BESS and Microgrids: A Practical Guide to Power, Energy and Control

A BESS cannot be sized responsibly from a monthly bill or a desired container count. First define the service: demand reduction, energy shifting, renewable firming, backup, frequency response, capacity support, power-quality assistance or microgrid operation. Each service has a different response time, power, energy, reserve, availability and dispatch requirement. The US DOE Energy Storage Handbook organizes storage around technologies, engineered systems and applications for exactly this reason [1].

Start by writing the operating statement in plain language. “Hold grid import below 600 kW during the utility demand interval” is testable. “Provide two hours of critical-load support after loss of grid” is testable once the critical-load profile and transition requirement are defined. “Install a 1 MWh battery” is not a use case. It is an equipment quantity with no operating promise attached.

A project may pursue several value streams, but they can compete. Energy reserved for an outage cannot also be committed to daily arbitrage. High cycling for one service can accelerate degradation and reduce future energy available to another. Dispatch priority and commercial priority must therefore be written together.

Power, expressed in kW or MW, describes the instantaneous rate of charge or discharge. Energy, expressed in kWh or MWh, describes how much can be delivered over time. A 1 MW / 2 MWh system has a nominal two-hour duration at rated power before accounting for the usable state-of-charge window, auxiliary consumption, conversion loss, temperature, degradation and operating reserve.

Nameplate energy is not the same as energy delivered at the point of interconnection. The model should distinguish DC battery capacity, usable DC window, PCS efficiency, transformer and cable loss, HVAC and controls, minimum reserve and end-of-life guarantee. NREL notes that power and energy together define BESS size and that oversizing can strand capital [2].

State of charge is an estimate inside an operating window

State of charge, or SOC, expresses the estimated remaining charge relative to a defined usable reference. It is not a direct fuel-gauge measurement. The BMS estimates SOC from electrical and temperature behavior using its supported methods, and the accuracy of that estimate can vary with operating history, calibration and cell condition.

Operators also track state of health, which describes how present capability compares with an agreed beginning-of-life reference. A system at 90 percent SOC late in life does not necessarily contain the same usable energy it held at 90 percent when new. Warranty and dispatch models should therefore state whether limits and guarantees apply at cell, rack, DC bus, PCS output or point of interconnection.

Cells form modules, modules form racks and racks form DC subsystems monitored by the BMS. The BMS supervises voltage, current, temperature, SOC estimates and protective limits. The PCS converts between DC and AC and controls active and reactive power within its capability. The EMS or microgrid controller translates site objectives into permitted setpoints while coordinating forecasts, meters, generators and operating constraints.

These controls operate at different speeds and authority levels. Cell voltage or temperature protection cannot wait for an economic optimizer. The BMS should constrain permitted charge and discharge. The PCS should respect those limits while controlling AC behavior. The EMS may choose the desired dispatch only inside the permitted envelope. Independent protection should isolate faults under the approved design even if a supervisory communication link is unavailable.

Protection, switchgear, transformer, grounding, metering, communications, HVAC, fire detection, enclosure and auxiliary power are not accessories. They determine whether the system can connect, survive faults, maintain temperature, stop safely and be operated after handover.

Reading the converter: the ratings that decide what the plant can do

The PCS is a bidirectional converter. It rectifies AC to DC to charge the battery and inverts DC to AC to discharge it, and every kilowatt the plant delivers passes through it. A battery cannot do anything its converter cannot, so the converter’s datasheet decides more about the plant’s behavior than the battery’s does. Seven lines on that sheet matter to a cooperative.

Kilowatts against kilovolt-amperes. Real power, in kW, does the work. Apparent power, in kVA, is what the converter’s electronics and the cables actually carry, and it includes the reactive power that motors and transformers draw to build their magnetic fields without consuming it. Power factor is the ratio of the two. A converter rated 105 kW with a power factor adjustable across the full range, from fully leading to fully lagging (the datasheet’s -1 to +1), which is what Bescore publishes for its microgrid cabinet, can supply reactive power to hold up voltage on a weak feeder, but every kilovolt-ampere-reactive it supplies at a given moment is capacity that is not available as kilowatts at that moment. Some sheets state the overload rating in kVA rather than kW; the 2024 Bescore container sheet gives 250 kW rated and 275 kVA overload for one unit. When comparing two offers, ask which figure is which, and at what power factor it applies.

