Key takeaways
- Mount pad flatness of 0.2 mm and hole position of ±0.5 mm are the two dimensions that cause call-outs; overall length and width rarely are.
- Weld sequence from the centre outward with balanced 200 mm runs holds a 2,400 mm frame to ±2 mm on the diagonal and 1 mm of bow, against 3–4 mm of bow when the seams are run end to end.
- A 4-cylinder set at 1,500 rpm fires at 50 Hz; the mounted system should have a natural frequency near 7 Hz so the transmissibility stays below 0.2.
- Drain holes of at least 12 mm at the lowest point of every closed section are the difference between a frame that lasts ten years and one that rusts from inside.
Genset base frame manufacturing is a fixture problem with a welding problem attached, and it is usually quoted by weight as though it were neither. The frame looks like the easiest part on a set: four channels, a few gussets, four mount pads and two lifting eyes, and it carries no pressure, no heat and no moving contact. It is also the part that decides whether the set sits level on a plinth 900 km away, whether the alternator coupling stays aligned after a year of running, and whether the canopy doors still close.
The dimensions that cause call-outs are not the ones a first-time designer marks. Overall length and width can be 2 mm out and nobody notices. Mount pad flatness of 0.2 mm, or a hole centre displaced ±0.5 mm from the engine drawing, produces a set that rocks, a mount that runs hot, and a service visit.
Both numbers are set by the fixture and the weld sequence rather than by the welder's eye, which is why the route below is written in the order it is. What follows is the cut, the fixture, the sequence that holds the numbers, the coating, and the four measurements that release a frame.
What the engine drawing fixes and what it leaves open
The engine and alternator maker supplies mount positions, pad sizes and the mass at each corner. Those positions are datum, not suggestion: an anti-vibration mount designed for a 4 mm static deflection cannot absorb a 2 mm positional error, and forcing it in preloads the rubber in shear.
What the engine drawing usually leaves open is the frame's own stiffness. A frame that flexes between the pads turns the set into a spring-mass system with a second mode in the 30–60 Hz band, and the firing frequency of a 4-cylinder set at 1,500 rpm is 50 Hz. When the frame mode and the firing frequency meet, the whole skid buzzes and the canopy panels join in.
The working rule: aim for a first frame mode above 60 Hz, keep the mounted system natural frequency at 6–8 Hz so the transmissibility at 25 Hz and 50 Hz stays below 0.2, and let the mounts do the isolating. Stiffness is cheap in a base frame; a 100 mm × 50 mm × 5 mm channel costs far less than an extra mount or a heavier engine bed.
A worked check makes the sizing concrete. Take a 250 kVA set with an engine and alternator mass of 2,600 kg plus 220 kg of frame, silencer brackets and fluids, so 2,820 kg in total on four mounts. Each mount sees 705 kg. A mount selected for 6 mm static deflection at that load has a vertical stiffness of about 1.15 kN/mm, and the mounted system natural frequency works out at 6.4 Hz — near the middle of the target band, and far enough from the 25 Hz running speed that transmissibility is roughly 0.07.
The frame still has to carry 705 kg at each corner without deflecting into contact with the plinth, and that is where the section size is actually decided. Treat one longitudinal member as a simply supported beam over the 1,800 mm between mounts, with the 705 kg corner load at mid-span. The load is F = 705 × 9.81 = 6,916 N, and a 100 × 50 × 5 mm cold-formed channel has I = 1.43 × 10⁶ mm⁴ about its strong axis.
δ = FL³ / 48EI = (6,916 × 1,800³) / (48 × 206,000 × 1.43 × 10⁶) = 4.03 × 10¹³ / 1.41 × 10¹³ = 2.85 mm.
That is nearly six times the 0.5 mm a set can absorb, and it is the honest reason a frame priced by weight is priced wrong. The bending stress is not the problem: M/Z = (6,916 × 1,800 / 4) / 28,600 = 109 MPa, which is 44% of the 250 MPa yield of E250. Deflection, not stress, is the design driver, and the numbers say so plainly.
