clicSAND for OSeMOSYS: A User-Friendly Interface Using Open-Source Optimisation Software for Energy System Modelling Analysis

Abstract

Energy modelling plays a crucial role in assisting governmental and policymaking bodies to strategise long-term investments within the context of energy transition. Among the well-established open-source optimisation models, OSeMOSYS—the Open-Source Energy Modelling System—stands out. This paper introduces clicSAND, a novel user interface designed for OSeMOSYS, aimed at reducing the learning curve and supporting novice energy modelers in efficiently conducting long-term investment analyses. clicSAND, freely available and open-source, features a user-friendly Excel interface for data input, integrated solvers, and a visualisation dashboard for result interpretation. The outcomes, projected up to 2070, hold the potential to inform policy decisions and mobilise financial resources for sustainable development endeavors, such as ensuring affordable and secure energy supply and mitigating climate change impacts. This advancement not only democratises access to energy modelling tools but also empowers policymakers and stakeholders to conduct thorough long-term investment analyses with ease. This paper elaborates on clicSAND’s key advantages, architecture, and functionalities. Additionally, it discusses the evolutionary journey from clicSAND 1.0 to 3.0, emphasising a commitment to continuous improvement and user-centric adaptation, thereby enhancing its utility and relevance. The inclusion of a South African case study, conducted during the EMP-A (Energy Modelling Platform for Africa) 2021 international capacity-building event, showcases clicSAND’s efficacy in facilitating knowledge transfer and skill development among inexperienced users, while providing a tangible example of its application in addressing specific regional energy challenges and policy contexts. Finally, current applications and future extensions of the software are also presented.

1. Scientific Significance and Purpose of the Study

The climate change crisis demands bold and courageous action from individuals and governments. One such action is the ‘energy transition’, which entails shifting away from a carbon-based economy towards less-polluting options like renewable energy sources. This transition necessitates multi-billion-dollar investments and long-term scenario planning extending to 2050, 2070, and beyond. To accomplish this, governments, academics, international organisations, and private-sector consultancies increasingly rely on energy models for long-term investment planning. These models offer valuable insights for efficient and reliable energy production aligned with climate change mitigation efforts. However, they are often complex and demanding, requiring a significant investment of time and resources to acquire the necessary skills. Hence, this study aims to explore how advancements in user interface design within energy modelling software can enhance accessibility, efficiency, and democratisation of access. This, in turn, can facilitate informed decision-making and sustainable development initiatives among policymakers and stakeholders.

1.1. A Focus on OSeMOSYS—The Open-Source Energy Modelling System

While there are numerous energy modelling tools available, this study focuses specifically on the Open-Source Energy Modelling System (OSeMOSYS) [1]. OSeMOSYS was selected due to its widespread adoption [2], robustness, and several distinct features that set it apart within the field. One notable feature is its recognition as being particularly well-suited for analysing energy scenarios in developing and emerging countries [3]. OSeMOSYS has been identified as the second-highest energy modelling tool used for analysis, especially for examining energy supply pathways and reducing greenhouse gas (GHG) emissions while effectively capturing the unique features of developing countries’ energy systems [3]. Moreover, another systematic literature review on energy system optimisation models (ESOMs) identified OSeMOSYS as “the most advanced in open science due to its diverse channels to communicate with users and developers and a further scope within the energy policy community” and of being “at the forefront of open-source codes due to its straightforward, elegant, and transparent setting, the adaptation for simple refinements, and the ability to conduct sophisticated analyses” [4]. Initially developed at the Royal Institute of Technology (KTH) in collaboration with various institutions and organisations [5], OSeMOSYS is a long-term bottom-up optimisation model for energy system analysis leading to better investment planning, as demonstrated by its use in various applications [6,7,8] and in the newly developed Data-to-Deal (D2D) approach [9].

By grounding our study within the context of OSeMOSYS, we acknowledge the significance of leveraging existing advancements in energy modelling. Our focus on enhancing user interface design within this established framework aims to build upon its strengths, further improving accessibility and usability. In doing so, we aspire to contribute to the broader goal of enabling informed decision-making and driving the transition towards sustainable energy systems.

State of the Art of OSeMOSYS User Interfaces

Throughout the continued development of OSeMOSYS over the past 15 years, developers focused on improving the user interface and user-friendliness, which enable skills to be developed faster and accelerate the uptake of models towards the formulation of sustainable development policies. Initially, users had to interact directly with the code and the input data file via text editor software and the command line, which was time-consuming and difficult for non-advanced users. As a result, a stand-alone open-source interface called MoManI (Model Management Infrastructure) was developed [10]. However, this interface had significant limitations in terms of data management [11]. The OSeMOSYS community tried to replicate the same functionality within an Excel workbook, but this informal process led to duplication of effort and scattered resources.

