Methodology
How embodied carbon is calculated and where the data comes from.
What is Embodied Carbon?
Embodied carbon is the total greenhouse gas emissions (measured in kgCO2e) released during the manufacturing, transportation, and end-of-life disposal of hardware — everything except operational energy use.
For on-premises infrastructure, embodied carbon typically represents 20–40% of total lifecycle emissions. Unlike operational carbon (which can be reduced by switching to renewable energy), embodied carbon is locked in at the point of purchase.
Every physical appliance replaced by a software-defined equivalent avoids its manufacturing emissions entirely. VCF Green Insight quantifies this impact for each transformation.
The VCF Transformation Story
VMware Cloud Foundation replaces dedicated physical appliances with software-defined services running on shared server infrastructure. The table below maps each transformation modelled in this tool.
| Function | Physical appliance | VCF replacement | Carbon impact |
|---|---|---|---|
| Storage | FC switches + SAN + shelves | vSAN (local NVMe) | Eliminates FC fabric and external storage |
| Routing | Cisco Nexus routers | NSX Edge virtual routers | ~200W overhead vs ~1,920W physical |
| Firewalling | FortiGate 600E | vDefend gateway firewall | Removes appliances, minimal CPU overhead |
| Load balancing | F5 BIG-IP i5800 | Avi Load Balancer | Avi service engines on existing hosts |
| IDS/IPS | Cisco Firepower 2130 | vDefend IDS | Inspection in hypervisor data path |
| WAF | Imperva SecureSphere | Avi WAF | WAF rules on Avi service engines |
How this affects emissions
- • Scope 3 — each eliminated appliance avoids its full manufacturing footprint
- • Scope 2 — reduced power draw, net of virtual overhead (10–15% of original)
- • Virtual overhead is not free — it is accounted for as a “Virtual Workloads” line item in the dashboard
GHG Protocol Scopes
The Greenhouse Gas Protocol defines three scopes of emissions. Embodied carbon falls under Scope 3.
| Scope | What it covers | Data centre examples | In this tool? |
|---|---|---|---|
| Scope 1 | Direct emissions from owned sources | On-site diesel generators, refrigerant leaks | No |
| Scope 2 | Indirect emissions from purchased energy | Electricity for servers & cooling (via PUE) | Yes |
| Scope 3 | All other indirect emissions in the value chain | Hardware manufacturing, transport, disposal — this is what VCF Green Insight calculates | Yes |
Lifecycle Phases
The Product Carbon Footprint (PCF) of hardware is broken down into lifecycle phases defined by ISO 14040/14044.
| Phase | Description | Typical share | Tracked? |
|---|---|---|---|
| Manufacturing | Raw material extraction, component fabrication, assembly | 20–40% of total lifecycle | Primary |
| Transport | Shipping from factory to data centre | ~4% of total | When available |
| Use | Electricity consumed during operational lifetime | 50–70% of total lifecycle | Scope 2 |
| End-of-Life | Recycling, landfill, incineration | ~4% of total | When available |
Calculation Methods
This tool uses three methods to determine embodied carbon, depending on data availability.
1. Published Product Carbon Footprint (PCF)
When a vendor publishes verified carbon data for a specific product.
Used for: Dell servers (Boavizta), Cisco switches, Brocade FC switches, NetApp storage, Fortinet desktop firewalls
2. Component Build-Up (Boavizta)
For modular products (servers), the total is built from individual component carbon factors.
+ (cpu_kgco2e × cpu_count)
+ (dimm_kgco2e × dimm_count)
+ (ssd_kgco2e × ssd_count)
+ (hdd_kgco2e × hdd_count)
+ (psu_kgco2e × psu_count)
Component factors are sourced from the Boavizta component impact database.
3. ADEME Weight-Based Estimate
When no published PCF or component data exists, the weight of the equipment is used with the ADEME manufacturing factor.
The factor of 80.7 kgCO2e/kg comes from the ADEME (French Agency for Ecological Transition) average manufacturing intensity for electronic equipment. Used for products without vendor-published data (e.g. firewalls, load balancers, IDS/IPS appliances).
Scope 2 — Operational Energy Calculation
Scope 2 estimates the carbon emissions from electricity consumed by hardware over its operational lifetime, including cooling overhead captured by the Power Usage Effectiveness (PUE) factor.
