# Cost Analysis Excel File

All\_parameters

| index | Category | Parameter | Symbol | Units | Value | SourceType | SourceRef | WhyItMatters | Status |
| ---| ---| ---| ---| ---| ---| ---| ---| ---| --- |
| 35 | AI | Additional CO₂ savings vs rule-based | ΔCO2^(S2-S1) | kg/year |  | ModelOutput | Emission factors + model | Shows climate benefit of AI beyond existing pilot. | Missing |
| 34 | AI | Additional gas savings vs rule-based | ΔV\_gas^(S2-S1) | m³/year |  | ModelOutput | Thermal model + AI scheduling | Extra gas reduction thanks to smarter use of FC heat. | Missing |
| 31 | AI | Average effective electricity price paid for H₂ | p\_H2\_el^(S2) | €/kWh |  | ModelOutput | Use hourly tariffs and AI schedule | Feeds into levelised cost of hydrogen. | Missing |
| 32 | AI | Cost of electricity from hydrogen (AI) | C\_H2^(S2) | €/kWh |  | ModelOutput | From E\_el, η\_FC, p\_H2\_el | Compare vs peak grid price to see if H₂ is cheaper. | Missing |
| 30 | AI | Energy shifted from peak to off-peak | ΔE\_shift | MWh/year | ModelOutput | Simulation of net grid imports | Key quantity for calculating € savings from off-peak shifting. | Missing |
| 29 | AI | Increase in off-peak H₂ production vs S1 | Δβ\_off | percentage points | ModelOutput | Compare S2 vs S1 | Shows how much extra off-peak shifting AI achieves. | Missing |
| 33 | AI | Peak demand reduction (peak shaving) | ΔP\_peak | kW |  | ModelOutput | Net load analysis | Reduced contracted capacity/peak charges → extra savings. | Missing |
| 28 | AI | Share of H₂ production done off-peak | β\_off^(S2) | % |  | ModelOutput | RL/AI simulation with ToU tariffs | Core KPI for off-peak generation savings. | Missing |
| 10 | Demand | Annual electricity demand of hospital | E\_hosp | MWh/year | 1000 | Public | Rijnstate Elst hydrogen project description – 1,000 MWh/year consumption | Base load used to scale PV/H2 coverage and express % cost and CO₂ savings. | Confirmed |
| 14 | Demand | Baseline annual gas consumption | V\_gas\_0 | m³/year |  | Internal | Gas utility bills | Baseline fossil gas use to compare against H₂-based heating and FC heat recovery. | Missing |
| 12 | Demand | Hourly load profile | L\_t | kW (time series) | Internal | SCADA / metering data | Essential to simulate operation of PV, H₂, battery and grid. | Missing |
| 11 | Demand | Peak electrical demand | P\_peak | kW |  | Internal | Load curve / SCADA | Needed to quantify peak shaving from H₂ and batteries. | Missing |
| 13 | Demand | Share of critical loads | α\_crit | % of total load | Internal | Hospital engineering department | Used to quantify resilience and guaranteed power to ICUs etc. | Missing |
| 23 | Pilot | Annual gas savings (rule-based pilot) | ΔV\_gas^(S1) | m³/year | 42000 | Public | Rijnstate H₂ pilot article | Reference performance of current P2P system vs old baseline. | Confirmed |
| 24 | Pilot | CO₂ avoided (rule-based pilot, annualised) | ΔCO2^(S1) | kg/year | 133333 | Derived | 2,000,000 kg CO₂ over 15 years reported by PowiDian → ≈133,333 kg/year | Baseline CO₂ savings from rule-based system. | Derived |
| 47 | Pilot | Cumulative CO₂ avoided over 15 years (rule-based pilot) | m\_CO2\_avoid\_15y | kg CO₂ | 2000000 | Public | Rijnstate pilot figures – 2,000,000 kg CO₂ avoided over 15 years. | Headline climate impact metric for the existing system, used as baseline to compare AI gains. | Confirmed |
| 46 | Pilot | Cumulative gas savings over 15 years (rule-based pilot) | V\_gas\_save\_15y | m³ | 600000 | Public | Rijnstate pilot figures – approx. 600,000 m³ gas saved over 15 years. | Long-horizon view of gas savings; useful for NPV and business case. | Confirmed |
