A process simulator that runs in your browser: steady-state, pressure-driven and dynamic flowsheets, rigorous thermodynamics, and an AI copilot that builds and explains them. There is nothing to install.

Does the AI calculate the results?

No. The copilot proposes a flowsheet or a change and explains it; the deterministic solver computes every temperature, duty and composition you see. A flowsheet from the copilot that does not validate is not applied.

Will it feel familiar if I have used Aspen Plus or HYSYS?

That is the intent. The workbench has a ribbon, an All Items tree, a form for each block and the three run modes — steady state, pressure driven and dynamics — so the habits carry over.

Can it open my Aspen or HYSYS cases?

Aspen Plus .bkp and .apw, HYSYS .hsc and DWSIM files import with deterministic parsing and a report of what was recovered. A HYSYS .hsc import is partial — unit names and types, not connectivity — and the report says so.

Which thermodynamic models are included?

Peng-Robinson and SRK (including Gᴱ mixing rules), NRTL, UNIQUAC, UNIFAC, PC-SAFT, CPA, GERG-2008, electrolyte NRTL, Pitzer and the IAPWS steam tables — more than 30 property packages in all.

Yes. The Python SDK (pip install flowsim-sdk) creates, runs and reads simulations, and its results are the same objects the canvas shows.

How much does it cost?

It is free to start, with no credit card. Paid plans add higher limits and team features; the pricing page has the details.

A process simulator that runs in your browser: steady-state, pressure-driven and dynamic flowsheets, rigorous thermodynamics, and an AI copilot that builds and explains them. There is nothing to install.

Does the AI calculate the results?

No. The copilot proposes a flowsheet or a change and explains it; the deterministic solver computes every temperature, duty and composition you see. A flowsheet from the copilot that does not validate is not applied.

Will it feel familiar if I have used Aspen Plus or HYSYS?

That is the intent. The workbench has a ribbon, an All Items tree, a form for each block and the three run modes — steady state, pressure driven and dynamics — so the habits carry over.

Can it open my Aspen or HYSYS cases?

Aspen Plus .bkp and .apw, HYSYS .hsc and DWSIM files import with deterministic parsing and a report of what was recovered. A HYSYS .hsc import is partial — unit names and types, not connectivity — and the report says so.

Which thermodynamic models are included?

Peng-Robinson and SRK (including Gᴱ mixing rules), NRTL, UNIQUAC, UNIFAC, PC-SAFT, CPA, GERG-2008, electrolyte NRTL, Pitzer and the IAPWS steam tables — more than 30 property packages in all.

Yes. The Python SDK (pip install flowsim-sdk) creates, runs and reads simulations, and its results are the same objects the canvas shows.

How much does it cost?

It is free to start, with no credit card. Paid plans add higher limits and team features; the pricing page has the details.

Process Economics — CAPEX & OPEX

Turton/Couper power-law purchased-equipment costs escalated by CEPCI and rolled up through bare-module → contingency/fee → working capital to total capital investment, composed with utility/feedstock OPEX and revenue into gross margin, simple payback, and ROI — screening estimates, not an AACE Class-3 quote.

economicscapexopexturton

Concept

Order-of-magnitude capital and operating cost from a solved simulation — deterministically, never a number the AI invents (principle #9: it may propose a material or price, the numbers always come from the correlations). CAPEX comes from Turton/Couper power-law correlations keyed off one size attribute per unit (duty for a pump/compressor/heater, rated area for a heat exchanger, a flow-residence volume proxy for vessels/columns), escalated by CEPCI and rolled through bare-module → contingency/fee → working capital. OPEX prices the same solved duties and feed/product streams the sustainability layer already classifies by utility type, so the two views of the plant always agree on which unit draws what.

The math

Show the governing equations

\log_{10}C_p = K_1 + K_2\log_{10}A + K_3(\log_{10}A)^2 \quad \text{(Turton heat-exchanger area cost)}

C_{installed} = C_p \cdot F_{BM} \cdot \frac{CEPCI_{now}}{CEPCI_{base}} \cdot F_{loc}\,F_{mat}\,F_{yr}

TCI = \underbrace{\sum C_{installed}}_{\text{bare module}}\,(1 + f_{cont} + f_{fee})\,(1 + f_{wc})

\text{ROI} = 100\cdot\frac{\text{revenue} - (\text{feed} + \text{utilities})}{TCI}, \quad \text{payback} = \frac{TCI}{\text{margin}}

Bare-module factors F_BM run 2.0-4.0 by unit type (columns/reactors highest); contingency 15%, contractor fee 3%, working capital 15% of the total module cost — a standard Turton screening breakdown, not vendor quotes.

Execution

Where it actually lives

Open Economics ▸ Capital & Operating Cost on a solved simulation — inputs are operating hours per year, electricity and process-heat prices, a location index, a material factor and a year factor. A named material such as ss316 (the Lang/Turton table lookup) and a CEPCI override are passed from Python.

# In the simulation's notebook — Simulator is pre-loaded and bound to it.
capex = Simulator.cost(material="ss316", location_index=1.2)
print(capex["capital_breakdown"]["total_capital_investment_usd"])

# CAPEX + OPEX + the KPIs, with the sheet's own overrides:
kpis = Simulator.economics_table(operating_hours=8400,
                                 prices={"electricity_usd_per_kwh": 0.09})
kpis.loc[["simple_payback_years", "roi_percent"]]   # payback is NaN when there is no margin