Optimizing the Sweet Potato Supply Chain in Zimbabwe Using Discrete Event Simulation: A Focus on Production, Distribution, and Market Dynamics

Abstract

This study leverages a Discrete Event Simulation (DES) model to optimize the sweet potato supply chain in Zimbabwe, focusing on production, distribution, and market dynamics under varying climate conditions. The integration of climate data into the simulation model reveals significant insights into the resilience of different sweet potato varieties, particularly highlighting the suitability of yellow-skinned sweet potatoes for harsh climates due to their high resilience and drought resistance. However, market preferences still favor white-skinned varieties despite their vulnerability to climate extremes. The DES model identifies key bottlenecks, particularly in cultivation and transportation, that hinder supply chain efficiency. To address these challenges, the study emphasizes the importance of targeted interventions, such as improving access to irrigation, strengthening pest management, and adopting community-based resource-sharing approaches. These strategies are critical for enhancing both the resilience and efficiency of the supply chain. Additionally, the study highlights the urgent need for adaptive strategies to mitigate the effects of drought on agricultural productivity, especially in regions that heavily rely on crops like sweet potatoes. Overall, this research offers strategic insights for policymakers and stakeholders aiming to improve food security and agricultural productivity in Zimbabwe, as well as in other countries with similar climate challenges.

1. Introduction

Discrete event simulation (DES) has become a critical tool for optimizing agricultural supply chains, offering detailed insights into complex systems by modeling the operations and interactions of various components within these networks [1]. HoHoeks (1994) defines simulation as the construction and use of a computer-based model of a part of the real-world, serving as a substitute for experimentation and behavior prediction [2]. In recent years, DES has garnered increasing attention for its ability to address the multifaceted challenges agricultural supply chains face, such as seasonal variability, perishability of goods, and the impact of unpredictable weather patterns. Studies by Pierreval et al. (2007) and Guo et al. (2017) demonstrate the effectiveness of DES in identifying bottlenecks, optimizing resource allocation, and enhancing decision-making processes [3,4]. Additionally, foundational work by Banks et al. (2010) and Law (2007) emphasizes DES’s strength in modeling individual events and their interactions over time, offering valuable insights into system dynamics and potential improvements [5,6]. These studies underscore the dynamic nature of DES, allowing stakeholders to simulate real-world processes and anticipate the impact of changes in production, logistics, and market demand. DES has been successfully applied across various fields, including manufacturing, healthcare, and transportation, proving its versatility and effectiveness in optimizing operations and decision-making. In the context of agricultural supply chains, DES is particularly promising for its ability to enhance sustainability by enabling more efficient use of resources and reducing wastage through improved planning and management. One of its key strengths lies in its capacity to incorporate randomness and variability, providing more accurate predictions of system behavior under diverse scenarios.

In the evolution of DES within agricultural supply chains, recent literature points to an integration of simulation with other analytical tools, such as Geographic Information Systems (GIS) and Artificial Intelligence (AI), to further enhance decision-making and strategic planning. For instance, research by Lemire et al. (2019) demonstrates how combining DES with GIS can optimize the routing and scheduling of transportation in the distribution of agricultural products, leading to significant reductions in carbon emissions and costs [7]. Moreover, the integration of DES with AI, as explored by Zhang et al. (2019), facilitates predictive modeling and real-time adjustments in supply chain operations, catering to the dynamic nature of agricultural production and market demands [8]. These advancements indicate a shift towards more holistic and adaptive approaches in managing agricultural supply chains, emphasizing the role of DES not just in optimization, but also in fostering resilience and adaptability in the face of global challenges such as climate change and market volatility.

The scholarly examination of sweet potato supply chains has increasingly focused on understanding the intricacies of production, distribution, and market dynamics, reflecting the crop’s significance for food security and economic development in various regions. Previous studies, such as those conducted by Mutai et al., Kikulwe et al. and Munyaka et al., have primarily explored the socio-economic factors influencing sweet potato production, including access to quality planting materials, adoption of improved agricultural practices, and market access [9,10]. These studies shed light on the critical role of smallholder farmers in the sweet potato supply chain and the challenges they face, such as climate-related challenges, limited financial resources, vulnerability to pests and diseases, and fluctuating market prices. Furthermore, research has also delved into the post-harvest handling and processing aspects, identifying the logistics, and the significant losses due to inadequate storage and processing facilities, which not only affect the income of farmers but also the availability of sweet potatoes in markets [11]. However, these studies often highlight a gap in comprehensive supply chain analyses, focusing more on individual segments rather than the integration and optimization of the entire chain.

Existing research on sweet potatoes often lacks a holistic, systems-oriented approach, with many studies isolating specific components without considering the interdependencies between stages of the supply chain. For example, logistics and transportation, critical for connecting farmers with markets and ensuring timely delivery of produce, are often underexplored [12]. Furthermore, there is a scarcity of research on modern technologies that could enhance supply chain efficiency and resilience, such as digitalization, traceability systems, and sustainable farming techniques [13]. Despite a growing body of literature on agricultural supply chain optimization through DES, a conspicuous gap remains in research specifically tailored to Zimbabwe’s context.

Sweet potato production in Zimbabwe is a vital component of the agricultural sector, contributing significantly to food security and the rural economy [14]. This crop thrives in Zimbabwe’s varied climates, from the high rainfall areas to the drier regions, making it a versatile and resilient choice for many smallholder farmers [15]. Historically, prior to Zimbabwe’s independence in 1980, sweet potatoes were predominantly cultivated as a supplementary crop by women farmers in rural areas. However, over time, sweet potatoes have gained significance as a primary food source throughout the country. According to Smith (2004), the average national yield of sweet potatoes is 6 tons per hectare, with irrigated varieties producing up to 25 tons per hectare [16]. Over the past two decades, annual national sweet potato production has increased from less than 10,000 tons to over 200,000 tons across various varieties.

This increase in production can be attributed to formal policies supporting cultivation at the national level. Historically classified as an orphan crop, sweet potatoes lacked formal policy support, until Zimbabwe’s National Development Strategy 1 (NDS1) identified research and development, particularly in seed production and multiplication, as critical to advancing the sweet potato value chain [17]. A comprehensive study by Mutandwa (2008), involving 133 smallholder farmers in the Hwedza District, demonstrated the effectiveness of tissue-cultured sweet potatoes compared to traditional varieties [18]. Farmers adopting tissue-cultured varieties achieved 1.8 tons per hectare, while those using traditional varieties produced only 0.5 tons per hectare [19]. This stark difference highlights tissue-cultured varieties’ potential to enhance productivity, particularly in regions facing climate variability and soil degradation challenges.

