I David Taylor Research CenterBethesda, MD 20084-5000

CY DTRC-PAS--90/17 June 1990

Propulsion and Auxiliary Systems Department* Research & Development Report

A Condition Based Maintenance Monitoring Systemfor Naval Shipboard MachinerybyChristopher P. NemarichWayne W. Boblitt

E David W. Harrell

! .

~00 E DCTE•h ELECTE> AUG2800

C31Approved for public. release; distribution is unlimited.

,, . , ,

CODE 011 DIRECTOR OF TECHNOLOGY, PLANS AND ASSESSMENT

12 SHIP SYSTEMS INTEGRATION DEPARTMENT

14 SHIP ELECTROMAGNETIC SIGNATURES DEPARTMENT

15 SHIP HYDIOMECHANICS DEPARTMENT

16 AVIATION DEPARTMEN'

17 SHIP STRUCTURES AND PROTECTION DEPARTMENT

18 COMPUTATION, MATHEMATICS & LOGISTICS DEPARTMENT

19 SHIP ACOUSTICS DEPARTMENT

27 PROPULSION AND AUXILIARY SYSTEMS DEPARTMENT

28 SHIP MATERIALS ENGINEERNG DEPARTMENT

DTRC ISSUES THREE TYPES OF REPORTS:

1. DTRC reports, a formal series, contain information of permanent technical value.They carry a consecutive numerical identification regardless of their classification or theoriginating department.

2. Departmental reports, a semiformal series, contain information of a preliminary,temporary, or proprietary nature or of limnited interest or significance. They carry adepartmental alphanurnerical idertiiicat:on.

3. Technical memoranda, an informal series, contain technical documentation oflimited use and interest. They are primarily working papers intended for internal use. Theycarry an identifying number which indicates their type and the numerical code of theoriginating department. Any distribution outside DTRC must be approved by the head ofthe originating department on a case-by-case basis.

NDW-DTNSRDO5560/51 (Rev 2-U8)

UNCLASSIFIED

SIECURITY CLASSIFICATION OF THIS PAGE

REPORT DOCUMENTATION PAGEle. RLPORI SF-CURIIY CLASSIFICATION lb. RESTRICTIE MARKINGS

lJnclIsQsilied None2a. SECURITY CLASSIFICATION AUTHORITY 3 DISTRIBUT ION/AVAILABILITY OF REPORT

2BE. DECLASSIFICATIONI)OOWNORADING SCHEDULE Approved for public release; distribution is unlimited.

4. PERFORMING ORGANIZATION REPORT NUMBER(S) 5. MONITORING ORGANIZATION REPORT NUMBER(S)

1)TRC-PASý-90/l7

Ea NAME OF PERFORMING OPGAtIIZATION 6b. OFFICL SYMBOL 7a NAME OF MONITORING ORGANIZAF ION(/t apploasol)

David Taylor Research Center Code 2758

61. ADDRESS (Cry, State. arid ZIP C-de) 7b. ADDRESS (Cay. State, end ZIP Code)

Bethesda, MID 2008Z4-5000

Sa NAME OF: FUNITING/SPONSORING 8bB. OFFIC' SYMBOL 8. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBERORGANIZATION [ (it app~labl.)

O ffice of Naval Technology OT226

Sc. ADDRESS (C¾Stare. and ZIP Coda) 10 SOURCE OF FI)NDING NUMBERS

PROGRAM PROJECT TASK WORK UNT1ELEMENT NO. NO. NO. ACCESSýION NO

Washington, D.C. 22217- 5000623NR 60 R -7( 1 -1

ii. TITLE (intrade Secu'ay Classfficanon)

A Condition Based Maintenance Monitoring System for Naval Shipboard Machinery

2a PERSONAL AUTHOR(S)

C"hrisronheT P Nemarich Wayne W Robhltt, David W. Harrill

13a TYPE OF REPORT 135 TME COVERED 14. DATE OF REPORT (YEAR. MONTH. DAYT T 5 AE ONTechnical FROM_ BE050T vl___TBO2B_ 19 arch 6 116. SUPPLEMENTARY NOTATION

Presented at the 44th meeting of the Mechanical Failures Prevention Group (MFT'PG). Api 2-5, 1990, Virginia Bezich, VA