Overload, and whether the motors will start. An induction motor started direct on line draws several times its running current, commonly five to eight times, for about a second. Island loads in the Philippines are motor heavy: deep-well pumps, rice mills, ice plants, cold rooms. Whether the plant can start them when the battery is the only source is decided by the converter’s short-time overload figure, stated as a percentage of rated power for a stated time. Bescore’s published figures for the cabinet classes Metro Power lists are modest: a maximum power about ten percent above rated for the microgrid cabinet (115 kW against 105 kW), twenty percent for the liquid-cooled cabinet (150 kW against 125 kW), and 110 percent long-term at 35 °C and 120 percent for one minute for the medium-voltage conversion station. Bescore publishes this unit at 50 Hz. A 60 Hz version is not yet confirmed. Now read that against a real load. A 30 kW pump starting at six times its running current asks for something on the order of 180 kVA or more for a moment. A 105 kW cabinet with about 10 kW of margin cannot start it alone. The options are a soft starter or variable-frequency drive on the motor, staggered starting of several motors, a generator on the bus during the start, or a larger converter. Whichever is chosen, the starting current of the largest motor belongs in the readiness data, and the start is demonstrated at commissioning.

Control modes. The shorthand on the sheet says what the converter can be told to do. PQ follows the grid and accepts real and reactive setpoints. VF makes the voltage and frequency itself and is what an island needs. VSG makes them the way a rotating generator would, with inertia and droop, so that the converter and a diesel set can share load without fighting. A converter published with PQ only cannot run an island. A converter published with VF or VSG has stated that it can, and that statement is then proven in the factory and on site. Chapter 6 explains where each mode is needed.

Ride-through. Low-voltage, high-voltage and zero-voltage ride-through describe whether the converter stays connected and keeps operating during a voltage dip or swell for a defined time, instead of tripping at the first disturbance. Rural feeders dip often. A storage plant that trips on every dip drops its support at exactly the moment the feeder needs it and then has to resynchronize. The Philippine Grid Code and Philippine Distribution Code set ride-through requirements by class of connected generation; which class covers storage at the site’s size is confirmed with the distribution utility for the current edition. Bescore publishes ride-through as a function of its medium-voltage station; it is not published for the 400 V cabinets, so ask for it.

Distortion. Converters switch current in rapid pulses, and the leftover harmonics heat transformers and disturb sensitive loads. The two lines to read are total harmonic distortion of current (THDi) and of voltage (THDv). Bescore publishes THDi under 3 percent for its hybrid inverter units and, for the OmniStax inverter, THDv under 2 percent on linear load. Bescore publishes this unit at 50 Hz. A 60 Hz version is not yet confirmed. The utility sets the limit at the point of connection; the site’s own loads set the limit inside the fence.

What 0.5P means. The P-rate is rated power divided by rated energy, per hour. A 0.5P system charges or discharges fully in about two hours; a 1P system in about one hour; a 0.25P system in about four. Bescore publishes 0.5P for its cabinets and its battery container, and the arithmetic shows it: 105 kW with 215 kWh, 125 kW with 261 kWh. The P-rate is a design choice, not a property of the battery. A 200 kWh battery behind a 50 kW converter is a 0.25P system that runs four hours but never delivers more than 50 kW; the same battery behind a 100 kW converter is a 0.5P system with half the duration and twice the power. The rate also sets the current through the cells, which is why the pack rating reads “280 Ah at 0.5P at 25 ± 2 °C”: the capacity was measured at that rate and that temperature. Work the battery harder and it delivers less, heats more and ages faster, which is where Chapter 7 continues.

Efficiency. A published maximum converter efficiency, 98 percent in the case of Bescore’s medium-voltage station, is measured at the most favorable operating point, in one direction. The plant’s round-trip efficiency includes two conversions, the battery, the cooling and standby consumption, and it is lower. Bescore does not publish a round-trip efficiency for any product, and none should be inferred from a converter figure. Figure 9A shows the difference.

Peak shaving requires a demand forecast and enough power and energy to hold the site below a target through the peak interval. Arbitrage requires tariff spread large enough to cover efficiency loss, degradation, operating cost and financing. Renewable firming needs a forecast, ramp objective and curtailment logic. Resilience needs an explicit reserve policy and critical-load model.

The shape of the peak matters. A 1 MW spike lasting five minutes can require high power but relatively little energy. A 400 kW exceedance lasting three hours needs less instantaneous power but much more energy. Monthly bills normally reveal the peak charge and total energy, not the interval shape needed to distinguish these cases.

The EMS should define priority when services conflict, recovery after communication failure, SOC targets, generator coordination, import and export limits, maintenance states and manual authority. Cybersecurity matters because the controller can influence substantial real power.

When a design needs an energy management system, and when it does not

The EMS is the plant’s decision layer. In a full plant it selects the operating mode for the day (self-consumption, demand limiting, tariff shifting, backup reserve, generator support); sets the charge and discharge setpoints each interval; holds back a state-of-charge reserve for outages; starts and stops the generator so that it runs at good load and is not cycled needlessly; orders load shedding in an island; curtails solar when the battery is full and the island cannot absorb the surplus; runs the black-start and resynchronization sequences; and raises alarms. It may use forecasts of solar and load to plan ahead rather than react. That is a long list, and not every site needs all of it.

