Smart distributed generation and energy storage for nearly zero-energy ports: cold ironing optimisation within a collaborative quality framework

Highlights

• A port energy community framework integrating ISO 50001 energy management and ISO 37101 sustainability governance is developed for coordinating distributed PV-storage investment in cold ironing infrastructure.
• Berthed vessels are modelled as high-capacity electric vehicles (2–24 MW), connecting maritime shore power to the smart grid and EV charging discourse.
• A genetic algorithm optimizer sizes PV and BESS across twelve scenarios at the Port of Ancona, achieving an LCOE of 0.211 EUR/kWh and carbon footprint reduction of 56.04%.
• Technology readiness assessment identifies integrated PV-BESS-CI with smart energy management at TRL six to seven, establishing pilot-scale validation as the critical deployment milestone.
• A dock-by-dock clustered electrification strategy outperforms monolithic port-level sizing in per-unit decarbonization, grid stability, and spatial feasibility.

1 Introduction

The ongoing transformation of power systems from centralized, unidirectional architectures toward decentralized, bidirectional smart grids is driven by four concurrent developments: the proliferation of distributed generation (DG) from various renewable sources (Savvakis et al., 2023; Laskaratou and Farantouris, 2023; Savvakis et al., 2022), the deployment of advanced metering infrastructure (AMI) and smart energy management systems, the integration of battery energy storage systems (BESS) for grid services, and the rapid growth of electric vehicle (EV) charging demand (Farhangi, 2010). Each of these developments introduces both opportunities and challenges. Distributed generation improves energy security and reduces transmission losses but introduces reverse power flows and voltage regulation issues (Ramakumar and Chiradeja, 2002). Smart metering enables real-time demand response and predictive analytics but raises concerns about data privacy and cybersecurity (Nambundo et al., 2025). Energy storage provides grid flexibility and peak shaving (Springer, 2026) but requires careful sizing to balance capital cost against utilization (Petkova, 2024). Electric vehicle charging, when unmanaged, creates localized demand peaks that strain distribution infrastructure (Nezamuddin et al., 2022), but when coordinated through vehicle-to-grid (V2G) and vehicle-to-everything (V2X) protocols, can provide valuable grid services (Shaheen et al., 2024).

These four developments converge in the maritime port environment, which presents a uniquely challenging context for smart grid deployment. Ports are energy-intensive facilities where distributed generation potential (large rooftop and ground-mount areas), storage requirements (temporal mismatch between solar production and vessel demand), smart metering needs (heterogeneous loads from diverse tenants and vessel types), and large-scale EV charging demand (cold ironing of berthed vessels) coexist at scales that test the limits of conventional grid management approaches (Cholidis et al., 2025a; Sifakis et al., 2023). While these power requirements vary significantly depending on the vessel’s specific duty cycle, such as the high-frequency commercial operations of electric ferries in Norway or at the Port of Dover, UK (Kirca et al., 2026), a single passenger ferry at berth typically draws between 2–16 MW for 4–48 h, a load equivalent to 500–4,000 passenger EVs charging simultaneously (Tseng and Pilcher, 2015). This observation positions cold ironing (CI) as a form of large-scale EV infrastructure (Qiu et al., 2022), directly linking maritime electrification to the broader EV integration and smart grid research agenda (Sifakis et al., 2026a). Consequently, within the simulation framework developed in this study, berthed vessels are treated mathematically and operationally as high-capacity EVs charging at the port. To maintain conceptual clarity, the term ‘berthed vessels’ is used consistently to denote the physical ships, distinguishing them from traditional road-based EVs while leveraging the same advanced smart-grid dispatch logic.

From a power quality perspective, CI introduces challenges that are distinct from those of conventional distributed generation or road EV charging (Innes and Monios, 2018). The connection and disconnection of individual vessels at shore power creates step-load transients of several megawatts within seconds, generating voltage fluctuations and flicker at the point of common coupling with the distribution network (Sciberras et al., 2015). Frequency conversion equipment (required where vessel systems operate at 60 Hz on a 50 Hz grid, or vice versa) introduces harmonic distortion into the port’s electrical distribution (Paul et al., 2012). These power quality challenges are concentrated at specific berths rather than distributed across a network, making them amenable to localized mitigation through BESS and power electronic controls, but they require dedicated monitoring through advanced metering infrastructure tailored to the port environment. The integration of AMI within port microgrids enables real-time visibility of vessel connection status (Rashed Mohassel et al., 2014), PV generation, BESS state of charge, and grid power quality parameters, providing the data foundation for smart energy management algorithms that coordinate dispatch across these heterogeneous assets (Zadeh et al., 2023; Sifakis and Tsoutsos, 2020).

The European regulatory landscape reinforces this convergence. Regulation (EU) 2023/1805 (FuelEU Maritime) (Christodoulou and Cullinane, 2022) mandates progressive reductions in the greenhouse gas intensity of energy used by vessels at European ports (Regulation, 2023a). The Alternative Fuels Infrastructure Regulation (AFIR) requires shore-side electricity at all TEN-T (Trans-European Transport Network) core and comprehensive maritime ports by 2030 (Niinimaa, 2025), explicitly framing CI as alternative fuel infrastructure alongside EV charging stations (Piccoli et al., 2021; Regulation, 2023b; Sifakis et al., 2024). The Adriatic-Ionian region, with over 350 registered ports, faces a coordinated infrastructure deployment challenge that cannot be addressed through individual port-level planning alone.

The design, organization, and management of this energy infrastructure through collaborative formations offers a promising institutional pathway. Energy communities, as defined in EU Directive 2019/944, provide a governance model for collective investment, operation, and benefit-sharing among multiple stakeholders: port authorities, terminal operators, vessel operators, and grid operators (Directive 2019/944, 2019). When structured as port energy communities, these formations can pool capital for distributed generation and storage, coordinate smart energy management across heterogeneous loads, and negotiate grid connections and tariff structures collectively. Quality and sustainability standards provide the normative framework for such collaboration. ISO 50001 establishes requirements for energy management systems, including the Plan-Do-Check-Act (PDCA) cycle for continuous performance improvement (Marimon and Casadesús, 2017; ISO 50001:2018, 2026). ISO 37101 addresses sustainable development in communities, providing governance indicators for evaluating environmental, social, and economic performance (Ruggiero et al., 2024; ISO 37101:2016, 2026). Technology readiness level (TRL) assessment, widely adopted in EU innovation policy, provides a structured metric for evaluating the maturity of the constituent technologies and the integrated system (Mankins, 2009).

This study positions cold ironing (CI) as the high-intensity boundary of the electric vehicle (EV) charging spectrum (Sifakis et al., 2026b). To transition this analogy from qualitative framing to a quantitative benchmark, the power requirements are significant: a single passenger ferry at berth can draw 8–16 MW for 6–14 h. While these requirements vary significantly depending on the specific vessel duty cycle and route profile, such as the high-frequency commercial operations of electric ferries in Norway or the Port of Dover (Kirca et al., 2026) the maximum loads remain energetically equivalent to 500–4,000 Level 2 EV charging sessions occurring simultaneously at a single grid node. While CI shares structurally analogous technical interfaces and dispatch hierarchies with road electrification, governed by the convergence of IEC 80005–1 and IEC 61851 standards, it remains a distinct engineering domain due to its high load concentration at single points of common coupling.

To address these challenges, this study develops a collaborative port energy community framework that integrates these institutional instruments with technical optimization (Cholidis et al., 2025b). A genetic algorithm (GA) optimizer sizes photovoltaic (PV) distributed generation and lithium iron phosphate (LFP) battery storage for CI infrastructure at the Port of Ancona, operating within quality constraints derived from ISO 50001 performance targets and ISO 37101 sustainability criteria. Smart energy management dispatch logic governs the hourly simulation. Twelve scenarios across four electrification scales and three technology configurations are evaluated. The technology readiness level (TRL) of individual components and the integrated system is assessed formally. Ultimately, the study aims to contribute to the current research literature by connecting maritime vessel electrification to the smart grid, distributed generation (DG), energy storage systems (ESS), and EV discourse through a governance framework grounded in collaborative quality standards. While the technical integration of distributed energy resources with cold ironing is an established concept, this study departs from conventional feasibility assessments by embedding the technical sizing (GA optimization) within a collaborative port energy community framework. By uniquely applying ISO 50001 performance targets and ISO 37101 sustainability criteria as hard constraints, this research shifts the focus from purely technical evaluation to actionable, multi-stakeholder infrastructure governance.

The remainder of this paper is organized as follows. Section 2 reviews the relevant literature. Section 3 presents the methodological framework. Section 4 reports the results. Section 5 provides a discussion of the findings and their broader implications. Section 6 offers conclusions and future research directions.

2 Literature review

2.1 Distributed generation and energy storage in port microgrids

The Port of Amsterdam operates 17 MW of PV across warehouse rooftops, Rotterdam hosts a 25 MWp solar park with plans to expand to 89 MWp, and Los Angeles has installed 12 MW with 20 MW under development (AIMPRESS, 2025). These installations function as distributed generation assets embedded within the port’s distribution network, introducing bidirectional power flows and the associated challenges of voltage regulation and reverse power flow management that characterize DG integration in distribution systems generally. Recent literature has increasingly highlighted the importance of dynamic energy management over various scenarios, including the integration of local generation with highly variable Cold Ironing duty cycles (Kirca et al., 2026).

Battery energy storage addresses the temporal mismatch between solar generation profiles and port demand. LFP chemistry has emerged as the preferred technology for stationary port applications due to its thermal stability, cycle life exceeding 4,000 equivalent full cycles, and declining capital costs reaching approximately 480 EUR/kWh in 2024 (Bloomber gNEF, 2024). Iris and Lam reviewed port energy management systems and concluded that coordinated dispatch strategies integrating vessel schedules, grid tariffs, and peak demand constraints are required for effective DG-storage coupling (Iris and Lam, 2019). Acciaro et al. proposed an energy management framework for seaports that identified storage as a critical DG integration enabler but did not develop a sizing methodology (Acciaro et al., 2014). While AMI and smart metering are standard components of urban distribution networks, their application to port microgrids presents distinct challenges (Chakraborty and Sharma, 2016). Port loads are highly erratic, spanning tens of kilowatts to over 10 MW. Managing this unpredictable demand requires advanced smart metering to capture detailed power metrics (like harmonics and power factor) for Smart Energy Management Systems (SEMS) (Nthontho et al., 2011).

Recent literature has explored flexible energy management for ships to reduce carbon footprints (Barone et al., 2025) and proposed multi-agent frameworks for decentralized port terminal operations (Dash et al., 2025). While these studies effectively address multi-stakeholder scheduling and operational optimization, they largely optimize for unconstrained economic or emissions objectives without linking the capacity sizing to formal institutional quality standards.