Two things bring it back. The corner load is shared between two longitudinal members, which halves δ to 1.43 mm, and the pads sit over the cross members and end plates, so the load reaches the plinth in bearing instead of through mid-span bending — the real unsupported length between a pad and the nearest cross member is 300–500 mm, not 1,800 mm.
Where a mount genuinely lands mid-span — a long 12-cylinder bed, or a skid with the mounts inboard of the ends — the section has to be sized on the formula rather than picked from a table. Holding 0.5 mm at 6,916 N over 1,800 mm needs I ≥ FL³ / 48Eδ = 4.03 × 10¹³ / (48 × 206,000 × 0.5) = 8.2 × 10⁶ mm⁴. No 100 mm channel reaches that; a 180 × 75 × 8 mm channel at I ≈ 1.18 × 10⁷ mm⁴ gives 0.34 mm with margin.
The numbers that go on the frame drawing
| Parameter | Typical value | Ganesh limit or note | Consequence if missed |
|---|---|---|---|
| Channel section | 100 × 50 × 5 mm, IS 2062 E250 | Cut on the 12 kW laser or the iron worker for plate | Undersized web flexes between pads |
| Frame mass for a 250 kVA set | 180–260 kg | Weighed and declared with the quote | Lifting and shipping calculations |
| Mount pad flatness | 0.2 mm over a 150 mm pad | Checked on a surface plate; 6 mm pads are faced after welding, not welded to 0.2 mm | Rubber preloaded in shear, vibration passes |
| Mount hole position | ±0.5 mm from engine datum | Located with a drill jig in the fixture | Mount cannot be fitted without slotting |
| Overall diagonal | ±2 mm on a 2,400 mm frame | Held by balanced weld sequence; ISO 13920 class A allows ±4 mm at this length, so ±2 mm is a specific callout | Set sits twisted on the plinth |
| Squareness at corners | 90° ± 0.5° over a 100 mm leg (0.87 mm) | Fixture stops, checked before release | Canopy panels do not close |
| Fillet weld leg | 5 mm, ISO 5817 level C | Robotic MIG on the CLOOS cells | Undercut becomes a fatigue crack start |
| Lifting eye capacity | 2 × the frame plus set mass | Designed, not estimated, at the sling angle | Eye yields during a two-point lift |
| Drain hole | 12 mm minimum at each low point | Drilled before coating, cleared after pretreatment | Water sits inside the section |
| Powder DFT | 80–100 µm | Seven-tank pretreatment, 18 ft × 14 ft booth | Rust bleed at weld toes within a season |
| Frame flatness on the plinth | ≤ 1 mm per metre | Checked with a straight edge across pads | Shim stacks and a set that rocks |
How a genset base frame is manufactured
Cutting plate and channel
Gussets, pads, end plates and lifting-eye bosses are cut from plate on the 12 kW fibre laser, which handles up to 20 mm; heavier sections go to the iron worker, which cuts plate to 30–40 mm. Laser-cut plate has a square edge and a 0.2–0.5 mm heat-affected zone that welds cleanly without a bevel on anything under 8 mm.
Channel is cut to length with the web vertical, so the cut face stays square and the two flanges do not splay. A 100 mm channel cut on a bandsaw without a stop comes out 0.5 mm out of square, and four of those errors close a 2,400 mm frame by 2 mm on the diagonal before a single weld is laid.
Fixture, then weld
The frame is assembled in a fixture that sets the outer rectangle, the diagonal and the mount pad positions. Clamps hold the section against fixed stops; the pads are tacked only after the rectangle is checked.
That sequencing is the whole game. Weld the rectangle first, let it move, then fit the pads to the moved frame, and the pads end up in the right place on a frame that is out of square. Weld the pads into a frame held square, and the closed rectangle and the pad pattern are both right when the clamps come off.
