ON-SITE CHILLER MONITORING SYSTEM FOR PREDICTION, DIAGNOSTIC AND
OPTIMIZATION At
HP DEPOT ROAD PLANT
Trang 2ABSTRACT
As large commercial and plants owners become more conscious and critical on chilled water system that not only consume a significant proportion of electric
energy in a built environment, but also contribute to the bottom line of the operations,
be it for human comfort or meeting process requirements for business continuity, an accurate, reliable and yet sufficiently comprehensive chiller monitoring system will become an integral part of today and future centralized chillers, which previously often being regarded as black-box
In view of these, Hewlett-Packard (HP) Depot Road plant chiller monitoring system is being commissioned with the ultimate aim to predict, diagnose and optimize the respective chillers in the system with non intrusive measurements consistent with the approach of Gordon and Ng Data collected from the field is systematically being filtered by a macro program for steady state operating parameters which are then being processed with a MS Solver for critical irreversibility parameters such as internal entropy generation, finite rate heat exchange and heat leaks, which are used to track the chiller performance Concomitantly, multiple linear regressions are performed to statistically determine the threshold of these parameters to minimize any false representations
Despite the large and yet complex chillers operations, equipped with built-in heat recovery such as economizer and intricate control mechanisms such as feed- forward adaptive control, the analytical model developed by Gordon and Ng from the basic principles of thermodynamics has been successfully tested and served as a basis for the analysis and evaluation of this research project
Trang 3ACKNOWLEDMENT
The author wishes to express sincere appreciation to project supervisor, Prof KC Ng, and Mr Jayaprakash the postgraduate PhD student in the Department of Mechanical Engineering, National University of Singapore for their invaluable guidance, support and encouragement during the planning, execution and assessment of the project Also, thanks to Mr Sacadevan, from Heat Transfer laboratory, for the site verification of the field RTD temperature measurements Also, acknowledging the contribution from other colleagues in HP, namely Chen Fei and Khee Boon for their help and advises during the project implementation and instrumentation and control set up
Trang 4Table of Contents Chapter 1 Introduction LI Back proud information cacscescencaseveonasenensmccresmannserinescnnreuseumenesmeeed El 1.2 Motivationortereseafchifoj6GEu.ssssesee.ceoeaseenniniisdonannoeakseseoe 1-3 1/21 EHSTEY ŒŨUSEbnonsoisgsb1G00880543GSD380G400340383396054030358œs@œuœaassssii 1.2.2 Environmental impaet -. -2+ s2 2+522+2z+z+xzxrzezrerxerxererre 1-5 1.2.3 System efficiency 1.2.4 System reliability -ô-s-côsxseeretreerererrrrrrerrrrrrrrrrrrrrree T~, 13 (QBGHVBonectnnerintiiicoagT.01015138015120030180L1000216354g10050066315.pnua0stsvssndsucudlLLSđ) 14 Scope of research and organization repOrt . -e -cc. .- Ï=Ø Chapter? THEỌW¿ystoiortotoiltitqHGGOODOEGGGHGINIENDENGEEODNAIRSSWARHW.NG 2-1 21 Thermodynamic and operational fundamentals of mechanical chiller
2.2 Universal Chiller Model by Gordon and Ng - 5-5-2 2-5 Chapter3 Equipment and Instrument Detail s-«555+c<xexeeeeeeeeeeeeee 3= 3.1 GillettdetlliÌBioeeereaeeniooroniissvGi01813511236036131610004052350413805360040030080p5oE58gvsvz2SL 32 Field instrument and data acquisition c c+ceccsccc-e.ce.e.- 3= 33 Measurement unCertainty . ¿- 5-5-5225 5s+xesszeexereerrrxrrrrreereevee T7 Chapter 4 Fault Scenarios and Data Processing -+ e + - đcÏ
4.1 D))0042 6T 4-1
4.2 Steady state data filtration
4.3 Determining COP prediction and irreversible parameters .- 4-3 Chapter 5 Data Analysis and Discussion .c0.cecssesseseeseeeeeseeeseeneeeeseeeereeeeneee 7d 51 Condenser tubes fouling - ¿- 55-52 52s éSzketekerkrkrrkrkerkrrree 5-1 52 High condenser coolant inlet tetaperafUre ‹-s- 5-55 ssexvzexeex 5-6
Trang 5Chapter 6 Conclusion and Recommendation .sccessecesseseseeseeeeteseeeesteeeseeeeses 6-1 Appendix A HP Depot Road Chilled Water System Equipment Schedule A-1 Appendix B Manufacturer Documented Chiller Part Load Efficiency B-1 Appendix C Measurement UncertaintY .- - 5+ ++svs++xe+zxerrxexererrerxrree C-1 Appendix D Field Data s65 sxsxseseeeerrrerrrrrrrrrrrrrrrrrrrrrrrrrreee TS AppendixE.Macro PfopramimiifE:‹‹-‹ se‹s csssszsss2,s21615422121140160156 0 00 10050080156508880668066 E-1
