🧪 GreenAmmonia-Opt — Green Ammonia Process Design & Techno-Economic Optimization

Engineering screening + decision-support dashboard for a green ammonia process using PEM electrolysis, nitrogen separation and Haber–Bosch synthesis.

👉 Launch Interactive Dashboard

GreenAmmonia-Opt is a reproducible chemical-engineering screening model for a 20,000 t/y green ammonia plant. It combines material balances, electrolyzer energy demand, nitrogen recovery, Haber–Bosch synthesis/recycle, compression, heat allowance, techno-economics, carbon intensity, deterministic optimization, sensitivity analysis and an interactive Streamlit dashboard.

The repository is deliberately transparent: assumptions are explicit, calculations are implemented in Python, outputs are reproducible, and an Aspen Plus validation workflow is included without fabricating Aspen results.

Scope: This is an engineering screening model, not a replacement for a rigorous Aspen Plus flowsheet, detailed equipment design, vendor quotations, HAZOP/SIL study, or plant commissioning analysis.

• 🎯 Product Overview
• 🔍 Problem Overview
• 💡 Project Objective
• 🏭 Process Overview
• 🧩 Process Architecture
• ⚙️ Key Features
• 📐 Baseline Design Basis
• 📊 Core Results
• 💰 Techno-Economic Analysis
• 🌱 Environmental Analysis
• 🔎 Sensitivity & Optimization
• 📈 Dashboard & Deployment
• 🧪 Aspen Plus Validation
• 🧪 Testing & Reproducibility
• 📁 Repository Structure
• 🚀 Quick Start
• ☁️ Deployment
• ⚠️ Engineering Assumptions & Limitations
• 🗣️ Interview Talking Points
• 📚 References
• 👤 Author

GreenAmmonia-Opt is an early-stage process design and decision-support tool that answers:

"Given a green-ammonia production target, what process assumptions drive energy consumption, cost and carbon intensity, and which screening-level design point looks economically attractive?"

The tool connects chemical-engineering calculations to practical engineering decisions:

Production target → Material balance → Energy model → Economics → Carbon intensity → Optimization → Validation path

It also provides an interactive dashboard so a reviewer can change key assumptions without editing Python source code.

• Chemical/process engineering students
• Process design and simulation learners
• Techno-economic analysis teams
• Energy-transition researchers
• Recruiters evaluating engineering + analytical problem solving
• Engineers preparing a first-pass Aspen Plus validation

Green ammonia can eliminate fossil-derived hydrogen, but its economics and energy footprint are strongly influenced by:

1. Hydrogen electricity consumption from water electrolysis.
2. Nitrogen recovery and separation energy.
3. Haber–Bosch single-pass conversion, which determines recycle requirements.
4. High-pressure compression requirements.
5. Electricity price, which can dominate operating cost.
6. Electricity carbon intensity, which changes the climate benefit of the product.
7. CAPEX and utilization, which affect levelized production cost.
8. Uncertainty in screening assumptions, which makes sensitivity analysis essential.

A useful early-stage model therefore needs to do more than calculate one material balance. It should expose the assumptions, quantify trade-offs and identify where rigorous simulation or experimental validation is required.

The project was designed around five engineering questions:

The conceptual process integrates:

1. PEM water electrolysis — produces H₂.
2. Nitrogen separation — supplies N₂ with a specified recovery.
3. Gas compression — brings synthesis gases to the selected pressure.
4. Haber–Bosch synthesis — converts H₂ and N₂ to NH₃.
5. NH₃ cooling/condensation — recovers product.
6. Recycle + purge — returns unreacted H₂/N₂ while preventing indefinite accumulation.
7. Utilities and economics — converts process performance into energy, cost and emissions metrics.

[ N_2 + 3H_2 \rightleftharpoons 2NH_3 ]

flowchart LR
    A[Water] --> B[PEM Electrolysis]
    B --> C[H₂ Compression]
    D[Air / N₂ Source] --> E[N₂ Separation]
    E --> F[N₂ Compression]
    C --> G[Mixer]
    F --> G
    G --> H[Haber–Bosch Synthesis]
    H --> I[NH₃ Cooling & Condensation]
    I --> J[NH₃ Product]
    I --> K[Recycle + Purge]
    K --> G
    H --> L[Heat / Utility Accounting]
    B --> M[Energy Model]
    L --> M
    M --> N[TEA + CO₂ + Optimization]

