Extending BlockSim with energy and carbon footprint modeling for sustainable blockchain evaluation

1 Introduction

Blockchain technology has become a crucial infrastructure for decentralized applications, facilitating trustless coordination within distributed systems. Following Bitcoin’s inception, blockchains have seen widespread adoption in sectors including finance, supply chain management, and digital governance. Nevertheless, the swift expansion of blockchain networks has prompted considerable apprehension regarding their energy consumption and environmental consequences, especially concerning consensus mechanisms reliant on computationally intensive processes like Proof of Work (PoW) (Nakamoto, 2008).

Numerous studies have shown that large-scale blockchain systems can consume energy levels comparable to those of entire nations, leading to significant carbon dioxide (CO₂) emissions (de Vries, 2018), (Stoll et al., 2019). These environmental issues have spurred the development of more energy-efficient consensus protocols and the necessity for robust evaluation tools capable of assessing sustainability alongside conventional performance and security metrics.

In this context, analysis using simulations is crucial. It allows for controlled experiments, avoiding the costs and risks associated with using real blockchain networks.

BlockSim, developed by Alharby and van Moorsel, is a prominent example of a blockchain simulator, distinguished by its adaptability and layered design. This architectural framework facilitates the examination of design compromises spanning network, consensus, and incentive layers via discrete-event simulation (Alharby and van Moorsel, 2020). BlockSim offers functionality for assessing a range of system-level metrics, such as latency, throughput, and fork rates, and has been extensively employed to analyze blockchain performance across different configurations. Nevertheless, like numerous blockchain simulators, BlockSim does not inherently model energy consumption or carbon emissions, which limits its applicability in research centered on sustainability.

Concurrently, energy modeling has been effectively incorporated into versatile network simulators like ns-3, thereby illustrating the capacity of discrete-event simulation to accurately represent power and energy behaviors within distributed systems (Sin et al., 2022). This observation implies that analogous modeling methodologies could be applied to blockchain simulators.

This study expands upon BlockSim by incorporating energy consumption and carbon footprint metrics, which are closely integrated with its event-driven execution model. This enhancement facilitates a quantitative assessment of blockchain protocols, considering both performance and environmental impacts, thus furnishing researchers and protocol designers with a comprehensive framework for the sustainable analysis of blockchain systems.

It is important to clarify the energy characteristics of the consensus mechanisms considered. Proof-of-Work (PoW) energy consumption is economically driven: a rational miner invests in electricity up to the fiat value of the expected block reward, so total PoW energy depends primarily on the cryptocurrency price, the block subsidy, transaction fees, the electricity price, mining hardware efficiency, and the difficulty/hashrate equilibrium, rather than on the raw number of miners (Sedlmeir et al., 2020), (Alharby and van Moorsel, 2020). Proof-of-Stake (PoS), by contrast, secures the network through stake and its energy is dominated by the number of validators kept online and their hardware power, essentially independent of coin price (Platt et al., 2021). Accordingly, this paper treats PoW and PoS as distinct energy regimes and presents its experiments as generic PoW and PoS consensus scenarios rather than tying them to any specific cryptocurrency.

2 Background and related work

Blockchain systems are often evaluated using analytical models, empirical measurements, or simulation-based approaches. Simulation, in particular, has gained prominence due to its capacity to effectively model complex interactions among networking, consensus, and incentive mechanisms within controlled and reproducible settings. Consequently, as blockchain protocols evolve, simulators have become essential tools for examining scalability, security, and performance trade-offs, thus eliminating the necessity for costly real-world infrastructure implementations.

Numerous blockchain simulators have been created in existing research. For example, SimBlock focuses on how blocks spread and how forks behave in large peer-to-peer networks. This makes it useful for studying network delays and their impact on the security of consensus mechanisms (Aoki et al., 2019).

JABS builds upon this research by incorporating support for diverse consensus mechanisms and accommodating extensive network scales, thereby prioritizing both performance and fault tolerance (Yajam et al., 2020). In contrast, other simulators, including those developed on the ns-3 platform, offer high-precision network modeling; however, they necessitate considerable resources to integrate blockchain logic and consensus processes (Gervais et al., 2016).

