Working on a complex lease audit here in Phoenix involving APS demand charges. Tenant operates a data center that creates significant demand spikes, but they're being charged for building-wide demand under the master meter. The issue is that their peak demand occurs at different times than the rest of the building. Should demand charges be allocated based on individual tenant peak demand or building-wide peak? Their lease just says "proportionate share of utility costs."
Arizona Commercial Lease - APS Demand Charges Question
Sarah, demand charge allocation is one of the trickiest aspects of commercial lease audits. Here in Memphis with MLGW, I always look at the actual load curves to determine each tenant's contribution to the building peak. If your client's demand peak is off-peak for the building, they shouldn't pay the same rate as tenants who contribute to the system peak. Do you have access to interval meter data showing time-of-use patterns?
Randy's absolutely right about load curve analysis. Here in Tucson, also dealing with APS territory, I've found that data centers often have completely different demand profiles than typical office tenants. Your client might be creating high demand during off-peak hours when the building's overall demand is low. This means they're not contributing to the utility's peak demand charges that get billed to the building. Have you requested the 15-minute interval data from APS?
This is fascinating because it's the opposite of what we usually see. Typically tenants complain about being charged for demand they don't create. But in your case, the data center might actually be creating demand that doesn't coincide with billing demand periods. Seattle City Light has similar time-of-use issues. The key is proving when the building's billable demand actually occurs versus when your client's demand peaks. Sub-metering data would be ideal here.
David makes an excellent point about timing. Here with Rocky Mountain Power in Salt Lake, demand charges are based on the 15-minute interval with the highest usage during billing periods. If your data center peaks during off-peak hours, they might not be contributing to the demand charge at all. I'd request a full year of interval data and map your client's usage patterns against the building's peak demand periods.
Connie's approach is solid. Here in Spokane with Avista, we had a similar case with a medical facility that ran equipment during overnight hours. Their peak demand never coincided with the building peak, but they were paying 40% of all demand charges based on square footage. We got interval data proving their non-coincident contribution and reduced their allocation from 40% to about 12%. Saved the client $180,000 over three years.
Lee, that's exactly the type of analysis I need to do. APS provided 12 months of interval data and I can see that the data center's peak loads occur primarily between 2-6 AM when the building's overall demand is at its lowest. The building's peak demand occurs during standard business hours when the data center is running at about 60% capacity. This suggests they're being significantly overcharged for demand costs.
Sarah, that's perfect data to support your case. Here in Sacramento with SMUD, I use a coincidence factor calculation to determine each tenant's actual contribution to peak demand. If the data center is only running at 60% during building peak hours, their demand allocation should reflect that coincidence factor, not their total capacity or square footage. This could result in substantial savings for your client.
Jennifer's coincidence factor approach is exactly right. Here in San Antonio with CPS Energy, I've used this methodology to prove that 24/7 operations like data centers often have lower coincidence factors than traditional office tenants. The math can get complex, but the savings are usually substantial. Make sure you're calculating both coincident and non-coincident demand charges correctly - APS bills both components differently.
Angela brings up a crucial point about APS's demand charge structure. Here in New Orleans with Entergy, we have similar dual demand charges. The coincident demand charge is based on the customer's contribution to the utility's system peak, while non-coincident is based on the individual customer's peak regardless of timing. Your data center might have high non-coincident demand but very low coincident demand contribution.
This thread is incredibly valuable for understanding demand charge allocations. Here in Carson City, we don't see as many complex data center situations, but the principles apply to any tenant with unique load patterns. Sarah, have you considered proposing a separate demand charge allocation methodology in the lease amendment? Some landlords will agree to usage-based allocations rather than square footage when presented with solid interval data analysis.
Franklin, that's a great suggestion about proposing an amended allocation method. Based on the interval data analysis, the data center's coincident demand factor is only 0.23 compared to the building average of 0.85. This means they should pay about 27% of what they're currently being charged for coincident demand charges. I'm preparing a proposal for the landlord to implement a usage-based allocation starting with the next lease renewal.
Sarah, those numbers look very reasonable for a data center operation. The 0.23 coincidence factor aligns with what I've seen for similar facilities here in Tucson. Make sure your proposal includes language about ongoing interval data monitoring to ensure the allocation remains accurate as the data center's operations change. APS is usually pretty cooperative about providing monthly interval data for these types of specialized allocations. Keep us posted on the landlord's response!