I was twenty-two years old and six months into my first real job in a machine shop. The shop specialized in hydraulic components for construction equipment. I had been trained on three gun drilling machines and had run about two hundred parts without any serious problems. The foreman gave me a job that I had done before: drill a 12mm diameter hole through a 600mm long 4140 steel shaft. The shaft was for a hydraulic cylinder rod. The material was annealed to about 28 HRC. I had run this exact job three times in the previous month. I was confident, maybe too confident.
The Setup
The machine was a single-spindle gun drilling machine built in 1987. It had a 30 horsepower spindle drive and a coolant system rated at 80 bar maximum pressure. The coolant tank held about 800 liters of oil. The coolant pump was a positive displacement piston pump that had been rebuilt twice in the previous year. I had checked the oil level at the start of the shift and it was full. I had checked the filter pressure gauge and it showed 2 bar, which was normal.
I installed a 12mm diameter single-flute gun drill with a carbide tip. The drill had a total length of 800mm and a flute length of 600mm. I set the guide bushing to 12mm and aligned it to within 0.03mm of the spindle centerline using a dial indicator. I set the spindle speed to 3500 RPM, which gave me a cutting speed of roughly 132 meters per minute. I set the feed rate to 0.06mm per revolution, which was a standard parameter for 4140 steel at that hardness.
What I did not check was the coolant pressure at the spindle outlet. The machine had a pressure gauge at the pump, but there was a separate gauge at the spindle rotary union that I normally checked before every job. I had skipped that check because the previous operator had run the machine that morning and I assumed the pressure was fine.
The Warning Signs
The drill entered the shaft and started cutting normally. The first 100mm produced good chips: short, broken segments about 1mm thick that flushed out of the hole cleanly. I was watching the spindle load meter, which stayed steady at around 12 percent. Everything looked fine.
At about 200mm depth, the chips started changing. They came out as long, stringy ribbons instead of broken segments. The ribbons were about 50mm long and curled tightly. I should have recognized that as a sign of inadequate coolant flow. When coolant pressure drops below the level needed to fully lubricate the cutting edge, the chip formation changes from shear to tear, and the chips come out stringy instead of broken. I did not recognize it at the time. I thought the material had a hard spot.
At 350mm depth, the chip ribbons started coming out tangled. I noticed the coolant flow at the chip tray looked thinner than usual. The stream was intermittent rather than continuous. The spindle load had climbed from 12 percent to 22 percent. I checked the coolant pump pressure gauge and it read 75 bar, which was within range. What I did not know was that the pressure gauge at the pump was not showing the pressure drop across the rotary union seal, which had started leaking internally.
| Parameter | Normal Range | Value at Time of Crash | What It Indicated |
|---|---|---|---|
| Coolant pressure at pump | 70-80 bar | 75 bar | Appeared normal |
| Coolant pressure at spindle | 65-75 bar | 40 bar (estimated) | Not checked |
| Coolant flow at chip tray | Steady stream | Intermittent | Restricted flow |
| Spindle load | 10-15% | 22% | Rising tool load |
| Chip form | Short, broken segments | Long, stringy ribbons | Inadequate lubrication |
| Flute clearance | Full | Packed with chips | Evacuation failure |
The Crash
At 380mm depth, the spindle load jumped from 22 percent to 55 percent in about three seconds. I hit the feed hold button. The machine stopped advancing but the spindle was still running. I heard a high-pitched squeal from the drill, which was the sound of the carbide tip rubbing against packed chips in the flute. About one second later, the drill snapped at the shank, about 20mm above the guide bushing.
I stopped the spindle and stood there staring at the broken drill sticking out of the shaft. The foreman walked over, looked at the machine, and looked at the chip tray full of stringy ribbons. He did not yell. He walked to the coolant pressure gauge at the spindle rotary union and tapped it. The needle did not move. He turned the isolation valve and the gauge jumped to 75 bar. The valve had been partially closed, likely bumped by the previous operator during a tool change.
The foreman turned to me and said: “Check the pressure first. Always.”
The Extraction Process
It took us about three hours to extract the broken drill from the shaft. The drill had snapped about 20mm above the guide bushing, leaving about 580mm of the tool stuck inside the 600mm bore. The carbide tip was wedged at the bottom of the hole with packed chips surrounding it. We tried pulling the drill out from the entry end first, using a set of pliers on the exposed shank. It did not budge. The chips had packed so tightly that the drill was effectively locked in place.
