Development of a Visualization Platform for Power Generation Analysis in Urban Building-Integrated Photovoltaic Systems

Abstract

Urban high-density planning and the rise of super-high-rise buildings have significantly limited the development of distributed photovoltaic (PV) systems, creating an urgent need for optimized three-dimensional (3D) layout strategies within urban building spaces. Given that PV power generation is influenced by environmental factors and building spatial configurations, a 3D panoramic visualization tool is essential to intuitively display relevant data and support decision-making for government planners and PV operators. To address this, we developed a visualization platform to assess the integrated PV power generation potential of buildings at both city and single-building levels. The platform enables a 3D spatial panoramic display, where building surfaces are color-coded to clearly represent key performance metrics, such as power generation capacity, installation costs, and potential electricity savings. This intuitive visualization allows stakeholders to identify optimal PV installation areas and evaluate economic benefits effectively. This article details the implementation of the visualization platform across four key aspects: data generation and input, power generation and economic calculation, building model creation and data mapping, and visual interface design, aiming to facilitate the efficient planning and deployment of distributed photovoltaic systems in complex urban environments.

1. Introduction

Limited land availability and the complexity of urban environments present significant challenges to the development of urban photovoltaic systems. Compared to centralized photovoltaics, distributed photovoltaic systems offer a more efficient solution by maximizing the use of available urban spaces, delivering higher carbon emission reduction potential, and achieving lower per-kilowatt-hour costs [1,2,3]. With the continuous decline in photovoltaic module costs and the growing availability of materials suitable for building facades, facade photovoltaics are becoming increasingly competitive with rooftop installations [4,5,6]. As a result, integrating photovoltaic systems into building facades has emerged as a key direction for advancing urban renewable energy utilization. However, photovoltaic power generation is highly sensitive to environmental variables such as climate conditions, building designs, lighting availability, and urban spatial layouts, which vary considerably across different cities [7,8,9,10]. These factors introduce significant uncertainties regarding power generation performance, economic viability, and the investment return cycle of photovoltaic systems [11,12]. Traditional visualizations relying solely on charts or two-dimensional maps are insufficient to comprehensively and intuitively represent these dynamic and spatial changes [13,14,15]. To address this challenge, the development of a city-level three-dimensional visualization platform for integrated-building photovoltaics has become particularly critical.

Currently, a variety of building photovoltaic visualization tools are available, such as PVsyst, Easy PV, PVcase, HelioScope, SolarGIS, and SolarMapper. PVsyst provides city-scale data primarily through charts while integrating individual-building photovoltaic planning with 3D modeling software [16]. SolarGIS focuses on large-scale photovoltaic planning using 2D GIS-based visualization techniques [17]. Easy PV, PVcase, SolarMapper, and HelioScope are designed for smaller regions or single buildings, with PVcase and HelioScope relying heavily on 2D GIS visualization methods [18,19,20]. With advancements in drone technology and remote sensing data, software for urban photovoltaic potential assessment increasingly incorporates high-resolution imagery and 3D modeling techniques. For example, SolarMapper [21] leverages remote sensing data combined with LiDAR technology to deliver city-wide visualization of solar potential, while Easy PV [22] uses satellite remote sensing and radiation modeling for predicting building-level photovoltaic generation capacity.

The growing influence of the gaming industry has also contributed to the evolution of building photovoltaic visualization software. Game engines such as Unity and Unreal Engine provide advanced visualization capabilities at a significantly reduced cost [23,24]. These engines enable features such as highly realistic 3D scene construction, real-time rendering, interactive simulations, terrain modeling, and physical environment simulations [25,26].

Compared with GIS, BIM, or energy simulation software that offer similar functionalities, the key contributions of this platform are as follows:

An integrated building–energy–economy coupling model: Unlike tools such as URBANopt or OpenStudio, which focus primarily on regional-building energy system simulation, this platform further incorporates economic evaluation indicators. By setting thresholds for economic parameters, the platform enables the generation of visualized investment planning schemes.

User-friendly interface design: Compared with traditional software such as DesignBuilder or EnergyPLAN, this platform offers interactive functionalities for users. Without the need to manually construct building or energy system models, users can quickly access and understand the BIPV potential—including economic power generation and installed capacity—at various spatial scales, such as individual buildings, neighborhoods, and cities. This makes the platform easier to use, with a relatively smooth learning curve.