When no separate EMS is needed. The simplest case is one converter, one battery and one site. A hybrid inverter serving an off-grid village makes the simple versions of these decisions internally: charge from solar first, start the generator on low battery, stop it when charged. A single storage cabinet with its own controller does the same for a grid-tied shop or clinic. The cabinet diagrams Bescore publishes show how its products differ in this respect. Its microgrid variant carries its own management unit, which talks to the generator over a serial link and exposes one external interface for a site controller or the cloud. Its liquid-cooled cabinet carries an EMS block inside, between the battery controller and the PCS. Its plain grid-tied cabinet carries only the battery controller and expects an external EMS over Ethernet. The question to put to a bidder is therefore not “is there an EMS” but “does this cabinet make its own operating decisions, and if so which ones, and what does it expose to my control room”.

When a separate EMS is unavoidable. Several cabinets in parallel, a diesel set, solar with its own inverter, a grid connection that comes and goes, and a tariff worth optimizing against: someone has to coordinate all of that, and no single converter’s internal logic will. This is the plant that needs an EMS with a written dispatch strategy, generator rules, reserve rules and fallback behavior. Forecast-driven strategies, which schedule the day from a solar and load forecast rather than reacting interval by interval, belong here too.

Local or cloud. The EMS can run on a controller at the site, in a cloud service, or in both with the site controller holding the safety-critical logic. A cloud-only controller adds a communications dependency. On an island served by one cellular tower, that dependency is a design risk, and the site must be able to keep running its essential logic when the link is down. Figure 5 already asks what happens on communications loss; for a cloud-hosted controller the answer must be a local fallback that has been tested.

The supervisory protocols. The cooperative’s control room and the plant have to speak the same language. Modbus TCP is the simplest, a register-based protocol almost every device supports. IEC 60870-5-104 is the telecontrol protocol most distribution control centers already use for substations, so a plant that speaks it appears in the existing SCADA without a gateway. IEC 61850 is the substation automation standard, with richer data models, used at utility scale. MQTT is a lightweight publish-and-subscribe protocol used for cloud platforms and dashboards. Inside the cabinet, the battery management system usually talks to the PCS over CAN, and the auxiliaries (air conditioner, fire panel, generator) over RS485. Bescore publishes MQTT, IEC 104 and Modbus TCP on its microgrid cabinets and IEC 61850, Modbus and CAN on its medium-voltage station. Ask for the points list (the set of measurements, states and commands actually exposed), not only the protocol name; two devices that both say Modbus can still show the control room nothing useful.

Decide who controls the asset and how it appears in the control room before the specification is written. The generator rules are where the diesel saving comes from, and the protocol choice is what makes the plant visible to the people who must operate it.

Electrochemical hazards depend on chemistry, cell condition, enclosure, propagation behavior, ventilation, detection, suppression strategy, separation, emergency access and operating response. The US DOE Energy Storage Handbook addresses performance testing, electrochemical safety, physical security, cybersecurity, procurement and commissioning as distinct engineering topics [1].

A thermal event can progress through abnormal cell behavior, heating, venting and propagation depending on chemistry, design and failure conditions. Detection must be connected to an approved response. An alarm that reaches an unattended dashboard has not created emergency readiness. Site personnel and responders need clear isolation boundaries, hazards, access information and authority.

The project must identify applicable current codes and authority requirements, then translate them into site design, product evidence, hazard analysis, emergency procedures, training, inspection and maintenance. A foreign listing or report can be relevant evidence but does not, by itself, establish Philippine approval.

Air cooling and liquid cooling: what changes for the people who run it

Cells need to stay in a narrow temperature band and at the same temperature as their neighbors. Cooling exists to make that true inside a steel box in the sun. The two ways of doing it change behavior, maintenance and siting in ways an operator will notice every week.

How each works. An air-cooled cabinet moves conditioned air through the racks with fans, usually from an air conditioner mounted in the cabinet. A liquid-cooled cabinet circulates coolant through plates under the packs, with a chiller unit rejecting the heat outside. Bescore publishes air cooling for its microgrid cabinets and liquid cooling for its compact commercial cabinet and its battery container.

What differs in behavior. Liquid holds the cells closer to one temperature. Bescore publishes a DC-side temperature difference held within about 1.5 °C on its liquid-cooled cabinet and a pack difference of 2 °C or less on its container; it publishes no equivalent figure for its air-cooled cabinets, which is itself informative, because air cannot hold a spread that tight. An even temperature means the cells age together, so the rack’s capacity is not set by its hottest cell. Both kinds derate above the same threshold in Bescore’s sheets, 45 °C ambient, and the published upper operating limits are 55 °C for the air-cooled microgrid cabinets and 50 °C for the liquid-cooled compact cabinet. Read those against the site: a cabinet in direct sun on a concrete pad in April is above the shade temperature the forecast quotes, and shading and orientation are design decisions, not afterthoughts. Air cooling uses more auxiliary energy as the ambient rises, and that energy comes off the plant’s round-trip efficiency; liquid cooling uses less in steady state but has a chiller that must run.