| 25 | Pilot | Energy self-sufficiency (rule-based) | SS^(S1) | % of annual demand | 60 | Public | Rijnstate hydrogen system articles – target up to 60% self-sufficiency | Indicates share of electricity demand covered by local PV+H₂+FC under rule-based control. | Confirmed |
| 27 | Pilot | Fuel cell power profile (rule-based) | P\_FC\_t^(S1) | kW |  | Internal/Model | SCADA / H2B2-PowiDian control | Indicates when H₂ displaces grid power today. | Missing |
| 26 | Pilot | Hydrogen production profile (rule-based) | m\_H2\_t^(S1) | kg/hour |  | Internal/Model | SCADA / H2B2 control logic | Used for comparison with AI-controlled operation. | Missing |
| 15 | Prices | Average electricity price | p\_el\_avg | €/kWh | 0,18 | Secondary | Anchored on NL non-household medium electricity prices ~0.15 €/kWh (Eurostat/TradingEconomics) plus network/taxes | For high-level cost savings estimation. | Estimated |
| 22 | Prices | Carbon price / shadow price | p\_CO2 | €/ton CO₂ | 70 | Secondary | EU ETS allowance prices in 2024–2025 mostly in €65–75/tCO₂ range; mid-value 70 €/t used | Optional: lets us monetise CO₂ reductions. | Estimated |
| 20 | Prices | Grid CO₂ emission factor | ε\_el | kg CO₂/kWh | 0,315 | Public | PowiDian Rijnstate case: Dutch grid emits on average 0.315 kg CO₂ per kWh consumed | Used to convert reduced grid imports into CO₂ savings. | Confirmed |
| 48 | Prices | Grid electricity CO₂ emission factor (Netherlands) | ε\_el | kg CO₂/kWh | 0,315 | Public | PowiDian case – Dutch grid emits on average 0.315 kg CO₂ per kWh. | Used to translate reduced grid imports into CO₂ savings. | Confirmed |
| 18 | Prices | Grid tariff reduction for off-peak | grid\_disc\_off | % | 65 | Secondary | TenneT flexible off-peak contracts can reduce grid tariffs by up to ~65% for large users | Impacts savings from shifting load away from congested periods. | Estimated |
| 21 | Prices | Natural gas CO₂ emission factor | ε\_gas | kg CO₂/m³ | 1,79 | Secondary | RVO NL "CO2 emission factors for fuels" list: 56.5 kg CO₂/GJ and 31.65 MJ/m³ → ≈1.79 kg CO₂/m³ | Used to quantify CO₂ savings from reduced gas usage. | Estimated |
| 19 | Prices | Natural gas price | p\_gas | €/m³ | 0,6 | Secondary | Based on NL non-household gas prices ~0.04–0.08 €/kWh (Eurostat/CEIC) and ~10 kWh per m³ → ~0.4–0.8 €/m³; mid-range 0.60 €/m³ used | Gas savings (€) = gas saved (m³) × p\_gas. | Estimated |
| 17 | Prices | Off-peak electricity price | p\_off | €/kWh | 0,14 | Secondary | Assumed ~20% below average NL non-household electricity price for valley hours | Used to value electricity used for off-peak hydrogen production. | Estimated |
| 16 | Prices | Peak electricity price | p\_peak | €/kWh | 0,22 | Secondary | Assumed ~20% above average NL non-household electricity price to represent peak ToU tariff | Used to value avoided grid imports at peak times. | Estimated |
| 8 | System | Battery energy capacity | E\_bat | kWh | 500 | Secondary | Order-of-magnitude assumption based on similar PV–H2 microgrid projects (hundreds of kWh Li-ion BESS) | Short-term storage for intra-day balancing and peak shaving. | Estimated |
| 9 | System | Battery round-trip efficiency | η\_bat | \- | 0,9 | Secondary | Li-ion stationary storage typically achieves 85–95% round-trip efficiency | Impacts whether AI should prefer battery vs hydrogen. | Estimated |