Trang 6List of Figures
Figure 1: WW BI Chiller Chilled Water System Schematic . 1-2 Figure 2: WW B1 Chiller Condenser Water System Schematic - |=2
Figure 3: Singapore LT Electricity Tariff vs Pegged Fuel Price
Figure 4: HP Depot Road Plant Energy Profile esesccesssscssseseeeseseseeeeeeseseeeneeeees 1-4 Figure 5: HPSG Carbon Footprint and Intensity .0 cseseeeeseeeeeseseeeeeerseeeseeenee LAS, Figure 6: Chiller 4 Condenser Approach Temperature after Online Tube Cleaning 1-7 Figure 7: Chiller 4 kW/RT after Online Tube Cleaning -. -. -.-« Ï=Đ Figure 8: Schematic of Reversible Carnot Refrigerant CycÌe -. « 2-2 Figure 9: Schematic of a Real Vapor Compression Mechanical Chiller with Complete Operational System :.cccesceseeesesessessseseseseeesseneecscscecseerseensnsnsesssesseaceeeceresseseteseesee 274 Eigure 10: General CVHE and CVHG Unit Componens -. -e- 7 Figure 11: 2-stage Economizer with 3-stage ChilÌe ¿ -c<ccc+cv-c e.- 3~2)
Figure 12: Manufacture Performance Data of 850 RT Chiller, kW/RT vs Load 3-4
Figure 13: Typical SCADA Graphic for Chiller ccc cece escent 28
Figure 14: Chiller 1 and 2 Before/After Tube Cleaning (COPpreq vs COP measured &
1/COP pred VS 1/Qevap) Charts 5-2
Figure 15: Chiller 1 and 2 Thermal Resistance with 1 and 2-Mean Standard Error 5-4 Figure 16: Chiller 1 and 2 Internal Entropy Generation with 1 and 2-Mean Standard EITƠE1566669021556058010S0GBGI2i0GI3G4ISSS3SHEI0iliSvbSisGiSR1310GxStiil2388@xsysygsisgasoÐ=5
Figure 17: Chiller 1 at Various Condenser Coolant Inlet Temperature (COPprea vs COP Measure and 1/COP prea VS 1/Qevap) Charts
Figure 18: Chiller 2,at Various Condenser Coolant Inlet Temperature (COPprea VS
COP Measure and 1/COPpwa vs 1/Qevap) ChartS 55555<55ccccc 2=
Trang 7Figure 19: Chiller 4 at Various Condenser Coolant Inlet Temperature (COP prea vs
COP Measure and 1/COPpred VS 1/Q¿¿ap) ChaTts 5-5255 25+2c2cezzerxerxrrrr 5-9
Figure 20: Chiller 5 at Various Condenser Coolant Inlet Temperature (COP prea vs
COP Measure and 1/COPpza vs 1/Qvap) Charts -25-25-555<cc<<c<cscc 5-10
Figure 21: Chiller 1, 2, 4 and 5 Thermal Resistance with 1 and 2-Mean Standard Error
Figure 22: Chiller 1, 2, 4 and 5 Internal Entropy Generation with 1 and 2-Mean
Standard Ervot ccssssssssssssessssssssssssssssssessssessssssssssssssesssscsssseasscesssesneasssessscssesseeeeee S71 Figure 23: Chiller 4 and 5 at Normal and Low Condenser Coolant Flow Rate (COP prea
vs COP Measure and 1/COP prea VS 1/Qevap) Charts .scesscsssssessessesssesecssecessseseeneesees 5-14
Figure 24: Chiller 4 and 5 Thermal Resistance with 1 and 2-Mean Standard Error 5-16 Figure 25: Chiller 4 and 5 Internal Entropy Generation with 1 and 2-Mean Standard
ETTOT 2S thành HH1 2101111111 111 rrrưn 5-17
Figure 26: Typical Plot with Experimental and OEM as Designed Data for 1/COP prea
Trang 8List of Tables
Table 1: 2005 Crude Oil and Natural Gas Reserves, Production and Consumption 1-4
Table 2: Chiller 1 and 2 Critical Irreversibility Parameters Before/After Tubes
Cleaning
Table 3: Chiller 1, 2, 4 and 5 Irreversibility Parameters for Various Condenser
Coolant Inlet Tem perature’s:sscisscsescceosecasnevnsinensasveonnaveuvanvnecusminssceccansuasearseanussnancsnensse 5-6 Table 4: Chiller 4 and 5 Irreversibility Parameters at Normal/Low Condenser Coolant ElbW RR[E:::zssgg 806 tEEL0001G3196018058G0001GG818804tA3853 12L GNEGRGIRSoSSngultsseremiiD-15
Trang 9HP HPSG RTD kW/RT kWh kV ATC STM coi Qhot ARI COP ASint gies eqy CHWS/R CWS/R Nomenclature Hewlett-Packard Hewlett-Packard Singapore
resistance temperature detector
kilowatt per refrigeration ton kilowatt-hour
kilovolt
auto tube cleaning
simple thermodynamic model work input
heat transfer
heat removal to cold reservoir heat removal to hot reservoir American Refrigeration Institute Coefficient of Performance rate of internal entropy generation
effective thermal resistance of heat exchangers equivalent heat leak parameters of a chiller chilled water supply/return
condenser water supply/return root mean square
Trang 10Chapter 1 Introduction
1.1 Background information
In a typical plant operation, chilled water system is often regarded to as energy center, and their “uptime” or fully operational period also plays an important role in ensuring interruption free business operation It is therefore imperative to ensure that these equipments are constantly being monitored with a reliable and accurate monitoring system