Sweet potatoes are increasingly seen as a viable strategy for climate change adaptation and resilience due to their lower water and fertilizer requirements compared to crops like maize [20,21,22]. The National Agricultural Policy Framework (2019–2030) also emphasizes sweet potato bio-fortification to increase access to micronutrients, especially vitamin A, for children [23,24]. However, drought, a severe environmental challenge, affects crop durability and productivity, with climate variability exacerbating these issues [25,26]. While drought-tolerant sweet potato varieties can maintain or boost yields in harsh conditions, drought-sensitive varieties may experience a 25–50% reduction in yield [27].

Despite rising production volumes, fundamental challenges such as logistical inefficiencies, wastage, and demand-supply mismatches remain, disrupting the flow of goods and services in the sweet potato supply chain [28,29]. While previous studies have explored these challenges, many do not fully account for the socio-economic, environmental, and infrastructural issues specific to Zimbabwe, such as limited access to advanced agricultural technologies and the impact of climate conditions [30,31].

This oversight underscores the pressing need for a comprehensive simulation model that is not only cognizant of the global best practices in supply chain optimization but is also deeply rooted in the understanding of local nuances. Such a model would be essential in enhancing supply chain efficiency and resilience in the face of Zimbabwe’s unique challenges, ultimately contributing to improved food security and economic development.

The primary aim of this study is to improve farmers’ decision-making regarding sweet potato variety supply chains in Zimbabwe, particularly under varying climate conditions. To enhance the sustainability and resilience of sweet potato farming in Zimbabwe and ensure better livelihoods for all farmers amidst increasing climate challenges, the study will:

• Assess the key factors influencing farmers’ decisions regarding sweet potato varieties in Zimbabwe.
• Simulate the entire sweet potato supply chain by integrating different sweet potato varieties’ production, resilience, and marketability.
• Utilize spatial analysis to provide data-driven insights for decision-making.
• Propose actions to enhance the sustainability and resilience of sweet potato farming in Zimbabwe.

2. Materials and Methods

2.1. DES Modelling Overview

The DES approach optimizes agricultural supply chains by developing a computational model that simulates real-world operations, focusing on key events like planting, harvesting, and transportation. Utilizing Discrete Event Simulation (DES) software version 2024, such as Simul8 2024, the model simulates these events to analyze the supply chain’s efficiency, resilience, and performance under various scenarios reflecting changes in external (weather, market demand) and internal (inputs, equipment, and routes) factors. Simul8 will aid decision-making and communicate the effects of these factors. The supply chain of sweet potatoes, as shown in Figure 1, is intricate and integrates several factors including climate conditions. It begins with land preparation and clearing, then vines are acquired for cultivation, which involves traditional knowledge and practices passed down through generations, alongside modern agricultural techniques to enhance yield. After harvest, sweet potatoes are distributed through local and regional markets, often involving informal traders who play a crucial role in linking producers with consumers. However, challenges such as limited access to high-quality planting materials, pests, and diseases, as well as post-harvest losses affect the efficiency of the supply chain.

Calibration and validation of the model with real-world data ensure its accuracy, allowing for the analysis of potential improvements and the impact of strategic decisions. This methodology aids in identifying inefficiencies, forecasting outcomes, and guiding optimization efforts towards a resilient and sustainable supply chain.

2.2. Data Collection

To comprehensively assess the dynamics of agricultural supply chains, our methodology encompasses a two-pronged approach to data collection, focusing on both primary and secondary data sources. Primary data is systematically gathered directly from key stakeholders across the supply chain, including farms, processing units, and retail points. This involves conducting surveys, and direct observations to collect real-time and context-specific information on production volumes, processing capacities, inventory levels, logistics and transportation practices, and sales data. Field visits to farms and processing units enabled the collection of detailed operational data, while interactions with retailers provided insights into market demand, consumer preferences, and sales trends. Additionally, this approach includes the use of digital data collection tools such as Kobo Collect to ensure accuracy and efficiency in gathering and analyzing data from diverse sources. Kobo Collect was deployed in Goromonzi district, focusing on Domboshava (Ward 1, 2, 3, 4 and 7), situated in East Mashonaland province, Zimbabwe, as shown in Figure 2. This location lies about 32 km southeast of Harare. The primary data contributed to both the assessment of key factors influencing farmer’s decisions regarding sweet potato varieties and in the simulation of the entire sweet potato supply chain.

Goromonzi encompasses 13 commercial agricultural areas, 11 communal areas, and 1 small-scale farming area. Fertile soils support diverse agricultural activities across various altitudes, with temperatures ranging from 15 to 20 degrees Celsius and an average annual rainfall of 800 mm to 1000 mm, primarily during the summer months. Land tenure comprises freehold, communal, and state ownership, with major land uses including Large Scale Commercial Farming Areas (LSCFA), communal lands, Small Scale Commercial Farming Areas (SSCFA), and urban areas.

The GPS data, extracted in CSV format, was cleaned and processed using the Pandas library and then geo-processed in QGIS. The purpose of this GPS tracking was to highlight the specific itineraries and mobility challenges encountered by farmers transporting sweet potato harvests to markets such as Domboshava and Mbare, as well as to agribusiness locations [29,30,31]. The GPS data provided insights into the time and speed of the transport means used, the distances traveled, and the elevation changes along the routes [32].

The GPS data collected was then integrated into a simulation model to better understand the transportation dynamics and explore potential solutions to the mobility challenges [33]. This research demonstrates the value of leveraging GPS tracking technology to gain a detailed understanding of the transportation challenges faced by smallholder farmers in developing regions, which can inform the development of more effective logistics and infrastructure solutions.

Furthermore, satellite imagery obtained from various sensors, including MODIS and Landsat imagery, is utilized to generate the Standardized Precipitation Index (SPI) for drought conditions assessment. These datasets, sourced from the United States Geological Survey [34], provide essential information on the land surface temperature, vegetation cover, and rainfall—critical elements for assessing the impact of drought on sweet potato agriculture [35]. The datasets amalgamated information from four distinct satellites: Landsat 4 and 5, with 28 satellite images; Landsat 7, including 26 satellite images; and Landsat 8, with 11 satellite images [36]. Supplementary datasets were acquired from Humanitarian Exchange, encompassing details on administrative boundaries, local/regional road networks, transport infrastructures, and populated plateaus [37]. SPI datasets are integrated with environmental factors for a comprehensive sweet potato variety dynamic.

Table 1. Sweet potato data collected from Goromonzi and description.

SN	Categories	Type	Description of Data Collected	Source
Zimbabwe Administrative Boundary	ADMIN_0, 1, 2, 3, 4	Polygon	Representing the country, provinces (10), Districts (64), Wards (1970)	[34]
Survey	Farmers from Ward 1, 2,3,4 and 7	CSV	Surveyed sweet potato production	Farmers working in the fields
		POI	Capture of Farm locations	GPS in Kobo Collect
Tracks	GPS Trackers	Line	Farmers routes to Threshing floors and Markets	Columbus 900 Trackers
	Open Street Maps Tracks	Line	Routes connecting Farms to Vegetable Markets and Sweet Potatoes Agri-business	OpenTopoMap [35]
Landsat 7	SPI datasets	Imagery	Land surface temperature, vegetation cover, and rainfall	USGS [33]
Reports	2021–2023	Text	Second-round crop and livestock assessment	Ministry of lands, agriculture, fisheries, water and rural development

below presents the data collected on sweet potatoes from primary and secondary sources in the Goromonzi district, including detailed descriptions. The data collection process involved surveying farmers and capturing farm geolocation details, followed by a data cleaning procedure to ensure the accuracy and reliability of the results, as detailed in Table 1.

