Bake Hardening Response of DP800 and the Influence on the 'In Service Performance'

D. Faras MT06.39

Department of Mechanical Engineering

Bake Hardening Response of DP800 and the Influence on the 'In Service Performance'

D. Faras August 2006 MT

Bake Hardening Response of DP800 and the Influence on the 'In Service Performance'

Master Thesis Report

D. Faras MT06.39 September 2006

Technische Universiteit Eindhoven

Department of Mechanical Engineering Division of Computational and Experimental Mechanics

Committee members:
Dr. Ir. Jens de Kanter (Corus, mentor) Dr. Ir. Willem Witteman (TU/e, mentor) Prof. Dr. Ir. Marc Geers (TU/e,)

Co advisor:
Dr. Stephen Carless (Corus)

ABSTRACT
The trend for the use of AHSS in automotive industry has grown in the last decades in an exponential way. This has encouraged research departments all around the world to study processes than enhance the characteristics of actual steel, in order to optimize weight for fuel economy and structural stiffness in safety. One of these processes is Bake Hardening (BH). Bake Hardening uses the temperature of the paint baking cycle in the automotive process to raise the strength of steel by the diffusion of Carbon atoms into dislocations. The effect of bake hardening has been measured especially for low carbon steels, but as the use of AHSS in auto parts has increased considerably, measuring the BH response of these steels has gained importance. Commercial reasons have focused this project on the BH response of a Dual Phase steel (DP800). The idea was to measure how does the BH affect the material in crash performance, and determining if this effect is considerable enough to include it in actual FE models of this material. To do this, a number of tensile tests of DP800 samples were tested. In these tests, conditions like strain path, pre strain level and baking temperature were varied in order to measure their influence on the BH response of this DP. The results of these tests were related to a first approach of the influence of BH in crash tests. These crash tests consisted of 3-point bending tests of Closed Top Hat sections and axial crash tests of Closed Top Hat sections and Hydro formed boxes. Paint baking is going to continue being a part of the automotive process; in that sense, BH will have an effect in all the material present in the cars structure. Although the influence of BH seemed to be small crash performance for the steel in this project, more work is recommended to be able to predict with more accuracy the effect of BH on other steels. Measuring BH on compression, bending and multi axial strain conditions can help build a BH model for future work after developing a strain path and strain rate dependent material model for crash performance on FE.

LIST OF SYMBOLS
A80 Ag AHSS BH CP D DP E Fmax Fmean IF L RD ReH , ReL Rm Rp S UTS Elongation until 80% total extension Elongation until UTS Advanced High Strength Steels Bake Hardening Complex Phase Steel Direction of Tensile Specimen: 90 with respect RD Dual Phase Steel Elasticity Modulus Peak force Average or mean force Interstitial Free Steel Direction of Tensile Specimen: 0 with respect RD Rolling Direction Upper Yield Point Lower Yield Point Ultimate Tensile Stress Yield Stress Direction of Tensile Specimen: 45 with respect RD Ultimate Tensile Stress

TABLE OF CONTENTS
CHAPTER 1: INTRODUCTION ...................................................................................6
1.1 Automotive Safety.................................................................................................... 6

CHAPTER 2: LITERATURE RESEARCH...................................................................9

2.1 Strain Ageing and Bake Hardening?.......................................................................... 9 2.2 Bake Hardenable steels ......................................................................................... 12 2.3 Dual Phase High Strength Steels ............................................................................ 14

CHAPTER 3: TENSILE TESTING.............................................................................18

3.1 Introduction ........................................................................................................... 18 3.2 Experimental Details .............................................................................................. 18 3.2.1 Materials ......................................................................................................... 18 3.2.2 Specimens ...................................................................................................... 19 3.2.3 Test setup ....................................................................................................... 22 3.2.4 Data files ......................................................................................................... 23 3.3 Results.................................................................................................................. 24 3.3.1 BH response with no pre strain......................................................................... 25 3.3.2 Ageing at room temperature ............................................................................. 26 3.3.3 Uniaxial Pre strain tensile test........................................................................... 30 3.3.4 Overview uniaxial test ...................................................................................... 35 3.3.5 Tests from hydro formed boxes ........................................................................ 41 3.3.6 Microstructure ................................................................................................. 44 3.4 Conclusions ........................................................................................................... 45

CHAPTER 4: CRASH TESTING................................................................................48

4.1 Introduction ........................................................................................................... 48 4.2 Three Point Bending Tests ..................................................................................... 48 4.2.1 Experimental details......................................................................................... 49 4.2.2 Results ........................................................................................................... 51 4.2.3 Empirical Fit .................................................................................................... 54 4.2.4 Conclusions .................................................................................................... 56 4.3 Axial Crash Tests................................................................................................... 57 4.3.1 Experimental Details ........................................................................................ 60 4.3.2 Res ults ........................................................................................................... 62 4.3.3 Conclusions .................................................................................................... 64

CHAPTER 5: IS IT IMPORTANT TO INCLUDE BH BEHAVIOUR OF DP800 ON CRASH SIMULATIONS?............................................................................................66

5.1 Why is not pertinent to include BH in crash performance? ........................................ 66

CHAPTER 6: CONCLUSIONS AND RECOMMENDATIONS...................................72 REFERENCES.............................................................................................................75 ACKNOWLEDGEMENTS ..........................................................................................77 APPENDIX...................................................................................................................78

A. TENSILE TEST RESULTS FOR OTHER AHSS ........................................................ 78 B. 3 POINT BENDING TEST FOR OTHER AHSS ......................................................... 83

CHAPTER 1: INTRODUCTION

1.1 Automotive Safety

One of the most important, if not the most, transportation media for people all a round the world is the automobile. Everywhere people use cars on daily basis to get their jobs, to go to schools and universities; to go anyplace. Each day more people are using cars as transport, and because of this, more people are involved in car accidents. Still, many of these accidents have very serious consequences. Only in the European Union, 40000 people die every year and 500 thousand hospital admissions are registered because of automotive accidents. Although it sounds cruel to label human life with a money value, all this accidents cost the governments something around 160 billion Euros [14]. And this numbers are getting larger worldwide, especially in countries with a developing automotive industry. In order to reduce fatalities or injuries, and even reduce the amount of accidents, car manufactures, governments and private institutes around the world are working together to find possible solutions to this enormous problem. The ideal condition concerning automotive accidents would be of course to be able to prevent them from happening at all. Car manufactures have developed many features that help drivers avoid having an accident (Electronic Stability programs, Braking systems, and Radar systems among others), at the same time Governments have reinforced policies to have a better control of human behavior as a driver (speeding limits, drug and alcohol controls) as well as worked hard to improve all the environment and driving conditions (better roads and its surroundings). But unfortunately, accidents still occur, and there is still going to be quite a long time before Intelligent Automobiles and Roads are capable enough to prevent accidents from happening. In the meantime, those same entities are working very hard to try to make cars safer, and especially to protect the passengers from any injury or fatality. This is where researchers in the automotive industry and material science come into the picture: lighter, stronger, and more resistant materials (among other characteristics) are needed to u se them as parts in vehicles. Frames, chassis, panels, engine, drive train are just a few parts that have taken advantage of the material development since the day the first car was built. But this improvement has been especially notorious in the last 30-40 years, where fuel consumption, emissions, weight and safety started to become mayor issues in the automotive industry; in this same period, the steel industry has focused in the development of High Strength Steels (HSS), which has been a mayor advanced not only for the automotive industry, but also other fields as science and technology. Focusing in the automotive industry and crash worthiness, one of the biggest problems and most lethal accidents for the passengers are the side impact ones, especially the ones poles and/or trees. The reason is very simple: in the side impact there is no space available in the vehicle structure to install a protection device for the driver and passengers. There is an average of 10 to 20 centimeters space between the impact object and the person inside the car, so the available distance to absorb the energy of the impact is not the same as the one available in front impact as in this case in the general case there is over 1 meter distance between the front bumper and the passengers. Also, the time the intrusion object takes to get to the passengers in the side impact is much less than the front case. Although side, door and curtain airbags have also started to be important for this kind of accidents, the improvement in the outcome of the crash consequences on the passenger is

not close to satisfactory yet. That is why development of HSS (Advanced High Strength Steels and/or Ultra High Strength Steels AHSS & UHSS) and its application on chassis structures, A, B and C pillars and side impact bars have become more important in vehicle design. But not only the development of new material is important main focus of researchers around the world: New designs, new processing, joining and forming techniques, as well as new heat and surface treatments have become major research topics in the automotive industry. All these possibilities give designers new choices to make cars lighter and safer. Modern cars always have to pass through a paint baking cycle. This paint baking cycle is basically a process that consists in getting the car in an oven to dry and bond the paint to the cars structure. Depending on the quality required (as well as the luxury level of the car), this process can take place several times, depending on the amount of paint layers required. It was discovered that this process not only helped the paint job, but also had some influence on the mechanical properties of the materials in the car structure. This process is called Bake Hardening. Not all the material present the same level of bake hardening response; if fact, in some there is no effect at all. The steels that present a higher level of influence of the Bake Hardening process are called bake hardenable steels, and this steels are generally used in outer panels because of their dent resistance characteristics. But the car structure consists in a huge amount of different parts that need different characteristics and properties: that also means that is easy to find in modern cars several kinds of steels, from bake hardenable to UHSS. The selection of the material is done due to the specific purpose of a certain car component. For example, level of strength required, or formability, welding, cost among others, are some of the aspects that are taken into account.

Fig 1.1: Formability Chart: Material Based on Strength and Elongation [16]

Now, when a car goes to the oven for the bake hardening process, it involves the entire car and not just the bake hardenable steel parts of the structure. Bake hardening probably has some effect in the other kind of steels as well, and because of this, it will also have some influence on the In service performance of the structure. The main purpose of this project is to measure the effect of the bake hardening process in AHSS, in this case more specifically Dual Phase 800. The effect of bake hardening DP800 is going to be taken into consideration in applications involving vehicle safety. Also, it is going to be important to take into account that FEA models contained within commercial FEA code used by automotive manufacturers cannot describe the influence of bake hardening and forming history (strain-path dependency). Using FEA simulations (PAM CRASH) the strain-paths and strain-rates observed in a crash event will be determined and this information used to develop experiments to measure the mechanical properties of DP800 under different loading conditions. The project will also assess the capability of current and state-of-art material models to describe the mechanical properties of DP800. At the present time, bake-hardening response is not taken into account in existing material models. A first approach to a similar project has already been done at Corus [20], but the main material was DP600. The conclusion was that the BH effect should not be included in the F2C analyses and mentioned that BH should be modeled as an extra step if required by a customer, using a DP600 material model with strain rates switched off. But, literature shows that the response of BH response on DP600 is lower than the one seen on DP800, and in the previous study, no pre strain levels, nor strain path or strain history, or temperature (or others among a set of possible variables that can affect BH response) were taken into account. In Chapter 2, literature review about the bake hardening process and Dual Phase steels is reported. Then in Chapter 3 the BH response of DP800 is studied on different sets of tensile tests; the link between BH and characteristics like temperature, strain path, rolling direction, pre strain levels and microstructure is studied. Concerning crash performance, BH influence on 3 Point Bending tests and Axial Crash test is measured in Chapter 4. Chapter 5 tries to establish a link between all the tests and their BH response, and finally in Chapter 6 the conclusion of this work is presented with some recommendations for future research.

CHAPTER 2: LITERATURE RESEARCH

2.1 Strain Ageing and Bake Hardening?

Industry will always be looking for stronger and lighter materials. The level of strength of steels depends on the kind of microstructure. It is inversely related to dislocation mobility, and even in high-purity single crystals, there are a number of possible factors that can affect the strength and mechanical performance. The base strength level (and its temperature dependence) is determined by the crystal structure. The purity and method of preparation determine the initial dislocation density and substructure. All this characteristics make the mechanical behavior a complex (and not completely accurate) function of strain, strain rate, temperature, and stress rate [7]. However, still greater complexity is needed in order to be able to maximize the strength and usefulness of steels. Fine grain size, large additions of solute atoms and fine particles are some of the mechanisms utilized to increase the strength. One of these mechanisms is Bake Hardening (BH). In a general way, Bake Hardening is a strengthening mechanism with underlying ageing effects exploited for the controlled diffusion of interstitial atoms in order to block dislocations and thus raise the materials yield strength [2]. In automotive industry, after press forming, painted parts are passed through drying furnaces at temperatures that vary between 140 and 200C where the different layers of coating are baked for 10 to 30 minutes. It has been observed that there is an increase in the strength of the steel parts due to this procedure. Taking into account the new environmental pressures and new technology developments to make cars lighter in order to lower the fuel consumption, and the increasing demand of safety regulations in the automotive industry, effects like BH have to be studied with more detail. Increasing the strength of the base material of the structural components of the car can bring along development of new designs, new sizes, new weights and new features, all heading to increase the reliability and safety of modern cars at the same time that makes it more environmental friendly. It is easier (and cheaper) to get steel with good formability (low yield stress, r value greater than 1, no yield point elongation, high n value) to develop a part and then do some bake hardening, than to trying to form complex shapes with HHS. This shows the importance of the bake hardening effect, and why it should be studied more carefully. Metallurgically speaking, the increase in strength of the steel is due to the diffusion of dissolved interstitial atoms (this is limited to C) to dislocations and grain boundaries and/or ecarbide precipitation. It is assumed that the bake hardening process is a kind of strain ageing caused by segregation of carbon and/or nitrogen atoms to the dislocations generated by press forming. Strain ageing is interpreted as the migration of carbon atoms to free dislocations; it is a type of behavior usually associated with yield point phenomenon (yield point elongation), in which the yield strength of a metal is increased and the ductility decreased on heating at relatively low temperature after cold working. The rate of ageing depends upon the concentration of carbon in solution, the time and the temperature (bake hardening) [1]. Bake hardening of steel is supposed to be controlled by the same mechanisms responsible for strain ageing. Strain ageing can be divided in 3 stages in time; these stages are continuous and successive and show a gradual change from one stage to the next one. Amount of mobile dislocations, concentration of solute and mobility of solute at room temperature are the mechanisms that typically control static strain ageing; the bake hardening process may be more accurately

characterized as occurring in three or two (the first step is too fast and can be ignored) steps. The step process mechanism of bake hardening is schematically illustrated in Figure 2.1:

Fig. 2.1: 3 stage ageing process. Taken from [1].

In the first stage an initial rapid increase in the upper yield point can be seen. This is an effect of the stress-induced rearrangement of the interstitial atoms in the stress field of dislocations. This effect is called Snoek effect, and it can be described like this: Initially, there is a random distribution of interstitials in the matrix. After an applied deformation, the interstitials in the stress field of a dislocation attempt to minimize the strain energy in the region of the dislocation by moving from random to minimum energy site positions. Only a very small amount of interstitials takes part in the Snoek rearrangement process and the time required is less than the time required for a normal interstitial jump. This is a short time process. In fact it is too fast to be measured. This step causes an increase in yield strength, but not the ultimate tensile strength, by pinning the dislocations [1, 15]. In this mechanism, solute nitrogen and carbon diffuse and interact with the strain fields of mobile dislocations and hence form atmospheres around them. These atmospheres are known as Cottrell atmospheres. These atmospheres would constitute regions in which the elastic strain field of the dislocation is partially relaxed, and because of that its energy reduced, so that the solutes would effectively lock the dislocations. This will either increase the stress required to unlock and move dislocations, or immobilize dislocations and thus require generation of new dislocations for subsequent plastic flow. Either mechanism results in an increase in strength and a return of discontinuous yielding. The second mechanism is thought to be more likely, due to the fact that this would explain the return of discontinuous yielding behavior during the tensile test [15]. The last stage of the bake hardening process is the precipitation of e- carbides, process which is supposed to occur at dislocation regions. This will cause an increase in yield strength and ultimate tensile strength. However, with continued solute segregation to dislocation cores, the increased local solute concentration leads to the formation of clusters and then precipitates which can eventually saturate the dislocation sites; when the ageing time is too long, a

10

decrease is observed due to coarsening of the carbides [1]. This over ageing effect was seen on DP, but not in TRIP [2]. The main factors of microstructure influencing the hardening effect are the solute carbon content, ferrite grain size and the amount of pre-strain. The amount of carbon controls the possibility for pinning mobile dislocations. Consequently, with more solute carbon, more dislocations can be pinned and a higher hardening effect can be obtained. As mentioned earlier, strain ageing and subsequently the bake hardening effect, are very sensitive to the amount of dissolved interstitial atoms, primarily carbon and nitrogen. In rimmed steels of the past, the bake hardening phenomenon was mainly due to dissolved nitrogen. Modern automotive steel sheets are aluminum-killed (or more recently IF steel), which means all nitrogen atoms are combined as aluminum nitride. Then, the bake hardenability in low and ultra low carbon steels is exclusively caused by dissolved carbon. To maximize the strength increase associated with bake hardening, it is necessary to have as much free carbon as possible. However, as the amount of solute increases, the resistance to room temperature aging decreases. This can be a none-desirable effect, especially for outer panel body parts. To determine the amount of free carbon that may be used in bake hardenable steel, it is necessary to examine room temperature ageing resistance. It is generally considered that if yield point elongation is 0.2% or less in a uniaxial tensile test, stretcher strain problems will not arise during panel forming. The times and temperatures of room temperature aging depend on the time and temperature at which the steel is stored between production and forming. Resistance to aging at 30C for 90 days is commonly used as a guideline for the upper limit of room temperature aging. The maximum increase in strength was found with higher pre-strain levels. The availability of more nucleation sites on dislocations as the density increases would lead to the formation of smaller particles, which would be expected to increase the yield strength. This is true initially, after which the observed decrease in strength was attributed to the formation of coherent particles that can more easily be cut by dislocations. Decreasing particle size would then represent less of a barrier to dislocation movement. The reason why bake hardenability depends on grain size is not clear, but it is inferred that the influence of dissolved carbon on bake hardenability differs depending on the location of carbon. Different effect of dissolved carbon was reported on the bake hardenability depending on its location, at grain boundary and inside grains. For a given carbon content, the bake hardenability increases with a decrease in grain size and the dependence on grain size increases with an increase in solute carbon. While the explanation of this effect is not complete, data suggests that free carbon located near grain boundaries, which is not detectable by internal friction measurements, has a more profound influence on strength than free carbon located within the grain interior. The effect of ferrite grain refinement on increase of bake hardenability is associated with the location of solute carbon. It is assumed that during cooling, the carbon atoms diffuse to the grain boundaries. Solute carbon positioned at the grain boundaries, so called hidden carbon atoms, cannot be detected by internal friction measurement, but it is supposed that this carbon makes a contribution to the BH effect. The smaller the grains are, the more carbon should be in the grain boundaries because of shorter diffusion paths. Thus, although the same overall solute carbon content can be measured, the contributed carbon content as well as bake hardenability can be higher in case of fine grains as the hidden carbon is more in finer grains. It is important to mention that Nitrogen has a more important roll than carbon in the ageing of iron because it has a higher solubility and diffusion coefficient and produces less complete precipitation during slow cooling. In order to control the ageing, it is usually desirable to lower the amount of carbon and nitrogen in solution by adding elements that will take part of the interstitials out of solutions by forming stable carbides and/or nitrides. The elements used for this purpose are aluminum, vanadium, titanium columbium and boron.

