NPS Microgrid Planner is a peer-reviewed, open-source research platform with selectable energy-management methods and component disturbance-and-repair analysis; version 4.0 (September 2026) added a lifecycle cost analysis and time-of-use energy rates, and now requires a login. In our July 2026 guest test of version 3.1, MicrogridModeler reached a full-year result with fewer steps and exposed more planning evidence in one place: chronology, lifecycle cost, sizing alternatives and exports. The two tools overlap; which one fits depends on hosting, access and the analysis you need.
Key takeaways
- NPS Microgrid Planner deserves credit for publishing deployable research methods, including selectable EMS logic and component disturbance-and-repair resilience.
- The public NPS guest path required several catalog and analysis pages; MicrogridModeler produced a complete, editable study in one browser workspace with no account.
- NPS 4.0 (September 2026) added a cost analysis over one representative dispatch year and time-of-use energy rates; MicrogridModeler re-dispatches every project year and adds constraint-enforced sizing with a Pareto decision set.
- This is a workflow and capability comparison, not a claim that unlike input cases should produce matching fuel, renewable, or unmet-load totals.
- Neither tool replaces transient stability, protection coordination, detailed AC power flow, interconnection, or constructability engineering.
Comparison matrix
| Criterion | MicrogridModeler 3.0.0 | NPS Microgrid Planner (3.1 tested; 4.0 notes) | Live finding | Decision note |
|---|---|---|---|---|
| Public access | Complete local model, sizing, economics, resilience, and exports work without an account. | In the July 2026 test (3.1), a limited guest account was available. Version 4.0 removes the built-in guest identity and requires a login; it can also be self-hosted from the open repository. | The 3.1 guest account could simulate its stored example, but no sizing template or disturbance was available to complete those analyses. | MicrogridModeler is easier to evaluate and teach without provisioning users. |
| Path to first result | Open a computed project, choose a benchmark, and resolve a postal code in one guided workspace. | Build or select components, a microgrid, a powerload, an EMS, a location, and a date window across separate pages. | The equivalent NPS guest journey moved through Components, Microgrids, Powerloads, and Simulate before results. | NPS makes its entity model explicit; MicrogridModeler reduces navigation and repeated context switching. |
| Chronological data | Runs 8,760 hourly steps for every project year; imports timestamped CSV/TSV and NPS exports with calendar alignment and interval-energy preservation. | Stores timestamped powerloads and supports user-selected simulation start and end dates. | NPS had the stronger date-aware load concept before this audit; MicrogridModeler closed that gap and added transparent import diagnostics. | Both support chronological work; MicrogridModeler now provides the clearer import and provenance trail. |
| Dispatch / EMS | Four selectable policies: smart battery-first, cycle charging, SOC maintaining, and diesel first, with explicit energy-balance, SOC, fuel, starts, and curtailment evidence. | Selectable battery-favoring, diesel-favoring, and maximum-SOC methods, plus legacy diesel-before-battery logic in source. | The policy intents overlap; assumptions and generator efficiency formulations differ. | Compare policies on frozen inputs before treating output differences as engine error. |
| DER sizing | Exhaustive declared grid, four-seed joint refinement, dispatch-policy co-optimization, adverse scenarios, hard feasibility, and full Pareto-efficient alternatives. | Automated sizing since version 2.0 (April 2024). | The public guest sizing route had no selectable template in the observed session. | MicrogridModeler provides the more complete immediately testable sizing decision workflow. |
| Cost modeling | Lifecycle NPC, LCOE, capex, fuel, O&M, replacements, escalation, incentives, salvage, decommissioning, carbon, tariffs, cash flow, payback, IRR, and matched baseline. | Version 4.0 (September 2026) adds a cost analysis: one representative dispatch year projected over the analysis period, with capital, O&M, fuel, purchased energy, demand charges, replacements, export revenue and residual value, and a system LCOE. | The 3.1 guest simulation we ran reported physical outcomes only; the 4.0 cost method was not exercised in our test. | NPS 4.0 projects one dispatch year; MicrogridModeler re-dispatches every year with degradation and load growth. Compare on frozen inputs before choosing. |
| Resilience | Outage survival plus NPS-style fixed, local, and global component disturbance methods; deterministic or seeded stochastic failures, repair restoration, ENS, P95, and recovery curves. | Version 3.0 added a resilience framework, disturbance and repair data, and visualizations. | The guest account exposed zero disturbances, so the resilience run could not be completed side by side. | NPS supplied important methods; MicrogridModeler makes them immediately runnable and adds reproducibility evidence. |