| 41 | System | Effective electrical energy per kg of stored H₂ | e\_H2,eff | kWh/kg H₂ | 33 | Derived | 6.6 MWh / 200 kg ≈ 33 kWh/kg usable electrical energy equivalent. | Helps approximate coverage time and compare with other storage technologies. | Approximate |
| 36 | System | Electrolyzer hydrogen production rate | Ṽ\_H2,el | Nm³ H₂ / h | 20 | Public | PowiDian / Rijnstate article – 20 Nm³/h electrolyzer | Links kW input to hydrogen flow; used to derive kWh/kg H₂ and cost per kg. | Confirmed |
| 37 | System | Electrolyzer hydrogen production rate (mass) | ṁ\_H2,el | kg H₂ / h | 1,8 | Derived | 20 Nm³/h × 0.0899 kg/Nm³ ≈ 1.8 kg/h | Used together with power to compute specific energy consumption and H₂ availability per hour. | Approximate |
| 2 | System | Electrolyzer rated power | P\_el | kW | 100 | Public | Rijnstate electrolyzer spec (H2B2) | Limits rate of converting cheap electricity into hydrogen. | Confirmed |
| 3 | System | Electrolyzer specific energy consumption | E\_el | kWh/kg H₂ | 55,6 | Derived | 100 kW electrolyzer / 1.8 kg H₂ per hour ≈ 55.6 kWh/kg | Determines cost and efficiency of converting electricity into hydrogen. | Approximate |
| 50 | System | Environmental classification of hydrogen installation | env\_class | \- | Milieuklasse 3 | Public | Rijnstate article – installation falls under environmental class 3 as closed system. | Indicates permitting complexity and replicability in similar zoning contexts. | Confirmed |
| 6 | System | Fuel cell electrical efficiency | η\_FC | \- | 0,6 | Secondary | Typical PEM fuel cell electrical efficiency for stationary 100 kW systems (~50–60%) | Used to compute kWh output from each kg of H₂. | Estimated |
| 42 | System | Fuel cell inlet hydrogen pressure | p\_FC\_in | bar | 8 | Public | Rijnstate description – pressure reduced from 30 bar to 8 bar for FC. | Relevant for modelling regulation valves, efficiency and safety. | Confirmed |
| 5 | System | Fuel cell rated power | P\_FC | kW | 100 | Public | Rijnstate fuel cell spec (PowiDian) | Maximum power that can be supplied from H₂ instead of grid. | Confirmed |
| 43 | System | Fuel cell waste heat temperature | T\_FC\_heat | °C | 80 | Public | Rijnstate article – fuel cell heat available at ~80°C. | Determines usefulness of waste heat for domestic hot water and space heating. | Confirmed |
| 44 | System | Hot water supply temperature from FC heat | T\_HW\_supply | °C | 50–60 | Public | Rijnstate description – heat exchanger cools to 50–60°C for building systems. | Indicates compatibility with hospital hot water and heating circuits. | Confirmed |
| 45 | System | Hydrogen pipeline diameter to fuel cell | D\_pipe | mm | 12 | Public | Rijnstate article – 12 mm hydrogen pipeline from tanks to FC. | Relevant for flow constraints and pressure drop in the hydrogen supply line. | Confirmed |
| 4 | System | Hydrogen storage capacity | M\_H2\_max | kg | 200 | Public | Rijnstate H₂ storage spec (PowiDian) | Defines how much off-peak energy can be buffered as hydrogen. | Confirmed |
| 39 | System | Hydrogen storage pressure | p\_store | bar | 30 | Public | Rijnstate H₂ storage at 30 bar | Determines storage density and safety classification; relevant for modelling compression losses. | Confirmed |
| 51 | System | Hydrogen system remote monitoring and leak detection | H2\_monitor | yes/no | Yes | Public | Rijnstate description – remote monitoring and H₂ leak sensors around installation. | Supports claims around safety, resilience and AI-enabled monitoring. | Confirmed |
