Vo l. 31. No. 260. Au g u s t 1945.
T H E PR EC ISIO N AND ACCURACY OF VISCOMETRY USING B.S.I. TUBES.
B y Th e Vi s c o s i t y Pa n e l o f St a n d a r d i z a t i o n Su b-Co m m i t t e e
N o. 6—Lu b r i c a n t s.
At th e F irst W orld P etroleum Congress held in London in 1933 a resolution was adopted th a t viscosities should preferably be expressed as kinem atic viscosities, an d th a t international co-operation was necessary to ensure th a t th e same values of units of kinem atic viscosity are in use in different countries.
A com parison of th e u n it of kinem atic viscosity—th e centistoke—is a m atter of considerable difficulty, as th ere is no certain ty th a t oils will retain th eir viscosity unchanged under different storage conditions. D r.
B arr has long felt th a t in tern atio n al standardization of viscosities should he m ade w ith viscom eters of know n characteristics, as these are m uch less likely to change th a n sta n d a rd viscosity oils.
A t a m eeting of a P an el of th e In te rn atio n al Association for Testing Materials a t th e F irs t W orld Petroleum Congress, it was arranged th a t two glass U -tube viscom eters, num bers 2 and 3, conforming to th e B ritish Standard Specification No. 188 of 1929, should be calibrated in a num ber of th e natio n al standardizing or associated laboratories. The same in stru ments were circulated to th e following laboratories.
(1) T he N atio n al P hysical L aboratory, Teddington.
(2) Physikalisch-technische R eichsanstalt, Berlin-C harlottenburg.
(3) Ecole N ationale Supérieure du Petrole, Strasbourg.
(4) Professor B ingham , L afay ette College, E aston, P a.
(5) D r. R . N . J . Saal, L aboratorium v an N. V. de B ataafsch Petroleum M aatschappij, A m sterdam .
Different m ethods of calibrating were used in each laboratory, b u t th e methods of te s t were those laid down in th e B ritish Standards Specification.
Calibration constants obtained were n o t reported u n til 1937, as there was considerable delay in th e circulation of th e instrum ents from one laboratory to another. T he absolute values of th e constants reported by th e five laboratories differed in some cases by more th a n 2 per cent, from th e mean, a n d fu rth er absolute and com parative experim ents were therefore required before satisfactory agreem ent could be reached on practical standards of kinem atic viscosity.
In th e m eantim e, th e standardization of m ethods of te s t was tra n s ferred to th e In te rn a tio n a l Standards Association and its Sub-Com m ittee 28 A/18.
A t a m eeting held a t Brussels, 2 6 th -2 8 th Ja n u a ry , 1939, a t a conference on viscosity m easurem ents, it was agreed th a t th e In s titu te of Petroleum should order a num ber of Ostw ald viscometers B .S.I. Nos. 2 and 3, which would be calibrated by th e N ational Physical L ab o rato ry and th en sup
plied to th e countries participating in th e tests. A rrangem ents were m ade sim ultaneously to supply a set of samples of three oils, described as
Oil 38, Oil 25, an d Oil 39, prepared by th e lab o rato ry of th e B ataafsch P etroleum M aatsch ap p ij, of w hich tw o oils were to be ru n on each in s tru m e n t; th e viscosities were a b o u t 17, 80, an d 530 centistokes a t 20 C.
G reat care h a d been ta k e n in th e p rep aratio n of th e oils, an d it was recom m ended t h a t th e y should be k e p t in th e d a rk in a cool, d ry place u n til th e viscosity determ inations were m ade. N o less th a n fourteen laboratories in th e U n ited S tates, Sweden, D enm ark, G erm any, France, H olland, an d E n g lan d h a d arranged to ta k e p a rt in these tests, an d all these laboratories were to determ ine flow tim es in th e viscom eters supplied to th em on th e same dates. A t th e sam e tim e th e oils were also to have th e ir viscosities determ ined, by those n atio n al stan d ard izin g bodies taking p a rt, in th e ir own in stru m en ts. T he arrangem ents for these te sts were ' m ade b y th e I n s titu te of P etroleum V iscosity P anel, an d a series of th irty -tw o viscom eters, sixteen of No. 2, 145/9 to 160/9 inclusive and sixteen of N o. 3, 129/9 to 144/9 inclusive, were sent to th e N .P .L . for calibration. T he flow tim es o f tw o o th er oils of different viscosities at 25° C. were observed by th e N .P .L . in each of th irty -tw o tubes, and the results were n o t com m unicated to an y of th e p a rticip an ts. Times were determ ined in trip licate a n d m ean values were d e d u c e d : th e readings agreed w ith ±0*1 per cent.
I t h a d been arranged t h a t th e whole of th e results should be sent to th e N .P .L . and, w hen com pleted, should be exam ined a n d reported upon b y th e In s titu te of P etroleum V iscosity P anel, b u t owing to th e Second W orld W ar, only a few of th e laboratories tak in g p a r t re tu rn e d their test results. I t was, however, felt th a t some account of th e w ork should be given as a record of w h at progress has been m ade.
T he N .P .L . constructed a ta b le giving th e ratio s of th e m ean flow times in each of th e viscom eters of range 3 to th e m ean flow tim es of th e same oils in relation to one viscom eter No. 140/9, an d of th e m ean flow times in each viscom eter of range 2 to those in viscom eter No. 150/9. This ta b le is p resented here as T able I. T he oils h a d viscosities, m easured in N .P .L . sta n d a rd tu b es a t 25° C. as follows (June 1939) :—
Oil A ... 122-7 cs.
Oil 200 . . . . 41-3 cs.
Oil 40 . . . . 5-15 cs.
I t is sta te d t h a t Oil 40 was calib rated in a No. 2 tu b e (40 per cent, sucrose) w hich gave th e viscosity of an oil (No. 25) as 78-93 cs. when its viscosity in a No. 3 tu b e (60 per cent, sucrose) was 78-63 cs. T hus Oil 40 m ay possess a viscosity of 5-14 cs. based on 60 per cent, sucrose. The o th er tw o oils were calibrated iii a No. 3 tu b e using 60 per cent, sucrose solution.
All th e tubes, except Nos. 140/9 a n d 150/9, were th e n despatched in pairs to different laboratories b o th in E urope an d in th e U.S.A. Nos.
140/9 an d 150/9 were retain ed a t th e N .P .L . T he viscosities of th e three oils supplied to each lab o ra to ry were determ ined by th e N .P .L . (October 1939) :—
Oil 38 viscosity a t 20° C. . . 17-75 cs.
Oil 25 viscosity a t 20° C. . . . 78-63 cs.
Oil 39 viscosity a t 20° C. . . . 526-0 cs.
Y ISC O M ETR Y USES’ G B .S .I . T U B E S . 241
T a b u e I . O il A in tu b e X o . 140/9.
