Understanding your batch process data the ANA Way
Who needs measurements? Aren’t the calculations too complicated to be of any practical value to anyone who isn’t a scientist?
That’s a fair question, so let’s take a simple example and explore. Spoiler alert: there’s barely a calculation in sight. Use the principles of Electronic Batch Processing to see the answers easily.
Consider a tank, equipped with an agitator/mill, and let this tank sit on load cells. That combination of equipment, when connected with the right measuring tools, provides three parameters. These are the frequency and power of the mill and the weight of the tank. Let’s assume that we gave an operator a recipe, a batch procedure, and left him to execute the process. While we appreciatively receive his documentation upon completion, we seek further information.
We want the data to speak. Our intention is not to judge but to learn what happened.
Figure 1 shows a plot of the 30 hour batch procedure.
Figure 1: the three parameters adjusted so that they all fit on scale.
The first thing we might notice is that the mill was turned on to a fixed frequency early in the process, was reduced in frequency and then went off for a long time before being turned on again. During the off-time the weight of the tank was constant. Right away we can see that this process was shut down for 15 hours. Perhaps a batch of this material can be left to sit overnight; perhaps we would have preferred that it be stirred. We see that it was static.
Another point we notice is that as the weight was increasing – due, of course, to the addition of material – the power used by the mill to mix the batch increased. We wonder, did the batch became increasingly viscous as material was added – or is it just that the extra mass of the material makes it harder to push? For an agitator at the bottom of the vessel, a prop, we would be seeing a viscosity effect; for a top to bottom sweep, it would be both viscosity and bulk weight. From this alone we can’t tell.
But what happened just before hour five? The weight remained the same for an hour, meaning nothing was added. Is this what we wanted? Perhaps the process required a pause to mix. But if it did we notice no change in the viscosity and we may wonder what exactly the delay accomplished. That was a whole hour.
And notice that added weight and increased power use go together here. In the three hour interval before the pause the power increased 3000 (arbitrary) units as the weight increased 5000 (arbitrary) units: 0.6 In the two hour post-pause time, the power increased 2000 units as the weight increased 3000 units: 0.7. These are similar effects, and we refer to our prior experience with reading graphs based on these materials to ask ourselves if this is normal. We have a warm feeling, that “yes, this seems right.”
But speaking of warm, aren’t we missing a key parameter? Temperature. We make a note to ourselves to add another parameter; after all, viscosity of aqueous-based solutions usually decreases with increasing temperature.
Curiously, the data shows that just as the last of the added material reached the tank, the mill was turned off. Odd, perhaps. Maybe it might have been better to let the full tank mix well before shutting it down.
Next, we ask ourselves what the effect of sitting unstirred for 15 hours might have had on the batch and look to Figure 2 to find out.
Figure 2. The data of Fig 1 is re-plotted, this time with the 15 hours of nothing removed. We’ve spliced the two manufacturing sections.
Well, the powers are not the same! Assuming the temperatures are the same, the next-day batch is more viscous.
And look further… at point 549 of Fig 2 (hour 24 of Fig 1) slightly more weight is added without impacting the power usage; maybe even decreasing it.
And, finally, as the batch is drained from the tank, the power consumption goes down: symmetrical with the power consumption on the way up.
Yes, it’s a sweep, not a prop that is pushing the fluid.
All this and no calculations!