by Dr. David M. Anderson, P.E.,
Cost Reduction Strategy (home page) Seminars Consulting Credentials Client List Articles Books Site Map
Excerpts from the Book: Build-to-Order & Mass
Copyright © 2017 by David M. Anderson
There are many opportunities to reduce total cost in supply chains, which are responsible for many unnecessary overhead costs to generate forecasts, count inventory on-hand, generate purchase order inputs through MRP (Material Requirement Planning) systems, place purchase orders, wait for parts to arrive, expedite those that are late, receive (and maybe inspect) materials, warehouse, group into kits for scheduled production, and distribute within the plant. These costly and time-consuming steps can be avoided with a spontaneous supply chain, which is able to pull in materials and parts on-demand.
The concepts presented herein are labeled as the resupply of parts and
materials as opposed to procurement or purchasing. This is to emphasize that
most of supply chain management is a resupply function of parts and
materials that have been procured before. Therefore, the role of
procurement/purchasing departments will change.
In a build-to-forecast batch environment (mass production), Purchasing’s role was to order parts based on forecasts and MRP data and then expedite part shortages due to forecasts being off, bill-of-material errors, inventory count errors, and late deliveries. For spontaneouls supply chains, Purchasing’s new role will be to:
$Encourage and maybe drive standardization of parts and raw materials for current products and help new product development teams design around aggressively standardized parts and materials.
$Identify standard raw materials and parts available from multiple sources; arrange breadtruck replenishment.
$Arrange steady flows of standard parts and raw materials; nurture supplier/partner relationships with the focus on delivery.
$Establish kanbans and pull signal arrangement with suppliers.
Most of the spontaneous part and material resupply will be automatic or
manually triggered by production personnel, not by MRP and purchasing.
For the 25 finalists in Industry Week’s annual Best Plants rating, 55.6% have implemented pull systems compared to the average of 18.6% for all 2,511 manufacturers surveyed.1 The best plants used pull systems three times more than the average plant.
There are two aspects of the spontaneous supply chain: raw material resupply and part resupply.
Identify the most common and least common parts by sorting usage in Pareto format. Standardize on the most common materials, even if they are a higher grade or tolerance; the total cost savings in material overhead and purchasing leverage should compensate for any perceived cost increase. New products should be designed around these standard materials. Existing designs could be converted for any "better than" substitutions. Unusual and seldom-used materials should be discouraged in product development and phased out after conversion on existing products.
For very flexible environments like lean, build-to-order, and mass
customization, the standardization should be very aggressive.
Too many types of raw materials can thwart spontaneity and present
manufacturers with the dilemma of stocking all types of materials or ordering
them and waiting for delivery. The effective procedures presented in the
Standardization article should be used to aggressively standardize raw
materials. Standardization reduces the incoming variety which helps enable the
spontaneous supply chain. Purchasing leverage and other material overhead
savings compensate for any cut-off waste and for some products getting "better"
material than they need. Standardization can also protect a manufacturer from
shortages, which are more likely to occur if production depends on many unusual
Raw materials could be automatically resupplied using the following techniques (in order if increasing variety):
Steady Flow of Standard Raw Materials. The ultimate scenario for spontaneous resupply is to reduce the number of raw material types within each category to one, in which case steady flows can be arranged for each standard raw material. Ideally, there should only be only one type of each material. Then forecasting multiple types would be unnecessary and "ordering" would be as simple as matching the tonnage in to the tonnage out; in other words, the incoming flow of the standard raw materials would be equal to the monthly consumption of the plant. These will be used one way or another. Multiple types of materials in each category would allow the same spontaneity if the ratio is constant or predictable and usage is segregated (if not, material changeover would have to be quick and hopefully automatic).