O il 200 in tu b e N o. 140/9.
O il 200 in tu b e X o. 150/9.
O il 40 in tu b e X o. 150/9.
485-2 sec.
163-4 sec.
909- 2 sec.
113-1 sec.
R a n g e 3.
R e la tiv e F lo w T im e. R an g e 2.
R e la tiv e F lo w T im e.
I n s tr . X o. O il A. O il 200. I n s tr . Xo. O il 200. OU 40.
129/9 0-8884 0-8915 145/9 0-8544 0-8540
130/9 0-8811 0-8861 146/9 0-7795 0-7796
131/9 0-8855 0-8881 147/9 0-8838 0-8821
132/9 0-9237 0-9262 148/9 0-8935 0-8925
133 9 0-8652 0-8669 149/9 0-9774 0-9777
134/9 0-9353 0-9397 150/9 1-0000 1-0000
135/9 0-9449 0-9469 151/9 0-7850 0-7857
136/9 1-0268 1-0284 152/9 0-7986 0-7980
137/9 0-9536 0-9543 153/9 0-9236 0-9221
138/9 1-0648 1-0670 154/9 0-8683 0-8675
139/9 0-9768 0-9763 155/9 0-7664 0-7669
140/9 1-0000 1-0000 156/9 0-7155 0-7129
141/9 0-8913 0-8930 157/9 1-0389 1-0403
142'9 0-8556 0-8581 158/9 1-0070 1-0053
143/9 0-8822 0"8S3o 159/9 1-0168 1-0177
144/9 0-9100 0-9124 160/9 0-9918 0-9928
Ta b u e ü .
T im e o f
T ube X o. OU X o. F lo w , sees. L a b o ra to ry . D a te .
a t 20° C.
135/9 25 291-7
135/9 39 1974-1 1 F ra n c e (U n iv ersité d e S tra s- _
146/9 25 1336-8 f bo u rg ).
146/9 38 301-8
129/9 25 292-0 T
129/9 39 1950-7 1 D e n m ark (A ssociation D an o is _
160/9 25 1361-7 f d e S ta n d a rd isa tio n ).
160/9 38 305-4
144/9 25 282-9
144/9 39 1892-0 1 S ta n d . In sp e c tio n L ab s. X o v . 1939
148/9 25 1545-0 U .S .A . (Esso).
148/9 38 347-4
140/9 25 311-0 ]
140/9 39 2071-3
^ X .P .L . O ct. 1939
150/9 25 1731-6
150/9 38 389-1
141/9 25 277-0
141/9 145/9
39 25
1859-0
1482-6 / E sso E u ro p e a n L ab s. —
145/9 38 332-4
J
130/9 25 277-0 i
130/9 159/9
39 25
1850-5
1774-0 j- B u rm a h OU Co. J a n . 1940
159/9 38 395-5 J
Oils 25 an d 39 were tested in a Xo. 3 tu b e calibrated b y 60 per cent, sucrose. Oil 38 was te ste d against 40 p er cent, sucrose in a Xo. 2 tu b e
and, sim ilarly to Oil 40, m ay therefore have a viscosity of 17-71 cs. against 60 per cent. N o t all th e laboratories have been able to re tu rn te s t results, as th e w ork was in itia te d in 1939. T able I I gives th e results available.
C ertain of these laboratories fu rth e r gave viscosity values for these oils a t 20° C. as m easured b y th e ir own viscom eters. All figures given are for a tem p e ratu re of 20° C. These figures are given in T able I I I . These th ree tab les comprise all th e d a ta available from these tests.
Ta b l e I I I .
L a b o ra to ry .
V isc o sity o f O il a t 20° C.,
R e m a rk s.
O il 25. O il 38. O il 39.
N .P .L . 78-63 H - 7 ^ 5 526-0 O c to b er, 1939.
U .S .A .
S ta n d . In s p . L ab o ra - . to r y
P e n n . S ta te
N a tio n a l B u re a u of S ta n d a rd s
79-00 78-89 79-08
17-77 17-76 17-79
528-7 527-6 528-5
A .S.T.M . su sp en d e d level (N ov., 1939).
F e n sk e-C an n o n .
B in g h a m a n d P y k n o m e te r.
D e n m a rk (Assoc. D an o is de S ta n d .)
77-92 17-47 521-0 R a a s c h o u r V iscom eter.
E sso E u ro p e a n L a b o r a to rie s
78-77 17-70 527-0 N .P .L . C a lib ra te d Ost- w ald .
T ex as O il Co., L td . — 17-65 529-8 C o m p an y ’s M aster Ost- w a ld tu b es.
An a l y s i s o p Er r o r s a n d Pr e c i s i o n.
I t was found t h a t T able I does n o t rep resen t th e b est w ay of pre
senting th e results. All tim es recorded are divided b y th e tim es of flow of th e sam e oils in tw o a rb itra rily chosen tubes, 140/9 a n d 150/9, of which th e accuracy is unknow n. T he ra tio of th e tim es of flow of tw o oils in the sam e tu b e was therefore calculated, a n d T able IV gives th e ratio for different tu b es which are also presented on Eigs. 1 a n d 2.
As precision in th e tim e reading was rep o rte d to ± 0 - 1 per cent., it is obviously useless to specify a higher precision on th e agreem ent between th e different tubes. I f dz 0 1 p er cent, be m ade th e m axim um allowable divergence from th e m ean value, it will be seen from Figs. 1 an d 2 th a t in b o th cases a num ber of tubes, ju s t over h a lf th e to ta l, give results w ith in these lim its. T he m axim um divergence for all tu b es in b o th series is of th e order of ± 0 - 3 per cent, from th e m ean value.
Tw o im p o rta n t conclusions follows :—
(1) I t is possible to lim it th e m axim um error to dz 0T p er cent, from th e m ean value in m easuring th e viscosity of an oil 2 b y m eans of U-tube viscom eters of th e B .S.I. ty p e which h av e been calib rated b y m eans of an oil 1, if a num ber of such tu b es are first subjected to a sim ple q u ality - control te st. This quality-control te s t consists of f i n d i n g th e ra tio of the tim e of flow of a t least tw o oils in a num ber of tubes, rejecting all tubes
VISC O M ETK Y U S IN G B .S .I . T U B E S . 243
£-.980
? 9 6 0
2-940
a-010
( O
) O
4- 0 • ! % 0
r \ M f AN ©
--- e--- 0 ~ ( )
)---e — e ---
<
0
>
_ 2 _ L 2 £ --- 0 -----
129/3 . 13^ T o b e N o . 13
F i g . 1.
9 /9 I'M
8-0 50
i-Oio
O
0
O
G + 0-1%
0
O G
O
G ME/\N
)
° O ()
C) - 0 - 1 %
0 (
I4S/3 I So / 9
T u b e No. Fi g. 2 .
