Documents — what each file contains

In simple terms

Everything the app gives you as a file is one of a few documents. Each one is made from one stage of one project — prefeasibility or feasibility — and says so inside. You get them from the Export ▾ button at the top of every page of a stage; a few also sit beside the tool that makes them. Two kinds can come back in: an assumption book (your edited inputs) and a project bundle (a whole project). What comes back in arrives as a new version or a new project, so nothing you had is overwritten.

In the app, the foot of every Export ▾ menu links here (“What’s in each file”).

What you can export, by stage

What Export ▾ offers on each stage, generated from the app’s own document catalogue (backend/core/documents/catalog.py) — the list the menu itself reads. In the menu, a document the project cannot produce yet is greyed out, with the reason under it.

Prefeasibility

Document	What is in it	Formats
Assumption book	Every parameter this stage reads, with its value, unit, allowed range and source. For checking or editing the inputs in Excel; it imports back.	Excel
Design	What the plant IS at this stage: capacities, storage, technology, how it runs, the grid set-up, the site and datasets, each product with its promise. One sheet, read by the Excel package; the app cannot import it back yet.	Excel
Timeseries	The hourly wind and solar capacity factors this stage runs on, one year, with the site’s name, year and source. The same workbook as the site library’s template: change the values and import it as a new site. Withheld when the data is confidential.	Excel
Results	The outputs of this stage’s design: cost per tonne and per kg, the cost build-up, the energy balance, production. For an analyst redoing the sums.	Excel
Summary report	The resource, the design, the results and the cost in a few pages, with the standing best-estimate note. PDF to circulate; Word to mark up; PowerPoint to present. Same numbers in all three.	PDF · Word · PowerPoint
Project bundle	The whole project, portable: this stage’s assumption book, results workbook and report to read, and project.json with every stage’s design, assumptions, versions, snapshots and saved runs. Import project on the portfolio page restores it on another machine; measured series are never inside, so that machine needs the same datasets. For a backup or a handover.	Zip

Feasibility

Document	What is in it	Formats
Assumption book	Every parameter this stage reads, with its value, unit, allowed range and source. For checking or editing the inputs in Excel; it imports back.	Excel
Design	What the plant IS at this stage: capacities, storage, technology, how it runs, the grid set-up, the site and datasets, each product with its promise. One sheet, read by the Excel package; the app cannot import it back yet.	Excel
Timeseries	The hourly wind and solar capacity factors this stage runs on, one year, with the site’s name, year and source. The same workbook as the site library’s template: change the values and import it as a new site. Withheld when the data is confidential.	Excel
Results	The outputs of this stage’s design: cost per tonne and per kg, the cost build-up, the energy balance, production, and the hourly-dispatch sheet when a feasibility solve exists. For an analyst redoing the sums.	Excel
Summary report	The resource, the design, the results and the cost in a few pages, with the standing best-estimate note. PDF to circulate; Word to mark up; PowerPoint to present. Same numbers in all three.	PDF · Word · PowerPoint
Project bundle	The whole project, portable: this stage’s assumption book, results workbook and report to read, and project.json with every stage’s design, assumptions, versions, snapshots and saved runs, plus the hourly frame when one exists. Import project on the portfolio page restores it on another machine; measured series are never inside, so that machine needs the same datasets. For a backup or a handover.	Zip
Hourly data	The 8,760-hour frame of the last feasibility solve — every series the charts are drawn from. Parquet as stored; CSV with the provenance as header lines; Zip adds the solve’s annual totals and engine identity.	Parquet · CSV · Zip
Hourly data · lifetime rollup	A multi-year solve’s summary across its years: each year’s totals, the aging multipliers and the lifetime cost. The hours of one year are the entry above with a year chosen.	JSON
Study · report	The design search as a document: the recommended design, what else costs the same, what reliability costs, and the basis it was solved under. For an internal decision.	PDF · Word · PowerPoint
Study · search run log	Every design the search simulated, one row each: sizes, output, cost, load factors, storage use. The full table the page only shows the top of.	CSV
Study · every combination	Every combination solved, every indicator and every check, one row per design. The full table behind the winner.	Excel · CSV
Excel package	This stage as a standalone Excel edition: a cockpit workbook, the assumption book and design to edit, the resource series, and the app’s own engine (with Python for Windows) to run one year, one component, the plant’s scale and every combination rule-based on your own machine, with a check against the app. A large zip: the engine travels inside it.	Zip

