ON-DEMAND LEAN PRODUCTION

by Dr. David M. Anderson, P.E., CMC
Build-to-Order Consulting
Copyright © 2010 by David M. Anderson

    The ability to build mass-customized and standard products on-demand is the payoff for lean production programs.   Lean Production is taught in the context of build-to-order and mass customization  through Dr. Anderson's in-house seminars and implemented through his leading-edge consulting.

    Lean production¹ is a key prerequisite for Build-to-order and mass customization. The key prerequisites of lean production are product line rationalization and standardization which simplify both the supply chain and manufacturing operations. This will make implementation easier and faster and ensure the success of lean production as well as build-to-order and mass customization.

    There are two types of lean production: replacement and spontaneous build-to-order. In replacement lean production parts are common enough to be already built and available to be pulled into assembly from kanban bins.  If not, then parts are made by spontaneous build-to-Order with common parts made available through kanban and the non-common parts built on-demand from standard raw materials by CNC machine tools or manually from on-line instructions.

    The most important attribute of lean production is the ability to build products quickly and efficiently in batch-size-of-one. In order to do that, all setup must be eliminated including any delays to kit parts, find and load parts, position workpieces, adjust machine settings, change equipment programs, and find and understand instructions.

    Mass Production deals with setup by accepting it as a "necessary evil" and then spread it over as many products as possible in batches or lots to minimize the set-up "charge" per part. For decades, industrial engineering have used formulas to try to calculate the "Economic Order Quantity" (EOQ). However, manufacturing in batches drastically raises costs and lead times because of the following considerations:

C Space. Batched parts occupy much more space than a single piece flow, especially if batches are so heavy that fork lift aisles are needed.

C Throughput times. Batching parts slows throughput because, at each step, the batches wait in line for setup changes and the processing time of all the parts in the batch.

C WIP inventory. Batches create WIP inventory; inventory carrying costs are 25% of the value of the parts per year.

C Defects. Parts made in batches could be made with recurring defects which may not be determined until after hundreds of defective parts have been made.

C Disruptions. "Rush jobs" can cause major disruptions to scheduled production by adding two additional setups for every affected part. "Treasure hunts" may be needed if there are any part shortages due to errors in forecasting, bills-of-materials, or inventory counts. Womack and Jones conclude that treasure hunts "are the distribution equivalent of the expediting always necessary in batch-and-queue production operation."²

C Flexibility. Because of all of the above, batch & queue manufacture is not flexible and therefore does not support build-to-order and mass customization.

    For an summary about the shortcomings of mass production click here to see the editorial, "End of the Line for Mass Production; No Time for Batches and Queues."

SETUP & BATCH ELIMINATION ³

    If successive products are to be unique and different, there cannot be any significant setup delays to get parts, change dies and fixtures, download programs, find instructions, or any kind of manual measurement, adjusting settings, or positioning of parts or fixtures. For a plant to mass customize or spontaneously build products to-order, all product setup must be eliminated, not just the low-hanging fruit or reduce setup as "much as you can." Definition: "eliminating" setup means that setup is reduced to the point where it is still feasible to operate efficiently in a batch-size-of-one mode.

Note that much setup is designed into the product and process.

C Distribute parts at all points of use (eliminate kitting)

If part variety is too excessive to allow distribution at all points of use, then enough parts for a batch must be assembled into a kit which is put together in the raw materials warehouse, delivered to the assembly area, and then distributed to part bins or automation machines. This kitting is a set-up which will inhibit flexibility.

Tools to eliminate kitting are aggressive standardization and the concurrent engineering of products and processes that minimizes the number of different parts at any assembly station.

C Tool & tooling setup

Plan A: Eliminate setup. Design the product/processes to eliminate the need for tooling changes for cutting tools, dies, molds, tool plates, and fixtures. Tool plates and fixtures can be designed to be versatile enough to accept a common blank which then can be customized by computer controlled machine tools or robots, as shown in both illustrations in the mass customization article.  The blank must quickly be positioned in the fixture without any need for measuring and manual positioning. Thus the blank must be designed with common fixturing geometries.