Material Cut-to-Length/Shape. Raw materials can be cut to-length or
to-shape on-demand from the longest version or standard sizes by programmable
CNC (Computer Numerically Controlled) equipment, such as laser cutters and screw
machines. For example sheet metal can be cut-to-shape on-demand by the
“source cell” shown below:
If the sheet metal can be standardized to predominantly one type for the factory, then sheet can be ordered on a coil, at much less material expense, and fed from a coil dispenser and straightener. The leading edge of the coil would then be programmably positioned in the CNC shear to cut any size sheet in any quantity on-demand (a) for subsequent processing in nearby machine tools, like the laser/plasma cutters shown or (b) as a kanban source for all users of that sheet in the plant. Note that after the first cut, the sheet can be rotated for second cut made in the open space shown in the foreground of the same shear, either automatically or manually positioned against a programmable stop, with appropriate safety assurance with sheet grippers and two-handed switches.
Benefits of Sheet Metal Source Cell:
• Flexibility advantage of having sheet metal cut on-demand without setup delays or costs
• High utilization potential of all equipment in cell and receiving equipment
• Cell can be a kanban source between on-demand tasks
• Raw material savings when:
• buying from larger suppliers or directly from the mill
• getting quantity discounts
• bypassing intermediaries suppliers
• avoiding sheet cutting charges, which is part of the cost of all cut sheets.
more efficient nesting when patterns are only constrained by width
• less waste from inefficient use of 4’ by 8’ sheets
• avoiding damage to storing/handing partial sheet remnants
• not needing to store, retrieve, and set up partial sheet remnants or discarding them
• avoiding chance of loading the wrong partial sheet if identifying marks are missing from a remnant
• unlimited long lengths possible
• variable sheet length gives the ability to keep a one-piece flow of sheet sets (for each product), rather than trying to fill up a 4' x 8' sheet by nesting many parts from several products, which is more only suitable for mass production batches.
The result is on-demand shearing of low-cost sheet that is always available.
Such sheets can travel in one-piece flow to flexible fabrication cells
or kanbans or to any other users BTO or MC users
Linear Cut-Off. Raw material variety can be greatly reduced by cutting off linear materials on-demand at the points of use or as kanban parts resupplied automatically to all the points of use. Linear materials include all forms of bar stock, extrusions, strips, tubing, hose, wire, rope, cable, chain, and so forth.
Min/Max Stacks. In the "min/max" technique, often used for raw material like sheet metal, material is consumed until the stack reaches the "min" (minimum) level, usually marked on the rack or wall. This triggers a reorder of the material to bring it up the "max" (maximum) level without the usual purchasing costs. Price and delivery arrangements, based on average usage data, could be negotiated on a long term basis for greater purchasing leverage. This can be done by single axis programmable cut-off machines, or from less automated tools directed by on-line instructions displayed on monitors]
Kanban. For raw materials that come in bins rather than stacks, the kanban resupply technique could be utilized. It may be that the source is one of the cut-off operations mentioned above (for more on kanban, see the illustrated discussion below).
Strategic Stockpiles. Until the above techniques can be implemented, it may be necessary to have strategic stockpiles of certain materials. The manufacturer could use selective stockpiles to temporarily compensate for any parts or materials that cannot be "pulled" or for temporary availability problems on standard materials. Stockpile ordering would have to be based on some kind of forecasts, but if the material was standardized, then the forecast would be easier to make for the aggregated demand for all consumption.
Order Material After Receipt of Product Order. Spontaneous resupply
may not be feasible for unusual or seldom-used raw materials, especially on any
products with inherently high diversity of materials. If material order times
are less than build times, these materials could be ordered after receipt of the
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
"on-demand." A special-case variation of this approach is the Dell model where
suppliers warehouse their parts next to the assembler’s factories. Not many
assemblers are big enough and powerful enough to force their suppliers into such
However, this is not really a pull-based supply chain. Parts availability would depend on assemblers’ forecasts, which are becoming increasingly less accurate, and the supplier’s inventory, which is costly to carry and prone to obsolescence.
Part resupply strategy depends on the variety of the parts. At one end of the spectrum, very standard parts that are used in almost all products could arrive in a steady flow like standard raw materials.
At the other end of the spectrum, parts with high variety would be built on-demand using the techniques presented herein including flexible CNC fabrication and manual equivalents.
