Site
Jbeil Municipality
Simulated time

AI Advisor

Continuous analysis of PV production, battery state, generator runs and fuel, EDL availability, load profiles, tariffs, alarms, maintenance records and weather forecasts. Every recommendation cites the measurements behind it.

Identified, not yet actioned
$1,852
per month across 7 active recommendations
Captured so far
$0
per month from 0 implemented · $0/mo in tracking
Share of opportunity actioned
0%
of $1,852/mo total identified
Load shiftinghigh confidencemedium effort

An 18 kW load never switches off — most of it is cooling empty rooms

Overnight (00:00–05:00) and weekend demand holds flat at 18 kW, which is 22% of total site consumption over the last 30 days. Sub-metering attributes 7.4 kW to server room cooling running at a 19 °C setpoint, and 6.1 kW to HVAC in the second-floor archive, the council chamber and the west wing training room — three zones with no recorded occupancy after 18:00 in the last 21 days. Raising the server room setpoint to 24 °C and scheduling the three zones off out of hours removes an estimated 11 kW of the baseline.

Load meterHVAC sub-metersOccupancy scheduleServer room sensors
Estimated saving
$680
per month
$8,160 / year
Site loadAverage weekday profile — the floor never falls below 18 kW.
Evidence window: 13 Jul – 11 Aug 2026 · 30 days
Generator optimisationhigh confidencelow effort

Stop running Generator 1 overnight — the battery can carry the night load

Between 5 and 11 August, Generator 1 ran an average of 3.2 hours after 21:00 on 6 of the 7 nights. Across those runs the measured load averaged 12 kW against a 250 kVA rating, so the set sat at 14% loading, the region where its specific fuel consumption is worst: 0.62 L/kWh versus 0.28 L/kWh at 70% load. The same nights ended with the battery above 63% state of charge, meaning 76 kWh of usable storage went unused while the genset burned 24 L/night.

Genset 1 controllerBattery BMSLoad meterFuel flow sensor
Estimated saving
$412
per month
$4,944 / year
Genset kWh after 21:00Battery kWh left unusedNightly genset energy against idle battery capacity, 5–11 Aug.
Evidence window: 5–11 Aug 2026 · 7 nights
Load shiftingmedium confidencelow effort

Start HVAC pre-cooling at 09:30 so solar pays for it instead of diesel

The building management system starts pre-cooling at 06:00, when EDL is off and the site is on generator at a blended $0.41/kWh. Solar output does not cross the 40 kW mark until 09:15. Measured indoor temperature at 09:30 on the 14 days sampled averaged 23.4 °C, and on the four days when pre-cooling started late for maintenance reasons the 11:00 temperature was within 0.6 °C of the normal profile. Moving 21 kWh/day of pre-cooling from $0.41/kWh diesel to $0.04/kWh solar costs nothing in comfort.

BMS schedulePV inverterIndoor temperature sensorsTariff table
Estimated saving
$260
per month
$3,120 / year
Solar availableHVAC demandPre-cooling runs two hours before solar can serve it.
Evidence window: 29 Jul – 11 Aug 2026 · 14 days
Maintenancemedium confidencemedium effort

Generator 2 is 9% thirstier than it was two months ago — service the injectors

Comparing runs at 55–75% load only, Generator 2 consumed 0.293 L/kWh in mid-June and 0.319 L/kWh in the first week of August, a 9% degradation over 60 days. Generator 1 moved 0.6% over the same period, so ambient temperature and fuel batch are ruled out. Exhaust gas temperature on cylinder 3 also runs 41 °C above the bank average. The pattern is consistent with injector fouling, which typically compounds to 15–20% before failure.

Parts and labour for an injector service are estimated at $340, recovered in under two months.

Genset 2 controllerFuel flow sensorEGT probesMaintenance log
Estimated saving
$180
per month
$2,160 / year
Genset 2Genset 1Specific fuel consumption at comparable load bands.
Evidence window: 13 Jun – 11 Aug 2026 · 60 days
Battery strategyhigh confidencelow effort

The battery is barely being used — deepen the discharge and burn less diesel

Depth of discharge has not exceeded 38% on any day in the last 60 days; the median daily minimum state of charge is 63% and the lowest point in 30 days was 62%. The dispatch controller currently reserves everything below 60% as backup. Releasing usable depth to 80% frees roughly 87 kWh per cycle, which displaces about 140 litres of diesel a month at the site's measured 0.34 L/kWh genset average. For this LFP pack, rated 6,000 cycles at 80% DoD, the added wear is under 2% of expected life over five years.

Battery BMSDispatch controllerFuel flow sensorManufacturer cycle curve
Estimated saving
$129
per month
$1,548 / year
Daily minimum SoCDaily minimum state of charge never approaches the 20% floor.
Evidence window: 13 Jun – 11 Aug 2026 · 60 days
Investmentmedium confidencehigh effort

Adding 100 kWh of storage pays for itself in about 3.4 years

Over the last 60 days the site curtailed no solar but ran the gensets for 214 hours in windows where an additional 100 kWh of usable storage would have carried the load. At the measured 0.34 L/kWh genset average, that is roughly 1,150 litres of diesel a year, about $1,035 at $0.90/L, plus $220 of avoided maintenance from 190 fewer running hours. Installed cost for a 100 kWh LFP rack with the existing inverter headroom is estimated at $4,300.

Sensitivity: diesel delivered at $0.70/L pushes payback to 4.3 years; at $1.30/L it falls to 2.4 years. Lebanese pump prices have moved more than 30% within a single year, so treat the mid case as indicative.

Battery BMSGenset controllersFuel invoicesLoad meterTariff table
Estimated saving
$105
per month
$1,255 / year
$4,300 capex
3.4 yr payback
PaybackPayback sensitivity to the delivered diesel price.
Evidence window: 13 Jun – 11 Aug 2026 · 60 days
Solar performancehigh confidencelow effort

Strings 3 and 4 are producing 14% below the rest of the array

For 11 consecutive days, strings 3 and 4 have delivered 14% less specific yield than the array average, measured at matched irradiance and module temperature. The shortfall is flat across the day rather than concentrated in the morning or afternoon, which points to soiling rather than a new shading obstruction; no inverter faults or DC arc events were logged in the period. Recovering the gap returns an estimated 210 kWh/month.

PV inverterString monitorsPyranometerAlarm log
Estimated saving
$86
per month
$1,032 / year
Specific yieldEleven-day specific yield by string.
Evidence window: 1–11 Aug 2026 · 11 days