Formula
scope2_kgco2e = annual_kwh × PUE × grid_carbon_intensity × lifespan_years
| Parameter | Default | Description |
|---|---|---|
| Service Life | 5 years | Expected operational lifetime. Presets: 3, 5, or 7 years. |
| PUE | 1.6 | Power Usage Effectiveness — the ratio of total facility energy to IT equipment energy. A PUE of 1.6 means 60% overhead for cooling, lighting, and UPS losses. Industry average is ~1.58 (Uptime Institute 2023). |
| Grid Carbon Intensity | 0.233 kgCO2e/kWh | The carbon intensity of grid electricity. Default is the UK average (DESNZ/BEIS 2023 conversion factor). Varies by country and energy mix. |
Power data sources
- • Servers — Dell PowerEdge R760 typical power from vendor datasheet
- • Network & Security — typical power from vendor datasheets in data/manual_seed_data.json
- • Power values assume typical workload, not idle or peak
Virtual workload overhead
When physical appliances (routers, firewalls, load balancers, IDS/IPS, WAF) are replaced by virtual equivalents, the workload shifts to the server fleet. This increases server power draw. The tool estimates this overhead at roughly 10–15% of the original appliance’s power consumption — virtual workloads are far more efficient than dedicated hardware with its own PSUs, fans, and unused capacity, but they are not free. The overhead appears as a “Virtual Workloads” line item in the infrastructure table when any virtualisation transform is active.
Manufacturing Estimation Ratios
When a product has a total lifecycle carbon value but no manufacturing-specific breakdown, manufacturing is estimated using category-specific ratios derived from published PCF reports.
| Category | Manufacturing ratio | Meaning |
|---|---|---|
| Server | 30% | Higher ratio due to complex component assembly |
| Storage controller / shelf | 40% | Dense storage media has high manufacturing intensity |
| Ethernet switch | 22% | Lower manufacturing share; use-phase power dominates |
| FC switch | 22% | Similar profile to ethernet switches |
| Firewall / IDS / Load balancer / WAF | 22% | Network appliance baseline |
Data Sources & Terminology
| Term | Full name | Description |
|---|---|---|
| PAIA | Product Attribute to Impact Algorithm | An MIT/industry methodology that estimates carbon impact from product attributes (weight, component count, power). Used by Cisco, Dell, and others for ICT equipment. Data sourced from the Boavizta database. |
| EPD | Environmental Product Declaration | A standardised, third-party verified document (ISO 14025) reporting the environmental impact of a product. More rigorous than PAIA. Used by Fortinet for the FortiGate 40F/50G. |
| LCA | Life Cycle Assessment | A comprehensive study (ISO 14040/14044) of all environmental impacts across a product's full lifecycle. The gold standard for carbon data. Used by Cisco for the Meraki MX67. |
| ADEME | French Agency for Ecological Transition | Publishes average carbon intensity factors for electronic equipment manufacturing (80.7 kgCO2e per kg). Used as a fallback when no vendor-specific data is available. |
| Boavizta | Boavizta Project | An open-source project providing environmental impact data for ICT equipment. Primary source for server PCF data and component-level carbon factors (CPU, DIMM, SSD, HDD, PSU) in this tool. |
| PCF | Product Carbon Footprint | The total greenhouse gas emissions attributable to a product across specified lifecycle stages. Measured in kgCO2e (kilograms of CO2 equivalent). |
| GWP | Global Warming Potential | A measure of how much heat a greenhouse gas traps relative to CO2 over a given time period (usually 100 years). All emissions in this tool are expressed as kgCO2e using GWP-100 conversion factors. |
| kgCO2e | Kilograms of CO2 equivalent | The standard unit for expressing greenhouse gas emissions. Converts all GHGs (methane, N2O, etc.) into their CO2-equivalent warming impact. |
| SCI | Software Carbon Intensity | A Green Software Foundation specification (ISO/IEC 21031:2024) for measuring the carbon intensity of software. Formula: SCI = ((E × I) + M) per R — where E = energy, I = carbon intensity of electricity, M = embodied emissions, R = functional unit. This tool focuses on the M (embodied) component. |
Data Quality & Limitations
| Data quality | Source | Products |
|---|---|---|
| High | Vendor EPD / third-party verified LCA | Fortinet FortiGate 40F/50G, Cisco Meraki MX67 |
| Medium | Vendor PAIA / Boavizta database | Dell servers, Cisco switches, Brocade FC, NetApp storage |
| Estimate | ADEME weight-based factor (80.7 kgCO2e/kg) | FortiGate 600E, F5 BIG-IP, Cisco Firepower, Imperva WAF |
Key limitations
- • Scope 2 uses typical power draw — actual power varies with workload, temperature, and configuration
- • ADEME estimates have ~30% uncertainty; use vendor-published PCF data when available
- • Transport and end-of-life data is incomplete for many products
- • Component-level data (CPUs, DIMMs, SSDs) uses Boavizta averages, not chip-specific measurements
- • A single service life is applied to all equipment. In practice, servers are typically refreshed every 3–5 years while network switches may run 7–10 years
- • All values represent manufacturing carbon at time of production; actual values vary by factory location and energy mix