Skip to main content
Docs / Project Modelling
Docs Menu

Dispatch Calendars and Sizing

Set operating bounds, test decision-price and forecast assumptions, and compare a small set of asset sizes in Project Modelling → Options.

Set a calendar bound

  1. Under Dispatch calendars, select Add calendar.
  2. Enter Name, Unit and Default value (blank = scalar bound).
  3. Select Add time window. Enter Months (1–12), Weekdays (Mon 0–Sun 6), Start minute, End minute and Value.
  4. Under Site limits, select the calendar for Site import calendar or Site export calendar.
  5. Save a case revision. Use aligned, timestamped profiles for a chronological run.
Site import calendar with a 1000 kW default and an 800 kW time window
This input example uses an 800 kW import ceiling from minute 600 to minute 720, with a 1,000 kW default.

The example window is 10:00–12:00 in the displayed case timezone, UTC. All 12 months and all 7 weekdays are selected. These are example operating-limit assumptions. The scalar connection limit remains a hard limit.

Attach battery operating bounds

Add a Battery storage asset under Design first. Then return to Options. Use these four operating-bound selectors:

SelectorMeaning
Minimum SOC / Maximum SOCPercent of energy capacity at the end of an interval.
Minimum net charge / Maximum net chargekW, or percent of rated power. Positive means charge; negative means discharge.

Our configuration example selects 20% minimum SOC, 90% maximum SOC, −100 kW minimum net charge and 100 kW maximum net charge. These are input bounds, not a calculated dispatch result. The battery’s scalar limits still apply.

Battery minimum and maximum SOC and net-charge calendar selections
Four separate calendar selections define the battery envelope.
Blank calendar values keep scalar bounds. Changes inside an interval are rejected. Contradictory or unreachable bounds stop a run. The chronological method does not prove global scheduling feasibility. Site shortages are reported as unserved load; excess generation is curtailed.

Reuse an exact library calendar

  1. In Profiles → Shared input library, create or select a dispatch_calendar item.
  2. Copy its pin: itemId, versionId and contentDigest.
  3. In Options, expand Reuse a shared calendar. Paste the pin into Calendar library pin JSON.
  4. Select Copy pinned calendar, then choose it in a site or battery selector.
Copied immutable dispatch calendar pin
The product confirms that pinned calendar values were copied into the draft.

Our reusable Export ceiling calendar has unit kW, default 500 and no time windows. Copying it does not change the library. Later library edits do not update the case. Save a case revision to retain the copied values. A calendar ID must be unique in the case.

Set a decision-price experiment

  1. Under Dispatch price experiment, select Enable price experiment.
  2. Choose Price information and Threshold horizon.
  3. For Rolling elapsed hours, enter Rolling hours. Set Trailing smoothing intervals.
  4. Use a Detailed Engineering Case with suitable price inputs. Review results before retaining the assumption.
Past-observation decision prices with a rolling 24-hour horizon
The configuration example uses past observations, a rolling 24-hour horizon and 2 trailing intervals.

Perfect information — actual prices known in advance is an idealised assumption. Past observations only uses causal information. Calendar horizons use the case timezone. Smoothing can cross a horizon boundary. SOC carries continuously; it does not reset at each horizon.

This changes decision prices, not settlement prices, load or generation inputs. It is a price-threshold heuristic, not an optimiser. A supplied forecast needs a distinct, aligned price profile with explicit currency units and a valid frozen issue time; ordinary values alone are not enough. Use Disable price experiment to remove it.

Plan ahead with rolling look-ahead

  1. In Options → Dispatch price experiment, select Enable price experiment.
  2. Set Dispatch method to Bounded rolling look-ahead — beam-search heuristic. Enter Look-ahead intervals and Beam width.
  3. Choose the Source for Load forecast and Generation forecast. For Signed error on strictly past observations, set Signed error (%), Prior observation intervals and Cold-start forecast (kWh per interval).
  4. For supplied forecasts, select an aligned power or energy profile and its issue time. Save a case revision and run Detailed.

A signed forecast uses the mean of strictly earlier observations, multiplied by (1 + signed error / 100). The first decision uses the stated cold-start assumption. Current and future actuals are excluded. This is a user assumption, not a measured accuracy claim. Annual-normalised load and generation need supplied forecasts. Generation forecasts describe aggregate generation, not a capacity factor.