In between, there are several other strategies such as kanban for somewhat standardized, medium-variety parts and breadtruck resupply for small, low-cost commodity parts such as fasteners. Inflexible parts, such as casting and plastic parts, can be consolidated into versatile parts that can be used in many products. Parts could be automatically resupplied by the following techniques (listed in order, with the easiest first):
Steady Flow of Parts. As with standard materials, steady flows could be arranged for very standard parts, which would be used one way or another. The criteria for steady flow of parts would be standardization and widespread use.
Breadtruck Resupply. The easiest and "lowest hanging fruit" in
material logistics is the breadtruck (sometimes called "free stock")
delivery system for small, inexpensive parts, like fasteners. Instead of
counting on forecasts to trigger an MRP system to generate purchase orders, all
the "jellybean" parts can be made available in bins at all the points of
use. A local supplier is contracted to simply keep the bins full and bill the
company monthly for what has been used, much like the way bread is resupplied by
the breadtruck to a market.
All the MRP/purchasing expense is eliminated and this type of delivery can assure a constant supply of parts, thus avoiding work stoppages. Being off the forecast/MRP system, the supply of these parts can be assured for "forecast-less" operations such as build-to-order and mass customization. Typical parts suitable for breadtruck deliveries are fasteners, hardware, and almost any small, inexpensive part.
As companies become more agile, they may include slightly more expensive and slightly larger parts into the breadtruck system. The more expensive parts may incur some inventory carrying cost, but that should be outweighed by savings in purchasing, material overhead, expediting, and avoiding work stoppages. The criteria for breadtruck deliveries would be:
$A reliable supplier can be contracted. Many suppliers welcome such business and want to perform well, since they usually get all the business for their categories of parts and raw materials.
$Parts can be distributed at all points of use, without cluttering assembly areas with too many parts. Of course, part standardization will help achieve this goal.
$Parts are small enough and cheap enough so that sufficient parts will always be on hand. Bin count can be set high enough to preclude any chance of running out.
$Parts are not likely to go obsolete or deteriorate while waiting to be used.
$The breadtruck parts are not so "attractive" as to create a significant pilferage problem, since, generally, companies do not correlate part consumption with product sales. However, making breadtruck parts freely available for R&D prototypes and factory improvements may encourage innovation and the use of standard parts.
$Manual reorders are not anticipated to occur. The supplier should be in a continuous improvement mode and be constantly adjusting bin count to correspond to prevailing demand. The factory should alert the supplier about any anticipated "spikes" in demand.
Kanban Resupply. In kanban resupply, parts with limited
variety are made, maybe in batches, and resupplied automatically to replenish
parts bins based on part consumption. This is one of the many pull systems
used to "pull" parts into assembly operations.2 The resupply is
automatic once the pull signal gets to the supplier. There are many simple ways
to do this without complex information systems such as MRP or ERP. Thus, kanban
resupply avoids the uncertainly of forecasting, the cost of purchasing, and the
cost and risks of inventory.
Kanban works best for semi-standard parts without too much variety, which would increase work-in-process (WIP) inventory and clutter assembly stations with too many part bins. Kanban parts can be made in mass-produced batches, thus reaping the benefits of economies of scale. Part manufacturers may have to implement setup and batch size reduction to be able to economically make batches small enough for kanban deliveries.
The principles of kanban can be best explained using the two-bin system as illustrated below, which shows two rows of part bins which are set up for resupply. Initial assembly starts with all bins full of parts.
When the part bin nearest the worker is depleted, the full bin behind moves forward, as shown by the empty space in the illustration. The empty part bin then is returned to its "source," which could be the machine that made the part, a subassembly workstation that assembled the part, or a supplier. The source fills the bin and returns it to this assembly workstation behind its counterpart which is still dispensing parts.
The beauty of Kanban resupply is that the system ensures an uninterrupted
supply of parts without forecasts or high-overhead-cost ordering procedures.
