Why monthly consumption produces the wrong answer
Dividing a hospital’s monthly kilowatt-hours by thirty produces an average day. It does not show which loads must remain on, their peak coincidence, their starting current, how quickly a source must respond or what can be shed during a long outage. Two hospitals with the same monthly bill can require very different storage because one has a well-segregated essential bus while the other has mixed clinical and comfort loads on common feeders.
The sizing boundary also matters. Battery nameplate energy is DC inventory at a reference condition. The hospital receives usable AC energy after the allowed state-of-charge window, ageing allowance, conversion loss and auxiliary consumption are considered. If the PCS, transformer, cooling or controls are excluded from the boundary, a seemingly precise kWh figure can be misleading.
Start with power: the kW question
The first calculation asks whether the power-conversion system can hold the protected bus. Continuous kW must cover the intended steady load with reserve. Short-duration overload and dynamic response must be checked against motor starts, transformer energization, imaging equipment, compressors, pumps and load restoration steps. A specification that lists only battery kWh leaves this fundamental duty unresolved.
Power quality and fault behavior belong in the same review. Voltage unbalance, harmonics, reactive demand and low power factor can consume inverter capacity. In island mode, available fault current may be much lower and differently shaped than utility fault current. Protection that worked on the grid may not operate as intended in an inverter-dominated island. These are study and test questions, not assumptions to bury inside a vendor model.
- Measured peak and step-load profile on the proposed protected bus.
- Motor and transformer inrush, sequencing and restart logic.
- PCS continuous, transient and reactive-power capability.
- Island-mode fault contribution and protection sensitivity.
- Redundancy after one module, PCS block or control component is unavailable.
Then calculate energy: the usable kWh question
Energy sizing integrates the critical-load profile over the required scenario. The calculation should use the actual time series where possible, not one average kW value. It then adds the approved reserve, conversion losses, auxiliaries, temperature effects, expected degradation and any design contingency. The result is usable energy at the hospital bus, which can be translated into the required installed DC capacity for candidate technologies.
A credible model keeps emergency reserve visible during normal operation. If the BESS performs daily peak shaving, its dispatch controller needs a rule for preserving or restoring the resilience reserve. That rule may depend on weather, generator readiness, outage warnings or operating policy. A savings model that cycles the battery to its lower limit immediately before a brownout has not delivered the promised resilience.
| Capacity layer | Meaning | Why it changes |
|---|---|---|
| Nominal DC kWh | Catalogue battery inventory | Reference temperature and beginning-of-life condition |
| Allowed SOC window | Operational range permitted by design | Reserve, cell limits and warranty rules |
| Degradation allowance | Capacity expected later in life | Calendar age, cycles, temperature and duty |
| Conversion + auxiliaries | PCS, transformer, HVAC and controls | Efficiency varies with load and environment |
| Usable AC kWh | Energy available at the defined bus | The quantity an autonomy model should consume |
Use 4, 8, 12 and 24 hours as scenarios—not universal legal promises
Four-hour, eight-hour, twelve-hour and twenty-four-hour cases are useful commercial and engineering scenarios because they expose different architectures. They are not presented here as universal PEC autonomy requirements. The current licensed code, hospital type, designated functions and AHJ interpretation must determine any mandatory duration and source arrangement for the actual project.
A four-hour case may test whether storage can bridge frequent distribution interruptions without starting a generator for every event. An eight- or twelve-hour case may cover an overnight event or a long feeder outage with scheduled load shedding. A twenty-four-hour case tests fuel, solar variability, staff procedures, recharge capacity and the feasibility of carrying only a tightly defined clinical bus. Longer events usually become an energy-supply problem rather than a battery-only problem.
| Scenario | What it helps test | Common omitted issue |
|---|---|---|
| 4 hours | Frequent outage bridge and short island | PCS overload and reserve after peak shaving |
| 8 hours | Extended feeder interruption | Night load, cooling and auxiliary energy |
| 12 hours | Overnight critical service | Generator failed start and staff handover |
| 24 hours | Full-day resilience plan | PV uncertainty, fuel logistics and recharge |
| Multi-day | Disaster continuity | Load rationing, maintenance, replenishment and recovery |
Generator and PV availability change the optimum
A reliable generator can reduce the energy capacity needed from the BESS, while the BESS can reduce generator starts, absorb step loads and allow the engine to operate in a more stable loading band. The model must include failed-start probability, maintenance periods, fuel on site, refuelling access, minimum loading, start time and the charging power available after the generator connects.
PV can replenish storage during a long daytime outage only if the island architecture permits it and the grid-forming source can regulate the bus while accepting variable solar power. The study should use conservative weather sequences and allow for curtailment when batteries are full or load is low. A yearly solar-yield estimate is not an outage guarantee.
Size the recovery, not only the outage
An autonomy model is incomplete when it ends at zero state of charge. The hospital needs a recovery strategy: which source recharges the battery, how much headroom remains for clinical loads, how quickly deferred loads return, and what happens if the utility supply is unstable after restoration. Charging too aggressively from a generator can overload it; restoring every motor together can create a second interruption.
A defendable specification therefore lists acceptance tests for representative states. Tests should include loss of utility, permitted transfer time, step loading, low SOC, generator start and synchronization where applicable, PV curtailment, communications loss, manual fallback and controlled restoration. Capacity is accepted as a delivered operating result, not as a number on a nameplate.
The S3 sizing deliverable
S3 begins with a critical-load register and measured profile, then builds a scenario matrix. Each scenario records protected functions, peak kW, usable AC kWh, allowable interruption, autonomy, source availability, reserve, failure assumptions and recovery. Options can then be compared on lifecycle cost and residual risk.
The result is normally a range rather than false precision: a minimum technically viable configuration, a recommended operating case and the consequences of a higher-resilience option. Final equipment selection follows only after the architecture, current PEC route, BFP strategy, utility interface and Philippine professional responsibilities are confirmed.