Turning Or Milling: Process Choice For Aluminum Flange Parts

The Pain Points

Process choice on aluminum flange parts is usually made by habit rather than by analysis. A shop with three lathes turns every flange, including the bolt circles, which means interpolating holes with a boring bar or living with the runout that a second operation introduces. A shop built around vertical mills mills every flange, including the sealing face, which means facing a disc that could have been turned in a fraction of the time. Both shops believe they are being efficient, and both are quietly paying for the decision in cycle time, tolerance stack, and scrap. The symptoms are easy to recognize: bolt holes that sit a few hundredths off the pitch circle because they were drilled after the part was re-chucked; sealing faces that are flat in the fixture and dished when released; a drawing that calls for 0.05 mm concentricity between bore and bolt circle that the process cannot hold without a fourth operation. Worse, the choice is often locked in at quotation, when nobody has time to model the routing, so the shop bids a price based on the wrong machine and discovers the cost only when the parts run. On a component as simple as an aluminum flange, process routing is the difference between a two-operation part and a four-operation part, and four operations rarely survive a competitive quote.

The Principle

The geometry of a flange explains the routing. A flange is a rotationally symmetric body: an outer diameter, a bore, a sealing face, a back face, a groove or a raised boss, and a pattern of holes on a bolt circle. Turning is the natural process for every feature that shares the axis, because a single setup produces the outer diameter, the bore, and the face perpendicular to them with concentricity that is built into the machine rather than chased by the operator. Milling is the natural process for every feature that does not share the axis: bolt holes, slots, lugs, milled flats, cross ports, and lightening pockets. The reason to respect that division is error accumulation. Every additional setup adds a new datum transition, and each transition carries chuck runout, fixture location error, and the spring of a part that was held one way and released another. On a thin aluminum flange, the clamping itself is a source of error: three-jaw chucks distort a flexible rim, so the bore machined in that condition is round only while the jaws are loaded. Milling, by contrast, loads the part axially against a plate, which is gentler on a thin disc but slower for facing a full diameter. The process that wins is therefore the one that completes the most features in the fewest setups while keeping the cutting force aligned with the part's stiffness.

Equipment Classification

The routing options form a ladder. A two-axis CNC lathe with a collet chuck or precision three-jaw is the cheapest way to turn faces and diameters, and for a simple flange with four holes it is often enough when combined with a drill fixture. A lathe with live tooling and a C-axis adds the ability to drill and tap the bolt circle without a second machine; the holes are positioned by the same spindle that turned the bore, so concentricity is excellent, but live-tool stations are slower and less rigid than a dedicated mill. A mill-turn center, with a B-axis turret and a subspindle, completes the whole flange in one setup and is the right answer for families of flanges with tight position tolerances or for volumes that justify the capital. On the milling side, a three-axis vertical machining center with a rotary table or a trunnion handles bolt circles on larger flanges; a five-axis machine adds the ability to reach both faces and the OD in one clamping. Workholding decides as much as the machine: collet chucks and machined soft jaws hold a turned flange concentrically without distorting it, vacuum plates hold thin discs flat, and a dedicated fixture plate with a locating bore lets a mill pick up the turned datum repeatably. The measurement layer, a CMM or a dedicated runout fixture, is what proves the routing actually delivered the tolerance.

Practical Guide

Build the routing from the datum backward. First, choose the face and bore as the primary datum pair, and machine them in the same operation so their relationship is a machine error, not a setup error. Second, turn the first side: face, outer diameter, bore, and any groove or step, holding the part on the as-supplied stock or on a rough-bored hub. Third, turn the second side on soft jaws machined to the finished diameter, or on a collet, so the part is located on the diameter that was just cut rather than on raw stock. Fourth, if the machine has live tooling and the tolerance allows, drill and tap the bolt circle in the same setup, referencing the spindle axis. Fifth, if a separate mill is needed, locate on the finished bore with a plug or a fixture plate, clamp axially, and drill the pattern in one cycle so every hole shares a common datum. Sixth, control clamping force: on flanges thinner than 8 mm, use distributed vacuum or low-pressure air chucks, because jaw pressure alone will dish the face. Finally, inspect concentricity and position off the machine, on a CMM or a purpose-built fixture, and record the result against the setup that produced it, so the next order starts from measured routing rather than memory.

Field Experience

A shop producing 250 mm aluminum flange parts for a pump housing ran a two-machine route: turn the OD, bore, and faces on a lathe, then drill eight bolt holes on a vertical mill located by the bore. The drawing asked for 0.08 mm true position on the hole pattern and 0.05 mm concentricity between bore and outer diameter. The route held concentricity comfortably but produced periodic true-position failures around 0.12 mm, always on the holes furthest from the locating plug, because the plug had clearance and the part was clamped on a rim that had already sprung. The fix was not a better operator but a different route: the shop moved the family to a mill-turn center, completed the faces, bore, and bolt circle in one setup, and the true-position spread collapsed to 0.03 mm. Cycle time fell by 18 percent as a side effect, since the second setup and the re-fixturing disappeared.

A second case went the other way. A thin 4 mm flange was being turned in a three-jaw chuck at 40 Nm of jaw pressure; the bore measured round on the machine and 0.04 mm oval after release, which pushed the bolt circle out of position when the part was milled later. The shop switched to a collet with a machined stop and reduced the jaw pressure, then milled the holes against the bore with a low-profile axial clamp. Both the ovality and the position error disappeared, and the operator learned a rule that now appears on every traveler: on aluminum flange parts, the clamping force is a process parameter, not a convenience.

Summary

Turning and milling are not competing religions; they are two halves of one routing. Turning owns the features that share the axis, milling owns the features that do not, and the cheapest capable process is the one that completes both in the fewest setups. Every extra setup adds a datum transition, and on flexible aluminum flanges a datum transition is where tolerance goes to die. A shop that routes by geometry rather than by habit cuts cycle time, holds position, and stops explaining away the same defect on every order.

Outlook

Routing is moving toward single-setup completion. Mill-turn centers with integrated probing and bar feeders are becoming affordable at the small-shop scale, and CAM software now simulates the whole flange, including the second-op soft jaws, before the first chip is cut. On-machine probing of the bore and face lets the control correct the bolt-circle position without an operator touching an offset. In the near term, the practical advantage will belong to shops that can prove their routing with measured capability data, because buyers of aluminum flange parts increasingly ask for a process flow diagram and a first-article report rather than a promise. The shop that can show a two-setup route with a Cpk above 1.33 on true position will win the order from the shop that still routes by tradition.