1 5 5 / 9 1 €0/9
giving a divergence from th e m ean exceeding di OT per cent, and re ta in ing th e rest. This conclusion implies th a t th e excessive divergence is due to im perceptible fau lts in th e tubes, and n o t in th e procedure. This im plication is justified, as all th e tubes were subjected to th e same pro cedure a n d gave repeatable results in th e same laboratory. I t appears
Ta b l e IV .
T u b e No.
T im e o f F lo w of
O il A a t 25° C. T u b e No.
T im e o f F lo w of.
O il 200 a t 25° C.
T im e of F low of Oil 200 a t 25° C.
T im e o f F lo w o f O il 40 a t 25° C.
129/9 2-959 145/9 8-043
130/9 2-952 146/9 8-039
131/9 2-959 147/9 8-056
132/9 2-959 148/9 8-049
133/9 2-960 149/9 8-038
134/9 2-954 150/9 8-039
135/9 2-961 151/9 8-047
136/9 2-962 152/9 8-045
137/9 2-968 153/9 8-054
138/9 2-965 154/9 8-047
139/9 2-971 155/9 8-037
140/9 2-969 156/9 8-067
141/9 2-963 157/9 8-032
142/9 2-959 158/9 8-056
143/9 ■ 2-965 159/9 8-047
144/9 2-961 160/9 8-032
th a t th is te st, an a c tu al flow te s t an d n o t m erely subjecting th e tubes to dim ensional specifications, should be th e final check for tu b es for accurate viscom etry.
(2) F o r w ell-constructed tubes, te ste d b y th e dynam ic te s t outlined above and found acceptable, th e co n stan t of calibration is independent of th e tim e of flow w ith in th e lim its of th e precision im posed in th e dynamic te st. This conclusion is ap p a re n tly in co ntradiction of th e evidence in T able I, w hich shows a consistent increase in th e ra tio of tim es of flow of oils in a tu b e to th e tim es in 140/9. T able I, however, is misleading, as th e basic tu b e chosen, 140/9 for range 3, is a n u nsuitable basis of com
parison. Fig. 1 shows th a t tu b e 140/9 is am ong th e w orst rejects. Tube
Ta b l e V.
T u b e No.
R e la tiv e F lo w T im e s o f O il in T u b e t o T im e o f S am e O il in T u b e 133/9 a t 25° C.
O il A. O il 200.
129/9 1-027 1-028
131/9 1-023 1-024
132/9 1-068 1-069
133/9 1-000 1-000
135/9 1-092 1-092
136/9 1-187 1-186
141/9 1-030 1-030
142/9 0-9891 0-9895
144/9 1-052 1-052
150/9 is ju st on th e border-line. I t will be seen t h a t for range 2—i.e., where th e basis of com parison (150/9) is acceptable—th e ratio s o f times in Table I differ b o th on th e positive and negative side of th e m ean by sm all am ounts. T he m axim um divergence is for tu b e 156/9, giving 0 003
VISC O M ETR Y U S IN G B .S .I . T U B E S . 245 in 0-715 or 0-4 per cent, higher value for th e more viscous oil. T ube 156/9, however, is th e w orst am ong th e tubes tested by th e dynam ic quality-control te st ju s t described, an d would be rejected. None of th e tubes of range 2 accepted by th is dynam ic te st shows more th a n 0-1 per cent, difference in th e ra tio from th e m ean value of th e tw o oils, and these differences are n o t consistently in favour of either high or low rates of flow for eightfold range in rates. In th e case of range 3 th e tu b e to be used as a basis should be a n y one of th e tubes which passed th e dynam ic quality-control test. P ro b ab ly tu b e 133/9 gives th e best basis of com
parison, as it falls on th e exact average in Fig. 1. Thus Table V gives th e appropriate d a ta for all th e tubes which passed th e dynam ic quality- control test.
I t is seen th a t th e change in th e ratio is of th e order of 0-05 per cent, from th e average, an d even though more results show higher ratio w ith the more viscous oil, th e deviation is below th e 0-1 per cent, precision limit, and th u s is w ith o u t significance. I t is interesting to note th a t by correctly choosing 133/9 an d n o t 140/9 as a stan d ard tube, even tubes 139/9 and 140/9 (the tw o giving th e .worse results on th e dynam ic quality-control test) give th e following reasonably constant ratios :—-
Oil. R a tio for O il A . R a tio fo r Oil 200.
139/9 1-129 1-126
140/9 1-155 1-153
The deviation, however, is such th a t a low er-calibration constant is obtained w ith th e more viscous oil. T hus it is concluded th a t th e cali
bration constant for tubes which pass th e dynam ic quality-control te s t is independent of th e tim e of flow of th e calibrating oil for both range 2 and range 3 tubes an d for ratios of viscosity of oils used up to 8 : 1.
The above tw o conclusions combine into th e corollary th a t it is possible to find th e viscosity of an oil, 3, b y m eans of stepping-up procedure using
Ta b l e V I.
T ube No. R a tio of F lo w T im e o f O il 39 a t
20° C. to T im e o f O il 25 a t 20° C. R em a rk s.
129/9 130/9 135/9 140/9 141/9 144/9
6-681 6-681 6-768 6-660 6-711 6-688
D en m ark . B u rm a h O il Co.*
F ra n c e.
N .P .L .*
E sso E u ro p e a n L ab o ra to rie s.
E sso., U .S.A .
T ube No. T im e o f O il 25 a t 20° C. to T im e
o f O il 38 a t 20° C. R em ark s.
145/9 146/9 148/9 150/9 159/9 160/9
4-460 4-429 4-447 4-450 4-486 4-459
E sso E u ro p e a n L ab o ra to rie s.
F ra n c e.
E sso ., U .S.A . N .P .L . B u rm a h Oil Co.
D e n m a rk .f
* T u b e is re je c t on F ig . 1. t T u b e is re je c t on F ig. 2.
tw o o th er oils, 2 an d 1, a n d tw o U -tubes of th e B .S.I. ty p es to w ithin
± 0-2 per cent, m axim um error from a m ean value, provided t h a t b o th tubes have passed th e dynam ic quality-control te s t a n d t h a t th e experi
m ents are perform ed in th e sam e lab o rato ry a n d a t one tem p eratu re. I t is to be no ted th a t no th in g in T able I or th e b e tte r conversions Tables IV an d V can give assurance t h a t th e calibration co n stan t is n o t influenced b y th e tem p eratu re of th e b a th or th e p articu la r la b o rato ry an d country o f te st. U n fo rtu n ately , nothing in th is w ork can be used to evaluate th e influence of tem p e ratu re on th e calibration co n stan t. However, Tables I I an d I I I m ay be analysed for th e discrepancies in viscosities rep o rted by different laboratories. T he dynam ic q uality-control te st is applied to th e contents of T able I I to find th e discrepancies between laboratories using th e same oils a n d different tubes.