This platform innovatively combines the high-precision solar radiation calculation of Ladybug with the real-time 3D rendering capabilities of Unreal Engine, supporting dynamic sunlight simulation and interactive perspective adjustment. It realizes “roaming-style” visualization simulation for city-level building photovoltaics. This technological breakthrough allows users to intuitively observe the changes in photovoltaic performance across different time periods and seasons, greatly enhancing the realism of the simulation and the interactive experience. It achieves integrated decision support for ‘technology-economics-esthetics’.

The proposed platform enables a city-scale panoramic visualization of photovoltaic power generation potential at both the building level and city level. It can intuitively display key parameters such as power generation, installation costs, electricity savings, and economic returns using color mapping on building models. The platform provides data-driven insights, actionable solutions, and visual representations to government agencies and investors, thereby supporting informed decision-making for urban renewable energy development. This paper describes the four main components of the proposed visualization platform: data resources, calculation methods, building model generation and parameter mapping, and visual interface design. The platform is then validated using Wuhan, China, as the case study to demonstrate its effectiveness and applicability.

User Demand Survey

The primary user groups of the platform include related commissioners in government departments, experts in the power industry, and photovoltaic investors. To gather insights, we selected 15 representative users, comprising 5 individuals from each category, to explore the platform’s purpose of use, data requirements, display preferences, and interaction mechanisms. To gain a comprehensive understanding of user needs, we conducted semi-structured interviews using a semi-open format. The consolidated findings are presented in

Table 1. User demand survey for the platform.

		Government Department Commissioner	Power Industry Expert	Photovoltaic Investor	Total
Purpose of Use	Regional Photovoltaic Investment Planning	5	5	5	15
	Large-Screen Report Presentation	5	3	2	10
Data Content	Estimated Power Generation	5	5	5	15
	Estimated Installation Cost	5	5	5	15
	Estimated Installed Capacity	5	5	5	15
	Estimated Electricity Cost Savings	5	5	5	15
	Building Data Information	5	5	5	15
Display Preferences	Panoramic 3D Visualization	5	4	4	13
	High-Precision Models for Significant Buildings	3	3	5	11
	Color-Coded System with Clear Data Correlations	5	5	5	15
	Interface Design for Large Screens	5	3	2	10
	Interface Design for Laptops	2	4	3	9
	Customized Photovoltaic Layout Plan	5	4	5	14
Interaction Methods	Investment Plan Generation	5	4	5	14
	Scene Roaming and Query	5	4	4	13
	Scene Display Switching	5	5	4	14
	Individual Buildings’ Detailed Information	5	5	5	15
	Adjustment of Installation Plans for Key Buildings	2	3	5	10
	Dynamic Effects for Interface Interaction	5	3	3	9

. Across all user categories, the overarching goal is to support regional photovoltaic investment planning. The data content requirements are generally consistent across groups, although specific needs vary. Government and power-sector decision-makers prioritize reports and presentation capabilities, photovoltaic investors emphasize the need for precise installation planning, and power industry experts value efficient access to detailed data.

2. Platform Architecture

Based on the findings from platform research and software selection, we established a clear workflow for platform development. As illustrated in Figure 1, the development process consists of three main components: basic data calculation, building surface data mapping, and interface display and interaction. The workflow begins with generating and acquiring data, which includes obtaining basic parameters, calculating solar radiation intensity, analyzing photovoltaic power generation potential, and converting these data for use in Unreal Engine (UE). Next, building models are generated using UE and Cesium for Unreal, followed by building surface segmentation and data mapping to achieve panoramic 3D visualization. This step also incorporates mapping data onto building surfaces through color representation. Finally, the platform’s information architecture, user interaction flow, and interactive interface display are developed to complete the visualization platform.

2.1. Data Sources

In large-scale urban environments, the construction of 3D building models and solar radiation simulation calculations become significantly more challenging. Therefore, this study divided the research area based on the fourth-level administrative boundaries (i.e., neighborhood blocks). Within each block, meteorological data collection and radiation simulation were conducted separately, which significantly improved the computational efficiency of the model. To simulate urban building-integrated photovoltaic (BIPV) power generation, this study utilized high-resolution meteorological data from NREL’s NSRDB 2020 dataset and detailed building data from Amap.

The NSRDB dataset provides hourly meteorological data for China in 2020 with a spatial resolution of 2 km. In this study, based on the latitude and longitude coordinates of the centroid of each street, Python 3.9 programming was used to call the API provided by the official NSRDB website to download hourly meteorological data—including latitude, longitude, elevation, annual solar radiation, temperature, and humidity—in CSV format. Then, the CSV files were further converted into EnergyPlus Weather (EPW) files according to the EPW format specifications. Finally, the “import EPW” component in Grasshopper was used to import the EPW files.