| 49 | System | Number of hydrogen system containers | N\_containers | \- | 2 | Public | Rijnstate description – two containers: one for production, one for FC and power conversion. | Gives physical layout; relevant for replication at other hospitals (footprint). | Confirmed |
| 38 | System | Number of hydrogen tanks | N\_tanks | \- | 2 | Public | Rijnstate hydrogen system description – two 100 kg tanks | Clarifies modularity and possible expansion (space for a third tank). | Confirmed |
| 0 | System | PV area | A\_PV | m² | 2300 | Public | Rijnstate H2 P2P case | Used to estimate PV capacity and annual kWh production. | Confirmed |
| 1 | System | PV capacity (approx.) | P\_PV | kWp | 460 | Secondary | Estimated from 2,300 m² PV area × ~0.2 kWp/m² (typical crystalline PV specific power) | Needed to simulate PV generation profile. | Estimated |
| 7 | System | Recoverable heat fraction from FC | η\_heat | \- | 0,45 | Secondary | Typical CHP fuel cell installations recover ~40–50% of input as usable heat | If heat replaces gas boiler heat, adds extra gas savings. | Estimated |
| 40 | System | Usable stored hydrogen energy | E\_H2,store | MWh | 6,6 | Public | Rijnstate article – 200 kg at 30 bar corresponds to 6.6 MWh. | Quantifies how long the hospital can run on stored hydrogen at a given power. | Confirmed |

Missing parameters

| index | Category | Parameter | Symbol | Units | Value | SourceType | SourceRef | WhyItMatters | Status |
| ---| ---| ---| ---| ---| ---| ---| ---| ---| --- |
| 35 | AI | Additional CO₂ savings vs rule-based | ΔCO2^(S2-S1) | kg/year |  | ModelOutput | Emission factors + model | Shows climate benefit of AI beyond existing pilot. | Missing |
| 34 | AI | Additional gas savings vs rule-based | ΔV\_gas^(S2-S1) | m³/year |  | ModelOutput | Thermal model + AI scheduling | Extra gas reduction thanks to smarter use of FC heat. | Missing |
| 31 | AI | Average effective electricity price paid for H₂ | p\_H2\_el^(S2) | €/kWh |  | ModelOutput | Use hourly tariffs and AI schedule | Feeds into levelised cost of hydrogen. | Missing |
| 32 | AI | Cost of electricity from hydrogen (AI) | C\_H2^(S2) | €/kWh |  | ModelOutput | From E\_el, η\_FC, p\_H2\_el | Compare vs peak grid price to see if H₂ is cheaper. | Missing |
| 30 | AI | Energy shifted from peak to off-peak | ΔE\_shift | MWh/year |  | ModelOutput | Simulation of net grid imports | Key quantity for calculating € savings from off-peak shifting. | Missing |
| 29 | AI | Increase in off-peak H₂ production vs S1 | Δβ\_off | percentage points |  | ModelOutput | Compare S2 vs S1 | Shows how much extra off-peak shifting AI achieves. | Missing |
| 33 | AI | Peak demand reduction (peak shaving) | ΔP\_peak | kW |  | ModelOutput | Net load analysis | Reduced contracted capacity/peak charges → extra savings. | Missing |
| 28 | AI | Share of H₂ production done off-peak | β\_off^(S2) | % |  | ModelOutput | RL/AI simulation with ToU tariffs | Core KPI for off-peak generation savings. | Missing |
| 14 | Demand | Baseline annual gas consumption | V\_gas\_0 | m³/year |  | Internal | Gas utility bills | Baseline fossil gas use to compare against H₂-based heating and FC heat recovery. | Missing |
| 12 | Demand | Hourly load profile | L\_t | kW (time series) |  | Internal | SCADA / metering data | Essential to simulate operation of PV, H₂, battery and grid. | Missing |
| 11 | Demand | Peak electrical demand | P\_peak | kW |  | Internal | Load curve / SCADA | Needed to quantify peak shaving from H₂ and batteries. | Missing |
| 13 | Demand | Share of critical loads | α\_crit | % of total load |  | Internal | Hospital engineering department | Used to quantify resilience and guaranteed power to ICUs etc. | Missing |
