Picked up my first data center client in Atlanta. Georgia Power Rate PLM. The facility draws 2,800 kW with a load factor above 95%. Demand is essentially flat 24/7. The demand charge component of their bill is $42,000 per month. Because the load is so flat theres virtually no demand optimization opportunity ? no peaks to shave, no loads to shift, no ratchet issue because every month is essentially the same demand. Where do I even look for savings on a flat-load data center?
Demand charges and data centers ? a different animal entirely
Derek ? data centers are counterintuitive for demand auditors. The demand charge itself is usually correct because the load is consistent. Your opportunities are: 1) Rate classification ? verify theyre on the optimal tariff for their load factor. High load factor accounts often benefit from industrial tariffs with lower demand charges but higher energy rates because the math favors them. 2) Voltage level ? primary or transmission voltage delivery has lower rates. 3) Power factor ? data centers with UPS systems sometimes have PF issues.
Marcus ? good direction. The data center is metered at secondary voltage (480V). Georgia Power Rate PLM has a provision for primary voltage delivery (4,160V or higher) with a discount of approximately $0.80 per kW on the demand charge. At 2,800 kW thats $2,240 per month or $26,880 annually. But taking primary voltage means the client owns and maintains the transformer.
Derek ? data center operators in Jacksonville are increasingly taking power at primary or transmission voltage specifically for the rate discount. The transformer maintenance cost is $3,000-5,000 annually. Against $26,880 in demand charge savings the economics are overwhelming. JEA offers a similar voltage discount. The barrier isnt cost ? its that the data center has to provide transformer switchyard space and accept maintenance responsibility.
Another angle for data centers ? check the UPS system efficiency. Older double-conversion UPS systems waste 8-12% of throughput as heat. Newer high-efficiency UPS systems waste only 2-3%. For a 2,800 kW facility the difference between 10% and 3% UPS losses is 196 kW of wasted demand. Upgrading the UPS reduces both demand charges and energy charges.
Hector ? UPS efficiency is a great point. This facility has older Liebert UPS systems running at about 91% efficiency. Modern units achieve 97%. The 6% improvement on 2,800 kW is 168 kW of demand reduction. At Georgia Power demand rates thats $2,520 per month. UPS replacement is extremely expensive but if theyre planning an upgrade anyway, the demand charge savings accelerate the payback.
Data centers also have cooling demand that IS optimizable. Raised the cold aisle containment question? Hot aisle/cold aisle containment reduces HVAC demand by preventing mixing. Free cooling economizers in cooler months can shut down mechanical cooling entirely. A Chattanooga data center I consulted on reduced cooling demand by 320 kW through containment and economizers.
Vernon ? cooling optimization is the one area where data center demand IS variable and reducible. This facility has no aisle containment and relies entirely on mechanical cooling. The cooling load is roughly 900 kW. Containment plus economizers in Atlantas climate could potentially reduce that by 200-300 kW for 4-5 months of the year. Ill include this in the audit recommendations.
Excellent thread on data center auditing. Key takeaway: for high load factor facilities, traditional demand shaving doesnt apply. Instead focus on rate optimization (voltage level, tariff selection), infrastructure efficiency (UPS, cooling), and power factor. Derek is building a comprehensive audit package that addresses each angle. The combined potential savings from voltage discount, cooling optimization, and future UPS upgrade easily justify the audit engagement even without traditional billing errors.