Professional Documents
Culture Documents
Major Stretch
From Channel
∆ FLC0
FLC0
As-Received
Minor
Stretch
Prepared by
B.S. Levy of B.S. Levy, Consultants
D.E. Green of Industrial Research + Development Institute
for
October, 2002
i
Auto/Steel Partnership Members
This publication is for general information only. The material contained herein should not be used without
first securing competent advice with respect to its suitability for any given application. This publication is
not intended as a representation or warranty on the part of Auto/Steel Partnership – or any other person
named herein – that the information is suitable for any general or particular use, or free from infringement
of any patent or patents. Anyone making use of the information assumes all liability arising from such use.
ii
Executive Summary
The Enhanced Formability Project was initiated on the basis of press shop observations that parts
with strains substantially in excess of conventional as-received forming limit curves (FLCs)
could be successfully produced in large volume without breakage. In these cases, the apparent
increase in the as-received FLC were observed in areas of the part that had been subjected to
bending and straightening through drawbeads as steel moved off a binder surface or bending and
straightening as steel moved off a post (punch). It was recognized that if advantage could be
taken of this additional formability, it would be possible to produce higher quality panels and/or
eliminate unnecessary die modifications in press shop tryout. This required experimental work to
quantify the increase in FLCs due to bending and straightening and to provide some mechanistic
basis for this behavior. This increase in an FLC after bending and straightening is described as
the enhanced FLC effect, which is represented by ∆FLCo.
The experimental concept for quantifying the enhanced FLC effect is based on the design and
construction of a channel draw die, which produces parts large enough for subsequent
experimental determination of FLCs. Since tooling radii and drawbead penetration are the
predominant factors affecting the magnitude of bending strain, considerable attention was given
to designing systems for using inserts for the drawbead and the die entry radius and for designing
and using variable penetration drawbeads. To provide back tension at the drawbead, the die was
designed with outboard drawbeads. These design approaches significantly reduced overall
experimental cost.
Since quantification of the enhanced FLC effect requires predictive relations, stringent
requirements for accuracy in determining ∆FLCo were established. Considerable development
was needed to meet these requirements.
Back tension and actual bending radii are key variables controlling the enhanced FLC effect, and
experimental methods had to be developed for measuring these variables. For back tension, a
hydraulic cylinder with a load cell was added as an auxiliary to the channel draw die.
Determining actual bending radii was more difficult. Initial attempts to measure radii from the
channel draw test pieces provided accurate measurements, but it was not possible to account for
springback. As a result, considerable work was done to develop an in-situ system for measuring
the actual radii in drawbeads. For the die entry radius, it is assumed that the actual bending
radius equals the tooling radius.
Planning for the experimental work was done by the Project Team based on available funding.
The experimental work was done over four years. Work was planned on a yearly basis, and each
year’s work was based on what had been learned and changing needs as determined by the
Project Team. Many of the ideas and analytic methods included in this report are the direct result
of input from the Project Team.
Initially, AKDQ steel and round drawbeads were evaluated. As work progressed, square, double
round and double square drawbeads, and BH210, HSLA, and DP600 steels were evaluated.
iii
Predictive relations were developed for use in press shops and to provide a mechanistic basis for
understanding the enhanced FLC effect. For use in press shops, ∆FLCo is predicted by the net
thinning strain resulting from bending and straightening. For a mechanistic understanding both
∆FLCo and net thinning strain are predicted using a bending strain factor (BSF) and the ratio of
the restraining stress to the yield strength for the operative ratio of minor to major strain (BTR).
The predictive relations were evaluated using both conventional statistical methods and special
techniques that determine variation due to measurement uncertainty. Considerable effort was
devoted to generating and analyzing raw data to statistically determine the measurement
uncertainty for all the variables in the predictive equations. The results of these analyses indicate
that the predictive equations for ∆FLCo are reliable and that 70% of the variation between
predicted and actual values are the result of measurement error. The unexplained variation is
equivalent to ∆FLCo of ± 1.5 strain percent, which is well within the engineering requirements
for press shop applications. This unexplained variation is the result of several engineering
approximations that were made to control experimental costs. These approximations are
identified in the text.
The mechanical basis of the enhanced FLC effect is changes in dislocation substructure resulting
from cyclic strain hardening. BSF and BTR, which are known to effect dislocation substructure,
are found to accurately predict ∆FLCo and net thinning strain. Furthermore, net thinning strain
can be used to predict prestrained yield strength with reasonable accuracy. The internal
consistency of these relationships provide an underlying rationale for the enhanced FLC effect.
Press shop applications for the enhanced FLC effect are (1) steel that has moved through a
drawbead and over a die entry radius and (2) steel that has moved off a post (punch) with an
entry radius to steel thickness ratio of about 5 or less. Details of how to use net thinning strain to
predict ∆FLCo are included in the text.
It is shown in this study that 60% of the net thinning strain from bending and straightening does
not affect the subsequent deformation limit in stamping processes. However, the corollary of this
observation is that 40% of the net thinning strain reduces the capability for subsequent
deformation. Thus, the importance of the enhanced FLC effect is that it adjusts conventional
FLCs so that they are not an overly conservative failure limit for breakage.
The enhanced FLC effect has been verified for AKDQ, BH210, HSLA, and DP600 steels in a
thickness range from 0.7 to 1.2 mm. On the basis of the underlying mechanisms, the enhanced
FLC effect should apply to other steels with comparable microstructures and cyclic stress-strain
behavior at high strain amplitudes. It is also reasonable to expect that the thickness range for
using the enhanced FLC effect can be extended, but precise limits can not be determined at this
time.
This report describes the quantification of the enhanced FLC effect and discusses its fundamental
basis. A companion report, which includes case studies, describes the application of the
enhanced FLC effect in press shops and is intended for use by press shop personnel.
iv
TABLE OF CONTENTS
Page No.
2.2.1 Overview.............................................................................................................. 8
2.2.2 Press .................................................................................................................. 9
2.2.3 Die .................................................................................................................. 9
2.2.4 Sample Preparation ............................................................................................ 20
2.2.5 Forming Channel Sections................................................................................. 21
2.2.6 Hydraulic System for Determining DBRF ........................................................ 23
2.2.7 Reproducibility Statistics for DBRF...................................................................26
1
Page No.
2
1.0 Introduction
The Enhanced Forming Limit Curve Project was initiated on the basis of press shop observations
that parts with strains substantially in excess of the conventional as-received forming limit curve
(FLC) could be successfully produced in large volume without breakage. In these cases, the
apparent increase in the as-received FLC were in areas of the part that had been subjected to
bending and straightening as steel moved off a binder surface or off a post (punch). It was
recognized that if advantage could be taken of this additional formability, it would be possible to
produce higher quality panels and/or eliminate unnecessary die modifications in press shop
tryout. This required experimental work to quantify the increase in FLCs due to bending and
straightening and to provide some mechanistic basis for this behavior. The increase in an FLC
after bending and straightening is described as the enhanced FLC effect.
The experimental concept for quantifying the enhanced FLC effect is based on a channel draw
die which produces parts large enough for subsequent determination of an FLC using the
Marciniak double blank method. Since drawbeads and die and punch entry radii are the
predominant source of bending and straightening in stamping dies, considerable attention was
given to a system for using inserts for the drawbead and the die entry radius so that a range of
geometries could be evaluated. In order to provide back tension at the drawbead, the die was
designed with outboard drawbeads. This unconventional approach was selected because of its
low cost.
With the basic experimental approach determined, several experimental methods were developed
to obtain the needed data quality. The first requirement was developing a more accurate system
for determining FLCs because determining the enhanced FLC effect requires comparing an
experimental enhanced FLC with the experimental as-received FLC. Since accurate predictive
relations were needed, stringent requirements for accuracy were established.
Since the original experimental planning had identified back tension and actual bending radius as
key variables, experimental methods were developed for determining these variables. For back
tension, a hydraulic cylinder with a load cell was added as an auxiliary to the channel draw die.
Determining actual bending radius was more difficult. Initial attempts to measure radii from the
channel draw test pieces provided accurate measurements, but it was not possible to account for
springback. As a result, considerable work was done to successfully develop an in-situ system
for measuring the actual radii in drawbeads. For the die entry radius, it is assumed that the actual
bending radius equals the tooling radius.
Planning for the experimental work was done by the Project Team based on available funding.
The experimental work was done over four years. Work was planned on a yearly basis, and each
year’s work was based on what had been learned and changing needs as determined by the
Project Team. Many of the ideas and analytic methods included in this report are the direct result
of input from the Project Team.
Initially, work was planned using AKDQ steel and round drawbeads. As work progressed,
square, double round, and double square drawbeads were included in the program. Another key
change was the design and use of variable penetration drawbeads so that different drawbead
3
conditions could be evaluated more economically. Also, in the last year of the program it was
decided to evaluate BH210, HSLA and DP600 steels.
Since the project objective was developing quantitative relationships for predicting the enhanced
FLC effect, considerable effort was devoted to determining the experimental variability of all
measured quantities used in its prediction. This information is used to evaluate the reliability of
the predictive equations.
Since the results of the project are for industrial use, the work is reported in two parts. This
report describes the quantification of the enhanced FLC effect and discusses its fundamental
basis. A companion report, which includes cases studies, describes how the enhanced FLC effect
should be applied in press shops and is intended for use by press shop personnel. Thus, the
Applications section of this report describes in technical terms, some of the factors that must be
considered for press shop work. This report also includes some discussion of the use of the
enhanced FLC effect in FEA work.
The experimental work done in this project is more extensive than is included in this report. The
complete data set with detailed description of experimental procedures, measurements and
results is available from the Auto/Steel Partnership on 2 CD-ROMs referred to as:
• A/SP Enhanced Formability Project: Database (February 28, 2002)
• A/SP Enhanced Formability Project: Bead Wrap Videos (February 28, 2002)
4
2.0 Experimental Procedures and Materials
2.1 Materials
Four drawing quality, two BH210, two HSLA, and one DP600 steel were evaluated. These steels
are low carbon, aluminum-killed products with the following microstructures:
Of the nine lots of steel that were evaluated; there were three cold rolled, four electrogalvanized,
and two hot dip galvanized. These steels are identified in the body of the report by nominal
thickness, coating and steel grade; e.g., 0.7 mm EG AKDQ. Tabular data on composition, coil
widths, coatings, and surface topography are included in Appendix 2.1.
The average [(L+T+2D)/4] tensile properties for the steels evaluated in this project are shown in
Table 2.1.1. The tensile testing was done in triplicate at various steel company laboratories using
various modifications of ASTM A370 (E8) with initial crosshead velocities ranging from
3.175 mm/min to 25.4 mm/min. For tests run with slower initial crosshead velocities, crosshead
velocity was increased at strains between 2% to 5% (or the end of yield point elongation if it is
higher) to crosshead velocities ranging from 12.7 to 25.4 mm/min.
It can be seen from Table 2.1.1 that for each lot within a steel grade, (1) tensile properties are
similar, and (2) thickness is in two ranges, 0.716/0.787 and 1.168/1.198 mm. The only major
discrepancy is that one HSLA lot exhibits substantial yield point elongation, and the other does
not. All other lots exhibit no significant discontinuous yielding under standard ASTM test
conditions.
5
Table 2.1.1
Average* Tensile Properties of As-received Steels
Drawing Quality BH210 HSLA DP600
0.7 0.7 1.2 1.2 0.7 0.8 0.8 1.2 1.2
CR EG CR EG EG EG HDG CR HDG
Thickness (mm) 0.716 0.734 1.198 1.233 0.732 0.780 0.787 1.168 1.180
Yield Strength (MPa) 178 182 178 180 245 274 416 380 417
Yield Point
0 0 0 0 0 0 5.9 0 0
Elongation (%)
Tensile Strength
325 306 317 314 366 351 460 493 618
(MPa)
Uniform Elongation
0.198 N/A 0.197 N/A N/A N/A 0.153 0.138 0.131
(True Strain)
n-Value (True Strain) 0.212 0.211 0.209 0.205 0.178 0.156 0.168 0.149 0.136
K-Value (MPa) 559 N/A 544 N/A N/A N/A 736 762 937
Range for n and K
10/20 10/20 10/20 10/20 10/20 10/20 ** ** 10/UE
(%)
Total Elongation (%) 40.7 40.8 42.5 41.6 33.4 33.8 28.9 27.9 24.4
R-Value 1.70 1.78 1.65 1.56 1.83 1.55 1.05 1.06 1.12
R-Value Extension
15 15 15 15 15 15 ** ** 15
(%)
* Average = (L + T + 2D)/4
** Not consistent
In order to facilitate a comparison of the tensile properties of the four steel grades, average data
for selected properties for each steel grade are shown in Table 2.1.2.
Table 2.1.2
Summary of Average As-received Tensile Properties for Each Steel Grade
Steel Grade
Property*
Drawing Quality BH210 HSLA DP600
Yield Strength MPa 178/182 245/274 380/416 417
Tensile Strength MPa 306/325 351/366 460/493 618
n-Value 0.205/0.211 0.156/0.178 0.149/0.168 0.136
Total Elongation (%) 40.8/42.5 33.4/33.8 27.9/28.9 24.4
R-Value 1.56/1.78 1.55/1.88 1.05/1.06 1.12
Number of Lots 4 2 2 1
* Average = (L + T + 2D)/4
It can be seen from Table 2.1.2 that yield strengths range from 180 to 420 MPa, tensile strengths
range from 300 to 600 MPa, n-values range from 0.136 to 0.211, total elongations range from 24
to 42% and R values range from 1.0 to 1.9. As expected, as strength increases, strain hardening
rate (n-value) and total elongation decrease. For R , the drawing quality and BH210 steels exhibit
6
values in the range 1.55 to 1.88, and the HSLA and DP600 lots exhibit values in the range of
1.05 to 1.12.
The directional R values of the four steel grades also exhibit different patterns of anisotropy. As
shown in Table 2.1.3, the drawing quality and BH210 steels exhibit low R values in the diagonal
direction and positive values of ∆R. In contrast, the HSLA lots exhibit low directional R values
in the rolling direction and negative values of ∆R. The DP600 lot exhibits a greater degree of
planar isotropy.
Table 2.1.3
Directional R Values for As-received Steels
RL RD RT R ∆R
Drawing Quality
0.7 CR 1.86 1.38 2.20 1.70 +0.65
0.7 EG 1.85 1.48 2.29 1.78 +0.59
1.2 CR 1.92 1.24 2.20 1.65 +0.82
1.2 EG 1.77 1.19 2.08 1.56 +0.74
BH210
0.7 EG 1.71 1.66 2.28 1.83 +0.34
0.8 EG 1.60 1.36 1.86 1.55 +0.37
HSLA
0.8 HDG 0.62 1.30 0.98 1.05 –0.50
1.2 CR 0.72 1.23 1.04 1.06 –0.34
DP600
1.2 HDG 1.04 1.06 1.34 1.12 +0.13
Table 2.1.4 shows the experimental values for FLCo in the as-received condition in both
engineering and true strain. The as-received forming limit curves (FLCs) are included in
Appendix 3.1. The experimental method for determining FLCs is discussed in a subsequent
section.
Table 2.1.4
Experimental and Predicted FLCo for As-Received Steels
Experimental FLCo (%) Predicted FLCo* (%) Predicted-Experimental (%)
Drawing Quality
0.7 CR 32.3 (0.280)** 33.8 (0.291)** +1.5 (0.011)**
0.7 EG 33.7 (0.290) 33.9 (0.292) +0.2 (0.002)
1.2 CR 35.5 (0.304) 40.1 (0.337) +4.6 (0.033)
1.2 EG 36.6 (0.312) 40.0 (0.336) +3.4 (0.024)
BH210
0.7 EG 24.4 (0.218) 38.5 (0.251) +4.1 (0.033)
0.8 EG 26.6 (0.236) 25.5 (0.227) –1.1 (0.009)
7
Table 2.1.4 (Continued)
Experimental and Predicted FLCo for As-Received Steels
Experimental FLCo (%) Predicted FLCo* (%) Predicted-Experimental (%)
HSLA
0.8 HDG 22.2 (0.200) 27.6 (0.244) +5.4 (0.044)
1.2 CR 27.0 (0.239) 28.3 (0.249) +1.3 (0.010)
DP600
1.2 HDG 25.0 (0.223) 25.9 (0.230) +0.90 (0.007)
* Predicted using Keeler-Brazier relationship (Equation 2.1.1)
** True Strain
FLCo can also be predicted from the Keeler-Brazier1 Equation as shown below:
The predicted values of FLCo using the average n-value and the difference between the
experimental and predicted values of FLCo are also shown in Table 2.1.4. It can be seen from
Table 2.1.4 that the predicted values of FLCo are consistently larger than the experimental
values. On average, the calculated values are 2.5 strain percent higher with a range from –1.1 to
+5.4 strain percent. From discussions between Stu Keeler and Daniel Green, it has been
established that the IRDI experimental method2 is more sensitive than that used by Keeler and
Brazier. Thus, it is reasonable that the predicted results should be higher than experimental
results determined using the IRDI method. Thus, it is concluded that the IRDI experimental
results confirm the Keeler-Brazier Equation with an average negative bias of 2.5 strain percent.
In examining the difference between experimental and predicted values of as-received FLCo, it
can be seen that the lowest difference is for the 0.8 EG BH210, and the three highest differences
are for the 1.2 CR AKDQ, 0.7 EG BH210, and the 0.8 HDG HSLA steels. This indicates that
yield point elongation or steel grade does not necessarily bias variations between predicted and
experimental values of FLCo.
2.2.1 Overview
1
S. P. Keeler & W. G. Brazier, (1977); “Relationship between laboratory material characterization and press shop
formability”, Proceedings of Microalloying 75, New York, pp.517-530
2
D. E. Green & K. C. Black, (2002); “A visual technique to determine the forming limit for sheet materials”, S.A.E.
Technical Paper 2002-01-1062
8
The Tooling and Press Operations section describes the equipment and
experimental methods used to produce channel draw test pieces and to
determine draw bead restraining force. The equipment consists of a
press, a channel draw die, inserts for the channel draw die, and an
instrumented hydraulic system which is used to determine draw bead
restraining force. Thickness and surface strain measurements are
determined on channel draw test pieces. Test pieces for determining
FLCs and for tensile testing are taken from the wall of channel draw
pieces.
2.2.2 Press
A self-contained, 300 ton hydraulic press with the following features was used:
2.2.3 Die
The die was designed with two objectives; (1) to use an outer set of draw beads as an
inexpensive way of producing controlled back tension and (2) to provide sufficient width for
plane strain deformation. Since constant deformation conditions are required in the test pieces
used for determining FLCs, the outer draw beads are located so that material which passes
9
through the outer draw bead never enters the inner draw bead. Also, kiss blocks are used so that
the only friction between the test strip and die is in the draw beads and over the die entry radius.
10
Punch
Upper Die
Inboar d Outboar d
Bead Bead
Figure 2.2.4
Scaled Schematic of Channel Draw Die Showing the Post and
the Binder Position at the Completion of the Press Stroke
11
Figure 2.2.5
Scaled Schematic of the Channel Draw Tooling Stack without the Post
12
Male Bead
Entry Exit
R2 R3
R1 R4
Female Insert
Figure 2.2.6
System for Identifying Radii in a Single Square Drawbead
Penetration is described schematically in Figures 2.2.7 and 2.2.8. For this study, zero penetration
is defined as the point where the male bead is in contact with the sheet material, but the sheet is
not subject to any bending. In contrast, 100% penetration is defined as when the centers of
curvature for both the male and female beads lie on the same horizontal plane. Partial penetration
is determined by the distance of the center of curvature of the male bead from the plane of zero
penetration, divided by the distance between the planes defining zero and full penetration.
Depending on the design of the drawbead, penetrations greater than 100% are possible.
Figure 2.2.7
Schematic Defining 0% Drawbead Penetration
13
Upper Insert with Male
Sheet Material Bead
Figure 2.2.8
Schematic Defining 100% Drawbead Penetration
Drawbead clearance is defined as the distance between the male and female bead at 100%
penetration. Unless otherwise specified, the reported clearance is per side. Air gap is defined as
clearance minus steel thickness. Unless otherwise specified, air gap is based on original
thickness. Schematics showing examples of clearance are shown in Figures 2.2.9 and 2.2.10.
These figures also illustrate how male and female bead widths are defined.
10.8 mm
1.4 mm
Clearance
13.6 mm
Figure 2.2.9
Non-Scale Schematic of a Single Square Drawbead
14
8 mm 8 mm
1.4 mm 2.8 mm
Clearance Clearance
10.8 mm 13.6
Figure 2.2.10
Non-Scale Schematic of Round Male Drawbead Inserts with Different Clearances
In the course of the project, various inner and outer drawbead inserts were used. In the initial
tests, fixed penetration drawbeads were used. Once the need for a range of drawbead
penetrations was recognized, adjustable draw bead inserts were designed and produced. A
description of all draw bead inserts produced for this project is provided in Table 2.2.1. It can be
seen from Table 2.2.1 that more male inserts were produced than female inserts, because in some
cases, a number of male inserts can be used with a single female insert.
15
Table 2.2.1
Description of Drawbeads Inserts
16
Table 2.2.1 (Continued)
Description of Drawbeads Inserts
Notes: 1) Square male beads have a flat land of 2.8 mm between the two radii
2) Penetration depends on material thickness and the kiss gap. Calculations are made assuming a kiss gap of 0.015 inches
17
Drawbead inserts were produced from either Cast Cut or D2 tool steel. Cast Cut inserts were
surface ground to the final geometry with a medium grade grinding wheel, polished, and flame
hardened to 60 – 65 HRC. D2 inserts were machined from stock at a hardness of 50 – 55 HRC,
surface ground to the final geometry with a medium-grade grinding wheel and polished. The
working surfaces of the male beads were then ion-nitrided to a hardness of at least 60 HRC by
the Exactatherm process with a case depth > 0.15 mm and then repolished. It should be noted
that the die entry radius on the inboard drawbead inserts was not ion-nitrided.
Details of construction of outboard and inboard inserts with adjustable penetration are shown
respectively in Figures 2.2.11 and 2.2.12. Further details of square and round male beads are
shown in Figure 2.2.13, and a picture of an inboard bead insert is shown in Figure 2.2.14. A
schematic of an adjustable penetration inboard double drawbead insert is shown in Figure 2.2.15.