Robotic MIG on the three CLOOS cells runs the long fillets with a repeatable 5 mm leg and a consistent travel speed. Manual stations handle the gussets, the lifting-eye bosses and any joint the robot cannot reach without a re-fixture. All positioners are set so the weld runs flat, because a vertical-up 5 mm fillet puts in 20–25% more heat per unit length than the same fillet flat.
Weld sequence for a 2,400 mm frame
Weld from the centre outward, in 150–250 mm runs, alternating sides. Run one side end to end and the frame bows 3–4 mm in the middle; the balanced pattern holds it within 1 mm on the same section and the same heat input.
For the closed rectangle, weld the four corners in a diagonal order: one corner, then the opposite, then the remaining pair. That distributes shrinkage evenly and avoids pulling one end square while the other opens.
Backstep welding, where each run starts 50 mm behind the previous one and works backward, is the answer for a long seam that cannot be balanced. It costs about 15% in arc time and removes most of the bow.
Shot blasting, coating and the inside of the section
After welding, the frame goes through the 16 ft × 16 ft shot blast chamber, which removes mill scale and weld discolouration and gives the pretreatment line a clean surface to work on. Skipping blast on a frame with heavy mill scale leaves a coating that fails at the scale boundary.
The seven-tank pretreatment follows: degrease, rinse, descale, rinse, conversion coat, rinse, demineralised rinse. Then powder at 80–100 µm in the 18 ft × 14 ft booth.
The inside of a closed channel is the part nobody inspects and everybody inherits. Two things make the difference: drain holes at every low point, drilled before coating and reamed clear after pretreatment, and a coat of the same powder or a spray of cavity wax on the internal faces of sections that will sit outdoors. A 12 mm hole at each end of a 2,400 mm channel is enough to let condensation out and air in.
How the frame is checked before it leaves
Four measurements release a base frame. Pad flatness is checked with a straight edge and feeler across each pad in two directions, with 0.2 mm as the reject line. Mount hole position is checked on a pin plate built to the engine datum, which takes about four minutes and catches a jig that has drifted. Diagonal is measured corner to corner on the top face, before and after coating, and the two readings must agree within 1 mm. The ±2 mm band is tighter than ISO 13920 class A, which allows ±4 mm in the 2,000–4,000 mm range and ±6 mm in class B, so it has to appear on the drawing as a specific dimension. A general welding class will not deliver it.
The fourth is the fit check that nobody puts on a drawing: the frame is stood on a surface table and a dial indicator is run along the pad faces. A frame can pass flatness pad by pad and still have two pads 1.5 mm apart in height, and that is the condition that puts a set on shims. Pad-to-pad height within 0.5 mm across the rectangle is the number to hold.
Records go with the first article and then to an agreed frequency. A frame that is 1 mm out on the diagonal is not a scrap decision on its own; it is a decision that depends on whether the canopy is built to the same frame and whether the plinth on site can absorb it.
Where base frames fail
Pads that are not flat. The weld shrinks the pad face and cups it by 0.3–0.8 mm on a 6 mm pad. The fix is to weld the pad with a balanced pattern on both sides, or to fit a thicker pad and face it after welding. Anything above 0.2 mm and the mount sits on its edges.
Diagonals that open during coating. A frame that measures 1 mm on the diagonal before coating and 3 mm after has been heated unevenly in the oven or hung from one corner. Hanging points for a 200 kg frame belong on the fixture drawing, not on the operator's judgement.
Lifting eyes welded to the flange. An eye on a 5 mm flange with a 4 mm fillet will tear the flange before the eye yields. The boss needs to sit over a web or a doubler plate, with the weld leg sized to the eye's rated load.
Water in the section. Frames come back with rust streaks at the welds after two monsoons, and the cause is a drain hole that was drilled before blasting and plugged with shot.
Mount positions that move because the engine changed. A frame is often built to a revision that was superseded. The mount pattern should be confirmed against the engine drawing at the time of the weld, not at the time of the quote.
