← Back to the plan

Demand-limiting alternatives for TRC (1660 9th St, Santa Monica, SCE)

Researched 2026-09-06. Goal: guarantee 15-minute average demand never exceeds ~19 kW, and see demand live with alerts.

0. The finding that changes the plan: soft starters cannot fix a 15-minute demand number

Confidence: HIGH. Victor and Farzad are right; the Ari/Greg theory is wrong.

SCE bills demand as the highest 15-minute average kW in the billing period. Compressor inrush (locked-rotor amps) lasts roughly 0.2-2 seconds. The arithmetic:

Sixty watts. A soft starter on all seven units changes the billed 15-minute demand by an amount the meter cannot resolve. Soft starters are real and useful, they cut mechanical/electrical stress, stop light flicker, and let a battery or generator start a compressor. They are not a demand-charge fix.

What actually produced 19.5 kW on 2026-01-22 was coincident running load, sustained for a full quarter hour:

Four AC units running together (12 kW) plus the sauna (12.3 kW) is 24 kW on its own. That is the real exposure, and it is why the $4,500 seven-soft-starter quote from Dennis should not be bought as a demand solution.

Corollary: the staff "stagger AC turn-ons by 15 minutes" policy also does not work as intended. Staggering starts spreads inrush, but the units are still all running an hour later. It helps only if the stagger is long enough that units finish their pull-down and cycle off before the next starts, which is not what a 15-minute stagger achieves.

1. Real-time demand monitoring, "the electrical tracker"

The requirement that eliminates most consumer products: a true 15-minute rolling demand calculation with a threshold alert, not an instantaneous-watts alert.

Monitoring pick: Eyedro EB6-EW-E3 with 200 A industrial sensors, $349 + MyEyedro Pro. It is the only sub-$500 device that natively computes an averaged demand alert, handles Delta as well as Wye, and does not care whether the ACs turn out to be 3-phase. Add an Emporia Vue 3 with 16 branch CTs (~$250) as a second, circuit-level meter so you can see which unit is running, that is the diagnostic you need to argue the PCOC case to SCE. Total hardware under $700.

Important: monitoring alone provides zero guarantee. An alert on a phone at 11pm does not stop the meter.

2. The hard guarantee, automatic demand limiting

Ranked by how literally they guarantee the ceiling.

Tier 1, Energy Sentry 9388B (Brayden Automation), the actual guarantee

energysentry.com/PP-9388B.php · technical manual PDF

This is a purpose-built demand controller, in the field since 1978, and it does exactly the job:

• Demand averaging period: 15, 30, or 60 minutes , matches SCE's interval exactly.
• Demand calculation update: under 1 second. It computes the rate of rise toward the limit and sheds before the interval average can cross it, rather than reacting after the fact.
• 8 relays standard, expandable to 16; up to eight 30 A DPST-NC relays at 300 VAC, plus low-power 3 A remote relays sized for HVAC 24 V control circuits, the right way to interrupt a compressor is through its thermostat call, not its line power.
• Strict shed priority order: least important shed first, restored last.
• Audible alarm approaching the limit; 12 seasonal limit changes per year; NEMA 3R enclosure.
• Model variants take 3-phase CTs or a KYZ meter pulse.

The published case studies are precisely this problem. Jim's Automotive Machine Shop used a 9388B to stay under a 25 kW rate-class threshold on Xcel; WOW! Children's Museum used one on five A/C units plus a dryer to get under 25 kW and move from the SG rate to the C rate (client profiles). That is the identical maneuver TRC wants with SCE's PCOC.

Cost: quote-only, Brayden does not publish pricing. Budget estimate for LA, stated as an estimate: $2,500-4,000 hardware (controller, CTs, 4-8 relays) + $3,000-6,000 C-10 installation, panel work, 24 V control wiring pulled to seven condensers, permit. Call it $6,000-10,000 installed. Contact Brayden Automation directly at brayden.com; they scope the load configuration with you.

This is the recommendation for the guarantee. It is the only option on this list whose entire design premise is "the 15-minute average shall not exceed N."

Tier 2 , SiteSage + DemandSmart (Powerhouse Dynamics)

Genuine commercial demand limiting with HVAC control, circuit monitoring, and a managed subscription that bundles hardware and install. Claims 10-25% peak demand reduction. Downside: it is a portfolio product sold to chains, TRC is one small site, and it is a recurring cost forever rather than a $7k box that runs unattended for 20 years. Worth one call for a price; do not wait on it.

