Predicting the benefits of “Lean Actions”

In the TPS + 1 ENGINEERING group on LinkedIn, Hela Hassine asked  “How can we predict and quantify the profit of lean actions before implementing them?”

I see three types of what Hela Hassine call “actions”:

  1. For some, you can do a complete discounted cash flow analysis before implementing. Cellularizing a job-shop falls into this category.
  2. For others, you cannot calculate the benefits ahead of time, but you can measure them afterwards. When you improve quality, first you can’t tell ahead of time by how much it will actually improve, and second, you can’t tell how much good this improvement will do to your business. After you have improved quality, you know by how much, and you can also measure the market impact of the improved quality, which is its dominant benefit. There is no way you can justify quality improvement ahead of time through cost-of-quality analysis.
  3. For the rest, the benefits are too diffuse to be measurable. 5S falls is in this category.

This has obvious consequences on implementation sequencing, that are often overlooked. Projects that lend themselves to a-priori justification are easiest to sell to management, and success in such projects gives you the credibility you need to undertake others with less tangible benefits. In other words, you are better off starting with cells than with 5S.

A Tour of Canon’s Suzhou facility | WhatTheyThink | Eric Vessels

See on Scoop.itlean manufacturing

In addition to the Tokyo headquarters visit I wrote about Friday, analysts and editors were also invited to Shanghai before leaving Asia to attend a plant tour of Canon’s Suzhou facility.  The facility is located in the Jiangsu province, an hour and a half bus ride west of Shanghai.

See on

The baton-touch approach

The following question came this morning from Diogo Cardoso:

What is baton-touch in terms of product oriented manufacturing systems? I have made a deep research about this on Science Direct and other resources but I can find nothing more than an inconclusive paragraph.

Your researched the wrong sources. You could have found your answer in Working with Machines, pp. 140-142. Baton-touch is one of three approaches used to design operator jobs in cells, the other two being the caravan/rabbit-chase and bucket-brigades. The key differences are as follows:

  • In the baton-touch method, each operator performs a fixed subset of the cell’s operations, organized in a fixed sequence. It is commonly used in cells requiring three or more operators making a narrow range of products with similar work content.
    The baton-touch

    The baton-touch

  • In the caravan or rabbit-chase method, the operators follow each other through the entire sequence of operations in the cell. It requires each operator to be skilled in all the operations of the cell, and works well with up to two operators but breaks down with three or more operators, as they queue behind the slowest member of the team.
    The caravan/rabbit chase

    The caravan/rabbit chase

  • In the bucket-brigade method, the operators are in sequence, but the scope of each operator’s tasks varies. When the last operator finishes a unit, he or she takes over the next unit from the preceding operator, who in turn takes over from his or her predecessor, and so on, until the first operator, who starts the next unit. Bucket-brigades are used with a broad mix of custom or configurable products, and work when the faster operators are always downstream from the slower ones. For details, see John Bartholdi’s article on bucket brigades.
    Bucket brigades

    Bucket brigades

Questions from an Industrial Engineer in an Automotive Machine-Shop

I received the following questions from an Industrial Engineer (IE) who has recently moved from vehicle assembly to the machining of car engine parts, blocks, heads, crankshafts, etc., activities that all new to him:

  1. Any reading material you would recommend?
  2. Is takt based off the slowest machine or the machine in the line that makes the least parts?
  3. Knowing cycle times and uptimes of a 30 machine line how do you calculate system uptime?
  4. Should there be more overspeed for machines at the beginning half of the machine line?

Following are my answers:

Any reading material you would recommend?

Industrial Engineering, as taught in universities, is generically about how people work, and gives you no process-specific knowledge. Manufacturing Engineering (ME), on the other hand, is heavily focused on metal working, as if these were the only processes worthy of the name “manufacturing.” To be effective as an IE in a machine shop, you need some familiarity with whatever processes are performed in your shop, such as turning, milling, drilling, reaming, broaching, grinding, or heat treatment. You don’t need to master them, but you need to know them enough to have a meaningful conversation with specialists. And you also need to know about the key operational issues of the machines used to perform these processes, such as lathes, machining centers, drill presses, etc. including how parts are loaded and unloaded, jigs and fixtures, cutting tools, and NC programs.

You will find more than you need to know in books written for MEs, like Degarmo’s Materials and Processes in Manufacturing. I would not attempt to read it cover to cover but instead use it as a reference, to cram on any process you are actually working on. There are other similar books, but this one was co-authored by J.T. Black, who was quite possibly the first American academic to recognize the significance of Lean and make it central to his teachings. My own book Working with Machines is about all types of manufacturing activities that involve the interaction of people and machines which includes automotive machine shops. It includes discussions of takt time, OEE, and availability.

The two main industrial applications of machining are automotive and aerospace, and the two are quite different. In automotive, you remove small amounts of metals from many parts that have been cast or forged near their final shape, in commonly available alloys and in high volume; in aerospace, by contrast, you remove 90%+ of the metal from slabs or forging that look like caskets in exotic alloys and in low volumes. Your needs are in automotive, so don’t waste your time studying approaches that are only used in aerospace. The literature does not always make this distinction obvious, so you have to be on the lookout.

Is takt based off the slowest machine or the machine in the line that makes the least parts?

The takt time is not based on machines but on demand and net available production time. If you have a line that puts out completed parts one unit at a time at fixed intervals, the takt time is the length this interval must have in order to meet the demand within the net available production time, which is the time you can count on the machine actually processing parts. It is not a parameter of your slowest machine but a requirement that even it has to meet.

Knowing cycle times and uptimes of a 30 machine line how do you calculate system uptime?