11

The standards for measuring BH behavior are described in prEN10325: a defined pre-strain of 2% are applied to specimens and a subsequent BH treatment is set to 170C and 20 minutes. After this, final tensile testing till fracture delivers the mechanical characteristics of the samples. This reloading is performed in the same direction of the pre strain. BH response is a strengthening mechanism that allows working with low strength formable steels, and then getting a higher strength. It is done on finished parts, which is a major advantage. According to literature, bake-hardening effect can be seen especially in yield strength: very regularly it presents a yield point elongation phenomena, as well as a considerable increase in strength, specially depending on the amount of carbon in the material. This factor influences the adequate measurement of Rp, giving instead values for ReL and ReH , which are respectively the Lower and Upper Yield point values seen as a consequence of the elongation phenomena (Fig. 2.2).

Fig. 2.2: Yield point elongation phenomena. Taken from: www.a-sp.org/database/ viewsec.asp?sec=246

2.2 Bake Hardenable steels

As it was mentioned previously, not all steels have a bake hardening response, and certainly have not the same level of response. Bake hardenable steel is any steel that exhibits a capacity for a significant increase in strength through the combination of work hardening during part formation and strain ageing during a subsequent thermal cycle, such as a paintbaking operation. Any steel with adequate carbon and/or nitrogen in solution to cause strain ageing may be classified as bake hardenable. In general, bake hardenable steels are aluminum-killed steels with a precise amount of aluminum to combine with the nitrogen as Aluminum Nitride (AlN). A combination of relatively low yield strength prior to manufacturing and a high in-part strength after forming and paint baking makes BH steels ideal for applications where dent and palm printing resistance is important. This material can be used in relatively deep draw or stretching operations. Due to the high in-part strength, BH parts are also good candidates for down gageing, which is important for weight reduction efforts.

12

When using BH steel, the amount of strain introduced during the forming process will largely dictate the final strength of the part. Since automotive parts, specifically exposed body panels, have a wide array of designs, there will be a corresponding variation in the amount of strain introduced in these varying geometries. As a result, when using BH steel, it is important to design an adequate amount of strain into a part in order to fully utilize this materials dent resistant characteristics. Besides this, there are other important characteristics of BH steels: Weldability: Low carbon level makes BH steel a good welding candidate. Fatigue Performance: If used properly, BH steels have high yield strength after forming and baking, which means it will have a good resistance to fatigue. Denting: BH steels were designed for dent resistance. Applications: BH materials provide customers with a material that is capable of reducing the amount of dent and dings found on todays cars. These materials have the formability requirements needed to produce most exterior applications. These exterior parts benefit from the work and bake hardening kicks that is experienced during processing. These parts include doors, deck lids, quarter panels, fenders, hoods, and roofs.

As it was mentioned in section 2.1, a defined pre-strain of 2% are applied to specimens and a subsequent BH treatment is set to 170C and 20 minutes. Fig 2.3 shows the behavior of a BH steel after a bake hardening procedure. The diagram helps differentiate the increase of strength in the steel due to work hardening (strain) and the bake hardening.

Fig. 2.3: Measurement of BH response. Taken from http://ussautomotive.com/

BH steels were designed to have high dent resistance; Dent resistant steels are widely used in the automotive industry. Exposed quality exterior automotive body panels such as doors, hoods, deck lids, fenders that require high dent resistance are commonly manufactured using these grades. Selection between non-BH and BH grades depends on formability and final part performance (dent resistance) requirements. Depending upon the amount of strain induced in the part during stamping, non BH grades can achieve similar final part strength as compared to corresponding BH grades. Designing and manufacturing of parts should utilize the unique properties of individual grades.

13

Most dent resistant grades are low carbon grades, with several having ultra-low carbon compositions. Non BH grades are non-ageing and their properties do not degrade over time. In contrast, BH grades have carbon/nitrogen in solution to develop the bake hardening response. The ageing phenomenon in BH grades is controlled through a combination of composition and processing. These steels are strengthened with Manganese, Phosphorus, Titanium and Niobium. In addition to outer body -panels, all dent resistant steels can also be used for structural applications. In fact, the higher strength (280 and 300MPa grades) can be used to replace micro-alloyed HSLA 300 and 350 type products, respectively. Typical mechanical properties of a BH steel are yield strength between 180-280MPa, tensile strength between 300-400MPa, total elongation range 30-40%, r between 1.0-1.6 and n-value 0.16-0.2.

2.3 Dual Phase High Strength Steels

Different types of steels are commonly used in the body structure of a car. From low carbon to UHSS, depending on the specific body part and requirement. When a car is subjected to paint baking, there is a bake hardening response of the steels in the structure. Plenty of research has been done on bake hardening response, but only on the ones that have low carbon weight percentage. Bake hardening response of AHSS has to be studied further, and on this project more specifically, on Dual Phase High Strength Steel with UTS of 800MPa (DP800). One of the first commercial dual phase products was marketed in the USA. It was based on a pearlite-reduced vanadium micro alloyed strip steel, commonly encountered in the 1970's, plus an intercritical annealing process. The steel exhibited the tensile strength of 650MPa, while having yield strength of 350MPa and total elongation of more than 27%. The main application of this dual phase steel was in bumper reinforcements, but it was also used in other automotive parts including passenger car wheels [42]. Dual phase (DP) steels are quickly becoming one of the most popular and versatile materials in today's automotive industry. Currently these steels are most commonly used in structural applications where they have replaced more conventional HSLA steels. They offer a great opportunity for part weight reduction. The improved formability, capacity to absorb crash energy, and ability to resist fatigue has driven this substitution. Today's applications include front and rear rails, crush cans, rocker reinforcements, B/C pillar reinforcements, back panels, cross members, bumpers, and door intrusion beams. Recently dual phase steels are gaining popularity in automotive closures. Dual phase steels present higher formability than microalloyed steels of comparable strength. Because of growing popularity of DP steels, companies like Corus are encouraged to conduct projects in this area. In general, dual phase steels are characterized by a matrix of fine ferrite containing small islands of martensite (Fig. 2.4). The hard martensite particles provide substantial strengthening while the ductile ferrite matrix gives good formability. In some cases, there is a possible addition of bainite.

14

Fig. 2.4: Optical image of typical Dual Phase microstructure. Taken from [41]

The martensite/ferrite dual phase mixture acts like a particle-reinforced composite. The tensile strength of the composite can be approximated by a simple rule of mixtures:

DP = VM M + VF F

[41]

where, V is the volume fraction of the phases and s ? is tensile strength. The subscripts DP, F and M indicate the composite dual phase structure and the ferrite and martensite phases respectively. The micro structural characteristic of the dual phase has on influence on the stress-strain curve. The yield strength is determined by the onset of plastic flow in the ferrite phase. At this stage the hard phase is still in the elastic region. With higher applied stress the material exhibits a high work hardening behavior according to a rule for mixtures for two phase microstructures. The strain distribution in the two phases is different, such that in the soft phase the strain and in the hard phase the stress is above the mean value of the composite. This remains even when the hard phase becomes plastic if in a later stage of deformation [42]. As martensite is the stronger phase, increasing the volume fraction increases the strength of the dual phase material. Unfortunately increasing the martensite content reduces ductility and for this reason the volume fraction is usually restricted in the range 10 to 20%, where the martensite remains as discrete isolated islands in an interconnected ferrite matrix. Further, to maintain high ductility in these steels it is desirable not only to limit the martensite fraction, but also to restrict the carbon content of this phase. At lower carbon contents (<0.4wt%) lath martensite forms, rather than high carbon twinned martensite. Lath martensite is desirable as it will deform to a limited degree and this helps to delay void formation at martensite/ferrite interfaces when the material is strained. Therefore the carbon content of dual phase steels is usually limited to ~0.1wt% maximum. For these typical compositions, with martensite volume fraction of ~15%, dual phase steels give tensile strengths in the range 550 to 650MPa in the as-rolled condition. To produce a dual phase microstructure, the equilibrium pearlite phase needs to be eliminated, with austenite being encouraged to form martensite by rapid cooling. The simplest method for producing dual phase microstructures is to anneal a ferrite/pearlite steel in the intercritical temperature range. The annealing temperature is controlled within the ferrite plus austenite two-phase region, such that much of the room temperature ferrite phase remains. The pearlite reverts to carbon rich austenite. When the steel is then quenched from the annealing temperature the austenite proportion is sufficiently hardenable to transform to martensite. By controlling the steel chemistry and annealing temperature, both the phase fractions and the properties of martensite can be adjusted.

15

The need for a post rolling heat treatment adds significantly to costs, so methods have been developed to produce dual phase microstructures by direct rolling and cooling. This has been achieved through a combination of process control and new alloy design. The new chemistries are low in carbon and rich in hardenability enhancing elements. The low carbon content encourages ferrite to form at temperatures above that necessary for pearlite formation. Other additions delay the pearlite transformation and therefore increase the potential to form martensite. The post rolling cooling process is controlled to allow the desired volume fraction of ferrite to form before rapid cooling. The later stage-cooling rate must be sufficient to quench the remaining austenite to martensite. First alloy concepts for strip to avoid the intercritical heat treatment were based on Mn, Si, Cr and Mo alloyed steels. Owing to the rather high alloy content, these steels were expensive and alternative grades without molybdenum have also been developed. The major application remained to be the automotive wheel and a typical alloy design for such hot strip material with a tensile strength level over 550MPa is: 0.08%C, 0.50%Mn, 0.30%Si, 0.50%Cr and 0.07%P [42]. The continuous annealing of cold rolled sheet has become more common in recent years and these facilities are very suitable for producing a dual phase microstructure by the intercritical annealing process. Figure 2.5 describes the processing route and the resulting mechanical properties in the production of a side impact beam for a passenger car using dual phase sheet steel. The ferrite plus martensite dual phase sheet exhibits a high work hardening potential and therefore a remarkable yield strength increase after the tube forming process. A further strength increase is related to bake hardening obtained in the standard painting process of the car. In order to guarantee a fine grained microstructure these sheet steels also contain a niobium micro alloy addition.

Fig.2.5: Yield Strength in the production of a side impact beam for a passenger car using dual phase sheet steel. Taken from [42]

Companies like Arcelor, TKS and SSAB have started to do some research on the influence of bake hardening on DP steels. There results state that steels like DP800 can achieve an increase of 300 to 400MPa in yield strength as a consequence of Bake Hardening and Work

16

(strain) Hardening [1,8,14]. From these results, approximately 80 to 100Mpa are BH response. These same references have also obtained similar BH plus WH response on different AHSS, being the BH response of the Complex Phase steel specially high (almost 200MPa). These characteristics should be taken into account to obtain benefits from the material in terms of formability and energy absorption capability, for example by reaching higher work hardening on forming processes like hydro forming, which involves high pre strain levels. DP has showed strain rate sensitivity, which decreases with increasing strength; it also has very high work hardening behavior until 5%, after which it decreases seriously and the curve turns almost flat [29].

17

CHAPTER 3: TENSILE TESTING

3.1 Introduction

The literature research [2, 3, 8] claimed that the BH response of an AHSS as DP800 could be something in the range of 300 to 400MPa increase in yield strength; this strength as well was including the increase in strength due to work hardening of the material. This is a considerable amount of strengthening, but it is associated directly with reduction in the maximum elongation, therefore in reduction of formability and ductility too. The characterization of a material always starts with a standard tensile test. In this chapter, different configurations for tensile samples are going to be discussed, in order to show how the BH of the samples affect the tensile response of the material. As described in Chapter 2.1, the standard test for measuring BH response includes pre strain, temperature and path dependency; these characteristics and their relation with BH are going to be evaluated as well. Previous studies also suggest that the BH response of steel can vary depending on the grade of AHSS, and its strength level. Although the research is going to be focused on DP800, it is considered helpful to compare the response of not only DP, but also some other AHSS. Not only because the different microstructure can help explain the BH response difference, but perhaps the influence of BH is more significant for other kinds of steel, and can be a subject of future research. The other AHSS that were subjected to the tensile test with BH were DP1000, TRIP800, TRIP600, CP800 and another DP800. The results of these tests are included in appendix A. During the development of the project, it was discovered that ageing at room temperature was a characteristic that had to be taken into account; a first approach to measure this effect was done, but further work is going to be recommended in order to get a better relation between pre strain level and ageing time.

3.2 Experimental Details

3.2.1 Materials As it was mentioned before, a Dual Phase steel is going to be the main subject of the study; more specifically, DP800. This one was provided from an external producer, SSAB, and its general chemical composition and mechanical properties are described in table 3.1 and 3.2.

18

Table 3.1: Chemical composition of DP800

C Mn Si Alzo N P S Nb V Ti

[10-3 %] [10-3 %] [10-3 %] [10-3 %] [ppm] [10-3 %] [10-3 %] [10-3 %] [10-3 %] [10-3 %]

149 1830 207 29 54 13 2 12 8 2

Table 3.2: Mechanical properties of DP800

Rp0.2 Rm A80 r

[MPa] [MPa] [%]

672 869 15.5 0.93

3.2.2 Specimens Two different kind of tensile samples were prepared for the tests: the standard size sample, and a small size sample. The small size sample was necessary in order to be able to test specimens taken from already formed boxes (hydro formed boxes); the small size samples were also useful to be able to pre strain in one direction and reload in another direction: doing this on standard size samples would require much more material. The geometry of the samples can be seen on figure 3.1.

Fig. 3.1a: Small tensile sample (measurements in mm)

Fig. 3.2a: Large tensile sample (measurements in mm)

Standard Size: For the standard size samples, in case BH was needed, it was done on an oil bath, at the stated temperature (170 or 220C) for a period of 20 minutes. Also, when the samples had to be pre strained, reloading was done in the same direction of the pre strain. Furthermore, results from the same kind of tests and size of other DP800, DP1000, TRIP600, TRIP800 and CP800 are going to be used in this project, and can be seen in appendix A. The standard size samples were cut from the strip product in the direction needed: Longitudinal, Diagonal and Transverse (L, S or D), then pre strained (0, 2, 5 or 10%) and finally BH (No BH, 170 or 220C) for the tests. Each of these tests was done on triplicate. Small Size: As for the small samples, these ones were used basically to be able to study the strain path dependency; pre strain and reload on different directions. The BH was done in an oven (air heated). The pre strain for these samples was always done in the rolling direction.

19

The small samples were used to analyze the strain path dependency in relation with BH temperature and pre strain level (uniaxial tensile test). These one were prepared the following way: 36 rectangular samples of DP800 were cut (figure 3.2a). The 350mm side was in the rolling direction of the material. Those samples were pre strained in the rolling direction to 0, 2, 5 and 10% (9 samples at each pre strain level). Because the ratio of the dimensions is 90:350, the strain generated can be assumed as uniaxial. Then a set of samples (3 rectangles of each pre strain condition) was Bake Hardened at 170 and another at 220C. Also, one set of samples was left without any BH at all (3 of each pre strain level). From each of the already pre-strained and baked rectangular samples, 4 small size bone probes were taken. These small probes were cut in the rolling direction, 45 to RD and 90 to RD. Then all the samples were reloaded until fracture. Figure 3.2a shows the way the samples were cut. Each of the conditions for the test was done in duplicate. Between the pre strain step and the BH step for the small samples, there was a period of approximately one month; and between the BH and the time when the samples were cut in shape to be tested, there was another period of approximately one month.

0, 2, 5 & 10% PS

RD

L S AR, BH170 & BH220 (x4) D

Fig. 3.2a: Preparation of small size tensile samples for uniaxial test

20

Weld

RD 1 3 2 AR, BH170 & BH220

D (x4)

L (x4)

Fig. 3.2b: Preparation of small size tensile samples from hydro formed boxes

A more detailed description of the samples tested in the tension test can be seen on table 3.3a. Comment A refers to a difference on the two samples tested: One was reloaded immediately after pre stain, while the other one was reloaded two weeks after pre strain. The details of this will be discussed later. Furthermore, another set of small tensile samples was tested: In order to measure the influence of BH on a component, small tensile samples were taken from hydro-formed boxes. 2 hydro-formed boxes (one with initial tube diameter of 61mm and another one of 70mm) were taken, and tensile samples were cut as shown in figure 3.2b. The L direction of the box is the rolling direction of the material as well. From each of the faces of the box (except the face that has the weld, no samples can be taken from this side), 4 samples were taken. From face 1 and 2 the samples were taken on the transverse direction (D), while from face 3 the samples were taken in longitudinal direction (L); a total of 12 samples were taken form Box 71mm (8 D direction and 4 L direction) and 11 samples were taken from Box 60mm (8 D direction, 3 L direction). The selection of face and direction in which the samples were cut was random. As the pre strain level was expected to be higher on D direction due to the way the hydro forming takes place, it was decided that two faces would provide D direction samples. After cutting the samples from the boxes, they were bake hardened. Table 3.3b has a description of how many tensile samples from each box and each direction were BH at a determined temperature. On one of the 60mm tube hydro-formed boxes, only 3 samples could be taken from face 3, so there is no sample for BH 220C L direction. All the samples were taken from bottom part (undeformed section) of boxes that had been previously crushed on quasi-static axial tests (0.33mm/s).

