functional modeling of control systems

63
Functional Modeling of Control systems Morten Lind, Prof. Emeritus Automation and Control

Upload: others

Post on 07-Jan-2022

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Functional Modeling of Control systems

Functional Modeling of Control systems

Morten Lind, Prof. Emeritus Automation and Control

Page 2: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Agenda

2 15 February 2014

• Status of Multilevel Flow Modeling • Control functions: some new challenges • Action theory and functional modeling • Preliminary findings

Page 3: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

circulation of water

production of power

distribution of power

Multilevel Flow Modeling The basic idea

conversion of energy

Functions of systems and subsystems are described in relation to a context of use or purpose

Page 4: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

circulation of water

production of oil and gas

distribution of gas

The idea apply to most engineering domains

conversion of energy

Functions of systems and subsystems are described in relation to a context of use or purpose

Page 5: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Research in Multilevel Flow Modeling

5 15 February 2014

Action theory

Page 6: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

6 15 February 2014

Functions are Context Dependent

supply of power

production of power

circulation of water

conversion of energy

A function of the pump impeller in the context of water circulation

A function of the power plant in the context of power supply

A function of the pump in the context of power production

Page 7: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

The Means-End Relation and Functions

The grains are transported and ground by rotation of the runner stone The runner stone is rotated by energy produced by the waterwheel The waterwheel produce energy by filling the buckets with water from the flume

by = ”by means of”

Functions of a Watermill

The concept of function is related to the Means-End relation

Page 8: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Motivations and Background

• Operator Decision support • Integrating process and automation

design with operation • Intelligent control (autonomy)

8 15 February 2014

Page 9: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Motivations

Concepts of means and ends, goals and functions play a significant role in human understanding of complex systems (and mundane reality)

– Support of operators decision making in diagnosis and counteraction planning

• Information presentation • Situation assessment and decision making • Knowledge representation and reasoning support

– Analysis of control and safety requirements for complex systems

– Integrated process and automation design – Intelligent Automation (agent systems)

9 15 February

2014

Page 10: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Conceptual foundations

• Complexity and concept of function • The means-end relation

10 15 February 2014

Page 11: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

11 15 February 2014

Complex Industrial Systems and Infrastructures

From a Technology View To a Functional View

Page 12: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Two types of operational complexity

12 15 February 2014

The complexity of purposes of operation: The objective of a control agent is dependent on the overall operational goals and the objectives of other agents. Goals may change and be conflicting.

The complexity of dynamic nonlinear physical interactions: The contol agent must manage the dynamic interactions with the process and the other agents

Page 13: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

What is a power plant?

13 15 February 2014

The natural framework The power plant is a natural object Modeling the power plant by the structural relations and the physical and chemical laws that govern its behaviour e.g. by differential equations

The social framework The power plant is a man made object - an artifact The power plant is purposeful i.e. Its functions are directed towards the satisfaction of human and societal needs. The purpose of automation systems is to ensure that design intentions are achieved

The framework for functional modeling (MFM)

Page 14: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Maintain water level withinsafe limits

Maintain conditionfor energy transport

Keep room temperaturewithin limits

Transport ofwater fromsupply to expansion tank

Circulationof water

Transport ofenergy fromboiler toradiator

Radiator

Pump

Water supply

Boiler Valve

Expansiontank

Maintain comfort

Com

pone

nts

Func

tions

Obj

ectiv

es

ENDS

MEANS

Goa

ls

Ends

Means

Purpose

Function

Behaviour

Structure

The Means-end relation and Functions

Page 15: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Multilevel Flow Modelling

• Basic principles • Control functions • Safety functions • Operating Modes

15 15 February 2014

Page 16: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Multilevel Flow Modeling

16 15 February 2014

• Process functions are represented by elementary flow and control functions interconnected to form function structures representing a particular goal oriented view of the system.

• MFM is founded on fundamental

concepts of action: Each of the elementary flow and control functions can be seen as instances of more generic action types.

