This action might not be possible to undo. Are you sure you want to continue?
Chapter 2. Fundamental Models
In this chapter, a methodology for developing dynamic models of chemical processes is presented. After studying this chapter, the reader should be able to do the following: Write balance equations using the integral or instantaneous methods Incorporate appropriate constitutive relationships into the equations Determine the state, input, and output variables and the parameters for a particular model (set of equations) Determine the necessary information to solve a system of dynamic equations Linearize a set of nonlinear equations to find the state space model The major sections of this chapter are as follows: 2.1 Background 2.2 Balance Equations 2.3 Material Balances 2.4 Constitutive Relationships 2.5 Material and Energy Balances 2.6 Form of Dynamic Models 2.7 Linear Models and Deviation Variables 2.8 Summary [ Team LiB ]
[ Team LiB ]
Reasons for Modeling
There are many reasons for developing process models. Improving or understanding chemical process operation is a major overall objective for developing a dynamic process model. These models are often used for (i) operator training, (ii) process design, (iii) safety system analysis, or (iv) process control. Operator training: The people responsible for the operation of a chemical manufacturing process are known as process operators. A dynamic process model can be used to perform simulations to train process operators, in the same fashion that flight simulators are used to train airplane pilots. Process operators can learn the proper response to upset conditions, before having to experience them on the actual process. Process design: A dynamic process model can be used to properly design chemical-process equipment for a desired production rate. For example, a model of a batch chemical reactor can be used to determine the appropriate size of the reactor to produce a certain product at a desired rate. Safety: Dynamic process models can also be used to design safety systems. For example, they can be used to determine how long it will take, after a valve fails, for a system to reach a certain pressure. Process control: Feedback control systems are used to maintain process variables at desirable values. For example, a control system may measure a product temperature (an output) and adjust the steam flow rate (an input) to maintain that desired temperature. For complex systems, particularly those with many inputs and outputs, it is necessary to base the control-system design on a process model. Also, before a complex control system is implemented on a process, it is normally tested by simulation. It should be noted that no single model of a process exists, since a model only approximates the process behavior. The desired accuracy and resulting complexity of a process model depends on the final use of the model. Usually more-complex models will require much more data and effort to develop than simplified models, since more model parameters will need to be determined. The focus of this textbook is on process control, so model development is provided with this in mind.
Lumped Parameter System Models
The models developed in this textbook are known as lumped parameter systems models. These models consist of initial-value ordinary differential equations, often based on a perfect mixing assumption. The models have the form
where x is the vector of state variables, the vector of state variables derivatives with respect to time equal to dx/dt, u the vector of input variables, p the vector of parameters, y the vector of output variables, and, f(x,u,p) and g(x,u,p) the vectors of functions. State variables are variables that naturally appear in the derivative term of ordinary differential equation models. Common states resulting from overall material balance equations include total mass, volume, level for liquid-phase processes, and pressure for gas-phase processes. Component compositions are the most common states that arise from component material balances. Temperature is the most common state arising from an energy balance modeling equation. This state-variable representation seems very abstract at this juncture, and it generally takes students some time to become comfortable with it. The easiest way is to work through some simple examples to begin to associate the notion of states, parameters, inputs, and outputs with the physical variables associated with chemical processes. Throughout the text we use matrix and vector notation; you may wish to review any standard linear algebra book to become familiar with this notation. A concise review is also provided in the M ATLAB module (Module 1). [ Team LiB ]
the rate of energy accumulation has the form dE/dt. then dNi/dt represents the molar rate of accumulation of component i in the system. One viewpoint is based on an integral balance. . If Ni is used to represent the moles of component i in a system. Integral balances are particularly useful when developing models for distributed parameter systems. while the other is based on an instantaneous balance. where E is the total energy in a system. Figure 2-1. for example). consider the system shown in Figure 2-1.[ Team LiB ] 2. Similarly. In some cases the system may be microscopic in nature (a differential element. Another viewpoint is the instantaneous balance where the time rate of change is written directly. which result in partial differential equations. while in other cases it may be macroscopic in nature (the liquid content of a mixing tank.2 Balance Equations In this text. particularly for component balances. Integral Balances An integral balance is developed by viewing a system at two different "snapshots" in time. The mean-value theorems of integral and differential calculus are then used to reduce the equations to differential equations. Also. while the other boundary represents the mass in the system at t + ∆t. Conceptual material balance problem. For example. The in and out terms would then include the generation and conversion of species by chemical reaction. we are interested in dynamic balances that have the form This equation is deceptively simple because there may be many in and out terms. where one boundary represents the mass in the system at time t. ∆t. When solving a problem. the focus in this text is on ordinary differential equation-based models. when developing a dynamic model we can take one of two general viewpoints. it is important to specify what is meant by system. The rate of mass accumulation in a system has the form dM/dt. for example). Consider a finite time interval. and perform a material balance over that time interval. where M is the total mass in the system. respectively.
2 where 0 < α < 1. as Equation 2. We can write the right-hand side of Equation (2. using the mean-value theorem of integral calculus. Substituting the right-hand side of Equation (2.1). while and represent the mass rates entering and leaving the system.3 .1 where M represents the total mass in the system. respectively. we find Equation 2.An integral balance on the total mass in the system is written in the form Mathematically this is written as Equation 2.2) into Equation (2.1).
leaving the system is equal to the density of Instantaneous Balances Here we write the dynamic balance equations directly. we find Equation 2.6 and taking the limit as ∆t goes to zero.5 and by substituting Equation (2. and F is a volumetric flow rate (volume/time).4 Equation 2. Equation 2. we obtain the equation Equation 2. where V is the volume. Equation 2.4). is the mass density (mass/volume).By dividing Equation (2. • as Fin in and as • Fout . and using the mean value theorem of differential calculus (0 < β < 1) for the left-hand side.5) into Equation (2.7 Representing the • total mass • as M = V . so that the density of material • • material in the system ( out = ).3) by ∆t.8 Note that we have assumed that the system is perfectly mixed. based on an instantaneous rate-of-change .
10 Steady-state relationships are often used for process design and determination of optimal operating conditions. Equation 2.Equation 2. the integral balance method is probably clearer when developing distributed parameter (partial differential equationbased) models.7). Although the integral balance takes longer to arrive at the same result as the instantaneous balance method.7 which can also be written as Equation 2. [ Team LiB ] .8 This is the same result obtained using an integral balance. the derivative with respect to time is zero. Steady State At steady state. so from Equation (2.9 or from Equation (2. by definition.8). Equation 2.
PV = nRT. Here we develop a model that describes how the drum pressure varies with the inlet and outlet flow rates. R. Figure 2-2. Let V = volume of the drum and n = the total amount of gas (moles) contained in the drum.1: Gas Surge Drum Surge drums are often used as intermediate storage capacity for gas streams that are transferred between chemical process units. In this section we use two process examples to illustrate the modeling techniques used. Gas surge drum.12 From the ideal gas law. Recall that a model for a liquid surge vessel was developed in Chapter 1 (Example 1. Consider a drum depicted below (Figure 2-2). Assuming that the molecular weight is constant. .11 where MW represents the molecular weight. T is temperature (absolute scale). since V.3 Material Balances The simplest modeling problems consist of material balances. Example 2.[ Team LiB ] 2. and R is the ideal gas constant. and T are assumed constant. The rate of accumulation of the mass of gas in the drum is described by the material balance Equation 2. where P is pressure. A typical control problem would be to manipulate one flow rate (either in or out) to maintain a desired drum pressure.3). Assumption: The pressure-volume relationship is characterized by the ideal gas law. where qi is the inlet molar flow rate and q is the outlet molar flow rate. we can write Equation 2.
so we find the steady-state relationship . dP/dt = 0. then the pressure increases without bound. and the initial condition P(0).3.15 At steady state. For example. we model that case now. if the inlet molar flow rate increases while the outlet flow rate stays constant. the parameters R. although q is the molar rate out of the drum. Indeed.14 So the dynamic modeling equation is Equation 2. and V. this is an integrating system. we must know the inputs q i and q. Ph. T. then Equation 2. If the flow/pressure difference relationship is linear. It should be noted that just like the liquid level process discussed in Example 1.13 To solve this equation for the state variable P. Once again. an increase in pressure would most likely cause an increase in outlet molar flow rate (owing to the increased driving force for flow out of the drum). Let β represent a flow coefficient.so which can be rewritten Equation 2. Outlet Flow as a Function of Gas Drum Pressure Consider the case where the outlet molar flow rate is proportional to the difference in gas drum pressure and the pressure in the downstream header piping. In reality. we consider it to be an input in terms of solving the model.
Equation 2. since a change in an input variable eventually leads to a new steady-state value of the output variable. There are many control loops that a process engineer must consider at the design stage of a process. Example 2. Figure 2-3. that a feedback controller can be designed to regulate the inventory levels (liquid volume or gas pressure). Contrast self-regulating systems with integrating systems that do not achieve a new steady state (the output "integrates" until a vessel overflows or a tank overpressures).2: An Isothermal Chemical Reactor Ethylene oxide ( A) is reacted with water (B) in a continuously stirred tank reactor (CSTR) to form ethylene glycol (P). If there is an imbalance in the inlet and outlet flow rates. which are integrating by nature. Solving explicitly for Ps.16 where we use the subscript s to indicate a steady-state solution. we find Equation 2.3 and 2. It is the independence of the flow rates that can cause this problem. This type of system is known as self-regulating . In Chapter 15. these must receive the highest level of consideration. Because of the critical nature of inventory loops. the inventory material (liquid or gas) can easily increase or decrease beyond desirable limits. An increase in qis will lead to an increased value of Ps. For the gas drum we found that pressure was the most appropriate state variable. Notice.1 were based on writing an overall material balance. The stoichiometric equation is Here we develop a model (Figure 2-3) to find the concentration of each species as a function of time. however. The next example illustrates the use of modeling for reactor design. In the case of a liquid vessel we found that either liquid volume or height could serve as an appropriate state variable. The modeling equations for Examples 1. . Assume that the CSTR is maintained at a constant temperature and that the water is in large excess. A feedback controller manipulates a stream flow rate to maintain a desired inventory level. Isothermal stirred tank reactor. we find that inventory loops must be closed before other loops are considered.17 which is a linear relationship. Liquid level and gas pressure vessels represent inventory problems.
ρ. The vessel liquid (and outlet) density is then equal to the inlet stream density.18) as (notice this is the same result as Example 1. so and we can write Equation (2.Overall Material Balance The overall mass balance (since the tank is perfectly mixed) is Equation 2.19 Component Material Balances It is convenient to work in molar units when writing component balances. is not a function of concentration.18 Assumption: The liquid-phase density.3) Equation 2. particularly if chemical reactions are involved. Let C A and CP represent the molar concentrations of A and P .
20a).22 Expanding the left-hand side of Equation (2. and (2. Since the water is in large excess its concentration does not change significantly.24 Similarly. Equation 2. Each mole of A reacts with a mole of B (from the stoichiometric equation) and produces one mole of P. and the reaction rate is first order with respect to the concentration of ethylene oxide. and CAi represents the inlet concentration of species A.19).21 where k is the reaction rate constant and the minus sign indicates that A is consumed in the reaction. we find Equation 2. so the rate of generation of P (per unit volume) is Equation 2.(moles/volume).23).23 Combining Equations (2.20a). Equation 2. The component material balance equations are Equation 2.20b where r A and rP represent the rate of generation of species A and P per unit volume.20a Equation 2. the concentration P can be written as . (2.