The power quality impacts of high-penetration DG in ports remain underexplored (Bajaj and Singh, 2020). During low demand, large PV installations trigger reverse power flows and voltage rise (Mariscotti, 2021). While BESS can mitigate this, optimizing its size for both power quality and energy arbitrage is an unaddressed multi-objective challenge (Kumar and Panda, 2023). To solve such sizing problems, genetic algorithm (GA) optimization is widely used for renewable systems, demonstrating success in policy-constrained microgrids (Cholidis et al., 2025a) and marine-based systems (Sifakis et al., 2025b). However, these GA studies lack the specific PV-storage- CI coupling and collaborative governance needed for modern port communities.

2.2 Cold ironing as electric vehicle infrastructure

CI, the supply of shore-side electricity to berthed vessels through standardized connections (IEC/ISO/IEEE 80005), enables the shutdown of on-board auxiliary diesel generators during hotelling (Zis, 2019). The IEC 80005 standard specifies high-voltage (6.6/11 kV) and low-voltage (440 V) connection systems with frequency conversion where 50/60 Hz mismatch exists, directly paralleling the standardization efforts in EV charging (IEC/ISO/IEEE, 2019; International Electrotechnical Commission, 2019). Winkel et al. estimated 50%–70% CO2 reduction potential at European ports (Winkel et al., 2016). Vaishnav et al. quantified health benefits exceeding infrastructure costs at US ports (Vaishnav et al., 2016). Sciberras et al. found CI superior to LNG bunkering when supplied by renewable-rich grids (Sciberras et al., 2015). Ballini and Bozzo assessed the socio-economic benefits of CI in port cities (Ballini and Bozzo, 2015).

The parallel of CI with EV infrastructure extends beyond the charging analogy to encompass scheduling, load balancing, and grid service provision (Matanov and Zahov, 2020). In road EV charging, smart charging algorithms schedule vehicle charging sessions to minimize grid impact by shifting demand to off-peak hours or periods of high renewable generation (Singh et al., 2022). An analogous opportunity exists in CI: where vessel schedules permit flexibility (e.g., ro-ro vessels with overnight stays), the timing and power level of shore power draw can be coordinated with PV generation availability and BESS state of charge (Pruyn and Willeijns, 2022). This coordination requires the same data infrastructure (AMI, vessel scheduling integration, real-time SoC monitoring) that underpins smart EV charging networks.

Future vessel-to-grid (V2G) concepts, in which vessels with on-board battery systems provide grid services while at berth, mirror current V2G and V2X research for road EVs (Sovacool et al., 2018). During long shore connections, the unutilized capacity of large hybrid and electric ferry batteries (1–10 MWh) could provide frequency regulation, peak shaving, and reactive power support to port grids (Li et al., 2020). While V2G for vessels remains at an early conceptual stage (TRL 2–3), the shore connection infrastructure deployed for CI would serve as the physical interface for such bidirectional energy exchange, making current CI investment forward-compatible with future V2G capabilities. The IEC 80005 standard, which governs shore connections, does not currently address bidirectional power flow, but its extension to accommodate V2G protocols is a foreseeable development (Hsu et al., 2024).

Furthermore, while commercial CI systems for large cruise vessels are increasingly established due to their long, predictable hoteling durations, high-frequency passenger ferries and Ro-Ro vessels present a distinct operational challenge. Their shorter, more stochastic berthing windows require dynamic energy management and BESS buffering to mitigate sudden grid impacts, a dynamic largely unaddressed in standard cruise-focused CI literature. The gap in literature is the absence of studies that couple CI demand analysis with distributed generation and storage optimization within a smart energy management framework, and that embed this technical optimization within a collaborative governance structure. Most existing studies treat CI as a fixed grid load without considering on-site generation potential, storage coordination, or the institutional arrangements needed to manage shared energy assets among multiple port stakeholders.

2.3 Energy communities and collaborative governance for energy infrastructure

Energy communities, as defined in Article 2 (16) of EU Directive 2019/944, are legal entities governed by voluntary, open participation principles whose primary purpose is providing environmental, economic, or social community benefits. The Renewable Energy Directive (EU 2018/2001) further defines renewable energy communities with emphasis on local ownership, proximity to generation assets, and autonomy from commercial energy supply (Directive 2018/2001, 2018).

When applied to port environments, energy community structures enable the pooling of capital for DG and storage investment, collective smart energy management across heterogeneous loads, and shared benefit from reduced electricity costs (Attou et al., 2021). Port environments are natural candidates for the community model because they already involve multiple co-located stakeholders (port authorities, terminal operators, shipping lines, industrial tenants) who share common infrastructure and energy supply points. The multi-stakeholder structure of port operations creates both a need for coordinated governance (no single actor controls all loads, generation assets, and grid connections) and an opportunity for collective efficiency (shared storage and generation assets achieve higher utilization rates than independently operated installations).

The institutional dimension of port electrification extends beyond technical optimization to encompass quality management and sustainability governance, a dimension largely absent from the existing optimization literature. Energy clusters, a related concept in industrial ecology, organize co-located enterprises into cooperative formations that share energy infrastructure, exchange waste heat, and jointly procure renewable electricity (Sifakis et al., 2019; Chalkiadakis et al., 2018; Sifakis et al., 2020). Port energy communities are specialized clusters comprising both fixed tenants and transient vessels. This transient participation creates unique governance challenges, requiring models that accommodate intermittent membership, variable benefit-sharing, and standardized, plug-and-play shore connections.

2.4 Quality management and sustainability standards for energy infrastructure

ISO 50001 establishes a management system for energy performance based on the Plan-Do-Check-Act (PDCA) cycle (ISO - ISO 50001, 2026). Applied to port energy infrastructure, the PDCA cycle structures the optimization process: Plan (energy baseline establishment, performance target setting, sizing optimization), Do (system deployment and operation), Check (monitoring through smart metering, KPI evaluation against targets), Act (corrective measures, system reconfiguration). This study operationalizes the Plan phase through GA optimization with quality-constrained fitness evaluation.

ISO 37101 provides a sustainable development management system for communities. For port energy communities, their most relevant purposes are responsible resource use (renewable energy deployment), environmental preservation (emissions reduction), and resilience (energy security). Ultimately, the standard serves as a framework to evaluate whether technical solutions meet these broader community objectives.

Additionally, TRL assessment provides a nine-level scale for evaluating technology maturity. In the context of collaborative energy systems, TRL applies not only to individual hardware components but to the entire integrated system, encompassing technology, governance, and operational procedures. This multi-level perspective aligns perfectly with the collaborative innovation approach emphasized in European energy transition policy.

2.5 Research gaps and contributions

The preceding review identifies four gaps.

• CI demand has not been coupled with distributed generation and storage optimization within a smart energy management framework.
• Port energy system studies lack the institutional governance dimension: energy community organization, quality management (ISO 50001), and sustainability assessment (ISO 37101) are absent from the optimization literature.
• The connection between maritime vessel electrification and the broader smart grid, DG, ESS, and EV research programme has not been systematically established.
• TRL assessment has not been applied to the integrated CI energy system as a collaborative technology innovation.

This study addresses these gaps through four contributions. First, it develops a GA optimization framework that sizes PV and BESS for CI demand profiles within ISO 50001 quality constraints and ISO 37101 sustainability targets. Second, it moves beyond standard technical assessments by proposing a port energy community governance model for multi-stakeholder coordination of distributed generation and storage investment, bridging the gap between hardware deployment and institutional readiness. Third, it frames CI as large-scale EV infrastructure, linking maritime electrification to the smart grid and EV discourse. Fourth, it provides a formal TRL assessment of the integrated system and its constituent components.

3 Methodology

3.1 Conceptual framework: ISO 50001 PDCA integration

The methodological framework is structured according to the ISO 50001 PDCA cycle, which provides a systematic approach to energy performance improvement within the port energy community.

• Plan phase (Sections 3.2–3.8): Energy baseline establishment from historical port consumption data; AMI data architecture design; definition of energy performance indicators (LCOE (Levelized Cost of Energy), carbon footprint, IRR, payback period); GA optimization of PV and BESS capacities; quality criteria definition aligned with ISO 50001 performance improvement targets and ISO 37101 sustainability purposes.
• Do phase (Section 4.9 implementation pathway): Phased deployment of the optimized PV-BESS-CI system according to the recommended four-phase implementation strategy; establishment of the energy community governance structure.
• Check phase (Sections 3.8–3.9 and 4.5): Evaluation of scenario results against quality criteria thresholds; TRL assessment of individual components and the integrated system; identification of performance gaps and compliance status.
• Act phase (Sections 4.7, 5): Corrective recommendations for configurations that underperform targets; identification of additional technology interventions (hydrogen, multi-RES) for future improvement cycles; policy recommendations for energy community governance.

This PDCA integration ensures that the technical optimization is not conducted in isolation but is embedded within a structured management framework that facilitates iterative improvement and stakeholder accountability.

3.2 Case study: port of Ancona

The Port of Ancona is a multi-functional maritime facility on the central Adriatic coast of Italy (43.62 deg N, 13.50 deg E), encompassing approximately 1.4 million m². The port accommodates passenger ferries, ro-ro cargo, cruise vessels, and commercial shipping. The 2019 energy audit recorded an annual consumption of approximately 53.7 GWh, of which maritime traffic accounted for 47%, industrial facilities for approximately 50%, and the commercial terminal for 2.4% (AIMPRESS, 2025).

The vessel fleet exhibits heterogeneous electrical characteristics. Passenger ferries (Superfast XI, Olympic Champion) operate generators at 8–16 MW, 6.6/11 kV, 50 Hz with hoteling durations of 6–14 h. Ro-ro vessels (AF Claudia, Florencia) draw 1–4 MW during 8–24 h berth stays. Cruise vessels impose peak loads exceeding 10 MW seasonally. The aggregate CI demand profile exhibits daily peaks of approximately 300 MWh during the summer ferry season. CI infrastructure commissioning is planned for 2026 in alignment with the AFIR deadline. The site receives annual GHI of approximately 1,520 kWh/m2. Meteorological data were obtained from the NASA POWER API at hourly resolution for 2023. The AIMPRESS project (Adriatic-Ionian Maritime Operational Efficiency and Sustainable Scheme, IPA-ADRION00527) provides the institutional framework, with Ancona as the pilot site and Split and Marina Bar as replication candidates.

3.3 System architecture and smart energy management

The proposed system integrates grid-connected PV distributed generation, centralized LFP BESS, and shore-side CI connections within a behind-the-meter microgrid topology. A SEMS governs dispatch according to a five-priority logic.

• Direct self-consumption: PV output is allocated to instantaneous demand.
• Storage buffering: Surplus PV beyond instantaneous demand charges the BESS, subject to SoC upper limits.
• Grid export: Excess generation beyond BESS absorption capacity is exported to the distribution grid.
• Peak shaving and grid support: BESS discharges to reduce grid imports during demand peaks, providing local grid stability services.
• Grid import: Residual demand is supplied by the distribution grid.