HP Depot Road building is served by two chiller systems, one at West Wing Basement 1 (WW BI); the other one at Central Utilities Building Level 2 (CUB L2) The former serves the entire Depot Road building load for thermal comfort and process cooling purposes Whilst the latter serves only the Jetmos and Tij 4 wafer fab clean rooms and their other related cooling processes This project will focus on the non intrusive measurement and analysis of the WW B1 chillers
The WW BI chiller system consists of 2 x 500 RT and 3 x 850 RT Trane R123 centrifugal chillers with primary and secondary chilled water loop The chillers are supported by 4 cooling towers located at the West Wing Roof Top (WW LS) The cooling towers are of counter flow type with cooling capacity of 1500 RT each Both the chiller and cooling tower capacity is based on N+1 During normal operation, 2 x 500 RT and 2 x 850 RT chillers are in operation with 1 x 850 RT chiller as standby Figure 1 and 2 depict the HP Depot Road WW BI chiller chilled and condenser water system schematic The detailed equipment schedule is outlined in Appendix A
Trang 121.2 Motivation for the research project 1.2.1 Energy costs
With escalation crude oil prices which reached a record high of US$100/- a barrel in the later half of 2007 LH, Singapore’s economy which hinges on the fossil fuel would not be spared The Singapore’s households energy cost has reached a high of S$0.2138/kWh for the period of October to December 2007 The common industrial high tension large (22 kV) rate has also recorded a new high of S$0.1888/kWh for peak period (7 am to 11 pm) for the same period Figure 3 shows the recent changes in the fuel and electricity tariffs of Singapore 100 100 Fuel Oil Prices (S$) Vs Low Tension Tariff 74g 8852 ard “ a "4e moms A , # _ ‘en TY Ly = 70 2° we 20 7 g see ` eB ASP gs” 4242 4308438148 129/524 “ ® : re 30 40 a * 1087 on a9 ULV og oo 46 gp 1601 1651 4524 8" s56 aSat eas 158 188 10201673 5008 a 4957 210220 49711521 6429 00 4 992052 2161 cv an 2 — tg Og tg Pegged Fuel ice —-LT Tait } 19 Ott ‘Apr-Jul Oct- Jan- Apte Jub Octe Jane Apte Jule Oct Jan- Agr Jul Oct- Jan- Apr tL Ơi 01 01 02 02 02 02 03 03 03 03 04 04 04 04 05 05 05 05 Source: SP Services
Figure 3: Singapore LT Electricity Tariff vs Pegged Fuel Price
The quest for economic growth will inadvertently increase the energy requirements and costs as the majority of the energy is produced by fossil fuels, which are of finite natural resources The proven recoverable reserves and total consumption as reported by the World Energy Council’s Survey of Energy
Resources 2007 ”! is shown in Table 1 Based on these data, the crude oil and
natural gas is projected to diminish in about 43 and 63 years, respectively
Tariff
Trang 13Table 1: 2005 Crude Oil and Natural Gas Reserves, Production and Consumption
Total proved Total production 'Total consumption recoverable reserves
Crude oil + natural 159,644 3,897 3,725
gas liquid (million tones) Natural gas (billion 176,462 2,883 2,796 cubic meters)
Source: World Energy Council’s Survey of Energy Resources 2007
The increase in fuel costs has had a significant impact to HP Depot Road plant operation, even with contestable rate since November 2006 The energy cost
has increased 37% from November 2005 to November 2007, whereas the total
Trang 14This has prompted the management to seriously consider the implementation of energy efficiency and optimization projects Besides the chiller monitoring project, many other projects have been implemented
1.2.2 Environmental impact
Global warming and climate change poses similar pressing issues to energy studies Thus, the chiller plant performance improvement could
significantly help to reduce the carbon emission footprint Even tough, the carbon intensity of HP Singapore has declined as compared to the year 2002, owing to business expansion, good business growth and energy conservation program, the carbon footprint has been steadily increased as shown in Figure 5
HPSG Cabon Footprint and Intensity 4080 |———————— — a 61% 150,000 ——— 02 eq (tonnes) 100,000 FY02 Fy0 FY04 FYOS FY0 FY
—®—HPSG CO2 eq _‹- LJMS CO2eq ~#-HPSG Cabon Inlensiy compare lo FY02 level
Trang 15The HP management has recognized the need to reduce the contribution to carbon emission so as to be aligned with the environmental management system, ISO14k, a tool that has been put in place and for good corporate citizenship of the
company
1.2.3 System efficiency
