The TIA-942 Tier IV standard mandates that redundant power distribution paths be physically separated to prevent a single physical event (fire, water leak, structural failure) from disabling both feeds. This separation is quantified through the Minimum Separation Distance (MSD) between the A-feed and B-feed conduits. For power cables carrying up to 600 V, the NEC Article 300.20(B) requires that the A and B conduits maintain a minimum of 25 mm (1 inch) of concrete encasement between them when routed in the same slab, or be separated by a 2-hour fire-rated barrier. In practice, Tier IV designs specify 0.9-1.2 m (3-4 feet) of physical separation to accommodate maintenance access and to prevent a localized fire (e.g., from a faulty splice in one conduit) from igniting the adjacent conduit. The separation distance is mathematically linked to the adiabatic temperature rise in a fault condition: for a 200 kA fault arc lasting 6 cycles (100 ms at 60 Hz), the arc temperature reaches 5,000-15,000°C, radiating heat at a rate of I²R_loss × t_arc. The thermal flux at a distance d from the arc is Φ = I²R_loss × t_arc / (4πd²). For an inter-conduit spacing of 0.3 m (1 foot), the thermal flux in a 200 kA fault at 480 V with R_arc = 0.1 Ω is 200,000² × 0.1 × 0.1 / (4π × 0.3²) = 4 × 10¹⁰ × 0.01 / 1.13 = 354 MW/m²—sufficient to vaporize copper conduit within 1 second. At 0.9 m (3 feet), the flux drops to 39 MW/m², which the adjacent conduit can withstand for 12-15 seconds before its structural integrity is compromised, exceeding the 100 ms fault clearing time by a factor of 120×.

The conduit fill ratio and ampacity derating interaction is the second physical isolation constraint. When multiple power cables share the same conduit, the NEC Table 310.15(B)(3)(a) requires ampacity derating based on the number of current-carrying conductors. For 4-6 conductors in a single conduit, the ampacity is derated to 80% of the base rating; for 7-9 conductors, to 70%. In a 2N redundant PDU architecture feeding two separate Power Distribution Units (PDU-A and PDU-B) from two separate UPS systems, the conduits must be kept separate not only for fault isolation but also to avoid exceeding the 4-conductor derating limit in a shared conduit. If the A-feed (3 phases + neutral + ground = 4 CCCs) and B-feed (3 phases + neutral + ground = 4 CCCs) share the same conduit, the conduit contains 8 CCCs, requiring a 70% derating. For a 600 MCM copper conductor rated at 420 A at 90°C, the derated capacity is 420 × 0.70 = 294 A—insufficient for a 400 A PDU feed, requiring an upgrade to two parallel 350 MCM conductors per phase (which adds 8 more CCCs, dropping the derating to 50% and making the installation both dangerous and inefficient). The practical implication is that conduit diversity is not just an availability requirement but also a conductor sizing constraint that the redundancy model must capture when computing the physical feeder cost.

The common-mode failure probability for dual-feed conduits is the ultimate reliability metric that the Markov chain model must incorporate to accurately reflect Tier IV availability. Given two conduits separated by distance d, the probability that a single external event (e.g., a backhoe excavation strike) damages both conduits is P_CM = P_excavation × P_width_over_d where P_excavation is the annual probability of an excavation event in the vicinity (typically 0.01-0.05 for urban campus environments per km of conduit run) and P_width_over_d is the probability that the excavation trench width w exceeds the separation distance d. For a standard backhoe bucket width w = 0.6 m and separation d = 0.9 m, P_width_over_d = 0.6/0.9 = 0.67, giving P_CM = 0.03 × 0.67 = 0.02 (2% annual probability). For a separation of d = 1.5 m, P_width_over_d = 0.6/1.5 = 0.4, giving P_CM = 0.03 × 0.4 = 0.012 (1.2% annual probability). The Tier IV requirement that P_CM be below 0.1% (one event per 1,000 years) can only be met by routing the two feeds through physically separate trench paths on opposite sides of the building or campus, with a trench separation of at least 3-5 meters.

The automated transfer switch (ATS) zone overlap is the final physical isolation consideration. In a Tier IV facility with dual-bus architecture (Bus A and Bus B), the ATS for each bus must be physically separated in different electrical rooms or different quadrants of the same room with a physical firewall between them. The IEC 60947-6-1 standard for ATS equipment requires that the two power sources within the same ATS enclosure maintain at least 25 mm of air clearance between the source inlet terminals (creepage distance for 600 V, pollution degree 3). However, when a mechanical ATS opens its switching contacts during a source transfer, a 0.5-2 ms arc can bridge the clearance, causing a phase-to-phase fault between the A and B sources. The arc energy at 480 V and 2,000 A prospective fault current is E_arc = V × I × t_arc = 480 × 2,000 × 0.002 = 1,920 J. This arc must be contained within the ATS enclosure without compromising the physical isolation between the two power sources. ATS designs that achieve Tier IV certification use vacuum interrupters or arc chutes with insulated barriers to prevent arc propagation to the alternate source terminals. Our redundancy model includes an ATS common-mode failure factor (typically 10⁻⁶ per ATS switching operation) based on the manufacturer's type-test data for arc containment, and incorporates this into the overall dual-feed path availability calculation. This provides the operator with a comprehensive physical isolation assessment that combines the conduit separation, trench distance, and ATS design into a single availability figure for each redundant power path.