Perhaps no industry has more operationally failing wastewater treatment plants than those deployed in the oil and petrochemical market segments. These plants have some of the toughest influent contaminant profiles to treat, and suffer from a range of contaminant loadings. The toughest area of wastewater process control in oil refining, of course, lies in the biological treatment areas, due to the vast number of process control variables when compared to neighboring chemical processes. Industry-specific operating conditions further aggravate these plants' process control efforts, such as:

1. The presence of recalcitrant aromatics, PNAs, and PAHs.

2. The presence of a variety of nitrification inhibitors.

3. Eight or even more independent sources of influent, all with varying characteristics, coming from oil processing units with poor communication between units and with wastewater treatment plant.

4. Lack of sufficient early warning resources to detect loading changes or inhibitory toxics.

5. Lack of sufficient equalization.

6. Insufficient emergency diversion systems.

7. Operation in upset mode more of the time than in design mode.

8. Operation in a dynamic mode more often than in a steady state.

To make matters worse, most refinery wastewater plants are not empirically designed, but rather are based on the theoretical design data from someone else's wastewater plant, under steady state conditions (which, of course, only applies to part of the time in an oil refinery). Under the conditions which a refinery biological wastewater plant has to operate, these assumptions more often than not fail in the real world of process control.

For example, to control its activated sludge process with the most current mean cell retention time (MCRT) methodology, the facility must establish an optimum range within which to operate for the realization of the specific effluent treatment goals. But how can this be accomplished without the biokinetic constants and coefficients for the specific plant design, and the specific range of influent profiles? The best that can be hoped for in such an application is an educated guess. Historically, such guesses work well in some industries, but in oil refining, these guesses have more often been the number one cause of the inability to control the biological treatment process, which of course culminates in effluent permit excursions. With biokinetic constants, engineers can mathematically precisely determine the MCRT control parameters, and all related real time process control adjustments.

Moreover, with the tremendous number of variables in a biological wastewater treatment system, it is difficult to quantitatively interpret the results observed from experimental or unintentional operational excursions from the normal control ranges. The use of biokinetic models as a measurement tool provides an absolute metric format, upon which the success or failure of an operational change can finally be based, without questions raised as to potential interferences in interpretation.

Last but not least, without the actual field determination of the specific biokinetic constants for a given plant, many attempts to operate a true MCRT control program fall short, such that the end result resembles more of a food-to-mass (F:M) control program based on trial and error operational strategy, with its considerable built-in error based on analytical methods available for the measurement of the food. This error is more than significant in oil refinery wastewater due to the dependence of the F:M calculations upon BOD and its nonlinear relationship to COD or other quick-test substitute parameters.

MCRT vs. F:M

In some wastewater applications, the use of the F:M strategy to control the activated sludge process works well. However, in many types of industrial settings, this strategy falls short of adequate due to:

1. Wide ranges of relative biodegradability of the substrate (food) in the influent.

2. Wide variability in the influent.

3. The intermittent presence of biologically toxic and inhibitory compounds in the influent.

Inherently, the actual calculation of F:M has several pitfalls:

1. In oil refinery wastewater, there is no representative quick test for the substrate. BOD5 would be representative, but does not meet quick adjustment turnaround times. COD, TOC or TPH do not have a consistent linear relationship to BOD5 in refinery wastewater. As such, considerable error in process control enters right into the calculation itself. Conversely, the use of the MCRT strategy does not depend on measuring the substrate.

2. Unlike the use of the MCRT strategy, F:M cannot be directly, mathematically related to microbial growth rates. As such, most of the operational and process control benefits of biokinetic modeling cannot effectively be achieved with F:M. Only MCRT can capture the entire spectrum of benefits, which translate to operational cost savings.

3. Unlike the MCRT strategy, the process for determination of the optimum target control ranges for F:M is not practical under the conditions that oil refinery activated sludge processes operate. As such, the optimum target F:M ranges are usually based on another plant's design and characteristics.

4. Adjustment of sludge wasting rates to control the F:M ratio is a trial and error process. With the use of the MCRT strategy, sludge wasting is precisely calculated and mathematically administered to hit the target control range.

More miles per bug

The use of biological treatment has a narrow fit in the overall scheme of available technologies applicable for processing wastewater and its related sludge. However, when biotechnology does fit, there is no alternative more cost-effective. Within the range of various biological treatment designs, there is no process more efficient and more controllable than the activated sludge process.

Based on this premise, a worthwhile goal of refinery wastewater plants would be to initiate a path directed toward maximizing the return on investment into the activated sludge system. In other words, making a long-term, concerted effort toward trying to have the activated sludge system consume as much of the refinery's wastes as possible. But how much can the activated sludge system take? What is its true operational capacity based on what really comes down the sewer? How is such a limit determined?

Although biological treatment has proven its effectiveness, trying to determine the parameters for optimizing its performance is complex. Because of the tremendous number of process variables involved in biological wastewater treatment, performance and control are not always straightforward. In many cases, metrics deployed to attempt to make these determinations appear indicative on a macro view level. But, because there are many other non-biological chemical mechanisms occurring simultaneously in a bioreactor, on a micro view level, what may appear to be the result of biological treatment may in fact be facilitated by enzymatic reactions, chemical oxidation, precipitation, adsorption, ligand complex formation reactions, sludge entrapment, air stripping and more. Now add to that problem the fact that, especially with oil refinery wastewater, steady-state conditions are not always prevalent. The textbook stoichiometric biomass relationships frequently are skewed by the presence of inhibitory toxic compounds.