Often other simplifying assumptions are made to reduce the number of differential equations to make them easier to analyze and faster to solve.27 or.24.2 has three differential equations.26 Using state-variable notation. Assuming a constant volume (dV/dt = 0). three state variables (V. the inputs (F i.28 Simplifying Assumptions The reactor model presented in Example 2. 2. perhaps owing to a feedback controller. Equation 2.25) and. input. F . CA(0). The resulting differential equations (since we assumed dV/dt = 0. The state. therefore. and the parameter (k).25 This model consists of three differential equations (2. the model has the form Equation 2. reduces the number of equations by one. CA. 2. F = Fi) are . and CP(0)]. and parameter vectors are Equation 2.Equation 2. To solve these equations we must specify the initial conditions [ V(0).19. and CP). CAi ) as a function of time.
we find that the conversion is related to the space velocity. The concept of conversion is important in chemical-reaction engineering.31 So. The Damkohler number is the ratio of the characteristic residence time to the characteristic reaction time and is widely used by chemical-reaction engineers to understand reactor behavior.29 Steady-State Solution At steady state. The space velocity can be thought of as the number of reactor volumes that "change over" per unit time. we find the following relationships (where the subscript s represents a steadystate solution): Equation 2. It is inversely related to the fluid residence time (V/Fs). Equation 2.30 Notice that the concentrations are a function of the space velocity (F s/V). from Equations (2. Two different chemical-reaction systems can have the same conversion if their Damkohler numbers are the same. A system with a large rate constant and low residence time can have the same conversion as a system with a small rate constant and high residence time.Equation 2.30) and (2. which has units of inverse time.32 Notice that the conversion is a function of the dimensionless term kV /F s. which has units of time and can be thought of as the average time that an element of fluid spends in the reactor. . Equation 2.31). which is known as the Damkohler number. The conversion of reactant A is defined as the fraction of the feed-stream component that is reacted.
while the bottom plots have conversion as the dependent variable. Design Objective It is desired to produce 100 million pounds per year of ethylene glycol.Numerical Example Using an Experimentally Determined Rate Constant Laboratory chemists have determined that the reaction rate constant at 55°C is k = 0. The design flow-rate calculation is shown below.8 lb/ft3. Here we find the steady-state concentrations of ethylene oxide (A) and ethylene glycol (P) as a function of the steady-state space velocity and residence time. Since 80% of the ethylene oxide is converted to ethylene glycol.4 lbmol/ft3 [see Equation (2. while the right-hand plots have residence time as the independent variable. Figure 2-4.5 lbmol/ft3 and an 80% conversion of ethylene oxide has been determined to be reasonable. the independent variable is the space velocity.32)]. At low space velocities (large residence times) there is nearly complete conversion of ethylene oxide to ethylene glycol. while at high space velocities (low residence times) there is little conversion. The operating flow rate is . Steady-state relationships for ethylene glycol reactor. the mass concentration is 24. the ethylene glycol concentration is 0. The feed-stream concentration is 0. The plots in Figure 2-4 all illustrate the same basic concept. On the left-hand plots. The top plots have concentrations as the dependent variables. Since the molecular weight is 62 lb/lbmol.311 min -1. assume 350 days/year of operation. What volume of reactor should be specified to meet the production rate requirement? Since process plants often have a shutdown period every 18 months or so.
000–20. We also see that it takes roughly 10 minutes for the reactor concentrations to achieve new steady-state values. Dynamic Response Assume that a control strategy will be specified to maintain the desired ethylene glycol concentration in the reactor by manipulating the reactor feed flow rate. In order to design the controller.29) using the techniques presented in Section 2.32) for reactor volume.000 gallons) are usually special orders involving on-site construction (or offsite with rail or truck delivery). For the remainder of this problem we assume that the reactor is operated with a volume of 769 gallons.6.Solving Equation (2. we find that the required volume is 102.0778–0. Larger scale reactors (greater than roughly 10. Response of ethylene oxide and ethylene glycol concentrations to a step change in space velocity of 5% (from F/V = 0. it is important to understand the dynamic response between an input change and the observed output(s). An increase in the space velocity (corresponding to a decrease in residence time) results in a decrease in the conversion of A to P. Figure 2-5. These simulations were performed by integrating differential Equations (2. It should be noted that reactors of this size range can be purchased in standard sizes. Most likely the engineer would have a choice of 750-gallon or 1000-gallon models and would choose the 1000-gallon model for expansion capability. A step change of 5% in the space velocity (F /V) yields the responses in the ethylene oxide and ethylene glycol concentrations shown in Figure 2-5. regardless of its maximum capacity. .0817 min-1).9 ft3 or 769 gallons.
where temperature is often a state variable. In Section 2. however.2 illustrate the use of material balances to develop models.5. [ Team LiB ] .Examples 2. Material balance equations are rarely adequate to develop most models of interest. we review the development of energy balance models.4. the state variable of interest was the drum pressure. the state variables of interest were the concentrations of ethylene oxide and ethylene glycol. we cover the basic idea of constitutive relationships in Section 2. In the isothermal ethylene glycol reactor. First. In the gas drum example.1 and 2.
consider the reaction A + 2B reaction of A is first order in both A and B. Van Ness. see a thermodynamics text. Gas Law Process systems containing a gas will often need a gas-law expression in the model. These required relationships are known as constitutive equations. For consistency in the units for rA. molar volume ( ). Reaction rates are normally expressed in terms of generation of a species. we find that k has units of (vol/mol B * time). Then we can write .[ Team LiB ] 2.35 where r A is the rate of reaction of A (mol A/volume time). such as Smith.34 For other gas laws. and Abbott (2001). k the reaction rate constant. For example. Chemical Reactions The rate of reaction per unit volume (mol/volume*time) is usually a function of the concentration C + 3D.4 Constitutive Relationships Examples 2. and CB the concentration of B (mol B/volume). The ideal gas law is commonly used to relate pressure (P). and temperature (T): Equation 2. we use the expression Equation 2. The minus sign indicates that A is consumed in the reaction above. several examples of constitutive equations are shown in this section. It is good practice to associate the units with all parameters in a model.33 The van der Waal's P T relationship contains two parameters (a and b) that are system specific: Equation 2.2 required more than simple material balances to define the modeling equations. If the rate of the of the reacting species.1 and 2. CA the concentration of A (mol A/volume). Notice that 2 mol of B react for each 1 mol of A.
k0 the frequency factor or preexponential factor.) The frequency factor and activation energy can be estimated based on data of the reaction constant as a function of reaction temperature. The most commonly used representation is the Arrhenius rate law. can be represented by Equation 2.37 and we see that k0 and E can be found from the slope and intercept of a plot of (ln k) vs.and vapor-phase compositions of component i .39 . we find Equation 2.38 where yi is the vapor-phase mole fraction of component i . E the activation energy (cal/gmol). Equilibrium Relationships The relationship between the liquid. Taking the natural logarithm of the Arrhenius rate law. xi the liquid-phase mole fraction of component i . as a function of temperature.Usually. Equation 2. Here. and T the absolute temperature scale (K or R. The simplest assumption for the calculation of an equilibrium constant is to use Raoult's law. Equation 2. the reaction rate coefficient is a function of temperature.36 where k(T) is the reaction rate constant. and Ki the vapor/liquid equilibrium constant for component i. when the phases are in equilibrium. (1/T).987 cal/gmol K). The equilibrium constant is a function of composition and temperature. R the ideal gas constant (1.
At the design stage the overall heat transfer coefficient can be estimated from correlations.40 Often. y the vapor-phase mole fraction of light component. A the area for heat transfer. U the overall heat transfer coefficient. for example) can be described by Equation 2. it is a function of fluid properties and velocities. the relationship often used between the vapor and liquid phases for the light component is Equation 2. . The individual film heat transfer coefficients (h i and ho). we will see a constant relative volatility assumption made. In a binary system.42 where Q is the rate of heat transfer from hot to cold fluid. and a fouling factor (f) can be used to determine the overall heat transfer coefficient from the relationship Equation 2.43 The individual film coefficients are a strong function of fluid properties and velocities. ∆x). and α the relative volatility (α > 1). to simplify vapor/liquid equilibrium models. and ∆T the difference between hot and cold temperatures. The overall heat transfer coefficient is often estimated from experimental data. the metal conductivity (k. and thickness.where the pure component vapor (saturation) pressure often has the following form: Equation 2.41 where x is the liquid-phase mole fraction of light component. Heat Transfer The rate of heat transfer through a vessel wall separating two fluids (a jacketed reactor.
g. the opposite is true for an equal-percentage valve. For an equal-percentage valve. as a function of x). f(x) = α x . The equal-percentage valve is commonly used in chemical processes because of desirable characteristics when installed . These flow characteristics are plotted in Figure 2-6. Figure 2-6. Valve flow characteristics.Flow Through Valves The flow through valve is often described by the relationship Equation 2. f(x) = x. ∆Pv the pressure drop across the valve. For a linear valve. The sensitivity or valve "gain" is related to the slope of the curve. Three common valve characteristics are (i) linear. and f(x) the flow characteristic (varies from 0 to 1. . the specific gravity of the fluid. For a quick-opening valve. The sensitivity of a linear valve does not change as a function of valve position. s. x the fraction the valve is open (0 x 1). and (iii) quick-opening.1. the sensitivity (or "gain") of flow to valve position is high at low openings and low at high openings.44 where F is the volumetric flow rate. Notice that for the quick-opening valve. Cv the valve coefficient. (ii) equal-percentage.
[ Team LiB ] .in piping systems where a significant piping pressure drop occurs at high flow rates. Knowledge of these characteristics will be important when developing feedback control systems. Flow control is discussed in detail in Module 15.
[ Team LiB ]
2.5 Material and Energy Balances
Section 2.3 covered models which consist of material balances only. These are useful if thermal effects are not important, where system properties, reaction rates, etc. do not depend on temperature, or if the system is truly isothermal (constant temperature). Many chemical processes have important thermal effects, so it is necessary to develop material and energy balance models. One key is that a basis must always be selected when evaluating an intensive property such as enthalpy.
Review of Thermodynamics
Developing correct energy balance equations is not trivial and the chemical engineering literature contains many incorrect derivations. Chapter 5 of the book by Denn (1986) points out numerous examples where incorrect energy balances were used to develop process models. The total energy (TE) of a system consists of internal ( U), kinetic (KE), and potential energy (PE),
where the kinetic and potential energy terms are
For most chemical processes where there are thermal effects, the kinetic and potential energy terms can be neglected, because their contribution is generally at least an order of magnitude less than that of the internal energy term. When dealing with flowing systems, we usually work with enthalpy. Total enthalpy is defined as
The heat capacity is defined as the partial derivative of enthalpy with respect to temperature, at constant pressure. The heat capacity, on a unit mass basis, is
where the overbar indicates that the enthalpy is on a unit mass basis. We make use of this relationship in the following example. The goal of the following example is severalfold: develop a model consisting of both material and energy balances illustrate the steady-state effect of the input on the output illustrate the effect of process "size" (magnitude of flow rate, for example) illustrate dynamic behavior
Example 2.3: Heated Mixing Tank
Consider a perfectly mixed stirred-tank heater, with a single feed stream and a single product stream, as shown below. Assuming that the flow rate and temperature of the inlet stream can vary, that the tank is perfectly insulated, and that the rate of heat added per unit time (Q) can vary, develop a model (Figure 2-7) to find the tank liquid temperature as a function of time.