BlockSim, as described by Alharby and van Moorsel, utilizes a layered discrete-event simulation model, which differentiates among network, consensus, and incentive components (Alharby and van Moorsel, 2020). This architectural framework allows for adaptable examination of protocol design alternatives, and it has been employed in various studies to evaluate throughput, latency, and fork rates. Moreover, BlockSim’s modular structure makes it particularly well-suited for extensions and experimentation, thereby positioning it as a valuable foundation for future enhancements.

Despite these advancements, blockchain simulation frameworks still largely overlook energy consumption and environmental effects. Although many studies have quantified the energy use and carbon footprint of blockchain networks, particularly permissionless PoW networks (de Vries, 2018), (Stoll et al., 2019), using real-world data and statistical models, these analyses are often retrospective and lack the flexibility needed for “what-if” experiments.

General-purpose simulators, like ns-3, include energy frameworks that can model power consumption in networked systems (Wu et al., 2012); however, most blockchain-specific simulators do not have these features built-in.

Recent investigations have underscored the necessity for evaluation instruments that incorporate sustainability considerations, thereby integrating performance, security, and environmental metrics within a unified framework (Stoll et al., 2019). Nevertheless, existing methodologies frequently depend on external post-processing or imprecise approximations, rather than employing tightly integrated simulation models. This deficiency underscores the imperative to augment blockchain simulators, including BlockSim, with explicit modeling of energy consumption and carbon emissions, thus facilitating systematic and comparative sustainability assessments throughout the protocol design process.

2.1 Energy, permissionlessness, and sybil resistance

The energy gap between PoW and PoS is tied to how each mechanism resists Sybil attacks in a permissionless setting. PoW achieves Sybil resistance by binding influence to an external, costly resource (computation and therefore electricity), so PoW is not “free”: its permissionlessness is purchased through continuous resource expenditure, which is precisely why its energy use is large and coupled to coin price (Sedlmeir et al., 2020), (Alharby and van Moorsel, 2020). PoS instead binds influence to internal economic stake and validator participation, removing most computational waste but introducing different trust and economic assumptions (Platt et al., 2021). Platt, Platt and McBurney formalise these tensions as a Sybil-attack vulnerability trilemma among permissionlessness, Sybil resistance, and freeness (Platt et al., 2024), and De Vries shows how Ethereum’s transition from PoW to PoS after The Merge illustrates why current PoS systems should not be modelled as historical PoW systems (de Vries, 2023). These distinctions motivate treating PoW and PoS as two separate energy models in the remainder of the paper.

3 Overview of BlockSim (Alharby and van Moorsel)

BlockSim is a discrete-event simulation framework that facilitates the systematic assessment of blockchain systems by abstractly modeling their fundamental architectural layers, while also allowing for extensibility. Developed by Alharby and van Moorsel, BlockSim was conceived to overcome the constraints inherent in analytical and empirical evaluation techniques when examining intricate blockchain design trade-offs across diverse configurations (Alharby and van Moorsel, 2020).

The simulator’s architecture is structured in layers, dividing the blockchain system into three main components: the network layer, the consensus layer, and the incentive layer. The network layer simulates message propagation delays and peer connectivity, which allows for the analysis of latency and communication overhead.

The consensus layer, which includes the processes of block creation, validation, and fork resolution, allows researchers to study how protocol parameters, such as block intervals and the distribution of mining power, affect the system.

The incentive layer models economic aspects, including rewards and penalties, providing insight into miner behavior and system stability (Alharby and van Moorsel, 2020).

BlockSim, constructed using Python, operates within a discrete-event framework, wherein system dynamics unfold through pre-scheduled events encompassing block creation, message dissemination, and chain modifications. This methodology facilitates repeatable experimentation and precise manipulation of simulation variables, while simultaneously providing the necessary abstraction to accommodate extensive scenario analysis. Prior research has demonstrated BlockSim’s efficacy in assessing throughput, block propagation latency, and stale block frequencies across diverse blockchain architectures (Alharby and van Moorsel, 2020), (Lund et al., 2019).