Plan B: make setup changes as quickly as possible. There has been much progress and much written about rapid die changes.⁴ Shigeo Shingo developed the methodology called "Single Minute Exchange of Dies" (SMED) for Toyota.⁵ Clever die and mold mounting geometries have been developed to facilitate quick changeovers.⁶ Conveyors and carousels, that were first applied to moving parts and products, are now being applied to moving dies quickly in and out of presses and molding machines.⁷

C Consolidate inflexible parts. Parts needing dies for molding, casting, or stampings should be designed to be versatile enough to accommodate all products that are supplied by each production machine.

C Use of CNC. CNC machine tools are very versatile tools to eliminate setup since programs can be changed quickly and electronically. However, physical setup must be eliminated, for example, workpiece positioning and tool changes. Products may need to be designed for CNC to completely eliminate setup.

C Maximize the amount of dimensional variation done with CNC

C Standardize raw workpieces and fixturing to eliminate setup

C Quick and automatic program change

C Standard cutting tools within tool changing capability

C Automatic material feed and eject (optional)

C For sheet metal, nesting optimization (can evolve over time)

C For unusual and low-volume parts, using a CNC machine to "hog out" parts may lower total cost if it avoids (a) stocking a high variety of low-volume parts or (b) complications and delays in the supply chain for low-volume castings or moldings.

C To make the right decisions on flexible use of CNC, total cost must be used to include machine time, material cost, and all related overhead costs.

C Manual processing setup. All of the above setup elimination strategies apply to manual processing. But a setup that applies uniquely to manual assembly is finding and understanding instructions. This setup can be eliminated by displaying instruction on monitors that instantly and clearly show what is to be done at that area to any product being worked on.

C Consolidated parts. Inherently inflexible parts (like castings, moldings, and bare PC boards) may need to be consolidated into very versatile standard parts which can be ordered as a steady flow with confidence that, because of their versatile, they will be used one way or another.

C Arrange kanban resupply. Parts that qualify for kanban resupply can be made in batches if the combination of setup time, run time, and delivery time is short enough.

    The typical response when suppliers are asked to deliver parts just-in-time to their customers’ pull signals is to keep building the parts in large batches, try to stock enough in their finished goods inventory, and meter them out "just in time." However, this is not really just-in-time and it is certainly not conducive to spontaneous BTO. Parts availability would depend on the assemblers’ forecasts, which are becoming increasingly less accurate, and the supplier’s inventory, which is costly to carry, especially as obsolescence risks increase. There are four basic techniques that contribute to a spontaneous supply chain:

(a) Kanban resupply. As mentioned in the third point above, parts that qualify for kanban resupply, and the related techniques of min/max and breadtruck, can be made in batches as long as the response time and bin (or delivery) size is adequate. Even though parts are made in batches, this still qualifies for a spontaneous supply technique because the batch (a bin’s worth of parts) is made upon the pull signal that the current bin has emptied. Of course, the parts manufacturers may have to implement setup reduction to make small batch production economical. Thus, kanban resupply avoids the hazards of forecasting, the cost and delays of purchasing, and the cost and risk of inventory. The resupply is automatic once the pull signal gets to the supplier.

(b) Spontaneous build-to-order of parts. For parts that do not qualify for kanban, suppliers themselves would need to implement spontaneous BTO so that they could actually build on-demand to their customers’ pull signals. This is the only way to supply mass-customized parts on-demand, which may be needed for mass-customized products. Spontaneous BTO of parts may require (1) the development of vendor-partner relationships for suppliers to establish the ability to build parts in any quantity on-demand and (2) versatile information systems to process and distribute the necessary information.

(c) In-house part fabrication. In order for spontaneous BTO to work, all parts and materials must be available on-demand. If there are any key parts that are not suitable for kanban and no supplier can build them to your pull signal, then you might have to bring those operations in-house. Companies that have outsourced certain operations in the interest of focusing on functional "core competencies" may have to reevaluate their strategies. Unfortunately, most outsourcing is a batch operation which does not lend itself to spontaneous BTO.

    If the new core competency is to be spontaneous BTO or mass customization, then the manufacturer will need a complete supply chain that can build products and all their parts on-demand. This may require the selective "reintegration" of certain key steps.⁸  One of the author’s clients, Badger Meter, of Milwaukee, Wisconsin, found it was able to build a wide variety of water meters flexibly except the printing of the face plates, which had to cope with several ways of measuring water flow plus the logo of every customer (utility). So they learned how to print face plates in small quantities to complete the picture.

