You are on page 1of 6

Copyright © IFAC 10t h Trienn ial World Con gress. 1.2- 3 MI N I NG.

MI N ERA L AND
Mun ich . FRG. 1987 T\! ETAL PROCESS I NG: SPEC IA L
APPLICATI ONS

CEMEXPERT: AN EXPERT SYSTEM FOR THE


CEMENT GRINDING PROCESS
L. Keviczky*, Cs. Banyasz*, R. Haber* and M. Hilger**
" Cvlllp"ln and A lllvma livn I lISlilll le. H ll ngarian Arad"IIIY '1' Seim cl'S.
H -1502 . Bllda/JeSI, Kmde 11 13-17. H llnKlII)'
**Cmlml Rn Nll'ch lllld D('sig n I lIslilllle for Silicllle I ndl/.lll)·. H -1300. Budap,'sl.
Becsi ul 126/1 28. HllllgW)'

Abstract. Grinding i n the cement industry is a very energy-exacting


process, therefore it is essential that these systems should operate
yith the highest possible effiCiency. The CEtEXPERT grinding expert
system assist s the operator (and the t!xp!t1ring engineer) to set the
technological parameters to opt imal work i ng pOints, whi le working
under variable circumstances . The system consists of several
subsyst ems wh i ch are placed on three h i erarch i ca I layers, as: p I ant
operator layer, industrial expert layer and academical expert layer .

Keywords . Cement grinding process, neY control technologiesje xpert


systems, kna...ledge based systems.

IHTROOUCTIOH To overcome these d i fficulties Ye initiated to


develop an expert system yith a decision
Gr i nding is a very energy-exacting process . support i ng or operator assisting capabil i ty on
Almost ~O% of the energy consumpt ion in the the plant control level and with an efficient
cement production is appl ied for grinding. It is exp I orat ory feat ure for process des i gners . Our
essent i a I that the gr i nd I ng system shou I d paper deals with the cOllllleted CEMEXPERT system.
operate yith the highest possible efficiency, as
even a 5IllQ11 improvement of the efficiency of
the process may lead to a considerable saving in DEUELOMHT OF CEtEXPERT
cost and consumption.
The COIllluter and Automation Inst i tute , Hungarian
Dur i ng operat ion the grindabi Ity of the ra... Ac ademy of Sciences (CAI) and the Central
material is changing, the balls in the mi II are Aesearch and DeSign Inst itute for Si I icate
attrited, etc . , consequently the static and Industry ( CRDIS!) together aimed to develop on
dynamic features of the mill - separator system expert syst em to decrease the spec if i c energy
al~o change . Accordingly, by optimizing the consumpt ion of the grinding eQuipments i n the
technological pol icy it is not enough to create cement i ndust ry . The task 0 f t he expert syst em
opt ima I cond i t ions once - e . g . for an average is to process the measured data, to analyse the
grindabi I ity - but a repeated adjustment has to operat ing situat "ons and to support the
be ensured . operators ' decis ions . Furthermore it helps the
exploratory york of i ndustrial engineer i ng
The theory of grinding is most Iy conSidered as experts of grinding accl.l1l.llating their long
an emp i rical science . The operat ions of the experimental kna...ledge .
gr i nd i ng t echno I og i ca I un its are QU i t e COllllI e x,
inspite of the i r relatively s imple physical 1. The control engineering fundamentals of
arrangements and the physical phenomena . Only gr i nd i ng systems Yere s\.l'llllClr i zed in Kev i czky
very feY behaviours can be explaned and and coworkers ( 1985) , based on about a
descr i bed by exact re I at i onsh i ps and form; I as . ten-year interdiscipl inary research work .
The experiences and process responses ( stat i C 2. Based on this theoretical research and on the
and dynamic evaluat ions) accl.l1l.llated during ana lyses 0 f operat i ng measurement sin cement
normo I and spec i a I operat i ng c i rcumst ances are i ndustrial grinding eQuipments, CADISI
very import ant for bot h p I ant cont r o I operat ors elaborated and classif ied 011 the
and process deSigners. Therefore a Ye l l tra i ned physico-chem ical and empirical re lationships
"if ft!r wit h long exper i ence is most I y higher ~ich are necessary to an efficient
evaluated than a sophisticated control system. accOllllI i s!'ment 0 f the tasks to be so I ved by

