Mixing tank concerns in a scale-down

Q: Our company is trying to optimize many of our processes. When I went to the pilot plant, I noticed that most of the mixing tanks were glass-lined, unlike many of our full-scale vessels. They also have impellers that look much different. What do I need to be concerned about when using these tanks to scale-down our current processes, and then how do we scale-up any process improvements we discover along the way?

A: Tom Post responds:

It's a good idea to think scale-down when optimizing an existing process instead of trying to do it full-scale. You use fewer chemicals, and equipment changes will be less costly. The smaller equipment will make it easier to make modifications, too. Flexibility is key for the optimization process.

There is no rule that scale-up or scale-down must be made with equipment that is geometrically similar, but that is the way it is most often done. Flow patterns and shearing actions will be the same. All the power that a mixing impeller provides to a process creates flow, shear and turbulence. The trick is to find out the optimum balance of these that the process needs.

I am assuming that your full-scale vessels are metal, since you mention that they are much different than your glass-lined pilot plant tanks. At the pilot scale, mixing tanks need to be able to suit many processes, and so in the pharmaceutical world, the tanks are often glass-lined. On full-scale, the mixing vessels are often dedicated to a single process, so the materials of fabrication have been selected to be compatible with the process. Since most pharmaceutical mixing processes are low in viscosity, I will assume that for the rest of this discussion.

The big disadvantage of most glass-lined mixing tanks is that they are poorly baffled. This means that most of the power from the mixer translates into tangential flow and creates almost no shear or turbulence. When you look into the tank from the top, you will see a distinctive swirl and a vortex in the center going down the shaft. Instead of good mixing, you have a poor centrifuge. The deeper the vortex, the better centrifuge you have, not the better mixer.

To optimize a process, you need to know what the process wants from the mixer. Most of us don't know this in advance and some carefully prepared experiments can help. First, run a small-scale experiment that you know works. Make sure the mixer you are using allows you to monitor not only impeller speed, but also power consumption. Note the amount of power required for a successful batch.

Now we can test what the process requires from the mixer. First of all, your tanks must be baffled. Four tests usually are all you need. They should be all running at the same power, not speed. The following are guidelines to test the effect of:

1. High flow and very low shear: A large diameter axial flow hydrofoil spinning at a low speed;
   
   
2. Moderately high flow and low shear: A small diameter axial flow hydrofoil spinning at a higher speed;
   
   
3. Moderately low flow and high shear: A large diameter radial flow turbine spinning at low speed;
   
   
4. Low flow and very high shear: A small diameter radial flow turbine with small blade width spinning at high speed.

One of these tests will most likely give you the best results, which may mean faster batch time, better selectivity, or better yield. These four tests should tell you what your process needs and how you should design the rest of your experiments to further optimize the process.

In regard to your question, the glass-lined vessels in your pilot plant basically offer high flow and very low shear. Most single-phase processes that require just blending will work just fine in these, but any of the four above may prove to be even better. As soon as your process has more than one phase, more shear and turbulence will be required. I think each pilot plant should either have four vessels equipped with the above or have one tank that can be easily modified with the four designs.

When you are ready to scale-up your results, take a look at the results from your pilot work and you should know which direction you need to go. Not all mixers give you the same results and this is a good way to get yourself started in the right direction to optimizing your full-scale processes.

If you want further clarification on any of the points made above, or if you have a specific mixing question you would like answered, feel free to drop me a follow-up question.