On 19th February 1982, over half a million dollars worth of hardware and software were shipped from Fenwick Software’s Melbourne office to Boyne Island, near Gladstone in Queensland.
The hardware is a HP3000/33 computer with a 120MB disc. The software is a system called IMPAC (Integrated Metal Production and Analytical Control) which controls the operation of the casting shop and warehouse of the new aluminium smelter owned by the Boyne Smelter consortium (BSL).
IMPAC comprises over 100,000 lines of source code in over 250 modules, which were developed over a two-year period in approximately 9,000 man hours.
Just another medium-sized project on a mini-Computer? Not quite! IMPAC has several most unusual features.
A New Problem
Molten aluminium is produced from alumina by a complex electrolytic process. As aluminium is difficult to transport in this molten state, it is transferred to large furnaces, cast into a variety of shapes and the solid metal is then warehoused and shipped.
The high-technology smelting operation has been subject to process control computers for over 15 years, but the simpler casting operations are controlled manually. Batch systems recording inventory and shipping information exist in other smelters, but IMPAC is the first on-line production and inventory control system for the casting operations of an aluminium smelter.
|
Production control systems are generally more individual than, say, payroll or general ledger, but the task of producing a world first presented a challenging (and at times frightening) prospect.
A Successful Solution
The most obvious criteria for success is that a project is completed within budget and schedule. But projects which meet these criteria are rare. IMPAC was comfortably within its budget, and effectively met its time schedule.
The original deadline for the system was 1st December 1981, which was about two months after the projected start-up date of the smelter operations. Delays in the smelter construction allowed us to relax this deadline. When a team member left the company in September, we elected to continue with reduced manpower rather than replace him, and most team members took holidays over the Christmas period.
The final deadline was chosen mostly to suit the client, with the shipment being timed so that the smelter start-up and system start-up coincided. It allowed us the extraordinary luxury of the project leader taking two weeks' leave three weeks prior to shipment.
Of course, as Jerry Weinberg says, if a system doesn't have to be reliable, it can meet any other objective. Does the fact that IMPAC met its deadlines simply mean that it didn't meet specification or that weskimped on testing and that it is full of bugs? The answer to these questions can only come from the users. A letter dated 8th May 1982 stated: "IMPAC continues to operate satisfactorily."
|