The penetration of both single and double drawbeads was adjusted by placing appropriate shims
between the male bead and the cap on the back of each insert. The resulting variation ranges
from a minimum of approximately 20% to a maximum of in some cases somewhat more than
100% penetration.
Figure 2.2.11
Schematic of Adjustable Penetration Outboard Drawbead Insert
18
Figure 2.2.12
Schematic of Adjustable Penetration Inboard Single Drawbead Insert
Figure 2.2.13
Schematic of Round and Square Male Beads Used with the
Adjustable Penetration Inner Drawbead Insert
19
Male Bead
Die
entry
radius
Figure 2.2.14
Picture of Adjustable Penetration Inner Drawbead Insert
Figure 2.2.15
Schematic of Adjustable Penetration Inboard Double Drawbead Insert
Blanks were produced by shearing to a width of 254 mm with the exception of the 0.8 mm EG
BH210 where the blank width is 229 mm. Blank lengths ranged from 1525 to 1650 mm. The
blank width is in the rolling direction for all steels except the DP600 where it is transverse to the
rolling direction.
20
When the blank width is oriented in the rolling direction, the prestraining direction in producing
channel draw pieces and the loading direction in producing samples for FLC evaluation are
transverse to the rolling direction.
Prior to forming, blanks were marked with a 0.1 inch diameter circle grid using the following
procedure:
1. The blank surface is degreased using Varsol and then dried with a squeegee
2. The blank surface is cleaned with all-purpose cleaner and then dried with a squeegee
3. The blank is covered with electrolyte (Lectroetch A112 for the galvanized sheets and
Lectroetch 53NC for the uncoated steel sheets)
4. A stencil (with a 0.1 inch circle grid) is placed on the blank and any trapped air is removed
5. A pad soaked in electrolyte is placed over the stencil
6. The variable transformer on the AC power unit is set to 60-70% of the maximum voltage
(100% equals 24 Volts). This will achieve a clear etch without unduly reducing stencil life
7. A roller is moved up and down the pad (with sufficient force to push electrolyte out of the
pad) for 30-45 seconds. This etches the grid on the blank
8. The blank is treated with cleaner/neutralizer
9. The lubricant (Quaker Ferrocote 61MAL HCL-1 or Ameriform 1351 Dryfilm) is applied to
both sides of the blank with a paint roller which prevents oxidation of the steel surface. No
additional lubrication was applied prior to forming test pieces.
In order to guarantee a clear, crisp grid marking, the following additional procedures were used:
– To prevent clogging, the stencils were soaked in diluted CLR for several hours to
unclog the stencil
– A clean stencil was used every 20 blanks or when grids were no longer crisp
– Fresh electrolyte was used approximately every 50 blanks or when the electrolyte became
visually contaminated
– Due to shrinkage and soiling, the pad was replaced every 75 blanks
Etched blanks were stored grid side up and covered with paper for protection against dirt and
particles. Thirty-three gridded blanks were required for each prestrain condition (28 blanks for
channel draw prestrain tests and five blanks for drawbead restraining force tests). Unmarked
blanks (20 blanks for each prestrain condition) were also cleaned, lubricated and stacked in the
same way. These unmarked blanks were used for “warm-up” to condition the die prior to
forming gridded channel blanks.
For each prestrain condition, 15 to 20 warm up pieces were produced followed by 28 gridded
pieces.
The A/SP channel draw die was used to prestrain sheet samples by forming each blank into an
open-ended channel section as shown in Figure 2.2.3. As a channel is being formed, the floating
21
binder maintains a constant pressure against the upper
die by means of pressure pins that are connected to a
pressure cushion underneath the bolster. Four “kiss
blocks” mounted on the binder guarantee a constant
clearance of material thickness plus 0.4 mm between
the binder and the upper die. Material on either side of
the punch flows through drawbeads which are shown
in Figure 2.2.2. The outboard bead inserts, A and D,
are located 254 mm behind each inboard drawbead.
The outboard drawbead creates the back tension on
the material flowing through the inboard drawbead
with no material which passes through the outer
drawbead, passing through the inner drawbead.
• Cushion Pressure (in each four 300 mm diameter cylinders): 2500 psi
• Pressing Stroke: 240 mm
• Stroking Cycle: Automatic
• Die Opening: Maximum
• The required drawbead inserts were set in the die for each prestrain condition.
• The working surfaces of the drawbead inserts were polished in the direction of metal motion
prior to each test condition. For some test conditions, material build-up occurred on the
drawbeads or die entry radius. In these cases, the inserts were cleaned and polished to
minimize scoring of subsequent channel
pieces. The drawbeads were polished by hand
with 600 grit SiC paper in the longitudinal
direction of the bead as shown in Figure 2.2.16.
• As shown in Figure 2.2.17, a bracket made of
angle iron was adjusted and bolted into
position so as to consistently locate each blank
prior to producing channel draw pieces.
• The correct kiss blocks were mounted at the
four corners of the binder as shown in Figure
2.2.17. Pieces of solder were placed at each
corner of the binder and over the draw bead
inserts to check the kiss gap distance prior to
forming channels.
22
• Great care was taken during die setup to ensure that each
side of the channels would be equally prestrained.
However, even though the drawbeads on both sides of
the channel nominally have identical geometry, they are
never exactly the same. Thus, the major strain in each
sidewall can be slightly different.
• The formed channel sections were stored on their edge
in such a way that the right and left hand sidewalls could
be identified. The right hand sidewall was used for FLC
and tensile testing.
• The formed channel sections were marked with a pencil
scribe to identify the prestrain condition.
• After the 28 gridded channel pieces were formed, the
bead height was measured at 4 points along the length of
each male bead. The average of the 4 bead height
measurements was recorded in the database. This
measurement is used in the calculation of drawbead
penetration.
• Large pieces to be used for subsequent tensile or
Marciniak Double Blank testing were removed from the
channel sidewall by nibbling. Actual test pieces were then sheared to final size.
DBRF is determined with a device designed and constructed at IRDI. Essential elements of this
apparatus are shown in Figure 2.2.18. In this device, a 127 mm diameter hydraulic cylinder with
a 305 mm stroke is mounted on a rigid frame and bolted to the press. A hydraulic wedge grip
capable of clamping a 254 mm wide blank is
attached to the cylinder rod as shown in
Figure 2.2.19. Two parallel guide bars enable the
hydraulic clamping unit to slide back and forth
with little friction. A hydraulic power unit with a
20 HP motor and a pump with a maximum flow
rate of 50 l/min are used to move the hydraulic
cylinder at speeds of up to 65 mm/s for a 254
mm wide blank.
23
with the die closed, the pulling force and the displacement of the clamp are recorded with data
acquisition software developed at IRDI.
Considerable effort was put into positioning the DBRF apparatus in the press because it is
essential that the pulling direction of the hydraulic clamp be in the same horizontal plane as the
binder surface and that it remain perpendicular to the longitudinal axis of the drawbeads
throughout the entire stroke. In spite of efforts to accurately position the DBRF apparatus on the
side of the press, for each phase of the project, minor differences in setup were unavoidable.
These small differences in setup contribute to the differences in DBRF in repeat tests.
DBRF tests with fixed outboard drawbeads were performed to determine the total restraining
force required to pull a 254 mm wide blank through each specific combination of outboard
drawbead used in this study. These DBRF tests were done immediately after producing the 28
channel draw test pieces used for subsequent testing. The purpose of these tests is to determine
the restraining force on the inboard drawbead.
• As shown in Figure 2.2.20, the upper binder is raised slightly above the zero position of the
punch and support blocks are placed under the binder to maintain the same clearance as used
when producing channel draw test pieces (Figure 2.2.20).
• The drawbead inserts B, C, and D are removed from the die, and flat inserts (no beads) are
installed in their place (Figure 2.2.20).
• The hydraulic wedge grip is brought “in” towards the die, and a blank is clamped
• The press ram is brought down onto the binder until the pressure on the support blocks
reaches 10 MPa.
• With the data acquisition program in recording mode, the horizontal actuator is retracted and
the strip is pulled out of the left hand side of the die at a velocity of about 30 mm/s which is
the approximate drawing rate for producing channel draw test pieces.
• For fixed drawbeads, the experimental data (force, displacement, and time) are recorded for
approximately five pieces for each test condition.
• The die and the hydraulic clamp are opened, and the deformed strip is removed.
24
It should be noted that the hydraulic actuator with the hydraulic wedge grip attached was
checked at the beginning of the 200, 300, 400, and 500 test series. This was to ensure that the
pulling direction was in the same horizontal plane as the binder face and also perpendicular to
the axis of the drawbeads.
Figure 2.2.21, which shows DBRF as a function of sampling rate, exhibits a number of
interesting features. The load peak at point A is related to the force necessary to close the bead.
The yield drop at Point B relates to the transition from static to dynamic friction as strip begins to
move through the drawbead. The reported DBRF is an average taken through the steady state
region of the curve.
50
30
of stroke
20
Figure 2.2.21
Example of a Plot of Drawbead Restraining Force as a Function of Sampling Rate
In the latter phases of the project, test numbers 300 and higher, adjustable penetration drawbead
inserts were used in the outer position. For these cases, DBRF was determined using the
following procedure. For each material/lubricant combination, the outboard bead penetration was
incremented 8 to 10 times from approximately 20% to 100% penetration. At each penetration, a
series of five DBRF tests were carried out according to the procedure previously described for
fixed drawbeads. The DBRF versus penetration curve was determined by regression analysis.
These curves were used to determine the outboard bead penetration required to achieve a
prescribed level of back tension for channel draw tests. In these cases, back tension was not
measured during the production of channel draw test pieces, but was calculated from the DBRF
versus penetration curves.
The results of this work are shown in the following figures. For this work, the prelube is Quaker
Ferrocote 61 MAL HCL-1, and the dry film is Ameriform 1351. The regression results for
these tests are included in these figures.
25
2.2.7 Reproducibility Statistics
As indicated previously, the drawbead restraining force is calculated from data points in the
steady state region of the DBRF versus sampling rate curve. These curves exhibit fluctuations
which result from differences in friction, various dynamic effects, and other variations in the test
system. All these factors are considered in determining variability in measuring DBRF. The
rationale for this approach is that the critical local neck occurs at a specific point in the channel
draw wall that relates to conditions when that point passed through the drawbead. Thus the
average standard deviation is calculated from the data in the steady state region of the five
replicate tests, for the data shown in Figures 2.2.22 through 2.2.26.
Standard deviations can be additive or proportional; that is, the higher the average DBRF, the
higher its standard deviation. Figure 2.2.27 shows the relationship between standard deviation
and average DBRF for each test condition. It can be seen from Figure 2.2.27 that on an overall
basis, the standard deviation does not depend on the average DBRF. However, Figure 2.2.27
masks patterns specific to each steel-lubricant combination. This effect is shown in Figure 2.2.28
where the relationship between the standard deviation and average DBRF for each steel-lubricant
combination is shown separately. It can be seen from Figure 2.2.28 that for each steel-lubricant
combination, the standard deviation increases as the average DBRF increases, but that the rate of
increase varies substantially between steel-lubricant combinations. It can also be seen from
Figure 2.2.28 that at constant average DBRF, there is a substantial variation in standard deviation
that depends on the steel-lubricant combination. Thus, it is necessary to consider if each steel-
lubricant combination should be treated separately or if the data should be combined.
26
35,000
30,000
3 2
y = -22.36x + 339.9x + 1053x + 7210
25,000 2
R = 0.996
20,000
0
0.0 2.0 4.0 6.0 8.0 10.0 12.0
Penetration (mm)
Figure 2.2.22
Drawbead Restraining Force Versus Penetration for 0.7 mm Cold Rolled AKDQ
Showing the Effect of Prelube and Dry Film Lubrication
27
80,000
velocity: 3 cm/s (72 in/min)
70,000
Draw Bead Restraining Force (N)
60,000 3 2
y = -137.4x + 2157x - 2760x + 23520
2
R = 0.997
50,000
20,000 3 2
y = -93.27x + 1484x - 1541x + 14840
2
R = 0.992
10,000
0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
Penetration (mm)
Figure 2.2.23
Drawbead Restraining Force Versus Penetration for 1.2 mm Cold Rolled AKDQ
Showing the Effect of Prelube and Dry Film Lubrication
28
35,000
30,000
25,000
20,000
15,000
y = -52.42x3 + 748.8x2 - 186.3x + 9335
R2 = 0.997
10,000
5,000
0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
Penetration (mm)
Figure 2.2.24
Drawbead Restraining Force Versus Penetration for 0.8 mm
Electrogalvanized BH210 with Prelube
29
100,000
Draw Bead Restraining Force (N)
40,000
4 3 2
y = 9.168x - 266.8x + 2409x - 4325x + 16707
20,000 2
R = 0.997
0
0.0 2.0 4.0 6.0 8.0 10.0 12.0
Penetration (mm)
Figure 2.2.25
Drawbead Restraining Force Versus Penetration for 0.8 mm
Hot Dip Galvanized and 1.2 mm Cold Rolled HSLA with Prelube
30
120,000
80,000
60,000
3 2
y = 7.02x + 142.7x + 8617x + 11646
2
R = 0.993
40,000
20,000
0
0.0 2.0 4.0 6.0 8.0 10.0
Penetration (mm)
Figure 2.2.26
Drawbead Restraining Force Versus Penetration for 1.2 mm Hot Dip Galvanized DP600 with Prelube
31
900
800
700
Standard Deviation (N)
600
500
400
300
200
100
0
0 20,000 40,000 60,000 80,000 100,000 120,000
Average Drawbead Restraining Force (N)
Figure 2.2.27
Standard Deviation Versus Average Drawbead Restraining Force for all Test Conditions
32
900
800
700
Standard Deviation (N)
600
500
400
300
200
0.7 mm AKDQ - Prelube 0.7 mm AKDQ - Dry Film
100 1.2 mm AKDQ - Prelube 1.2 mm AKDQ - Dry Film
1.2 mm DP600 - Prelube 0.8 mm BH210 - Prelube
0 0.8 mm HSLA - Prelube 1.2 mm HSLA - Prelube
0 20,000 40,000 60,000 80,000 100,000 120,000
Average Drawbead Restraining Force (N)
Figure 2.2.28
Standard Deviation Versus Average Drawbead Restraining for Each Thickness-Steel Grade Combination
33
While it is possible that the observed behavior of the standard deviation is steel dependent, there
is no clear pattern with any lubrication condition or material property that would explain the
observed differences. Another possible explanation is differences in test setup. That is, the DBRF
versus penetration curves were each determined in a separate series of tests. While not definitive,
it seems more likely that the different behavior of each steel-lubricant combination is related to
test setup. Also, since the purpose of analyzing the variation in DBRF is to determine
experimental noise in comparing predicted versus actual results, it would be very difficult to
consider each lot of steel separately. Thus, it was decided to analyze all the data as a group.
In examining the sample population of standard deviation of DBRF (69 data points), its average
is 382.0 newtons with a standard deviation of 155.4 newtons. These same data are shown as a
cumulative frequency distribution in Figure 2.2.29. At ± two sigma, the standard deviations
should range from 71 to 693 newtons. It can be seen from Figure 2.2.29 that as expected about
95% of the data points are within this range. However, it can also be seen that the lower tail of
the distribution is not normally distributed because at minus three sigma, the standard deviation
would be negative which is not physically possible.
Given the nature of the statistical distribution of standard deviations, the average standard
deviation is the best estimate of the population standard deviation.
Other data on the standard deviation of DBRF are available from tests on the fixed outer
drawbeads. Since these results are very limited and fall within the range established by the
variable penetration drawbeads, these data are not incorporated in the analysis of standard
deviation of DBRF
34
100
90
80
70
Frequency (%)
60
50
40
30
20
10
0
0 100 200 300 400 500 600 700 800 900 More
Standard Deviation of Drawbead Restraining Forces
Figure 2.2.29
Experimental Cumulative Frequency Distribution for the Standard Deviation of Drawbead
Restraining Force for all Test Conditions Shown in Figure 2.2.27
35
2.3 Relating DBRF to BTR
The back tension ratio BTR is DBRF divided by the product of the original thickness times the
plane strain yield strength. Plane strain yield strength is used because deformation in straight
drawbeads is in plane strain rather than in the tensile deformation mode. Comparing the DBRF
stress (DBRF/original thickness times original width) with the plane strain yield strength more
clearly indicates the likelihood of plastic deformation.
While for certain test conditions, the outer drawbead produces thinning, steel which passes
through the outer drawbead does not enter the inner drawbead. Thus, the original thickness is an
appropriate quantity to convert DBRF to drawbead restraining stress.
Tensile yield strength is converted to plane strain yield strength using Hill 48 with planar
isotropy and normal anisotropy as shown in Equation 2.3.1:
EQ 2.3.1 [ (
YS PS = A / A 2 − 1 )0.5
]
YS Tens
Where YS PS is the plane strain yield strength
YS Tens is the average tensile yield strength
A = (R + 1) R
R is the average plastic strain ratio
The average tensile yield strength and plastic strain ratio are computed using the expression
(L+2D+T)/4. This average is used because plane strain is a biaxial deformation process where it
is not appropriate to use a specific directional tensile test. It should also be noted that in bending
there is a through thickness strain gradient.
To determine the variation in BTR for a given variation in DBRF, Equation 2.3.2 can be
rewritten
The variability in BTR can be calculated from the variability in DBRF starting with Equation
2.3.3. The result is shown in Equations 2.3.4 and 2.3.5.
EQ 2.3.4 d(BTR)/d(DBRF) = C
36
EQ 2.3.5 ∆(BTR) = C ∆DBRF
Where C = (YSPS to Wo)–1
It can be seen from Equation 2.3.5 that the variation in BTR depends on the magnitude of the
term, YSPS to Wo as well as the variation in DBRF itself.
For any lot of steel, the initial width and thickness can be taken as a constant. By substituting in
Equation 2.3.5, the variation in BTR can be calculated for each lot of steel. The result is shown
in Table 2.3.1. It can be seen from Table 2.3.1 that the standard deviation for BTR ranges from
± 0.0029 to ± 0.0091.
Table 2.3.1
Calculated Values for Variation in BTR Based on the Average Variation in DBRF
Std.
to Plane Strain Yield Strength Std. Deviation
Steel Deviation Wo (mm)
(mm) (MPa) BTR
DBRF n
AKDQ
0.7 CR 382 0.716 254 229.6 0.0091
0.7 EG 382 0.734 254 237.1 0.0086
1.2 CR 382 1.198 254 227.4 0.0055
1.2 EG 382 1.223 254 227.6 0.0054
BH210
0.7 EG 382 0.732 254 321.1 0.0064
0.8 EG 382 0.780 254 345.5 0.0054
HSLA
0.8 HDG 382 0.787 254 484.4 0.0039
1.2 CR 382 1.168 254 482.4 0.0029
DP600
1.2 HDG 382 1.180 229 491.3 0.0029
Strain measurements were made at three regions of the channel draw sidewall. These regions are
shown in Figure 2.4.1. Region A is the area of uniform deformation and corresponds to the
location of samples taken for determining prestrained FLCs. Most measurements were done in
this region. Regions B and C are respectively the area between the bead and the die impact line
and the area between the die impact line and the top of the sidewall.
37
Area C
Area B
Area A
Figure 2.4.1
Identification of Measurement Locations Used to Determine Net Thickness Strain and
Major and Minor In-Plane Strains in the Channel Draw Test Piece Sidewall
For each test condition, three gridded channel pieces were selected for strain measurement.
Strains were measured on both sides of the channel draw piece, but only the data for the right
hand side, which corresponds to the location of the FLC tests is reported. In Region A, thickness
measurements were made with both a digital micrometer and an ultrasonic thickness tester. In
Regions B and C, only an ultrasonic thickness tester was used. In all cases, the original thickness
of the blank is used to calculate thickness strain. The in-plane major and minor strains were
calculated using a steel rule over an original gauge length of 101.6 mm determined from the
initial grid spacing.
38
Area C
Die impact line
Area B
Bead impact line
Area A
ε1
ε2
50.8 mm
25.4 mm
~ 68 mm 25.4 mm
254 mm
Figure 2.4.2
Strain Measurement Positions from Channel Draw Sidewall
The pattern for thickness testing in Regions A, B and C is shown for the 200 to 400 series in
Figure 2.4.1 and for the 500 series in Figure 2.4.2. For Region A, thickness measurements were
taken at 20 locations; specifically every 50.8 mm from left to right and top from bottom. As
shown in Figures 2.4.1 and 2.4.2, the average thickness strain is determined from the 6 points in
the center of the sidewall. The major and minor strains were also taken from this region. The
complete data set is available on a pair of CD-ROMs which includes all the experimental data
from this study.
The year-to-year variability for net thinning strain is shown in Table 2.4.1. These numbers
correspond to the year the work was done.
39
Table 2.4.1
Comparison of Net Thickness Strain Measurements in Successive Years
for Comparable Test Conditions
Comparison 1st Year 2nd Year 1st Year – 2nd Year
200 Series (Year 1) vs 300 Series (Year 2)
223 vs 307 0.097 0.109 – 0.012
200 Series (Year 1) vs 400 Series (Year 3)
203 vs 435 0.251 0.260 – 0.009
219 vs 437 0.269 0.260 + 0.009
223 vs 438 0.097 0.105 – 0.008
224 vs 439 0.128 0.145 – 0.017
225 vs 440 0.188 0.212 – 0.024
200 Series (Year 1) vs 500 Series (Year 4)
223 vs 501 0.097 0.122 –0.025
300 Series (Year 2) vs 400 Series (Year 3)
303 vs 441 0.132 0.150 – 0.018
305 vs 442 0.061 0.082 – 0.021
307 vs 438 0.109 0.105 + 0.004
311 vs 431 0.179 0.176 + 0.003
400 Series (Year 3) vs 500 Series (Year 4)
438 vs 501 0.105 0.122 – 0.017
Average – 0.0113
Standard Deviation ±0.0114
Range – 0.024/+0.004
Number + Values 3
Number – Values 9
Thus, years 1, 2, 3, and 4 of the test program correspond respectively to the 200, 300, 400, and
500 test series. It can be seen from Table 2.4.1 that thickness strain measurements increase with
time. Since the data are limited, and the pattern with time is inconsistent, no attempt is made to
correct the data for a time effect. As shown in Table 2.4.1, the resulting standard deviation is ±
0.0114. Given the relatively small sample size for repeat conditions, ± 0.0114 may be an
underestimate of the population distribution, particularly since there are no repeat conditions for
a number of drawbead geometries and conditions.