Tier 3, DIY relay control (Shelly + Home Assistant)

Shelly Pro EM measures, Shelly relays interrupt the 24 V thermostat calls, Home Assistant runs the shed logic. ~$800-1,500 in parts plus your own engineering. Fully capable in principle, and you would write the 15-minute-integral logic yourself. Real risk: this becomes an unmaintained script that silently stops working, and you find out on the bill. Acceptable as a layer on top of the Energy Sentry, unacceptable as the guarantee itself.

Tier 4 , Consumer smart panels: do not buy for this

• Emporia Peak Demand Management (help article) is real and does prioritized shedding to a kW threshold , but it can only control Emporia smart plugs, Emporia EV chargers, and compatible thermostats. None of those switch a 220 V condenser. Useful for the mini fridges only.
• SPAN panel (tech note) does control to a current setpoint under NEC 750.30 / 220.70, but the setpoint is an ampacity/overload protection limit factory-set at 80% of the main breaker, not a user-chosen kW demand target, and it is a residential 200 A panel swap at $5,000-8,500. Wrong tool.
• Lumin (Smart Power Mode) sheds on instantaneous watts against a Power Limit, with 12 controllable circuits, $3,000-5,500 installed. Instantaneous shedding does reduce a 15-minute average, but it is tuned for inverter protection during outages, not for demand billing, and it is a residential product.

HVAC-side controls (complements, not substitutes)

• Pelican Wireless has a documented demand-authorization API (demand control attributes): each thermostat requests a cool cycle and waits for external authorization before starting. That is an exact, code-level interlock for "at most N of 7 compressors may run." Their Z8/Z24 zone controllers also carry a built-in 4-minute anti-short-cycle timer. This is the cleanest software path to compressor sequencing, and it composes with the Energy Sentry rather than competing.
• 75F MyStat (spec sheet) does demand response and 2-stage control, needs the 75F CCU ecosystem.
• ICM102 delay-on-make timer (icmcontrols.com), $18.91 each, 0.03-10 minute adjustable, universal 18-240 VAC, 2-wire, explicitly marketed for compressor staging. Seven of them is $133 in parts and a couple of HVAC-tech hours. Set staggered delays so no two condensers can energize together after a power event. Buy these regardless, they are nearly free and they do fix genuine simultaneous-start problems. They do not limit the 15-minute average, because a delayed compressor still runs.

3. Battery peak shaving , available now

SGIP is effectively dead for this project. Confidence: HIGH. SCE's own page states the ratepayer budgets are closed to new applicants (sce.com). The Large-Scale Storage budget closed to new applications after 2025-12-30; General Market non-residential continues but is largely waitlisted with 3-9 month queues. Do not underwrite a battery on SGIP. The 30% federal ITC still applies.

Installable within 60 days by a licensed LA contractor:

Sizing for this problem: shaving ~5 kW for a 4-hour daily peak needs ~20 kWh usable and ~5 kW continuous, so realistically two Powerwall-class units, $30k-40k installed. Compared with a $7k demand controller that guarantees the same outcome, a battery is a poor first purchase here. It earns its place only if outage resilience for the studios is independently worth $30k. Kora Power is covered separately; note only that a product still in "certification progress" as of 2026-08-07 has no place on the critical path.

4. HVAC efficiency route, replace, don't retrofit

The running-kW difference is large and it is the number that actually matters:

Sources: EngineerSkill, PickComfort. Inverter units also start at only 1.2-1.5x running current, so soft starters become moot.

Replacing seven aging single-stage 3-ton condensers with inverter units plausibly takes coincident AC load from ~21 kW to ~12-14 kW, and because inverters modulate instead of cycling at 100%, the coincident peak falls further than the average does.

Cost: $8,000-14,000 per unit installed in LA for a 3-ton inverter split system. Seven units: $60k-95k. This is the most expensive path and the slowest.

Rebates that are actually open in 2026: SCE's Commercial Energy Efficiency Program, administered by Willdan, reports ~$25,000,000 remaining for 2026 (willdanefficiency.com/rebates). Relevant measure: Packaged or Split HVAC Replacement, Gas to Heat Pump (eTRM SWHC046-06), $400-925/ton for units under 65 kBTU/hr at SEER2 14.4+ / HSPF2 6.7+. Seven 3-ton units = 21 tons = $8,400-19,400. Critical rule: the rebate must be reserved BEFORE installation. TECH Clean California commercial incentives are fully reserved and closed since 2025-10-31 (techcleancalifornia.net) , do not count on them.