By integrating primary data from direct field observations and surveys with comprehensive secondary data from reputable sources, the research aims to build a robust and multidimensional dataset. This dataset will serve as the foundation for developing a DES model that accurately reflects the complexities and nuances of the agricultural supply chain, enabling targeted strategies for optimization and improvement.

2.3. Model Development

2.3.1. Discrete Event Simulation of the Sweet Potato Supply Chain

The sweet potato supply chain, like other agricultural chains, involves multiple interconnected activities that are influenced by various factors, such as climate conditions, labor availability, equipment usage, and market demand (see Figure 3). A DES model is ideal for capturing these complexities because it can simulate each event or stage in the process as it occurs over time, allowing for a detailed examination of how changes in one part of the chain affect the entire system.

In developing the DES model for the sweet potato supply chain, the study carefully delineates the stages of the supply chain to be modeled, identifies the critical assumptions underpinning the model, and outlines the key variables and parameters that drive the simulation. Agricultural activities such as planting schedules, harvesting times, and market distribution are inherently time-sensitive. The DES model’s ability to simulate these processes at specific points in time (discrete events) ensures that the study can accurately assess the impact of timing on the overall efficiency and effectiveness of the supply chain.

The model integrates a wide range of variables, including land size, vine count, harvest times, packaging durations, storage periods, and transportation logistics. By doing so, the DES model provides a holistic view of the supply chain, enabling the study to explore how different factors interact and influence final outcomes. Additionally, the DES model allows for the simulation of different scenarios, such as varying levels of workforce availability, equipment usage, or transportation options. This flexibility is crucial for testing the resilience of the supply chain under various conditions, aligning with the study’s objective of improving the efficiency and resilience of the sweet potato supply chain in Zimbabwe.

In the targeted wards, selected in Zimbabwe, farmers cultivate a variety of sweet potato types. The selection of these sweet potato varieties is influenced by multiple factors. Key considerations include the variety’s exposure and ability to withstand local climate conditions, its resilience against environmental stresses, and its popularity among both the farming community and consumers. Farmers prioritize varieties that can thrive in their specific climatic conditions, ensuring a stable and reliable crop yield. Additionally, they consider the preferences of their customers, aiming to grow sweet potato types that are in high demand in the market. This multifaceted approach to variety selection helps farmers optimize their production and will be integrated into the model.

The model incorporates both deterministic and stochastic variables to capture the dynamics of the sweet potato supply chain accurately. Deterministic variables include the planting (cultivation) and harvesting schedules, while the stochastic variables address the inherent uncertainties within the supply chain, such as yield variability due to weather conditions, fluctuations in market demand, and variations in transportation times due to logistical challenges. Parameters used in the model include yield per hectare, transportation costs, and market prices, among others. These variables and parameters are calibrated with real-world data collected from the targeted wards, ensuring that the simulation accurately reflects the operational realities of the sweet potato supply chain. By mapping out these stages, assumptions, variables, and parameters, the model aims to provide comprehensive insights into the supply chain’s performance, identify bottlenecks, and evaluate the impact of strategic interventions on overall efficiency and resilience.

The mapped stages were then interpreted into a mathematical model as a means to optimize outputs. This model considered the interdependencies shown in Figure 3 and included inputs such as climate conditions, vines, labor, pesticides, equipment, and more. Below is the definition of the stage outputs integrated into the model:

A.

Cultivation Output ${(Y}_c)$: Represents the growth phase, which depends on the climate condition, land, sweet potato vines, and input from farmers. The cultivation output is defined in (1) below:

$$Y_c=\mathrm{\mu } \left(V, L, P, E, C\right)$$

(1)

where $\mathrm{\mu }$ is a function representing the interaction between vines, land, pesticide usage, equipment, and climate conditions.

B.

Harvesting Output ($Y_{ha}$): Represents the collection of sweet potatoes from farms. This operation is influenced by the efficiency of the harvesters and the labor involved. The harvesting output is defined in (2) below:

$$Y_{ha}=\partial \left(Y_c,H,LB,E,C\right)$$

(2)

where $\partial$ is a function representing how the output from cultivation, combined with harvester efficiency, labor, equipment, and climate impacts the harvesting phase.

C.

Market, Home Consumption, and Decay ${(Y}_{mkt})$: The final stage, where sweet potatoes are either sold in the market or consumed at home. The market, home consumption, and decay are defined in (3) below:

$$Y_{mkt}=\omega \left(Y_{ha},PK,T,VH,M,HC,C\right)$$

(3)

where $\omega$ is a function representing how the harvested sweet potatoes are packaged, transported, and either consumed at home or sold in the market, including the influence of climate.

Initially, let us assume that these functions are linear combinations of their inputs, as shown in (4), (5), and (6) below.

$$\mathrm{\mu }(V, L, P,E,C)=aV+bL+cP+dE+eC$$

(4)

$$\partial (Y_c,H,LB,E,C)=f{Y}_c+gH+hLB+iE+jC$$

(5)

$$\omega \left(Y_{ha},PK,T,VH,M,HC,C\right)=k{Y}_{ha}+lPK+mT+nVH+oM+pHC+qD+sC$$

(6)

where a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, s are coefficients that will be estimated based on empirical data collected in the Goromonzi district from sweet potato farmers.

The variables are defined as follows:

C: Climate condition

V: Vines health and growth

L: Land available for cultivation

LB: Labor available for farming and harvesting

P: Pesticide usage

E: Equipment effectiveness

H: Harvester availability and efficiency

PK: Packaging efficiency

TT: Types of packaging materials available

VH: Vehicle availability for transportation

M: Marketplace

D: Decayed sweet potato tubers

HC: Home consumption rate

2.3.2. Integration of Vine Varieties Decision

After integrating the various stages of the sweet potato supply chain, including key factors such as climate conditions, labor, and equipment availability, the model further incorporates a series of evaluation steps. These steps, particularly focused on the Standardized Precipitation Index (SPI), assess drought conditions that significantly affect the productivity and resilience of different sweet potato varieties. This approach ensures that the model closely mirrors real-world scenarios, providing actionable insights for enhancing resilience in the sweet potato supply chain.

A critical component of the model is the selection of sweet potato varieties based on skin color, which is influenced by several factors, including market demand, climate resilience, and pest resistance. Farmers consider the popularity of certain varieties among consumers, their adaptability to local climatic conditions, and their susceptibility to pests and diseases when making these decisions [9,10,11].