21

Table 3.3a: Summary of tensile test samples. A = Immediate and 2-week reloading

PRE STRAIN DIR 0 2 5 10 L S D #

X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

PRE STRAIN DIR 0 2 5 10 L S D #

X X X X X X X X X X X X 3 3 3 2 2 2

COMMENTS

No BH

LARGE

SMALL

X X X

No BH

A A A

X X

X X X X X X X X X X X X X X X X X X X X X X X X X X X X

3 3 2 3 2 3 2 3 3 3 2 3 2 3 2 3

170C

170C
X X

220C

Table 3.3b: Description of tensile samples from hydro formed boxes

BOX 71mm 71mm 71mm 71mm 71mm 71mm 60mm 60mm 60mm 60mm 60mm 3.2.3 Test setup

Direction L L L D D D L L D D D

220C

BH (C) NO 170 220 NO 170 220 NO 170 NO 170 220

Number of samples 2 1 1 2 2 2 2 1 2 2 2

For the standard size samples, the tensile machine pulls the sample apart while measuring extension (mm) and width (mm) with the help of an extensometer; besides, the force (N) and crosshead displacement (mm) and time (s) are also measured. The data file obtained from this test (MUS-file) takes this information, and with a manual approximation of elasticity modulus based on the force displacement curve, it calculates engineering strain, engineering stress, and gives values for parameters as Rp, Rm, Ae, Ag, n and r among others. All the tests done on the standard size samples were done at IJTC, 300kN MTS hydraulic machine. The strain rate in this test varies from the elastic to the plastic zone: the elastic range is done at 30MPa/s, which equals to approximately 0.4mm/s, and the plastic range is done at 1mm/s. This setup is done according to standard norm ISO 6892. From the small tensile sample tests, the output values are time (s), position of the crosshead (mm), force (kN) and strain gauge elongation (mm). In this case the output file is a csv -file, and it doesnt calculate the engineering strain and stress nor gives the values of the other

22

parameters. This was done manually. These samples were tested at IJTC, on a 50kN MTS hydraulic machine, at a speed of 0.33mm/sec. 3.2.4 Data files As mentioned before, the output file for the standard sample test already gives the engineering strain and stress values. The small samples output file does not, so for these cases the engineering strain and stress had to be calculated. The following equations were used for this purpose:

eng = eng
where

lext 10.5 F = w t

l ext (mm) is the extension of the extensometer, 10.5 (mm) is the fixed initial value of

length of the extensometer, F is the load (KN), and w and t are the width and thickness of the sample (both in mm). After doing this, all the sample data have values of engineering stress and strain. For the engineering stress the original cross section of the tensile specimen is used. However, during the tensile test the cross section of the specimen is reduced due to contraction. Therefore, if an accurate measure of the material performance is needed, that contraction should be taken into account. The stress that takes the current cross section into account is called the true stress. Because the cross section is usually not measured, the calculation of stress is done using other information. It is assumed that the volume of the specimen does not change (true for plastic deformation) and that the cross section remains constant over the measured length. So the current cross section can be related to the change in length of the specimen. After the start of necking the assumption that the cross section is constant over the measured length is no longer valid. This means that the equation for true stress may only be used until necking starts. The general equation for the true stress remains valid, but then a measurement of the cross section at the location of the neck is needed. Besides a true value for the stress, a true value for the strain is also needed. The reason is that if an intermittent tensile test were to be done (loading unloading loading of the specimen), the test would be stopped after a certain amount of strain. At reloading, a new value for initial length should be taken, and the strains of the first and second part cannot be added if engineering values are being used. Similar to the stress, the calculation of the strain based on the current length would require a differential equation: the increase in true strain is the increase in the length divided by the current length. In order to calculate the true strain based on the data from the extensometer, it is assumed that the change in length along the gauge length is constant. This means that when necking occurs and locally the strains (meaning the change in length of the extensometer) start to vary too much, the equation cant be used anymore. The engineering stress strain data obtained from the output files was transformed to true plastics strain true stress through these equations:

true = ln(1 + eng ) true = eng (1 + eng )

23

Finally, true plastic strain was obtained:

engPlastic = eng

eng E

truePlastic = ln(1 + engPlastic )

With the true values calculated, it is possible to measure the yield stress (Rp is calculated as true stress at 0.002 {0.2%} true plastic strain) and plastic energy. The plastic energy is calculated as the area under the true stress true plastic strain curve. Besides that, the new data is required to create the input file for FE analysis later on. The ultimate tensile stress value is calculated as the stress at maximum force value in the engineering stress strain curve. Figure 3.3 gives an idea what happens when the engineering stress strain curve is converted to true stress true strain. The engineering curve is shown up to the point of fracture. The red curve shows the true stress true strain curve. Where the red curve ends and the grey curve starts is the point where necking starts (which is the maximum in force displacement curve). The grey curve shown here is the result of converting all the engineering data to true data; but as the equations are only valid until the sample starts necking, the grey curve does not have any physical meaning. What the true stress true strain curve looks like beyond this point cant be determined from the engineering curve. More data is needed about the cross section of the specimen at the neck.

450 400
True stress - True strain ? ? ? ? ? Engineering stress - strain

350 Stress [MPa] 300 250 200 150 100 50 0 0 10 20 Strain [%]
Fig. 3.3: Engineering Curve vs True Curve

30

40

50

3.3 Results
In this chapter, a general overview of the results of the tensile tests is going to be exposed. The tensile samples tested have been influenced by factors as BH temperature, rolling direction, work hardening (pre strain level), time (strain ageing) and strain direction (strain path). These parameters are going to be analyzed separately.

24

3.3.1 BH response with no pre strain The standard test to measure BH response is described on 2.1; it includes 2% uniaxial pre strain prior to the BH and reloading. But it is still important to analyze the effect of BH without pre strain, due to the fact that in an actual vehicle component, there are regions where there is hardly any pre strain after forming operations. Further more, after the material is temper rolled, small strain effects of around 1% are generated. These strain levels are already generating dislocations which can hold the diffused atoms resulting in a strengthening effect. Figures 3.4 and 3.5, and table 3.4 show the results of the BH without influence of pre strain for L and D directions. The results presented in these figures were done on standard (LARGE) size samples.
1000 900 800 700 True Stress (Mpa) 600 500 400 300 200 100 0 0 0,02 0,04 0,06 True Strain (-) 0,08 0,1 0,12 L L BH170 L BH220

Fig. 3.4: BH response rolling direction without Pre strain

1000 900 800 700
True Stress (Mpa)

600 500 400 300 200 100 0 0 0,02 0,04 0,06 0,08 0,1
True Strain (-)

D D BH170 D BH 220

Fig. 3.5: BH response 90 to rolling direction without Pre strain

25

Table 3.4: Yield strength and UTS values for standard samples with and without BH on L and D directions

L D

Rp (Mpa) AR 662 696

Rp (Mpa) BH170 692 726

ReL (Mpa) BH220 750 770

ReH (Mpa) BH220 770 790

Rm (MPa) AR 869 880

Rm (MPa) BH170 869 882

Rm (MPa) BH220 878 891

From the previous 2 graphs and table 3.4, three aspects are important to mention: the first one is that the as received material presents different behavior on L and D direction (more than 30MPa difference). Additionally, the increase on strength throughout the curves due to BH170 is not significantly noticeable; even though on D is more clear the beginning of yield point elongation (probably due to the fact that the strength level is higher in this direction). Finally, the most evident effect, on both directions yield elongation phenomena can be seen after BH at 220C. In general, the graphs are very similar, with a small variation on the strength level. Concerning the elongation, it can be said that the BH had no significant influence on Ag or A80; all the samples (either direction and BH temperature) had an approximate Ag of 10% (1% scatter) and A80 of 14% (3% scatter); there was no reduction in elongation as a consequence of the strengthening of the material. 3.3.2 Ageing at room temperature BH effect has been generally described as an accelerated strain ageing; the strengthening mechanism consists of the diffusion of atoms into dislocations. The only difference is that the baking temperature accelerates the diffusion of these atoms. Theoretically, with enough time, the ageing steel will have the same response at room temperature as it would have after a BH process. Ageing at room temperature is a characteristic that has to be taken into account due to the fact that it can affect the mechanical properties of the steel. Usually, car manufactures demand a 3 -month period warranty of non-ageing effect on their steel (room temperature storage). Unfortunately, the room temperature ageing effect was not considered at the beginning of this project (time between pre strain and reload was not considered an issue), so an adequate schedule for testing this effect was not possible due to the projects deadline. Only a short period of 2 weeks ageing was measured during this project. This test consisted of pre straining samples in the rolling direction at 0, 2, 5 and 10%, and posterior reloading until fracture on the same direction. It was done on standard size samples (See comment A on table 3.3). One set of samples was reloaded immediately after the pre strain step; the other set was stored at room temperature for two weeks and then reloaded until fracture. The results of this test are shown on figure 3.5, and a detail of the same graph can be seen on figure 3.6:

26

1000

800
True Stress (MPa)

AR 600 2% Inmediate 5% Inmediate 400 10% Inmediate 2% 2 weeks 200 5% 2 weeks 10% 2 weeks

0 0 0,02 0,04 0,06 0,08 0,1

True Strain (-)
Fig.3.5: 2-week room temperature ageing effect on DP800

0,12

0,14

0,16

0,18

0,2

Fig. 3.6: Detail figure 3.5

From figure 3.5 and 3.6, it can be clearly noticed that there is in fact a marked room temperature ageing effect; the effect increases with increasing pre strain level, which seems logical due to dislocation density. At 2% the increase on Rp after 2 week ageing is around 5MPa, while at 5% is approximately 20MPa and at 10% is almost 30MPa. There was another side effect to analyze after seeing the results of the ageing test. The samples that were pre strained 5 and 10% showed a very drastic dropping effect of the curve just after the elastic region; on figure 3.6, a red ellipse highlights this effect on both samples at 5% pre strain. At first, the dropping effect of the graph was thought to be a consequence of yield elongation phenomena. Then, the curves showed a sudden increase effect (highlighted by the blue ellipse).

27

After analyzing carefully the samples data, it was discovered that both the dropping and the increase effect of the curves (red and blue ellipses in figure 3.6) were not really a yield elongation phenomena. It was found that this effect was a consequence of strain rate change on the test setup. The standard automatic program setup for the tensile test has a variation on the strain rate of the test from the elastic region to the plastic region. For HSS, the elastic region has a speed of 30MPa/s (0.4mm/s), and as soon as it passes to the plastic region it changes to 1mm/sec. The program automatically changes the speed of the test as soon as it detects what it considers as yield point. But, there is also an additional change. The standard also has a setup for the materials that present yield point elongation. When the programs detects what it thinks is a yield elongation, it goes from the speed of the elastic region to an intermediate speed, and as soon as it detects the end of the elongation phenomena, it changes to the pre determined speed of the plastic region. The p roblem is that sometimes the program can mistakenly think that the material is presenting yield elongation phenomena. This can be because in some cases it is difficult to determine properly the elasticity modulus and the Rp of a sample, especially if it is dealing with materials like DP. In figure 3.7 the strain rate (left axis) and the true stress (right axis) are graphed as functions of true plastic strain of the 2 -week room temperature aged sample (5% pre strain). The points where the strain stress curve shows the drop down and the pull up are the same where the strain rate has a plateau; that plateau tries to take into account a non-existing yield elongation phenomena. To describe the real behavior of the material, the strain rate should have kept the same increase path as the one that it can be seen after the plateau. So that means that the stress strain curve should not have the drop and the increase as well. Without the automatic change in velocity of the test, the real material behavior during the plateau would approximately be the one described by the interrupted lines shown on the graph.

Fig. 3.7:Effect of automatic velocity change due to standard setup

From the graphs presented in this section it is important to mention that the complete set of data was graphed (beyond UTS inclusive) to help getting a more clear view of the ageing effect. Also, the as received material (no pre strain) was tested with six months or even one year time difference, and no variation of the results on the tensile tests was noticed. During the testing, another fact showing the importance of ageing at room temperature was discovered: For L and D directions, when the samples were not pre strained (0%) and then

28

baked, the stress strain curves were the same for the standard size and small size samples; still there was a difference seen after the UTS, where the small sample presented a larger elongation (figure 3.8a). This is a normal result due to the fact that a fixed amount of elongation after necking would be in percentage larger for the small samples than the standard size samples. This was expected as the mechanical properties of the material should be the same, but it is important to clarify this point because of the differences in the testing setup of both sets of samples: For the small ones, there is a fixed velocity through the whole test, but the large ones run as stated in the standard test. BH for the small samples was done by means of hot air in an oven, while the large ones were done on an oil bath. The machine in which the tests were performed is different

Still, besides the difference between on the total elongation between the small and large samples, another distinction was found between the their engineering stress strain curves: when the samples had been pre strained at any level, then BH and finally reloaded on L direction until fracture, it was seen that there was a difference in the strength curve for both types of samples. This effect is shown on figure 3.8b:

1000 900 800 Eng Stress (MPa) 700 600 500 400 300 200 100 0 0 0,05 0,1 0,15 Eng Strain (mm/mm) 0,2 0,25 0,3 AR Large BH170 Large BH220 Large AR Small BH170 Small BH220 Small

Fig. 3.8a: Small and Large samples L direction with BH (no pre strain)

29

1050 1000 Eng Stress (MPa) 950 900 850 800 750 700 0 0,05 0,1 0,15 0,2

BH170 small BH220 small BH170 LARGE BH220 LARGE

0,25

Eng Strain (mm/mm)

Fig. 3.8b Small and Large samples L direction with 2% pre strain and BH

It is clear from the previous graph that the behavior of the material is very similar for both conditions; the difference is a shifting of the curves from the large samples to the small samples of around 45MPa. This behavior was seen as well on the samples that had 5 and 10% pre strain (L direction, BH at both temperatures). As mentioned previously, there were some differences between the testing of the small and the standard size samples. But this effect couldnt be explained because of the differences mentioned before, because the results for the test without pre strain were basically identical until UTS (figure 3.8a). Then the only difference left to analyze is time. Between the pre strain and final reload, the small samples had a period of approximately two months, while the large samples at most had one day difference. This confirms the very significant room temperature ageing effect described on the previous section.

3.3.3 Uniaxial Pre strain tensile test As mentioned previously, in real components a whole range of pre strain levels in different directions can be seen on every sample. This is mainly a consequence of the strains generated in the forming operations to obtain the part. Then, the influence of pre strain level on BH response should be measured. Not only the standard BH response measurement has to include a previous uniaxial pre strain, but the BH response stated in different literature sources included the effect of work hardening as well. A first approach to a rigorous study of influence of pre strain can be done by means of a uniaxial test. Pre strain level and direction, and their influence on BH response were measured. The samples used were the small ones described I section 3.2.2. The results for the uniaxial pre strain test after reloading on L, S and D are shown on figures 3.9, 3.10 and 3.11.

30

1200 1000
True Stress (MPa)
AR BH170

800 600 400 200 0 0 0,02 0,04 0,06 0,08 0,1

BH220 2% AR 2% BH170 2% BH220 5% AR 5% BH170 5% BH220 10% AR 10% BH170 10% BH220

0,12

0,14

True Strain (-)

Fig. 3.9: Pre strain RD, reload L direction

1200

1000
AR BH170 BH220 2% AR

800
True Stress (Mpa)

600

2% BH170 2% BH220 5% AR

400

5% BH170 5% BH220

200

10% AR 10% BH170 10% BH220

0 0 0,02 0,04 0,06 0,08 0,1 0,12 0,14

True Strain (-)
Fig. 3.10: Pre strain RD, reload S direction

31

1200

1000

True Stress (Mpa)

800

AR BH170 BH220

600

2% AR 2% BH170 2% BH220

400

5% AR 5% BH170 5% BH220

200

10% AR 10% BH170 10% BH220

0 0 0,02 0,04 0,06

True Strain (-)

0,08

0,1

0,12

0,14

Fig. 3.11: Pre strain RD, reload D direction

The first thing that is clear after comparing figures 3.9, 3.10 and 3.11 is that changing the direction of the reloading after having pre strained all the samples on the RD has a drastic influence on the results of the tensile tests. When reloading was done on the same direction of the pre strain (L), the total uniform elongation (elongation until UTS) was very small compared to the one obtained on the other two directions, having an increasing uniform elongation from 0 to 90 to RD. This result seems logical, as the immobile dislocations generated during pre strain have been locked in the direction of the pre strain; when reloaded in the same direction the material presents strengthening and it reaches its maximum strength faster because all the dislocations are on the same slip plane. When reloaded in other direction the dislocation is not locked in the latter one, so it is still free to move, and this effect would be expected to be clearer as the reloading direction is farther from the pre strain direction. Strain Path In order to try to explain more effects of the strain path dependency, figures 3.12a and 3.12b are shown; they are details of fig 3.9 and 3.11 respectively.