Page 17: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

ht

Vcfi

V0

fo

Function structure of a simple tank process

source sink storage

transport

Page 18: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

MFM model of a watermill

The grains are transported and ground by rotation of the runner stone The runner stone is rotated by energy produced by the waterwheel The waterwheel is producing energy by the filling the buckets with water from the flume

Explanation:

Page 19: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

The Watermill

mfs2

so3 bal2st2tr7 tr8 tr9 tr10st3

Filling the buckets with water from the flume

Page 20: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

The Watermill

so2

mfs2

so3 bal2st2tr7 tr8 tr9 tr10st3

Producing energy filling the buckets with water from the flume

Page 21: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

The Watermill

efs2

so2 tr4

mfs2

so3 bal2

st1

st2

tr5

tr6 si4

si3

tr7 tr8 tr9 tr10st3

Rotating the stone by energy produced by the waterwheel

Page 22: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

The Watermill

tr1

efs2

so2 tr4

mfs2

so3 bal2

st1

st2

tr5

tr6 si4

si3

tr7 tr8 tr9 tr10st3

Transporting the grain by rotating the runner stone

Page 23: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

mfs1

tr1

tr2

tr3

so1 bal1

si1

si2

efs2

so2 tr4

mfs2

so3 bal2

st1

st2

tr5

tr6 si4

si3

tr7 tr8 tr9 tr10st3

The Watermill The grains are transported and ground by rotation of the runner stone

Page 24: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

ob1

mfs1

tr1

tr2

tr3

so1 bal1

si1

si2

efs2

so2 tr4

mfs2

so3 bal2

st1

st2

tr5

tr6 si4

si3

tr7 tr8 tr9 tr10st3

The Watermill

Page 25: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Another example: A heat transfer loop

MFM model without control functions

Page 26: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Control functions

Page 27: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

27 15 February 2014

Cascade control A Control loop in MFM

Page 28: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

28 15 February 2014

The heat transfer loop extended with controls

MFM model with flow regulator MFM model with flow and

temperature regulator

Page 29: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Cause - effect reasoning in MFM

Applications •Fault analysis •Alarm filtering •Rasoning about control •Counteraction planning …

Page 30: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Event propagation in MFM models

Event propagation within a flow structure

Event propagation across flow structures

Page 31: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

The MFM Workbench: An integrated MFM model editor and reasoning platform

31 15 February 2014

Page 32: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Integrating process, control design and operation

Process design Control design

Ope

ratio

n

Page 33: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

MFM Application Examples

• Nuclear Power • Oil and Gas

33 15 February 2014

Page 34: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

MONJU Nuclear Power Plant

34 15 February 2014

Page 35: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Control System for MONJU Plant

35 15 February 2014

T

S F

n

B C

B C

F

T

+ -

+ - + - + - + - + -

+ - +

+ -

T P

F

B C

Δ P

B C

Power demand master

Reactor power

program

Reactor vessel outlet sodium temperature

program

PHTS flow

program

SHTS flow

program

Feed water flow

program

Reactor power

controller

CRDM controller

PHTS flow

controller

PHTS circulation

pump controller

SHTS flow

controller

SHTS Circulation

pump controller

EV outlet steam temperature

setting

Main steam pressure setting

Rector

Control rod

IHX

EV

SH

SHTS circulation pump

PHTS circulation pump

EV outlet steam temperature controller

Feed water flow

controller

SH outlet steam temperature

setting

Main steam temperature controller

Main steam pressure controller

turbine controller

Feed water control valve Feed water pump

Condenser

Steam control valve

Feed water control valve differential

pressure controller

Feed water control valve differential

pressure setting

Moisture separator

(average)

Page 36: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

MFM of Monju Breeder Reactor

36 15 February 2014

Model developed with Prof. Yoshikawa (HEU) as part of Chinese 111 project on MFM based risk monitoring of NPP. Topics of special interest • Modeling safety barriers

and defense in depth. • Modeling operating

modes

Page 37: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

37 15 February 2014

Page 38: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

38 15 February 2014

Water heating functions

Water circulation functions

Feed water pumping and control functions

MFM model of Monju of hot water warming mode

Page 39: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Mode transitions and MFM models

39 15 February 2014

?? ??

