Work

Decision-making model on the evaluation of high-powered computing based on the fuzzy comprehensive evaluation and analytic hierarchy process method

Year: 2026
Type: article
Abstract: High-performance computing (HPC) exerts complex, multi-constrained environmental impacts. This work integrates the analytic hierarchy process (AHP) and fuzzy comprehensive evaluation (FCE) to construct a decision-making model on the evaluation of high-powered computing. Nonlinear regression estimates the 2027 annual power consumption of the target HPC at 1.4×10 4 kWh. Six evaluation indices, namely water resource, electronic waste, land use, air quality, chemical usage and resource consumption indices, are established, followed by data collection, normalization and imputation. AHP ranking yields XD HPC > PKU HPC > THU HPC > SJTU HPC > CSU HPC > TC HPC. The more objective FCE further identifies PKU, THU and SJTU HPCs as environmentally favorable, whereas CSU and TC HPCs show adverse overall environmental effects. A logistic growth model is adopted to project HPC expansion, energy demand and energy mix scenarios toward 2030; energy demand gradually rises and saturates under HPC-scale and environmental constraints. Carbon emission abatement is quantified for higher renewable energy penetration, and sensitivity analysis examines how adjusted water resource indices alter overall scores. Upon implementing a proposed mitigation measure, ordinary and partial differential equations are incorporated to update the judgment matrix, re-derive indicator weights and re-run the AHP-FCE evaluation. Validated on five HPC cases, the integrated AHP-FCE framework supports HPC impact assessment. A recommendation is submitted to the advisory board to add detailed HPC environmental targets to the 2030 roadmap.
Language: English
Open Access status: hybrid
OpenAlex ID: W7236573164
FWCI:
Cites: 66
Cited by:
Related to: 10