The number of parts in a bin is based on the highest expected usage rate and
the longest resupply time. The size of each bin is determined by the bin
quantity and size of the parts. For large parts, some companies use two-truck
kanbans, in which parts are drawn from one truck trailer while the other trailer
goes back to the supplier for more parts. Alternate systems include kanban
squares for larger parts and a two-card system where the cards travel (or are
faxed) back to the source instead of the bins. Electronic equivalents can also
In order for Kanban systems to work, there must be enough room to dispense all parts at all the points of use. This, again, emphasis the importance of part standardization.
Yasuhiro Monden, in an updated version of his classic Toyota Production System,3 states that “the kanban system’s most remarkable feature [is] its adaptability to sudden demand changes or exigencies [urgencies] of production.” This is exactly what is needed for build-to-order environments, which are based on demand, not on planned production schedules.
Spontaneous Build-to-Order of Parts. For parts that are too varied for kanban, the assembler or the suppliers themselves would need to implement spontaneous build-to-order so that they could actually build on-demand to their customers’ (the assemblers’) pull signals. This is the only way to supply mass-customized parts on-demand for mass-customized products. Parts can be made on-demand in-house or by nearby agile suppliers.
It may appear that spontaneous build-to-order of parts may cost more than mass production, but in reality, a complete BTO operation is very cost-effective when measured on a total cost basis.
Parts Made On-Demand by Suppliers. Hopefully, it may be possible to find suppliers who can implement these techniques to make your parts on-demand in response to your pull signals. Pull signals need to be initiated early enough and response time needs to be quick enough so that parts arrive without causing assembly delays.
Spontaneous build-to-order of parts may require the development of supplier/partner relationships in which suppliers establish the ability to build parts in any quantity on-demand. The distance to the supplier must not be so great so that part delivery delays product delivery.
Be wary of suppliers that are “pulling” parts from inventory because of the risk of not having enough (which they may blame on your inadequate forecasts) and higher than necessary cost for inventory carrying costs.
Parts Made On-Demand In-House. In order for spontaneous build-to-order to work, all parts must be available on-demand. If there are any key parts that are not suitable for kanban and no supplier can build them and ship them quickly enough to your pull signal, then you might have to bring those operations in-house, as discussed in the article on outsourcing and vertical integration.
Flexible Processing. Regardless of the sources of parts, spontaneous part manufacturing operations must be able to make parts on-demand efficiently in a batch-size-of-one mode without setup or inventory. CNC programmable machine tools and flexible assembly can produce a high variety of parts without setup costs and delays from standard raw materials. Similarly, manual assembly can be made flexible, as shown in the mass customization article. This may require concurrent engineering of product families, parts, and processing to eliminate all setup changes, as discussed in the article on designing products for lean production and BTO.
Strategic Stockpiles. Until the above techniques can be implemented, it may be necessary to have strategic stockpiles of certain parts. The assembler could use selective stockpiles to temporarily compensate for any aspect of the supply chain that cannot be pulled or for temporary availability problems on standard parts.
Order Parts After Receipt of Product Order. Spontaneous resupply may not be feasible for unusual or seldom-used parts, especially on capital equipment with an inherently high diversity of parts. If parts order times are less than build times, these parts could be ordered after receipt of the product order.
Dock-to-Receiving-to-IQC-to-Warehouse-To-Kitting. In most plants incoming
parts a slow and expensive procedure starts in the receiving department (where
they are logged in), then to the incoming quality control (IQC) department
(where they are inspected), then to the raw material warehouse (where they are
inventoried), and then the kitting department (where they are counted and
grouped into batches).
Dock-to-Line Deliveries. To be truly agile, incoming parts and materials must flow directly to the all points of use without all the steps listed above; this is called dock-to-line delivery. This is sometimes referred to using the more common, but less accurate phrase, dock-to-stock, which technically means parts go to some kind of internal warehouse, hopefully without incoming inspection, before being distributed to the line.
Dock-to-line may be more easy to implement after implementing part and material standardization, product line rationalization, breadtruck deliveries, and kanban resupply of appropriate parts. Freeing up floor space by inventory reduction efforts will make room for internal distribution at all points of use. Lean environments require much less raw materials inventory than batch oriented operations, so there will still be a net reduction in floor space requirements after implementing lean production. In addition, part warehouse space may now be more available.
