WHAT IS PROCESS VARIATION?
Process variations are undesired changes in production processes that can affect final product quality characteristics. These effects can be seen as fluctuations in: edge size, edge shape, geometry (sagging), wing radius, optics, and residual stresses. In order to control and eliminate any process variation, all key indicators must be identified, monitored, measured and adjusted accordingly. Key process variable #1 - temperature The most important key process variable in gravity bending processes is, naturally, temperature. The glass is sagged to its final shape with heat-treatment using infrared (IR) heating elements. The windshield is annealed and cooled at a controlled rate using a combination of natural airflow and forced cooling. In the gravity bending process, temperatures are controlled by pyrometers and thermocouples. The pyrometer measures the glass surface temperature and controls automation. Thermocouples are used to control empty chamber temperatures, both as a safety function and as a process controlling device in such case where the glass surface cannot be measured due to printing, loading location or shape of the glass. Thermocouples are very precise instruments.
However, in bending furnace technology, the applications cannot be considered as reliable response variables, because they are merely an indication of ambient temperature information in individual locations of the furnace affected by various conditions making the reading at...

WHAT IS PROCESS VARIATION?
Process variations are undesired changes in production processes that can affect final product quality characteristics. These effects can be seen as fluctuations in: edge size, edge shape, geometry (sagging), wing radius, optics, and residual stresses. In order to control and eliminate any process variation, all key indicators must be identified, monitored, measured and adjusted accordingly. Key process variable #1 - temperature The most important key process variable in gravity bending processes is, naturally, temperature. The glass is sagged to its final shape with heat-treatment using infrared (IR) heating elements. The windshield is annealed and cooled at a controlled rate using a combination of natural airflow and forced cooling. In the gravity bending process, temperatures are controlled by pyrometers and thermocouples. The pyrometer measures the glass surface temperature and controls automation. Thermocouples are used to control empty chamber temperatures, both as a safety function and as a process controlling device in such case where the glass surface cannot be measured due to printing, loading location or shape of the glass. Thermocouples are very precise instruments.
However, in bending furnace technology, the applications cannot be considered as reliable response variables, because they are merely an indication of ambient temperature information in individual locations of the furnace affected by various conditions making the reading at times very unpredictable.
Glass surface temperature measurements via pyrometer are reliable and accurate. However, without calibration and maintenance, the accuracy can suffer over a period of time. Although considered reliable, the measurements can be affected by various conditions:
• incorrect angle of the pyrometer in relation to the glass surface due to glass shape;
• pyrometer positioned over the printed area of glass;
• electrical noise in the system;
• insufficient cooling;
• dirt or moisture in the lens.
Although glass surface temperature is the number one key process variable in the gravity bending process, process controlling should rely on the combination of visual reference or/and measured process variables. For example, in bending of large windshields, this combination is preferably:
• glass surface temperature;
• visual reference during bending from mould tooling;
• hinge point status.
In serial production of car-, coach- and truck windshields, the combination is preferably:
• glass surface temperature;
• measured sagging after production cycles to correct bending time/ready temperature.
Temperature variations between bending cycles occur in bending, annealing and cooling stages. Variations in these stages can affect final product characteristics. Temperature variations will detrimentally affect: edge shape, edge size, geometry, optics and residual stresses.
Heating variations primarily affect the repeatability of sagging and edge size. Heating process variations can also be influenced by another key process variable – time – especially in relation to serial type bending furnaces. Typically, in single chamber production, timevariations are related to the thermal conditions, whereas in serial furnace production, they are related to bending cycle time variations. Variation in the annealing curve can also affect residual stress levels, edge shapes as well as optics. Firstly, if annealing is started at the wrong time, product sagging will be either flat or too deep. Secondly, when the cooling rate varies it will affect sagging again at temperatures over 550°C and residual stresses at temperatures roughly over 400°C. Modern automotive serial bending furnaces are equipped with automation that can reduce existing variation to a certain extent in heating and cooling conditions between cycles. However, single chamber bending furnaces and larger serial bending furnaces for bus windshields are rarely set-up to repeat heating curves cycle after cycle. Modern automation has enabled single chamber bending furnaces to provide detailed heating curves but in practice many manufacturers operate their furnaces without the full engagement of the available/supplied automated control systems. The primary reasons for this are lack of resources, know-how and a general lack of training. Luckily, with extra training and professional assistance the windshield manufacturers can be able to operate their furnaces with a greater degree of capability, which would enhance their output, quality and profitability.
Variations in temperature conditions can be caused by several sources, such as, for example:
changes in ambient condition, furnace insulation condition, heating element condition, loading time variations, varying operator actions and temperature control stability.

