I85PREP
⌘ K
LCA Technology study materials
LCA Technology study resources

OpenLCA JSON-LD to ILCD round trip and model versioning

LCA Technology · Put your knowledge to work in context.

All case studies

OpenLCA JSON-LD to ILCD round trip and model versioning

A consultancy models a cradle-to-gate LCA of a heat pump in openLCA 2.0 and exports the product system as openLCA JSON-LD for a client who uses SimaPro 9.5 and ILCD. The model contains a parameter 'grid_share' (0.6 wind, 0.4 grid) and links to ecoinvent 3.10 APOS. After the round trip, the client reports GWP of 1,510 kg CO2e versus the consultancy's 1,240 kg. Inspection shows: the parameter became fixed values; the client's database is ecoinvent 3.9.1 cut-off; one flow 'electricity, low voltage' was mapped by UUID to a different dataset; the LCIA method is ReCiPe 2016 v1.1 in the original and v1.2 at the client; and the functional unit '1 unit (heat pump, 8 kW)' lost its reference flow property in ILCD. The client asks whether the 22% difference is a real modelling difference and how to make future exchanges reproducible.

1. Diagnose the likely causes of the 22% GWP difference and outline a stepwise reconciliation.

Read the worked answer

The 22% gap is probably a combination of causes, not one error. Ranked by likely effect:

  1. Parameter fixation: 'grid_share' 0.6/0.4 became fixed values and may not match the intended mix.
  2. System model change: APOS versus cut-off changes waste and by-product burdens, often 5-15% for electricity-intensive products.
  3. UUID mapping: 'electricity, low voltage' mapped to a different dataset (voltage, geography or year), which can change GWP substantially.
  4. LCIA method version: ReCiPe 2016 v1.1 versus v1.2 has revised characterisation factors for some flows.
  5. Reference flow property loss in ILCD can alter unit conversion for the 8 kW unit.

Reconcile stepwise: run both models with identical dataset version, method version and fixed parameter values; compare per-flow LCIA contributions (top 10 flows) and mass/energy balances; validate the ILCD package; inspect the UUID mapping table. Only after harmonising these should you judge whether any residual difference is a real modelling difference.

2. Design a versioning and exchange workflow that would make future transfers reproducible.

Read the worked answer

Use a versioned exchange package:

  1. Freeze and declare ecoinvent version and system model, LCIA method version, and openLCA/SimaPro versions.
  2. Export a manifest listing every dataset UUID, name, version and unit.
  3. Do not rely on parameters surviving ILCD. Export openLCA JSON-LD for the full parameterised model and a separate ILCD package for the client, with a parameter sheet (grid_share = 0.6 wind / 0.4 grid).
  4. Maintain a UUID mapping table and require manual review of every mapped flow.
  5. Run a round-trip test before delivery and compare GWP within a tolerance (for example <1%).
  6. Use the openLCA API or a script to generate exports, checksums and a change log, stored in version control.
  7. State the model version in the final report.

With this workflow, the 1,240 versus 1,510 kg difference becomes explainable and reproducible rather than an unexplained 22% discrepancy.