The standard deviation of the 18 thickness measurements for each test condition represents
measurement error and non-uniformity of deformation. Since the critical local neck is position
specific, both sources of variation must be considered. For the 86 test conditions that were
evaluated, the average standard deviation is ± 0.0055 and the standard deviation of the standard
deviation is ±0.0040. The cumulative frequency distribution for the raw data is shown in Figure
2.4.3. It can be seen from Figure 2.4.3 that standard deviations are generally small with
occasional cases where the standard deviation is large.
40
30
25
20
Frequency
15
10
0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02 0.022 0.024 0.026 More
Std. Dev.
Figure 2.4.3
Frequency Distribution for the Standard Deviation in Measuring Net Thickness Strain
in the Channel Draw Test Piece Sidewall
41
It can be seen from Table 2.4.1 that the year-to-year differences in thickness strain are
consistently greater than the expected standard deviations shown in Figure 2.4.3. Thus, it seems
likely that the year-to-year differences relate to differences in test set-up for ostensibly similar
test conditions, rather than differences in measuring thickness. However, the ultimate purpose for
determining reproducibility of the thickness strain is to evaluate the validity of predictive
equations. Consequently, it is appropriate to combine variability in test set-up with variability in
measuring thickness strain.
Combining standard deviations can be done by the rule of sum of squares as shown in Equation
2.4.1.
σ = (σ T2 + σ M2 )
0.5
EQ 2.4.1
Where σ T is the standard deviation from the year-to-year variation
σ M is the standard deviation from variation in thickness strain measurement for a
given test condition
σ is the combined standard deviation
The resulting combined standard deviation for net thinning strain is ± 0.0127.
2.4.2 Drawbeads
A digital point micrometer and an ultrasonic thickness tester were used to measure thickness
strains on the sample at the entry and exit of the drawbeads. Digital point micrometer readings
were also taken at positions between radii in the drawbead. Since the ultrasonic thickness
measurements taken at the entry to the drawbead do not always agree with the as-received strip
thickness used to measure thickness strains in the sidewall, the as-received thickness is used to
calculate strain through the inboard drawbead. This procedure is justified on the following basis;
(1) material which has passed through the outer drawbead does not enter the inner drawbead, and
(2) the measurements of as-received thickness are more reliable than the more limited ultrasonic
thickness measurements at the entry of the drawbead, and (3) this approach guarantees a self
consistent series of strain measurements at the exit of the inner drawbead and the channel draw
sidewall.
With regard to measurement frequency, one channel was evaluated for the 200, 300, and 400 test
series, and 3 channels were evaluated for the 500 test series. Thus, it should be recognized that as
a consequence of the limited testing, the drawbead thickness strain data are less reliable than the
thickness strain data at position A on the channel draw sidewall. Thus, no quantitative analysis of
variability in drawbead thickness strain data is included in this report.
42
2.5 Determining Bending Strain
Depending on the size of the bend and the light intensity, the total depth of field is 6 to 10 mm.
Control of focus and light intensity is critical because excessive light can result in an artificial
thickening of the sheet which can compromise radius measurements. This control requires
adjusting the zoom lens and the lighting for each test condition. Typically, the distance between
the sheet edge and the end of the bead is about 3 mm.
For the 200, 300 and 400 test series, rollover from shearing was minimized by stoning to provide
a crisper edge condition. For the 500 test series, edges were machined, and the lubricant was
removed for about one inch from each strip edge to prevent lubricant from blurring the image.
Thus, crisper images were obtained for the 500 test series.
A series of bead wrap images were recorded; two from channel draw samples and one from a
draw bead strip pulled through both the outer and inner draw beads. In performing the tests, one
person operated the press while an operator at the computer recorded the images.
The bead wrap images were recorded at a rate of 2-3 images/second and a series of images
generally included the following stages:
• Initial setting of the bead prior to the part being formed (no material movement)
• Forming of the part (approximately 300 mm of material pulled through the bead)
• Opening of the die after the part was formed
3
D. E. Green, (2001); “An experimental technique to determine the behaviour of sheet metal in a drawbead”, S.A.E.
Technical Paper 2001-01-1136
43
In order to determine the bending radii of the sheet, one image from each series was selected. An
image was considered suitable for analysis if it was taken during the steady state operation and
before lubricant had time to build up and conceal the edges of the sheet. The complete set of
bead wrap image data is available on a CD-ROM from the Auto/Steel Partnership.
To calculate bending strain, the following image analysis system was used:
• A CAD drawing of each draw bead configuration was created in AutoCAD with the specified
bead dimensions and clearances. Each CAD drawing was created with the actual kiss gap and
the bead penetration that was measured during the tests.
• An image selected from a given test series was imported into AutoCAD. The image was
overlaid on the CAD drawing, then scaled and translated until the bead wrap image coincided
with the CAD drawing of the drawbead. It should be noted that the image was always scaled
by the same amount in both directions.
• Once the image was appropriately scaled and positioned on the CAD drawing, an arc
segment was fitted by eye with the AutoCAD drawing tools at each bend and in the area of
minimum radius. The radius of each arc segment was then determined with the AutoCAD
dimensioning tools and recorded.
• The minimum bend radii were recorded for each of the three parts formed. The average radii
in each bend were then calculated for the three images that were analyzed.
The sample size for measuring radii is three. For a small sample size, the parameter
range/average is a useful measure of test accuracy. The distribution of this parameter can be
approximated by a log normal distribution (Figure 2.5.2). The resulting data for Radius 1, Radius
2, Radius 4, and all three radii combined are shown in Table 2.5.1, where it can be seen that most
values are reasonably small.
44
Figure 2.5.2
Cumulative Frequency Distribution for all the Data for the
Parameter Range/Average for Drawbead Radius Measurements
Table 2.5.1
Distribution of the Parameter, Range/Average for Radius Measurements
Condition N Average Std Deviation Range
Radius 1 74 0.0912 0.0642 0.0115/0.311
Radius 2 74 0.0704 0.0635 0.0000/0.428
Radius 4 74 0.0943 0.0674 0.0081/0.484
Radii 1, 2, and 4 222 0.0853 0.0659 0.0000/0.484
R 1 t
EQ 2.5.1 ε = ln o = 1n 1 +
b R 2 R
n i
Where R o = R i + t
t is the strip thickness
45
R n is the radius at the neutral axis
A possible shift of the neutral axis is accounted for in Equation 1 according to the expression:
EQ 2.5.2 R n = R iR o
Every bend-unbend cycle contributes to the overall work hardening of a material as it is pulled
through a drawbead. Assuming that the cyclic work hardening of the strip during steady state
deformation can be expressed as the sum of the bending strains in each cycle, a cumulative
bending strain factor is calculated which is shown in Equation 2.5.3:
EQ 2.5.3 ∑ε( b )= εb 1 + εb 2
+ ε − ε
b 3 b
+ε
unbend
b 4
Where the bending strains are numbered from entry to exit, and the term
ε − ε
where it exists, represents the bend on the exit side of a
b 3 b unbend
male square drawbead corrected for any partial unbending of the strip. If the
penetration is insufficient, and the strip merely unbends after initial contact with
the male bead, this term equals zero.
For double draw beads, Equation 2.5.3 is extended to include both unbending between the two
draw beads and the radii in the second draw bead.
( )
∑ ε b is an important variable for predicting the enhanced FLC effect and is described as the
bending strain factor (BSF).
One approach to estimating the accuracy of the bending strain factor is to compare results from
repeat test conditions. Table 2.5.2 is such a comparison where BSF from a later test is subtracted
from BSF from an earlier test. Eight such comparisons for aluminum killed steels are available.
These comparisons primarily represent conditions with a steel thickness of 0.7 mm in a single
round drawbead. It can be seen from Table 2.5.2 that in contrast to net thinning strain there is no
consistent variation in BSF with time, and that the standard deviation of the difference
population for BSF is ± 0.0061.
46
Table 2.5.2
Analysis of the Variability in Bending Strain Factor (BSF) Using Repeat Conditions
Steel DB Type Comparison 1st Value 2nd Value Difference
0.7 EG Round 203 vs 435 0.459 0.468 –0.009
0.7 EG Round 224 vs 439 0.335 0.329 +0.006
0.7 EG Round 219 vs 437 0.343 0.345 –0.002
0.7 EG Round 223* vs 501 0.262 0.273 –0.011
0.7 CR Round 303 vs 441 0.245 0.244 +0.001
0.7 CR Square 305 vs 442 0.178 0.184 –0.006
1.2 CR Round 225 vs 440 0.377 0.374 +0.003
1.2 CR Round 311 vs 431 0.268 0.266 +0.002
Average -- -- –0.002
Standard Deviation -- -- ±0.0061
Range –0.011/+0.006
Number of + Values 4
Number of – Values 4
* Conditions 307 and 403 were not measured separately
In extending the results from Table 2.5.2, the following seems reasonable:
• The results for aluminum killed steels should apply to the other steel grades because the
properties of the steel should not affect the measurement of radius
• With regard to drawbead type, there is a progressive increase in number of radii from
round (3) to square (4) to double round (6) to double square (8). If the variation in
measuring bending radius is consistent, an increasing number of bends increases the
variation in determining BSF. The extent of consistent versus random variation in
measuring BSF is not known
• With regard to drawbead penetration, qualitative examination of the data indicates
variability is greater for shallower penetrations
• In summary, the standard deviation for measuring BSF is greater, by some indeterminate
amount than the ± 0.0061 calculated in Table 2.5.2.
As an alternative method for estimating the bending strain factor, the effect of variation in
measuring radius can be assessed. For purpose of discussion, the standard deviation for the
parameter, range/average, is taken as 0.09. It can be seen that the effect of the parameter,
range/average, depends on the radius. As examples, for 4 and 10 mm radii, the comparable
standard deviations would be respectively ±0.36 and ±0.9 mm. The effect of these standard
deviations on bending strain can be calculated using Equation 2.5.1, and the results are shown in
Table 2.5.3. It can be seen from Table 2.5.3 that the +/- two sigma range in maximum bending
strain for a single radius is 0.012 to 0.043. If the measurement error is consistent from radius to
radius, for a round drawbead, +/- two sigma would range from 0.036 to 0.129, and for square or
double drawbeads, the range in two sigma would be even greater.
47
Table 2.5.3
Effect of a Two Standard Deviation Variation in Radius Measurement on
the Maximum Bending Strain
Maximum Minimum
Radius Thickness ± Two Sigma
Bending Bending
(mm) (mm) Range
Strain Strain
4 0.7 0.097 0.069 0.028
1.2 0.156 0.113 0.043
10 0.7 0.041 0.029 0.012
1.2 0.068 0.048 0.020
While it is likely that the variation in measuring radii is more random than consistent, it can be
seen that the population standard deviation estimated from the repeat conditions (Table 2.5.2) is
too small. Since the proportion of random to consistent variation in measuring radius is not
known, further quantitative calculation is not productive, and the standard deviation associated
with measuring BSF must be determined on an arbitrary basis. Therefore, it seems reasonable to
double the standard deviation from the difference table, and the population standard deviation
associated with measuring BSF is assumed to be ± 0.0122. Given the nature of the variation in
measuring radius, it also seems likely that occasionally there is greater error in measuring BSF.
FLCs were determined using the Marciniak Double Blank method in a hydraulic press at a
constant punch velocity. A schematic of the Marciniak Double Blank test is shown in
Figure 2.6.1. The die set includes a binder and a 100 mm diameter flat bottom punch with a 6.4
mm punch profile radius. The working surface of the punch is hardened and non-directionally
polished with a 1200 grit polishing compound. Clearance between the binder and the punch is
5.0 mm per side. The binder has a lock bead and a 12.7 mm die entry radius.
Figure 2.6.1
Schematic of Die Used for the Marciniak Double Blank Test
The Marciniak Double Blank method uses a carrier blank which is placed under the test piece.
The carrier blank has a centrally located hole. The test piece is placed on top of the carrier blank
48
and the two pieces are securely clamped in the die. As stretching proceeds, the hole in the carrier
blank increases in diameter allowing it to more easily slide over the punch profile radius. The
friction between the carrier blank and the test piece pulls the test piece over the punch radius
with the deformation centered over the hole in the carrier blank. The extent of deformation is
controlled by punch travel, and the blank width determines minor strain. Blank length is
approximately 200 mm and blank widths ranging from 100 to 130 mm were used for evaluating
the left side of the FLC. Two or three sample widths were used in determining the left side of the
FLC. In determining the right side of the FLC, blank widths of up to 200 mm were used.
All test samples were electrochemically etched prior to forming the channel draw pieces with a
grid of 2.54 mm diameter circles as described previously. Undeformed grids are periodically
measured 3 times by an operator. Using this quality control process, it was found that the as-
applied circle diameter is 2.54 ± 0.076 mm.
The carrier blanks that were used with each steel are shown in Table 2.6.1. Holes were machined
to minimize edge damage and avoid splitting before the end of a test. Carrier blank hole
diameters are 50.8 mm for test pieces with widths between 100 and 130 mm. For larger widths,
the carrier blank hole diameter is 34 mm.
Table 2.6.1
Material Used for Carrier Blanks in Marciniak Double Blank Tests
Carrier Blank Properties
Test
Thickness Yield Strength Tensile Strength Total n-
Material Material
(mm) (MPa (MPa) Elongation Value
AKDQ
0.7 EG IF 0.787 177 315 43.1 0.216
0.7 CR IF 0.749 107 298 50.4 0.271
1.2 CR IF 1.23 181 314 41.6 0.205
BH210
0.7 EG IF 0.787 177 315 43.1 0.216
0.8 EG IF 0.749 107 298 50.4 0.271
HSLA
0.8 HDG IF 0.749 107 298 50.4 0.271
1.2 CR IF 1.53 137 309 51.1 0.243
DP600
1.2 HDG IF 1.53 137 309 51.1 0.243
In running tests, lubrication is applied to the surface of the carrier blank in contact with the
punch. The carrier blank, test piece combination is aligned and clamped in the die with sufficient
binder force so that it is locked on the binder.
Tests were terminated at a punch height that corresponded to the onset of local necking. Punch
height is controllable to a resolution of 0.03 mm. To provide sufficient samples for evaluation,
7 or 8 test pieces were produced at each blank width.
49
The Marciniak Double Blank tests were carried out for each of the as-received materials and also
for the material taken from the sidewalls of channel sections prestrained under various
conditions. These tests were carried out at the following facilities:
• For the 0.7 mm material in the 200 and 300 test series, the University of Waterloo using the
200 ton hydraulic press that was designed by Prof. John Schey for formability testing.
• For the 1.2 mm material in the 200 and 300 test series, Dofasco’s R&D Centre using an MTS
formability press capable of forming thicker gauge materials.
• For the 400 and 500 test series, IRDI’s formability press
In evaluating test pieces, a light oil film was applied to the surface to enhance the appearance of
surface features. Identification of local necks was done visually. The first observable stage in
neck development is surface roughening. Such roughening is not necking. As deformation
proceeds, a regular pattern of very shallow, narrow, long depressions form in the central region
of samples. These depressions are a field of incipient necks. Samples with this condition were
evaluated to determine critical local necks. These critical local necks are the failure criteria in
this study. Test pieces deformed beyond this point were not used to determine FLCs.
Neck orientation was observed to depend on minor strain (width of test piece). Specific
observations follow:
• Width 100 to 110 mm: A crosshatched pattern with necks oriented 50° to 60° to the direction
of loading (Figure 2.6.2)
• Width 110 to 125 mm: A less apparent crosshatched pattern with necks oriented
approximately 80° to the direction of loading (Figure 2.6.3)
• Width 125 to 130 mm (close to plane strain): Usually a single neck forms oriented at 90° to
the direction of loading (Figure 2.6.4)
• Width greater than 130 mm: A uniform pattern across the entire flat surface that is oriented
90° to the direction of loading (Figure 2.6.5)
50
Figure 2.6.2 Pattern of Incipient Necks at a Sample Width of 100 to 110 mm
Figure 2.6.3 Pattern of Incipient Necks at a Sample Width of 110 to 125 mm
Figure 2.6.4 Pattern of Incipient Necks at a Sample Width of 125 to 130 mm
Figure 2.6.5 Pattern of Incipient Necks at a Sample Width of more than 130 mm
In addition to the patterns that have been described, for the as-
received 0.8 mm hot dip galvanized HSLA, as a sample was
progressively deformed in the Marciniak double blank test,
one or more long necks would form parallel to the width of a
sample (i.e., perpendicular to the major strain axis). The necks
having this non-typical appearance were designated as a
secondary type of neck. At the same time as these secondary
necks were formed, the surface of the sample would dimple,
then the dimples would connect and form parallel necks at an
orientation that depended on the width of the sample.
Therefore, as shown in Figure 2.6.6, two types of necks are
observed. It should be noted that this material exhibited a
yield point elongation of about 6%, though this may be
coincidence rather than cause and effect.
After samples were removed from the press, deformed electro-etched circles were determined to
be safe or necked and identified inside the circle with a light pencil mark. In most cases, circles
51
lie across a critical local neck. However, in some cases, particularly for narrow samples,
deformed circles do not lie across a critical local neck, but were within a field of necks. Such
deformed circles were classified separately.
Strain measurement on identified deformed circles was done by placing a test piece on an X-Y
translating table with adjustable Mitutoyo Digital Verniers that have an accuracy of ± 0.001 mm.
This apparatus is shown in Figure 2.6.7. A Leica MZ8 optical microscope is focused on the X-Y
translating table. A digital camera with its horizontal and vertical axes parallel to the axes of the
X-Y table is used to produce an image. This image is displayed on an 1270 mm television
monitor which produces a magnification of about 140X. This strain measuring system is shown
in Figure 2.6.8.
Deformed circles are measured using perpendicular cross hairs that intersect at the center of the
television monitor. Each axis of the deformed circle (ellipse) is measured sequentially by
visually aligning the cross hairs on the monitor with the center of the grid line on one side of the
ellipse and then displacing the X-Y translating table with the verniers until the cross hairs are
aligned with the center of the grid line on the other side of the ellipse.
For each sample width (strain path), 33 to 67 deformed circles were measured which provides
sufficient data to produce statistically valid FLCs. In plotting FLCs, the left side is assumed to be
a line of constant thickness. A typical FLC is shown in Figure 2.6.9. Since all necked data points
must lie above the FLC, accuracy in determining individual data points is important to accurately
determining FLCs. This is particularly important when sample sizes are large because the more
data points, the more likely it is to obtain a low value of minor and/or major strain. Thus, the
repeatability and reproducibility of the strain measurement system has an effect on the accuracy
of FLC results.
52
90
80
70
60
40
30
0
-30 -20 -10 0 10 20 30
Minor Strain (%)
Figure 2.6.9
Typical Forming Limit Curve
The repeatability and reproducibility of the strain measuring system was evaluated using
procedures outlined in the automotive SPC reference manual. Specifically, three operators (one
of which had never used the measuring system prior to this study) measured major and minor
strains of five circles, three different times at 15 minute intervals. The results follow:
According to the SPC reference manual, errors of less than 10% indicate a measuring system is
in control. Thus, with a 4.43% error, the strain measuring system used for FLC work is in
excellent control.
53
The raw data from the repeatability and reproducibility study can also be used to directly
determine the standard deviation associated with strain measurement. The average value of the
major and minor strains from this study range from –20.9 to +65.7%. The average and standard
deviation for the nine measurements for each average strain are shown in Table 2.6.2. Normally,
errors in measuring strain are thought to be additive, but for measuring deformed circles, the
effect of deformation on line width must be considered. It can be seen from Table 2.6.2 that for
positive strains, where there is line broadening, the average standard deviation is 1.29 strain
percent. In contrast, for compressive strains, the comparable standard deviation is 0.37 strain
percent.
Table 2.6.2
Statistical Evaluation of Raw Data from R and R Study on Strain Measurements for
Determining Forming Limit Curves
Condition Average Standard Deviation Range
1 –20.87 0.38 –21.40/–20.30
2 –19.56 0.38 –20.40/–19.20
3 –18.93 0.61 –19.90/–18.10
4 –15.72 0.17 –15.90/–15.35
5 –12.08 0.29 –12.50/–11.60
6 +29.97 0.76 28.70/31.40
7 +42.85 2.14 39.60/45.45
8 +60.18 1.52 57.25/61.70
9 +60.59 1.00 59.20/62.65
10 +65.74 1.05 64.15/67.30
In determining the effect of variation in strain measurement on experimental FLCo, both the
major and minor strains must be considered. This requires knowledge of the slope of the left side
of the FLC. If it is assumed that the slope of the left side of the FLC is a straight line in true
strain, the following analysis is possible.
It can be seen from Equation 2.6.3 that the slope of the left side of the FLC depends on both the
material (FLCo) and the minor strain. FLCo for any test condition depends on both the as-
received FLCo and the magnitude of the enhanced FLC effect. For the steels and test conditions
in this study, FLCo in true strain, ranges from 0.18 to 0.47 (20 to 60% engineering strain).
54
For illustration, values of the slope can be calculated for FLCs of 20, 40 and 60% and minor
strains of 5, 10, and 20%. The results range from -1.26 to -1.92. In combining the effect of major
and minor strain, two assumptions can be made;(1) the variation in both strains are systematic (ie
they are both high or both low) or (2) the variations are random. If the measurement errors are
totally systematic, they are additive, and the estimated population standard deviations range from
+/- 1.76 to 2.00 strain percent. In contrast, if the errors are totally random, the combined
estimated population standard deviation can be calculated from the square root of the sum of
squares of the standard deviations for the major and minor strains. The resulting range is +/- 1.56
to 1.74 strain percent. In the absence of specific data, it is assumed that there is a 50-50 mixture
of systematic and random variation. For this assumption, the estimated population standard
deviation for FLCo ranges from 1.66 to 1.87 strain percent.