Figure 2-7. Stirred-tank heater.
Neglecting changes in density due to temperature, we find
Here we neglect the kinetic and potential energy contributions,
We write the total work done on the system as a combination of the shaft work (WS) and the energy added to the system to get the fluid into the tank and the energy that the system performs on the surroundings to force the fluid out.
This allows us to write Equation (2.49) as
and since H = U + pV, and
, we can rewrite Equation (2.51) as
Neglecting pressure*volume changes, we find
53 Often we assume that the heat capacity is constant.52). we select an arbitrary reference temperature (Tref) for enthalpy Equation 2. we have or from Equation (2. Expanding the derivative term and assuming that the density is constant.54a Equation 2. The change in the pV term was neglected. so Equation 2.We must remember the assumptions that went into the development of Equation (2. or calculated at an average temperature. The kinetic and potential energy effects were neglected.52) in the following fashion.48) . The total enthalpy term is and assuming no phase change.54b We now write the energy balance (2.
Also.Tref)/dt = dT/dt. so d(T . F .55 but Tref is a constant.59 .58) to 0. and the initial conditions V(0) and T(0). and Ti (as a function of time).56 which yields the two modeling equations Equation 2.Canceling common terms gives Equation 2. neglecting WS (which is significant only for very viscous fluids). the inputs Fi. The resulting relationship between the manipulated power and the outlet temperature is Equation 2.57 Equation 2. Q. we can write Equation 2. we must specify the parameters ρ and cp. Steady-State Behavior and the Effect of Scale (Size) The steady-state solution can be found by setting the derivative terms in Equations (2.57) and (2.58 In order to solve this problem.
Equation 2. the relationship between heater power and outlet temperature is linear. The slope is the sensitivity (also known as the "gain") of the output with respect to the input. Here we consider a stream of water entering a stirred-tank heater at 20°C. Notice that for a given steady-state flow rate. The outlet temperature as a function of heater power [Equation (2. 10 liters/minute (household shower). This sensitivity is also known as the process gain and is defined as the partial derivative of the output with respect to the input. and 100 liters/minute (small car wash). evaluated at steady state.59). because a given change in power will have a much larger affect on a low flow rate than a high flow rate stream. while the higher flow rate operation has a low sensitivity. As expected from Equation (2. Also.59)] is plotted in Figure 2-8 for each of the three cases. The lower flow rate operation has a high sensitivity (slope) of temperature to power. Figure 2-8. the curves are linear. This makes physical sense. Outlet temperature as a function of heater power. the volume of the vessel has no effect on the steady-state temperature (volume has a solely dynamic effect).60 .where the subscript s is used to indicate a steady-state value. at three possible flow rates: 1 liter/minute (espresso machine).
62 Figure 2-9. For example. The discussion thus far has centered on the steady-state behavior of stirred-tank heaters. if a scaled steady-state input is defined as Equation 2. It is often useful to work with scaled variables.61 then all three input flow rates have the same steady-state sensitivity of the output to the scaled input.It is clear from Equation (2. however. on the dynamic behavior of a stirred-tank heater. The volume has a major impact.60) that larger flow rates will have smaller gains (slopes or sensitivities). This is shown in Figure 2-9 and the following equation: Equation 2. and we found that the volume had no effect. Outlet temperature as a function of scaled heater power. .
longer residence times have a slower response time than shorter residence-time systems. Response of temperature to step change in scaled heat addition rate. As expected. These curves were obtained by integrating Equation (2. as a function of the residence time (V/F ). [ Team LiB ] .Dynamic Behavior Volume has a considerable effect on the dynamic behavior of this process.58) for the three different residence times (with V assumed constant). The response of the temperature to a step change in the scaled heat input is shown in Figure 2-10. The initial steady-state values are T = Ti = 20°C and Qscaled = 0. Figure 2-10. Qscaled is stepped from 0° to 1°C. At t = 0.
relating states. and pi is a parameter. temperature is the common state variable that arises from a dynamic energy balance. nx state variables. nu inputs.6 Form of Dynamic Models The dynamic models derived in this chapter consist of a set of first-order (only first derivatives with respect to time). Input variables typically include flow rates of streams entering or .63 where xi is a state variable. and np parameters.64 State Variables A state variable is a variable that arises naturally in the accumulation term of a dynamic material or energy balance. The notation is used to represent dxi/dt. Inputs are normally specified by an engineer. based on knowledge of the process being considered. A state variable is a measurable (at least conceptually) quantity that indicates the state of a system. A representation of a set of first-order differential equations is Equation 2. Input Variables An input variable is a variable that normally must be specified before a problem can be solved or a process can be operated. ordinary differential equations. For example. explicit. Concentration is a state variable that arises when dynamic component balances are written. initial-value. Also included in these models is a set of algebraic equations. nonlinear. Equation 2.[ Team LiB ] 2. and parameters to output variables. Notice that there are nx equations. ui is an input variable. inputs.
p the vector of parameters. that is.66 . For example. Examples include density. heat transfer coefficient. and mass-transfer coefficient. Other times the outputs are a nonlinear function of the states (or even inputs). reactor volume may be an important design parameter. When designing a process. the reaction chemistry. Parameters A parameter is typically a physical or chemical property value that must be specified or known to mathematically solve a problem. Equation 2. Parameters are often fixed by nature.63) and (2. since That is. Input variables are often manipulated (by process controllers) in order to achieve desired performance. molecular structure.leaving a process (notice that the flow rate of an outlet stream might be considered an input variable!). Vector Notation The set of differential and algebraic Equations (2. viscosity. Steady-State Solutions Notice that dynamic models (2. and y the vector of output variables. a parameter might be "adjusted" to achieve some desired performance.64) can be written more compactly in vector form. and so forth. Compositions or temperatures of streams entering a process are also typical input variables. existing vessel configuration.63) can also be used to solve steady-state problems. operation. u the vector of input variables. thermal conductivity. Very often the measured outputs are simply a subset of the state variables. Output Variables An output variable is often a state variable that is measured. Equation 2. particularly for control purposes.65 where x is the vector of state variables.
For more details on how to use MATLAB integration routines. leaving the nx state values to be solved for. ∆t. This approach is illustrated in Appendix 2.68 The explicit Euler integration technique involves specifying the integration step size. resulting in a set of algebraic equations. see Module 3. Equation 2. In this case. until a steady state is reached.67 if the derivative term is approximated (where k represents a time index) as Equation 2. more accurate integration routines using a variable step size are used.2. Numerical Integration Here we briefly consider numerical methods to integrate ordinary differential equations. Equation 2. In practice. all inputs and parameters would be specified. Note that differential equation solvers can also be used to solve for the steady state of stable systems. by simply integrating from an initial value for the states for a long period of time. [ Team LiB ] .for steady-state processes. The focus of this text is not on the development of numerical methods. that is.69 and marching sequentially from one time step to another. nx equations in nx unknowns must be solved. so we briefly cover the basic idea in Appendix 2.1.
perturbations from a steady-state operating condition. The reader should show that if we define the following deviation variables Equation 2.[ Team LiB ] 2.3).72 or Equation 2. when the volume." that is.71 then Equation (2.7 Linear Models and Deviation Variables Consider the stirred-tank heater model (Example 2. and inlet temperature are constant at their steady-state values (indicated by the subscript s): Equation 2.70 Deviation Variable Formulation Control engineers like to think in terms of "deviation variables.73 where the new parameters that appear are . flow rate.70) can be written in the form Equation 2.
75 Since the steady-state operating point is chosen as the point of linearization. resulting in the following approximate equation: Equation 2. then. is the same as the residence time.73) is one of the most widely used models to describe the dynamic behavior of chemical processes. while commonly used control strategies are based on linear systems theory. The quadratic and higher order terms are neglected.74 The value of this function can be approximated using a Taylor series expansion of the form where the subscript s is used to indicate the point of linearization (usually the steady-state operating point). One State Variable Consider a single variable function (equation) Equation 2. The Taylor series approximation is based on the steadystate operating point of the process. The method that we use to form linear models is based on a Taylor series approximation to the nonlinear model. Equation (2. to be able to linearize nonlinear models for control system design and analysis purposes. Linearization of Nonlinear Models The material and energy balance models that describe the behavior of chemical processes are generally nonlinear. in this case. then [by definition of . It is important.Notice that the process gain is the same as the sensitivity shown in Equation (2.60) and the time constant.
single-input equation. dropping the "approximately equal" notation where x' = x . The quadratic and higher order terms are neglected.79 . resulting in the following approximate equation Equation 2.xs represents a deviation variable. we can write the following form Equation 2. is the derivative of the function One State and One Input Consider now the following single-state. and evaluated at the steady-state value.76 and since xs is a constant value.78 The value of this function can be approximated using a Taylor series expansion of the form where the subscript s is used to indicate the point of linearization (usually the steady-state operating point). Equation 2.a steady state.77 or. f(xs) = 0] Equation 2.
These basic ideas are illustrated in the following example.81 and since ys = g(xs.xs. us) = 0] where x' = x . f(xs. .us) where y' = y . and and are the derivatives of the function with respect to the state and input. evaluated at the steady-state value. and and are the derivatives of the function with respect to the state and input.xs represents a deviation variable. evaluated at the steady-state value. and u' = u .ys.80 A Taylor series expansion about the state and input yields (after neglecting higher order derivatives) Equation 2.us represent deviation variables. x' = x .Since the steady-state operating point is chosen as the point of linearization. then [by definition of a steady state. Output Variable Consider now the expression for an output variable Equation 2.
assuming constant volume and density is Here the state variable is CA and the input variable is F .4: A Second-Order Reaction Consider a CSTR with a single. consider the concentration of A to be the output variable so and the state space model is where the state. .Example 2. input. The modeling equation. and output (in deviation variable form) are For the following parameters. second-order reaction. A Taylor series expansion performed at the steady-state solution yields Now.
ny outputs.A steady-state operating point is and the partial derivatives are and the linear model is Generalization Consider the general nonlinear model with nx states. nu inputs. and np parameters The elements of the linearization matrices are defined as .
A hot fluid circulated through the jacket (which is assumed to be perfectly mixed).where ij subscripts refer to the i th row and jth column of the corresponding matrix. and heat flow between the jacket and vessel increases the energy content of the vessel fluid. . Jacketed stirred-tank heater. Example 2. For example. element Bij refers to the effect of the jth input on the i th state derivative. The rate of heat transfer from the jacket fluid to the vessel fluid is Figure 2-11. The linear state space form is where the deviation variables are defined as perturbations from their steady-state values In future chapters we normally drop the prime (') notation for deviation variables and assume that a state space model is always in deviation variable form.5: Jacketed Heater Consider the jacketed stirred-tank heater shown in Figure 2-11.
