Standard models cover the common operating envelope; once the demand curve, site conditions or control interface no longer fit, customization becomes necessary. But customization does not mean anything can be changed — its value lies in translating site constraints into a design that can be built and verified. Judging whether a customized nitrogen generator system is sound rests on three questions: are the requirements clearly defined, are the trade-offs justified, and is delivery verifiable? The sections below cover applicability, requirement inputs, design trade-offs, structural limits, control interfaces, site condition checks, delivery verification and misconceptions.

　　First, when customization is needed. Five situations are common. The first is an unusual demand profile, such as a large flow for a short period and almost nothing the rest of the time, where a standard unit lacks buffering capacity. The second is a purity requirement outside the usual range, needing different adsorbent and switching timing plus a fresh check on dew point and oil content control. The third is a constrained site where standard dimensions cannot pass the equipment room or maintenance access. The fourth is integration with an existing control system whose protocol and interlock logic differ from standard. The fifth is integration with other equipment, such as sharing control with the compressor station, dryer or pipe rack. Where requirements fall inside the standard envelope, customization only adds cost and verification burden. Start from the standard model, compare differences item by item, and enter customization only where a real conflict exists, recording the conflict in the requirement document.

　　Second, what the requirements must state. At least six items: required purity and acceptable fluctuation; average and peak demand with the demand curve; number of points of use, distances and pressure requirements; intake conditions including pressure, temperature, oil and water content; installation site dimensions, load bearing, ventilation and power supply; and control interface and data requirements. Only with these six can a design have a basis. The requirement document should also note where the data came from and when it was collected, so departments do not supply conflicting figures. A frequent failure is giving a single daily consumption figure without peaks or a curve, producing a unit sized too small or too large, with the mismatch discovered only after commissioning. Exporting historical demand records from existing instrumentation markedly improves sizing accuracy and provides a basis for later expansion.

　　Third, design trade-offs. The first is purity against energy: higher purity raises energy per unit of gas, so confirm whether every point of use needs high purity, splitting supply where appropriate. The second is flow against buffering: larger tanks shave peaks but raise footprint and cost, while smaller tanks place the burden on fast response. The third is redundancy: a single unit costs less but loses gas the moment it stops, while parallel or backup units raise availability at added investment. The form of redundancy also needs choosing — main unit backup, key component backup, or temporary external supply — with markedly different response and cost. The fourth is control approach: simple start-stop control costs less, while load-following control saves energy but complicates commissioning. Every trade-off should carry a written rationale rather than relying on experience. The trade-off document should also note which parts can be extended and which require replacement if requirements change, supporting later planning.

　　Fourth, structural customization limits. Customization often involves frame dimensions, pipe entry and exit positions, maintenance door orientation, split arrangement, protection rating and materials. Structural changes affect transport, lifting and maintenance access, and should be checked against site conditions during design. Splitting the main unit from storage capacity suits dispersed sites or constrained transport, but piping connections and pressure loss must be recalculated. Integrated skid solutions take little space and connect quickly, yet may limit maintenance access and often require whole-unit replacement if capacity must grow. Either way, adequate maintenance and replacement space should be preserved rather than sacrificed for compactness. Structural customization should also account for turning radii and doorway dimensions along the transport route, details more easily overlooked than the equipment itself.

　　Fifth, control and data interfaces. Customized systems often connect to plant monitoring. Protocol, point list, alarm levels and interlock logic should be defined: whether low purity triggers alarm, shutdown or switchover to backup; how the system reacts to low intake pressure; and how remote and local control priorities are set. These should be written into the functional description and confirmed by both parties, avoiding repeated changes at commissioning. For data logging, define which parameters are recorded, how long they are retained and whether export is supported, supporting later analysis of energy use and degradation. Where the plant already has an energy management platform, confirm whether data can be uploaded in its existing format to avoid creating another silo.

　　Sixth, checking interfaces and site conditions. Interfaces are where customized designs most often fail: intake pipe diameter and pressure, power capacity and grounding, drainage and discharge destinations, foundation bearing and vibration isolation, noise limits. If these are not checked at design stage, site modifications follow after delivery, adding cost and delay. Drainage destination is especially easy to miss; without a suitable discharge point, commissioning stalls. A site condition confirmation sheet signed item by item by the user, as part of design input, is advisable. It should carry measured data rather than drawing-based estimates, particularly for power capacity and floor bearing.

　　Seventh, delivery and verification. Customized projects are verified in three layers: factory testing for purity, flow, pressure and alarm logic; site commissioning for interfaces, interlocks and running stability; and a trial period observing purity fluctuation, dew point, drainage and energy use. All three records should be retained and compared item by item with the design. Acceptance criteria should be fixed at contract stage, stating measurement methods and positions, so disputes do not arise when the same parameter reads differently in different places. Deviations should be explained and their impact assessed rather than recorded as simply commissioned. If purity fluctuation exceeds expectations during the trial, check intake conditions and drainage first before adjusting timing parameters.

　　Eighth, common misconceptions. First, giving average demand without peaks or a curve. Second, supplying every point of use at the highest purity, wasting energy. Third, sacrificing maintenance space for compactness. Fourth, control logic not confirmed in writing, causing repeated changes at commissioning. Fifth, site conditions unchecked until modifications are needed after delivery. Sixth, redundancy not considered, so a stop means no gas. Seventh, acceptance looking only at whether gas is produced, not at fluctuation and stability. Eighth, customized portions not verified specifically, relying on standard model reports. Ninth, requirement changes undocumented, so which design version matches which equipment cannot later be traced.