Chapter 1

Natural Response
A signicant portion of these notes are concerned with the study of nitedimensional, linear time-invariant (LTI) systems. We will dene this term with more care in section 1.3.2. Such systems can be described by niteorder linear constant coecient dierential equations. Such models are widely applicable to physical systems. In this chapter, we will be primarily concerned with the natural response of such models, which is dened as the response which occurs solely from initial conditions with no other inputs. The natural response is also known as the unforced response or characteristic response. The model dierential equation for such a system is homogeneous, in that there is no forcing term. There is a beautiful property of LTI systems: the homogeneous or natural response can be very simply found. It is composed of weighted sums of functions est , where s is possibly complex (or most generally such functions multiplied by polynomials in the time variable t). This is a statement about the solution of dierential equations. However, it is a remarkable empirical result that such dierential equations well-describe many physical systems. Said another way, the types of natural responses discussed below can be easily observed in an experimental context, and in observations of many physical phenomena. The natural response ties things together. A further surprising result is that real-world systems are frequently able to be represented in terms of very simple models of rst- or second-order. When higher-order models are required, these systems have responses com posed of sums of rst- and second-order responses. So its very worthwhile to understand the building-block rst- and second-order responses in depth. This chapter is organized as follows: We present rst-order systems, and their natural response, starting with a mechanical example. The charac 5

CHAPTER 1. NATURAL RESPONSE

teristics of such rst-order responses in time are discussed in detail. These responses involve only real functions and thus use only real mathematics. Next we present the similar rst-order responses encountered in electrical, thermal, and uidic systems. Second-order systems in general have complex-valued natural responses. Thus the section on second-order systems starts with a review of complex numbers. The natural responses for a second-order mechanical system are presented, with individual attention to the overdamped, critically-damped, and underdamped cases. Section 1.2 presents second-order system natural responses. Analogous electrical, thermal, and uidic second-order systems are discussed next. Finally, the chapter concludes with a discussion of the natural response of higher-order systems, and a discussion of linearity.

1.1

First-order systems
dy (t) + y (t) = 0, dt

The canonical1 homogeneous rst-order dierential equation is (1.1)

where we assume = 0. The variable is the system time constant and has units of seconds. Here we have explicitly shown the time dependence of y (t). It is also acceptable and more compact to use the form dy + y = 0. dt (1.2)

The response of a such an unforced rst-order system is always of the form y (t) = cest . This is a simple and beautiful result, easy to remember, and extends to higher-order systems in a natural way. The variable s has units of frequency (sec1 ). The dierential equation (1.1) will only allow one value of s = 1 . We call 1 the characteristic frequency or equivalently the eigenvalue of the system (1.1). In this rst-order case, 1 is a real number, but in higher-order cases the eigenvalues are more generally complex-valued. The constant c is a real number with the same units as y ; it is used to set the value of the function at some point in time, typically t = 0. The value at t = 0 is called the initial condition of the homogeneous response. You can nd the homogeneous solution as follows: First, substitute the assumed form y (t) = cest into the dierential equation. The deriviative
1

prototypical

1.1. FIRST-ORDER SYSTEMS operation just brings down a multiplicative term s, and so you have scest + cest = 0. This can be factored as ( s + 1)cest = 0.

(1.3) (1.4)

Setting the initial condition c = 0 satises this equation but is not very interesting, since this gives y = 0 for all time. The function est is nonzero for nite s and t, and thus can be divided out to give the characteristic equation ( s + 1) = 0. This has the solution s = 1 = 1/ , which is the one and only characteristic frequency (eigenvalue)2 associated with this rst-order system. Thus we have arrived at the homogeneous solution y (t) = ce .
t

(1.5)

