Laylaa Okike

HYBRID IMPLEMENTATION OF ARTIFICIAL INTELLIGENCE WITHIN THE WATER RESOURCE FRAMEWORK OF SOUTH AFRICA - A REPORT AUTHOR: KATLEGO THAKADU1 SHAYM NATHA1 1 SDM CONSULTING ENGINEERS AND PROJECT MANAGERS, JOHANNESBURG, SOUTH AFRICA KATLEGO THAKADU: BSC (ENG): CIVIL AND EVIRONMENTAL ENGINEERING (HONS), MSC (ENG): CIVIL ENGINEERING, PROFESSIONAL CIVIL ENGINEER (PR. ENG.) ECSA, PROJECT MANAGEMENT PROFESSIONAL (PMP®) PMI, PROFESSIONAL CONSTRUCTION MANAGER (PR.CM) SACPCMP E-MAIL:- SHAYM NATHA: B. ENG: CIVIL ENGINEERING, M. ENG: CIVIL ENGINEERING, PROJECT MANAGEMENT PROFESSIONAL (PMP®) PMI E-MAIL:- 1. INTRODUCTION: Water is an economic and social good [1], that sustains life and promotes the efficient functioning of the South African economy. The total volume of fresh water reserves, which accounts for 2.5% of the total water available on the earth’s surface [2], is a finite but renewable resource, that has been adversely affected by climate change, through the redistribution of annual rainfall patterns, causing a flux of available fresh water within any region, and the mismanagement of water resources. This flux has imposed significant social and economic stresses on communities, who rely on the availability of freshwater resources to sustain their livelihoods. In a region such as South Africa, which is a semi-arid country, categorised by inequitable distribution of water resources [3]. The latest projection of water-demand to freshwater availability, conducted by the Council of Scientific and Industrial Research (CSIR), indicate that by the year 2030, the country will not be able to meet the freshwater demand [4]. However, it is fundamental to highlight that most challenges encountered through the management of freshwater, are because of increased pollution, fragmented water policies and institutions, lack of financial 1 transparency from state institutions, lack of infrastructure maintenance and inequitable water tariff estimations [5], conducted by water governing institutions and local municipalities. The concept of Artificial Intelligence (AI), centres around the concept of autonomous systems of information accumulation, verification, analysis and decision making. AI is an adaptive system that simulates human brain functions to make decisions, based on information and resources available, devoid of any political, financial, and social prejudice, which plagues our government’s ability to deliver services to the public. The implementation of this concept has both, positive and negative impacts on the South African economic sectors, and social clusters of the already wealth disparaged country, with a Gini Coefficient of 6.3 [6], as shown in Figure 1. Figure 1 - Class Sizes in South Africa [7] The distribution of resources has contributed significantly to the socioeconomic and political climate of the country, which is evident in the provision, quality, and maintenance of water resources in the country, with the elite, middle and vulnerable classes having access to portable water systems within their households, which account for 48% of the population, whereas, the remaining 62% of the population (49% chronic poor and 13% transient poor), typically do not have access to efficient infrastructure within their households, and whatever infrastructure is available, is inadequate as maintenance budgets are plagued by budget cuts, rampant corruption, and lack of capacity to manage water resources, due to a lack of technical expertise within local municipalities. Additionally, infrastructure development and maintenance projects fail to meet socio-economic needs of the local communities, as implementing agencies predominantly use a Top-Down engagement, when conducting/delivering services to these communities, which means that they 2 miss out on the benefits of participatory approaches, thus limiting the benefits of building capacity and increasing resilience within the communities that they serve. 2. WATER RESOURCE DISTRIBUTIONS ACCORDING TO THE SECTORS OF THE LOCAL ECONOMY South Africa’s water resources are, in comparison to global averages, scarce and extremely limited [3]. The country does not have large water bodies and exhibits a significantly higher evaporation rate, as compared to the annual rainfall recharge. The ground water distribution is uneven and not enough information is available on the resources, to incorporate it into the existing water resource networks as a supply reserve. The country mainly depends on surface water resources to meet water demand requirements [3], and shares four main rivers with neighbouring countries, which are used to feed the current reticulation systems. South Africa has also implemented a freshwater trade agreement with Lesotho through the Highlands Project, to increase the current capacities of freshwater reserves of the country. Data compiled by the Department of Water and Sanitations (DWS) [4] indicates that there is a misalignment of the current freshwater distribution, between the sectors that require the resource, and their contribution to the GDP, as depicted in Figure 2 below. Figure 2 - Water Use Per Sector in South Africa [4] 3 The agricultural sector has the significantly largest withdrawal, 61%, of freshwater resources, whereas it only has a GDP contribution of 5%, thus the volume to equity ratio of the water distribution, is disproportionate, when compared to the other economic sectors. The mining industry has a withdrawal of 2%, whereas its contribution to the GDP is 8% [8]. Thus, taking into consideration the full economic cost of water, South Africa experiences significant Opportunity Costs. 