What differs in maintenance. Air-cooled cabinets breathe the site’s air. Filters clog, fans wear, and on a coastal site salt and dust come in with the air and settle on everything. The work is routine and a cooperative’s own maintenance crew can do it: change filters, clean coils, replace a fan. Liquid-cooled cabinets add a coolant loop, leak detection, coolant quality and level checks, and a chiller with a compressor and refrigerant. That work needs a refrigeration technician and a service contract, or a trained person on staff. Neither is better in the abstract; the question is which set of tasks the cooperative or its contractor can actually keep doing in year six.

What differs in siting. Both make noise; Bescore publishes 65 dB for its cabinets and under 75 dB for its container, which matters within a few tens of meters of housing. Air-cooled cabinets need clear space for intake and exhaust and suffer if that space is enclosed. Liquid-cooled cabinets can be placed closer together as far as cooling is concerned; Bescore describes side-by-side placement for its liquid-cooled cabinet. Separation for fire, however, is decided by the fire authority and the propagation evidence, not by the cooling method, and that decision is covered below.

Choosing. Daily deep cycling, a hot exposed site, a tight footprint and a maintenance contract in place point to liquid. Light or occasional duty, a shaded site, a limited budget and a crew that can change a filter but not service a chiller point to air. Write the choice down with its reason, so that whoever maintains the cabinet in year six knows why it was made.

Which document proves which layer: the standards behind battery fire safety

The Bureau of Fire Protection and the local building official will ask for documents. A bidder will answer with a list of standards. Knowing what each standard actually covers turns that list into questions the cooperative can put back, and it is the difference between a certificate list and a safety case.

The layers. Figure 8 shows five layers. In the hardware and on the site, they take these forms: cell design and venting, so a failing cell releases gas in a controlled way; pack-level detection of temperature, gas and smoke, with fusing that isolates a pack; cabinet-level suppression; venting and forced ventilation so that released gas cannot accumulate to an explosive mix; fire-resistant cabinet construction; separation distances between cabinets and from buildings; and the site measures of access, signage and an emergency plan that responders have seen before the day they need it.

Aerosol and water. The cabinets Bescore publishes carry both, and the two do different jobs. A condensed aerosol agent floods the enclosure with fine particles that interrupt the flame chemistry within seconds; it is compact, needs no water supply and leaves little residue, but it does not cool the cells, and a lithium cell that has gone into thermal runaway will reignite once the agent disperses. Water, as a spray or mist, cools: it takes heat out of the cells and the neighboring packs and is what prevents the fire from restarting and spreading, at the cost of a water supply, drainage and wetting of live equipment. A design that uses both knocks down the flame with aerosol and then cools with water. Ask which agent is in the pack, which is in the cabinet, how each is triggered, and what happens to the water afterwards.

What each standard covers. These are the standards that appear on storage datasheets. Naming a standard says what was tested. It does not say that a given product passed, and it does not apply in the Philippines by itself.

1. UL 1973covers the battery: cells, modules and battery systems for stationary use, tested for electrical, mechanical and thermal abuse.
2. UL 9540covers the energy storage system as a whole: battery, PCS, BMS and their integration, as one product.
3. UL 9540Ais not a pass-or-fail standard. It is a test method that deliberately drives a cell into thermal runaway and measures whether, and how far, the fire propagates at cell, module, unit and installation level. It produces a report with data. That report is what a fire authority reads to decide whether cabinets may stand closer together than the default spacing. A supplier who lists UL 9540A has had the test performed; the result is in the report, not in the listing.
4. NFPA 855is an installation standard, written for the site rather than the product. It sets separation distances, limits on energy per unit and per room, and requirements for detection, suppression, ventilation, explosion control and emergency planning, and it uses UL 9540 listings and UL 9540A data as inputs.
5. IEC 62619covers the safety of secondary lithium cells and batteries for industrial use, including stationary storage: the cell and battery level, not the whole system.
6. IEC 62477covers the safety of power electronic converter systems, which is the PCS.
7. IEC 62933is the series for electrical energy storage systems; its safety part covers the grid-connected system with an electrochemical battery, at system level.
8. UN 38.3covers transport: the tests a lithium battery must pass before it may be shipped by sea or air. It says nothing about operation.