| 27 | Pilot | Fuel cell power profile (rule-based) | P\_FC\_t^(S1) | kW |  | Internal/Model | SCADA / H2B2-PowiDian control | Indicates when H₂ displaces grid power today. | Missing |
| 26 | Pilot | Hydrogen production profile (rule-based) | m\_H2\_t^(S1) | kg/hour |  | Internal/Model | SCADA / H2B2 control logic | Used for comparison with AI-controlled operation. | Missing |

Directly derrived values and secondary souced values

| index | Category | Parameter | Symbol | Units | Value | SourceType | SourceRef | WhyItMatters | Status |
| ---| ---| ---| ---| ---| ---| ---| ---| ---| --- |
| 10 | Demand | Annual electricity demand of hospital | E\_hosp | MWh/year | 1000 | Public | Rijnstate Elst hydrogen project description – 1,000 MWh/year consumption | Base load used to scale PV/H2 coverage and express % cost and CO₂ savings. | Confirmed |
| 23 | Pilot | Annual gas savings (rule-based pilot) | ΔV\_gas^(S1) | m³/year | 42000 | Public | Rijnstate H₂ pilot article | Reference performance of current P2P system vs old baseline. | Confirmed |
| 24 | Pilot | CO₂ avoided (rule-based pilot, annualised) | ΔCO2^(S1) | kg/year | 133333 | Derived | 2,000,000 kg CO₂ over 15 years reported by PowiDian → ≈133,333 kg/year | Baseline CO₂ savings from rule-based system. | Derived |
| 47 | Pilot | Cumulative CO₂ avoided over 15 years (rule-based pilot) | m\_CO2\_avoid\_15y | kg CO₂ | 2000000 | Public | Rijnstate pilot figures – 2,000,000 kg CO₂ avoided over 15 years. | Headline climate impact metric for the existing system, used as baseline to compare AI gains. | Confirmed |
| 46 | Pilot | Cumulative gas savings over 15 years (rule-based pilot) | V\_gas\_save\_15y | m³ | 600000 | Public | Rijnstate pilot figures – approx. 600,000 m³ gas saved over 15 years. | Long-horizon view of gas savings; useful for NPV and business case. | Confirmed |
| 25 | Pilot | Energy self-sufficiency (rule-based) | SS^(S1) | % of annual demand | 60 | Public | Rijnstate hydrogen system articles – target up to 60% self-sufficiency | Indicates share of electricity demand covered by local PV+H₂+FC under rule-based control. | Confirmed |
| 20 | Prices | Grid CO₂ emission factor | ε\_el | kg CO₂/kWh | 0,315 | Public | PowiDian Rijnstate case: Dutch grid emits on average 0.315 kg CO₂ per kWh consumed | Used to convert reduced grid imports into CO₂ savings. | Confirmed |
| 48 | Prices | Grid electricity CO₂ emission factor (Netherlands) | ε\_el | kg CO₂/kWh | 0,315 | Public | PowiDian case – Dutch grid emits on average 0.315 kg CO₂ per kWh. | Used to translate reduced grid imports into CO₂ savings. | Confirmed |
| 41 | System | Effective electrical energy per kg of stored H₂ | e\_H2,eff | kWh/kg H₂ | 33 | Derived | 6.6 MWh / 200 kg ≈ 33 kWh/kg usable electrical energy equivalent. | Helps approximate coverage time and compare with other storage technologies. | Approximate |
| 36 | System | Electrolyzer hydrogen production rate | Ṽ\_H2,el | Nm³ H₂ / h | 20 | Public | PowiDian / Rijnstate article – 20 Nm³/h electrolyzer | Links kW input to hydrogen flow; used to derive kWh/kg H₂ and cost per kg. | Confirmed |
| 37 | System | Electrolyzer hydrogen production rate (mass) | ṁ\_H2,el | kg H₂ / h | 1,8 | Derived | 20 Nm³/h × 0.0899 kg/Nm³ ≈ 1.8 kg/h | Used together with power to compute specific energy consumption and H₂ availability per hour. | Approximate |
| 2 | System | Electrolyzer rated power | P\_el | kW | 100 | Public | Rijnstate electrolyzer spec (H2B2) | Limits rate of converting cheap electricity into hydrogen. | Confirmed |
