Column Archive: May 2012

Dear Gemba Coach,

How do you suggest we should do lean in a foundry?

The glib answer could be: just as anywhere else – but I’ve done a couple of gemba walks in foundries and I have to say that I find the environment somewhat daunting. The foundries I’ve seen vary considerably from one another, but one thing for sure: it’s a demanding process. On the one hand the process is quite heavy going – it’s physically demanding, dirty and, well, hot – and on the other precision casting require a sophisticated handling of technical parameters.  This is definitely a case where I would recommend both 5S and a robust six sigma program. One place I would not start is what I’ve seen on a few occasions: kaizen events to eliminate waste by reducing operators’ walking distances. This is not negative, as such, of course, but is unlikely to deliver the hoped for results.

Like most process industries, foundries have large equipment and relatively few workers spread around the place – although I’d have to distinguish the pattern crafting part of the operations from the actual casting side of things. Lean experts coming from an assembly work tend to start with a “seven wastes” perspective and immediately show that operators and parts are moved around quite a bit, which is sheer waste. This is certainly true, but has an element of a solution looking for a problem, like looking for the lost key under the lamppost because the light is there. If you’ve considered starting by cutting headcount costs by reducing operator movements, I’d suggest you take a step back and, well, think. The question, as always, is “what is the problem we’re trying to solve?” and hence, how do we see our real problems as opposed to what is immediately apparent.

The issue with wearing lean glasses in any new setting is that wastes will jump out everywhere, but I don’t have any experience base to distinguish real problems (with a high-pay off potential on sales, cash and bottom-line) from mundane issues (indifferent as regards sales, cash, and the bottom-line), no matter how “obvious” the waste might be, such as operators walking all over the foundry for very little apparent value-added time. For what it’s worth, I do have a standard or sorts to approach lean in an unknown environment:

1. Safeguard the people
2. Protect customers
3. Control the lead-time
4. Reduce the lead-time
5. Eliminate costs at the source

Forging Ahead Safely

How would that standard apply to a foundry? To start with, we have to accept that a foundry is an aggressive working environment. Those I’ve visited tend to be dark, dusty and filled with very hot objects. I rarely recommend starting a lean initiative with 5S because of the many disappointments with this over the years, but this is clearly a case where a 5S drive applies. The problem we’re trying to solve is not cleanliness per se, but involving operators in improving their own working conditions. Many kaizen workshops tend to be structure heavy with one or two “alibi” operators. In this case, we should think of a workshop plan by and for operators. The idea is to organize events so that every operator of the foundry has participated at least in one event in the year. There is no need for more than one facilitator per workshop, and the idea is to use the 5S methodology with a special focus on safety.

As a plant manger, as you progress through this plan set aside a small budget for “cleaning up” each area touched by the event – fixing what is clearly broken down, a coat of fresh paint, maybe – this should be running costs, not investment. Overall, getting a foundry spick and span is a daunting task, so we break it down in kaizen steps: the ideal is that over one year, every part of the foundry has been spruced up and every operator has participated in the effort to improve the working environment. As a plant manager, you can make a point of showing up at the end of the workshop and personally agreeing to what is feasible and what is not. You’ll find that this will considerably benefit your relationship with your workers.

What’s Kilning Delivery

Protecting customers is of course about quality and delivery. Before we tackle the quality issue, let’s start with delivery. Many manufacturers I know keep complaining about the poor delivery performance of their cast parts suppliers. This issue is not specific to foundries but tend to touch every flow process business, where the batch mentality still rules (mostly because of the difficulty and high perceived cost of change-overs). To understand delivery issues, the first step is to create a group to handle a weekly production plan based on understanding real customer demand. The trick is separating high-volume parts (even in a low-volume environment, you can safely bet that less than ten percent of all product references will account for more than 50% of total volume) from low volume ones, and then schedule the high-volume parts every day. This might (or might not) put a strain on the process flexibility, but it will have the merit to clarify your understanding of real demand. What I usually find is that if you focus on making the high-demand parts every day to a leveled schedule (since they’re high demand, demand variation is overall lower), it leaves ample production time to produce make-to-order one-offs as well.

