Calibration to Shared Meters

Roovie calibrates simulation models against actual utility data.

For individual buildings, this means comparing simulated monthly energy use against utility bills and measuring how well the model tracks reality. For campus and portfolio settings where multiple buildings share a single meter, Roovie goes further: it uses simulation-based allocation to disaggregate the shared meter reading back to individual buildings.

Both workflows produce calibration metrics aligned with ASHRAE Guideline 14, giving teams a quantitative measure of model accuracy.

In One Line

Real utility bills in. Calibration metrics out. Works even when buildings share a meter.

How Individual Calibration Works

For buildings with their own utility meter, calibration is straightforward:

1. Enter 12 months of utility data — electricity in kWh and natural gas in therms, CCF, or kWh
2. Run a simulation for the same building
3. Roovie compares simulated results against actuals month by month
4. Calibration metrics are calculated automatically

The comparison tracks three fuel modes:

• Electricity only — kWh comparison with its own NMBE and CV-RMSE
• Natural gas only — comparison in native units (therms or CCF) and kWh, with its own metrics
• Combined energy — total energy comparison across both fuel types

What The Metrics Mean

NMBE (Normalized Mean Bias Error) measures systematic over- or under-prediction. A positive NMBE means the model consistently overestimates; negative means it underestimates. The ASHRAE Guideline 14 threshold for monthly data is plus or minus 5 percent.

CV-RMSE (Coefficient of Variation of Root Mean Square Error) measures the variability of prediction errors. It captures both bias and random error. The ASHRAE Guideline 14 threshold for monthly data is 15 percent or less.

Both thresholds are configurable. The system shows pass/fail status for electricity, natural gas, and the overall comparison.

What The Comparison Shows

• Annual totals: actual versus simulated for each fuel type
• Monthly bar chart: side-by-side actual and simulated values for each month
• Monthly table: per-month delta in kWh and percentage
• Metric gauges: NMBE and CV-RMSE with visual progress against thresholds
• Status indicators: green for passing, amber for review, gray for missing data

How Shared Meter Calibration Works

Many real buildings — campuses, schools, healthcare facilities, multi-building properties — do not have individual utility meters. They share a single meter at the site level.

Roovie's shared meter system handles this by using simulation results to allocate the aggregate meter reading back to individual buildings.

The Allocation Method

Shared meter reading (monthly kWh)
  × Building's simulation share of total group simulation
  = Building's allocated energy for that month

For each month, the system calculates what percentage of the total group simulation each building represents, then applies that percentage to the actual meter reading.

This is called simulated monthly share allocation.

Step-by-Step Workflow

1. Create a building group representing the buildings on the shared meter
2. Ensure each building has a completed simulation — this is what drives the allocation ratios
3. Enable the shared meter for the group
4. Select one designated simulation per building — the simulation whose monthly results will be used for proportional allocation
5. Enter the shared meter actuals — 12 monthly values for the aggregate meter, in combined or separate electricity and gas mode
6. Optionally configure residual load — a percentage reserved for unmodeled campus loads like site lighting, central plants, or EV charging
7. Optionally apply monthly overrides — manual percentage adjustments for specific buildings and months
8. Optionally lock a building's allocation — pin one building to its current simulation output while the remaining buildings rebalance proportionally
9. Save and recalculate — the system allocates actuals and computes calibration metrics for each building

Residual and Common Load

Not all energy consumed at a campus goes to the buildings. Site lighting, distribution losses, central plants, and other campus-level loads may not be modeled.

The residual load setting (0 to 100 percent per month) reserves a portion of the meter reading for these unmodeled loads before allocating the remainder to buildings. This keeps the building-level allocations honest.

Applied Simulation Locks

In some cases, one building's model is well-calibrated and should not be disturbed. An applied lock pins that building's allocated actuals to its simulation output. The remaining buildings then rebalance proportionally from what is left.

Locks persist across sessions and are explicitly cleared when no longer needed.

Stale State Detection

The system automatically detects when the shared meter allocation is stale — when a building is added or removed from the group, or when a designated simulation selection changes. The interface prompts recalculation to restore consistency.

Utility Rate Structures

Calibration is not just about energy quantities. Roovie also tracks costs through a flexible tariff system.

Supported Rate Structures

• Simple scalar rates — a single price per kWh or per therm, used as the default when no variable rates are defined
• Monthly variable rates — different rates per billing month, the most common structure for utilities with seasonal pricing
• Time-of-use rates — rates that vary by month, day of week, and time of day simultaneously
• Seasonal rates — explicit summer, winter, and shoulder season pricing
• Demand charges — infrastructure for peak-demand-based pricing

How Tariffs Are Structured

A utility rate plan contains rules. Each rule specifies:

• A rate (price per unit)
• Optional effective date range
• Optional month filter (for seasonal pricing)
• Optional day-of-week filter (weekday versus weekend)
• Optional time windows (for intraday peak and off-peak periods)
• Optional season label (summer, winter, shoulder)
• Priority for resolving overlapping rules

Multiple rules combine to create complex rate structures like PG&E's E-1 or Con Edison's tiered commercial rates.

Rate Context Tracking

After a simulation is costed, the system records a rate context showing:

• Whether variable rates were applied or whether the system fell back to scalar rates
• How many tariff rules were applied
• The period-average rate for electricity and natural gas
• Sample applied rules with plan ID, rule ID, and rate value

This makes cost calculations transparent. Users can see exactly which tariff rules drove the cost results, rather than trusting a black-box cost figure.

End-Use Breakdown

The calibration panel includes end-use comparison between the simulation model and historical data:

• HVAC, heating, cooling, fans, ventilation, pumps
• Lighting, equipment, refrigeration, hot water, other

The system highlights significant mismatches — for example, if the model shows 35 percent heating but utility records suggest 40 percent — so users can focus calibration efforts on the categories with the largest discrepancies.

What Makes This Different

Most calibration tools assume each building has its own meter. When buildings share a meter, teams are left to do allocation manually in spreadsheets.

Roovie handles the shared meter case directly:

• Allocation is model-based, using simulation results as the proportional driver
• Residual load accounts for unmodeled campus consumption
• Applied locks let teams pin well-calibrated buildings while adjusting others
• Stale detection prevents outdated allocations from persisting silently
• All results are explicitly labeled as allocated, not as true submetered data

The system does not pretend allocated actuals are the same as real meter readings. It labels them clearly as diagnostic estimates while still making them useful for calibration and comparison.

Bottom Line

Roovie's calibration system closes the loop between simulation and reality. It takes real utility data — from individual meters or shared campus meters — and produces ASHRAE Guideline 14 calibration metrics that quantify how well the model matches the building.

For shared meters, Roovie uses simulation-based allocation to disaggregate a single meter reading across multiple buildings, making portfolio-scale calibration practical even when submetering does not exist.

Combined with multi-dimensional tariff support and transparent rate context tracking, the system ensures that both energy and cost calibration are grounded in real utility data.