Cannone [11] addressed this problem by developing an Excel-based interface for OSeMOSYS called SAND (Simple And Nearly Done). Although the Excel interface improved the collection and input of data, the user still had to interact with the command line via simple lines of code to run the models, and computation time could be a few hours. Thus, stemming from the Excel SAND Interface, clicSAND is a user-friendly software interface designed for OSeMOSYS, created to overcome these limitations. It adds integrated solvers, one-click installation, and a graphical user-interface. clicSAND provides the OSeMOSYS community with an easy-to-use interface for energy modelling analysis and has been widely used for several applications (see Section 4.2). The clicSAND software helps model alternative long-term pathways of a selected region, identifying the optimal energy mix configuration, including allocation of energy resources and associated financial investment. Moreover, clicSAND is available for Windows and Mac users [12], increasing the accessibility of this tool to a wider audience.

The scientific novelty of the presented clicSAND software lies in its user interface innovation and its capacity to enhance accessibility and efficiency in energy modelling tasks, particularly for novice users. By introducing a user-friendly Excel interface coupled with integrated solvers and a visualisation dashboard, clicSAND significantly reduces the learning curve associated with complex energy modelling systems like OSeMOSYS. This advancement not only democratises access to energy modelling tools but also empowers policymakers and stakeholders to conduct thorough long-term investment analyses with ease.

The remainder of the paper is structured as follows: Section 2 presents clicSAND’s architecture and main functionalities. Section 3 provides a brief, illustrative example of clicSAND use. Finally, Section 4 concludes by presenting current and past software applications, extensions, and future work.

2. Software Description

This section outlines the key features of the clicSAND software in its initial version, clicSAND 1.0, and highlights the enhancements introduced in its third iteration, clicSAND 3.0. A brief mention will also be made of the macOS version of the software, clicSAND 2.0.

2.1. Key Features of clicSAND 1.0 Software

A basic overview of the clicSAND 1.0 software and its main functionalities is provided in Figure 1, while the primary advantages and limitations are summarised in

Table 1. Advantages and limitations of the clicSAND 1.0. Software.

clicSAND 1.0. Software
Advantages	Limitations
Users do not need to interact with the command line at any stage. Features a user-friendly SAND Interface, based on Excel, making it accessible to most IT users. A single spreadsheet with filtering capabilities is used for inputting all necessary data. Installation is straightforward and swift. Provides full compatibility with the Starter Data Kits models. Has a free, self-learn tutorial in the form of an Open University course with certification. All OSeMOSYS parameters are represented. It is freely available as a public resource for everyone. It is compatible with both Windows and macOS operating systems. It operates entirely offline.	Troubleshooting proved time-consuming as the software did not point out errors. A maximum of 96 time slices could be employed for seasonal representations of intermittent renewable technologies. The software supported a maximum of 200 technologies, 50 commodities, and 5 types of emissions. It has not been tested for multiregional studies. Vertical copy-pasting could not be done when filters were applied to the Excel file.

. It utilises the SAND Excel interface as the sole data entry point for the user. A significant usability advantage is that no interaction with the command line is necessary during the modelling process. Furthermore, the software operates entirely offline, which is especially beneficial for regions with low-bandwidth connections.

A one-click installation package containing all the necessary components of the software is freely available for download on GitHub [13]. Instructions on correctly installing the clicSAND 1.0 software and the two free solvers can be found on Zenodo [14]. Additionally, a step-by-step online certified course hosted on the Open Learn Create (OLC) Platform is available as a public good [15]. This course combines theoretical lessons with practical exercises, teaching users the theory behind energy modelling for investment planning and how to build an energy model from scratch using the clicSAND 1.0 software for OSeMOSYS. The authors recommend that first-time users of the clicSAND 1.0 software enrol in this online course and carefully follow the instructions provided in practical exercises 1, 2, and 3:

• Hands-on 1: Download and installation of the clicSAND 1.0 software and the solvers (GLPK and CBC).
• Hands-on 2: Best practices for inputting data into the Excel SAND Interface and practical examples for one technology.
• Hands-on 3: Instructions on how to save, run, and visualise results using the Microsoft Access database and the Excel template provided with the software.

Another asset of the clicSAND 1.0 software lies in its compatibility with a related project aimed at creating Starter Data Kits (SDKs) [16,17]. These SDKs consist of national datasets and simple energy system models, currently available for 69 countries in Africa, East Asia, and South America (Appendix A). Notably, the models in the Starter Data Kits were constructed using the Excel SAND Interface, and the results were obtained by running the clicSAND 1.0 software. The scientific significance of the SDK lies in its facilitation of access to national datasets and energy system models, thereby streamlining research processes and enhancing the efficiency of energy analysis. By providing a comprehensive set of predefined parameters within the SAND file, the SDK eliminates the arduous task of manual data entry, allowing researchers to focus more on analysis and interpretation.

One key requirement of this first version of the software is the need for at least 8 GB of RAM for smooth operation, along with a Windows computer with Microsoft Access and Excel installed, both of which are commercial software. However, to broaden the accessibility of this public good and to overcome the need for a Windows operating system, a second version of the clicSAND software (clicSAND 2.0), which is compatible with macOS and does not require Microsoft Access, was recently released. Details regarding this are provided in Section 4.1.1 of this paper. Similarly, to address the 8 GB RAM requirement, a third version of the software, clicSAND 3.0, was developed. This version offers the flexibility to execute a model on an online cloud platform, thereby eliminating the need for dependence on the computer’s memory.