Building information was extracted from official Shapefiles (.shp) at 1 m spatial resolution, containing attributes such as coordinates, height, area, number of floors, building names, Shape_Area, and Shape_Length. After preprocessing, these data enabled the construction of detailed 3D urban models for accurate PV performance analysis.

Overall, this study achieves detailed segmentation at the building level and parallel partitioned computation at the urban level, making the visualization of high-spatiotemporal-resolution BIPV simulation results at the city scale more feasible.

2.2. Software and Plugin Selection

Based on data analysis and visualization requirements, specialized toolchains were selected for different tasks (Figure 2). For building environment simulations, Rhino’s Grasshopper and the Ladybug plugin were used to calculate photovoltaic generation and assess thermal performance, offering precise solar radiation mapping and adjustable window-to-wall ratios for diverse building types. Unreal Engine (UE) served as the core platform for high-fidelity 3D interactive visualization, selected for its real-time rendering capabilities, cross-platform support, and blueprint visual programming features. To address geospatial data processing challenges, the Cesium Lab toolchain was introduced for efficient conversion of .shp files into UE-compatible formats, ensuring accurate large-scale city model integration. Blender was used for detailed modeling of key buildings, while City Engine facilitated rapid generation of roads, water bodies, and urban base layers. High-fidelity UI/UX prototypes were developed in Figma to ensure consistent user experience across the platform.

2.3. Interface Interaction

The photovoltaic (PV) visualization platform adopts a multi-scale interactive design to address the decision-making needs of users at different levels.

At the macro scale, government agencies can use the 3D globe navigation interface to intuitively grasp the global distribution of solar energy potential. By clicking on a specific city, users can access detailed solar radiation data and theoretical power generation estimates, supporting early-stage policy planning and renewable energy target setting.

At the city and regional scale, the platform currently covers data for 77 cities across China. Government authorities and investors can conduct in-depth analyses within highly detailed 3D urban environments. The platform supports free roaming, customized area selection, and administrative boundary overlays, allowing users to quickly assess the PV generation potential of specific districts. For example, to support a city’s 2035 renewable energy targets, municipal planners can select the relevant administrative area, input parameters such as investment budget, expected electricity savings, and target PV generation, and submit them through the platform. The system then recommends suitable buildings for deployment based on estimated rooftop potential and cost-effectiveness. Users receive visual feedback with highlighted buildings and can view detailed attributes for each structure and export the building list to assist in investment prioritization.

At the building scale, individual property owners and designers can perform detailed PV planning. Users input target power outputs, budget constraints, and solar panel specifications, and the system generates customized simulation reports, including installation layouts and projected performance.

To ensure a seamless user experience, the platform incorporates several technological innovations. CesiumLab’s streaming terrain technology ensures rapid loading of large-scale urban models, while Level of Detail (LOD) techniques dynamically adjust model resolution to maintain smooth performance across different hardware configurations. Cross-platform data sharing allows users to view reports generated on PCs directly on mobile devices. These design choices make complex PV data analysis intuitive and accessible, truly enabling data-driven decision-making across multiple user groups.

3. Methods

3.1. Solar Radiation Intensity Calculation and Photovoltaic Power Generation Potential Analysis

This study constructs 3D models of urban buildings on the Rhino-Grasshopper platform based on building footprint data and planar segmentation methods. By integrating meteorological data and ray tracing techniques, a BIPV solar radiation and photovoltaic power generation model is developed, which accounts for mutual shading between buildings. The model calculates both the power output and the installed capacity of the BIPV system. Finally, based on the cost of photovoltaic systems and economic parameters such as electricity prices, an economic evaluation indicator for the BIPV system is established to quantify its economic potential. It is worth noting that Xie et al. [27] have already demonstrated the accuracy of using Ladybug for calculating solar radiation potential. Therefore, this study does not validate the simulation results against real façade PV power generation data or other standard simulation models. The detailed modeling process is described as follows (Figure 3):

3.1.1. Construction of City-Scale 3D Building Models

Building footprint data in Shapefile format, containing height attributes, is first imported into the Rhino platform. Using the “Weaverbird’s Mesh Thicken” component in Grasshopper, the footprint data are extruded according to their height to generate 3D building models represented by mesh surfaces (Figure 4a). Subsequently, the “Mesh Brep” component is applied to subdivide the building models into finer planar segments by setting constraints on the maximum and minimum edge lengths (Figure 4b).