As you know, uptime ratios are multiplicative, so that, if you have a line of 30 machines, each of which is up 85% of the time, you line is up 85\%^{30}= 0.7\% , which is obviously not workable. But 99% uptime on each machine still only gives you 99\%^{30}= 74\% , which is still too low. So what do you do?

First, you don’t put 30 machines in line. machining cells usually have 5 to 10 machines, including simple, auxiliary machines that rarely break down. And you have buffers between cells that are managed by pull. A cell of 10 machines with 99% uptime will be up 90% of the time. With 5 machines, 98% uptime on each machine is enough to give you 90% on the whole.

Second, you work on improving the machines and customizing them to your needs so that they have fewer breakdowns and can be changed over faster, and you use these improvements to increase the number of machines in line.

Should there be more overspeed for machines at the beginning half of the machine line?

The takt time is set for the entire line. The line meets this requirement if, and only if, the last machine puts out one unit of the product in question like clockwork at the end of every takt interval. For this to happen, you must not only make sure that this last machine is up and running but also that it has a part to work on, and one way to ensure this is to give all the upstream operations a modicum of slack. This strategy, however, works better in manual assembly, where much of the work can be reassigned backwards and forwards among assembly stations in minuscule increments, where you cannot ask a lathe to do milling and vice versa.

The minimum takt time a machining line can support is determined by the capacity of its bottleneck machine, which is usually not last in line.

Increasing Subassembly Productivity

See on Scoop.itlean manufacturing

In my time spent onsite with the customer implementing PFEP (Plan-For-Every-Part) and advanced material flow techniques, I often was pulled into other projects. One of these projects was an effort …

Michel Baudin‘s insight:

This is a rare post on assembly engineering, dealing with the layout of subasembly cells for a mixed-flow line. This is the red meat of Lean, ignored in most of the English-language literature on the subject. Kudos to Kelcy Monday for getting involved.

Reading this, I can’t help but thinking of many issues I would have handled differently, but I have not seen the product of the shop floor. In any case, this is the right opportunity to work on, with order-of-magnitude performance improvements at stake, as opposed to the 5% others might have nibbled by  applying 5S on the old layout.

See on

5S and multiskilled operators

In a Lean plant, we expect to see a tidy, uncluttered shop floor with high visibility as a result of 5S, and skills matrices on performance boards that track the cross training of the operators in the different tasks performed in that shop. 5S and multiskilled operators are both features of Lean that we do not, a priori, consider as linked. But in fact they are, and the feasibility of implementing certain aspects of 5S is in fact contingent on having multiskilled operators.

For example, assume you are running a traditional machining job-shop. You have a turning center, a milling center, a drilling center, a grinding center, etc. In each of these centers, you have a farm of machines performing only one type of operation and working in parallel. Each job follows its own path from center to center, with a document called traveler showing the list of operations with check marks for the operations done to date. And each center has single-skilled operators, usually able to operate just one machine, or a bank of identical machines, as seen in Figure 1, with the orange areas showing WIP locations.

Figure 1. Machining job-shop

If you try to implement 5S in this context, you will be telling a machinist with 15 years on the same machine to put hand tools on a shadow board and label every location. But the machinist knows where everything is, and sees no value in this exercise. The only clear point is that 5S would make it easier for somebody else to take over the job. And since this machinist doesn’t know how to do anything else in the plant, it is not an attractive proposition.

On the other hand, assume you first set up cells in which each job makes a machinist operate several machines, and the cell operators rotate between jobs, as shown in Figures 2 and 3.

Figure 2. Machine shop with cellular layout

Figure 3. Operator jobs in a cell

Then the shadow boards and labels come in handy and are well received. The tooling is shared among several operators, none of whom “owns” any of the machines (See Figure 4).

Figure 4. Labeled tooling positions in a cell

In other words, if you try to have assigned and labeled locations for tooling in a traditional job-shop, you will get nowhere with the machinists. On the other hand, it is indispensable when you operate with multiskilled operators, and they will cooperate in making it happen.

Fab manager tries Lean with no support from the top, by starting with 5S…

See on Scoop.itlean manufacturing

Tim Heston reports a conversation with the manager of a low-volume/high-mix fabrication shop who wants to implement Lean, without top management support, and starting with 5S, and it’s not working.

Two thirds of the article are not just about 5S but about the tool hoarding behavior of operators. Yes, organizing workstations with commonly used tools makes sense, but, if the manager starts by addressing this head-on, he will have a mutiny on his hands and his bosses won’t back him up.

To be successful, changes in tool management policies should be part of more major changes, such as the implementation of SMED on a machine, or the development of a machining cell. Once you have a team of operators who move between stations and rotate positions, then  tools naturally become attached to stations rather than individuals.

What should the manager do? I am currently reading Art Byrne’s Lean Turnaround, and, maybe, getting his CEO to take a look at it might be a good idea to get him or her on board. Next, he should get better advice getting started than focusing on 5S. Much of the literature recommends it because it looks easy. It’s not, and it almost never works as a first step.

See on

A Lean Journey: Meet-up: Michel Baudin

See on Scoop.itlean manufacturing

Interview on Tim McMahon’s A Lean Journey.
See on

Cellular manufacturing for web-customized wedding dresses

See on Scoop.itlean manufacturing

“…rather than be part of a large group working on an assembly line, workers are divided into small teams, with each one headed by a high-skilled leader. The cells each produce three or four dresses a day according to very specific designs that require a lot of handwork. Taken in total, each company can produce about 200 dresses a day that way, with each cell producing a very different dress…”

See on