A Tier-1 automotive supplier in Querétaro runs a 24/6 manufacturing campus and spends about $3.1 million a year on electricity under CFE's GDMTH tariff. Mexico Energy Partners modeled a stacked program for the site. It sets $4,550,000 of capital against $884,100 of projected annual savings, an 18% blended IRR and a 5.1 year payback. That saving equals 28% of the plant's annual energy spend.
Every figure below is an engineering projection rather than metered performance after construction. The question it helps a CFO and a plant director settle is whether to commit efficiency capital before generation is sized, and what site data is needed to test that sequence anywhere else.
What the plant faced
Four conditions set the scope of the work. The site bills on GDMTH, where the demand charge is built from measured peak behavior rather than from total consumption, so the monthly bill was hard to forecast. Grid voltage sags were causing micro-interruptions in sensitive equipment and costing production time. A corporate Scope 2 mandate required a measurable reduction plan. An aging 400-ton chiller and a fixed-speed air compressor were both becoming reliability liabilities.
The voltage problem is not only an uptime question in Mexico. Every load center connected at medium or high voltage carries Código de Red obligations, set by the Comisión Reguladora de Energía in RES/151/2016 and published in the Diario Oficial de la Federación on 8 April 2016. Power quality measurement at a site with recurring sags is a compliance item, not housekeeping.
Efficiency first, then generation
The sequencing is the part of this case worth copying. Waste was removed before generation was sized, so the array and the battery were specified against a reduced load. Fewer megawatts of PV are then needed to cover the same share of demand, and the generation capex falls with them.
Compressed air and chillers
An audit of the compressed air system found leaks and system pressure set higher than the plant needed. Replacing a fixed-speed compressor with a variable speed drive unit and resetting pressure is modeled at $97,500 in annual savings against $220,000 of capex, a payback of 2.3 years. Leak loads of that scale are ordinary. The US Department of Energy's sourcebook Improving Compressed Air System Performance reports that a plant without an active leak program typically loses around 20% of its compressed air production capacity to leaks, a pattern MEP also found when it tagged more than 80 leaks at a single plant.
Compressed air work clears a capital committee faster than generation for a structural reason. It needs no interconnection study, no generation permit and no production shutdown, so the approval path is short and the risk is contained inside the fence line.
The second measure replaces the 400-ton chiller with a magnetic-bearing machine and adds VFDs to the pumps and fans. It is modeled at 820,000 kWh a year, worth $123,000. Capex is $580,000. The implied blended electricity price is about $0.15 per kWh, and that is the first number to check against your own invoices, because it drives every energy saving in the model.
Solar and storage
A 1.5 MW rooftop array is modeled to displace over 2.4 million kWh of grid purchases a year. That works out at roughly 1,600 kWh per kWp of installed capacity. Querétaro irradiance supports that range, but the result depends on tilt, soiling, degradation and the DC to AC ratio. None of those assumptions are stated in the model.
A 1 MW / 2 MWh battery energy storage system sits alongside it. Its duty is peak shaving. It charges during the day on solar output, discharges at 1 MW during the CFE punta window, and holds the plant's grid draw below a set line. The model credits it with removing 1,000 kW from billed demand every month.
Two Mexican specifics decide whether that duty is deliverable at another site. CFE publishes the punta, intermedio and base windows by tariff region and season in its GDMTH schedule, so a battery sized for a short punta in one region will not cover a longer window in another. And GDMTH bills two demand charges rather than one. The cargo por capacidad follows the maximum demand registered in punta. The cargo por distribución follows the maximum demand registered in the month across all periods. Cutting punta demand by 1,000 kW moves the first charge. It moves the second only if the month's overall maximum falls with it.
Array size also changes the permitting route. The Ley del Sector Eléctrico, published in the Diario Oficial de la Federación on 18 March 2025, sets 0.7 MW as the ceiling for generation that connects without a permit. A 1.5 MW array sits above that line and requires a generation permit from the Comisión Nacional de Energía, through the simplified procedure the law provides for capacity between 0.7 MW and 20 MW. Budget the study and permit time before the equipment lead time.
What the model shows
Capex, first-year savings, payback, IRR and avoided emissions by component. All currency figures are USD.
| Project Component | Capex (USD) | Annual Savings (USD) | Payback (Yrs) | IRR (%) | CO₂e Reduction (t/yr) |
| Compressed Air Optimization | 220,000 | 97,500 | 2.3 | 42% | 298 |
| Chiller Plant & HVAC Upgrade | 580,000 | 123,000 | 4.7 | 20% | 376 |
| Onsite Solar PV + BESS | 3,750,000 | 663,600 | 5.6 | 16% | 2,511 |
| Total / Blended Project | $4,550,000 | $884,100 | 5.1 | 18% | 3,185 |
The blended return is an 18% IRR on $4,550,000 of capital, and the model shows 3,185 tonnes of CO₂e avoided a year. A second reference point for the sector sits in MEP's automotive electricity cost case study.