Another approach to estimating the reproducibility of FLCo is to compare FLCo values for
similar test conditions in the 300, 400, and 500 test series as shown below:
In examining the differences in FLCo for repeat conditions, it can be seen that the maximum
difference is -2.4 strain percent, and the range is from -2.4 to +1.0 strain percent. For a sample
size of 4, these differences are consistent with an estimated population standard deviation
ranging from 1.66 to 1.87 strain percent which at two sigma translates to a range of 3.32
to 3.74 strain percent.
Since ∆FLCo is the actual parameter of interest, variation in both the prestrained and as-received
FLCs must be considered. Given that the condition of the grids for the as-received tests is much
clearer than for the prestrained grid, a combined standard deviation of half that for prestrained
FLCs is reasonable. It is also reasonable that the measurement of the as-received and prestrained
FLCs is random.
Selection of a standard deviation for as-received FLCo’s is based on the average of 50% of the
values for the standard deviations for a prestrained FLC with FLCo equal 40%, minor strains of
–10 and –20%, and considering both systematic and random deviation between major and minor
strain. The result is a combined standard deviation of 0.65%.
The resulting estimated combined standard deviation for ∆FLCo can be computed from the sum
of squares rule as shown in Equation 2.6.4.
EQ 2.6.4 σ 2 = σ AR
2
+ σ P2
Where σ AR represents the as-received FLC and,
σ P represents the prestrained FLC
55
The result is ± 1.63%, which is thought to be a reasonable approximation of measurement error.
Thus, a two sigma range for ∆FLCo is ±6.52 strain percent.
Three test conditions (218, 219, 223) that were used in Year 1 were repeated one or more times
in Years 2 through 4. The results for the increase in FLCo are shown below:
Using the foregoing analysis a correction of +1.6% strain percent is applied to all the values of
FLCo that were determined in the 200 test series. In this regard, no correction is made to any
values of as-received FLCo because there is less difficulty in identifying critical local necks so
that a correction is not needed.
Tensile tests were taken from the same sidewall as the test pieces for determining FLCs. Test
pieces were taken from the edge of the sidewall. The edges of the sidewall do not deform in
plane strain as there is a moderate negative minor strain.
Tensile testing followed ASTM A370 (E8) using samples with a 50.8 mm gauge length and a
63.5 mm reduced section at a constant crosshead speed of 12.7 mm/min. For the 200 to 400 test
series, samples were tested in triplicate. The orientation of these samples is parallel to the
direction of metal movement into the die cavity which corresponds to the transverse direction of
the original coil stock. For the 500 test series, duplicate tests were taken transverse and parallel
to the rolling direction. However, it should be noted that for the 1.2 HDG DP600, the length of
the strips used to produce channel draw test pieces is in the rolling direction of the original coil
stock. This change was made because it was the only way to obtain the needed sample length.
56
3.0 Results
Table 3.1 shows selected set-up conditions, the summary causative variables, and the
performance factors for the steels evaluated in this project. Note that for convenience, all
thinning strains are shown as positive values. This table is organized by increasing thickness for
each steel grade. Secondary breakdowns are by bead type, and decreasing bending strain factor.
The experimental value for prestrained FLCo’s are also included in Table 3.1. Additional set-up
data for the outboard and inboard drawbeads are shown respectively in Tables 3.2 and 3.3. The
individual as-received and prestrained FLCs are shown in Appendix 3.1.
57
Table 3.1
Summary of Test Conditions and Results
58
Table 3.1 (Continued)
Summary of Test Conditions and Results
59
Table 3.1 (Continued)
Summary of Test Conditions and Results
60
Table 3.2
Details of Test Set-up for the Outboard Drawbead
Outboard
Female Kiss Block Outboard Clearance per
Prestrain Steel Grade Male Bead Variable Bead
Insert Gap Bead Radii Side
Condition Identifier Material or Fixed Penetration
Material
(mm) (mm) (mm) (x t0) (%) (mm)
432 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.394 4 1.4 1.96 31.6 2.76
308 AKDQ-CR 0.7 Cast Cut Cast Cut Fixed 0.381
422 AKDQ-CR 0.7 Cast Cut Cast Cut Fixed 0.364
302 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.381 4 1.4 1.96 22.4 1.95
423 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.384 4 1.4 1.96 22.7 1.98
303 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.381 4 1.4 1.96 22.6 1.97
427 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.354 4 1.4 1.96 45.7 3.98
426 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.354 4 1.4 1.96 74.6 6.51
434 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.374 4 1.4 1.96 48.9 4.26
433 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.384 4 1.4 1.96 57.0 4.97
301 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.381 4 1.4 1.96 71.2 6.21
425 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.354 4 1.4 1.96 87.9 7.66
401 AKDQ-CR 0.7 Cast Cut Cast Cut Fixed 0.349
402 AKDQ-CR 0.7 Cast Cut Cast Cut Fixed 0.349
403 AKDQ-CR 0.7 Cast Cut Cast Cut Fixed 0.359
404 AKDQ-CR 0.7 Cast Cut Cast Cut Fixed 0.349
411 AKDQ-CR 0.7 Cast Cut Cast Cut Fixed 0.369
412 AKDQ-CR 0.7 Cast Cut Cast Cut Fixed 0.369
405 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.364 4 1.4 1.96 45.6 3.98
406 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.364 4 1.4 1.96 45.6 3.98
306 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.381 4 1.4 1.96 22.2 1.94
305 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.381 4 1.4 1.96 22.4 1.95
418 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.374 4 1.4 1.96 94.9 8.27
417 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.364 4 1.4 1.96 58.3 5.08
304 AKDQ-CR 0.7 Cast Cut Nitrided D2 Adjustable 0.381 4 1.4 1.96 71.2 6.21
226 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381
218 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 4 1.4 1.91 86.1 7.52
436 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 4 1.4 1.91 86.1 7.52
216 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 3 1.4 1.91 83.0 5.59
217 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 2 1.4 1.91 76.0 3.60
212 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 4 1.4 1.91 86.1 7.52
211 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 3 1.4 1.91 83.0 5.59
209 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 2 1.4 1.91 76.0 3.60
61
Table 3.2 (Continued)
Details of Test Set-up for the Outboard Drawbead
Outboard
Female Kiss Block Outboard Clearance per
Prestrain Steel Grade Male Bead Variable Bead
Insert Gap Bead Radii Side
Condition Identifier Material or Fixed Penetration
Material
(mm) (mm) (mm) (x t0) (%) (mm)
213 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 4 1.4 1.91 86.1 7.52
215 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 3 1.4 1.91 83.0 5.59
201 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 2 1.4 1.91 76.0 3.60
203 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381
224 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381
221 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381
220 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381
219 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 4 3.4 4.63 53.4 3.60
437 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381 4 3.4 4.63 53.4 3.60
222 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381
223 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381
307 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381
438 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.381
501 AKDQ-EG 0.7 Cast Cut Cast Cut Fixed 0.401
407 AKDQ-CR 1.2 Cast Cut Cast Cut Fixed 0.367
408 AKDQ-CR 1.2 Cast Cut Cast Cut Fixed 0.382
409 AKDQ-CR 1.2 Cast Cut Cast Cut Fixed 0.382
431 AKDQ-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.412 4 1.4 1.17 21.0 1.93
311 AKDQ-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.381 4 1.4 1.17 21.2 1.95
310 AKDQ-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.381 4 1.4 1.17 21.5 1.98
430 AKDQ-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.397 4 1.4 1.17 42.9 3.94
410 AKDQ-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.377 4 1.4 1.17 64.9 5.97
429 AKDQ-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.397 4 1.4 1.17 47.8 4.39
428 AKDQ-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.392 4 1.4 1.17 70.3 6.47
309 AKDQ-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.381 4 1.4 1.17 50.5 4.65
413 AKDQ-CR 1.2 Cast Cut Cast Cut Fixed 0.452
414 AKDQ-CR 1.2 Cast Cut Cast Cut Fixed 0.452
420 AKDQ-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.427 4 1.4 1.17 20.8 1.92
214 AKDQ-EG 1.2 Cast Cut Cast Cut Fixed 0.381 4 2.8 2.27 82.0 7.57
225 AKDQ-EG 1.2 Cast Cut Cast Cut Fixed 0.381
62
Table 3.2 (Continued)
Details of Test Set-up for the Outboard Drawbead
Outboard
Female Kiss Block Outboard Clearance per
Prestrain Steel Grade Male Bead Variable Bead
Insert Gap Bead Radii Side
Condition Identifier Material or Fixed Penetration
Material
(mm) (mm) (mm) (x t0) (%) (mm)
207 BH-210-EG 0.7 Cast Cut Cast Cut Fixed 0.381 2 1.4 1.91 76.1 3.60
206 BH-210-EG 0.7 Cast Cut Cast Cut Fixed 0.381
512 BH210-EG 0.8 Cast Cut Cast Cut Fixed 0.415
514 BH210-EG 0.8 Cast Cut Nitrided D2 Adjustable 0.415 4 1.4 1.79 41.0 3.60
516 BH210-EG 0.8 Cast Cut Nitrided D2 Adjustable 0.415 4 1.4 1.79 71.1 6.24
515 BH210-EG 0.8 Cast Cut Cast Cut Fixed 0.415 3 1.4 1.79 80.4 5.45
513 BH210-EG 0.8 Cast Cut Nitrided D2 Adjustable 0.415 4 1.4 1.79 56.2 4.93
518 DP600-HDG 1.2 Cast Cut Cast Cut Fixed 0.395
521 DP600-HDG 1.2 Cast Cut Nitrided D2 Adjustable 0.395 4 1.4 1.19 32.2 2.96
520 DP600-HDG 1.2 Cast Cut Nitrided D2 Adjustable 0.405 4 1.4 1.19 44.6 4.09
522 DP600-HDG 1.2 Cast Cut Nitrided D2 Adjustable 0.39 4 1.4 1.19 73.6 6.76
519 DP600-HDG 1.2 Cast Cut Cast Cut Fixed 0.395
523 DP600-HDG 1.2 Cast Cut Nitrided D2 Adjustable 0.395 4 1.4 1.19 53.9 4.95
517 DP600-HDG 1.2 Cast Cut Cast Cut Fixed 0.395
502 HSLA-HDG 0.8 Cast Cut Cast Cut Fixed 0.408
503 HSLA-HDG 0.8 Cast Cut Nitrided D2 Adjustable 0.408 4 1.4 1.78 44.9 3.94
505 HSLA-HDG 0.8 Cast Cut Nitrided D2 Adjustable 0.408 4 1.4 1.78 99.6 8.75
504 HSLA-HDG 0.8 Cast Cut Nitrided D2 Adjustable 0.408 4 1.4 1.78 22.1 1.94
415 HSLA-HDG 0.8 Cast Cut Cast Cut Fixed 0.453
416 HSLA-HDG 0.8 Cast Cut Cast Cut Fixed 0.453
421 HSLA-HDG 0.8 Cast Cut Nitrided D2 Adjustable 0.408 4 1.4 1.78 93.8 8.24
507 HSLA-CR 1.2 Cast Cut Cast Cut Fixed 0.407
509 HSLA-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.412 4 1.4 1.20 32.0 2.94
510 HSLA-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.407 4 1.4 1.20 64.7 5.93
508 HSLA-CR 1.2 Cast Cut Cast Cut Fixed 0.407
511 HSLA-CR 1.2 Cast Cut Nitrided D2 Adjustable 0.407 4 1.4 1.20 44.6 4.09
506 HSLA-CR 1.2 Cast Cut Cast Cut Fixed 0.407
63
Table 3.3
Details of Test Set-up for Inboard Drawbead
64
Table 3.3 (Continued)
Details of Test Set-up for Inboard Drawbead
65
Table 3.3 (Continued)
Details of Test Set-up for Inboard Drawbead
66
In doing empirical work, it is important to determine the dispersion of data in the range of
interest. Since BTR and BSF are the important causative variables, values of these parameters for
tests on AKDQ steel are shown in Figure 3.1. It can be seen from Figure 3.1 that for the 0.7 mm
steels, there is extensive data for BTR equal zero and a good dispersion of data for higher values
of BTR and BSF. The gap in the data for lower values of BTR and BSF was purposeful. The
Project Team de-emphasized this region of experimental space because higher values of BSF and
BTR result in higher values of ∆FLCo. The higher values of ∆FLCo are of more direct interest
and have more value in developing useful regression equations.
67
0.8 AKDQ (0.7mm)
AKDQ (1.2mm)
0.7
0.6
0.5
BSF
0.4
0.3
0.2
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
BTR
Figure 3.1
Experimental Combinations of BSF and BTR for all AKDQ Steels
68
With regard to the 1.2 mm AKDQ steel, it can be seen from Figure 3.1 that the more limited data
for the 1.2 mm exhibits a reasonable overlap with the results for the 0.7 mm steel. Thus, it is
meaningful to analyze the 0.7 and 1.2 mm thicknesses as a group.
Much of the experimental work on the BH210, HSLA, and DP600 steels was done in the latter
stages of the program after the effect of BTR and BSF on ∆FLCo, net thickness strain and
increase in yield strength for the AKDQ steels had been determined. Thus, for maximum
experimental efficiency, three cases were studied; BTR/BSF equals 0, 1 and 2. This approach is
designed to more efficiently examine the experimental space for BTR and BSF. For the same
results as explained for the AKDQ steels lower values of BTR and BSF were not evaluated. The
actual experimental conditions for these three steel grades are shown in Figure 3.2. It can be seen
from Figure 3.2 that the actual results closely follow the experimental plan.
69
0.7
BTR/BSR = 0 BTR/BSR = 1
0.6
0.5
0.4
BSF
BTR/BSR = 2
0.3
0.2
BH210 (0.7mm)
BH210 (0.8mm)
0.1 DP600 (1.2mm)
HLSA (0.8mm)
0.0 HSLA (1.2mm)
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
BTR
Figure 3.2
Experimental Combination of BSF and BTR for all BH210, USLA, and DP600 Steels
70
With regard to ∆FLCo, net thickness strain and increase in yield strength, it can be seen from
Table 3.1 that there is a wide range of change in these properties as a result of the changes in
BTR and BSF. For ∆FLCo, increases of as much as 25.8 strain percent are observed and for net
true thinning strain, values as low as –0.381 are observed. It can also be seen that there are
significant increases in yield strength after prestraining.
More detailed information on strains in channel draw test pieces are shown in Tables 3.4 to 3.6.
Table 3.4 shows true thickness strains in regions A, B, and C of the channel draw sidewalls. The
relationship between the true thickness strain in regions B and A is shown in Figure 3.3. It can be
seen from this figure that for a flat drawbead, the strains in regions A and B are virtually
identical. This is a reasonable result because in effect, there is no drawbead impact line because
there is no drawbead. For experimental conditions with drawbeads, it can be seen that (1) there is
a great deal of scatter in the data and (2) that the strain in region B increases with the strain in
region A up to a thickness strain of about 0.16. Thereafter within a wide scatter band, the strain
in region B remains constant while the strain in region A increases.
Table 3.4
True Thickness Strain Data from the Channel Draw Sidewall at Regions A, B and C
Inboard Inboard
Back Drawbead Bead True Thickness Strains
Prestrain Steel Grade Lubricant
Tension Radii Penetration
Condition Identifier Type
Ratio
(mm) (%) Area A Area B Area C
432 AKDQ-CR 0.7 Prelube 0.290 2 65.3 0.246 0.057 0.033
308 AKDQ-CR 0.7 Prelube 0.000 4 87.0 0.106 0.046 0.016
422 AKDQ-CR 0.7 Prelube 0.000 4 100.4 0.101 0.043 0.006
302 AKDQ-CR 0.7 Prelube 0.246 4 48.7 0.097 0.038 0.019
