Before commencing, it is worth spending a few minutes describing the environments to which these processes and systems are addressed. They range from simple stockist type environments to highly complex manufactured products, manufacturing processes and distribution networks. They also range from intermittent / low volume, or one-off complex assemblies or mixtures, to very high volume mass produced products of varying complexity. It is not surprising therefore that a wide range of techniques have been developed to accommodate this range of applications. So let us start by simply listing the techniques available:
The techniques fall into two themes:
- “Push” systems (triggered by interpretation of the expected demand and scheduling of supply to meet that demand)
- “Pull” systems (replenishment systems triggered by the usage or depletion of stock)
|All Time Buy||Replacement (see Previous Technique T015: Replacement Systems)|
|Project Manufacturing||Top up point of use / Vendor Managed Inventory|
|Period Batch Control||Input / Output Control (see Kanban Systems)|
|Material Requirements Planning (MRP1)||Two Bin|
|Manufacturing Resources Planning (MRP2)||Three Bin|
|Drum Buffer Rope (although the “rope”, it could be argued, is pulling, the “drum” is pushing)||Kanban|
|Advanced Planning & Scheduling (APS)||Reorder Point (ROP)|
There is however a third grouping which we call complementary, in that they utilise, or are an adaptation of, some of the above techniques, (principally but not exclusively in repetitive environments), but apply them in a unique way. These include:
- Consignment stocks
- Outsourced Warehousing
- Visual control systems (see Previous Technique T036 Visual Control Systems)
- The use of bar code reading / RFID in stock movement
- Electronic Data Interchange (EDI) / Internet enabled supplier integration
- The use of “Back-flushing” techniques in MRP1 systems
- ERP systems
As a final part of this introduction it is worth positioning “Just In Time (JIT)” in these groupings. The supply of items just in time for their immediate use is an objective ofall materials planning and control systems. It could be said that most of the techniques described above have this objective, except perhaps “all time buy”. However Just In Time has become associated with “Kanban” Systems.
Follow the links on the left of the page for further detailed articles for items coloured red. Now to an overview of each:
Dealing with the Pull systems first, they are a group of techniques that aim to ensure that when stock is used, it is replenished. It is important to note at this point that pull systems do not plan, they react! This is fine if your processes and logistics are infinitely flexible, uncomplicated and do not vary. If they are not, some degree of capacity or materials buffering will be necessary! For example, there is a need to reduce buffer stock for declining demand and prime stock for sales initiatives. In addition the New Product Introduction process and product change process (see Previous Best Practice B022 Product Change Process) must somehow feed in the new and feed out the old.
Secondly most of these “pull” techniques except reorder point (see Previous Technique T014: Reorder Point Systems) operate principally using visual controls (see Previous Technique T036 Visual Control Systems) rather than computer systems and are therefore well aligned to “lean manufacturing“.
(Also see Previous Technique T015: Replacement Systems)
In it’s simplest form there is “Replacement“, which simply replaces used items. This technique is still very common for many simple applications and is commonly used for maintenance spares, grocery shelf restocking, stationery and my partner’s re-supply of our bathroom consumables, and was probably used by cave dwellers to replenish food stocks.
This is also the underlying principle of all pull systems.
It is now very common for suppliers to manage customer’s inventory by managing the stock levels themselves as a value added service to customers. Usually this replacement is based on simple top up to predetermined levels or by a “Fax-Ban”, (see Kanban).
Also it is now very common for suppliers to manage customer’s inventory, replacing used items by delivering directly to point of use, avoiding the stores management process, as a value added service to customers. Alternatively this method can be used by stores personnel to replenish line-side stock from a central stores for items in repetitive / continuous use.
Input / Output Control
(see Kanban Systems Also see Input / Output Capacity Control)
This simply delivers (inputs) to a process what has been produced (output). When applied to Work in Process as a whole, this technique has come to be known as a “Conwip” (Constant WIP) Kanban system.
(Also see Previous Technique T010: Two Bin Systems)
Again an ancient technique for which confusingly there are 3 “bins” in the system, 2 bins at the point of use and a third elsewhere, (a “bin” can be any fixed quantity container including a plastic bag, work trolley, or kitting tray):
- Bin 1 is in use in the shop.
- Bin 2 is held adjacent to bin 1 (but is not in use) and is used to replace bin 1 by operatives when bin 1 is depleted, without formal authorisation.
- Bin 3 can be somewhere else (maybe in a stores, WIP, or at the supplier) and is used to replace bin 2 when it is swapped for bin 1.
