A framework for distributed control and building performance simulation

CoMetS'12, June 25-27, Toulouse

Azzedine Yahiaoui1 and Abd-El-Kader Sahraoui2

1Center for Building & Systems TNO – TU/e,

Technische Universiteit Eindhoven, The Netherlands 2Systems Engineering & Integration Group,

LAAS-CNRS of Toulouse , France

A Framework for Distributed

Control and Building

Performance Simulation

2

Outlines

Problem Statement

Building Automation and Control Systems

Classical and Advanced Control Systems

Integration Possibilities

Development and Implementation of a Distributed Dynamic Simulation Mechanism

Modelling: Example of Application

Simulation Results

Conclusions and Future Work

3

Problem Statement (1)

Domain based Control Modeling Environment (e.g., Matlab/Simulink)

very advanced in control systems design

but still limited in building performance simulation

Domain specific Building Performance Simulation (e.g., ESP-r) relatively based on modeling of building zone, plant ...

Marrying two approaches by run-time coupling

integrated building performance assessments

4

Problem Statement (2)

Extending this potential by distributing one or more ESP-r(s) with Matlab/Simulink over a network

typical pattern of distributed simulation between control systems and building performance applications

as qualified by similarity to BACS architecture

5

Building Automation and Control Systems

1

2

3

Central computer

Netwerk

Substation(computer)

+

-

Substation

Figure 6.1

Building Automation System.

Control &

Management

Protocol

Building &

Plants

General architecture of BACS Modern BACS Architecture

‘LonMark’

‘BACnet’

Protocols

6

Classical Control Applications

Control

law Room

ri Reference

command

ei yi

di

+

-

Input

disturbance

ui

1 2 3

MATLAB ESP-r

Communication

mechanism

Send Rec.

Send Rec.

•Optimal Control

•Predictive Control

•Fuzzy Logic

•Neural Nets

•…

Building Control Application

•Optimal Control

•Predictive Control

•Fuzzy Logic

•Neural Nets

•…

Advanced Control Applications

7

Integration Possibilities

Struct function

ESP-r

control data

MEX-file

Matlab/Simulink

control data

run-time coupling infrastructure

These APIs must present the highest

possible abstraction meaningful to

run-time coupled software tools

• Shared file

• Pipes

• Shared Memory

• Sockets

(XML/SOAP)

• CORBA

• HLA

• …

8

Classification Different Possibilities

Trade-Off Analysis

Requirements

IPC

Flexibility Reliability Concurrency Scalability Transparency Applicability

Pipe + + + + + −+

Standard File + − + + −+ −+

Shared memory ++ ++ + ++ ++ +

Socket ++ ++ ++ ++ ++ ++

CORBA ++ ++ ++ ++ ++ −

Performance comparison and ease use indicated

Sockets were best suited for Development and

Implementation of run-time coupling between two

or more software tools on different platforms

9

SE Structured Approach to run-

time coupling Development (1)

Sub-System Level

System Level

User’s

Requirements

System

Requirements

System Architecture

and Design

Acceptance

Test

System

Test

Integration

Test

Technical

Requirements

High Level

Design

Technical

Requirements

High Level

Design

Sub-System

Test

Integration

Test

Sub-System

Test

Integration

Test

Component Level

Software

Requirements

High Level

Design

Integration

Test

Unit

Test

Integration

Software

Requirements

High Level

Design

Integration

Communication

Requirements

High Level

Design

Integration

Test

Unit

Test

Unit

Test

Integration

Test

Implementation

Building Model Controller Run-Time Coupling

Mechanism

Time

Level of

Abstraction

10

SE Structured Approach to run-

time coupling Development (2)

11

Run-Time Coupling Implementation

network

ESP-r

Un

ix-v

arian

t

or

Win

dow

s

Un

ix-v

arian

t

or

Win

dow

s

Matlab/Simulink

network

ESP-r

Un

ix-v

arian

t

or

Win

dow

s

Un

ix-v

arian

t

or

Win

dow

s

Matlab/Simulink

Run-time coupling between ESP-r

and Matlab/Simulink

advantage of modeling building model and its control

systems separately and using different platforms

Distributed Control and Building Performance Simulation

1 2 3

12

Conceptual Design & User Interfaces

Matlab Side ESP-r Side

13

Extension of Run-Time Coupling

to Represent BACS in Simulation

Extension Possibilities for

Multiplexing (using select() function)

Broadcasting (using SO_BROADCAST API)

Multitasking (using POSIX treads – library)

Run-time Coupling Infrastructure

ESP-r (1) MatlabESP-r (9)

Conventional representation Equivalence in V-lifecycle model

Conceptual view of a Distributed simulation between

Matlab/Simulink and one or more ESP-r (s)

14

Distributed Simulation Mechanism

for BACS representation

1

2

3

15

Run-Time Coupling Options

Data Exchange Formats:

ASCII format

Binary format

Web-Services (SOAP/XML) format

Communication Modes:

Synchronous

Asynchronous

Partially Synchronous

16

Modelling: Example of Application

Control of Building Heating System

17

Building

Control systems

Sensors

Actu

ato

rsReferences

Modelling: Example of Application

Control of Building Heating System

Matlab/Simulink ESP-r

18

Simulation Results: Continuous and

Digital Control Systems

Simulation results obtained with continuous PI control system

Simulation results obtained with discrete PI control system

19

Conclusion & Future Work

A distributed simulation mechanism for BACS technology by run-time coupling Matlab/Simulink and one or more ESP-r(s) is implemented as a practical solution for improving control applications in ABs

Using a SE methodology to define all required functionalities in the development, implementation, validation, and operation of run-time coupling between ESP-r and Matlab/Simulink early in V-moel

Future work will envisage to analyze and simulate complex and large-scale building control applications involving the utilization of multiple ESP-r(s) by run-time coupling to Matlab/Simulink

20

Any Questions & Comments

Thank you for your

kind attention