3. WATER GOVERNANCE OVERVIEW In South Africa, water institutions continue to implement a fragmented approach to water resource management, where legislature, policies, and procedures, are not reflective of the conditions on the ground. The National Water Act (Act 36 of 1998) {NWA} recognises four broad categories of water usage [9], namely, domestic, industrial, agricultural, and recreational use. The quality requirements specific to each category are specified under corresponding subcategories outlined in the Act. Governmental agencies lack adequate technical capacity and political will to implement the regulations listed under the Acts and contribute to the water conservation challenges which continue to plague the strained resources within the country. These administrative challenges are the main reasons that country remains water stressed, with water quality deteriorating at an unprecedented rate. The lack of effective water management has resulted in a deterioration of water resource quality over time, due to the poorly regulated release of effluent into water courses, leading to extensive damage to the existing ecosystems and threatening their biodiversity. Additionally, untreated acid mine drainage has contaminated groundwater supply within regions with closed mine shafts that were unrehabilitated after decommissioning, hence even with the scarce availability, existing ground water resources are becoming non-viable within some regions. Industrial and domestic effluent management remains the most critical drawback to sustainable water resource management within the country. There exists a lack of political will to implement policies and procedures, which address and implement strategies to manage grey and green water resources, private institutions such as the mining industry, have implemented successful 4 programmes, making use of grey and green water, to reduce their water footprint and manage their freshwater withdrawals. 4. IMPLEMENTING A HYBRID AI SYSTEM WITHIN THE SOUTH AFRICAN SOCIO POLICTAL LANDSCAPE According to the South African Institution of Civil Engineering (SAICE) Infrastructure Report Card, South Africa's infrastructure is in a poor state, with the majority key infrastructure, such as water reservoirs, major traffic bridges and dams, scoring a grade of D or lower. The report card evaluates each sector based on various criteria, including the level of investment, maintenance and upgrades, sustainability, resilience, safety, and innovation. The grades are assigned on a scale from A to D, with A representing excellent condition, B representing good, C representing satisfactory, and D representing poor. The report additionally notes that, the construction industry faces multiple challenges, including delays in project delivery, inadequate funding, a shortage of skilled workers, and poor project management. The implementation of an AI based Water Resource Management Framework in the context of the South African economy, requires efficient, equitable, sustainable and transparent, socio-economic legislature, implemented into unilateral water institutions, working in close proximity with the DWS and the National Treasury (NT) of South Africa, to reduce freshwater demand on the available resources. AI has the potential to help alleviate some of these bottlenecks in service delivery in the South African construction industry through: 1. Predictive Analytics: AI can be used to analyse past water resource construction projects and predict potential issues that may arise in future projects. This can help project managers plan better and avoid delays caused by unforeseen problems. 2. Machine Learning: By using machine learning algorithms, AI can analyse vast amounts of data and identify patterns that humans may miss. This can be used to identify areas where service delivery projects are likely to face bottlenecks and address them proactively. 3. Automation: AI-powered automation can help speed up many of the repetitive tasks in construction, such as data entry and project scheduling. This can free up workers to focus on more complex tasks that require human expertise. 