1T. COSATI CODES 18. SUBJECT TERMS (Cortnr; on rov-e, If necessay aid identify by block nutiwrb)

FIELD GROUP -UIe_5_ - Automated machine montitoring. Expert systems, Machine health, Condition based mainwz-I., nance, Fault detection, Process central, Distributed monitoring, Fault diagnosis. Prognosis,

~. ASTRAT -~Shipboard machinery mtonitoring

19 B )TRCT r.-bnui on revers It necessary 'ki~en~tt by blodi number)

A demionstr-ation model of a Condition Based Maintenaiice (CBMI) nicatirorng, system has been developed and installed on a highpressure compressor. The CBM model is a distributed, micioprocessor-based system incorporating graphical mian/machine interface, ex-pert system and local area network software. The CBM model integrates commercially available hardware and software packages withsoftware, sensors and monitoring techniquies currently uinder development. The intent is to apply CBM broadly to naval shipboard Ma-chinery for fault detection, diagnosis and prognosis (DD&P). The knowledge gained by implemnenting increasingly, sophisticated systemswill be_ used to develop generic CBM guidelines for naval machinery and ship systemi designs. This report presents a detailed descriptionof the CBM monitoring system and discusses design issues. This report also presents future development plans for CBM.

2E. DISTRIBUTIONIAVAIL.ABILITY OF ABSTRACT 21. ABS5TRACT SECURITY CLASSIFICATION

12UNCL.ASIFIED/TNLIMITED 53 SAME AS RPT C3 D71 USERS Unclassified22a, NAME OF RESPONSIIBLE INDIVIDUAL 22b. 1ELE PHONE (IndmMk Area Coda) 22c. OFFICE SYMBOL

Christopher P. Nern aiich (301) 267-2163 Code 2758ý

DD FORM 1473. JUN 86 Previous editions are ob.,ouiev. SECURITY CLASSIFICATION OF I HIS PAGE

UNCLASSIFIED01 02--LF--O1 4-6602

CONTENTS

Page

Abstract ............................................................ .1

Introduction ........................................................ 1

Background ......................................................... 1

CBM System Concept ............................................... 2

CBM Demonstration M odel ........................................... 3

Graphic M an/M achine Interface ....................................... 5

Expert System Diagnosis .............................................. 8

Expert System Prognosis .............................................. 9

F uture P lans ............................ ........................... 9

C onclusion .......................................................... 9

R eferences .......................................................... 11

FIGURES

1. CBM system concept............................................ 3

2. CBM demonstration model ......... .................................. 4

3. A nalog display screen ................................................ 6

4. A iu Spy in I: s el ll .................................................... 6

5. Oil system screen................................................7

6. W ater system screen ... ............................................. 7

7. Example data display screen ........................................... 8

Acce••ion ForNTTS 'qRA&I

Oiiam a unC &Unneunted 0• [

Justtficat1on

Distribution/-

Avallahility CodesD~s- Vail and/°r-

Dist Spoaial.

DTRC-PAS-90/17 iii

/ .' '4I

ABSTRACT

A demonstration model of a Condition Based Maintenance (CBM) monitoring sys-tern has been developed and installed on a high pressure ai- compressor. The CBM modelis a distributed, microprocessor-based system incorporating graphical man/machine in-terface, expert system and local area network soft'ware. The CBM model integratescommercially available hardware and softnware packages with software, sensors and mon-itoring techniques currently under development. The intent is to apply CBM broadly tonaval shipboard machinery for fault detection, diagnosis and prognosis (DD&P). Theknowledge gained by implementing increasingly sophisticated systems will be 'sed to de-velop generic CBM guidelines for naval machineiy' and ship syst- n d signs. This r.portpresents a dctailed description of the CBM monitoring systenm _ad discusses designissues. This report also presents future developmentplansfor CBM.

INTRODUCTION

One of the roles of the David Taylor Research Center (DTRC) is to conduct researchand development in the area of naval shipboard propulsion and auxiliary systems. DTRCalso develops sensors and controls for these systems. Because of this expertise, the Logis-tics Block Program manager tasked the Center to create a technology push in the area ofCondition Based Maintenance (CBM) monitoring. The expected benefits of CBM areimpruved maintenance procedures and scheduling, increased machinery operationalreadiness, and reduced logistics support cost.

