China's first smart-manufacturing satellite plant, operated by CASIC Space Engineering Development in Wuhan, has been publicly stated to produce up to 240 satellites per year off a single integrated line since commissioning on 2021-05-13 [S6]. That throughput number is the reference datum every Western smallsat planner is now benchmarking against when sizing a new pressure sensor and AIT-floor footprint for 2026.
The capacity question has changed: with LEO broadband constellations such as Starlink, OneWeb, Kuiper and Guowang pulling forward build orders, the limiting resource is no longer aluminium billet or composite layup, but cleanroom-integrated assembly, integration and test (AIT) hours, PCBA panelization yield, and qualified flow meter test stands for propulsion bench runs. Capacity-planning models must therefore be hybrid: a discrete-event model for the bus line, paired with a finite-capacity MRP loop for propulsion and electronics sub-assemblies.
What "capacity" means in a satellite plant, and which standard governs the math
Production capacity is the maximum output a fixed-asset system can sustain under a defined organization, shift, and product mix; for a satellite line it is normally expressed in satellites per year or in equivalent AIT-hour budgets per quarter [S8]. The 2021-05-13 CASIC disclosure of 240 units/year per line is equivalent to roughly one satellite per 1.5 working shifts on a single-shift, 240-day calendar, or 1.0 satellites/day on a two-shift, 300-day calendar [S6].
Capacity-planning methodology itself is described in classical terms as the process of determining what hardware and software configuration is required to meet application needs, and is explicitly not an exact science: every application is different and every user behaviour differs [S1]. Applied to satellites, that translates into a per-program re-baseline: a 200 kg LEO bus and a 6U CubeSat share almost no common bottleneck, so the capacity envelope has to be recomputed when the product mix shifts.
Selection criteria: discrete-event vs. MRP/finite-capacity vs. constraint-based
Three capacity-planning styles are in active use across the aerospace ERP ecosystem. SAP PP/PP-PI and SAP APO with DDMRP covers discrete, process, and repetitive manufacturing via work-center routing, control recipes, process orders, and S&OP, with 1084-page reference texts documenting the full workflow [S3]. Oracle NetSuite Advanced Manufacturing defines capacity as the maximum amount of work a work center can complete in a given period, then evaluates available capacity against planned orders to flag overload [S7]. The third style is constraint-based "theory of constraints" drum-buffer-rope, used inside many smallsat AIT cells where the thermal-vacuum chamber is the drum.
Selection gates are concrete: choose SAP PP-PI when batch traceability, MRP re-explosion, and S&OP are required across multiple plants; choose NetSuite Advanced Manufacturing when the operation is a single SME site with cloud-native deployment and QuickBooks-style finance integration [S5][S7]; choose a constraint-based in-house tool when the thermal-vacuum chamber, the industrial valve test rig, or the anechoic chamber is the single physical bottleneck dominating takt. Capacity-planning software rated for African SME use in 2026 is dominated by Fishbowl, Katana, and similar cloud MRP suites, illustrating that the planning layer has fully moved off spreadsheets even in mid-tier factories [S5].
Who satellite capacity planning is FOR, and who it is not for
It is for integrators running 12+ satellites per year with a fixed bus architecture, plus a propulsion line feeding both commercial and government programs, plus a pressure transmitter calibration loop that must be auditable per AS9100. It is also for any smallsat constellation prime trying to model 50-200 vehicle/year ramp curves where the AIT floor is the gating asset. A practical sibling discipline here is drone production line design: the cell-layout, takt, and compliance math transfers almost directly, because a 50 kg class II UAV line and a 100 kg smallsat bus line share more workflow DNA than either shares with traditional aerospace. [S6]
It is NOT for one-off science missions, university CubeSat programs building 1-2 vehicles per year, or for organizations without a controlled MRP feed. For those, a single Excel takt sheet plus an AIT Gantt is more honest than over-engineering an SAP APO rollout. A related cost-side question is the drone manufacturing cost breakdown: the BOM drivers (composites, IMU/gyro stack, flight servo motor, PCBAs) are close enough that a planner can sanity-check smallsat avionics costs against mature drone data.
Comparison of the three main capacity-planning styles on decision criteria
On four engineering criteria, the styles line up as follows. Coverage scope: SAP PP/APO spans discrete + process + repetitive; NetSuite Advanced Manufacturing is discrete-first with cloud MRP; constraint-based is single-bottleneck focused [S3][S7]. Implementation cost and lead time: SAP is 6-18 months with a partner; NetSuite is weeks; constraint-based is days on a whiteboard. Required data fidelity: SAP requires work-center routing, control recipes, and S&OP master data; NetSuite requires work-center capacity definitions; constraint-based requires only drum cycle time and buffer sizing. Suitable plant scale: SAP fits multi-plant primes; NetSuite fits single-site SMEs; constraint-based fits any single-AIT-cell operation including a university cleanroom. The capacity-planning questions Oracle lists for server sizing (transaction concurrency, SSL overhead, cluster failover) [S1] map almost one-to-one onto satellite line sizing: concurrent AIT slots, encrypted downlink bench time, and cleanroom failover when a HEPA bank trips.
Real use cases: CASIC 240/yr, Western smallsat lines, and the propulsion bottleneck
CASIC Space Engineering Development's Wuhan line, with a stated 240-satellite annual capacity from a smart-manufacturing plant template, is the most-cited throughput benchmark for 2026 capacity reviews [S6]. Western primes are pushing smallsat lines toward similar numbers by replacing serial AIT with parallel AIT cells, and by moving PCBA panelization from 100 mm x 100 mm panels to 300 mm panels that share PLC-controlled pick-and-place lines with adjacent drone and automotive programs.
For drone-program planners, the cross-over is direct: the industrial drone adoption pattern, with spec maps and selection gates for payload, endurance, and BVLOS, is mirrored in satellite bus selection, where the same spec-map discipline is applied to payload mass, orbit altitude, and link margin. The practical signal in 2026 is that a satellite line targeting more than 50 vehicles/year without parallel AIT cells will queue, while a line that consolidates propulsion test onto a single shared flow meter rig will be capacity-bound on the test cell, not the bus line.
Limitations, failure modes, and sourcing constraints
The dominant failure mode in 2026 satellite capacity planning is cleanroom and thermal-vacuum chamber saturation, not raw machining; a second is qualified-part allocation for rad-hard pressure transmitter and IMU devices, which gates ramp curves independently of line count. A third is composite layup yield, where a 2-3% scrap rate silently removes 5-7 vehicles/year off a 240/year target [S6]. The sourcing side interacts with broader industrial trends: PCB panel sourcing split between Chinese fabs and regional EMS now drives avionics lead time, and the related PCB manufacturer market share data set is the right reference to bracket lead-time risk in 2026.
Trackable signals into Q4 2026: (a) whether any Western prime publicly commits to a 200+/year smallsat line, matching the CASIC Wuhan disclosure [S6]; (b) whether AS9100D audits in 2026 start citing capacity-planning software validation as a finding, mirroring how ISO 9001 audits now cite MRP records; (c) the next update of NetSuite Advanced Manufacturing capacity documentation, which currently defines capacity as maximum work per work center per period [S7]. A single composite layup cell, a single TVAC chamber, and a single propulsion test stand will continue to set the line ceiling for most primes through 2026.