~ an Advanced Laboratory Imaging and Analysis System
~
ALIAS Accuracy
Cascade Data Systems finds that, under proper conditions,* ALIAS typically provides measurements for the average sample length and average sample diameter that are comparable to the measurements that can be obtained with calipers.** * - Proper Conditions:
1. The number of particles being scanned is large (preferably up to around 200 particles).
2. Particles in a given sample all have the same geometric shape (cylinder, trilobe, bilobe, quadlobe, or sphere). Particles in a given sample all have approximately the same diameter (within +/- 10%) and are of the same species of catalysts. (Mixing different species of catalysts--say two catalyst types that have different color, or shapes, or typical diameters--can degrade ALIAS accuracy. If you have a mixed catalyst sample, Cascade Data Systems recommends you separate the particles into their various species and measure them separately.)
3. The operator has ensured that the dimensions of the individual particles have been properly assigned according to which is length and which is diameter.
4. Particles are uniformly scattered on the scanner (but no closer than 1.5 inch to either the left or right edge of the scanner glass). All particles should be laid on their sides, an issue that typically arises only when measuring short particles, ones that have a chance of coming to rest on their ends.
5. No particles are touching each other.
6. Particle diameters lie within the range for which ALIAS is tuned (between 0.85 mm and 3.5 mm).
7. Catalyst particles also need to have fracture geometries and other properties that are similar to those catalysts on which ALIAS was tuned. Particles should be smooth along their long sides, but may have some jaggedness on either end. The more jagged the ends of a particle, the tougher it is to ascribe an accurate length measurement to it.
8. The Standard Error (the standard deviation divided by the square root of the number of particles in the sample) of the sample should be small for the dimension of interest.
9. The ALIAS hardware and ALIAS software has been properly installed and the scanner calibration information has been properly entered.
Additional Notes:
--At all times ALIAS strives to reproduce the measurements that a trained technician would achieve when measuring samples with calipers.**
--As the diameter of particles increase, typically so does the magnitude of the difference between caliper-measured length and ALIAS-measured length. (And also, as the diameter of the particles increase, the variability from one set of measurements done with calipers when compared to another set of measurements done with the same calipers also tends to increase: this is due to the particles sitting differently within the narrow jaws of the calipers.)
--ALIAS was tuned using about twenty catalyst samples that are taken to be representative of catalysts in use by the oil industry. Obviously, though, these catalyst that have been studied cannot be guaranteed to be adequate representations of all possible catalyst species that might ever be scanned with ALIAS. As such, Cascade Data Systems recommends that whenever a user scans a new catalyst species--one that they haven't yet scanned with the current version of ALIAS--that they take extra steps (as described in the ALIAS Help file) to ensure that ALIAS is achieving the accuracy that they desire. Some catalyst characteristics that might pose measurement challenges are catalysts with highly reflective (high-luster) surfaces, catalysts with severe curvature and twisting, catalysts with significantly different cleavage geometries from the fractured cleavages that were studied (other types of cleavage geometries that weren't studied would include catalysts with rounded, pointed, crisply cut, or eroded end geometries), catalysts with different surface coarseness than what was studied, catalysts that vary significantly in diameter along the length of a given particle, etc.
** - When comparing caliper measurements and ALIAS measurements, we suggest you use the following procedure:
1. Have at least three different caliper-trained technicians measure the same (sufficiently-large) sample with calibrated calipers;
2. Ensure no chipping or breakage of the catalyst occurs during or between measurements;
3. Average the results of these three (or more) samplings together.
NOTES:
--It's common for caliper measurements of sample average length to differ by 0.1 mm or more from measurement to measurement, and for measurements of sample average diameter to differ by 0.05 mm. Multiple measurements are therefore encouraged. The variation in sample average measurements when using ALIAS is typically seen to be smaller than the variation in sample average measurements when using calipers, even when particles are redistributed between successive ALIAS scans. And ALIAS sample average measurements are normally operator-independent if the proper conditions discussed above are met.
--As any technician who has ever used calipers to measure the lengths and diameters of catalysts can tell you, it can be quite tricky determining what the actual diameter and actual length of a given particle is. As the jaws of the caliper close on a catalyst to measure its length, either or both of the jaws might fall into a fracture crevice, or on a fracture peak, or anywhere in between. The particle also might not be perfectly parallel to the long axis of the calipers, and this is particularly difficult to achieve if the particle has any curves or twists in it, features which are quite common. Similarly, when measuring diameters, the calipers might not be placed precisely perpendicular to the long-axis of the particle at the point where the calipers touch the particle, or the calipers might not be placed at the exact midpoint of the particle, or the calipers might have been closed a little too strongly, causing the sharp jaws of the calipers to dent the particle slightly. This problem is exacerbated with trilobes, where there are three possible diameters to be measured; or with quadlobes, where there are four possible diameters. And even a cylinder might yield a different diameter if the particle is rotated inside the jaws of the calipers.
--Due to the complexities of measuring dimensions, and the countless deformations from the straight and true that may occur in a single particle of catalyst, Cascade Data Systems can make no absolute claim about accuracy. That being said, know that Cascade Data Systems has worked hard to make ALIAS as accurate as we know how for the particle species that we have calibrated against, leveraging our 25 years of experience with the ALIAS product to refine and improve our algorithms. All that's left is for ALIAS customers to investigate the results that ALIAS provides (through the process described in the ALIAS Help file) and see if they are of sufficient accuracy for their needs.
Cascade Data Systems specifies no accuracy thresholds for area, curvature and color values:
1. Area is the number of pixels each particle covers multiplied by the area of each pixel.
2. The Curvature algorithm tells, in increments of diameter, the amount by which the particle deviates from the straight and true. If a particle has a curvature of 0.33, then it differs from a straight line by an amount equal to a third of the average diameter of the particle. In layman terms what this means is that if you wanted to construct a box to hold the particle, you would have to make the box 33% wider than the average diameter of the particle in order for the particle to fit into it. Or looked at another way, the curvature value gives you an approximation of the gap that would form between the center of the particle and a tabletop if the particle curved with a nice smooth arc, and if that particle were laid on its tips so that it formed an arch. The particle in the image below has a reported curvature of 0.418, or a bounding box that needs to be 41.8% taller than the average particle diameter in order to accomodate the particle.
3. The Roundness measurement, which only applies to spheres, is the standard deviation (of "a sample", that is, by using the "N-1" denominator inside the square root) of several diameters drawn at uniform angular increments through the center of the sphere. (The number of diameters used can be determined by counting the number of yellow diameter lines--or "double" spokes--drawn through the particle after it is scanned, highlighted and magnified in the Scan window.) The roundness of the sphere shown below was computed using eight different diameters.
