Chemical Engineering Science (2011), Vol. 66, pp 440-447

In this study the mixing time in shake flasks was determined using a colorimetric method combined with a rotating camera. The experimental setup was consistent with the normal movement of a flask on an orbital shaker with a variable shaking diameter. With the camera rotating around the flask, the movement of the fluid could be observed during the entire experiment.

The test bulk solution was based on either deionized water or a solution of 55 g/L polyvinypyrrolidone (PVP) in deionized water with the pH indicator bromthymol blue added. The pH of the test bulk solution was adjusted to 7.5, resulting in blue colour.

Mixing experiments were performed in standard unbaffled Erlenmeyer flasks by slowly feeding a drop of sulphuric acid through a needle mounted in the center of the flask headspace. During one experiment a single drop flew from the needle tip to the surface of the rotating bulk liquid under the influence of centrifugal force with the rotating camera taking images of the entire mixing process.

The colour of the test bulk solution changed from blue to yellow, below pH 5.5, with the addition of acid. Therefore the exact mixing time could be determined by analysing the last trace of the original blue colour in the experimental video recording and the exact time from the acid drop hitting the test solution to the disappearance of the last trace of blue could be discerned.

Impressive experimental photos showed that the mixing process always took place from the “tail” to the “head” of the rotating bulk solution. The acid drop penetrated the liquid sickle at the back corresponding with the rotational direction and mixing continued to the front where the last trace of blue colour disappeared.

Regarding the effect of different filling volume to nominal shake flask volume ratio, the authors could show that at filling ratios larger than 8% no influence on mixing time was detectable. Filling ratios up to 20% were tested and mixing time did not exceed 3 seconds (200 rpm shaking frequency; 50 mm shaking diameter; flask volume 100 – 500 mL; bulk solution distilled water).

The influence of shaking frequency on mixing time was investigated in water and PVP solution with a viscosity of 38 mPa s. It was shown that mixing time decreased with increasing shaking frequencies. At 100 rpm mixing in water was complete after 6 seconds while at 350 rpm it took only 0.5 seconds. The more viscous system took longer to mix, from 10 seconds at 100 rpm decreasing to 3 seconds at 350 rpm (filling volume 10% of nominal flask volume; shaking diameter 25 mm; flask volume 100 – 500 mL).

In another attempt it was shown that the shaking diameter had no significant effect on mixing time in both test solutions under the chosen conditions (filling volume 10% of nominal flask volume; shaking diameter 25 or 50 mm; flask volume 100 – 500 mL).

Depending on the Reynolds number, flow behaviour in shake flasks could be laminar, transition and turbulent (Re ~ 60.000) with a dimensionless mixing number referred as a function of flow regime. The effects of Reynolds number on mixing number in shake flasks were investigated in this work, both in water and viscous PVP solution. Transferring the fluid from laminar to turbulent conditions with increasing Reynolds numbers, shortened the mixing time. At high Reynolds numbers the dimenensionless mixing number remained constant, indicating the independence of mixing time from Reynolds number under the chosen conditions (filling volume 10% of nominal flask volume; shaking diameter 25 or 50 mm; flask volume 100 – 500 mL).

For successful scale-up, shake flask mixing times were compared in this study with typical stirred tanks at equal specific power input conditions of about 2 kW/m³. Mixing times in shake flasks and stirred tanks with comparable operating conditions fitted well with the trend that mixing time decreases with the scale of the reactors.