This dispatch hierarchy is functionally analogous to the demand-side management strategies employed in smart grid contexts with DG, storage, and EV charging. The SEMS monitors PV production, BESS state of charge, vessel connection status, and grid conditions in real time, paralleling the monitoring functions of AMI in distribution networks. The data acquisition layer comprises revenue-grade smart meters at the grid connection point, generation meters at each PV inverter string, bidirectional meters at the BESS power conversion system, and dedicated shore connection meters at each berth. These metering points provide the 1-s resolution data required for real-time dispatch decisions and the 15-min interval data required for grid billing and energy community benefit allocation.

The SEMS incorporates grid support functions that extend beyond energy management to address power quality. During periods of high PV generation and low demand, the BESS absorbs surplus power not only to store energy for later dispatch but also to prevent voltage rise at the medium-voltage bus. During vessel connection transients, the BESS can provide instantaneous power injection to compensate for the step-load demand, reducing voltage sag at neighboring connection points. These grid support functions are consistent with the ancillary service capabilities that distribution system operators increasingly require from behind-the-meter storage installations in smart grid architectures.

The system operates under a “prosumer” paradigm for port facility loads and a “service provider” paradigm for CI loads. In the prosumer paradigm, the port authority both generates and consumes electricity, with surplus exported under net billing arrangements. In the service provider paradigm, shore-side electricity is sold to vessel operators at a regulated tariff, analogous to the pricing structure of public EV charging networks where the infrastructure operator recovers investment through per-kWh charges to vehicle users.

3.4 AMI data architecture for port smart energy management

The SEMS requires a structured advanced metering infrastructure to supply the data inputs for real-time dispatch and energy community accounting. The AMI architecture comprises four metering layers, each serving distinct functional requirements.

• Layer 1. Grid interface metering.

A revenue-grade IEC 62053-class power quality analyser is installed at the medium-voltage point of common coupling (PCC) with the distribution network. This meter records active/reactive power (P, Q), voltage and current waveforms, total harmonic distortion (THD), and power factor at 1-s resolution. The data stream feeds the SEMS grid interaction controller and provides the basis for monitoring compliance with the distribution system operator’s connection agreement, particularly regarding reverse power flow limits and harmonic injection thresholds. Communication follows the IEC 61850 standard for substation automation, enabling interoperability with the DSO’s supervisory control and data acquisition (SCADA) system.

• Layer 2. Generation metering

Each PV inverter string is equipped with a generation meter recording AC output power, DC input voltage and current, and inverter efficiency at 1-s resolution. Aggregate generation data are transmitted to the SEMS via Modbus TCP or IEC 61850 GOOSE messaging. These data enable the SEMS to compute instantaneous surplus or deficit relative to demand and to trigger the appropriate dispatch priority (self-consumption, storage charging, or grid export).

• Layer 3. Storage metering

The BESS power conversion system incorporates bidirectional metering of charge/discharge power, cumulative throughput (kWh), state of charge (%), cell temperature, and internal resistance. These parameters are transmitted to the SEMS at 1-s intervals. The SoC and temperature data inform the dispatch algorithm’s enforcement of the 40% minimum SoC constraint and the 1C rate limit, while the cumulative throughput feeds the degradation model for lifecycle performance estimation.

• Layer 4. Berth-side vessel metering

Each shore connection point is equipped with a dedicated smart meter recording the active power, reactive power, power factor, and harmonic spectrum of the vessel load. Connection and disconnection events are time-stamped and linked to vessel identity data from the port management information system (PMIS). This per-berth metering serves two functions: it provides the SEMS with real-time demand data for dispatch decisions, and it generates the consumption records required for per-vessel billing within the energy community’s service provider business model.

The four-layer architecture produces a combined data throughput of approximately 50–100 data points per second across all metering nodes. Data are aggregated at 15-min intervals for billing, daily intervals for KPI evaluation, and annual intervals for ISO 50001 energy performance reporting. The communication backbone uses a dedicated Ethernet/IP network within the port, with the IEC 61850 protocol stack ensuring interoperability across metering devices from different manufacturers.

3.5 Component mathematical modelling

3.5.1 Photovoltaic distributed generation module

The hourly PV power output is calculated as:

P PV t = P rated × f derate × G t G 1000 × 1 + γ × T cell t − T STC (1)

In Equation 1, P_PV(t) is the hourly PV array power output (kW) at time t; P_rated is the installed PV capacity (kWp); f_derate is the dimensionless system derating factor that aggregates inverter, mismatch, soiling, and DC/AC losses (set to 0.885); G(t) is the hourly global horizontal irradiance (GHI) at the site (W/m²); G_STC = 1,000 W/m² is the reference irradiance at standard test conditions (STC); γ is the power temperature coefficient of the PV module (−0.26%/°C); T_cell(t) is the operating cell temperature (°C) at time t; and T_STC = 25 °C is the cell temperature at STC. The cell temperature T_cell(t) is estimated via the NOCT (Normal Operating Cell Temperature) method:

T cell t = T amb t + NOCT − 20 800 × G t (2)

In Equation 2, T_amb(t) is the ambient air temperature (°C) at time t recorded by the on-site meteorological station; NOCT = 45 °C is the Normal Operating Cell Temperature (defined as the cell temperature reached under 800 W/m² irradiance, 20 °C ambient, and 1 m/s wind); and G(t) is as defined for Equation 1. The constants 800 (W/m²) and 20 (°C) in the denominator and offset correspond to the standard NOCT reference conditions. Module efficiency is 24.8%, area requirement is 5 m²/kWp, and land-use efficiency is 24.7 W/m².

3.5.2 Battery energy storage system module

The BESS is modelled using the Kinetic Battery Model (KBM). State of charge:

S o C t = E s t o r e d t E n o m i n a l (3)

In Equation 3, SoC(t) is the dimensionless state of charge of the BESS at time t; E_stored(t) is the energy currently stored in the battery (kWh); and E_nominal is the rated nominal energy capacity of the BESS (kWh). SoC(t) is constrained between SoCmin = 40% and SoCmax = 100%. While 40% is conservatively high for LFP chemistry purely from a deep-discharge perspective, this high floor is intentionally set to preserve a substantial energy reserve margin for unscheduled vessel arrivals. Regarding the upper bound, while physical BESS controllers typically limit operational charge to 90%–95% to mitigate high-SoC ageing mechanisms, 100% is utilized in this macro-sizing model to represent the fully available nominal capacity allocated for the sizing algorithm. The BESS operates under a maximum 1C charge/discharge rate, with a round-trip efficiency η_rt = 94%. The required ESS capacity is then determined via the Max Rolling Sum Method:

E rolling w = ∑ t ∈ w max 0 , D t − P PV t (4)

E ESS = max E rolling w 1 − SoC min × η rt (5)

In Equation 4, E_rolling(w) is the cumulative net energy deficit (kWh) accumulated over the rolling time window w; D(t) is the aggregate port-side load demand (kW) at time t; and P_PV(t) is the instantaneous PV output as defined for Equation 1. The max (0, ·) operator restricts the integral to intervals in which load exceeds instantaneous PV generation, so only un-self-consumed deficits contribute. In Equation 5, E_ESS is the required nominal ESS energy capacity (kWh); max (E_rolling(w)) is the worst-case (largest) rolling deficit observed across the annual hourly simulation; SoC_min and η_rt are as defined for Equation 3; and w is the rolling-window length, calibrated to the average vessel hotelling duration. ESS specifications: LFP chemistry, 2,064 kWh unit capacity, 480 EUR/kWh, 0.072 m²/kWh footprint.

3.5.3 Degradation modelling

PV soiling is modelled as a 0.15%/day transmittance loss, capped at 30% maximum opacity, with rainfall-cleaning thresholds of 1 mm (partial cleaning) and 20 mm (full cleaning). PV aging combines a 1.0% one-off Light-Induced Degradation (LID) loss with a 0.35%/year linear degradation rate and an Arrhenius thermal-acceleration term (activation energy E_a = 0.9 eV):

G eff t = G t × 1 − α LID × 1 − β age × t × exp − E a k B × T cell t (6)

In Equation 6, G_eff(t) is the effective in-plane irradiance reaching the active layer of the PV module (W/m²) after accounting for degradation; G(t) is the incident GHI as defined for Equation 1; α_LID is the dimensionless light-induced degradation factor (0.01); β_age is the linear annual degradation coefficient (0.0035 yr⁻¹); t is the elapsed operating time of the array (years); E_a = 0.9 eV is the activation energy of the dominant thermal-degradation mechanism; k_B = 8.617 × 10⁻⁵ eV/K is the Boltzmann constant; and T_cell(t) is the cell temperature, expressed in Kelvin for the Arrhenius factor (T_cell [K] = T_cell [°C] + 273.15).

ESS degradation is modelled as a 0.02% capacity fade per equivalent full cycle (EFC), augmented by Rainflow Counting to capture mechanical fatigue effects from partial-depth cycles:

C f a d e = 0.02 % × N E F C (7)

In Equation 7, C_fade is the cumulative ESS capacity fade (%) attributable to throughput cycling, and N_EFC is the cumulative number of equivalent full cycles experienced by the BESS, where one EFC corresponds to a complete charge–discharge of the rated nominal capacity E_nominal. The 0.02%/EFC coefficient reflects manufacturer-warranted LFP cycle-life data for stationary applications. This rate closely correlates with empirical aging models reported in the academic literature for LFP chemistries under moderate C-rates and controlled thermal environments, which consistently demonstrate a 20% capacity fade over approximately 1,000 to 2,000 deep equivalent full cycles (Wang et al., 2011). The Rainflow Counting algorithm decomposes the SoC time-series into cycle pairs (amplitude, mean) and aggregates equivalent damage to the chemical and mechanical fatigue model.

3.6 Genetic algorithm optimization

The optimization minimizes a weighted objective combining LCOE and lifecycle carbon footprint:

f x = w L C O E × L C O E x + w C F × C a r b o n F o o t p r i n t x (8)

In Equation 8, f(x) is the scalarized weighted objective evaluated for the candidate decision vector x = [P_PV, E_ESS], where P_PV is the installed PV capacity (kWp) and E_ESS is the BESS energy capacity (kWh); LCOE(x) is the levelised cost of energy (EUR/kWh) computed for configuration x over the 25-year analysis horizon; CarbonFootprint(x) is the lifecycle GHG footprint (kg CO₂-eq/kWh) of configuration x; and w_LCOE = 0.6 and w_CF = 0.4 are the dimensionless objective weights, satisfying w_LCOE + w_CF = 1. The LCOE weight is dominant because investment decisions in port infrastructure are primarily governed by economic return. Because the algorithm minimizes LCOE while ensuring the system can handle stochastic, multi-megawatt maritime loads, it naturally sizes the PV array to produce substantial excess generation during periods without vessel traffic. Exporting this unavoidable surplus is not a flaw indicative of oversizing, but rather a vital secondary economic mechanism; selling this excess to the grid generates critical revenue that offsets the high capital costs of the BESS and shore connection infrastructure, thereby driving down the final LCOE. A sensitivity check at w_LCOE/w_CF = 0.5/0.5 and 0.7/0.3 yielded PV capacity changes within ±5% and LCOE changes within ±2%, confirming the robustness of this weighting scheme.