For each building surface, the tilt angle and azimuth angle are the two most critical parameters. In our model, the tilt angle of vertical façades was set to 90°, while the tilt angle of rooftops was set equal to the local latitude. The azimuth angle of each surface was computed based on its normal vector.

3.1.2. Solar Radiation Model with Shading Consideration

First, a sky matrix was generated based on meteorological data to determine the solar position for each of the 8760 h in a year. Then, direct solar radiation rays between the sun and the centroid of each building surface were constructed as vectors, as illustrated in Figure 5.

Without considering shading, the total solar radiation received by each surface at time t can be calculated as follows:

$$Total\_Ra{d}_t=Dir\_Ra{d}_t\ast cos{q}_t+Dif\_Ra{d}_t+Ground\_Ra{d}_t$$

(1)

where $Dir\_Rad$, $Dif\_Ra{d}_t$, and $Ground\_Ra{d}_t$ represent the direct, diffuse, and ground-reflected solar radiation at time t, respectively, and $\mathit{cos}{\theta }_t$ is the correction factor for the angle between the direct solar incidence direction and the surface normal at time t.

To account for shading effects between buildings, all surrounding buildings near the target building are added into the 3D model as shading elements. Ray tracing is used to determine whether the direct solar radiation ray to each building surface intersects with any of the shading elements. If an intersection is detected, it indicates that the surface is shaded and does not receive direct solar radiation at time t. If the ray is blocked, its contribution to the surface’s received radiation is set to 0; otherwise, it is set to 1. This can be expressed as follows:

$$State\_Ra{y}_t=\left\{\begin{array}{cc}1,& noshade\\ 0,& shade\end{array}\right.$$

(2)

After accounting for shading, the total solar radiation received by each surface at time t is modified as follows:

$$Total\_Ra{d}_t=Dir\_Ra{d}_t\ast State\_Ra{y}_t\ast cos{q}_t+Dif\_Ra{d}_t+Ground\_Ra{d}_t$$

(3)

3.1.3. Photovoltaic Power Generation Potential Analysis

For the photovoltaic power generation potential analysis, the conversion efficiency of photovoltaic panels is set, and the expected annual photovoltaic power generation on both building roofs and facades is simulated. This simulation incorporates solar radiation intensity, panel power, and installation parameters. The installed capacity is estimated based on the panel area and power, and the corresponding installation cost is calculated by multiplying the installed capacity by the unit capacity price.

The power output of a single PV module is calculated as follows [28]:

$$P=I\cdot V\cdot {\beta }_{inv}\cdot {\beta }_{sys}$$

(4)

where I and V are the operating current and voltage of the PV module, respectively, determined by ambient temperature, wind speed, atmospheric pressure, and the actual solar radiation received by the module [29]. The meteorological data used are sourced from the NSRDB dataset developed by NREL [30]. Here, ${\beta }_{inv}$ represents the inverter efficiency, and ${\beta }_{sys}$ denotes the photovoltaic system efficiency, assumed to be 80.96% [31].

The installed capacity and actual power generation of the building-integrated PV system are expressed as follows:

$$Ca{p}_{PV}^{In}=\sum \left(Ca{p}_{single}^{In}\cdot {\beta }_i^{Use}\cdot A_i/A_{single}\right)$$

(5)

$$Powe{r}_t^{PV}=\sum \left(Ca{p}_{single}^{In}\cdot {\beta }_i^{Use}\cdot A_i/A_{single}\right)\cdot Powe{r}_{i,t}^{PV}$$

(6)

where $Ca{p}_{single}^{In}$ is the rated capacity of a single PV module, $A_i$ is the area of building surface i, $A_{single}$ is the area of a single PV module (using the SolarWorld Sunmodule Pro-Series 250 W polycrystalline solar panel [32]), and ${\beta }_i^{Use}$ is the utilization factor of building surface i, assumed to be 80% for all surfaces. $Powe{r}_{i,t}^{PV}$ represents the photovoltaic power generation of surface i at time t, which can be calculated by Equation (4).