T able V I gives th e results for all th e laboratories. I t is evident th a t th e results from F ran ce disagree from all others a n d can be neglected u n til a m ore detailed stu d y of th e causes of difference can be realized.
W hen neglecting th e F rench results as well as those o b tain ed w ith tubes w hich are unsuitable for accurate viscom etry, th e discrepancies between th e different laboratories using good tu b es reach a m axim um of d; 0-25 per cent, from th e m ean for all laboratories for range 3 tu b es an d d: 0-7 per cent, for range 2 tubes. As sam ples of th e sam e oils were tested at approxim ately th e sam e d a te in tu b es w hich hav e yielded good results in a single laboratory, it becomes ap p a re n t t h a t th ere is need to stu d y the factors which m ake laboratories differ so m uch from each other. I t is to be rem em bered, however, t h a t for m ost purposes, th e calculation of V .I. included, te sts would norm ally be done in a single laboratory.
Ac c u r a c y o f Vi s c o m e t r y.
T able I I I affords a possibility of studying th e accuracy of viscometric m ethods used. (N ot all th e laboratories w hich re tu rn e d results tabulated in Table I I re tu rn ed viscosity values.) I t is ev id en t t h a t th e Danish viscom etric m ethod gives low values w hen com pared w ith B ritish and U.S.A. m ethods. R esults from th e T exas Oil Co., L td ., show a relatively larger difference from th e m ean B ritish viscom etric results th a n either of th e tw o others, hence for a first com parison th e y are excluded. The N .P .L . an d Esso E u ro p ean L aboratories agree to w ith in 1 in 800 m axim um difference. As b o th results are based on N .P .L . calibration, th is is not gre at significance from th e view -point of accuracy. T h e U .S.A . results use th ree distinct m ethods, a n d th e agreem ent betw een th e th ree results is to w ithin 1 in 500 m axim um difference giving d: 0-1 per cent, deviation from th e m ean. The averages for th e U.S.A. an d B ritish results (excluding th e Texas Oil C om pany’s results) are as follows.
I f we ta k e th e average results for B ritish values for all th re e laboratories concerned we o btain :—
Oil . . 25 38 39
Viscosity, cs. . 78-70 17-69 527-3
This gives deviations of - f 0-36, + 0-45, a n d + 0 - 1 9 p er cent. The average deviation is 0-33 per cent.
VTSCOMETR.Y U S IN G B .S .I . T U B E S . 247
Ta b l e V I I .
V iscosity of Oil a t 20° C., cs.
Oil 25. O il 38. O il 39.
U .S .A ...
B ritis h . . . . .
Diff. = U .S .A . — B ritis h .
% diff. - d lff' X 100 m ean
79-09 78-70
17-77 17-71
528-3 526-5 + 0-29
+ 0-36
+ 0-06 + 0-34
+ 1-S + 0-34
T hus, it appears either differences in th e fundam ental scales of th e dimensions mass, length, and tim e and of tem p eratu re as used in th e tw o countries m ay exist giving unidirectional errors or th e practical units of viscosity used in th e tw o countries m ay be different. The basis of th e American u n it is th e viscosity of pure w ater a t 20° C., th a t in B ritain is th e viscosity of 60 per cent, sucrose a t 25° C. I t is to be no ted th a t th e N .P.L. adopted th e 60 per cent, sucrose as sta n d a rd on th e basis of D r.
B arr’s experim ents in which he has shown deviations betw een different absolute m ethods am ounting to 0-3 per cent. Viscosity results calculated on a basis of calibrating w ith a 40 per cent, sucrose would m ake th e British an d A m erican results agree to less th a n 0-1 per cent, error. O ther work, however, has shown th e 40 per 'cent, sucrose viscosity to be unreliable as a standard.
Co n c l u s i o n s.
(1) I t is n o t satisfactory to rely solely on th e present dim ensional specifications for B .S.I. tubes if viscosities are to be determ ined w ith
± 0-1 per cent, m axim um error from a m ean value for results obtained in different laboratories using th e same basis for th e p ractical unit.
(2) In any one lab o rato ry agreem ent in viscosity determ inations w ithin 0-2 per cent, can be obtained if th e ratio of tim es of flow of tw o oils are determ ined in th e tu b e under te s t in com parison w ith a tu b e shown to be satisfactory in a dynam ic control te s t outlined above, such th a t th e ratio of tim e of flow of tw o oils is w ithin ± 0 - 1 per cent, of th e m ean of a large num ber of tubes (15-20).
(3) I n an y one laboratory, tu b es calibrated b y th e procedure under No. 2 above can be used for stepping-up viscosities and should give agreem ent w ith ¿ 0 - 2 per cent.
(4) F u rth e r investigation is required to o btain closer agreem ent between different laboratories an d p articu larly as regards :
(а) tem p eratu re m easurem ent, and (б) tim e m easurem ent.
(5) V iscosity m easurem ents in th e U.S.A. by fundam entally different m ethods, on th e same oil show m axim um differences of i 0-12 per cent, from th e m ean value.
(6) B ritish m ean values of viscosities on these oils are 0-35 per cent, below th e m ean values obtained in th e U.S., suggesting th a t th e fu n d a
m ental u n it of centistokes has a different practical value in th e two countries.
T H E TESTIN G OF GREASES FO R BALL-BEARINGS.*
B y S. R. P e t h b i c k , B .Sc.
St t m m a b y.
C o n v en tio n al m eth o d s fo r te s tin g greases a re o f little a ssista n c e in th e selectio n o f greases fo r p a rtic u la r a p p lic atio n s .
T es ts on lim e-base greases sh o u ld in clu d e : (1) flow m e a su re m e n ts a t v a rio u s ra te s o f sh e a r a n d a t v a rio u s te m p e ra tu re s , (2 ) a n o x id a tio n te s t w ith a n e x a m in a tio n o f th e o x id atio n p ro d u c t, a n d (3) a sy n eresis t e s t in v o lv in g c a p illa ry a ctio n .
S oda-base a n d lith ia b a se greases v a r y w id ely in te x tu r e , cohesion, a n d ad h esio n , a n d som e m e th o d o f m ea su rin g th e s e p ro p e rtie s is d esirab le. T he co m b in ed cohesion a n d a d h esio n p ro p e rtie s c a n b e ex am in ed b y m ea su re m e n ts o f th e to rq u e e x e rte d on th e h o u sin g o f a ro ta tin g b earin g , w hen th e la t t e r is p a r tia lly filled w ith g rease. A su ita b le a p p a r a tu s fo r to rq u e m e a su re m e n ts is d escribed.