HP Depot Road chiller plant has been audited by a third party energy consultant for its facilities energy usage The energy indices for the chiller, cooling tower, chilled water and condenser water pumps are 0.554, 0.057, 0.065 and 0.106 kW/RT, respectively By the industry standards, this is considered to be a good performance status but there are rooms for improvement
1.2.4 System reliability
There is direct relationship between chiller efficiency and its uptime It is a known fact that cleaner condenser water tubes would not only increase the chiller efficiency but also ensure the chiller would not surge and trip due to lower heat rejection rates This can be easily proven with effectiveness of the heat exchanger where, e=l—e *“”, and e"⁄ =1! with T, =T.-T,
ch
Ta, Tc and Tạ are condenser approach, saturated refrigerant and coolant
Trang 16where a reduction of 1°C in condenser approach temperature when the particular chiller condenser tubes being cleaned online and on-load with brush and basket type of auto tube cleaning system, and Figure 7 depicts that the chiller kW/RT has reduced by more than 10% This is a classic example of the capability of the online monitoring and ATC system
(8/2908 294 11 PM) 18" ‘Civ CONDENSER APPROACH TEIAP (PWEESVRITET_I61 3701 204.07) Y-cwrseit} 040) 15 BE 14 ‘oye (8/2908 294 11 PM) 01 'C -17%ÊQ sy2,009800) 7n 32 BA t aman Ta
BS Kxx enlmnbn Pesnloe 6n Ietmimsise [owe Gene
Gia 3 KewstPV hae 4 wa PV TAS 85k 4463 0y 1 1 Decl Womeererab 227060 327670 Cyc 3324001185 DươnM MpErovvvMe 3/890 57870 Cyoke | SABRIOS 1405
Grater «owner Pv —— 05 Sa ố ỐỀẼ saz006 1105,
R CHOCORDENSER APFROACHTE — CHLLERJOWHIURT PY ——= 65 08H9IƠNG+ — Dẹng Wppsresehb 3/880 37870CvME 5320081105 03 28% 1 Dong Wpprewvvnb 27880 27670 Cys $906 1105, .~ € EPRRENEBSVR ———_]CHACONGOISDRAPPROACHT— 82 24°C 1 Dachnal Wppeborver -2276R8 227674 Cyehe 592086 1285
Trang 17
HALLER S CW RVVRT PV (IPWEESURTON 101 too 34 OD) ‘yet 8/2998 29£11 P0) OSKMNTORS Y-cwso(t) 05] 35 [D4] 00/40MTGNS 58/2005 20641 PM 05 MTOR BORAITONS (1 days, 00:00) 1 ar [ANH GÀ H2, xe rgfnJd13, kh, 4Ba,-cnue Pot nad “onl - SAR ERE AST n C[ e " dz
sài TAO Soman 3i AC Tae 7
hư 4Ị HNESẸR — G8xsMomMPV + HANEBSVR Đam —_ Cher 4 Howat PY lesnlum JaeselEsme ]Joxse_— 1kw T0 7 1 fmomg Wgmrevekb - Becta Wp prcerveib —]
DRHEMERSR ——_ CIRLER ACHR PY 00 0 RIWTONS! — Deemg Wpgsewvehe 321670 Cyete
3 CÍSNEBSVR —)CHMLLrR3CWRWRTPV —— 9 teMMTO.t Đechual "Hppebserver 227688 32T6T.O Cyclic 5192006 11105
SP HVEBSVR ODCOREOEER AMROAGHTE 90 10% t—=—(Decinal WWppisevwMb 327280 207670 Gyete 1920081106 XTET-IBI-MU-S ĐỊT HEREĐSR CH CONDENSER APPROACH TE 00 40° 1 Dong Wipgtnanrar 227680 327670 Cycke 3920061105
Figure 7: Chiller 4 kW/RT after Online Tube Cleaning
1.3 Objectives
There is a saying “you cannot manage what you do not measure” This is absolutely true There are 3 main objectives for this project:
1 To set up a reliable and accurate chiller monitoring system,
2 To analyze field data for the prediction, diagnostic and optimization of the chillers as well as the chilled and condenser water pumping systems, 3 To serve as a learning curve for an on-site real-time chiller prediction,
diagnostic and optimization
Trang 181.4 Scope of research and organization report
The scope of this research project can be categorized into 2 main categories:
1 Installation of the appropriate chiller monitoring system and verify the
measurement accuracy and systematic uncertainties of the measurement
2 Investigate of the chiller performance under various predetermined fault scenarios based on proven theoretical methodology
The report is structured in 6 chapters The subsequent Chapter 2 describes
the basic theory of thermodynamic for a vapor compression cycle and introduces
the generally acceptable simple thermodynamic model (STM) The model is developed by JM Gordon and KC Ng for the predictive, diagnostic and optimization for real centrifugal chillers Chapter 3 provides the insight on the detail of the chillers to be investigated and field instrumentation set up The data acquisition system and measurement uncertainties are presented in this chapter This is followed by Chapter 4 which aims at providing the predetermined faults scenarios and details on the data processing methodology towards establishing the chiller’s irreversibility parameters and performance prediction Detailed discussion on the findings from the operational scenarios is presented in Chapter
5, with tables and charts Finally, Chapter 6 summarizes the results, reveals the