This comprehensive approach not only improves the efficiency of the supply chain, but also provides a strategic framework for addressing the impacts of climate variability on agricultural practices in Zimbabwe. Specifically, climate data from targeted wards, collected through SPI using Landsat imagery from 2021 to 2023, were integrated into the simulation model. This integration aims to estimate the impact of extreme climate conditions on sweet potato production in Zimbabwe. The objective is to align climate conditions, particularly drought severity, with the production and resilience of various sweet potato varieties. The evaluation of these varieties, based on their exposure to extreme climate conditions and their resilience, follows the structured steps illustrated in Figure 4.

The model evaluates sweet potato varieties based on their exposure to extreme climate conditions (drought) and their resilience. It also takes into account other environmental exposures and the popularity of each variety.

• Step 1 

The SPI, developed by McKee et al. (1993), is a referent index in drought monitoring [36]. It is statistically consistent and displays abnormal wetness and dryness. The SPI shows rainfall quantity deficit in timescales and can identify short- to long-term drought impacts in variable spatiotemporal scales. The index can have a positive and negative value. When the SPI value is negative, it means that a drought event occurs, whereas the SPI becomes positive when the drought ends.

Table 2. Drought class values, climate hazards group Infrared precipitation with station data.

Positive SPI		Negative SPI
Wet Class	Values	Dry Class	Values
Extreme	>2	Extreme	<−2
Severe	1.5–1.99	Severe	−1.5 to−1.99
Moderate	1.0–1.49	Moderate	−1 to−1.49
Mild	−0.99 to 0.99	Mild	-
No value	-	No value	-

below shows a different classification used in SPI.

The SPI is a score deviation from the mean units of standard deviation, calculated for each pixel area of a composite period for the period from 2021 to 2023. The equation below explains how the SPI is generated, as shown in Equation (7):

$${SPI}_{ijk}={\displaystyle \frac{P_{ijk}-\overline{P_{ij}}}{{\sigma }_{ij}}}$$

(7)

where ${SPI}_{ijk}$ is the z-value for the pixel i during timeframe j for year k, $\overline{P_{ij}}$ is the precipitation value for pixel i during timeframe j for year k, $\overline{P_{ij}}$ is the mean for pixel $i$ during timeframe j over $n$ years, and ${\sigma }_{ij}$ is the standard deviation of pixel $i$ during week $j$ over $n$ years.

Favorable weather for sweet potatoes generally means adequate sunlight, moderate rainfall, and warm temperatures throughout the growing season. Excessive rain can lead to rot, while too little can stress the plants. Different varieties of sweet potatoes (white, red, yellow, and purple) have slightly different growth requirements and yield potentials depending on their production and resilience to extreme climate conditions. While all sweet potato varieties in Zimbabwe face some level of exposure to extreme weather, varieties like Yellow-Skinned (Chingova) and Purple-Skinned (Kenspot) sweet potatoes are bred to be more resilient, whereas White-Skinned and Red-Skinned (Beauregard) tend to be more susceptible.

Table 3. Sweet potatoes varieties, climate conditions (exposure and resilience), and popularity.

Sweet Potato Variety	Extreme Climate Exposure		Extreme Climate Resilience	Other Exposures	Popularity
	VHI	SPI
White-Skinned	Mild	Severe	Good	Pests and diseases during wet conditions	Very Popular
Red-Skinned	Moderate	Extreme	Moderate	Root rot from excessive rainfall	Moderately Popular
Yellow-Skinned	High	Severe	High	Diseases during wet conditions	Not so Popular
Purple-Skinned	High	None	High	None	Rare

below shows the exposure and resilience of these different varieties.

• Steps 2 and 3 

A formula-based model was developed to evaluate and integrate the climate conditions and production of the different sweet potato varieties. The average yield and performance of each sweet potato variety under different drought severity levels uses the following variables:

$D_s:$ Drought severity score

$R_v:$ Resilience score of the variety

$E_v:$ Exposure score of the variety to other factors (e.g., pests, diseases)

$P_v:$ Popularity score of the variety

The overall score $S_v$ for each sweet potato variety can be calculated using a weighted sum of the different factors, as shown in Equation (8):

$$S_v=w_1.{\displaystyle \frac{1}{D_s}}+w_2.R_v+w_3.\left(1-E_v\right)+w_4.P_v$$

(8)

where $w_1,$ $w_2$, $w_3$, $w_4$ are the weights assigned to each factor, reflecting their relative importance.

Weights are assigned, based on the relative importance of each factor. The weight assessment was based on historical data and expert input. The assessment revealed that resilience and drought severity weight were the two most important ones. Environmental exposure also contributes a sizeable to sweet potato production, followed by the marketability of the sweet potato varieties, as shown in

Table 4. Drought intensity weight.

Weight	$\mathbf{Drought} \mathbf{Severity} \mathbf{Weight} \mathbf{(}{\mathit{w}}_1\mathbf{)}$	$\mathbf{Resilience} \mathbf{Weight} \mathbf{\left(}{\mathit{w}}_2\mathbf{\right)}$	$\mathbf{Exposure} \mathbf{Weight} \mathbf{\left(}{\mathit{w}}_3\mathbf{\right)}$	$\mathbf{Popularity} \mathbf{Weight} \mathbf{(}{\mathit{w}}_4\mathbf{)}$
Percentage	30%	40%	20%	10%

.

• Step 4 

A scoring system to rank each sweet potato variety based on its performance, resilience, and other exposures was developed. Then, scores were assigned to each variable in

Table 5. Scores.

Scores	$\mathbf{Drought} \mathbf{Severity} \mathbf{Score} \mathbf{\left(}{\mathit{D}}_{\mathit{s}}\mathbf{\right)}$	$\mathbf{Resilience} \mathbf{Score} \mathbf{\left(}{\mathit{R}}_{\mathit{v}}\mathbf{\right)}$	$\mathbf{Exposure} \mathbf{Score} \mathbf{\left(}{\mathit{E}}_{\mathit{v}}\mathbf{\right)}$	$\mathbf{Popularity} \mathbf{Score} \mathbf{\left(}{\mathit{P}}_{\mathit{v}}\mathbf{\right)}$
1	Extreme	Low	High exposure to other factors	Rare
2	Severe	-	High exposure to pests/diseases	-
3	Moderate	moderate	Moderate exposure	Moderately popular
4	Mild	-	-	-
5	No drought	High	Low exposure	Very popular

below, with “1” representing “extreme” drought severity score, “low” resilience score, “high” exposure to other factors exposure and “rare” popularity score.

• Step 5 

Based on the calculated scores, the sweet potato varieties are ranked in terms of suitability under given climate conditions. A decision matrix is created to visualize and compare the performance of each sweet potato variety. The use of the matrix helps to assist farmers to make informed decisions on the best varieties to cultivate under different climate scenarios.