How it applies here. None of these standards is Philippine law. The Fire Code of the Philippines, as enforced by the Bureau of Fire Protection, the Philippine Electrical Code and the local government’s building permit process are the controlling requirements, and the authority decides what foreign evidence it will accept in support of a design. The current editions and the specific requirements for storage installations are confirmed with the authority for the site.

What to ask a bidder. For each standard named: the certificate number, the issuing body, the exact model and configuration it covers, and the date. For UL 9540A: the report itself, and the unit-level and installation-level results, because that is the document the fire officer will want. For the suppression: the agents, the triggers, the water supply and drainage. Metro Power does not display certification for storage products on this site until the certificates are in hand and their scope has been checked against the model offered; a standard named on a datasheet is a claim to verify, not evidence.

A grid-connected BESS does not automatically form an island. Microgrid operation requires a defined boundary, isolation method, grid-forming source or compatible control strategy, protection changes, grounding treatment, load shedding, generator and renewable coordination, black-start sequence where required, resynchronization and utility approval.

Grid-following operation generally synchronizes to an existing voltage and frequency reference. Grid-forming operation can establish or regulate that reference within a supported design. These phrases do not prove black-start capability, seamless transfer or compatibility with every generator and inverter. The exact PCS mode, controller sequence, protection and tested configuration determine what can be claimed.

Three operating modes, three sets of equipment

A cooperative meets storage in three situations: a battery on a constrained feeder that stays connected to the grid; a hospital, water system or plant that must keep running when the feeder fails; and an island or missionary-area barangay that has no feeder at all. These look like one product with a different setting. They are three electrical situations, and each demands something different from the converter, the switching device, the grounding and the approval path. Most of the cost difference between two storage proposals is explained by which situation each was designed for.

Grid-tied. While the grid is present it sets the voltage and frequency for everything connected to it. The PCS follows that reference and pushes current in or out at the setpoint it is given; this is grid-following operation. The design needs an interconnection agreement with the distribution utility, protection that disconnects the system when the grid is lost, and revenue metering at the point of connection. The anti-islanding protection is not optional: a converter that keeps energizing a dead feeder puts the line crew repairing it at risk. Export may need to be limited or prevented under the utility’s rules. No transfer device is needed, because the system never runs alone.

Islanded from a grid-tied site. When the feeder fails, something must open the connection to the utility, and the PCS must change from following a voltage to making one. The device that opens the connection is the transfer device. A static transfer switch (STS) is built from power electronics and transfers in a few to a few tens of milliseconds; Bescore publishes 20 milliseconds or less for the microgrid cabinet that carries the switch inside. At that speed most computers, controls and motors ride through the change. A mechanical breaker or automatic transfer switch takes longer, and the loads see an interruption. Which one to specify is a question about the loads: what on the site can tolerate a short blackout, and what cannot. The island can only carry what the battery and any generator can supply, so feeders the island cannot hold are dropped at or before the transfer. When the grid returns, the island is resynchronized to it before the connection is closed again, because closing onto a grid that is out of phase damages equipment.

The isolation transformer. In an island the utility’s neutral and ground reference are gone. An isolation transformer between the PCS and the island’s loads gives the island its own neutral point to ground, so that protective devices still see a fault current path and a live-to-ground fault still trips something. It also separates the battery’s DC-side circuits from the load side. Some cabinets offer it as an option, some site designs place it in the switchgear, and some designs do without it because the PCS topology and the site grounding already provide what is needed. Whether one is needed is decided by the grounding study for the island, not by the cabinet brochure. Bescore publishes an isolation transformer as optional on its container classes and does not list it on the smaller cabinets.

Grid-forming, and why it matters when there is no grid to follow. A grid-forming converter sets the voltage and frequency itself and holds them steady as loads switch on and off. A grid-following converter cannot operate without a reference to follow. When the grid is gone, only a grid-forming source can hold the island: the PCS, a diesel generator, or both sharing the load. Datasheets name the control modes in shorthand. PQ is grid-following, taking real and reactive power setpoints. VF is grid-forming, acting as the voltage and frequency source. VSG, the virtual synchronous generator, is a grid-forming mode that imitates the inertia and droop of a rotating machine so that the converter shares load with a diesel set and rides through disturbances the way a generator would. Bescore publishes these mode names, together with black start, for its medium-voltage conversion station. Bescore publishes this unit at 50 Hz. A 60 Hz version is not yet confirmed. For the 400 V cabinets, the published evidence of island capability is the built-in transfer switch and on-grid to off-grid switching on the microgrid variant; the mode names are not published. Grid-forming capability and its limits are therefore confirmed per model and witnessed at commissioning, not assumed from the word microgrid.