Secondly: quality. Quality in any process industry is a matter of finding out which are the right parameters to focus on. This is by no means obvious, and certainly six sigma will help. To get started, you can task a technical group to set up SPC tracking of key parameters not, at this stage, to reduce variation, but to figure out which parameters have the greatest influence on quality. Sure, temperature control is important. Certainly, sand quality matters. But this can be true of any foundry. The challenge is to figure out what parameters matter in your specific foundry, and this is by no means obvious.

EPEI vs. MRP

Once you have a better handle on working conditions, quality and delivery, you can start on the next phase, which is controlling lead-time. In the cases I’ve seen this comes down essentially to two key problems (1) getting the right rotation of products through casting and (2) getting patterns in time from crafting. The lean tool to understand crafting programming is Every Part Every Interval (EPEI): the set sequence of parts production so that the lead-time of any individual part can be known. This goes counter to the MRP logic of programming parts according to calculated instantaneous demand, but is hugely helpful in settling down the process and seeking “economies of repetition”. Exploring Every Part Every Interval will also lead you to much better understand how various parts you produce behave, from the high runners to the occasional repeater to outlandish exotics.

Pattern making remains in many cases a craft operation and I’d start there with straightforward production analysis boards to see whether we’re making patterns at the expected rhythm in order to control the overall lead-time. If handled properly by checking the boards several times a day and discussing comments with the pattern-makers, the various causes for production loss should become visible, and highlight the many opportunities for kaizen.

Once lead-time is under control, you’ll have a better idea of what value streams are going through your foundry, and how different parts behave with different lead-time conditions – these different parts need not be treated the same in terms of scheduling – high runners can be pulled, and make-to-order programmed, for instance. Once your management team has clearer ideas about how to handle these value streams, it’s time to bite the bullet and reduce lead-time. In casting, as with most flow operations, this is a matter of increasing the frequency of change-overs – which, in turn, means single-minute exchange of die (SMED), no two ways about it. At this stage I don’t know any other way than finding a local SMED consultant and setting up a robust SMED program with the aim of systematically reducing the Intervals of the EPEI – this will highlight many shortcomings and deficiencies of your technical process, but there’s nothing doing – no pain, no gain.

But what about cost reduction? Keep in mind that the full lean principle is cost reduction by waste elimination. Obvious wastes, such as operator movements backs and forth, are not necessarily high potential wastes, such as poor scheduling. The previous exercises should keep you busy for a couple of years and, more importantly, should lead you and your team to a more detailed understanding of what they key wastes occur in your operation as a result of the way you currently run things. I have no doubt that as you get more expert at recognizing the wastes you create you will also find countermeasures to eliminate these wastes and your costs will come down. I realize there’s something of a leap of faith in this statement, but your costs will come down.

Lean as usual you might say - maybe.  However, the foundry remains a rather extreme environment where people’s involvement is critical, so before you roll out the plan, let me back track to the essential point of starting with a 5S/safety workshop for each operator team and a six sigma project program for technicians. The immediate aim is to get people working together in taking ownership for their physical and technical environment, and work improvement strategies from a basis of teamwork and shared challenges.

Dear Gemba Coach,

It seems that lean these days is all about learning and coaching. What about the good ole-fashioned just-in-time stuff of the earlier days?

Touché – and I plead guilty. The other day I was on the gemba of a plant that makes industrial equipment and had exactly the same discussion with the plant manager. This fellow has an excellent, intuitive grasp of respect and a deft touch with people. He’s really good at getting people to work together and has a gift for fostering both change and goodwill. But his corporate management is asking for immediate productivity results, by the book and by the numbers. Hmm.

Now, in a low-volume high-variety business run by an MRP, quick productivity results are hard to find. There is no high-runner cell with ten operators that can be quickly kaizen-ed down to five. There are a lot of grinding, turning, milling machines, a monument of a paint shop, a few craftsmen-like assembly stations, and many operators milling about according to where the supervisor needs them. Not surprisingly, on-time-delivery is a struggle, and productivity is poor. This is not seen as a problem since the margins per product are high, but we all know what senior execs are like when it comes to productivity figures (and their own bonuses).

