Computer-Use Agent Taxonomy中文
R07.08.01.03 · TASK CLASS

Building energy simulation

Model building energy and HVAC control and report results.

20 project-authored proposals · taxonomy v1.1 · 0 new benchmark runs. Verifiers and partial-credit rules below are designs, not executed results.

Typical applications
EnergyPlus · OpenStudio · DesignBuilder
Class GUI-need label
CL · CLI-sufficient

Task proposals · 20

Expand to inspect the complete original wording and proposed checks. Original proposal text remains English in both interfaces.

R07.08.01.03-T01 Run an EnergyPlus office model and report EUI and peak demand

Original instruction

Run EnergyPlus 24.1 on fixtures/small_office.idf with input/chicago_ohare_tmy3.epw. Write output/summary.json with annual site EUI (kWh/m2), total electricity (kWh), total natural gas (kWh) and the annual peak electric demand (kW) with its timestamp.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP07 · Simulation result
Conditions
None declared
Proposed difficulty
2 / 5

Proposed verifier

Grader runs the same IDF/EPW; EUI and energy totals within 0.5%, peak within 1% and timestamp exact. IDF hash must be unchanged.

Proposed partial credit

20% simulation completes; 40% EUI/energy totals; 40% peak value and timestamp.

R07.08.01.03-T02 Apply a glazing upgrade to every window and quantify savings

Original instruction

In fixtures/midrise_apartment.idf, replace all exterior window constructions with a SimpleGlazingSystem of U=1.4 W/m2K and SHGC=0.28, run baseline and upgrade with input/phoenix.epw, and write output/glazing.csv with heating, cooling and total site energy for both plus percent savings.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus · eppy
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP07 · Simulation result
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

Grader parses the modified IDF: every FenestrationSurface:Detailed window uses the new construction, no other objects changed; reruns both cases; energies within 0.5%.

Proposed partial credit

30% all windows updated, no collateral edits; 20% both runs; 50% table values within tolerance.

R07.08.01.03-T03 Generate an ASHRAE 90.1 Appendix G baseline and compute PCI

Original instruction

From the proposed design fixtures/proposed_lab.osm (OpenStudio), produce the ASHRAE 90.1-2019 Appendix G baseline (system type, envelope, LPD, rotated orientations), simulate all four baseline orientations and the proposed case, and report the Performance Cost Index with the tariff in input/tariff.json.

Platforms
mac · macOS win · Windows lin · Linux
Applications
OpenStudio · EnergyPlus
Interfaces
desk · Desktop GUI term · Terminal
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
click · Point & click type · Text & forms code · Code & commands
Work product
WP07 · Simulation result
Conditions
long · Long-horizon
Proposed difficulty
5 / 5

Proposed verifier

Grader checks baseline model rules (HVAC system type per climate/area, envelope U-values, LPD by space type, 4 rotations) and reruns; PCI within 1% of reference baseline workflow.

Proposed partial credit

30% baseline rules correctly applied; 20% four orientations simulated; 50% PCI within tolerance.

R07.08.01.03-T04 Write a supervisory controller that cuts July energy cost

Original instruction

Using the EnergyPlus Python API, control fixtures/retail_mpc.idf with input/chicago_july.epw for July 22-28 under the TOU tariff in input/tou.json. Adjust zone cooling setpoints (22-27 C) each timestep to cut cost at least 12% vs the fixed 23 C baseline with <=10 occupied hours outside the comfort band.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus · Python
Interfaces
term · Terminal api · API/SDK
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
AP · API-sufficient
Interaction primitives
code · Code & commands
Work product
WP07 · Simulation result
Conditions
long · Long-horizon
Proposed difficulty
5 / 5

Proposed verifier

Grader runs the submitted controller twice (determinism) plus the baseline: cost reduction >=12% and comfort violations <=10 occupied hours; setpoints never outside 22-27 C (actuator log).

Proposed partial credit

20% controller runs deterministically; 20% setpoint bounds respected; 30% comfort constraint; 30% cost reduction (pro-rata up to the 12 pct target).

R07.08.01.03-T05 Build a two-storey office from a floor plan in DesignBuilder

Original instruction

In DesignBuilder, model the office in input/office_plans.pdf: 2 storeys of 3.6 m, 6 perimeter and 2 core zones per floor, windows per elevation drawing, ASHRAE 90.1 constructions template. Run annual simulation with input/london.epw and export results; save output/office.dsb.