5. Operational and near-free measures, do these this week

1. The sauna is probably your single largest controllable load. A commercial 3-phase Harvia Club-class heater is 12.3 kW at 34.1 A (spec). Verify the nameplate. If it is 12 kW, then the sauna plus four AC units alone exceeds 19 kW. Put it on a timer/contactor that hard-locks it out during the AC-heavy window, and make it the Energy Sentry's first shed priority. Highest-leverage single action available.
2. EV charging: permanently and physically excluded. 11.3 kW on a 19 kW ceiling is not survivable. Kill the circuit or move it off this meter. No policy, no signage, a breaker.
3. Seven ICM102 delay-on-make timers, $133 total, staggered 0/1/2/3/4/5/6 minutes, so a power blip cannot restart all seven condensers together.
4. LED wall: a wall's nameplate is all-white-pixels peak; typical content runs 15-35% of that. Get the actual measured draw off the new monitor before assuming it is a problem, then set a brightness cap. Cheap kW if it turns out to matter.
5. Mini fridges, SCE flagged these. They are small individually but numerous and always on. Consolidate; Emporia smart plugs can schedule them, and unlike the ACs, Emporia's Peak Demand Management genuinely can control them.
6. Reset the staff stagger policy. Replace "stagger by 15 minutes" with a hard rule on how many condensers may be running at once. That is the variable the meter sees.

6. Local contractors

Found via web search; license numbers as published by each firm, not independently verified against CSLB. Verify at cslb.ca.gov before signing.

Who is licensed to do what

• Victor's Supco / 5-2-1 Compressor Saver job: HVAC tech, no electrician needed. It is a hard-start capacitor wired inside the condenser's own control compartment, squarely C-20 (warm-air heating/AC) scope. Victor at $190/hr, 2-hour minimum, can do it. (Whether you should buy it at all: see section 0. It will not move the bill.)
• A demand controller: C-10 electrical contractor, required. It lands in the main service panel, takes CTs around service conductors, mounts a NEMA 3R enclosure, and pulls new 24 V control wiring to seven condensers. That is licensed electrical work with an LADBS/Santa Monica permit. Ideally the C-10 and the HVAC tech are on site the same day, since the control wiring terminates inside the condensers.
• Note the Zinsco panel. The insurance carrier waived replacement, but Zinsco panels have a documented history of breakers failing to trip. Any contractor asked to add CTs and a controller to that panel will raise it, and may decline to warrant work in it. Get that opinion in writing early, because it could turn a $7k job into a $7k job plus a panel replacement.

Ranked recommendation

(a) Monitoring, order this week, ~$700, installed in days Eyedro EB6-EW-E3 3-phase meter with 200 A industrial sensors, $349, plus MyEyedro Pro for the averaged-demand alerts. Add an Emporia Vue 3 with 16 branch CTs, ~$250, for per-circuit visibility. Installed by any C-10 in 2-3 hours. Set an alert at 16 kW average. Run this for 2-4 weeks before buying anything else, it will tell you definitively whether the sauna, the ACs, or something unmeasured is driving the peak, and it produces the evidence SCE will want with the PCOC declaration.

(b) The hard guarantee, Energy Sentry 9388B, ~$6,000-10,000 installed, 4-8 weeks Purpose-built, 15-minute averaging, sub-second update, rate-of-rise prediction, prioritized relay shedding. Shed order: sauna first, then AC units 7 → 1 in reverse importance, then water heater. Quote from Brayden Automation, install by RG Electric or Optimum. Layer Pelican thermostats later if you want per-zone compressor authorization, and buy the seven $19 ICM102 timers immediately regardless.

(c) Battery, optional, deferred $30k-40k for two Powerwall-class units, no SGIP, 30% ITC only. Do not put it on the critical path. Revisit if outage resilience for the studios is worth the money on its own.

(d) HVAC replacement, the right long-term answer, staged $60k-95k for seven inverter condensers, offset by $8,400-19,400 in SCE/Willdan rebates that must be reserved before installation. Do not do all seven at once. Use the monitoring data from (a) to identify the two or three worst units (Ari already fingered units 4 and 7) and replace those first. Each inverter swap permanently lowers the floor the demand controller has to work against.

What not to buy: the $4,500 seven-soft-starter package as a demand fix. If Victor wants to add hard-start kits to units that are struggling mechanically, that is a legitimate maintenance call at his hourly rate. It is not a $4,500 demand-charge project, and buying it will not move the number that got you a $3,600 bill.