3.1.4. Economic Assessment

The payback period and the self-sufficiency rate are used to evaluate the economic performance of the photovoltaic system. First, the system’s annual revenue and initial investment cost can be calculated as follows:

$$Cos{t}_{invest}=Cos{t}_{cap}\cdot Ca{p}_{PV}^{In}$$

(7)

$$Incom{e}_{year}=Cos{t}_{power}\cdot {Power}^{PV}-Cos{t}_{\mathrm{OM}}\cdot Ca{p}_{PV}^{In}$$

(8)

where $Cos{t}_{cap}$, $Cos{t}_{power}$, and $Cos{t}_{\mathrm{OM}}$ represent the unit capital cost of installed capacity, electricity purchase price, and unit operation and maintenance (O&M) cost, respectively. The economic parameters of the photovoltaic system are derived from China’s 2050 Photovoltaic Development Outlook (2019) [33], while the electricity price data are obtained from the 2020 report published by the Provincial Development and Reform Commission [34,35].

Accordingly, the payback period and the photovoltaic self-sufficiency rate can be calculated as follows:

$$PBP=Cos{t}_{invest}/Incom{e}_{year}$$

(9)

$${SelfUsage=Power}^{PV}/{Power}^{Load}$$

(10)

Finally, the cost per unit of electricity is estimated based on the approximate investment cost. All resulting data, including solar radiation, power generation potential, and economic metrics, are exported in bulk to .shp format files for further use.

3.1.5. Simulation Parameter Settings

A series of assumptions were made in this study. For instance, it was assumed that the photovoltaic system’s power generation efficiency and the inverter conversion efficiency remain constant, and that the available photovoltaic installation area coefficient is consistent across building surfaces of different orientations. The detailed system simulation parameter settings are shown in

Table 2. Simulation parameter settings.

Parameters	Value
${\beta }_{inv}$	90%
${\beta }_{sys}$	80.96%
${\beta }_i^{Use}$	80%
$Cos{t}_{cap}$	3.16 RMB/kW
$Cos{t}_{\mathrm{OM}}$	0.047 RMB/kW
$Cos{t}_{power}$ for Residential	0.5580 RMB/kW
$Cos{t}_{power}$ for Commercial	0.6907 RMB/kW

.

3.2. Data Analysis and Data Interoperability

Since the platform was developed using UE, the batch-exported data must be accurately imported and processed within UE. This is achieved by utilizing the SHPOpen function from the GIS open-source library Shapelib to open .shp files. The SHPGetInfo and SHPReadObject functions are used to retrieve data entries within these files, while the DBFGetFieldInfo function iteratively reads field data for each entry, including attributes such as area, floor count, building name, Shape_Area, ID, object ID, Shape_Length, installed cost, electricity cost savings, effective installed capacity, and kWh cost.

The retrieved data are categorized into three groups: geographical information for reproducing urban architectural models, basic data for calculating power generation, and computed results for display mapping. UE reproduces urban building models based on geometric data from the .shp file, leveraging latitude and longitude coordinates to generate models in the correct geographic location.

Basic data for calculating power generation and installed capacity, as detailed in Table 2, include values such as CenterXYZ for plane centering, VectorXYZ for plane orientation, and power generation calculations using “Rad × Area × efficiency coefficient” for each plane. Installed capacity is computed by dividing Area by the area of a single photovoltaic panel and multiplying it by the panel’s capacity.

Data for display mapping include ten calculated values from RoofCap to SouthPower (shown in

Table 3. Analysis of batch-exported .shp data entries.