A c h u rn in g te s t follow ed b y to rq u e m ea su re m e n ts gives u sefu l in fo rm a tio n on th e s ta b ility o f a grease.
In recent years the Institute of Petroleum has endeavoured to achieve a high degree of standardization of the methods for testing petroleum products.
T he m ethods so fa r stan d ard ized for testin g lu b ricatin g greases are of value for th e routine checking of a p a rtic u la r p ro d u ct, b u t th e results of such te sts do n o t render it possible to select a grease for a particular ap plication w ith o u t a full-scale te s t in th e equipm ent to be lubricated.
I t is well know n th a t a num b er of greases giving th e sam e penetration value can be m ade b y com pounding a soap w ith a n u m b er of oils of v ary in g viscosities, an d sim ilarly, by incorporating a num b er of different soaps in an oil, several products having th e sam e consistency can he produced.
I t is extrem ely unlikely t h a t all th e pro d u cts m ade b y th e various m ethods will give sim ilar perform ances in lu bricating a bearing. Con
sistency m easurem ents are therefore of little assistance in th e selection of greases.
The oth er te s t w hich m u st be used w ith a know ledge of its lim itations is th e drop-point determ ination. A m inim um d ro p -p o in t requirem ent is a simple m eans o f elim inating obviously u n su itab le pro d u cts, b u t the fa c t t h a t a grease has a drop-point above th e m inim um req u irem en t does n o t m ean t h a t th e lu b rica n t will be satisfacto ry even a t te m p eratu res well below its drop-point.
T he Sub-C om m ittee responsible for stan d ard izin g th e d ro p-point test have sta te d t h a t it should n o t necessarily be considered as having any bearing on service perform ance.
There is therefore a need for sta n d a rd m ethods w hich can be used for th e selection of greases for various purposes.
Such m ethods could be classified as te sts for greases suitable for use a t (a) low speeds an d m oderate tem p eratu res an d (b) high speeds an d high
* A ck n o w led g m en ts a re d u e to M r. H . S. W o o d a n d M r. R . H . U n s w o rth fo r a s s is t
an ce in th e e x p e rim e n ta l w ork o n Which th is p a p e r is based.
tem peratures. In to category (a) would fall general purpose greases of th e lime-base ty p e, an d category (6) would include greases of th e soda-base type.
The m ethods of te s t should sim ulate practical conditions as far as possible.
In th e sam e way th a t th e sta n d ard m ethod for fuel testing uses an actu al engine for th e d eterm ination of octane num ber, so a sta n d a rd m ethod for testing greases should em ploy a bearing, unless th e properties can be precisely defined by th e results of fundam ental m easurem ents.
The requirem ents for a desirable grease are : (1) th e ability to lubricate a bearing for long periods a t all tem peratures and speeds a t which the equipm ent is to be o p erated ; (2) com plete stab ility and retention of.
properties on storage, b o th in its container and in a b e a rin g ; (3) capability of being dispensed by norm al grease guns and lubricators.
In order to lubricate a bearing, th e grease m ust rem ain in th e place where it is m ost required. T he surfaces of a ball bearing which require lubrication are those of th e balls an d tracks. Grease which rem ains in any other position in th e bearing does no useful w ork unless it acts as a reservoir for th e replenishm ent of th e tracks.
N orm al lime-base greases, if used a t tem peratures below 50° C. and in bearings of low periphery speeds, te n d to m ain tain lubrication by flowing into th e tracks. A m oderate ra te of shear is sufficient to cause such greases to a tta in some degree of fluidity. H igh rate s of shear te n d to reduce th e ap p aren t viscosity to a value which is relatively close to th a t of the base oil, an d th e lu b rican t is soon lost in a high-speed bearing.
This can be illustrated by th e ap p aren t viscosities of a lime-base grease which were found to be 80 poises a t a rate of shear of 15 secs.-1 an d 3-3 poises a t a ra te of shear of 10,000 secs."1 The viscosity of th e base oil was 2-1 poises.
I t is obvious th a t some form of viscom etric m easurem ent will provide useful inform ation on lime-base greases. A pressure viscom eter which has been developed in th e U.S.A. is probably one of th e m ost convenient instrum ents for such m easurem ents. I t consists of a cylinder fitted w ith a capillary orifice th ro u g h which th e grease is forced by a piston operated hydrauhcally a t a know n pressure. B y m easuring th e pressure required to cause th e grease to flow a t a given ra te and knowing th e dim ensions of the capillary th e a p p aren t viscosity can be calculated.
Since th e cohesion an d adhesion of m ost lime-base greases are similar, the product which will be retained in th e bearing to th e greatest ex ten t can be selected by flow m easurem ents w ithout tests in a bearing.
U nfortunately, lim e-base greases which depend on w ater for th eir structure are very prone to be unstable in use an d in storage. D espite the num ber of h e a t te sts which have been proposed for m easuring stab ility , it cannot y e t be sta te d th a t a satisfactory solution has been found.
Two types of in stab ility occur in service. In one case either a sticky resinous substance or a h ard varnish is form ed in th e bearing. This type of failure is undoubtedly due to th e oxidation and polym erisation of u n satu rated constituents. I t is usually found th a t greases of this n atu re absorb oxygen in a bom b fairly rapidly a t 100° C. a n d become fluid, whilst greases n o t prone to this ty p e of failure absorb oxygen slowly and retain th e ir consistency.
A rapid m ethod of elim inating oxidizable greases consists of heating a
P E T H B IC K : T H E T E ST IN G OF G E E A SE S FO B B A L L -B E A E IN G S . 249
film of th e grease in a bearing a t a tem p eratu re of 120-150° C. an d m easu r
ing th e to rq u e required to tu rn th e cooled bearing th ro u g h th e first revolution a fte r a definite period of heating.
T he other ty p e of failure arises from greases w hich on testin g b y norm al h e a t te sts app ear to be stable, a n d in fa c t are stable w hen sto red in bulk.
W hen applied to a bearing, capillary forces a p p ear to e x tra c t th e oil from th e soap an d leave a w hite deposit u n d er th e balls of th e bearing.
The failure is often associated w ith excess of alkali w hich ten d s to p re cip itate th e soap.
I t has been found th a t if a bearing packed w ith such a grease is placed on an absorbent p a d an d h eated for ab o u t a week a t a te m p eratu re ab o u t 10° C. below th e drop-point, th e grease becomes of a cheese-like n a tu re an d th e bearing is no longer free tu rn in g .
The a u th o r has m et tw o extrem e cases of failure of th e above types.
In th e first case th e grease was slightly acid a n d all-steel bearings filled w ith th e grease were fou n d to seize com pletely in six weeks.