challenges on STM and suggests future improvements to on-site energy monitoring and control programs
Trang 19Chapter 2 Theory
2.1 Thermodynamic and operational fundamentals of
mechanical chiller
A Carnot refrigeration cycle is an ideal reversible thermodynamic cycle where both the internal and external irreversibilities are assumed zero It comprises 4 reversible branches, namely isentropic compression; isothermal cooling or condensation; isentropic expansion; and isothermal heating or vaporization These four processes are elaborated as follows and shown in Figure 8:
1 Work W is input, adiabatically compressing the refrigerant and raising its
temperature
2 The refrigerant rejects heat Qhot, isothermally to a hot reservoir at temperature Tho, which is typically at 35.0°C, in accordance to ARI
Standard 550
3 The refrigerant is expanded adiabatically and isentropic
4 Heat Qcoia is removed from cold reservoir at temperature Teoig by
isothermal transfer to the refrigerant T.oig is usually designed at 6.7°C, according to ARI Standard 550
For ideal cycle, W + Ĩ „„ = Ở,„„
Trang 20Heat Rejection \ Werk Input, Ww Refrigeration Cycle (Chiller)
Figure 8: Schematic of Reversible Carnot Refrigerant Cycle
In comparison, the real COP of a chiller is usually much lower than the Carnot cycle’s COP In a real chiller, many dissipative losses are incurred For the purpose of determining the efficiency of a real chiller system, the Coefficient of Performance COP is introduced It is defined as ratio of useful effect or cooling capacity to the power input to achieve that useful effect as shown in following expression: COP = Qeota Ww And, it is also noted that COP is related to kW/RT by the following expression: kW _3.517 RT COP
A mechanical chiller consists of 4 major components namely compressor, condenser, thermal expansion or throttling valve and evaporator as shown in Figure 9 Other auxiliaries could be incorporated within the chillers, depending on the design and application requirements, such as the economizer, non condensable purge unit, oil tank and pump, inlet guide vane, etc Coolant systems are the
Trang 22Cooling Tower Flowmeter Condenser Water Pump Chiller Condenser Throttling Valve Compressor Evaporator Flowmeter Chilled Water Pump
Air Handling Unit
Figure 9: Schematic of a Real Vapor Compression Mechanical Chiller with Complete Operational System
Trang 23Non intrusive measurements such as flow and temperature sensors are
installed in the coolant circuits together with the power monitoring to determine the work input to the compressor
As opposed to the ideal cycle, a real cycle is effected by the following irreversibilities processes and conditions:
1 finite size heat exchangers and hence the finite rate heat transfer losses 2 non isentropic compression
3 no isentropic expansion via throttling valve 4 pressure and mechanical friction losses 5 heat leak
6 and, other operational faults which will be discussed in Chapter 4 and 5
Hence, actual COP is far below the Carnot limit
2.2 Universal Chiller Model by Gordon and Ng
Cooling capacity and COP are important deterministic performance parameters for chiller manufacturer; ACMV system designers and integrators; users and plant owners in designing and selection of chillers and centralized chilled water system
In a simplified model and assuming steady state operation, there are 3 critical irreversible parameters which govern the performance of a typical mechanical chiller, namely rate of internal dissipation or entropy production, AS,,, (kW/K); effective thermal resistance of heat exchangers, R (K/kW) and equivalent heat leak parameter of a chiller, O° dạy (kW) These parameters can be derived from the 1“ and 2™ Law of Thermodynamic as shown in Equation 5.5 of Cool Thermodynamic by Jeffrey Gordon and KC Ng BỊ,
Trang 24T* 7 HS leak Te 7h R ni i cond CoP Quvap Tend Qevap Toon cop Equation (1) Where,
T:"_, is evaporator heat exchanger coolant inlet temperature evap T7; 1S condenser heat exchanger coolant inlet temperature «
AS int is rate of internal entropy generation
leak = s :
Q is equivalent heat leak parameter of a chiller R is effective thermal resistance of heat exchangers Q.,ay is evaporator heat removal rate
Rearranging,
in in leak ‘in in
Te |- 1 |e in, Eat ns Teg), RO E 1 Su cond COP] Quy = Ting T5 GĨP Equation (2) in leak ¢apin in Tay ASu | | oe al all -1
And, | 1 | | uy Toon ‘evap
Trang 25G71” C=O leak * evap