The response decays to zero with increasing time if > 0; if the natural response of a system always decays to zero with increasing time for any initial conditions, we say that the system is stable. If the response goes o to innity with increasing time for some initial conditions, the system is unstable. The response (1.5) has the initial value y (0) = c. The graph of this response is shown in Figure 1.1 for an initial condition c = 1 with the four values = 2, 1, 0.5, 0.1. As you can see, represents the characteristic time for the response to decay toward zero; smaller values of correspond with faster responses. Because such a response is widely applicable in real engineering systems, we will take a bit of time to understand it in more depth. Your eorts here to internalize an understanding of this response and its characteristics will pay dividends throughout your engineering studies and practice. Specically, in an interval of one time constant, the response shown decays to a value of 0.37 times the value at the start of the interval. This is so because e1 = 0.3679 0.37. Since this response has an initial value of 1, the response decays to a value of 0.37 in one time constant, 0.372 in two time constants, and a value of 0.37n in n time constants. You should verify this result to graphical accuracy for all four of the time constant values; they pass through the dashed line y = 0.37 in an interval equal to . And, in 3 seconds, the response for = 1 sec passes through a value of 0.373 0.05; we would
2 The terms characteristic frequency and eigenvalue are equivalent, and will be used interchangeably herein.

CHAPTER 1. NATURAL RESPONSE

1 0.9
y (t) 0.8

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0  = 0.1 0.5 1 1.5  = 0.5

decreasing 

 =2  =1

0.37

3 0.37=0.05

2.5 t(sec)

Figure 1.1: First-order system response to an initial condition c = 1 with the four values = 2, 1, 0.5, 0.1. say that this response has settled to within 5% in 3 time constants. How many time constants would it take to settle to within 1%? (You should be sure you can answer this before going on.) Meanwhile, in 3 seconds, the = 0.1 sec response has passed through 30 time constants, and has a value of e30 = 9.4 1014 . This is pretty close to zero, but in theory the response never quite gets to zero, no matter how long you wait; it just keeps decaying by further factors of 0.37. So we see that the eigenvalue captures the time-scale of the rst-order response. This idea extends to the higher-order systems considered later. In these more general cases, the eigenvalues may have imaginary components; in the rst-order case considered above they are pure real. Because of the primary importance of the eigenvalues of a system, it is common in prac tice to graphically plot the eigenvalue locations on a plane with horizontal axis Re{s}, and vertical axis Im{s}. This complex plane is referred to as the s-plane, and the eigenvalue locations are called poles. For example, in the rst-order system considered above, the eigenvalue is 1 = 1/ ; this system is thus also said to have a pole at s = 1 = 1/ . In the complex plane, poles are plotted as xs; for the rst-order system, the pole diagram

1.1. FIRST-ORDER SYSTEMS

Im{s}

decreasing  X
1 = 1 

Re{s}

Figure 1.2: First-order system pole location as a function of . Arrow indicates the direction of decreasing .

appears as shown in Figure 1.2. Decreasing values of result in the pole moving to the left along the negative real axis. Thus, faster systems have poles located further from the origin in the s-plane. A dierential equation of the form (1.1) occurs as the mathematical model for systems of many dierent physical principles. In the sections below, we show the process of modeling rst-order systems from the me chanical, electrical, thermal, and uidic domains. In these domains, ho mogeneous responses such as shown in Figure 1.1 occur with a variety of associated physical units. The beauty of system theory is rst that it is found to be applicable to many classes of real-world systems, and secondly that we can thereby understand these systems from a common mathematical framework.

10

CHAPTER 1. NATURAL RESPONSE

Figure 1.3: Picture of mechanical system which can be modeled as rstorder.

1.1.1

Mechanical translational rst-order system

Consider the mechanical system shown in the picture of Figure 1.3 as used in Lab 1. This consists of a spring-steel beam rigidly xed at one end, and attached to an air cylinder damper on the other end. We will consider this as a translational system, with the point of translation corresponding to the nearly straight line motion of the end of the beam where it joins the air piston damper. The air piston damper3 consists of a graphite piston sliding in a precisely t glass cylinder as shown in Figure 1.4. The knob at the near end controls an adjustable orice to set the resistance to ow in and out of the damper, and thereby set the damping coecient. Figure 1.5 shows experimental data taken from this system via videotap ing at 20 frames per second, as well as data from a model adjusted to match this response. The measured data points are shown in blue, with asterisks at the data points taken every 1/20th of a second. The red curve is a plot of the rst-order response (1.5) with the parameters adjusted to reasonably
3

Also known as an Airpot, which is a trademark of the Airpot Corporation, Norwalk,

CT.