F i = F . If the outputs. the inputs are the jacket flow rate. feed temperature.where U is the overall heat transfer coefficient and A is the area for heat transfer. Energy balances on the vessel and jacket fluids result in the following equations. feed flow rate. the linearized model is . in deviation variable form. states. Assuming that the volume and density are constant. Here the outputs are the vessel and jacket temperatures. and jacket inlet temperature. which are also the states. are Then. and inputs.
the reader should show that Exercise 8 is a numerical example for this problem. [ Team LiB ] .Similarly.
which is The ODE module provides a tutorial on the use of MATLAB to integrate ordinary differential equations. States appear in the "accumulation" (derivative with respect to time) term. outputs are often a subset of the states. The models have the form where the states. pressure (total material balances on gases). some important characteristics were discussed. inputs can be either manipulated or disturbance variables.[ Team LiB ] 2. States are often concentrations (from the component balances). An understanding of dynamic behavior is obtained by using linear state-space models.u.p) = 0 which can be found numerically.8 Summary We have focused primarily on the development of ordinary differential equation models that describe the dynamic behavior of processes where perfect mixing can be assumed. outputs. and p. Process gain is the . which are used later for control system design. For the processes studied. such as density or kinetic rate parameters. or temperature (energy balances). To integrate the differential equations numerically. volume (total material balance on liquids). In the next chapter we further analyze linear state space models to obtain an understanding of dynamic behavior. since it is the ratio of a characteristic reaction time to residence time. inputs. respectively. u. which have the form where the prime notation (') is used to indicate a deviation variable. we use a variant of Euler integration. These state space models are converted to Laplace transfer function models. and parameters are often physical intensive variables. particularly for nonreacting systems. residence time (volume/flow rate) is often a good indicator of the relative "speed" of the process dynamics. and parameters are x. the Damkohler number is an important parameter. For example. The modeling equations are based on material (component or total) and energy balances. or f(x. y. . using M ATLAB The steady-state solution is functions such as fzero and fsolve (which require the optimization toolbox) as shown in the MATLAB tutorial module. For chemical reactors. The concept of process gain is extremely important for process control design.
3 Heated Mixing Tank 2.5 Jacketed Heater [ Team LiB ] . process gain is the ratio of the long-term (steady-state) change in process output to the change in process input.4 A Second-Order Reaction 2. The resulting first-order model has the form where kp is the process gain and τp is the process time constant. The dynamic behavior of firstorder processes will be discussed in the next chapter. Many process systems can be represented as single-state models. where the output is the state. The examples used in this chapter were 2.2 An Isothermal Chemical Reactor 2.sensitivity of the process output to a manipulated input.1 Gas Surge Drum 2. That is.
McGraw-Hill.R. R. Longman. Elementary Principles of Chemical Processes. and M. Rousseau. Process Dynamics: Modeling. Upper Saddle River.. An excellent textbook for an introduction to chemical engineering thermodynamics is Smith. New York (1986). Denn. J.. Process Modeling. M. Excellent discussions of the issues involved in modeling a mixing tank. 3rd ed. C. S. H.. incorporating density effects and energy balances. 2nd ed.. Van Ness. and M. F... B. NJ (1992).W. Prentice Hall. NJ (1998). Prentice Hall. Prentice Hall. Analysis and Simulation.. The reaction of ethylene oxide and water to form ethylene glycol is discussed in Fogler. New York (1971). An introduction to chemical reaction engineering is Fogler.[ Team LiB ] Suggested Reading A nice introduction to chemical engineering calculations is provided by Felder. Russell.. Chemical Engineering Thermodynamics. New York (1986). 2nd ed. H. NJ (1999). Abbott.. M. Elements of Chemical Reaction Engineering. Englewood Cliffs. Introduction to Chemical Engineering Analysis. T. M. New York (2001). S. M. A more detailed discussion of process dynamics is provided in the textbook Bequette. H.. Elements of Chemical Reaction Engineering. M. and R. M. 6th ed. Wiley. [ Team LiB ] . Upper Saddle River. Wiley. is provided in the following two books: Denn.
2: Consider the heated mixing tank example. Q. 3: Often liquid surge tanks (particularly those containing hydrocarbons) will have a gas "blanket" of nitrogen or carbon dioxide to prevent the accumulation of explosive vapors above the liquid. respectively: a. sketch the steady-state input (qis) -output (Ps) curve. which had the modeling equations For steady-state inlet and outlet flow rates of 100 liters/minute. Use Euler integration with a integration step size of 0. Consider a step inlet temperature change from 20° to 22°C. with the form so the modeling equation is Discuss whether this is now a self-regulating process. a liquid volume of 500 liters and inlet and outlet temperatures of 20° and 40°C. based on the flow coefficient. β = 1 mol · s -1 · atm-1/2 and a constant header pressure. Ph = 1 atm. Find the steady-state heating rate. as depicted below. Often the outlet flow relationship is actually nonlinear. b.1.[ Team LiB ] Student Exercises 1: Consider the gas drum in Example 2. Also.5 minutes to find the vessel temperature response for the first 2 minutes. Compare this with MATLAB ode45 (see the ODE module). .
and P the gas pressure.5 liter/(mol hr). Compare your plots with those shown in Figure 2-4. mol/liter 5: 6: 7: . If two 800-gallon reactors are placed in series. There is. Let qf and q represent the inlet and outlet gas molar flow rates. The state variables should be volume and concentration of reactant A. Show that the modeling equations are and state any other assumptions. The stream flow rate is 100 liter/hour and two available reactors (400 and 2000 liters) have been placed in series (the smaller reactor is placed before the larger one). and a continous feed of reactant is started. Solve the dynamic equations using ode45 (see the ODE module). Assume constant volume and constant density.2). with a concentration of 1 mol/liter.0817 min-1. W. Vl the liquid volume. To meet Environmental Protection Agency and state standards. however. V the constant (total) volume. Let Cw1 = concentration in reactor 1.2). no outlet stream. A stream contains a waste chemical. 4: Consider the ethylene glycol problem (Example 2. Write the modeling equations for the concentration of the waste chemical. Semibatch reactors are operated as a cross between batch and continuous reactors. at least 90% of the chemical must be removed by reaction.Develop the modeling equations with gas pressure and liquid volume as the state variables. Develop the modeling equations for a single first-order reaction. Ff and F the liquid volumetric flow rates. Assume the ideal gas law. a. what volumetric flow rate is necessary to produce 100 million lb/year of ethylene glycol? What percentage savings is this compared with using a single 800gallon reactor? Consider the ethylene glycol problem (Example 2. The chemical decomposes by a second-order reaction with a rate constant of 1. A semibatch reactor is initially charged with a volume of material.0778 to 0. for a step change in space velocity from 0.
liter/hr V1 = liquid volume in reactor 1.33333 mol/liter (reactor 1) and 0. Show that the A and B matrices are (also.09005). show that the C and D matrices are f. liter/(mol hr) b. using ode45. show the units associated with each coefficient). Would better steady-state removal of W be obtained if the order of the reaction vessels was reversed? Why or why not? (Show your calculations. Is the removal specification still obtained? h. Show that the steady-state concentrations are 0. Compare the linear and nonlinear variables on the same plots (make certain you convert from deviation to physical variables for the linear results) for a step change in the flow rate from 100 liters/hour to 110 liters/hour. so the specification is met.) 8: Consider Example 2. the stirred-tank heater example. Find the eigenvalues of A using the MATLAB eig function.A) = 0.Cw2 = concentration in reactor 2.09005 mol/liter (reactor 2). Compare the linear and nonlinear responses of the reactor concentrations. liters V2 = liquid volume in reactor 2. Read and work through the .) c.3333 and 0. Solve a for the nonlinear differential equations. Linearize at steady state and develop the state space model (analytical) of the form where d. Assuming that each state is an output. Assume the initial concentrations are the steady-state values (0. by solving det (λI . (Hint: You need to solve quadratic equations to obtain the concentrations. liters k = second-order rate constant. mol/liter F = volumetric flow rate. and find the eigenvalues by hand. g.5. e.
to be used with ode45 (Module 3) to solve the two nonlinear ordinary differential equations. perform simulations for small and large step changes in the jacket flow rate. Here we discover the effect of vessel scale on the heat transfer removal capability of a vessel.1 ft3/min in jacket flow rate. Now. assuming that the jacket volume is 0. so the volume of the new vessel is 100 ft3.3 Btu/(°F · ft3) V = 10 ft 3 Vj = 2. find the new steady-state jacket flow rate. First. Also. d. We now consider the important issue of scale-up: Pilot plants are small-scale (intermediate between lab-scale and full-size manufacturing-scale) chemical processes. Discuss the effect of process scale-up on the operating conditions. Find the values of the matrices in the state space model. heater. How large can the vessel become before the jacket temperature is too high (it approaches the inlet jacket temperature)? h. Assume that the heat transfer coefficient (U) remains constant. Use the following parameters and steady-state values: Fs = 1 ft3/min Tis = 50°F Tjis = 200 °F ρcp = 61. find the new state space model. Find the value of UA for the larger vessel. Calculate the eigenvalues of the A matrix.25 times the vessel volume. verify that the following values obtained for UA (overall heat transfer coefficient * area for heat transfer) and F js (steady-state jacket flow rate) are correct: b. used to understand process operating behavior before the manufacturing process is designed. 10 ft3/min rather than 1 ft3/min). How do they compare with the smaller vessel? i. f. Assume that the vessel can be modeled as a cylinder. verify that the steady-state state variable values are correct by simulating the process with no change in the jacket flow rate. Use the MATLAB step command to solve for the linear state space model. By solving the steady-state equations. but that the heat transfer area changes.m (example shown below). Consider now a vessel that can handle 10 times the throughput of the previous vessel (that is. Find the step response for the nonlinear and linear (state space) systems for a step increase of 0. Realize that the step results are based on deviation variables and for a unit step change in input one (jacket flow rate). and convert the linear states to physical variable form. Hint: Solve the two modeling equations at steady state to obtain these values. with the height (L) = 2*diameter ( D). For the larger vessel. g. For the larger vessel. c. find the new steady-state value of jacket temperature that must be used to maintain the vessel temperature at T = 125°F.5 ft3 a. Write a function file. How do these compare with the smaller vessel? . e.example.3 Btu/(°F · ft3) Ts = 125 °F Tjs = 150 °F ρjcpj = 61. Assume that the same residence time is maintained (V/F = 10 minutes).