em ~ Sevap pin opin cond — * evap Equation (5) [ Peo int Equation (6)
These 3 parameters of the simple thermodynamic model can be regressed either by intrusive measurement of the refrigerant parameters, or by non intrusive measurement at the coolants sides In order not to interrupt the chiller internal refrigerant circuit and for better overall system performance monitoring, the latter will be deployed with systematic data processing methodology With a handful set of field performance data, they could be analyzed or regressed and worked towards determining the predicted chiller COP Details of such analysis will be discussed in sections 4.2 and 4.3
Trang 26Chapter 3 Equipment and Instrument Detail 3.7 Chiller details
The chiller plant at HP comprises 5 chillers They are Trane 3-stage centrifugal chillers, direct drive, water cooled with R123 refrigerant, equipped with inter-stage economizer These models are the CVHE 590 and CVHG 670 for 2 x 500 RT and 3 x 850 RT chillers, respectively Figure 10 shows the typical major components diagram These chillers are designed and tested according to ARI Standard 550 — 92
Oil Tank and
Refigerant Pump
Figure 10: General CVHE and CVHG Unit Components
Trang 28For 3-stage chiller with 2-stage economizer and refrigerant orifice system as depicted in Figure 11, liquid refrigerant leaving the condenser at state point 6 flows through the orifice plat A and enters the high pressure side of the economizer The purpose of this orifice and economizer is to pre-flash a small amount refrigerant at an intermediate pressure, P1 P1 is between evaporator and condenser pressures Pre-flashing some liquid refrigerant cools the remaining liquid to state point 7 The cooler refrigerant gas from this high pressure side mixed with the entering gas in the third stage compressor to lower the enthalpy, near state point 4 Refrigerant leaving the first stage economizer flows through the second orifice B and enters the second stage economizer Some refrigerant is pre- flashed at intermediate pressure P2 Pre-flashing the liquid refrigerant cools the remaining liquid to state point 8 The cooler refrigerant gas from this low pressure side mixed with the entering gas in the second stage compressor to lower the enthalpy, near state point 3 To complete the operating cycle, liquid refrigerant leaving the economizer at state point 8 flows through a third orifice C where refrigerant pressure and temperature are reduced to evaporator conditions at state point 1
Another benefit of flashing refrigerant is to increase the total evaporator refrigeration effect from RE’ to RE this is believe to provide up to 7% energy saving compared to chiller without economizer
Trang 29850 RT Performance Index 4.200 4.000 =— 0800 “ œ © 04600 - = 0.400 SS =
ý = 3.4526E-1ĐẺ - 1.2082E-131Ẻ + 1.7226E-10x' -1.2878E-07¢ +
0.200 5.3691E-05ý - 1.2360E-02x+ 1.8624E+00
0.000 RẺ=9.9086E-01
0 200 400 600 800 1000
Load (RT)
Figure 12: Manufacture Performance Data of 850 RT Chiller, kW/RT vs Load
Figure 12 shows the rated chiller efficiency at part load Appendix B shows the detail 850 and 500 RT chiller performance as designed by the original equipment manufacturer, embedded in MS Excel file
3.2 Field instrument and data acquisition
All the necessary data for the chiller monitoring, data acquisition and
analysis are based on non intrusive, continuous online measurement The
measured data are coolants flow rates and temperatures; and in power input to the
compressor
1 For flow measurements, the E+H electromagnetic flow sensors are installed The model of the instrument is PROline Promag 10W, consists
Trang 30of various sizes to measure the CHW and CW flow for the 5 chillers The
system accuracy is +0.5%
For temperature measurements, the E+H TR10 PT100 temperature sensors
are installed, to measure CHWS/R and CWS/R temperatures for the 5
chillers The sensor accuracy is Class 1/3 DIN B
For power measurements, the Schneider Electric power transducer and PM500 Merlin Gerin power meter is used The reported uncertainty of the sensor is within +1% of the value measured
Data acquisition from field instruments is achieved via an Allen Bradley PLC, model SLC-505 All the analog signal inputs from field instruments such as RTD for temperature measurement, 4-20mA for flow and power measurement are being connected to the respective I/O cards in the PLC which will then be routed via Hub Switch by 2 pair fiber optic cables to the respective servers which is installed with Wonderware Intouch software version 8 SP 1, for data storage and retrieval The system has a sampling capability of every 10 sec interval Figure 13 depicts a typical the SCADA graphic used for the monitoring and data acquisition of the chiller facility
