Introduction
In an instrument designed as a tool for
improving machine reliability it is of the essence that the readings
provided by MIDAS are themselves always reliable and meaningful. To this
end each MIDAS in supplied with a calibration tube with which the
calibration point may be confirmed as within allowable limits or requires
recalibration.
The nature of this calibration
technique is discussed here with a view to informing the interested user and
building confidence in the MIDAS debris analyser. Some topics relating to
MIDAS sensitivity are aired in 'sensitivity
considerations'.
Calibration standard
The calibration sample tube contains a
single ball bearing of 0.5 mm (+/-0.13 µm) diameter fixed in position and has the
nominal 2 ml filling of clean mineral oil. It permits MIDAS reading to be
directly related to a repeatable standard quantity, viz the mass of
ferromagnetic material. The balls used conform to the appropriate
standards for bearing steel material (and therefore its magnetic
properties) and tolerance. This forms a practical transfer standard. The
repeatability of this approach is supported by measurements taken with
many different balls from different manufacturers. Furthermore different
sizes of balls have been shown to give an output directly proportional to
mass. This confirms expectations from theoretical considerations and
magnetic simulation. Furthermore combinations of balls give rise to a
reading which is the simple sum of each individual ball's
reading (spaced apart).
It is woth mentioning that an alternative method to verify MIDAS calibration has been rejected.
A solution of paramagnetic ions could be used as the standard. This would have an advantage
in that a repeatable concentration could be made, occupying the whole tube homogeneously.
The demerit of that approach has is that such a paramagnetic solution is inherently temperature sensitive,
its susceptibility varying as the inverse of absolute temperature. The MIDAS reading then varies at ~-1%
for an increase of just 3K (degrees Celsius). MIDAS features very much better
temperature stability than this, typically less than
0.5% change in reading for 20K operating temperature change.
Relation to debris suspension
The magnetic field within the MIDAS sample
tube aperture is not the same along its length: it varies along the tube
length axis. Moving a single ball sample (of a suitably small size) along
the axis of the tube allows the output in that position to be plotted.
This data combined with the tube dimensions permits a numerical
integration of sensitivity across the entire sample volume to be
calculated. This calculation relates the point sensitivity where the
calibration sphere is positioned to the overall response for a
homogeneous suspension of particle occupying the sample volume. Any need
for precise sample filling volume is avoided as the field at the top of
the sample when inserted is much less than the peak field in the sensed
volume. For example, if the output reading at the nominal upper position
were one tenth of maximum for example, and the sample level varied by +/-
10%, then the total reading would vary by only around +/- 1%. Again the
experimental data has confirmed magnetic simulation results. The simple
volume / sensitivity calculation described above then relates mass of
(ferro)magnetic material per unit volume to the repeatable calibration
datum, avoiding the need to introduce a new and arbitrary 'MIDAS' unit.
An alternative would be to produce a reproducible transfer standard with
a homogeneous distribution of magnetic material. Whilst this would
avoid the implied reliance on similar field patterns in all MIDAS units,
it is in practice much a more difficult standard to consistently achieve.
Each calibration standard would than have its own calibration with the
disadvantage that each unit would demand an associated calibration tube,
and recalibration if that standard were lost or mislaid.
Calibration caveats
Calibration of this magnetic field based
instrument is naturally subject to the magnetic properties of the
material used. The relation of MIDAS reading to mass of material is
only strictly true for steel of permeability of the calibration ball. In
practice the readings for much ferromagnetic debris are likely to be
somewhat higher, mass for mass, due to typically higher permeability of the debris
compared with the intact sphere. Of course, this variation is of little
consequence when comparing similar installations or trending data.
Responding as it does magnetically, MIDAS has only low sensitivity to
most stainless steels, and is similarly responds relatively slightly to chemically
combined iron in the form of rust or from reaction of small particulate
with reactive species from additives or combustion products. These
factors can cause difficulty in correlating MIDAS readings with those
from spectroscopic iron contamination. (In fact as Fe ions are paramagnetic
MIDAS does respond to their presence, but at a much lower level mass for mass.) This is compounded by the well
known limitation of spectroscopy, in its under reading of particles of
greater than about 8 micron diameter. MIDAS is concerned with wear debris
measurement and so this is seen as a strength of the inductive method,
and a weakness of using a spectrometer in many circumstances. Orientation
of particles may also significant. Non-spherical particles with longer
axis along the sensor field read more that if across the field. The
random orientation of particles in an agitated fluid sample tends to
cancel the effect orientation has on the calibration. The calibration
also assumes that particles are randomly distributed throughout the fluid
sample, and have substantially not settled to the base of the
tube.
Conclusions
In the final analysis the preferred
calibration point provides a straightforward and repeatable means of
establishing a reference point for readings on all MIDAS instruments. The
approach taken avoids the introduction of an arbitrary unit, relating as
it does to mass of the calibration sphere. In daily use the main value of
oil debris analysis is as a trending and comparative tool and the actual
calibration unit is not of prime importance.