As part of this effort, DTRC has developed a demonstration model of a CBM moni-toring system. This demonstration model integrates commercially available hardware andsoftWae With ,Cvhot,, ,.+UJiuoljcs. Tilhe m-'odel de..o.sat....s l1Id.(Ji1le'i 'IdIL . cicc-

tion, diagnosis and prognosis (DD&P) capabilities as applied to naval shipboard systems.The demonstration model makes use of new sensor types, programmable logic control,and industrial process control hardware and software. The model also includes local areanetworking, graphical man/machine interface and diagnostic software. Expert systemsoftware and the results from research in machinery modeling will be used for DD&P.

The near term goals of this program are to successively develop CBM systems forexisting machineries and integrate them to form larger systems. The long range goal is todevelop generic CBM guidelines for futuie machinery developments and naval ship sys-tem designs.

This paper discusses the CBM demonstration system and some issues involved withthe design of naval shipboard monitoring systems.

BACKGROUNDThe United States Navy now uses the Planned Maintenance System (PMS) for the

upkeep of shipboard system machinery. This involves the periodic overhauling of ma-chinery on a scheduled basis. 'Tie operation and maintenancc of propulsion and auxiliarysystems has steadily become more difficult, costly and labor intensive with the increasingdiversity and complexity of these systems. In this era of shrinking maintenance budgetsand skilled labor pool, the current means of maintaining and operating Navy stips hascome under attack.

DTRG-PAS-90/17

Efforts are being made to reduce the number of scheduled overhauls through theperiodic monitoring of shipboard machinery health and performance. Data obtained fromperiodic monitoring helps the ships' forces diagnose machinery problems, determine ma-chinery wear condition, and reduce maintenance procedures. "Tis approach, however,will not solve the Navy's operating and maintenance problems.

The solution is an integrated shipboard machinery monitoring and control systemthat includes CBM. The Condition Based Maintenance concept entnumpasses more thanjust periodic condition moritoring. CBM is an approach that incorporates all aspects ofautomated machinery failure detection, diagnosis and prognosis. It has been shown by atask group of the Advanced Testing Technology Committee 1I] that applying CBM toshipboaid hull, mechanical and electrical systems can produce the following benefits:

"* reduce maintenance induced failures - 50%

"* reduce maintenance actions - 35%

"* increase availability -- 20%

"* reduce inspection and repair hours - 20%

"9 reduce spare parts provisioning - 20%

"* reduce good parts removal - 10%

"* extend equipment life/overhaul cycle - 10%

CBM cian produce these benefits but it requires an effective resolution of the issues in-volved with automated shipboard monitoring. The Navy has determined that CBM is an"imperativc Characieristic" ,-n

CBM SYSTEM CONCEPr

An established Lpproach is ,o develop the top-down system concept identifying thepertinent design issues. Once this lia, been done, the system is implemented from the bot-tom-up [3]. This is the approach that DTRC is taking.

Sonic of the issues addressed in the design of the CBM system are the man machineinterface, distributed versus centralized processing and data base management, designflexibility and adaptability to shipboard environments.

rhe overall CBM shiphard system concept developnprd is shown in Fig. 1.hms sys

tem has four levels. Level I is the progranmmable logic controllers (PLC). The PLCsreside at the machinery and provide digital machine control and collect and digitize sen-sor data.

Tne industrialized PC/ATs at level 2 reside with the PLCs in the machinery spaces.They provide the operator at the graphical man/machine interface with machine controlthrough the PLCs . The operators have access to all information, alarms and control nec-essary to the operation of their machinery space.

Level 3 is a supervisory level consisting of PC/AT type workstations. These work-stations, located in the engineering spaces, contain the historical database for the shipmachinery and provide a high level of access to ship engineering system data and control.

2 DTRC-PAS-90/17

W - VAX/VMS or similarminicomputer system.