The Excel SAND interface is specifically designed for beginner and intermediate users, allowing them to work on middle-sized models with basic regional and temporal resolution. As explained in Section 2.3, the Excel SAND Interface serves as the platform where users can intuitively input data for up to 200 technologies, 50 commodities, five types of emissions, one region, and 96 time slices.

It is important to note that when initially using the Excel Interface to input data, users need to be cautious when pasting data while a filter is applied, as Excel will copy the data to the subsequent row and not to the next filtered row. Nevertheless, to assist users in avoiding mistakes during model compilation, step-by-step video guidance has been released on YouTube, demonstrating best practices for effectively inputting data [18].

Another weakness of the first version of the clicSAND software for Windows was its inability to detect errors if the model failed during the run. In fact, the software platform depicted in Figure 1 does not specify the exact error that might occur; it simply indicates that the model is not functioning. This issue also arose when running OSeMOSYS models with other interfaces. To address this limitation and enhance the troubleshooting experience, clicSAND 3.0 has been equipped with functionality to identify in which step of the process the model is failing, as detailed in Section 2.2 of this paper.

All versions of the clicSAND software are fully compatible with the existing OSeMOSYS architecture, meaning that the constraints, variables, and parameters are all represented. Additionally, two widely used solvers to find a model’s optimal solution, GLPK (GNU Linear Programming Kit) [19] and CBC (COIN-OR branch and cut) [20], are included as part of the open-source and freely available clicSAND software.

Moreover, the results produced by the software are fully compatible with other tools such as CLEWS (Climate, Land, Food, Energy, and Water systems approach) [21], the IRENA FlexTool for analysing system flexibility, MUSE (Modular energy system Simulation Environment) [22], and other OSeMOSYS platforms, including the OSeMOSYS Cloud platform [23] (which provides online solvers through a cloud service). Further details on this compatibility are provided in Section 2.4.2 of this paper.

2.2. Enhancements in the clicSAND 1.0 Software: The Release of clicSAND 3.0

The decision to enhance the functionalities of clicSAND 1.0 and release an improved version, clicSAND 3.0, arose from a continuous improvement process. This process involved collecting feedback from users through surveys distributed at the conclusion of various capacity-building events. During these events, inexperienced users were tasked with constructing basic national OSeMOSYS models within a three-week timeframe using clicSAND 1.0. As indicated in

Table A2. Survey results—clicSAND 1.0.

Statement	Rating (Out of 100)	Number of Responses
The SAND Excel Interface was easy to use and user-friendly	74	50/70
The clicSAND solver was easy to use and user-friendly	75	50/70
The clicSAND solver produced results in a timely manner	60	50/70
Running the data file extracted from the SAND Interface on the OSeMOSYS Cloud was fast and easy to use	70	50/70

(Appendix B), the feedback received confirmed the limitations that developers had already identified, primarily concerning extended computational times and the laborious process of visualising results through an Access database.

To address these limitations, the authors developed, tested, and implemented a new version of the clicSAND software, known as clicSAND 3.0. Drawing inspiration from the innovations in clicSANDMac (clicSAND 2.0), this version of the software embeds a Python code that effectively reduces the data file size (both matrix and LP-file) by pre-filtering unused elements—available for download on GitHub [24]. This optimisation significantly accelerates the solving process. Additionally, this feature eliminates the need for an Access database to visualise results, which will be further elaborated upon in Section 2.5.2. Instructions to install clicSAND 3.0 and run the model on the OSeMOSYS Cloud are available on Zenodo [25]. Specific teaching material and exercises using clicSAND 3.0 are also freely available as a public good [26].

As discussed in the preceding section, clicSAND 1.0 had a specific requirement, necessitating a minimum of 8 GB of RAM for smooth model execution. However, clicSAND 3.0 has comprehensively addressed this requirement by introducing the capability to run models not only offline but also online through a cloud-based platform known as OSeMOSYS Cloud. This innovative approach circumvents the need to rely entirely on the computer’s internal memory, as elaborated further in Section 2.4.2.

To facilitate this novel functionality, significant modifications were made to the user interface of clicSAND 3.0. As shown in Figure 2, it now includes a button labelled “Generate OSeMOSYS Cloud input”. This button streamlines the process of formatting the .txt file generated by the Excel SAND interface into a compatible format tailored specifically for seamless integration with the OSeMOSYS Cloud platform. This transformative enhancement substantially expands the accessibility and flexibility of clicSAND 3.0, empowering users to tap into cloud resources without being restricted by local hardware limitations.

The integration with the OSeMOSYS Cloud platform greatly enhances the software’s convenience and adaptability, allowing users to access models and results from virtually anywhere, at any time. This development marks a significant step forward in making clicSAND 3.0 a versatile and user-centric tool in the field.

As highlighted in Table 1, one of the prominent limitations of clicSAND 1.0 revolved around troubleshooting and debugging challenges. In this earlier version, inexperienced users encountered significant difficulty in identifying errors because the user platform did not provide clear indications of where the error occurred, making the debugging process a challenging task.

However, with the introduction of clicSAND 3.0, this challenge has been effectively addressed through two key enhancements. First, when running a model offline, the user platform now displays the specific line where the error occurred, greatly simplifying the process of identifying and rectifying mistakes. Second, when utilising the OSeMOSYS Cloud to run a model, users have the ability to monitor the entire solving process in real-time, allowing them to pinpoint precisely where the process encounters an issue.