| 3 | System | Electrolyzer specific energy consumption | E\_el | kWh/kg H₂ | 55,6 | Derived | 100 kW electrolyzer / 1.8 kg H₂ per hour ≈ 55.6 kWh/kg | Determines cost and efficiency of converting electricity into hydrogen. | Approximate |
| 50 | System | Environmental classification of hydrogen installation | env\_class | \- | Milieuklasse 3 | Public | Rijnstate article – installation falls under environmental class 3 as closed system. | Indicates permitting complexity and replicability in similar zoning contexts. | Confirmed |
| 42 | System | Fuel cell inlet hydrogen pressure | p\_FC\_in | bar | 8 | Public | Rijnstate description – pressure reduced from 30 bar to 8 bar for FC. | Relevant for modelling regulation valves, efficiency and safety. | Confirmed |
| 5 | System | Fuel cell rated power | P\_FC | kW | 100 | Public | Rijnstate fuel cell spec (PowiDian) | Maximum power that can be supplied from H₂ instead of grid. | Confirmed |
| 43 | System | Fuel cell waste heat temperature | T\_FC\_heat | °C | 80 | Public | Rijnstate article – fuel cell heat available at ~80°C. | Determines usefulness of waste heat for domestic hot water and space heating. | Confirmed |
| 44 | System | Hot water supply temperature from FC heat | T\_HW\_supply | °C | 50–60 | Public | Rijnstate description – heat exchanger cools to 50–60°C for building systems. | Indicates compatibility with hospital hot water and heating circuits. | Confirmed |
| 45 | System | Hydrogen pipeline diameter to fuel cell | D\_pipe | mm | 12 | Public | Rijnstate article – 12 mm hydrogen pipeline from tanks to FC. | Relevant for flow constraints and pressure drop in the hydrogen supply line. | Confirmed |
| 4 | System | Hydrogen storage capacity | M\_H2\_max | kg | 200 | Public | Rijnstate H₂ storage spec (PowiDian) | Defines how much off-peak energy can be buffered as hydrogen. | Confirmed |
| 39 | System | Hydrogen storage pressure | p\_store | bar | 30 | Public | Rijnstate H₂ storage at 30 bar | Determines storage density and safety classification; relevant for modelling compression losses. | Confirmed |
| 51 | System | Hydrogen system remote monitoring and leak detection | H2\_monitor | yes/no | Yes | Public | Rijnstate description – remote monitoring and H₂ leak sensors around installation. | Supports claims around safety, resilience and AI-enabled monitoring. | Confirmed |
| 49 | System | Number of hydrogen system containers | N\_containers | \- | 2 | Public | Rijnstate description – two containers: one for production, one for FC and power conversion. | Gives physical layout; relevant for replication at other hospitals (footprint). | Confirmed |
| 38 | System | Number of hydrogen tanks | N\_tanks | \- | 2 | Public | Rijnstate hydrogen system description – two 100 kg tanks | Clarifies modularity and possible expansion (space for a third tank). | Confirmed |
| 0 | System | PV area | A\_PV | m² | 2300 | Public | Rijnstate H2 P2P case | Used to estimate PV capacity and annual kWh production. | Confirmed |
| 40 | System | Usable stored hydrogen energy | E\_H2,store | MWh | 6,6 | Public | Rijnstate article – 200 kg at 30 bar corresponds to 6.6 MWh. | Quantifies how long the hospital can run on stored hydrogen at a given power. | Confirmed |
| 15 | Prices | Average electricity price | p\_el\_avg | €/kWh | 0,18 | Secondary | Anchored on NL non-household medium electricity prices ~0.15 €/kWh (Eurostat/TradingEconomics) plus network/taxes | For high-level cost savings estimation. | Estimated |
