Background and Setting

Fiberglass insulation is produced in mats several inches thick. Perforated rotating drums , “spinners”, receive molten glass from furnaces and spray glass fibers, evenly and at constant volume across the width of moving conveyer belts eight to ten feet wide. The speed of the belt therefore controls the thickness of the mat. As the mat is carried down the conveyer belt it is slit into varying sized strips by stationary knives. A particular combination of strip sizes remains unchanged until knives are repositioned during a “changeover”. The production time lost in a “changeover” incurs a “set-up cost”. Each strip can be “finished” differently downstream and additional set-up costs are incurred from changing down stream equipment. The unusable width of glass at the edge of the mat after it has been slit is called “trim”. Production is primarily to inventory but occasionally to order. Upstream and downstream manufacturing constraints forbid certain combinations of finished products and strip widths.

Client Business Objectives 

Given current inventory levels, forecasted and known demand for finished goods determine the optimal combinations of strip widths and run lengths to:

  • Minimize the number of set-ups
  • Minimize trim loss
  • Minimize inventory holding costs
  • Provide required customer service levels

How did UIA Help?

The production planning module first determines all allowable combinations of products. The module also establishes safety stock levels for finished goods which provide required service levels.
The production scheduling module is run daily with a four day planning horizon. Schedules can be modified as needed to deal with contingencies.
Production schedules are then passed to supervisory process control equipment automatically. Production schedules are also forwarded to BPCS ( SSA’s IBM AS400 manufacturing control system ) to update order entry files, and work order files used for cost control.
The system applies advanced techniques in statistical analysis, optimization (mixed integer programming) and dynamic programming. It runs over a LAN/WAN on a combination of IBM AS 400, IBM RISC 6000, and IBM PC hardware.

How the Client Benefited

  • A saving of $12 mm in reduced inventory and outside warehousing leases.
  • An $ 800,000 annual saving in time lost on set-ups.
  • A focus by R&D on relaxing the most costly manufacturing constraints as identified by the optimization model.