15 7
the expert system. decimols will be displayed) . Two forms of
3. CAI specified the har~re and basic softwore representation ore used: tabular and graphic. In
requirements with special regard to data graphic mode .....e distinguish the representation
hand ling, mon-mcch i ne cClfllrun i cat i on and of variables (functions) and schemes. The latter
graphic 1/0. one means a graphic scheme of the technology by
4. The system design of CEMEXPERT wos the result indicating the measured/calculated values (e.g.
of a team work, which contained the mot er i 0 I f IO'w'). Most 0 f the resu Its are given in
subsystems, the algorittmical blocks and graph i c represent at ion, at first appearance of
their connection with data bases, console the functions the type of axes (linear or
communication and graphic representation Iogar i t tm i c) i s fixed and the sca ling is
(cant rary tot he usua I syst em des i gns in the automctic . Then the user con change types and
field of process control, which contain all scales of both axes by interactive cClfllrunication
the detai Is in bit-/evel::». on the graphical screen.
5. The syst em was acc~ I i shed by t he end 0 f
1985. The first t .....o systems were ilTl' Iemented From any displayed result (including drO'w'ings) a
in t he CA I and CRD I SI. In spr i ng 1986 dat a hard copy can be mode whose size (f ormot Ai)
measured in the Hejocsaba Cement Factory he Ips t he easy document at ion. The a I phanumer i c
(Hungary) .....ere evaluated, then the system was tables as well os the graphic pictures can be
implemented on the spot. stored for further processing ( to display later
or to edit into a report). A more detailed
descr i pt ion 0 f the conso Ie corrfT~.m i cat i on con be
HARDWARE CONFIGURATION found in Almasy and co.....orkers (1986).

The CBMEXPERT system is realized on a Hungarian


mode persona I comput er M06X. Th i s c~ut er is a CHARACTERISTICS OF THE CEMEKPERT
CPft1 compatible machine . . . ith a medium resolution SYSTEM
graphic opt ion and ..... ith a double floppy unit .
(An add i t i ono I RAM disc opt i on con be used to The CEMEXPERT system has numerous features which
speed up the system.) A letter qual ity dot distinguish it from other expert systems.
motrix printer is used for hard copy needs. The t rad i t i ona I expert syst ems cons i st 0 f , in
general, a kno..... ledge based port and a collection
The se Iect ion 0 f t he above HUJ .....as mode 011 the of logical rules. Such are e.g . the medical
bas i s 0 fit s 10'w' pr i ce, . . . i de ava i I ob i lit y, fast d i agnost i c syst ems in wh i ch knO'ft'Tl s!,lllPt oms 0 f
and good service possibi I it ies provided by the diseases ore stored thus the c~uter helps
producer. The good software support and the recognizing the disease in a given case. The
integrated graphic capabi I it ies were also key CEMEXPERT syst em d i ff ers, first 0 fa", from a
points in the selection. usua I expert syst em in that it does not cont a i n
a co II ect ion 0 flog i ca I ru Ies but i nc Iudes
process models, parameter estimators, simulators
SOFTUJARE EliU I AOtHNT and predictors . . . ith algoritl-rnic calculat ions and
assists the user (operator or e~plorer) . . . ith
Ninety percent of the system is written in dedicated paradigms making the best decision
FORTRAN (being easi Iy portable to other machine poss i b le . These ca Icu Iat ions i nc Iude sc i ent i f i c
or to other language), e~cept the graphic kernel and statistic analyses of present and old-,
system and the extremely ::>aie data base system, measured and ca Icu Iat ed dat a, as we II as
which are written in assembler. technical calculations and graphic
representation of the results . The user can
a I t er the operat i ng cond i t ion 0 f t he mi I I in the
MAN-MACHINE COMMUNICATION direction of the optimal production . Such expert
system is col led deci sion supporting or operator
The CEMEXPERT system has a user friendly assist ing system (Krigman, 1985). Essent ial
mon-moch i ne comnun I cat ion. Furt hermore t he user modules of this system are the simulators . . .hich
does not need any computer knowledge on the t .....o help to test certain technological alterat ions
lower layers, on Iy quest ions posed by the It'i t hout the don gel' and expense 0 f i nt erven i ng
moch i ne have to be ans.....ered. I n genera I the into the actual technology (8ristol, lC;B5),
anS'W'e1' means a se Iect i on from a fu II screen menu ensuring a very efficient exploratory process
or typing some data . The consol communication is deSign .
i nt eract i ve, t he cursor can be dr i ven in the
full screen display to the chosen data to be The CEMEXPERT system IS not a program on Iy, it
a It ered by t he correspond i ng but tons. The dot 0 consists of subsystems whose madules are on
of the display cClfllrunication are stol'ed so they hierarchical layers (see Fig . 1) , i . e. plant
can be displayed during the next run of the operator layer, industrial expert layer,
program. Both input and output data are handled academical expert layer .
in free fONTlClt . In case of output data this
means that the pre fixed f ormct wi I I be 0 I t ered The lower layer i nc Iudes programs t hat are used
according to the order of the mJ'nber (e.g . less in t he rout i ne p Iant operat or work in a cement
Cell1expen