Permanently off grid. With no grid at all, the converter forms the grid all the time. For a village or resort island of a few hundred kilowatt-hours, the usual design is a single hybrid inverter: battery input, solar charge controller and generator input in one unit, running its own logic. It charges from solar first, starts the generator when the battery is low or the load is high, and supplies the loads. There is no separate energy management system because there is one converter making one set of decisions. The design still needs a grounding system and protection designed for a source that never had a utility neutral, and it needs no interconnection protection because there is nothing to interconnect with. One caution for a Philippine site: the hybrid units Bescore currently publishes for this pattern are the CF2-50 and OmniStax. Bescore publishes these units at 50 Hz. A 60 Hz version is not yet confirmed. Philippine loads and any future grid connection are 60 Hz, so Metro Power teaches the pattern here and confirms the 60 Hz version with Bescore before either unit is proposed for a Philippine site.

Write the mode into the operating statement before asking for prices: “hold import below 400 kW while grid-tied and carry the pump house and the clinic for four hours when the feeder fails” is a specification; “a microgrid-ready battery” is not. Metro Power then asks for the loads that must survive an outage, the largest motor among them, the existing solar and its inverter, and the site’s distance from the coast.

Where the solar joins: DC coupling against AC coupling

Solar can enter a storage system in two places, and the choice is one of the earliest and most expensive forks in a design.

DC coupling. The array connects through maximum power point tracking charge controllers onto the battery’s DC bus, and the one PCS serves both the battery and the solar. Solar energy that goes into the battery is converted once instead of twice, so less is lost. The array can be larger than the converter: the midday peak the converter cannot pass either charges the battery or is clipped, while the larger array gains morning and evening energy. Suppliers publish this as an oversizing ratio. The converter’s rating caps what the solar can export to the grid or the load at any moment. Bescore’s microgrid variant is DC-coupled in exactly this way: it publishes two or four solar inputs of 55 kW each, with a 250 to 650 volt input range, feeding the battery bus inside the cabinet, which is why its sheet states a solar input power at all.

AC coupling. The array has its own solar inverter on the AC side, and the storage system sees it as a source on the same bus. This is the natural arrangement where solar already exists, including the net-metered rooftop systems many candidate sites already carry, and it is how utility-scale plants are built: Bescore’s own utility arrangement shows the solar plant with its own inverters and the storage with its own conversion station meeting at medium voltage. The cost is a second conversion for any solar energy that goes into the battery, and a control problem in an island: when the battery is full and the load is light, the solar inverter must be told to back off, otherwise the island’s voltage and frequency climb. That curtailment is done either by the storage converter shifting the island frequency upward, which standard solar inverters respond to by reducing output, or by a communication link to the solar inverter. Either method must be specified and tested; without one, an AC-coupled island trips itself on a sunny Sunday.

Which to choose. A new island or off-grid site with no existing solar: DC coupling, one converter, one controller, fewer losses. An existing site with working solar under a net-metering agreement: AC coupling, keep the array and its inverter, add curtailment control and check whether the interconnection agreement needs revising. A large plant: AC coupling, because the solar and storage are separate plants with separate inverters and separate protection. A site where the existing solar inverter is old, or cannot be curtailed, is the hard case: it may be cheaper to re-engineer the array onto the DC bus than to make the old inverter behave in an island. That decision needs the inverter’s make, model and firmware, which is why the readiness list now asks for them.

Working with the diesel set: minimum loading, cycling and black start

In a diesel hybrid the battery earns its keep by changing how the generator runs, and the fuel saving lives in the control logic rather than in the battery’s size.

Minimum loading. A diesel generator burns fuel inefficiently and wears faster at low load. Run for long periods below a fraction of its rating, commonly quoted around 30 percent though the generator’s own manual is the controlling figure, unburned fuel and carbon accumulate in the exhaust system, a condition known as wet stacking, and maintenance intervals shorten. Without storage, a village set idles all night at a fraction of its rating because someone needs a light on. With storage, the controller runs the set at a high steady load for a few hours to carry the load and charge the battery, then stops it and lets the battery carry the night. Each start costs fuel and wear, so the controller also aims to minimize the number of starts and stops per day, and it uses the battery to absorb the sudden load steps that a generator handles badly, such as a pump starting.

What to specify and witness. The saving is a property of the strategy, so the strategy is specified: the state of charge at which the generator starts and stops, the load at which it runs, the maximum starts per day, and the behavior when the battery is empty and the sun is down. Ask the bidder for the expected generator hours and starts per day under the site’s own load profile, and test it at commissioning with the real set: the generator’s governor, voltage regulator and droop settings must be compatible with the converter, and a VSG-mode converter is designed to share load with a set the way a second generator would. A generator that hunts against the converter, or a converter that trips the set on every load step, is a commissioning failure, not a product defect to be lived with.

