1 development of empirical dynamic models from step response data some processes too complicated...

23
1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to model using physical principles material, energy balances flow dynamics physical properties (often unknown) thermodynamics

Upload: kristina-lamb

Post on 20-Jan-2018

217 views

Category:

Documents


0 download

DESCRIPTION

3 Step Input Step response is the easiest to use but may upset the plant manager Other methods –impulse - dye injection, tracer –random - PRBS (pseudo random binary sequences) –sinusoidal - theoretical approach –frequency response - modest usage (incl. pulse testing) –on-line (under FB control)

TRANSCRIPT

Page 1: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

1

Development of Empirical Dynamic Models from Step Response Data

Some processes too complicated to model using physical principles

• material, energy balances• flow dynamics• physical properties (often unknown)• thermodynamics

Page 2: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

2

Black Box Models

Page 3: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

3

Step Input

• Step response is the easiest to use but may upset the plant manager

• Other methods– impulse - dye injection, tracer– random - PRBS (pseudo random binary sequences)– sinusoidal - theoretical approach– frequency response - modest usage (incl. pulse testing)– on-line (under FB control)

Page 4: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

4

Cha

pter

7

Page 5: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

5

Fitting of 1st-Order Model

/

0

11

0.632

1 1

t

t

K MG s U ss s

y t KM e

y KM

dyKM dt

Page 6: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

6

Cha

pter

7

(θ = 0)

Page 7: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

7

FOPDT and SOPDT Models

2 2

First-Order-Plus-Dead-Time (FOPDT) Model

1

Second-Order-Plus-Dead-Time (SOPDT) Model

2 1

s

s

KeG ss

KeG ss s

Page 8: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

8

For a 1st order model, we note the following characteristics in step response:

1. The response attains 63.2% of its final response at one time constant (t = ).

2. The line drawn tangent to the response at maximum slope (t = ) intersects the 100% line at (t = ).

There are 3 generally accepted graphical techniques for determining the first-order system parameters and .

( )1

sKeG ss

Fitting of FOPDT ModelC

hapt

er 7

Page 9: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

9

Cha

pter

7

Page 10: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

10

Method 1: Sundaresan & Krishnaswany (1978)

1. Find K from stead-state response.2. Normalize step response by dividing all data with KM (t =

0, y = 0; t →∞, y = 1)3. Use 35.3% and 85.3 % response times (t1 and t2), i.e.

4. Calculate = 1.3 t1 – 0.29 t2

= 0.67 (t2 – t1)

Cha

pter

7

1

2

0.353

0.853

y t KM

y t KM

Page 11: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

11

Method 2: Numerical Fitting

(1) Find and in ( ) 1

to fit data of vs.

(2) Find and in ln

to fit data of ln vs.

t

y t KM e

y t

KM y t tKM

KM y tt

KM

Page 12: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

12

Cha

pter

7Method 3: Fitting an Integrator Model

to Step Response Data

In Chapter 5 we considered the response of a first-order process to a step change in input of magnitude M:

/1 1 ty t KM e

For short times, t < , the exponential term can be approximated by

/ τ 1τ

t te

so that the approximate response is:

1 1 1τt KMy t KM t

Page 13: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

13

Cha

pter

7is virtually indistinguishable from the step response of the integrating element

22 (7-23)KG s

s

In the time domain, the step response of an integrator is

2 2 (7-24)y t K Mt

Hence an approximate way of modeling a first-order process is to find the single parameter

2 (7-25)τKK

that matches the early ramp-like response to a step change in input.

Page 14: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

14

Cha

pter

7

Figure 7.10. Comparison of step responses for a FOPTD model (solid line) and the approximate integrator plus time delay model (dashed line).

Page 15: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

15

Fitting 2nd-Order Models

1 21 1

MU ss

KG ss s

Page 16: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

16

Cha

pter

7Harriot’s Method

1.3

0.73

Page 17: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

17

0.39

0.26

1 2

Page 18: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

18

Harriot’s Method

0.73

0.73

0.731 2

0.5 1 2

0.5 0.5

1) Determine experimentally to satisfy 0.73

2) Calculate 1.3

3) Calculate 0.5

4) Determine from experimental data

4) From Figure 7.6, det

ty t KM

t

t

y y t

1 2.ermine and then calculate

Page 19: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

19

Smith’s Method

60

20

20

60

1) Determine t and t experimentally so that 0.6

0.2

2) Fig 7.7 ,

y t KM

y t KM

tt

Page 20: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

20

Cha

pter

7

Page 21: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

21

Cha

pter

7

Page 22: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

22

1.3=

1.79= 8.260

t

84.081.3

2

1

1 2 Sum of squares S 3.81 0.84 0.0757

NLR (θ=0) 2.99 1.92 0.000028 FOPTD (θ = 0.7) 4.60 - 0.0760

Smith’s Method20% response: t20 = 1.8560% response: t60 = 5.0t20 / t60 = 0.37from graph

Solving,

122

121

Cha

pter

7

Page 23: 1 Development of Empirical Dynamic Models from Step Response Data Some processes too complicated to…

23

Cha

pter

7