(389b) Mixing Scale-up for An Air Oxidation Process

Air oxidation reactions represent an important category of commercial reaction processes used in the manufacture, for example, of organic hydroperoxides and organic acids. Proper agitation of those reactors, providing adequate gas dispersion and sufficient residence time to allow enough surface area for mass transfer and residence time for reaction, is critical. Efficient utilization of the oxygen contained in the air feed allows economic use of the compressed air and also helps to prevent the presence of flammable gas/vapor mixtures in the reactor headspace.

Sabic worked with an agitator vendor to perform scale-up trials for an agitator design for reactors to be installed as part of an air oxidation process. Reactions had previously been performed at a 2-liter lab scale and also at a 1m³ pilot plant scale to finalize the reaction conditions and procedure. The agitation trials were necessary to finalize the details of the agitator and baffling for the 23m³ commercial reactors. A preliminary design for the reactor cooling coils had been developed and the agitation trials were also designed to confirm the adequacy of that design, considering the overall heat transfer coefficient as well as the effect of the coils on the vessel agitation.

Agitation trials were carried out utilizing a clear plastic vessel, 1 m in diameter by approximately 1.5 m straight side height. Geometric similarity was maintained with the proposed design. Two concentric sets of cooling coils (1” OD x 9 wraps each) were installed, along with a 5% baffle located between the two sets of coils. Two stages of impellers (0.45m in dia.) were provided, a radial flow impeller close to the lower tangent line of the vessel and near the point of introduction of the air, and a combination radial/axial flow impeller located 0.45m above the lower impeller. A sparge pipe was provided to inject air just under the lower impeller. Water was utilized as the agitated fluid, and the top of the vessel was open to facilitate insertion of baffles, piping, and instrumentation and for visual observation of the quality of agitation. Compressed air was provided, along with flow control, to allow study of the agitation under varying degrees of aeration. Scaling of the airflow was done to compensate for the operating pressure and temperature differences between the ambient atmospheric agitation trials and the commercial reactor. This scaling was done in order to more closely duplicate the gas void fraction that would be experienced in the commercial reactor. The scale of the trial reactor was 1:2.8 in length (1:22 in volume) compared with the commercial reactor, and the geometric scaling was preserved within 5% on critical dimensions. Agitator rotational speeds of 0 to 230 RPM corresponded to 0 to 110 RPM on the commercial reactor, considering equal power per unit volume.

Following is the list of studies performed during the agitation studies:

 

• Baffling.  The initial baffling arrangement of the preliminary design was found to be insufficient in preventing vortex formation. The power draw was found to be only 60-70% of theoretical when operating with the vortex. Gas from the headspace was also entrained into the liquid by the action of the vortex. The addition of four 10% baffles just inside of the inner cooling coil was found to provide adequate baffling to prevent this vortex formation, when added to the existing four 5% baffles between the two sets of coils.
• Impeller Spacing. Point velocity measurements and void volume measurements were made for two locations of the lower impeller, one slightly above the centerline of the lower coil wrap (~8% of the agitator dia.), and the other approximately even with the centerline of the lower coil wrap. Higher fluid velocities were observed between the cooling coils (0.37 vs. 0.18 m/sec) and a slightly higher air hold-up (6.4% vs 5.6%) when the lower impeller was in the lower position. This position was selected for scale-up for improved heat and mass transfer.
• Power Consumption and Gas Dispersion. Once the baffle configuration and impeller spacing were fixed, trials were conducted to identify the flooding point for the agitator at the scaled down design airflow rate. The flooding point is defined as the minimum rotational speed that is required to completely disperse the air at a given airflow rate. The flooding point was determined visually by observation of the dispersion and circulation of the bubbles as viewed through the vessel walls, as well as by checking for “spouting” at the surface. Spouting was observed as larger gas pockets (undispersed air or air bubbles which had coalesced) breaking through the upper surface. The flooding point for the 1m³ test unit was determined, along with the power consumption at the flooding point. Knowledge of the flooding point allowed selection of the range of agitator speeds over which the remainder of the trials would be conducted.
• Mixing Time Trials. Mixing times were established as a function of airflow and impeller speed, using a visual chemical method (starch iodine complex). The mixing time for the plant scale at 110 RPM, the maximum rotational speed planned for the plant scale agitator, was estimated based on these results.
• Mass Transfer Trials. Mass transfer, relative to the specific compositions and conditions used in the test, was evaluated vs. rate of airflow and rotational speed. Dissolved oxygen depletion in the presence of sodium sulfite and the time required to re-oxygenate the liquid phase containing a known starting quantity of sodium sulfite was the method used to measure the value of k_La. The accepted relationship between mass transfer and the variables of superficial gas velocity and power per unit volume was verified to be accurate for this agitation geometry, allowing scaling of the mass transfer performance in the 1m³ actual pilot reactions to the commercial scale.
• Heat Transfer Trials. Outside film heat transfer coefficients were determined through cooling trials in the agitation test unit. Separate trials were conducted for the inner and outer cooling coils. Measurement of cooling water flow, inlet and outlet cooling water temperature, as well as the temperature of the bulk fluid were made as a function of time during trials at various rotational speeds and air flows. The overall heat transfer coefficients were determined and then the film coefficients were calculated based on well known correlations for the inside film coefficient and the resistance of the metal tube wall. The trial data were fit to standard correlations for film coefficients for heating/cooling coils in agitated vessels to determine the best constants to describe the heat transfer with the type of agitation and geometry being utilized in this reactor design. Those correlations were then used to validate the coil design for the commercial reactors.

Results from the agitation trials were used to finalize the details of the agitator, baffling, and the cooling coil design for the commercial reactors.

After installation of the plant reactors, and before plant start-up, water runs were conducted as a part of plant commissioning. Mixing trials were conducted at various agitator speeds and airflows to determine the flooding point. This was necessary in order to set the minimum recommended agitator speed that would insure adequate gas-liquid contact and avoid “bypassing” of un-reacted oxygen directly to the reactor headspace. This was important for conservation of air as well as being one of several measures necessary for the avoidance of flammable mixtures in the reactor headspace.

During the plant commissioning some of the U-bolts used to attach the cooling coils to their supports were found to be fatiguing to the point of failure. A finite element analysis indicated that the coil supports needed to be improved in order to increase the stability of the coils against vibration under agitation conditions to avoid fatigue and failure of the U-bolts. Insufficient communication between the vessel manufacturer and the agitator supplier regarding the stresses induced on the coils by the agitated fluid led to the under design, illustrating a critical communication that needs to take place during the design phase for agitated vessels with complex internals. Redesign of the supporting brackets and U-bolts, based on the findings of the finite element analysis, led to an installation that has been successfully operated with no further fatigue related issues.

After start-up of the oxidation reactors, data were taken to determine the actual heat transfer, and the available heat transfer was compared with the required heat transfer and that predicted by the correlations developed during the agitation trials. Observations were also made regarding the concentration of oxygen in the reactor headspace and the implications regarding the quality of the agitation vs. that required by the reaction and mass transfer rates.