VMS Shipwide supervisorydata management

Microprocessor Workstations

hlI.I'ii," Pu. ( , GraphicalIN Man/Machine

InterfaceHistorical TrendingEthernet

Industrialized PC/ATS Graphical

NlMar/Machine

i i~.+I i i nterface

1 a ... .. .. Realtim e

Data CollectionRS-232 Communication

PLCs M ..i1 ]f . Lil

Machinery SystemsFig. 1. CBM system concept.

Level 4, the highest level, provides managerial access to each workstation and in-

dustrial PC/AT. This level consists of a central VAX/VMS system or equivalent. Itperfonis ship wide data management and control and other tasks required of the ship'smost senior officers. Other ship functions would tie into this level.

This is a general system concept. It addresses the issues of distributed control anddatabase management and can be applied easily to a ship environment. It is flexible and isnot limited to the technologies listed here. This is the framework from wilich the CBMsystem will evolve.

CBM DEMONSTRATION MODEL

A Worthington high pressure air compressor (IPAC) installed in the laboratory pro-vides a means of effectively demonstrating CBM on an actual shipboard auxiliary. TheHPAC is a motor-driven, oil-lubricated, water-cooled, four-cylinder air compressor.

DTRC-PAS-90/17 3

Because it is complex and has a variety of subsys.tems, the HPAC is ideally suited fordemonstrating CBM.

The CBM demonstration model is shown in Fig. 2. It consi:,ts of a pair of i386 PCIAT computers and a Gould 984 programmable logic controller with 32 channels of A/D.A commercially available process control software package by lutellution, Inc. calledFIX/DMACS (TM) is used for the system control, data namagement, and display. It sup-ports the communication with the PLC, the diagnosis and prognosis, software, and th,local area network (LAN) software.

The CBM model developed to date works as follows. Analog signals from iheHPAC sensor suite are digitized by the A/D modules in the Gould 984 PLC. The HPAC isoutfitted with the standard shipboard gages and controls .ugmented with easily addedtransducers for data acquisition purposes. In order to aid in the acceptance of the new dis-plays and controls provided by CBM, it is installed alongside the original HPAC gagesamd meters which have not been removed. The PLC provides limited microprocessor ca-pabilities for data messaging, buffering and control. It communicates with the indusnialPC/AT through the IBM realtime coprocessor.

The IBM realtime coprocessor is referred to as an Artic card. The Artic card, via itsRS-232 serial port, reads the bank of memory addresses corresponding to the varioustransducer data channels. It comprises one of the three nodes in the CBM model. Therealtime coprocessor executes the SCADA portion of FLX/DMACS software. SCADA isa mnemonic for Scan, Control, Alarm and Data Acquisition. The FIX/DMACS softwareis installed on both PC/ATs and provides communication between the PCs and the Atticcard. This forms a three node FIX/DMACS system.

PC/AT Industrial PC/AT386i387 316-13•/a3

Ethernet RS-232

Graphical Human Interface Interface between Polls PLC / -,

Historical Data Collection Realtime Coprocessor Maintains Database I'LLDiagnostic Software and other DMACS nodes

Graphical Human interface

"F " Sensors I

LII!_•/Signals

High Pressure Air Compressor

Fig. 2. CBM demonstration model.

4 DTAC-PAS-90/17

HPAC condition information is processed by FIXJDMACS and displayed oil theEGA screen for the ItPAC operator. This in-fornation is also available remotely at tileworkstation which is in another area of the building. The communication between thePC/AT at the I IPAC and tic remote workstation is via Ethernet and 3Comn hardware.

The remote workstation maintains and stores the I PAC database and provides theoperator with access to all information available to the machine operator. Messages cambe passed between these nodes.

Ultimately, tile worksta-ion will also rtui a rule-based expert system for tile detec-tion, diagnosis and prognosis of HPAC faults. Failure prediction algorithms are beingdeveloped for this. At the present time, only C-coded diagnostic rules run at this node.

GRAPHIC MAN/MACHINE INTERFACE

A crucial portion of any automated monitoring system is the nian/machinc interlace.The information flow from machine to man must be dealt with effectively to prevent in-fornation overload. Yet it must provide all the necessary data to the system operator. 'hisis one of the central issues to the demonstration CGM system.