A significant advantage of BlockSim is its inherent modularity and capacity for extension. The framework is deliberately structured to empower researchers to modify or augment internal models, encompassing node functionality, transaction processing, and consensus mechanisms, without necessitating a complete overhaul of the simulator.

BlockSim’s design makes it particularly well-suited for “what-if” analyses and comparative studies of different protocol designs (Lund et al., 2019). While it is versatile, BlockSim mainly focuses on performance, security, and incentive-related metrics. The original framework does not explicitly model environmental factors like energy use and carbon emissions. Given the growing importance of sustainable blockchain design and the proven ability to integrate energy models into discrete-event simulators (Wu et al., 2012), BlockSim offers a solid foundation for extensions that include energy and carbon-aware evaluations while preserving its original architectural principles.

4 Energy consumption modeling extension

Energy consumption in blockchain systems is primarily driven by computational operations associated with consensus execution and communication activities related to block and transaction propagation. To support sustainability-oriented analysis, this work extends BlockSim by integrating an explicit energy consumption model into its discrete-event simulation framework.

Following established practices in energy-aware modeling of distributed systems, energy usage is expressed as a function of discrete events and their associated resource costs (Feeney et al., 2001), (Wu et al., 2012). In the proposed extension, each node i maintains a cumulative energy counter E i, which is incremented whenever a computation or communication event is executed.

4.1 Computational energy model

Computational energy consumption is modeled based on consensus-related operations such as block generation and validation. For a node i, the computational energy consumed during a simulation interval is defined as Equation 1:where N i c o m p denotes the number of computational events executed by node i, and e c o m p k represents the energy cost of the k-th computational event.

E i c o m p = ∑ k = 1 N i c o m p e c o m p k (1)

For Proof of Work–based mechanisms, e c o m p is proportional to the number of hash attempts (Equation 2):where H i is the number of hash computations performed by node i, and α is the energy cost per hash operation, consistent with prior empirical analyses of blockchain energy usage (de Vries, 2018). For less computation-intensive consensus mechanisms, fixed or lower per-event energy coefficients are applied, reflecting reduced processing requirements (Xiao et al., 2020).

e c o m p = α · H i (2)

4.2 Communication energy model

Communication energy consumption accounts for message transmission and reception events in the network layer. The communication energy for node i is defined as Equation 3:where N i t x and N i r x represent the number of transmitted and received messages, respectively, and e t x, e r x denote the energy costs of transmission and reception. This formulation aligns with packet-level energy models commonly used in network simulators such as ns-3 (Wu et al., 2012).

E i c o m m = N i t x · e t x + N i r x · e r x (3)

Although defined here, communication energy was not analysed in the previous version. In this revision it is instrumented with explicit transmit/receive counters for block and transaction messages and analysed in Section 7 (Results and analysis) as a function of transaction rate, block size, and peer connectivity, and reported separately from consensus energy.

4.3 Total energy consumption

The total energy consumption per node and for the entire network is computed as Equation 4:

E i = E i c o m p + E i c o m m , E t o t a l = ∑ i = 1 N E i (4)

This integrated energy model enables BlockSim to quantify energy usage per block, per transaction, and per simulation run, forming a foundation for subsequent carbon footprint analysis.

4.4 Proof-of-work economic energy model

Because PoW energy is bounded by mining economics, we add an economic model. The expected per-block reward in fiat is given by Equation 5, where B t is the block subsidy, F t the transaction fees, and P t the coin price. A rational mining market spends a fraction κ ∈ [0,1] of this reward on electricity, giving the economic energy budget per block in Equation 6, where C elec , t is the electricity price per kWh. When hardware data are available, a technical bound E technical = H block · ε hash is derived from the expected hashes per block (H block = D · 2 32) and the per-hash efficiency ε hash, and the network per-block energy is E PoW , t = min E technical , E budget , t. The miner-level allocation by hashpower share s i , t is given by Equation 7. Thus PoW energy depends on coin price, reward, electricity price, efficiency, and difficulty/hashrate, and only the distribution depends on the number of miners.