Furthermore, the integration with the cloud offers an exceptional feature for debugging: as a clicSAND model runs, OSeMOSYS Cloud generates a visualisation of the Reference Energy System for that model. This empowers users to effortlessly visualise their input data with a single click [23]. Any missing connections within the model are highlighted in red, providing valuable assistance to users in identifying and rectifying bugs within their models. These advancements collectively enhance the user experience, making troubleshooting and debugging significantly more accessible and efficient in clicSAND 3.0.

As summarised in

Table 2. Advantages and limitations of the improved version of the software, clicSAND 3.0.

clicSAND 3.0 Software
Advantages	Limitations
Users do not need to interact with the command line at any stage. Features a user-friendly SAND Interface, based on Excel, making it accessible to most IT users. A single spreadsheet with filtering capabilities is used for inputting all necessary data. Installation is straightforward and swift. Provides full compatibility with the Starter Data Kits models. Has a free, self-learn tutorial in the form of an Open University course with certification. Additional step-by-step video lectures are available. All OSeMOSYS parameters are represented. It is freely available as a public resource for everyone. It is compatible with both Windows and macOS operating systems. Embedded Python code reduces data file size and consequently speeds up computational time. The software provides the option to operate completely offline. It also allows users to run the model using an online cloud-based service, eliminating the need to rely on the machine’s internal memory (RAM). There is no need for an Access database for results visualisation. A new offline Excel visualisation template is included, generating high-quality graphs. It seamlessly integrates with the OSeMOSYS Cloud and provides the option to visualise results offline.	A maximum of 96 time slices could be employed for seasonal representations of intermittent renewable technologies. The software supports a maximum of 200 technologies, 50 commodities, and 5 types of emissions. It has not been tested for multiregional studies. Vertical copy-pasting cannot be done when filters are applied to the Excel file.

, clicSAND 3.0 maintains the benefits introduced in clicSAND 1.0 while introducing enhanced features. These include embedded Python code, which reduces data file size and accelerates computation (with clicSAND 3.0, a full Starter Data Kit runs in less than 10 min). Users now have the flexibility to work offline or access a cloud-based service, eliminating the need for dependence on local memory. The tool also streamlines result visualisation with an offline Excel template and seamlessly integrates with the OSeMOSYS Cloud, enabling offline result visualisation.

2.3. Software Licenses

The clicSAND software and its components are licensed under an MIT license (open-source and free of charge), which permits the public to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the software. Users can do so by giving credit, linking to the license, and indicating if they have made any changes to the work in all copies or substantial portions of the software. clicSAND is written using the C# language, enabling easy uptake, improvements, and adaptability.

2.4. Operating the Software

2.4.1. clicSAND 1.0

As mentioned earlier, the user downloads the one-click installation package from GitHub [13], runs the executable, and follows the instructions for installing the GLPK solver and CBC solver available on Zenodo [14], either as part of the OLC online course [15] or in the ReadMe file of the clicSAND repository [24]. The one-click installation package contains the following components:

• User platform (Figure 1);
• Excel SAND Interface to input data (see Section 2.5.1);
• Access database to import results;
• Excel template to visualise the results;
• OSeMOSYS code needed by the solvers.

The user populates the SAND Interface on Excel using a single spreadsheet where all the OSeMOSYS sets and parameters can be defined. Additionally, all parameters are initially set to default values to save time on entering non-variable data. The spreadsheet uses filters to guide the user in compiling the data file. The flowchart in Figure 3 describes all the process steps, while specific functionalities of the SAND Interface are presented in Section 2.5.1.

Next, the user saves and closes the Excel file to open the clicSAND executable (Figure 1). The user selects the SAND Excel file and then the OSeMOSYS code by clicking respectively on the “Data Source (xls)” button […] and “Model” button […] as presented in Figure 1. The OSeMOSYS code contains the mathematical model with the definition of sets, parameters, variables, and constraints. By using the arrows of the “Ratio (CBC)” button, the user can, if needed, change the accuracy of the model solution (a 0.05 ratio, which is usually used, means that a 5% range of error on the optimal solution is accepted). A lower ratio will result in a longer computational time.

A VBA (Visual Basic for Applications) Macro in the selected SAND Excel workbook is executed by clicking ‘Run’. The macro copies the last tab in SAND, named “ToDataFile”, into a new text file. The first solver, GLPK, will generate the matrix based on the data file and code selected. Then the second solver, CBC, will find the optimal solution and produce results in a text file format. If the model successfully runs, the large text box in Figure 1 will show lines of text describing the process followed and ends up with a sentence saying the “Optimal Solution was found”.

At that point, the results obtained will have been saved in a .txt file and should then be imported into the Access database. The last step to visualise the results requires the user to link the Excel Result Visualisation template with the Access database. The authors highly recommend following the step-by-step instructions on how to run and visualise the results provided in Hands-on 3 of the OLC Course [27] and the YouTube video [18].

