Scheduling Departments

[Mark]

Something I have run into in many factories is that the function of scheduling production is centrally controlled by a department (or a person) who is highly resistant to letting go of it in a transition to pull.

Ideally, of course, this person would lead the transition and assume responsibility for managing the internal pull loops that make everything work.

But realistically I can also see that this function undergoes one of the most dramatic changes, and this has to be threatening.

Thoughts?
Insights?

[Jimi F]

Mark:

Here were I work the whole production process is run by the Planners. When I say "whole production process", I really mean that.

The Planners do the following:
They purchase any outside parts
They cut work orders for the manufacture of parts in house
They cut work orders for the assembly of our product
They kit the jobs for assembly
They move parts around the plant
They prioritize which parts get worked on next by attaching numbered tags to the WIP (work in progress)

So consequently the Planners are the heart of our manufacturing process. And the success of any lean progress is dependant on them.

Currently I try to Kaizen small parts of the process. But the overall operating system resists change and therefore most of the changes made during a Kaizen event are not maintained. Someday I want to step back and view the overall plant operation and change how jobs are planned and carried out.

[Steve Fonseca]

Hang in there Jimi F. How much of a chance do you get to work with the Production Schedulers? Are there glaring problems with their schedules?

[Jimi F]

I never start improvements with the Production Schedulers. But I always end up including them because all roads lead to them.

Here’s the big picture in a nut shell. We manufacture solenoid valves. We have about 300 final assemblies that we manufacture. Each assembly is made up of about 50 individual parts. Some of the individual parts are made in-house and some are purchased. Some parts go out to vendors for special processes (i.e. heat treat). All 300 assemblies are assembled and tested by us.

Our many customers want these assemblies at different shipping rates. None of the shipping rates are consistent. They all vary constantly. So you can imagine what our Gemba looks like. We are constantly doing part of a job while we wait for parts to arrive. There is stuff sitting all over the place. We move workers around all the time. The flow through each cell is go, stop, hurry up, slow down, etc. I’d like to blame the whole mess on the Production Schedulers. But that will get me nowhere.

Consequently all I’ve been able to do is work on small improvements in different process areas of the plant.

[Steve Fonseca]

Jimi F, How big are these solenoids? I'm thinking they'll fit in a shoe box. Maybe all their parts will fit in a shoe box too. And how much variation is there in shipping orders 10pcs - 5000pcs? Assuming all these wild guesses are relevant, maybe you can have a bin system... when all the parts are there... you start building. Maybe use one box per solenoid, a set of boxes represents a customer order. Your planners can make you a check list for the parts and give you an ETA. Waiting to start the work until all the the pieces are there minimizes WIP and puts urgency on getting the necessary parts. Wow that's scary... if one of the pieces is defective you are back to square one.

[Gary Stiefel]

We have pretty much the identical situation that Mark describes in the initial post. I don't know if it is comforting or discouraging to think that this is the norm.

I have worked closely with one of our plants to implement flow through the value streams. We have had some success in implementing flow from machine to machine, but ultimately I believe we have been limited by the inability to change the ingrained mindset of the scheduling process. The FIFO lanes look beautiful, but we still schedule the plant in one large batch on a weekly basis, complete with a back-and-forth negotiating process with the corporate planning function that consumes several days. I guess on the bright side, we have reduced the frequency of the batch scheduling from 2 to 1 weeks. But even that is often viewed as more work with little benefit.

My role is at the corporate CI level, and I have worked closely with this plant to get to the point we are today. After many attempts to address this scheduling dilemma, here is what we are doing now:
We have chosen one product (that's one product within a larger value stream) to pilot a pull system. I just spent two days at this plant working with the local CI resource to develop the plan for the pilot. The pull system will consist of a finished goods supermarket at the plant which supplies 9 warehouses in our distribution network. The whole idea of this pilot is to go an "inch wide and mile deep vs. a mile wide and inch deep." I have been tracking the inventory (both at the plant and warehouses) and service level of this product on a daily basis. We will use this data to check the results of the pilot.

