A dimensional tolerance controls how big something is. A geometric tolerance controls how it is shaped, how it is oriented, and how one feature relates to another — whether a face is flat, whether two faces are parallel, whether a bore is square to a surface. On fabricated steel components, geometric control is where most disputes originate and where a great deal of avoidable cost is spent, because a welded assembly cannot be made geometrically accurate by welding alone.
This guide explains what geometric tolerances control, why they behave differently on fabricated components than on machined ones, which surfaces justify them, and how to state them so that they can actually be met and checked.
Dimensional vs Geometric Control
The two kinds of control answer different questions, and a drawing needs both:
- Dimensional control answers “how big” — the length of a part, the diameter of a hole, the distance between two holes. Its tolerances are expressed as plus and minus limits on a size or position.
- Geometric control answers “what shape, and how does it sit relative to something else” — flatness, straightness, parallelism, perpendicularity (squareness) and the profile of a surface.
The distinction matters because the two are independent. A face can be exactly the right distance from a reference and still not be flat. A hole can be the right diameter and still not be square to the face it enters. A pair of faces can each be the right thickness and still not be parallel to one another. Size and shape have to be controlled separately, because they are produced and verified separately.
On a fabricated component, the geometric controls are the ones that usually decide whether the part works: whether a mounting face sits flat against its support, whether a bore aligns with the shaft that has to pass through it, whether a frame is square enough for the equipment that mounts on it.
Why a Welded Component Is Not Geometrically Accurate as Welded
Welding introduces heat, and heat moves steel. Even when every part going into an assembly is cut and prepared correctly, the act of welding changes the shape of the finished component: joints pull as they cool, long welds draw a structure out of square, and heat applied unevenly on one side of a neutral axis bends the part toward the other. The result is that the geometry of an as-welded fabrication is inherently variable, and it varies with factors that are hard to control precisely — the amount of weld, the sequence, the restraint, the thickness.
This is not a defect in the process; it is what welding does. The practical consequence is that geometric accuracy and the as-welded condition are in tension. A component can be fabricated to close dimensional control and still show measurable movement in its overall shape.
The Consequence: A Geometric Requirement Usually Means Machining
Because welded geometry moves, the reliable way to produce an accurate geometric feature on a fabricated component is to make it after welding, by machining. A face that has to be flat, a bore that has to be square, a mounting surface that has to sit true — these are normally finished by a cutting operation performed on the welded assembly, not left to the welding process.
This is the single most useful thing a buyer can keep in mind when writing geometric requirements: specifying geometry on a fabricated part is, in most cases, a decision to have that surface machined. Stating a tight flatness or parallelism requirement on a surface that is expected to stay as-welded creates a requirement the process cannot guarantee, which leads either to a part that is rejected on paper or to a supplier who quietly ignores the requirement.
How machining fits into the sequence of cutting, welding and finishing is covered in CNC machining for heavy steel components.
Which Surfaces Justify a Geometric Control
Not every surface needs one. Geometric controls cost money — they require machining, and they require inspection — so they belong on the surfaces whose function actually depends on them. In heavy fabricated equipment, that usually means:
- Mounting and mating faces — surfaces that seat against another component, a base, a foundation or a mating part, where flatness determines how the load is transferred.
- Surfaces that locate or align — faces and bores that set the position of a shaft, a bearing, a rail or another assembly.
- Bearing and sealing surfaces — where flatness or parallelism directly affects fit and function.
- Interfaces to purchased equipment — where the component has to match the mounting dimensions of a motor, gearbox, cylinder or machine supplied by someone else.
Surfaces with no such duty can usually stay in the as-cut or as-welded condition. Applying a geometric requirement to a surface that merely looks important adds machining and inspection without adding function.
Datums: The Foundation of Any Geometric Requirement
A geometric tolerance is meaningless without a datum — a reference feature from which the requirement is measured. “Flat within a limit” says nothing unless it is clear what the flatness is relative to, and a parallelism requirement needs two things: the surface being controlled, and the surface or axis it must be parallel to.
On fabricated components, the datum structure deserves deliberate thought, because the natural reference features are often not available or not reliable:
- The datum has to be something that can be established physically on the finished component, not just something convenient on the drawing.
- It has to survive the manufacturing sequence. If a datum is machined, then the part is welded and the datum moves, the requirements built on it no longer mean what the drawing intended.
- It should be a feature that matters. Where possible, the datum should be a functional surface — a mounting face, for example — so that the control is measured from something that matters to the part.
How machining allowance and datums interact in a fabricated component is introduced in heavy fabrication tolerances, which covers the tolerance stack from cutting through to final machining.
Stating Geometric Requirements That Can Be Met and Checked
A geometric requirement is only useful if the supplier can achieve it and the buyer can verify it. In practice that means:
- Put the control on the right surface. Apply it to the functional surface, not to a convenient one.
- Give it a datum. State what the requirement is measured from, and choose a datum that can exist on the finished part.
- Match the requirement to the function. Ask what the surface has to do, then specify the control that serves it, rather than the tightest value that seems safe.
- Confirm it can be measured. A requirement on a feature that cannot be reached or set up for inspection is a requirement that cannot be verified, which helps no one.
- Keep the small number of surfaces that matter. A drawing that applies geometry everywhere usually means the buyer has not decided which surfaces are functional — and the cost of the ambiguity is paid by everyone.
Common Mistakes
A few patterns account for most of the friction on geometric requirements in fabricated components:
- Applying geometry to as-welded surfaces. The requirement cannot be guaranteed without machining, but nothing says the surface will be machined.
- No datum, or a datum that moves. The requirement is stated but cannot be interpreted or verified consistently.
- Over-tightening. A flatness or parallelism value tighter than the function needs buys additional operations and inspection for no benefit.
- Conflicting requirements. A geometric control that contradicts a dimensional one, or two requirements that cannot both hold on the same feature.
- Forgetting the sequence. Specifying a geometric relationship that assumes an earlier stage will hold, when a later stage — welding, or a second machining operation — will disturb it.
How VHOFAB Approaches Geometric Requirements
VHOFAB fabricates heavy steel components where geometric accuracy and fabrication are planned together, rather than treated as separate steps. Cutting, welding and machining are carried out in one workshop, so a geometric requirement can be met by machining the surface after welding — the sequence that makes the requirement achievable — instead of being specified on a surface that welding will move.
Requirements are reviewed from the drawing before production, so the functional surfaces, the datum and the machining sequence are settled early. Where a requirement is ambiguous or cannot be verified, that is raised before the component is made rather than discovered at inspection.
To discuss a component with geometric requirements, send the drawing and technical requirements through the contact form for a manufacturing review. The same logic applies across heavy plate processing and, on a specific class of part, to welded frames and equipment bases; for the wider question of what to put on a drawing, see how to choose a steel fabrication supplier.