5 4. Remote Monitoring: With the help of AI, remote monitoring of natural water resources can be conducted more efficiently, providing real-time insights into the quality, quantity, and potential bottlenecks in distribution. This can help project managers make informed decisions and take corrective action as needed. AI based IWRM can thus alleviate significant bottlenecks within the country’s water sector, improve water supply volumes and quality, promote sustainable use of water resources, and unlock new opportunities within the construction industry. However, caution is required in the implementation of autonomous technology within third world countries, as most of their population still lives below the extreme poverty line, as defined by the World Bank [6]. Therefore, AI advancements are currently considered a threat, to the livelihoods of non-skilled labourers, as most labour-intensive employment that is currently used in the country, can be undertaken using robotics, which will reduce bottlenecks and overheads. However, autonomous labour will also cause substantial job losses and increase the wealth disparity. It is with consideration of these key matters, that it is concluded that for South Africa, a hybrid implementation system would be the best option, to alleviate the bottlenecks and wastage due to mismanagement of water resources, lack of maintenance of infrastructure, unequitable water tariff estimations and collections and poorly regulated pollution of water resources. And promote shared benefits from the increased government expenditure in maintaining water resources and providing infrastructure to all citizens. Participatory programmes will need to be implanted to get the buy in from all government personnel and their constituent communities. Training and awareness programs will have to developed and administered in order promote knowledge, build resilience and showcase the opportunities that will be borne from the implementation of the AI IWRM system. Extensive coordination, public participation, transparency, integrated water resource management and political will be integral to implementing a sustainable AI IWRM, that will reduce freshwater withdrawals and pollution to ensure sustainable use, which meets future demands. 5. WORKFLOW PROCESS OF AI IWRM Implementing an IA based water resource monitoring and reporting system in South Africa will require several workflow processes to be developed, in line with the current infrastructure available. The first step will be to identify the requirements and objectives of the system, which 6 may include real-time monitoring, data analysis, and reporting. Next, the appropriate hardware and software components must be selected, such as sensors, data storage devices, and IA algorithms. Research, Investigations and Feasibility Assessments Resource Monitoring and Data Analysis Requirements and Scope Definition Story Board Definitions and Algorithm Compilation Model Deployment Model Testing and Optimisation Data Review and Preperation Model Developement Figure 3 – IWRM AI Workflow Process The system should also be designed to comply with local regulations and standards, such as the South African National Water Act and the South African Bureau of Standards. The system should also consider the data privacy and security requirements outlined by the Protection of Personal Information Act (POPIA). Once the system is designed, it must be assessed and validated to ensure that it meets the requirements and objectives of the initial scope requirements. This process involves simulating various scenarios to determine the system's performance and reliability. Finally, the system should be deployed and maintained to ensure ongoing functionality and accuracy. This includes regular maintenance, updating the system with new data and algorithms, and conducting periodic reviews of the system's performance and effectiveness. 7 6. PROS AND CONS OF AI APPLCIATIONS OR THE DIFFERENT CATEGORIES Integration of technological solutions such as AI in IWRM has the good and adverse effects as detailed below. Pros include: • Data capturing and data quality improvements (near error free and diminishes high human error). • 24/7 remote monitoring and quicker information access for decision makers (targeting interpretation processes and transformation objectives). • System failures (burst frequency, no water supply etc.), business management (work process, profitability etc) and policy making trends (asset management etc) can be identified quickly for gap fulfilment and decision making in infrastructure master planning can be amplified; and • Mitigates failure and system breakdown risk as they can be used in areas which may be hazardous, inaccessible and/or distant from the current administrative outposts (improving turnaround times). Cons include: • Increased unemployment (traditional job roles/repetitive tasks will be restructured and phased away due to AI solution performing work of multiple employees). • Lack of innovation and creativity in problem resolutions (solutions derived form past trends and successes and struggles in challenging situations). • High investment cost for AI system implementation (hardware, software and training component expenses required for basic operation). • Technical and ethical dilemmas/issues are inherent (purely logical programming without human work ethic and moral values are difficult to be incorporated). • Unknown work territories with undefined AI Governance in SA; and • Increases laziness potential of existing staff members (AI reliance on machine learning/algorithms impacts our ability to problem solve). AI solutions should not be allowed to fully replace human work interaction (it can make incredible improvements through integration but is not perfect as standalone technology). In IWRM predictive analytics and remote monitoring should be the basis for AI solutions to facilitate data 8 extraction and compilation for operational, strategic and financial management and ultimately improving water service delivery in SA context. 