Key process variable #2 time
The longer the processed glass pair remainsunder IR-heating at temperatures over the glass transitions range the more it will bend. Time is not a controlling factor in single chamber furnace production, but in serial furnace production time plays an essential role. To perform with great repeatability, bending times need to be conducted within a certain time frame, otherwise the process will fail as the variations in bending times affect all the heating chambers as well as the annealing/cooling track. In single chamber furnaces, the most important source of variation is the actual thermal condition between each production cycle. If the cycle intervals vary greatly the thermal conditions change accordingly.

Key process variable # 3 mould tooling
Mould tooling is potentially a source of several process variations such as loading position changes, structural fatigue, and faulty adjustments to name a few (more sources listed in Chart 1). For example, mould tooling that enables visual reference on the sagging provides the bending operator important information about the bending, enabling the operator to control the sagging precisely. If this visual information (response variable) of sagging is not available or not reliable, the bending results will vary greatly cycle after cycle. This will cause poor repeatability as the final products geometry and edge sizes vary after every production cycle. In the worst case, poor visual reference, along with other process variations such as non-reliable temperature information, the glass can be overheated and drop out of the mould. Obviously, the solution is not just plain simple updating of mould tooling. As said earlier, the existing sources of other process variations can contribute and, in unison, create a negative synergy.
Repeatability in bending processes is usually further affected with several sources of process variation such as ambient conditions, various operation related actions, and bending furnace condition and set-up, to name a few.

IDENTIFICATION OF SOURCES CAUSING PROCESS VARIATION
The first step towards a well-controlled gravity bending process is the identification of all existing process variations. Chart 1 indicates most of the typical sources of process variations, showing the relatively simple items, but their existence and effects are usually neglected. All of these conditions are potential sources of process variations and majority of these conditions can be found in every existing bending furnace. How well they are identified and controlled depend on the windshield manufacturer. It is not impossible to have all these conditions affecting your production.
Control and elimination of process variation Chart 2 indicates the counter actions required to improve controls and eliminate existing process variations. Define, Measure, Analyze, Improve and Control (DMAIC) are the key to reliable longterm results with process variations. After the sources of process variations are identified, the work for control and elimination can commence. It is important to understand that certain sources causing process variation cannot be eliminated completely with existing technology, but significant improvements in repeatability, yield and constant quality can be made controlling the identified sources of process variation.
For example, standardized working practices including efficient and pre-set short cycle intervals in single chamber furnaces can decrease the cost of operation as the heat mass left into the roof heating elements, tooling and wagon will reduce the required heating power in successive bending cycles.
Additionally, controlled loading intervals within constant ambient temperatures, no draft conditions and by processing the glass pairs within similar heat mass conditions enables better and more precise heating and cooling process repeatability.

CONCLUSIONS
Single chamber bending furnace operations with current technology leave more room for process variation. Up to date serial bending furnaces, especially car windshield furnaces repeat process conditions continuously cycle after cycle simply due to more advanced technology and the nature of production method. However, we must consider that similar process variation still exists in both production equipment types. Several factors affect process repeatability and, most importantly, the negative synergy these variables together can create.
Defining, measuring, analyzing and elimination of the sources of process variation are essential parts of continuous gravity bending process development. The best counter action to the negative synergy created by process variations is to create a similar positive synergy using a combination of reliable response variables to improve repeatability. Furthermore, the elimination of arising production defects is easy and efficient only when the bending process is well controlled and repeated in a strict standardized manner.
Safety Glass Experts provides a fast lane for improvement with expert guidance for manufacturers seeking ways to improve their production, achieve better gross profit per produced piece, producing continuous quality and high yields. More information on this and other subjects will be available at the author’s eight-hour workshop: The Art of Gravity Bending Windshields, to take place at GPD 2011, followed by the author’s upcoming book: The Secrets of the Art of Gravity Bending.