Vol. 2,
Issue 6
Nov - Dec 2004
Nov - Dec 2004
|
|
|
Vol. 2,
Issue 6
Nov - Dec 2004 |
|
Return to the
Front Page
Customer Base Dynamics
Scott Gardiner
As a regional steel solutions company, with assets in Australia, New Zealand,
USA, China, and South East Asia, BlueScope Steel was developing big plans for
the future. A sustained period of significant demand growth combined with
strong corporate performance underpinned a program of capacity development at
both existing as well as new manufacturing sites. The challenge was to develop
a solid understanding of the demands, such an expansion that would be placed on
the internal supply chains and on the organisation's ability to satisfy its
target markets.
Traditionally, the approach to this sort of problem was to develop a push-style
manufacturing model using spreadsheets. The scale of the corporate question
invariably leads to tradeoffs in levels of timing and detail, and in the degree
of integration. An alternative approach was sought. |
|
| isee systems, inc.'s iThink® simulation software was known in various parts of the organisation, and its combination of simulation capability, simple user interfaces, and inherent visual animations made it an ideal candidate for the research model. The focus of the project was to be on industrial simulation, rather than a 'systems thinking' basis, so there were some unknowns and risks involved in the choice, but it was felt these were outweighed by the strengths. |
|
After selecting the platform, the basic elements of
the system model needed to be determined. From the early stages, it was
apparent there were going to be challenges in scale and scope, as the business
operates in a complex supply network of interdependence as well as
independence. In order to bring a sharp focus to the customer demand aspects,
it was decided that the key behaviour of the system was to be representative of
pull, rather than push, manufacturing. Market priorities were set in four
levels, to nominally represent domestic, core market long-term export, new
market development, and other. Product groups were limited to ten high level
end products - the minimum number required to get an acceptable map to
manufacturing unit capability and routing. In order to accommodate the key
operational aspects, it would be necessary to include eighty manufacturing and
despatch units across twenty sites (including three sites that are not yet
built). The timescales needed to cover a horizon of up to five years, due to
the long construction leadtimes required to build new facilities, and it was
concluded the basic time units would need to be in weeks, to provide the best
link to major equipment maintenance cycles and the timing of new capacity.
The flows and reservoirs nature of iThink ® was a natural fit to matching the manufacturing processes being modelled. Customer demand was presented as a desired flow rate for a corresponding despatch flow. These despatches were supplied from stocks that were in turn connected to various upstream manufacturing operations and stocks. A mixture of modelling techniques combined with the basic reservoir and flow options provided a capability to model the required mass balance. This is the interaction between feed availability (via non-negative reservoirs), unit capacity (via separate convertors for how many hours a unit was available for a week and how quickly the unit could run if otherwise unrestricted), unit scrap losses (via unit conversion on the flows) and downstream stock limits (combining a reservoir for inventory accumulation and a convertor for the target maximum quantity to hold). If-then-else calculations were used to set final flows to the lesser of available feed, available room for downstream stock, or planned rate. The largest modelling challenge was related to properly reflecting market priority at each manufacturing decision point. This was achieved using an array structure throughout the model where the elements of the array represented the market priorities. This was combined with modelling each week's available operating time on a manufacturing unit as a 'reservoir of time'. The 'time stock' was then consumed by 'flows' to the market priorities, in array sequence and subject to the correspondingly arrayed upstream, downstream, and unconstrained flow limits. With the basic operational techniques in place, the focus shifted to enhancing the user interfaces. High-level performance summaries were placed on the interface layer, and navigation buttons allowed a user to move to regional summaries or to drill down into the site and unit details on the model layer. The equations have been extracted and imported into a database, where they have been parsed, documented, and cross-referenced. This corporate model allows supply chain planners to examine and test various scenarios involving demand (quantity, timing, market priority), supply options (macro switches have been set to select different aspects of internal and external supply, and convertors provide supply flow limits), and manufacturing capability (rate, timing, new startup). The key strengths of the model are in its integration and animation. The physical construction of the iThink ® model assures unit-to-unit flow integration, through the obvious linking of reservoirs to flows. Site-to-site flows are integrated via ghosted connections. As the simulation runs, the graphs on the interface layer provide continuous feedback on key factors including market despatches and shortfalls, inventory levels, and operating utilisation. Drilling down to the model layer, reservoirs can be seen filling and draining, clearly pointing out where inventory is plentiful and where it is not. More graphs provide supporting details of the products flowing to customers, and data on manufacturing conditions and performance. Some avenues for future development are being discussed, and two key areas are on the radar. One involves extending the manufacturing units to bring in more upstream and downstream details. The other area is concerned with adding manufacturing variability factors and studying the impacts on market satisfaction. This could provide support for facility upgrades or assist in determining product offers to the different markets. The system is only beginning to transition from its research and development phase into its application phase. The modelling concepts and operational implementation of the system are yet to be tested in detail, and therefore the deliverables are still more potential than actual. There are, in all likelihood, several iterations yet to be completed before overall value will be determined. |