1.1. FIRST-ORDER SYSTEMS

11

Figure 1.4: Picture of air piston damper. t the data. The tted model response is y (t) = 1.5 102 e1.65t [m], (1.6)

and thus the time constant is = 1/1.65 = 0.61 sec. The initial condition is c = 1.5 cm. The rst-order model (1.5) ts this response very well. The experimental data is a bit noisy as might be expected. The primary noise source is that the video camera frame rate is not very constant. This could be improved with better video hardware, but is not important for this experiment. The simplest lumped mechanical model which ts this response is the rst-order mechanical spring-damper system shown in Figure 1.6. Here we assume that the link can only move in the x-direction. The cantilever beam acts as a spring which is linear for moderate deections. The spring con stant k for this beam can be calculated from rst principles. With this calculated spring constant we can compute the damping coecient equiva lent b for the air piston damper. As shown in the gure, the system consists of a spring and damper attached to a rigid massless link. The link represents the connection between the spring and damper, but contributes no dynamics of its own. The position of the link is denoted as x. The zero of position is indicated in the gure by the vertical line connecting to the arrow which indicates the direction

12

CHAPTER 1. NATURAL RESPONSE

Position versus time 1.5

1 Position (cm) 0.5 0 0

0.5

1.5 Time (sec)

2.5

Figure 1.5: Experimental natural response of beam/air piston system, and rst-order model response.

1.1. FIRST-ORDER SYSTEMS

13

Rigid, massless link

Figure 1.6: First-order mechanical system model. of increasing x. This choice of zero accounts for the rest position (zero force length) of the spring. The spring is moved by a force proportional to motion in the x-direction, Fk = kx. The damper is moved by a force which is proportional to velocity in the x-direction, Fb = b dx/dt. Newtons second law states that F = ma = mx , where F is the sum of the forces acting on a mass. This relationship also applies to the massless link, but since the link is massless, the forces must instantaneously sum to zero. For any mass element, or massless assembly from a system, Newtons second law can be captured in the form of a free-body diagram. For this system the free-body diagram appears as shown in Figure 1.7. Summing forces acting on the link and applying Newtons second law yields the system equation of motion Fk Fb = kx b dx = 0. dt (1.7)

The minus signs appear here for the forces Fk and Fb since they act on the link in the x-direction. The zero term on the right is due to the fact that the link is massless. The governing dierential equation can be rewritten as b dx + x = 0, k dt (1.8)

If we dene = b/k , this is in the form of (1.1). The natural response is thus as calculated in section 1.1, with its associated gures.

14

CHAPTER 1. NATURAL RESPONSE

x
k Fk

Fb

Systemcuthere

Forcesactingonelements

Figure 1.7: Free body diagram for massless link of rst-order system. We can calculate that for the dimensions of this beam k = 170 N/m. With this value in hand the model damping coecient is given by b = k = 170 0.61 = 104 [N sec/m] (1.9)

Dynamic systems can be studied at a number of levels of detail. Models of greater complexity could be readily justied for the beam air pot system if it were studied in more depth. For example, the distributed mass and compliance of the beam would lead to the existence of vibratory modes on the beam itself. These modes would require a high dimensional or innite dimensional model that could more accurately capture some of the transient behavior of the beam. Further, we have ignored the compressibility of the air in the cylinder of the damper. With nite compressibility, the air pot is a thermodynamic system in that the temperature of the air contained within the cylinder is a reection of the work done on the air by the piston, as well as inputs/losses of heat from the outside world. Such considerations are important to understand the behavior in many cases, but are well out side the scope of topics for this text. You need to study fundamentals of thermodynamics to fully understand this issue. Meanwhile, within the right range of time scales and accuracy require ments, a simple rst order lumped model well-captures the dominant dy

