"Modular or stick-built" is argued as a construction preference. It is not. It is a design constraint that runs backwards from the road.
A module has to fit on a truck, pass under the bridges on the route, and be liftable by the crane that will be on site. Those three limits are fixed before the first frame is detailed, and they decide what a module can contain. A design that is modular in principle but 5 m (16 ft) too wide never leaves the yard.
1. Start from the transport envelope, not the plant layout
The module's size comes from four numbers: the width allowed on the route, the height allowed under the lowest obstruction, the length a trailer can carry, and the weight the site crane can pick.
Width is usually the binding one. A load in the order of 3 to 3.5 m (10 to 11.5 ft) wide travels widely; beyond that, permits and escorts appear, and the route may have to change. Height under bridges is often more generous than people assume, but it is checked per route, not assumed.
Working backwards from those limits produces the module grid: a rack bay that is 6 m (20 ft) wide and 12 m (40 ft) long fits the envelope comfortably, while a bay that is 9 m wide does not. Once the grid is set, the piping layout has to live inside it, which is why modular design is a layout discipline as much as a fabrication one.
2. What a module should contain, and what it should not
The value of a module is that work moves from a site to a yard where people, tools and weather are controlled. That value is lost if the module contains the items that are hardest to align.
Structure, pipe, tray and supports travel well. Instruments, valves with long stems, and anything that needs field calibration travel badly. Tie-in spools are a special case: the module should stop short of the final connection, leaving a short piece that is cut and fitted on site to absorb tolerance. A module that tries to include the tie-in will not fit.
The other item to leave out is anything that would make the module too heavy for the crane. A module that is structurally complete but cannot be lifted is a module that has to be split on site, which defeats the purpose.
3. Connection details decide whether modular works
A modular project succeeds or fails at the interfaces, and the interfaces are mostly between adjacent modules.
Two modules that must bolt together need their connection points to be set out accurately enough that the bolts pass — in practice, within a few millimetres over a length of several metres. Achieving that in a fabrication yard is possible with the right jig; achieving it when the two modules are built by different suppliers is where projects lose days.
The practical answers are to use a joint that tolerates movement — slotted holes, a packer, a splice designed with adjustment — and to build a trial assembly of at least two adjacent modules before shipping. A trial assembly costs a day in the yard and saves weeks on site.
Lifting points deserve the same attention. A module is picked by four slings from four points, and those points should be designed into the frame rather than found on the day by wrapping a sling around a brace. A module lifted from a point that was never intended will distort, and a distorted module will not close its joints, which returns the problem to the interface that was supposed to be solved in the yard.
4. Stick-built is not the old way; it is the right way on some sites
Where the site has a crane, a workforce and a yard, stick-built is often better.
Its advantages are real: no transport limit on the module size, no trial assembly, no interface between fabrication yards, and a frame that can be adjusted as it goes up. On a site with good access, a competent erector and a long programme, stick-built is simpler to manage.
Its disadvantage appears when site labour is expensive, scarce, or unreliable in the rainy season — which is the situation modular exists to solve. The honest comparison is not cost per tonne of steel; it is the number of site hours, the risk of those hours being rained off, and the quality of the connection that results.
5. The hybrid that most projects actually need
In practice most projects land in between: modules for the repetitive, congested parts, and stick-built for the parts that have to fit existing structures.
A rack that runs from a new module to an existing pipe bridge cannot be modular at its far end, because the existing structure will not be where the drawing says. So the last bay is stick-built, measured on site, and the module stops one bay short. That is not a compromise; it is the correct answer, and it should be designed as such rather than discovered.
The same applies at height. Anything above the reach of the site crane is either modular or it does not get built, so the module boundaries follow the crane's envelope, not a neat division of the process.
6. What I want before drawings
- The transport route and its width, height and weight limits, checked rather than assumed.
- The site crane's capacity at the radius it will work, not its maximum.
- The piping layout, to see which parts are congested enough to justify a module.
- The interface with any existing structure, and who measures it.
- Whether a trial assembly is possible, and where.
- The wind speed to be used, taken from the applicable code at the project site rather than a round number — a module is designed for transport and for lifting as well as for service, and the lifting case is often the one that governs the bracing.
Modular is a decision made by the road and the crane. Make it at concept stage, and the design follows. Make it at fabrication stage, and the modules follow the design into a yard they cannot leave.