| 22 | Prices | Carbon price / shadow price | p\_CO2 | €/ton CO₂ | 70 | Secondary | EU ETS allowance prices in 2024–2025 mostly in €65–75/tCO₂ range; mid-value 70 €/t used | Optional: lets us monetise CO₂ reductions. | Estimated |
| 18 | Prices | Grid tariff reduction for off-peak | grid\_disc\_off | % | 65 | Secondary | TenneT flexible off-peak contracts can reduce grid tariffs by up to ~65% for large users | Impacts savings from shifting load away from congested periods. | Estimated |
| 21 | Prices | Natural gas CO₂ emission factor | ε\_gas | kg CO₂/m³ | 1,79 | Secondary | RVO NL "CO2 emission factors for fuels" list: 56.5 kg CO₂/GJ and 31.65 MJ/m³ → ≈1.79 kg CO₂/m³ | Used to quantify CO₂ savings from reduced gas usage. | Estimated |
| 19 | Prices | Natural gas price | p\_gas | €/m³ | 0,6 | Secondary | Based on NL non-household gas prices ~0.04–0.08 €/kWh (Eurostat/CEIC) and ~10 kWh per m³ → ~0.4–0.8 €/m³; mid-range 0.60 €/m³ used | Gas savings (€) = gas saved (m³) × p\_gas. | Estimated |
| 17 | Prices | Off-peak electricity price | p\_off | €/kWh | 0,14 | Secondary | Assumed ~20% below average NL non-household electricity price for valley hours | Used to value electricity used for off-peak hydrogen production. | Estimated |
| 16 | Prices | Peak electricity price | p\_peak | €/kWh | 0,22 | Secondary | Assumed ~20% above average NL non-household electricity price to represent peak ToU tariff | Used to value avoided grid imports at peak times. | Estimated |
| 8 | System | Battery energy capacity | E\_bat | kWh | 500 | Secondary | Order-of-magnitude assumption based on similar PV–H2 microgrid projects (hundreds of kWh Li-ion BESS) | Short-term storage for intra-day balancing and peak shaving. | Estimated |
| 9 | System | Battery round-trip efficiency | η\_bat | \- | 0,9 | Secondary | Li-ion stationary storage typically achieves 85–95% round-trip efficiency | Impacts whether AI should prefer battery vs hydrogen. | Estimated |
| 6 | System | Fuel cell electrical efficiency | η\_FC | \- | 0,6 | Secondary | Typical PEM fuel cell electrical efficiency for stationary 100 kW systems (~50–60%) | Used to compute kWh output from each kg of H₂. | Estimated |
| 1 | System | PV capacity (approx.) | P\_PV | kWp | 460 | Secondary | Estimated from 2,300 m² PV area × ~0.2 kWp/m² (typical crystalline PV specific power) | Needed to simulate PV generation profile. | Estimated |
| 7 | System | Recoverable heat fraction from FC | η\_heat | \- | 0,45 | Secondary | Typical CHP fuel cell installations recover ~40–50% of input as usable heat | If heat replaces gas boiler heat, adds extra gas savings. | Estimated |
| 52 | Cost\_savings | Annual gas bill savings (rule-based pilot vs baseline) | C\_gas\_save^(S1) | €/year | 25200 | Derived | ΔV\_gas^(S1) \* p\_gas | Monetises the pilot’s gas savings baseline, used to benchmark AI gains. | Derived |
| 53 | Cost\_savings | Annual CO₂ value (rule-based pilot vs baseline) | C\_CO2\_save^(S1) | €/year | 9333,31 | Derived | ΔCO2^(S1) / 1000 \* p\_CO2 | Values the rule-based CO₂ reductions using a CO₂ price. | Derived |
| 54 | AI\_scenario | Additional annual gas savings enabled by AI vs rule-based | ΔV\_gas^(S2-S1) | m³/year |  | Scenario | To be set based on AI controller simulations. | Captures extra gas savings potential relative to the current pilot. | To be filled (AI scenario) |
| 55 | Cost\_savings | Annual gas bill savings (AI-optimised, total vs baseline) | C\_gas\_save^(S2) | €/year | 25200 | Derived | Uses rule-based gas savings plus additional AI gas savings times gas price. | Shows total gas bill reduction when AI is deployed. | Derived |