Basis of the numbers
The 28% in the title is $884,100 of modeled annual savings divided by the site's stated annual energy spend of about $3.1 million. It is not a measured reduction in a CFE invoice. It combines three different effects: avoided grid purchases from onsite solar, lower consumption from the efficiency measures, and a lower demand charge from peak shaving. Only the third touches the demand component of the bill.
Nothing in this article was measured after construction. Every line comes from a pro-forma built during the site assessment in 2025. The baseline is twelve months of the plant's CFE invoices. The savings are first-year estimates for each measure, not a multi-year average and not a verified result under a measurement and verification protocol. Payback and IRR are published without a project life, a discount rate, a tariff escalation assumption, an O&M line or a battery replacement allowance. Ask for those five inputs before you compare this return with a machinery investment.
Two internal checks do not close, and the article states them rather than smoothing them over. First, a 1 MW / 2 MWh battery delivers about two hours at full discharge before depth of discharge limits and round-trip losses are counted. It cannot hold a 1,000 kW shave across a three-hour punta window such as the 7 PM to 10 PM example used here. The model reconciles only if the punta window at this site is about two hours, or if solar output and the reduced load carry the remainder. That assumption is not stated in the source model.
Second, the emissions figures do not tie to the energy figures. Taking the compressed air saving at the same $0.15 per kWh, the three measures avoid roughly 3.9 GWh of grid electricity a year. SEMARNAT set the factor de emisión del Sistema Eléctrico Nacional for 2024 at 0.444 tCO₂e per MWh, in an aviso dated 28 February 2025. At that factor the same energy yields about 1,700 tonnes a year, not 3,185. The model's implied factor is close to 0.82 tCO₂e per MWh, and its source is not given. Anyone loading this into a corporate inventory should rebuild the tonnage from the published factor.
One more variable is missing. CFE bills in pesos, and the model states neither the MXN to USD rate used nor the date it was taken.
What to test before copying it
Three checks decide whether this sequence works at your plant.
Start with the demand charge. Pull twelve consecutive CFE invoices and split the billed demand into cargo por capacidad and cargo por distribución. Then model the change in each one separately. Removing 500 kW to 1,000 kW from your punta maximum reduces the capacity charge, but it reduces the distribution charge only if the month's overall maximum falls too. That gap is where most peak shaving cases lose their return.
Check the roof before the capital. On this campus 1.5 MW of array took close to the whole usable roof area, so roof area set the ceiling on generation, not the budget. A structural assessment against the added dead load and the local wind loading comes before any array is sized.
Then benchmark the three loads that carry most industrial waste: compressed air, chillers and HVAC, and process heating. On the engagements Mexico Energy Partners has run, a Level 2 audit as defined in ANSI/ASHRAE/ACCA Standard 211-2018 has surfaced measures worth 10% to 15% of site electricity spend. That is our own observed range across client sites, not a published benchmark.
One limitation frames all three. This model assumes a punta window and a set of CFE rates specific to one tariff region and one season. Returns move once the site sits in a different region or a different seasonal schedule.
Test this model against your own CFE bills
Send twelve consecutive CFE GDMTH invoices and, where you have it, 15-minute interval demand data for one full billing cycle. Mexico Energy Partners will return your billed demand split between cargo por capacidad and cargo por distribución, the kW a peak shaving battery would actually remove from each, and the capex range that would put payback inside five years. If the roof is a candidate, tell us the building age and whether structural drawings exist. We confirm receipt and tell you what else is needed within two business days. Nothing is promised on savings or eligibility until your own data has been reviewed.
Sources
- Diario Oficial de la Federación, Ley del Sector Eléctrico, 18 March 2025. Sets the 0.7 MW threshold for permit-free generation and the simplified permit route for 0.7 MW to 20 MW.
- Comisión Reguladora de Energía, RES/151/2016, Código de Red, Diario Oficial de la Federación, 8 April 2016. Applies connection requirements to load centers at medium and high voltage.
- SEMARNAT, Aviso: Factor de Emisión del Sistema Eléctrico Nacional 2024, 28 February 2025. Factor of 0.444 tCO₂e per MWh for indirect emissions from electricity use.
- US Department of Energy, Improving Compressed Air System Performance: A Sourcebook for Industry, compressed air systems fact sheet on leaks. Leak load of about 20% of production capacity in plants without an active leak program.
- ANSI/ASHRAE/ACCA Standard 211-2018, Standard for Commercial Building Energy Audits, reaffirmed 2023. Definition and scope of a Level 2 audit.
- Comisión Federal de Electricidad, Gran Demanda en Media Tensión Horaria tariff schedule, as published and in force in August 2025. Charge components, the split between cargo por capacidad and cargo por distribución, and the publication of punta, intermedio and base windows by region and season.
- Consejo Consultivo de Eficiencia Energética Aplicada, published explanation of the GDMTH tariff. Capacity charge driven by maximum demand in punta and distribution charge driven by the month's maximum demand.
- Mexico Energy Partners site assessment pro-forma for the Querétaro plant, 2025. Source of all capex, savings, payback, IRR and emissions figures in the table.