| Results and exports | Exact selected-window and full dispatch CSV, lifecycle CSV, complete JSON package, print/PDF, content fingerprint, point inspection, and SVG/2× PNG charts. | Compact summary plus interactive power, energy-share, and battery charts; observed chart menus offered SVG, PNG, and CSV. | NPS chart exports are good. MicrogridModeler adds whole-run provenance and result reconciliation around them. | Prefer MicrogridModeler when another analyst must reproduce the complete decision, not only reuse a figure. |
| Declared boundary | Hourly, single-bus energy planning; no transient, protection, detailed feeder power flow, or subhour controller claim. | Early-stage microgrid planning suite; scope should not be confused with final electrical design. | Neither observed workflow establishes construction-ready network or protection results. | Escalate shortlisted designs to appropriate electrical and controls engineering. |
How we ran the live comparison
Update, September 25, 2026: NPS released Microgrid Planner 4.0 in September 2026, adding a cost analysis method, energy-rate handling, containerized deployment and a required login. The test below was run on version 3.1 and was not repeated on 4.0; rows that 4.0 changes are marked in the table above.
We used both public production applications as ordinary guest users on July 20, 2026. The task was practical: select a PV, battery, and diesel system; choose battery-first operation; attach a chronological load; place the project in Monterey, California; run a simulation; understand fuel, deficits, and state of charge; find a better size; inspect resilience; and export evidence.
On NPS Microgrid Planner, we used the public “no wind” microgrid, “Favor battery energy storage system usage,” “Powerload 1,” Monterey, and the available September 1–15, 2023 window. On MicrogridModeler, we selected the “Remote clinic (small)” benchmark, entered postal code 93940, accepted the PVWatts typical-year resource (modeled from NSRDB satellite data), and ran the full-year study before inspecting a 72-hour window.
Those are deliberately described as two user journeys, not one calibration dataset. Component ratings, load shapes, generator fuel models, reserves, converters, and study horizons were not identical. A valid numerical engine benchmark would freeze the same inputs and compare hourly balances under mapped policy assumptions.
Sources for this section: NPS Microgrid Planner live application · MicrogridModeler live application
What NPS Microgrid Planner genuinely does well
Microgrid Planner is not a toy competitor to dismiss. NPS describes it as a peer-reviewed, open-source suite for the early stages of microgrid planning. Its source separates components, grids, loads, simulations, sizing, disturbances, and repairs; its release history documents NSRDB and PySAM resource work, selectable energy-management methods, automated sizing, simulation caching, and a resilience framework.
Its most important contribution is methodological transfer. Research logic that might otherwise stay in a paper is available in a running web application and public repository. MicrogridModeler’s component disturbance-and-repair analysis and fixed, local, and global resilience scores explicitly credit that lineage in our public methodology.
The observed results page was also compact and useful. It surfaced renewable and diesel shares, unmet energy, diesel consumption, carbon dioxide, wet-stacking hours, unmet-power hours, power traces, energy proportions, and battery state of charge. Its chart controls for zooming, panning, and SVG, PNG, or CSV export set a practical usability bar.
Sources for this section: Microgrid Planner release notes and platform description · Microgrid Planner source repository · Peer-reviewed INFORMS paper
What users actually saw in the two runs
The NPS guest run completed a 14-day simulation and reported 24.8% photovoltaic contribution, 74.6% diesel contribution, −2.3% BESS net contribution, 2.8% unmet energy, 457.6 gallons of diesel, 10,272.3 pounds of carbon dioxide, 11 wet-stacking hours, and 23.5 unmet-power hours. The negative battery share is a net-energy reporting convention: storage absorbs energy before returning less after losses; it is not negative physical generation.
MicrogridModeler resolved postal code 93940 to Monterey, loaded PVWatts typical-year solar data, and produced the full chronological and lifecycle result in the same session. Its selected 72-hour window reconciled 686 kWh of demand: 91 kWh served directly by renewables, 364 kWh from battery discharge, and 231 kWh from diesel, with 100% served load, a 21% minimum SOC, 221 liters of fuel, 24 generator hours, and 12 starts.