A variety of graphic display screens have been designed for CBM. Figures 3, 4, 5,and 6 show the major HFAC system screens. Figure 3 is tihe ANALOG DISPLAY screen.This screen is the main display and shows the HPAC operator information in an easy toread graphical format. At this screen, air temperatures and pressures for each stage ofcompression are displayed. The values are displayed relative to their respective alarm andwarning states in a har graph format. A digital screen, not shown, is also avaiiable which"displayS the ViueS of"teu's ,,sINCri ii, a jimuei icai format. I nhe dieitaii screen shows allsensor data including the AC motor speed, current, voltage and power and rate .of oillubrication to the compression cylinders.

If any parameter exceeds either a warning or alarm threshold, the Machine Healthbox, located in the upper right corner of the display, changes color. Green is normal. Yet--low is warning. Red is alarm. The operator can reach any major screen by selecting anyof the large boxes at the bottom with the touch screen. The diagnostic box will changecolor to instruct the operator to go to the diagnostic screen to receive furter instructionsgenerated by the expert system.

The AIR SYSTEM screen is shown in Fig. 4. This screen provides a mimic of the'PACi air system. The color of the boxes at each stage of compression represent the con-

dition. Again, the operator can reach any otlier screen from here.

The OIL SYSTEM screen iL shown in Fig. 5. The -PAC oil system consists of apump for distributing oil to main bearing and crank assembly and an auxiliary oil distri-bution system for cylinder lubrication. The concept of this screen is similar to the others.This screen, however, incorporates data from a unique fiber optic drop counter sensor.This sensor monitors the rate of drop lubrication to the tIPAC cylinders from the auxilia-ry lube system.

The WATER SYSIEM screen, shown in Fig. 6, displays cooling water inlet tempex-ature and discharge temperature. The pressure at the condensate drain accumulator is alsoshown.

DTRC-PAS-90/17 5

Analog DisplayHlcc th L

PRESSURE Thr0r, INPUI TEIP, OUTPUT

1 2 3 4 1 2 3 4 1 2 3 4 TACJH

R~EDai'

YELLOW

GREEN

YELLOW

I H OIL L.ý I ER I N14 LOG DI GIT L- DIAGNO0TIC

Fig. 3. Analog display screen.

MochnmeAir Systemn LEoltH

.11 5 4 325 '33 1 6 2• 1 035 11

ۥ[ OI WNTER IAINiTLOG DIGITAL DIAlGNOSTI C

ivi

EDCII DI

FIg. 4. Air system screen.

6 DTRC--PAS-90 17

Oil System o

Lur, 1 30 r-

LubE 0,.1 Lubt

RIF 01L lJ", 1 Ek ilNO LOC D"I C.I 10 L D I :.N :I'S 1 1

Fig. 5. Oil system screen.

79wa ter Sstem Heoo Ih

PS T1'j.

U-T,

I----' IV

432?_i I_ __ _ _ 2

A"UýF, I Tt

CIL ACC ~ r~L M 1

RIR 01 L 4I-0TER ANAO DICITA.L DIAGNOSTIC

FIg. 6. Water system screen.

Dl'RC-PAS-90/17 7

I~ccnu IOycrcx~r

From thie rmmajor screens,* a graphical limec history of a sciisý'r value may bie dis-played. The user simply touches the screen eknicnt associated with that setnsor. The data.is displayed as in Fig. 7.

EIXPERT SYSTEM DIAGNOSIS

The CJ3M model is designed to use at-. expert systCni software package. NF'~EPFRT(TMI). to anialyze sensor dmataamd provide a diagnosis of'HPAC faults. Vic inicution is fortinc expert sy'stemi to obtaitn sensor data fiomi the SCADA node dat~ibasc. FIX/MACSprovides access to the database via C function calls which are executable on any FIX/DMIACS node. With the appropriate knowledge base, NEXPERT (-an analyt-e tile dataand commut-nicate its results back to FIXJDMACS via C funtction calls.

