Eight Heat Exchangers, One Package: Supplying a Cluster to GS E&C Under ASME U-Stamp and KGS

The backdrop behind the RFQs

In the first half of 2026, Yeosu — Korea's largest petrochemical complex — moved into serious restructuring. Seoul has committed roughly ₩2.1 trillion (about USD 1.5 billion) to support consolidation of the sector. LG Chem and GS Caltex have been in late-stage talks on a joint venture covering the Yeosu naphtha crackers. Industry estimates suggest this round of consolidation could remove as much as ₩6.7 trillion of industrial output in 2026.

Three drivers stack on top of each other:

• Margin. New ethylene capacity out of the Middle East and China has squeezed Korean NCC cracking margins for years. In 2026 they turned negative again. The top three producers have reportedly cut new hiring by around 20%.
• Policy. Seoul is funding restructuring directly, and consolidation of the petrochemical base is being pushed on multiple fronts.
• Energy efficiency. Korea's efficiency framework — administered by KEMCO, the Korea Energy Agency, across roughly 44 categories of industrial and building equipment — keeps tightening, and it shapes equipment selection in both revamps and new builds.

One clarification belongs here, because it gets confused constantly: KEMCO is an energy-efficiency and labelling body, not a manufacturing qualification. It tells you whether a machine is efficient enough. It does not tell you whether a shop is allowed to build it.

Put the three drivers together and you get the instruction that now governs Korean fixed-equipment procurement: cut cost, but do not touch compliance. The procurement manager has to take CAPEX out of the package while holding two lines that cannot be crossed — ASME as the technical basis, and KGS as the entry approval. That is why qualified overseas shops holding both matter more than they did three years ago.

The case: eight exchangers, one package, for GS E&C

GS Engineering & Construction (GS E&C) is one of Korea's established EPC contractors, with a long record in refining and olefins. For their MFC — Mixed Feed Cracker — programme, tied to the GS Caltex Yeosu olefins chain, we supplied the heat exchanger package.

It was not a single unit. It was a cluster of 8 shell-and-tube exchangers delivered as one package, with a wide spread of sizes and two different material families:

Item	Qty	Size DN × wall × length (mm)	Unit weight (kg)	Materials (shell / tube)	Certification
Large unit	1	DN1000 × 16 × 10,725	15,758	SA516 GR.70 / SA106	ASME U-Stamp + KGS
Mid units	2	DN800 × 14 × 7,453	11,656	SA240 304L / SA312 TP304L	ASME U-Stamp + KGS
Heavy-wall unit	1	DN1200 × 20/30 × 6,008	6,519	SA516 GR.70 / SA106	ASME U-Stamp + KGS
Small unit A	1	DN406.4 × 12.7 × 5,040	2,239	SA516 GR.70 / SA106	ASME U-Stamp + KGS
Small unit B	1	DN323.8 × 17.48 × 2,046	994	SA516 GR.70 / SA106	ASME U-Stamp + KGS
Stainless units	2	DN168 × 7.11 × 1,835	536	SA240 304L / SA312 TP304L	ASME U-Stamp + KGS

Diameters run from DN168 to DN1200. The largest unit is DN1000, 10.7 metres long, 15.76 tonnes. All eight were designed to ASME Boiler & Pressure Vessel Code Section VIII Division 1, carry the U-Stamp, and hold KGS certification — which is what the Korean plant needs in order to accept them.

Why one package carries two material families

The mix of carbon steel and stainless follows the process, not indecision.

Where the duty is conventional process gas or a utility service, SA516 GR.70 shells with SA106 tubes give controlled cost and reliable pressure performance. Where the medium is corrosive, or surface cleanliness matters, the specification moves to SA240 304L shells with SA312 TP304L tubes. Inside one cracker unit, material follows the medium. More expensive is not automatically better — it is simply wrong in the places where carbon steel does the job.

What is demanding is carrying both families inside one package while keeping each inside its own welding qualification and inspection regime. That is a manufacturing-management problem more than a welding problem.

What the material spread means for the shop floor

Two families in one package multiply the control points rather than add to them:

• Two welding qualification sets. SA516 GR.70 to SA106 and SA240 304L to SA312 TP304L are separate WPS/PQR families. Welder qualification is verified per combination.
• Two consumable streams. Filler metals, backing gas and purge control differ. Stainless root passes on tube-to-tubesheet joints need argon purge discipline that carbon steel work does not.
• Two heat-treatment paths. The carbon steel heavy-wall unit at 20/30 mm follows a PWHT route the stainless units do not take at all.
• Two inspection routes. RT/UT coverage, PT on stainless welds, and the acceptance criteria that go with each.