(d) Strategic stockpiles. Strategic stocks may be necessary until one of the above three techniques can be applied. As far as overall inventory strategy is concerned, this could be considered temporarily moving one step backwards after moving twenty steps forward. Hopefully these parts are standardized and consolidated so that there would be few to stock and each would have a good chance of being used one way or another.

    If setup can be eliminated or reduced enough to eliminate the need to manufacture in batches, then parts, sub-assemblies, and products can flow one piece at a time. One-piece flow may be essential when building to-order a wide variety of standard or mass-customized products.

    It also eliminates much of the waste of batch-and-queue manufacturing: waiting, interruptions, overproduction, extra handling, recurring defects, and other non-value-added activities.

    One-piece flow has a distinct advantage for assuring quality at the source. First, flow manufacturing eliminates the possibility that recurring defects may be built into several batches before being caught at a downstream inspection step. Second, people working in flow manufacturing look for any visible deviation as each part is handed to "its customer" (the next station). Further, if the part doesn’t fit or work in the next operation, the feedback will be immediate leading to quick rectification of the problem.

    In flow manufacturing, parts may be manually handed to the next station, which may be very close, thus eliminating the need for mechanized conveyance or fork lifts, whose aisles may occupy as much space as the production line.

    One-piece lines are usually sequential, sometimes breaking into parallel routes when needed to balance the line (see next section). Rather then laying out "lines" in a literal straight line, it may be advantageous create a U-shaped line which bends the line into the shape of a "U" for the following reasons:

C Visual control. Everyone in the line can see the whole operation, enhancing visual control, thus resulting in greater group ownership, continuous improvement (kaizen), and problem solving. Visual control can be further enhanced with clearly visible andon (warning or status) lights and display boards.

C Problems heard. When everyone in the line works close together, problems at all stations will be heard by the entire line, thus leading to faster problem identification and resolution.

C Helping out. If one worker gets behind, nearby workers can help out, even from end to beginning.

C Skipping steps. Having work stations closer together makes it easier to process orders that skip steps.

    In sequential one-piece flow, when one production machine breaks down, the whole line will go down. Therefore proactive equipment maintenance is important to prevent unexpected production interruptions. A good TPM program should assure this. Inventory buffers may give an allusion of protection, but may still require special measures, like overtime, to recover.

    Equipment maintenance can be more responsive and less costly with standardization of all replaceable parts: belts, motors, fuses, controllers, etc.

    Ideally, to achieve optimal machine tool and work station utilization, one-piece flow lines should be balanced so that the time to do the required tasks at each station, called the takt time, is fairly constant.

C If takt time at each station = station capacity, arrange into sequential line.

C If takt time does not equal station capacities, but does not vary with products:

C Upgrade appropriate capacities or find faster machines to achieve balance.

C Group machines/stations into series/parallel paths to achieve better balance, perhaps 3 of one feeding 2 of another.

C If underutilized machines are not expensive, don’t worry about balancing if the entire system can provide high value.

C If takt time varies with different products,

C Make stations/machines flexible enough to share workload.

C Sequence jobs to compensate for imbalances

C Size the line based on the most expensive machine and provide excess capacity for the less expensive machines.

    Another way to balance lines is to make certain stations become kanban sources, so that they make kanban parts during times when they have excess capacity.

    Flexible operations work best with dedicated cells or lines for every product family. Cells can be permanently configured so that within a product family, all setup has been eliminated. This strategy work best with many simpler dedicated machines instead of a single "mega-machine, unless the mega-machine can handle a very large family -- enough to justify its expense. In some cases older or "obsolete" machines may be used to provide complete set of machines for the cell; this was one of the solutions covered in Eli Goldratt’s The Goal.⁹ Remember that speed or capacity may not be as important as flexibility.

    Total cost analysis must be used taking into account all related overhead costs in addition to the usual material and processing cost. In some cases, cells may be installed even if the cell alone can not be justified by traditional analyses, but if the cell completes a valuable plant capability like build-to-order. The guiding strategy for cell design is flexibility and setup elimination.

LEVELING PRODUCTION

Artificially induced irregularities.

