ICEBO 2013, MONTREAL, CANADAoaktrust.library.tamu.edu/bitstream/handle/1969.1/151438/ESL-IC-13...

Development of an Online Expert Rule based Automated Fault Detection and Diagnostic (AFDD) tool for Air Handling Units: Beta Test Results

ICEBO 2013, MONTREAL, CANADA

Ken Bruton, Daniel Coakley, Peter O’Donovan, Marcus M. Keane & D. T. J. O'Sullivan

ESL-IC-13-10-30a

Proceedings of the 13th International Conference for Enhanced Building Operations, Montreal, Quebec, October 8-11, 2013

• Why does industry need this work?

• Why HVAC?

• An automated solution…..

• AFDD techniques

• Early Results

Agenda

ESL-IC-13-10-30a


• Focus on production output in lieu of efficiency

• Difficult to track efficiency due to islands of information

• Competition from lower cost economies for investment

Energy Consumption remains relatively constant though production volume has decreased significantly

Why does industry need this research work?

ESL-IC-13-10-30a


• Typically greater than 20% of an industrial site’s energy consumption

• HVAC systems get less efficient over time

• 20‐30% energy savings are achievable by re‐commissioning HVAC

0

5

10

15

20

25

30

35

Manufacturing HVAC CHW LTHW Others

Why Focus on HVAC Systems?

ESL-IC-13-10-30a


Operation & Management

Control

Process

Equipment

Energy Service Requirement

How can AFDD help save HVAC Energy?

ESL-IC-13-10-30a



Control

Process

Equipment


How can AFDD help save HVAC Energy?

Air Change Reduction

ESL-IC-13-10-30a



Control

Process

Equipment


How can AFDD help save HVAC Energy?• Inefficient control strategies• Poor loop tuning• Free cooling• Incorporation of Deadbands

• Simultaneous heating & cooling

• Set point adjustments• Manual operation• Sub optimal performance of

equipment

ESL-IC-13-10-30a


Why an automated tool?

ESL-IC-13-10-30a


AFDD tool requirements

ESL-IC-13-10-30a


• FDD automates the process of detecting and diagnosing faults

• A rule based FDD tool can be developed and implemented in industry relatively quickly utilising existing equipment

FDD Methods

Quantitative Model Based

Detailed Physical Models

Simplified Physical Models

Qualitative Model Based

Rule Based

Qualitative Physics Based

Process History Based

Black Box

Gray Box

FDD Techniques

ESL-IC-13-10-30a


• Business layer expanded on existing knowledge based rule sets by;

– Applicable fresh air & recirculation AHU’s as well as return air units

– Detecting issues when the AHU is off

– Calculating virtualised readings

– Improved error threshold calculation

• A server side application performs the mode checks, calculates the virtual values, applies the business layer rules, and stores the results in the database

The Business Layer

ESL-IC-13-10-30a


• > 200 AHU’s available on 7 industrial and commercial sites

• AHU’s were selected with– Different component and sensor layouts

– Varying levels of instrumentation

Design of Test Study – AHU Selection

SiteNo. AHUsin pilot

Type Type of zone(s)supplied

Operatinghours perannum

BMS/DataloggingPlatform

Frequency oflogged data

1 2 Constant volume Office & canteen 8760 Trend 15 minutes

2 4 Constant volume Manufacturing Floor 8760 Tridium 15 minutes

3 9 Variable Volume Manufacturing Floor 6240 Cylon 15 minutes

4 4 Constant Volume Commercial officespace

6240 Cylon 7.5 minutes

5 3 Variable Volume Commercial officespace

6240 Schneider 15 minutes

6 2 Constant Volume Manufacturing Floor 8760 Qlikview 15 minutes

ESL-IC-13-10-30a


Design of Test Study ‐ Feedback

ESL-IC-13-10-30a


AFDD tool in Alpha testing

ESL-IC-13-10-30a


Alpha to Beta Testing

632012‐08‐18 18:15 Sat 19

Hour 13Day 2Week 0

TRUE TRUETRUE TRUE Zone

TRUE V V

V V

V + ‐

No. fault instances 157… as % of period 47%… as % of run hours 47%

Automated Diagnosis, Costing and Prioritisation under development

14.3m3/s

Description of current fault

Component has been design load (fully open) for a

number of consecutive hours

VSD 20%Flow 12.3m3/s Flow

Go to ……….. fault instanceVSD 26% 0% 34%

11.4°C 21.7°C59%

Fault selection 24.8°C 25.8°C 18.9°C 18.7°C 9.2°C

List faults by… 10.0°C 10.4°C

Return air Supply airRise 1.0°C 0% Rise 1.0°C

100% 100%

Faults in this hour:10.0°C

OutOfControl_zonT_Hi, ComponentPosition_AtCapa

city_cooV

site3 : ahu9Exhaust air Outside air Mean Weather

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

18 00 06 12 18 00 06 12

Fraction

ope

n

Hour of day

cooV

Temps

Humidities Virtual valueOther cond

Oldest

ComponentPosition_AtCapacity_coo

NextNewest Previous

Valve/damper position

…frequency …cost

Reset schematic diagram

Browser based GUI

Development of web based configuration tool

Secure HTTPS communication

ESL-IC-13-10-30a


AFDD tool in Beta testing

April 2012 November 2012

February 2013 March 2013

ESL-IC-13-10-30a


AFDD tool in Beta testing

ESL-IC-13-10-30a


Requirements Versus Developments

Rationale Requirement(s) Framework/ToolDeveloped

Data Access Layer Flexibility Compatibility with any BMS type or age A generic data access tool was developed

Business Layer Flexibility Compatible with various combinations of sensors and components

Calculation of virtual values coupled with “rule libraries”

Reliability Low number of false positives/negatives A error threshold applied to each rule based onrule makeup

Usability User friendly graphical user interface (GUI) A browser based GUI was developed

Fault Priority Quantification and prioritisation of the diagnosed faults

Each fault is prioritised in terms of cost andfrequency of occurrence

Scalability Rapid setup time per AHU A web based site configuration tool was developed

Low Cost Ability to use existing measurementsFirst principal techniques and engineering computation utilised to calculate readings where none are present

ESL-IC-13-10-30a


Results

ESL-IC-13-10-30a


Next Steps• Develop OPC functionality to expand data capture capabilty

• Expand the business layer to incorporate humidity control AFDD

• Further develop the fault costing & prioritisation methodology

• Improve diagnostic and future prognostic capabilities

• Automate the link to planned maintenance systems

ESL-IC-13-10-30a


Ken Bruton

i2e2 Researcher

Department of Civil & Environmental Engineering,

University College Cork

[email protected]

If we knew what it was we were doing, it would not be called research, would it?Albert Einstein

ESL-IC-13-10-30a


ICEBO 2013, MONTREAL, CANADAoaktrust.library.tamu.edu/bitstream/handle/1969.1/151438/ESL-IC-13...

Documents

Transcript of ICEBO 2013, MONTREAL, CANADAoaktrust.library.tamu.edu/bitstream/handle/1969.1/151438/ESL-IC-13...