423 AKDQ-CR 0.7 Prelube 0.247 4 100.5 0.153 0.055 0.014
303 AKDQ-CR 0.7 Prelube 0.247 4 79.6 0.151 0.065 0.028
427 AKDQ-CR 0.7 Prelube 0.365 4 50.7 0.114 0.047 0.012
426 AKDQ-CR 0.7 Dry Film 0.367 4 50.7 0.106 0.043 0.004
434 AKDQ-CR 0.7 Prelube 0.384 4 86.7 0.200 0.074 0.034
433 AKDQ-CR 0.7 Prelube 0.430 4 86.7 0.215 0.083 0.041
301 AKDQ-CR 0.7 Prelube 0.514 4 49.3 0.173 0.070 0.030
425 AKDQ-CR 0.7 Prelube 0.601 4 50.7 0.203 0.072 0.033
401 AKDQ-CR 0.7 Dry Film 0.000 4 100.4 0.295 0.082 0.031
402 AKDQ-CR 0.7 Dry Film 0.000 4 74.9 0.243 0.058 0.020
403 AKDQ-CR 0.7 Dry Film 0.000 4 50.5 0.113 0.033 0.008
404 AKDQ-CR 0.7 Prelube 0.000 4 50.8 0.159 0.054 0.010
411 AKDQ-CR 0.7 Dry Film 0.000 4 100.1 0.072 0.031 0.007
412 AKDQ-CR 0.7 Prelube 0.000 4 100.1 0.105 0.063 0.027
405 AKDQ-CR 0.7 Dry Film 0.232 4 75.5 0.381 0.088 0.074
406 AKDQ-CR 0.7 Dry Film 0.232 4 50.7 0.199 0.046 0.017
306 AKDQ-CR 0.7 Prelube 0.245 4 80.5 0.135 0.065 0.022
305 AKDQ-CR 0.7 Prelube 0.246 4 49.7 0.081 0.041 0.014
418 AKDQ-CR 0.7 Dry Film 0.437 4 100.5 0.229 0.054 0.020
417 AKDQ-CR 0.7 Prelube 0.438 4 100.5 0.256 0.094 0.042
304 AKDQ-CR 0.7 Prelube 0.514 4 49.6 0.148 0.081 0.031
226 AKDQ-EG 0.7 Prelube 0.000 0.001 0.004 0.005
218 AKDQ-EG 0.7 Prelube 0.486 0.075 0.077 0.014
71
Table 3.4 (Continued)
True Thickness Strain Data from the Channel Draw Sidewall at Regions A, B and C
Inboard Inboard
Back Drawbead Bead True Thickness Strains
Prestrain Steel Grade Lubricant
Tension Radii Penetration
Condition Identifier Type
Ratio
(mm) (%) Area A Area B Area C
436 AKDQ-EG 0.7 Prelube 0.557 0.075 0.077 0.014
216 AKDQ-EG 0.7 Prelube 0.571 0.083 0.087 0.012
217 AKDQ-EG 0.7 Prelube 0.769 0.101 0.108 0.017
212 AKDQ-EG 0.7 Prelube 0.486 0.039 0.042 0.012
211 AKDQ-EG 0.7 Prelube 0.571 0.046 0.048 0.012
209 AKDQ-EG 0.7 Prelube 0.769 0.058 0.058 0.017
213 AKDQ-EG 0.7 Prelube 0.486 0.025 0.032 0.009
215 AKDQ-EG 0.7 Prelube 0.571 0.025 0.029 0.013
201 AKDQ-EG 0.7 Prelube 0.769 0.030 0.034 0.007
203 AKDQ-EG 0.7 Prelube 0.000 2 76.0 0.274 0.121 0.029
224 AKDQ-EG 0.7 Prelube 0.000 2 64.8 0.151 0.063 0.013
221 AKDQ-EG 0.7 Prelube 0.000 3 82.0 0.175 0.093 0.019
220 AKDQ-EG 0.7 Prelube 0.000 3 81.7 0.139 0.071 0.015
219 AKDQ-EG 0.7 Prelube 0.336 3 81.7 0.275 0.086 0.035
437 AKDQ-EG 0.7 Prelube 0.336 3 81.7 0.275 0.086 0.035
222 AKDQ-EG 0.7 Prelube 0.000 4 86.8 0.187 0.108 0.029
223 AKDQ-EG 0.7 Prelube 0.000 4 86.8 0.107 0.060 0.008
307 AKDQ-EG 0.7 Prelube 0.000 4 86.8 0.116 0.061 0.027
438 AKDQ-EG 0.7 Prelube 0.000 4 86.8 0.107 0.060 0.008
501 AKDQ-EG 0.7 Prelube 0.000 4 86.5 0.122 0.052 0.016
407 AKDQ-CR 1.2 Dry Film 0.000 4 71.2 0.366 0.106 0.042
408 AKDQ-CR 1.2 Dry Film 0.000 4 35.3 0.158 0.073 0.027
409 AKDQ-CR 1.2 Prelube 0.000 4 35.3 0.190 0.087 0.026
431 AKDQ-CR 1.2 Prelube 0.364 4 50.2 0.183 0.095 0.029
311 AKDQ-CR 1.2 Prelube 0.364 4 50.5 0.186 0.091 0.031
310 AKDQ-CR 1.2 Prelube 0.366 4 37.1 0.129 0.082 0.026
430 AKDQ-CR 1.2 Dry Film 0.374 4 50.5 0.171 0.079 0.020
410 AKDQ-CR 1.2 Dry Film 0.554 4 35.5 0.316 0.110 0.046
429 AKDQ-CR 1.2 Prelube 0.595 4 50.5 0.255 0.118 0.050
428 AKDQ-CR 1.2 Dry Film 0.598 4 50.6 0.234 0.106 0.032
309 AKDQ-CR 1.2 Prelube 0.626 4 36.6 0.197 0.093 0.038
413 AKDQ-CR 1.2 Dry Film 0.000 4 99.5 0.238 0.079 0.030
414 AKDQ-CR 1.2 Prelube 0.000 4 99.5 0.299 0.107 0.046
420 AKDQ-CR 1.2 Dry Film 0.239 4 99.0 0.329 0.104 0.037
214 AKDQ-EG 1.2 Prelube 0.644 0.041 0.044 0.002
225 AKDQ-EG 1.2 Prelube 0.000 4 82.2 0.215 0.139 0.030
207 BH-210-EG 0.7 Prelube 0.644 0.059 0.063 0.025
206 BH-210-EG 0.7 Prelube 0.000 2 76.1 0.260 0.111 0.019
512 BH210-EG 0.8 Prelube 0.000 3 80.2 0.177 0.070 0.006
514 BH210-EG 0.8 Prelube 0.259 4 57.0 0.167 0.064 0.008
516 BH210-EG 0.8 Prelube 0.399 4 34.2 0.130 0.048 0.004
515 BH210-EG 0.8 Prelube 0.515 4 51.3 0.268 0.068 0.017
513 BH210-EG 0.8 Prelube 0.330 4 97.7 0.309 0.058 0.017
518 DP600-HDG 1.2 Prelube 0.000 4 81.4 0.171 0.087 0.011
521 DP600-HDG 1.2 Prelube 0.262 4 48.1 0.147 0.061 0.006
520 DP600-HDG 1.2 Prelube 0.338 4 88.3 0.294 0.102 0.008
522 DP600-HDG 1.2 Prelube 0.533 4 50.3 0.265 0.073 0.010
519 DP600-HDG 1.2 Prelube 0.000 6 48.3 0.059 0.034 0.003
72
Table 3.4 (Continued)
True Thickness Strain Data from the Channel Draw Sidewall at Regions A, B and C
Inboard Inboard
Back Drawbead Bead True Thickness Strains
Prestrain Steel Grade Lubricant
Tension Radii Penetration
Condition Identifier Type
Ratio
(mm) (%) Area A Area B Area C
523 DP600-HDG 1.2 Prelube 0.398 6 39.7 0.154 0.071 0.008
517 DP600-HDG 1.2 Prelube 0.000 4 100.5 0.237 0.081 0.008
502 HSLA-HDG 0.8 Prelube 0.000 2 73.8 0.323 0.135 0.008
503 HSLA-HDG 0.8 Prelube 0.238 4 76.8 0.192 0.036 0.000
505 HSLA-HDG 0.8 Prelube 0.395 4 49.9 0.247 0.066 0.004
504 HSLA-HDG 0.8 Prelube 0.161 6 52.5 0.076 0.042 0.002
415 HSLA-HDG 0.8 Dry Film 0.000 4 98.3 0.071 0.022 0.000
416 HSLA-HDG 0.8 Prelube 0.000 4 98.3 0.107 0.025 0.000
421 HSLA-HDG 0.8 Dry Film 0.322 4 99.0 0.291 0.084 0.002
507 HSLA-CR 1.2 Prelube 0.000 4 81.6 0.185 0.103 0.001
509 HSLA-CR 1.2 Prelube 0.319 4 93.0 0.308 0.084 0.001
510 HSLA-CR 1.2 Prelube 0.534 4 51.9 0.281 0.099 0.010
508 HSLA-CR 1.2 Prelube 0.000 6 48.2 0.068 0.044 0.003
511 HSLA-CR 1.2 Prelube 0.395 6 39.6 0.152 0.076 0.000
506 HSLA-CR 1.2 Prelube 0.000 4 100.4 0.250 0.070 0.000
73
Table 3.5
Major, Minor and Thickness Strains for Region A of the Channel Draw Sidewall (True Strain)
Inboard Inboard
Drawbead Bead True Strain Strain True Hill 48
Prestrain Steel Grade Lubricant
BTR Radii Penetration Ratio Thickness Effective
Condition Identifier Type
E2/E1 Strain Strain
(mm) (%) Major Minor
432 AKDQ-CR 0.7 Prelube 0.290 2 65.3 0.257 -0.011 -0.043 0.246 0.322
308 AKDQ-CR 0.7 Prelube 4 87.0 0.118 -0.005 -0.043 0.106 0.148
422 AKDQ-CR 0.7 Prelube 4 100.4 0.119 -0.006 -0.051 0.101 0.148
302 AKDQ-CR 0.7 Prelube 0.246 4 48.7 0.098 -0.006 -0.061 0.097 0.121
423 AKDQ-CR 0.7 Prelube 0.247 4 100.5 0.180 -0.010 -0.056 0.153 0.224
303 AKDQ-CR 0.7 Prelube 0.247 4 79.6 0.152 -0.006 -0.040 0.151 0.191
427 AKDQ-CR 0.7 Prelube 0.365 4 50.7 0.135 -0.008 -0.059 0.114 0.168
426 AKDQ-CR 0.7 Dry Film 0.367 4 50.7 0.125 -0.009 -0.072 0.106 0.154
434 AKDQ-CR 0.7 Prelube 0.384 4 86.7 0.213 -0.011 -0.052 0.200 0.266
433 AKDQ-CR 0.7 Prelube 0.430 4 86.7 0.225 -0.012 -0.054 0.215 0.280
301 AKDQ-CR 0.7 Prelube 0.514 4 49.3 0.186 -0.010 -0.054 0.173 0.232
425 AKDQ-CR 0.7 Prelube 0.601 4 50.7 0.228 -0.013 -0.057 0.203 0.283
401 AKDQ-CR 0.7 Dry Film 4 100.4 0.317 -0.019 -0.061 0.295 0.393
402 AKDQ-CR 0.7 Dry Film 4 74.9 0.268 -0.015 -0.056 0.243 0.334
403 AKDQ-CR 0.7 Dry Film 4 50.5 0.138 -0.010 -0.073 0.113 0.170
404 AKDQ-CR 0.7 Prelube 4 50.8 0.194 -0.012 -0.062 0.159 0.240
411 AKDQ-CR 0.7 Dry Film 4 100.1 0.079 -0.005 -0.064 0.072 0.098
412 AKDQ-CR 0.7 Prelube 4 100.1 0.111 -0.006 -0.054 0.105 0.138
405 AKDQ-CR 0.7 Dry Film 0.232 4 75.5 0.405 -0.025 -0.062 0.381 0.502
406 AKDQ-CR 0.7 Dry Film 0.232 4 50.7 0.221 -0.014 -0.064 0.199 0.273
306 AKDQ-CR 0.7 Prelube 0.245 4 80.5 0.149 -0.008 -0.054 0.135 0.186
305 AKDQ-CR 0.7 Prelube 0.246 4 49.7 0.085 -0.003 -0.035 0.081 0.107
418 AKDQ-CR 0.7 Dry Film 0.437 4 100.5 0.249 -0.011 -0.044 0.229 0.312
417 AKDQ-CR 0.7 Prelube 0.438 4 100.5 0.280 -0.017 -0.061 0.256 0.347
304 AKDQ-CR 0.7 Prelube 0.514 4 49.6 0.164 -0.007 -0.043 0.148 0.205
226 AKDQ-EG 0.7 Prelube 0.000 -0.005 0.000 0.001 0.007
218 AKDQ-EG 0.7 Prelube 0.486 0.079 -0.005 -0.064 0.075 0.098
436 AKDQ-EG 0.7 Prelube 0.557 0.079 -0.005 -0.064 0.075 0.098
216 AKDQ-EG 0.7 Prelube 0.571 0.091 -0.005 -0.055 0.083 0.114
217 AKDQ-EG 0.7 Prelube 0.769 0.112 -0.008 -0.072 0.101 0.139
212 AKDQ-EG 0.7 Prelube 0.486 0.045 -0.006 -0.134 0.039 0.054
211 AKDQ-EG 0.7 Prelube 0.571 0.049 -0.006 -0.123 0.046 0.059
74
Table 3.5 (Continued)
Major, Minor and Thickness Strains for Region A of the Channel Draw Sidewall (True Strain)
Inboard Inboard
Drawbead Bead True Strain Strain True Hill 48
Prestrain Steel Grade Lubricant
BTR Radii Penetration Ratio Thickness Effective
Condition Identifier Type
E2/E1 Strain Strain
(mm) (%) Major Minor
209 AKDQ-EG 0.7 Prelube 0.769 0.067 -0.009 -0.135 0.058 0.080
213 AKDQ-EG 0.7 Prelube 0.486 0.025 -0.006 -0.244 0.025 0.028
215 AKDQ-EG 0.7 Prelube 0.571 0.028 -0.005 -0.182 0.025 0.032
201 AKDQ-EG 0.7 Prelube 0.769 0.042 -0.009 -0.215 0.030 0.048
203 AKDQ-EG 0.7 Prelube 2 76.0 0.282 -0.007 -0.025 0.274 0.361
224 AKDQ-EG 0.7 Prelube 2 64.8 0.160 -0.007 -0.044 0.151 0.203
221 AKDQ-EG 0.7 Prelube 3 82.0 0.189 -0.009 -0.048 0.175 0.238
220 AKDQ-EG 0.7 Prelube 3 81.7 0.160 -0.005 -0.031 0.139 0.203
219 AKDQ-EG 0.7 Prelube 0.336 3 81.7 0.282 -0.010 -0.036 0.275 0.359
437 AKDQ-EG 0.7 Prelube 0.336 3 81.7 0.282 -0.010 -0.036 0.275 0.359
222 AKDQ-EG 0.7 Prelube 4 86.8 0.196 -0.006 -0.031 0.187 0.249
223 AKDQ-EG 0.7 Prelube 4 86.8 0.115 -0.006 -0.052 0.107 0.145
307 AKDQ-EG 0.7 Prelube 4 86.8 0.115 -0.006 -0.052 0.116 0.145
438 AKDQ-EG 0.7 Prelube 4 86.8 0.115 -0.006 -0.052 0.107 0.145
501 AKDQ-EG 0.7 Prelube 4 86.5 0.119 -0.002 -0.017 0.122 0.153
407 AKDQ-CR 1.2 Dry Film 4 71.2 0.374 -0.019 -0.051 0.366 0.463
408 AKDQ-CR 1.2 Dry Film 4 35.3 0.156 -0.010 -0.064 0.158 0.192
409 AKDQ-CR 1.2 Prelube 4 35.3 0.194 -0.011 -0.057 0.190 0.239
431 AKDQ-CR 1.2 Prelube 0.364 4 50.2 0.191 -0.008 -0.042 0.183 0.237
311 AKDQ-CR 1.2 Prelube 0.364 4 50.5 0.187 -0.008 -0.043 0.186 0.233
310 AKDQ-CR 1.2 Prelube 0.366 4 37.1 0.131 -0.007 -0.054 0.129 0.162
430 AKDQ-CR 1.2 Dry Film 0.374 4 50.5 0.181 -0.008 -0.044 0.171 0.225
410 AKDQ-CR 1.2 Dry Film 0.554 4 35.5 0.328 -0.021 -0.065 0.316 0.403
429 AKDQ-CR 1.2 Prelube 0.595 4 50.5 0.270 -0.015 -0.056 0.255 0.333
428 AKDQ-CR 1.2 Dry Film 0.598 4 50.6 0.245 -0.012 -0.049 0.234 0.303
309 AKDQ-CR 1.2 Prelube 0.626 4 36.6 0.194 -0.011 -0.057 0.197 0.239
413 AKDQ-CR 1.2 Dry Film 4 99.5 0.233 -0.013 -0.056 0.238 0.288
414 AKDQ-CR 1.2 Prelube 4 99.5 0.302 -0.012 -0.040 0.299 0.377
420 AKDQ-CR 1.2 Dry Film 0.239 4 99.0 0.330 -0.013 -0.040 0.329 0.412
214 AKDQ-EG 1.2 Prelube 0.644 0.0 0.044 -0.007 -0.160 0.041 0.051
225 AKDQ-EG 1.2 Prelube 4 82.2 0.233 -0.009 -0.039 0.215 0.288
75
Table 3.5 (Continued)
Major, Minor and Thickness Strains for Region A of the Channel Draw Sidewall (True Strain)
Inboard Inboard
Drawbead Bead True Strain Strain True Hill 48
Prestrain Steel Grade Lubricant
BTR Radii Penetration Ratio Thickness Effective
Condition Identifier Type
E2/E1 Strain Strain
(mm) (%) Major Minor
207 BH-210-EG 0.7 Prelube 0.644 0.0 0.056 -0.005 -0.089 0.059 0.070
206 BH-210-EG 0.7 Prelube 2 76.1 0.266 -0.009 -0.034 0.260 0.341
512 BH210-EG 0.8 Prelube 3 80.2 0.176 -0.005 -0.028 0.177 0.218
514 BH210-EG 0.8 Prelube 0.259 4 57.0 0.171 -0.005 -0.029 0.167 0.212
516 BH210-EG 0.8 Prelube 0.399 4 34.2 0.142 -0.006 -0.042 0.130 0.175
515 BH210-EG 0.8 Prelube 0.515 4 51.3 0.283 -0.011 -0.039 0.268 0.348
513 BH210-EG 0.8 Prelube 0.330 4 97.7 0.321 -0.014 -0.044 0.309 0.393
518 DP600-HDG 1.2 Prelube 4 81.4 0.176 -0.005 -0.028 0.171 0.205
521 DP600-HDG 1.2 Prelube 0.262 4 48.1 0.156 -0.005 -0.032 0.147 0.181
520 DP600-HDG 1.2 Prelube 0.338 4 88.3 0.309 -0.009 -0.029 0.294 0.359
522 DP600-HDG 1.2 Prelube 0.533 4 50.3 0.277 -0.009 -0.033 0.265 0.321
519 DP600-HDG 1.2 Prelube 6 48.3 0.065 -0.002 -0.031 0.059 0.075
523 DP600-HDG 1.2 Prelube 0.398 6 39.7 0.165 -0.006 -0.037 0.154 0.190
517 DP600-HDG 1.2 Prelube 4 100.5 0.250 -0.008 -0.032 0.237 0.290
502 HSLA-HDG 0.8 Prelube 2 73.8 0.335 -0.006 -0.018 0.323 0.387
503 HSLA-HDG 0.8 Prelube 0.238 4 76.8 0.218 -0.005 -0.023 0.192 0.251
505 HSLA-HDG 0.8 Prelube 0.395 4 49.9 0.262 -0.006 -0.023 0.247 0.302
504 HSLA-HDG 0.8 Prelube 0.161 6 52.5 0.088 -0.002 -0.023 0.076 0.101
415 HSLA-HDG 0.8 Dry Film 4 98.3 0.098 -0.008 -0.082 0.071 0.110
416 HSLA-HDG 0.8 Prelube 4 98.3 0.111 -0.007 -0.063 0.107 0.125
421 HSLA-HDG 0.8 Dry Film 0.322 4 99.0 0.321 -0.013 -0.041 0.291 0.367
507 HSLA-CR 1.2 Prelube 4 81.6 0.202 -0.005 -0.025 0.185 0.233
509 HSLA-CR 1.2 Prelube 0.319 4 93.0 0.329 -0.010 -0.031 0.308 0.377
510 HSLA-CR 1.2 Prelube 0.534 4 51.9 0.298 -0.010 -0.034 0.281 0.341
508 HSLA-CR 1.2 Prelube 6 48.2 0.080 -0.003 -0.038 0.068 0.091
511 HSLA-CR 1.2 Prelube 0.395 6 39.6 0.165 -0.004 -0.024 0.152 0.190
506 HSLA-CR 1.2 Prelube 4 100.4 0.269 -0.008 -0.030 0.250 0.309
76
Table 3.6
Entry and Exit Thickness and True Thickness Strain for Inboard Drawbeads for the Channel Draw Test Pieces
77
Table 3.6 (Continued)
Entry and Exit Thickness and True Thickness Strain for Inboard Drawbeads for the Channel Draw Test Pieces
78
Table 3.6 (Continued)
Entry and Exit Thickness and True Thickness Strain for Inboard Drawbeads for the Channel Draw Test Pieces
79
0.16
y=x
Thickness Strain Area B (True Strain)
0.10
0.08
0.06
Figure 3.3
Relationship between the True Thickness Strain in Regions B and A of the Channel Draw Sidewall
80
0.12 AKDQ & BH210
DP600
Thickness Strain Area C (True Strain)
y=x HSLA
0.10
0.08
0.06
0.04
0.02
0.00
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Thickness Strain Area B (True Strain)
Figure 3.4
Relationship between the True Thickness Strain in Regions C and B of the Channel Draw Sidewall
81
A similar relationship between the strain in regions B and C is shown in Figure 3.4. In contrast to
Figure 3.3, the strain in region C compared to region B depends on steel grade. For the AKDQ
and BH210 steels, the true thickness strain in region C increases at about half the rate of increase
of the true thickness strain in region B. In contrast, for the DP600 steel the increase in true
thickness strain in region C compared to region B is very modest. And for the HSLA steels, the
true thickness strain in region C is essentially zero.
The data on true major, minor and thickness strains in region A is shown in Table 3.5. In
interpreting the results of the ratio of true minor to major strain, it should be recognized that for
small values of major strain, the result is statistically questionable. Also, the Hill 48 effective
strain is calculated from the three principal strains shown in this table.
Thickness measurement at the entry and exit of the inboard drawbead and the calculated true
thickness strain from the channel draw sidewall is shown in Table 3.6. More detailed thickness
strain measurements in the drawbead are included in Appendix 3.2.
A comparison of the true thickness strain at the exit from the drawbead with the strain in region
A of the sidewall shows the strain produced as metal moves over the die entry radius. The back
tension at the entry to the die entry radius also affects the net thinning strain resulting from metal
moving over the die entry radius. Data on this restraining force is shown in Table 3.7.
The bead wrap image analysis data used to compute BSF is shown in Table 3.8 and calculated
tangent-to-tangent wrap angles are shown in Table 3.9. Representative pictures of actual wrap of
the steel in the inner drawbead are shown in Appendix 3.3.
The tensile properties taken in the direction of metal movement into the channel draw die cavity
are shown in Table 3.10. With the exception of the DP600, this orientation is transverse to the
rolling direction (metal movement is in the rolling direction of the coil for the DP600). For the
500 series, tensile test results are available for orientations parallel and transverse to the rolling
direction. These results are shown in Appendix 3.4.
