Click here to load reader
Upload
wd
View
215
Download
0
Embed Size (px)
Citation preview
140 ABSTRACTS OF PAPERS
QUANTITATIVE TESTS ON FINISHED MOULDINGS
By W. D. OWEN, Associate Member.
[ABSTRACT of a Measurements Section paper (based on Report Ref. B/T40 of the British Electrical and Allied Industries Research Association) whichwas published in December, 1947, in Part II of the Journal.]
Moulded insulating materials made from a given base by heatand pressure differ appreciably according to the type of mouldand the moulding technique used. To avoid anomalies from thiscause when comparing alternative materials, manufacturers ofelectrical equipment have adopted a standard test and test-specimen for each of the salient properties. The values obtainedin this manner are, by common consent, deemed to be those ofthe materials. It is recognized that these values may be higheror lower than those manifested by commercial mouldings, andthat it is sometimes necessary to know the actual values. Sinceit is usually not possible to apply the standard tests to non-standard test-specimens, a new set of tests and test-specimens isrequired, the latter being cut from commercial mouldings. Atthe request of the B.S.I. anjnvestigation was put in hand by theE.R.A. to explore how this might be done.
The whole of the investigation work was carried out at theUniversity of London, under the direction of Prof. W. John, byB. C. Fleming-Williams, Z. Rogowsky and D. J. Strong, whosubmitted a number of interim reports. These were consideredby a Technical Committee of the E.R.A. and subsequentlycollated by B. Shearman, who prepared an omnibus reportpublished by the E.R.A. as Ref. B/T40—1944, and by the B.S.I.as B.S. 1330: 1946. In this omnibus report the suggestion wasmade that the proposed tests "should be tried out under industrialconditions and revised in the light of experience." The presentpaper describes the work in an abridged and modified form andwas submitted as a basis for discussion.
With the object of determining the state of the art, a study wasfirst made of the published work of previous investigators and aquestionnaire was addressed to persons and organizations knownto be interested. These led to a useful bibliography and to anexhaustive study of the German Dynstat—a machine for applyingdynamic and static loads to miniature specimens.
The investigation began with a metrical examination of severalhundred commercial mouldings in order to ascertain whatpercentage of them yielded test-specimens of specified dimensions.The results are given in a series of statistical curves. Usefulinformation is also given on the preparation of specimens fromfinished mouldings.
The properties dealt with in current Standard Specificationsare tabulated, and a group of standard test-specimens is illustratedin order to facilitate comparisons. It is noted that only 9 of the17 principal properties are covered by the new technique, namely
Mr. Owen is with the British Electrical and Allied Industries Research Association.
resistance to impact and slow bend (using the Dynstat), crushingstrength, electric strength, surface resistivity, plastic yield,resistance to heat, specific gravity and water absorption.
For each of these an appropriate method of test is describedusing miniature test-specimens cut from commercial mouldingsas distinct from standard mouldings specially designed for thepurpose. In several cases a test-jig or machine had to bedevised. Some of the machines made up for the investigationare described and illustrated.
Work done on the Continent, previously, had led to the generaladoption of the Schopper Dynstat for impact and slow bend.This was fully exploited in this investigation as a time-savingexpedient, and consequently is featured in the report despite theconclusion that a simpler machine is recommended.
Owing to the system of grading by yield temperature used bythe electrical industry in this country (see Journal I.E.E., 1937,81, p. 553), the property known as plastic yield or heat stabilityis regarded as an important one. For this reason, two methodsof test were explored, one using a small specimen loaded as abeam, the other using a sensitive penetrometer. The relativemerits of each are discussed.
The paper includes a short dissertation on resistance tocrushing, with particular reference to the effect of the size of thespecimen and the effect of the position in the moulding fromwhich the specimen was cut.
Electric strength is determined by means of a ball-and-planetest, the specimen being a disc 8 mm in diameter and 1 mm thick,fixed to an anti-flashover shield suitable for a maximum voltageof 14 kV, determined by experience.