Each of those is routine on its own. Running them concurrently, on eight units, against one delivery date, is where a package is won or lost.

What Section VIII Div.1 actually fixes on a cluster

A code stamp is often read as a single yes/no line on a bid sheet. On a multi-unit package it is closer to a set of design decisions that have to be taken consistently across eight drawings, and those decisions drive both cost and schedule.

Shell thickness and the corrosion allowance. The heavy-wall unit in this package runs 20 mm on the shell and 30 mm at the heavier section. That step is not decoration — it follows the design pressure and the corrosion allowance the process licensor specified. Where a client carries 3 mm corrosion allowance on carbon steel and none on the stainless units, the two families diverge in plate procurement, in forming, and in the post-weld heat treatment decision.

Tube-to-tubesheet joints. Expanded, seal-welded, or strength-welded is a duty decision, not a preference. On the carbon-steel units in conventional process gas service, expanded plus seal weld is normally sufficient. Where leakage between shell and tube side cannot be tolerated, the joint moves to strength welding with its own qualification and its own inspection. Deciding this late is one of the more common causes of drawing revision loops.

Baffle cut and support spacing. These come out of the thermal design, but they land on the fabrication floor as fit-up tolerance. On a 10.7 m bundle, cumulative baffle misalignment is what makes a tube bundle refuse to draw. It is checked at fit-up, not at final inspection, because at final inspection it is too late.

Differential expansion. Fixed tubesheet designs need the shell-side and tube-side expansion difference resolved — by an expansion joint, or by moving to a floating head or U-tube configuration. On a cluster spanning DN168 to DN1200 with two material families, the coefficients differ between the carbon and stainless units, so this is settled unit by unit rather than once for the package.

None of this is exotic. The point is that eight units means eight sets of these decisions, taken against one delivery date and one inspection release.

What a Korean EPC is buying when it consolidates a cluster

When GS E&C put all eight exchangers with one shop, price was not the whole story. The hidden cost of splitting a package rarely shows up on the bid comparison sheet:

• Quality interface. Four shops means four quality systems, four sets of welding procedure qualifications, four interpretations of the same note on the same drawing.
• Material traceability. Multiple mills, multiple certificate formats, and no single party accountable when one heat number cannot be reconciled.
• Schedule coordination. The package ships when the slowest shop ships. The EPC absorbs the float.
• Site interface. Nozzle orientation, support saddles, tie-in dimensions — anything that has to line up on site becomes an inter-supplier negotiation the EPC has to chair.

Consolidating the cluster into one qualified shop absorbs those interfaces internally. Our position on this project is worth stating plainly: we are an equipment supplier to GS E&C. They build the plant. Our scope was to build eight exchangers to their drawings, their code basis and their inspection plan, test them, certify them, and get them to Korea in one piece.

KGS: the approval that cannot be obtained after award

ASME is a technical foundation. It is not entry into Korea. KGS — the Korea Gas Safety Corporation, under the Ministry of Trade, Industry and Energy — requires its own manufacturer approval and product type approval. Holding Section VIII Div.1 capability is the precondition, not the permission.

From the published requirements, the path runs roughly like this:

1. Application — application forms, business licence, ISO 9001 certificate, quality manual (Korean or English), shop layout, equipment list, personnel qualification records.
2. Factory audit — the initial audit typically takes 3–4 days, depending on how many product categories and standards are covered by the application.
3. Technical review and close-out of findings.
4. Certificate issued.

The cycle typically takes 6 to 12 months, depending on documentation quality, audit scheduling, and how fast findings are closed. A tender window is usually a few weeks. KGS cannot be started when the RFQ lands.

That is the structural reason an already-approved shop carries a premium for an EPC: the approval is not something the winner can go and get afterwards. In our case the ASME U-Stamp and the KGS approval were in hand before the GS E&C enquiry arrived. That was the precondition for being on the list at all.

Cluster delivery is a supply-chain discipline

Eight units. Two material families. Dozens of material certificates. It does not go wrong at the welding arc. These are the control points that decide it:

Material traceability. Every batch of SA516 GR.70, SA106, SA240 304L and SA312 TP304L needs a traceable mill certificate to the EN 10204 3.1/3.2 convention. On a cluster, one unreconciled heat number does not stall one exchanger — it stalls the inspection release of the whole batch.

Separated workshops for carbon and stainless. Cutting, fit-up and welding for stainless happen away from carbon steel. Shared grinding wheels, shared layout tables and shared handling gear cause iron contamination on stainless surfaces, and it typically shows up long after anyone can trace it. This is a physical layout decision rather than a procedural one, which is why we run separated bays.

Welding qualification. Every material combination has its own WPS/PQR. Welder qualifications are verified per combination, not per shop.