82
Table 3.7
Drawbead Restraining Force for Strip Pulled Through Both the Outboard and Inboard Drawbeads
83
Table 3.7 (Continued)
Drawbead Restraining Force for Strip Pulled Through Both the Outboard and Inboard Drawbeads
84
Table 3.7 (Continued)
Drawbead Restraining Force for Strip Pulled Through Both the Outboard and Inboard Drawbeads
85
Table 3.8
Calculated Radii for the Inboard Drawbead
Inboard Inboard
Back Bead Bead Average Measured Bend Radii during Channel Draw and DBRF Tests Bending
Prestrain Steel Grade Lubricant
Tension Radii Penetration Strain
Condition Identifier Type
Ratio Factor
(mm) (%) R1 R2 Unbend R3 R4 Unbend R5 R6 Unbend R7 R8
432 AKDQ-CR 0.7 Prelube 0.290 2 65.3 3.54 1.99 3.99 0.357
308 AKDQ-CR 0.7 Prelube 0.000 4 87.0 4.62 3.85 4.53 0.260
422 AKDQ-CR 0.7 Prelube 0.000 4 100.4 4.46 4.04 4.43 0.260
302 AKDQ-CR 0.7 Prelube 0.246 4 48.7 8.22 4.18 8.00 0.193
423 AKDQ-CR 0.7 Prelube 0.247 4 100.5 3.92 4.20 4.19 0.270
303 AKDQ-CR 0.7 Prelube 0.247 4 79.6 4.93 4.15 4.86 0.245
427 AKDQ-CR 0.7 Prelube 0.365 4 50.7 7.49 4.24 7.65 0.197
426 AKDQ-CR 0.7 Dry Film 0.367 4 50.7 6.54 4.00 7.08 0.211
434 AKDQ-CR 0.7 Prelube 0.384 4 86.7 4.41 4.06 4.65 0.257
433 AKDQ-CR 0.7 Prelube 0.430 4 86.7 4.23 4.01 4.60 0.262
301 AKDQ-CR 0.7 Prelube 0.514 4 49.3 7.14 4.40 6.61 0.204
425 AKDQ-CR 0.7 Prelube 0.601 4 50.7 6.08 4.19 6.42 0.216
401 AKDQ-CR 0.7 Dry Film 0.000 4 100.4 4.40 4.32 7.64 5.35 4.13 29.00 4.14 3.95 12.34 4.06 4.19 0.562
402 AKDQ-CR 0.7 Dry Film 0.000 4 74.9 5.28 4.50 8.19 6.42 4.91 37.34 4.56 4.50 10.02 4.57 4.39 0.497
403 AKDQ-CR 0.7 Dry Film 0.000 4 50.5 8.09 5.51 6.93 47.08 7.70 5.22 6.92 0.332
404 AKDQ-CR 0.7 Prelube 0.000 4 50.8 7.73 5.45 6.29 40.99 7.52 4.84 5.95 0.352
411 AKDQ-CR 0.7 Dry Film 0.000 4 100.1 4.95 4.55 6.84 5.41 4.18 0.261
412 AKDQ-CR 0.7 Prelube 0.000 4 100.1 4.90 4.49 8.29 5.05 4.01 0.278
405 AKDQ-CR 0.7 Dry Film 0.232 4 75.5 4.15 4.38 9.35 5.32 4.35 57.04 4.19 4.17 12.21 4.16 4.46 0.564
406 AKDQ-CR 0.7 Dry Film 0.232 4 50.7 8.26 5.42 7.52 38.97 8.22 5.01 9.58 7.45 5.93 0.344
306 AKDQ-CR 0.7 Prelube 0.245 4 80.5 4.85 4.70 10.00 5.53 4.51 0.269
305 AKDQ-CR 0.7 Prelube 0.246 4 49.7 8.42 5.75 6.91 0.178
418 AKDQ-CR 0.7 Dry Film 0.437 4 100.5 4.40 4.23 10.58 3.97 5.18 0.298
417 AKDQ-CR 0.7 Prelube 0.438 4 100.5 4.34 4.25 11.74 3.76 4.66 0.312
304 AKDQ-CR 0.7 Prelube 0.514 4 49.6 7.75 6.85 6.95 0.172
226 AKDQ-EG 0.7 Prelube 0.000 0.109
218 AKDQ-EG 0.7 Prelube 0.486 0.109
436 AKDQ-EG 0.7 Prelube 0.557 0.109
216 AKDQ-EG 0.7 Prelube 0.571 0.109
217 AKDQ-EG 0.7 Prelube 0.769 0.109
212 AKDQ-EG 0.7 Prelube 0.486 0.058
211 AKDQ-EG 0.7 Prelube 0.571 0.058
86
Table 3.8 (Continued)
Calculated Radii for the Inboard Drawbead
Inboard Inboard
Back Bead Bead Average Measured Bend Radii during Channel Draw and DBRF Tests Bending
Prestrain Steel Grade Lubricant
Tension Radii Penetration Strain
Condition Identifier Type
Ratio Factor
(mm) (%) R1 R2 Unbend R3 R4 Unbend R5 R6 Unbend R7 R8
209 AKDQ-EG 0.7 Prelube 0.769 0.058
213 AKDQ-EG 0.7 Prelube 0.486 0.030
215 AKDQ-EG 0.7 Prelube 0.571 0.030
201 AKDQ-EG 0.7 Prelube 0.769 0.030
203 AKDQ-EG 0.7 Prelube 0.000 2 76.0 3.17 1.97 3.87 0.459
224 AKDQ-EG 0.7 Prelube 0.000 2 64.8 3.86 2.09 5.01 0.335
221 AKDQ-EG 0.7 Prelube 0.000 3 82.0 4.70 3.23 4.40 0.361
220 AKDQ-EG 0.7 Prelube 0.000 3 81.7 4.45 3.18 4.36 0.316
219 AKDQ-EG 0.7 Prelube 0.336 3 81.7 4.00 3.05 3.59 0.343
437 AKDQ-EG 0.7 Prelube 0.336 3 81.7 3.97 3.10 3.48 0.345
222 AKDQ-EG 0.7 Prelube 0.000 4 86.8 4.61 4.09 4.12 0.348
223 AKDQ-EG 0.7 Prelube 0.000 4 86.8 4.74 4.01 4.41 0.262
307 AKDQ-EG 0.7 Prelube 0.000 4 86.8 4.74 4.01 4.41 0.262
438 AKDQ-EG 0.7 Prelube 0.000 4 86.8 4.77 4.09 4.33 0.262
501 AKDQ-EG 0.7 Prelube 0.000 4 86.5 4.34 3.97 4.24 0.273
407 AKDQ-CR 1.2 Dry Film 0.000 4 71.2 7.57 3.93 6.90 7.78 3.97 6.70 0.620
408 AKDQ-CR 1.2 Dry Film 0.000 4 35.3 20.79 7.97 7.80 23.70 7.62 10.86 0.367
409 AKDQ-CR 1.2 Prelube 0.000 4 35.3 19.05 7.76 8.05 22.14 6.66 14.37 0.368
431 AKDQ-CR 1.2 Prelube 0.364 4 50.2 12.88 4.14 12.24 0.266
311 AKDQ-CR 1.2 Prelube 0.364 4 50.5 11.20 4.24 12.72 0.268
310 AKDQ-CR 1.2 Prelube 0.366 4 37.1 17.62 7.06 13.23 0.202
430 AKDQ-CR 1.2 Dry Film 0.374 4 50.5 12.72 4.18 12.98 0.263
410 AKDQ-CR 1.2 Dry Film 0.554 4 35.5 18.37 5.78 15.95 20.68 5.85 18.19 0.363
429 AKDQ-CR 1.2 Prelube 0.595 4 50.5 11.44 4.26 12.14 0.268
428 AKDQ-CR 1.2 Dry Film 0.598 4 50.6 12.22 4.33 12.49 0.262
309 AKDQ-CR 1.2 Prelube 0.626 4 36.6 16.96 6.19 17.08 0.204
413 AKDQ-CR 1.2 Dry Film 0.000 4 99.5 4.52 4.14 7.69 4.28 4.24 0.468
414 AKDQ-CR 1.2 Prelube 0.000 4 99.5 4.20 4.15 8.95 4.04 4.27 0.491
420 AKDQ-CR 1.2 Dry Film 0.239 4 99.0 4.45 3.97 7.96 3.98 4.22 0.485
214 AKDQ-EG 1.2 Prelube 0.644 0.049
87
Table 3.8 (Continued)
Calculated Radii for the Inboard Drawbead
Inboard Inboard
Back Bead Bead Average Measured Bend Radii during Channel Draw and DBRF Tests Bending
Prestrain Steel Grade Lubricant
Tension Radii Penetration Strain
Condition Identifier Type
Ratio Factor
(mm) (%) R1 R2 Unbend R3 R4 Unbend R5 R6 Unbend R7 R8
225 AKDQ-EG 1.2 Prelube 0.000 4 82.2 8.09 3.92 7.56 0.377
207 BH-210-EG 0.7 Prelube 0.644 0.058
206 BH-210-EG 0.7 Prelube 0.000 2 76.1 3.21 2.20 3.59 0.448
512 BH210-EG 0.8 Prelube 0.000 3 80.2 3.91 3.06 3.71 0.361
514 BH210-EG 0.8 Prelube 0.259 4 57.0 5.72 4.14 5.55 0.247
516 BH210-EG 0.8 Prelube 0.399 4 34.2 12.45 4.59 11.27 0.174
515 BH210-EG 0.8 Prelube 0.515 4 51.3 5.72 4.09 4.30 0.266
513 BH210-EG 0.8 Prelube 0.330 4 97.7 4.26 4.14 13.54 4.14 4.20 0.346
518 DP600-HDG 1.2 Prelube 0.000 4 81.4 7.86 4.20 6.32 0.369
521 DP600-HDG 1.2 Prelube 0.262 4 48.1 10.95 4.73 10.17 0.264
520 DP600-HDG 1.2 Prelube 0.338 4 88.3 5.77 4.16 5.07 0.369
522 DP600-HDG 1.2 Prelube 0.533 4 50.3 11.50 4.79 10.53 0.259
519 DP600-HDG 1.2 Prelube 0.000 6 48.3 15.55 6.85 10.35 0.217
523 DP600-HDG 1.2 Prelube 0.398 6 39.7 16.82 6.69 14.83 0.200
517 DP600-HDG 1.2 Prelube 0.000 4 100.5 4.07 4.02 7.57 4.38 4.06 0.477
502 HSLA-HDG 0.8 Prelube 0.000 2 73.8 2.75 2.00 2.81 0.532
503 HSLA-HDG 0.8 Prelube 0.238 4 76.8 4.27 4.00 4.14 0.293
505 HSLA-HDG 0.8 Prelube 0.395 4 49.9 5.85 4.42 5.95 0.239
504 HSLA-HDG 0.8 Prelube 0.161 6 52.5 10.85 6.30 10.02 0.163
415 HSLA-HDG 0.8 Dry Film 0.000 4 98.3 5.02 4.22 7.29 5.93 4.43 0.283
416 HSLA-HDG 0.8 Prelube 0.000 4 98.3 4.98 4.19 7.25 5.16 4.46 0.292
421 HSLA-HDG 0.8 Dry Film 0.322 4 99.0 4.59 4.23 9.84 4.17 4.58 0.324
507 HSLA-CR 1.2 Prelube 0.000 4 81.6 8.29 4.05 7.25 0.356
509 HSLA-CR 1.2 Prelube 0.319 4 93.0 5.58 4.03 4.64 0.381
510 HSLA-CR 1.2 Prelube 0.534 4 51.9 10.73 4.68 8.38 0.275
508 HSLA-CR 1.2 Prelube 0.000 6 48.2 11.86 7.78 10.29 0.217
511 HSLA-CR 1.2 Prelube 0.395 6 39.6 16.25 6.72 16.62 0.195
506 HSLA-CR 1.2 Prelube 0.000 4 100.4 4.58 4.23 10.01 4.44 4.32 0.463
88
Table 3.9
Calculated Tangent-to-Tangent Wrap Angle on Drawbead Radii
R1 R2
Drawbead Bead
Prestrain Steel Grade Male Bead Theoretical tangent-to-
Radii Penetration
Condition Identifier Type tangent Wrap Angle
89
Table 3.9 (Continued)
Calculated Tangent-to-Tangent Wrap Angle on Drawbead Radii
R1 R2
Drawbead Bead
Prestrain Steel Grade Male Bead Theoretical tangent-to-
Radii Penetration
Condition Identifier Type tangent Wrap Angle
90
Table 3.10
Tensile Properties in Channel Sidewalls for Tests Oriented Parallel to
the Direction of Metal Motion into the Channel Draw Die
Plane
0.2% Overall
Prestrain Steel Grade Male Bead Strain Tensile Uniform Total
Yield Stress-Strain
Condition Identifier Type Yield Stress Elongation Elongation
Stress relation
Stress
91
Table 3.10 (Continued)
Tensile Properties in Channel Sidewalls for Tests Oriented Parallel to
the Direction of Metal Motion into the Channel Draw Die
Plane
0.2% Overall
Prestrain Steel Grade Male Bead Strain Tensile Uniform Total
Yield Stress-Strain
Condition Identifier Type Yield Stress Elongation Elongation
Stress relation
Stress
92
Table 3.10 (Continued)
Tensile Properties in Channel Sidewalls for Tests Oriented Parallel to
the Direction of Metal Motion into the Channel Draw Die
93
4.0 Analysis
4.1 Overview
Prestrain does not affect composition, and as an engineering approximation, it does not affect
microstructure. For higher prestrains, some changes in mechanical and crystallographic
anisotropy are likely. Nonetheless, the primary effect of prestrain is on changes in dislocation
density and morphology, and the internal stresses that result from these changes in dislocation
substructure. Since anisotropy and dislocation substructure were not experimentally determined,
the data are analyzed using an empirical approach based on fundamental principles.
For prestraining in the channel draw die, the major factors affecting dislocation substructure are;
(1) the as-received characteristics of the material, (2) BTR, and (3) BSF. BSF is a measure of
cumulative strain that sums the cyclic bending strains as metal passes through a drawbead and
over a die entry radius. Conceptually, BSF is an engineering approximation that does not
consider the number of bending cycles or the sequence of bending strain. Thus, regardless of
cyclic history, equal values of BSF have an equal effect. This approach is justified on the basis
that very little is known about the cyclic stress-strain behavior for these steels at the operative
strain amplitudes. Furthermore, since experimental error in determining bending strain is in part
random, total BSF is more accurate than its individual components.
BTR is a ratio of the restraining stress to the plane strain yield strength. With each successive
bending cycle, the restraining stress increases by an undetermined amount. In addition, there is
the conceptual issue of whether the as-received plane strain yield strength should be increased by
the resulting strain hardening if BTR is adjusted for each bending cycle. Given the lack of
experimental data, these factors were not considered in developing predictive equations.
In considering bending under tension, increasing back tension increases the maximum bending
strain (outside surface) and decreases the minimum bending strain (inside surface). With
sufficient back tension, the minimum bending strain becomes positive. Thus, for higher values of
BTR, the bending strain gradient can be from a higher to a lower positive major strain. In such a
case, the bending strain gradient is entirely compressive on unbending.
Since factors which affect dislocation substructure should affect the enhanced FLC effect, the net
thinning strain, and the increase in yield strength, the effect of BTR and BSF on these parameters
should be related. Since net thinning strain can be measured in a press shop, it can be used to
predict both the enhanced FLC effect and the increase in yield strength from prestraining. Also,
since lubrication affects friction as metal passes through drawbeads and over a die entry radius,
reducing friction, reduces the increase in back tension.
94
4.2 Regression Analysis
Using the fundamental understanding described in the previous section, a variety of empirical
relations were examined to determine the best predictive equations for (1) ∆FLCo and net
thinning strain in terms of BTR and BSF and (2) ∆FLCo in terms of net thinning strain. In
evaluating relationships, the criteria were that (1) the relationship be reasonable at its limits and
(2) minimize the difference between predicted and actual values. The resulting predictive
relations using all the data (sample size equal 89) are shown in Equations 4.2.1, 4.2.2, and 4.2.3.
In examining the limits of Equations 4.2.1 and 4.2.2, it can be seen that (1) negative values of
∆FLCo and εt are not possible, (2) ∆FLCo is zero when BSF is zero, and (3) positive values of
∆FLCo are possible for BTR = 0. The first two observations are reasonable, but the third
observation requires explanation. The absence of tension across a bend is generally described as
a free bend. In free bending, the net thinning strain is zero, but Equation 4.2.2 predicts net
thinning strain. The proposed explanation for this behavior is that BTR is taken at the entry to the
drawbead, and that BTR develops as metal is bent and straightened as it moves through a
drawbead. Also, for cases where the steel is not in contact with the tooling, bending and
straightening strains may not be identical. On a practical basis, the only limitation to the use of
Equations 4.2.1 and 4.2.2 is the case of a die entry radius with no back tension.
While not strictly a limitation, the simplifications underlying Equations 4.2.1 and 4.2.2 should be
recognized. These simplifications are (1) all steels exhibit a similar response to BTR and BSF,
(2) BTR is not adjusted for successive bending and straightening cycles, and (3) BSF is
independent of cyclic strain history.
With regard to Equation 4.2.3, it can be seen that it predicts a negative ∆FLCo for zero net
thinning strain. The predicted relationship for Equation 4.2.3 and the actual data points are
shown in Figure 4.2.1. It can be seen from Figure 4.2.1 that there is only one data point with a
negative value of ∆FLCo. Close inspection of Figure 4.2.1 suggests that for small net thinning
strains the relation is non-linear and approaches zero. Furthermore, the data suggest that at larger
net thinning strains, there are also indications of non-linearity. From these observations, it is
recommended that (1) predicted negative values of ∆FLCo be taken as zero, (2) that Equation
4.2.3 not be extended beyond the data from which it is derived (εt ≅ 0.4), and (3) that predicted
values of ∆FLCo for εt more than 0.3 be considered as conservative.
95
30
25
15
10
-5
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
True Thickness Strain, εt
Figure 4.2.1
∆FLCo as a Function of εt Using Equation 4.2.3
96
It can be seen from Equation 4.2.3, that its slope is 59.7. This slope is a measure of the difference
in magnitude between ∆FLCo and εt. Thus, if Equation 4.2.2 is multiplied by 59.7, it can be
compared with Equation 4.2.1. The result is shown in Equation 4.2.4.
It can be seen that the constants in Equation 4.2.1 and 4.2.4 are similar. This result is consistent
with the hypothesis that BTR and BSF are the determining factors for describing the enhanced
FLC effect.
In evaluating Equations 4.2.1 and 4.2.2, it is of interest to determine the effect of BTR and BSF
on ∆FLCo and εt. These effects are shown respectively in Figures 4.2.2 and 4.2.3. The
pronounced effect of BTR at higher values of BSF is apparent from these figures.
97
50
45
BTR = 1.33
40
BTR = 1.00
35
30
∆FLCo
BTR = 0.67
25
20
BTR = 0.33
15
BTR = 0
10
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
BSF
Figure 4.2.2
Effect of BSF on ∆FLCo for Various Values of BTR Using Equation 4.2.1
98
0.8
0.7
BTR = 1.33
0.6
BTR = 1.00
0.5
BTR = 0.67
0.4
-εt
BTR = 0.33
0.3
BTR = 0
0.2
0.1
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
BSF
Figure 4.2.3
Effect of BSF on εt for Various Values of BTR Using Equation 4.2.2
99
4.2.2 Statistical Analysis
While R2 is one measure of statistical validity, added insight is provided by comparing predicted
and actual results. In this analysis, use is made of the data on experimental uncertainty
determined in Section 2 to predict the expected difference between predicted and actual values.
The first step in this analysis is to write difference equations for Equations 4.2.1 to 4.2.3 as
shown in Equations 4.2.5 to 4.2.7.
The experimental uncertainty for ∆ can be determined from the square root of the sum of the
squares of the partial derivative of each experimental variable times the uncertainty associated
with that experimental variable. Equations 4.2.5 to 4.2.7 can be transformed respectively to
Equations 4.2.8 to 4.2.10 where U represents the statistical uncertainty.
0.5
∂∆ 2 ∂∆
2
∂∆
2
U = 2 + 2 + 2
∆ ∂BTR BTR ∂BSF BSF ∆FLCo
EQ 4.2.8 U U U
∆FLCo
0.5
∂∆ 2 2
∂∆
2
2 ∂∆ 2 U 2
EQ 4.2.9 U ∆ = +
BTR
U U BSF + εt
∂BTR
∂BSF ∂ε t
0.5
∂∆
2
∂∆
2
(U ∆FLCo )
2 2
EQ 4.2.10 U ∆ = U +
εt
∂ε t ∂∆FLCo
Using the results from Section 2 as the best estimates of the uncertainty, the standard deviation
for each experiment variable, is shown below:
UBSF = ± 0.0122
100
U∆FLCo = ± 1.63%
Uεt = ± 0.0127
Since UBTR is steel grade dependent, values were calculated separately for each steel grade-
thickness combination. The results are shown in Table 4.2.1 where values of UBTR can be seen to
range from 0.0029 to 0.0090. For simplicity, a weighted average is used with the result that UBTR
equals 0.0069.
Table 4.2.1
Determining the Weighted Average of the Experiment Uncertainty in Measuring BTR
Steel Grade Sample Size Uncertainty in Measuring BTR (UBTR)
0.7 AKDQ* 47 0.0090
1.2 AKDQ* 15 0.0055
BH210** 7 0.0059
0.8 HSLA 7 0.0039
1.2 HSLA 6 0.0029
DP600 7 0.0029
Weighted Average 89 0.0069
* Average of Results for EG and Cold Rolled
** Average of Results for 0.7 and 0.8 EG
∂∆
= 104BSF 1.54
∂BTR
∂∆
= 1.54(41.7 + 104 BTR )BSF 0.54
∂BSF
∂∆
= −1
∂∆FLCo
∂∆
= 1.20 BSF 1.27
∂BTR
101
∂∆
= 1.27(0.624 + 1.20 BTR )BSF 0.27
∂BSF
∂∆
= −1
∂ε t
∂∆
= 59.7
∂ε t
∂∆
= −1
∂∆FLCo
For Equations 4.2.8 and 4.2.9, it can be seen that U∆ depends on the nominal values of BTR and
BSF. To understand the effect of BTR and BSF on U∆, these equations were evaluated for all
combinations of BTR equal 0, 0.3, and 0.6 and BSF equal 0.1, 0.2, 0.3, and 0.4. For Equation
4.2.8, U∆ for two standard deviations ranges from ± 3.29 to ± 4.05 with U∆FLCo having the
predominant effect, UBSF having a secondary effect, and UBTR having essentially no effect on U∆.
For Equation 4.2.9, U∆ for two standard deviations ranges from ±0.0254 to ±0.0284. With Uεt
having the predominant effect, UBSF having a secondary effect and UBTR having essentially no
effect.
Quantitatively, the importance of U∆FLCo and Uεt on U∆ can also be seen by comparing their
effect to the total experimental uncertainty. Specifically for Equation 4.2.8, at two standard
deviations, U∆FLCo is ± 3.26% versus a total experimental range of 3.29 to 4.05. For Equation
4.2.9, at two standard deviations, Uεt equals ± 0.0254 versus a total experimental range of
± 0.0254 to ± 0.0284.
For Equation 4.2.10, the calculation of U∆ is more straightforward. At two standard deviations,
U∆ is ± 3.60%.
The expected difference in ∆ based on experimental uncertainty can be compared to the actual
values of ∆ by using plots of observed versus predicted values. Such plots are shown in Figures
4.2.4, 4.2.5, and 4.2.6, respectively for Equations 4.2.1, 4.2.2, and 4.2.3.
102
30
∆FLCo = (41.7 + 104 BTR) BSF1.54
R2 = 94.5%
25
Predicted ∆FLCo (%)
20
15
10
0
0 5 10 15 20 25 30
Observed ∆FLCo (%)
Figure 4.2.4
Predicted Versus Actual Observations for Equation 4.2.1
103
0.5
εt = (0.624 + 1.2 BTR) BSF1.27
2
R = 98.5%
0.4
Predicted Thickness Strain
0.3
0.2
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5
Observed Thickness Strain
Figure 4.2.5
Predicted Versus Actual Observations for Equation 4.2.2
104
30
∆FLCo = 59.7εt - 1.24
R2 = 81.6%
25
Predicted ∆FLCo (%)
20
15
10
0
0 5 10 15 20 25 30
Observed ∆FLCo (%)
Figure 4.2.6
Predicted Versus Actual Observations for Equation 4.2.3
105
For Figure 4.2.4, it can be seen that 95% of the observations are within about ± 5 strain percent
of the regression line, while the expected experimental uncertainty ranges from ± 3.29 to ±
4.05%. This indicates that Equation 4.2.1 explains about 70% of the total variation. That is,
about ± 1.5 strain percent of the variation is not accounted for by Equation 4.2.1. This result is
reasonable, since as indicated earlier in this section, a number of simplifications were required to
work with the available data.
As shown in Figure 4.2.6, which represents Equation 4.2.3, it can be seen that 95% of the
observations are also within about ± 5 strain percent of the regression line. Since the
experimental uncertainty is ± 3.6 strain percent, Equation 4.2.1 and 4.2.3 explain about the same
amount of the total variation.
With regard to predicting net thinning strain, it can be seen in Figure 4.2.5 that the actual
variation in predicted-actual net thinning strain for 95% of the observations is about ± 0.05
strain. This is considerably greater than the expected variation from experimental uncertainty,
which ranges from ± 0.0254 to ± 0.0284. Thus, Equation 4.2.8 explains less of the actual
variation than is the case for Equations 4.2.1 and 4.2.3.
Regressions similar to Equations 4.2.1 and 4.2.2 were also run for each steel grade-thickness
combination. These results are shown in Tables 4.3.1 to 4.3.4. For ease in presentation, these
equations are characterized as shown in Equations 4.3.1 and 4.3.2.