In order for dock-to-line to work, quality must be assured at the source by suppliers whose processes are so in control that their customers (the assemblers) do not need to inspect incoming parts. Further supply chain cost could be saved if the suppliers’ systems we so in control that the suppliers did not have to inspect them either.
Dock-to-stock deliveries can be either triggered by purchase orders that come from MRP systems or hopefully automatic pull signals like kanban. Dock-to-line deliveries can be an essential part of a lean production program or may be instituted primarily to save cost and improve throughput. Industry Week’s “Best Plant” survey of the 25 top performing candidates indicated that 68% of suppliers deliver parts to the point of use in the plant.4
The Problems with Incoming Inspections. The big paradigm shift required for dock-to-line deliveries is the elimination of incoming inspections of parts and raw materials. Incoming inspections are impractical for two reasons: time and cost. Just-in-Time deliveries, as the name suggests, should be just in time. Having to go through incoming inspections, usually at central receiving stations, would cause too many delays for a fast-moving lean environment. JIT deliveries may be smaller and occur more often than the traditional large order that is delivered infrequently. Consequently, inspecting many small orders would be very inefficient and costly because of the inspection setup which include finding getting up to speed on quality standards and procedures, setting up and calibrating test and inspection equipment, and dealing with problems via MRB (Material Review Boards).
Eliminating Incoming Inspections. But incoming inspections cannot simply be eliminated without some way of assuring that the incoming parts and raw materials will have adequate quality. If a manufacturer simply dictates new standards for incoming part quality from suppliers, the suppliers may respond by shifting inspection from the manufacturer’s receiving to the end of the suppliers operations. This may screen out bad parts but at too great a cost in money and agility. In addition, when companies try to achieve quality by rejecting out-of-spec parts from a “wide” bell curve, the result is that the parts that do pass will have a high proportion close to the “hairy edge” of not working, which may cause more worst case failures.5
Suppliers, internal and external, need to adopt the Six-Sigma philosophy of assuring quality at the source. If part manufacturing and raw material processing is sufficiently in control, quality will be assured by the process, not by subsequent inspections. Statistical Process Control (SPC) is a proven tool for assuring quality by process controls.6 Even though SPC is firmly founded on statistical principles, its implementation does not require the proverbial Ph.D. in statistics. Control charts, available from the American Society of Quality (ASQ), have the statistics built into the charts so that factory workers can use them by literally filling in the blanks and performing some simple arithmetic computations.
Certification. Suppliers that can prove that their processes are in control and, thus, can deliver good parts directly to the line are certified by the manufacturer. Similar certifications may also be applicable for raw material suppliers to allow them to ship metals, plastics, and chemicals directly to the points of use without incoming inspections.
ENDNOTES/REFERENCES (See below)
For more information call or e-mail:
Dr. David M. Anderson, P.E., fASME, CMC
Cost Reduction Strategy (home page) Seminars Consulting Credentials Client List Articles Books Site Map
1. David Drickhamer, “Aim High; How Industry Week’s Best Plants Measure Up,”
Industry Week, October 2002, pp. 67-70. This article is available online at
2. Yasuhiro Monden, Toyota Production System, An Integrated Approach to Just-in-Time, Second Edition (1993, Industrial Engineering and Management Press, IIE), Ch. 2, “Adaptable Kanban System Maintains JIT Production,” and Ch. 3, “Supplier Kanban and the Sequence Schedule Used by Suppliers.”
3. Monden, Toyota Production System, p. 27.
4. Industry Week, The Complete Guide to America’s Best Plants, (1995, Penton Publishing).
5. David M. Anderson, 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, (2010, CIM Press, 456 pages; 805-924-0200), Chapter 10, “Design for Quality.”
6. Robert Amsden, Howard Butler, and Davida Amsden, SPC Simplified, (1989, Quality Resources, New York, NY).