A GA was selected for this optimization due to its proven robustness in solving non-linear, multi-objective sizing problems with discrete constraints, where deterministic solvers frequently trap in local optima (Cholidis et al., 2026). The specific hyperparameters were chosen based on preliminary empirical tuning to balance computational efficiency with global exploration: The GA employs a population size of 100 individuals, tournament selection, single-point crossover (probability = 0.85), and adaptive mutation. The initial mutation rate is set to 0.01 and is adaptively scaled according to σ = σ₀ (1 - g/G)^0.5, where g is the current generation and G is the maximum number of generations (set to 500). To prevent premature convergence, an adaptive restart mechanism replaces 80% of the population when the best fitness fails to improve for 50 consecutive generations. Under these parameters, all scenarios converged within 150–200 generations.

Each candidate solution is evaluated through a continuous, chronological annual simulation at a 1-h time step, resulting in 8,760 sequential intervals. Within this framework, the BESS State of Charge (SoC) is managed as a continuous state variable that carries over from day to day across the entire year. This continuous integration ensures realistic charge sustainability and inter-day energy shifting, avoiding artificial daily resets, while focusing strictly on energy balances rather than sub-hourly electrical dynamics. Crucially, by evaluating the full 8,760-h continuous meteorological and operational dataset rather than an aggregated ‘average day’ profile, the optimization framework explicitly captures extreme weather variations, sudden cloud cover, and seasonal production deficits. During periods of low solar availability or peak winter demand, the smart energy management system ensures absolute load reliability by defaulting to grid imports, which are fully accounted for in the system’s LCOE and carbon footprint calculations. Fitness function:

f i t n e s s = 1 1 + f x + p e n a l t y (9)

Unlike standard DER optimization models that minimize cost in a theoretical vacuum, this penalty-driven fitness landscape ensures the algorithm only converges on solutions that satisfy the strict ISO 50001 and ISO 37101 quality constraints, guaranteeing that the mathematical optimum is also institutionally deployable. In Equation 9, fitness is the dimensionless GA fitness score, mapped to the interval (0, 1] with the value one corresponding to a hypothetical zero-cost zero-emission optimum; f(x) is the weighted objective from Equation 8; and penalty is a non-negative penalty term that enforces the minimum renewable-penetration threshold, the SoC_min/SoC_max bounds, and the grid-export limit, computed as 10× the maximum unconstrained objective value for any single constraint violation. The transformation 1/(1 + f(x) + penalty) maps a constrained minimization problem onto a maximization-friendly fitness landscape suitable for the GA tournament-selection operator.

The lifecycle CF calculation in Equation 11 is governed by the grid emission factor (GEF). This factor is calculated as the ratio of total direct greenhouse gas emissions from national gross electricity production to the total gross electricity consumption (including grid losses and net imports). For this analysis, we utilize the average Italian grid emission factor of 0.233 kg CO2-eq/kWh, derived from the latest available national electricity mix data published by the Italian Institute for Environmental Protection and Research (Change, 2026).

3.7 Scenario design

The preliminary feasibility assessment (Section 4.1) established that a monolithic electrification strategy, sizing a single PV-BESS system against the aggregated 53.7 GWh port demand, is spatially and economically infeasible, requiring over 43 MWp of installed capacity and more than 213,000 m² of surface area that exceeds the port boundary. This finding necessitated a clustered, dock-by-dock sizing strategy in which the CI demand is disaggregated into discrete load clusters based on energy consumption intensity. Analysis of historical vessel traffic and berth-level power requirements identified three distinct maritime demand clusters, low, major, and full traffic CI docks, alongside the stationary port facilities baseload. The intersection of these four demand profiles with three progressive levels of technological integration produces a 4 × 3 scenario matrix comprising twelve distinct configurations.

The four demand profiles are defined as follows.

• Use Case A–Port Facilities Prosumer (Baseline Decarbonisation):

Addresses the continuous, predictable baseload of stationary infrastructure including administrative buildings, terminal lighting, and reefer racks (∼1.7 GWh/year, ∼550 kW peak, no CI). The port authority operates as a standard prosumer, with revenue deriving from avoided grid tariffs and surplus injection. This use case serves as the low-risk entry point for the energy community, validating the operational baseline before maritime loads are introduced.

• Use Case B–Targeted Maritime Electrification (Low-Traffic CI Docks):

Extends the system to serve CI loads at low-consumption docks, predominantly servicing Ro-Ro cargo vessels with extended berthing windows (∼21.3 GWh/year, 16 MW shore connection). The port transitions from prosumer to energy service provider, generating revenue from green tariffs charged to shipowners. This configuration targets the operational regime where the temporal alignment between solar generation and vessel demand is most favourable, yielding the deepest carbon reduction per unit of installed capacity.

• Use Case C–High-Intensity Maritime Electrification (Major CI Docks):

Targets high-consumption docks servicing a diverse mix of container ships and high-frequency Ro-Ro traffic (∼28.6 GWh/year, 22 MW shore connection). This use case introduces substantially greater operational complexity, requiring sophisticated SEMS scheduling to manage the temporal mismatch between PV generation peaks and stochastic vessel arrivals. It stress-tests system flexibility and exposes the decarbonisation ceiling where storage capacity begins to saturate against volatile maritime loads.

• Use Case D–Full CI Electrification (All CI Docks):

Represents the maximum electrification boundary, encompassing all designated CI docks for a 100% maritime decarbonisation target (∼52 GWh/year, 24 MW shore connection). This use case demonstrates the green economy of scale, achieving the lowest levelised cost through utility-scale deployment, but confirms that full neutrality requires remote generation arrangements due to spatial requirements (151,000 + m²) that exceed the port perimeter.

Each demand profile is assessed under three technological configurations: (i) a grid-only baseline representing the business-as-usual reference for both cost and emissions; (ii) grid + PV, which introduces the solar array to test the capacity of intermittent renewable generation to offset grid reliance without dispatchable backup (Sifakis, 2026); and (iii) grid + PV + BESS, which incorporates energy storage to provide peak shaving, energy shifting, and a defined window of operational autonomy. The grid-only configuration anchors the economic and environmental comparison by establishing the marginal grid tariff and the unmitigated carbon footprint for each demand scale.

The transition from Use Case A to D follows a logical deployment sequencing: the low-risk facility baseload (A) is prioritised to validate the digital monitoring and energy management infrastructure before expanding into targeted (B), major (C), and eventually full (D) maritime electrification (Figure 1). This stepwise approach allows the port energy community to manage capital requirements incrementally while scaling SEMS capabilities in tandem with physical infrastructure expansion, ensuring that operational experience from earlier phases directly informs the management of more volatile, high-intensity maritime loads.

The diagram details the evolution of the required PV and BESS infrastructure, the shifting on-site and grid power flows, and the institutional transition from a facilities prosumer model to a port-wide energy service provider.

3.8 Quality assessment framework

The quality assessment framework translates the normative requirements of ISO 50001 (energy management systems) and ISO 37101 (sustainable development in communities) into quantitative decision criteria that are embedded directly within the optimisation objective. Consistent with the ISO 50001 PDCA Check phase, each scenario is evaluated against formal quality thresholds organised into three assessment domains: energy performance, operational reliability, and economic viability. These thresholds serve a dual function: they provide the port energy community with a contractual baseline for evaluating system adequacy during the pilot phase, and they establish the benchmark against which follower ports calibrate their own deployments, ensuring methodological consistency across the replication framework.

The economic and environmental assessment computes four primary indicators. LCOE:

L C O E = C A P E X + Σ O P E X t 1 + r t Σ E d e l i v e r e d , t 1 + r t (10)

In Equation 10, LCOE is the levelised cost of energy of the PV-BESS-CI configuration (EUR/kWh); CAPEX is the total capital expenditure incurred at year 0 (EUR), aggregating PV, BESS, balance-of-system, and shore-connection investment; OPEX_t is the operational and maintenance expenditure in year t (EUR), including replacement reserves; E_delivered,t is the useful electrical energy delivered to port loads and shore connections in year t (kWh); r is the dimensionless discount rate applied for net-present-value normalisation; and t indexes the operating year over the project lifetime (t = 1, … , T, with T = 25 years). This 25-year horizon aligns with standard PV performance warranties; to account for the shorter lifespan of the storage components, a complete BESS replacement is financially modelled at Year 12, as detailed in the economic assumptions (Section 3.9).

Carbon footprint:

C F = ∑ t E g r i d t × G E F (11)

In Equation 11, CF is the annual operational carbon footprint of the configuration (kg CO₂-eq), summed over all hourly intervals t in the simulation year; E_grid(t) is the electrical energy imported from the public grid at hour t (kWh) — i.e., the residual demand not met by on-site PV generation or BESS discharge; and GEF is the grid emission factor of the host network (kg CO₂-eq/kWh), set to the Italian average value of 0.233 kg CO₂-eq/kWh derived from the national electricity mix (ISPRA, 2023).

Consistent with the ISO 50001 PDCA Check phase, each scenario is evaluated against quality thresholds derived using a conservative engineering approach: the minimum acceptable value is intentionally set below the simulated optimum to account for real-world deviations and uncertainties, whilst remaining sufficiently ambitious to ensure meaningful environmental and economic outcomes. The quality criteria span the following three domains.

• Energy performance criteria evaluate whether the PV and BESS infrastructure converts and delivers energy at rates consistent with the design intent. These include minimum PV performance ratio (≥75%), self-consumption ratio thresholds differentiated by sector (≥30% for facilities, ≥20% for CI to reflect the temporal mismatch between solar generation and vessel arrivals), annual specific yield (≥1,100 kWh/kWp, site-calibrated for the Adriatic climate), battery round-trip efficiency (≥88% for LFP systems), and carbon footprint reduction (≥25% for PV-only configurations, ≥40% for PV + BESS, aligned with the ISO 37101 environmental preservation purpose).
• Operational reliability criteria assess whether the system delivers energy when needed and maintains structural integrity under stress. These include system availability (≥95% over any 12-month rolling window), battery state-of-health at end-of-period (≥80%, the industry-standard end-of-life threshold for LFP systems), grid self-sufficiency hours (≥1,500 h/year for facilities, ≥800 h/year for CI configurations), peak demand reduction (≥10% relative to the grid-only baseline, a prerequisite for DSO interconnection approval), and ramp volatility reduction (measurable decrease in the standard deviation of hour-to-hour power fluctuations injected into the grid).
• Economic viability criteria determine whether the investment generates sufficient financial return to sustain the system over its operational lifetime. These include: LCOE below 0.35 EUR/kWh for facilities and strictly below the grid-only tariff for CI configurations (ISO 50001 energy performance improvement target); simple payback period not exceeding 8 years, aligned with EU infrastructure funding cycles (ISO 37101 responsible resource use purpose); internal rate of return at minimum 10%, set conservatively below simulated results of 20%–47% (energy community financial viability threshold); and positive net present value over the 25-year project lifetime at a 5% discount rate.

