pended solids
and that the results can be expressed as
L=La+ KS (14)
Li - L~
- 0-311 + 0.779 E (19)
where: L = total C O D m g l - ~ ; La = dissolved C O D Si
mg 1-1; K = a coefficient; and S = suspended solids i.e. K~ = 1'09 and K~ = 0.779.
concentration mg 1- a. Equation (15) was based on the whole data and
Consideration of the influent data from tl~e pilot using the same data range as plotted in Fig. 11 (i.e.
plant arranged in groups as shown in Table 4 gives a Si = 200-800 mg 1- t) the relationship between influent
set of points, plotted in Fig. 10 which produce a satis- C O D and SS becomes
factory linear relationship,
L i = 342 + 1"072 Sv (20)
L~ = 411 + 0"894Si (15)
This equation can be compared with equation (19)
If it is assumed that the dissolved C O D varies with the which gives Ks = 1.09 and the fact that both equations
(19) and (20) give similar values of K~ (1.09 and 1.07)
would suggest that m = 0. This would inply that the
dissolved COD remains constant and does not vary
15OO -- x
with suspended solids concentration. It must be
emphasized that this implication is derived from the
pilot-plant data only and its more general application
would require detailed examination of data from other
IOOO -- sources,
L
~'
/- ~- L,=411+ 0'894 S
If it can be accepted that the average C O D equival-
ent of the effluent suspended solids is not influenced by
/× the removal efficiency it is then possible to compute
/× the effluent COD, viz.:
500 /x
L~.=La+K,,S,,=La+[I -E)K,,S~ (21)
The C O D removal efficiency (E D is then given by
( K ~ - K,,) + K,,E
I i t EL = (-£;,,,,,'fj+,,;+ K, (22)
500 1000 1500
Simg ( J
which gives Ea as a linear function of E.
Fig. 10. Relationship between COD and SS in influent for Using the pilot plant data from which equation (20)
pilot-scale results. was derived, equation (22) has been used to calculate
Performance relationships for primary sedimentation 3551
L,--L e ×
S, =0.311 + 0 7 7 9 E
0 e©
060
0.40
020 I i I
)20 040 0-60
E= S i - - S e
S,
Fig. 11. Relationship between COD and SS removal efficiencies for pilot-scale results.
Table 5. COD Removal efficiency data since it also suggests a linear dependence of C O D
removal upon SS removal. The CIRIA relationship is
EL
Si interval from however quite different since it involves a marked
(mg 1-1) E EL equation (22) dependence upon SS removal efficiency.
The use of C O D as a measure of organic strength in
200-300 0.260 0.220 0.212 the present investigation was based on the relative
301-400 0-400 0.314 0.302
speed of this determination as compared with the
401-500 0.482 0.376 0.372
501-600 0.530 0.375 0.425 BOD test. Although there is no universal relation
601-800 0.627 0.543 0-500 between C O D and BOD it does not seem likely that
primary sedimentation would significantly affect the
C O D / B O D ratio as does biological treatment for
the C O D removal efficiency which is compared with
example. It is postulated therefore that the C O D remo-
the observed values in Table 5, the maximum error
val efficiency by primary settlement is comparable to
between the two sets of removals being 13%.
the BOD removal in the process.
Published relationships for the removal of
organic matter in primary sedimentation are relatively
CONCLUSION
uncommon and those that are available are based on
B O D data. Full-scale data reported by Water Pollu- Pilot-scale studies on the performance of a primary
tion Control Federation (1959) shows a considerable sedimentation tank have shown that it is possible to
scatter but since they are for relatively constant in- formulate a simple mathematical model to describe the
fluent suspended solids levels (Si = 230 mg 1-1) the average removal of suspended solids in terms of sur-
data have been examined in groups of E values with face overflow rate and influent suspended solids. The
a class interval of 0'10. The resultant average values are effect of influent suspended solids concentration on
shown in Fig. 12 on which is also shown the CIRIA removal efficiency is, because of the effect it has on in-
(1973) relationship for BOD removal (Ea) fluent density and the degree of flocculation, consider-
ably more important than overflow rate in the range
E 8 = 0-86 E 2 - 0-029 E (23)
examined (25-150 m 3 m - 2 day-1). The results suggest
and a plot of equation (22) using the pilot plant coeffi- that conventionally designed tanks are likely to be
cients with Si = 230 mg 1- ~. Over a range of E values considerably underutilized and that significantly
of 0.20-0.65 the Water Pollution Control Federation shorter retention times could be adopted with little
data bears a considerable similarity to the pilot plant loss in removal efficiency. A preliminary model of
C O D removal in primary settlement has also been for-
0"601 mulated but there is need to examine more full-scale
× data before the relationship derived could be consi-
dered to have general application.
Hamlin M. J. and Tebbutt T. H. Y. (1973) Sedimentation. systems. J. Sal~it, Engng. Dil:. Am. Soc. Cit~. Engrs. 95, SAI,
Proc. Syrup. on the Use of Mathematical Models in Wat. 117.
Pollut. Control, University of Newcastle-upon-Tyne Steel E. W. (196(t) Water Supply and Sewerage, 4th Edition,
March R. P. (1967) The performance of rectangular sedi- McGraw-Hill, New York, 501.
mentation tanks, Birmingham University Ph.D. thesis. Tebbutt T. H. Y. (1969) The performance of circular sedi-
March R. P. and Hamlin M. J. (t969) An investigation into mentation tanks. War. Pollut. Control 68, 467.
the performance of a full-scale sedimentation tank. J. Water Pollution Control Federation (1959) Manual of Prac-
Proe Inst. Sew. Pur!¢~ (2), 118. tice No. 8 Sewage Treatment Plant Desiqn, Wat. Pollut.
Smith R. (1969) Preliminary design of wastewater treatment Control Fed. Washington, 93.