tspan. V = 10.9. Ti = 50.delFj).T)/(V*rhocp).5.j.delFj). % where tspan = [0 30]. Bequette % 8 July 94 % 23 Jan 01 . Find the step response for the nonlinear and linear (state space) systems for a step increase of 10% in the jacket flow rate. dTjdt = (Fj/Vj)*(Tji .x. % % generate simulations using % [t. T = x(1).revised for jacket step change in argument list % % x(1) = T = temperature in tank % x(2) = Tj = temperature in jacket % delFj = change in jacket flowrate % F = Tank flowrate % Tin = Tank inlet temperature % Tji = Jacket inlet temperature % V = Tank volume % Vj = Jacket volume % rhocp = density*heat capacity % rhocpj = density*heat capacity.flag.x] = ode45('heater'.T) + UA*(Tj . % % odes % dTdt = (F/V)*(Ti . rhocpj= 61.W. UA = 183. Vj = 2. for example . rhocp = 61.3. xdot = [dTdt.x0.5. % % Dynamics of a stirred tank heater % (c) 1994 . Tji = 200. % % step change in jacket flow % Fj = Fjs + delFj.B.UA*(Tj .jacket fluid % % parameter and steady-state variable values are: % F = 1.T)/(Vj*rhocpj). How do these compare with the smaller vessel? Example m-file to be used by ode45 function xdot = heater(t. Fjs = 1. Tj= x(2).Tj) .3.. dTjdt].
min') ylabel('temp.flag. [t.2). % plot in physical variable form figure(3) subplot(2.tspan. sys = ss(a.plot(tlin. min') subplot(2. deg F') % % state space model with unit step change a = [-0.1).'--') .1). b = [0. unit step.x(:. d = 0.options.b.1)) title('nonlinear.c.delFj).1.20]. % figure(1) subplot(2.2)) xlabel('time.d). you may wish to generate and run the following script file: % % % % % % % % % % runs stirred tank heater example .1e-6.1.125+ylinscale(:.1.ylin(:.1.tlin] = step(sys). min') % % scale deviation by delFj ylinscale = ylin*delFj.0 1].1. [ylin.x.plot(tlin.delFj). c = [1 0.23 Jan 01 step changes to stirred tank heater make certain you enter the step change value (delFj) before running this file also.1)) title('linear state space.1.[1e-6 1e-6]). min') % compare linear and nonlinear on same plot figure(4) subplot(2.plot(tlin. % figure(2) subplot(2.1).2).2).plot(tlin.8].3.1.ylin(:.1)) title('linear state space.1). min') ylabel('jacket temp.2)) xlabel('time.2)) xlabel('time.125+ylinscale(:. generate function file with the following first line: function xdot = heater(t.1). options = odeset('RelTol'.tlin.x(:.150+ylinscale(:.To make the runs go faster. min') subplot(2.plot(t.x0. deg F') subplot(2.plot(t.2 -1.1). physical magnitude') xlabel('time.4 0.x] = ode45('heater'. deviation') xlabel('time.x(:. using ode45') xlabel('time.'AbsTol'.plot(t.1.
plot(t.2). deg F') legend('nonlinear'.'linear') subplot(2. linear') xlabel('time.title('nonlinear vs.x(:.2).tlin. deg F') legend('nonlinear'. min') ylabel('temp. min') ylabel('jacket temp.1.150+ylinscale(:.'--') xlabel('time.2).'linear') [ Team LiB ] .
4 In this text we do not focus on the solution of algebraic equations. Various quasi-Newton techniques provide approximations to the Jacobian and do not require as many function evaluations. we use the form Equation A. the Jacobian matrix in Equation (A.1: Solving Algebraic Equations Fortunately. [ Team LiB ] . using Equation A.3 The ij element of the Jacobian represents the partial derivative of equation i with respect to variable j. The most commonly used numerical techniques are related to Newton-Raphson iteration. If analytical derivatives are not available.2) is not inverted.[ Team LiB ] Appendix 2. The "guess" for iteration k + 1 is determined from the value at iteration k. requiring n + 1 function evaluations for an n-equation system of equations. In practice. the MATLAB fsolve function is easy to use for solving algebraic equations.2 where f[x(k)] is the vector of function evaluations at iteration k.66) with a fixed p and u. and J(k) is the Jacobian matrix Equation A. Rather.1 obtained from Equation (2. elements of the Jacobian are obtained by perturbation of the state variable. See the text by Bequette (1998) for more details on these techniques. Equation A. reducing computational time. For a simplified presentation. a set of linear algebraic equations is solved for x(k+1).
0816 min-1 we find. An example is shown next.2: An Isothermal Chemical Reactor. where k represents the time-step index. Substituting the numerical values for an integration step size of 0.2.2: Integrating Ordinary Differential Equations The Euler integration method often requires very small integration step sizes to obtain a desired level of accuracy.[ Team LiB ] Appendix 2. for the first time step. continued Consider the ethylene glycol reactor problem in Example 2. For example. and a space velocity (F /V) = 0.4.1 and CP(0) = 0. The Euler integration algorithm results in the following two equations. so that it is not confused with the time-step index. Note that x is a vector of n state variables at each time step. For C A(0) = 0. we find . We use k1 to represent the reaction rate constant.5 minutes. Example 2. if there are two states (two differential equations) we have left the inputs and parameters out of the function variable list for convenience.
MATLAB has a suite of routines for integrating differential equations.5 minutes Values can be obtained at future times by continuing to march forward in time. [ Team LiB ] . These are covered in Module 3.resulting in the concentration values at t = 0.
Dynamic Behavior The goal of this chapter is to understand dynamic behavior. first-order + dead time and integrating system step responses Understand second-order under-damped behavior Understand the effect of pole and zero values on step responses Convert state space models to transfer functions The major sections of this chapter are as follows: 3.7 Second-Order Behavior 3.5 First-Order Behavior 3. The main advantage to Laplace transforms is that they allow us to analyze behavior exhibited by linear differential equations by using simple algebraic manipulations.12 Converting State Space Models to Transfer Functions 3.2 Linear State Space Models 3.13 MATLAB and SIMULINK 3. We then introduce Laplace transforms.8 Lead-Lag Behavior 3. such as those developed in Chapter 2. We begin by working with linear state space models.14 Summary . which are the basis for many control system design techniques.9 Poles and Zeros 3.[ Team LiB ] Chapter 3. often obtained by linearizing a nonlinear model.4 Transfer Functions 3.1 Background 3.3 Introduction to Laplace Transforms 3.11 Padé Approximation for Dead Time 3. After studying this chapter. Laplace transforms are used to create transfer function models. the reader should be able to: Apply the initial and final value theorems of Laplace transforms Understand first-order.6 Integrating System 3.10 Processes with Dead Time 3.
[ Team LiB ] .
The dashed lines are those of negative gain processes. In this plot. The curves in Figure 3-1c are known as underdamped or oscillatory responses. then apply a step change to an input variable. (b) integrating. . Figure 3-1. Assuming a positive step change. distillation columns. The curves in Figure 3-1b are indicative of "integrating" processes. Assume that the process is initially at steady state. a prime example is a liquid surge vessel. and some adiabatic plug flow reactors. (a) Overdamped or first order. The curves in Figure 3-1a show a monotonic change in the output. that is. The solid curves are examples of "positive gain" processes. and (d) inverse response. this behavior is generally known as overdamped. The behavior shown in Figure 3-1d is known as "inverse response" and is seen in steam drums. and the dashed curves are indicative of "negative gain" processes. we assume that a positive step increase has been made to the input variable of interest. the process output increases for an increase in the input.[ Team LiB ] 3. This type of behavior may occur in exothermic chemical reactors or biochemical reactors. the solid curves are illustrative of "positive gain" processes. The majority of chemical processes will exhibit one of the responses shown in Figure 31. It more often occurs in processes that are under feedback control.1 Background One of the major goals of this chapter is to obtain an understanding of process dynamics. In the sections that follow. (c) underdamped or oscillatory. we discuss the characteristics of process models that lead to each of the behaviors shown in Figure 3-1. Process engineers tend to think of process dynamics in terms of the response of a process to a step input change. where the level continues to change when there is an imbalance in the inlet and outlet flow rates. particularly if the controller is poorly tuned. Common responses of process outputs to step changes in process inputs.
[ Team LiB ] .
Numerically. and outputs (y) are all perturbation or deviation variables Equation 3. return to zero. since they are deviation variables. Element cij relates the effect of state xj on output yi. In a matrixvector operation y = Cx. which is simply the perturbation of a variable from its steady-state value.[ Team LiB ] 3. When matrices multiply vectors. which were normally nonlinear in nature. Also. The process is stable if after a period of time. we can determine the stability of a state space model by finding the eigenvalues of . inputs (u). Stability One of the first basic concepts that we need to cover is the notion of stability. The shorthand notation for Equation (3. the first subscript refers to the row and the second subscript refers to the column. State space models have the following form.1) is Equation 3.1 Recall that in matrix notation. which relates the states to the outputs. while the column corresponds to a particular input of the operation. This led to the notion of a perturbation or deviation variable. Consider the C matrix. where the states (x ).2 Linear State Space Models In Chapter 2 we developed fundamental models. We then developed state space models that were based on linearizing the fundamental models at a steady-state solution. each row corresponds to a particular output of the multiplication. This means that the state variables. the number of columns in C must be equal to the number of elements in x. Consider a process where one or more states have been perturbed from the steady-state solution or operating point. the variables return to the steady-state values.2 It is important to always check for dimensional consistency in matrix operations. the number of rows in C must be equal to the number of elements in y.
as shown at the bottom.3). A system with all but one eigenvalues negative and with one eigenvalue equal to zero is called an integrating system and is characteristic of processes with liquid levels or gas drum pressures that can vary. Figure 3-2. There are analytical solutions for two. If an eigenvalue is real and positive. A is an n x n matrix. the system is unstable . Mathematically.3 where λ is known as an eigenvalue. (a) Monotonic and (b) oscillatory. Unstable responses. the system oscillates (with ever increasing amplitude). the eigenvalues of the A matrix are found from the roots of the characteristic polynomial Equation 3. and there will be n solutions (eigenvalues) of Equation (3. In two-state systems. For a state space model with n states. If there are complex conjugate eigenvalues.and three-state systems.the state space A matrix. if any single eigenvalue is positive. the system response is that shown in the top curves. the two-state solution is shown below. Remember that the A matrix is simply the matrix of derivatives of the dynamic modeling equations with respect to the state variables. then the system is stable. with positive real portions. Examples of unstable systems are shown in Figure 3-2. The determinant can be found by . and I is the identity matrix. If all of the eigenvalues are negative.
A more general method of qualitatively checking the stability. CSTR are detailed in Module 8. the first state is the concentration and the second state is the reactor temperature.5)] .4 The roots can be found using the quadratic formula Equation 3. the A matrix is (the time unit is hours) Operating condition 1 Operating condition 2 and the eigenvalues for operating condition 1 can be found using the following steps with the solutions [using the quadratic formula (3. For a particular reactor with two different operating conditions. Example 3.5 It is easy to show that if and a11 + a22 < 0 and a11a22 >a12a21. the roots (eigenvalues) are negative and the system is stable. is shown in Chapter 5. For a two-state representation.1: Exothermic CSTR Models for an exothermic. known as the Routh stability criterion.and the eigenvalues are found as the two solutions (roots) to Equation 3.