Taken together, these fifteen quality criteria translate the normative requirements of ISO 50001 and ISO 37101 into a multi-dimensional decision framework for the optimisation. A scenario passes the quality gate only when all applicable thresholds are satisfied simultaneously, preventing single-metric optimisation from compromising overall system adequacy. The differentiation of thresholds by use case (facilities versus CI) acknowledges the fundamentally different operational regimes: the predictable baseload of stationary infrastructure versus the stochastic, high-power demand of maritime vessel electrification.

3.9 Technology readiness level assessment methodology

TRL assessment follows the European Commission nine-level scale (EU Funding and Tenders Portal) and is applied at three distinct assessment levels to capture the full spectrum of deployment readiness, from individual hardware to institutional arrangements. This disaggregation is motivated by the observation that port energy system deployment is typically constrained not by component maturity but by the integration gap between proven hardware and the organisational structures required to operate it. Each level employs specific evidence criteria:

Component TRL assesses the maturity of individual hardware elements, PV modules, BESS units, shore connection systems, and frequency conversion equipment, against documented commercial deployments. This high maturity level is evidenced by various existing CI installations across the EU and globally (such as in the Baltic Sea and Norway), some of which already extend beyond standard hoteling loads to include the high-power megawatt charging of fully electric ferries. The evidence basis includes the number of operational installations in comparable port environments, conformance with applicable standards (IEC 80005–1 for high-voltage shore connection, IEC 62619 for LFP battery safety), and demonstrated operational lifetimes under relevant environmental conditions. Components that have been commercially deployed at more than ten ports and have accumulated multiple years of operational data are classified at TRL 9. Components with commercial deployment but limited port-specific track records are classified at TRL 8.

System TRL assesses the maturity of the integrated PV-BESS-CI system with coordinated SEMS, evaluated against the degree of system-level validation in operational environments. The critical distinction from component TRL is that system-level readiness depends on demonstrated interoperability: the capacity of the SEMS to dispatch storage in response to real-time vessel arrivals, coordinate bidirectional grid flows under varying demand conditions, and execute the five-priority dispatch hierarchy (self-consumption, storage charging, grid export, peak shaving, grid import) under operational constraints. A system validated through simulation against representative demand profiles but not yet deployed in a physical port environment is classified at TRL 6–7. This classification corresponds to TRL 6 under the European Commission definition: “technology demonstrated in relevant environment through representative model or prototype”. In this study, the 8,760-step annual simulation, utilizing site-specific hourly meteorological data and historical vessel traffic profiles, serves as the high-fidelity representative model. This methodology provides a robust assessment of long-term energy balances and economic viability, representing the transition point between lab-scale validation and full operational deployment. Subsystems at earlier stages of development, such as predictive SEMS with vessel scheduling integration (TRL 4–5) or vessel-to-grid bidirectional power exchange (TRL 2–3), are assessed independently, as their maturity constrains the overall system readiness.

Governance TRL assesses the maturity of the energy community governance model, organisational structure, benefit-sharing protocols, regulatory compliance, and multi-stakeholder coordination mechanisms, against documented implementations in comparable infrastructure contexts. This level is anchored in the institutional requirements of EU Directive 2019/944 for citizen energy communities and the national transposition frameworks governing collective self-consumption. The evidence base includes: the existence of operational energy communities in other sectors (residential, industrial), the availability of standardised legal templates for port-specific multi-stakeholder arrangements, the maturity of benefit-sharing protocols that accommodate both permanent members (port authority, terminal operators) and transient participants (vessel operators with intermittent membership), and the degree of regulatory precedent for bidirectional grid connection agreements at the medium-voltage level. ISO 50001 certification for port-level distributed generation and storage is classified at TRL 7–8 where certified organisations exist but not for community-managed generation, whilst ISO 37101 application to port communities remains at TRL 4–5, with municipal adoption documented but port-specific piloting pending.

This three-level approach reflects the collaborative innovation perspective emphasised in European energy transition policy, where technology deployment depends not only on hardware maturity but also on organisational and institutional readiness. The framework is designed to identify the binding constraint on deployment: where component TRL is high but governance TRL is low, the critical path runs through regulatory and institutional development rather than hardware procurement. Conversely, where governance readiness outpaces system integration, the critical path lies in pilot demonstration of the coordinated PV-BESS-CI-SEMS architecture under real operational conditions.

3.10 Limitations and assumptions

The simulation relies on a single year of hourly meteorological data (2023, NASA POWER API satellite-derived). Ground-level irradiance measurements were not available. The CI demand profile is a static projection from historical vessel traffic. The 1-h temporal resolution smooths sub-hourly transients. While power quality phenomena, such as step-load transients and harmonic distortion, are fundamental to the operational reality of cold ironing, the simulation framework developed in this study functions exclusively as a macro-level capacity sizing model. Real-world vessel connection dynamics, which unfold over sub-second intervals, are beyond the scope of this 1-h resolution energy balance approach. Consequently, detailed transient stability modeling represents a distinct, subsequent phase of engineering for the pilot-scale deployment. While sufficient for long-term capacity sizing and LCOE estimation (supporting the TRL 6–7 assessment as a “representative model”), these dynamics represent the primary engineering challenge for the pilot-scale deployment (TRL 7–8) at the Port of Ancona, where real-time SEMS dispatch will be required to manage live vessel arrivals. PV and BESS degradation are applied as post-simulation corrections. Grid electricity costs and emission factors are treated as constant over the system lifetime. For configurations exceeding 30 MWp, the required PV area (150,000 + m2) exceeds available in-boundary surfaces, necessitating remote generation arrangements not assessed in detail.

Furthermore, the economic model operates under the assumption that the distribution grid can accommodate the simulated import and export capacities; consequently, potential grid reinforcement and infrastructure upgrade costs are excluded from the current LCOE calculation.

The following economic assumptions, benchmarked against European and international standards, underpin the financial modeling.

• Discount rate: 5%, aligned with the European Commission (2021) Guide to Cost-Benefit Analysis for EU-funded infrastructure.
• PV system lifetime: 25 years, consistent with IEC 61215 standards and manufacturer performance warranties.
• BESS replacement: Scheduled at Year 12, based on reaching the 80% State-of-Health threshold for LFP chemistry at the simulated cycling rates (IRENA, 2023).
• PV O&M costs: 1.5% of CAPEX/year, following IEA-PVPS Task 13 (2024) benchmarks.
• BESS O&M costs: 0.5% of CAPEX/year, as established by IRENA (2023) for stationary storage.
• CI infrastructure O&M: 2.0% of CAPEX/year, based on IEC 80005 operator guidelines.

A methodological limitation of the environmental analysis is the use of a static average rather than a dynamic (or marginal) hourly grid emission factor. While marginal factors in the Italian grid can be 2–3 times higher during peak demand hours, potentially yielding significantly higher reported carbon benefits for BESS-coordinated peak shaving, the average factor was maintained to ensure consistency with the constant-parameter optimization framework. The reported 52%–56% carbon footprint reductions should therefore be viewed as a conservative lower bound of the system’s actual environmental impact.

4 Results

This section presents the optimization results, energy balance analysis, quality compliance assessment, and TRL evaluation.

4.1 Preliminary feasibility assessment

A preliminary assessment established that the port facilities demand (∼1.7 GWh/year) requires 970 kWp of PV, spatially and economically feasible. The CI demand (∼52 GWh/year) requires 42,600 kWp (213,000 m²), exceeding port boundaries. The aggregated demand requires 43,550 kWp. These results confirmed that monolithic port-level sizing is infeasible and motivated the multi-scenario, dock-by-dock strategy.

4.2 Multi-scenario optimization results

Table 1 presents the complete results across all twelve scenarios. The corresponding annual PV production profiles are shown in Figure 2a (facilities-scale array, 1,050 kWp) and Figure 2b (full-CI array, 30,200 kWp).

TABLE 1. Multi-scenario optimization results for the Port of Ancona.

TABLE 2. Incremental impact of PV and BESS integration across use cases.

TABLE 3. Use case synthesis with quality criteria validation.

4.2.1 Port facilities (S1-S3)

The grid-only baseline (S1) establishes a reference LCOE of 0.360 EUR/kWh, reflecting the Italian commercial electricity tariff for medium-voltage industrial consumers, and a carbon footprint defined by the full grid import of 1,748,503 kWh/year. This baseline provides the energy performance reference point required under the ISO 50001 Plan phase. The addition of 1,050 kWp of distributed PV generation (S2) reduces the LCOE to 0.279 EUR/kWh (22.5% reduction) with a carbon footprint reduction of 28.47%. Grid imports decrease to 1,020,028 kWh/year (a 41.7% reduction from the baseline), while grid exports reach 1,325,156 kWh/year. The export volume exceeding the import reduction indicates that the GA optimizer has identified an economically optimal PV capacity that exceeds the facility’s instantaneous self-consumption capacity during peak solar hours. This surplus generation, exported to the distribution network, generates revenue that offsets additional PV capital cost, a finding consistent with the economic logic of net metering and net billing policies. The payback period of 3.31 years and IRR of 30.16% confirm the financial attractiveness of this configuration under current Italian tariff structures.

The addition of 2,064 kWh of LFP storage (S3) reduces the LCOE further to 0.238 EUR/kWh and raises the carbon footprint reduction to 51.78%. The mechanism is quantifiable in the energy balance: grid exports decrease from 1,325,156 to 883,469 kWh/year (a 33.4% reduction), indicating that the BESS absorbs 441,687 kWh/year of surplus PV generation that would otherwise have been exported. This stored energy is discharged during evening and overnight hours, reducing grid imports from 1,020,028 to 807,987 kWh/year (a 20.8% further reduction). From a grid interaction perspective, the BESS smooths the port’s net load profile, reducing both the peak export power during midday and the peak import power during evening demand. This load flattening provides implicit grid support benefits analogous to those valued in smart grid demand-side management programmes.

4.2.2 Low-intensity CI (S4-S6)

The CI grid-only baseline (S4) exhibits an LCOE of 0.551 EUR/kWh, substantially higher than the facilities baseline. The elevated cost reflects the amortization of IEC 80005–1 compliant shore connection infrastructure (frequency conversion, cable management, safety interlocking) and the dedicated 16 MW medium-voltage grid connection required for CI. From a power systems perspective, this 16 MW connection represents a significant point load on the distribution network, comparable in magnitude to a medium-sized industrial park or a major EV charging hub. PV integration alone (S5) achieves a 45.0% LCOE reduction to 0.303 EUR/kWh with a 2.10-year payback and 47.54% IRR, the strongest economic case of any scenario. The 12,900 kWp PV array generates approximately 16.9 GWh/year, displacing 10.2 GWh of grid imports and exporting 12.55 GWh. The high IRR reflects the large cost differential between PV-generated electricity and the elevated CI grid tariff.