I just finished reading the book "Toyota Kata" that I know Mark highly recommends, and I can see how both the Improvement Kata and Coaching Kata will apply to this initiative. I am already seeing how following the 5 questions outlined in "Kata" is better than our traditional approach of developing long lists of action items.

I am interested to hear what anyone thinks about this approach and any advice as we go forward.

[Jimi F]

Steve: If you want to get a good idea what we make go to ramcompany.com. You can even watch a 5 minute video found in the "about ram co." section. I'm in the video, as the Lean Manager, doing my Lean thing.

Actually I've made the situation sound worse than it is. About 50% of our orders are consistent and planned very well. It's those other orders that mess up the system. We have a 90% on time shipping record. We do kit our jobs. We have a red tag system. If the kitted order has a red tag that means the job is waiting for parts. As you walk around the plant you'll see red tags on about 50% of the jobs. The supervisors are constantly making decisions on what jobs to do next. The saving grace is a year ago we instituted a morning scheduling meeting. The weekly schedule is discussed in that meeting. And this meeting is what has kept our deliveries at 80% to 90% on time.

My complaint is that every time I try to apply lean principles to a process I run into scheduling issues. Getting the parts to assembly on time always seems to be the obstacle to making improvements.

[Mark]

Jim -
For the 50% of your orders that are consistent, are there regular "runners" that have more or less constant volumes, either as individual items or as families of related or similar items?

[Jimi F]

Mark:

There are very few runners with constant volumes. There are a few products, 5% (10 out of 200) that we ship in constant volumes. These are our “bread and butter” products. Those products that we make in larger volumes that are on a more constant volume schedule.

We have about 15 major processes in the plant and product families share some processes but not all processes. It’s a pretty complex situation. And being in the areospace business we have a lot of quality requirements.

A couple of years ago when we were learning value stream mapping we did a family product matrix. We uncovered those products that used most of the processes. And we mapped a few of those products through the whole plant. We have our major processes laid out in a logical flow pattern. But most of our product does not flow smoothly, like it could, through the plant. Too many variations. Basically we make about 200 items on a consistent basis but the shipping volumes are not constant.

[Steve Fonseca]

Jimi F,

There is a professor at OSU, Dr Irani, that may have interesting technical scheduling analysis tools for you to consider. I disagree with his approach of advocating/creating a flavor of Lean for high mix low volume plants. In general I also avoid computers as much as possible such as using scheduling software; BUT maybe your plant is able to use the tool. I've been to his graduate student's final presentations. The students presented the data from their internships at real plants. They input the data into the program and analyzed the results. It numerically combined shared-resources-tools, spaghetti diagrams, and the different types of runners, costs/revenue, and provided results. I believe that the program was developed with government funding so it may be affordable. He probably has a new and improved projects and I'm sure he would be willing to talk to you. He is currently planning a Conference and info can be found on the web. I found this 2008 email announcement from him...

The federal contract to develop the PFAST (Production Flow Analysis and Simplification Toolkit) software for facility layout and material flow analysis terminated in 2008. Now, we can freely disseminate a demo version of this software. Initially, we developed PFAST to implement Burbidge’s method of Production Flow Analysis (PFA) which allows a radically different approach to design of facility layouts. To some extent, PFAST can be used to design layouts at many levels --- supplier network for a complex product, a factory, any shop in the factory, any cell in a shop and a single machine. Later, we found that PFAST helps to implement JobshopLean, a body of best practices that adapt and extend Lean for small-to-medium sized manufacturers.

There are no strings attached to this offer that is being extended to faculty, students, industry practitioners, etc. J On receipt of your email request for the software, you will be sent a .zip file that contains the .exe file that you must run to auto-install PFAST. Additional installation instructions will also be provided in that email. Depending on your affiliation and residency, we may be able to release the professional version of PFAST to you. In that case, you will simply need to send us your computer’s System ID#, after which we will send you the Serial No. that will activate the professional version of PFAST on your computer. Note that you could install PFAST on as many machines as you wish ex. in a student laboratory using this executable.

And, if you are a faculty at a US university and would like me to teach a 1 or 2-day tutorial on the software and its applications in industry to your class, I would be willing to do that if you bore just the costs for my trip.

Shahrukh A. Irani

Anyway, this is just a cautious thought.