Black start. A black start re-energizes a dead system from its own resources, with no grid to lean on. Two things can do it: the PCS in its voltage-and-frequency-forming mode using energy in the battery, or the generator. Either way the controls need their own auxiliary power to run while the bus is dead, and the sequence is written and rehearsed: energize the bus, close the feeders in priority order so that the first loads connected are the ones that must come back first and none of them is a large motor that would collapse the fresh island, then start the generator if the battery cannot carry what follows. Figure 9 shows where this sits in the operating sequence. Bescore publishes black start as a function of its medium-voltage station; for the 400 V cabinets it is not published, and any proposal that relies on it names which device performs it and demonstrates it at commissioning by shutting the whole site down and bringing it back.

Critical-load resilience must use an interval or scenario profile, not merely the building peak. Motor starting, step load, inrush, harmonics, minimum generator loading and communications availability can determine whether the island remains stable.

Economics should separate demand reduction, energy shifting, avoided outage cost, renewable utilization and other services. A benefit must have a baseline, dispatch rule, tariff or valuation basis, adoption timing and responsible owner. Avoid double counting energy that serves two commitments in the same interval.

Round-trip efficiency is also conditional. It depends on power level, temperature, auxiliaries, SOC range and where energy is measured. A headline PCS efficiency does not equal site-level round-trip efficiency. The economic model should use the warranted or tested boundary relevant to billing and include HVAC, controls and standby energy.

Lithium iron phosphate, and how to read a cycle-life claim

Every storage product Bescore publishes uses lithium iron phosphate cells, written LFP or LiFePO4: prismatic cells rated 3.2 volts nominal and 280 or 314 ampere-hours. LFP dominates stationary storage for reasons that matter to a cooperative. It is more stable at high temperature than the nickel-based chemistries used in most vehicles, it tolerates more cycles, and its lower energy density, which makes it heavier per kilowatt-hour, costs little when the battery sits in a cabinet on the ground. Its voltage curve is flat across most of the charge range, which is good for the converter and awkward for the battery management system: the BMS cannot read state of charge from voltage alone, so it estimates from current and time and periodically recalibrates by holding the battery at a known state. That is why state of charge is an estimate, as Chapter 2 says, and why a system that is never fully charged drifts.

Reading the configuration line. A datasheet writes the battery as so many cells in series and in parallel. Multiply the series count by 3.2 volts for the string voltage, and the string voltage by the cell capacity for the energy. Bescore publishes 240S1P with 280 Ah cells: 240 times 3.2 is 768 volts, times 280 ampere-hours is about 215 kilowatt-hours, which is the rack’s rated energy. Its container publishes 416S12P with 314 Ah cells: 416 times 3.2 is 1,331.2 volts, and twelve parallel strings of 1,331.2 volts times 314 ampere-hours is about 5,016 kilowatt-hours. High string voltages keep the current down for a given power, which is why cabinet batteries run at 700 to 1,500 volts DC and why the DC side is treated as high voltage work.

The cycle-life claim. A cycle-life figure counts full charge-and-discharge cycles until capacity falls to a stated fraction of new, under a stated temperature, rate and depth of discharge. Bescore’s pack and rack sheets state it completely: 8,000 or more cycles at 25 ± 2 °C, at 0.5P, at 100 percent depth of discharge, to 70 percent end of life. Each condition changes the number. Temperature: the test is at 25 °C; a cabinet interior that runs hotter ages faster, and in a hot climate calendar aging (time at temperature, whether or not the battery is cycled) can matter as much as cycling. Rate: 0.5P is a two-hour charge or discharge; faster work heats the cells more per cycle. Depth of discharge: 100 percent means the whole capacity is used each cycle; a claim at 80 percent depth uses less of the battery per cycle and gives a higher count for the same cells. End of life: 70 percent means the count stops when the battery holds 70 percent of its original energy; a claim to 80 percent stops earlier and gives a smaller count. A figure such as “8,000+ cycles” on a web page, without those four conditions, cannot be compared with anything. Bescore’s web feature lists for its cabinets carry it in that bare form, while its pack and rack sheets carry the full statement. Ask for the full statement.

Reading it properly. Put two offers on the same conditions before comparing them: the same depth, the same temperature, the same end-of-life fraction. Then convert cycles to energy throughput: cycles multiplied by the usable energy per cycle. Then divide by the site’s own daily throughput from the dispatch model in Chapter 4 to get years, and compare that with calendar life at the site’s temperature; the shorter of the two governs. As an illustration only: a 215 kWh rack cycled once a day at 80 percent depth moves about 172 kWh a day; 8,000 such cycles is about 1,376 MWh of throughput, which at 172 kWh a day would take about 22 years, so at that duty calendar aging and temperature, not cycle count, would set the life. Cycle the same rack twice a day at full depth in a hot cabinet and the answer changes in every term.

