Machine uptime is crucial to the efficiency of all our operations. Machines are intertwined with all areas of our operations. We enter data into them and they process the data and communicate information and reports. We load raw material into them. They cut or weld or assemble and provide us with a processed part. When these machines are down or buggy they create barriers to getting products out. Clients often ask us to help them monitor and measure the performance of their machines, then use the data to drive improvement. In this article I will share what I have learned in my years trying to help companies improve their machine uptime.
Most of what I will share here is based on the physical machinery world. I find the nature of software downtime very different from physical machinery. If there is interest I can write a separate article on this topic. Drop me a note if this is of interest to you.
First, I want to talk about data collection. Ideally, companies should buy machines that have up and downtime indicators built in. This is the best case for collecting good data and it is even better if the machine is capable of assigning its own downtime codes. In practice, I observe two problems. The first is that the company does not purchase the software that the OEM provides to monitor other machines. The second is that they do not learn how to extract the data to make good use of this data for improvement. These machines are expensive, the cost of downtime is not just wages of the people working the machine, but the opportunity cost of the lost revenue or the lead time impact to the customer. Properly estimating these costs allows the company to justify proper machine maintenance, enhancements, and upgrades. It will prove that the investment in good data collection will have an immediate impact and set the right mindset on keeping the machines producing efficiently.
There are many machines where data collection is not built in and for those instances there are plenty of 3rd party machine monitoring solutions. I am happy to share my experience with different tools for anyone interested. Again, just send me a note.
We have also been in environments of high security where machine monitoring tools cannot be used. In these cases we have to set up manual tracking with some success. Unfortunately, it is an imperfect solution. People don’t accurately track downtime and that creates friction when we start to discuss treads, problem areas, and root cause issues. Still, even manual solutions provide useful information and set the tone for machine management.
It is important to remember that you will not get away from human interaction completely when capturing machine downtime. Most systems require a human to enter downtime codes of some type. A machine may be able to tell you that it failed due to a jam or senor issue, but it will not tell you that it is down due to a labor shortage or a lunch break. It is critical to know this difference when analyzing the data. This means that part of the data capture is working with the operators to teach them good practices in collecting data and helping them to understand how the data will be used and why data quality is important.
When it comes to analysis I find that clients want to talk about calculating OEE (Overall Equipment Efficiency) but often lack the understanding of the information required to produce a proper OEE measurement. OEE is really an index of multiple factors and these factors can be difficult to capture in practice. In my opinion, it is far worse to misuse OEE due to poor input data than to use a simple percent downtime or percent availability metric. The ultimate goal is to create urgency on the production floor when machines are down and provide enough historical trends for systemic root cause analysis. Any tracking can accomplish this, so I think it is best to keep things simple.
If you do find that OEE is a must for your organization then make sure you have thought through the two pieces of OEE that are most often incorrect. First, let us discuss the standard cycle for production. This is needed to calculate a performance target and whether the machine is producing at the optimum rate. Machine rates are certainly important; unfortunately, I find in practice they are hard to gauge and compare across product mixes. I prefer to monitor these on hour by hour boards, but if you can get it straight from the machine and the product standards are consistent enough, then you can calculate OEE. Second, what is a consistent measure of acceptable quality? Now this may seem simple enough for the operator to track. Of course some machines can measure and reject inline. But what about part defects that are not found until further down the production line? This is often missed and leads to an inflated OEE. Again, I prefer to use other methods to manage and address part quality.
As I mentioned, the purpose of all this is to get the most out of our machinery. I prefer to spend less time creating the perfect measure and more time figuring out how to best apply learnings from the data to actionable improvement. In our practice we find three things very crucial to driving machine output.
- Simple reports that can be displayed on the production floor and provided to leadership.
- A weekly review to discuss the performance and review the first symptom of downtime.
- Action tracking with clear ownership, status, and target closure dates.
For example, machine jams might be captured in the data log. The discussion is, do we see a trend up or down in machine jams. If the trend is up what might be the second level cause (think 5 whys). Do we have a new operator that is not following the loading procedure? Are the belts due for replacement? Are we keeping up with PM work? All this data collection still leads to changes to procedures and practices. Getting to this point and creating routines that result in changes to human habits and procedures is where the value add occurs.
In closing, it is important to note that we will see more machine driven processes in the future. Learning how to manage the machine – human interaction to drive efficiency is crucial for all businesses. The automation we see will not just be in physical production, but in monitoring of the process itself. New video capture technology can tell if the machine is running, if the machinery is properly staffed, and if people are away from the machine due to lunch break or due to some unplanned form of downtime. These systems are not yet widespread but many large companies are experimenting with them. All companies need to start experimenting now. Small experiments now means that you will be ahead of the curve in the future when this becomes the way that we monitor machine uptime and ensure we are getting the most efficiency out of our production equipment.