Ask a BESS engineer how long a first credible design takes and the honest answer is "it depends, but weeks." Ask a software vendor and the answer is a percentage with no methodology behind it. This page tries to do better. We itemize the standard spreadsheet and CAD workflow step by step, with stated assumptions about the hours each step consumes, and then argue a hypothesis about where the time actually goes and why a single-model tool collapses it. Every assumption is listed so you can substitute your own numbers. When we have client project data, we will publish a measured case study against this baseline; until then, this is an argued estimate and is labeled as one.
By initial design we mean the deliverable that lets a project advance to procurement and detailed engineering: a container count sized to a usable-energy target with the full derate stack applied, an inverter or PCS pairing, a 20-year augmentation plan, and a site layout on the real parcel showing that the equipment, roads, and setbacks actually fit. Not a napkin estimate, and not detailed engineering. Think of an ordinary standalone utility-scale system in 2026, on the order of 100 MW and a few hundred MWh.
These hour figures assume an experienced engineer with existing templates, not a first-timer building from scratch. They cover working time; calendar time stretches longer because the steps interleave with everything else the engineer owns.
| Step | What it involves | Working time |
|---|---|---|
| Requirements and equipment selection | Fix the POI target, duration, and delivery year; collect OEM datasheets for candidate batteries and inverters; extract usable energy, SOC window, RTE, auxiliary specs, and degradation tables into a comparable form. | 4 to 8 hours |
| Loss and derate stack | Build or adapt the sizing spreadsheet: DoD and SOC window, RTE across battery and conversion stages, auxiliary and parasitic load, temperature derate at the site's conditions, day-one degradation margin. Validate the chain against a known design. | 8 to 16 hours |
| Sizing iterations | Iterate candidate architectures (AC block vs DC block, container models, PCS ratios) to a container count that nets the usable target. Each architecture swap touches most formulas. | 8 to 16 hours |
| Augmentation model | Extend the spreadsheet across 20 years with the OEM fade curve; derive the year-by-year additions that hold usable energy flat; sanity-check against warranty terms. | 8 to 16 hours |
| Site layout in CAD | Import the parcel DXF; draw battery and PCS blocks to spacing rules; place rows and clusters; draw access roads with turning radii; route MV feeders; respect setbacks and exclusion zones; count what fits and reconcile with the sized count. | 16 to 40 hours |
| The reconciliation loop | The count that fits the parcel rarely equals the count the spreadsheet sized. Adjust one side, redo the other, repeat. Two or three loops is normal for a first design. | 8 to 24 hours |
Total: roughly 52 to 120 working hours, call it two to four working weeks of an engineer's time for the first credible design. The wide range is the point; the workflow's cost is dominated by the reconciliation loop, and how many loops you need is unknowable in advance. Every subsequent change, a new battery model, a tightened setback, a revised target, re-enters the loop and costs hours to days.
The hours above are not caused by hard math. Each step is arithmetic an engineer could do on paper. Our hypothesis is that the time goes to three structural problems. First, state lives in two tools that do not share it, so the spreadsheet's container count and the CAD drawing's container count drift apart and must be reconciled by hand. Second, the derate stack compounds, so a change to any early assumption invalidates everything downstream, and spreadsheets do not track which cells are downstream. Third, layout constraints are geometric and sizing constraints are algebraic, and neither tool can see the other's constraints, which is why the reconciliation loop exists at all.
If the hypothesis is right, a tool that holds both kinds of constraint in one model does not just do the same steps faster; it deletes the loop entirely. The sized count and the placed count are the same number by construction, so there is nothing to reconcile. That is the design thesis behind FluxPilot: sizing, the derate stack, the 20-year augmentation schedule, and the DXF site layout live in one project file, a changed input re-runs the whole chain in one pass, and a full design iteration runs in minutes rather than weeks. The tenth iteration costs the same minutes as the first, which is where the compounding advantage sits, because early-stage projects are nothing but iterations.
Two things. If most of the manual hours turned out to be in gathering inputs rather than in the loop, a single-model tool would not help much; but input gathering (parcel DXF, OEM datasheets, an interconnection target) is common to both workflows and excluded from the table above. Or if reconciliation were rare, the loop would not dominate; every storage engineer we have talked to reports the opposite. We hold the hypothesis with confidence but state it as one: the honest test is measured client project data against this baseline, and that case study, with real numbers, is what we will publish once our first client projects have been through the tool.
For the engineering reasoning inside the sizing pass, see how to size a utility-scale BESS. For the derate stack, see BESS sizing software. For the layout stage, see BESS site layout software. For the 20-year plan, see BESS augmentation explained. To compare tools in this category, see BESS design software compared. To time your own project, book a demo.