    Raw material comes out of the ground in a steady flow. Most products are ultimately consumed in a steady flow. Most irregularities in factory workload are artificially induced.¹⁰ Sources of irregular factory workload include:

C Production quotas for end of the month, quarter, and year.

C Promotions, usually to move built but unsold finished goods or to meet sales quotas. This situation is compounded when customers hold off purchases until the expected "sales."

C Quantity discount and "deal making" encourage large batch purchases. Ironically, this may cause the producer to work overtime to deliver the large batches and cause the customer to incur inventory carrying costs once the batch is received.

C Lack of dealer confidence of product availability, leading suppliers to build up inventory.

C Lack of customer confidence in product availability, leading consumers to "stock up" when they can.

C The "business cycle." Half of the effects of downturns are caused by working off excess inventories; half the upturns are caused by building up inventories for anticipated upswings in demand.¹¹

Seasonal irregularities. Some irregularities in factory workload are seasonal, such as Christmas, back-to-school, and other events. But these can be dealt with, since these are predictable. Notice how grocery stores know how many turkey’s to order for the holidays and how much beer and snack foods to order for major sporting events.

C Eliminated artificially induced irregularities listed above.

C If demand exceeds daily capacity for a line, prioritize scheduling into categories (next-day, two-day, time available within the week) and charge accordingly, either a premium for next day or a discount for slower delivery.

C For short-range peak demand beyond capacity of a line or cell:

C Shift production to another line if second line is flexible enough

C Consider overtime

C For long-range peak demands beyond capacity, expand capacity and/or outsource the least efficient or least compatible operation. Pre-assign efficiency ratings & work from the top of the list.

C Avoid unexpected loss of capacity with preventive maintenance (TPM) and quality assurance programs, like TQM and process controls, to avoid interruptions and products looping back.

DEVELOPING PRODUCTS FOR LEAN MANUFACTURE ¹²

C There may be too many different products and, thus, too many different parts to distribute at all the points of use.

C Even within the ideal group of products, there may be a needless and crippling proliferation of parts and materials.

C Specified parts may be hard to get quickly.

C The product/process may have too many setups designed in.

C Quality may not be designed into the product/process which results in disruptions when failures loop back for correction.

C Product/process design may not make optimal use of CNC. Most CNC equipment is used in a batch mode, not flexibly.

To be successful at designing products for a lean environment, product development teams must:

C Proactively plan product portfolios for compatibility with lean operations

C Design products in synergistic product families

C Design around aggressively standardized parts and raw materials

C Make sure specified parts are quickly available

C Consolidate inflexible parts into very versatile standardized parts

C Assure quality by design and by concurrently designed process controls

C Concurrently engineer product families and flexible processes.¹³

C proactively specify all processes; not doing so gets the defaults, which may not be lean

C design overall process flow

C design the process and the tools at each workstation, ensuring no setup

C Make all decisions based on total cost.

C Design around common cutting tools, bending mandrels, punches, etc. Keep tool variety within tool changing capability for the entire product family.

C Design versatile fixtures at each workstation that eliminate setup to locate parts or change fixtures.

C Develop flexible assembly with on-line assembly instructions

C Make sure part count does not exceed available tool capacity or space at each work station.

Maximize use of available programmable fabrication and assembly tools, without expensive setup delays. Programmable CNC tools include CNC machining, spot welding, robotic painting, printed circuit board assembly, and sheet metal laser cutting, bending, and punching.

Setup eliminated þ batch-size-of-one flexibility þ one piece flow þ WIP elimination

C Decrease throughput time to approach the "labor standard" (actual minutes actually spent to fabricate and assemble a product).

C Eliminate setup delays on expensive equipment and thus improve machine tool utilization

C Save setup labor costs

C Eliminate inspection time and scrap costs verifying the first parts made after setup

C Allow build-to-order and mass customization

C Allow dock-to-stock part delivery (see discussion later), thus eliminating:

C Inventory carrying costs of raw parts inventory (floor space, administration, etc.)

C The cost and delays of incoming inspection and logging in parts

C Eliminate the penalties of kitting:

C Labor cost to kit

C Floor space for kitting area

C Production delays and expediting costs when kits are short

C Waste or restocking costs when kits are over

C Improve quality with rapid feedback to catch and rectify quality problems fast. Large batches of parts can all be made with the same defect so that many parts will have to be scrapped or reworked.