Classification	Data Header	Specific Value	Representative Meaning
Geographic Data for Reproducing Urban Architectural Models	Geometry Information Stored in .shp file	The longitude and latitude of each vertex of the building.	Different 3D modeling software uses this longitude and latitude information to generate models of the same geographical location.
	ID	1	Identity Number of Each Building
	Elevation (m)	3	Building Height (determine the height of each building during white mold model slice generation in CesiumLab v0.25, Cesium GS, Inc., Salt Lake City, UT, USA.)
Basic Data for Power Generation Calculations	FaceNum	8	Number of Building Planes
	CenterX/Y/Z	11,355,522.75, 11,355,506.25, 11,355,522.75, 11,355,506.25, 11,355,523.5, 11,355,531.0, 11,355,522.0, 11,355,498.0, 2,215,085.1875, 2,215,092.125, 2,215,085.1875, 2,215,092.125, 2,215,086.75, 2,215,081.75, 2,215,083.625, 2,215,095.625, 2,215,093.75, 2,215,090.5, 0.0, 0.0, 3.0, 3.0, 1.5, 1.5, 1.5, 1.5, 1.5, 1.5	Plane Center Point Coordinates
	VectorX/Y/Z	0.0, 0.0, 0.0, 0.0, 0.3807, 0.8321, −0.3887, −0.8517,0.0, 0.0, 0.0, 0.0, 0.9247, −0.5547, −0.9214, 0.5241, 0.9162, −0.9247, −1.0, −1.0, 1.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0	Plane Normal Vector
	Rad (kWh/m2)	283.7222, 283.7222, 1598.8638, 1598.8384, 425.9791, 922.2407, 960.1941, 665.7827	Solar Radiation Results for Each Plane (Considering occlusions)
	Area (m2)	61.875, 64.125, 61.875, 64.125, 55.1543, 10.8167, 52.0967, 11.4483	Total Area of Each Plane
	Direction	Ground, Ground, Roof, Roof, North, East, South, West, North, South	The Orientation of Each Plane (Identified using VectorXYZ in Python 3.9)
Computed Data for Display Mapping	RoofCap (kW)	16.30588235	Roof Facade Installed Capacity
	NorthCap (kW)	13.9178470588235	North Facade Installed Capacity
	EastCap (kW)	1.39980823529412	East Facade Installed Capacity
	WestCap (kW)	1.48154470588235	West Facade Installed Capacity
	SouthCap (kW)	13.8795411764706	South Facade Installed Capacity
	RoofPower (kWh)	37384.48032	Roof Power Generation
	NorthPower (kWh)	8008.86552001925	North Facade Power Generation
	EastPower (kWh)	1852.73434408509	East Facade Power Generation
	WestPower (kWh)	1404.61523382367	West Facade Power Generation
	SouthPower (kWh)	19,217.9880733783	South Facade Power Generation
	BuildType	1	1 for Residential Buildings, 2 for Public Buildings.

), which are directly mapped in UE. Electricity cost savings and installation costs are determined based on whether the building type is residential or commercial, with calculations performed directly within UE.

3.3. Building Model Generation and Data Mapping

The building information generated in UE was obtained from .shp files, which were processed in Rhino to create the computational results for 3D slices. The Cesium for Unreal plugin was then used to generate the urban buildings. As shown in Table 2, each building has a unique ID corresponding to a data entry, including electricity generation and installed capacity for the east, west, south, north, and roof surfaces. Therefore, each building must be divided into five surfaces in UE to apply the appropriate color materials.

3.3.1. Building Model Slicing and Generation

To create urban building models in UE, the ShapeFile data are converted into basic 3D model grid slices using the CesiumLab toolkit and its geographic data processing platform. First, the basic white model grid slice is selected, ensuring the index field attributes are verified and vertex data are not compressed. The Prj file is then loaded, and the model height is bound to the Floor or Height field attributes. Once the field attributes are successfully bound, the model slices can be accessed using the Cesium for Unreal plugin in Unreal Engine. The relevant field data are subsequently obtained through UE’s Blueprint Visual Scripting material system for further processing and visualization.

3.3.2. Building Model Facade Splitting

Since the basic white model generated by CesiumLab treats each building as a whole, the model needs to be split into five surfaces. In the material script, the direction of each surface is determined by calculating the dot product between the surface normal and the unit vector representing the basic directions (east, west, north, south, and vertical up). The dot product operation compares the surface normal with the reference direction vector, setting a threshold of 0.5 to determine the surface orientation (for example, for the east-facing surface: normal·{1, 0, 0} > 0.5) and using the world coordinate system to ensure accuracy in the direction judgment.

3.3.3. Material and Color Mapping

This study achieves the dynamic visualization of photovoltaic data through Unreal Engine’s Cesium plugin. The key steps include adding the Cesium3DTileset component to import the city model, using CesiumFeaturesMetadata to automatically associate building attribute data, and developing a dedicated material system to map parameters such as electricity generation into color gradients via linear interpolation (Lerp nodes). This method supports querying of photovoltaic data when clicking on buildings, enabling the efficient rendering of hundreds of thousands of building instances.

4. Case Study

4.1. Overview

Wuhan, China, was selected as the research location due to its representative climate and urban layout. The researchers are based in Wuhan, a city located in the central plains of China. Wuhan’s climate is characterized by cold winters and hot, humid summers, making it ideal for studying solar energy potential. Meteorological data were sourced from typical-year datasets available through EPW maps, with the Wuhan weather station located at (30.60° N, 114.05° E) at an altitude of 34.4 m. Wuhan, situated in central China’s plains, experiences cold winters and hot summers, with annual solar radiation ranging between 1163 and 1393 kWh/m²/y. Typical-year meteorological records show temperatures varying from −8 °C to 39 °C and daily radiation levels ranging from 0 to 8 kWh/m² [36,37].

