System Modelling
Answers about building models in TianGong LCA and how it compares with other LCA tools.
How do I build a model in TianGong LCA?
- Create or select the required unit processes in My Data → Processes, ensuring inputs and outputs are complete.
- Open My Data → Models, click
+, and add the processes to the canvas. - Connect inputs and outputs, set the reference process, and define the reference flow quantity.
- Save the model—see Create My Data for the full workflow.
If saving a model reports a missing reference process, what should I do?
If saving shows “The model is missing reference process information. Complete or rebind the
reference process, then try again.”, the current model does not have a usable
referenceToReferenceProcess. Return to the model canvas, confirm that the node used as the
reference process still exists, then set the reference process and reference flow quantity again
before saving.
How does the platform choose providers for product inputs?
When a product input exchange is not explicitly linked to one provider, TianGong
LCA uses an implicit regional supply mix in the solver snapshot.
exchange.location on the product input is used first as the supply region. If
no usable exchange location exists, the consumer process location is used as the
default supply-region anchor. If the consumer process has a model_id and the
same model contains an eligible reference-output provider for the product flow,
the system first uses that same-model provider subset. If no same-model provider
exists, it falls back to the broader regional supply mix. The system then chooses
the best geography tier and weights providers inside that tier by their annual
supply or production volume. Automatic linking treats only a quantitative
reference output as an eligible provider for that product flow; a
same-flow_id non-reference output does not automatically enter the provider
set only because it has an amount or allocation fraction.
This allocation decides which providers carry one input demand; it does not
increase or reduce the total amount of that input. If a provider has no valid
annual volume, the system keeps it with a default positive weight and exposes
the fallback in snapshot coverage or provider-linking diagnostics.
If every same-flow candidate is a non-reference output, diagnostics expose the
input as rejected_non_reference_only; repair the reference output, add
provider data, or model the market / co-product process explicitly.
How does TianGong LCA differ from other LCA platforms?
| Aspect | TianGong LCA | Traditional desktop tools (SimaPro, GaBi, Umberto, openLCA, etc.) |
|---|---|---|
| Data management | Cloud-native with team collaboration and review workflows | Primarily local projects; collaboration often manual |
| Model construction | Assemble processes from shared repositories; publish to teams | Typically import or maintain private databases |
| Data format | Uses the open TIDAS JSON schema for transparency and integration | Proprietary database formats |
| Version control | Built-in drafts, history, and status tracking | Depends on external file management |
Can I interoperate with other software?
- JSON export: Download datasets as TIDAS JSON, then map fields for import into other tools.
- MCP integration: Use the TianGong MCP services to call datasets from external applications.
- Caveats: Unit names and property definitions may differ between systems; plan for a mapping or conversion step.
For bespoke integration or enterprise requirements, contact the TianGong team for partnership options.