In h e a t te sts a t 130° C. th e grease failed to re-set on cooling, a n d it was th o u g h t th a t such a te s t would provide an ad eq u ate safeguard against seizure of bearings. Sam ples o f grease were ob tain ed w hich solidified a fte r h eating an d suffered little loss of oil w hen h eated an d in v erted in th e dish a t room te m p e ratu re for a week. Bearings were packed w ith th e greases an d stored in grease-proof p ap er in cardboard boxes a t 40° C.
W hen th e bearings were inspected afte r th ree weeks, th e bearing packed w ith th e grease w hich was th o u g h t to be th e m ost stable h a d seized com pletely, an d a w hite deposit was found under each ball.
A fter a num ber of te sts, it was found th a t only te sts involving capillary action w ould elim inate greases of th is ty p e.
T he m ethods for testin g lim e-base greases should therefore include (1) flow m easurem ents a t various ra te s of shear a n d various tem peratures, (2) a n oxidation te s t w ith a n ex am ination of th e oxidation p roduct, and (3) a syneresis te s t involving capillary action.
O ther ty p es of grease, such as soda base an d lith ia base, v a ry widely in th e ir te x tu re , cohesion, an d adhesion, an d it is on such properties th a t th e lubrication o f a high speed bearing depends.
I t is w ith high speed a n d h igh-tem perature lu b rican ts t h a t th e need for p ractical testin g is clearly indicated.
I f a grease has no adhesion to m e ta l little lu b ricatio n is possible;
w hilst if it possesses good adhesion b u t little cohesion, high-speed ro ta tio n will cause th e bulk of th e grease to b reak aw ay from th e film adhering to th e balls a n d th ere will be no reservoir of lu b rican t available for th e replenishm ent of th e track s. I f b o th adhesion a n d cohesion are m oderately high a strong link is set up betw een th e film on th e balls an d th e bulk of th e grease a n d a continual circulation of lu b rican t ta k e s place.
T he question now arises as to how th e com bined cohesion an d adhesion properties are to be m easured.
I t is well know n th a t a lightly loaded bearing w hich contains no grease exerts little to rq u e on th e housing w hen it ro tates, b u t if grease is applied a high to rq u e is exerted during th e first ro tatio n . T he m ag n itu d e of this to rq u e depends to a large e x te n t on th e consistency of th e grease. D uring subsequent rotations, if th e grease channels, due to a low cohesion, th e
[Crow n Copyright Reserved
[To fa ce p . 250 J
Fia. 2.
to rq u e will fall rap id ly ; on th e other hand, if th e grease replenishes the track s th e torque will rem ain high and any fall will be due to a breakdow n of th e stru ctu re under shearing forces.
T orque m easurem ents therefore provide a m eans of studying th e com
bined effects of cohesion and adhesion.
A suitable a p p ara tu s is shown in th e accom panying photographs (Figs.
1 an d 2). I t consists of a B .R .L .020 bearing, B , m ounted in a drum , A , on a ro ta ta b le shaft. The torque tra n sm itte d to th e drum is m easured by m eans of a th rea d , D, connected to a calibrated spring torque m eter, E .
In use, th e bearing is filled w ith a know n weight of grease by forcing it in from one side. T he bearing is m ounted in th e drum and on th e sh aft in a n electrically h eated box. The box is h eated to a tem p eratu re of 37-8° C. (100° F .), an d m aintained a t this tem p eratu re for 2 hours. A t th e end of th e first hour th e sh aft is ro ta te d a t a speed of 600 r.p.m . and readings of to rq u e ta k e n after 15 seconds, 5 m inutes, and 60 m inutes.
The bearing is cooled, wiped on th e outside, a n d re-weighed for th e d eterm ination of grease loss.
The bearing is rem ounted in th e ap p aratu s a n d th e tem p eratu re raised to a value slightly higher th a n th a t to which th e grease is to be subjected in service. This tem p eratu re is m aintained for 2 hours. A t th e end of th e first h our th e sh aft is ro ta te d a t a speed approxim ating to th a t operating in th e equipm ent to be lubricated. Torque readings are not tak en during th is ro tatio n .
I n order to ascertain th e changes th a t have been produced a t th e higher tem perature, th e grease is cooled to 100° F . (37-8° C.), and torque readings are ta k en as before. The bearing is re-weighed for th e determ ination of the to ta l grease loss.
I f th e tw o sets of to rq u e readings are of a similar order it is clear th a t the grease has rem ained in th e trac k s an d th a t little change in cohesion and adhesion has tak e n place.
The rep eatab ility of th e te s t is good provided th a t th e bearings used are in a n unw orn condition.
The following results were obtained on a grease when using three different new bearings selected a t random from a batch of a dozen. The torque readings were ta k e n a t 37-8° C. and 600 r.p.m . both before and after ro ta tio n a t 176° C. a n d 2500 r.p.m .
PE T H R IC K : TH E T E STIN G OF G R E A SES FOR B A L L -B E A R IN G S. 251
(a) I n itia l ru n .
(b) A fte r h ig h te m p e ra tu re ru n .
B earing.
T o rq u e, gm . cm. G rease Losses.
15 seconds. 5 m in u tes. 60 m in u tes. R u n (a), % . R u n (6), % .
1 (a) 1067 640-5 427 40-3 --
(&) 1037 701-5 427 — 74-2
2 (a) 1037 610 457-5 37-9 —
(b) 1006 701-5 427 — 72-4
3 (a) 1098 610 457 37-9 —
(b) 1067 640 442 --- 72-4
In Fig. 3 th e results of tests on a num ber of greases have been p lo tted to show how th e te s t differentiates betw een channelling an d non-channelling
greases. . . ,
T he selection of th e m ost suitable to rq u e lim its m u st be influenced by th e knowledge th a t if th e cohesion is too high an abnorm al tem p era tu re rise will occur in service, owing to th e w ork w hich has to be done in over
coming th e cohesion. F o r th e p articu lar bearing used a value of 300-700 gm. cm. after 5 m inutes is p robably satisfactory, since greases having this to rq u e value have given good perform ances in service.
[Crown C opyright Reserved.
Fi g. 3.
C L A S S IF IC A T IO N O F G R E A S E S I N T O C H A N N E L L IN G A N D N O N - C H A N N E L L IN G T Y P E S B Y T O R Q U E T E S T S .
The o th er criterion of perform ance w hich m u st be considered when exam ining th e results of th e te st, is t h a t of ra te of change of torque with tim e. A change of 50 per cent, from th e 15 seconds value to th e 5-minute value is n o t excessive, b u t th e fall from th e 5-mirfute figure to th e final to rq u e should n o t be m ore th a n 40 per cent. T he changes produced by th e higher tem p eratu re ro ta tio n should be as sm all as possible. Values betw een 70 and 150 per cent, are n o t excessive. T he above te sts are pro b ab ly adequate for m ost purposes, b u t it is advisable to supplem ent th em by a te s t of longer duration.