The PV + BESS configuration (S6) achieves the highest carbon footprint reduction of any scenario: 56.04%, at an LCOE of 0.267 EUR/kWh and a payback of 3.08 years. As depicted in Figure 2c and in detail in Figure 4, the SEMS dispatch reveals a structured daily cycle: battery charging during 08:00–11:00 at approximately 4.5 MWh/hour average, and discharge during 17:00–20:00, effectively shifting solar generation to cover the evening vessel demand transition. The ESS-to-demand ratio of 0.058% enables meaningful daily cycling, and the near-symmetry of the charge/discharge profile indicates high storage utilization. Grid reliance during overnight hours (00:00–06:00) is reduced to 0.2–0.5 MWh/hour, demonstrating that the targeted, dock-specific sizing achieves a degree of grid independence that the full electrification scenario cannot match.

4.2.3 Major vessel CI (S7-S9)

The highest grid-only baseline (0.556 EUR/kWh) is reduced to 0.299 EUR/kWh with PV + BESS. The GA reduces PV capacity from 17,250 kWp (S8) to 17,100 kWp (S9) when ESS is available, substituting storage for marginal generation capacity and reducing curtailment.

4.2.4 Full port electrification (S10-S12)

The grid-only baseline (S10) exhibits an LCOE of 0.482 EUR/kWh, lower than the dock-specific CI baselines (S4: 0.551, S7: 0.556) because the fixed shore connection and grid infrastructure costs are amortized across the largest energy volume (51,954,296 kWh/year). This cost dilution through economies of scale is the primary mechanism driving the superior economic performance of the full electrification configuration. PV integration alone (S11) with 30,250 kWp achieves an LCOE of 0.240 EUR/kWh (50.2% reduction) and a carbon footprint reduction of 45.27%. The PV array requires 151,250 m2 of installation area (approximately 15 hectares), establishing the spatial ceiling for in-boundary DG deployment. Grid imports decrease from 51,954,296 to 26,454,838 kWh/year (49.1% reduction), while grid exports reach 29,902,773 kWh/year, the highest absolute export volume of any scenario. This substantial bidirectional power flow has implications for the port’s grid connection agreement and underlines the need for coordinated planning with the distribution system operator.

The full PV + BESS configuration (S12) achieves the lowest LCOE of any scenario: 0.211 EUR/kWh, with a payback of 2.77 years, an IRR of 36.12%, and a carbon footprint reduction of 52.48%. As shown in Figure 2d and in detail in Figure 3, HRES production peaks at approximately 20 MWh/hour at midday against an average hourly demand of 7-9 MWh/hour. The 18,576 kWh BESS handles approximately 8-10 MWh/day throughput, but the overnight grid reliance of 1-4 MWh/hour persists due to an ESS-to-demand ratio of only 0.036%. From a smart grid perspective, the persistent overnight grid draw of 22,812 MWh/year (44% of baseline) constitutes the irreducible grid dependence under a PV-only DG topology, and its reduction would require either complementary generation sources (wind, marine) or substantially larger storage capacity.

4.3 Energy balance analysis

A comparative examination of the optimal CI configurations S6 and S12 quantifies the scale-dependent ESS effectiveness. Figure 3 presents the average daily energy balance for the full port electrification scenario (S12), and Figure 4 presents the corresponding profile for the low-intensity CI scenario (S6). Each figure displays the hourly average of five energy components: HRES (PV) production, grid energy usage, battery charging, battery discharging, and load (energy demand).

Figure 3 details the cyclical generation patterns versus the variable daily vessel traffic. The green shaded area represents average HRES (PV) production, the blue line represents the CI demand load, and the orange and purple traces indicate battery charging and discharging events, respectively.

The diurnal PV generation cycle (sunrise to sunset) is clearly visible, with production peaks reaching approximately 15–20 MWh during midday hours. Demand exhibits a weekly pattern driven by ferry schedules, with pronounced peaks on weekdays and reduced activity over weekends. Battery charging concentrates during late-morning solar surplus periods, while discharge occurs during the evening demand transition.

The profile (Figure 4) exhibits the same structural characteristics as S12 but at a reduced scale, with PV production peaks of approximately 6-8 MWh. The demand-to-generation ratio is more favourable at this scale: PV output frequently exceeds CI demand during midday hours on all days of the week, resulting in consistent battery charging opportunities. Battery discharge events are more uniformly distributed across the evening and night-time hours compared to S12, indicating more effective temporal shifting of solar generation. The Friday-to-Monday ferry traffic pattern is visible in the demand trace, with Saturday representing the lowest demand period of the week.

Figure 5 presents the side-by-side comparison of both configurations for direct visual assessment of the scale difference.

Since Figure 5 illustrates an annual average profile, trace amounts of grid import appear concurrently with peak midday PV generation. This visual overlap reflects seasonal climatic variations: while the system exports significant surplus power at midday during clear summer months, it relies on grid imports during the overcast midday hours of winter. Consequently, the aggregation of 365 days into a single 24-h profile naturally superimposes these distinct seasonal operational states.

The comparison reveals the ESS saturation effect: in S6, the battery charge/discharge volumes are approximately proportional to daily demand, while in S12, the battery contribution is disproportionately small relative to the increased demand volume. In S6, the ratio of daily BESS throughput to daily demand is approximately 0.35, meaning the battery handles roughly one-third of average daily energy requirements. The charge/discharge cycle exhibits near-symmetry, indicating high utilization. In S12, this ratio drops to approximately 0.12, confirming ESS saturation: the 18,576 kWh battery, though 50% larger than in S6, operates against a 2.4 times larger demand volume. The overnight grid reliance in S12 (1-4 MWh/hour) is approximately four times that of S6 (0.2–0.5 MWh/hour), constituting the primary remaining carbon emission source.

The weekly operational pattern reveals a Friday-Monday demand cycle driven by ferry schedules, with Saturday troughs permitting deeper battery cycling. This pattern has implications for the SEMS: predictive scheduling algorithms incorporating vessel arrival data (available from port management systems) could pre-position battery SoC to optimize weekend coverage.

4.4 Technology integration impact

PV distributed generation delivers the largest reductions across all scales (Table 2). BESS CF contribution diminishes from 23.31 percentage points (UC-A, ESS-to-demand ratio 0.121%) to 7.21 percentage points (UC-D, ratio 0.036%). BESS LCOE contribution is more stable (11.3%–14.7%), as the economic value of peak shaving and grid arbitrage scales proportionally with system size.

4.5 Quality compliance assessment under ISO 50001 and ISO 37101

All four use cases satisfy all quality criteria (Table 3). The ISO, 50001 energy performance improvement target (LCOE <0.35 EUR/kWh) is met with margins ranging from 14.6% (UC-C) to 39.7% (UC-D). The ISO, 37101 environmental preservation target (CF, reduction >40%) is exceeded by 11.17–16.04 percentage points. Financial viability thresholds (payback <8 years, IRR >10%) are satisfied with substantial margins.

A scale-efficiency trade-off is apparent. UC-D achieves the lowest LCOE (0.211 EUR/kWh) but a CF reduction (52.48%) that is 3.56 percentage points lower than UC-B (56.04%). Port energy communities can use this trade-off to calibrate their investment strategy according to their prioritized ISO 37101 sustainability purpose: responsible resource use (minimize cost, favour UC-D) or environmental preservation (maximize decarbonization, favour UC-B).

4.6 Technology readiness assessment

The TRL assessment (Table 4) reveals that the principal deployment bottleneck is not component maturity (TRL 8–9) but system integration (TRL 6–7) and governance readiness (TRL 4–6). The technical validation gap can be addressed through pilot deployment at the Port of Ancona under the AIMPRESS project, which would advance the system TRL from 6-7 to 8 by demonstrating the integrated PV-BESS-CI operation with the five-priority SEMS dispatch in an operational port environment. The governance gap requires parallel regulatory work: establishment of the energy community legal entity under national transpositions of EU Directive 2019/944, development of benefit-sharing protocols that accommodate both permanent and transient community members, negotiation of grid connection agreements addressing bidirectional power flows and ancillary service provision, and formal ISO 50001 certification of the port energy management system. These governance milestones are prerequisites for replication across the ADRION region, where the 350+ registered ports represent a potential market for the framework developed in this study (Papadaki et al., 2025).

4.7 Comparative analysis with existing literature

Table 5 positions the results of this study relative to published findings for port energy system optimization with renewable integration and CI. Several observations emerge from the comparison. First, the carbon footprint reductions achieved in this study (52%–56%) are lower than those reported by Colarossi and Principi, who found 87.4% CO2 reduction at the same Port of Ancona using a Life Cycle Cost optimization (Colarossi and Principi, 2023).

TABLE 4. Three-level TRL assessment.

TABLE 5. Comparative analysis with existing literature.

This discrepancy is attributable to methodological differences: their study compared shore-side electricity directly against on-board diesel generator emissions, while the present study measures CF reduction relative to the grid-only scenario (which already uses the Italian electricity mix with its own carbon intensity). The two metrics are complementary rather than contradictory: this study quantifies the incremental benefit of adding DG and storage to shore power, while their study quantifies the total benefit of shore power itself.

Second, the LCOE values achieved here (0.211–0.267 EUR/kWh) are higher than those reported for multi-RES port systems without CI (0.080–0.129 EUR/kWh, Sifakis et al., 2021), reflecting the additional amortized cost of CI infrastructure (IEC 80005 shore connections, frequency converters, dedicated grid connections). When CI infrastructure costs are excluded, the PV-BESS generation cost alone is competitive with the multi-RES values.

Third, the payback periods achieved in this study (2.10–4.79 years) are superior to those reported by Yuksel et al. (2025) for PV/battery systems (6.98–8.78 years) and substantially faster than those of hydrogen-based systems (Yuksel et al., 2025). This favourable payback reflects both the maturity and declining costs of PV and LFP technology relative to hydrogen electrolysis and fuel cell systems, and the high baseline CI electricity costs that amplify the return on DG investment.

Furthermore, while existing models excel at multi-agent operational scheduling and flexible energy management (Barone et al., 2025; Dash et al., 2025), the GA optimization framework developed in this study performs ‘better’ from a deployment readiness perspective. By embedding ISO 50001 and ISO 37101 thresholds directly into the fitness penalty function, the proposed model guarantees that the resulting infrastructure sizing strictly complies with the governance requirements of a Port Energy Community, bridging the gap between theoretical optima and actionable policy.

Fourth, the energy community dimension addressed in this study has only one direct precedent in the port energy literature: Buonomano et al., who modelled renewable energy communities in Mediterranean ports and reported payback periods of 3 years and lifecycle savings of EUR 6 million. Their study, however, did not address CI demand or PV-BESS capacity optimization, and it did not integrate ISO 50001/37,101 quality governance or TRL assessment. Abu Bakar et al. examined CI “cooperative” with two-stage optimization but focused on operational scheduling rather than capacity sizing or institutional governance. The present study is, to the authors’ knowledge, the first to integrate GA-based DG-storage sizing for CI within a collaborative energy community framework governed by ISO quality and sustainability standards.