The useful comparison is interpretation friction. NPS gave a concise physical result after the user assembled catalog entities and selected an analysis window. MicrogridModeler started from a computed project, let the user change location and load in place, retained the full-year context, and connected the same result to constraints, sizing alternatives, lifecycle economics, risk cases, and a downloadable evidence package.
| Observed item | NPS Microgrid Planner guest | MicrogridModeler guest |
|---|---|---|
| Study computed | 14-day stored example window | Full 8,760-hour year plus 72-hour inspection window |
| Location / resource | Country, region, and Monterey selections | Postal code 93940 resolved to Monterey; PVWatts typical-year data loaded |
| Result depth | Energy shares, fuel, CO₂, wet stacking, unmet load, power and SOC charts | Dispatch reconciliation, constraints, fuel and starts, economics, sizing, risk, resilience, provenance, and exports |
| Sizing / resilience as guest | No sizing template and zero disturbances in the observed account | Both runnable from the same project without registration |
Accuracy is an evidence chain, not a marketing adjective
A real accuracy claim begins with aligned inputs. The side-by-side test exposed one meaningful NPS advantage: its stored loads preserved timestamps and simulations accepted exact date windows. MicrogridModeler now imports timestamped CSV or TSV files and NPS exports, detects named value columns, time-weights variable intervals into energy-preserving hourly values, calendar-aligns partial profiles with the resource year, warns about gaps or assumptions, and preserves that custom load when site details change.
The engine then checks hourly energy balance and battery SOC under all four EMS policies, re-dispatches every project year, and exposes fuel, starts, wet stacking, curtailment, unmet load, converter treatment, reserves, and binding feasibility limits. Component disturbance runs are deterministic or seed-reproducible. The test harness checks energy balance and SOC bounds on all 20 synthetic reference cases, load interval preservation, calendar alignment and window-to-full-series reconciliation; it checks the NPS-derived controllers for invariants, not against Microgrid Planner outputs.
That is stronger evidence inside the supported boundary, not proof that any planner is universally more accurate. MicrogridModeler uses an explicit part-load generator fuel curve while the observed NPS example used a constant-efficiency generator. Without mapping that and every other assumption, a difference in gallons or unmet energy cannot diagnose which engine is right.
Sources for this section: MicrogridModeler implemented methodology and limits · Microgrid Planner source repository
Cost analysis: what NPS 4.0 added and how the approaches differ
In July 2026 the NPS homepage listed cost modeling and enhanced DER sizing as future plans. Version 4.0 (September 2026) shipped a cost analysis method: it runs one representative year of dispatch and projects it over the analysis period with a nominal discount rate and inflation, reporting net lifecycle cost and a system LCOE. Its documentation lists demand ratchets, tiered rates, fixed customer charges, minimum bills, taxes, degradation and financing fees as not modeled. NPS also states that the release meets the Unified Facilities Criteria 3-550-04 requirements for microgrid design modeling tools.
The cost engine computes technology and balance-of-system capital, development, contingency, operating and maintenance escalation, insurance and asset management, fuel, carbon, replacements with optional learning, incentives, salvage, decommissioning, tariffs, payback, internal rate of return, net present cost, levelized cost, and a dispatch-modeled diesel or grid baseline. Annual cash flows are downloadable, and each project year is re-dispatched instead of multiplying one annual result by project life.
Sizing begins with an exhaustive declared PV × battery-energy × genset grid, retains infeasible candidates and reasons, refines several promising basins jointly across PV, wind, battery energy, battery power, genset, and EMS policy, and can screen finalists under high-load and renewable-drought scenarios. It returns the full Pareto-efficient decision set across NPC, upfront cost, fuel, and renewable share, with explicit lowest-cost, lowest-capex, lowest-fuel, and highest-renewable anchors.
Sources for this section: Microgrid Planner release notes (version 4.0, September 2026) · MicrogridModeler economics and sizing methodology
How the MicrogridModeler workflow is organized for review
NPS’s multi-page catalog design is logical for a database application, but it asks a new user to understand components, microgrids, powerloads, analyses, locations, and permissions before seeing the whole decision. MicrogridModeler uses one authoritative project underneath Guided and Expert views. Guided mode exposes the decision-driving fields; Expert mode opens the complete component, reliability, tariff, resource, and economic assumptions without switching engines.
