Anda di halaman 1dari 9

INTERNATIONAL INSTITUTE OF WELDING

Slovenian Delegation

Doc. 212-985-00

THE CHAOTIC BEHAVIOR OF THE MATERIAL TRANSFER IN GMAW


- preliminary report -

by Marjan Suban

Welding Institute, Ptujska 19, 1000 Ljubljana, Slovenia

Presented to Study Group 212 at the


Annual Assembly of the International Institute of Welding
Florence, July 2000
Abstract

The paper presents some thoughts on the analysis of the material transfer during gas-shielded
metal-arc welding based on nonlinear dynamics. At the begining of the paper, an introduction
to chaos theory is given. Based on this theory some extensive research work on dripping
faucet has been done by many researcher, which is presented in the second section. Some
similarities can be found between the material transfer in GMAW and the water drops falling
from a dripping faucet, which is one of the basic models of chaotic system. Based on
experimental work, we can examined time series of welding voltage and current, identifying
signatures of weld material droplet separation in the short-circuit, globular and spray
transfer mode. Identification of chaotic behavior in the material transfer could permit the
development of improved control of welding through well-known methods of control of chaos.
Improved control can lead to improved arc stability, which is important for determination of
weld quality.

Key words: gas-shielded metal-arc welding, material transfer, dripping faucet, chaos

1 INTRODUCTION

A dynamical system consists of an abstract phase space or state space, whose coordinates
describe the dynamical state at any instant; and a dynamical rule which specifies the
immediate future trend of all state variables gives only the present values of those same state
variables. Phase space is collection of possible states of a dynamical system. A phase space
can be finite (e.g. for the ideal coin toss, we have two states heads and tails), countably
infinite (e.g. state variables are integers), or uncountably infinite (e.g. state variables are real
numbers).

Dynamical systems are "deterministic" if there is a unique consequent to every state, and
"stochastic" or "random" if there is more than one consequent chosen from some probability
distribution (the "perfect" coin toss has two consequents with equal probability for each initial
state). But there also exists a system, which contains "deterministic" and "stohastic" properties
and is called "chaotic system".

Deterministic chaos was brought to the fore of our scientific awareness by happenstance in
the early 1960s by meteorologist Edward Lorenz [1]. Actually, many properties of chaos had
been worked out prior to Lorenz's discovery by Henri Poincare [2] in the nineteenth century,
by Georg Birkhoff [3] in the 1920s and many other scientists. The word "chaos" was first
used in a technical context by T.Y. Li and J. Yorke [4].

It has been said that "Chaos is a name for any order that produces confusion in our minds."
(George Santayana, thanks to Fred Klingener for finding this). However, the mathematical
definition is, roughly speaking: chaos; effectively unpredictable long time behavior arising in
a deterministic dynamical system because of sensitivity to initial conditions.

Lorenz was attempting to model convection in the atmosphere by computer calculations.


Lorenz truncated the complicated differential equations which describe the model of

1
atmosphere convection and obtained the three coupled, first order, nonlinear differential
equations of the so-called Lorenz model:

X& = −σ ⋅ X + σ ⋅ Y (1.a),
Y& = r ⋅ X − Y − X ⋅ Z (1.b),
Z& = X ⋅ Y − b ⋅ Z (1.c),

where σ and b are dimensionless constants which characterize the system, and r is the control
parameter. A numerical analysis of this apparently simple set of nonlinear differential
equations shows that its variables X, Y and Z can exhibit chaotic motion above threshold value
rc (see Fig. 1).

Fig. 1: The Lorenz attractor

Some nonlinear systems which display deterministic chaos are: forced pendulum, lasers,
nonlinear optics devices, chemical reactions, plasmas with interacting nonlinear waves,
stimulated heart cells,...

One of the basic properties which deterministic chaos manifests is "a sensitive dependence on
initial conditions" (the butterfly effect). It becomes therefore practically impossible to predict
the long-time behavior of these systems. It must be emphasized that a deterministic dynamical
system is perfectly predictable given perfect knowledge of the initial condition, and is in
practice always predictable in the short term. The key to long-term unpredictability is a
property known as sensitivity to (or sensitive dependence on) initial conditions.

Often chaotic dynamical systems exhibit fractal structures in phase space. However, there is
no direct relation. There are chaotic systems that have nonfractal limit and fractal structures
that can arise in nonchaotic dynamics.

The second property of the chaotic systems is an attractor. Informally an attractor is simply a
state into which a system settles (thus dissipation is needed). Thus in the long term, a
dissipative dynamical system may settle into an attractor. Before Chaos (BC), the only known
attractors were fixed points, periodic orbits (limit cycles), and invariant tori (quasiperiodic

2
orbits). In a famous paper in 1963, Ed Lorenz discovered that simple systems of three
differential equations can have complicated attractors. The Lorenz attractor (with its butterfly
wings reminding us of sensitive dependence - see Fig. 1) is the "icon" of chaos. Lorenz
showed that his attractor was chaotic, since it exhibited sensitive dependence. Moreover, his
attractor is also "strange," which means that it is a fractal.