                    ┌────────────────────┐
                    │     Parameters     │
                    └─────────┬──────────┘
                              ↓
┌──────────────┐     ┌────────────────────┐
│ Material     │ ──→ │ Energy & Utilities │
│ Balances     │     └─────────┬──────────┘
└──────────────┘               ↓
                       ┌──────────────────┐
                       │ Economics + LCOA │
                       └────────┬─────────┘
                                ↓
                       ┌──────────────────┐
                       │ CO₂ Intensity    │
                       └────────┬─────────┘
                                ↓
                       ┌──────────────────┐
                       │ Optimization     │
                       └────────┬─────────┘
                                ↓
                   ┌─────────────────────────┐
                   │ CSV Results + Figures   │
                   │ + Streamlit Dashboard   │
                   └─────────────────────────┘

• Stoichiometric H₂/N₂/NH₃ material balances using the 3:1 H₂:N₂ reaction ratio.
• Explicit product-recovery loss accounting.
• Explicit synthesis conversion and recycle/purge accounting.
• PEM electrolyzer electricity model using kWh/kg H₂.
• Nitrogen recovery and gross N₂ feed calculation.
• Compression electricity screening.
• Synthesis heat allowance.
• CAPEX/OPEX screening.
• Capital-recovery-factor annualization.
• LCOA calculation.
• Renewable/grid electricity carbon-intensity scenarios.
• Deterministic grid optimization across five engineering/economic variables.
• Electricity-price sensitivity analysis.
• Reproducible CSV outputs and figures.
• Interactive Streamlit dashboard.
• Automated Python tests.
• GitHub Actions CI workflow.
• Aspen Plus build-and-validation workflow.
• Engineering assumptions and interview guide.

Running:

python -m src.run_all

produces the current deterministic baseline.

The fresh H₂ flow is lower than the reactor-loop H₂ flow because the synthesis model explicitly represents single-pass conversion plus recycle.

The baseline energy model is dominated by hydrogen production:

• Electrolyzer: 25,312.26 kWh/h
• Compression: 893.85 kWh/h
• Synthesis heat allowance: 875.00 kWh/h

This is an important engineering conclusion: electricity consumption for H₂ production is the main screening-level energy driver.

The screening LCOA includes:

[ LCOA = \frac{Annual\ Electricity + Fixed\ OPEX + Variable\ OPEX + Annualized\ CAPEX} {Annual\ NH_3\ Production} ]

These are screening assumptions, not vendor quotations or bankable project economics.

The model calculates electricity-related cradle-to-gate CO₂ intensity:

[ CO_2\ intensity = \frac{Electricity\ consumption \times Carbon\ factor} {NH_3\ production} ]

The baseline renewable scenario uses 0.05 kg CO₂e/kWh.

The model also supports a grid scenario using 0.45 kg CO₂e/kWh.

Included:

• Electricity-related operational emissions.

Excluded:

• Embodied equipment emissions.
• Upstream material emissions.
• Construction emissions.
• Transportation.
• End-of-life emissions.
• Full life-cycle assessment.

Therefore, the environmental result should be interpreted as an electricity-related screening metric, not a complete LCA.

The current sensitivity run evaluates:

$20, $40, $60, $80, $100 and $120/MWh

At the baseline configuration:

The optimizer evaluates:

• Electrolyzer consumption: 48–56 kWh/kg H₂
• Synthesis pressure: 100–200 bar
• Single-pass conversion: 12–28%
• N₂ recovery: 88–96%
• Electricity price: $30–90/MWh

The current grid search contains 1,875 design points.

The lowest-LCOA point in the predefined screening grid is:

Important: this is a mathematical result inside the predefined screening grid, not an optimized industrial design. In particular, the current pressure variable is retained as a design-space variable for screening, but the present economics model does not assign a pressure-dependent compressor penalty or reactor capital penalty. This is a known model limitation and a priority for a future rigorous model.

The repository includes an interactive Streamlit dashboard in app.py. The dashboard is designed for scenario exploration without editing the engineering model source code.

The sidebar exposes:

• NH₃ production target
• Operating hours
• Electrolyzer electricity consumption
• N₂ recovery
• Single-pass conversion
• Synthesis pressure
• Electricity carbon factor
• Electricity price

The sidebar also provides a dedicated Display & units section:

• Display currency: USD, EUR, GBP, INR or JPY
• FX rate: selected-currency units per USD
• Energy display: kWh/h, MWh/h or GWh/y
• Mass-flow display: kg/h, t/h, kg/d or t/d

All engineering calculations remain in their native internal units. Currency and unit conversion is applied only to dashboard presentation. Electricity price is an explicit engineering-model input in USD/MWh; the selected display currency is shown as a converted reference value. This prevents display-unit changes from silently changing the engineering equations.

