352
Discussion
2. Measuring Technique
Soluble organic carbon was measured by infrared analysis. This technique was more direct and
possessed smaller variance than techniques which measure substrate in terms of oxygen equivalents.
The use of oxygen or respiratory equivalents as quantitative indices of organic substrate levels would
appear to have rather nebulous fundamental justification. An estimation of the extent of oxygen
participation in the removal of organic carbon from the fluid phase would be extremely difficult and
of questionable value.
3. Modeling Technique
This work attempts to demonstrate the sensitivity of the "kinetic" description of a batch biological
reactor to the modeling techniques used to smooth the raw data. A large number of mathematical
models could be used to describe a given set of data. The Monod model is such an example. The data
suggest, however, that statistically equivalent descriptions of a given set of data can result in the
prediction of required reactor volumes (costs) differing by from 100% to 300%. Should one have
enough faith in any "kinetic" model to justify a reactor size (cost) which may be four times too large?
This would demonstrate a confidence in the model which the data suggest is not warranted.
The data of this investigation indicate that during the course of a single batch run, microorganism
"growth" is observed when measured by the mass of suspended solids retained on a 0.45JU pore size
filter, but it is not observed when measured by the rate of removal of organic carbon, dC c /dt. This may
indicate that the number of microorganisms does not change over the course of a single batch test.
However, it is conceivable that there is variation in the initial number of microorganisms for the
constituent runs of this investigation. This would account for the independence of dC c /dt on Cg for
each run, and the significant correlation of dC c /dt with CgQ over the runs of this study.
This work further indicates that there is no evidence to suggest that the rate of removal of organic
carbon is dependent on the mass concentration of carbon or oxygen at the levels used in this study.
The variations in specific growth rate estimates result from different models of the raw data.
The use of "percentage removal" as a performance parameter has questionable validity. The
amount of dextrose in the steady-state region was probably negligible. Every run was characterized by
a nonzero stable steady-state level of soluble organic carbon. The fact that this residual organic carbon
is biologically inert over a period of days suggests that its removal in a conventional biological
treatment system is not economically feasible. These steady-state data should not be modeled to
obtain design information or coupled with the influent to describe process efficiency.
The experimental practice of wasting 75% of the mixed reactor contents after each run, refilling
the reactor with a solution of nutrients in tap water, and waiting for several hours before commencing
the next run, ensured a chemical environment in which dextrose was the growth-limiting constitutent.
The schedule of feeding employed in this study may be considered analogous to pulse loadings on a
reactor. The pulses ranged from 300 mg to 1 gm of carbon for the 3.5 litre container.
Since the specific growth rate of bench scale reactors operating under tightly controlled environmental conditions cannot be predicted without a larg? degree of uncertainty, the feasibility of
characterizing prototype reactors by specific growth rate estimates would appear to be suspect.
P.H. Brouzes, France.
In studying the biological growth rate on a bench scale to find the kinetics that apply to a large
plant, we must reproduce the condition that exists in the supply of food to the floe. In a large plant
the mean value of the time period between successive passages of one floe near the food source is
around 10 minutes. So to reproduce in a little vessel the variation in food supply to a floe that occurs
in a large plant, we must have periodic feeding.
Reply
The degree of mixing in a reactor should be estimated independent of the "reaction kinetics".
The experimental schedule of this investigation employed pulse feedings approximately 24 hours
apart, combined with daily wastage of mixed liquor and replenishment of nutrient solution and trace
elements. The data were collected immediately after each pulse was applied.
Discussion
2. Measuring Technique
Soluble organic carbon was measured by infrared analysis. This technique was more direct and
possessed smaller variance than techniques which measure substrate in terms of oxygen equivalents.
The use of oxygen or respiratory equivalents as quantitative indices of organic substrate levels would
appear to have rather nebulous fundamental justification. An estimation of the extent of oxygen
participation in the removal of organic carbon from the fluid phase would be extremely difficult and
of questionable value.
3. Modeling Technique
This work attempts to demonstrate the sensitivity of the "kinetic" description of a batch biological
reactor to the modeling techniques used to smooth the raw data. A large number of mathematical
models could be used to describe a given set of data. The Monod model is such an example. The data
suggest, however, that statistically equivalent descriptions of a given set of data can result in the
prediction of required reactor volumes (costs) differing by from 100% to 300%. Should one have
enough faith in any "kinetic" model to justify a reactor size (cost) which may be four times too large?
This would demonstrate a confidence in the model which the data suggest is not warranted.
The data of this investigation indicate that during the course of a single batch run, microorganism
"growth" is observed when measured by the mass of suspended solids retained on a 0.45JU pore size
filter, but it is not observed when measured by the rate of removal of organic carbon, dC c /dt. This may
indicate that the number of microorganisms does not change over the course of a single batch test.
However, it is conceivable that there is variation in the initial number of microorganisms for the
constituent runs of this investigation. This would account for the independence of dC c /dt on Cg for
each run, and the significant correlation of dC c /dt with CgQ over the runs of this study.
This work further indicates that there is no evidence to suggest that the rate of removal of organic
carbon is dependent on the mass concentration of carbon or oxygen at the levels used in this study.
The variations in specific growth rate estimates result from different models of the raw data.
The use of "percentage removal" as a performance parameter has questionable validity. The
amount of dextrose in the steady-state region was probably negligible. Every run was characterized by
a nonzero stable steady-state level of soluble organic carbon. The fact that this residual organic carbon
is biologically inert over a period of days suggests that its removal in a conventional biological
treatment system is not economically feasible. These steady-state data should not be modeled to
obtain design information or coupled with the influent to describe process efficiency.
The experimental practice of wasting 75% of the mixed reactor contents after each run, refilling
the reactor with a solution of nutrients in tap water, and waiting for several hours before commencing
the next run, ensured a chemical environment in which dextrose was the growth-limiting constitutent.
The schedule of feeding employed in this study may be considered analogous to pulse loadings on a
reactor. The pulses ranged from 300 mg to 1 gm of carbon for the 3.5 litre container.
Since the specific growth rate of bench scale reactors operating under tightly controlled environmental conditions cannot be predicted without a larg? degree of uncertainty, the feasibility of
characterizing prototype reactors by specific growth rate estimates would appear to be suspect.
P.H. Brouzes, France.
In studying the biological growth rate on a bench scale to find the kinetics that apply to a large
plant, we must reproduce the condition that exists in the supply of food to the floe. In a large plant
the mean value of the time period between successive passages of one floe near the food source is
around 10 minutes. So to reproduce in a little vessel the variation in food supply to a floe that occurs
in a large plant, we must have periodic feeding.
Reply
The degree of mixing in a reactor should be estimated independent of the "reaction kinetics".
The experimental schedule of this investigation employed pulse feedings approximately 24 hours
apart, combined with daily wastage of mixed liquor and replenishment of nutrient solution and trace
elements. The data were collected immediately after each pulse was applied.