Energy Experience: Interdisciplinary Insights

• Energy Experience is a multifaceted concept defined by how energy data is perceived, integrated, and acted upon in social, architectural, and computational contexts.
• It bridges research in HCI, built-environment monitoring, and service optimization by focusing on design, measurement, and user feedback mechanisms.
• The literature emphasizes practical applications such as immersive visualizations, optimization algorithms, and energy monitoring tools to enhance literacy and system performance.

In the cited literature, “Energy Experience” does not denote a single standardized construct. It is used for several related but non-identical research programs concerned with how energy becomes perceptible, interpretable, actionable, and optimizable across built environments, households, software systems, and infrastructural services. In HCI-oriented work, it explicitly shifts attention from conventional user experience toward “social experiences involving energy design,” and toward the “socially mediated, material, provisioning, earthbound problems of the journey” (Sutherland, 7 Aug 2025). In built-environment and energy-informatics research, it refers to embedding energy data into spatial, architectural, or domestic context so that users can inspect, compare, and act on it (Safikhani et al., 2024, Tchatchoua et al., 12 May 2025). In software, HPC, and service-optimization literatures, it is operationalized through energy measurement, path-dependent execution, consumer satisfaction, or Quality of Experience (QoE), indicating that “experience” may refer either to human users’ perceived service quality or to the practical and analytic conditions under which energy behavior is observed and controlled (Wegmeth et al., 2024, Li et al., 2019, Rajput et al., 17 Mar 2026).

1. Conceptual foundations

A direct definition is given in “Energy Experience Design” (Sutherland, 7 Aug 2025). There, energy experience design is distinguished from a narrow consumerist conception of usability and instead focuses on the social experiences created by powering, maintaining, repairing, and provisioning devices. The emphasis falls on longevity, interoperability, repairability, standards-based components, reduced dependence on critical minerals, and, in more radical cases, batteryless sustainable-energy architectures. The same work treats batteries as a paradigmatic site where everyday convenience intersects with material waste, extraction, and climate justice, and therefore makes energy experience inseparable from storage media, maintenance practices, and material supply chains.

A second conceptual strand comes from household energy monitoring. “Energy personas in Danish households” (Tchatchoua et al., 12 May 2025) argues that energy experience is produced through the intersection of household routines, social organization, material devices, and prior experience with monitoring technology. On this view, households do not simply consume electricity; they use electricity while carrying out cooking, laundry, dishwashing, EV charging, and the coordination of family life. Real-time monitoring therefore does not merely convey information. It changes what can be noticed, which routines can be shifted, and which activities remain non-negotiable.

A third strand appears in service and infrastructure optimization. In isolated microgrids, user experience is formalized as a consumer satisfaction indicator that measures the adequacy of total supplied power relative to total load over the dispatch horizon (Li et al., 2019). In crowdsourced IoT energy services, QoE becomes an importance-weighted or satisfaction-weighted measure of how well energy requests are fulfilled over time (Abusafia et al., 2022, &&&10&&&). This suggests that, outside HCI, “energy experience” often denotes a schedulable proxy for perceived service quality rather than a phenomenological account of energy awareness.

2. Spatialization, visibility, and immersion

One major research direction makes energy visible by placing it back into spatial context. “VR4UrbanDev: An Immersive Virtual Reality Experience for Energy Data Visualization” (Safikhani et al., 2024) is exemplary. It presents a VR environment designed to facilitate interaction with energy-related information through two complementary modes: “world in miniature” for large-scale exploration and “first-person” for real-world-scale local inspection. In the miniature mode, a circular two-tier interaction desk combines a 3D campus or city model with gadgets and visualization-mode buttons; touching a building opens a detailed model, and graphs for cooling, heating, and electricity consumption are projected onto the room walls. In the first-person mode, users are transferred to a 1:1 local view for contextual inspection of buildings and energy-related information. The paper’s central contribution is to make energy data explorable as place-bound, embodied, and cross-scale rather than dashboard-bound.