FX note: the dashboard uses an explicit, editable FX assumption rather than a live exchange-rate API. This keeps the deployment deterministic and usable without external API credentials. Update the FX rate when a different market assumption is required.

The dashboard displays:

• Production rate with the selected mass-flow unit
• Specific energy
• LCOA in the selected currency per tonne NH₃
• CO₂ intensity
• Material and utility balance with selected flow/energy units
• Energy breakdown with the selected energy unit and percentage share
• Economics table with dynamic currency headers such as INR/year and INR/t NH₃, with annual costs and per-tonne costs populated consistently; LCOA is shown only as a per-tonne metric
• Optimization result with converted economic values
• Best-design CSV download

streamlit run app.py

The dashboard should be started from the repository root.

See the complete guide:

docs/deployment.md

The recommended deployment target is Streamlit Community Cloud because the repository contains a root-level app.py, requirements.txt and .streamlit/config.toml.

The dashboard deliberately separates engineering units from display units. The model calculates internally using USD, kg, kWh and the existing engineering basis. The UI converts values only after the calculations are complete.

If the model calculates:

LCOA = $1,025.11/t NH₃

and the dashboard is set to INR with:

1 USD = 83 INR

the displayed value becomes approximately:

₹85,084.24/t NH₃

Likewise, an energy rate of 27,081.10 kWh/h becomes approximately 27.081 MWh/h or 216.649 GWh/y at 8,000 operating hours. The underlying model value does not change.

This design fixes a common dashboard error where the number changes but the table heading or chart axis still shows the old unit. The selected currency/unit is now used consistently in metric cards, table headers, charts, material balances and optimization downloads.

The repository does not fabricate Aspen Plus results.

See:

aspen/Aspen_Plus_Validation_Workflow.md

H₂ Feed ─┐
         ├→ MIX → COMP → HB REACTOR → COOLER → FLASH → NH₃ PRODUCT
N₂ Feed ─┘                         ↑               │
                                  └── RECYCLE ────┘
                                      │
                                    PURGE

Compare Python and Aspen results for:

• NH₃ product kg/h
• H₂ feed kg/h
• N₂ feed kg/h
• NH₃ synthesis rate
• H₂/N₂ recycle
• Compressor duty
• Reactor duty
• Condenser duty
• Product purity
• Overall mass balance

Differences are expected because Aspen uses rigorous thermodynamics and the Python repository uses transparent screening correlations.

Run:

python -m pytest -q

The current suite contains 20 automated tests covering:

• Production-rate scaling
• Stoichiometric relationships
• Nitrogen recovery
• Water-feed positivity
• Energy positivity
• Electrolyzer energy dominance
• LCOA positivity
• Renewable vs. grid emissions
• Conversion bounds
• Model validation
• Optimization result structure
• Optimization minimum selection
• Deterministic optimization
• Production scaling

Run:

python -m src.run_all

This regenerates:

• results/baseline_results.csv
• results/optimization_results.csv
• results/best_design.csv
• results/electricity_sensitivity.csv
• results/energy_breakdown.csv
• figures/lcoa_electricity_sensitivity.png
• figures/energy_breakdown.png

GitHub Actions runs:

1. Dependency installation.
2. Automated tests.
3. Full reproducible model execution.

Workflow:

.github/workflows/ci.yml

green-ammonia-process-design/
│
├── app.py                              # Streamlit decision-support dashboard
├── requirements.txt                    # Python dependencies
├── .gitignore
├── .streamlit/
│   └── config.toml                     # Streamlit configuration
│
├── src/
│   ├── __init__.py
│   ├── model.py                        # Material, energy, economics, emissions
│   ├── optimize.py                     # Deterministic grid optimization
│   ├── report.py                       # CSV + figure generation
│   └── run_all.py                      # Reproducible project entry point
│
├── tests/
│   └── test_model.py                   # Automated engineering tests
│
├── results/
│   ├── baseline_results.csv
│   ├── optimization_results.csv
│   ├── best_design.csv
│   ├── electricity_sensitivity.csv
│   └── energy_breakdown.csv
│
├── figures/
│   ├── lcoa_electricity_sensitivity.png
│   └── energy_breakdown.png
│
├── docs/
│   ├── engineering_basis.md
│   ├── deployment.md
│   ├── interview_guide.md
│   └── references.md
│
├── aspen/
│   └── Aspen_Plus_Validation_Workflow.md
│
└── matlab/
    └── sensitivity_notes.m