The operating window. Recommended windows narrower than the full range extend life. Bescore publishes a permitted state-of-charge range of 0 to 100 percent and a recommended window of 5 to 95 percent for its container, which is the manufacturer saying that the top and bottom five percent cost more life than they are worth. The dispatch model and the acceptance clauses should use the recommended window, not the maximum. A factory state of charge of 30 to 40 percent, also published, is a shipping condition; the first full charge and capacity check belong to commissioning.

The four conditions above are what the degradation and acceptance clauses of a supply agreement should reference, because a promise of years and cycles without them is a promise no one can test.

Capacity and efficiency change with age, cycles, calendar time, temperature and operating window. The model should state beginning and end-of-life performance, augmentation, replacements, availability, downtime, auxiliary energy, warranty throughput and residual value.

The Philippine DOE issued an ESS policy framework under DC2019-08-0012, and current project treatment depends on application, ownership, connection and applicable market or utility requirements [3]. Confirm the current DOE, ERC, National Grid Corporation of the Philippines (NGCP) or distribution-utility pathway, system-impact requirements, environmental and local permits, fire and electrical approvals, and the authority having jurisdiction for the exact project.

Procurement should define the guaranteed power and energy at an agreed condition and measurement point, usable SOC window, efficiency method, auxiliary treatment, availability, degradation curve, throughput or cycle conditions, augmentation, warranty exclusions and acceptance tests [4]. A warranty stating only years and cycles may not explain how operating temperature, depth of discharge, C-rate (charge or discharge rate relative to battery capacity) and annual throughput affect coverage.

The environmental lines that matter on a Philippine island: corrosion class first

Salt air is why outdoor electrical equipment fails early on Philippine islands and shorelines. Four lines on an outdoor cabinet’s datasheet describe how it will cope, and the corrosion class is the one to read first because it is the cheapest line to specify and the most expensive to get wrong.

Corrosion class. ISO 12944 grades the corrosivity of an atmosphere from C1 (heated indoor spaces) to C5 (very high: coastal areas with high salinity and industrial areas with aggressive atmospheres), with a CX category beyond that for offshore and the most extreme sites. The class on the datasheet says which category the enclosure’s paint and material system has been designed and tested for. Bescore publishes C5 for its cabinets and its battery container and C4 for its medium-voltage station. A C5 enclosure is the right starting specification for any site within sight of the sea, and for most Philippine sites that is a large share of them. What the class does not cover: the air an air-cooled cabinet draws in, the fins of a heat exchanger, and the fasteners and hinges if they were not part of the tested system. Ask for the coating system and the salt-spray test duration behind the class, and ask what the intake path does with salt.

Ingress protection. The IP code from IEC 60529 has two digits: the first for solid objects and dust, the second for water. IP55, which Bescore publishes for its cabinets, means dust-protected (dust may enter but not enough to interfere) and protected against water jets from any direction. IP67, published for the packs inside the container, means dust-tight and able to survive temporary immersion. IP20 and IP2X, published for indoor racks and packs, mean finger-safe and nothing more, which is why those items live inside an enclosure and are never installed bare outdoors.

Humidity. The published range, 0 to 95 percent non-condensing, hides a condition. Condensation forms inside a cabinet when its internal surfaces fall below the dew point, which happens when cooling runs hard on a humid night and then stops. Dew on a busbar is a fault. The design answer is a heater, a dehumidifier or a control strategy that keeps the interior above the dew point, and the question to a bidder is how the cabinet manages condensation, not whether the range includes 95 percent.

Altitude. Derating above 2,000 meters, which most Bescore storage sheets state, matters where thin air cools less and insulates less. Almost no Philippine site is that high, so this line seldom applies here. The temperature line matters everywhere: the published derating above 45 °C ambient is reached inside a cabinet enclosure on a hot afternoon well before the weather station reads it.

The lines to write into a specification. Corrosion class C5 to ISO 12944 for every outdoor enclosure, with the coating system and test duration stated; IP55 or better for cabinets and the rating of any pack or rack that could be exposed during maintenance; the ambient temperature at which full output is available without derating, and the derating curve above it; the condensation control method; and the noise level at a stated distance.

Starting small without building a dead end

Cabinet-type systems can be paralleled, which lets a cooperative install what it can fund this year and add cabinets as load grows. Paralleling is not free of engineering: the converters must share load and the controllers must coordinate, and suppliers state a maximum paralleled energy and whether extra combining or synchronizing equipment is needed to reach it. Bescore publishes parallel operation to 4 MWh or more for its 400 V cabinets, without external combining equipment, and four to five units per group for its container classes. The practical rule is to design the site’s electrical and civil works, the transformer, the switchgear, the pad and the cable routes, for the final count, and to install the first cabinets into that design. A first phase designed only for itself is the phase that cannot be extended.