Embedded Controllers for Increasing HVAC Energy Efficiency Increasing HVAC Energy Efficiency by Automated Fault Diagnostics J. Vass, J. Trojanová, R. Fišera, J. Rojíček Honeywell Prague Laboratory Automation & Control Solutions GREEMBED Workshop, Stockholm, April 12th 2010
26
Embed
Embedded Controllers for Increasing HVAC EnergyIncreasing …€¦ · Automation & Control Solutions GREEMBED Workshop, Stockholm, April 12th 2010. Outline • Honeywell Building
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Embedded Controllers for Increasing HVAC Energy EfficiencyIncreasing HVAC Energy Efficiency
by Automated Fault Diagnostics
J. Vass, J. Trojanová, R. Fišera, J. Rojíček, j , , j
Honeywell Prague LaboratoryAutomation & Control Solutions
GREEMBED Workshop, Stockholm, April 12th 2010
Outline
• Honeywell Building Solutions (HBS)
• JACE controller (embedded system)
• HVAC diagnostics- Air Handling Unit (AHU)
• AFDD algorithmD t Cl i- Data Cleansing
- Mode Detection- System Observationy- Fault Isolation
• Examples
Honeywell
Transport & Power12% Aerospace
Specialty 12% Aerospace
40%Materials
16%
Automation and Control SolutionsSolutions
32%
Automation & Control Solutions (ACS)
More than 50,000 employeesProfileProfile
HoneywellSecurity
HoneywellSecurity
HoneywellBuilding Solutions
HoneywellBuilding Solutions
HoneywellProcess Solutions
HoneywellProcess Solutions
HoneywellSensing & Control
HoneywellSensing & Control
HoneywellE & C Control
HoneywellE & C Control
HoneywellLife SafetyHoneywellLife Safety
Embedded System (JACE)
• JACE = JAVA® Application Control Engine• Integration Controller
Integrator- Integrator Integration of communication protocols (LON, BACnet, Modbus etc.)Graphical interface & Built-in web serverData processing capability (e.g. AFDD)
( 1) if ( ) 2% and ( ) 2%( ) s i iCCh HCtc UT t tU
Counter for
( g ) g
( ) ( ) ( )( )
0 otherwise
( 1) if ( ) 2% and ( ) 2%
s i i
s i i
CC
c CC
h HCh
HC
c
U
t
t T t tc U
Counter for Heating mode
Counter for ( ) ( ) % ( ) %( )
0 otherwise
( 1) if ( ) 2% and ( ) 2%
s i ic CCc
C
H
C
C
i iCHc
t
t T t tU
c
U
Counter for Cooling mode
Counter for ( 1) if ( ) 2% and ( ) 2%( )
0 otherwises C
vCi iCv Hc
ct T t t
tU U
Counter for Ventilation mode
Module 3: System Observation
Vector of Measurements: CCHCSASSAMA UUTTT ,,,,
Vector of Measurements:
Observation o2 Function of supply air temp & setpoint
Th ibl l 1 0 1
if1
Three possible values: 1, 0, -1
T
T
2 if 0if 1
)(
SASSA
SASSA
TTTT
o setpoint is met
above setpoint
T
T2
if 1)(
SASSA
SASSA
TTp
below setpoint
Observed State
• Mapping between observations & states- Each state defined by particular combination of o1, o2 & o3
• 45 possible states of the AHU (states are mutually exclusive)• 45 possible states of the AHU (states are mutually exclusive)- States are classified as normal and abnormal
• Mapping between abnormal states & faults- Binary diagnostic matrix- Represents the expert knowledge
1 f lt1 = fault0 = don't know
State s12 implies faultsF2 F3 & F7F2, F3 & F7
State s1implies faultsF2 & F6
Fault Aggregation
• Fault Relevance is updated in time using CUSUM: )1(),(,0max)( tRksfmtR ijii )(),(,)( f ijii
Previousvalue
Currentvalue
Forgetting factor k=0 3
Mapping function output 0 or 1 valuevalue factor, k=0.3output, 0 or 1
vanc
et R
elev
Faul
t
Example 1: Stuck Heating Valve
• Air was heated despite UHC = 0 (heating control signal)• Controller compensates by increasing UCC (cooling control signal)• Simultaneous heating & cooling wasted energy financial losses• Simultaneous heating & cooling wasted energy financial losses
detected by d t l i
coolingdata cleansing
e
ventilation
Most likelyl
evan
ce
Most likely fault is F2(stuck heating
only one AHU state
at each time
ault
Rel heating
valve)instant
(confirmedFa
correct fault cannot be isolated
(confirmed by building technician)
Example 2: Oscilating control signalC With D t Cl iWithout Data Cleansing With Data Cleansing
Data cleansing needed to avoid detection of non-existing faults
Summary
• AFDD algorithm for HVAC equipment- HVAC system = Major energy consumer in buildings
Red ce energ asting b a tomated detection of HVAC fa lts- Reduce energy wasting by automated detection of HVAC faults Abrupt hardware faults; Performance degradation
- Performed by Honeywell JACE controller (embedded system)- Graphical visualization in JACE AHU scheme, Measured data, Observed state, Fault relevances
AHU diagnostics• AHU diagnostics- Rule-based diagnostics (based on APAR by Schein et al.)- Data cleansing moduleData cleansing moduleDetect raw data errors (outliers, oscillations, etc.) Protects the AFDD algorithm from wrong decision (hoax faults)
- Diagnostic mapping tableMeasurements → Observations → States → Faults
Fault aggregation- Fault aggregation Fault relevance (of each fault) is updated in time