Windows:

python -m venv .venv
.venv\Scripts\activate

Linux/macOS:

python -m venv .venv
source .venv/bin/activate

pip install -r requirements.txt

python -m pytest -q

python -m src.run_all

streamlit run app.py

1. Push the complete repository to GitHub.
2. Sign in to Streamlit Community Cloud with GitHub.
3. Create a new app.
4. Select the repository.
5. Select the main branch.
6. Set the entrypoint to:

app.py

7. Deploy.
8. Test the dashboard using the verification checklist in docs/deployment.md.

The root-level requirements.txt declares the Python dependencies required by the application, and .streamlit/config.toml contains the repository configuration.

If this project is already uploaded to GitHub, do not recreate the repository. Replace/add these files:

Replace

README.md
requirements.txt
.gitignore

Add

app.py
.streamlit/config.toml
docs/deployment.md
.github/workflows/ci.yml

Keep the existing src/, tests/, results/, figures/, docs/, aspen/ and matlab/ content unless you specifically want to replace it with this final package.

After pushing:

git add .
git commit -m "Add tested Streamlit dashboard and deployment workflow"
git push

This project is intentionally a transparent screening model.

• The synthesis loop uses a simplified single-pass conversion representation.
• A fixed purge fraction is used.
• Detailed inert accumulation is not modeled.
• Detailed gas-liquid equilibrium is not modeled.

• Electrolyzer demand is represented by a fixed kWh/kg H₂ assumption.
• Compression is a screening correlation rather than a compressor train design.
• Synthesis heat is represented as an allowance.
• Pressure-dependent compressor work is not yet rigorously coupled to the optimization objective.

• CAPEX is a screening assumption.
• OPEX is simplified.
• No detailed equipment sizing/cost correlations are included.
• No financing structure, taxes, depreciation or escalation model is included.
• LCOA is not a bankable project-finance estimate.

The equilibrium-conversion function is a transparent screening correlation. It is not a fitted industrial catalyst model.

No Aspen Plus numerical result should be claimed until the actual Aspen workflow has been executed and the exported results are documented.

This project can support discussion around both chemical engineering and analytical decision making.

• Why is the H₂:N₂ feed ratio 3:1?
• Why is recycle required?
• Why is a purge required?
• Why does electrolysis dominate energy demand?
• Why does pressure influence ammonia synthesis?
• What is the role of NH₃ condensation?
• How would you validate the process in Aspen Plus?

• Which assumptions dominate LCOA?
• Why use deterministic grid search?
• How would you handle uncertainty?
• How would you introduce a pressure-dependent compressor model?
• How would you optimize for cost and carbon simultaneously?
• How would you model renewable intermittency?
• What would you change before calling this a bankable TEA?

"I intentionally separated the transparent screening layer from the rigorous validation layer. The Python model is useful for fast scenario analysis and optimization, while Aspen Plus should be used to validate thermodynamics, recycle convergence, phase behavior and unit-operation duties before treating the results as design-grade."

See docs/references.md.

Key background sources include:

1. Conceptual process design and technoeconomic analysis of an e-ammonia plant: Green H₂ and cryogenic air separation coupled with Haber–Bosch process, International Journal of Hydrogen Energy (2024).
2. Performance of a Small-Scale Haber Process: A Techno-Economic Analysis, ACS Sustainable Chemistry & Engineering (2020).
3. Recent green-ammonia techno-economic optimization literature listed in docs/references.md.

Abhishek Kumar Gond
B.Tech in Chemical Engineering @ IIT Guwahati

• Email : mr.abhishekaaa@gmail.com
• LinkedIn

This repository demonstrates:

Process Engineering + Python Modeling + Techno-Economic Analysis + Optimization + Reproducibility + Deployment

The central screening insight is straightforward:

For green ammonia, the cost and energy story is strongly driven by electricity-intensive hydrogen production.

The value of this project is not a single "optimal" number. It is the reproducible chain from engineering assumptions → process balances → energy → economics → emissions → optimization → validation.