- Bin 1 is then refilled to become bin 3 and held waiting for the depletion of the bin in use, for the system to recycle.
When feeding several points of use, the 3rd bin can be a central stores stock.
3 Bin Systems
3 bin systems operate in the same way as 2 bin systems except that a safety stock (perhaps a handful of items in a plastic bag) is stored separately in bin 2. Breaking into this safety stock, if the system has not recycled in time, triggers an expedite request and a review of bin sizes. To maintain stock rotation the safety stock must be refreshed on replenishment, which means that to enjoy the opportunity of an expedite signal you need a bit more administration.
(Also see Kanban)
Attributed originally to Toyota in the 1980’s, Kanban Systems are an extension of 2 bin and 3 bin systems, where there is a semi-fixed number of Kanbans (typically but not necessarily primed containers) in the system, which may be spread throughout the system and which are replenished when depleted.
(Also see Previous Technique T014: Reorder Point Systems)
“Reorder-point” systems, invented by operational researchers, in the Second World War, made more sophisticated the previously simple replenishment processes by taking into account:
- The variability of supply and demand (introducing “safety stock”, (calculated using statistical probability)
- Lead-time to replenish, which orders replacements at a point (the “reorder-point”) where allowing for the lead-time and recent historical demand, replacements would be available before depleting the safety stock, with statistical certainty.
- Various forecasting algorithms were introduced to more accurately reflect usage within future lead-time rather than recent history. (Trend, Seasonality, Seasonality with Trend, Intermittent etc.) (We select from 6 methods in our training course SSC03 Advanced Forecasting & Inventory Modelling.)
- “Safety time” began to be used instead of “safety stock” in some early MRP systems in circumstances where lead-time rather than quantity was the major variable.
This is an ongoing area of research for us and our latest research:
- Characterises 7 generic methods to be selected on a “best fit” basis in order to reduce average and maximum stockholding in pull systems. (This will be covered by a future article.)
- Identifies the characteristics of a further uniquely identifiable forecasting method to incorporate into our “advanced forecasting” methodology and training.
Perhaps the original method of ordering, you buy everything you think you are likely to need. The obvious disadvantage being that it is difficult to forecast all time requirements. However for items which are likely to be difficult to acquire in the future and are of low value, this is still a widely used technique, particularly in the aerospace and electronics industries towards the end of original equipment production.
(Also see Previous Technique T004: Chunking)
Very different in concept to MRP systems (below), and much better suited to small volume production, job shops, engineer-to-order environments, or capital / development projects, this system schedules both parts and processes. Bills of material are created for each “chunk” (work package) of the project, and capacity planning may include technical / engineering resource as well as manufacturing resource. Items are made and purchased (“multi-level pegged”) for individual projects / final products. Actual costs are accumulated during manufacture, with estimates to completion (time and cost) routinely calculated. The more sophisticated versions employ “critical path” (“PERT”) network or Advanced Planning & Scheduling (APS) scheduling.
In the 1930’s, but more famously during the second world war, the first simple scheduling technique to adopt a JIT philosophy was introduced by RJ Gigli to deliver to the next stage in the process all the parts required for the next period’s work in a UK aircraft manufacturing plant. This technique (“Period Batch Control” by the late Jack Burbidge, who I had the privilege of working with at Lucas Industries) synchronised the arrival of components for sub-assembly and sub assemblies for final assembly by allowing a fixed period for the production of all the parts required for the next stage.
Using a weekly period as an example:
Week 1: Order parts from supplier
Week 2: Deliver to the component manufacturing process by the end of the week
Week 3: Make components by the end of the week
Week 4: Make sub assemblies by the end of the week
Week 5: Make final assemblies and deliver to customer by the end of the week
In this example the lead-time is fixed and is 5 weeks.
The technique was a huge step forward in coordinating the supply of parts for “A”, and “X” type products (see Postponement & Mass Customisation. It required a Bill of material (parts-list), to drive the schedule for the preceding stage. This concept was later adapted by Lucas Industries in 1991 to incorporate variable work packages of fixed duration (without Bill of Material restructuring) (see T038: Bills of Materials Simplification) in their JIT-MRP project which integrated the two concepts in a single philosophy.
In the automotive industry, computer based scheduling based on multiplying demand by exploded Bills of Material was starting to be used to produce forward fixed weekly schedules for suppliers by the mid 1970’s, using commercially available computers. These fixed cycle schedules, are still the most common method of communicating medium term demand (advisory schedules) to suppliers in the automotive and aerospace industry today, and the technique has come to be known as “cyclic re-ordering“. Also although this technique has largely been overshadowed by the wide adoption of MRP systems by software suppliers (see later), it is now enjoying a comeback due to its simplicity and the ability to live comfortably with Kanban systems. (See below.)