In this study, Wuhan City is divided into research areas according to the fourth-level administrative boundaries (neighborhood blocks), resulting in a total of 182 building blocks and over 200,000 buildings, with a commercial-to-residential building ratio of approximately 5.5:4.5. Since Wuhan lies in the northern hemisphere, solar radiation is generally higher on south-facing surfaces and lower on north-facing surfaces. The platform displays critical metrics such as estimated annual power generation, total installed cost, estimated electricity savings, effective installed capacity, and average electricity cost per kilowatt-hour for all buildings. Users can also generate optimized power generation plans by setting electricity cost savings targets, budget constraints, and photovoltaic panel models.

4.2. Results

The platform calculations show that the total installed capacity of rooftop and facade photovoltaic systems for all buildings in Wuhan is 70.59 GW, with an annual electricity generation of 46.17 TWh, covering 61.1% of the city’s electricity demand. Among this, rooftop photovoltaics contribute the most (22.94 TWh), followed by the east facade (10.54 TWh). The total project investment is 223 billion CNY, with an annual revenue of 42.2 billion CNY. The investment payback period is 5.3 years, showing good economic feasibility, with a self-consumption rate of 61.2%, indicating a high-energy-self-sufficiency potential (

Table 4. Total values for Wuhan.

Generation of Each Face (TWh/year)					Capacity of Each Face (GW)
Roof	South	East	West	North	Roof	South	East	West	North
22.94	4.52	10.54	4.20	3.97	17.98	17.75	17.71	8.61	8.54
Building PV Costs (Billion RMB)		Building PV Incomes (Billion RMB/year)		PBP (year)		Power Demand (TWh/year)		Self-Use Rate (%)
223		42.2		5.3		75.5		61.2

).

The annual daily photovoltaic generation of a single solar panel in Wuhan shows a trend of lower generation in summer and higher generation in winter, as shown in Figure 6. The highest daily generation is 3.85 kWh/m², occurring on March 2. Despite the strong solar radiation in Wuhan during the summer, the high working temperature reduces its efficiency, leading to lower power generation.

The seasonal electricity generation of a single photovoltaic panel on each wall of Wuhan is shown in Figure 7. Due to temperature and radiation effects, the rooftop generation is relatively balanced throughout the seasons (70–80 kWh/m²). The electricity generation characteristics of the east and west walls are similar in each season, with higher generation in spring and summer, and the lowest generation in winter. The south wall exhibits more significant seasonal fluctuations in power generation, with the lowest generation in summer (only 25 kWh/m²). This is due to the severe deviation of the solar incidence angle from the optimal tilt angle during summer in low-latitude regions. In winter, the generation is highest (60 kWh/m²), approximately 2.4 times that of summer generation, and the difference from the rooftop winter generation is smaller. This is because the winter temperature is lower, and the solar incidence angle on the south wall in winter is closer to the optimal tilt angle, leading to higher direct solar radiation reception.

4.3. Interface Display Content and Effects

When users click on Wuhan in the two-dimensional map of China, the interface transitions to a three-dimensional visualization of Wuhan’s spatial layout and buildings. As illustrated in Figure 8, users can freely navigate the virtual environment to explore the estimated annual power generation for the rooftops and four facades of each building. The annual power generation is represented using a color gradient ranging from blue (low values) to red (high values). Hovering the mouse over any building reveals detailed information, including building name, coordinates, estimated annual power generation, total installed cost, estimated electricity savings, effective installed capacity, and average electricity cost. Additionally, the platform identifies and highlights the top 100 buildings with the highest total power generation, and the relevant detailed information of each building can be exported (

Table 5. Information on the top 10 buildings in Wuhan.