2.4.2. clicSAND 3.0

Figure 4 illustrates the primary operational contrast between the third and first versions of the software. Specifically, it outlines the steps for a fully offline run with clicSAND 3.0. Meanwhile, Figure 5 details the procedures for a partially online run, which does not rely on the computer’s memory but utilises the OSeMOSYS Cloud platform.

Offline Run

When it comes to solving the model in clicSAND 3.0, clicking the “Run” button initiates the execution of an embedded Python code. This Python code, seamlessly integrated into the software, eliminates the need for users to possess Python skills and to engage with any coding. The code, which is also available in the folder “Python scripts” of the GitHub repository [24], scans the existing data combinations in the “ToDataFile” tab of the Excel SAND interface, where user-added data are stored. Consequently, it generates a considerably smaller matrix and data file that exclusively includes the existing combinations of technologies and fuels, resulting in a significant reduction in data file size. While a data file in clicSAND 1.0 was typically around 50MB, in clicSAND 3.0, it is approximately 1MB. This reduction in data file size significantly accelerates processing by the solvers, with a Starter Data Kit now taking only 5–6 min, as opposed to the previous 1.5–2 h. Another noteworthy difference in clicSAND 3.0 is that, upon solver completion, an automatic Result File in .csv format is generated. This .csv file can be directly imported into the newly created Excel “Results Template” with just a single click. Consequently, there is no longer a need for the Microsoft Access database and for the time-consuming process associated with it required in clicSAND 1.0. The “Results Template” facilitates the comparison of multiple scenarios and allows for the visualisation of up to 25 graphs (more details in Section 2.5.2).

Partially Online Run

This section outlines what happens in the clicSAND 3.0 interface when users opt to utilise the OSeMOSYS Cloud platform to run their models rather than relying on their machine’s internal RAM. If this option is chosen, users need to initiate the process by clicking on the newly introduced “Generate OSeMOSYS Cloud Input” button, as shown in Figure 2. Similarly to the offline run, the embedded Python code is executed, resulting in the generation of a compressed data file in the format required by the OSeMOSYS Cloud to run an OSeMOSYS model. At this point, users move to the OSeMOSYS Cloud platform. After creating a new run, they select the OSeMOSYS code and the compressed data file, then click “Run”. This initiates a process involving the creation of the matrix and so forth, ultimately producing the results. To view these results, users must download the Results.txt file from the cloud. Subsequently, they can employ the Results Converter app to transform the .txt file into a .csv format, which is ready for one-click import into the new Excel Results template, as per the offline version explained earlier. The benefits of running online are that it does not burden the internal RAM, and users can run multiple models simultaneously. However, it necessitates the additional step of using the results converter to visualise the graphs.

2.5. Software Functionalities

2.5.1. Input Data through the SAND Excel Interface

The clicSAND software comprises several components, but in this section, we will specifically discuss two of them—the SAND Excel interface and the Result Visualisation template—since users primarily interact with these for data management. Illustrated in Figure 6, the SAND Interface is an Excel workbook containing four sheets: Naming, SETS, Parameters, and ToDataFile.

In the “Naming” sheet, users find descriptions of all parameters utilised to construct and constrain the model. The “SETS” sheet enables users to define the names of up to 200 technologies, 50 commodities, and five types of emissions within their model. Users can revisit this sheet and make unlimited changes without losing progress. Any naming conventions added in the “SETS” sheet are automatically inserted into the other two sheets: “Parameters” and “ToDataFile”.

The “Parameters” sheet (see screenshot in Figure 6) is a worksheet containing 48,757 rows, encompassing all the OSeMOSYS parameters necessary for building a model. Each column is equipped with filters to enhance the interface’s usability. This enables users to apply multiple filters to pinpoint the desired data entry point. For instance, to input fixed costs for renewable technologies like geothermal, solar, wind, and hydro (as depicted in Figure 6), users access the Parameters sheet, filter out the fixed costs in column A, then filter out the selected technologies in column C, and finally, insert the data into columns K to BN (representing years 2015 to 2070).

The user follows this process for all data inputs in their model. The interface provides a modelling period spanning up to 55 years (from 2015 to 2070) and up to 96 time slices. Nevertheless, users have the option to model a shorter period or decrease the temporal resolution (number of time slices) as per their requirements (refer to guided exercises [28,29] and video recordings [30,31]).

In all versions of clicSAND, the user does not directly interact with the “ToDataFile” sheet to input modelling data. This sheet is pre-formatted to meet the requirements of the solvers, automatically capturing all data added from SETS and the Parameters sheet. When the user clicks ‘Run’ in clicSAND 1.0 (Figure 1), a VBA macro embedded in the Excel SAND Interface is triggered. This macro opens the SAND Interface, retrieves the “ToDataFile” sheet, and saves it in a new .txt file used by the solvers to find the optimal solution.

However, in clicSAND 3.0, an additional step is performed when clicking ‘Run’. The .txt file generated by the VBA macro undergoes processing with an embedded Python code. This code selectively includes only the existing combinations with non-default values, significantly reducing the size of the .txt file. As a result, the computational time is greatly shortened compared to that of clicSAND 1.0, as explained earlier in Section 2.4.2.