A m ethod which gives valuable inform ation consists of churning one h a lf p o u n d o f th e grease, b y ro tatin g a bearing in it, a t a tem p eratu re ab o u t 20° C. below its drop-point. The bearing should be ru n a t high speed, an d th e te s t m ay be conveniently ru n for 7 hours. The grease is th en cooled an d inspected for liquefaction, oil separation, an d granulation.
The sam ple is th e n subjected to th e torque te sts previously described.
The results obtained b y such a test are illu strated by th e d a ta obtained on th e grease used in th e rep eatab ility tests m entioned earlier.
PE T H R IC K : T H E T E ST IN G OF G R E A SE S FO R B A L L -B E A R IN G S . 253
R u n .
T o rq u e, gm . cm. G rease Losses.
15 seconds. 5 m in u te s. 60 m in u tes. R u n (a), % . R u n (6), % . (а) .
(б) •
1174 1067
823 732
549 488
53-3
80-6
I t is obvious th a t th e grease has retained its basic stru ctu re and th a t its cohesion and adhesion have undergone little change.
F o r m ost greases th e above te sts should reveal any tendency tow ards instability, b u t one exception has been found. One grease, which behaved well in th e to rq u e tests, w hen h eated in th e presence of a copper strip showed a m arked degree of liquefaction. This shows th e difficulty of devising a te s t which is of universal application.
The te s t m ethods outlined above cover th e basic properties of m ost greases, b u t o th er tests are necessary if th e grease is to operate under special conditions, such as in th e presence of w ater or a t low tem perature.
W ater, b y em ulsifying w ith or dissolving constituents of th e grease, m ay radically change th e cohesion an d adhesion, w hilst high cohesion and adhesion will be undesirable a t low tem peratures.
The fa c t th a t a grease has sufficient cohesion and adhesion to give a reasonable perform ance in th e high-speed tests does n o t m ean th a t th e grease will cause undue trouble in startin g th e equipm ent a t low ground tem peratures.
The following values of th e pro d u ct of torque and tim e for th e first revolution produced by th is to rq u e were obtained on th e grease which has been used as an exam ple earlier in th is paper.
T e m p e ra tu re , °C. P ro d u c t o f T o rq u e a n d T im e, gm . cm . sees.
25 320-25
20 457-5
15 732
10 1,098
0 2,074
— 5 4,727
- 1 0 7,808
- 1 5 10,705
The bearing is reasonably free tu rning if th e to rq u e-tim e product does no t exceed a value of 20,000, so th a t th e grease will allow an easy s ta rt a t tem peratures down to a t least — 15° C.
I t is of in terest to n o te th a t if th e above results are p lo tte d on a semi-log.
scale, th e points lie fairly close to a stra ig h t line.
Sum m arizing, th e m ethods for testing w hich are ad v o cated are as follows :—
(a) Flow m easurem ents a t different ra te s of shear an d a t various tem peratures.
(b) O xidation te sts w ith an exam ination of th e oxidized product.
(c) Bleeding te sts involving capillary action.
(d) T orque te sts a t tw o speeds an d tem p eratu res.
(e) Churning te s t a t high speed an d tem p e ratu re.
(/) Special tests for p a rticu lar applications w hich involve conditions likely to m odify th e properties of th e grease on w hich its perform ance in th e above te sts depends.
255
A RATIONAL BASIS FO R TH E VISCOSITY IN D E X SYSTEM. PA RT I.
E . W . Ha r d i m a n, B.Sc., M .Inst.P et., and Al f r e d H . Ni s s a n, D.Sc., M .Inst.P et.
In t r o d u c t i o n.
Th e necessity for providing a satisfactory m ethod of comparing th e viscosity-tem perature characteristics, or change o f viscosity w ith te m perature, of oils has long been recognized, p articularly in th e field of autom otive and aviation lubricants, and has resulted in th e presentation o f a large num ber of suggested m ethods whereby this p ro p erty of an oil could be expressed. Chief am ong these are th e Viscosity In d ex system , the viscosity pole height, an d th e so-called A.S.T.M. slope. The Viscosity Index system ,1 which has been adopted b y th e In stitu te of Petroleum a n d Am erican Society for T esting M aterials Com m ittee D-2, is based on a com parison of th e viscosity of th e oil under te st a t 100° F . an d 210° E., w ith tw o sets of reference oil viscosity figures representing upper and lower extrem es in viscosity-tem perature susceptibility. Viscosity pole height,2 which has been in use in G erm any for several years, was based on th e assum ption, now found to be only p a rtly tru e, th a t when viscosity- tem perature relationships are p lo tted on a suitable scale to give a straig h t line, such straig h t lines for oils of th e same origin or chemical ty p e will m eet a t a common po in t know n as th e viscosity “ pole.” The distance of this “ pole ” from th e base line is know n as th e “ pole h eight,” and for ordinary m ineral lubricating oils th is value ranges from 1-0 to 6-0, in creasing values representing increasing steepness in th e viscosity-tem perature curve. A.S.T.M. slope characterizes an oil in term s of th e angle of the straig h t viscosity-tem perature line obtained b y plotting viscosity- tem perature d a ta on th e A.S.T.M. C hart for liquid petroleum products, D 341-39.3 This last system has lately increased in popularity in America, being used p articu larly for those oils having good viscosity-tem perature characteristics, i.e., showing flat viscosity-tem perature curves, such as aviation hydraulic oils, for which other m ethods are, a t present, less suitable. These three system s all have certain disadvantages and anomalies as a result of which no one of them has been universally adopted, although the Viscosity In d ex system is b y far th e m ost widely used of th e three.
The purpose of th e w ork described in th is paper was to m ake a careful exam ination of th e system m ost widely in use—nam ely, th e Viscosity Index system , a n d endeavour to eradicate those disadvantages and anomalies which present lim itations to its usefulness and universal accept
ance; in so doing an a tte m p t was m ade to derive a m athem atical basis for th e Viscosity In d ex system . In view, however, of th e very large background of testing an d research b uilt up using th e Viscosity Index system , which has now been in use f6r some fifteen years, it was con
sidered th a t an y modifications which m ight be found necessary should be such as to interfere as little as possible w ith th e present system between th e V .I. values of 0 a n d 100.
u
Vi s c o s i t y In d e x.
The viscosity index of an oil, as propounded b y E . W . D ean a n d G. H . B.