Table 4.3.1
Effect of Steel Grade on the Constants for Equations 4.3.1 and 4.3.2
R2 (%) for
Steel Grade N A1 B1 A2 B2 A3 B3
EQ 4.3.1 EQ 4.3.2
0.7/0.8 mm
AKDQ 47 39.7 0.58 101.1 1.10 1.64 1.24 93.8 98.7
BH210 7 47.8 0.73 108.9 1.64 1.62 1.34 99.3 99.7
HSLA 7 58.1 0.94 111.7 4.04 1.44 1.76 98.4 98.4
1.2 mm
AKDQ 15 36.2 0.63 67.6 0.85 1.24 1.20 99.1 99.4
HSLA 6 51.8 0.66 111.7 1.11 1.44 1.26 99.4 99.6
DP600 7 41.7 0.63 138.2 1.71 1.65 1.37 97.5 99.1
106
Table 4.3.2
Effect of Thickness on the Constants for Equations 4.3.1 and 4.3.2
R2 (%) for
Steel Grade Thickness (mm) N A1 B1 A2 B2 A3 B3
EQ 4.3.1 EQ 4.3.2
0.7 47 39.7 0.58 101.1 1.10 1.64 1.24 93.3 98.7
AKDQ
1.2 15 36.2 0.63 67.6 0.85 1.24 1.20 99.1 99.4
HSLA 0.8 7 58.1 0.94 111.7 4.04 1.44 1.76 98.1 98.4
1.2 6 51.8 0.66 111.7 1.11 1.46 1.26 99.4 99.6
Table 4.3.3
Effect of Steel Grade on the Constants for Equation 4.2.3
Steel Grade N Constant Coefficient R2 (%)
0.7/0.8 mm
AKDQ 47 – 0.91 52.6 75.7
BH210 7 –1.65 54.9 97.3
HSLA 7 +2.70 57.5 87.0
1.2 mm
AKDQ 15 +0.25 59.0 90.1
HSLA 6 – 4.64 77.4 96.7
DP600 7 – 1.87 59.4 93.3
Table 4.3.4
Effect of Thickness on the Constants for Equation 4.2.3
Steel Grade Thickness (mm) N Constant Coefficient R2 (%)
0.7 47 – 0.91 52.6 75.7
AKDQ
1.2 15 + 0.25 59.0 90.1
HSLA 0.8 7 + 2.70 57.5 87.0
1.2 6 – 4.64 77.4 96.7
From inspection of Tables 4.3.1 to 4.3.4, it can be seen that there are no consistent trends with
steel grade-thickness combinations. Since only conventional as-received tensile properties are
available, and there is no detailed information on microstructure, it is difficult to characterize
these materials to analyze the effect of steel grade and thickness on the constants, coefficients,
and exponents in Equations 4.2.1, 4.2.2, and 4.2.3. Another difficulty is in some cases, small
sample sizes. Given the enumerated difficulties, no attempt is made to incorporate the effect of
steel grade in predicting ∆FLCo and net thinning strain. However, it seems likely that the effect
of steel grade and thickness on ∆FLCo and net thinning strain accounts for some of the
unexplained variation in Equations 4.2.1, 4.2.2, and 4.2.3.
As part of the experimental program, direct comparisons between prelube and dry film
lubrication are available for six test conditions. In all cases, the drawbead radii are 4 mm, and the
die entry radius is 12 mm. These data are presented in Table 4.4.1. While the coefficient of
107
friction (COF) for the prelube and dry film lubricants was not directly measured in the channel
draw die, it is evident that the COF for the dry film lubricant should be considerably less than the
COF for the prelube. Thus, the increase in back tension as a sheet moves through a drawbead
should be less with dry film lubrication than with prelube.
In examining Table 4.4.1, it can be seen that for identical die conditions, the result of using dry
film lubrication in place of a prelube is smaller values of BSF and net thinning strain. These
observations indicate that the increase in back tension in the drawbead affects the resulting
values of BSF and net thinning strain.
Table 4.4.1
Comparison of Prelube and Dry Film Lubrication for a
Drawbead Radius of 4 mm and a Die Entry Radius of 12 mm
Test DB Prestrained Prestrained
Lube BTR Penetration BSF εt
Condition Type* FLCo YS
0.7 CR AKDQ
412 Prelube SQ 0 100.1 0.278 0.105 38.3 376
411 Dry SQ 0 100.1 0.261 0.072 35.0 326
Film
Difference 0.017 0.033 3.3 50
404 Prelube DBL 0 50.8/51.3 0.352 0.159 41.8 418
SQ
403 Dry DBL 0 50.5/51.2 0.332 0.113 42.7 380
Film SQ
Difference 0.020 0.046 – 0.9 38
1.2 mm CR AKDQ
429 Prelube RD 0.595 50.5 0.268 0.255 52.1 453
428 Dry RD 0.598 50.6 0.262 0.234 50.0 436
Film
0.006 0.021 2.1 17
414 Prelube SQ 0 99.5 0.491 0.299 50.5 469
413 Dry SQ 0 99.5 0.468 0.238 47.0 428
Film
0.023 0.061 3.5 41
409 Prelube DBL 0 35.3/34.7 0.368 0.190 45.8 404
RD
408 Dry DBL 0 35.3/34.7 0.367 0.158 48.0 382
Film RD
0.001 0.032 –2.2 22
0.8 mm HDG
416 Prelube SQ 0 98.3 0.292 0.107 32.6 501
415 Dry SQ 0 98.3 0.283 0.071 30.4 490
Film
0.009 0.036 2.4 11
* SQ = Square drawbead, DBL SQ = Double square drawbead, RD = Round drawbead, and DBL RD =
Double Round Drawbead
** Penetration respectively for first and second drawbead
108
These observations confirm that Equations 4.2.1 and 4.2.2 are a simplification of actual
conditions because they do not include the effect of increasing back tension with successive
bending and straightening. In contrast, Equation 4.2.3 is unaffected because ∆FLCo is predicted
from the measured net thinning strain. Thus, Equation 4.2.3 can be used to determine if the
difference in net thinning strain between prelube and dry film lubrication is consistent with the
differences in ∆FLCo. The data for this analysis is shown in Table 4.4.2, where considering
experimental variation, it can be seen that with one exception, that this is the case. This single
exception may be the result of an outlier in measuring ∆FLCo.
Table 4.4.2
Predicting Changes in ∆FLCo from Changes in εt in Comparing
Prelube and Dry Film Lubrication by Using Equation 4.2.3
εt Changes in ∆FLCo
Test Condition
Prelube Dry Film Difference Predicted* Actual Predicted-Actual
412 vs 411 0.105 0.072 0.033 2.0 3.3 – 1.3
404 vs 403 0.159 0.113 0.046 2.7 – 0.9 1.8
429 vs 428 0.255 0.234 0.021 1.3 2.1 – 0.8
414 vs 413 0.299 0.238 0.061 3.6 3.5 0.1
409 vs 408 0.190 0.158 0.032 1.9 – 2.2 4.1
416 vs 415 0.107 0.071 0.036 2.1 2.4 – 0.3
Average 0.6
* Equation 4.2.3
The effect of increasing back tension on net thinning strain can be analyzed by using Equation
4.2.2 to predict the difference in net thinning strain between prelube and dry film lubrication.
Table 4.4.3 shows BSF, the predicted and actual difference in net thinning strain, and the
difference between these quantities. It can be seen from Table 4.4.3 that for BTR equal zero, the
predicted difference in net thinning strain is consistently less than the actual difference.
Table 4.4.3
Predicting Changes in εt from Changes in BSF in Comparing Prelube
and Dry Film Lubrication Using Equation 4.2.2 with Known Values of BTR
BSF/BSF1.54 Changes in εt
Test Condition Predicted-Actual
Prelube Dry Film Predicted* Actual
BTR = 0
412 vs 441 0.278/0.139 0.261/0.126 0.008 0.033 – 0.025
404 vs 403 0.352/0.200 0.332/0.183 0.011 0.046 – 0.035
414 vs 413 0.491/0.334 0.468/0.311 0.014 0.061 – 0.047
409 vs 408 0.368/0.214 0.367/0.214 0 0.032 – 0.032
416 vs 415 0.292/0.150 0.283/0.143 0.004 0.036 – 0.032
BTR = 0.6
429 vs 428 0.286/0.145 0.262/0.127 0.024 0.021 + 0.003
* Equation 4.2.3
In contrast, for BTR equal 0.6, the predicted difference is virtually identical to the actual
difference. These observations suggest that for higher BTR, the increase in back tension with
successive bending and straightening is less important.
109
Examination of Table 4.4.3 also reveals that for zero back tension, the difference between
predicted and actual net thinning strain is related to BSF with prelube. This relationship is shown
in Figure 4.4.1 where the line in the figure is forced through zero. The slopes for the data points
range from 0.0899 to 0.1233 and the average slope is 0.10. The explanation for this behavior is
that BTR has a much smaller effect on actual bend radius in a drawbead than it does on net
thinning strain. Since BSF depends on the actual bending radius, it is reasonable that the
difference in BSF does not explain the difference in net thinning strain when increasing back
tension through successive bending and straightening is important.
In summarizing the effect of a lower COF on increasing back stress through a drawbead, it can
be seen that ignoring this effect is more detrimental to predicting thinning strain (Equation 4.2.2)
than it is for predicting ∆FLCo (Equation 4.2.1). This behavior contributes to the greater
unexplained variation in Equation 4.2.2 compared to Equation 4.2.1.
110
0.06
∆ = Predicted - Actual change in εt
0.05
0.04
0.03
0.02
0.01
0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6
BSF with Prelube
Figure 4.4.1
Predicted Minus Actual Change in εt as a Function of Measured BSF with Prelube
for Comparisons Between Prelube and Dry Film Lubrication
111
4.5 Predicting Prestrained Yield Strength
Prestrained yield strength can be predicted using σ = Kεn with as-received K and n and net
thinning strain. As an engineering approximation, deformation in the channel draw wall is
assumed to be plane strain. For plane strain, net thinning strain can be converted to effective
strain using Hill 48 with planar isotropy and normal anisotropy. The resulting conversion factor
is shown in Equation 4.5.1.
CF = (1 + R ) ) / (1 + 2R )
0.5
EQ 4.5.1
Where CF is the conversion factor
Detailed results are shown in Appendix 4.1, and the results are summarized in Table 4.5.1. The
EG AKDQ and BH210 steels are not included because K values are not available.
Table 4.5.1
Comparison of Predicted Versus Actual Prestrained Yield Strength
For AKDQ, HSLA, and DP600 Steels
Predicted-Actual Yield Strength (MPa)
Steel Grade Sample Size
Average Range
0.7 AKDQ CR 25 – 15.4 – 41/+ 3
1.2 AKDQ CR 14 – 4.6 – 16/+ 14
0.8 HSLA HDG 7 – 12.0 – 32/+ 15
1.2 HSLA CR 6 –12.5 – 37/+ 15
1.2 DP600 HDG 7 + 68.0 – 34/+ 138
It can be seen from Table 4.5.1 that the predictions of the prestrained yield strength for the
AKDQ and HSLA steels are reasonably accurate with a general trend for the predicted
prestrained yield strength to be on average 5 to 15 MPa less than the actual measured values.
Though no statistical analysis was done, empirically, the range of predicted minus actual values
seems reasonable for the combined effect of variations in measuring n, K, and net thinning strain.
In contrast to the AKDQ and HSLA steels, it can be seen that for the DP600 steel that the
average predicted prestrained yield strength is 68 MPa more than the actual prestrained yield
strength, and the scatter is considerably greater.
From a fundamental perspective, the results for the AKDQ and HSLA steels are reasonably
consistent with the hypothesis that the same changes in dislocation substructure control ∆FLCo,
net thinning strain, and prestrained yield strength. The small, but consistent observation that
predicted prestrained yield strength for the AKDQ and HSLA steels is most likely related to
cyclic strain hardening. With regard to the DP600, the most likely explanation is a combined
effect of higher cyclic strain hardening and stress-strain curves that do not follow conventional
parabolic strain hardening. Since the detailed stress-strain curves are not available, it is not
possible to test this hypothesis.
112
4.6 Predicting Bending Radii in the Draw Die
Various types of regression equations were evaluated for predicting bending radii in a drawbead.
Some conclusions from this work follow:
The regressions for predicting curvature (C) are in terms of the actual curvature divided by the
curvature of the drawbeads. All available data are used in the following regression equations.
The term θ is the tangent-to-tangent wrap angle in degrees.
Since curvature is the inverse of radius, it can be shown that the curvature ratio, C, must be less
than or equal to 1.00. Otherwise the actual bending radius would be less than tooling radius.
Several attempts were made to fit the data with non-linear regressions with a form such that the
curvature ratio, C, could not exceed one. However, it was found that the statistical fit of these
regression equations were inferior to the linear relations of Equations 4.6.1, 4.6.3 and 4.6.4.
It can be seen that the R2 values for Equations 4.6.1, 4.6.2 and 4.6.4 are lower than the R2 values
for Equations 4.2.1 to 4.2.3. Nonetheless, the equations for predicting the curvature ratio can be
used to provide insight on actual bending radii in drawbeads. Since most of the data are for radii
one, two, and four, the curvature ratios from Equations 4.6.1, 4.6.2 and 4.6.4 are plotted as a
113
function of the tangent-to-tangent wrap angle in Figure 4.6.1. For radius two, the actual wrap
angle is used in the calculation of curvature ratio, but is divided by two for the sake of graphical
presentation.
It can be seen from Figure 4.6.1 and Equations 4.6.1, 4.6.2 and 4.6.3 that the curvature ratio for
radius two rapidly approaches the geometric curvature of the tooling and that BTR has no
significant effect on the result. In contrast, the curvature ratio for radii one and four are much
more sensitive to wrap angle. It can also be seen that the curvature ratio at radius one and four
are similar.
While data are more limited, it is also of interest to compare curvature ratios at radii two and
three using Equations 4.6.2 and 4.6.3. This comparison is shown in Figure 4.6.2, where it can be
seen that the curvature ratio at radius three is highly dependent on BTR. By comparing the
curvature ratios for radii one and four in Figure 4.6.1 with the curvature ratio for radius three in
Figure 4.6.2, it can be seen that for BTR equal 0.6 the curvature ratio for radius three is greater
than for radius one and four. In contrast, for BTR equal zero, the curvature ratio at radius three is
less than the curvature ratio for radius one and four.
114
1.0
C1, BTR = 0.6
0.9
C2 C1, BTR = 0
0.8
0.7
Curvature Ratio
0.6
0.5
C4
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70 80 90
Wrap Angle (θ)
Figure 4.6.1
Effect of Wrap Angle on Curvature Ratio for 0≤ BTR = 0.6 for Radii One, Two, and Three
115
1.0
0.9
C2
0.8
0.7
C3, BTR = 0.6 C3, BTR = 0
Curvature Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70 80 90
Wrap Angle (θ)
Figure 4.6.2
Effect of Wrap Angle on Curvature Ratio for 0≤ BTR = 0.6 for Radii Two and Three
116
The results shown in Figures 4.6.1 and 4.6.2 can be qualitatively explained by using the available
pictorial evidence of strip shape in a drawbead. These details are shown in Appendix 3.3. These
observations for radii one to four can be summarized as follows for cases where the actual
bending radii do not conform to the tooling radii:
Radius one: The strip initially contacts the entry shoulder of the drawbead, but the bend is
essentially in air. Particularly at small tangent-to-tangent wrap angles, unbending is also in air.
For a free bend, it is mechanically reasonable for drawbead geometry to have a controlling
effect. It also seems intuitive that BTR can have a modest effect in forcing the strip to conform
more closely with the tooling at the entry to the drawbead.
Radius two: The nature of the drawbead geometry suggests that at radius two, the strip is most
constrained by the tooling. At shallower penetrations, the strip may only contact radius two at a
single point on the actual arc of the drawbead and unbending is in air. As the tangent-to-tangent
wrap angle increases, the contact length increases so that the strip conforms to the shape of the
drawbead.
Radius three: Radius three is the exit of a square drawbead. The flat between radius two and
three for the male bead is similar for all experimental work in this project. Therefore, the
applicability of the following analysis to other male bead configurations is uncertain. Since the
flat on the male drawbead is short, at small tangent-to-tangent wrap angles, the male drawbead
acts as a round drawbead and therefore radius three does not exist. In this case, the strip
gradually unbends in air between radius two and radius four. For the drawbead geometries that
were studied, increasing tangent-to-tangent wrap angle should have a similar effect to that at
radii one and four. The similar value of the coefficient of the tangent-to-tangent wrap angle in
Equations 4.6.1 and 4.6.3 is consistent with this observation.
The difference between Equations 4.6.1 (radius one) and 4.6.3 (radius 3) is the constant term and
the coefficient of BTR. For radius one, the strip is initially constrained between the binder and
the upper die. In contrast, at radius three, there is no such constraint and even at deep
penetrations the strip does not conform to the geometry of the square bead. Thus, BTR has a
much greater effect on the curvature ratio at radius three because BTR has a significant effect on
forcing conformity to the radius. This effect is described in Equation 4.6.3 by a negative constant
term coupled with a larger coefficient of BTR.
Radius four: For most tangent-to-tangent wrap angles, free bending is observed. The effect of
tangent-to-tangent wrap angle is most likely associated with the curvature of the strip at radius
three affecting the angle at which the strip approaches radius four. It has been indicated that at
radius one and three, the effect of BTR is to aid in forcing conformity of the strip to a flat at the
entry to the radius. Since this geometric factor is not relevant at radius two and four, there should
be no effect of BTR on curvature ratio for these two radii. Indeed, for the range of conditions that
were studied, BTR does not seem to have an effect on curvature ratio at radius four. Nonetheless,
curvature ratios for radii one and four appear to be similar.
117
Equations 4.6.1 to 4.6.4 can also be evaluated by using them to compute BSF, and comparing the
result with BSF measured directly from the experimental observations. This can be done using
the following analytical approach:
EQ 4.6.5 CA = CDB C
Where CDB is the curvature of the drawbead radius
C is the curvature ratio
CA is the actual curvature
Equation 4.6.5 is applicable to all radii. Maximum bending strain is calculated using Equation
4.6.6.
EQ 4.6.6 εB = 0.5 ln (1 + CA t)
Where εB is the maximum bending strain and
t is the initial strip thickness
For the die entry radius, it is assumed that the strip conforms to the geometric radius. The result
of calculating BSF for single round drawbeads is shown in Figure 4.6.3. It should also be noted
that Figure 4.6.3 is based on the full data set used to develop the underlying equations.
It can be seen from Figure 4.6.3 that for nominal thicknesses of 0.7 and 0.8 mm, the predictions
are reasonable. In contrast, for a nominal thickness of 1.2 mm, predicted values of BSF are
frequently larger than the actual values of BSF. These results would seem to suggest that more
accurate prediction of the curvature ratio would require consideration of a thickness effect. Such
an effect is logical because thickness affects stiffness, and stiffness should affect conformity to
tooling.
To evaluate the thickness effect for round drawbeads, regressions were run for all thicknesses
and separately for the nominal 0.7 and 1.2 mm thicknesses. It was found that for radius two, the
R2 for the 0.7 mm nominal thickness is 32%. Since radius two exhibits the largest bending strain,
no further attempt was made to analyze the data.
Though the data are limited, it can also be seen from Figure 4.6.3 that there is no significant
effect of steel grade. However, care must be taken in extending this observation to much higher
strength steels because the stiffness modulus includes the elastic modulus and for plastic
bending, use of the secant modulus may be more appropriate. Nonetheless, for high strength
steels this is probably a second order effect.
118
Thickness 0.716 to 0.780 mm Thickness 1.180 to 1.233 mm
0.6 0.6
AKDQ AKDQ
y=x y=x
BH 210 DP600
HSLA HSLA
0.5 0.5
0.4 0.4
Predicted BSF
Predicted BSF
0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6
Actual BSF Actual BSF
Figure 4.6.3
Relationship between Predicted and Actual BSF for a Thickness Range of 0.716 to 1.233 mm
119
5.0 Applications
In applying the enhanced FLC concept to metal flowing off a binder surface, it is necessary to
consider metal moving through a drawbead and over a die entry radius. From a fundamental
perspective, BTR and BSF are the key factors. In the channel draw die, back tension was applied
to metal moving into the die cavity. Back tensions were significantly less than the plane strain
yield strength so that metal moving across the binder and into the die cavity was not plastically
deformed. In contrast, metal movement in production dies can be more complex. The following
analysis describes these complexities and discusses issues relevant to use of the enhanced FLC
concept with FEA. This discussion also provides background on the use of enhanced FLC in a
press shop.
Production dies can be built with or without stop blocks. If stop blocks are used, the only back
tension on a drawbead results from deformation as metal flows from the edges of a blank toward
the die cavity. Such deformation can be observed when the shape of a blank perimeter becomes
uneven as the stamping process progresses. One manifestation of such behavior is wrinkling.
Wrinkling on a binder is always associated with plastic deformation and positive thickness
strains. Wrinkling can also be observed in cup drawing of steels with planar isotropy for which
earing (non-uniform deformation of the blank perimeter) is not observed. For dies without stop
blocks, grinding on the binder surface is often needed to adjust for metal thickening. Also, when
dies are designed without stop blocks, there is friction between the blank and the binder surfaces
that increases back tension on a drawbead. In summarizing deformation conditions behind a
drawbead, plastic strains can range from 0 to an estimated 0.05 effective strain, and elastic
strains can range from negligible to that associated with an effective strain of 0.05. Ratios of
minor to major strains can vary over a range from perhaps – 1.2 to about 0.
In the channel draw die, the ratio of minor to major strain is 0; i.e., plane strain. In the absence of
experimental data for back tension in non-plane strain conditions, the following approach seems
consistent with the metallurgical factors underlying enhanced FLC behavior:
Determine the ratio of minor to major strain and the magnitude of the effective strain using
either FEA prediction or experimental strain measurements for press shop work.
As appropriate, use existing methodology to compute the principal stress orthogonal to the
direction of metal movement.
Compute BTR from the drawbead restraining force divided by the actual thickness at the
entry to the drawbead. Given the typical magnitude of thickness strain at the entry to a
drawbead, the use of as-received thickness is usually a reasonable engineering assumption.
With regard to determining BSF, for use in FEA, it is necessary to predict the actual bending
radii in a drawbead. It should also be noted that all experimental work in this project was done
with a straight drawbead and production drawbeads can be curved.
120
For gentle plan view draw bead curvature, it would probably be a reasonable engineering
approximation to apply results from straight drawbeads. However, for sharply curved drawbeads,
it is unlikely that BSF could be calculated using predictive equations for straight drawbeads.
It should also be noted that drawbead radii in production dies may be non-uniform. Also, the
range of drawbead geometries evaluated in this study is more limited than those used in
production drawbeads.
In summary, it is clear that more work is needed before Equations 4.2.1 and 4.2.2 can be used for
FEA or other analytic work on production dies. It should also be noted that most FEA
formability analysis treats a drawbead as a line of force. As a result, thinning strain is
underestimated and use of the enhanced FLC concept would result in an additional overestimate
of formability.
