Alignment Tolerances Carved in Stone
Jane Alexander | December 19, 2018
Details of new ANSI standard provide clarity for an important industry issue.
The recently released Alignment Standard (ANSI/ASA S2.75-2017) from the American National Standards Institute, Washington (ansi.org) took nearly three years to develop. A committee of alignment experts discussed every aspect from safety procedures to the mathematics involved in defining the new standard. Alan Luedeking, CMRP, CRL, of Ludeca Inc., Miami (ludeca.com) a member of that committee, has been involved in the development of alignment standards for significantly longer than three years.
Luedeking remembers the “old days” well. Back then, personnel simply aligned components to the best of their abilities with straightedges or dial indicators. “Those alignments,” he said, “usually weren’t that good, due to sag, span limitations, obstructions to rotation, or whatever. But you did the best you could, and that was good enough because it was all you could do.”
In 1982, Ludeca introduced the world’s first computerized dial indicator alignment system (from the now-defunct Industrial Maintenance Systems Inc.), followed in 1984 by the world’s first laser-alignment system. With the improved measurement resolution and accuracy afforded by the laser sensor, a more precise definition of what constituted a good alignment became necessary. So, according to Luedeking, after poring through the sparse alignment literature that existed, Ludeca developed tolerance tables for short and spacer couplings, which, for lack of anything better, end users readily accepted. “Over time,” he noted, “these tolerance values came to be accepted as the U.S. industry standard and were adopted as the official tolerance standard by various corporate and government entities, including NASA and the U.S. Navy.”
So what are alignment tolerances, and why are they important? As Luedeking described them, “Tolerances exist because absolute perfection does not. No matter how hard you try and how long you work, you will never get a shaft alignment absolutely perfect.” He offered the following detailed discussion as to why, along with some expert advice on the meaning of the new standard and how it can help you improve your operations.
“Suppose you have a rusty steel block and your objective is simply to remove the rust. You take a sandpaper block and whoosh, a cloud of rust flies off. So you do it again. Another orange cloud of rust flies off. You repeat, over and over, until no orange cloud flies off. Now you switch to a finer-grain sandpaper and keep whooshing away at the block. Eventually you have a nice polished steel block that you could look into and almost shave in its reflection. Are you done?
Consider that if all you wanted to do was remove the rust, you went too far. You could have stopped when you had the rust-free gray block. Still, you went ahead and polished it to a near-mirror finish. You got it way better than it needed to be and expended much time and energy doing so. But wait: Suppose you were grinding the mirror of a space telescope? Now this block would still be terrible. You’d have to scrape a few million more times with ever finer media until you had a mirror so perfect it might scare you to look into it. Yet, if you took a good microscope and looked closely at that surface, you’d still see many peaks and valleys. The surface wouldn’t be absolutely, perfectly flat, and no matter how hard you tried, and how long you worked, you’d never get it there.”
“All of this,” Luedeking said, begs the question, ‘when is enough enough?’ How good is good enough? The same applies to alignment efforts, which leads us to try and find a way to accurately quantify and describe misalignment so that it makes sense to all and allows us to define acceptable limits.”
Luedeking encourages personnel to consider how they perceive shaft misalignment. In short, if one shaft is higher than or displaced to the side a bit from the other, it’s said to be in an “offset” condition. If the shafts aren’t parallel to one another, meaning they’re angled to each other, the condition is referred to as “angularity,” i.e., nothing more than the rate at which the offset changes between the shaft centerlines as you travel along their lengths.
“By defining how much offset we’re allowed to see at the coupling,” Luedeking explained, “and how much this offset is allowed to change from that point on (the angularity), we create a tolerance envelope.” (See Fig. 1)
As he pointed out, suppose a machine that’s to be aligned is allowed to have as much as two- thousandths of an inch (0.002 in.) offset vertically and horizontally. “Specifying that the machine may separately be allowed to have this much offset both vertically and horizontally at the same time,” he said, “means the shaft centerline is actually allowed to be more than two thousandths offset. This is a called a standard tolerance definition.” (Tolerance limits or boundaries are defined by the square shown in Fig. 2.)
If, though, the intent is that the shaft never be misaligned by more than two-thousandths offset in any direction, the tolerance limit, i.e., boundary, is defined by the circle in Fig. 2. Referred to as “vector” tolerance, this reflects a more conservative alignment approach than “standard” tolerance.
“Of course,” as Luedeking put it, “this is just the tip of the iceberg. You could also specify alignment tolerances in terms of sliding velocities of the internal moving components within a flexible coupling, or by other geometric constructs.” The complexity of the overall issue, he said, is why it took the alignment committee three years of painstaking discussions to reach a consensus and produce the ANSI tolerance tables that were ultimately published in the new standard.
Since virtually all alignments are performed by millwrights and mechanics, not highly trained engineers or mathematicians, two simple tables were deemed the best way to define tolerances. Their three-tier formats cover minimal, standard, and precision alignment values, related to machine rpm. As Luedeking observed, the faster machines turn, the tighter alignment must be (see Table I, based on information published as Table 4 in the ANSI standard). Table I illustrates the tolerance values for “short” couplings or flexible couplings with distances between flex planes of 3 in. or less. The “minimal” standard actually represents the “maximum” values by which shafts can be misaligned and still be minimally acceptable.
The values are presented as a maximum permissible offset overall, and a maximum permissible angle between the driver and driven shaft centerlines of rotation at the short coupling center, grouped into the three referenced tiers (minimal, standard, and precision) and segregated into four speed ranges defining the minimum rotational speed for which these values should be applied, up to the next higher speed. In other words, from 1,800 to 3,599 rpm, apply the values in the “1,800” column.
Regarding Table I, Luedeking noted that the original description of the information published in the ANSI Standard as Table 4 was in error because it referred to the values as applying to coupling spans “greater than” 3 in. between flex planes. Table I above correctly notes the values in inches apply to coupling spans between flex planes of 3 in. or less. All true flexible couplings have two flex planes. The key to understanding how to properly apply tolerances to them is to be cognizant of the distance between these planes. If it is greater than 3 in., the coupling should be treated as a spacer coupling, not as a short coupling, and the commensurate tolerances applied. The mathematical justifications for this are beyond the scope of this article but were considered and taken into account by the ANSI panel in charge of this topic. Thus, a second tolerance table was published, with suggested values for spacer couplings (see Table II here, based on information published as Table 3 in the ANSI standard). What does this mean?
Consider the distance between the flexible-coupling flex planes in Figs. 5 and 6. The gear couplings in these two figures are essentially identical, with the only difference being in the distance between their flex planes, or the locations where the outer floating hub interacts with the inner solid hub, i.e., at their respective ring gears. This means any offset and angle between driver and driven shafts is converted to angularity at each flex plane.
Referring to Fig. 7, Luedeking said, “Since all misalignment between driver and driven shafts is nothing but angularity at each flex plane, that’s all you have to limit.” The longer the spacer length the more overall misalignment you can accommodate at the same maximum permissible angles, and that’s the reason why some machines are installed with long spacers between them. “They grow a lot due to thermal expansion, or need that extra cushion to absorb dynamic movement.” In the end, he concluded, “This means you can define your tolerances in terms of maximum permissible angularity at each flex plane.” That’s what the information in Table II does. Apply these values any time the distance between flex planes is greater than 3 in. EP