Trang 323.3 Measurement uncertainty
Based on manufacturers’ data and field verification on the measurement instrument, the root mean square (RMS) error for COP monitoring is estimated to
be 3.1% and 16.23%, respectively However, as the site flow verification is based
on pump curve which is subjected of ambiguity, and that all the new flow meters are supplied with calibration cert, hence the flow rate accuracy is to be based on
manufacturer’s data of +0.5% As a result, based on the combination of site
verification (for temperature and power) and manufacturer data (for flow), the maximum total uncertainty for determining COP experimentally based on rms error is about 6.22% Appendix C shows the calculation for the change COPs
Trang 33Chapter 4 Fault Scenarios and Data Processing
4.1 Fault scenarios
Followings are the predetermined fault scenarios for the analysis:
Chiller 1 — before and after condenser tube cleaning on 1“ and 15" May 2007 Chiller 2 — before and after condenser tube cleaning on 30" Apr and 15" May 2007 Chiller 1 — at preset condenser water supply temperature at 29.0, 29.5 and 30.0°C on 12", 6" and 13" June 2007 Chiller 2 — at preset condenser water supply temperature at 29.0, 29.5 and 30.0°C on 12", 6" and 13" June 2007 Chiller 4 — at preset condenser water supply temperature at 29.0, 29.5 and 30.0°C on 12", 6" and 13" June 2007 Chiller 5 — at preset condenser water supply temperature at 29.0, 29.5 and 30.0°C on 12", 6" and 13" June 2007
Chiller 4 — at low and normal condenser water flow-rate at 2000 and 2400 USgpm on 13" and 15"" May 2007
Chiller 5 — at low and normal condenser water flow-rate at 2000 and 2400 USgpm on 13" and 15" May 2007
Scenarios 1 and 2 represent the condenser tubes fouling, whereas scenarios 3,
4, 5 and 6 represent high condenser coolant inlet temperature due to cooling tower under performance or ambient web bulb temperature, as a deterministic factor for cooling tower performance, exceeded the normal operating range
Trang 34And finally, scenarios 7 and 8 provide the fault scenarios due to condenser coolant flow deficiency
Followings are some observations related to the operation for the scenarios: i) ii) iii) iv)
Condenser water supply temperature is very much depending on the ambient
wet bulb temperature It may fluctuate within +0.5°C of the preset temperature Cooling tower variable speed drive will regulate the speed of the fan to meet the requisite set point
Cooling load is subjected to the operational demand; hence it’s not being controlled i.e chilled water return temperature will fluctuate according to load demand
Coolant flow rates are the nominal Actual flow rates are at the most 6% deviate from these nominal flow rates
All relevant data sets captured as close as possible to the scenarios, in term of timing, for better representation and to prevent other time lapse interferences
4.2 Steady state data filtration Woi
this
Date analysis is aided by the ActiveFactory suite version 8.0.3 in the nderware which is set to sample each data at every 10 sec interval Base on sampling time, with the 7 chiller parameters being measured by the non- intrusive method such as power input (Pin); condenser water supply and return
tem;
T),
peratures (CWS/R T); chilled water supply and return temperature (CHWS/R
chilled and condenser water flow rate (CHW/CW FP), a total of 60,780 data
Trang 35MS Excel, and averaged for every 1 minute to bring down the number of data to 10,080 data for easy handling Appendix D shows typical data extracted from the excel file
In order to distinguish the steady states from the transients, a Macro programming is being developed to determine the steady state operating parameters for power input (Pin); chilled water return temperature (CHWR T); chilled water supply temperature (CHWS 1); chilled water flow-rate (CHW F) The predetermined boundary parameters for steady state are Pj, and CHWR T
Initial boundary conditions of +0.5% for Pi, and +0.1°C for CHWR T in 15
minutes of steady state yielded minimum steady state data A more loosen parameters was agreed upon, and changed to +2% and +0.15°C for Pin and CHWR T, respectively, and with the 5 minutes steady state, these boundary limits provide more data for assertion But this could be at the expense of data representation and accuracy in the later stage of development work Appendix E shows the macro programming details for the steady state filtering