| 56 | Cost\_savings | Additional annual gas bill savings from AI vs rule-based | ΔC\_gas\_save^(S2-S1) | €/year | 0 | Derived | C\_gas\_save^(S2) - C\_gas\_save^(S1) | Explicitly quantifies the incremental euro savings from AI on the gas bill. | Derived |
| 57 | AI\_scenario | Additional CO₂ avoided by AI vs rule-based | ΔCO2^(S2-S1) | kg/year |  | Scenario | To be set based on AI controller simulations. | Captures extra CO₂ savings potential relative to the current pilot. | To be filled (AI scenario) |
| 58 | Cost\_savings | Annual CO₂ value (AI-optimised, total vs baseline) | C\_CO2\_save^(S2) | €/year | 9333,31 | Derived | Uses rule-based CO₂ savings plus additional AI CO₂ savings times CO₂ price. | Shows total CO₂-related value when AI is deployed. | Derived |
| 59 | Cost\_savings | Additional annual CO₂ value from AI vs rule-based | ΔC\_CO2\_save^(S2-S1) | €/year | 0 | Derived | C\_CO2\_save^(S2) - C\_CO2\_save^(S1) | Shows extra monetised CO₂ savings due to AI. | Derived |
| 60 | AI\_scenario | Energy shifted from peak to off-peak by AI (vs rule-based) | ΔE\_shift | MWh/year |  | Scenario | To be set based on AI controller simulations. | Quantifies how much load AI moves from expensive to cheap hours. | To be filled (AI scenario) |
| 61 | Cost\_savings | Annual energy price savings from peak→off-peak shifting (AI vs rule-based) | C\_shift | €/year | 0 | Derived | ΔE\_shift \* 1000 \* (p\_peak - p\_off) | Values the benefit of buying more electricity off-peak instead of at peak prices. | Derived |
| 62 | AI\_scenario | Reduction in grid peak demand enabled by AI vs rule-based | ΔP\_peak | kW |  | Scenario | To be set based on AI controller simulations. | Captures reduction in contracted/measured peak load due to AI scheduling. | To be filled (AI scenario) |
| 63 | Prices | Grid capacity tariff (indicative, adjust to contract) | p\_cap | €/kW/year | 100 | Secondary | Indicative Dutch-style capacity tariff; replace with hospital-specific value. | Allows monetisation of ΔP\_peak as annual euro savings. | Estimated |
| 64 | Cost\_savings | Annual capacity tariff savings from AI vs rule-based | C\_cap\_save | €/year | 0 | Derived | ΔP\_peak \* p\_cap | Values the benefit of reducing grid peak demand through AI scheduling. | Derived |

Cost Savings (Needs the missing values to show valuable info)

| Metric | Symbol | Units | Baseline S0 (no H₂) | Rule-based S1 (pilot) | AI-optimised S2 | Δ S1–S0 | Δ S2–S1 |
| ---| ---| ---| ---| ---| ---| ---| --- |
| Annual gas savings vs baseline | ΔV\_gas | m³/year | 0 | 42000 | 42000 | 42000 | 0 |
| Gas bill savings vs baseline | C\_gas\_save | €/year | 0 | 25200 | 25200 | 25200 | 0 |
| CO₂ avoided vs baseline | ΔCO2 | kg/year | 0 | 133333 | 133333 | 133333 | 0 |
| CO₂ value vs baseline | C\_CO2\_save | €/year | 0 | 9333,31 | 9333,31 | 9333,31 | 0 |
| Energy price savings from peak→off-peak shifting | C\_shift | €/year | 0 | 0 | 0 | 0 | 0 |
| Capacity tariff savings | C\_cap\_save | €/year | 0 | 0 | 0 | 0 | 0 |
| Total annual impact (gas + CO₂ value + shifting + capacity) | C\_total | €/year | 0 | 167866,31 | 167866,31 | 0 | 0 |

AI scenario

| C\_shift^(S2-S1) \[€/year\] | C\_cap\_save^(S2-S1) \[€/year\] | ΔC\_tot^(S2-S1) \[€/year\] | C\_tot^(S1-S0) \[€/year\] | C\_tot^(S2-S0) \[€/year\] |
| ---| ---| ---| ---| --- |
| 4000 | 5000 | 10523,13 | 167866,31 | 178389,44 |
| 8000 | 10000 | 21046,26 | 167866,31 | 188912,57 |
| 16000 | 15000 | 35569,39 | 167866,31 | 203435,7 |