The fixture, the robot and the blast line behind a base frame
Base frames and skid bases for Mahindra Powerol sets are cut on the 12 kW fibre laser with its 40 ft × 16 ft bed, formed on the 200 T, 5-axis brake where a folded section replaces a channel, and welded in fixtures built in our own tool room on three CLOOS robotic MIG cells, with more than 30 manual MIG, spot and SPM stations taking the gussets and the lifting points. Blast, seven-tank pretreatment and powder coating run in sequence, and the frame is checked for pad flatness, hole position and diagonal before it is released to the canopy line. The frames ship with the canopies as a kit through genset parts. The sheet metal weight calculator gives a first estimate of frame and set mass for the lifting plan, and the genset canopy RFQ checklist lists the fields a canopy or frame enquiry needs.
The joint-level detail is in welded genset base frames, the distortion mechanics in controlling weld distortion in frames, and the heavy-section practice in welding for heavy gauge material handling parts. Frame and canopy guides are collected under genset, canopy and acoustic enclosures.
What to put on the drawing
- Engine and alternator mass, and the mass at each mount, from the OEM drawing with its revision.
- Mount hole pattern, pad size, pad flatness and the datum those dimensions come from.
- Section sizes and grade, and whether a folded section is acceptable in place of a rolled channel.
- Lifting arrangement: number of points, sling angle, rated capacity and the doubler detail.
- Weld specification: fillet leg, quality level, and which welds are visual only.
- Drain hole positions and the internal coating requirement.
- Finish: pretreatment stages, powder type, DFT, colour, and whether the inside faces get the same treatment.
- Inspection record: which dimensions are recorded on the first article and on what frequency afterwards.
When a welded steel base frame is the wrong choice
A steel frame is the wrong choice where the set sits inside a building on an inertia block that the structural engineer has already designed. In that case the frame is a transport skid, and a lighter folded frame with no lifting eyes is enough.
It is also wrong for a set above about 2,500 kVA. At that size the frame is a structural fabrication that needs a designed weld procedure, NDT on the primary welds and a stress analysis, and it should be engineered as such rather than quoted from a drawing that was scaled up from a 250 kVA part.
And where the customer wants the base frame to double as a fuel tank, the tank rules apply: baffles, a welded and tested seam, a fill point and a vent. We build skid frames and we build welded enclosures, but a fuel tank is a pressure-retaining item with its own test regime, and it should not be treated as a frame with a lid.
Send the frame drawing with the engine mount detail and the lifting requirement. Request a quote
Frequently asked questions
What steel is used for a genset base frame?
IS 2062 E250 channel or plate is the standard choice for frames up to about 3,000 kg, in 5–6 mm web thickness. Grade E250 gives 250 MPa minimum yield, which is ample for the static load; the thickness is chosen for weldability and stiffness, not for stress.
Does a genset base frame need machining after welding?
Only if the drawing calls for it. On a 6 mm pad, welding cups the face by 0.3–0.8 mm, so 0.2 mm is not held by welding alone: the pad is either 12–16 mm thick, welded with a balanced fillet on both sides and left with its cut face, or it is welded, released from the fixture and faced in one pass. Where the pads must be within 0.05 mm, plan for face milling after stress relief.
How many anti-vibration mounts does a genset need?
Four is normal up to about 400 kVA and six or eight above that, so that no single mount carries more than 25% of the mass. Each mount position is located from the engine maker's drawing, not scaled from the frame.
Why does the base frame rust from the inside?
Condensation and wash water collect in the closed channel sections. A 12 mm drain hole at each low point, drilled before coating and blown through after pretreatment, is the only reliable fix.
Sources and further reading
- ISO 13920 — Welding, general tolerances for welded constructions · reviewed 3 Nov 2026
- IS 2062 — Hot rolled medium and high tensile structural steel · reviewed 3 Nov 2026
- ISO 5817 — Quality levels for imperfections in fusion-welded steel · reviewed 3 Nov 2026
Quoting a base frame or a full canopy kit?
Send the frame drawing, the engine and alternator mass, the mount positions and the finish. The weld sequence, fixture and inspection plan are reviewed before the price is fixed.
Request a quote