Price information is separate. Perfect information — actual prices known in advance still assumes future prices are known. Choose Past observations only for causal decision prices.

Up to 16 storage and dispatchable-generator assets share a bounded decision. Only the first action is executed; the next decision starts from actual SOC, generator output and remaining energy budgets. Static generator availability is required. Complete year-start battery and renewable curves and generator running-cost curves are supported. Scheduled storage, startup and idle fuel, unit commitment and within-year aging are not modelled. Six actions per asset and a four-million candidate-interval cap can miss a feasible or cheaper path. This is not demand-charge or network optimisation.

Keep SOC physical

Continuous SOC — no reset (default) carries inventory forward. For a real end target, choose Physical window-end target — real energy required, then enter Independent boundary window (intervals) and Window-end SOC (%). Blank target SOC uses each battery's initial SOC. Real charge and discharge must meet all power, efficiency, SOC, connection and calendar limits.

Only prior observations enter the signed-error forecast. A zero error setting does not reveal future actual demand.
Only prior observations enter the signed-error forecast. A zero error setting does not reveal future actual demand.

Compare bounded sizes

  1. Under Bounded financial sizing, choose Asset, Mode and Rank by.
  2. Enter Minimum, Maximum and Steps (including endpoints) for each size dimension.
  3. Select Check run complexity. Review the candidate count and work limit.
  4. Select Compare candidates. Read engineering status before the financial rank.
  5. For a candidate to retain, select Adopt as new draft. This creates a new saved draft; it does not replace the source case.
Two quick solar sizing candidates ranked by NPV
Two supplied solar sizes were compared. Both remain at engineering status review.

Worked example: two solar sizes

We selected solar, Quick — annual assumptions and NPV (higher). The range was 100–200 kW with 2 steps. The product completed 2 of 2 candidates:

CapacityNPV (AUD)IRRDiscounted payback
200 kW220,429.6512.71%12.11 years
100 kW−175,863.853.38%Not available

The case used AUD, a 20-year horizon, 8% discount rate, 2.5% inflation, 2% energy and export escalation, and 2.5% opex escalation. Annual load was 1,000,000 kWh, peak demand 250 kW, solar yield 1,500 kWh/kW and degradation 0.5% a year. Import value was 0.25 AUD/kWh, export value 0.08 AUD/kWh and quick self-consumption 70%. Installed capex stayed at 500,000 AUD and annual opex at 5,000 AUD for both candidates. There was no replacement cost, network, point of connection or snapshot. Other value streams were zero. These were the model assumptions for the two-candidate comparison.

The product warned that fixed total capex stays constant across sizes. Review costs before adoption. We adopted the 200 kW candidate as a new draft.

At most 25 discrete candidates are compared. The best tested point is not a global optimum. Quick mode has no interval dispatch or network validation. Detailed mode uses the existing chronological engine and network screen. Missing financial metrics are not zero. Use Measure runtime estimate (one candidate) for a measured projection. A changed input invalidates that estimate. These sizing ranges replace option sweeps only for this comparison.

Measure a sizing runtime estimate

  1. Set the sizing range and select Check run complexity.
  2. Select Measure runtime estimate (one candidate). This runs one real candidate; it is not a fixed speed claim.
  3. Read the measured projection and work counts before choosing Compare candidates.
Measured sizing runtime
The estimate excludes queue and startup time. Load and candidate behaviour can change it.

Our two-candidate Detailed example showed 4.43 seconds as its measured projection. The complexity result showed 8,760 intervals per candidate, 17,520 candidate intervals, 8 dispatch passes including evidence reruns, and 70,080 interval evaluations against a 500,000 limit. These are the actual counts for this example, not limits for all study types.

Use a fast sizing estimate

  1. Under Bounded financial sizing, set Mode to Fast estimate — first-year dispatch, lifetime extrapolation.
  2. Enter the size ranges.
  3. Select Measure runtime estimate (one candidate).
  4. Select Compare candidates.
  5. To use a size, choose a scenario in Apply sizes to scenario, or select Adopt as new draft. Save the case to keep applied sizes.
Bounded financial sizing with a measured runtime estimate
Two battery candidates: 35,040 of 500,000 interval evaluations. The measured runtime estimate appears beside the candidate count.
Fast estimate candidate comparison ranked by NPV
Fast estimate results ranked by NPV. Both candidates have engineering status review.