Thiis plan was postponied because of problems getting F1X/D.MACS and N EXPERTto coexist on the samec DOS node due to systernm tneorv constraints. A temporary solu-tion hias been adpe.rugoicnles are hadCdn ihna C proglram. The Cprogram- replaces the expert systemi with 24 hiard coded rules arraniged ini order of de.-creasnig severity. These rules are developed fromn the I-WAC systeni niatial andc quipmnent experts. Once a r-ule fites, indicating a fault, a diagnostic numiber utiquely inl-dicating the fault is immnediately sent lo FIX/DMACS for display on the screenl. The colorof the diagnmostic box Is controlled by this nuniber. A yellowv or red dilagnostic box indi-cates the fault severity and directs the operator to the diagnostic screen. Even thoughmiore than one fault may be present, only the most severe is reported.

T 10 Mrac[- i pF1 emper.3ture F rst StI gf- Ju Ipul

RED

425

YkELLOW-

338~ GREEN329YELLOW4

2.79

0eon j'O 140 1'0 16 '-0-~140 8 0 0 5 40 30 20 10

A 1L PVJ TER ANALOG DIGITAL DIAGCNOSTIC

Fig. 7. Example data display screen.

DTrRG-PAS-90/1 7

The C program provides piecewise linear scaling of data for the ANALOG DIS-PLAY. and controls the color of the MACHINE HEALTH box and major screen boxes oneach display. Future CBM systems will use an expert system rather than hard coded rulesin order to support larger rule bases.

EXPERT SYSTEM PROGNOSIS

The final element in the CBM model is the prognosis capability. Prognosis projectsfuture machine health based on current and past condition. The expert system softwarewill be extended to provide prognosis. It will use a knowledge base developed for the ma-chine along with a probabilistic model of the HPAC to provide predictions of futuremachine health. Probabilities will be assigned to these predictions to give the operator anindication of the degree of confidence with which the prediction is made. The develop-ment of probabilistic machine models are crucial in order to provide more than a limitedprognosis capability.

Work is being done at Columbia University on the method of state space machinerymodels. Data collected from the machine will be used to establish normal and abnormalstate spaces. Algorithms will be developed to construct a multidimensional state spacemodel of the H-AC, Although beyond the scope of this paper, this method has the poten-tial of providing sophisticated prognostics.

FUTURE PLANS

The plans for CBM are to expand the present system to include the extended diag-nosis and the prognosis capabilities discussed. Work will continue on the integration ofthe, prcess¢ contro~l an~d ex,. system softare. GEM may be extended to other operatingsystems with greater multitasking capabilities and larger memory models in order to ac-complish this.

CBM will use the knowledge gained in developing the diagnosis and prognosis fea-hires to expand and refine the HPAC sensor suite. Sensor technologies will be developedas necessary for Ihe detection of machinery health problems. Finally, CBM will be con-tinually refined and expanded to include other shipboard machineries.

CONCLUSION

The CBM demonstration model will provide the Navy with the vehicle to improveits maintenance procedures and reduce costs. The challenge is to integrate CEM into thenaval shipboard env:ronment by using the knowledge gained from this effort to generatethe guidelines for future machinery development and ship system designs.

DTRC-PAS-90/17 9

REFERENCES

1. "Industry/Joint Services ATE Project," Final Report, Advanced Testing Thchnol-ogy Cornmittee, Task Group I d, Electronic Test of Non-Electronic Equipment,Revision C (1 Apr 1979).

2. Laird, W.C., "U.S. Navy Testing/Diagnostic RDT&E Program Plan," NOSCTechnical Document 1646, Naval Ocean Systems Center, San Diego, CA.92152-5000 (Sep 1989).

3. "Human Intcgrated Manufacturing in CIM Applications," Chilton's IC&S, Vol.62, No. 11, pp. 67-70 (Nov 1989).

DTRC-PAS-90/17 11

INITIAL DISTRIBUTION

Copies CENTER DISTRIBUTION

2 ONT Copies Code Name

CMDR A. Bavier 2 0116 R. Brengs

Code 226 1 27 L. Argiro

Wash., D.C. 22217-5000 1 272 J Ward

12 DTIC 1 275 F. Walker10 272Z G. Nickerson

10 2722 B. Boblitt

1 2753 G. Garduno

5 2753 D. Harrell

1 2758 E. Petnsko

15 2758 C. Nemanch

1 3411 M. Knox

1 522.1 TIC (C)

1 522.2 TIC (A)

1 5231 Office Services

DTRC-PAS-90/17 13