1.1. FIRST-ORDER SYSTEMS

15

Figure 1.8: Picture of mechanical rotational system which can be modeled as rst-order. namics of the airpot/beam system, as veried by the experimental data shown above. For information on expanding models to include such addi tional detailed eects, in this and many other systems, take a look at the Masters thesis of Katie Lilienkamp [1]. In particular, the beam/air-piston system is treated in great detail in section 3.3 of this reference.

1.1.2

Mechanical rotational rst-order system

Consider the mechanical rotational system shown in the picture of Fig ure 1.8. This system is described in more detail on the Activlab pages under the heading of Lab 2; you can see a video of it in motion on these pages. This system consists of a shaft rotating about a vertical axis. The axis of rotation is constrained by a pair of air bearings, which use pressurized air to create a nearly-frictionless rotational/translational bearing. Since the air bearings do not constrain axial motion, the shaft rests on a ball bearing resting on a hardened at. This ball-on-at acts as the axial bearing for the system. If the rotational axis is properly aligned perpendicular to the at, then this axial bearing exhibits very little friction. The rotating shaft carries a brass ywheel which serves as an additional rotational inertia. This ywheel can be placed on the hub to increase the inertia, or removed to decrease the inertia. Figure 1.9 shows the brass

16

CHAPTER 1. NATURAL RESPONSE

Figure 1.9: Picture of brass ywheel being placed on top of shaft of rotational system.

ywheel being put in place on the top of the hub. A line drawing of the system is shown in Figure 1.10. Here we can observe the shaft located in the air bearings. The axis of rotation is vertical in this gure. At the top of the shaft is the ywheel hub which is shown with the brass weight removed. At the bottom of the shaft there is a cup lled with a viscous liquid. In the present case this liquid is honey. More detail of the bottom end of the shaft is shown in Figure 1.11. Here you can see the ball bearing which is mounted on to the end of the shaft and rotates with the shaft. The ball bearing rests on the hardened at shown at the bottom of the gure. Honey is lled within the chamber to a depth L and has an annular thickness t. Figure 1.12 shows experimental data taken from this system. The re sponse shown in the gure looks reasonably modeled as rst-order. At the level of modeling that we require, we can then think of this system as com posed of a rotational inertia spinning on a rotational damper a shown in Figure 1.13. The rotational equivalent of Newtons second law is = J . The only torque acting on the inertia is due to the the viscous drag of the rota tional damper = b . Summing torques acting on the inertia yields the

1.1. FIRST-ORDER SYSTEMS

17

Figure 1.10: Line drawing of rotational system.

Figure 1.11: Cross-section at bottom of shaft showing ball bearing on at, and honey used for viscous damping.

18

CHAPTER 1. NATURAL RESPONSE

Figure 1.12: Experimental natural response of mechanical rotational system.

J b

Figure 1.13: Model for rotational system.

1.1. FIRST-ORDER SYSTEMS dierential equation. J d + b = 0 dt

19

(1.10)

which can be rewritten in standard form as J d + =0 b dt This equation has the solution (t) = cet/ (1.12) (1.11)

1 b if we dene = J/b [sec]. Thus the system has an eigenvalue 1 = = J . Integrating this result allows us to solve for the associated angular position as t 0

dt =
0

cet/
t 0

(1.13) (1.14) (1.15) (1.16) ]

(t) (0) =

cet/

= c[et/ 1] = c[1 e which gives (t) = (0) + c[1 et/ ].