32

1100 1050 1000 True Stress (MPa) 950 900 850 800 750 700 0,015
AR 2% AR 2% BH170 2% BH220 5% AR 5% BH170 5% BH220

0,035

0,055 True Strain (-)

0,075

0,095

Fig. 3.12a: Stain path effect of uniaxial test: Reloading direction L

1100
AR

1050 1000
True Stress (Mpa)

2% AR 2% BH170 2% BH220 5% AR 5% BH170 5% BH220

950 900 850 800 750 700 0,015

0,035

0,055

True Strain (-)

0,075

0,095

Fig. 3.12b: Stain path effect of uniaxial test: Reloading direction D

In figure 3.12a it can be seen that the ageing effect described on section 3.3.2 happens here as well; as the samples were reloaded on the same direction of the pre strain, it would be expected that the curves of 2% AR and 5% AR lie on top of the curve of the AR material. But instead of that, a clear increase in the curve can be seen. Taking into consideration the time

33

that passed between the pre strain and the reload of the samples (approximately two months), it was concluded that ageing at room temperature was the factor that caused this increase. Figure 3.12b shows the response of the material when reloaded in a different direction of the pre strain. All the curves are much smoother, showing a continuous yield point characteristic. This effect was also seen when the reload was done at 45 to RD, but it is more pronounced when the reload is done at 90 to RD. Even more, on figures 3.13a and b, it is clear that the material seems to soften as it is reloaded in a direction different from the pre strain direction; in this two graphs, yield point of the samples reloaded in D direction are much lower than the one seen on the samples reloaded in L direction. This softening behavior seen on these two cases can be attributed to Bauschinger effect. Strictly speaking, Bauschinger effect is defined as the lowering of the yield stress when deformation in one direction is followed by deformation in the opposite direction. As the slip direction is reversed, dislocations of the opposite sign could be created at the same sources that produced the dislocation responsible for the strain hardening in the pre strain step. Both sets of dislocations basically cancel each other, and the net effect of this is softening of the lattice. This explains the softening behavior on the curve [7]. Although Bauschinger effect is defined for complete reversal of the slip direction (this would mean compression after first straining), an increasing trend of this effect was seen as the samples were reloaded on 45 and then 90 to RD. This can be explained as during the tensile test there is contraction in the direction perpendicular to the direction of the test; then, the reloading of the samples on D direction until fracture would imply strain reversal. Besides, it is known that all metals exhibit Bauschinger effect. Figures 3.13a and b help understanding the conclusion made previously by graphing specimens with different strain path history for a fixed BH temperature.

1200 1000 True Stress (Mpa) 800

AR L

600 400 200 0 0 0,02 0,04 0,06 0,08 True Strain (-) 0,1

2% BH170 L 5% BH170 L 10% BH170 L AR D 2% BH170 D 5% BH170 D 10% BH170 D

0,12

0,14

Fig. 3.13a: Strain path change for BH170

34

1200 1000 True Stress (Mpa) 800

AR L

600 400 200 0 0 0,02 0,04 0,06 0,08 True Strain (-) 0,1 0,12

2% BH220 L 5% BH220 L 10% BH220 L AR D 2% BH220 D 5% BH220 D 10% BH220 D

0,14

Fig. 3.13b: Strain path change for BH220

In figures 3.13a and b it can be seen clearly the drastic reduction in uniform elongation when reloading in the same direction of the pre strain. Also the softening behavior (decrease in Rp) described previously is evident for the samples reloaded in D direction. 3.3.4 Overview uniaxial test Table 3.5 shows the calculated Rp, Rm and Plastic energy from the complete set of results of the uniaxial pre strain tests. These parameters are calculated as described on chapter 3.2.4:
Table 3.5: Overview uniaxial tensile test results

Sample AR 0% 0 AR 0% 45 AR 0% 90 AR 2% 0 AR 2% 45 AR 2% 90 AR 5% 0 AR 5% 45 AR 5% 90 AR 10% 0 AR 10% 45 AR 10% 90 BH170 0% 0 BH170 0% 45 BH170 0% 90 BH170 2% 0 BH170 2% 45 BH170 2% 90

Rp (Mpa) 664 677 695 840 729 633 920 785 665 996 824 678 682 684 705 887 744 677

Rm (Mpa) 869 858 878 898 880 895 931 903 908 1006 956 951 870 858 879 909 885 898

Plastic Xx Sample Energy () Xx 89 85 88 70 74 76 39 46 49 6 26 35 90 85 85 61 17 68 Xx BH170 5% 0 Xx BH170 5% 45 Xx BH170 5% 90 Xx BH170 10% 0 Xx BH170 10% 45 Xx BH170 10% 90 Xx BH220 0% 0 Xx BH220 0% 45 Xx BH220 0% 90 Xx BH220 2% 0 Xx BH220 2% 45 Xx BH220 2% 90 Xx BH220 5% 0 Xx BH220 5% 45 Xx BH220 5% 90 Xx BH220 10% 0 Xx BH220 10% 45 Xx BH220 10% 90

Rp Rm Plastic (Mpa) (Mpa) Energy () 951 786 699 1024 879 791 732 753 765 925 782 747 997 861 755 1070 933 848 966 916 926 1044 996 996 878 878 889 939 914 936 995 972 970 1073 1044 1045 6 28 36 7 22 25 88 81 77 56 54 54 1 41 40 6 29 34

35

The results for reloading at 0 to RD show in general an expected trend: increase in Rp and decrease in plastic energy with increasing BH temperature and amount of pre strain. For Rp, an estimated increase of 30 to 40MPa as a consequence of BH at 170C independent of pre strain level, and between 70 and 85MPa when BH at 220C. Still, is has to be taken into account that all the Rp values showed on table 3.5 were calculated as defined in 3.2.5, even when yield point elongation was seen on the specimens. Concerning the energy reduction, it is important to emphasize some aspects: from figures 3.8 to 3.10 it is evident that there is a reduction in the length of the curves, and could be assumed at first as a drastic decrease on energy absorption capability. But, it has to be kept in mind that those graphs only show the data until UTS; that doesnt mean that the samples fractured at that point, but that their uniform elongation finished in that point. After that, the samples presented necking (not uniform elongation), and, as discussed in section 3.2.4, that data is removed from the results. The apparent reduction in ductility is particularly clear in the samples that were reloaded in the same direction as the pre strain (L). The engineering stress strain data from the uniaxial test on L direction was taken in order to see the effect of BH on total elongation, and is presented of figures 3.14 to 3.17:
1200

1000

Eng Stress (MPa)

800

600
AR BH170 BH220

400

200

0 0 0,05 0,1 0,15 Eng Strain (mm/mm)

Fig. 3.14: Engineering Stress Strain curve in L direction with BH and NO pre strain

0,2

0,25

0,3

36

1200

1000

Eng Stress (MPa)

800

600

400

AR BH170 BH220

200

0 0 0,05 0,1 0,15 Eng Strain (mm/mm)

Fig. 3.15: Engineering Stress Strain curve in L direction with BH and 2% pre strain

0,2

0,25

0,3

1200

1000

Eng Stress (MPa)

800

600
AR BH170 BH220

400

200

0 0 0,05 0,1 0,15 Eng Strain (mm/mm)

Fig. 3.16: Engineering Stress Strain curve in L direction with BH and 5% pre strain

0,2

0,25

0,3

37

1200

1000

Eng Stress (MPa)

800

600
AR

400

BH170 BH220

200

0 0 0,05 0,1 0,15 Eng Strain (mm/mm)

Fig. 3.17: Engineering Stress Strain curve in L direction with BH and 10% pre strain

0,2

0,25

0,3

From figures 3.14 to 3.17 it is noticeable that BH at 170 has no major influence on the reduction of total elongation. In fact, the only time that a serious reduction of elongation was seen was when the samples had 10% pre strain, reloaded in the same direction and BH at 220C. But this could also have been noticed on figure 3.5; the sample that was aged at room temperature for 2 weeks after been pre strain 10% and then reloaded till fracture, presented this drastic reduction in ductility. This can be a consequence of having its Rp as its maximum strength value as well: over 1000MPa yield strength for a material that has Rm in as received condition of 850MPa. Besides, the total elongation for this DP800 is around 15%; the reduction in ductility after pre straining 10% (value at which the material has passed already its UTS) is completely normal. Figures 3.18 to 3.23 are a graphic representations of the results summarized on table 3.5: On this graphs WH represents Work Hardening, or just the increase on Rp or Rm in the samples that were not BH.

38

1100 1050 1000 950 900 850 800 750 700 650 600 0% 2% 4% 6% Pre Strain % 8%

Yield Strength (MPa)

WH 170C 220C

10%

Fig. 3.18: Rp as a function of pre strain level in L direction with and without BH

1100 1050 1000 950 900 850 800 750 700 650 600 0%

Yield Strength (MPa)

WH 170C 220C 2% 4% 6% 8% 10%

Pre Strain %
Fig. 3.19: Rp as a function of pre strain level in S direction with and without BH

Yield Strength (MPa)

1100 1050 1000 950 900 850 800 750 700 650 600 0%

WH 170C 220C 2% 4% 6% 8% 10%

Pre Strain %

Fig. 3.20: Rp as a function of pre strain level in D direction with and without BH

39

1100 1050 UTS (MPa) 1000 950 900 850 800 0% 2% 4% 6% 8% 10% Pre Strain %
Fig. 3.21: Rm as a function of pre strain level in L direction with and without BH

WH 170C 220C

1100 1050 UTS (MPa) 1000 950 900 850 800 0%

WH 170C 220C

2%

4%

6%

8%

10%

Pre Strain %
Fig. 3.22: Rm as a function of pre strain level in S direction with and without BH

1100 1050 UTS (MPa) 1000 950 900 850 800 0% 2% 4% 6% 8% 10% Pre Strain %
Fig. 3.23: Rm as a function of pre strain level in D direction with and without BH

WH 170C 220C

40

3.3.5 Tests from hydro formed boxes Table 3.6 shows the calculated Rp, Rm and Plastic energy from the complete set of results of the small tensile samples taken from the hydro formed boxes.
Table 3.6: Overview results from samples taken from hydro formed boxes

Sample 71mm No BH L 71mm BH170 L 71mm BH220 L 71mm No BH D 71mm BH170 D 71mm BH220 D 60mm No BH L 60mm BH170 L 60mm No BH D 60mm BH170 D 60mm BH220 D

Rp (Mpa) 632 701 833 743 783 832 738 781 1059 1088 1129

Rm (Mpa) 894 904 936 885 900 917 980 1034 1078 1116 1151

Plastic Energy () 80 69 61 85 81 73 35 27 8 8 5

Various trends can be seen from the results on table 3.6. First, it can be noticed that BH, especially at 220C, does show an increase on Rp and Rm. Also, there is an increase of strength from L to D direction, and a significant increase on the results of the samples that were taken form the 60mm tube diameter hydro formed boxes. Except for the samples from 71mm on L direction, the results showed an increase of 30 to 50MPa in Rp after BH at 170C, and between 70 to 90MPa in Rp after BH at 220C. Still, it has to be said that due to the continuous yielding seen on the L direction samples (figure 3.24 and 3.26), the measurement of E and Rp were difficult, and variations of 0.01% of true strain gave differences of more than 5MPa on Rp. In fact, for L direction on both diameters on the as received condition, the scatter of the Rp measurement was around 100MPa, and only one sample was tested with BH170 an BH220. Figure 3.24 to 3.27 help explain these trends:
1200

1000

True Stress (MPa)

800

600

71mm L AR 71mm L BH170 71mm L BH220 71mm D AR 71mm D BH170

400

200

71MM D BH 220

0 0 0,01 0,02 0,03 0,04 0,05

True Strain (-)

0,06

0,07

0,08

0,09

0,1

Fig. 3. 24: BH response on D and L for hydro formed samples from initial tube diameter 71mm

41

In graph 3.24 it can be seen that the values for D direction are higher than on L direction. It is important to see that on L direction, the AR curve presents clearly continuous yielding, and the BH220 curve present yield point elongation.

1200

1000

True Stress (MPa)

800 60mm L AR 600 60mm L BH170 60mm D AR 60mm D BH170 60mm D BH220 400

200

0 0 0,005 0,01 0,015 0,02 0,025

True Strain (-)

0,03

0,035

0,04

0,045

0,05

Fig. 3.25: BH response on D and L for hydro formed samples from initial tube diameter 60mm

On graph 3.25 it is clear the significant increase of Rp and a drastic reduction of uniform plastic elongation when the direction of the samples is changed from L to D. There is no yield point phenomenon, and continuous yielding can be seen on the samples in L direction. All the samples on D direction already presented Rp values of over 1000MPa.
1200

1000

True Stress (MPa)

800
71mm L AR 71mm L BH170 71mm L BH220 60mm L AR 60mm L BH170

600

400

200

0 0 0,02 0,04 0,06 True Strain (-) 0,08 0,1

Fig. 3.26: Comparison between samples with BH on L direction on samples with different initial tube diameter.

42

1200

1000

True Stress (MPa)

800
71mm D AR

600

71mm D BH170 71MM D BH 220 60mm D AR

400

60mm D BH170 60mm D BH220

200

0 0 0,02 0,04 0,06 True Strain (-) 0,08 0,1

Fig. 3.27: Comparison between samples with BH on D direction on samples with different initial tube diameter.

Graphs 3.26 and 3.27 compare the results of the tests on the same direction but with different initial tube diameter. The curves of the samples from the 60mm tube are always higher than the ones of the 71mm diameter, but the most significant effect is seen in D direction: the increase on Rp is over 300MPa for same conditions (direction and BH temperature). Also, the uniform elongation is drastically reduced. The differences in the results seen between L and D directions can be explained by two reasons: the first is the fact that was seen on chapter 3.3.1, where it was shown that the strength of the material in D direction was higher than the one on L direction. The other reason (and more important one) is due to the nature of the forming process. In hydro forming, a tube profile of the material is blown with high-pressured water to shape a square profile. For these tests tube of 71 and 60mm diameter were used; the final shape is independent of initial tube diameter, the smaller tube requires more stretching. This implies that the pre strain levels are higher on the box that is formed from the 60mm diameter tube. The effect is clear on the transverse direction of the tube the pre strain levels, as it is the direction in which the tube is deforming. In the longitudinal direction, the deformation is not that noticeable, even though in some cases end feed is used; besides, the axial feed is generating pre strain in compression. This also brings the possibility of having Bauschinger effect on these samples (D direction samples). Using PHAST measurements of strain, an equivalent strain values was calculated for hydroformed boxes from both diameters (60 and 71mm). These values can be seen on table 3.7
Table 3.7: Pre strain values for samples from hydro formed boxes

Equivalent strain

60mm (40ax endfeed) 0.20 (20%)

71mm 0.06 (6%)

43

3.3.6 Microstructure Another thing that should be mentioned concerns the microstructure of the steel; after having a discussion with metallurgists at Corus RD&T, it was surprising to see the response of this DP800, even on the as received condition. This was because usually Dual Phase steels have a smaller Rp - Rm ratio (around 0.5 or less) to the one seen on this on this specific DP800 (672 / 889 = 0.77). Besides this, Dual Phase steels are characterized by a continuous yielding point: a much smother and longer transition from the elastic region to the plastic one which is graphically seen as almost an absence of a fixed yield point (for example, figure 3.28).
1000 900 800 Engineering stress (MPa) 700 600 500 400 300 200
04-029 as received r 04-029 BH0 r 04-029 BH5

100 0 0 5 10

04-029 BH2 r

15

20

25

Engineering strain (%)

Fig. 3.28: Dual phase 800: The stress strain response, before and after baking at 170C for 20 minutes, of each of the material conditions tested. In the legend, the suffix 0,2,3, or 5 denotes the level of pre-strain applied prior to baking. The suffix "r" denotes that the data resulting from the duplicate test has been used. Taken from [40].

In addition to the continuous yielding on the as received condition mentioned before, the results shown on figure 3.28 also illustrate BH response of other DP800. It is important to see that there is no yield point elongation for the BH samples, and even more, the BH sample without any pre strain has a softening behavior instead of a hardening one (blue curve on figure 3.28 is underneath the red curve). Nevertheless, the BH response cannot be determined from this figure, because the data is presented in engineering stress strain values, and not in true stress strain values. Also, the yield point elongation seen on the samples of the DP800 of this project was not evident (even non existing) when the samples were BH at 170C, but it was at 220C. In [40], no samples were treated at 220C. The high Rp/Rm value and the difference of graphs the AR and BH tensile samples of the DP800 used in this project compared to the one studied in [40] could be explained after checking the microstructure (Fig 3.29).

44

Fig. 3.29: Microstructure analysis of the DP800 used in this project

Normally, DP steels have a 20% fraction of second phase (martensite) in their structure. This one has 60%, which means that the material would start behaving more like a martensitic steel than a DP; it is still a dual phase, but with a bigger influence of the stronger second phase.