Worked example: two battery sizes

At North depot we chose the battery, power 1–2 kW in 2 steps, and duration 2 hours. The runtime estimate was 5.54 seconds (measured projection). Each candidate had 8,760 intervals, with 4 dispatch passes and 35,040 interval evaluations in total.

RankBatteryNPV (AUD)IRRDiscounted payback
12 kW, 4 kWh3,899.546.13%2.42 years
21 kW, 2 kWh3,373.8141.51%2.69 years

The product warned that the battery's fixed total capex stays the same across sizes. Check costs before you adopt a size.

A fast estimate models first-year operating physics only. Future years use scalar financial projections. They do not repeat dispatch or curve aging. It is not a lifetime physical validation. For a full review, run the chosen size in standard Detailed mode. Under Detailed run fidelity, a saved run can also use Fast estimate — first-year dispatch, lifetime extrapolation.

Save a separate participant dispatch goal

  1. Open Value → Bounded participant dispatch.
  2. Enter Dispatch request JSON with explicit participants, selected parties, tariff rates and allocation fractions, interval timestamps and hours, fixed load, flexible-load bounds, required energy and import limits. Include currency and kW/kWh units.
  3. Set direction to minimize or maximize. Select Solve bounded dispatch.
  4. Read the objective, residual and Full dispatch result JSON.
  5. Select Apply bounded dispatch to case, then Save case revision, to retain this separate experiment.
Bounded participant goal result
This solver is separate from the main feasibility dispatch and connection-study ranking.

Worked example: minimise an allocated tariff cost

We supplied two one-hour UTC intervals on 1 January 2026. Each had 1 kW fixed load, flexible bounds 0–2 kW and a 3 kW import limit. Required flexible energy was 2 kWh. Energy rates were 0.1 and 0.5 AUD/kWh. The selected customer had 75% of the tariff; the owner had 25%.

The product returned optimal, with signed objective 0.60 AUD and energy residual 0 kWh. Applying and saving retained the request when we returned to the panel.

This is import-only optimisation of one divisible flexible load with a fixed energy requirement and linear tariffs. There is no storage, generation, export, demand-charge, load-shedding or network-feasibility model. Maximise means maximum selected tariff cost, not profit. Per-tariff objectiveDirection can reverse a signed contribution; it does not reverse the actual bill. The saved experiment does not change the main case dispatch.

Use a participant goal in physical dispatch

  1. In Value → Participant value allocation, define and apply your participants.
  2. Leave Storage charging value blank to use the import rate. This is required for a case-energy goal and for joint optimisation.
  3. In Design, set storage Control policy to Tariff arbitrage.
  4. In Value → Physical dispatch goals, choose Case goal, the selected participants, each tariff's direction and explicit cost shares.
  5. Select Apply physical dispatch goals, then Save case revision. Run the Detailed Engineering Case.
The case-energy goal assigns 60% of import energy cost to the site host and 40% to the battery owner. Both objectives are selected.
The case-energy goal assigns 60% of import energy cost to the site host and 40% to the battery owner. Both objectives are selected.

Unlike ordinary value allocation and the separate bounded flexible-load experiment, this goal changes main-case dispatch. Maximise allocated energy cost means higher cost, not higher profit. Actual bills retain their original rates. The default physical goal is heuristic dispatch, not a global optimum. Only supported interval-energy tariff goals are admitted; demand, netted tariffs and saved commercial terms need the separate joint path where supported.

Worked example: a saved case-energy goal

Our North depot case has 2 kW solar, a 2 kW / 4 kWh battery, 8,760 kWh annual load, import value 0.25 AUD/kWh and export value 0.08 AUD/kWh. Storage charging value was blank. We minimised both selected participants' case import energy cost with 60% and 40% shares.

Saved physical dispatch objective evidence showed about 1,399.52 AUD as the signed contribution and allocated energy cost. The completed run showed 4,380 kWh generation, 3,161.93 kWh onsite use, 5,598.07 kWh import and 1,113.76 kWh export. NPV was A$3,899, IRR 46.12% and discounted payback 2.42 years. The engineering status remained review; no electrical network or POC was supplied.