4.8 Sensitivity analysis

To assess the robustness of the optimization results against uncertain input parameters, a local (one-at-a-time) sensitivity analysis was conducted. This analysis focuses specifically on Scenario S12 (full port electrification with PV and BESS), as it represents the most comprehensive and capital-intensive configuration, making it the ideal baseline to stress-test system vulnerabilities. Four key techno-economic drivers were varied around their S12 baseline values to evaluate their impact on the LCOE, lifecycle CF reduction, and optimal system sizing. The results are summarized below.

• Discount Rate (3% to 7%): Variations in the discount rate resulted in an LCOE fluctuation of ±4.2% from the S12 baseline. Notably, the optimal PV capacity remained unchanged, indicating that the sizing algorithm’s preference for maximum solar generation is robust across standard financing conditions.
• BESS Capital Cost (350 to 600 EUR/kWh): Varying the cost of storage yielded an LCOE variance of ±6.1%. Furthermore, BESS capital cost emerged as the dominant driver for system sizing; at the upper bound constraint (600 EUR/kWh), the optimizer reduced the optimal BESS capacity by approximately 8% to maintain economic viability.
• Grid Electricity Tariff (±20% from baseline): The grid tariff proved to be the most highly sensitive economic driver. A ±20% variation caused the LCOE to fluctuate by ±8.3% and resulted in a broad Internal Rate of Return (IRR) range of 28%–44% for the S12 configuration.
• PV Derating Factor (0.80 to 0.92): To account for operational uncertainties, the PV derating factor was varied, resulting in a ±3.1% variance in carbon footprint reduction and a ±2.4% variance in LCOE.

Overall, the sensitivity analysis confirms the stability of the proposed framework. While the grid tariff dictates the baseline economic performance and BESS capital cost heavily influences optimal storage sizing, the environmental benefits (carbon reduction) remain highly resilient to economic parameter variations, being primarily governed by the physical PV capacity. These findings reinforce the core conclusions drawn from the baseline optimization. The relative sensitivity of the LCOE to these parameter variations for Scenario S12 is illustrated in the tornado diagram in Figure 6.

4.9 Business model, implementation, and study replication pathway

For port facility loads (UC-A), a prosumer model applies: the port authority invests, consumes, and exports surplus. For CI loads (UC-B-D), a service provider model is required, with energy sold to vessel operators at tariffs competitive with on-board generation (∼0.25–0.35 EUR/kWh; Zis, 2019). The achieved LCOEs of 0.211–0.267 EUR/kWh are within this competitive range. The energy community governance model facilitates both paradigms through a special purpose vehicle or cooperative. Revenue is distributed by consumption and capital contribution. The ISO 37101-aligned governance structure provides the transparent accountability framework needed for multi-stakeholder investment decisions.

A phased implementation pathway is recommended.

• Phase 1 (years 1–2): Deploy PV + BESS for port facilities (UC-A); validate SEMS dispatch logic; establish energy community legal entity; initiate ISO 50001 certification.
• Phase 2 (years 2–4): Extend to low-intensity CI docks (UC-B); activate service provider model; install smart metering infrastructure; begin ISO 37101 community sustainability assessment.
• Phase 3 (years 4–7): Scale to major vessel berths (UC-C); evaluate remote generation hubs; negotiate grid reinforcement with DSO.
• Phase 4 (years 7+): Full port electrification (UC-D), contingent on spatial constraints, grid capacity, and regulatory approvals.

Furthermore, to facilitate the replication of this study’s results, the following computational replication pathway must be adhered to.

• Data Acquisition: Extract hourly global horizontal irradiance (GHI) and ambient temperature for the target port via an API.
• Load Profiling: Map the target port’s vessel traffic schedules and baseloads to the four CI demand clusters (Use Cases A–D) established in Section 3.6.
• System Simulation: Implement the component and degradation models (Equations 1–7) using the exact techno-economic boundary conditions and discount rates defined in Section 3.9.
• Optimization Execution: Initialize the Genetic Algorithm using the specified hyperparameters. It is critical that the objective function incorporates the penalty logic (Equation 9) so that the optimizer strictly enforces the ISO 50001 and ISO 37101 quality criteria during the sizing process.

5 Discussion

The results demonstrate that PV distributed generation and battery storage integration with CI infrastructure is technically feasible and economically attractive under the conditions examined. The ISO 50001 and ISO 37101 quality compliance is achieved with substantial margins across all configurations, validating the approach of embedding quality constraints directly within the optimization objective.

5.1 Quantitative equivalence between CI and EV charging

CI shares fundamental power systems characteristics with EV charging: variable-duration grid connections with high power demand, the need for smart scheduling to manage peak loads, and the potential for bidirectional energy exchange (V2G). To substantiate this parallel quantitatively, Table 6 presents a direct comparison of CI load characteristics with equivalent EV fleet charging scenarios at comparable aggregate power levels.

TABLE 6. Quantitative comparison of CI and equivalent EV fleet charging characteristics.

The comparison reveals that a single ferry at 8 MW imposes the same aggregate energy demand as 1,081 Level 2 EV chargers or 53 DC fast chargers, but concentrates this demand at a single point of common coupling. The grid impact is therefore qualitatively different: dispersed EV charging creates gradual, statistically smoothed load profiles across the distribution network, while CI creates step-load transients at the berth-level PCC. This concentration makes CI both more challenging (higher per-point grid impact) and more amenable to centralized mitigation (a single co-located BESS can address the entire load). The SEMS dispatch logic developed in this study, with its five-priority hierarchy (self-consumption, storage, export, peak shaving, import), maps directly onto the demand-side management taxonomies used in smart EV charging research, but operates at power levels that require medium-voltage rather than low-voltage infrastructure.

The smart metering requirements identified in this study (1-s resolution for dispatch, 15-min intervals for billing, per-berth vessel metering) are consistent with the AMI architectures being deployed in advanced distribution management systems across European smart grids. The data granularity required for effective SEMS operation in the port context exceeds that of conventional commercial metering but is well within the capabilities of current AMI technology. The integration of vessel scheduling data from port management information systems (PMIS) with the SEMS represents a domain-specific extension of the demand forecasting approaches used in smart grid contexts, where EV charging session predictions are derived from mobility patterns and user behavior analytics.

5.2 Power quality implications

The simulation results, while focused on energy balance and economic performance, carry implicit power quality implications. In S12, the PV system generates approximately 20 MWh/hour at midday against a demand of 7-9 MWh/hour, creating a surplus of 11-13 MW that must be either absorbed by the BESS (limited to 18.6 MW at 1C rate), exported to the grid, or curtailed. During periods of simultaneous high PV output and low vessel demand, the reverse power flow on the distribution feeder could reach 10–15 MW, a level that requires dedicated assessment of voltage regulation and protection system coordination. As established in the optimization results, this reverse power flow is a necessary operational reality of sizing for high-intensity, stochastic maritime loads rather than a flaw in capacity planning. While the BESS buffers a portion of this energy, the remaining surplus must be exported. This transforms an unavoidable physical consequence of intermittent vessel schedules into a critical revenue stream, validating the port’s role as a proactive energy prosumer. The BESS, by absorbing a portion of this surplus, provides inherent power quality benefits: it reduces the magnitude of reverse power flow, dampens voltage fluctuations, and provides a buffer against the step-load transients associated with vessel connection events. These ancillary benefits, while not monetized in the current LCOE calculation, represent additional value that would strengthen the economic case for BESS investment if captured through grid service revenue streams.

5.3 Spatial constraints and the case for dock-level electrification

The full port electrification scenario (S12) requires 151,000 m² for a 30.2 MWp PV array, an area that exceeds the surfaces available within port boundaries. This spatial infeasibility renders S12 a theoretical upper-bound reference case rather than a directly deployable configuration. The actionable recommendation emerging from this study is the dock-level electrification strategy represented by S6 (UC-B): at 12.9 MWp PV and 12.4 MWh BESS, the required area of 64,500 m² is within the range of available rooftop and adjacent ground-mount surfaces, the LCOE of 0.267 EUR/kWh is competitive with on-board generation, the payback of 3.08 years is commercially attractive, and the CF reduction of 56.04% exceeds that of S12 (52.48%) in per-unit decarbonization. S6 therefore represents the configuration with the highest practical viability and environmental effectiveness. Full port electrification (S12) remains relevant as a benchmark for understanding scale effects and as a long-term objective contingent on remote generation hubs (off-site solar farms, floating PV) or complementary marine renewables (wave, tidal, offshore wind), but investment and deployment planning should begin with dock-level configurations.

5.4 Energy community governance implications

The multi-stakeholder nature of port operations, involving port authorities, terminal operators, vessel operators, and grid operators, creates both a need and an opportunity for collaborative governance. The energy community model, structured under EU Directive 2019/944, provides the legal and organizational framework for joint investment, shared operation, and equitable benefit distribution. The TRL assessment (Section 4.6) reveals that governance readiness (TRL 4–6) lags behind technology readiness (TRL 8–9), indicating that institutional innovation is as critical as technological validation for deployment.

The governance gap manifests in several concrete challenges. First, the legal entity structure for a port energy community must accommodate both permanent members (port authority, terminal operators with long-term concessions) and transient participants (vessel operators connecting for hours or days). This dual membership structure has no direct precedent in existing energy community implementations, which typically involve residential or industrial members with stable, long-term participation. Second, benefit-sharing protocols must allocate the economic savings from PV-BESS integration (reduced grid costs, avoided emissions penalties under FuelEU Maritime) across stakeholders with differing investment contributions and consumption patterns. ISO 37101 provides the governance framework for structuring these allocation decisions according to transparent sustainability criteria, but the specific application to port contexts requires pilot validation. Third, the interaction between the port energy community and the distribution system operator must be formalized through connection agreements that address bidirectional power flow, grid service provision, and the technical requirements for DG and storage integration at the medium-voltage level. Regulatory sandboxes, such as those established under national implementations of the EU Clean Energy Package, could provide the institutional space for testing these governance arrangements before full-scale deployment.

5.5 ESS saturation and alternative storage pathways

The diminishing BESS effectiveness at scale (ESS-to-demand ratio declining from 0.121% to 0.036%) establishes an upper bound on the decarbonization achievable with LFP batteries alone. Green hydrogen production via electrolysis of surplus PV generation offers a complementary pathway: hydrogen storage can buffer multi-day and seasonal mismatches that exceed LFP battery capacity, and hydrogen can be re-electrified via fuel cells or sold as clean fuel to vessels.