3.4 Conclusions
In this chapter tensile test properties of DP800 with variables as BH, time, pre strain and strain path/direction have been measured. One of the first and most clear conclusions that can be made from this is that ageing at room temperature is a characteristic that has to be taken into account properly in order to be able to define the behavior of the material. The DP that was analyzed has a very pronounced response to ageing time at room temperature after pre strain. The time-temperature relation to determine strengthening should be measured more extensively. Theoretically, it could be said that after an infinite time, the room temperature ageing effect would have the same response as the BH at 170 or 220C or other temperature but in a shorter period. An upper limit in strengthening should be determined as a function of time, temperature and pre strain level. It was clear that for higher pre strains shorter periods of time and lower temperatures are needed to see some increase on strength, but it is important to see how far it can go, or at which point the response stabilizes. According to the ageing tests, a component with high pre strain levels will present a hardening behavior in a short period of time without any baking. This could imply that even without the BH, the material would eventually reach the strength levels seen on the baked samples. From figures 3.18 to 3.23 it could be said that BH response is approximately constant for a defined temperature, independent of pre strain level or reloading direction (remember that BH response is measured as an increase on Rp); the shape of the BH curves for each set of samples seems to be determined by the shape of the WH sample. The BH curves at both temperatures could be approximated as parallel to the WH curve for every case. An

45

estimation of the increase in yield strength due to BH at 170C can be approximated between 30 and 50MPa, while BH at 220C shows an increase of approximately 70 to 90MPa. This is an important conclusion, because it would mean that BH response of this material (at a fixed temperature) could be measured in a single direction with a single pre strain level, and could be assumed the same independent of strain path changes or strain levels. It would be just a matter of measuring the response on the standard BH test; in terms of material modeling, the response would be determined by strain path dependent model, which would have to be developed. But, when the uniaxial test was done, time was not taken into consideration as an important factor, so it was managed carelessly. Although all the samples were measured under the same ageing time, it is not possible to determine a real effect of temperature and time. The results presented as WH are not only WH really; rigorously speaking they also include ageing at room temperature. When the two-week ageing test was done (fig. 3.6), it could be seen the effect was very large on the 10% pre strained sample, and just slightly seen for the 2% pre strained one. Although the material shows strengthening as result of ageing (at room temperature and BH), the total elongation is not affected significantly due to temperature; so the ductility of the material would be determined by the pre strain history and not the BH effect. Only in very extreme cases (very large pre deformation) a considerable decrease was seen as a consequence of BH at 220C. This should be taken into account only for very specific application, were you find loadings in the same direction of the pre strain levels generated in the forming of the part. From the results of the samples taken from the hydro formed boxes several aspects can be mentioned. First, it is important to see how different the response of the samples from the box can be depending on the direction that is been measured. This because the pre strain levels on the directions are very different. It is important to emphasize this aspect because the increase on the strength levels is more influenced by the forming history of the part, and not really by BH process: while the increase in strength due to BH is less than 50 and 90MPa for 170 and 220C respectively, the increase from the 71 to the 60mm initial diameter tube is approximately 300MPa, value which can be considered constant for the three temperature levels. This also confirms the fact that the BH response can be considered constant independent of the pre strain level. Also, it can be said that the response of the component to a loading condition will be localized, because it can be behaving plastically in one direction, while the material is still in the elastic region in another one. It can also be mentioned that the Bauschinger effect (softening when the reloading is done on other direction than the pre strain) described on the uniaxial test can also be seen on the samples from 71mm with no BH on the L direction. The value for the measured Rp on this condition (632MPa) is less than the value of Rp on L direction without pre strain or BH (662Mpa). In fact, the measured results for the mentioned condition are practically identical to the ones obtained on the uniaxial tensile test for the AR 2% 90 samples (see tables 3.5 and 3.6). In Appendix A, results from tensile tests of other AHSS are found; among them, there is a DP800 (Figure A.1). This one has a smoother AR curve (more alike the DP800 results presented in [40]). Because of that characteristic continuous yielding, when calculated the BH response of the materials without any pre strain, values of over 100Mpa were obtained after BH at 220C; but as the pre strain level increased, the BH response appeared to be lower. This is a consequence of the difficulty of measuring Rp and E for materials with this kind of behaviors. Still, the average increase due to BH was around 60MPa at 220C, and 30MPa at 170. The total elongation didnt seem to be directly compromised by the BH treatment; even more, it is interesting to mention that the samples that had pre strain of 10% and were baked at 220C had a longer uniform plastic elongation (as well as total elongation) than the ones that had 10% pre strain and were baked at 170C.

46

As for the other AHSS tested, both TRIP grades showed a clear yield point elongation phenomena at both temperatures; after it, the tensile curve seemed to stabilize with the characteristic AR curve. The BH response of these two steels was the lowest of the grades tested. By the other hand, the DP1000 presented a clear reduction on uniform elongation due to BH already at 2% pre strain. With pre strain and subsequent BH, these samples reached strength levels higher than its initial UTS, which makes them jump their uniform plastic elongation region when reloaded. Unfortunately, data for BH response of CP with pre strain was not available. In general it could be said that the BH response of is higher on DP than TRIP steel, but CP should be studied in detail. It was also seen that the DP grade of this research behaves more like a martensitic steel (stronger steel) than an actual DP, and its BH response is apparently higher than the other DP tested.

47

CHAPTER 4: CRASH TESTING

4.1 Introduction
In an impact event, the car structure can be subject to front, rear, side impact or roll-over. Consequently, the deformation of structural members is often a complex combination of forces in different directions. This can bring different types of collapse on the structure parts, which is a very important characteristic to take into account to be able to improve a nd optimize the energy absorption capability of a member during crash. In order to understand the influence of various design parameters on crash performance, it is often easier to investigate the two predominant collapse modes: These are axial and bending collapse. So far, the response of DP800 to BH has been studied in literature and through experimentation in a set of tensile tests. But as the main concern of the project is to ascertain the influence of the process on the In service performance, a lot more conditions and variables have to be taken into account. Although tensile tests are very important in material characterization, the relation between the results of this kind of test may not necessarily be directly linked to the crash performance of a certain material. Strain rate, energy absorption, ultimate tensile stress, strain path, uniform elongation, thickness, geometry, compression strength and collapse mode among others are characteristics that have to be taken into account in order to measure the crash performance of an automotive component. In this chapter, the influence of BH on DP800 is going to be measured in crash experiments consisting of 3 point bending tests and axial crash tests. Some data obtained from other research measuring the influence of BH on other AHSS is going to be used as well.

4.2 Three Point Bending Tests

When talking about a 3 Point Bending test, it is possible to divide the bending behavior of a fixed geometry in elastic and plastic bending. In elastic bending, the displacement of the structure depends on the elasticity modulus of the material and the geometry (because it determines the Inertia modulus). As for the plastic bending, the displacement depends on the bending moment, which is a function of the flow stress of the material and the thickness. The stiffness of the component needs to be increased, and it is a compromise of the two last characteristics: By increasing the thickness the stiffness is increased, but the weight of the part as well. And by increasing the strength of the material, the resistance of the section to bending will increase, but that would compromise the elongation capability of the material. Previous studies have concluded that as the yield strength of the material increases, the peak load (load that generates the buckling) increase. Testing different materials, with an increase in yield stress of 70%, an increase of approximately 70% on maximum peak load was obtained [31]. This would also bring an increase in the capacity of the part to absorb energy during impact, measured as the area under the force displacement curve obtained in the test. But it was also found that for samples that had width to thickness ratios greater than 70 (w/t>70), a point where no more improvement of load bearing capacity was found due to increase of yield strength [9, 11].

48

4.2.1 Experimental details 4.2.1.1 Specimens Besides the DP800 of the research, CTH sections of DP1000, TRIP800 and CP800 were also tested. All the specimens tested were CTH sections, with the same geometry described in figure 4.1. The only difference between samples was the thickness of the CTH, which depended on the as received material thickness. MATERIAL DP800 DP1000 TRIP800 CP800
60 Ro t

THICKNESS (mm) 1.486 1.486 1.23 1.16

60

Ro

107.5
15 20
Fig. 4.1: Cross section of CTH sample

Figure 4.1 gives the cross section geometry of the CTH section. The dimensions of the section are 60*60 mm with a flange width of 20 mm. The outer radius R0 of the corners is 4.5 mm. The hat is closed with a plate (width 100 mm) using spot welding to join the U section to the closing plate. The weld pitch is 30 mm and started 10 mm from the end of the crash box. Given a box length of 500 mm the total amount of spot welds becomes 34 (17 each side). 4.2.1.2 Test Setup The bending tests were done at IJTC on a 300kN MTS hydraulic machine. The setting of the samples can be seen on figure 4.2. All samples on as received (AR) and BH at 170C condition were done on duplicate, while the 220C samples were done on one sample only (no more samples were available). The velocity of displacement of the indenter in the test is 20mm/min, and a maximum fixed displacement of 80mm was chosen.

Fig. 4.2 Three Point Bending Test setup

49

Previous tests done on DP and TRIP steels have shown that they are extremely sensitive to spot-weld quality, position and orientation [6]. In Closed Top Hat (CTH) components the bending test can be performed in two different ways taking into account the surface that is in contact with the indenter. The indenter can be in contact with the plate that is welded to the U section (Backing plate upper) (Fig. 4.3), or this plate can be in contact with the two supports (Backing plate lower) (Fig 4.4). In the Backing plate Upper orientation, the collapse force was very sensitive to spot weld pitch, position and quality; in the Backing plate Lower orientations, the measurements of the force didnt show any sensitivity to those parameters.

Fig. 4.3 Backing plate upper i.e. plate in contact with indenter

Fig. 4.4 Backing plate lower i.e. plate in contact with supports

The purpose of this research is to measure BH response of the material tested, and not variations due to effect of the spot -weld; therefore it was decided to set Backing plate lower as standard position for bending test of the BH samples (figure 4.5).

Fig 4.5 Samples of DP800 after 3PB test

4.2.1.3 Data file The output file for this test is a csv-file, in which the time (s), the crosshead displacement (mm) and the load (k N) are measured throughout the test. From this file, graphs relating force and displacement of the test can be made in order to determine peak load (load at which the buckling of the sample occurs), and energy absorption capacity of the sample (area under the force-displacement curve).

50

4.2.2 Results The results of the bending tests are shown in Force vs Displacement curves (e.g. Fig. 4.6). From these ones, it is important to take the peak load and the energy absorption capacity. The peak load can be seen as the first local maximum in the graphs, and its level is calculated as the highest value of force before 15mm displacement. The 15mm parameter was chosen with help of the graphs, due to the fact that for all the samples tested, the first peak load (yield) was reached before the displacement reached 15mm; also, the force hadnt build up again in the curve at this time for all the samples. For the energy absorption parameter, the area under the curve for the complete test was measured. The results of the test for DP800 were as follow:

60 50 40 30 20 10 0 0 20 40 DISPLACEMENT (mm)
Fig. 4.6: 3 point bending test of DP800 with and without BH
DP800 DP800 BH170 DP800 BH220

FORCE (kN)

60

80

Before the first 10mm displacement of the test, all the behavior of the sample is elastic. There is no buckle, and if the load were released there would be no permanent deformation on the CTH box. At the first local maximum (peak load), the buckling starts developing. The CTH sections display a collapse mode consisting of a peak load corresponding to the section buckling, and a second rise in load as the section folds around the punch. From the figure 4.6 it can be noticed that there is a small increase in the force level throughout the complete test for the BH samples; in addition, the variation of the BH temperature apparently didnt affect force levels in the test, as the results for 170 and 220C are practically the same. Figures 4.7a, b and c, and table 4.1 show the results of the bending test concerning the change in peak load and energy levels registered:

51

26,0 25,5 Peak load (KN) 25,0 24,5 24,0 23,5 23,0 22,5 DP800 AR DP800 BH170 DP800 BH220

Fig. 4.7a. BH influence on peak load on 3 point bending test for DP800

2700 2650 2600 2550 2500 2450 2400 DP800 AR DP800 BH170 DP800 BH220

Energy (J)

Fig. 4.7b. BH influence on Energy absorption on 3 point bending test for DP800

52

8 7 6 % increase 5 4 3 2 1 0 DP800 BH170 DP800 BH220

Peak Load Energy

Fig. 4.7c Normalized results of BH influence on 3 point bending test for DP800 Table 4.1: Summary result 3PB for DP800 MATERIAL DP800 AR DP800 BH170 DP800 BH220 Peak Load (KN) TOTAL ENERGY 23,8 25,0 25,5 2503,0 2634,3 2658,2 %F 0,0 4,8 7,2 %Energy 0,0 5,2 6,2 Rp 679 709 757 Rm 874 876 885

It is important to see that although the increase due to BH effects in peak load are between 1 and 2KN and for energy between 100 and 150J, which could be considered as a small amount of increase, in percentage all these values fluctuate around 5%. Apparently, no significant difference due to changing the BH temperature is characteristic behavior of Dual Phase steels in the bending test; TRIP steel showed no significant influence of BH at all, while Complex Phase showed the highest influence (increase up to 13% in peak load and 20% in Energy absorption) and as well as temperature dependent response (see appendix B). There was no noteworthy difference either in the fracture of the samples (nor size or position); they all presented small cracks on the corner outer side of the folding (Fig. 4.8). This aspect should be analyzed more carefully (more precise measurement of crack size), because although no clear difference was seen, an increase in strength of any material is usually associated with a decrease in ductility, affecting this way the size of the crack. Taking into account that the corners of the CTH are the places where the samples have the highest level of pre strain due to forming operations, and that as well are the places where the highest level of stress is seen measured during the bending test, BH can possibly generate a brittle behavior of the sample. This could be seen if the cracks were bigger in the BH samples, but it was not the case.

53

Fig. 4.8: Location of fractures on CTH sections after bending test

4.2.3 Empirical Fit After performing the experiments, an empirical fitting of the bake hardening effect on peak load was possible using approximation equations from previous studies. The peak load of the 3-point bending test can be estimated by equations relating the yield strength of the material and the thickness of the CTH box. The Peak load (KN) is calculated by

FMAX = C R p t B
A

[9,13,14]

where R p is the measured Yield strength (MPa), t is the thickness (mm) and A, B and C are constants resulting from the empirical fitting data. Previous work done b y Jones [34] and Carless [35], gave 0.7, 1.85 and 0.13 as values for A, B and C respectively, while having a correlation parameter of 97% (r squared = 0.97) between the measured data and the fitted equation. This fit is valid for formable and HSS grades having thickness between 1.2 to 1.5mm, and the backing plate upper orientation. Never the less a new empirical fit had to be made because the data used for this equation was taken from 3 point bending tests done on backing plate upper orientation. As discussed in 4.1.1, this orientation was changed to avoid the effects of weld position on the peak force measurement on the CTH sections. The new multiple regression analysis fit gave 0.58, 1.91 and 0.26 as values for A, B and C respectively, while having a correlation parameter of 97% (r squared = 0.9685). The complete set of results for this regression analysis is shown in Table 4.2. To run this fitting analysis, the results of 15 tested samples were used and they are listed on table 4.3.
Table 4.2: Regression analysis for 3PB on CTH

VALUES WITHOUT BH A 0,5839 B 1,9148 C 0,2632 Iterations 26 Standard Error 1,1373 R2 0,9685 SSR 477,2314 SSE 15,5223

54

Table 4.3: Experimental measurements of Peak Load on 3PB without BH

Material HSS HSS HSS HSS HSS HSS HSS HSS HSS HSS HSS DP800 DP1000 TRIP800 CP800

BH Temp (C) No BH No BH No BH No BH No BH No BH No BH No BH No BH No BH No BH No BH No BH No BH No BH

Measured Peak Load (kN) 6,9 8,2 7,3 10,1 11,9 11,8 14,2 19,4 17,0 15,6 15,5 23,8 28,3 14,7 15,5

Using the values of A, B and C obtained in the regression analysis, a prediction of the Peak Load of the samples that had been BH was made. The yield stress and thickness for each of the four materials baked at both temperatures were used as i nput for the equation. The values of Rp for the BH samples of DP1000, TRIP800 and CP800 can be found on Appendix A, and the calculations are presented on table 4.4. It can be seen that only for the DP1000, the prediction of the peak load was lower than the measured on the experiments; although no results of BH samples were taken into account to obtain the values of the constants in the prediction formula, it seems fair to say that the estimation of the peak load for the BH samples is accurate. Figure 4.9 shows these results, as well as a linear fitting of the non-BH samples used for the regression analysis.

Table 4.4: Comparison of prediction and experiments for 3PB on BH samples

PEAK LOAD (kN)

Material BH Temp (C) Predicted Measured Variation (kN) % Variation DP800 170 25,4 25,2 0,2 0,9 DP800 220 26,7 25,5 1,2 4,8 DP1000 170 27,4 29,0 -1,6 5,4 DP1000 220 29,2 29,8 -0,6 1,9 TRIP800 170 15,1 15,1 0,0 0,3 TRIP800 220 15,8 15,3 0,5 3,0 CP800 170 17,7 16,4 1,3 7,9 CP800 220 18,1 17,5 0,6 3,2

55

35 No BH BH 170C BH 220C Linear (No BH)

30

Measured (KN)

25

20 y=x R = 0,9685
2

15

10

5 5 10 15 20 Calculated (KN)
Fig. 4.9: Accuracy of prediction vs measurements (First fit)

25

30

35

It is important to mention that all the values for yield point were calculated as described on 3.2.4 (see appendix A for Rp values of materials different of DP800); This can be a factor to study more carefully, due to the fact that all the materials presented yield point elongation phenomena when subjected to BH, especially visible at 220C. But, considering that A (exponent of Rp in the equation) is relatively low value, and that the difference between Upper and Lower yield points was less than 30MPa for all the materials tested (having Rp value somewhere in between them), the accuracy of the analytical regression can be considered still to be very high.

4.2.4 Conclusions In Chapter 3 the influence of BH on Rp was shown for DP800 (the specific one in this research); as the peak load is proportional to Rp, it is normal to see an increase on the peak load of the bending test. It is important to measure this peak load because it is the one that determines the buckling of the sample; it marks the point where the performance of the sample goes form elastic to plastic. The influence of BH on 3PB for DP800 was studied in this chapter. The results show that there is a small effect of BH on the samples, as the influence on the peak load and energy absorption was around 5% with a scatter of 2% (for BH at 170C). This can probably be explained by the fact that even though the first local maximum in the force displacement curve (Peak Load) can be predicted very accurately by an analytical equation function of the yield stress, the exponent affecting the Rp is less than 1, while the exponent affecting the thickness is almost 2. So, the increase in yield strength due to BH is not high enough to generate significant increase in the bending performance. But also that is probably a very rough conclusion: the problem is that the scatter of the results of the measurements of peak load of the samples is around 0.3kN. The increase in peak load due to BH is 1 or 2kN depending on the temperature. Because the peak load for the non-BH sample is just below 24kN, an increase of 1 and 2kN would represent an increment around 4 and 8%.

56

Now, all the experimental results depend on the size and shape of the indenter; even the maximum displacement was fixed because of it, as after 80mm the CTH sample would have already fold around the indenter cylinder and would start touching the hydraulic machine in which the test is performed. The buckle formed on the CTH is completely det ermined by the shape of the indenter, and this influences the force and displacement measurements. From the complete set of tests done not only on DP800 but also on the other HSS (appendix B), two other conclusions can be made: with the exception of CP800, the change in the force displacement curve in the tests seem constant through out the whole test. Approximately the same amount of increase can be seen for the complete displacement. Also (again with the exception of CP), there were no significant differences seen on the results due to the change in the tests temperature. The CP grade showed clearly the highest response to BH in this test, and future research on this material is suggested. Ideally, at least 3 tests on each condition should be done to verify the results; this to be able to show repeatability of the tests. It would also be useful to see the influence of BH on the bending test of another geometry, like a hydro-formed box for example; as hydro formed samples have higher levels of pre strain, it would be interesting to see if a serious reduction in ductility is seen after BH (like was seen on the tensile sample L with 10% pre strain after BH at 220C).