Who each file is for

Document	Who reads it	What they do with it
Summary report	a reader, a lender, a partner	reads it: the PDF to circulate, Word to mark up, PowerPoint to present — the same numbers in all three
Results	an analyst	redoes the sums: every figure the stage’s pages show, unrounded, one sheet per section
Assumption book	whoever owns the inputs	checks every value with its unit, range and source; edits it in Excel and imports it back
Design	an engineer; the Excel package	reads what the plant is — sizes, technology, how it runs, the site; the Excel package runs it
Study	the project team	chooses between designs: the search, what else costs the same, what reliability costs
Hourly data	an engineer or an analyst	works on the hours themselves, outside the app
Project bundle	you, or a colleague	keeps a backup, or hands the whole project over; Import project restores it
Excel package	a colleague without the app	runs the feasibility rules in Excel on their own computer, with a check against the app

What is inside

The Results workbook and the Summary report

Both are built from one content model, in the same reading order, so a figure is in both or in neither. The workbook has one sheet per section after its Cover; the report has the same sections:

Sheet (Results workbook)	Section (Summary report)
Resource	The resource — what the site gives you
Design	The design — inputs, selectors, site, versions
Production	Production and the binding constraint
Energy balance	Energy balance and curtailment
Water	Water
Feasibility	Hourly dispatch (feasibility)
Reliability	Reliability — when the plant runs short
Storage	Storage — behaviour, value, hydrogen failures
Grid	Grid and back-up power
Emissions	Emissions — the RFNBO screen per product
Costs	Cost — build-up, CAPEX and levelised cost
Lifetime	Plant lifetime — year by year
Flows	Mass flows — annual totals per stream
Thermal	Process heat (recorded, not supplied)
Provenance	Provenance — what produced these numbers

A section the project has no answer for yet (no feasibility solve, no grid run) is not left out: it holds one line saying why. Both documents carry totals and summaries only — annual and monthly figures, duration curves, shortfall episodes — never the hours themselves; the hours are the Hourly data document. When the page you export from shows one of the charts made for reports, the report carries that chart too.

The Assumption book

The standard workbook of the app: a Cover (the stamp, below), a hidden locked sheet that checks the structure, and one sheet per area, each row Parameter | Key | Value | Unit | Allowed range | Source | Notes. Only Value and Notes are meant to be edited; Excel refuses a value outside the allowed range. The sheets, and how many values each holds:

Domain (Excel sheet)	Parameters
Generation	34
Hydrogen	38
Water	18
Derivatives	116
CO2	14
Infrastructure	18
Financials	6
Total	244

The Design

One sheet, Design, in the same standard: the capacities, the storage, the electrolyser technology, how the stage runs (engine, battery strategy, H₂ buffer policy, promised output, horizon), the grid set-up, the site and its datasets, and each product with its promise (the demand_t_yr the Reliability tab saves).

The Project bundle

A zip you can open. It holds:

• assumption_book.xlsx
• results.xlsx
• report.html
• project.json

with manifest.json (what is inside, with the stamp) and, when the feasibility stage has a stored solve, dispatch.parquet — the hourly frame — unless the project runs on confidential data (below). The three readable files describe the stage you exported from; project.json holds every stage: designs, assumption books, both version histories, snapshots, saved runs. A bundle never holds a measured resource series: it names the datasets, and the computer that imports it must have them already.

A snapshot’s bundle (the Snapshots tab) is the same zip, made again from that snapshot.

Hourly data

The year of the last feasibility solve, hour by hour — every series the charts are drawn from. Parquet as it is stored; CSV with the stamp as its first lines; Zip adds the solve’s annual totals and the engine that made it. After a multi-year solve you choose the plant year in the menu, and the summary across the years is its own entry (JSON).