Platforms
win · Windows
Applications
DesignBuilder · EnergyPlus
Interfaces
desk · Desktop GUI
GUI need · task setting
G2 · GUI-essential
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
click · Point & click path · Continuous path geom · Precise geometry
Work product
WP04 · 3D/spatial model
Conditions
licensed · Licensed pro app long · Long-horizon
Proposed difficulty
4 / 5

Proposed verifier

Grader exports the IDF from the saved .dsb with DesignBuilder's batch exporter: zone count and floor areas within 2%, window areas per facade within 5% of drawing, annual energy within 3% of reference. Imported IDF/gbXML geometry is detected via model history and fails.

Proposed partial credit

35% zoning/geometry; 20% windows per facade; 15% template applied; 30% results match.

R07.08.01.03-T06 Assign space types and constructions in the OpenStudio Application

Original instruction

Open fixtures/school.osm in the OpenStudio Application. Assign space types (Classroom, Corridor, Gym, Office) per input/space_map.csv, apply the 'ASHRAE 189.1-2009 ClimateZone 4' construction set, run the 'Reduce LPD by 20%' measure and simulate. Save output/school_v2.osm and report site EUI.

Platforms
mac · macOS win · Windows
Applications
OpenStudio Application · EnergyPlus
Interfaces
desk · Desktop GUI
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
click · Point & click drag · Drag & drop type · Text & forms
Work product
WP07 · Simulation result
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

Grader loads the .osm with the OpenStudio SDK: all 48 spaces have mapped types, default construction set exact, LPD reduced 20% via the measure (workflow.osw lists it); EUI within 0.5% of rerun.

Proposed partial credit

35% space types; 20% construction set; 20% measure applied; 25% EUI.

R07.08.01.03-T07 Calibrate a model to utility bills per ASHRAE Guideline 14

Original instruction

Calibrate fixtures/clinic.idf to the 12 monthly electricity and gas bills in input/bills_2025.csv using input/actual_2025.epw. Tune only infiltration, plug-load density, schedules and boiler efficiency within the ranges in input/tuning_bounds.json until CV(RMSE) <=15% and NMBE within +/-5% for both fuels.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP07 · Simulation result
Conditions
None declared
Proposed difficulty
4 / 5

Proposed verifier

Grader reruns the submitted IDF: monthly CV(RMSE) and NMBE computed vs bills for both fuels; IDF diff restricted to allowed parameters within bounds.

Proposed partial credit

20% only allowed params changed within bounds; 40% electricity metrics met; 40% gas metrics met.

R07.08.01.03-T08 Run a three-level wall insulation parametric study

Original instruction

Create three variants of fixtures/house.idf with exterior wall insulation of 50, 100 and 150 mm mineral wool (k=0.035), run each with input/montreal.epw and write output/insul.csv with annual heating energy (kWh) per variant.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP02 · Records/table
Conditions
None declared
Proposed difficulty
2 / 5

Proposed verifier

Grader checks material thickness in each variant IDF and reruns; heating energies within 0.5%.

Proposed partial credit

30% variants built correctly; 70% heating values within tolerance.

R07.08.01.03-T09 Add a VAV-with-reheat system with economizer via the OpenStudio SDK

Original instruction

Using the OpenStudio SDK (Ruby or Python), add a VAV system with hot-water reheat, a gas boiler loop and a water-cooled chiller loop serving all conditioned zones of fixtures/office_ideal.osm (currently ideal loads), with a differential dry-bulb economizer. Autosize, simulate and save output/office_vav.osm.

Platforms
mac · macOS win · Windows lin · Linux
Applications
OpenStudio · EnergyPlus
Interfaces
term · Terminal api · API/SDK
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
AP · API-sufficient
Interaction primitives
code · Code & commands
Work product
WP06 · Code/engineering
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

Grader inspects the .osm: one AirLoopHVAC with VAV reheat terminals on every conditioned zone, plant loops with boiler and chiller, economizer type set; simulation runs with zero severe errors and unmet hours <300.

Proposed partial credit

40% air loop topology; 25% plant loops; 15% economizer; 20% clean run with unmet <300 h.

R07.08.01.03-T10 Fix excessive unmet hours in an EnergyPlus school model

Original instruction

fixtures/school_unmet.idf reports over 1,900 occupied unmet hours. Diagnose the causes (undersized hardsized coils, schedule mismatch, setpoint deadband) and fix them so heating plus cooling unmet occupied hours fall below 300, without changing envelope or internal loads. Document fixes in output/fixes.md.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP06 · Code/engineering
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

Grader reruns: unmet occupied hours <300; IDF diff shows no envelope or internal-load changes; fixes.md names the three planted causes.