4.3 Axial Crash Tests

For the past several decades car manufactures and material suppliers for the automotive industry have made a considerable amount of work in research for crash performance. Even though huge advances have been accomplished, the complexity of the crash problems, due mainly because of the highly nonlinear behavior of the structures, brings the fact that the methods used to quantitatively asses the problem are not yet reliable design tools. Besides, the continual changes in vehicle regulations and material research have kept structural crash behavior a current and important design requirement. But these experimental approaches are still expensive and difficult to utilize in development of optimal structures. Some of the parameters that are important to investigate in the structural collapse design are collapse load, energy absorption, and instability. The collapse load can be defined as the load required to initiate significant permanent deflection of the structure. This parameter is usually normalized for weight or size to compare different designs. The area under the force displacement curve in an axial crash is a measurement of the energy absorbed by the structure. This one is also divided by the weight of the part to have a specific energy absorption parameter. The energy absorbed during the collapse of a structural member during axial crash depends on the material mechanical properties, the geometry of the member and the collapse mode. From the force displacement curve, peak load and mean load or average load can be measured as well. The instability of the structure is the basic concern of the design, in order to be able to optimize the amount of energy absorbed, as well as it regulates the way of measuring and modeling the performance of the structure [22, 23, 24]. It is not adequate to rank the material contribution to specific energy absorption based on terms of the tensile specific energy absorption (derived from a conventional tensile stress strain curve), because the latter one is governed by necking, which limits the extension, and, hence, the energy absorbed [22]. As described in chapter 3, the stress strain curve is only valid until UTS, even though there is still extension of the sample, and then energy absorption. In a collapsing structure, the energy absorption can go on over this design limit (failure either by extensive plastic deformation of even fracture) and still perform its function, provided that no unstable collapse occurs [24]. Failure of the material does not necessarily lead to failure of the structure for its design purpose, since the structure can continue to

57

absorb energy to its limit provided that the failure zone remains relatively localized. Besides that, it also must be taken into account that the primary load on the structure may not be tensile in nature, but bending or compressive. In several literature sources it has been stated that the specific energy parameter measured in axial crash tests is linearly dependent on specific ultimate tensile strength (for fixed geometry and collapse mode). But at very high UTS values (for AHSS), it has been found that the collapse mode tends to change from folding (plastic deformation) to fracture, which can seriously affect the energy absorbing capacity. Nevertheless, the parameter that fitted best the relation between the mechanical properties of the material and the absorbed energy was UTS [9,13,14]. A summary of different formulas to predict energy, peak and mean loads in axial crash found in literature are presented now; this estimation was compared with the actual crash test. It is important to realize that these formulas are a statistical approach to the parameters that are been calculated, and not really a mathematical explanation of the physical phenomena describing the performance.
A Energy = C * Rm * t B , for dynamic axial crash [9,13, 14,]. The

values for A

and B were around 0.5 and 1.5 respectively.

.382 FMAX = 20.2 * R 0 * t 0.86 , for dynamic axial crash [14]. This one was the p

only relation with peak load found in literature. It is important to mention that it had a wide scatter.
A FMEAN = K * Rm * t B , for static and dynamic axial crash [9, 14, 21, 30, 32].

The values of the constants change depending on the test setup; that is if the test is dynamic or static.
FMEAN = K * Areaunder _ tensile _ curve * t

for dynamic and static crash [30].

The advantage of this equation is that the values of the same constants are valid for different impact velocities (K = 1.135, a = 0.75 and = 2.09). The area under the tensile curve was considered for this case between 0 and 10% strain. In general, for the calculation of the Fmean using the measured data, in the static test the mean force is equal to the absorbed energy (area under f-d) divided by maximum displacement; in the dynamic test the mean force is calculated as the potential energy divided by crushed length [29]. Fmean is calculated in a different way for those tests because during the dynamic test the velocity is not constant and the displacement is not fixed (then the potential energy is know before hand by selecting the height and weight that is used), while in the static and intermediate test both velocity and displacement are constants.

Although all the formulas presented previously were developed for CTH sections, it should be mentioned that the geometry of the section could vary from source to source; the flange length varied, as well as the distance between spot welds. Also, in the sources where BH was taken into account, it was mentioned that even though an increase in peak load could be seen, there was no significant increase in energy absorption. Still, the increase in peak load could be used for design of other components like the structure behind an energy -absorbing member. Wierzbicki, Jones and Abramowicz introduced theoretical formulations for axial collapse of thin-walled structures that were applicable to various geometries, load cases and collapse types. Specifically in square tubes, the so-called Yield Hinge Model of Progressive folding was identified and studied (Figures 4.10a and b). This study led to the introduction of the Basic Folding Element, which combines kinematical considerations and plasticity theory [36].

58

One of the most important aspects of the formulations is that the thin walled structural member is considered a cluster of crushing corner elements (Basic Folding Elements), and that they were analyzed separately. In each of these Basic Folding Elements, energy absorption was divided then into three main mechanisms: 1) Bending of the material stationary yield hinge lines, 2) Bending of the material around moving yield hinge lines, and 3) Stretching of the material over a toroidal surface (Fig 4.10b). These three mechanisms approximately contributed in similar proportions to the total energy absorption [36].

Fig. 4.10a: Yield Hinge Model of Progressive folding. 1) Initial undeformed shape. 2) Collapse. Taken from [36]

Fig 4.10b: Basic folding element and its deformation mechanisms: 1) Floating toroidal surface. 2) Bending along horizontal stationary hinge lines. 3) Rolling deformation along moving inclined hinge lines. Taken from [36].

In the progressive folding model, approximately two thirds of the energy is dissipated in compressive deformation of the material; this may occur in bending (stationary horizontal hinge lines), pure compression (flow of material through toroidal surface) or bending and subsequent unbending (rolling deformation in the moving inclined hinge lines). This concept gives importance to the understanding of the compressive behavior of the material [36].

59

4.3.1 Experimental Details 4.3.1.1 Specimens All the specimens tested were made of DP800. There were three different sets of specimens for this part. One set was the CTH section, just like the one described on chapter 4.2.1.2, but with a total length of 125mm instead of 500mm. 9 samples were tested, 3 for each temperature (AR 170 and 220).
Table 4.5: Number of CTH sections tested in axial crash with and without BH

Samples CTH

AR 3

BH170 3

BH220 3

The other two sets were hydro-formed sections, one group taken from a 60mm diameter tube and the other one taken from a 71mm diameter tube. The dimensions of the section are 60*60mm. The outer radius R0 of the corners is 4.5 mm. The total length of these specimens was 125mm (fig. 4.11). 4 samples were tested in total:
Table 4.6: Number of hydro formed sections tested in axial crash with and without BH

Hydro formed samples Initial tube diameter 60mm Initial tube diameter 71mm

AR 1 1

BH170 1 1

Fig. 4.11: Cross section of hydro-formed sample

The thickness of the square section samples made from the 60mm tube samples was 1.433mm, and the thickness of the samples made from the 71mm tube was 1.493mm. 4.3.1.2 Test Setup The axial crash tests were done at IJTC on a 300kN MTS hydraulic machine. As in the bending test, the peak load of an axial crash test is related to the increase in yield strength. The samples of DP800 with and without BH were tested to find peak load. This peak load is different depending on the type of axial test that is required; this test can be done at dynamic (50km/h = 13888.89mm/s), intermediate (100mm/sec) and static (20mm/min = 0.33mm/s) velocities. The results of this test are usually exposed in a force-displacement curve (same kind of graph as for the bending test). The results of intermediate and static are very similar, while the dynamic test has a first peak load approximately two times higher than the one seen on the other test velocities. Also, the differences due to noise of this test can also be seen clearly in the graphs (figures 4.12 and 4.13). The measurement of the peak load in the dynamic test depends on the points of the setup of the test where the data is recorded, so the variation can be high. Besides, as the intermediate and the static tests are done on a hydraulic machine (constant velocity), while the dynamic tests are done in a drop tower

60

(changing velocity), the resonance and vibrations also have an effect on the measurements made. Also, the intermediate and static test give the possibility of getting an insight into the folding mechanisms developed during the collapse process in an easier way than in the dynamic test. That is why in dynamic test the mean load is calculated, and not the peak load. The differences between the results of the three test velocities are shown in figures 4.12 and 4.13. Peak load measurements of axial crash tests of several materials done at Corus showed that the variations in the dynamic test are a good deal, so a choice had to be made between the intermediate and the static test to make sure that reproducibility of the test could also be taken into account. Arbitrarily, the intermediate setup was chosen.

300 250 200 Force [kN] 150 100 50 0 0 -50 Displacement [mm]
Fig.4.12: Axial crash at 3 velocities for CTH of DP800

0.33mm/s 100mm/s 13888.89mm/s

50

100

150

200

250

300

300 250 200 Force [kN] 150 100 50 0 0 -50 Displacement [mm]
Fig. 4.13: Detail of figure 4.12

0.33mm/s 100mm/s 13888.89mm/s

10

15

20

25

30

35

40

61

As the load generating the first buckle was the parameter to measure, there was no need for crashing a whole CTH box. For the tests with the C TH sections, a fixed displacement of 20mm was chosen. For this same reason, it is important to state that the samples that were tested did not have a triggering mechanism in their profile; the trigger affects the peak load (it makes it lower). BH response is measured as the increase on Rp, and Rp influences peak load, so it was necessary to avoid the triggering so that the difference measured would be an effect of BH only. According to [36], it is possible to assume that the length of the boxes does not affect the measurement of peak load for CTH sections, especially for HSS; even more, longer specimens (untriggered ones) had higher chances of presenting bending collapse, while the shorter specimens usually presented more stable axial collapse mode. This apparently also holds for square sections (KEB); still, the shortest length used in that study was 175mm, and in this set of experiments the length of the boxes is 125mm. 4.3.1.3 Data file The output file for this test is a csv-file, in which the time (s), the crosshead displacement (mm) and the load (kN) are measured throughout the test. From this file, graphs relating force and displacement of the test can be made in order to determine peak load (load at which the buckling of the sample occurs. 4.3.2 Results For the CTH sections results, an average of the three experimental results of the force displacement curves was calculated for each temperature. Those results can be seen in figure 4.14. The scatter of the calculation of Peak Load of the data in these tests can be seen in figure 4.15, while in table 4.7 the whole set of results are available.

250 200 Force (kN) 150

AR BH170 BH220

100 50

0 0 2 4 6 8 10 Displacement (mm)

Fig. 4.14: Intermediate velocity Axial Crash for CTH with and without BH

62

230 225 Peak Load (kN) 220 215 210 205 200 195 190 AR BH 170 BH 220

Fig. 4.15: Scatter of the data for Peak Load for CTH on axial crash with and without BH

Table 4.7: Measured Peak Loads for CTH

AR BH 170 BH 220

Measured Peak Load (kN) 191,2 194,7 198,5 202,7 210,0 202,4 214,5 215,9 228,5

Average (kN) Increase (%) 194,8 0,0 205,0 5,3 219,6 12,8

As for the hydro-formed boxes, the force displacement curve for the crash test can be seen in figure 4.16. As only one sample was tested for each condition, there is no determination of the scatter of the results. Table 4.8 shows the Peak loads measured for each sample.

180 160 140 AR, 71mm BH170, 71mm AR 60mm BH170 60mm

Force (kN)

120 100 80 60 40 20 0 0 2 4 6

10

Displacement (mm)
Fig. 4.16: Intermediate velocity Axial Crash for Hydro formed boxes with and without BH

63

Table 4.8: Measured Peak Loads for Hydro formed sections

Thickness (mm) Peak AR 60mm 1,433 BH170 60mm 1,433 AR 71mm 1,493 BH170 71mm 1,493

Load (kN) 160,4 169,2 152,0 158,9

Increase (%) -5,5 -4,6

From table 4.7 it is important to call attention on the fact that even though the thickness of the box that was formed from the 60mm diameter tube is less than the one that was made from the 71mm tube, the peak load that the first one can handle is higher than the peak load of the latter one. This fact is true independent of the BH effect, and can be explained by the behavior of the tensile samples taken from the hydro-formed boxes (table 3.6). As the pre strain levels on the 60mm hydro-formed box are higher than the ones obtained on the 71mm box (on both directions, but specially on D), the Rp measurement is higher. The measurement of Fmax can be estimated as a function of Rp and t; comparing the 71 and the 60mm boxes, there is an increase of approximately 5% on Fmax even as the thickness is reduced 4%. So in this case the increase on Fmax has to be a product of the increase on Rp: but, as mentioned on section 3.3.7, the increase on Rp depends on the direction that is being measured. Rp is approximately 15% higher on L, but it reaches almost 40% more on D.

4.3.3 Conclusions According to the results from the tests performed on the CTH section, there is an average increase of the peak load of approximately 5% and 13% for the BH samples at 170 and 220C respectively (10 and 25kN increase). This amount of increase could be considered significant, especially at 220C, but some other factors such as collapse mode and fracture or failure should be considered as well; Although there was a fixed displacement for the axial test (it was not a dynamic test), and this one was just enough to generate the first buckle (but not a complete fold) in the sample, the specimens already appeared to showed unstable collapse modes (Figure 4.17, samples 2 3 and 6 from left to right). This can be explained due to the lack of triggering mechanisms on the samples. That unstable folding pattern seen on the test can probably explain why there was a scatter of 14kN on the measured peak load for the samples at 220C; this difference is important because it changes the percent increase on peak load due to BH220 from 10 to 17%. But also it should be mentioned that at the as received condition and at 170C the scatter of the measured peak load was just over 7kN.

Fig 4.17: Intermediate axial crash test: CTH at AR, BH170 and BH220, 2 samples each from left to right.

In relation with the results of the hydro-formed boxes, an increment around 5% on the peak load by BH at 170C could be seen; unfortunately, only one sample was tested in each condition, so it is not possible what accuracy can be expected form the experiments. Still, some conclusions can be made: Like in the CTH, the increment seen by BH 170C was around 5%; even though the forming processes are different for the three cases (CTH, hydro formed 71 and 60mm) apparently the effect of a baking at 170C is the same percentage-wise. This fact could imply different statements: one could be that the effect of BH is independent from the geometry of the

64

component (and by this meaning the stiffness of the part). Also it could be expected that the increase in peak load for the BH samples at 220C would be around 13%. Or even that if pre strain levels are analyzed, it could be said that BH response is independent of pre strain level; this last one was also concluded form the tensile samples from the hydro formed boxes on section 3.3.7. It had been stated that providing the boxes with a trigger mechanism would affect the peak load measurement, and then so the BH response. Triggering the box would give a measurement of the bending load, while without a trigger the buckling load is measured. According to several sources, samples without trigger are more likely to develop unstable collapse modes. Also, fracture or failure at the spot welds was seen on the CTH sections (Figures 4.18 and 4.19), without even completing one fold, while the hydro-formed samples didnt present problems in the weld.

Fig. 4.18: Spot weld fracture

Fig. 4.19: Spot weld failure

It should also be mentioned that in all the axial tests (CTH and hydro formed sections) for both BH temperatures, the difference generated by BH effect on the force displacement curves of the experiment were only seen on the first part of the graph: after the first 2 to 3 mm of displacement (after the peak load was reached) all the graphs seemed to have an identical response to the one seen on the as received material. Apparently, in the axial crash, the BH is only influencing the peak load.

65

CHAPTER 5: IS IT IMPORTANT TO INCLUDE BH BEHAVIOUR OF DP800 ON CRASH SIMULATIONS?

The main purpose of this project was to answer if it is important to include BH response on crash simulations; even more, if it is necessary to develop a new material model to include BH in FE simulations. The answer would be no. By only performing bending and axial crash tests of DP800 with and without the influence of BH, and then analyzing the differences in the results, the same conclusion could have been made. But this approach would have a limited number of characteristics to take into account for the tests: Concerning the geometry, for example, the samples could have been a CTH, or square box, or a tube. Already only in the CTH, the flange length and the distance between spot welds (or even joining technique between parts) can vary. Also the presence or absence of a triggering mechanism gives other condition. And the setup can keep on varying if characteristics like time and temperature of the BH, or velocities of the test (in general test setup) are taken into account. By this what it wants to be explained is that there is a very big number of different conditions that should be taken into account to try to measure the performance of a component, and it is simply not possible to test them all (too many and just extremely expensive). That is why the idea of the project is trying to relate the BH response of DP800 on tensile tests to explain the results obtained in one of the possible setups for crash, which apparently shows that the BH effect on the performance is small enough to be able to consider it as part of the scatter of the results. It is easier and cheaper to change conditions like pre strain level, strain path, strain rate, time and temperature of the BH on a tensile test, and then try to generalize the results for different setups on the actual crash condition. Still, the base of a material model used on FE analysis will always be the tensile test.

5.1 Why is not pertinent to include BH in crash performance?

According to the results obtained in this project, it could be estimated that the BH response of DP800 can be summarized on table 5.1. This table shows the calculated percent increase on the measured values of Rp and Fmax as a result of BH at 170 and 220C.
Table 5.1: Percentage increase in Rp or Fmax due to BH. *These values are taken from the uniaxial tensile test at 2% pre strain for the 3 directions.

