In short. An industrial microgrid coordinates grid supply, diesel or gas generators, battery storage and rooftop solar under one controller, so a factory can cut demand charges in normal operation and keep its process safe when the DISCOM fails. It must be designed around the production sequence, not the meter: which loads must stay on, which can be shed, and in what order the plant restarts.
Start with the process, not the energy bill
A pharmaceutical plant in Genome Valley, a plastics moulder in Jeedimetla and a food processor near Shamshabad may draw similar maximum demand from the DISCOM, yet need entirely different microgrids. The first loses a batch if HVAC and process water stop; the second freezes material in its barrels if heaters drop for too long; the third loses product if cold rooms warm up. A design that begins with the electricity bill tends to optimise kWh and kVA demand, and then discovers during the first outage that the plant could not survive the transition. Begin instead with a load classification that the production manager signs.
| Class | Typical loads | Microgrid behaviour on grid failure |
|---|---|---|
| Must stay on | PLC and DCS, safety systems, instrument air for safe shutdown | UPS for no-break, generator behind it |
| Short break acceptable | Process cooling, critical pumps, clean room HVAC | Generator or BESS picks up within seconds |
| Restart in sequence | Compressors, extruders, chillers, large motors | Restored one step at a time once sources are stable |
| Sheddable | Comfort cooling, non-process lighting, some production lines | Dropped on islanding, restored if capacity allows |
The classification decides generator capacity, battery power rating, UPS scope and the shedding list. Manufacturing power gives examples by sector.
Grid-connected operation: where the savings come from
Most of the time an industrial microgrid runs in parallel with the grid, and its value comes from controlling what the factory draws. Telangana HT consumers pay a demand charge on recorded maximum demand in kVA, so a short peak during a shift change can set the charge for the month. A BESS that discharges during those peaks, or a generator that synchronises and shares load briefly, reduces recorded demand. Solar reduces daytime energy import. Power factor correction and harmonic filtering keep kVA close to kW.
P_bess = P_site − P_limit, and E_bess = Σ (P_site − P_limit) × Δt
P_site is the instantaneous site demand, P_limit the demand target, P_bess the battery discharge power needed and E_bess the energy over every interval Δt in which demand exceeds the target. Worked example (illustrative): a plant peaks at 1,450 kVA for about 40 minutes each morning and wants to hold 1,250 kVA. At unity power factor from the battery, it needs roughly 200 kW of discharge and about 135 kWh of usable energy for that peak, more with margin for a longer peak. Generator-based peak shaving requires grid paralleling approval and protection agreed with the DISCOM; see peak-shaving power systems.
Island mode: the hard part
When the grid fails, the microgrid must open its point of common coupling, detect the island, form a stable bus from its own sources and reconnect loads in order. Each step has an engineering requirement. Islanding detection must operate before the DISCOM’s auto-reclose closes back onto an energised, unsynchronised plant. The forming source, either a generator or a grid-forming inverter, must hold voltage and frequency while loads return. Motor restarts must be staggered so the combined inrush stays within the sources’ capability. Protection must work with the lower fault current of an inverter-held bus and the higher fault current of generators. Reconnection to the grid must be synchronised, never a dead closure onto a live plant.
Restoration order after islanding
- Control power and safety systems, which should never have dropped if UPS-supported.
- Process cooling and critical pumps, so the process returns to a safe state.
- Instrument and process air compressors, one at a time.
- Production lines in the order the process engineer specifies.
- Comfort loads last, only if spare capacity remains.
The controller should own this sequence centrally, using breaker positions, voltage and frequency measurements and source status, rather than scattering it across unrelated logic in different panels. See microgrid controllers and islandable microgrids.
Production risk is part of the business case
An energy-saving strategy that interrupts production is not an optimisation. If peak shaving discharges the battery to its floor at 11:00 and the grid fails at 11:20, the plant has less backup than before the project. The microgrid business case must therefore value downtime as well as tariff savings, and the controller must reserve enough battery energy and generator readiness for the island transition. For many factories, the honest answer is a modest battery for peak management and ride-through support, with generators for sustained islanding, rather than a large battery expected to do everything.
Common mistakes
- Designing around the monthly bill and treating islanding as an afterthought. The first grid failure shows the plant cannot restart its compressors on the microgrid.
- Letting every motor restart automatically when the bus returns. The inrush collapses the island and the plant suffers a second blackout.
- Running generators in parallel with the grid without DISCOM approval and proper reverse power and loss of mains protection. The arrangement is unsafe for line staff and may breach the supply agreement.
- Spending the battery’s reserve on peak shaving. The site has no stored energy when the outage arrives.
What to send us
Send twelve months of HT bills with recorded maximum demand, a load list with motor ratings and starting methods, the single-line diagram, details of existing generators and their controllers, and a process engineer’s note on which loads must stay on and in what order the plant restarts. Fifteen-minute demand data from the meter is ideal. Send your application details. Remote sites with similar logic are covered under telecom microgrids.
Questions engineers ask
Can our existing diesel generators become part of a microgrid?
Often yes, if their controllers support synchronising and load sharing or can be upgraded to do so. Older sets with basic AMF controllers usually need new control panels. Engine condition, CPCB compliance status and running hours decide whether upgrading is worthwhile.
Do we need DISCOM permission for an industrial microgrid?
Any arrangement that parallels on-site generation or battery export with the grid needs approval and protection agreed with the DISCOM. A microgrid that only islands, with no parallel operation, has fewer interface requirements, but the changeover arrangement must still prevent backfeed.
How large should the battery be for a factory?
Size it for its assigned job. Peak management needs power and energy matched to recorded peaks; ride-through support needs power for seconds to minutes; generator-free islanding needs hours of energy and is rarely economic at industrial scale.
What is the payback of an industrial microgrid?
It varies with the tariff, demand profile, outage frequency and value of lost production. Treat any generic payback figure with caution; a defensible number comes from interval meter data and a costed downtime estimate for your plant.
Technical content reviewed by Srinivas Reddy, Managing Partner, Reddy Generators. Last updated 19 October 2026.