The preceding discussion has highlighted many of the issues in using Equations 4.2.1 and 4.2.2.
for FEA. Many of these same issues apply in the press shop and are compounded by
experimental difficulties in measuring actual bending radii in drawbeads and in determining back
tension. Thus, for press shop work, net thinning strain should be determined and ∆FLCo
calculated using Equation 4.2.3. Since critical local necks result from non-plane strain
deformation paths, as an engineering approximation, an adjusted FLC should be determined by
adding ∆FLCo to the standard FLCo for the material, and assuming that the standard curve shape
applies.
The enhanced FLC effect applies to material subject to bending and straightening. Using
Equation 4.2.2, it can be seen that net thinning strain depends on BSF. In prior discussion, it has
been shown that BSF increases with each bending and straightening step. Thus, the maximum
enhanced FLC effect is observed in material that has passed through the drawbead and over the
die entry radius. The following approach can be used to determine if it is important to consider
the effect of die entry radius. For stretch-draw dies, strip generally conforms to the die entry
radius. Thus bending strain for the die entry radius can be calculated using Equation 5.2.1.
The bending strain for the die entry radius calculated using Equation 5.2.1 is an incremental
BSF. Its effect on ∆FLCo can be estimated from Equation 4.2.1. It can be seen from Figure 4.2.4,
which is a graphical depiction of Equation 4.2.1 that ∆FLCo is highly dependent on BTR. In the
absence of information on BTR, it would be too conservative to assume that BTR equals zero.
Therefore, given the effect of a drawbead, engineering experience should be used to estimate
BTR.
121
If more specific information on BTR is desired, it can sometimes be inferred from thickness
measurements. In this approach, the thickness at the entry and exit from the drawbead can be
used to calculate strain. Measuring elastic strain is not practical, but if plastic strain can be
measured, then BTR is at least one.
It is also evident that BSF increases as strip moves through a drawbead. However, on a practical
basis, it is unlikely that this progression through a drawbead can be measured with sufficient
accuracy in an actual stamping. Thus, on a practical basis, the regions of potential interest are 1)
strip that has moved over just the die entry radius, 2) strip that has moved through the drawbead
and over the die entry radius, and 3) strip that has moved through the drawbead. For strip that has
just moved over the die entry radius, ∆FLC can be estimated using the procedures that were
described for determining if the die entry radius has a significant effect on ∆FLCo. However, it is
common practice to place the drawbead close to the die entry radius, therefore this region would
normally be quite small.
Metal movement off a post occurs when restraining force on the binder surface is high so that
metal is pulled off the surface of the post and into a die wall or addendum area. In this case,
metal movement over the post entry radius produces the bending strain factor (BSF). Since metal
movement off the post is a consequence of plastic deformation, BTR is high.
In plan view, the perimeter of a post is typically composed of straight lines and convex arcs,
though in some cases concave arcs are observed, e.g., the windshield side of a hood. If it is
assumed that the strain path for metal moving off the post and over the post radius is co-linear,
then in principle, it would seem reasonable to approximate BTR using a ratio of the principal
stress in the direction of metal movement divided by the yield strength for that strain path. Using
this approximation, Equation 4.2.1 can be used to estimate the enhanced FLC effect.
It is evident from Equation 4.2.1 that the post radius has a major effect on the increase in ∆FLCo.
This effect is shown graphically in Figure 5.3.1 which illustrates the relationship between
∆FLCo and the ratio of bending radius to thickness. For small R/T, it can be seen that values of
∆FLCo are substantial and that BTR has a significant effect on the resulting ∆FLCo. However,
as R/T increases, ∆FLCo diminishes rapidly and the effect of BTR becomes insignificant. Given
that strains on the post are usually small, significant increases in ∆FLCo are expected to be
observed for R/T ratios of about 5 or less.
122
18
16
14
12
∆FLCo (%)
10
8
BTR = 1.5
6
4
BTR = 1.0
2
0
0 2 4 6 8 10 12 14 16 18 20 22
Radius / Thickness
Figure 5.3.1
Effect of Radius to Thickness Ratio on ∆FLCo for Metal Movement off a Post Where BTR is High
123
In predicting the increase in ∆FLCo during die design, a side calculation can be made using the
results of conventional FEA. In such a calculation, BSF can be calculated by assuming the strip
conforms to the post radius. If thinning on the post (punch) should be significant, a conservative
approach would be to use the deformed thickness rather than the as-received thickness. In
calculating BTR, the principal stress in the direction of metal movement could be calculated
from the major and minor strain, a yield criteria that includes the effect of crystallographic
anisotropy (R value), a hardening rule and the Levy-Von Mises equation. As discussed
previously, this flow stress would be divided by the yield strength for the appropriate strain path.
For press shop work, it should be possible to use either Equation 4.2.1 or 4.2.3 to calculate the
increase in ∆FLCo. In using Equation 4.2.1, procedures would be similar to those described
previously for die design except that radii and strains would be measured rather than taken from
FEA.
In using Equation 4.2.3, thickness would be measured for material that had flowed off the punch.
However, as discussed previously, a conservative approach for calculating net thinning strain
would be to use the thickness of the material prior to its moving over the entry radius on the post.
It should also be noted that the discussion related to metal flow off a post or a punch is based on
extrapolating Equations 4.2.1 or 4.2.3 outside the range of the data on which they are based. That
is, there is no experimental data for radius to thickness ratios less than or equal to five for BTR
more than or equal to one. Nonetheless, given that these equations are consistent with
fundamental factors, this use of Equations 4.2.1 or 4.2.3 is a reasonable engineering
approximation.
It can be seen from Equation 4.2.3 that for each one strain percent increment in net thinning
strain due to bending and straightening, ∆FLCo only increases by 0.6 strain percent Thus, the
loss in FLCo from thinning from bending and straightening is 40% not 100%. This difference is
the basis for the enhanced FLC effect. However, given this 40% loss, it is evident that at some
point the thinning strain attributable to bending and straightening will lead to failure. This
behavior is illustrated schematically in Figure 5.4.1.
124
Deformation Due
εt as a measure of potential failure
to Bending and
Straightening
Failure
Adjusted FLC
∆FLCo
As Received FLC
Figure 5.4.1
Schematic Illustrating the Factors which determine the Limitation to the Enhanced Forming Limit Effect
125
The horizontal axis in Figure 5.4.1 is the true thinning strain due to bending and straightening,
and the vertical axis is true thinning strain as a measure of failure. The net thinning strain can be
represented in Figure 5.4.1 by a 45° line; i.e., Y = X. The issue is to compute the adjusted
forming limit curve which accounts for the enhanced FLC effect. This adjusted forming limit
curve has two components; the as-received FLC and ∆FLC.
In expressing an FLC in terms of thickness strain, the left side of the FLC is approximated as a
line of pure shear in true strain. Since thickness strain is zero for pure shear, the left side of an
FLC is the thickness strain for plane strain when strain is expressed in true strain. Thus, the left
side of an as-received FLC can be described in terms of true thickness strain as a horizontal line
that is independent of the net thinning strain due to bending and straightening. In contrast,
∆FLCo is a function of net thinning strain from bending and straightening. These curves are
included in Figure 5.4.1.
It can be seen from Figure 5.4.1 that failure occurs when the line representing thinning from
bending and straightening intersects the adjusted FLC. At this point a critical local neck forms
which is the failure criteria used in this work.
Given the mix of engineering and true strain, some analytic manipulations are required to put all
the various terms in true strain. For convenience in following the algebra, Y is used for true
thinning strain as a measure of potential failure and X is used for true thinning strain due to
bending and straightening.
∆FLCo is calculated using Equation 4.2.3. Substituting for FLCo and ∆FLCo in Equation 5.4.1
results in Equation 5.4.3.
It can be seen from Equation 5.4.4 that if X is less than Y deformation can proceed without
failure and Y minus X represents the remaining thinning strain after unbending that is available
for further deformation. This relationship is illustrated in Figure 5.4.2 for a steel with an n value
126
of 0.21 and a thickness of 1.2 mm. For this example, if the thinning strain exceeds 0.55, no
further deformation would be possible without formation of a critical local neck. Furthermore, on
a practical basis, it would seem that if Y – X is small, a stamping process would probably be
unstable.
As shown in Figure 5.4.2, increasing bending and straightening is not a method for improving
formability because the capacity for subsequent deformation decreases. What the enhanced
forming limit does describe is the fact that approximately 60% of the thinning due to bending
and straightening does not affect the capacity for subsequent thinning as forming progresses.
The limit strain for the enhanced FLC effect can be calculated directly from Equation 5.4.4 using
iterative techniques. However, it should be noted that this requires extending Equation 4.2.2 to
net thinning strains that were not experimentally evaluated. Thus, calculated limit strains may be
subject to error, but it is clear that limit strains do exist.
127
0.9
Thinning Strain as a measure of Failure
0.8
0.7
0.6
Y<X
FLC0' No Additional
0.5
Deformation
Possible
0.4
0.3 Y>X
Additional
0.2 Deformation Increase in Thickness Strain From
Possible Bending and Straightening
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
εt from Bending and Straightening
Figure 5.4.2
Relationship between increasing Thickness Strain from Bending and Straightening and the increase in FLCo
Due to the Enhanced Forming Limit Effect. The example is for a steel with nt = 0.21 and thickness = 1.2 mm
128
5.5 Optimization
The use of the enhanced FLC concept for optimization requires consideration of the die design
process. If blank dimensions and die cavity geometry are fixed, BTR is fixed because the
movement of the metal across the binder surface and into the die cavity determines the extent
and pattern of deformation which in turn creates the back tension on a drawbead.
Drawbeads are used to control the flow of metal into a die cavity. In general, the required pattern
of drawbead restraining force would seem to be fixed once the blank dimensions and die cavity
geometry is fixed. Thus, optimization using the enhanced FLC concept reduces to minimizing
the net thinning strain produced by the drawbead while maintaining a constant drawbead
restraining force. The need to minimize the net thinning strain produced by a drawbead is a
direct consequence of the fact that approximately 40% of the net thinning strain is no longer
available for subsequent deformation.
In simplistic form, the drawbead restraining force is a function of force multiplication of back
stress due to friction and the force needed for bending and straightening. The analysis of the
effect of lubrication indicates that reduced friction reduces the build-up of back tension, and this
reduces net thinning strain. Similarly, the data on drawbead restraining force shows that reduced
friction reduces drawbead restraining force. An analysis of the effect of friction on the relative
rate of these changes is not possible with the available data from this study as would be a
detailed study of the effect of the sequence of bending operations. Thus, further work would be
required to determine if the enhanced FLC concept could be used to optimize drawbead design to
improve formability.
Current practice in using FLCs as a failure criteria for breakage is to use a safety margin of ten
strain percent. This safety margin accounts for variations in material, process, and the inherent
variation in the Keeler-Brazier Equation (Equation 5.4.2). In using Equations 4.2.1 and 4.2.3, the
regression technique provides the best estimate of a relationship given data with experimental
variation in measured parameters.
The total variation between predicted and actual ∆FLCo is ± 5 strain percent. It has been shown
that about ± 70% of this variation can be explained by a statistical analysis of the effect of
experimental variation in the measured parameters. The remaining variation represents
inaccuracies in the predictive equations. However, the magnitude of this variation, approximately
± 1.5 strain percent is probably similar to the underlying variation in the Keeler-Brazier
Equation. Since the experimental variation in ∆FLCo includes the variation in determining the
as-received FLC for each lot of steel, it is recommended that the same 10 strain percent safety
margin be applied when using an enhanced forming limit curve.
129
5.7 Applicability to Other Steel Grades
On the basis of the underlying mechanisms for the enhanced FLC effect, the concept should
apply to other steel grades with similar microstructures and cyclic stress-strain behavior at high
strain amplitudes. Given the limited information on these characteristics, it is not possible to
determine with any certainty the extent to which Equations 4.2.1 to 4.2.3 can be extended to
other steel grades. Furthermore, it is clear that caution is needed in applying this work to TRIP or
complex phase steels. Also it is known that commercial dual phase steels exhibit a range of
microstructures. Since the microstructure of the DP600 steel used in this study is not available,
the applicability of Equations 4.2.1 to 4.2.3 to all dual phase steels is not known with any
certainty.
Clean interstitial free steels may exhibit different cyclic stress-strain behavior. However, such
steels are more likely to exhibit a greater tendency for dislocation annihilation which would tend
to result in an increased enhanced FLC effect. However, it must be emphasized that this
conclusion is speculative.
With regard to thickness, it seems reasonable that the applicable thickness range can be extended
beyond the 0.7 to 1.2 mm range evaluated experimentally in this study. On a speculative basis,
extending Equation 4.2.3 to a thickness range of 0.55 to 1.65 mm would seem reasonable.
130
6.0 Summary
For press shop evaluation, the enhanced forming limit effect can be predicted by using the
net thinning strain resulting from bending and straightening in a drawbead and over die radii.
This predictive equation has been verified for AKDQ, bake hardening, HSLA, and Dual
Phase (DP600) grades at thicknesses ranging from 0.7 to 1.2 mm.
Applications for this predictive equation include (1) steel that has moved through a drawbead
and over a die entry radius and (2) steel that has moved off a post (punch) with an entry
radius to steel thickness ratio of about 5 or less.
The enhanced FLC effect can be rationalized on the basis of the effect of stress reversals and
back tension on changes in the dislocation substructure that affect the deformation behavior
of steel. These factors are described by a back tension ratio (BTR) which is the back tension
divided by an approximate yield strength and a bending strain factor (BSF) which is the sum
of the maximum bending strains based on the actual bending radii.
An internally consistent set of empirical equations have been developed that predict the net
thinning strain and the enhanced FLC effect in terms of BTR and BSF, and the yield strength
after forming in terms of net thinning strain and material properties. The internal consistency
of these equations is indicative of a common underlying mechanism.
The physical basis of the enhanced FLC effect is that 60% of the net thinning strain
associated with bending and straightening is available for subsequent deformation. However,
the 40% of the net thinning strain that is not available for subsequent deformation limits the
extent of the enhanced FLC effect. A procedure to predict this limit is included in the text.
The underlying basis for the limit to the enhanced FLC effect means that its importance is
that it adjusts as-received FLCs so that they are not overly conservative failure criteria.
Statistical analysis of the experimental error associated with measured quantities accounts for
about 70% of the difference between predicted and actual values of ∆FLCo. This translates to
an unexplained difference of about ± 1.5 strain percent.
Statistical analysis of the experimental error associated with measured quantities accounts for
about 50% of the difference between predicted and actual values of net thinning strain.
The remaining differences between predicted and actual values are thought to be the result of
engineering approximations used in developing the empirical equations or errors in
estimating experimental uncertainty. Specific engineering approximations that may cause the
unexplained differences between predicted and actual values are thought to include the
number of strain reversals, not accounting for the increase in back tension, and the sequence
of bending strain in successive bending and straightening steps.
Individual lots of steel had varying effects on net thinning strain and the magnitude of the
enhanced FLC effect. This behavior is not consistent with either steel grade or thickness. The
131
most likely cause is individual variations in cyclic stress-strain behavior at large strain
amplitudes. These individual differences by lot of steel is not thought to have a significant
effect on the overall validity of the predictive equations, but probably contributes to the
unexplained differences between predicted and actual values.
132
LIST OF TABLES
Section 2
Section 3
Section 4
4.2.1 Determining the Weighted Average of the Experimental Uncertainty in Measuring BTR
4.3.1 Effect of Steel Grade on the Constants for Equations 4.2.1 and 4.2.2
4.3.2 Effect of Thickness on the Constants for Equations 4.2.1 and 4.2.2
4.3.3 Effect of Steel Grade on the constants for Equation 4.2.3
4.3.4 Effect of Thickness on the Constants for Equation 4.2.3
133
4.4.1 Comparison of Prelube and Dry Film Lubrication for a Drawbead Radius of 4 mm and a
Die Entry Radius of 12 mm
4.4.2 Predicting Changes in ∆FLCo from Changes in εt in Comparing Prelube and Dry Film
Lubrication by Using Equation 4.2.3
4.4.3 Predicting Changes in εt from Changes in BSF in Comparing Prelube and Dry Film
Lubrication Using Equation 4.2.2 with Known Values of BTR
4.5.1 Comparisons of Predicted Versus Actual Prestrained Yield Strains for AKDQ, HSLA,
and DP600 Steels
134
LIST OF FIGURES
Section 2
Figure 2.2.1 Hydraulic Press Used to Produce Channel Draw Test Pieces
Figure 2.2.2 Non-Scale Schematic of Channel Draw Die Showing Location of Drawbead
Inserts
Figure 2.2.3 Formed Channel Draw Test Piece in the Channel Draw Die
Figure 2.2.4 Scaled Schematic of Channel Draw Die Showing the Post and the Binder Position
at the Completion of the Press Stroke
Figure 2.2.5 Scaled Schematic of the Channel Draw Tooling Stack without the Post
Figure 2.2.6 System for Identifying Radii in a Single Square Drawbead
Figure 2.2.7 Schematic Defining 0% Drawbead Penetration
Figure 2.2.8 Schematic Defining 100% Drawbead Penetration
Figure 2.2.9 Non-Scale Schematic of a Single Square Drawbead
Figure 2.2.10 Non-Scale Schematic of Round Male Drawbead Inserts with Different Clearances
Figure 2.2.11 Schematic of Adjustable Penetration Outboard Drawbead Insert
Figure 2.2.12 Schematic of Adjustable Penetration Inboard Single Drawbead Insert
Figure 2.2.13 Schematic of Round and Square Male Beads Used with the Adjustable
Penetration Inner Drawbead Insert
Figure 2.2.14 Picture of Adjustable Penetration Inner Drawbead Insert
Figure 2.2.15 Schematic of Adjustable Penetration Inboard Double Drawbead Insert
Figure 2.2.16 Polishing the Die Entry Radius of an Inboard Drawbead Insert
Figure 2.2.17 Positioning a Blank Prior to Forming a Channel Section
Figure 2.2.18 Picture of Drawbead Restraining Force Apparatus
Figure 2.2.19 Close-up View of the Hydraulic Clamp Used in the Apparatus to Determine
Drawbead Restraining Force
Figure 2.2.20 Non-scale Schematic of the Channel Draw Die Configured for Drawbead
Restraining Force Tests
Figure 2.2.21 Example of a Plot of Drawbead Restraining Force as a Function of Sampling Rate
Figure 2.2.22 Drawbead Restraining Force Versus Penetration for 0.7 mm Cold Rolled AKDQ
Showing the Effect of Prelube and Dry Film Lubrication
Figure 2.2.23 Drawbead Restraining Force Versus Penetration for 1.2 mm Cold Rolled AKDQ
Showing the Effect of Prelube and Dry Film Lubrication
Figure 2.2.24 Drawbead Restraining Force Versus Penetration for 0.8 mm Electrogalvanized
BH210 with Prelube
Figure 2.2.25 Drawbead Restraining Force Versus Penetration for 0.8 mm Hot Dip Galvanized
and 1.2 mm Cold Rolled HSLA with Prelube
Figure 2.2.26 Drawbead Restraining Force Versus Penetration for 1.2 mm Hot Dip Galvanized
DP600 with Prelube
Figure 2.2.27 Standard Deviation Versus Average Drawbead Restraining Force for all Test
Conditions
Figure 2.2.28 Standard Deviation Versus Average Drawbead Restraining for Each Thickness-
Steel Grade Combination
135
Figure 2.2.29 Experimental Cumulative Frequency Distribution for the Standard Deviation of
Drawbead Restraining Force for all Test Conditions Shown in Figure 2.2.27
Figure 2.4.1 Identification of Measurement Locations Used to Determine Net Thickness Strain
and Major and Minor In-Plane Strains in the Channel Draw Test Piece Sidewall
Figure 2.4.2 Strain Measurement Positions from Channel Draw Sidewall
Figure 2.4.3 Frequency Distribution for the Standard Deviation in Measuring Net Thickness
Strain in the Channel Draw Test Piece Sidewall
Figure 2.5.1 CCD Camera Mounted on the Binder of the Channel Draw Die and Focused on
the End of the Drawbead Insert
Figure 2.5.2 Cumulative Frequency Distribution for all the Data for the Parameter
Range/Average for Drawbead Radius Measurements
Figure 2.6.1 Schematic of Die Used for the Marciniak Double Blank Test
Figure 2.6.2 Pattern of Incipient Necks at a Sample Width of 100 to 110 mm
Figure 2.6.3 Pattern of Incipient Necks at a Sample Width of 110 to 125 mm
Figure 2.6.4 Pattern of Incipient Necks at a Sample Width of 125 to 130 mm
Figure 2.6.5 Pattern of Incipient Necks at a Sample Width of more than 130 mm
Figure 2.6.6 Pattern of Local Necks for the 0.8 mm Hot Dip Galvanized HSLA Steel
Figure 2.6.7 X-Y Table with Mitutoyo Vernier Used to Measure Strains in the Marciniak
Samples
Figure 2.6.8 Apparatus Used for Magnification and Measurement of Circle Grids Used for
Determining Forming Limit Curves
Figure 2.6.9 Typical Forming Limit Curve
Section 3
Figure 3.1 Experimental Combinations of BSF and BTR for all AKDQ Steels
Figure 3.2 Experimental Combination of BSF and BTR for all BH210, USLA, and DP600
Steels
Figure 3.3 Relationship between the True Thickness Strain in Regions B and A of the
Channel Draw Sidewall
Figure 3.4 Relationship between the True Thickness Strain in Regions C and B of the
Channel Draw Sidewall
Section 4
136
Figure 4.6.2 Effect of Wrap Angle on Curvature Ratio for 0≤ BTR = 0.6 for Radii Two and
Three
Figure 4.6.3 Relationship between Predicted and Actual BSF for a Thickness Range of 0.716
to 1.233 mm
Section 5
Figure 5.3.1 Effect of Radius to Thickness Ratio on ∆FLCo for Metal Movement off a Post
Where BTR is High
Figure 5.4.1 Schematic Illustrating the Factors which determine the Limitation to the
Enhanced Forming Limit Effect
Figure 5.4.2 Relationship between increasing Thickness Strain from Bending and
Straightening and the increase in FLCo Due to the Enhanced Forming Limit
Effect. The example is for a steel with nt = 0.21 and thickness = 1.2 mm
137
APPENDIX LIST
138