factory. The middle layer is the level of the by the user


techno I og i ca I ana I yses and the adv i sory o Technology specific simulation
evaluations ~ich help the industrial experts in - Steady-state and dynamiC material flow
the i r research work. At last, the t h i rd layer models of a closed-circuit bal I mi I I
includes general purpose programs independent - Dynami c part i c I e size d i str i but i on mode I
from the technology, ...-hich help to ,olve ta,ks of batch grinding
not covered by the technology-oriented - DynamiC particle size model of a
simulation and ident ification programs of the closed-circuit mi II
second layer, and to i nc I ude new t heor i es and o Analysis of a continuous homogenizing system
methods. The knowledge necessary to use the (experimental identification of silo
programs of the individual levels grows parameters and Qual ification of the si 10 by
upwrds, so cert a i n programs 0 f the p I ant I eve I qual ity patterns)
can be used without any computational, technical o Particle size distribution analysis (fitting
or ,i I icate industrial pr act ice. The middle di fferent analytical particle size
level assists the knowledge acquisition already d i st r i but ion f unct ions tot he measured si (""'"
in the branch engineering, the research and passing values)
adVisory act ivities with special programs for o DeteNllination of the average resident time
the given technology. The simulat ion and of the material in the mi II (measurements
i dent i f i cat i on programs 0 f t he upper Ieve I are are performed by isotop)
comp I et e Iy genera I. S(1 t hey can be app lied very 3. Academical expert layer
eas i 1Y in any t echno 1og i ca 1 t ask if the o Process identification (estimat ion of the
engineering knowledge is given. The alphanumeric parameters of a single or multivariable,
and graph i c report ed it or ond t he genera I dat a st at i c or dynami c, I i near or non I i near
management belong also to this level.(See the model)
logical scheme in Fig. 2) o Process simulator (simulat ion of a system
constructed from function blocks)
o Graphic report editor (editing alpham.neric
PARTS AND FUNCTIONS OF THE texts and graphic pictures stored in fi les
CEMEXPERT SYSTEM to make a report)
o Data base (read/write data into/from disc
1. Plant operator layer files, their modification, listing and
o Procession of plant operat ion data graphic display)
Init ial ization of data base For handl ing the plant operat ion data a general
- Read/write and modification of data purpose dot a management sys t em ....as used. A more
- List of plant operation data detai led descript ion can be found in Almasy and
- Graphic representation of data co...orkers (1956) .
o Technological calculations
- Evaluat ion of the mi II system
(calculation of the internal material DATA COMMUNICATION BETWEEN THE
flows) MODULES OF CEMEXPERT
- Ana Iys i s 0f the separat or syst em
(evaluation of the particle size Three t1l8X persona I comput er con f i gurat i on are
distribution, deteNllination of the i nst a II ed: ane in t he fact ory (i n He j ocsaba) ,
particle size belonging to the sharp one in CRDISI and one in CAI , enabl ing the
separat ion, ca Icu Iat ion 0 f the e ff i c i ency accomp I i shment 0 f tasks 0 f d i f f erent I eve I and
of the separat ion) the cooperation. The identical har~~re and
- Ca Icu Iat i on of the mi 1I diagrams (energy soft ...are ensure the simple transport of the data
utilization factor, specific energy on floppy discs. The analysis of the data
consumpt ion , newly produced surface; measured regu I~r Iy in the fact ory can be
samp Ies from t he ground mat er i a I are per f ONned on t ne spot as ....e I I as t he research
taken along the rnl I I axis) and des i gn (t echno logy spec if i c i dent i f i cat ion,
- Ca Icu Iat ion 0 f the ba I I charact er i st i cs simulat ion) can be done in CRDISI fr'om the same
(deteNllinat ion of the maximal ball size dat a . The dot a t rans fer bet '>'Ben the hi er arch i c
after BOND, calculat Ion of opt imol ball layers is rea I i zed from t he bott am upwards on
content) floppy discs. ...-h i le dO'W'fl't(Qrds t here Is no
- Read, write, modi f icat ion and list ing of data bearer as very fe.... and mostly verbal data
technological data are transferred (e . g. suggestion to increase the
- Read, 'tfT'ite, modi ficat ion and list ing of r. p . m. 0 f the c Iass if i er by 10 percent for a
plant specific data given kind of cement). These instructions
2. I ndust r i a I expert loyer concern the t echno Iog i ca I process it se If, so a
o Technology specific ident ification feedback is establ ished in the grinding
- Parameter·s of the separator technology. The local regulators are placed on
- Batch grinding (separation matrix) the very bottom of this system, thus the control
- Characterist ic of the air-classifier can be conSidered a cascade one. Thus the
(TRct1P curve) CEI'l:XPERT is a dec i s i on support i ng syst em but
- ParCllTlet ers 0 f a pre f erred mode I de fined the decisions have to made by the operators .
160 L. KC\'iczh Pt al.