The participant-goal run completed with a review gate.
The participant-goal run completed with a review gate.

Use joint Detailed optimisation

  1. Leave Storage charging value blank. Use exact tariff versions and save any commercial agreement version before pinning it.
  2. In Options, select Use joint optimisation in Detailed Analysis. Do not combine it with a forecast experiment.
  3. Save and run. In Results → Report, download the JSON manifest and inspect the joint-dispatch objective and diagnostics.
Enable joint optimisation in Options for a Detailed case.
Enable joint optimisation in Options for a Detailed case.

The joint model schedules multiple storage assets against shared import/export limits, efficiencies, SOC and calendar bounds, daily cycle budgets and a physical terminal SOC target. It uses known full-horizon inputs, not a forecast. Supported tariff rules, saved participant tariff shares and pinned commercial agreements can form one objective. Bounded demand formulas and decreasing demand blocks can shape the storage schedule. A selected participant objective can differ from whole-site economics.

Worked example: the same case with a physical terminal target

We cloned the participant-goal scenario and enabled joint dispatch. The saved objective artifact returned optimal, method joint_convex_lp (a linear optimisation with no yes/no decisions), objective about 1,399.85 AUD and the same tariff cost. There were no pinned contracts, so commercial settlement was 0. The results showed 4,380 kWh generation, 3,160.62 kWh onsite use, NPV A$3,301 and IRR 40.86%. This is not a like-for-like claim that one dispatch method always earns more: the joint model enforces a physical terminal inventory.

Worked example: rolling demand and decreasing blocks

We supplied four one-hour intervals on 1 January 2026 in UTC. The example load was 0, 3, 0 and 1 kW. A 1 kW / 2 kWh battery started and ended at 50% SOC. Both efficiencies were 100%. The import limit was 3 kW and the export limit was 0 kW.

With a two-interval rolling demand formula and an example rate of 20 AUD/kW, the saved objective was 20 AUD. Exact settlement was also 20 AUD. The saved method was joint_convex_lp and status was optimal.

We then used decreasing demand blocks: the first 0.5 kW at 8 AUD/kW, the next block up to 1.5 kW at 2 AUD/kW, and the remaining demand at 1 AUD/kW. The saved method was joint_bounded_milp, with 2 integer variables. This method is used when the problem needs yes/no (integer) decisions, such as demand blocks. The objective and exact settlement were both 6.5 AUD. Each example covers only its four supplied intervals.

In the JSON manifest, inspect joint-dispatch-objective. It records the method, status, tariff terms and exact-settlement reconciliation. The rolling and decreasing-block examples both returned a reconciliation difference of 0 AUD.

Run a NONPHYSICAL SOC reset experiment

This is not physically feasible dispatch. A free reset adds or removes stored inventory without charging or discharging. Do not use it for finance, carbon savings, ranking or network-feasibility claims. Use continuous SOC or a physical window-end target for physical studies.
  1. In Options, select Enable price experiment. Choose Bounded rolling look-ahead — beam-search heuristic.
  2. Supply valid load, generation and price forecasts. Set SOC boundary mode to Independent windows — NONPHYSICAL free reset.
  3. Set the independent window length and window-end SOC. Acknowledge the warning. Save and run.
  4. In Results, read the NONPHYSICAL banner, total adjustment and SOC boundary adjustments ledger. Check Energy and Economics separately.
This example resets simulated SOC to 50% every 24 intervals. Daily cycle use is not reset.
This example resets simulated SOC to 50% every 24 intervals. Daily cycle use is not reset.

Worked example: separate artificial support from physical energy

We reran the North depot reset scenario over 8,760 hourly intervals in 2026. The battery was 2 kW / 4 kWh. Look-ahead was 12 intervals, beam width 4, and load forecast used one prior observation, 0% signed error and a 0 kWh cold start. Generation used a separate assumed frozen 1 kW forecast issued on 31 December 2025. Actual solar generation remained the supplied resource profile. Annual load normalisation was off.

The product reported a net adjustment of 438 DC kWh. Each boundary added 1.2 DC kWh, taking simulated SOC from 0.8 to 2 kWh. Reset-supported load was about 171.45 AC kWh. Opening-inventory support was about 0.79 AC kWh. These are separate references, not ordinary unserved load or renewable supply.