2.5.2. Results Visualisation Templates

clicSAND 1.0

In the Results Visualisation Template file (a separate Excel workbook), users can visualise the results of their modelling exercises (refer to the available graphs in the green box of Figure 6). These graphs are generated using pivot tables and depend on an Access database to store the data in a compatible format. In the following section, an illustrative example of an analysis using this software will be presented.

clicSAND 3.0

In clicSAND 3.0, we have introduced a new visualisation template aimed at enhancing usability and flexibility. Figure 7 presents an example of the template applied to a case study with three different scenarios. This template, also Excel-based, eliminates the need for Microsoft Access, streamlining the entire process. In the case of an offline run, a ready-to-use .csv file is generated in the correct format for easy copying and pasting into the visualisation template. However, if the run is conducted via the OSeMOSYS Cloud, it produces a .txt file, which is subsequently processed using a dedicated tool called the Results Converter (See Figure 8), included in the installation package of clicSAND 3.0. This converter transforms the .txt format into an equivalent .csv format, making it compatible with the visualisation template. After inputting the results, users can easily generate a series of graphs automatically by following a few simple steps. One notable addition is the feature to graph any number of scenarios, making it effortless to compare them, as shown in Figure 7, with three scenarios analysed. Furthermore, each graph includes slicers that allow users to select any combination of parameters to be displayed, enabling a wide range of analyses. To assist users in using this new visualisation template, the authors have developed a training video that demonstrates its functionality [32].

3. Illustrative Example of the Application of the clicSAND Software

This section presents the outcomes of a case study conducted for South Africa as part of the EMP-A (Energy Modelling Platform for Africa) 2021 capacity-building event [33]. The inclusion of such illustrative examples is crucial in demonstrating the practical application and capabilities of the clicSAND software and its interface. These examples provide concrete scenarios where clicSAND can effectively be used for energy modelling analyses, thereby providing valuable insights for decision-making processes. Specifically, the South Africa case study illustrates how clicSAND can analyse and evaluate energy system dynamics, resource allocation, and investment strategies within a real-world context. Through simulations of various scenarios and policy interventions, stakeholders can assess potential impacts on energy supply, demand, and environmental sustainability, thus informing strategic decisions and policy formulation processes. Furthermore, the South Africa case study serves as a reference point for users to understand the functionalities of clicSAND and its relevance to specific regional or national contexts. It demonstrates the versatility of clicSAND in addressing diverse energy challenges and emphasises its applicability across various geographical locations and socioeconomic settings.

The results of this case study come from a three-week online training course centred on clicSAND 1.0 for OSeMOSYS, as the third version of the software was not released at that time. Participants had no prior knowledge of OSeMOSYS. It is notable that the format of the results presented in Figure 9 was achieved using an additional scientific plotting package called Veusz [34], for post-processing graphics production, which was fully compatible with the Excel template generated with clicSAND 1.0 software. At the time of the initial writing, an older version of clicSAND was in operation, requiring Veusz for visualisations. However, with clicSAND 3.0, users can now visualise high-quality results using the new Excel Visualisation Template, as explained in Section 2.4.2, without the need of the software Veusz.

3.1. Background and Modelling Questions

The South African power system relies significantly on coal [35]. As a result, an exercise was conducted to explore the potential implications of reduced coal dependence compared to existing policy and the least-cost scenario. In order to create a representative scenario, all coal capacity would be decommissioned by 2040. The exercise considered the following questions:

• How will the energy mix evolve?
• How will CO₂ emissions change?
• What roles will different technologies play?

3.2. Modelled Scenarios and Starter Data Kit

The modelled scenarios are summarised in

Table 3. Summary of scenarios modelled including description and critical assumptions.

Scenario Label	Scenario Description	Key Assumptions
Existing Policy	The power sector evolves with existing policy [38]	Coal capacity decommissions in line with existing expectations (Coal < 60%, 25% renewables by 2030) Peak Plateau Decline (PPD) trajectory
Least Cost	Least-cost evolution with no substantial upfront constraints	Least-cost optimal unconstrained path PPD trajectory
Game Change	No coal generation by 2040; renewable energy revolution	All coal capacity decommissioned by 2040—linear decline (over ten years) PPD trajectory Least-cost path with the above constraint

. The base model for the ‘Existing Policy’ scenario utilised an adjusted version of the available South African Starter Data Kit for implementation in OSeMOSYS [36,37], with further adjustments made (described in Table 3) to define the ‘Least Cost’ and ‘Game Change’ scenarios.

3.3. Discussion of Results

Compared to the Existing Policy scenario, both the Least Cost and Game Change scenarios show a trend towards a more diversified clean energy mix (Figure 9). In these scenarios, new-build capacity is predominantly comprised of solar photovoltaics and onshore wind, which serve to replace decommissioned coal capacity as per the planned transition, while also meeting the increasing energy demand. Transitioning from the Least Cost scenario to the Game Change scenario, there is a further increase in the deployment of solar photovoltaic and onshore wind technologies. Additionally, gas-fired generation capacity is introduced to supplement the energy gap left by the decommissioned coal capacity by 2040.

