Shops that machine aircraft parts have traditionally been asked to hold some of the tightest tolerances in the metalworking industry. With its Accurate Fuselage Assembly (AFA) program, Boeing now has upped the ante by requiring ultra-tight tolerances that allow fuselage components to literally snap in place like Lego parts.

To meet these increased accuracy demands for "snap-together" assembly, Northrop Grumman's Applied Digital Technologies Group (ADT) not only bought a new machine tool, but fitted it with an MP10 machine-tool probe from Renishaw Inc. (Schaumburg, IL) for in-cycle gaging of critical assemblyhole locations, reducing scrap and speeding throughput of the flightcritical components.

"Our working tolerances used to be 0.03" (0.76 mm] or better on hole location," says Jeff Howard, ADT manager for Northrop Grumman's Dallas-based Aerostructures Business Area. "To comply with Boeing's AFA program, we now must work with 0.01" [0.25 mm] true position, which is really half of 0.01" ."

These tight specifications tested the limits of conventional sheetmetal frame-fabrication methods. The challenge was to find a way to reduce the variability on formed sheetmetal to a level comparable to CNC machined parts.

Called "Z-frames," the parts in question are arc sections of the frame that make up the barrel for the 747 fuselage. Typically 10-18' (3-5.5 m) long, they are made from roll-formed sheetmetal that is subsequently stretch-formed, then trimmed and drilled. They have a Z-shaped cross section and are 0.063" (1.6-mm) thick.

When Boeing reduced the design tolerance as a part of AFA, Northrop Grumman's existing frame supplier could not produce parts to meet the specifications. After the forming and extruding processes, a typical sheetmetal part's dimensions varied widely. One day, parts might be on the high side of tolerance, and the next day, parts were on the low side. "If you're trying to hold them in a fixture, the best way you can do that is hold them at nominal," Matt Turner, Northrop Grumman Engineer, notes. "For example, say we have a raw extrusion that has a 0.03" tolerance and we're trying to put a hole in the center with a 0.01" true position. If the part is on the high or low side, and you're holding one end at nominal, you're not going to cut a good part."

As a result, Northrop Grumman decided to in-source the frame fabrication, and purchased a Komo three-axis router, measuring 28 (8. 5-m) long, 16 ' (4.8-m) wide, and Z of 24" (0.6 m), to produce the parts in compliance with AFA. On paper, the machine's tolerances showed it had the accuracy to do the job. Because of thermal changes and setup difficulties, however, the large sheetmetal parts could never be fixtured in a way that allowed Northrop to accurately or repeatably drill holes within specification.

That's where the MP10 probe came into play. Rather than setting up the part and probing the fixture, the fixtured part is measured with the probe before cutting, making machine axis compensations to put the part into the center of the tolerance band. "If the part is too wide or necked-down, we can compensate," says Turner. "There was no way we could do this through fixturing." With the new machine and Renishaw probe, Northrop has cut about 1700 parts to spec. "We establish, before machining, that the part can be a good part after cutting," he adds. "We don't waste time on a raw part that can't meet spec."

An MP10 touch-trigger probe acts as an omni-directional switch, effective in the +/- , Y and Z directions. The probe is automatically spindle-loaded, then driven against the workpiece or fixture. Each contact generates a signal, relating the contact point to a common datum.

Repeatability of the probe is +/-0.00004" (1 mu/m).

Northrop Grumman uses two MP10 probes on the Komo router. One uses a 2-mm stylus-smaller than traditional machine styliwhich the company made by necking-down a CMM stylus. "Our smallest holes have a diameter of 0.098" [2.5 mm] so we needed the 2-mm

[0.05"] styli." Turner describes the holes as "determinant" holes used to bolt the parts together, while the rest of the assembly is riveted together along mating surfaces.

"We have parts with as many as 50 holes," says Turner. "Right now, we're doing 100% inspection of the part with the probes, too, but we're planning to move to sampling as we firm up our SPC data and prove that we have a repeatable process. We're at a point of confidence where the post-machining probing will probably go away or be minimized," says Howard.

A second MP10 probe with a larger stylus probes the outer contour of the barrel. Those data are sent electronically to AFA program engineers for analysis on how any given part will mate up with other parts. The information is unique to individual parts, which are serialized with a work-order number.

"There is strong angularity on the outer contours, and to make sure we're not hitting the shank of the probe, we have to use the larger stylus," says Howard.

Northrop Grumman is currently working on software to streamline the process. It will take real-time SPC data on the floor and give operators instant feedback, and it will FTP the data directly to the customer, transparent to the operator.