Demand control for commercial buildings

A commercial electric bill has two large numbers on it. One is energy: the kilowatt-hours you used. The other is demand: the highest averaged draw you hit during the month. Many commercial tariffs use a 15-minute interval; others use 30.

Most people manage the first one. Lighting retrofits, high-efficiency rooftop units, occupancy sensors. All of that reduces kWh and all of it is worthwhile. Demand does not respond to any of it.

On a 15-minute commercial tariff, a billing month holds roughly 3,000 intervals, and your demand charge is set by exactly one. The highest one. You can run an efficient building for 2,999 intervals and still pay a full month of demand charges because three rooftop units, a water heater, and a kitchen line all came on inside the same quarter hour on a Tuesday in July.

On a residential bill, demand is often small or absent. On a commercial bill it frequently represents a large share of the total, especially in a building with a low load factor.

Here is what that costs in dollars. Under Arizona Public Service Rate Schedule E-32 M, a medium general service account pays $14.690 per kW for the first 100 kW at secondary voltage. On the smaller E-32 S schedule it is $12.393 per kW. Both effective March 8, 2024 under Decision No. 79293.

At $14.69 per kW, shaving 20 kW off a peak is worth roughly $294 a month, about $3,500 a year, on the demand line alone. You did not use less electricity. You just stopped paying for the worst interval of the month.

The Energy Sentry Demand Automation Computer watches the total electrical demand entering your building in real time and prevents your controlled loads from stacking up in the same interval.

It measures the whole service, not individual circuits. Current transformers on the incoming three-phase conductors, ahead of the main breaker, feed the controller a continuous picture of total demand. Where the utility restricts access to the CT cabinet, the controller can take a KYZ pulse input directly from the meter instead.

It predicts instead of reacting. The controller samples instantaneous demand a thousand times per second and tracks both the demand level and how fast it is climbing. If the rate of rise means the limit will be crossed before the interval closes, the system acts first. Simple load-shedding devices act after the spike, which is too late, because the spike is the billable event. Bill Brayden and his engineering partner spent roughly a year writing that algorithm and another year to eighteen months testing it against simulated load scenarios. It has been refined across three generations of hardware since.

It sequences, it does not cut. Loads get priorities during commissioning. A water heater reheat cycle can be held a few minutes and nobody notices. Rooftop units get staged against each other so they do not all pull inrush at once, with minimum off-time protection so compressors never short-cycle. A dryer someone just started sits at the top of the priority list and is the last thing touched.

It works in small blocks. Bill Brayden describes commercial load management as Tetris. A 25 kW furnace with three heating stages gets controlled as several small blocks rather than one large one, which produces a smooth demand curve instead of an oscillating one.

Nobody has to do anything. No app to check, no alert to answer, no staff procedure. The system runs continuously and the building operates normally.

Energy Sentry is not an energy monitor. It will not show you a dashboard and ask you to change your behavior. It is not a smart breaker and it does not trip circuits. It is a whole-service demand coordination system.

In a house, every major load sits within a few feet of the panel. In a commercial building they are spread out: six rooftop units on a roof, water heaters in three mechanical rooms, PTAC units in forty apartments. Running dedicated control wiring to all of them is a multi-day job with conduit, permits, ceiling access, and an electrician's hourly rate on every hour of it. On many buildings that cost alone kills the project before the savings math gets a hearing.

Power line carrier removes that job. A transmitter board mounts inside the controller at the main distribution panel and sends a 270 kHz signal out over the copper wiring already in the building. A receiver at each controlled load, inside that equipment's own electrical compartment, listens for its address and switches the load. Effective range is approximately 2,000 wire feet from the transmitter.

The install changes from pulling wire across a building to mounting receivers at equipment. That is the difference between a project that pencils and one that does not, which is why roughly 90 to 95 percent of our commercial installations use it. The transmitter has to sit at the center of the electrical distribution rather than out on a branch circuit, and confirming that is one of the things our site inspection covers.

There is a second reason commercial customers buy, and for smaller accounts it is often the stronger one.

Utilities sort commercial customers into rate classes by demand. Cross a threshold and you are reclassified into a different tariff with a different service charge, different energy rates, and in many cases a demand charge that did not exist on the rate below it. The reclassification is automatic, based on the utility's annual review of your peak, and you do not get to opt out.

The gap between classes can be severe. A small general service rate might carry a modest service charge and no demand component at all. The next class up can add a service charge many times higher plus a per-kW charge on every kilowatt. A customer at 52 kW on a rate that tops out at 50 kW pays a structurally different bill than the identical building next door running at 48 kW.

So the question is not "how much will I save per kW." It is "can I stay under the line," and the value of staying under is the whole difference between two tariffs. We see it most with churches near a small-general-service ceiling, offices and retail that grew into the next class after adding HVAC capacity, and shops that tripped a threshold by adding a bay. The system suits this well because it defends a hard configured limit rather than aiming at an average. If the number to stay under is 49 kW, that is the number it holds.

If you are within about 15 percent of a rate threshold, tell us. The analysis is different and the answer is often better.