(1.17)

To graphical accuracy, the experimental data of Figure 1.12 is reasonably well t by the function (t) = 1 + 8.5[1 et/0.1 ] rad, (1.18)

that is, with = 0.1 sec, (0) = 1 rad, and c = 85 rad/sec. This allows us to give the estimated velocity as a function of time as (t) = 85et/0.1 rad/sec. (1.19)

At this point we could develop a calculation of the rotational system inertia from rst principles. If we know the rotational inertia of this system, we can then use the time constant result = J/b to calculate the equivalent rotational damping. Alternately, we could experimentally measure the rota tional damping and thereby develop an estimate of the rotational inertia J .

20

CHAPTER 1. NATURAL RESPONSE

ic C

ir R

vc

vr

Figure 1.14: First-order parallel RC circuit diagram.

1.1.3

Electrical rst-order system

The circuit shown in Figure 1.14 is a parallel RC circuit which can be de scribed by a rst order dierential equation. The formulation of the dier ential equation goes as follows. First we need to account for each of the network elements. The resistor has a current-voltage relationship described by Ohms law vr = ir R. The capacitor has a current-voltage relationship c given by ic = C dv dt . The currents ir and ic must be equal and opposite so that their sum is equal to zero, since current cannot accumulate at their common node. That is, we must have vr dvc 0 = ir + ic = +C . (1.20) R dt Recognize further that since the two elements are connected in parallel, their voltages must be equal: vr = vc . You substitute this into (1.20), and multiply through by R to nd RC dvc + vc = 0. dt (1.21)

If we dene = RC , this is in the form of (1.1). The natural response is thus as calculated in section 1.1, with its associated gures. Specically, if the initial voltage on the capacitor is dened as vc (0) = v0 , then the voltage as a function of time varies as vc (t) = v0 et/RC [Volts]. (1.22)

For example, suppose that we set C = 100 F and R = 1 M. Then the time constant is = 100 sec; it will take the capacitor voltage 100 seconds to decay to 37% of its initial value.

1.1. FIRST-ORDER SYSTEMS

21

Figure 1.15: Sketch of bulb and relevant thermal elements.

1.1.4

Thermal rst-order system

For an example thermal system we study the desk lamp shown in the picture (to be added). This lamp bulb is electrically heated via the bulb lament. The resulting bulb temperature is measured with the infrared sensor shown in the gure (to be added). A sketch of the light bulb in the lamp is shown in the line drawing of Figure 1.15. We left the lamp on for a long enough time to reach steady state, and then turned o the lamp and measured the decay of temperature back to ambient. Data taken from this system is shown in tabular and graphical form in Figure 1.16 By inspection of this data, the bulb system is well-t by a rst-order model of the form of (1.1). An estimate of the associated time constant is about 3 minutes. But we need to have in seconds, so the system time constant is formally given as = 180 sec. An abstraction to a lumped model of this system is shown in Figure 1.17. Here the thermal capacitance of the bulb is summarized by the block of material labeled with the capacitance Cb with units of [J/ K]. The block is assumed to have a uniform temperature Tb [ K]. This block has a total stored thermal energy Wb = Cb Tb [J]. The change of thermal stored energy happens via heat ow dWb dTb = Cb . qb = (1.23) dt dt Here qb in units of watts represents heat ow into the bulb. As shown in the gure, we assume that the block is insulated on three sides, and so the

22

CHAPTER 1. NATURAL RESPONSE

Figure 1.16: Data from light bulb cooling experiment.

# B

4B

2B

4  4A

4HERMAL RESISTANCE TO OUTSIDE WORLD "ULB THERMAL CAPACITANCE

Figure 1.17: Lumped model for bulb cooling experiment.

