From Noise to a Symphony
In an earlier post The story in your code, I introduced the concept of noise when it comes to data. Essentially, data without context is just noise. I wanted to go into a bit more detail of what that means with an example using a batch manufacturing process.
Continuous vs Batch Manufacturing Processes
In the manufacturing industry, there are two general types of manufacturing processes:
- Continuous manufacturing process
- Batch manufacturing process
Selecting which type of manufacturing process is right for you depends on several factors. Typically, the higher up the supply chain you are, the more likely your process will be a continuous process. In that case, you’re trying to make bulk quantities of a base product.
The further down the supply chain you are, the more likely it is your process will be a batch process. Being so close to the end customer, you need to be more responsive to special product requests and fluctuating market conditions. Having a batch process gives you that flexibility.
In a continuous process, adjustments are made in real time to maintain the process within a fixed set of parameters. These parameters are called set points. Samples are taken at various points along the process. The results from these in-process samples drive decisions to keep the process in alignment, or bring the process back into alignment. Continuous manufacturing processes that have been optimized may require little input to remain in alignment, saving you on time and costs.
With batch manufacturing, you have the same set points; however, these set points are applied batch to batch. In this case, not only does the process need to be alignment with your set points, moreover, each batch needs to be in alignment with each other. The end customer should not be able to tell which batch (or lot) their desired product came from. With a batch manufacturing process, consistency is key!
Uncovering issues from batch to batch
In the chemicals industry, batch manufacturing could involve adding the feed chemical and an oxidizing agent to a reactor in a specific sequence. In the pharmaceutical industry, this could be adding cells and media to a bioreactor in a particular order.
As an example, let’s consider a multi-step batch process. You’ve started to notice that this batch that normally took 8.5 hours to complete, has been slowly creeping up closer to 12 hours. Every new batch seems to be taking longer and longer. Whoa!!!!
The first image (Figure 1) is the product of an Excel data dump of the overall times for batches 1 – 8. It’s helpful, yes, but it doesn’t really tell you which sub-step in your process is taking longer than usual.
The second image (Figure 2) is the product of what I would get from an SQL query. I introduce SQL in an earlier post: The story in your code. SQL is a powerful coding technique that allows you to efficiently drill down into the useful parts of your data. Instead of pulling, sorting and filtering large amounts of data in Excel, SQL query does all that for you in a fraction of the time.
In this example, I used SQL query to focus on the sub-steps that make up the overall batch. I also color-coded each sub-step to make it easier to visualize.
Question. Which sub-step is the culprit? (Hint: It’s not Step 1 or Step 3).
Uncovering issues within a batch
With this clarity from SQL, the noise of the first image is replaced with the symphony of context. Now, you can focus your time and energy onto Step 2. You can drill down even further into Step 2 to identify the root cause to these drifting batches. That’s time you’re not wasting on Step 1 or Step 3. That’s time you gaining to restart the process sooner than expected. And who has time for a 12-hour batch anyway?
What could cause Step 2 to start drifting away from the set point by so much?
- Could it be a leaking pump that is taking extra time to fill your reactor or vessel?
- Could it be the automation recipe is not sending the fill signal as designed?
- Could it be damaged wires on the automated valve at the pump discharge not receiving the signal at all?
Over time, as you keep digging to get to the root cause, you will start to see a pattern
- Do certain processing areas always seem to have this same issue?
- Does this issue only happen on certain product lines?
- Does the time of the year make a difference in how often you see this issue?
The impact of resolving the batch issue on time
The additional hours will impact your overall plant processing time. This is especially true if there are other processing steps that come before and after this problem batch. The more time you spend inefficiently troubleshooting the batch issue, the more it will take you to get your product to the customer. Depending on where you are in the supply chain, the ripple effects could be significant. Remember 2020, anyone?
A streamlined troubleshooting approach gets your team focused on fixing the right problem in a shorter amount of time. This means you can get your entire process back up and running in record time.
After you identify the root cause, document it!
Now that you’ve identified the root cause, here comes the tricky part. You may be tempted to apply the solution – repair the pump, replace the valve, replace the wires, or whatever you came up with – and keep it moving. Or, you could capture this new found knowledge in your Standard Operating Procedure (SOP) under the Troubleshooting Section. This way, your entire team now has an easily accessible play book of what to look for the next time the same or similar issue recurs.
You mean, you don’t have an SOP with a Troubleshooting Section?
We need to talk right away!
This article is also published on Linkedin.