R t = B t + F t · P t (5)

E budget , t = κ · R t C elec , t (6)

E i , t = s i , t · E PoW , t (7)

4.5 Proof-of-stake validator energy model

For PoS we add a validator-count model: the network energy is given by Equation 8, where P v is per-validator power (W), T the horizon (h), u v ∈ [0,1] the uptime, V the number of validators, and E comm the communication energy. PoS energy therefore scales with validator count and hardware power and is independent of coin price.

E PoS = ∑ v = 1 V P v · T · u v 1000 + E comm (8)

5 Carbon footprint modeling

While energy consumption provides a direct measure of resource usage, assessing the environmental impact of blockchain systems requires translating energy demand into associated carbon dioxide (CO₂) emissions. Carbon footprint modeling enables sustainability-oriented evaluation by linking simulated energy consumption to greenhouse gas emissions attributable to electricity generation.

In this work, carbon emissions are derived directly from the energy consumption metrics produced by the extended BlockSim framework. Following widely adopted carbon accounting practices, total CO₂ emissions are modeled as a linear function of energy usage and an emission factor that represents the carbon intensity of electricity generation (Peters et al., 2013), (IPCC, 2019).

5.1 Carbon emission model

Let E i denote the total energy consumption of node i, measured in kilowatt-hours (kWh), as obtained from the energy consumption model. The corresponding carbon emissions C i are calculated as Equation 9:where γ is the electricity emission factor expressed in kilograms of CO₂ per kilowatt-hour (kgCO₂/kWh). The value of γ depends on the energy mix of the assumed electricity grid and can be configured to represent global averages or region-specific scenarios, consistent with prior studies on blockchain carbon emissions (Stoll et al., 2019), (de Vries, 2019).

C i = E i · γ (9)

Because γ captures the carbon intensity of the underlying grid, the same energy yields very different emissions across regions and times. We therefore treat γ as an experimental variable with three scenarios: a low-carbon grid (γ = 0.05 kgCO₂e/kWh), a global-average grid (γ = 0.475), and a high-carbon grid (γ = 0.82); the sensitivity is reported in the Results.

The total carbon footprint of the blockchain network over a simulation run is computed as Equation 10:where N denotes the total number of participating nodes.

C t o t a l = ∑ i = 1 N C i = γ ∑ i = 1 N E i (10)

5.2 Granular emission metrics

To enable fine-grained sustainability analysis, carbon emissions are further normalized with respect to blockchain outputs. Carbon emissions per block C b l o c k and per transaction C t x are defined as Equation 11:where B is the total number of confirmed blocks and N t x is the total number of processed transactions during the simulation. These normalized metrics are commonly used in empirical blockchain sustainability studies to compare protocol efficiency across different workloads and configurations (Stoll et al., 2019), (Sedlmeir et al., 2020).

C b l o c k = C t o t a l B , C t x = C t o t a l N t x (11)

5.3 Integration with BlockSim

By embedding carbon footprint computation directly into BlockSim’s event-driven execution, the simulator produces time-series and aggregate emission statistics without external post-processing. This tight coupling enables comparative evaluation of blockchain protocols with respect to performance, energy efficiency, and environmental impact within a unified simulation framework.

6 Experimental setup

To evaluate the effectiveness of the proposed energy and carbon footprint extensions, a series of controlled simulation experiments were conducted using the extended BlockSim framework. The experimental design follows best practices for blockchain performance evaluation, ensuring reproducibility and comparability with prior studies (Alharby and van Moorsel, 2020), (Sedlmeir et al., 2020).

6.1 Network and system configuration

The simulated blockchain network consists of N nodes, where N ∈ 50 , 100 , 500 , 1000, representing small to moderately sized permissionless networks commonly analyzed in the literature (Aoki et al., 2019). Nodes are connected via a peer-to-peer overlay with randomly assigned propagation delays drawn from a bounded distribution, consistent with established modeling approaches for blockchain network latency (Gervais et al., 2016). BlockSim’s discrete-event scheduler is used to model message dissemination, block arrival, and chain updates.