Index	Height (m)	Area (m2)	PBP (Year)	Power (GWh/Year)					Capacity (MW)
				Roof	South	East	West	North	Roof	South	East	West	North
1	54	478,122	4.87	79.16	4.73	2.95	2.91	2.20	56.92	5.33	4.04	4.03	5.43
2	54	478,122	4.88	79.16	4.48	2.95	2.89	2.20	56.92	5.33	4.04	4.03	5.43
3	18	148,161	4.79	24.52	1.64	0.28	0.29	0.65	17.64	1.87	0.43	0.42	1.87
4	15	106,328	4.82	17.31	0.39	0.33	0.23	0.32	12.66	0.71	0.44	0.44	0.91
5	12	94,659	4.71	15.64	0.56	0.31	0.22	0.19	11.27	0.70	0.42	0.42	0.70
6	12	91,042	4.68	14.99	0.50	0.33	0.33	0.19	10.84	0.59	0.47	0.48	0.59
7	21	76,910	5.67	12.66	1.51	0.27	0.62	0.54	9.16	1.92	0.61	0.88	2.25
8	21	81,569	3.73	12.81	0.54	0.44	0.45	0.81	9.71	0.95	1.06	0.58	3.14
9	3	86,288	4.35	14.28	0.30	0.00	0.00	0.17	10.27	0.33	0.00	0.00	0.33
10	12	80,151	4.70	13.24	0.53	0.20	0.19	0.26	9.54	0.68	0.31	0.29	0.68

).

Users can also access a selection box to switch between different data visualizations. For instance, as shown in Figure 9, when the “Estimated Installed Cost” option is selected, the buildings are color-coded based on their total installation cost, using a gradient from yellow (lower cost) to red (higher cost). Alternatively, clicking the “Estimated Electricity Savings” option changes the building colors accordingly, distinguishing between commercial buildings (blue) and residential buildings (pink).

To optimize power generation within a specific budget, users can access the “Simulation Plan” feature (Figure 10). After entering parameters such as electricity savings target, budget, and preferred photovoltaic panel model, the platform calculates the optimal solution. The recommendations consider building power consumption types, annual power generation potential, and installation costs. The interface highlights the relevant buildings and displays specific details of the proposed plan.

For critical or landmark buildings, the platform supports importing high-precision three-dimensional models. As shown in Figure 11, users can interactively select and adjust photovoltaic panel paving areas on the building’s surfaces to achieve more accurate performance estimates.

5. Discussion and Conclusions

The platform integrates solar radiation, building models, and economic data through 3D visualization technology, providing electricity generation, installed capacity, and investment return analysis for urban-level photovoltaic deployment. It supports macro-level decision-making and investment optimization. Its core advantage lies in efficient urban-scale evaluation, but due to the research focus, it does not consider factors such as the type of photovoltaic components (thin-film modules, photovoltaic bricks, photovoltaic tiles, photovoltaic curtain walls, etc.), degradation of photovoltaic systems, performance of photovoltaic materials (monocrystalline silicon, polycrystalline silicon, etc.), rooftop layout details, and the impact of facade photovoltaics on building energy consumption. These simplifications aim to balance computational efficiency and generality, focusing on rapid evaluations needed for policy-making rather than detailed analysis of individual buildings.

The platform’s limitations primarily stem from its positioning for city-level planning, where it prioritizes overall data over micro-level variables. For instance, ignoring component differences and degradation effects reduces model complexity, while omitting the impacts of rooftop layouts and building energy consumption avoids coupling additional simulation tools. Future upgrades could gradually incorporate more details through modular enhancements, but a balance must be struck between computational efficiency and accuracy to ensure the platform remains practical and operable for energy planning.

The platform pre-calculates the relevant data for each building in all cities that can be viewed, significantly improving the platform’s computational speed. For example, Wuhan spent 1 h pre-calculating data in Rhino, which were then transferred to UE to accelerate display speed in UE. Tests showed that on a computer configured with Intel i5-13600KF/AMD RX 7900 XT/32 GB DDR5, running a city photovoltaic project took about 2 s to start, and loading an offline map level for a city with a road network took 3–4 s. Loading a level without a road network using an online map took less than 1 s, with most scenes running at a frame rate exceeding 90 frames per second. Switching between white model datasets took no more than 1 s.

Future upgrades to the platform will focus on three areas: first, integrating building electricity loads, photovoltaic component performance, and degradation models to optimize generation-demand matching; second, adding energy consumption impact assessments and rooftop layout optimization tools to address current limitations; third, enabling batch display of photovoltaic arrays and dynamic regional analysis to improve decision-making accuracy. In the future, the platform will incorporate more diverse analytical scenarios. For instance, at the single-building level, users will be able to set different PV panel types or installation coverage ratios to explore how different photovoltaic system parameters affect investment decisions. At the regional level, the platform will provide BIPV potential analyses across different building types—including residential, industrial, and commercial buildings—offering more realistic and policy-relevant insights for decision-makers. Through these multidimensional upgrades, the platform will balance macro-level planning with micro-level optimization needs, providing a more comprehensive solution for urban photovoltaic deployment.