2.4. Data Analysis

Simul8 Modelling

The Simul8 setup includes four basic objects, with two being active (Start and Activity) and the remaining two passive (Queue and End). In Simul8, both labels and resource are critical. The study then conducted 201 surveys from farmers. The survey questions covered various aspects such as the farmers’ wards, size of land for cultivation, source of sweet potato vines (slips), varieties of sweet potato skin planted, equipment used for land preparation, cultivation, harvest preparation, and harvesting sources and types, types of pesticides used, and the number of laborers. The survey data were incorporated into the Simul8 model to help identify bottlenecks in the chain and overall business outcomes.

(a)

Entry point

At the “Start” object, sweet potato vines from various origins, spacing between vines, and skin colors, enter the system, with an assumed distribution of one day on average. Three-quarters of the farmers surveyed obtain their sweet potato slips from their gardens (76.11%), and nearly 20% from neighboring farmers (19.90%). Sweet potato vines are typically packed in 50 kg sacks for both commercial and personal use. Since most vines are generated from the farmers’ own farms, it is often challenging to estimate the number of vines (in 50 kg sacks) used per hectare of land. With an average slip length of 20 cm and moderate thickness, each slip weighs approximately 50–100 g. This means there are about 20 slips per kilogram. Therefore, a 50 kg sack contains roughly 1000 vines.

Standard spacing and planting density are also critical in estimating the number of vines per hectare. This spacing allows the vines room to spread and grow. A common spacing for sweet potato vines in Zimbabwe is about 30 cm between slips and 90 cm between rows, representing a planting density of about 0.27 square meters. With one hectare equaling 10,000 square meters and 3.7 vines per square meter, the total number of vines per hectare is estimated at 37,000. These estimates provide a framework for simulating sweet potato plantations (per hectare) using Simul8.

The DES model, built using Simul8, simulates different scenarios and optimizes inputs to either maximize market output or minimize home consumption, depending on economic goals. For this study, sweet potato vines, categorized by skin varieties, are routed from the “Start” object to the “Cultivation” and “Harvesting” activity objects in Wards 1, 2, 3, 4, and 7, as shown in Figure 3. Among the 201 farmers involved in the study, 11.44% (23 farmers) planted sweet potatoes in Ward 1, covering 9.51 hectares. Similarly, 97 farmers (48.26%) in Ward 2 plant sweet potatoes on 30.4 hectares. In Ward 3, 31 farmers (15.42%) plant sweet potatoes on 18.26 hectares. In Ward 4, 8 farmers (4.98%) plant sweet potatoes on 5.8 hectares. Finally, in Ward 7, 45 farmers (22.39%) plant sweet potatoes on 18.58 hectares.

Table 6. Number of 50 kg sacks expected per ward.

Ward	Number of Hectares	Number of 50 kg Sacks	Number of Vines (1000 Vines/Sacks)
Ward 1	9.51	352	352,000
Ward 2	30.4	1125	1,124,800
Ward 3	18.26	676	675,620
Ward 4	5.8	215	214,600
Ward 7	18.58	688	687,460

below details the number of 50 kg sacks that are expected to be cultivated in each of the targeted wards.

(b)

Label and resources

After defining and assigning values to the model at the “entry point”, labels were integrated into the simulation in order to collect different results. A label is a way to attach attributes to Work Items in the” simulation. First, a gender label was developed into the DES model, denoted as “lbl_gender”. This label is applied in each activity in the model, as shown in Figure 3.

Table 7. Sweet potato farmer’s gender.

Ward	Male	Female
Ward 1	86.96	13.04
Ward 2	31.96	68.04
Ward 3	25.81	74.19
Ward 4	37.5	62.5
Ward 7	57.14	42.86

reveals the farmer’s gender rate integrated into the model:

Another label applied in the Simul8 model is sweet potato skin varieties, denoted as “lbl_vine_color”. A probability distribution value per the percentage in each ward is integrated into the DES model as shown in

Table 8. Sweet potato skin varieties used by farmers per ward.

Ward	White-Skinned (%)	Red-Skinned (%)	Yellow-Skinned (%)	Purple-Skinned (%)
Ward 1	95.65	0	4.35	0
Ward 2	78.35	19.59	2.06	0
Ward 3	80.65	16.13	0	3.23
Ward 4	100	0	0	0
Ward 7	95.56	4.44	0	0

below. This data provides a clear view of the distribution of sweet potato skin varieties across different wards, helping in understanding the preferences and cultivation patterns of the farmers in each ward in Goromonzi.

Sweet potato vine sources are critical because they impact the farmer’s budget, as 50 kg averages 5 dollars ($). The sources also limit the farmer’s sweet potato skin varieties options. For instance, it is common for farmers to have white-skinned and red-skinned sweet potato vines, it is therefore difficult to acquire other skin types regardless of their extreme climate resilience and their potential exposures as shown in Table 3.

Table 9. Sweet potato vine sources per ward.

Ward	Own Farm Vines (%)	Neighboring Farmers (%)	Commercial Vine Nurseries (%)	Dr—SS (%)
Ward 1	34.78	60.87	4.35	0
Ward 2	50.52	41.24	7.22	1.02
Ward 3	74.19	25.81	0	0
Ward 4	50	50	0	0
Ward 7	35.71	57.14	7.15	0

provides a detailed breakdown of the sources of sweet potato vines used by farmers in different wards. Unlike wards 1 and 7, the majority of farmers obtain their vines from their own farms. In the DES model, the following label “lbl_vine_sources” was created, added in the cultivation object and action as probability distribution using data provided in Table 9.

Sweet potato planting operations, whether in “field preparation”, “planting”, or “harvesting” activities, are highly labor-intensive. Both labor and equipment “resources” are useful for these operations.

Table 10. Tools and equipment used as per activities.

Tools and Equipment	Hoes Muttock	Mould Board Plough	Tractor	Wheelbarrow	Scotch Cart	Sprinkler
Land Preparation	✓	✓	✓
Cultivation	✓			✓	✓
Pesticides						✓
Harvesting	✓	✓		✓	✓

shows which tools and equipment were used as per upper mentioned activity.

Table 11. Labor hired by farmers.

Hired Labor	Hired Labor 0 (%)	Hired Labor 1 (%)	Hired Labor 2 (%)	Hired Labor 3 (%)	Hired Labor 4 (%)	Hired Labor 5 (%)	Hired Labor 6 (%)	Hired Labor 7 (%)	Hired Labor 8 (%)	Hired Labor 9 (%)	Hired Labor 10 or over (%)
Ward 1	30.43	0.00	13.04	4.35	13.04	8.70	17.39	0.00	4.35	0.00	4.35
Ward 2	45.36	11.34	21.65	9.28	2.06	3.09	5.15	0.00	1.03	0.00	1.03
Ward 3	51.61	16.13	16.13	9.68	6.45	0.00	0.00	0.00	0.00	0.00	0.00
Ward 4	37.50	0.00	25.00	25.00	12.50	0.00	0.00	0.00	0.00	0.00	0.00
Ward 7	15.56	4.44	4.44	17.78	4.44	17.78	4.44	0.00	4.44	0.00	20.00

and

Table 12. Tools and equipment recorded as per wards.

