Is Economic Lot Size Dead?

On page 19 of Shigeo Shingo’s masterpiece A Revolution in Manufacturing: The SMED System he writes an extremely strong statement:

Graeme Reeves
June 16, 2009 - 3:23 pm
Reply

Agree 110% because economic lot formulas are nothing but sophisticated overproduction formulas popularized by supply chain “experts” that rarely step foot on the gemba or their warehouses full of excess inventory.
Mark Graban
June 16, 2009 - 3:23 pm
Reply

Shingo is right. But it’s not “EOQ” that’s dead, it’s the assumption that changeover times are fixed that is dead.

EOQ is still a very powerful framework. You have to first reduce changeover time, THEN it makes sense to reduce batch sizes.

Saying “EOQ is dead” might mistakenly lead people to cut batch sizes arbitrarily (I’ve seen companies do this, unfortunately).

I wouldn’t quit teaching EOQ in school or even in lean courses. But you have to emphasize that changeover cost/time must be reduced through SMED principles.
Bill Waddell
June 16, 2009 - 5:29 pm
Reply

The optimium EOQ = customer demand. Anything produced more or less than that is pure waste.
Alan Shephard
June 16, 2009 - 7:09 pm
Reply

I second Graeme and Bill’s comments and Mark’s first sentence.
Ron Pereira
June 16, 2009 - 8:05 pm
Reply

Thanks for the comments everyone. I hope others will comment as well!

Hi Mark – What do you make of Shingo’s last sentence in particular when he said, “The economic lot size lost its entire reason for being when the SMED system was developed.” What do you think he meant by this?
Mark Graban
June 16, 2009 - 8:18 pm
Reply

So to say “EOQ sucks” isn’t very constructive, which seems to be the tone of some commenters. It’s not entirely groundless. If your setup time/cost is big, forcing smaller lot sizes than your system can handle is a path to ruin.

Bill, your comment isn’t the least bit practical. Customer demand over what period? And that translates to lot size in what way?

Ron, I think Shingo was engaging in a bit of hyperbole, as us bloggers also tend to do.
Bill Waddell
June 17, 2009 - 12:08 am
Reply

Mark,

The question is impractical. What kind of scheduling rythm is the plant on? Some of everything every day? Every week? Every month? If the plant attempts to produce some of everthing every week then the lot is size is customer demand for one week.

EOQ is an outdated, failed idea. The amount produced is based on demand over the scheduling period – not some mathematical algorithm like EOQ. It is not hyperbole. It is how you schedule a factory.
Jan Jochmann
June 17, 2009 - 2:46 am
Reply

I agree with Mark that “…it’s the assumption that changeover times are fixed that is dead.”
You can achieve dramatic reductions in setup times, based on that you can dramatically reduce batch size.

But even if you reduce the setup time to minimum, it doesn’t mean that you can economically produce any batch size or 1 piece for example. It is dependent from many other factors – type of process, process organization, workload, flow of product…
For example, it is a big difference from the economical point of view to produce 10 pcs of something on a right-sized-equipment which is a part of the production process with one-piece-flow production, than to produce the same 10 pcs in a shared equipment in a departmentaly organized factory. Even when those two equipments have a setup time of 1 minute.

Whether you call it EOQ or “common sense”, it doesn’t matter.
Mark Graban
June 17, 2009 - 4:46 am
Reply

So Bill, how do you determine if the Every Product Every Interval is monthly, weekly, daily, hourly, or in 6-minute pitches???

There must be some intuition or math to get you to that point. EOQ isn’t an “algorithm” in the sense of a computer optimization algorithm with 100 inputs (like some failed ERP system). It’s like F=MA. Is that also an “algorithm” that you have distaste for, a basic law of physics?

Waddell in a Grocery Store:

Clerk: “OK that will be 19.02”

Waddell: “Here is a 20”

Clerk: “The register says your change is….”