TENSILE (Rp) BH 170 BH 220 4.9 2%* 11.7 4.5%*

3PB (Fmax) 5 2% 7%

CTH AXIAL (Fmax) 5 4% 13 6%

HYDRO AXIAL (Fmax) 5 0.5% -

From table 5.1 is important to mention that the values for percentage increase in Rp from the tensile test were calculated from the tests that had 2% pre strain (small samples from the uniaxial test). This was done because the standard BH test is measured with 2% pre strain, and also because although the response was approximated as constant for each temperature and all the pre strain values and directions, certainly the percentage would be different for every pre strain level. No scatter is presented on 3PB with BH220, as only one sample was tested. Also, no data was presented for Hydro Axial with BH220 as no samples were available. In general, one of the most important things that can be observed form the results is that after BH at 170C, the increase on Rp and Fmax for all the tests done was around 5%. It could be said that the increase in Rp as a consequence of BH would determine the increase in peak load for the crash tests. Even more, if the same characteristics are evaluated for the tensile test and the axial crash of the CTH sections after BH at 220C (12% and 13% increase

66

respectively), it could also be estimated that the relation between Rp and Fmax holds as stated for the BH at 170 C. Unfortunately, there was not enough data for 220C available to establish a linear relationship between the increase in Rp and Fmax for all the tests. Besides, it should be remembered that the data used to measure the increase in the tensile tests shown in table 5.1 is the one taken from the small samples in the uniaxial test, which include some room temperature ageing. As mentioned in chapter 3, when the uniaxial test for the small samples was done, ageing at room temperature effects were not taken into account. Contrary to this, large samples has a maximum of 1 day room temperature ageing after being pre strained; usually, these samples were pre strained, baked and tested all on the same day. Still, the variation measured in the small and large samples due to BH was approximately the same (approximately same ?MPa between the Rp in the AR samples and the Rp in the BH samples); but as the small samples has an extra variation due to the nearly 2 month room temperature ageing, the strength level of these samples was higher than the one seen on the large samples. That fact makes the percentage increase in Rp of the large samples to be slightly higher than the percentage increase in Rp of the small samples (figure 5.1).

1100 1050 Yield Strength (MPa) 1000 950 900 850 800 750 700 650 600 0% 2% 4% 6% 8% 10% Pre Strain %
Fig. 5.1: Difference in Rp for different Pre strain levels and BH temperatures for Small and Large size samples

WH Large 170C Large 220C Large WH Small 170C Small 220C Small

The increase in peak load for the bending tests was in the range expected taking into account the empirical fit relating thickness and Rp found in literature: as the only parameter varying was Rp, an increase of 5% in yield stress would g ive an increase of 3% in Fmax, and an increase of 12% in yield stress would give an increase of 7% in Fmax, like shown in the next equation:
A .58 FMAX = C R p t B = K R p = K R0 ( Sec. 4.2.3) p A

BH 170 K (1.05R p ) 0.58 = 1.03FMAX BH 220 K (1.12R p ) 0.58 = 1. 07FMAX

Still, when the increase in Fmax was 5% the scatter was already 2%, so the BH effect in bending is small and can be ignored; other factors like spot weld quality, position and orientation of the CTH samples showed differences of over 15% in Fmax, and a completely different behavior of the force displacement curve [6].

67

The 3PB test also showed two things are worth to mention: the first one is that there is almost no difference at all in the BH response independent of the baking temperature; the response was practically the same after BH at both temperatures. But also, it was interesting to see that even though the change due to the baking was apparently insignificant, this response was practically constant trough out the whole displacement (not like in the axial crash where only the peak load seems to be affected). The static axial crash tests reviewed in literature are characterized by having a fixed displacement that is obtained after various crash steps that vary between 80 and 120mm depending on the size of the sample. In this kind of tests, the references studies showed on initial length of the box that varied between 250 and 500mm. Due to the short length of the samples available (125mm), it is not possible to perform displacement steps of this magnitude. This small displacement limits the measurements of the test to the peak load, as not even one complete fold was obtained from the tests. In order to be able to determine parameters like Fmean and absorbed energy, at least two complete folds should be obtained from the tests; this also would give the opportunity to relate the results to the equations mentioned on chapter 4 for axial crash tests. In crash behavior the initial peak-force or initial buckling strength is important. In reality, the peak-force of the front longitudinal should be high enough to prevent collapse of the structure under low energy impact thus ensuring all deformation is in the bumper and crush cans. On the other hand, it should be low enough to prevent overloading the back-up structure under more severe loading. Peak load appears to be controlled by trigger design mainly and not the residual effects of the forming process such as thinning and strain-hardening [35]. In the 3PB and the axial crash tests, the peak load measured is determined by buckling behavior of the sidewalls; the critical load that can be supported is a function of the tangent modulus. Literature research shows that in axial crash, the main focus has been given to the calculation and measurement of Fmean and energy absorption. There was no equation that measured a peak load, except the one found on [14], but its results are not reliable due to very wide scatter. This equation was meant for dynamic axial crash, which explains the wide scatter. In static tests the peak load is measured, while in the dynamic tests most of the times Fmax is not even mentioned. Even more, some of the references studied calculate net mean crush force instead of Fmean: The difference between these two is that in the net mean crush force, the data of the first 10-15mm of crushing distance in the test is ignored. This is done to eliminate the influence of first peak in the calculation of Fmean and energy. The effect of this removal is particularly high on dynamic tests and untriggered samples. The samples tested in this project had no trigger, and its force displacement curves are compared in figure 5.2 to the ones obtained in a intermediate velocity axial test (100mm/s), in complete and triggered CTH sections of DP800 (exact same material) done previously at Corus RD&T. In this figure it can be seen the difference in Fmax due to the triggering mechanism, but shortly after the first peak, the untriggered samples curves seemed to have the same behavior as the triggered boxes. This fact is important to mention, because it can be said that the trigger only affects the first peak of the curve, and then this same behavior can be seen on figures 4.14 and 4.16, where the effect of BH appears to have an effect only on the first peak of the curves for both CTH and hydro formed sections. However, this similarity between the BH effect and the lack of a triggering device is valid only if progressive stable collapse mode is obtained. It appears like only the first few mm (less than 4 or 5) of the displacement would be affected, and then the influence of BH should only be measured on this region. Then net values of energy and Fmean would not show any difference in the performance of BH samples in comparison to unbaked samples.

68

200
Trigger Trigger

150 Froce [kN]

Untrigger Untrigger

100

50

0 0 5 10 15 Displacement [mm] 20 25

Fig. 5.2: 100mm/s axial crash on triggered (500mm long) and untriggered (125mm long) samples of DP800

Unstable folding seen on the samples seemed to compromise the measurement of peak load in these tests. This can be the reason of the high scatter seen, especially on the samples BH at 220C. By using triggers on the samples the scatter may be reduced and a more stable collapse mode may be obtained, but it also means that a bending load is being measured, and not a buckling load. In general, although differences due to BH are measured in all the tests (increase in peak load, Rp and Energy), they dont seem the have a significant influence in the materials crash performance. One of the critical points was the reduction of ductility seen on very extreme cases of strain hardening and BH, but the localized behavior of the crash tests prevent the sections to fracture. Even more, one of the basic problems trying to relate the actual value of Rp from the tensile tests to the data obtained in the crash tests was the measurement of E; the E of all the tensile tests is calculated by hand from the force displacement curve obtained in the tensile test. It was very difficult to estimate accurately the elasticity modulus of the samples, specially the ones that had pre strain. Samples of the same condition (same amount of pre strain and same BH temperature), presented E that varied more than 40000MPa. In fact, in the whole set of samples, E values from 175000MPa to 250000Mpa were calculated. Those differences in the measurements of E of the samples affected as well the calculation of Rp, and then necessarily the standard measurement of BH effect. Although all the values of Rp were calculated in the same way as explained in Chapter 3, still more detailed analysis should be done on this subject, particularly when the samples present yield point elongation. In these cases, small variations on the proof strain (0.2%) and measurement of E, could produce differences of 20 or 30MPa on Rp, which is quite significant if a BH response of 30 to 40MPa was measured. The difference on the measurement of Rp can be considered as a standard material deviation in this case (DP), because of their characteristic continuous yielding behavior. Maybe a more adequate way of measuring BH response on this kind of materials could be based on the change of Rm from the tensile tests. Previous studies done at Corus and SSAB [8] state that the BH response of DP steels are minor on crash performance. Still, the DP studied at Corus was a DP600, and the report of SSAB was talking specifically about axial crash only. It has also been reported that although an increase in Peak Load can be seen after BH, no significant influence of it was noticed on energy absorption or fracture strength [9, 14]. Considering that the BH mechanism is the

69

same for all DP steels, it is not surprising to find that like in DP600, BH did not affect crash performance of DP800 significantly. As indicated in chapter 2.1, the BH response is measured with a two-step tensile test consisting in pre straining and posterior reloading till fracture. Different studies have shown that the BH response in the uniaxial tension is different from that in the uniaxial compression and the one in biaxial state. Also, it has been observed that the strain-hardening model derived from the uniaxial tensile test does not describe well the hardening behaviour in the biaxial stretching state for high strength steels or bake hardening steels [37, 38]. In this project, it was concluded that the BH response could be estimated as constant independent of the direction and the pre strain level; according to this, to be able to model the behaviour of BH response of DP, first it is necessary to develop a work hardening and strain path dependent model. Previous studies have shown that after examining the collapsed CTH structures, it was seen that the component was subjected to a significant amount of compressive strain, and the formation of folds include bending-unbending mode failure, which would result in the structure experimenting strain reversal [21]; still, the material selection is being made based on tensile properties. One of the concerns about the influence of the BH on crash was that in extreme cases (high pre strain level and high BH temperature), the tensile tests showed a significant reduction in the total elongation. But, the reduction in ductility (based on tensile tests) is not that important apparently either: After measuring the strains on an axial crashed CTH sample, it is interesting to note that tensile strains in the extensional outward-outward buckling mode reach maximum values of +0.40 or +40%, and that compressive strains in the toroidal corner zone reach maximum values of 0.45 or 45% (Figure 5.2). Both values exceed by far the rupture strain of the material determined in uniaxial tensile tests. This is possible because the behavior of a component (stress and strains) in crash is localized; the material has the ability to sustain very high strains (over 40%) in the region adjacent to fracture in a tensile test. Clearly, the behavior under biaxial and triaxial stresses is required [36].

Fig. 5.2: Measured strains in a crashed CTH section. Taken from [36].

Still, the reduction in ductility seen on the tensile tests was only when the reloading was done on the same direction of the pre strain (with high pre strain and high temperature). The

70

highest pre strain levels on the CTH boxes are seen on the corners (around 10 to 15%), the strains generated due to bending or axial crash are in other direction, so the effect BH on the ductility reduction should not be significant. The BH diffusing C to lock dislocations in the same direction of the pre stain, so if the loading is done in other direction, there is no evident change in strength or elongation. According to the yield hinge model theory, in progressive folding the total energy absorption during an axial crash is composed of three mechanisms, which are bending of the material around the stationary hinge lines, bending of the material around the moving yield lines and stretching over the toroidal surface; two thirds of this energy is dissipated in compressive deformation of the material [36]. Also, companies are trying to shorten the BH periods, and lower the temperatures; it saves time and money. Renault actually makes various cycles of 15-20 minutes, at temperatures between 140 and 170C. Still Ford has 10 to 15 minutes cycles at 220C. In general, the BH response seen on the tests was low (crash and even tensile) when the baking temperature was 170C, so it is expected to be lower for shorter cycles with lower temperatures. In [3] it is declared that no significant difference was found by varying the BH time between 15 to 27 minutes, and that was found for temperatures between 150 and 240C; they explained this by saying that the material had already reached its ageing limit at that time. During this research no variations were introduced to the baking time, but the baking was different. Standard size tensile samples were BH in oil baths and the small size samples (as well as all the crash boxes) were BH on air oven. Although no difference could be noticed due to this factor, still the relation between time and temperature was not established, and no BH limit (highest possible strengthening effect due to BH) was discovered yet; finding this limit can be an important factor to optimize time and temperature during the process. In [36] it was also mentioned that there was no evidence that the established analytical theory (based on the yield hinge model and the kinematic approach in plasticity) that has successfully been used for mild steel could be used for HSS; it was also concluded that the BH mechanism works the same way for mild steels and HSS. In previous studies like [21], different parameters like steel strength, sheet thickness, cross sectional area, temperature and joining technique have been varied on CTH sections, and the influence of each one of them on axial crash has been analyzed independently. One of the main conclusions of [21] was that strain rate sensitivity reduces with increased steel strength; this conclusion has been found in several sources throughout this project. Also, it was found that steel strength has a limited influence on collapse resistance statistically compared to other factors such as temperature (at which the crash test is performed) and thickness (this last one specially).

71

CHAPTER 6: CONCLUSIONS AND RECOMMENDATIONS

An approximate value to the measurement of BH response of DP800 was given in this project; besides that, an approach to the influence of this BH response to the in service performance was analyzed, to evaluate the necessity of developing a BH model for FE analysis. In average, it can be estimated from the tensile tests that the absolute increase in yield strength due to BH are summarized in table 6.1. The increase seen at 170C was closely related to the one seen in the crash tests, but it is low compared to its scatter; as for the 220C results, unfortunately there not enough crash data to form a relation.
Table 6.1: Average increase in Rp due to BH

BH 170 BH 220

?Rp (MPa) 41 10* 88 8*

?Rp (%) 6.1 13.1

From the tensile tests it was concluded that the BH response for this material could be assumed as constant for a determined baking temperature, independent of loading (reloading) direction and pre strain level. This would mean that the response of the material on the tensile test would be determined by the work or strain-hardening curve; the BH would just give a constant increase (for a fixed temperature) as a parallel shifting of the BH response. This fact would generate the importance of developing a strain hardening model, rather than a BH model, basically because the latter one would just be a constant that has to be added to the stress strain curve in every point (at least until UTS). One of the major issues that was discovered during this project and that had not been taken into account at the beginning was the ageing at room temperature. It was clearly demonstrated that the DP800 selected had a very pronounced ageing response; It would be useful to measure this temperature-time relation with precision. It is still not clear which are the limits to this ageing at room temperature. Even more, it is not known yet if ageing at room temperature for a long time, responses like the ones o btained at 220C (with a determined pre strain level) could be reached as well. On the ageing test the increase of Rp measured after 2% pre strain was very low compared to the one obtained at 10%, but in figure 5.1 it can be noticed that when the ageing time at room temperature is longer (approximately 2 months for the small samples and no ageing for large samples), the increase in Rp at every pre strain level appears to be constant. Even more, does the material continue its ageing process after being BH? Also, the ageing test was done only on one sample at each condition. Longer ageing periods like 2, 6 and even 12 months are recommended, with different pre strain levels. In chapter 5 it was concluded that BH on DP800 is not significant enough to take into account on crash performance, but it is still important to distinguish two factors here: First, saying that BH on this material is not significant enough on crash components does not mean that BH has any influence at all on the material. It is clear that there is a change on the tensile properties of the material when it is baked, and this should be taken into account for some other applications. An increase of over 10% on the yield strength of a component can be quite considerable and useful. Second, it is important also not to generalize DP800, because as was discovered on the microstructure analysis, the amount of Martensite on the subject material was very high for a regular DP grade. In the other DP800 tested, the results of the tensile tests presented smoother continuous yielding, even after BH at 170C (appendix A, figure A.1). When done at 220C, it presented a yield point elongation plateau, but never a critical ReH. It was also interesting to see that the elongation on this material was bigger, and even the uniform plastic elongation after pre strain 10% and BH at 220C was significant (longer than the one at

72

BH170). In this DP800, the BH response (increase in Rp) at 220C for the 2% and no pre strain samples appears to be bigger than for the higher pre strain level. This can be explained due to the change on the continuous yielding to a more clear yield point for the BH 220C. The main point is that the response of a material to BH depends on its the microstructure; in order to give a generalized BH response of DP800, other DP800 (maybe different providers) should be tested. Other aspect that should be discussed is the proper measurement of BH response; as mentioned on chapter 5, the measurement of Rp and E of the tensile samples was complicated, due to yield point elongation or continuous yielding behavior. In most of the cases this implied having differences in Rp (on same kind of samples) that were larger than the actual BH response than wanted to be measured; this makes very hard to determine the BH response. Maybe some other parameter to measure BH response should be established, in order to avoid this problem; for example measuring Rm, or measuring the maximum value of the true stress strain curve. A positive thing about Rm, is that the analytical calculations of energy absorption and mean load are based on this parameter. Also, it was seen that the standard tensile test sometimes gives results that can be influenced by the test setup and are not really a material effect. The automatic change of speed during the standard test sometimes makes the machine try to identify none existing yield point phenomena cases. The ideal case would be a test where a constant strain rate can be maintained. Besides the apparent small influence of BH on In service Performance, when dealing with complex structures, the simulation of strain rate, strain path and yield locus on crash performance are still greatly influenced by the material model used; differences up to 20% on energy absorption have been found for the same simulation depending on the material model being used [39]. These differences are already much higher than the change brought to the structure by BH, which makes it even less important to take it into account yet. First, strain path and strain rate dependent material models should be developed; these characteristics seem to be much more important to understand the crash behavior of DP800 than BH, and a proper development of a model including these factors will enable accurate FE analysis of materials that are gaining more importance every day in the automotive industry. In addition to that, the crash experiments done showed that the influence of BH on the components was not very significant, and in several cases, the scatter already was as extensive as one half of the BH response. Still, to confirm this it would be highly recommended to perform more crash tests (both axial and bending); this because in the bending test, only 2 CTH samples BH at 170C were tested, and only one sample at 220C, which does not give any certainty about the repeatability of the test results. And in the axial tests, even though three CTH sections were tested for each temperature, it was not possible to determine a relevant measurement energy absorption because the displacement was too small (due to the size of the component); ideally, the test should be done on complete specimens (500mm long) to have the possibility of analyzing the influence on the foldings, the mean load and the total energy that is possible to absorb. Even more, the test could be done dynamically as well to see the influence of BH at a different strain rate. It would also be preferred to have the tests on samples with triggering mechanisms: although the influence on peak load would not be accurately measured (it would be a bending load and not a buckling load), the mean load and energy absorption would represent very important parameters to be taken into account to compare BH influence on crash performance. Influence of steel strength on collapse resistance is not defined exclusively by quasi-static tensile properties, since strain-rate sensitivity and work hardening all have a role to play. In general, the influence of steel strength is reduced with increasing crash velocity owing to reduced sensitiv ity of higher strength steels to strain-rate. That is why another important factor that should be taken into account to try to determine the BH response is strain rate. The uniaxial tensile tests on this project was done at a fixed strain rate (low strain), but it can be important to determine the influence of a higher rate on the BH response, as in an actual crash the rate would be higher, and it is known that AHSS have a reduced strain rate sensitivity. However, more important than the increase in energy with strain-rate is the actual

73

energy absorbed which is dependent upon material strength, strain rate sensitivity and strain hardening behavior. Although tensile testing will always be important to determine material characteristics, to relate to in serve performance, other kind of testing should be done. Biaxial or triaxial straining, or behavior of steel sheets under combined compression and bending, which is relevant to progressive folding, is largely unexplored yet. More work has to be done to understand completely the material behavior to be able to predict better the collapse behavior of thin walled structures; and more specifically for this project, the influence of BH on these kind tests should be measured. This testing would probably give a more accurate insight about how does the BH affect the crash performance.