ITP. Hold, Witness and Review points aligned with the EPC and the third party up front. On the large DN1000 and DN1200 units, RT/UT coverage follows the drawing and the code.

FAT. Hydrotest, dimensional check, nameplate, coating — unit by unit, with KGS-related witness points scheduled in advance rather than negotiated at the door.

Export packing. An exchanger that survives a hydrotest can still lose the voyage. Nozzle capping, moisture protection, machined-face preservation, crate reinforcement, centre-of-gravity marking. Long ocean transits punish shortcuts here more than anywhere else.

Compressed into one line: cluster procurement is a contest over who can bring a dozen units, in mixed materials, through one traceability system, one inspection regime and one acceptance line, to a Korean site.

Third-party inspection on a Korean package

Korean projects run heavier witness requirements than most, and on a cluster the inspection load compounds. It is worth being concrete about who turns up and what they look at.

Who attends. On a package of this type there can be three separate interests in the shop: the EPC's own inspector, a third-party inspection agency appointed by the EPC or the owner, and the KGS-related witness activity tied to the certification basis. They do not necessarily attend on the same day, and they do not necessarily accept each other's sign-off.

What that does to the schedule. A hold point is not a half-day. It is a notification period, a travel window, the inspection itself, and the release. Multiply by eight units and by the number of hold points in the ITP, and the inspection calendar becomes the critical path rather than the welding.

The practical consequence is that the ITP has to be agreed before fabrication starts, and hold points have to be grouped deliberately. Two units reaching hydrotest on the same day is one inspection visit; two units reaching hydrotest three days apart is two visits, and on a package with a fixed shipping date that difference is real.

Documentation as a deliverable. On a cluster, the data book is a deliverable in its own right. Material certificates, WPS/PQR and welder qualification records, NDE reports, heat treatment charts, hydrotest records, dimensional reports, nameplate rubbings, coating records — per unit, cross-referenced, and reconciled against the drawings. An EPC that cannot close its own documentation review cannot release payment, whatever the state of the equipment.

The cost of splitting a package

Bid comparison sheets compare unit prices. The costs of a split award land elsewhere, mostly on the EPC's own team, which is why they are easy to miss at award and expensive later.

Cost line	Single-shop award	Split award
Drawing review	One drawing convention, one review cycle	Separate conventions per shop; the same comment raised repeatedly
Welding qualification review	One WPS/PQR package	One package per shop, each reviewed separately
Inspection visits	Hold points grouped in one shop	Travel and notification multiplied per shop
Material traceability	One reconciliation against one set of mill certificates	Multiple mills and certificate formats to reconcile
Schedule float	Managed inside the shop	Absorbed by the EPC; package ships when the slowest shop ships
Site fit-up	Interfaces settled on one set of drawings	Inter-supplier negotiation chaired by the EPC

A split award can still be the right answer — where capacity is genuinely constrained, or where one shop cannot cover the material range. The decision is worth taking on those grounds rather than on a unit-price comparison that leaves the coordination cost off the sheet.

The wider Korean record

The GS E&C package is one slice. Across the Korean market we have delivered fixed equipment repeatedly on the same ASME U-Stamp + KGS basis — including large exchange and vessel equipment for an Air Products project in Korea (columns in the DN4500 class, direct-contact aftercoolers in the tens of tonnes), and a batch of air receivers, desulphurisers, PSA vessels and steam generators tied to Samsung-related projects.

The point of listing that is this: entry into a Korean EPC supply chain is not a one-shot transaction — it can become steady repeat business, provided the dual qualification is real, the traceability holds, and cluster management does not slip.

An evaluation checklist for EPC procurement

For selecting an overseas shop for a heat exchanger cluster on a Korean unit, these six lines can go straight into an internal review:

Criterion	Pass line	Why it matters
ASME U-Stamp	Held and current	Technical basis. Missing = out.
KGS approval	Already held, not "we'll apply after award"	6–12 months to obtain — no tender window is that long.
Material traceability	EN 10204 3.1/3.2; carbon and stainless in separated bays	Multi-material cluster; one traceability gap holds the whole batch release.
Cluster delivery record	Documented multi-unit single-package delivery	Shows the shop can carry interfaces and coordination, not only build one shell.
ITP / FAT co-operation	Accepts EPC + third-party witness points	Korean projects run heavy witness requirements.
Export packing capability	Demonstrated long-haul preservation and securing	Where exchangers most often get damaged.

Unit price alone is not enough to shortlist on. Putting "is KGS already in hand" and "is there a real cluster delivery record" at the top of the sheet avoids the common outcome: the cheapest quotation that fails on schedule.