(Also see MRP1)
Chronologically Materials Requirements Planning (MRP1) came next in the late 1960’s, initially with fixed period lead-times (“bucketed” systems). This was the scheduling engine required for Period Batch Control, which utilised the Bill of Material within it and the scheduling rules (periods) to produce a schedule for the preceding stages, using the example above, five weeks in advance. At this time computers began to be used commercially to generate these schedules. Later variable lead-times (“bucket-less” systems) and safety stocks were accommodated.
(Also see MRP2)
This was followed in the 1970’s by Manufacturing Resources Planning (MRP2) (“Manufacturing Resource Planning: MRP2 Unlocking America’s Productivity Potential”: Oliver Wight), which combined MRP1 and capacity planning together with a control system. Whilst widely implemented, the faulty implementation of MRP2 systems became a scandal, with little regard for data accuracy, ownership, and accompanying (new) management processes.
One of the major criticisms of MRP is the concept of a staged Bill of Material, which represented the stages of manufacture and which also enshrined that into a sequential process, which ignored the potential for parallel working. Work by Burbidge, showed that by flattening the Bill of Material, and parallel working not only simplified the administration of production but also reduced lead-times. However this created tension between the engineering view of the product and the logistics view of the product which “phantom bills of material” did not fully resolve and which also complicated the question of bill of material ownership. This difficulty remains today in most commercially available ERP systems. (See Previous Best Practice B014: “Effective Bills of Material Design”, and Previous Technique T038: “Bills of Materials Simplification”)
However MRP1 is still the dominant technique used by most computer ERP software commercially available today. Whilst the concept of MRP2 is sound, the advent of “Just in Time” (JIT) and it’s control system Kanban created an attractive and significantly simpler mechanism. This for a while in the early 1980’s seemed to provide an alternative to the concept of MRP1, to an extent where the leading gurus of the time predicted the demise of MRP1, in favour of the more pragmatic approach. This was unfounded and in fact ignored the requirement to plan “A” and “X” type product (See Postponement & Mass Customisation) production in advance from both a materials and capacity viewpoint, which JIT did not provide. Even “I” and “T” and “V” type product production require a forward view to provide a planning ability. This led the computer software suppliers to attempt to integrate the two approaches with so called “Electronic Kanbans” (a replenishment signalling mechanism sent electronically). Although rarely needed for internal company signalling, this mechanism has found later usage in the inter-company replenishment signalling involved with “Agile” communications.
The commercialisation of computer packaged software, incorporating the mathematical approaches of MRP2 and later ERP / APS systems, unfortunately ignored the inherent weakness of computers, in that it is notoriously difficult to change computer programs quickly and reliably in the event of requirements, suppliers, or system changes and the lack of visibility of the process. This led to the expansion of parameters held within the computer system that could be changed to reflect different environments. This has now led to a process of “configuration” where the software can be “configured” to reflect different circumstances. Presently and unfortunately the process of reconfiguring can often be difficult and does not reflect the need in some cases to reflect hybrid requirements. (For example some items are repetitive whilst others are not.) Another trap that the software suppliers have fallen into, in the race to provide increased functionality, is to add complexity to the configuration process. This has resulted in many horror stories of incorrect configuration, and a demand for highly skilled “configurers” which for a time outstripped supply. These factors combined to constrain sales of the software, and has led to the sale of preconfigured or cut down versions of complex software. However this completely ignores the “KISS principle” (Keep It Simple Stupid). (See “Implementing ERP Systems“, and B018. Production Planning and Control The KISS Principle (Overview) (Or how to live without the computer))
MRP1 and JIT have now resolved into split roles. In a mixed MRP1/JIT environment MRP1 is primarily used for long term capacity planning and forward ordering of indicative material requirements with JIT pulling the next job when the previous one is finished or stocks depleted. In the absence of JIT, MRP1 schedules may be treated as firm.
In mixed mode operation MRP1 can provide firm schedules in parts of the supply chain not covered by JIT replenishment signals, and indicative schedules in parts of the supply chain where JIT replenishments are used as shown below. (See “Kanban“.)
The terms were defined by Goldratt and Cox in their book “The Goal” the early 1980’s. The “Drum” is the schedule for the bottleneck (drumbeat for the whole system). The “Buffer” is surplus stock put in front of the bottleneck to make sure it never runs out. The “Rope” is the coupling mechanism for ensuring that inputs (release of raw materials) do not exceed the bottleneck capacity, avoiding build-ups of unnecessary WIP.