	Hoes Muttock (%)	Mould Board Plough (%)	Tractor (%)	Scotch Cart (%)	Wheel Barrow (%)	Sprinkler (%)
Ward 1	100	13.04	0.00	0.00	30.43	0.00
Ward 2	100	11.34	0.00	11.34	26.80	0.00
Ward 3	100	3.23	0.00	12.90	25.81	0.00
Ward 4	100	0.00	0.00	25.00	25.00	0.00
Ward 7	93.33	26.67	6.67	26.67	33.33	0.00

respectively show the percentage of labor and equipment availability in each targeted ward. The labor and equipment resources are integrated into the Simul8 model as both labels, denoted as “lbl_equipment_used” for equipment and “lbl_labor” for labor and resources.

Labor resource varies based on the land size as well as on the influence of socio-cultural factors. Table 12 revealed that Wards 1 (69.5%) and 7 (84.44%) have higher hired labor utilization. Ward 4 has shown a moderate hired labor utilization (54.64%) while Wards 2 (54.64%) and 3 (48.39%) showed lower hired labor. These hired labor utilization data are integrated in Simul8 as “labor” resource 1 (1–10%), 2 (11–20%), etc.

The data reveals that the presence of tractors (6.67%) and a higher percentage of mouldboard ploughs (26.67%) suggest a more mechanized approach, likely leading to higher productivity compared to other wards. This ward’s diverse tool usage may allow for more efficient land preparation and management, reducing labor time. In contrast, Wards 1, 2, and 4 reliance on manual tools such as hoes and mattocks indicate a more labor-intensive and time-consuming process [38,39]. These wards may experience lower productivity due to the higher labor requirement and slower land preparation and planting processes. For Ward 2, while predominantly using hoes and mattocks, the presence of mouldboard ploughs, scotch carts, and wheelbarrows suggests a balance between manual and mechanized tools. Based on the percentages, ten resources are allocated to “hoes_muttock” across all the wards in the Simul8 model. Additionally, two resources are allocated to “mould_board_plough” in Wards 1 and 2, respectively. The average cost of using tractor services and animal tractor services for the smallholder sector was US$18.90 per hectare [40].

Furthermore, it takes about 8 to 10 days for one person to clear one hectare of land, in Zimbabwe, averaging 0.1 hectare per person per day, using traditional tools such as hoes, shovels, and mattocks. This rate improves with the use of mechanized equipment. However, mechanization incurs additional costs, and there is no government subsidy for equipment, as revealed by the survey findings in

Table 13. Farmer’s equipment sources.

Ward	Hired (%)	Own-Equipment (%)	Government Subsidy (%)	Borrowed (%)
Ward 1	33.33	66.67	0	0
Ward 2	17.09	77.78	0	5.12
Ward 3	11.76	85.29	0	2.94
Ward 4	36.36	63.64	0	0
Ward 7	28.07	70.18	0	1.75

. It revealed that the majority of farmers surveyed lack the funds to hire new equipment. The farmer’s equipment sources were labeled as “lbl_equipment_sources” in the Simul8 model and applied in the “field preparation”, “cultivation” and “harvesting” activities, based on the probability distribution value as per percentage shown in each ward in Table 13 below.

Planted sweet potatoes are often susceptible to pests and various diseases, depending on their skin varieties. The survey revealed a variety of indigenous mechanisms for combating these issues in sweet potato plantations. These mechanisms were integrated as an “Activity” into the DES model. Among the methods for combating pests and diseases, as shown in Figure 5, are spraying pesticides to repel infestations. Farmers also use ashes to deter insects. Rat poison is used to prevent rats and similar pests from devouring the growing plants. Other preventive mechanisms include trapping, watering, rotating fields, and eradication. A label, denoted “lbl_pesticides_app_types”, was applied in the simulation model.

For harvesting, it is estimated that one vine produces between three sweet potato tubers under less favorable conditions and up to eight tubers under favorable conditions. A quantity label “lbl_quantity” was created, set as a fixed distribution of one vine (value = 1) in the cultivation activities and as a rounded uniform distribution with values ranging between three to eight sweet potato tubers per vine introduced into the harvesting activities in the DES model. The rounded uniform value is then routed out as a “batch” to the next activity in the system.

Post-harvest processes are fundamental to the model. These processes include activities such as sorting, cleaning, curing, grading, packaging, and transporting marketable sweet potatoes to the appropriate markets. Remind those sweet potatoes are packaged in a 50 kg sack in Zimbabwe. With sweet potato tuber average weight estimated at 300 g, the number of tubers per kilogram is calculated as shown in Equation (9):

$$\mathrm{number} \mathrm{of} \mathrm{tubers} \mathrm{per}={\displaystyle \frac{1\mathrm{kg}}{\mathrm{Average} \mathrm{weight}/\mathrm{tuber}}}=3.33{\displaystyle \frac{\mathrm{tubers} }{\mathrm{kg}}}$$

(9)

Therefore, a 50 kg sack of sweet potatoes in Zimbabwe would contain approximately 167 sweet potato tubers, assuming an average tuber weight of 300 g. This number can vary depending on the actual average weight of the tubers. In the simulation model, a 50 kg sack, containing an average of 1000 vines to be cultivated, can harvest on average 6000 sweet potato tubers tons per hectare [15].

The estimated production value and packaging per 50 kg sack for one hectare is extrapolated to the following land sizes: Ward 1 (9.51 hectares), Ward 2 (30.4 hectares), Ward 3 (18.26 hectares), Ward 4 (5.8 hectares), and Ward 7 (18.58 hectares). In the DES model, a new quantity label “lbl_50 kg_sack_packaging” was created and applied in the packaging activity. This label is set as an average distribution with a value of 167 sweet potato tubers per 50 kg sack. The average value was then routed out as a “batch (50 kg sack)” to either the market, while the remaining sweet potato tubers, labeled “lbl_quantity”, will be sent to family consumption, or the waste bin.

Furthermore, sweet potato sacks are transported either to the market or to farmers’ homes. Information on distance, speed, road conditions, and transportation costs was incorporated into the simulation model. The condition of road infrastructure crucially affects the transportation of products to markets. Survey data shows that 10% of farmers in Goromonzi reach markets in Domboshava (Showground) or Mbare using tertiary roads (A3), which are gravel and poorly maintained, especially during the rainy season. Most farmers mainly utilize footpaths and tracks to reach the family houses (often also the threshing floor) or sweet potatoes meeting place to the market, reflecting differences in access to transportation means.

Table 14. Farmer’s transport means to the markets.