Domestic monitoring work addresses the same visibility problem at another scale. The Barry app, later integrated into Ewii, shows householders “exactly how green your electricity is and how much CO2 it emits, hour by hour – kWh for kWh,” along with electricity price levels, renewable share, and notifications about cheap or low-carbon periods (Tchatchoua et al., 12 May 2025). Repeated exposure teaches users recurring temporal patterns even when active checking declines, so the app functions not only as a display but also as a medium of energy literacy. The study’s four personas—dedicated, organised, sporadic, and convenient—show that energy visibility is filtered by household composition, gendered labor division, EV ownership, and the negotiability of routines. Across personas, EV charging, dishwashing, and laundry are relatively shiftable; cooking is repeatedly treated as non-flexible.

Residential appliance-level breakdown extends the same logic of visibility from grid conditions to end uses. “Active Collaborative Sensing for Energy Breakdown” (Jia et al., 2019) starts from the observation that a monthly bill tells a household how much was spent, but not why. It therefore models appliance-level monthly energy as a low-rank tensor over homes, appliances, and time, and uses active sensing to decide which <home, appliance> pairs should be instrumented so that non-instrumented homes can still receive appliance-level breakdown. In the Austin Dataport study, ActSense reaches a target Year RMSE of 50 with 3 new observations per month, whereas QBC requires 8 and Random 10. The significance is not only predictive accuracy. It is the prospect of making energy use legible and actionable for many more households without fully sub-metering every home.

3. QoE, consumer satisfaction, and service allocation

In several papers, energy experience is formalized as QoE. The most explicit microcell formulation appears in “Quality of Experience Optimization in IoT Energy Services” (Abusafia et al., 2022), where the super provider seeks to maximize an importance-weighted measure of demand fulfillment over time:

$$QoE(\mathcal{T}, BM, \mathcal{EDD}, STR) = \frac{1}{n} \sum_{i=1}^{n} \left( \frac{STR.Al_i}{EDD.r_i} \cdot BM.I_i \right).$$

Here, experience is not direct subjective report. It is an aggregate, business-aware indicator of how well limited crowdsourced wireless energy is allocated across time slots with different importance.

“Maximizing Consumer Satisfaction of IoT Energy Services” (Abusafia et al., 2022) refines this formulation by decomposing QoE into a Satisfaction Ratio and a Fulfillment Ratio, then combining them as

$$QoE(M) = \alpha \times \left(\sum_{i=1}^{m} SR_i \times \beta_i \right) + (1-\alpha) \times \left(\sum_{i=1}^{m} FR_i\times \gamma_i \right).$$

This design is important because it distinguishes breadth of service—how many consumers receive at least partial energy—from depth of service—how much of the requested energy is fulfilled. The paper’s Partial-Based and Demand-Based algorithms exploit the shareable and partially allocable nature of energy services; empirically, the Demand-Based approach performs best overall, while the Partial-Based approach is particularly effective when services are scarce and partial fulfillment of many consumers is preferable to complete fulfillment of a few.

The isolated microgrid literature adopts a related but more aggregate consumer-satisfaction proxy. “Incorporating energy storage and user experience in isolated microgrid dispatch using a multi-objective model” (Li et al., 2019) defines

$$F_3 = \frac{ \sum_{t=1}^{T} \left( \sum_{n=1}^{MG} P_{n,t}^{MT} + P_t^{PV} + P_t^{WT} + P_t^{DC} - P_t^{CH} \right) }{ \sum_{t=1}^{T} P_t^{L} } \times 100\%.$$

In that framework, better user experience requires more adequate net supply, but this conflicts with economy and emissions. The reported extreme solutions make the trade-off explicit: a cost-optimal schedule yields $F_3=90.5\%$, while a satisfaction-optimal schedule yields $F_3=99.9\%$ at higher cost and higher emissions.

