TPT September 2009

Deformation Control Technology Inc – USA Fax : +1 440 234 9140 Email : sales@deformationcontrol.com Website : www.deformationcontrol.com Pines Technology – USA Fax : +1 440 835 5556 Email : info@pinestech.com • Website : www.pinestech.com The results of the simulations and subsequent confirmation that simulation can provide an accurate predictive model are encouraging. Pines will continue to conduct further tests and simulations during 2009 when three CNC 250 HD machines, capable of bending 10-inch schedule 80 pipe, will be tested. The results will be published as soon as they are available. It is clear that the model results and the results from the trials collectively show that: Boosting changes the ovality direction so that the in-plane • (horizontal) tube diameter remains larger than the normal (vertical) diameter As the boost load is increased, the amount of outer wall thinning • is reduced However, the boost load must not exceed the load that could • cause buckling or separation of the tube from the inner wall of the bend die. This means that the boost load must decrease as the bend angle increases. It is also clear that an accurate model predicts tube bending results that are sufficiently accurate to be used for process design. Summary The study showed that pipe bending could be simulated successfully given accurate data for the material, interfacial friction, and bending conditions. Product development time can be reduced considerably by simulating the process first in order to establish the machine settings, such as the boost load as a function of bend angle. Getting close to an acceptable machine/tool set eliminates the trial and error process that can often prolong downtime during changeovers. Machines need to be capable of varying booster pressure during the bend cycle to minimise wall thinning while avoiding tube buckling. Carriage boosters and normal PDAs do not provide sufficient boost to overcome the material’s natural yield strength, especially as the inside wall thickness increases and the force required to deform the material becomes higher. Boosting at a high pressure for the first 60° of bend arm provided the best wall thinning ratio. However, maintaining the boost pressure at a high level after 102° of bend arm travel has two negative effects: Firstly, the pipe becomes detached from the die and, if continued, the booster will push the pipe out of the die completely, and negative ovality is caused. Pipe detachment from the die was clearly forecast by the simulation model. In agreement with experience, the model predicted negative ovality for its higher boost load. Both machines used in the tests, the Pines No. 4 and the Pines CNC 150 HD, are designed to bend 4-inch schedule 80 pipe. To bend pipe of that size with minimal outer wall thinning, it is clear that a high capacity booster is needed. The Pines No. 4 has a boost capacity of 25,000 pounds force and the Pines CNC 150 HD has a boost capability of 30,000 pounds force.

 Photo 5 : Cross section

of sample 12 at the bend location of 45° showing the outer wall thinning, the inner wall thickening, and a small amount of ovalling

Ovality Data

Sample 9

Sample 10

Sample 11

Angle of Bend

45°

125°

45°

125°

45°

125°

Vertical

1.965 1.925 1.970 1.980 1.970 1.977 2.000 2.035 1.976 1.985 1.980 1.980 -2% -6% 0% 0% -1% 0%

Horizontal

Ratio

 Table 2 : Tube ovality data Generally, over boosting will cause negative ovality. This condition exists when the horizontal axis is larger than the vertical. Conclusion A comparison of the inner and outer wall thickness change data for the bend trials in Figure 10 with the model results shown in Figures 6 and 7 shows that the simulation did achieve the objective of reduced outer wall thinning because of the greater boost load (Figure 9). However, comparison of Figure 8 predictions for ovality with the experimental ovality data in Table 2 shows that the simulation load schedule produced greater ovality than realised in the bend trials. Using the mentioned 4% limit for ovality, the model boost load schedule was predicted to produce ovality of about -3.5%, which is still acceptable but greater than the ovality measured for the trial bends (ignoring sample 9). Another simulation was run using a boost load schedule similar to that of test 8 to compare simulation results against bend tests for similar boost schedules. A comparison of inner and outer wall thick- ness changes is shown in Figure 11. For this simulation, the reduced boost load predicted that the outer wall thinning would be greater and the inner wall thickening would be less, and this is shown in Figure 11. Furthermore, the predicted wall thickness changes are in fair agreement with the measured changes from the bend trials.

 Figure 11 : Comparison of FE simulation results with booster bend trials

Wall Change, Percent

Bend Angle, Degrees

135

www.read-tpt.com

S eptember 2009

›