5.6 Replication of the proposed framework

While the specific optimal capacities, LCOE values, and spatial constraints calculated in this study are inherently tied to the Port of Ancona’s unique meteorological conditions and vessel traffic profiles, the underlying methodological framework is universally generalizable. The core innovation, embedding ISO 50001 energy management targets and ISO 37101 sustainability criteria as hard constraints within a Genetic Algorithm, creates a universally replicable template for port energy communities. By following the computational replication pathway detailed in Section 4.9, any maritime port can substitute localized irradiance, grid tariff, and vessel demand datasets to generate site-specific, institutionally compliant infrastructure configurations.

6 Conclusions

This study has developed a collaborative port energy community framework integrating ISO 50001 energy management, ISO 37101 community sustainability governance, and genetic algorithm optimization for the capacity sizing of distributed PV generation and LFP battery storage supporting CI infrastructure. The framework was applied to the Port of Ancona across twelve scenarios spanning four electrification scales (1.7–52 GWh/year) and three technology configurations.

The principal quantitative findings are as follows. PV distributed generation reduces LCOE by 22.5%–50.2% and carbon footprint by 28.5%–46.2% relative to grid-only baselines across all scales. The recommended configuration for near-term deployment is the dock-level electrification scenario S6 (12.9 MWp PV, 12.4 MWh BESS, 16 MW CI capacity), which achieves the highest carbon footprint reduction of any configuration (56.04%), an LCOE of 0.267 EUR/kWh, a payback period of 3.08 years, and an IRR of 32.47%, while requiring only 64,500 m2 of PV area that is achievable within and adjacent to port boundaries. The full-electrification scenario S12 (30.2 MWp PV, 18.6 MWh BESS) achieves the lowest absolute LCOE (0.211 EUR/kWh) but requires 151,000 m², exceeding available in-boundary surfaces; it is therefore positioned as a theoretical upper-bound reference case contingent on remote generation or marine renewable integration. All four use cases pass the quality criteria derived from the ISO 50001 performance improvement framework and the ISO 37101 sustainability purposes (environmental preservation, responsible resource use). BESS effectiveness diminishes at scale (ESS-to-demand ratio declining from 0.121% to 0.036%), establishing an upper bound on the decarbonization achievable with LFP storage alone and motivating investigation of complementary technologies (hydrogen, marine renewables).

The study establishes four contributions to the “Progress and Challenges in Power Systems: Distributed Generation, Smart Metering, Energy Storage and Electric Vehicles” research programme. First, CI is formally situated as large-scale EV infrastructure, with berthed vessels drawing 2–24 MW at power levels 1,000–10,000 times those of passenger EVs, linking maritime electrification to the smart grid and EV integration discourse. The load characteristics of CI (high power, variable duration, stochastic scheduling) create grid management challenges that are qualitatively similar to those of large EV charging hubs but require dedicated planning approaches due to the concentrated nature of the demand at individual berths.

Second, the smart energy management system developed in this study, with its five-priority dispatch hierarchy and real-time metering requirements, parallels the demand-side management strategies deployed in smart grid contexts with high DG and EV penetration. The SEMS architecture demonstrates that port microgrids require the same data infrastructure (AMI, predictive scheduling, real-time SoC monitoring) that characterizes advanced distribution management systems, extending the smart metering research programme to a new application domain.

Third, the ISO 50001 PDCA cycle is operationalized as a structuring methodology for energy infrastructure optimization, demonstrating that quality-constrained optimization produces results that are simultaneously technically optimal and institutionally compliant. The quality criteria thresholds (LCOE <0.35 EUR/kWh, CF reduction >40%, payback <8 years, IRR >10%) translate the normative requirements of ISO 50001 and ISO 37101 into quantitative parameters within the GA fitness function, establishing a methodological template for quality-governed infrastructure design applicable to energy communities beyond the port sector.

Fourth, the three-level TRL assessment (component, system, governance) identifies governance readiness (TRL 4–6) rather than technology maturity (TRL 8–9) as the principal deployment bottleneck. This finding redirects attention from technology development toward institutional innovation: the establishment of port energy community legal structures, the development of benefit-sharing protocols for heterogeneous stakeholder configurations, and the regulatory frameworks needed to accommodate bidirectional power flow and grid service provision from port-based DG-storage assets.

A dock-by-dock, clustered electrification strategy starting with UC-B (S6) is recommended as the actionable deployment pathway, as it achieves the highest per-unit decarbonization (56.04%), remains within spatial constraints (64,500 m²), delivers a commercially attractive payback (3.08 years), and aligns with the phased governance maturation required for port energy community establishment. Full port electrification (S12) is positioned as the long-term upper-bound objective, contingent on the resolution of spatial constraints through remote generation or marine renewable integration. The spatial ceiling on in-boundary DG (151,000 m² for 30 MWp) is a binding physical limit that reinforces the practical superiority of the dock-level approach for near-term investment decisions.

6.1 Future research directions

• Integration of complementary marine renewables (wave, tidal, offshore wind) to exceed the 56% CF reduction ceiling identified for PV-only systems.
• Green hydrogen production from surplus PV to provide multi-day and seasonal storage beyond LFP battery capacity.
• Dynamic grid emission factors and time-of-use tariff structures within the optimization for real-time environmental and economic assessment.
• Pilot-scale validation of the integrated PV-BESS-CI system with SEMS at the Port of Ancona, advancing system TRL from 6-7 to 8-9 and generating calibration data for predictive AI-based energy management.
• Vessel-to-grid (V2G) feasibility analysis for battery-equipped vessels, leveraging CI infrastructure for bidirectional energy exchange.
• Regulatory analysis for port energy community establishment under EU Directive 2019/944 across the ADRION region, addressing transposition differences between EU member states and IPA partner countries.
• Multi-year simulation incorporating inter-annual climate variability and vessel fleet evolution for robust long-term capacity planning.
• Power quality impact assessment of large-scale CI on port distribution networks, including harmonic analysis, voltage regulation, and flicker mitigation through BESS.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. Generative AI tools were used during manuscript preparation to refine language, syntax, and wording. All scientific content, methodology, analysis, and interpretations are entirely the authors’ own work.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Glossary

AMI Advanced Metering Infrastructure BESS Battery Energy Storage System CF Carbon Footprint CI Cold Ironing DG Distributed Generation EFC Equivalent Full Cycle g Current generation in Genetic Algorithm G Maximum number of generations (500) G(t) Hourly global horizontal irradiance (W/m²) GEF Grid Emission Factor (0.233 kg CO2-eq/kWh) GSTC Irradiance at standard test conditions (1,000 W/m2) (W/m²) IRR Internal Rate of Return (%) LCOE Levelized Cost of Energy (EUR/kWh) LFP Lithium Iron Phosphate NOCT Normal Operating Cell Temperature (45 °C) (°C) PPV(t) Hourly PV power output (kW) Prated Rated/Installed PV capacity (kWp) PV Photovoltaic SEMS Smart Energy Management System STC Standard Test Conditions TRL Technology Readiness Level TSTC Temperature at standard test conditions (25 °C) (°C) σ Adaptive mutation rate — σ0 Initial mutation rate (0.01) — P _**PV** (t) Hourly PV array power output at time t (kW) P _**rated** Installed PV array capacity (kWp) f _**derate** Dimensionless system derating factor aggregating inverter, mismatch, soiling and DC/AC losses (0.885) G(t) Hourly global horizontal irradiance (GHI) at the site (W/m2) G _**STC** Reference irradiance at standard test conditions (1,000W/m2) γ Power temperature coefficient of the PV module (−0.26%/°C) T _**cell** (t) Operating cell temperature at time t (°C; K in Equation 6 for the Arrhenius factor) T _**STC** Cell temperature at STC (25°C) T _**amb** (t) Ambient air temperature at time t recorded by the on-site meteorological station (°C) NOCT Normal Operating Cell Temperature, reached at 800W/m2, 20°C ambient and 1 m/s wind (45°C) SoC(t) Dimensionless state of charge of the BESS at time t E _**stored** (t) Energy currently stored in the battery (kWh) E _**nominal** Rated nominal energy capacity of the BESS (kWh) SoC _**min** Minimum allowed state of charge (40%) SoC _**max** Maximum allowed state of charge (100%) η _**rt** Round-trip efficiency of the BESS (94%) E _**rolling** (w) Cumulative net energy deficit accumulated over the rolling time window w (kWh) D(t) Aggregate port-side load demand at time t (kW) w Rolling-window length, calibrated to the average vessel hotelling duration E _**ESS** Required nominal ESS energy capacity (kWh) G _**eff** (t) Effective in-plane irradiance reaching the active layer of the PV module after degradation (W/m2) α _**LID** Dimensionless light-induced degradation (LID) factor (0.01) β _**age** Linear annual degradation coefficient of the PV array (0.0035 yr−1) t Elapsed time index — denotes the elapsed PV operating time (years) in Equation 6, and the discounting year index (t = 1, … , T, with T = 25 years) in Equation 10 E _**a** Activation energy of the dominant thermal-degradation mechanism (0.9eV) k _**B** Boltzmann constant (8.617 × 10⁻⁵ eV/K) C _**fade** Cumulative ESS capacity fade (%) attributable to throughput cycling N _**EFC** Cumulative number of equivalent full cycles experienced by the BESS f(x) Scalarized weighted GA objective evaluated for the candidate decision vector x x Decision vector x = [PPV, EESS] sized by the GA LCOE(x) Levelised cost of energy of configuration x over the 25-year horizon (EUR/kWh); see Equation 10 for its closed-form definition CarbonFootprint(x) Lifecycle GHG footprint of configuration x (kg CO₂-eq, or normalised in kg CO₂-eq/kWh); see Equation 11 for its closed-form definition w _**LCOE** Dimensionless objective weight on LCOE (0.6) w _**CF** Dimensionless objective weight on the carbon footprint (0.4) fitness Dimensionless GA fitness score in (0, 1] penalty Non-negative constraint-violation penalty term (10× the maximum unconstrained objective per violation) LCOE Levelised cost of energy of the PV-BESS-CI configuration (EUR/kWh) CAPEX Total capital expenditure incurred at year 0 (EUR), aggregating PV, BESS, balance-of-system, and shore-connection investment OPEX _**t** Operational and maintenance expenditure in yeart (EUR), including replacement reserves E _{**delivered,t**} Useful electrical energy delivered to port loads and shore connections in yeart (kWh) r Dimensionless discount rate applied for net-present-value normalisation T Project lifetime over which the LCOE integral is evaluated (25 years) CF Annual operational carbon footprint of the configuration (kg CO₂-eq) E _**grid** (t) Electrical energy imported from the public grid at hourt — i.e., demand not met by on-site PV or BESS (kWh) GEF Grid emission factor of the host network (kg CO₂-eq/kWh; Italian average 0.233).

Footnotes

Edited by: Ehsan Najafi, University College Dublin, Ireland

Reviewed by: Andrew McGordon, University of Warwick, United Kingdom

Linfeng Li, Guilin University of Aerospace Technology, China

Shitikantha Dash, National University of Singapore, Singapore