Can a Foundry Cobot Take Casting Grinding and Cut Silica Dust?

Can a foundry casting grinding cobot cut silica dust exposure?

Grinding and chipping castings is a known source of silica exposure in foundries. Castings come off the mold with sand still attached, and the fettling that removes it throws respirable silica into the breathing zone of whoever holds the tool. The cobot question is really an exposure question: can a robot take the grinding stroke off a person's hands while the dust stays contained at the source?

The honest short answer is that a cobot can help, but the gain comes from where the grinding happens and how the air moves, not from the arm alone. NIOSH tested a downdraft booth with a turntable on casting cleaning work and measured a 59 to 77 percent reduction in respirable silica exposure, and the agency said the booth may keep exposures below permissible limits. A cobot cell built on the same principle, with capture at the grinding point, is the design worth testing against your existing numbers.

Your safety manager is right to look for an alternative to hand grinding. Hand grinding keeps the operator's head close to the source for the whole shift, and every hour adds exposure time. A cell moves the person outside the dust during the cut, which is the core reason the idea deserves a controlled pilot rather than a yes or no decision on the whole floor.

What do OSHA's silica rules require at a foundry?

OSHA's respirable crystalline silica standard sets an 8-hour time-weighted average limit of 50 µg/m³, according to a legal summary of the rule by Reinhart. Foundries fall under the general industry standard, so the same limit applies to your grinding stations as it does to other general industry work. An action level of 25 µg/m³ triggers medical surveillance and increased monitoring, which means your sampling results decide how much monitoring and surveillance applies.

Silica has been a foundry focus for a long time. OSHA ran a Special Emphasis Program on silicosis in foundries starting in 1996, so grinding and fettling work has been under agency scrutiny for decades. For a safety manager, the practical first step is a documented exposure baseline for each grinding station, by task and by shift.

Keep the baseline in the same form you would use after the pilot. A before and after comparison only works if the sampling method, the tasks sampled and the shift patterns match. Sampling that changes between runs will make the cell look better or worse than it really is.

Does a grinding cell with a cobot actually reduce silica?

A cell reduces silica exposure when it does three things well: it contains dust at the source, it keeps the operator out of the breathing zone, and it controls the swarf and airflow around the casting. Tool choice and position also matter. A NIOSH study of a steel foundry found that tool type, the direction of grinding swarf, and whether the tool sat inside or outside the casting produced noticeable differences in exposure.

Some foundries have already built robotic grinding with closed-capture ventilation. One iron foundry reported that its robotic grinders brought operator fatigue and staffing gains, but its report gave no dust measurements. That is useful for layout ideas, not evidence of a silica number.

A UK Health and Safety Executive case on automated fettling of steel castings reported reduced noise, dust and fumes. The result is qualitative, so read it as direction rather than a figure you can plan around. We did not find a published cobot study measuring silica reduction on heavy casting grinding, and that gap is the reason to measure your own stations rather than trust a general claim about robots and dust.

Can a cobot hold the finish customers expect on castings?

This is the question that decides the pilot. A review of robotic grinding, which we could read only as a search summary, says a lightweight joint-type robot suits light-load fine grinding, while its lower stiffness limits high material removal rate work. The same review names steady normal force control as a main difficulty on uneven material and deforming workpieces, so finish consistency is the place to focus.

Castings are not uniform. Research on deburring cast iron notes that castings vary widely from nominal dimensions and that burrs can be of unknown size in unpredictable positions, so adaptive force feedback is needed. A 2018 deburring study reported that a human-mimicking strategy cut cycle time and improved finishing quality. The lesson for you is that control method matters as much as the arm you choose.

In practice, finish quality depends on the casting family. A cobot is more likely to hold a tight finish on a narrow set of parts with repeatable geometry than across a mixed job shop of castings. Ask your quality team which parts have the finish spec customers actually check, and start the pilot there, with inspection criteria written down before the first casting runs.

Where do cobots fit, and where do grinders still win?

Cobots are a good match for repeatable, moderate-force passes on consistent castings inside an enclosed cell. Heavy grinders still win on heavy stock removal, on uneven castings that need a human eye, and on judgment calls about blended surfaces. The comparison below sets out the trade without favoring either side.

Adoption is climbing, so there is more vendor and integrator experience to draw on than a few years ago. The Association for Advancing Automation, known as A3, reported 7,212 cobot units ordered in North America in 2025, which was 19.6 percent of the 36,766 total robots ordered. Design News reported that cobots reached 28.6 percent of robots ordered in Q4 2025, the highest quarterly share since A3 began tracking the category.

How many foundries and foundry workers are in the US?

Federal counts for foundries are hard to pin down, and the sources we found disagree. An undated career page attributed to the American Foundry Society puts the U.S. foundry count at about 1,750, down from 3,200 in 1991, and says about 75 percent of foundries employ fewer than 100 people. That mix matters. Small plants have fewer engineers and less room for a multi-week pilot, so the integration plan has to be lighter and the scope tighter.

On the workforce side, IBISWorld reports 53,517 people employed in nonferrous metal foundries as of 2025. Those figures cover a range of shop sizes and products, so use them to size the regional picture, not to forecast your own staffing. Your own grinding headcount, by station and shift, is the number that will shape the pilot.

What does getting a grinding cobot cell look like with us?

We are an integrator, not a single robot maker. We choose the arm that fits your castings, lease it through financing, deploy and integrate it into the cell, train your staff and service every unit. The fixed robot lease payment comes in a quote after a free site assessment, so you see the full picture before any commitment.

The site assessment is where the exposure work starts. We map the grinding stations, look at the capture and airflow, and review your sampling data so the pilot has a before number to beat. The lease quote then reflects the actual scope, including the enclosure work and the programming for each casting family.

Maintenance is included under our service plan, with on-site dispatch and remote triage, and regional service engineers in every state we service. Ask for the response scope in writing before you sign, including who services the cell after go live and how fast a unit can be replaced if it goes down during a critical run. That written scope is the document your safety manager and operations lead should both read.

What should you check before committing to a pilot?

A few facts will settle most of the question. Collect them now, because they change both the quote and the answer.

If most of your grinding is heavy and your castings vary widely, a cobot may fit only a slice of the work. A realistic plan can still lower operator exposure on the lighter, repeatable parts, and that is a legitimate result for a safety program. Pilot the parts with the clearest path first, measure the stations before and after, and let the data decide how far the cell goes.

• Respirable silica results for each grinding station, by task and shift, compared with the 50 µg/m³ limit and the 25 µg/m³ action level.
• Which casting families carry a tight finish spec, and which finish checks customers actually run.
• Whether the existing booth or hood can hold capture airflow with the arm inside it.
• How much stock each casting must lose, since heavy removal stays with grinders for now.
• Who services the cell after go live, and how fast that person can dispatch.