Results follow the same pattern. Every chart offers point inspection and an accessible text description; analyst-ready charts can be shared as SVG or 2× PNG; selected windows have exact KPIs and timestamped CSV; the entire run can be exported as JSON with inputs, provenance, outputs, and a content fingerprint. Sizing marks distinguish feasible, infeasible, Pareto-efficient, and recommended candidates, and any decision anchor can be applied with one click.
Ease of use also includes honest feedback. Import warnings explain resampling. Boundary flags identify a likely-too-small search domain. Source badges distinguish site datasets from synthetic resource inputs. P50/P90 results come from a stated uncertainty budget and are labelled as a planning screen, not lender-grade uncertainty analysis.
Where neither browser planner is the final engineering answer
MicrogridModeler’s local engine is an hourly, single-bus energy-planning model. It does not simulate electromagnetic transients, protection coordination, fault current, harmonics, detailed feeder voltage, unbalance, dynamic controls, construction sequencing, or interconnection approval. Subhour load readings are aggregated into energy-preserving hourly values; they do not become a subhour controller study.
NPS likewise describes Microgrid Planner as supporting the early stages of planning. The right handoff is to use these tools to screen sizes, economics, dispatch, and resilience; freeze the chosen assumptions and evidence; then perform the required power-flow, protection, controls, civil, environmental, procurement, cybersecurity, and commissioning work in the appropriate tools and disciplines.
Which to use
Choose NPS Microgrid Planner when you want to study or deploy its open research architecture, trace its published methods, or teach the relationship among stored components, loads, EMS choices, and resilience analyses. It is a credible primary source and its chart exports remain good.
Choose NPS Microgrid Planner 4.0 when you need a self-hostable, access-controlled tool with CAC support, or its UFC 3-550-04 statement matters to your program. Choose MicrogridModeler when a user must go from a location and load to a reviewable size, cost, dispatch and resilience result without an account, with every year re-dispatched and the whole run exportable. In our July test (on NPS 3.1) MicrogridModeler needed fewer steps to a full-year result; we have not re-run that test on 4.0.
The comparison also changed the product. NPS’s date-aware load workflow, chart interactions, and result compactness led directly to calendar-aligned imports, custom-load preservation, any-date KPI reconciliation, exact-window CSV, inspectable marks, complete control labels, and chart-grade SVG and PNG exports. That is the standard: when another tool does something better, adopt the lesson, test it, and make the full workflow better for the next user.
Keep exploring
Sources and review notes
This comparison is based on current public product and documentation pages. Verify licenses, modules, and pricing before making a procurement decision.
Product and company names on this site, such as HOMER, REopt, PVWatts, Xendee and Energy Toolbase, may be trademarks of their respective owners and are used only to identify those products and companies. MicrogridModeler is independent and is not affiliated with, sponsored by or endorsed by them, or by DOE or NLR (formerly NREL), unless a page says otherwise.
FAQ
Is NPS Microgrid Planner open source?
Yes. NPS publishes the source on GitHub under an MIT-style license that excludes its CodeShack-derived authentication code. Version 4.0 (September 2026) requires a login on the hosted application and can be self-hosted with the provided container setup.
Does MicrogridModeler implement NPS Microgrid Planner methods?
MicrogridModeler implements and credits the NPS selectable EMS concepts and the version 3 component disturbance-and-repair framework, including fixed, local, and global resilience scoring. It extends them with deterministic seeding, fleet-unit outcomes, recovery curves, lifecycle economics, robust sizing, and complete run provenance.
Which tool is easier to use without an account?
MicrogridModeler runs without registration. In the July 20, 2026 test of NPS 3.1, the guest account could run a stored simulation but did not expose a usable sizing template or disturbance case; version 4.0 removed the guest identity and requires a login.
Can the two observed result sets be compared for numerical accuracy?
No. The observed projects had different loads, component assumptions, fuel models, reserves, converters, and time windows. They compare user experience and reporting. A numerical calibration test must freeze identical inputs, map EMS behavior, and reconcile every hourly energy balance.
Does MicrogridModeler replace detailed electrical engineering?
No. It is an hourly, single-bus planning model. Shortlisted designs still require the appropriate power-flow, protection, transient, controls, interconnection, constructability, procurement, and commissioning studies.
Run the same decision journey yourself
Open a computed benchmark, change its location or load, inspect the exact hours, and export the complete evidence package.