Since both eigenvalues are negative. the stability is determined by the sign of the real portion of the complex number.2324. If the unit of time used in the differential equations is minutes.6837 1. there will be n eigenvalues. Generalization Notice that a solution of a second-order polynomial was required to find the eigenvalues of the two-state example. the larger magnitude (more negative) eigenvalues are faster. It is too complex to find these analytically for all but the simplest loworder systems. The reader should use the MATLAB command window to verify the following results for the second operating condition: » a2 = [-1. The reader should show that the eigenvalues of A2 are where the positive eigenvalue indicates that operating condition 2 is unstable . the positive eigenvalue indicates that the second operating condition is unstable. [ Team LiB ] . for example.8124 -0. » eig(a2) ans = -0.9. then the eigenvalues have min-1 as the unit. In this case. The values of the eigenvalues are related to the "speed of response. in MATLAB the eig function can be used to find eigenvalues. For stable systems (where all eigenvalues are negative). For the general case of an n x n matrix. MATLAB Eigenvalue Function The M ATLAB eig command can be used to quickly find eigenvalues of a matrix. As long as all real portions are negative. operating condition 1 is stable. the system is stable. some eigenvalues may occur in complex conjugate pairs." and the eigenvalue unit is inverse time. The simplest way to find eigenvalues is by using existing numerical analysis software.4697]. this resulted in two eigenvalues. For matrices that are 2 x 2 or larger. for example.4939 Again.8366 0.
Equation 3.8 . transfer functions allow us to make algebraic manipulations rather than working directly with linear differential equations (state space models). The Laplace transform of a time-domain function. we need the notion of the Laplace transform. so the Laplace transform of a constant ( C) multiplying a time-domain function is just that constant times the Laplace transform of the function. is represented by L[f(t)] and is defined as Equation 3. f(t).7 The Laplace transforms of a few common time-domain functions are shown next.6 The Laplace transform is a linear operation. Exponential Function Exponential functions appear often in the solution of linear differential equations. Basically. Although we could develop these procedures using the state space models.3 Introduction to Laplace Transforms Most control system analysis and design techniques are based on linear systems theory. it is generally easier to work with transfer functions. Here The transform is defined for t > 0 (we also use the identity that ex+y = exey) So we now have the following relationship: Equation 3.[ Team LiB ] 3. To create transfer functions.
.. If f(t) represents a particular function of time.9 For an nth derivative... or measurement sample delays. so we can write Equation 3. then f(t – θ) represents the value of the function θ time units in the past. we can derive Equation 3.Derivatives This will be important in transforming the derivative term in a dynamic equation to the Laplace domain (using integration by parts). Time Delays (Dead Time) Time delays often occur owing to fluid transport through pipes.10) are 0. if the system is initially at steady-state. f (n – 1) (0) One reason for using deviation variables is that all of the initial condition terms in Equation (3.10 n initial conditions are needed: f(0). Here we use θ to represent the time delay. .
We can use a change of variables.13 Pulse . the Laplace transform of a constant. because the function is defined as f(t) = 0 for t < 0. is Equation 3. t* = t – θ. A "unit" step function is defined as and using the definition of the Laplace transform. A step function is discontinuous at t = 0.11 So the Laplace transform of a function with a time delay ( θ) is simply e–θs times the Laplace transform of the nondelayed function. Step Functions Step functions are used to simulate the sudden change in an input variable (say a flow rate being rapidly changed from one value to another). so Equation 3. Notice that the lower limit of integration does not change. Equation 3. to integrate the function. C.12 Similarly.
The Laplace transform is Equation 3. where the pulse period is decreased while maintaining the pulse area. . Pulse function.15 Figure 3-4. as shown in Figure 3-4. Concept of an impulse function.Consider a pulse function. Figure 3-3. as tp approaches 0. The function is f(t) = P/t p for 0 < t < tp and f(t) = 0 for t > t p. In the limit. as shown in Figure 3-3. where a total integrated input of magnitude P is applied over tp time units. the pulse function becomes (using L'Hopital's rule) Equation 3.14 Impulse An impulse function can viewed as a pulse function.
then look for the Laplace domain function in the second column and write the corresponding time-domain function from the first column. This notion of the inverse Laplace transform can be written Equation 3.17 For example . Other Functions It is rare for one to derive the Laplace transform for a function. or dumping a bucket of fluid or bag of solids into a chemical reactor.If we denote a unit impulse as f(t) = δ. Table 3-1 presents solutions for most common functions. if your goal is to "invert" a Laplace domain function to the time domain. Similarly.16 Examples of common impulse inputs include a "bolus" (shot or injection) of a drug into a physiological system. rather a table of known transforms (and inverse transforms) can be used to solve most dynamic systems problems. If you desire to transform a function from the time domain to the Laplace domain. then look for the time-domain function in the first column and write the corresponding Laplace domain function from the second column. then the Laplace transform is Equation 3.
The long-term behavior of a time-domain function can be found by analyzing the Laplace domain behavior in the limit as the s variable approaches zero.L .Sn-1f(0) .θ) t tn sF( S) .f(0) Snf(S) .sf (n-2)(0) .Initial.and Final-Value Theorems The following theorems are very useful for determining limiting values in dynamics and control studies.f(n-1)(0) e-at te-at e-θsF (s) 1-e-t/τ sin ωt . The initial value of a time-domain function can be found by analyzing the Laplace domain behavior in the limit as s approaches infinity. Laplace Transforms for Common Time-Domain Functions Time-domain function f(t) δ(t) (unit impulse) F (s ) 1 Laplace domain function C (constant) f(t . Table 3-1.
y(t).Time-domain function cos ωt Laplace domain function e–at sin ωt e–atcos ωt where . The final-value theorem is Equation 3.18a The initial-value theorem is Equation 3.and Final-Value Theorems Find the long-term and short-term behavior of the time-domain function.2: Application of Initial. using the final. Example 3.and initial-value theorems on the Laplace domain function Y(s) (we see later that this arises from a step input applied to a second-order process): .18b It should be noted that these theorems only hold for stable systems.
and y(t 0). are consistent with the final. y(t). The reader should verify that the time-domain function. Transform each element of the differential equation to the Laplace domain.and initial-value General Solution Procedure To obtain analytical solutions for differential equation-based models. is found using the initial-value theorem. Start with an nth order linear differential equation and known initial conditions 2.The long-term behavior. y(t ). 1. . the general procedure is composed of several steps. is found using the final-value theorem. can be found by applying Table 3-1 to find and that the values of y(t theorems. The short-term behavior. y(t 0). ).
3. Invert each element back to the time domain to find the final solution for y(t). where a. where N(s) and D(s) are polynomials in s. Perform a partial fraction expansion to isolate the individual elements. For the case of repeated roots (say 3 that have a value of a). and so forth are the roots of D(s).3. Equation 3. 4. as shown in Example 3. b. Example 3. 5. Usually. the solution has the form It should be noted that chemical process systems are rarely described by an nth order differential equation. Use algebraic manipulations to solve for the transformed variable.19 .3: Second-Order Differential Equation Consider the following state space model of an isothermal CSTR (Module 5). If roots are repeated the partial fraction expansion has the form (for an example where three roots all have a value of a). a set of n first-order differential equations is transformed to a single nth order equation.
where x1 and x2 represent the concentrations of two components (in deviation variable form) in an isothermal reactor. Solving for Y(s) . 1. we find Here we consider a unit step input. Taking the Laplace transform of each element. we find  Alternatively. 3. 4. Performing a partial fraction expansion using the roots of the denominator polynomial.20 . and the input is initially u(0)= 0.19) and use algebraic substitution to find this second-order equation in s. U(s) = 1/s. one could take the Laplace transform of each equation in (3. and substituting back into the differential equation. the initial conditions are x1(0) = x2(0) = 0. The reader should show (see exercise 6) that this can be arranged to a second-order differential equation. Equation 3. 2. with the initial conditions y(0) = dy(0)/dt = 0. Solve for the output (concentration of component 2) response to a unit step input.
Solve for C1 by multiplying by s. solve for C2 by multiplying Equation (3. In this plot we notice that the concentration initially decreases. it is generally more pleasing to engineers to view a plot of a variable as it changes with time. See Equation (3. Response of the concentration as a function of time.21 Although it is nice to have an analytical solution. Setting s = 0 to find Similarly.20) by s + 2. before increasing to a new steady-state value.20) by s + 2. Inverting each element back to the time domain. .2376 which yields 5. solve for C3 by multiplying Equation (3.21). Equation 3.4053 Also. Figure 3-5.2376 and setting s = –2. as shown in Figure 3-5.4053 and setting s = –2.
The primary goal of this section was to provide background material to understand the transfer functions and dynamic responses shown in the next section. such as Boyce and DiPrima (1992) present many examples of applications of Laplace transforms to solve differential equations.In Example 3.3 we solved for the coefficients (C1. In practice. it is far easier to solve these numerically. [ Team LiB ] . 2. C3) by selecting values of s to minimize the multiplications performed during each step. Differential equations textbooks.20). and 3. for example) in Equation (3. it is rare for process engineers to seek analytical solutions to differential equations. C2. An alternative is to solve for three equations in three unknowns by using various values of s (say s = 1.
22a Since we are assuming that the model is based on deviation variables. we find The ratio of polynomials is called the transfer function .[ Team LiB ] 3. .22b Taking the Laplace transform of each term. Solving for Y(s). When it relates a manipulated input to an output it is commonly called a process transfer function. In general. Equation 3. the Laplace domain relationship between a manipulated input and output is called a process transfer function. In particular. we will use gp(s) to represent the process transfer function. Consider an nth order differential equation. and that the system is initially at steady state. the initial conditions are Equation 3.4 Transfer Functions A transfer function relates inputs to outputs in the Laplace domain.
[ Team LiB ] . Transfer functions are often used in block diagrams. the relationship between an input and output is shown as In the rest of this chapter we study the dynamic behavior of some commonly used transfer functions.Equation 3. and the roots of the denominator polynomial are call poles. In the rest of this text we generally use lowercase letters for all input and output variables . If the argument is s. The significance of poles and zeros are discussed in detail in Section 3. We have used capital letters to distinguish Laplace domain variables from the time-domain variables.23 In this case the process transfer function is Equation 3. Our focus is on Step Responses .9. For example. since process engineers often apply step changes to understand dynamic behavior.24 The roots of the numerator polynomial are known as zeros. then we assume that we are referring to the Laplace domain.
[ Team LiB ]
3.5 First-Order Behavior
Many chemical processes can be modeled as first-order systems. The differential equation for a linear first-order process is often written in the following form:
This can also be written as
where the parameters ( τp and kp) and variables (y and u) have the following names: τp is the process time constant (units of time), kp the process gain (units of output/input), y the output variable, and u the input variable. Taking the Laplace transform of each term (notice that we are now using lower-case variables to represent the Laplace domain input and output), and assuming that the initial condition is y(0) = 0,
So the Laplace transform of Equation (3.25) can be written
or solving for y(s) we find a first-order transfer function
Consider the case where the output is initially zero (steady state in deviation variable form), and the input is suddenly step changed by an amount ∆u. The Laplace transform of the input is
So Equation (3.26) can be written
Using a partial fraction expansion and inverting to the time domain, you should find (see Exercise 1)
Here the notion of a process gain is clear. After a substantial amount of time (t >> τp), we find, from Equation (3.29),
and, since y(0) = 0, we can think of y(t
) as ∆y, so
We can think of the process time constant as the amount of time it takes for 63.2% of the ultimate output change to occur, since when t = τp,
Remember that this holds true only for first-order systems.