6.2 Consensus and workload parameters

The primary evaluation scenario is based on a Proof of Work–style consensus mechanism, which serves as a baseline due to its well-documented energy characteristics (de Vries, 2018). Two PoW timing scenarios were evaluated: a long-interval PoW baseline scenario (Model 1) and a short-interval, high-throughput PoW scenario (Model 2). Block generation intervals follow the default targets of each scenario (long-interval: ∼10 min; short-interval: ∼12–13 s), which directly affects the number of block events observed during the same simulation horizon. Transaction workloads are modeled as a Poisson process, a common assumption in blockchain performance studies (Alharby and van Moorsel, 2020), (Croman et al., 2016). To assess scaling effects on sustainability metrics, miner population was varied while maintaining the same simulation time window.

To assess the sensitivity of energy and emission metrics, additional scenarios vary mining power distribution among nodes and transaction throughput. These variations allow for an analysis of how network size and the intensity of the workload affect results related to sustainability.

6.3 Energy and carbon parameters

Energy consumption is computed at runtime using mining-interval accounting (start/stop events). In the evaluated configuration, the PoW baseline scenario uses a hashrate-and-efficiency-based power derivation, while the high-throughput PoW scenario uses a direct fixed per-miner power configuration (e.g., 2500 W/miner). Carbon emissions are computed using a configurable electricity emission factor; in this revision it is varied across low, average, and high grid scenarios (see Results) rather than held fixed, following established blockchain carbon footprint analyses (Stoll et al., 2019). Concretely, carbon is derived via the standard emission-factor formulation (electricity use × emissions factor), consistent with common GHG accounting practice. For completeness and reproducibility, the complete set of experimental parameters used throughout this study is summarized in Supplementary Appendix Table A1 in Supplementary Appendix A.

6.4 Evaluation metrics and methodology

The subsequent metrics are gathered for each simulation scenario: aggregate energy consumption, energy expenditure per block, energy usage per transaction, total CO₂ emissions, and CO₂ emissions per block and transaction. Concurrently, performance metrics, including throughput, propagation delay, and stale-block rate, are recorded to analyze the trade-offs between efficiency and sustainability. Each scenario is replicated across 30 independent random seeds (generated deterministically as base + i, i = 0, …, 29). The random seeds affect stochastic simulator events, including peer-to-peer propagation delays, transaction arrivals, mining-event scheduling, and message ordering, resulting in small variations across simulation runs. Results are reported as the mean with sample standard deviation and 95% confidence intervals (Student’s t distribution).

The extended implementation and all code-level modifications to the BlockSim simulator (including the added energy-consumption and carbon-footprint modeling components) are made publicly available in a version-controlled GitHub repository at (Rasheed, 2026).

7 Results and analysis

The simulation results demonstrate that integrating energy and carbon footprint modeling into BlockSim enables meaningful sustainability analysis while preserving the simulator’s original performance evaluation capabilities. In the fixed-power comparative scenarios, total energy and CO₂ increased with network scale because additional active nodes contributed additional configured power. By contrast, under the economic PoW model, total network energy remained approximately invariant with miner count when the economic assumptions were held fixed, as discussed below (Alharby and van Moorsel, 2020), (Sedlmeir et al., 2020).

However, the magnitude difference between the two PoW scenarios in these runs is dominated by the selected power modeling assumptions, as discussed below.

7.1 Energy consumption analysis

The fixed-power comparative scenario exhibits increasing total energy with miner count because each additional miner contributes additional power. By contrast, the economic PoW model presented later is constrained by mining economics rather than miner population; therefore, its total network energy remains approximately invariant with respect to the number of miners.

In the fixed-power comparative scenario (simulation horizon = 10,000 s), the cumulative totals extracted from EnergyLog were:

• PoW baseline scenario: 3.93 kW h (50 miners) → 83.84 kW h (1,000 miners)
• High-throughput PoW scenario: 346.70 kW h (50 miners) → 6,934.13 kW h (1,000 miners)

This behavior aligns with earlier observations that computational effort scales with participating mining capacity in Proof of Work–based systems (de Vries, 2018). Notably, the high-throughput-scenario totals are ∼83–88× higher than the baseline-scenario totals at the same miner counts in this configuration (e.g., 6,934.13/83.84 ≈ 82.7× for 1,000 miners). The preceding results correspond to a fixed-power comparative scenario evaluated over a 10,000-s simulation horizon. By contrast, the economic Proof-of-Work (PoW) model presented below is normalized to a 24-h analytical horizon. Because these two analyses are based on different assumptions and serve different purposes, their numerical values should not be interpreted as directly comparable.