The mechanism later became confused with the software “OPT” and its successors ” Advanced Planning & Scheduling (APS)“. Here we have deliberately separated the two.
Drum, buffer, rope can be implemented in its own right manually. In this case:
- The Drum is an MPS for the system based on a Rough Cut Capacity Plan for the bottlenecks
- The Buffer needed can be calculated using statistical approaches
- The Rope can any one of five methods of making work flow (See Workflow Management / Scheduling)
Advanced Planning & Scheduling
(which are probably best generically described as, multiple constraint, finite capacity, network, planning systems (Also see Advanced Planning & Scheduling (APS)systems))
This type of system originated with the increasing availability of high speed computers in the 1960’s to perform combined, complex, materials and process, scheduling calculations for complex environments. They incorporate highly mathematical optimisation approaches to scheduling and are therefore intellectually appealing. (Research on the ideal scheduling algorithm is still going on.) Perhaps because of this, they have enjoyed, (based on our own experience and research), unjustified commercial success. But if your environment cannot be simplified, they have application. We have identified 11 different levels of sophistication in these models, but each software has its own methods and cannot simply be paraphrased, so I will not attempt to do that here. Suffice it to say that this methodology is:
- Sophisticated, (some would use the word “complex”)
- A computer based modelling system which is principally based on Bill of Materials, Routing, & Costs, but can be based on more complex scheduling / optimisation rules and parameters such as space, profitability etc.
- Uses critical path networking or optimisation routines
- JIT but not very lean
In this group, developments include:
Originally this technique was adopted by customers who wanted to lose inventory from their balance sheet, whilst still having it available, giving rise to bonded warehouses, where the stock was available but still belonged to the supplier until issued for use in the shop. However suppliers then recognised that this gave them more opportunity for increasing value added services by maintaining this stock for the customer. It also locked in the customer to a long term arrangement. It is now widely recognised that the benefit is largely in the supplier’s hands rather than with the customer. Also care must be taken to ensure that the financial transfer of ownership from supplier to customer is auditable, and that liability for risk of obsolescence in the event of change is agreed.
Sales and purchasing departments continue to use consignment stocks as a bargaining chip and a way of losing stock from the inventory respectively. Depending if the accountant belongs to the customer or supplier, each will argue: “a way of reducing stock and deferring payment”, or “increased administration” respectively.
In a lean organisation and to the inventory planner, the arguments are irrelevant, the stock is still there and it should be minimised to improve the whole supply chain.
I wrote in the year 2000, in the first version of this article: “A logical extension of this concept is that the major retail outlets become property owners rather than retailers with their aisles owned by the supplier!”. I think we have deviated from this concept now where the supplier still owns the stock due to extended credit terms and payment on use, and sophisticated off-balance sheet accounting of buildings is used to make the retailers tenants, so they do not own anything!
Again, in the year 2000, I wrote that off-site, (out-sourced) warehousing was an increasing trend. At that time it applied to Distribution Centres (DC’s) to consolidate supplies for JIT customers for onwards delivery to point-of-use destinations, primarily in the major automotive assemblers. This alleviates the logistics congestion which can arise in feeding hundreds of different items per day to an assembly line. However, now, out-sourcing is applied to many pre-assembly stores, most distribution networks and is also being applied to raw materials. Of course there has also been a massive growth in warehousing to accommodate increasingly imported goods. (We will cover this in a future article.)
Since the late 1990’s we have seen the out-sourcing of kitting from assembly operations. There are a few important considerations here, but in principle this is a perfectly valid method of managing inventory. Often combined with 2 bin systems (above) (where the kit is a bin), one kit can be prepared, whilst the previous one is being delivered. Kitting suppliers are either suppliers with dominant kit content, specialist kit consolidators, or increasingly part of a wholly outsourced stores operation.
Unlike the USA, delivery of goods in the UK was principally the responsibility of the supplier. The increasing use of third party carriers has given rise to significant economy of scale by having a full load on the return journey. “Ex-works costing” has facilitated a carrier to create a milk round, delivering and collecting goods on a daily or weekly cycle, ensuring full loads at all times. Also an in-house customer can often arrange to collect raw materials in the vicinity of their important customers, or other suppliers, to gain economy of scale. However the latest statistic I have seen for the UK, still suggests that 25% of all vehicles are running empty on our roads.