The fo11oo.'ing data fi le exist in the CEI'EXPERT Fig. 3.) Then using the above values and having
system: only one measured material floo.' the all
o factory specific constant data rerne i n i ng ones can be comput ed from the ba lance
o measured plant operation data equat ions. Spec i a I eQu i pment s (granu Iomet el's)
o data enabling the communication bet~en the based on d i ff ract i on met hod are more and more
prograns widely used for determination of the particle
o stored console communication data size distribution.(In our tests the laser
o output data of a program stored for graphic granulometer CILAS 715 wos used.) Their
presentation advantage against sieving is that they work
o data measured for identification purposes . automotically and makes possible to measure
part i c Ie sizes under 30 mi cromet el's wh i ch
Certain part of this data fi les does not occur practically determine the qual ity of the cement.
for the user, they are only auxi I iary fi les for
the convenient appl ication. Actually three kinds In these tests, however, a serious problem
of data have to be dealt ~ith. These are: arises, nOOlely that the sample is not
representative enough. The quasi-stationary
o plant operation data base state of the mi 1I system lasts only for a short
o plant specific constant data time, so to reduce the distort ion effect caused
o technological data base by the particle size distribution analysis, a
correct ion can be calculated from the material
1. Plant operation data base means the balance equations.
co II ect ion 0 f measurement s (approx. 40 dot a
per mil I) in regular intervals (e.g. by every A typical technological analysis includes the
hour).Since there are no data loggers folloo.'ing items :
i nst all ed an i nt ell i gent screen commun i cat ion o The analyt ic relat ionship between the Blaine
is elaborated for typing the data into files and the particle size distribution values has
(I.e. this is not a real- time system).One to be determined for the rni II product. By
record of the fi le contains the data means of this funct ion, the part icle size of
belonging to the SOOle time. The appl ied data the ideal separat ion can be calculated from
handl ing system ensures very safe data the specific surface reference value.
collection and possibi I ity for data selection o The outlet rate, the efficiency and the TROMP
for different purposes. curve can be computed from the analysis of
2. The plant specific constant data base gathers the air-classifier. The TROMP curve gives the
the data of the grinding system invariable in probabi I ity of the particle size getting into
time and the characteristics of the plant the reject (Beke, 1964).
(e.g. the size of the mill). These data can o The mi I1 diagram shows the effiCiency of the
also be modified by wish . grinding in different rnil I sections (particle
3. The technological data base includes the size distribution and specific surface along
measured data (e.g. the specific surface, the mill axis).
particle size distribution values) necessary o The optimal ball content of the mill (ball
to the technological computations (e.g . charges in each compartment) can be
material flws in the closed-circuit mi II - determined from the prescribed conditions
separator system, rni II diagram, ball charge given for the fractions (equal number or
charoct er i st i cs) . It i nc Iudes not on Iy equal surface of balls are required in each
