Space station power management, circa 1986

What the work claims

The paper claims two things. First, that a complete reference PMAD architecture for the then-future Space Station can be specified at the component level, and it proceeds to do so in thirteen figures: photovoltaic and solar-dynamic reference architectures, array regulation and power conversion behind the alpha and beta joints, main bus switching units, remote power controllers feeding user loads across the booms and keels, and an uninterruptible power path backed by regenerative fuel cells.¹ Second, that the open distribution-voltage and frequency question should be settled by hardware: the advanced-development plan lists power components, a power management integrated control network, system simulation, and a single-string 20 kHz AC demonstration at 40 kW scale, to be run in the Lewis Research Center Power Systems Test Facility.¹

Be honest about the artifact: this is a schematic-level conference record. It contains no efficiency figures, no mass estimates, no test results. Its claims are architectural, and its one quantitative anchor is the planned demonstration's power level.

How it works

The heart of the paper is Figure 1, a trade tree for distribution that still reads like a checklist. High-voltage DC offers simple distribution and high efficiency but needs power electronics for every voltage change and is difficult to switch; high-frequency AC changes voltage easily with a transformer and switches easily, at the cost of a more complex distribution network and lower conversion efficiency; within AC, high frequency is light and quiet but in 1986 not widely used, while low frequency (the paper cites 400 Hz) is heavier, possibly noisy, and mature.¹ The tree ends in a box that is the paper's real thesis: many concepts look attractive, and they need an unbiased assessment rather than advocacy.

The reference architecture resolves the joints first, because a station whose solar arrays track the Sun rotates against a fixed structure. Power crosses the beta gimbal of each array wing and the alpha joints of the transverse booms through rotary power transfer devices, then passes through array regulation and power conversion into inverter-fed AC buses on the booms. Main bus switching units (MBSUs) tie the buses together and to the keels; remote power controller units meter and switch power out to users, from reaction control systems and attached payloads to the communication and tracking antennas, the satellite refueling bay, and the mobile servicing system.¹ Inside the crewed module the same hierarchy repeats at lower power: an exterior power bus feeds lights, environmental control and life support, thermal control, an orbiter power supply, and workstations through local controllers, with an uninterruptible power supply and an energy storage controller supervising a regenerative fuel cell for eclipse survival. A parallel solar-dynamic reference architecture swaps the arrays for two solar-dynamic systems behind beta joints, feeding a conditioning unit and the same MBSU-centered bus.¹

The demonstration plan, Figure 13, closes the loop: a 40 kW solar array simulator drives a 40 kW resonant inverter that establishes a single-string 20 kHz AC bus, feeding 20 kW of loads and two DC-AC converter branches at 120 VDC, one into a fuel cell simulator and one into an electrolyzer simulator, each at 20 kW, all under a power management control layer with D-controllers.¹ The supporting cast was physical: Figure 12 shows the Lewis Power Systems Test Facility floor plan at 15,000 square feet, with test area, assembly and shop space, control and power processing laboratories, and battery storage.¹

Steelman: the best case for the 1986 approach

The architecture got the enduring things right. Concentrating switching and protection in remotely commanded bus units, separating power, data, and uninterruptible-power paths, treating rotary joints as first-class power components rather than afterthoughts, and giving storage its own controller are exactly the patterns that flew on the ISS and that every large platform PMAD since has reused. The regenerative fuel cell as eclipse storage beside a photovoltaic primary, with the solar-dynamic option kept warm in parallel, shows a genuinely neutral generation trade rather than a favored solution.

The process argument was equally sound. Figure 1's demand for an unbiased assessment, backed by a named facility with a floor plan and a single-string demonstration at meaningful power, is what disciplined architecture development looks like: settle the bus question with a resonant inverter at 40 kW and real simulators on both storage legs, not in a chart. Even the 20 kHz bet was rationally placed: distribution mass scales with magnetics and bus hardware, and raising frequency shrinks both, which is the same physics that drives modern high-frequency DC-DC converters and wireless power magnetics.

Where a skeptic should push

The load-bearing assumption is that high-frequency AC distribution would mature fast enough to bet a station on, and the paper flags the weakness itself: high frequency is, in its own words, not widely used.¹ An exotic bus frequency means every converter, every switchgear line, and every test rig in the program is custom; the component industrial base that makes 400 Hz or DC hardware cheap does not exist at 20 kHz. The efficiency penalty listed on the trade tree compounds across a station with dozens of conversion stages. And the demonstration as drawn proves integration, not performance: a single string at 40 kW with simulated array, simulated fuel cell, and simulated electrolyzer says nothing about ripple, fault clearing, electromagnetic compatibility at kilohertz, or connector behavior under load cycling, because the record reports no measurements at all.

TRL honesty matters here: nothing in this record is flown or even tested hardware as documented. It is a component-level design study plus a planned ground demonstration, at the equivalent of early component stage. Finally, the choice the station program eventually made is informative: the ISS as built distributes high-voltage DC through main bus switching units descended from these very drawings, the branch of Figure 1 marked simple distribution and efficient, with the difficult-switching problem solved by solid-state remote power controllers rather than avoided. That outcome does not invalidate the 1986 analysis, but it does mean the 20 kHz branch remained a demonstration plan, not a deployed standard.

What a 1986 bus schematic still teaches PMAD

Implication: the non-obvious lesson is that distribution standards are infrastructure lock-in decided before the loads exist. Teren's Figures 2 through 7 draw a station that is mostly empty payload bays wired to a bus nobody has built yet; whoever sets the bus voltage, frequency, and switching topology constrains every user that follows for the life of the platform. That is why Figure 1's plea for an unbiased assessment outranks any particular branch: the cost of a wrong PMAD standard is paid by every later payload in conversion mass and lost efficiency.¹

Threat: the genuine threat the record exposes is standard fragility on the wrong axis. A program that chooses a technically elegant bus frequency its suppliers cannot industrialize strands itself with custom components, custom test gear, and a qualification burden it must carry alone. For today's lunar surface grids and large in-space platforms, the same trap appears as tempting non-standard voltages or frequencies that optimize one trade study while taxing every piece of user hardware for decades.¹

Opportunity: the opportunity is that the hard parts identified here are where modern PMAD gains actually come from. Resonant inversion at tens of kilowatts, high-frequency magnetics, remotely commanded solid-state switchgear, and storage under its own control hierarchy are the ancestors of today's high-specific-power converters and of wireless power transfer hardware. The 1986 program treated the rotary joint, the bus switch, and the converter as co-designed components; any present-day effort wiring large slewing arrays, rotating habitats, or distributed surface grids faces the identical trio and can inherit the decomposition rather than rediscover it.¹

The bottom line

Calibrated reading: as engineering history, this record is a clean snapshot of PMAD thinking at the moment the Space Station stopped being a sketch and started being a wiring problem, and its component decomposition aged well. As evidence, it supports architectural claims only: no efficiency, mass, or reliability figure should be quoted from it, and the 20 kHz demonstration it describes is a plan, not a result, at early component maturity. The trade tree's bottom line survived contact with history: the simple, efficient, hard-to-switch branch (high-voltage DC with remotely switched protection) is what large platforms actually fly, while the light, quiet, immature branch stayed on the bench. Confidence is high that the framework still frames; confidence is nil on any number this record does not contain. What would have confirmed the high-frequency branch: published 40 kW demonstration data showing fault clearing and compatibility; what broke it instead was the industrial-base reality the paper itself flagged.¹