Provision is made for the determination of surface resistivityby setting up a potential difference between two Aquadagelectrodes drawn on the surface of the moulding with an ordinarysteel pen.
Heat resistance is an ambiguous term because there are severaldifferent heat hazards, ranging from the electric arc to radiantheat. Flame is convenient for test purposes but is not realistic.The investigators found the V.D.E. test a reasonably goodcompromise. This involves "contact with glowing hot body"—in this case a glow-bar heated to 950° C—and leads to a groupof five distinguishable grades.
Neither specific gravity nor water absorption presented anydifficulty.
The results are summarized in Table 1, which gives particularsof the degree of correlation with standard tests.
ABSTRACTS OF PAPERS 141
TABLE 1
Summary of Results
Property tested
Impactstrength
Cross-breakingstrength
Crushingstrength
Electricstrength
Surfaceresistivity
Resistance toheat (glow-bar test)
Plastic yield
Specificgravity
Water absorp-tion
Means or apparatus used
Dynstat with modifiedpendulum
Dynstat (unmodified)
Simple beam-typemachine giving loadup to 4"5 kg
Electrodes are a 2\ mmsteel ball and a brassplate respectively. Aspecial anti-flashovershield of ebonite issealed to the specimen
Aquadag line electrodesapplied with specialpen. Metal contactfingers and sensitiveballistic galvanometer
Glow-bar apparatus
Three-point load appar-atus
Vicat needle apparatus
Direct weighing andcomputation of di-mensions, or weighingin air and in liquid
Immersion and constanthumidity vessels.
Size of miniature specimen
Notched:—10 x 10 x 2 mm
Unnotched:—1 0 x 1 0 x 2 mm to10 X 5 x 1 mm
15 x 10 x 4 mm to10 x 5 x l-6mm
Recommended size:—10 X 5 x 2 mm
3 mm cube or cylinder
Disc 8 mm diameter andup to 1 mm thick
Surface of actual mould-ing between line elec-trodes 25 mm longand 2 mm apart
10 x 15 x 1 mm
5 x 10 x 1-5 mmor 10 x 10 X 1 mm
Disc 8 mm diameter,1 mm thick
Immaterial
Strip about 2 cm x 1 cmx 0-15 mm cut ascylindrical shavingfrom moulding, orpowder filed frommoulding
Percentage ofmouldings fromwhich specimenis obtainable
63%
63%97%
66%
63%
83%
85%
100%
Over 80%
85%70%
85%
100%
Nearly allmouldings
100%
Remarks
Striking distance shouldbe 1 • 75 timesspecimenthickness
Variation is largeAverage of a number of
tests should be taken
Tests may be made underoil, using shield, ifdesired
Surface resistivity varieswith gap width be-tween electrodes, andvariation differs fordifferent materials
Classificationintoclearlydefined groups reason-ably satisfactory
Short-time tests (tem-perature raised at50 deg C per hour)gave rough indicationof yield temperature.Long-time tests (stressapplied at constanttemperature) for moreaccurate work
Tests were made by im-mersion, by exposureto known relative hu-midity, and by expo-sure of powder toknown relative hu-midity
Correlation of results withstandard tests
No correlation expected orobserved
Cross-breaking strength canbe very approximately re-lated to all the dimen-sions. Hence approxi-mate correlation can beobtained
Approximate correlation isobtainedforgeometricallysimilar specimens usingformula Y = y + nylv', ifY=M30, and n = A/48,where h = height ofspecimen in mm
No correlation expected orobserved
Reasonable correlationwithin limits of experi-mental error
Since B.S. documents giveno classification, no cor-relation was possible
Reasonable correlation inshort-time tests, correla-tion to within 1 deg C inlong-time tests
Fair correlation in short-time tests, no correlationin long-time tests
Satisfactory correlation ob-tained
No clear correlation by anymethod
(Correlation not expected)