primari Iy measured but computed data, too compartment) (Clemonn, 1964).
(e.g. moterial flws). The analysis of the
technology is occasional. Considering the The above ment i oned ana Iyses are va lid for the
large Quant it Y 0 f dot a each ser ies 0 f steady-state behaviour of the grinding system.
measures are to be placed i nt 0 independent
fi les. The structure of the fi les is fixed. The dolo logged in quasi-stat ionary state are
Oedicated programs help typing, modifying and not enough to accompl ish the analysis necessary
displaying these data. for the determination of an optimal separation .
The most useful method is to make computer
simulations with di fferent static and dynamiC
APPLICATION AREAS OF CEMEXPERT models. Its precondition rei ies upon the
expel" iment a I i dent i f icat ion 0 f mode I parOOlet el's
A typical appl ication is a technological of the technology (mi 11, separator, batch
ana Iys is. Since t he we i ghs are very expens i ve grinding system, etc . ) . These tasks can be
for the determination of moss fl~ in a performed most efficient Iy ~ith the
closed-circuit rni II system, other m<:>t.hods have techno Iogy-spec if i c i dent i f i cat ion subsyst em of
widely been appl ied which are based upon the CEt£XPERT. The st at i c and dynam i c behav i our 0 f
mass f Ia'*' somp Ies taken in a QUos i -st at i onary the mil I system can be investigated by
state and then either sieving the samples or technology-specific simulation using
measuring their specific surface or particle mat hemat i ca I mode Is based on i dent i f i ed
size distribut ion the received values are parOO'leters, so the opt imol input (fresh) feed,
analysed.(The places of sompling are shwn in as wel I as the regulator parOO'leters can be
Cemexpen I (j I

designed. above mil I - separator system. The minimum value


of the TROHP curve (0.16), the part icle size
flQplication ex~ belonging to the medium value (125 micron at the
TRCM' value of 0.6) and the TAOHP value (0.866)
Example 1. Calculation of mass flows in the at the doub Ie part i c Ie size 250 mi cron can be
closed-circuit grinding process read on the measured TRCM' curve. The measured
and fitted TAOMP curves are seen in Fig. 6, the
The scheme of the grinding system is shown in corresponding data ore in Fig. 7. Here the
Fig. 3. The mi II is of an end-discharge type. technology-specific identification subsystem was
The mi II product (10) gets into tw air- appl ied.
cla:!!~jfiers through an elevator (1 and 2). The
rejects (grits) (3 and 4) are led again into the COtCLUS I OHS
mill together .... ith the fresh material . The fine
The firs t a i. 0 f the C8'EXPERT i s to r.e I p tr.e opera tors
product (5 and 6) together .... ith the dust (11) of in tr.e caent factories to control tr.e .i II - sepcrotor
the electric air fi Iter form the end product systell opt ilia II y. By IIeQI'lS of the sys tell tr.e operator
(12) of the grinding. The length of the mi II is con analyse t~ - . r e d teclY\ological data frooo
technological and enerqeti ca I point of view. Different
15 meters its diameter is 4.4 meter. Its plots referring to tr.e efficietlCO,j of the plant (e . g.
guarant ed spec i f i cat ion is 113 t on/hour at the alOl'19 ~ axis of tr.e . i l l ) r.elp to n.rt tr.e . i l l IIOI"e
61aine value 2800 cm2/gr or 71 ton/hour at 4000 aid IIOI"e op t i IKJ I . Techno I ogy-spec i f i c
prograti r.elp to silOUlate tr.e effect of changes in tr.e
s i IOU I at i on