Ward	Bicycle (%)	Motorcycle (%)	Minibus (%)	Pickup Truck (%)	Single Axle Van (%)	Tractor (%)	Animal (%)	Walking (%)
Ward 1	0	0	0	4.34	0	0	0	95.66
Ward 2	0	0	0	19.59	1.03	1.03	16.49	61.86
Ward 3	0	0	0	41.94	0	0	12.90	45.16
Ward 4	0	0	0	0	0	0	0	100
Ward 7	0	0	0	19.05	0	4.76	30.95	45.24

reveals that most farmers rely on pick-up trucks as mechanized means for swift market transportation. Farmers often have to coordinate product collection and engaging transporters based on market readiness and transport service availability, with costs ranging from $1 to $3 for a 50 kg sack, varying by distance, vehicle access and availability. Moreover, due to limited access to funds, most farmers are inclined to use animals (24.42%) and even walk (93.02%) to these destinations. Table 14 provides a breakdown of the various transport means used by farmers in different wards to get their sweet potatoes to the markets and the data was integrated into the DES model under the “lbl_travel_mean” label. The transportation methods include bicycles, motorcycles, minibuses, pickup trucks, single-axle vans, tractors, animals, and walking. Each ward shows different preferences based on the availability and accessibility of transportation means.

Farmers in Ward 1 predominantly walk to the market, with only a small percentage using pickup trucks. Figure 6 below shows the proximity of Ward 1 to Domboshava market, explaining the high reliance on walking. The limited number of mechanized transports suggests the need for more efficient methods to transport large quantities of sweet potatoes. In Ward 2, most farmers also walk to the market, but there is a significant use of pickup trucks and animal transport. This variety indicates access to multiple transportation options and varied-cost services. Similarly, farmers in Ward 4, due to its proximity to the Domboshava market, primarily walk to the market. Across all wards, a significant percentage of farmers rely on walking, particularly in Wards 1, 3, and 4. Conversely, pickup trucks and animals are notably used in Wards 2, 3, and 7, highlighting the need for improved supply chain efficiency based on available resources.

Constructing Figure 6, open-source datasets were utilized to delineate administrative boundaries and map existing networks (OSM Tracks). Kobo was used to geolocalise farms across wards. GPS devices assisted in identifying the routes farmers use to transport their harvested and packed sweet potatoes to the Domboshava and Mbare markets. GPS data were then imported and overlaid with OSM tracks to verify the routes’ existence, measure the distance farmers travel from their farms to the market, and calculate the average speed and travel time, as shown in

Table 15. Sweet potato supply chain routes from farms to Domboshava and Mbare markets.

Ward	Survey	Distance (km)	Speed (km/h)	Time (h)	Cost ($)	Destination (Market)
Ward 1	23	7–19	1–27	0.3–0.6	1–3	Domboshava
		46–58	3–53	0.92–1.19	1–3	Mbare
Ward 2	97	6–17	1–23	0.2–0.5	1–3	Domboshava
		45–56	3–64	0.93–1.04	1–3	Mbare
Ward 3	31	10–14	1–24	0.37–0.51	1–2	Domboshava
		46–48	3–64	0.98–1.08	1–2	Mbare
Ward 4	8	3–12	1–15	0.11–0.41	1–3	Domboshava
		42–47	3–64	0.74–0.98	1–3	Mbare
Ward 7	45	11–22	1–34	0.36–0.64	1–3	Domboshava
		34–45	3–54	0.58–0.91	1–3	Mbare

.

Figure 7 summarizes key variables and breaks down all the key components and an approximate estimate of the time it takes for one person to plant one hectare of sweet potatoes. This information is integrated into the simulation model.

The simulation model applies various scenarios to identify ways for enhancement. Tools, equipment, and worker efficiency can reduce this planting time, leading to higher market output and better economic outcomes for farmers. This detailed information is integrated into the simulation model.

2.5. Validation and Verification of the Model

The validation and verification process of the DES model is crucial to ensure its accuracy and reliability in representing the sweet potato supply chain. Verification, the first step in this process, involves checking that the model has been correctly implemented according to the conceptual design, with no errors in the code or logic of the simulation. This includes reviewing the model’s structure, flow of events, and interactions among different components to ensure they align with the intended design. Techniques such as code inspection, walkthroughs, and debugging are employed to identify and correct any discrepancies between the model’s programmed behavior and its conceptual specification. Verification ensures the model is a true technical representation of the conceptual model, free from implementation errors.

Following verification, validation is conducted to ascertain that the model accurately reflects the real-world system it aims to simulate. This involves comparing the model’s outputs with real-world data and observed behavior of the sweet potato supply chain. Validation is achieved through several methods, including historical data validation, where model predictions are compared against actual data from the past; face validation, involving subject matter experts who assess the model’s realism and relevance; and sensitivity analysis, which examines how changes in model inputs affect outputs. If discrepancies are found during validation, adjustments are made to the model’s structure, parameters, or assumptions, followed by re-testing until the model’s outputs are in reasonable agreement with real-world observations. This iterative process of validation and refinement continues until the model is deemed an accurate and trustworthy tool for exploring the dynamics of the sweet potato supply chain, capable of producing reliable insights for decision-making and strategic planning.

3. Results

3.1. Drought Severity Mapping

The SPI was applied to targeted wards to assess their vulnerability to drought, based on classifications from 2021 to 2023 (see Figure 8 and Figure 9). The primary objective was to determine the point at which weather and climate challenges significantly influence farmers’ decisions to predominantly plant sweet potatoes during the summer months. Despite the inherent resilience of sweet potato production, extreme weather and climate conditions continue to pose significant challenges. According to Aldridge (2002), the 1992 Southern African drought was the worst in Zimbabwe’s recorded history, leading to the drying up of many wells and even some perennial rivers [41,42].

Given that precipitation data is often not normally distributed, particularly over periods of 12 months or less, a transformation is typically applied to better analyze the data. This data is usually fitted to a gamma distribution function. The SPI for the period from 2021 to 2023 was used to calculate the time series of drought conditions. Figure 9 illustrates the area-zoning map based on SPI values, revealing that approximately 60% of the cultivated areas are under significant precipitation stress, categorized as severe dryness. This has directly affected soil conditions, leading to a noticeable decrease in sweet potato productivity (see Figure 8 below). This analysis highlights the critical need for adaptive strategies to mitigate the impact of drought on agricultural productivity, particularly in areas heavily reliant on crops like sweet potatoes.

3.2. Evaluation of Sweet Potato Varieties Suitability to Climate Conditions

The study evaluated different sweet potato varieties based on their suitability to climate conditions. Each variety is scored according to key factors such as resilience, drought resistance, and popularity, which contribute to their overall suitability in a given climate. The following factors were used for the variety’s evaluation as per

Table 16. Suitability to climate conditions.