In short, when attempting to quantitatively define all of the important metrics related to gauging the true performance of the specific microbiological population functioning in a given plant, there is only one way to accomplish that with 100-percent reliability, and that is through the use of biokinetic modeling tools. Only by knowing the true kinetic and metabolic reactions of microbial growth in a system can a facility truly control that process. This knowledge culminates in maximizing the true operational plant capacity, starting with maximization at the individual microbial cell level.

Fine-tuning biokinetic modeling

The future will likely find lab bench-scale, continuous-feed process simulation in multiple application areas throughout a refinery. The key indicator for reaping an operational benefit with this tool falls in target areas where the refinery has little room to play with process control variables due to the sensitive nature of the effluent quality, compliance risks, or effects on the production process itself. In other words, they cannot fully exploit process optimization too far from the middle of the established control ranges due to critical restraints.

Furthermore, rarely will the sole train of a full-scale plant be allowed to play with changes in more than one variable at one time. Unfortunately, the relationships of many biological treatment variables are influenced by multiple correlation phenomenon. Sometimes the productive operational adjustments do lie at the outer ranges of standard deviations from the normal ranges. And more importantly, in utilizing modern mathematical tools to the fullest extent, the evaluation of a plant's performance variables at the higher ranges of standard deviation are vastly valuable in fine-tuning the accuracy of the model itself. Most refineries do not have the budget to construct processes with standby units, which could be alternately used for experimentation. The use of a simple online bench scale simulator of the full-scale activated sludge system would solve the aforementioned problems, and add to the accuracy, effective predictability and overall value of the concurrent modeling efforts.

Conducting a field biokinetic modeling study

Using chemical engineering principles, bioengineers have quantitatively created reaction-rate mathematical models that have primarily been used for design and sizing calculations. Similar design models can be constructed based upon real-world observation, and collection of material and substrate balances across an operating activated sludge plant. This is predicated upon the proper analytical program to collect mass and volumetric flow data, along with the corresponding chemical analyses. This is not a trivial task in the confines of an operating unit wastewater operation compared to a controlled laboratory environment, but it can be achieved.

1. Determine the Biokinetic Constants and Variables including:

A. θ = Optimum mean cell retention time.

B. 1/θ = Growth rate of microbial population.

C. Y = Cell yield.

D. U = Specific substrate utilization rate.

E. Kd = Decay rate coefficient.

F. K = Specific substrate utilization coefficient.

G. k = Maximum substrate utilization rate.

H. Ks = Half saturation constant (effluent ---> ½ k).

The underlying biological kinetic expressions can be obtained from fitting operating data as a function of microbial growth rate. The data can be fitted using computerized linear regression analyses, which ensures the predictability of the model.

2. Build and calibrate the plant-specific biokinetic model and related equations, including:

A. Target MCRT model vs. effluent quality.

B. Construct a predictive what-if model that will allow manipulation of changes in influent flow and quality.

C. Assess the extent of inhibition biokinetic response and adjust the model accordingly.

D. Determine optimum steady-state operating conditions and determine operating strategy during non-steady state conditions.

E. Calculative model for process variable adjustments.

Once the biokinetic constants are determined, a comprehensive model across all normal and transient operating conditions can be predicted. Predictive control of complex biological systems is the next stage to maximum utilization of refinery treatment assets as well as maintaining effluent quality. Just as chemical reaction kinetics are used in every upstream refinery unit operation to optimize the process, biological kinetic modeling can be utilized to achieve a greater control over environmental stewardship as wells as return on investment.

Operational significance of the biokinetic constants

The key operational biokinetic constants for real-time process control considerations would be:

• Maximum substrate utilization rate (k) – This number defines the total contaminant-loading capacity of the entire biological plant, and can be calculated in terms of organic (carbonaceous) mass or nitrogenous (autotrophic) mass.

• Cell yield (Y) – This number defines the biomass production and CO₂ generation resulting from biological oxidation, and can be expanded to define the equilibrium shift between CO₂ and the produced biomass. The ability to measure this enables a plant to control and shift the equilibrium, thus placing a handle on such important factors as sludge disposal and oxygen utilization.

The key operational biokinetic equations for plant performance evaluations and what-if simulations would be:

• 1/θ = (Y)(U) - Kd [growth rate vs. substrate utilization]

• Se = (1/θ + Kd)/(Y)(K) [effluent substrate vs. MCRT]

• Y = (1/θ + Kd)/(K)(Se) [biomass generation vs. CO₂]

Case study

Enlarge this picture Table 1: Biokinetic modeling results from an oil refinery wastewater treatment plant.

Table 1 shows actual results obtained from a large U.S. oil refinery wastewater plant biokinetic modeling project. The refinery experimented with a process change that represented a controlled change in the metabolic activity of carbonaceous microorganisms deployed in the activated sludge aeration basin at the cellular level. The "Before" column represents operation under normal historical conditions. The "After" column represents operation during the experiment.

Significant knowledge was gained with the results obtained from the study:

1. The maximum substrate utilization rate increase demonstrated that the plant nearly doubled its total organic loading capacity by employing the operational changes during the experiment.

2. The cell yield increase demonstrated that the biomass production increased nearly 30 percent by employing the operational changes during the experiment.

From this, the plant was able to draw several conclusions about its future prospective alternatives. In particular, they learned that by deploying the experimental conditions permanently, they could gain a 50-percent increase in plant capacity, but at a cost of a 30-percent increase in biomass disposal costs. PE