cm2/gr. The produced cement was PCS 350/20 ( .... ith IIQSS floes (e.9 . in feeding), particle size
min. 2800 cm2/gr according to the Hungarian di tributions, etc.
parotleters which
TI-oe !IOdels include technological
can be det_ined by the
st andards) . As it ....as ment i oned ear I i er the techno I ogy-spec i f i c i dent if i cat i on prograIIS.
we i ghs are very expens i ve , there fore t hey are
The expert systell - which is bel ieved to be tr.e first
not placed in every measuring pOint, but to get one in tr.e caent grinding technology - di ffers frOll tr.e
opt ima I operat ion it is necessary to knO'ff the usua I exper t sys t...,. . I t con ta i ns and """"*"
I a tes the
internal material flO'ffs of the mi I I system. knOllledge of experts in this field, but instead of
logical rules, arithlletic operations. diagralls are used
Measur i ng on Iy t he fresh mat er i a I f IO'ff all the to coapute t~ decision supporting results displayed
other material flows can be calculated from the lIQinly graphically. CEI'EXPERT is bui It in hierarchical
IICIY, dividood into three I""""",:
particle size distributions measured on the
numbered places. Further technological values o Plant operator layer
are given in Fig. 4. o Industrial expert layer
o Acadellical expert layer.