3.4. Policy Implications and Future Work

Based on the findings of these representative scenarios (see Figure 9a), early preparation for a transitioned power sector in South Africa is recommended. The primary objective is to gradually move away from coal-fired power generation through phased implementation, incorporating geographically tailored policy interventions, and conducting assessments of associated socio-economic costs and benefits.

Figure 9b compares the total system and discounted costs across the scenarios. Given the potentially significant role of variable renewable energy (VRE) such as solar photovoltaics and wind in South Africa (refer to Figure 9a), it is advisable to prioritise the localisation of the VRE supply chain whenever feasible. However, this localisation may lead to increased electricity costs.

While the adoption of the considered technologies results in reduced CO₂ emissions (see Figure 9c), the transition to natural gas capacity from coal indicates a continued dependence on thermal fuels and imported fuel, raising concerns about energy security. This study did not delve into storage options (short- or long-duration) due to its limited scope for demonstration purposes and the absence of specific insights into the potential role of storage. However, considering the high penetration levels of VRE, particularly solar PV and wind (onshore/offshore), further exploration of various short-duration and long-duration storage options presents an exciting avenue for future research.

4. Software Extensions and Future Work

In conclusion, clicSAND presents a significant advancement in energy modelling, marked by its innovative user interface and enhanced accessibility for users across different expertise levels. The evolution from clicSAND 1.0 to the latest version, clicSAND 3.0, demonstrates a commitment to continual improvement and adaptation to user needs. By introducing a user-friendly Excel interface coupled with integrated solvers and a visualisation dashboard, clicSAND simplifies complex processes and empowers users to conduct thorough long-term investment analyses with unprecedented ease. The software’s integration of cutting-edge features such as embedded Python code, file size reduction, and real-time debugging through OSeMOSYS Cloud underscores its relevance and utility in the field. Notably, clicSAND’s transition from offline to cloud-based operations further enhances accessibility, enabling users to navigate complex energy models effortlessly. Moreover, clicSAND’s role in informing sustainable policies and mobilising resources highlights its importance in shaping a resilient and environmentally friendly energy future. The success stories showcased, such as the South African case study, underscore clicSAND’s effectiveness in facilitating knowledge transfer and skill development among participants of international capacity-building events. In essence, clicSAND represents a significant contribution to the field of energy system modelling, with promising implications for informed decision-making and sustainable development initiatives. In the remainder of this section, other parallel advancements of the clicSAND software will be presented, namely, its macOS version and the web interface OSeMOSYS UI. Finally, the chapter will showcase recent impacts and success stories of using clicSAND and will conclude with a view of future work and improvements.

4.1. Extensions and Software Updates

4.1.1. clicSAND for MacOS Users (clicSAND 2.0)

A second version of the software, called clicSAND for the Macintosh User Group (clicSANDMac), was developed [12,39,40]. This allows clicSAND and the Starter Data Kits [16] to be accessible and adaptable to a wider audience. The same principles and functionalities of clicSAND 1.0 for Windows and source code are maintained; however, a few changes are required for macOS compatibility. For example, Microsoft Access was replaced with a Python-based data manipulation approach (consequently embedded in the clicSAND 3.0 software for Windows). Additionally, data are run in a text file format rather than an Excel spreadsheet (however, the user still inputs the data through the Excel SAND Interface). An online course solely for OSeMOSYS MacOS users was released on the OLC Platform [41].

4.1.2. OSeMOSYS UI Development

As previously discussed, a notable feature of clicSAND is its utilisation of spreadsheets for data management. However, it is important to acknowledge that Microsoft Excel, a common choice for this purpose, has inherent limitations. These limitations encompass security vulnerabilities, error-prone operations, organisational challenges, and scalability issues, particularly when dealing with extensive projects. Recognising these challenges and building on the clicSAND software components, the development of an OSeMOSYS user-friendly interface has evolved into a comprehensive, standalone application known as OSeMOSYS UI. OSeMOSYS UI, developed by the United Nations Department of Economic and Social Affairs (UNDESA) [42], is available as a stand-alone platform on GitHub [43], empowering users to efficiently create, execute, and analyse OSeMOSYS models. Crucially, OSeMOSYS UI functions independently of any Microsoft software. It provides the flexibility to set up various units, construct models, and execute the solver, all within a unified interface. Furthermore, it streamlines the process of downloading and uploading models, facilitating seamless model-sharing among users. This not only promotes collaborative research but also simplifies the task of model updates over time, ensuring that research remains current and relevant.

Despite these advancements, it is important to highlight that the clicSAND software served as an initial iteration in improving user interaction with the OSeMOSYS model, predating the development of the web interface, OSeMOSYS UI. While the web interface represents a significant advancement in terms of accessibility and usability, it is essential to emphasise the continued relevance and benefits of the clicSAND software, particularly in regions with limited internet access. In low- to middle-income countries where internet bandwidth may be restricted or unreliable, the clicSAND software remains a valuable tool for energy modelling tasks. Its offline functionality allows users to work with the OSeMOSYS model without relying on internet connectivity, ensuring uninterrupted workflow and data manipulation even in areas with limited access to high-speed internet.