Consider now an impulse input of magnitude P, which has units of the input*time; if the input is a volumetric flow rate (volume/time), then the impulse input is a volume. The output response is
You should find that the time domain solution is
which has an immediate response of Pkp/τp followed by a first-order decay with time.
Example 3.4: Stirred-Tank Heater
Recall that an energy balance on a constant-volume stirred-tank heater (Example 2.3) yielded
where the subscript s is used to indicate that a particular variable remains at its steady-state value. Defining the following deviation variables
The equation can be written in the form
the resulting output change is [from Equation (3.29)] A plot of the step input and the resulting output are shown in Figure 3-6. . a process with a large volume-to-flow rate ratio is expected to have a slow response compared to a process with a small volume-to-flow rate ratio. since a given heat power input will have a larger effect on the small process than the large process. This makes physical sense. Step response of a stirred-tank heater. characteristic of a first-order system. For liquid water. The process gain and time constant are then Step Response If a step input change of 10 kW is made. Figure 3-6. A process with a large steady-state flow rate will have a low gain. compared to a process with a small steady-state flow rate. Consider a heater with a constant liquid volume of Vs = 50 liters and a constant volumetric flow rate of Fs = 10 liters/minute.or where the parameters of this first-order model are The gain and time constant are clearly a function of scale. Deviation variables. Similarly. the other parameters are ρcp = 1 kcal/ liter°C.
143°C (see Exercise 15).Remember that the inputs and outputs in this expression are in deviation variable form. [ Team LiB ] . the physical temperature response is Impulse Response If an impulse input of 30 kJ is made. you should be able to show that the temperature changes immediately by 0. If the steady-state values are (for an inlet temperature of 20°C) Then.
with y(0) = 0.[ Team LiB ] 3. the output ramps with a constant slope of k∆u.6 Integrating System We found in the previous chapter that material balances on liquid surge vessels or gas drums often yielded models with the following form: Equation 3. Step Response If a step input change of ∆u is made at t = 0.32 In the Laplace domain.33 Consider an integrating process initially at steady state. Impulse Response If an impulse input of magnitude P is made at t = 0.34 That is. and we find the time-domain value Equation 3. . this is Equation 3.
then the height as a function of time is Figure 3-7. If the steady-state height is 2 meters. A is the constant cross-sectional area of the tank. the output response is which is shown in Figure 3-7. .then the output immediately changes to a new steady-state value of Example 3. Defining the output and input in deviation variable form as For a constant cross-sectional area of 10 m2. the model is Step Response For a step input change of 0. F1 is the inlet flow rate. Step response of a liquid surge tank. Deviation variables. Assume that the outlet flow rate remains constant at a steadystate value of F2s.25 m3/min.5: Tank-Height Problem The mathematical model for a liquid surge tank is (see Example 1. and F2 is the outlet flow rate.3) where h is the liquid height.
the output response is which makes sense. because the cross-sectional area is 10 m3.Impulse Response For an impulse input of 1 m3. [ Team LiB ] .
35) or (3.35 This is often written in the form Equation 3. ζ the damping factor (dimensionless). The second-order model shown in Equation (3. Taking the Laplace transform of Equation (3. For a given second-order ODE. Equation 3.36 where (obviously a0 0) and the parameters are as follows: k is the gain (units of output/input).36).7 Second-Order Behavior Second-order models arise from systems that are modeled with two differential equations (two states). In this section we separately consider transfer functions that do not have "numerator" dynamics and those that do.37 where y(s) indicates the Laplace transformed variable. and τ the natural period (units of time). there are an infinite number of sets of two first-order (state-space) models that are equivalent.[ Team LiB ] 3.36) generally arises by changing a set of two first-order equations (state-space model) to a single second-order equation. Pure Second-Order Systems Consider a linear second-order ODE. . with constant parameters Equation 3.
We can find the roots (known as the poles) by using the quadratic formula which yields the following values for the roots: Equation 3. Step Responses Now. the denominator polynomial. The three possible cases are shown in Table 3-2. τ 2s2 + 2ζτs + 1. we consider the dynamic response of second-order systems to step inputs ( u(s) = ∆u/s). This implies that the real portions of p1 and p2 are negative and. that is . we find Equation 3.38 The characteristic equation of the second-order transfer function is τ2s2 + 2ζτs + 1. Equation 3.40 where ∆u represents the magnitude of the step change. Case 1: Overdamped (• > 1) For ζ > 1. therefore. the system is stable. can be factored into the form .39 The following analysis assumes that ζ >0 and τ > 0.Assuming initial conditions are zero.
we can divide by k∆u to develop a dimensionless output.where the time constants are Table 3-2. distinct roots Two real. as discussed next.41 Note that. Equation 3. Also. as in the case of first-order systems. Characteristic Behavior of Second-Order Transfer Functions Case 1 2 3 Damping factor ζ>1 ζ=1 ζ<1 Pole location Two real. Figure 3-8. Step response of a second-order overdamped system. equal roots Two complex conjugate roots Characteristic behavior Overdamped Critically damped Underdamped We can derive the following solution for step responses of overdamped systems. This is done in Figure 3-8. Most chemical processes exhibit overdamped behavior. the dimensionless time is t/τ and we can plot curves for dimensionless output as a function of ζ. which includes the critically damped case. .
whereas the first-order responses have their maximum slope initially. the step response of a system with a relative order greater than 1 has a zero slope at the time of the step input.42 Notice that the main difference between overdamped (or critically damped) step responses and first-order step responses is that the second-order step responses have an S shape with a maximum slope at an inflection point. The initial behavior for a step change is really dictated by the relative order of the system.43a .Case 2: Critically damped ( • = 1) The transition between overdamped and under damped is known as critically damped.39)] that the poles are complex. we find [from Equation (3. Equation 3. If the relative order is 1. We can derive the following for the step response of a critically damped system Equation 3. The relative order is the difference between the orders of the numerator and denominator polynomials in the transfer function. then output response has a nonzero slope at the time of the step input. Case 3: Underdamped (• < 1) For ζ < 1.
Interpolate between the curves for the behavior of other damping factor values. we can produce the dimensionless plot shown in Figure 3-9. dividing Equation (3.44b Again.where the real and imaginary contributions are Equation 3. Step response of a second-order underdamped system as a function of the damping factor (• ).44a) by k∆u. .44a where Equation 3. Figure 3-9.43b We can derive the following step response for an underdamped system. Equation 3.
(3) overshoot.39)] is As the imaginary/real ratio gets larger. Figure 3-10. (4) decay ratio. the response becomes faster. and defined below: (1) rise time. and (5) period of oscillation. As the real portion becomes larger in magnitude (more negative). We use these insights to interpret pole/zero plots in Section 3. Underdamped Step Response Characteristics The following common measures of underdamped second-order step responses are shown in Figure 3-10.A number of insights can be obtained from Figure 3-9 and from an analysis of the step response equations.9. Step response characteristics of underdamped secondorder processes. the ratio of the imaginary portion to the real portion of the pole [from Equation (3. (2) time to first peak. the response becomes more oscillatory (ζ becomes smaller). We also notice that a decreasing ratio corresponds to a larger negative value for the real portion. For ζ < 1. .
The reader should show that the pole-zero form is . Second-Order Systems with Numerator Dynamics The previous discussion involved pure second-order systems. as shown in Figure 310.Rise time is the amount of time it takes to first reach the new steady-state value. A b/a ratio of 1/4 is commonly called "quarter wave damping. Decay ratio is a measure of how rapidly the oscillations are decreasing. Overshoot is the distance between the first peak and the new steady state. where the relative order (difference between the denominator and numerator polynomial orders) was two. This is usually expressed as the overshoot ratio. Consider now a second-order system with numerator dynamics with the gain/time constant form Equation 3." Period of oscillation is the time between successive peaks.45 which is relative order one. Time to first peak is the time required to reach the first peak.
46 The reader should show that.6: Illustration of Numerator Dynamics Consider the following input-output relationship: The unit step responses are shown in Figure 3-11. Notice that negative numerator time constants (corresponding to positive zeros) yield a step response which initially decreases before increasing to the final steady state. the response is the same as a first-order process. Example 3.where the parameters are The gain/time constant form has the following time-domain response to a step input (see Exercise 4): Equation 3. Step responses of a second-order system with numerator dynamics. This type of response is known as inverse response and causes tough challenges for process control systems. acting in parallel (see Exercise 5). Positive zeros are often caused by two first-order transfer functions. if τn = τ2. Figure 3-11. with gains of opposite sign. .
Notice also that a numerator time constant that is greater than the denominator time constant causes overshoot before settling to the final steady state. [ Team LiB ] . Also notice that the inverse response becomes "deeper" as the numerator zero (–1/τn) approaches a value of 0 from the positive side.
Step responses of the lead-lag example. This can also be found by applying the initial value theorem to Equation (3. but many controllers exhibit such behavior. The responses to a step input at t = 0 are shown in Figure 3-12.47"/>) for a step input change (see Exercise 3).[ Team LiB ] 3.8 Lead-Lag Behavior Lead-lag transfer functions have the same order numerator polynomial as denominator polynomial.48) by kp∆u.47 For a step input of magnitude ∆u. and t/τp is a natural dimensionless time.48 A dimensionless output can be defined by dividing Equation (3. Notice that there is an immediate response that is equal to the τn/τp ratio. Figure 3-12. It is rare for processes to exhibit lead-lag behavior. In terms of the state space model (3. This occurs when the input has a direct effect on the output variable. this means that D 0. Consider a lead-lag transfer function where the numerator and denominator polynomials are first order: Equation 3. [ Team LiB ] .2). the output response is Equation 3.
9 Poles and Zeros There are a number of different ways to represent process transfer functions.51 where τni is a numerator time constant and τpi is a denominator time constant.49) to equal zero are known as the "poles" of the transfer function.50 and complex poles (or zeros) must occur in complex conjugate pairs.[ Team LiB ] 3. The "pole-zero" form is Equation 3. This form is normally used when the roots (poles) of the denominator polynomial are real. The "gain-time constant" form is the one that we use most often for control system design.49 The values of s that cause the numerator of Equation (3. Equation 3.49) to equal zero are known as the "zeros" of the transfer function. The "polynomial" form is Equation 3.7: Comparison of Various Transfer Function Forms Consider a transfer function with the following gain-time constant form . Example 3. The values of s that cause the denominator of Equation (3.
7 is shown in Figure 3-13 [the pole locations are (-1/3. there is no imaginary component and the poles and zeros lie on the real axis.0) and the zero location is (1/10. poles. that is. RHP zeros have a characteristic inverse response. For this system. Figure 3-13. with the coordinates (real. Pole-zero location plot for Example 3.0). as shown in Figure 3-11 for τn = -10 (which corresponds to a zero of +0. A positive zero is called a right-half-plane (RHP) zero. o. zero). This behavior is summarized in Figure 314. and increasing the magnitude of the imaginary portion makes the response more oscillatory. because it appears in the right half of the complex plane (with real and imaginary axes). A pole-zero plot of the transfer function in Example 3. and the poles are –1/3 and –1/15. RHP (positive) poles are unstable. Also notice that the poles are negative (left-half-plane). the system never settles to a steady state when a step input change .1).0) and (-1/15.imaginary)]. Notice that the zero for Example 3. Recall also that a process with a pole at the origin (and none in the RHP) is known as an integrating system. Recall that complex poles will yield an oscillatory response.The polynomial form is The gain-polynomial form is and the pole-zero form is The zero is 1/10. As poles move further to the left they yield a faster response. indicating a stable process.7 is positive.7 (x.