Proposed partial credit

50% unmet hours below threshold (pro-rata); 20% no forbidden edits; 30% causes identified.

R07.08.01.03-T11 Build a schematic school model with the eQUEST wizard

Original instruction

In eQUEST, use the Schematic Design Wizard to model the school in input/school_brief.pdf (2 floors, 6,200 m2, L-shaped footprint, packaged VAV with gas heat, 30% WWR) with weather input/atlanta.bin. Run the simulation, save output/school.pd2 and report annual electricity and gas.

Platforms
win · Windows
Applications
eQUEST
Interfaces
desk · Desktop GUI
GUI need · task setting
G2 · GUI-essential
GUI need · ordinary access
G2 · GUI-essential
Interaction primitives
click · Point & click type · Text & forms geom · Precise geometry
Work product
WP07 · Simulation result
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

Grader reads the generated .inp: floor area within 2%, two floors, L footprint vertices match brief, system type and WWR correct; reruns DOE-2 and compares totals within 1%. The standalone DOE-2.2 engine is not exposed to the agent, and .inp files written outside the wizard lack the .pd2 wizard state and fail.

Proposed partial credit

30% geometry/footprint; 30% HVAC and WWR; 40% energy totals.

R07.08.01.03-T12 Turn a Rhino massing into a Honeybee energy model and run it

Original instruction

In Rhino 8 with Grasshopper and Ladybug Tools, convert the massing in fixtures/mixed_use.3dm into a Honeybee model: one room per floor per use (retail ground, office above), 40% WWR on south and 25% elsewhere, programs and constructions per input/programs.json. Run EnergyPlus and export output/mixed_use.hbjson with results.

Platforms
mac · macOS win · Windows
Applications
Rhino · Grasshopper · Ladybug Tools
Interfaces
desk · Desktop GUI
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
click · Point & click drag · Drag & drop code · Code & commands
Work product
WP04 · 3D/spatial model
Conditions
licensed · Licensed pro app
Proposed difficulty
4 / 5

Proposed verifier

Grader validates the .hbjson (honeybee-schema): room count, programs by floor, aperture ratios within 1 pt per orientation; reruns simulation with honeybee-energy and compares EUI within 1%.

Proposed partial credit

30% rooms and programs; 25% apertures; 15% constructions; 30% EUI.

R07.08.01.03-T13 Compare cooling energy under a 2050 climate file

Original instruction

Convert input/future_2050_hourly.csv (dry bulb, dew point, RH, pressure, GHI, DNI, DHI, wind) into a valid EPW using the header of input/denver_tmy3.epw, run fixtures/warehouse.idf with both files and report the change in annual cooling energy in output/answer.txt.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus · Python
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP07 · Simulation result
Conditions
None declared
Proposed difficulty
2 / 5

Proposed verifier

Grader validates the EPW (8760 rows, field order, units) and reruns both; reported cooling change within 1%.

Proposed partial credit

40% valid EPW; 20% both runs; 40% cooling delta.

R07.08.01.03-T14 Add an EMS precooling and lighting-shed demand-response program

Original instruction

In fixtures/office_dr.idf, add EnergyManagementSystem programs that precool zones to 22 C from 13:00-16:00 and, during 16:00-19:00 on the event days in input/dr_events.csv, raise cooling setpoint to 26 C and dim lighting 20%. Report event-window peak reduction vs baseline in output/dr.json.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP06 · Code/engineering
Conditions
None declared
Proposed difficulty
4 / 5

Proposed verifier

Grader reruns baseline and DR models: EMS actuates only on listed event days (actuator output variables), average event-window demand reduction >=15%, reported reduction within 1 pt.

Proposed partial credit

30% EMS objects valid; 30% correct days/windows; 40% peak reduction >=15 pct and reported correctly.

R07.08.01.03-T15 Extract an end-use breakdown from eplusout.sql

Original instruction

From the EnergyPlus SQLite output fixtures/run1/eplusout.sql, produce output/enduse.csv with annual energy by end use and fuel (as in the ABUPS table) and output/enduse.png as a stacked bar chart by fuel.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus · Python · SQLite
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP02 · Records/table
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

CSV compared to ABUPS table values in the SQL (exact to 0.1 kWh); PNG checked for existence, dimensions and bar count equal to end uses with nonzero energy.

Proposed partial credit

70% CSV exact; 30% chart present with correct categories.