Example 2. Steady-state and dynamiC simulation C£I'EXPERT is installed at HejOcsaba CeIoent Factory in
of material flows in the closed- Hungary and .arks efficiently.
The s.ysteta indalled at the CADISI has considaNlbly
circuit ball mill system increased ~ knoeledge about tl-oese cetllent industrial
gr i nd i ng techno I og i ca I un i ts by rurYI i ng CEMEXPERT for
The dynamiC made I of the material flows which exploratory invest i gat ions.
The project .i II be continued to develop tr.e COIIpletely
approximates the operat ion of the mi II is used IX>rt~le version of the systetll for IBM AT cCMRpOtible
in the technology-specific simulation subsystem. professional PC-s.
The static characterist ic of the total feed -
fine product relationship can be seen in Fig. 5.
The explanation of the static characteristics is
that increasing the total teed over a certain It is t:Ar pleasure to express the appreciation of
limit the mi I I gets clogged. The step responses efforts and contributions !lOde by our colleagues G.
AI.a"Y, K. Hangos and F.Sipos (1986) in the preparation
of the fine material flO'ff for the increase of of CEMEXPERT. Tr.e authors are grateful for the valuable
fresh feed from 110 t/h to 120 t/h and CI"Id useful technological advices of J . Kolostori (1984).
thereafter to 125 t/h is shown in Fig. 6. The
F£FEF£tn:S
phys i ca I exp Ianat ion 0f the osc i I Iat ions
appear i ng in the st ep response is in the A I oosy, G., Cs. B6nyQsz, R. Haber, K. Hangos and F .
Si IX>s ( 1986 ) 4 FfR1ES So ftware f raIIe for
feedback loop. technological expert syst_s in
Hungar ion), Aut08lOt i zo las i Konferenc i a , Ny i regyh6za
(H), 49-58.
Beke, B . (1964) . Principles of COIIIItinution, Publ . House
Example 3. Identification of the TROHP curve of of tr.e HlIlgaI" i an AcadeIIy of Se i ences, Budapes t, 163.
the separator Bristol, E .H. (1985). Ihlt is AI? And should we
care?, Instrunoentation Technology, 32,4,37-41.
Clella\fl, J. (1984). Einfluss der RoIrIMi1lenbeluftung auf
The TAOMP curve shows the probabi I ity of getting die MahI.irkung,Zeaent-Kalk-Gi~,37,62B-635 .
a particle into the grit, by definition Keviczky, L . , M. Hilger and J. Kolostori (1985). Control
engineering theory of grinding,~t SZIKKTI.
Kr i gaoan, A . (1985 ). Expel" t sytees : I«Ju I d 'P.J trus t one?
T(i) = (G/I1) G(i)/I1(i) } Flleggi, R. (1967). fbsche i de-Effek te unci ~ i rksaaIke it von
Stre is i ch tarn f lA" Zeaen t.ato I cri I agen , Schooe i zer i sche
Bauzeitung, 85 , 70-74 .
where G is the grit flO'ff, M is the mil I product Seebach, H. M. (1972). IJerfciY'enstechnische Opti.ien.rog
flO'ff and G(i) and M(i) denote the corresponding ""'" Z--.tllClhlan- lagen,Ze.ent-l(alk-Gi~,25, 2,71-73 .
rooss flows of the i-th part icle size .... indO'olo'S. Slegten, P. and J . Mattan (1973). Cesent lIIill liners
segregate grinding lledia, Rock Products.76,3,68-71.
There are various analyt ical approximat ions of Tra.inski, M. (1976). Die IIOthetlatische Fonoul ierung der
the TAOMP curve (T rO'ff insk i, 1976), here that 0 f T~KLrVe, Aufbereit~tedYlik, 17, 248-254,
LYHCH is used for the separator Ho.l in the 449-459.
162

Academ ical Expert Layer

Industrial Expert Layer

Plant Operator Layer

Fig. 1 The CEnE XPEAT layers


MA 55 F'LOW 5 OF THE CLOSE D CIRCUtT 1. CEMENT MILL-SEPARATOR S'tSTEM
FrL~NAM£1 TC IMA21 . DAT 2l . ~5.1985. 12:30
f'IASS FLOW [t In J F i Q. Log I ca I .cheme 0 f CE~IEXPEAT

Cc:m2/g)

1 1 :::. :::: 15.2


203'3~(:

TECHtICL OGIC N.. nAT A OF" T1~ m OCEs!>

1:,:;0

TYFE 01" THE I'IILL (N,AAW, C r CF.:MENT),


GF:AIM. IA.)HOEn OF Tl~ HU.L '

~IE ":;URt:O FL Otl OF THE TOlfll. nRIT tt/hl,


AI::'IOLUT ID'" ,..EA tilI'llJT£ OF flE SE"ARAfDn C i 1 .. 1" I,
TYPE or. CCtl,,;rrT, PC 3::~/20

ftc::our::a[D D1.AltlE V(!IUIt;. 01'" THG F'II-IE ","ODUC T Cu, VO,.11 2800,0
POSITION OF' TIll:: gN""'" AT t HC OOUJ:ll. E F:OtAtr)~ MILL (ll,
POWEr. CC!-I\.it,)l'1PTIO .... "rWlt
,..rUJl>ORt IOW\1.. Ct'r::r.".Ut;(;Q At CALCU.. ATJOtol OF tH!! POWER,
F'MTICLE ItllE At 11)(;:1\1. &F.PM\ATIOIl (.11'''0--]1
OAT!:: OF' St.HPLHrO,
DA1 E COl' I-.'\.JN!-III./Ct ()QH., 0 1' Tl '" 1~ 1I.1 . 1

Fig . of
Fig . 3
8IMr>t..IF1F.O IWNAHIC HA!3(J FLOW 11001:1. 01" THI':
GIHPLIFtE:O 91::AOV OTATE J'1I\SG FLO"I Moon OF THE
CLOaEn CIRCU IT Df\LL I'IILL
CL Oil!!!) r:tnCUIT OALL 11ILI,

lot:HTJlltCR I 11.03 . 1'705.