C Eliminate fork lifts including the labor, equipment, and floor space for the aisles

C Foster psychological flow,¹⁴ improve job satisfaction, relieve boredom, and encourage continuous improvement.

C Eliminate WIP inventory carrying cost = 25% of value of inventory per year

C Cut floor space needs in half.^{15, 16, 17} This is especially important in times of growth, but floor space savings should always be assigned a value to encourage more efficient utilization of space.

Dr. Anderson is a California-based consultant specializing in training and consulting on build-to-order, mass customization, lean/flow production, design for manufacturability, and cost reduction. He is the author of  "Build-to-Order & Mass Customization, The Ultimate Supply Chain Management and Lean Manufacturing Strategy for Low-Cost On-Demand Production without Forecasts or Inventory" (2008, 512 pages; CIM Press, 1-805-924-0200, www.build-to-order-consulting.com/books.htm) and "Design for Manufacturability & Concurrent Engineering; How to Design for Low Cost, Design in High Quality, Design for Lean Manufacture, and Design Quickly for Fast Production" (2008, 448 pages; CIM Press, 1-805-924-0200; www.design4manufacturability.com/books.htm).  He can be reached at (805) 924-0100 or andersondm@aol.com; web-site: www.build-to-order-consulting.com.

1. An excellent reference on lean production is: Lean Thinking; Banish Waste and Create Wealth in Your Corporation, by James P. Womack and Daniel T. Jones (Simon & Schuster, 1996).

2. ibid., p. 83.

3. Much of this is drawn from the book "Build-to-Order & Mass Customization, the Ultimate Supply Chain and Lean Manufacturing Strategy for Low-Cost On-Demand Production without Forecasts or Inventory," (2004, 520 pages, CIM Press,1-805-924-0200; www.build-to-order-consulting.com/books.htm); Chapter 8, "On-Demand Lean Production."

4. Robert W. Hall, Zero Inventories, (Homewood, IL, Business One Irwin, 1983), Chapter 5.

5. Shigeo Shingo, A Revolution in Manufacturing, The SMED System, (Portland, OR, Productivity Press, 1985).

6. Kiyoshi Suzaki, The New Manufacturing Challenge, Techniques for Continuous Improvement, (New York, Free Press, 1987), Chapter 3.

7. Kiyoshi Suzaki, The New Manufacturing Challenge; Techniques for Continuous Improvement, Video program (Dearborn, MI, Society of Manufacturing Engineers).

8. Adrian J. Slywotzky and David J. Morrison, Profit Patterns, 30 Way to Profit from Strategic Forces Reshaping Your Business, (Times Business, Random House, 1999), "Reintegration," p. 117-123.

9. Eliyahu M. Goldratt, The Goal, (North River Press, Second Revised Edition, 1992).

10. James P. Womack and Daniel T. Jones, Lean Thinking; Banish Waste and Create Wealth in Your Corporation, (1996, Simon & Schuster), p. 87, "Is Chaos Real?"

11. Ibid. p. 88, "Do we really need a business cycle."

12. Much of this is drawn from the "Build-to-Order & Mass Customization, the Ultimate Supply Chain and Lean Manufacturing Strategy for Low-Cost On-Demand Production without Forecasts or Inventory," (2008, 512 pages, CIM Press,1-805-924-0200; www.build-to-order-consulting.com/books.htm); Chapter 10, "Product Development for Build-to-Order & Mass Customization."

13. For a general summary on agile product development, see the book Agile Product Development for Mass Customization, by David M. Anderson (McGraw-Hill, 1997); Chapter 9, "Agile Product Development for Mass Customization."

14. Mihaly Csikzentmihalyi, Flow: The Psychology of Optimal Experience (New York: Harper Perennial, 1990).

15. Jones, Daniel J., "JIT & the EOQ Model: Odd Couples No More!," Management Accounting v72, n8 (Feb 1991), pp. 54 - 57.

16. Richard J. Schonberger, World Class Manufacturing, The Lessons of Simplicity Applied, (New York, Free Press, 1986), p. 83.

17. ibid., pp. 229-236.

[Mass Customization and Build-to-Order home page]   [Article Directory]