D avis in 1929,4 was an expression of its v isco sity -te m p e ratu re charac
teristics in term s o f its S aybolt viscosities a t 100° E. an d 210 E . In effect, it was a com parison of th e viscosity a t 100° E . of th e oil under te s t w ith those of tw o reference oils representing, respectively, 0 an d 100 on th e V .I. scale, each having th e sam e viscosity a t 210° E. as th e oil u n d er te st. T he basic d a ta used in developing th e system were obtained from S aybolt viscosity determ inations a t 100° F . a n d 210J F . on two series of lubricating-oil fractions, designated th e L series a n d th e H series, an d derived, respectively, from tw o crudes chosen a t th e tim e as exhibiting th e g reatest an d th e sm allest change of viscosity w ith tem p eratu re. The H series of oils, derived from a P en n sy lv an ian crude, was arbitrarily assigned th e viscosity index value of 100, a n d th e value of 0 was given to th e L series. B y p lo ttin g viscosity d a ta a t 100° F . ag ain st those values ob tain ed a t 210° F ., curves were o b tain ed for b o th th e L series an d the H series which were ex trap o lated upw ards to 160 S ay b o lt seconds at 210° F ., th e lower lim it being fixed a t 40 seconds a t 210° F ., as represent
ing th e low est p ractical viscosity d e term in atio n obtainable w ith th e Saybolt in stru m en t.
T he viscosity index o f a given oil was th e n calculated from the eq u atio n :—
V .I. = X 1 0 0 ...(1) where U = V iscosity (S.U.) a t 100° F . of the oil
u n d er test.
L an d H = Viscosities (S.U.) a t 100° F . of th e series L an d H oils respectively, having the sam e viscosity a t 210° F . as the oil un d er test.
I n 1932 revised basic d a ta for th e L a n d H series were published by D avis, Lapeyrouse, a n d D ean 4 for th e viscosity range betw een 50 seconds a n d 40 seconds a t 210° F ., since it h ad been found t h a t th e original data did n o t yield consistent results, p ro b ab ly due to th e lim itations in the accuracy of th e S aybolt viscom eter for efflux tim es of less th a n 50 seconds.
A t a la te r d ate th e need becam e a p p a re n t for extension in to th e viscosity region below 40 seconds (S.U.) a t 210° F ., a n d again because of th e limita
tions of th e S aybolt viscom eter, a new basic V .I. ta b le was presented by D ean, B auer, an d B erglund in 1940,5 based on kinem atic viscosities in centistokes instead of S aybolt seconds in w hich th e lower viscosity limit was placed a t 2-0 cs. a t 210° F . a n d th e u p p er lim it ex ten d ed to 75-0 cs.
a t 210° F . This change in viscosity u n its was p a rtic u la rly opportune in view of th e rap id ly increasing use of th e glass-type k inem atic viscometers for oil testing. T he extension dow n to 2-0 cs. was of a n em pirical nature, an d it was found necessary to m ake some slight a d ju stm en ts in th e range already revised b y D avis, Lapeyrouse, an d D ean in order to preserve a degree of uniform ity in th e curves. These revised d a ta form th e basis of th e Viscosity In d ex system as used to-day.
TH E V IS C O SIT Y I N D E X SY ST E M . PA R T I. 257 Li m i t a t i o n s o f t h e Vi s c o s i t y In d e x Sy s t e m.
A lthough th e Viscosity In d ex system as briefly described has m et with a very wide degree of acceptance in m any countries, it has become ap p a re n t th a t in its present form its lim it of usefulness has been reached, and fu tu re lubricating oil requirem ents are likely to call either for a new m ethod of expressing viscosity-tem perature characteristics or for some degree of modification to th e present system .
The chief reason w hy these changes are necessary arises from th e need for assigning viscosity indices to oils having b e tte r v isco sity -tem p eratu re characteristics th a n those represented by 100 on th e V .I. scale— i.e., b e tte r th a n th e original P ennsylvanian oils used by D ean and D avis for their work. Im provem ents in m ethods of refining lubricating oils and th e increased use of viscosity index im provers have resulted in th e production of oils having viscosity indices up to 140, or even higher. A lthough equation (1) perm its a viscosity index higher th a n 100 to be determ ined for a given oil, these values over 100 are, from th e n a tu re of th e equation, com paratively meaningless, as can be well illu strated in a curve relating kinem atic viscosity a t 100° F . w ith kinem atic viscosity a t 210° F . for sets of oils of different viscosity indices. Fig. 1 shows th a t as th e viscosity index increases from 0, th e cu rv atu re of th e viscosity relationship decreases u ntil a t some po in t betw een 120 a n d 130 a straig h t line is obtained. A t still higher viscosity index values th e cu rv atu re is reversed u n til a t 150 V.I. th e anom aly is d em onstrated th a t tw o oils having th e same viscosity a t 100° F ., b u t different viscosities a t 210° F ., would b o th have th e same V .I. of 150. F o r exam ple, tw o oils, b o th having viscosities of 80-0 cs. a t 100° F ., b u t viscosities of 15-5 cs. and 41-5 cs. a t 210° F ., are each assigned the same viscosity index of 150. T his failure becomes even more m arked w ith increasing V .I. W ith oils having com paratively high viscosities a t 210° F ., th e anom aly becomes ap p a ren t m uch lower down th e V .I. scale, and, generally speaking, it could be sta te d th a t th e Viscosity Index system breaks dow n a t 130 V .I. This particu lar feature of th e Viscosity Index system is generally realized, a n d Larson and Schwaderer, for exam ple, directed a tte n tio n to it in th eir paper on “ yiscosim etry and N ew G raphical Viscosity Classifications.” 6
A second disadvantage to th e present system , though n o t perhaps of such a serious n a tu re as th e first, is th a t V .I. is based on tw o fixed te m peratures, 100° F . a n d 210° F . A p art from th e inconvenience a n d th e possibility of error introduced b y th e need for extrapolating back to these two tem p eratu res in those cases where other tem peratures are more com
monly specified or used for viscosity determ inations, th e fixing of 100° F . and 210° F . as basic tem peratures imposes a lim it to th e ty p e of product to which viscosity index can be applied, an d those products, such as diesel oils, kerosines, an d gasolines, which have viscosities lower th a n 2-0 cs. a t 210° F . cannot be included in th e system . I t is suggested th a t th e possi
bility of applying viscosity index to such products would be of considerable interest, a n d m ight increase our knowledge regarding th e inter-relation betw een com position an d behaviour, and would provide a convenient m ethod w hereby products of different com position and origin could be distinguished. Such an extension in th e application of viscosity index
would be o b tain ed if an equation were derived b y m eans of w hich V .I could be o btained from viscosity d a ta a t an y tw o tem p eratu res.
Fi g. 1.
D E A N A N D D A V IS V . I . S Y S T E M S H O W I N G A N O M A L IE S A T H I G H V I S C O S I T Y I N D E X .
A th ird ra th e r u n satisfacto ry fe atu re of th e p resen t V iscosity Index system lies in th e a rb itra ry n a tu re of th e L a n d H series basic tables for th e range below 8-0 cs. a t 210° F . A description of th e m eth o d s by which basic L a n d H d a ta were determ ined will help to illu stra te th is point.