A communication-systems variant appears in “Quality of Experience Optimization for Real-time XR Video Transmission with Energy Constraints” (Pan et al., 2024). There QoE is defined frame by frame as

$$q_f =(1-x_f)\left(v_f-\mu_1 |v_f-v_{f-1}|\right)-\mu_2 x_f,$$

with long-term average transmission energy constrained. The paper’s main conclusion is that, under real-time XR constraints, stable and timely delivery can dominate raw bitrate: the proposed algorithm improves average QoE by 0.04 to 0.46, reduces average video quality variation by 29% to 50%, and improves frame transmission success rate by 5% to 48%. Across these service literatures, energy experience is therefore an optimization target defined by reliability, smoothness, or fulfillment under scarcity.

4. Measurement, reporting, and the energy behavior of computation

Another major meaning of energy experience concerns the practical observability of computational energy. “EMERS: Energy Meter for Recommender Systems” (Wegmeth et al., 2024) treats the energy experience of experimentation as measurement, monitoring, organization, interpretation, and reporting. EMERS measures whole-system energy draw via smart plugs, supports integrated per-experiment logging and standalone background logging, and exposes a Flask-based UI with instantaneous power graphs, cumulative energy-over-time graphs, experiment-level summaries, and configurable cost and carbon views. Its empirical examples also show that idle baselines can be substantial and hardware-dependent: 69.15 ± 2.45 W for a Windows 11 workstation with Intel Xeon W-2255 and RTX 3090, 80.45 ± 3.45 W for a Windows 10 workstation with Intel Core i7-6700K and GTX 980 Ti, 18.55 ± 1.55 W for a 2022 Mac Studio, and 12.20 ± 1.3 W for a 2020 MacBook Pro.

“Energy Flow Graph: Modeling Software Energy Consumption” (Rajput et al., 17 Mar 2026) pushes this further by arguing that software energy is path-dependent rather than a single aggregate property. It defines an Energy Flow Graph

$$\mathcal{G} = (V, E, \mathcal{C}_s, \mathcal{C}_t, \mathcal{P})$$

with costs on both states and transitions, and total path energy

$$E_{\text{total}(\pi)} = \sum_{i=1}^{k} \mathcal{C}_s(v_i) + \sum_{i=1}^{k-1} \mathcal{C}_t(v_i, v_{i+1}).$$

In 3.5 million executions, 15.6% of solutions exhibited high path-dependent variance with $CV > 0.1$, and structural optimization revealed up to $705\times$ energy reduction. The paper therefore reframes the energy experience of software development as explicit reasoning about execution paths, transition overheads, and optimization interactions rather than aggregate profiling alone.

The CEEC experience report on European HPC systems shows the same observability problem at production scale (Kulkarni et al., 4 Nov 2025). Across LUMI, MareNostrum5, MeluXina, and JUWELS Booster, energy had to be harvested through different mechanisms, including SLURM ConsumedEnergy, HPE Cray PM counters, EAR, and LLview/DCGM-derived GPU power data. The case studies show large accelerator gains—for instance, waLBerla on one LUMI node falls from 154.74 kJ on LUMI-C to 25.67 kJ on LUMI-G—but they also show that energy and time are not perfectly aligned for every kernel, as in the reduction module where the GPU is faster but the CPU uses slightly less energy. The report’s strongest general point is methodological: energy measurement is much less standardized than timing, so exclusive-node use, repeated trials, explicit accounting boundaries, and clear reporting of measurement pathways are indispensable.