These absolute magnitudes are scenario-based model outputs, not real-world measurements, and the high-throughput PoW scenario is retained only as a comparative computational baseline and should not be interpreted as representing any existing blockchain network. Under the added economic PoW model, total network energy is governed by mining economics rather than miner count: holding the economics fixed, the network total is approximately 467.5 GW h over 24 h and is essentially invariant across 50–1,000 miners (95% CI ± 17.1 GW h), while mean per-miner energy falls approximately as 1/N. Energy scales linearly with coin price, rising from 155.8 GW h at 20,000 USD to 779.1 GW h at 100,000 USD, and inversely with electricity price, confirming the economic argument of Section 4.4. Figure 1 summarizes the sensitivity of PoW network energy to the number of miners under the revised economic model: the network total is essentially invariant to miner count, while the mean per-miner energy falls approximately as 1/N.

For Proof-of-Stake, the validator-count model yields far smaller totals: a 1000-validator network of standard 100 W servers consumes about 2,387 kW h over 24 h, scaling linearly with validator count (239 kW h at 100 validators). The PoW/PoS energy ratio is on the order of 196,846×, consistent with the order-of-magnitude reductions reported in the literature (Sedlmeir et al., 2020), (Platt et al., 2021) and the ∼99.95% reduction at Ethereum’s Merge (de Vries, 2023). As an analytical sanity check, fixed-power configurations reproduce the closed-form E = P · T exactly (maximum relative error 0.00e+00). As shown in Figure 2, PoS total energy scales linearly with the number of validators across the low-power, standard-server, and high-power hardware classes.

Communication energy was also analysed. At peer degree 8 with 1 MB blocks it grows from 15.0 kW h/day at 1 tx/s to 1,455 kW h/day at 100 tx/s. Relative to PoW consensus energy this is negligible (ratio ∼3e-06), but at high throughput it becomes comparable to PoS consensus energy, so communication energy matters mainly in PoS and high-throughput settings. Figure 3 compares computational (consensus) energy with communication energy under the evaluated scenarios on logarithmic axes.

Overall, these results show that energy scaling depends on the adopted power-model assumptions. In the fixed-power comparative scenario, increasing the number of active miners increases total configured network power and therefore total energy demand. Under the economic PoW model, however, holding the economic assumptions fixed, total network energy remains approximately invariant with miner count: increasing participating mining capacity increases total energy demand (de Vries, 2018), (Croman et al., 2016). Additionally, skewed mining power distributions (not shown here) tend to increase aggregate energy because dominant miners perform a larger share of work, consistent with centralization concerns that can worsen efficiency without necessarily improving throughput (Gervais et al., 2016).

7.2 Carbon footprint analysis

Carbon emissions closely follow energy consumption because emissions are computed directly from electricity usage via an emission factor that, in this revision, is varied across low/average/high grid scenarios rather than held fixed (Stoll et al., 2019). In the evaluated configuration, cumulative CO₂ totals increased proportionally with miner count:

• PoW baseline scenario: 1.75 kg CO₂e (50 miners) → 37.31 kg CO₂e (1000 miners)
• High-throughput PoW scenario: 154.28 kg CO₂e (50 miners) → 3,085.69 kg CO₂e (1000 miners)

As expected, the carbon curves preserve the same ordering and near-linear scaling as energy because the model uses a constant grid factor. This follows standard emissions accounting (kWh × kgCO₂e/kWh).

