Adaptive Control Of Glass Bottle Manufacturing

The Process

Bottles are made from melting silica sand, soda ash, limstone and other minor ingredients in a furnace. The molten glass is pulled through the throat of the furnace and through a distributor where it is fed to individual forehearths.

Each forehearth has three zones where the glass temperature is measured and controlled with a combined system of cooling dampers and heaters. These are used to bring the glass to its final temperature before it is sheared into gobs and delivered to the bottle blowmolding machine.

For the bottle to be molded properly with the desired volume and appearance, the gob of glass must have the correct weight; thus weight is a primary quality parameter for the finished bottle. Controlling the weight of the gob means controlling how much glass flows through the gob cutter.

Since the viscosity of glass is very sensitive to temperature, this means tightly controlling the temperature of the glass at the gob cutter.

Click on the image to see a larger version

The Control Objective

Different bottle sizes and shapes require different glass temperatures for proper molding. The control objective is to control forehearth glass temperature and to minimize the settling time when changing target temperature settings or reacting to process changes such as production rate changes, interruption to furnace gas supply, or a color change.

Conventional control systems require four to six hours to stabilize glass temperature following target or production changes, resulting in lost bottle production. Occasionally, temperature cycling persists for much longer periods and requires the attention of a knowledgeable instrument technician for several hours to stabilize the process.

Controlling forehearth glass temperature is very difficult because:

Changes in production rate effect process gain and time lag for the temperature control loop;

Long temperature response time causes conventional controllers to cycle on set point changes;

Combined heating/cooling control is non-linear;

Thermal and mechanical properties of glass change with temperature.

The existing control system consisted of three single loop controllers on a proprietary network communicating to an operator interface. In-glass probes measured the glass temperature in each zone of the forehearth.

The BrainWave Solution

Plant maintenance suggested trying BrainWave on the most difficult forehearth. This was the one directly in front of the furnace throat, which was most affected by fluctuations in the furnace temperature.

We used Dynamic Modeling Technology (DMT) to construct a mathematical model of the glass temperature process in each of the three zones. DMT models were also constructed to include the temperature of upstream sections as feedforward inputs to the downstream zone temperature controllers.

BrainWaves were implemented on a standard PC platform linked by a serial connection to the existing plant single loop controller network. The BrainWaves controlled the heating and cooling for the front and rear zones of the forehearth, using the DMT models to forecast the control adjustments required in each zone to reach and maintain the temperature setpoint.

The glass temperature of the front and rear zones was provided to the conditioning zone BrainWave as a feedforward signal. This allowed the BrainWave to anticipate the temperature control adjustments required to keep the glass temperature at setpoint.

The Results

BrainWave:

Reduced the settling time after a temperature set point change by 50%, from 4-6 hours to 2-3 hours;

Reduced the settling time after an interruption to the furnace gas supply by 33%, from 1 1/2 hours to 1 hour;

Increased acceptable production of standard container packs by up to 20%, and of specialty container packs with more precise temperature control requirements by 40%, based on two years of actual operating performance;

Improved control on set point changes resulting in an annual production increase of 4%.

Eliminated the need for control specialists to assist with setpoint changes.

The total annual savings are estimated at $150, 000, resulting in a return on investment of less than 1 year. The BrainWave is to be installed on all forehearths and the plant is considering using the BrainWave for the main furnace temperature control to reduce the temperature fluctuations at the furnace throat. This should provide a further improvement in production quality and reduce energy costs.

Click on an image to see a larger version