Effect of pole-zero location on dynamic behavior (x. for example. with two RHP zeros. the step response. switches direction. o. As the imaginary/real ratio increases. then switches back to the initial direction. poles. the response is faster. zero). As poles become more negative. Multiple RHP zeros cause multiple "changes in direction". the response becomes more oscillatory. Figure 3-14.is made. initially going in one direction. [ Team LiB ] .
10 Processes with Dead Time Many processes have a delayed response to a process input.4 we found that a time-delay of θ time units had a transfer function of e–θs. A first-order process combined with a time-delay has the transfer function Consider a process with kp = 1. We see that the time-delay shifts the response of the output. A unit step input change at t = 0 yields the response shown in Figure 3-15. Response of a first-order + dead time process. either due to transport lags (such as flow through pipes) or high-order effects. In Section 3. Figure 3-15.[ Team LiB ] 3. τp = 10. [ Team LiB ] .3. and θ = 5.
[ Team LiB ] 3. A first-order Padé approximation is A second-order Padé approximation is Example 3.8: Application of the Padé Approximations for Dead Time Consider the first-order + dead time transfer function. the Padé approximation for dead time is often used in this case. where θ is the time delay. Some control system design techniques require a rational transfer function. where the time-delay dominates the time constant The first-order Padé approximation yields the transfer function and the second-order Padé approximation yields .11 Padé Approximation for Dead Time As discussed in the previous section the transfer function for a pure time delay is e–θs.
g1(s). and g2(s) is shown in Figure 3-16. Notice that the first-order approximation has an inverse response. If you have a system of differential equations that has time delays. complex-conjugate zeros of the numerator transfer function of g2(s)." The reader should find that there is a single positive zero for g1(s). [ Team LiB ] . the Padé approximation can be used to convert them to delay-free differential equations. One of the many advantages to using SIMULINK is that time delays are easily handled so that no approximation is required. and there are two positive. Figure 3-16. Comparison of first-order + dead time response with first and second-order Padé approximations for dead time.A comparison of the step responses of g(s). Most ordinary differential equation numerical integrators require pure differential equations (with no time delays). which can then be numerically integrated. while the second-order approximation has a "double inverse response.
Starting with the state space model Take the Laplace transform of each term. the transfer function relating input j and output i is or in matrix form (with m inputs and r outputs) Example 3. assuming zero initial conditions Solving for x(s).12 Converting State Space Models to Transfer Functions A general state space model can be converted to transfer function form. using the following steps. then y(s) (it should be noted that often D = 0) where G(s) is a transfer function matrix.3 and Module 5.9: Isothermal CSTR Consider the isothermal CSTR shown in Example 3. For example.[ Team LiB ] 3. The state space model is .
manipulated) is the dilution rate ( F /V). B. The manipulated input-output process transfer function for the reactor is .The first input (u1. The sequence of steps used to find the transfer function matrix is and multiplying.4048 and –2. disturbance) is the feed concentration (C Af). the second input (u2. Also.2381 min-1. The eigenvalues of A (obtained by solving det(λI – A) = 0) are –2. and the output is the concentration of the intermediate component.
2381 and –2. [ Team LiB ] .and the disturbance input-output transfer function is The transfer function poles (–2.4048) are equal to the eigenvalues of the A matrix. An example of pole-zero cancellation is shown in Exercise 13. that is. Also. Sometimes the resulting transfer functions can be factored into lower order transfer functions because of pole-zero cancellation (a value of a pole is equal to a value of a zero). the positive zero (1/0. We see that it is straightforward to convert state space models to transfer function models. An nstate system results in transfer functions that have a denominator polynomial that is nth order in s.3549) in g11(s) yields the inverse response shown in Figure 3-5. with n poles.
3750 + 0. The following conv (convolve) command makes this straightforward. » g = conv([2 1.13 MATLAB and SIMULINK MATLAB and SIMULINK are ideal for analyzing and simulating dynamic behavior.2836i -0. Although the following multiplication is quite simple: (5s + 1)(3s + 1) = 15s2 + 8s + 1. » sysc = tf(num.0000 » p = roots(g) p = -2.0.5995i -0. The Control Toolbox uses LTI (linear.6. such as (2s2 + 1. These objects can be state space or transfer function models. the roots of the resulting polynomial are easily calculated using the roots command.0.2836i step First.[3 8 4 1]) g = 6.5 1].2786 .5000 23.2786 + 0. a more complex multiplication.3750 . define an object.0000 20. conv (Convolution) and roots Often it is necessary to multiply polynomials in s together. Here we first present some useful MATLAB functions not covered in other modules.0000 16.[ Team LiB ] 3.1095 -0. such one of the transfer functions in Example 3. » num = [-10 1].den) Transfer function: -10 s + 1 ----------------45 s^2 + 18 s + 1 . Also. den = [45 18 1].0000 5. time invariant) objects to represent dynamic models.5s + 1)(3s3 + 8s2 + 4s + 1) could be quite time consuming.5000 1. details on how to use these are presented in Module 4.5995i -0.
Simulink It is highly recommended that you interactively work through the initial sections of Module 2 to understand how SIMULINK can be used to simulate dynamic systems. such as pole and tzero are shown in Module 4.then plot the step response (use help step to view other options) » step(sysc) A number of other useful functions. [ Team LiB ] .
process. valve. A first-order + dead time model is the most common process representation. sensor. the system is stable. The process gain is a measure of the long-term change in a process output for a given input change. [ Team LiB ] . First and second-order Padé approximations are sometimes used for controller design.). characterized by damping factors less than 1. The concept of a transfer function is very useful for control system design and analysis. etc. the approximations lead to right-halfplane zeros and inverse response. If zeros are positive (in the right-half-plane). In future chapters we find that transfer functions allow the use of algebra rather than differential equations when analyzing control systems composed of a number of components (controller. The use of LTI objects (see Module 4) allows easy interconversion of models in MATLAB. State space models are easily converted to Laplace transfer function models. Complex eigenvalues (poles) correspond to underdamped (oscillatory) behavior. we will not be using them for this purpose during the rest of this textbook.[ Team LiB ] 3. step responses have a characteristic inverse response. The stability of a process is determined by the eigenvalues of the state space A matrix.14 Summary The main objective of this chapter was to develop an understanding of dynamic process behavior. The initial and final value theorems of Laplace transforms can be used to find the limiting behavior of transfer functions without having to obtain a full solution. Dynamic responses are also affected by the values of the zeros (roots of the numerator polynomial in s) of a transfer function. Time-delays are represented by an irrational term (e–θs) in the Laplace domain. or the poles of the Laplace transfer function: if all are negative. Although Laplace transforms can be used to obtain analytical solutions to differential equations.
[ Team LiB ] References For more details on analytical mathematical methods (including Laplace transforms) to solve differential equations. DiPrima. Wiley. New York (1992).. [ Team LiB ] . Ordinary Differential Equations and Boundary Value Problems. W.. 5th ed. and R. see: Boyce.
3: Use the initial value theorem to find the immediate response of a lead-lag transfer function to a step input change at t = 0. 5: Consider an input-output transfer function that mimics two first-order processes in parallel where . use the final value theorem to find the long-term response of a lead-lag transfer function to a step input change. solve for the timedomain output response to a step input change of magnitude ∆u at t =0. A second-order process with one pole at the origin has the transfer function 2: Find the output as a function of time. For kp = 1°C/Lpm. τn = 20 min find the peak temperature and the time that it occurs. Sketch the expected behavior. for a unit step input change.[ Team LiB ] Student Exercises 1: Solve for the time-domain output of a first-order transfer function to a step input change. τ1 = 3 min. Also. 4: For the following second-order process with numerator dynamics. τ2 = 15 min.
find the conditions (relationship between gains and time constants for the two transfer functions) that cause a right-half-plane zero (resulting in inverse response to a step input change) in gp(s). using the initial and final value theorems. Assume that the reactor was initially at a steady state.If the gain of g1 is positive and the gain of g2 is negative. and iii consistent? 8: As a process engineer with the Complex Pole Corporation. you decide to make a step change in the input variable. which is reactor temperature (notice that the reactor temperature is in °F). You obtain the following plot for the output variable. Find the output at t=0 and as t approaches infinity. 6: Consider the state space model Find the second-order differential equation in y. 7: Consider the following state-space model Which has the following input-output transfer function relationship For a unit step change in the input. Find the time domain solution. from 10°C to 15°C. you are assigned a unit with an exothermic chemical reactor. Are your results for i. the coolant temperature. y(t) iii. Hint: first solve for x 1 from the second equation. Sketch the time domain behavior of y(t) iv. then take the derivative and substitute into the first equation. In order to learn more about the dynamics of the process. Use Figure 3-9 to help answer the following questions. ii. . ii. u(s) = 1/s: i.
iii. Write the second-order transfer function. ii. Hint: Use Figure 3-9 to assist you. v. shown in the figure. What is the value of the process gain? (show units) ii. What is the period of oscillation? (show units) vi. What is the decay ratio? v. 9: Match the transfer functions with the responses to a unit step input. iv. What is the value of τ? (show units) iii. What is the value of ζ? (show units) iv.i. . i.
12: .75 mol/L. with an input flow rate of 5 gpm and an output of 0.m. You become distracted and do not have a chance to look at the output variable again. You note that the output has ceased to change and has achieved a new steady-state value of 0.10: Consider the following state space model: i. with units? Show your work. ii.8 mol/L.85 mol/L. Find (analytically) the time-domain output response to a unit step input change. You make a step increase of 0. Sketch the expected response to a unit step input change. until you leave for happy hour at a local watering hole at 6:30 p.m.5 gpm at 3:00 p. and do not observe any changes until 3:07 p.m. you decide to develop a first-order + time delay model of a process using a step test. The process is initially at steady state. At 3:20 p. Find the transfer function model.. 11: As a process engineer. What are the values of the process parameters.m. the value of the output is 0. Use the initial and final value theorems of Laplace transforms to determine the initial and final values of the process output for a unit step input change to the following transfer functions.
iii. Since there are two states (the A matrix is 2 x 2) we expect that the process transfer function will be second-order. ii. Show that pole-zero cancellation occurs. shown in the figure. v. 14: Match the transfer functions with the responses to a unit step input. i. ii. . 13: Consider the following state space model for a biochemical reactor. Find the values of the gain and time constant. iv.i. resulting in a first-order transfer function. iii.
4. find the output response as a function of time.15: Consider Example 3. Consider the following second-order transfer function 16: For a unit impulse input. What is the peak change and when does it occur? [ Team LiB ] . For an impulse input of 30 kJ. find the value of the vessel temperature immediately after the impulse input is applied.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.