R07.08.01.03-T16 Draw a chilled-water plant in DesignBuilder's HVAC schematic

Original instruction

In fixtures/hq.dsb (DesignBuilder), replace the ideal loads with detailed HVAC: a chilled-water loop with 2 equal chillers in parallel and a cooling tower, a hot-water loop with a condensing boiler, and 3 VAV AHUs serving floors 1-3. Autosize, simulate with input/singapore.epw and report chiller plant electricity.

Platforms
win · Windows
Applications
DesignBuilder
Interfaces
desk · Desktop GUI
GUI need · task setting
G2 · GUI-essential
GUI need · ordinary access
G3 · GUI-only
Interaction primitives
click · Point & click drag · Drag & drop geom · Precise geometry
Work product
WP04 · 3D/spatial model
Conditions
licensed · Licensed pro app long · Long-horizon
Proposed difficulty
5 / 5

Proposed verifier

Grader exports IDF from the saved .dsb: plant topology (2 chillers parallel, tower on condenser loop, boiler), 3 air loops mapped to floors; chiller electricity within 3% of reference. DesignBuilder's script manager is disabled in the agent license, and IDF imports into the model are detected via model history and fail.

Proposed partial credit

40% plant topology; 25% air loops; 10% autosize clean; 25% chiller energy.

R07.08.01.03-T17 Optimize window-to-wall ratio per orientation for energy cost

Original instruction

For fixtures/office_box.idf with daylighting controls in perimeter zones and input/denver.epw, search WWR 10-60% (5% steps) independently for N, E, S, W facades to minimize annual energy cost under input/tariff.json. Write output/wwr.json with the chosen WWRs and cost.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus · Python
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP07 · Simulation result
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

Grader reruns the chosen configuration and a reference grid search: chosen cost within 0.5% of the optimum; window geometry in IDF matches reported WWRs within 1 pt.

Proposed partial credit

20% WWR geometry correct; 30% runs reproducible; 50% cost within 0.5 pct of optimum (partial credit to 2 pct).

R07.08.01.03-T18 Set up a measure-based parametric analysis in OpenStudio PAT

Original instruction

In the OpenStudio Parametric Analysis Tool, create a project from fixtures/library.osm with measures 'Add rooftop PV' (60% roof), 'Reduce LPD 30%' and 'Increase wall R-value' (R-20, R-30), create the 4 design alternatives in input/alternatives.csv, run locally and export the results CSV.

Platforms
mac · macOS win · Windows
Applications
OpenStudio PAT
Interfaces
desk · Desktop GUI
GUI need · task setting
HY · Hybrid
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
click · Point & click drag · Drag & drop type · Text & forms
Work product
WP07 · Simulation result
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

Grader reads the PAT project (pat.json): measures and arguments exactly per alternative; results CSV EUIs within 0.5% of reruns of each alternative.

Proposed partial credit

35% measures/arguments configured; 25% four alternatives; 40% results match.

R07.08.01.03-T19 Debug a failing EnergyPlus model with four planted severe errors

Original instruction

fixtures/hospital_broken.idf terminates with severe errors. Fix the 4 planted problems (node connection, schedule type limit, unconnected plant branch, missing material) without changing design intent per input/design_notes.md, run to completion and write output/fixes.md with each error and fix.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP06 · Code/engineering
Conditions
None declared
Proposed difficulty
4 / 5

Proposed verifier

Grader reruns: zero severe errors; annual energy within 2% of the reference-fixed model; fixes.md matches the 4 planted defects.

Proposed partial credit

60% planted defects fixed (15 pts each); 25% results within 2 pct; 15% fixes documented.

R07.08.01.03-T20 Model a data hall with ITE objects and report annual PUE

Original instruction

In fixtures/datacenter.idf, model the 400 kW data hall with ElectricEquipment:ITE:AirCooled objects (input/ite_spec.json), a CRAH system and an airside economizer. Run with input/ashburn.epw and write output/pue.json with annual PUE and monthly PUE.

Platforms
mac · macOS win · Windows lin · Linux
Applications
EnergyPlus
Interfaces
term · Terminal
GUI need · task setting
CL · CLI-sufficient
GUI need · ordinary access
CL · CLI-sufficient
Interaction primitives
code · Code & commands
Work product
WP07 · Simulation result
Conditions
None declared
Proposed difficulty
3 / 5

Proposed verifier

Grader verifies ITE objects match spec and reruns: annual PUE within 0.01 and monthly values within 0.02.

Proposed partial credit

35% ITE objects correct; 15% economizer configured; 50% PUE values.

Mapped research references

Existing benchmark-to-class links from the research catalog. They are not record-level annotations, executions or product endorsements.

In this workflow family