Page 40: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Gas Separation Plant

40 15 February 2014

Page 41: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

MFM of three-phase separation process

Page 42: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Detailled MFM of separation function and reasoning results

42 15 February 2014

Sto3 lo (pressure is low)

Page 43: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Research challenges

• Modeling methodology and tools – ”From structure to function”

• Operating modes and transitions • Combining MFM with dynamic simulation models (DSM)

43 15 February 2014

Page 44: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Modeling the interaction between the process and control agents • Functions in Multilevel Flow Modeling are actions • Control agents act on the process Using Von Wrigth’s theory of action as a common basis for modeling process actions and control actions

Page 45: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Actions are defined by two situations (Von Wright)

Situation Explanation Illustration

Hypothetical state with no agent

The state of affairs changes from si to sh due to the dynamics of the environment

Actualized with one agent

The agent acts and the state of affairs changes from si to sA instead of sh

si

sh

A

si sA

sh

Page 46: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

The elementary change types

Change schema Description

~pTp p happens

pTp p remains

pT~p p disappear

~pT~p p remains absent

Page 47: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

The elementary interventions

Intervention

Change Action schema Description

~pTp (p happens)

~pT[pI~p] produce p

pTp (p remains)

pT[pI~p] maintain p

pT~p (p disappear)

pT[~pIp] destroy p

~pT~p (p remains absent)

~pT[~pIp] suppress p

Page 48: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Elementary actions and transitions

48 15 February 2014

Page 49: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Composite actions and transitions

49 15 February 2014

Page 50: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Control functions

50 15 February 2014

General Control function types: 1. Direct control (loops with setpoint control and disturbance rejection) 2. Start-up, shut down and transition between modes 3. Optimizing control

Direct control functions (type 1) are included in the current MFM function library Type 2 and 3 control functions should be included

Page 51: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

There is a control action type (1) corresponding to each of the elementary interventions

Intervention Control action

produce steer

maintain regulate

destroy trip

suppress interlock

Page 52: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Relations between elementary action types and control functions(type 1)

52 15 February 2014

Page 53: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Advantages of the action types

• They comprise a complete set • They correspond to control action types known from control

engineering • The existing MFM functions have a logical foundation in the action

types • They are generic – the specific meaning depends on the state of affairs

p

Page 54: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Control of mode transitions

54 15 February 2014

Abstract states representing the modes

Page 55: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Elementary actions and transitions

55 15 February 2014

Page 56: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Example: translocation of objects Intervention Omission

Move-to(x,y)

The object y is moved to location x(from somewhere)

x

? y? let-move-to(x,y)

The object y move by itself tolocation x x

? y

Keep-at(x,y)

The object y is kept y at location x.x

y

?

let-stay-at(x,y)

The object y stay by itself atlocation x

xy

Move-from(x,y)

The object y y is moved away fromlocation x (to somewhere)

x

?

y

let-move-from(x,y)

The object y move by itself awayfrom location x. x

?

y

Keep-away-from(x,y)

The object y is kept-away-from(x,y)

x

? y

?

let-stay-away-from(x,y)

The object y stay away by itselffrom location x

x

? y

?

Page 57: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Composite actions and transitions

57 15 February 2014

Page 58: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

The transfer relation

58 15 February 2014

Page 59: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Reasoning about control

59 15 February 2014

Page 60: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

The control relation

60 15 February 2014

What does it mean that an agent (system 1) control an object (system 2)?

p is the desired state of system 2

q is the actual state of system 2

Page 61: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Promoting: transitions

p maintainp~pmaintain

~p

producep

produce~p

let phappen

let ~phappen

let ~premain

let premain

Page 62: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

Opposing: transitions

p suppress~p~psuppress

p

destroy~p

destroyp

let ~pdisappear

let pdisappear

let premainabsent

let ~premainabsent

Page 63: Functional Modeling of Control systems

DTU Electrical Engineering, Technical University of Denmark

63 15 February 2014

Thank you for your attention