Design Philosophy for 20 MW Trigeneration fecility for Data Centres

Design Philosophy The 20 MW CRT Trigeneration demonstration will be designed around a simple systems-engineering principle: Every molecule has a destination. Every MW must produce useful work. The objective is therefore not merely to generate 20 MW of electricity. It is to maximise the useful output obtained from the primary energy entering the complete system. For an AI data-centre application, the energy pathway should be considered as an integrated chain: Primary Energy → Firm Power → Compute → Cooling → Workload → Useful AI Output This leads to an overarching performance metric: Useful AI Output per MW of Primary Energy Traditional generating efficiency measures the conversion of fuel or primary energy into electricity. Data-centre PUE measures the relationship between facility electricity consumption and IT electricity consumption. Neither metric alone describes the performance of the complete energy-to-compute system. CRT Trigeneration therefore proposes a broader systems approach in which electrical generation, carbon recycling, hydrogen, heat recovery, cooling and high-density AI computing are considered as one integrated energy architecture. Three nested performance levels can be measured: 1. CRT System Efficiency How effectively primary energy is converted into firm electricity and useful recoverable thermal energy. 2. Data-Centre Energy Efficiency How much delivered electrical energy reaches the computing equipment rather than auxiliary infrastructure. 3. Compute Productivity How much useful AI workload is completed for each MW of primary energy entering the overall system. The ultimate objective is not simply the lowest-carbon electron or the most efficient GPU considered independently. It is to maximise useful computational work from constrained primary energy while maintaining firm, continuous operation. This provides the engineering basis for the 20 MW CRT Trigeneration demonstration and a framework that can subsequently be scaled modularly to larger AI and industrial energy infrastructure.