We switched to the extraction method that the senior machinist recommended. We bored out the center of the broken drill using a carbide spade drill, starting at 3mm diameter and stepping up in 2mm increments until we reached 10mm. Each step required a new drill and careful alignment to avoid pushing the broken tool deeper into the shaft. At 10mm bore diameter, the remaining wall thickness of the broken drill was thin enough that it collapsed inward when we applied a slide hammer puller to the entry end.
The extraction cost about $400 in labor and tooling. The drill itself was another $85. The downtime on the machine was four hours, which meant I lost the production time for that shift. The shaft was salvageable, but it had to be re-drilled with a new 12mm drill and the bore diameter ended up at 12.2mm instead of 12.0mm. The customer accepted it because the tolerance allowed 12.5mm maximum.
The Rebuild of the Rotary Union
After the extraction, the foreman and I took the rotary union apart to find the root cause of the pressure loss. The rotary union is the component that transfers high-pressure coolant from the stationary supply line to the rotating spindle. It has a set of mechanical seals that wear over time. When we opened it, we found that the carbon face seal had a crack running across about a third of its surface. The cracked seal was allowing coolant to bypass the spindle and drain back into the tank instead of going through the drill.
The previous operator had reported a slight coolant leak the week before, about a drip every ten seconds from the rotary union drain port. He had mentioned it in the shift handover log but nobody had acted on it. The drip was the early warning sign of the seal crack. If I had checked the rotary union condition during my setup, I would have seen the drip and replaced the seal before starting the job.
We replaced the carbon seal, the mating ring, and all the O-rings in the rotary union. The parts cost about $120 and the replacement took about an hour. After the rebuild, the pressure at the spindle matched the pressure at the pump within 2 bar. I made a note in the machine maintenance log and added the rotary union seal condition to the weekly preventive maintenance checklist.
A New Protocol for Shift Handovers
That experience changed how I handle shift handovers. I now read the previous shift’s handover log before I start any setup, not just glance at it. I look for any mention of coolant leaks, unusual noises, pressure fluctuations, or chip form changes. If I see a flag, I investigate before I put a tool in the spindle.
I also write my own handover notes more carefully. I include the coolant pressure reading at the spindle, the chip form I observed on the last part, and any changes I made to the machine settings. I date and sign every entry. I have found that a good handover log prevents about half of the setup errors that happen between shifts.
What Changed After That Day
I have checked the coolant pressure at the spindle on every single job since that crash. Not just glanced at the gauge. I open and close the isolation valve to confirm the needle responds, and I record the pressure in the setup sheet. I also check the rotary union seal condition weekly instead of monthly.
I started paying attention to chip form as a real-time diagnostic signal. Stringy chips mean something is wrong, and the first suspect is always coolant flow. I train every new operator I work with on chip form recognition. I show them the difference between good broken chips and bad stringy chips with actual samples from the chip bin.
I also added a pre-start checklist to every machine on the floor. The checklist covers coolant pressure at the spindle, filter pressure drop, rotary union condition, guide bushing alignment, and drill condition. It takes about three minutes to complete. I have not had a coolant-related crash since I started using it.
| Prevention Step | Time Required | Cost | Estimated Crash Cost Saved |
|---|---|---|---|
| Check spindle coolant pressure | 30 seconds | $0 | $400+ per incident |
| Verify rotary union seal | 2 minutes weekly | $0 | $400+ per incident |
| Pre-start checklist | 3 minutes | $0 | $400+ per incident |
| Chip form monitoring | Continuous | $0 | $400+ per incident |
Key Takeaways
- The coolant pressure gauge at the pump does not tell you what the pressure is at the cutting edge. Always check the gauge closest to the spindle.
- The previous operator can bump a valve, change a setting, or leave the machine in a condition you do not expect. Never assume the setup is correct because someone else ran it before you.
- Stringy chips are not normal. They are a warning sign that the cutting edge is not getting enough lubrication or coolant flow. Stop the machine and investigate before the drill seizes.
- A three-minute pre-start checklist is cheap insurance against a three-hour tool extraction and a $400 repair bill.
- The foreman’s advice was short and correct. Check the pressure first. I have passed that advice to every operator I have trained. It has saved more drills than any other single habit I have adopted.
- Machine crashes happen fast. From the first warning sign to the broken drill, I had about thirty seconds. That is not enough time to troubleshoot. The troubleshooting has to happen before the drill enters the hole.