To make matters worse, in a high-variety environment, any increase in volume means an increase of one part or component missing, which requires a schedule change, leading to less OTD and productivity … so we won’t be saved by volume. That sucks. The strategy then, is to ship on-time-as much as possible, in order to get cash in hand from the customers and pacify corporate with cash and bottom-line. But how?

Not with Safety Glasses
We need to look at the same situation with several different pairs of goggles. One, the respect glasses, reveals how everything we ask operators to do that does not add value or does not contribute to shipping products is a lack of respect. Guys are here to work, not to fill in paperwork or recount kanban cards. The more focused their work, the more productive, and the more likely we are to ship on time. But how can we focus on the right problems?

We use the other pair of glasses: the just-in-time system, and a view of the entire factory as if it was a mechanical engine. We were looking at the value-stream maps his team had been doing, and they had the usual flaw of detailing the material flow and sketching the information flow as an afterthought. Value-stream mapping (VSM) is part of a lead-time analysis, not a process analysis. So two critical aspects of VSM are:

1. The number of references made on each machine, since production lead-time is calculated as the time between the last part of A from the previous batch to the first part of A from the next batch (with batches of B, C, D, etc. in between).
2. The frequency at which information is updated in the information flow: once a week? Once a day? Every hour? Whenever something happens, and so on.

The main problem of a high-variety environment is that although everyone is busy working, they’re usually working on the wrong part. As we walked the shop floor we checked the work orders: either ahead (parts needed for next week, aka overproduction), or late (parts needed yesterday, aka waiting). So we tried to look at the production process as a dynamic system of information converted into action, and then into parts. What tells who to make what when. The more frequent – and the more regular – the update, the better chances of making the right stuff at the right time.

Want Fries with that Kanban?
Developing the dynamic vision of just-in-time is by no means intuitive, and it requires learning the techniques and discipline of a pull system. For instance, they’d set up a vague form of kanban with a launcher that had many, many cards in it. Now, any restaurant works with kanban. The front waiter takes the orders when people sit down, and then puts them in front of the cook to make in the same order. The cook picks up each order and either (1) it’s a today’s special so she takes it from the supermarket of ready dishes or (2) it’s à la carte and she makes the dish from scratch. Then it’s shipped. Like a complex product, a table has to receive several different dishes at the same time, but hey, restaurants mostly get it right – so why not factories?

Back at the gemba we looked at the launcher and we calculated that the machining team had orders for the next three days. This is massive overproduction of information, and leads to a number of problems. First, how likely is it that the kanban cards of three days in the future truly represent what will really be needed then? Secondly, with three days of cards ahead of you, how tempted would you be to pick and chose the ones you feel are urgent, or easier to make? Thirdly, and this is where just-in-time dovetails with involvement and engagement, with three days of work ahead of you, how determined would you feel to get the work down rather than have another smoke before changing the tool setting once more? How involved would you feel with the company’s determination to ship to its customers?

There is a sly underhand trick to pull: it completely changes frontline management’s job. The supervisor can no longer decide what to do when; she’s got to follow the cards, just like the cook. The supervisor must now focus on doing the job right, much like the cook again. Some supervisors will love it, feeling liberated from self-defeating decisions, and some will hate feeling technically challenged and losing petty boss’ power.

Furthermore, this also means that information management and material handling is taken away from the operators’ job and given to a specific logistic function: the train that delivers parts and picks up cards, cards which then need to be sorted and redistributed. When this happens we suddenly realize quite how many tasks operators have to perform that have nothing to do with making parts.

The difficulty, of course, is to walk on two feet – to hold the two different visions – just-in-time and respect-for-humanity – at the same time and slowly learn to figure out how they converge. This was, I believe, already the point of the very first paper on lean published by Toyota veterans back in 1977, although at that time much of the paper was devoted to calculating kanban cards. We’ve learned the hard way that respect is as important as kanban, but maybe we’ve now overcorrected. Without both a system-level and card-level understanding of JIT, respect becomes wishful thinking and will not satisfy any of our stakeholders: neither the operators (we’ll “respect” them on the wrong topics – they’re not fooled that easily, involvement comes from sharing the company’s objectives) and neither corporate. Never easy, but let’s face it, never boring either!