The term "strange attractor" was introduced by Ruelle and Takens in 1970. Unfortunately, the
term "strange attractor" is often used for any chaotic attractor. However, the term should be
reserved for attractors that are "geometrically" strange, e.g. fractal. One can have chaotic
attractors that are not strange, but there are also strange, nonchaotic attractors.

The tool which is commonly used for representation od chaotic behavior of systems is a map
(Fig. 2). A map is simply a function f, on the phase space that gives the next state f(z) (the
image), of the system given its current state z. Often you will find the notation z' = f(z), where
the prime means the next point, not the derivative. A function must have a single value for
each state, but there could be several different states that give rise to the same image.

T n+1

Tn

Fig. 2: Example of the map (Henon map)

The most important indicators of chaotic behavior of dynamical systems are Lyapunov
exponents. Lyapunov exponents measure the rate at which nearby orbits converge or diverge.
There are as many Lyapunov exponents as there are dimensions in the state space of the
system, but the largest is usually the most important. Roughly speaking the (maximal)
Lyapunov exponent is the time constant λ, in the expression for the distance between two
nearby orbits exp(λ⋅t). If λ is negative, then the orbits converge in time, and the dynamical
system is insensitive to initial conditions. However, if λ is positive, then the distance between
nearby orbits grows exponentially in time, and the system exhibits sensitive dependence on
initial conditions (which predict chaos).

There are basically two ways to compute Lyapunov exponents. In one way one chooses two
nearby points, evolves them in time, measuring the growth rate of the distance between them.
This is useful when one has a time series, but has the disadvantage that the growth rate is
really not a local effect as the points separate. A better way is to measure the growth rate of

3
tangent vectors to a given orbit. More precisely, consider a map f in an m dimensional phase
space, and its derivative matrix Df(x). Let v be a tangent vector at the point x. Then we define
a function

1
L( x, v) = lim ln ( Df n ( x)v ) (2).
n→∞ n

2 THE CHAOTIC BEHAVIOR OF THE DRIPPING FAUCET

In 1977 Rössler suggested that the drops falling from the leaky faucet might provide a
familiar example of a dynamical system capable of exhibiting chaotic behavior [5]. The
dripping faucet is since then well known model system, used for illustrating the appearance of
chaotic behavior in nonlinear systems [6]. The experiment (see Fig. 3) studies the statistic of
liquid drops dripping from a leaky tap when the mean dripping rate (control parameter of
system) is increased.

Fig. 3: Schematic setup of the apparatus [6]

In the research work [2] the measured quantity was the drip interval (time interval between
succesive liquid drop detachments). The sequence of these drip intervals (t1, t2, t3, ..., tn, ...) is
usually used to describe the dynamics of the system. Plotting the results in time-delayed
coordinate system - map (tn+1 vs tn), the experiments revealed the following results:
• at low dripping rates the system is periodic (see Fig. 4a);
• above a critical dripping rate the system exhibits chaotic behavior characterized by
qualitatively different types of strange attractors (see Fig. 4b).

4
a) b)

T n+1 T n+1

Tn Tn

Fig. 4: Map of drop interval


a) period two drop regime
b) chaotic behavior - (1-d) map

In Fig. 4b no periodicity is apparent, but the system clearly is not totally random either; a
considerable degree of determinism connects one drop interval with the next. The time vs.
time map could be approximated by a parabola with some "pure stohastic" noise added.

The dripping faucet can also be presented as a naive analog model. The term "model" is
commonly used in physics to describe a representation of a physical system not involving all
of its relevant variables, but rather a much simpler set of variables and dynamics between
them which still manages to preserve some aspect of the qualitative behavior of the complete
system. Shaw [6] constructed a model of this type for the dripping faucet. The model is
sketched in Fig. 5. This simple model is so crude as to have perhaps little or no predictive
power, but it may be useful in describing to first order why the dripping faucet behaves as it
does.

A mass, representing the drop, grows linearly in time, stretching a spring, representing the
force of surface tension. This simple analog model shows that, as a control parameter m& is
increased, the analog model enters into periodic motion, then period doubles (Fig. 4a), then
goes chaotic (Fig. 4b).

5
F=ma
m ..
. g-kx=mx
m = const.
.
∆ m x x=x
0

X
m
X0
∆m

Fig. 5: Analog model for water drop at low flow rates [6]

3 SEARCH FOR TIME INTERVAL DYNAMICS IN ARC WELDING

We can find similarity between the dripping faucet and gas-shielded metal arc welding
(GMAW) in droplet processes, in which a material separates from the molten wire and "falls"
to the plate (material transfer). Principle transfer modes in GMAW are:
• short-circuit,
• globular,
• spray.