Waddell: “Whoa whoa, you and your fancy algorithms!! I bet that cash register was designed by Ivy Leaguers. I don’t trust it.”

Clerk: “But it says to give you 92 cents, I think that’s right….”

Waddell: “Just give me a handful of pennies…. that’s the perfect small batch money. A handful is probably close enough, I don’t have time for your fancy bow-tie thinking….”
Paul Cary
June 17, 2009 - 6:11 am
Reply

Shingo’s atatement was more of a mindset than and absolute. If we accept that
one piece flow or the quantity that the customer needs at any given time (not orders) is the goal (the pull) than we are relentlessly pursuing what stands in the way of achieving the goal i.e. set up, cycle times, supplier constraints, maintenance etc.. Lean thinkers and practitioners should never accept that one piece flow is impossible to achieve.
Del Ames
June 17, 2009 - 7:00 am
Reply

We have been using the Work Cell/Team concept for sometime now and have found that small lot size are better of process control, better flow, shorter cycle time, less WIP and less impact or exposure to bottom line. In a mass or high volume setting smaller lot sizes can have an impact if you continue to think large volume concept/handling. That what I feel Lean teaches you is to adjust and think outside the box.
Ron Jacques
June 17, 2009 - 7:02 am
Reply

I whole heartedly agree with what Paul is saying. We cannot be so fixated on just constraint management (SMED on a given piece of equipment) when the goal is servicing Takt. We need to balance our change over /cost reduction efforts with that of demand with the ultimate goal being to move towards one piece flow, which gives the ability to maximize mix, volume and customer value add and to go after more smaller customers instead of just large volume accounts. Let demand drive the improvement process and focus on keeping the process in balance instead of creating fire fights by moving constraints constantly. EOQ is just that, the “economic” order quantity. If it is economical to produce just one because we can change over fast enough to be economical, then one it shall be. Otherwise it is what it is given our current process capability. The goal should be an EOQ of 1.
Bill Waddell
June 17, 2009 - 6:45 pm
Reply

Mark,

Whatever the current set-up/changeover time is drives whether a factory can produce in monthly batchs, weekly, daily, hourly or 1 piece flow. That determination is a function of run time to meet demand plus the current chageover times relative to available capacity.

The fundamental objective of lean manufacturing is cycle time compression. The factory that can produce in weekly intervals has a much lower overall cost than the factory that can only produce in monthly batches due to excessive setup/changeover times. So the goal is to reduce the setup times to produce in shorter intervals. Continuous improvement is measured in progress toward shorter interval production. Once the factory achieves consitsent on time customer delivery in monthly intervals, their goal should be to ship on time with weekly scheduling intervals, then up the bar to daily intervals, etc… striving for the probably unattainable goal of 1 piece flow.

This is hiejunka 101.

EOQ is a meaningles calculation that rationalizes inventory, rather than reduce or eliminate it. It is driven by the “cost of setup time” which is unknown and unknowable since it is based entirely on accounting allocations and assumptions of fixed and variability that are accounting assumptions and have no basis in reality.

That is Lean Accounting 101.

This understanding of lot sizing and the relationship of setup reduction to factory flow and overal cost defines Shingo’s fundamental contribution to the Toyota Production System.

You can reject Shingo and toss childish insults at me all day long but it doesn’t make EOQ a relevant tool for a lean manufacturer, nor does it negate the fact that setup/changeover reduction should always be a high priority in achieving optimum overall cost.
Gary Leech
June 30, 2009 - 2:10 pm
Reply

Bill,
Succint I’m with you,

Mark…get a grip! When i went to school; 20-19.02 was 98 cents
Observer
July 12, 2009 - 12:27 am
Reply

I think the assumption here is very simplistic. Different factories, locations, equipment, scales have different constraints. SMED made set up time to be thought as reducible. In this new context if EOQ still makes sense to some one, let them follow it. Don’t try to make universal maxims.

Popular concepts generally apply to a large set of common situations. They do not become the rule.