4.3 Determining COP prediction and irreversible parameters
From the steady state data, the COP predicted and critical irreversibility parameters namely internal entropy production, AS,,, (kW/K); effective thermal resistance of heat exchangers, R (K/kW) and equivalent heat leak parameter of a chiller, oF (kW) is being determined by Microsoft Excel Solver For such
analysis, Equation 1 is rearranged into smaller terms as follows:
rin
Left term: Y=? wl -1
Trang 36Tova Right terms: X,=—“ , X, = evap Te cond’ Hence, ¥Y =AS,,X, + OR" eqv X, + RX, Equation 7
With 1/COP eq according to Equation 3, Solver is performed to obtain the
minimum root mean square (RMS) error for COP measure and predicted by varying AS,,, , R and os parameters with other standard Solver options such as 100 iterations for maximum of 100 seconds with precision of 0.000001 and tolerance of 5% A typical of Solver for one of the scenarios is embedded in Appendix F
With COPprea parameter, charts for 1/COPprea VS 1/Qevap and COP
Predicted vs COP Measured for various scenarios are plotted Each of these charts will be discussed in details in Chapter 5
From these critical irreversibility parameters, a multiple linear regression is performed on those steady state data via Equation 7 to obtain the 1 and 2-mean standard error for the irreversibility parameters as in Appendix F These values are important to provide the thresholds to the nominal values of the irreversibility
parameters to avoid or minimize false detection If 1-mean standard error is used,
then faults are likely to occur at approximately 68% confidence level if the any of the parameters exceeded the normal critical parameters thresholds Similarly, if based on 2-mean standard error, then the faults may happen at approximately 95% confidence level if any of the parameters exceeded the normal critical parameters
limits E1,
Trang 37Chapter 5 Data Analysis and Discussion
5.1 Condenser tubes fouling
Both Chiller 1 and 2 demonstrate better overall performance upon condenser tube cleaning The COP has increased about 1 to 5% depending on the chiller loading The higher the loading, the bigger the gap in COP, as can be seen from the gradient of the straight line curves in Figure 14
The COP model prediction RMS error percentage for Chiller 1 before and after tube cleaning is 0.3816 and 0.2424%, respectively; whereas for Chiller 2 is 0.5211 and 0.6888%, respectively The low RMS error percentages as compared to experimental error of 6.22% signify the representation of the 3 critical irreversibility parameters towards the COP prediction Table 2 provides the values of each parameter Table 2: Chiller 1 and 2 Critical Irreversibility Parameters Before/After Tubes Cleaning R(KIKW)|Qleak(KW}| AS (KWIK)| RMSE | #of data} %4RMSE| Qevap(RT) COP Ave | Wve
CH pre cleaning (1st May 2007) (1.00786) 0.9990484} 0.3718755} 0019623) 20) 0.381604) 37801) i2
CH 1 post cleaning 15h May 2007) (100664) 0.999867} 0.392125} 0.013243} 26} 0.242432) MAI 56
Changes wi pre cleaning Ahh} HH SH 10h) 5H
(CH2 pre ceaning (30th Apr 2007) 0.00831} 0.999835} 0.317044) 0.029765) 19} 0.521054 445.34) S71
ICH 2 post cleaning (15th May 2007) (0.00801) 0.999837 0.3087126) 0.039802) 21} 0.688804) 423.73) «579
(Changes wr pre cleaning 4h Đ 3 jj !h
Trang 39
However, even though Chiller 1’s heat exchangers thermal resistance can give an indication of chiller condenser tube fouling with 68% confidence level, as shown in Figure 15, this may be due to the increase in chiller loading In comparison, Chiller 2’s heat exchangers thermal resistance does not have any significant prediction to the tube fouling as shown in Figure 15 This could be due to the fact that Chiller 2 has slightly higher condenser coolant flow rate at about 1300 USgpm as compared to Chiller 1’s of about 1250 USgpm and that the original fouling may not be that significant
Nonetheless, it is certain that both chillers’ heat exchangers demonstrate a reduction in thermal resistance upon tube cleaning Also, as anticipated, the chiller condenser tube fouling does not reflected in the internal entropy generation rather this parameter is mainly a function of the chiller loading and other operating
conditions, as shown in Figure 16 and Table 2 And, the chiller system heat leak is