Furthermore, the clicSAND software may also cater to users who are more familiar and comfortable with traditional spreadsheet software like Microsoft Excel. Its interface provides a familiar environment for data input and manipulation, potentially reducing the learning curve for users transitioning to energy modelling tools. Therefore, while the OSeMOSYS UI represents a significant step forward in terms of accessibility and features, the clicSAND software continues to offer practical benefits, particularly in regions where internet access may be limited or unreliable, and for users who prefer a familiar spreadsheet-based environment.

4.2. Impact and Use of clicSAND Software

The clicSAND software has been employed in teaching, research analysis, and policy evaluation within a broader energy planning ecosystem [2]. It was initially used and tested with six Imperial College London master’s students for thesis projects in Kenya [44], Nigeria [45], Vietnam [46], Laos [47,48], and Armenia [49]. It proved to be a flexible tool supporting national energy planning strategies. These theses have expanded existing research and been incorporated into national government energy strategies (e.g., Vietnam [50]). clicSAND has also been used to develop 69 country Starter Data Kits (for countries in Africa, Asia, and South America), which are zero-order models with complete datasets freely available on Zenodo and pre-print papers on Research Square (complete list: Appendix A).

Furthermore, the clicSAND software was used to train over 250 energy analysts and academics during international capacity-building events, such as the Summer Schools at the Abdus Salam International Centre for Theoretical Physics (ICTP) [51,52,53], the EMP-A [33,54] and the Energy Modelling Platform for Latin America and the Caribbeans [55,56]. In these events, the skills previously taught using MoManI are now gained using the clicSAND software. As a result of these events, several researchers and government energy ministry analysts utilise clicSAND to build national energy strategies [57]. As explained in [2], the International Energy Agency (IEA) has made it a requirement to complete the OLC Platform online courses on OSeMOSYS, called “Energy and Flexibility Modelling”, using the clicSAND interface as a condition for taking part in its technical assistance program in Africa [58], utilising two of the above-mentioned capacity-building events, namely EMP-A 23 and ICTP 23.

clicSAND is the base of the practical exercises of the OLC Platform online course on “Energy and Flexibility Modelling” [15], which 444 people have currently completed and, in doing so, obtained an official certificate which was translated into Spanish. The GitHub supplementary repository is available on Zenodo [14].

Following participation in the OSeMOSYS&Flextool track of the online training event EMP-A 2021, a lecturer from Makerere University Business School, Faculty of Economics, Energy & Management Science, Department of Economics, incorporated two open-source tools into a postgraduate course syllabus. One of these tools was OSeMOSYS, along with its clicSAND software. clicSAND for OSeMOSYS was integrated into the syllabi of master’s and Ph.D. courses on Energy Economics and Governance at Makerere University Business School in Kampala, Uganda. Moreover, clicSAND is currently used to teach how to perform energy modelling analysis in the Climate Change Politics and Policy Master’s course [59] and the Climate Change Science and Management Degree [60] at Loughborough University [59], and to students of the Centre for Environmental Policy at Imperial College London working on their master’s thesis in consultation with the UK Government Energy Transition Council (ETC) [61].

To address the numerous requests from students for more teaching materials and video lectures, the authors have created a compilation of exercises which span applied case studies on clean cooking, decarbonisation of transport, and the integration of hydrogen and storage in OSeMOSYS using clicSAND 3.0 software (full list of available teaching material in Appendix C).

A survey conducted among participants in the capacity-building events, EMP-A 2021, and ICTP 2022, focusing on Energy and Flexibility Modelling, aimed at assessing the compatibility and user-friendliness of software tools. The survey results, presented in more detail in Appendix B, demonstrate that both the SAND Excel Interface and the initial version of clicSAND 1.0 received satisfactory ratings of 75/100 for their ease of use and user-friendliness. However, the results revealed certain weaknesses in terms of computational speed for the solvers. In response to the identified computational challenges, the development team introduced clicSAND 3.0, a solution designed to reduce data file sizes and significantly enhance solver performance (detailed information available in Section 2.4.2). As shown in Appendix B, following the integration of clicSAND 3.0 in international capacity-building events, subsequent feedback surveys indicated a noticeable improvement in both user satisfaction and software usability over time. These surveys consistently reported higher ratings, reflecting users’ growing perception of these tools as user-friendly and easy to use. Moreover, the speed and performance of the software and associated cloud platform demonstrated tangible improvements, marking a positive trend. Finally, clicSAND was an invaluable asset in the initial phase of OSEMOSYS UI development as a blueprint and prototype for data structure design.

4.3. Future Work

Looking forward, the future direction of clicSAND involves a comprehensive approach to enhance its functionalities and applicability in energy modelling. This includes improving features related to energy storage, integrating hydrogen, and refining transport system representation within the software to ensure a holistic view of energy transitions. Additionally, there is a need to explore advanced methods for representing transport data accurately, considering the complexities of mobility patterns and infrastructure requirements. Furthermore, integrating social aspects and parameters into the modelling framework is crucial for capturing societal impacts and preferences, thereby enhancing decision-making processes. Moreover, expanding the capability for multi-regional studies in clicSAND can offer valuable insights into interconnected energy systems, facilitating integrated planning and policy development across different geographical regions. Through these efforts, clicSAND aims to remain at the forefront of energy modelling, contributing to sustainable decision-making in the global energy sector.