FULL GUIDELINE:   5 August 2003

The bounds of this concerns a "manufacturing" entity, meaning a complex process demanding raw materials and delivering product in the hands of the end user.
In control terminology, it is a network with a topology consisting of a multiplicity of nodes and links.
Today, in business terms, we recognise a node as being characterised by a materials planning  process (e.g. MRP), and an elemental link as being a data connection between two nodes in the network.

In the example above, each of the nodes is itself part of other supply networks. These link to nodes in other networks which are not shown here.

If we can identify a sufficient description of a generalized link and its interaction with the node at each end, then by combining these, we can accurately build up and identify an entire network

This is the so-called single factory concept. Today's business systems can in fact perform this inherently, within a single enterprise, since they have access to all the data relevant to that enterprise. What they do not have today is unrestricted access to the data from external enterprises.

If it is possible to identify the necessary information flow needed across an elemental link between two different proprietary MRP processes, then the integration of a cooperating two-enterprise supply link can be achieved, and by extension, potentially the relevant supply network.

Using  this premise, the requirement for MRP-to-MRP bi-directional information flow can be standardised and is identified. This is nothing more than the supply & demand situation and inventory status/movement, by product.

The next stage is defining how to implement the elemental 2 linked-nodes model in reality.

Again, any node may be connected, using multiple links, to other nodes. This means that for practical, commercial and security reasons, we have been forced to restrict out partners to use data in the nodes which has been limited and filtered for that specific link.

The filter on the buyers side is a procurement system, which selects the data relating only to seller at the other end of the link. 

The filter on the sellers side is a planning system, which selects the data relating only to the buyer at the other end of the link. 

Typically these filters are very conservative, and block access for the other partner to additional information in that node, used for the MRP computations, and which would allow optimization of the overall process. It is for this reason that until quite recently , buying procurement was limited to issuing short term purchase orders and seller planning only to providing a best case match to these.

Business liability was 100%; any changes to demand/supply by either party required case-by-case negotiation.

What has now been realised is that by making much more information available to the external partner,  very impressive cost benefits can be realised, as the MRP systems become closer linked and optimisation of each elemental supply link is achieved.   The movement of data between partners increases massively, which can only be implemented with a very reliable and resilient data link.

Business liability is now shared between the partners.

Evidently these 2-node joint systems between partners should be standardised for maximum re-use and efficient implementation with all other partners. Until this is achieved, the individual links are implemented according to bilaterally agreed "service models". These are part of the dynamic process of gaining the confidence and building the capabilities of each partner, to move towards a solution of transparent linking and ultimately the "single factory".

Even then it is not finished, because we really need to optimize the relevant  network and not each individual link which makes up the supply network.  As outsourcing increases (in order to optimize other costs), there is no longer a unique "supply chain", but many different links which are controlled and  related by data movements in adjacent  links in the supply network. Examples are the use of contract manufacturing, remote warehousing, purchasing from distributors, etc. The beneficial effects of exchanging forecast and consumption data with related up-stream nodes is a well known academic exercise used in training materials managers.

This whole supply chain process should be a co-operative industry effort and can only happen if the overall objectives are understood by all parties, and adherence to agreed standards is appreciated and implemented. Customization is a temptation which has to be avoided as far as is possible.