Positive DC adjustments add inventory. Supported AC load is reported separately; the two are not interchangeable.
Positive DC adjustments add inventory. Supported AC load is reported separately; the two are not interchangeable.
Full-horizon resultProduct value
Generation4,380 kWh
Physical onsite useabout 2,068.33 kWh
Physical exportabout 2,311.67 kWh
Renewable utilisation100%

Onsite use plus export equals generation in this example. The energy preview reconciled all 2,000 returned intervals. That preview is capped; the summary covers all 8,760 simulated intervals. A full simulated horizon is not always a full calendar year.

Economics showed: “Finance excluded: Nonphysical independent-window SOC reset; artificial energy cannot support financial or carbon claims.” NPV and IRR were unavailable, not zero.

The result explains why cashflows and financial metrics are excluded.
The result explains why cashflows and financial metrics are excluded.
Source inventory uses proportional accounting, not physical tracing. Reset energy and unverified opening inventory are excluded from physical charge, discharge, onsite use, exports, losses and utilisation. No replacement grid imports are invented. Negative resets remove inventory by source. Grid-charged exports do not count as renewable utilisation. Older saved reset results show a rerun warning; their energy totals are not silently recalculated.

Plan fleet charging

Open Design and find Mobility planning. Enter the fleet in Fleet plan JSON. Give each vehicle its usable energy, opening energy, reserve and departure requirement. Add charger groups, dated connection sessions and site import and export limits.

In Preview profiles JSON, enter the interval timestamps, load, generation and import and export prices. Select Preview mobility dispatch. The Mobility preview report shows the modelled schedule, tariff cost and vehicle energy balance.

In this two-hour example, one vehicle starts empty and needs 4 kWh at departure. Its charger limit is 4 kW. Import prices are 1 then 2 AUD/kWh. The preview charges 4 kWh in the first hour and none in the second. The tariff cost is 4 AUD, closing energy is 4 kWh, and departure shortfall is 0 kWh.

The managed charging preview schedules energy in the lower-price hour.
The managed charging preview schedules energy in the lower-price hour.

Model bidirectional charging

Choose the bidirectional mode in the fleet plan. Supply vehicle consent and an export permission record with its dates, power limit and evidence reference. The schedule uses these caller-supplied assumptions together with the vehicle reserve and discharge efficiency.

In the second example, the vehicle starts and finishes at a 4 kWh reserve. Discharge efficiency is 80%, the export limit is 2 kW, and the second-hour export price is 3 AUD/kWh. The preview imports 2.5 kWh, exports 2 kWh and records 0.5 kWh of loss. Its modelled tariff cost is −3.5 AUD, with no departure shortfall.

The bidirectional preview retains the vehicle reserve while using the supplied export window.
The bidirectional preview retains the vehicle reserve while using the supplied export window.

Inspect Vehicle energy totals and Interval dispatch for the charge, discharge and state of charge. Use Apply mobility plan to draft to carry the plan into the case. Save the case and use Detailed Analysis with joint optimisation for the case run.

These examples use assumed prices and vehicle data, with caller-attested permission. Results are modelled schedules for the supplied intervals. Charging and exported energy are measured at the modelled charger boundary.

Apply connection access limits

Open Design and find Scheme sources and connection access. Under Explicit connection agreement, enter Connection preview JSON. Supply the agreement reference and digest, timezone, dates, import and export permission, rating and agreement caps. Add dated instructions or an access calendar for the study.

Select Preview access limits. The table shows import and export limits for each UTC interval, with the constraint that sets each limit. Use Apply access attachment to draft to keep the agreement inputs with the case.

This three-hour example has a 100 kW import rating, split into 40 kW firm and 60 kW flexible access, and a 90 kW agreement cap. Supplied instructions of 200, 20 and 60 kW produce modelled import limits of 90, 20 and 60 kW. The example has no export permission, so its export limit is 0 kW in all three intervals.

Each interval shows the limit set by the supplied agreement or instruction.
Each interval shows the limit set by the supplied agreement or instruction.
The example uses a caller-supplied agreement and instructions. Use your own dated agreement evidence for a project. The preview preserves the separate import and export permissions.

Related Guides

Can't find what you need? Email support@networkmap.energy or call 1300 936 116. Sign in to Network Map.