What Is Rapid Traverse in CNC Machining?

• 1
  What G00 actually commands on the machine
• 2
  How the controller plans each rapid move
• 3
  Rapid rate, acceleration and the real travel time
• 4
  When a rapid move is the wrong choice
• 5
  How we apply it on 5-axis and mill-turn work
• 6
  Rapid move versus cutting feed

A rapid move is the motion a machine makes when no material is being removed. In G-code it is G00. The tool leaves its current position and travels to the next programmed point as fast as the drives allow. No feed rate is commanded, so the machine uses its own rapid rate, typically given in the machine manual in mm/min or m/min.

The distinction matters because the controller treats rapid and feed moves differently. A G01 move follows the programmed feed rate and is expected to cut. A G00 move is expected to arrive. Feed-rate override on the operator panel usually changes G01 speed but leaves rapid rate alone, or scales it on a separate switch.

On a typical vertical machining center the axis drives hit their rapid rate only on long moves. Short hops between nearby holes never reach it. Acceleration and deceleration eat most of the distance. That is why two programs with the same number of moves can post very different cycle times.

Most controls plan G00 as a straight line in machine coordinates, then clamp each axis to its own maximum velocity. The slowest axis limits the move. On a diagonal path the tool still travels in a straight line, but the axes finish their travel at different moments unless the control is in a synchronized rapid mode.

Some builders default to a dogleg path instead. Each axis moves at its full rate, so the tool reaches the target in less time but traces a two-segment path. On a machine with a tall fixture or a rotary table, that path can clip a clamp. The fix is a G00 mode change in the parameters or an explicit intermediate point in the program.

Look-ahead changes the picture again. A modern control reads 100 or more blocks ahead, blends deceleration out of one move into acceleration of the next, and never fully stops at a corner if the following move allows it. In tight tool paths this blending saves more time than a higher rapid rate would.

Rapid rate itself is a parameter, not a law. Many builders ship a conservative default and let the integrator raise it after a ballbar test. Raising it without checking the servo tuning usually shows up as a servo alarm or a lost position, not as a faster cycle.

Published rapid rates on a 3-axis mill often sit between 20 m/min and 48 m/min. That figure describes steady-state travel. A 50 mm hop on a machine that accelerates at 5 m/s² takes far longer than the steady-state math suggests, because it spends most of the move speeding up and slowing down.

The useful formula is short. Time equals twice the square root of distance divided by acceleration, as long as the move never reaches top speed. For a 50 mm hop at 5 m/s² this works out near 0.2 s, and the tool covers those 50 mm at an average speed well under the rated rapid.

That is why adding a tool change or a longer clearance plane rarely costs what the programmer expects. A 100 mm hop at the same acceleration is roughly 0.28 s, not double the 50 mm figure. Acceleration dominates until the move gets long.

For a 4,000 mm machine travel on a large gantry, the math flips. A move of 1,000 mm or more does reach the rated rate, and there the published number is the honest one. On small parts, ignore it.

G00 is safe only when the path is clear. The control does not know where the clamps, the vise jaws or the finished walls are. It moves in a straight line between two points and trusts the programmer. Any fixture that sits above the part or beside it must be modeled into the setup sheet.

Never use G00 to approach a surface. The standard approach is a rapid to a clearance plane, then a feed move down to the cut. A clearance of 2 mm to 5 mm above the stock is common. Approaching at rapid rate has no margin, and the first contact is at full speed.

Plunge moves are the other trap. A G00 into the stock will break a small cutter, stall a spindle or scrap the part. Use G01 with a programmed plunge feed, usually 30% to 50% of the side-cutting feed for a carbide end mill.

There is also a thermal and wear cost. Every rapid is a full-torque acceleration and a full-torque stop. On a machine running three shifts, the rapid moves outnumber the cutting moves in many programs. Ball screw wear and guideway load follow move count, not spindle hours.

Materials do not care about rapid speed, but geometry does. A long, thin tool that survives a feed cut can still snap if it rapids into a wall at 30 m/min. The move that breaks it is not the cut.

GreatLight runs 127 high-precision CNC machines across three wholly-owned plants, including 16 simultaneous 5-axis machining centers and 16 mill-turn centers. On 5-axis work the rotary axes change the rapid picture. A 0.5° rotary move paired with a 100 mm linear move produces a tool-tip speed that neither axis rate describes on its own.

We post with a clearance plane and a retract plane set per setup, not per program. On a deep cavity the retract is small and the clearance is tight. On an open plate the clearance can be generous because the tool never travels far. This keeps the air-cutting part of the cycle near its floor.

For parts held to ±0.005 mm, thermal drift over a long cycle is a real variable. Shortening the non-cutting time reduces the number of heat cycles the spindle and ballscrews see. It is not only a cost issue. Cycle time and dimensional stability move together.

On mill-turn centers, rapids between the turning and milling stations are the main hidden cost. We check the posted air-cut percentage on every new program, and we re-post rather than accept a path that spends more time moving than cutting. Typical production work here runs 3 to 5 days from released program to shipped parts, with 100% inspection before shipment.