Varying the grid emission factor shows that geography dominates carbon outcomes for a given energy level. For the PoW reference, daily emissions range from 23,374 t at γ = 0.05 to 383,332 t at γ = 0.82 (about a 16× spread from grid choice alone), while the PoS reference spans only 119–1947 kg. A single fixed emission factor is therefore insufficient. Figure 4 reports the carbon-emission sensitivity under low, average, and high electricity emission factors γ.

By normalizing emissions at the block and transaction levels, the simulator enables direct comparison between protocol configurations, a feature that is often missing in retrospective carbon analyses (Stoll et al., 2019), (de Vries, 2019). These metrics provide actionable insights into how protocol parameters and workloads influence environmental impact.

7.3 Performance and sustainability trade-offs

When performance metrics are analyzed alongside sustainability indicators, the evaluated scenarios reveal differences in both throughput and environmental metrics. The high-throughput configuration also exhibits higher energy consumption and carbon emissions; however, because the scenarios use different power-model assumptions, this difference should not be interpreted as a causal effect of throughput alone (de Vries, 2018); (Croman et al., 2016). The integrated results demonstrate the value of a unified simulation framework that captures performance, energy, and carbon metrics simultaneously.

Overall, the results confirm that the proposed extensions enable BlockSim to serve as an effective tool for sustainability-aware blockchain evaluation, bridging the gap between performance analysis and environmental impact assessment.

8 Discussion and threats to validity

8.1 Discussion

The findings underscore the necessity of incorporating sustainability metrics directly into blockchain simulation frameworks. This study’s extension of BlockSim, incorporating energy consumption and carbon footprint modeling, facilitates a unified assessment of performance, energy efficiency, and environmental impact within a single experimental setting. This integrated approach is crucial, given that previous research has demonstrated that enhancements in throughput and scalability frequently entail increased energy consumption and emissions (Alharby and van Moorsel, 2020), (Croman et al., 2016).

A key observation in the comparison of the two PoW scenarios is that energy and CO₂ outcomes are highly sensitive to the chosen power model. In these experiments, the high-throughput PoW scenario is configured with a fixed high per-miner power, which causes total network power to scale directly with miner count. This makes the high-throughput scenario’s cumulative energy and carbon footprint substantially larger than the baseline scenario in absolute terms, even though it also produces many more blocks during the same simulation window. Methodologically, this emphasizes the need to clearly distinguish between (i) scenarios that represent a growing mining industry (more miners = more total watts) and (ii) scenarios that represent redistribution of a fixed global hashrate across more participants.

The results show that the scaling behavior of energy and carbon metrics depends on the evaluated scenario, particularly the adopted power model, network configuration, and workload assumptions, consistent with empirical data derived from actual blockchain systems (Stoll et al., 2019). Furthermore, the capacity to normalize emissions on a per-block and per-transaction basis offers a valuable framework for comparing protocol configurations, thereby mitigating a constraint observed in numerous retrospective sustainability analyses that depend on broad, system-wide estimations (de Vries, 2019). Consequently, this strengthens the efficacy of simulation-based methodologies for investigating “what-if” scenarios during the design phase of blockchain protocols.

Incorporating energy and carbon accounting into BlockSim, from a methodological standpoint, preserves its fundamental modularity and abstraction advantages, simultaneously expanding its applicability to sustainability-oriented research. Consequently, the simulator becomes an essential tool for evaluating innovative consensus mechanisms and optimization approaches aimed at reducing environmental effects, thus augmenting current performance-centered studies (Sedlmeir et al., 2020).

8.2 Threats to validity

Several threats to validity should be considered when interpreting the results. Modeling assumptions represent a primary internal validity concern. Energy consumption is approximated via mining-interval accounting and configurable coefficients/power assumptions rather than hardware-specific measurements, which may not fully capture the diversity of real-world mining and validation infrastructures. However, this abstraction is consistent with prior simulation-based energy studies and is necessary for scalable experimentation (Wu et al., 2012).

Although this revision evaluates three representative emission-factor scenarios (low, average, and high grid carbon intensity), these discrete values do not capture the full temporal and geographical variability of real electricity grids. Therefore, the reported carbon emissions should be interpreted as scenario-based estimates rather than location-specific operational measurements.