(Also see Previous Technique T036: Visual Control Systems)
Along with the introduction of lean thinking has come the re-introduction of visual control systems, whereby the control of stock is now becoming genuinely achievable through “eyeball control”, rather than sophisticated computer systems and heavy reliance on “Perpetual Inventory” stock checking.
Stock recording and stock movement recording has been greatly simplified and improved by the automation of the data entry by bar coding. If used as a common identifier between sales outlet and first tier manufacturer this becomes a significant element of “Agile Manufacturing“. Radio Frequency Identification (RFID) has now made an entrance into this arena. RFID is capable of identifying an item at radio frequency distances, as opposed to bar code scanning. However this may be distracting us from the simplification of stock control which can also be achieved by visible control systems, and 2 Bin / Kanban replenishment systems.
(Internet enabled supplier integration (Also see E-commerce: “E-nabling” Your Business)
The use of computers to communicate demands between organisations has grown significantly from the introduction of faxes which were used to send schedules, thenKanbans, to the use of electronic buying via the Internet and electronic payment settlement towards e-commerce. It is now common to communicate the output of a delinquent MRP system to a supplier, with much greater speed via electronic means! Genuinely seamless, fully integrated customer ordering, to supplier sales order processing and onwards through their manufacturing scheduling and out into the next tier of the supply chain is still rare but is becoming an interesting development area, but I feel that the technical barriers are much less significant than the commercial barriers.
The computer technique of deducting component parts from stock based on the arrival of a sub-assembly at some key measurement (“deduct”) point has been developed in response to the need to automate the stock issuing process in conjunction with the introduction of JIT techniques. This has also been used as a basis for paying suppliers. However “backflushing” has created a situation where the Bill of Material accuracy for these generally inexpensive parts must be totally accurate in order to avoid stock discrepancies. Often these parts’ usage is probabilistic anyway rather than deterministic (with selective assembly shims for example) (see Previous Question Q001. Managing “C” class items in a deliver to point-of-use situation), with average usage entered into the Bills of Material (“Planning Bills of Material”) making the usage approximate. But, these parts require minimum administration and supplier top up or two bin systems are usually more appropriate than issuing MRP1schedules for these parts which are notoriously unreliable. (See Malpractice M007: The Cost of the Costing System) Any significant delay in “backflushing” from the point at which these items become committed to sub-assembly can also cause the situation to become un-auditable.
(Also see Implementing ERP Systems)
These systems grew out of MRP2 and / or Inventory Management and / or Accounting systems in the early 1990’s depending on the pedigree of the software supplier into a “total” business control system, integrating the functions and data from various areas. For example: a goods receipt would automatically update outstanding purchase orders, reduce requirements generated by MRP, update the stock balance, create a creditor in the Purchase ledger and eventually update the general ledger with the new asset. Many ERP systems incorporate MRP2 planning processes. Fewer incorporate Project Manufacturing. Many also incorporate links to Advanced Planning & Scheduling (APS) as an optional extra avoiding the MRP2 module of the software.
In 1986 it was concluded by a European conference of leading MRP guru’s, which I attended, that the logical solution to the problem of Manufacturing Planning and Control was an appropriate use of techniques to match the circumstances. Unfortunately there are a number of inhibiting factors here:
- No one method is universally applicable!
- Any one business has an array of circumstances and the situation will change with product life cycles, product mix, volume changes, and process changes.
- No one except us has yet catalogued the control systems available in a logical way such that the strengths and weakness of the techniques are apparent and can be used in a mix and match way to solve an individual problem. (See “What control systems do you need?)
- It is not a simple task to change the technique.
- Each new technique is often viewed as universally applicable and better than the last!?
- It is easier to blame the system rather than the implementation and move on to the latest fad. In particular the success rate of MRP2 implementations is very poor, such that it may be replaced when it could have been easily corrected. (See Implementing ERP Systems).
- There are a number of vested interests and significant sunken investment.
- No method is “fit and forget”.
- This is an area steeped in conventional wisdom, poor design, implementation and operation, generating a “gap” between the potential and the delivered reality. (An interesting aspect of our current research!) (Gap Analysis available on request)
Choosing a manufacturing control system to suit your environment
This poses the question, in what circumstances is an approach valid? The correct choice is primarily determined by an individual operating environment. In fact in any one company there is likely to be a diverse range of requirements. (See Expert Systems and Development Tools: What control systems do you need?). You can often make a technique work in less than ideal circumstances (in an inefficient and ineffective way), but the correct choice of technique, implemented and maintained well is the key to realising ongoing bottom line benefits. A strategic, periodic review of the appropriateness of your current application of a technique is essential to avoid degeneracy! We call this process “reimplementation”.