1.1. FIRST-ORDER SYSTEMS

23

heat ow through those sides is zero. The block is connected to the outside ambient temperature via the thermal resistance Rb , such that qb = Ta Tb . Rb (1.24)

This resistance represents the ow of heat into the bulb as a linear function of the temperature dierence4 between the ambient and the bulb temperatures. Setting equality between the last two equations gives Cb dTb Ta Tb = . dt Rb (1.25)

Now, its convenient to dene a variable to represent the temperature dif ference between the bulb and ambient: T Tb Ta . Since the ambient temperature is constant, dT /dt = dTb /dt. Making these substituations and multiplying (1.25) through by Rb yields Rb Cb dT + T = 0. dt (1.26)

If we dene = Rb Cb , this is in the form of (1.1). The natural response is thus as calculated in section 1.1, with its associated gures. Specically, if the initial temperature dierence of the bulb is dened as T (0) = T0 , then the temperature dierence as a function of time varies as T (t) = T0 et/Rb Cb [K]. (1.27)

If you want to convert back to the absolute temperature of the bulb, re member that Tb = T + Ta .

1.1.5

Fluidic rst-order system

A uidic system which can be modeled with a rst-order dierential equation is shown in Figure 1.18. Here a tank lled with liquid drains through a long, thin pipe. The height of the liquid above the pipe inlet is dened as h. If we assume that the liquid has a density of [kg/m3 ], then the pressure Pt at
In real systems, more exact and likely nonlinear models can apply, but a linear model gives a rst understanding of this system response, and is well able to match the measured behavior. For example, pure radiative cooling varies as temperature dierence to the fourth power, which is highly nonlinear. There will certainly be signicant radiative heat ow in this system, however, the experimental data ts well to a linear heat ow model which suggests that radiative cooling is not highly signicant at the bulb envelope temperatures of 100 C .
4

24

CHAPTER 1. NATURAL RESPONSE

Figure 1.18: Liquid tank experiment. the inlet of the pipe is given by Pt = Pa + gh [N/m2 ]. In the SI system of units, the units of pressure are Pascals, i.e., 1 Pa = 1 N/m2 . Here Pa is the ambient pressure outside the system, and g is the acceleration of gravity. The pipe volumetric ow into the tank is dened as qt [m3 /s]. The ow is assumed to vary linearly with the pressure dierence as qt = Pa Pt gh = . R R (1.28)

Here R [Pa s/m3 ] is the uidic resistance of the pipe. If we assume that the tank has a constant cross-sectional area A, then the uid height varies with ow into the tank as dh qt = dt A (1.29)

We multiply through by A and set equality between the last two equa tions to give RA dh + h = 0. (1.30) g dt If we dene = RA/g , this is in the form of (1.1). The natural response is thus as calculated in section 1.1, with its associated gures. Specically,

1.2. SECOND-ORDER SYSTEMS

25

if the initial uid height is dened as h(0) = h0 , then the uid height as a function of time varies as h(t) = h0 etg/RA [m]. (1.31)

1.2

Second-order systems

In the previous sections, all the systems had only one energy storage element, and thus could be modeled by a rst-order dierential equation. In the case of the mechanical systems, energy was stored in a spring or an inertia. In the case of electrical systems, energy can be stored either in a capacitance or an inductance. In the basic linear models considered here, thermal systems store energy in thermal capacitance, but there is no thermal equivalent of a second means of storing energy. That is, there is no equivalent of a thermal inertia. Fluid systems store energy via pressure in uid capacitances, and via ow rate in uid inertia (inductance). In the following sections, we address models with two energy storage elements. The simple step of adding an additional energy storage element allows much greater variation in the types of responses we will encounter. The largest dierence is that systems can now exhibit oscillations in time in their natural response. These types of responses are suciently important that we will take time to characterize them in detail. We will rst consider a second-order mechanical system in some depth, and use this to introduce key ideas associated with second-order responses. We then consider secondorder electrical, thermal, and uid systems.

1.2.1

Complex numbers

In our consideration of second-order systems, the natural frequencies are in general complex-valued. We only need a limited set of complex mathematics, but you will need to have good facility with complex number manipulations and identities. For a review of complex numbers, take a look at the handout on the course web page.

1.2.2

Mechanical second-order system

The second-order system which we will study in this section is shown in Figure 1.19. As shown in the gure, the system consists of a spring and damper attached to a mass which moves laterally on a frictionless surface. The lateral position of the mass is denoted as x. As before, the zero of