74

REFERENCES
[1] Moreno, J.J. Metallurgical Phenomena Governing The Bake Hardening Response of Bake Hardenable Steels. 2001 [2] Bleck, W., Bruhl, S., Gerber, T. B ake-hardening properties of TRIP and Dual-Phase steels, in STEEL, FUTURE FOR THE AUTOMOTIVE INDUSTRY, pg. 489. June 2005. [3] Palkowski, H., Anke, T. Using Bake-hardening effect of hot rolled multiphase steels for weight optimized components, in STEEL, FUTURE FOR THE AUTOMOTIVE INDUSTRY, pg. 497. June 2005. [4] Bosch, M. Restraint Systems, in Vehicle Safety Lectures at TU/e. January 2005. [5] Smith, A.W.F. A review of the production and application of the automotive impact beams, Corus RD&T, November 2003. [6] Carless, S. 3 point bending tests on CTH sections. Corus RD&T internal report, 2005. [7] Dieter, G.E. Mechanical Metallurgy McGraw-Hill book Co. UK, 1988 [8] Carless, S. The Bake Hardening Response of DP800, DP1000 and TRIP800. Corus RD&T internal report, 2005 [9] Sperle, J., Lundh, H. Strength and Crash Resistance of Structural Members in High Strength Dual Phase Steel Sheet. Skandinavian Journal of Metallurgy 13, 1984. [10] Nesterova, E., Bacroix, B., Teodosiu, C. Microstructure and texture evolution under strain path changes in low carbon Interstitial-free Steel. Metallurgical and Materials Transactions A, Volumen 32A, October 2001 [11] De Winter, Heijne, J. Effect of pre deformation and strain rate on stress-strain data for an IF grade. Corus RD&T internal report, September 2004. [12] Carless, S., Heijne, J. Influence of Pre strain and Strain path change on the mechanical properties of Low carbon, Carbon manganese, Micro-alloyed and Dual Phase steels. Corus RD&T internal report, August 2002. [13] Carlsson, B., Larsson, J., Nilsson, T. Dual Phase Steels for Auto Body: Design forming and Welding Aspects. SSAB Tunnplat AB, Borlange, Sweeden. 1990 [14] Sperle, J., Olsson, K. High Strength Automotive Sheet Steels for Weight Reduction and Safety Applications. SSAB Tunnplat AB, Borlange, Sweeden. High Strength Steels for Automotive Symposium Proceedings Pg 65-77. 1994 [15] DeArdo, A., Garcia, I. New Ultra-Low Carbon High Strength Steels with Improved Bake Hardenability for Enhanced Stretch Formability and Dent Resistance. University of Pittsburg, 2003. [16] Baker,L., Daniel, S., Parker, J.D. Metallurgy and processing of ultra low carbon bake hardening steels; Materials Science and Technology; Vol.18; April 2002; 355-368. [17] Christen, J., Rubianes, J. ; The bake hardenable steels for automotive outer body panels: Correlation between the BH measurement and the dent resistance; 40th MWSP Proc. ISS; 1998; 77-81. [18] Kim, S.; Effect on the chemical composition and processing variables on the bake hardenability of ULC high strength steel; Research work in BAMPRI and POSCO. [19] Leslie, W.C., The physical metallurgy of steels; 1981 (first published). [20] Norman, D. The effect of BH on F2C in DP600 Corus RD&T internal report, 2004 [21] ECSC Report; Stiffness, energy absorption, fatigue of high strength steel structures in relation to applied joining and forming technologies and mechanical properties. 2002. [22] Thornton, P., Mahmood, H., Magee, C.; Energy absorption by structural collapse, IN: Jones, N., Wierzbicki, T., Structural Crashworthiness, Butterworths, London, 1983 [23] International Journal of Crashworthiness, 2003-2005 [24] Magee, C.L., Thornton, P.H.; Design Considerations in Energy Absorption by Structural Collapse, SAE Paper no. 780434, 1978. [25] Automotive Circle International Conference, EuroCarBody Frankfurt, October 2004 [26] Automotive Circle International Conference, Low cost concepts versus Lightweight innovations, Frankfurt, June 2005 [27] vStijn, I.; Influence of the hydroformed process on the crash performance of crash boxes. Corus RD&T internal report, October 2000. [28] vStijn, I., vGoethem, R., Heijne, J.; Hydroforming and crash. Corus RD&T internal report, August 2005.

75

[29] vStijn, I., vGoethem, R., Heijne, J.; Crash performance of Dual Phase and TRIP steels. Corus RD&T internal report, March 2006. [30] vStijn, I., vGoethem, R., Heijne, J.; Hydroforming and Crash of Dual Phase and TRIP material. Corus RD&T internal report, May 2006. [31] Carless, S. Influence of section design and steel strength on the resistance to bending under 3 point loading, Corus RD&T internal report, November 2000. [32] vStijn, I., vGoethem, R., Moolevliet, T..; Crash Performance of Dual Phase Materials. Corus RD&T internal report, January 2005. [33] Jones, N.; Structural Impact , Cambridge University Press, 1989 [34] ECSC Report; The relative impact performance of press formed, hydroformed, and roll formed structures and the application of patch piece and tailor welded blank technique for optimum mass efficiency. 1999 [35] ECSC Report; The relative impact performance of press formed, hydroformed, and roll formed structures and the application of patch piece and tailor welded blank technique for optimum mass efficiency. 2004 [36] Schneider, F. D.; Impact Behavior of Thin Walled Structural Sections made fro High Strength Steel, PhD Thesis, University of Liverpool. December 1999 [37] An, Y., Heijne, J., Elliot, L., dWolff, S., Vegter, H.; The Vegter material model for high strength and bake hardening steels. Corus RD&T internal report, September 2001. [38] An, Y., Heijne, J., dWolff, S.; Strain State Dependence of Bake Hardening Response in a BH220 (5PAB) Material. Corus RD&T internal report, April 2001. [39] Carless, S., tHorn, C., vStijn, I.; Influence of material model on crash simulations, in STEEL, FUTURE FOR THE AUTOMOTIVE INDUSTRY, pg. 547. June 2005. [40] Hanlon, D.; The Bake Hardening Response of Continuously Annealed 3C56. Corus RD&T internal report, March 2004. [41] http://www.corusautomotive.com/file_source/StaticFiles/Microsites/Automotive/Technical/Final _CBM_Dupla_Paper.pdf [42] http://www.us.cbmm.com.br/english/sources/techlib/info/dualph/dualphas.htm

76

ACKNOWLEDGEMENTS
First of all, I would like to thank Stephen Carless for giving me the chance of working with him (not once but twice!), and for his extracurricular support and understanding during this year. Besides, thanks to Jens de Kanter for his extreme effort and help to finish this report. Also, thanks to Willem Witteman and Marc Geers in Eindhoven for their help during the project. Also, thanks to everyone in the Corus team who helped me with the different testing involved, or their knowledge about the subject to make more sense of what I was supposed to write: Ronald v an Goethem, Jan Heijne, Yuguo An, Henk Vegter, Carel ten Horn, Toni Chezan, Dave Hanlon and Peter van Liempt. Thanks as well to the Dutch real people (Pascal, Ruud and Daan: hartelijk bedankt jongens!) and Foreign community of stagiers (French, German, Danish, Turkish) for all the fun times we had. Special thanks to Elizabeth and Sullivan (what a naughty couple!!), and the Corrosive Spanish Speaking community (Carlos, Fernando, Jose and Alessandro) for taking care of me in every possible way (thanks to them I never missed a meal, always had a place to sleep, and above all, always had somebody to talk to). Also to Ms. Bunk, whose blushing face gave the perfect amount of fun to start working every single day. To all the previously mentioned people, I hope they know that there will always be a small Colombian guy somewhere in the world who remembers you and who will always be there when needed: Pa las que sea. Maybe even closer than you thinkor wish for!! Gracias tambin: Oscar y Karin por convertirse en mis padres adoptivos y familia en Holanda (un milln de gracias!!!) A mis amigos del cartel latino en Eindhoven (Hirsa, Pancho, el Padrino, el Turco, Gilito, Carlos (marcos?), Armenzo) por su apoyo y compaa. A todos los que entendieron por lo que estoy pasando, fueron incondicionales y todava estn pendientes de m. Finalmente, a mi FAMILIA, por mostrarme este ao ms que nunca que es lo que de verdad significa esa palabra: FAMILIA. No conozco alguien ms afortunado que yo en ese aspecto, y por eso, como ya lo dije antes, GRACIAS, TOTALES!!!! Este trabajo esta dedicado a Camila, sencillamente porque creo que es imposible encontrar una mejor razn para hacer esto y para dar el siguiente paso. Solo espero llegar a ser todo lo que esperas de m, y ms

77

APPENDIX
A. TENSILE TEST RESULTS FOR OTHER AHSS
All the tests done on this appendix were done on standard size samples, reloading on the same direction of the pre strain (in pre strain cases), and the samples were taken on D direction. All tests presented were done on triplicate, under the same conditions described on 3.2 for standard size samples.

DP800 (20040516)

1000 800 600 400 200 0 0 0,02 0,04 0,06 0,08 0,1 0,12 0,14
True Strain
Fig. A.1: BH response of DP800 (20040516).

True Stress (MPa)

AR BH170 BH220 2% BH170 2% BH220 5% BH170 10% BH170 5% BH220 10% BH220

Table A.1: Overview tensile test DP800 (20040516)

DP800 Rp (MPa) Rm (MPa) Energy

AR BH170 BH220 2% BH170 2% BH220 5% BH170 5% BH220 10% BH170 10% BH220 540 586 683 750 812 836 901 924 983 837 838 846 853 877 890 933 959 1002 97 92 89 84 75 58 48 12 47

78

TRIP800 (20030305)

1000 True Stress (MPa) 800

AR

600 400 200 0 0 0,05 0,1 True Strain 0,15

BH170 BH220 2% BH170 2% BH220 5% BH170 10% BH170

0,2

Fig. A.2: BH response of TRIP800

Table A.2: Overview tensile test TRIP800

TRIP800 Rp (MPa) Rm (MPa) Energy

AR BH170 BH220 2% BH170 2% BH220 5% BH170 5% BH220 10% BH170 10% BH220 510 526 567 651 694 763 X 861 X 793 794 801 814 828 850 X 907 X 164 158 154 156 146 130 X 102 X

79

TRIP600 (20050094):

800 700 True Stress (MPa) 600 500 400 300 200 100 0 0 0,02 0,04 0,06 0,08 True Strain 0,1 0,12
AR BH170 BH220 2% BH170 2% BH220 5% BH170 10% BH170

Fig. A.3: BH response of TRIP600

Table A.3: Overview tensile test TRIP600

TRIP600 Rp (MPa) Rm (MPa) Energy

AR BH170 BH220 2% BH170 2% BH220 5% BH170 5% BH220 10% BH170 10% BH220 416 434 461 502 542 595 X 682 X 631 631 633 647 655 671 X 711 X 67 63 61 58 52 41 X 9 X

80

DP1000 (20050095):

1400 1200

True Stress (MPa)

1000 800
AR

600 400 200 0

BH170 BH220 2% BH170 2% BH220 5% BH170 10% BH170

0,02

0,04

0,06 0,08 True Strain

0,1

0,12

0,14

Fig. A.4: BH response of DP1000

Table A.4: Overview tensile test DP1000

DP1000 Rp (MPa) Rm (MPa) Energy

AR BH170 BH220 2% BH170 2% BH220 5% BH170 5% BH220 10% BH170 10% BH220 736 775 874 1051 1123 1106 X 1230 X 1063 1069 1080 1114 1136 1231 X 1321 X 62 57 58 4 3 6 X 5 X

81

CP800 (20050186):

1000 800 600

AR

True Stress (MPa)

400 200 0 0 0,01 0,02 0,03 0,04 0,05 True Strain 0,06

BH170 BH220

0,07

0,08

Fig. A.5: BH response of CP800

Table A.5: Overview tensile test CP800

CP800 Rp (MPa) Rm (MPa) Energy

AR 681 972 63

BH170 713 969 61

BH220 746 974 62

82

B. 3 POINT BENDING TEST FOR OTHER AHSS

For DP600:

50 45 40 35 FORCE (kN) 30 25 20 15 10 5 0 0 20 40 DISPLACEMENT (mm)

Fig. B.1: 3PB of DP600 with BH

DP600 AR 1 DP600 170 1 DP600 AR 2 DP600 170 2

60

80

In this case, it was not possible to perform the test at 220C, but 2 different grades of DP600 were tested.
Table B.1: Overview 3PB of DP600 with BH MATERIAL DP600 AR 1 DP600 BH170 1 DP600 AR 2 DP600 BH170 2 Peak Load (kN) Energy (J) %Fmax %Energy

15,6 16,0 21,5 22,3

1497,9 1563,8 2124,8 2199,0

0,0 2,6 0,0 3,7

0,0 4,4 0,0 3,5

Two DP600 (from different providers) were tested at 170C. The yield point of one is higher than the other, but basically the bake hardening response was the same: 1KN increase in peak load (less than 5%). It showed a very similar response as the DP800.

83

For DP1000:

70 60 FORCE (kN) 50 40 30 20 10 0 0 20 40 60 DISPLACEMENT (mm) 80

DP1000 AR DP1000 BH170 DP1000 BH220

Fig. B.2: 3PB of DP1000 with BH

Table B.2: Overview 3PB of DP1000 with BH MATERIAL DP1000 AR DP1000 BH170 DP1000 BH220 Peak Load (kN) Energy (J) %Fmax %Energy

28,3 28,8 29,8

2848,5 2911,1 3047,4

0,0 1,9 5,2

0,0 2,2 7,0

Although in this case a difference between the response due to temperatures can be seen, still the increase in peak load is less than 2KN and in the range of 5%.

84

For TRIP800:

35 30 FORCE (kN) 25 20 15 10 5 0 0 20 40 60 DISPLACEMENT (mm) 80

TRIP800 AR TRIP800 BH170 TRIP800 BH220

Fig. B.3: 3PB of TRIP800 with BH

Table B.3: Overview 3PB of TRIP800 with BH MATERIAL TRIP800 AR TRIP800 BH170 TRIP800 BH220 Peak Load (kN) Energy (J) %Fmax %Energy

14,7 15,2 15,3

1511,0 1526,0 1537,3

0,0 3,6 4,1

0,0 1,0 1,7

The TRIP steel showed the lowest response of all; in the graphs it can be seen that BH has hardly an influence in the behavior of the bending test.

85

For CP800:

50 45 40 FORCE (kN) 35 30 25 20 15 10 5 0 0 20 40 DISPLACEMENT (mm)

Fig. B.4: 3PB of CP800 with BH

CP800 AR CP800 BH170 CP800 BH220

60

80

Table B.4: Overview 3PB of CP800 with BH MATERIAL CP800 AR CP800 170 CP800 220 Peak Load (kN) Energy (J) %Fmax %Energy

15,5 16,4 17,5

1687,1 1869,3 2020,5

0,0 5,8 13,1

0,0 10,8 19,8

By the other hand, CP gave the biggest and clearer response to the BH of all the samples tested. The increase in peak load was over 10% and apparently the difference was increasing as the displacement increases as well. It also showed a difference depending on the temperature of the BH. To summarize, it can be seen that the Dual Phase grades presented the same kind and range of behavior; even more, the measurements made showed that the percentage increase in Peak load and Energy absorption for 15mm and 80mm (total) displacement were very much alike. Although the TRIP grade had a much lower response to the BH (almost no difference at all), it also maintained its performance for 15 and 80mm displacement. Unlike them, the Complex Phase (CP) showed as higher percent increase when the total displacement was considered.

86

14,0 12,0 Fmax % increase 10,0 8,0 6,0 4,0 2,0 0,0 DP600 DP600(2) DP800 DP1000 TRIP800 CP800 BH 170 BH 220

Fig. B.5: Normalized increase in Peak Load due to BH for various AHSS

25,0 20,0 Energy % increase 15,0 10,0 5,0 0,0 DP600 DP600(2) DP800 DP1000 TRIP800 CP800 BH 170 BH 220

Fig. B.6: Normalized increase in Energy absorption due to BH for various AHSS

As mentioned previously, the Dual phase steel showed a similar kind of response independent of there as received strength level (DP600, 800 and 1000); the TRIP practically didnt show BH response at all, while the Complex Phase showed the highest increase (%) as well as a difference in response due to temperature. As for the cracks, all the samples presented the same kind of behaviour as the one described for DP in 3.2.2, except the TRIP. All the TRIP samples presented cracks on the corner inner side of the samples (Fig B.7). Apparently the cracks were also bigger, but this was independent of the BH temperature, because the not BH samples showed the same results.

87

Fig. B.7: Fracture of CTH of TRIP after 3PB

88