The Excel package

A zip that is a standalone Excel edition of the feasibility stage:

Inside the folder PtX-<project-slug>-excel:

• READ ME FIRST.txt
• PtX Engine.xlsm
• Inputs/project-bundle.zip
• Inputs/Assumption book.xlsx
• Inputs/Design.xlsx
• Resources
• Resources/datasets.json
• Golden/golden.json
• engine
• engine/python/python.exe
• engine/ptx_engine.py
• engine/app
• engine/mac/ptx-run.scpt
• Runs

The engine inside is the app’s own code, copied when you export, with a Python for Windows, so it gives the app’s answers; Check my installation in the cockpit workbook proves it on your computer against the project’s current solve. It runs the rule-based engine only.

The stamp — every file says where it came from

Every document carries the same record: on the Cover sheet of a workbook, in manifest.json of a zip, in the subtitle of a report, and as # lines at the top of a CSV. A fact the app does not know is written ”—”, never guessed.

Stamp field	What it tells you
kind	which document this is
project_id, project_name	the project
stage	the stage it was made from
app_version	the version of the app that wrote it
dab_version	the default assumption book this stage was seeded from, or last pulled
book_version, book_edited	the assumption-book version, and whether it had unsaved edits
sizing_version, sizing_edited	the design version, and whether it had unsaved edits
site, data	the site and its resource datasets
engine	the engine behind the numbers, when a solve is inside
commitment	the promised output (firm, or a yearly tonnage), on a feasibility document
snapshot_id, snapshot_name	the snapshot it was made from, if any
generated_at	when it was made (UTC)

Every file is named the same way, <project>_<kind>_<stage>_<date>.<ext> — for example atlas_summary-report_feasibility_2026-11-01.pdf — so a folder of downloads reads by project.

Confidential data — what is held back, and why

A resource dataset can be marked confidential: measured data you may not pass on. A project that uses one, at any stage, keeps every total and summary in its documents (annual and monthly figures, duration curves, the costs) but never lets the hours leave:

• Hourly data is not offered; the menu says why.
• The project bundle, and a snapshot’s bundle, leave the hourly frame out and say so.
• The Results workbook and the Summary report state that the frame was held back.
• The Excel package carries no confidential series; its READ ME asks you to supply it.

Why the hours: the hourly frame holds the wind and solar power available each hour, and the installed capacity is in the same file — one division gives back the confidential series exactly. A total or an average cannot be turned back into it.

What comes back in

What	Where	What happens
Assumption book (Excel)	the stage’s Assumptions tab → Import Assumption book	a strict check of the structure, units and ranges; a preview of every change, the notes and the file’s Cover, with a warning when the file came from another project or stage; then Import as a new version: your unsaved work is saved first, the import lands as a new version of the book (changed values marked “user”), and switching back to the earlier version undoes it
Project bundle (zip)	the projects page → Import project	a preview (stages, versions, snapshots, which datasets this computer has); then a new project, under the bundle’s own id when that id is free; datasets are linked again only when this computer has them, and what could not be restored is listed
Default assumption book (Excel)	App settings → Default assumption book → Import	the same check and preview; the values go into the version you selected; existing projects do not change (their books were frozen when they were created)

The other documents are results: the app does not read them back. The Design workbook is read by the Excel package; the app has no import for it yet.

Where each one is

• Export ▾, top right of every page of both stages: every document of that stage.
• Beside the tool that makes it: the Study menu in Design lab › Find a design (the report and the search run log); Export the study in Design lab › Try every combination; Hourly data in How it works › What this run consumed.
• The Assumptions tab: export and import the stage’s assumption book.
• The Snapshots tab: each snapshot’s bundle.
• The project’s ⋯ menu in the sidebar: Export bundle (.zip).
• The projects page: Import project.
• App settings → Default assumption book: export and import a default book.

Related topics

• Assumptions
• Snapshots
• Settings — the default assumption book
• How feasibility works