7. FUTURE GOVERNANCE OF AI The technology exists where officials can start installing AI (sensors, data loggers, transmitters and data storage for review) each time they repair or maintain infrastructure to start collecting critical information about the country’s water catchment, storage, and distribution network. AI is a systematic mechanism in this integrated data management approach to water resource management which will require governance to cover everything from project ownership, accountability and continuous data-based technical decision making for improvements in infrastructure management (scalability, capacity and capability elements). South African governance on AI indicates an absence of immediate policy and regulation. Public Information (POPI) Act is an overarching law which will also have a minor impact on the usage of government data by public officials and third-party service providers. AI Governance will be required in terms of alignment between sustainable development and infrastructure strategic centred values from machine learning/algorithms, transparency and human oversight through justification in decision making, as well as data security and technical accountability, business environment versus undesirable industry practises when adopting and utilising AI workflow processes. The ability of AI profiling through algorithmic programmes and machine learning in IWRM can also discriminate/disobey municipal principles and human rights through loosing operational control of water supply and usage patterns, asset management trends, infrastructure failure modes through AI systems being implemented. Global AI implementation trends and regulatory frameworks are developing quicker, and South African legislative and governing frameworks need more regulation for usage in water and other sectors (finance, agriculture, IT etc.) and the Artificial Intelligence Institute of SA can help lay the foundation for AI governance. 8. CONCLUSION AND WAY FORWARD The implementation of efficient, equitable and environmentally sustainable water resource management strategies within South Africa, requires governing institutions to make use of the subsidiary principle, to ensure a homogenous water governance, from national, to local water 9 institutions, which engage the communities, to build capacity, improve resilience and distribute opportunities to vulnerable groups, to build equity. As with all innovative and sound modern solutions to key challenges, political will and technical development remains a key driver in the implementation and management of these strategies, thus, with the existing political landscape, the feasibility of the implementation of AI, remains unknown as there is not enough infrastructure and pilot tests that have been conducted, to be used as baselines. Thus, further research and development studies are required, to assess the feasibility and benefits realisations, needed, to drive the political governance framework that will support the establishment of an AI based IWRM system that can manage our water resources efficiently, equitably and sustainably. 9. REFERENCES [1]. Nations, U., 1992. Dublin - Rio Principles. Rio de Janeiro, UNCED. [2]. A, T., 2022. Global and National Water Situation, Johannesburg: University of the Witwatersrand. [3]. BS, M., 2011. Water Development in South Africa. Zaragoza, Spain, UNCED. [4]. Steyn M, G. M. B. I. T. M. R. t. W. C. R. F., 2020. A decision support tool for Industrial Water reuse in South Africa, Johannesburg: CSIR. [5]. A, T., 2022. Sustainable Value of Water in Use, Johannesburg: University of the Witwatersrand. [6]. World Bank, 2022. Gini index South Africa, s.l.: World Bank. [7]. Department of Statistics South Africa, 2022. Poverty and Inequality, South Africa: Department of Statistics South Africa. [8]. R, M., 2022. Gross Domestic Product (GDP) Q2:2022, Johannesburg: Department of Statistics South Africa. [9]. T, G., 2019. South Africa's Options for Mine-Impacted Water re-use: A review. Journal of the Southern African Institute of Mining and Metallurgy, 119(3). [10]. South African National Water Act, 1998. Available at: https://www.gov.za/documents/national-water-act [11]. South African Bureau of Standards. Available at: https://www.sabs.co.za/ [12]. Protection of Personal Information Act (POPIA), 2013. Available at: https://www.gov.za/documents/protection-personal-information-act 10