The physics of the droplet transfer process is more complicated than that of the dripping
faucet. In addition to surface tension and gravity, and the interaction with the power supply
dynamical response, the conducting droplets are subjected to complicated electromagnetic
forces, along with arc plasma forces. Nonetheless, generic properties of the droplet release
dynamics could just as well give rise to low dimensional dynamics as they can in the
conceptually simpler, dripping faucet case. Indeed, in the globular mode, where the droplets
are more massive, gravity and surface tension tend to dominate as they also do in the dripping
faucet case.

We can examine the dynamics of the droplet transfer process for GMAW using common
welding equipment. We can perform welding in short-circuit, globular and spray modes. In
the case of short-circuit mode we can measure two different time intervals: short-circuit
duration and arc duration (see Fig. 6a). The sum of these two times produces the time between
two sucessive short-circuits. In the case of globular and spray modes we can measure time
between two droplet separations (see Fig. 6b).

The droplet events can be identified from visual observation of droplet processes (as in a case
of dripping faucet - Fig. 3), or from air or plate-borne acoustic emission or from signature
effects in weld voltage and current (Fig. 6). Perhaps, the most problematic will be observation

6
of droplet separation in the spray mode because of the weakness of the event signatures.
Construction of map time series with some missing points or extra points is not very
competent.

tc
a) to tk
U

I t [ms]

t [ms]
t c [ms] - period
b) tc
t o [ms] - arc burning
U
t k [ms] - short-circuiting

I t [ms]

t [ms]

Fig. 6: Time intervals in GMAW


a) short-circuit mode
b) globular mode

Based on measured time intervals we can diagnosticate them, with the emphasize on chaotic
behavior.

The next step in our research work will be construction of an analog model for material
transfer in different modes based on a model for the dripping faucet.

One of the steps in our research could also be searching for chaotic behavior in known models
of GMAW processes, such as these fifth-order nonlinear differential equations developed by
the INEEL and ISU researchers [7]:

x& 1 = x 2 (3.a),
− K ⋅ x1 − B ⋅ x 2 + Ftot
x& 2 = (3.b),
x3
x& 3 = M R ⋅ ρ w (3.c),
MR
x& 4 = u1 − (3.d),
π ⋅ rw2
u − (Ra + Rs + RL ) ⋅ x5 − V0 − Ea ⋅ (C ⋅ T − x 4 )
x& 5 = 2 (3.e),
Ls

where x1 is droplet displacement, x2 is droplet velocity, x3 is droplet mass, x4 is stick-out, x5 is


current, Ftot is the total force acting on the droplet, and all other variables are constants or
factors based on welding parameters.

7
4 INSTEAD OF A CONCLUSION

The question is why we are trying to find chaotic behavior in material transfer? The answer is:
controlling chaos.

Even if the behavior is chaotic, it can be controlled once the rules are known. Within the
chaos exist unstable periodic orbits: state values that, if they were supplied and iterated with
infinite precision, would repeat over and over. Of course, any slight imprecision in this state
causes successive iterates to wander away from the cyclic order and take on the complexity of
the chaotic attractor. Nonetheless, in some simple situations the chaotic dynamics can be
forced to track arbitrarily close to a specified unstable orbit by aplying small, well-timed
perturbations. The control of chaos by unstable periodic orbits embedded in a chaotic attractor
was first originated in the work of Hubert [8] and Ott et al. [9].

Furthermore, it is possible to switch efficently from one unstable periodic orbit to another.

We recall, finally, that the possibility of design for chaos has long been advanced as a means
of controlling some nonlinear effects. Could this be done for welding, with well-known
techniques for controlling chaos used to advantage?

REFERENCES

1. Lorenz E.: Deterministic non-periodic flow. Journal of Atmosferic Science 20, 1963, p.
130-141.
2. Poincare H.: Les methodes nouvelles de la mechanique celeste. American Institute of
Physics, 1892-1899.
3. Birkhoff G.: Dynamical Systems. American Mathematical Society, 1927.
4. Li T.Y., Yorke J.: Period three implies chaos. American Mathematical Monthly, 82, 1975,
p. 985-992.
5. Rössler O. in : Synergetics - edited by Haken H. Springer Verlag, Berlin, 1977, p. 174-
183.
6. Shaw R.: The Dripping Faucet as a Model Chaotic System. Aerial, Santa Cruz, 1984
7. Moore K.L., Naidu D.S., Abdelrahman M.A., Yesildirek A.: Advanced welding control
project: Annual report FY96. Techical Report, ISU, Pocatello, 1996.
8. Hubler A.W.: Adaptive control of chaotic systems. Helv. Physics Acta 62, 1989, p. 343-
346.
9. Ott E., Grebogi C., Yorke J.A.: Controlling chaos. Physical Revue Letters 64, 1990, p.
1196-1199.
10. Neda Z., Bako B., Rees E.: The dripping faucet revisited. Chaos, 6, 1996, p. 59-62.
11. Cahalan R.F., Leidecker H., Cahalan G.D.: Chaotic Rhythms of a Dripping Faucet.
Computers in Physics, 8/9, 1990, p. 368-382.

Anda mungkin juga menyukai