Jim Connett, on February 05, 2019, 11:17 AM
E10 Unmasked – Part 4, Q & A
This is the final installment of our four-part series titled E10 Unmasked. We hope our surface-level journey through the E10 world has been useful to you and has challenged you to think about improvements needed in your systems. If you missed any of the previous blogs, please read Part 1, Part 2, and Part 3. In this final post of this series, I will attempt to answer common questions. The answers to each of the questions below are but one person’s opinion. You may agree or disagree! We welcome your follow-up questions, views, and ongoing dialog. Email us anytime and we will continue to update this article accordingly.
Q & A
In a fully automated fab, does it matter where we mark the transition into and out of a productive state?
Yes, it matters a great deal. Ideally, the transition point should occur when the tool indicates it is actually running the chosen recipe, and the transaction back to STANDBY should occur just after the completion of that recipe. Some customers include the track-in and track-out events to these same points as well, which provides a nice correlation. In the purest sense of the term, the tool is not PRODUCTIVE until the recipe starts. The period of time between the wafers’ arrival at the tool and loading/preparation for processing is not generally considered PRODUCTIVE time. Changing the tool state when the wafers arrive at the tool (and before they process on the tool) clouds the PRODUCTIVE and IDLE states and artificially inflates the total PRODUCTIVE time. Setting the best transition points involves the use of industry standards, an incorporation of your business rules, and a bit of trial-and-error. Over time, you will eventually reach a consensus regarding the best transition points for your site. The goal is to make the transition points the same for all toolsets in your facility – this will require a concerted effort.
Doesn’t it take a lot of data to paint an accurate E10 picture of my facility?
Not really. Simply tracking the transition between STANDBY and PRODUCTIVE states will give you data to measure productivity. Add to the mix the ENGINEERING state and you can clean up the PRODUCTIVE state to only include PRODUCTIVE time associated with sellable material. Same with SCHEDULED and UNSCHEDULED time. Most MES systems should have some type of event scheduler as well as a way to easily log a tool to a down state. These two “features” are all that you need to capture utilization and availability. Add to it the NONSCHEDULED time and you can purify the UNSCHEDULED time by removing all unscheduled downtime not related to the tool.
Ok, let’s say I want to start collecting E10 data today. I’ll have to wait a year or two before I’ll have enough data to see meaningful results, right?
Not at all. SYSTEMA’s OEE (Operational Equipment Efficiency) module allows you to go back in time (assuming your database stores historical transactions) and use the MES-related events to reconstruct the database history into an “E10 picture” using STANDBY, PRODUCTIVE, SCHEDULED and UNSCHEDULED times logged in the lot and equipment history database tables (NONSCHEDULED and ENGINEERING times are a bit more difficult to reconstruct, but we’ll certainly try!).
I’m still not convinced that we need to go to the effort of creating and tracking a NONSCHEDULED DOWNTIME category. What more can you say about the implementation of this E10 element? How can it be useful to me?
Every piece of equipment you buy contains some type of guarantee regarding productivity and/or efficiency. If the guarantee is violated, you may be in the position to ask for discounted field service costs or other discounts as a result of the violation. In order to properly enforce this guarantee, you (the owner of the equipment) need to have data to support your position. If you are unable to separate the downtime between internal tool failures (UNSCHEDULED downtime) and external tool failures (NONSCHEDULED downtime), then the downtime that is purported to be violating the agreement will be artificially inflated.
As stated earlier, NONSCHEDULED downtime is the most difficult to track because it is rarely an automated event-based transaction at the time of the transaction, but rather a post-event transaction that someone must remember to log. The first step is to properly track SCHEDULED and UNSCHEDULED downtime. Then, you should tackle the process and cultural change necessary to account for NONSCHEDULED downtime.
What other data can be extracted from tracking E10 states?
Hundreds of examples exist. Here are just a few:
- If your weekly STANDBY time is high versus the PRODUCTIVE time, it’s possible that you have underutilized tools.
- If your weekly PRODUCTIVE time is close to 100%, you may have a bottleneck in your process flow.
- If your SCHEDULED DOWNTIME for semi-annual PMs is elevated versus the SCHEDULED DOWNTIME for all other PMs, you may have parts supply problem or a procedure that may need to be examined.
- If your UNSCHEDULED DOWNTIME is elevated for a given week, you may have a performance problem on the tool.
My location has operated for 25 years without any E10 implementation. How would you recommend I “sell” this concept to my managers?
The answer to this question highlights the goal of every manufacturing site: produce the largest amount of the highest quality of parts for the least amount of money. To that end, E10 can help:
- Inform on and support the best capital purchases with real-life data
- Optimize human capital (both engineers and operators)
- Improve the effectiveness of the vendor supply chain and hold vendors accountable to your high standards
- Encourage better cycle times and more accurate delivery commitments to the customer by pinpointing capacity constraints through utilization data.
For more information regarding improving manufacturing efficiency through automation solutions, check out SYSTEMA’s guide to digital transformation.