FIIJE f,'n:l1t'1Uc:r tt/h]

1"+1).

130. ·· .. ..
_ •• -r - - - -r -. - -,.-
··
•• ••• :••••• :.-.:•.:..;
.. ..
•• --r----i'
••=--.;,..

60 , 0
110, .... - . --
'10 . 0

-~~.----~----~~----~~
l C'''. ZO \). ~·o o. !dlO.
T I/If'; rtlltl l
r.Ui'1 lNI"UT "' Ln~l tt/k)
r<EVOL lITIOII f>1::~ HJr.lUTG or. lW': GF.Pl\nr.tOIl IH)(! . ~ C I/ I"IH ' I
r.C'JOLUTIOIl "EA tI/l IU TE 0 " TH!: $1Zr>o.!:tllo lOI1 11011.0 CI/HTtll

Fig. 5 Fig . 6

F'ILE tr"~u: , 1C I I"II'l;;:I .O/)1 21 . t: ~. I <;I O:;"


FI LENflHt; 1 : I.'l~. I'?L'I :5 . H!)O I I" ltD v " U "r::::: f'I ,~SC D ( )I~ THE HI'IT Fr":II'IL pl\LArlCC
I n:: IHtF1ERI t'I~, .0~. I Q06 ,
!111Ufl'Ce: 01" t~e: DATA: ::

I. OI SPLIIVJ'JO AND t10 0 11'"IC AT I(JI~ OF ~\-IE tlCT ll l'I !,. OAl l\
:: . LOI'l'lINB OF , THE OATI'I S TOI'<F.:n I t I TH;' f [ (;I' NC'I.OG 1C"L
TV""e: (IF T~E SE"'~A T C" SVS Tr,:H: ::
I. G(P,.....< A T~::; 101 11'1-1 l~U! ELCCTO::O" :t.'(1'i; 1.r.::l'r.PI,\lcn .'3 01) ............ .
::. ~IO P .... I>l\ T01"":' ~1!'''''lIt T"~ I:l.Er.Tr,~"II. ·::n

Tyt-l:: r:ll' ti'r:: l ~ lIl'W curr./fO, :

O" T I'I rl r::CCE ;' TO TllF.: I t'£N Tl"IC "," l ION


~ . MO . . . . . . . -- .. . . . -- ..... ..

L. ~lrH I"IUI'1 V ~'-L'E ~ THE TROK ' Cl'~ ... t:: ,


:.' . f' t.tlTIC:.::: S I1 E AT I1r; OIUM rr';'Hr" 'J,',L'j:,
:;. Tr.'t\"lP VALU£ AT Tlr£ OOllp\. E 0"" ~ "E t-EOll'H f' lvn !cu: Sit!;: ,

• LOAD IN G,
- Or>:M' H 1C
L I STP'!; I\U;),tP
;) I S P\ .(l ~l l lG
r l =, ~ :r lr..n: C 1!

COl' 110'1: H ,C.,;> tu:;'. ;::


Lr- TIt(! 1 :("11'

· ' '' B
·····~·········· ....-.... ........
. ~:: ·· ·· ·· · · I ~ · '
"
>: :

~
- FI T T tI>.IQ WIT'" 1\ G I VE N FCoI::"' · :[i " =:r.
- AL~ I At-fl..lM'E nIC O ISPL Ci y ltlO C'" n 't R E 5~'LT!';.
- CRo'WHlC OISI'\. .:IVIt.'G er=- lI(E 'i:1;!>t..'\.~ 5

(N, Eh N PAIH!C I. !:: SI; r:: , ... \ 1"" 0" ]

Fig. 7 Fig . B

You might also like