Construct validity is, in the final analysis, shaped by the degree of abstraction present within BlockSim. Discrete-event simulation, although capable of accurately modeling system-level behavior, is inherently limited in its capacity to fully capture low-level hardware optimizations or the complex nuances of real-world network dynamics. Despite these constraints, this degree of abstraction is a commonly accepted compromise within blockchain research, facilitating systematic comparisons across a variety of protocol designs (Alharby and van Moorsel, 2020), (Sedlmeir et al., 2020).

Additional limitations follow from the revised scope. First, the framework is a scenario-based explorer; the absolute energy and carbon values are model outputs for the chosen assumptions and are not intended to predict the exact real-world consumption of any specific cryptocurrency.

Second, PoW energy is sensitive to coin price, block reward, transaction fees, electricity price, mining efficiency, and difficulty/hashrate dynamics, and to the spend ratio κ; PoS energy is sensitive to validator count, node hardware, uptime, redundancy, and communication overhead.

Third, carbon depends strongly on region- and time-specific grid mixes; the γ scenarios bracket but do not resolve this variability. Finally, the PoW results represent scenario configurations rather than any specific cryptocurrency, and communication energy, while small under PoW, can matter in PoS and high-throughput scenarios.

9 Conclusion and future work

This paper presented an extension of the BlockSim simulation framework that integrates energy consumption and carbon footprint modeling into its discrete-event architecture. By embedding mining-interval energy accounting and deriving carbon emissions using established emission-factor–based formulations, the proposed approach enables unified evaluation of blockchain systems in terms of performance, energy efficiency, and environmental impact. Unlike retrospective measurement studies, the extended simulator supports controlled “what-if” experimentation, allowing protocol designers to analyze sustainability implications during the early design phase.

The experimental results show that, under the fixed-power comparative scenario, total energy consumption increases with total configured network power and therefore with the number of active miners when per-miner power is held fixed. By contrast, under the economic Proof-of-Work model, total network energy is governed primarily by mining economics and remains approximately invariant with miner count when coin price, block reward, electricity price, mining efficiency, and difficulty/hashrate conditions are held fixed. Furthermore, the capacity to normalize environmental impact on a per-block and per-transaction basis offers a valuable foundation for evaluating different protocol designs, thereby overcoming a significant constraint observed in numerous current blockchain simulation tools (Alharby and van Moorsel, 2020). These results corroborate that augmenting BlockSim maintains its advantages in terms of flexibility and abstraction, while simultaneously expanding its utility for research centered on sustainability.

Several avenues for future research are suggested by the findings of this study. Initially, the energy model’s accuracy could be enhanced by integrating hardware-specific parameters obtained from empirical data on mining rigs, validator nodes, and networking infrastructure, thereby improving realism while preserving scalability. Furthermore, subsequent developments could incorporate dynamic and region-specific carbon intensity models, allowing for the simulation of diverse energy compositions and renewable energy adoption levels, which have demonstrably influenced blockchain carbon footprints (Stoll et al., 2019). Furthermore, the framework’s scope could be broadened to incorporate emerging low-energy consensus mechanisms, thus facilitating direct comparisons between traditional Proof of Work and alternative, sustainability-oriented designs (Xiao et al., 2020). Ultimately, integrating the improved BlockSim with more advanced network simulators or real-world data offers a potentially effective approach to reconcile the gap between theoretical modeling and practical blockchain applications.

Overall, this work contributes a practical and extensible foundation for sustainability-aware blockchain simulation, supporting future research on environmentally responsible decentralized systems.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. Generative AI was used to assist with language editing (grammar, clarity, and readability) and to refine phrasing of non-technical text. Tool: ChatGPT (OpenAI), model GPT-5.2. Thinking, accessed January 2026. The authors reviewed, edited, and verified all AI-assisted text and take full responsibility for the content, including the accuracy of statements and the integrity of citations. No AI tool was used to generate or alter data, results, or figures.

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.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbloc.2026.1804830/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Remo Pareschi, University of Molise, Italy

Reviewed by: Claudio Schifanella, University of Turin, Italy

Moritz Platt, King’s College London, United Kingdom