5. Pedagogy, professional adoption, and design practice

Energy experience is also pedagogical. “Teaching Energy-Efficient Software -- An Experience Report” (Christensen et al., 28 Apr 2025) describes three Danish university settings in which students learned to treat software energy consumption as an architectural quality attribute and an empirical object of study. At Roskilde University, the course combined experimental design, RAPL-based estimation, external measurement with a Siglent SPD3303X-E programmable power supply, statistics, repeatability, and automation. The questionnaire showed that 75% highly agreed and 25% agreed that the course increased awareness that technology choices, implementation styles, and hardware have a direct impact on energy consumption. The most important educational finding, however, was the gap between intuition and evidence: students repeatedly reported surprise that “optimized” code did not always consume less energy, and learned to question implausible measurements rather than accept them uncritically.

The professional-adoption problem appears equally clearly in immersive visualization. VR4UrbanDev’s interview study involved 10 interviews with energy researchers, planners, and facility managers (Safikhani et al., 2024). Participants found the prototype user-friendly and liked the virtual environment, but some struggled to see clear added value for everyday work. The feedback pointed to the need for simpler visualizations, improved interaction and locomotion, stronger support for the influence of design decisions, and interfaces tailored to stakeholder roles. A similar adoption lesson appears in the Danish household-monitoring study: engagement with Barry/Ewii was often strongest at introduction and then stabilized, faded, or became background knowledge depending on domestic routines and household organization (Tchatchoua et al., 12 May 2025).

These cases suggest a shared constraint. Energy experience systems can raise awareness effectively, but professional or domestic uptake depends on task fit, workflow integration, and the extent to which the system supports actual decisions rather than novelty alone. That conclusion is especially consistent with the CEEC call to “raise awareness, teach the community, and take actions toward more sustainable exascale computing” (Kulkarni et al., 4 Nov 2025).

6. Extensions, abstractions, and contested usages

Outside user-facing HCI and infrastructure literatures, the phrase extends into more abstract or technically specialized domains. In reinforcement learning, “Energy-Based Hindsight Experience Prioritization” (Zhao et al., 2018) uses the physical energy change of a manipulated object as a proxy for experience value. The trajectory energy

$$QoE(M) = \alpha \times \left(\sum_{i=1}^{m} SR_i \times \beta_i \right) + (1-\alpha) \times \left(\sum_{i=1}^{m} FR_i\times \gamma_i \right).$$0

determines replay probability, and the method improves final mean success rate on all four tested robotic manipulation tasks while remaining close to vanilla HER in computational time. Here, energy experience no longer means human experience of energy. It means experience selection based on energy transferred to an object.

In energy automation, the term becomes infrastructural and dependability-oriented. “Flexible Development of Dependability Services: An Experience Derived from Energy Automation Systems” (Florio et al., 2019) addresses a Primary Substation Automation System and the auto-exclusion function, introducing a configurable Redundant Watchdog service built with Ariel and assessed using Generalized Stochastic Petri Nets. The relevant “experience” is engineering experience derived from Electric Power System automation requirements, especially the trade-offs between availability-first and integrity-first watchdog policies.

More speculative usages appear in theoretical work. “The inner screen model of consciousness” (Ramstead et al., 2023) connects conscious experience to the free-energy principle, nested Markov blankets, and internal holographic screens, arguing that consciousness itself may be—or entail—classical information encoded on an internal Markov blanket. “An Energy Puzzle in Quantum Collapse” (Guerreiro et al., 2011) instead argues that projective measurement can change the expectation value of energy of a system-detector composite, leading to a trilemma involving energy extraction, energy absorption, or the incompleteness of standard conservation arguments. These papers employ “energy” and “experience” in ways that are conceptually distant from household monitoring or QoE optimization. This suggests that the term’s meaning is field-dependent and must be read with attention to local formalism.

Taken together, the literature presents Energy Experience as a broad interdisciplinary concern rather than a single doctrine. Its most stable core is the effort to make energy consequential at the level where decisions are actually made: in rooms, households, classrooms, schedulers, dispatch models, software pipelines, and material device ecologies. What changes across domains is the unit of analysis—building, household, microcell, job, execution path, or conscious system—and the formal proxy by which “experience” is rendered observable.