Membrane package sizing looks deceptively simple on a datasheet: pick a flow, pick a purity, order a skid. In practice the three variables interact, and the duty cycle underneath them decides whether the package delivers at the loading rack or only on paper. Engineers who size well start from the site, not the catalog, and the reasoning is worth walking through.
Flow: the variable everyone overestimates
Rated throughput is not the sizing flow. The vapor stream arrives in waves shaped by the loading schedule, and the package must handle the peak without being condemned to idle through the troughs. Oversizing wastes capital and, less obviously, hurts membrane performance at low flows; undersizing pushes vapor past the recovery stage during peak unloading. The honest sizing flow comes from measured or modeled loading patterns, not from a nameplate tank volume.
Purity: decide what the recovered product is for
| Purity target | Typical driver | What it costs in the design |
|---|---|---|
| Return to storage | Recovered fuel goes back into product tanks | Higher stages, tighter control, richer monitoring |
| Emission compliance only | Meeting the vapor limit at the stack | Fewer stages, simpler controls, lower cost |
| Both, site dependent | Mixed regulatory and commercial pressure | Configurable purity with documented switch points |
Duty cycle: the variable everyone underestimates
A package sized for steady flow will disappoint a batch-loading site. The duty cycle brings start-stop behavior, thermal swings and condensate waves, and the design must absorb all three:
- Build the loading profile across a full week, including the delivery waves after each drop;
- Match the package turndown to that profile rather than to a single rated point;
- Check start-stop frequency against what the membranes and controls tolerate;
- Size the condensate handling for the post-delivery wave, not the average day;
- Verify the hottest and coldest expected days both sit inside the operating envelope.
Where sizing goes wrong in real projects
The recurring failures are predictable. Sizing from nameplate instead of measured flow produces a skid that surges. Sizing purity without asking what happens to the recovered product produces either an overbuilt train or an underbuilt one followed by an expensive stage addition. And ignoring the duty cycle produces a package that passes a steady-state test and alarms its way through a real Tuesday. Every one of these is cheaper to fix on paper than on site, which is why the sizing conversation deserves its own meeting, separate from the commercial one.
Frequently asked questions
Can one package serve two loading bays with different products?
Often yes, provided the combined duty profile is modeled honestly and the controls can adapt to each product behavior. The failure mode is assuming the heavier product stream is a scaled copy of the lighter one, when it usually arrives with its own condensate and purity behavior.
Does higher purity always mean more membrane area?
Broadly yes, since each purity step demands more separation work, but compressor energy and stage arrangement matter as much as area. A well-configured two-stage design can reach targets a naive single-stage area stack cannot, at lower energy.
How much margin should sizing carry?
Enough margin to cover measurement uncertainty and modest future throughput growth, without stranding the package far from its efficient operating band. Ten to twenty percent, justified by the loading data, is a defensible engineering answer; a doubled nameplate is a commercial decision wearing an engineering costume.
Luoyang Wohong Petrochemical Equipment Co., Ltd. designs and manufactures vapor recovery units, membrane separation skids and condensers for fuel depots, terminals and retail sites. Contact us for duty assessment and equipment selection support.