Methodology

Calculation Methodology

Helios converts your cloud infrastructure data into actionable cost, energy, and carbon metrics using transparent, auditable calculations.

Cost

Real billing data with reference pricing fallback

Energy

Power consumption × runtime with PUE adjustment

Carbon

Energy × regional grid carbon intensity

Core Principles

Deterministic

Same inputs always produce same outputs

Transparent

Every assumption is visible and editable

Auditable

Full calculation trace for compliance

Confidence-Scored

Uncertainty is always quantified

How It Works

Step 1

Data Ingestion

Upload CSV, connect cloud APIs

Step 2

Normalization

Map to standard schema

Step 3

Calculation

Apply formulas with assumptions

Step 4

Results

Cost, energy, carbon metrics

Energy Calculation

Formula

Energy (kWh) = (Runtime × Power × PUE) / 1000

Runtime Hours

Actual workload duration from billing data, job logs, or query timestamps.

Power Draw (Watts)

Determined by priority:

1Customer measurementHighest confidence
2Instance type lookupReference database
3GPU TDP specsManufacturer data
4EstimationLowest confidence

PUE by Provider

1.135

AWS

1.10

GCP

1.18

Azure

1.58

Industry Avg

Carbon Calculation

Formula

Carbon (kgCO₂e) = Energy × Grid Intensity / 1000

Regional Carbon Intensity

Grid intensity varies by region based on energy mix. Lower values indicate cleaner grids.

eu-north-1 (Sweden)
28 g
eu-west-3 (France)
56 g
us-west-2 (Oregon)
117 g
us-east-1 (Virginia)
337 g
ap-south-1 (Mumbai)
632 g

Cost Calculation

Priority Order

Exact BillingReference PricingEstimation

Exact Billing

AWS CUR, GCP Billing Export, Azure Cost Management — highest confidence

Reference Pricing

On-demand pricing from reference database: Hourly Rate × Runtime Hours

Estimation

Based on CPU/GPU count and memory — lowest confidence

Confidence Scoring

Every calculation includes a confidence score (0-100%) based on data quality. Scores start at 50% and adjust based on available data.

0-39%

Unverified

40-59%

Low

60-79%

Medium

80-100%

High

Increases Confidence

  • +15-20%Exact billing data
  • +10-15%Customer measurements
  • +8%Known region
  • +5%Recognized instance

Decreases Confidence

  • -10-20%Estimated power
  • -10-15%Unknown region
  • -10-15%Missing fields
  • -15%Validation errors

Data Sources

Grid Carbon Intensity

  • EPA eGRID 2022 (US)
  • Ember 2023 (Europe, APAC)
  • IEA 2023 (Global)

PUE Values

  • AWS Sustainability Report 2023
  • Google Environmental Report 2023
  • Microsoft Sustainability Report 2023

Instance Specifications

  • Cloud provider documentation
  • Hardware manufacturer specs
  • Third-party benchmarks

Pricing Data

  • AWS, GCP, Azure pricing APIs
  • Updated monthly
  • On-demand rates

Limitations

Historical snapshots

Analysis reflects data at time of upload. Real-time changes are not captured.

Time-of-day variations

Grid carbon intensity varies throughout the day. We use annual averages.

Scope 3 emissions

Hardware manufacturing and supply chain emissions are not included.

Cooling efficiency

PUE is an average; actual efficiency may vary by workload and season.

Last updated: January 2026