Sweet Potato Varieties	Score		Suitability to Climate Conditions
	Dry	Wet
White-skinned sweet potatoes	2.375	2.9	high suitability in both dry and wet conditions, particularly due to their good resilience and high popularity. They are less exposed to severe climate conditions in the dry season but face challenges from pests and diseases during the wet season.
Yellow-skinned sweet potatoes	2.2	2.8	Highest overall suitability, particularly excelling in resilience and drought resistance. Despite being less popular among farmers, their strong resistance to climate challenges makes them a reliable option in both dry and wet seasons.
Red-skinned sweet potatoes	1.6	2.5	Moderate suitability, with a noticeable decline in performance under extreme wet conditions. Their moderate resilience and popularity, coupled with high susceptibility to root rot during excessive rainfall, make them a less favorable choice compared to other varieties.
Purple-skinned sweet potatoes	1.6	1.8	Moderate suitability, although resilient and resistant to climatic challenges, suffers from low popularity among farmers. This limits their overall suitability, especially in the wet season, where their performance slightly declines compared to dry conditions.

above:

• For white-skinned sweet potatoes, “mild” climate exposure for the dry season, which earns a score of 4 and “severe” climate exposure for wet season, earns a score of 2. Also, with a “good” resilience, with a score of 5. They also have a high exposure to pests and diseases, scoring 2, but are very popular among farmers and customers, with a score of 5.
• For red-skinned sweet potatoes, the evaluation considers “moderate” climate exposure for the dry season, with a score of 3 and “extreme” climate exposure for the wet season, earns a score of 1. Also, with a “moderate” resilience, also scoring 3. They have a high exposure to root rot due to excessive rainfall, scoring 1, and are somewhat popular among farmers and customers, with a score of 3.
• For yellow-skinned sweet potatoes, the assessment is based on “high” climate exposure, with a score of 1 and “severe” climate exposure for the wet season, which earns a score of 2. Also, with a “high” resilience, earning a score of 5. They also face a high exposure to pests and diseases during wet conditions, with a score of 2, and are not very popular among farmers and customers, with a score of 1.
• For purple-skinned sweet potatoes, the evaluation includes “high” climate exposure, with a score of 1 and “no value” climate exposure for the wet season, which earns a score of 5. Also, with a “high” resilience, scoring 5. They have no exposure to climatic conditions and diseases, earning a score of 5, but are rare among farmers and customers, with a score of 1.

Applying Equation (8) for the integration of sweet potato production and climate conditions, taking into account the above-mentioned variables, the results revealed that, based on the calculated scores, the sweet potato varieties rank as follows in terms of suitability under given climate conditions (see Figure 9).

Table 16 highlights that yellow-skinned sweet potatoes are the most suitable for harsh dry climate conditions due to their high resilience and drought resistance. However, white-skinned sweet potatoes are also highly suitable, benefitting from both good resilience and high popularity among farmers. Both red-skinned and purple-skinned sweet potatoes have moderate suitability, with the latter being slightly less preferred due to lower popularity despite its resilience. The finding from Table 16 prompted a focus on white-skinned sweet potatoes in the simulation modeling.

3.3. Model Application

3.3.1. Simulation of White-Skinned Sweet Potato Variety

The agro-ecological environment in Goromonzi District is generally well-suited for white sweet-skinned potato cultivation, with the crop thriving across the various soil types found in the region. White-skinned sweet potatoes are particularly well-adapted to both dry and wet conditions, thanks to their good resilience and high popularity among farmers and consumers. During the dry season, they are less exposed to severe climatic conditions, which contributes to their suitability. However, during the wet season, they face challenges from increased pest and disease exposure, which could impact their overall productivity. Despite these challenges, their adaptability and resilience make them a favorable crop choice in this district.

Current Sweet Potato Supply Chain Production

In Goromonzi, the sources of sweet potato vines, as shown in Table 9, reveal that most farmers still rely on “own farm vines” and “neighboring farmers” for their planting material. This indicates a prevalent use of unimproved varieties over tissue-cultured sweet potato varieties [14], which are typically sourced from “commercial vine nurseries” and specialized suppliers like “Dr-SS”. The reliance on traditional sources for vines highlights the need for greater access to improved varieties that can offer higher yields and better resilience against environmental stresses.

To analyze the current sweet potato supply chain, this study employed the constructed DES model, using Simul8 software version 2024 to simulate each event or stage in the supply chain process. As illustrated in Figure 10, a batch of 1000 vines (denoted as (1)k in the model) enters the simulation at the “start point” and is subsequently sent to a prepared field for planting. Following planting, the sweet potatoes go through the processes of harvesting, curing, sorting, and grading before being packed. The packed sweet potatoes are then distributed to markets and homes for sale and consumption, while any spoiled sweet potatoes are discarded as waste. This process flow highlights the critical stages in the supply chain and emphasizes the importance of each step in ensuring the quality and marketability of the final product.

The model in Figure S1 (Supplementary File) schematically illustrates the various variables and factors influencing the sweet potato supply chain. By understanding these factors and their impact on the efficiency and resilience of each stage—from cultivation to market distribution—the study aims to assist in optimizing the supply chain. This comprehensive understanding is crucial for enhancing the overall performance and sustainability of sweet potato production and distribution.

By simulating each stage of the supply chain, the model helps identify where delays or inefficiencies occur, such as during transportation or storage. This insight is essential for making targeted improvements. With a total of 201 farmers involved, the cultivation of 1000 vines leads to a harvest ranging from 13 sweet potato tubers under less favorable conditions to as many as 625 tubers under favorable conditions. However, bottlenecks have been identified, particularly in the “cultivation” activity. The processing time for the “cultivation” activity is influenced by two key factors: (1) the size of the land being cultivated (as shown in Table 6) and (2) the availability of resources used by farmers (as detailed in Table 11 and Table 12). Interestingly, the timing of field preparation did not impact the vines entering the system, as they were only brought to the field once the preparation was completed, regardless of the time taken to complete the preparation with or without resources. The detailed breakdown of Figure 10 emphasizes the multifaceted role of sweet potato production in both economic and subsistence contexts.

The findings revealed that from a cultivated land area of 82.55 hectares, a total harvest of 334 tonnes of sweet potatoes was achieved. Of this total, 206 tonnes were allocated for sale in the market, highlighting the crop’s economic value to the farmers. Additionally, 104 tonnes were reserved for family consumption, underscoring the crop’s importance in ensuring household food security. However, 4 tonnes of the harvest were lost due to decay, which points to potential issues in post-harvest handling or storage conditions. The remaining 20 tonnes are currently still undergoing processing, which could include cleaning, sorting, or preparing the produce for either sale or storage. This long processing time also highlights the need to optimize the sweet potato production line.

Leveraging the data on land size and production, the findings from Figure 10 indicate that Wards 2, 3, and 7 have been particularly affected by these bottlenecks, largely due to the larger land areas that farmers in these wards must cultivate. Specifically, Ward 2, with 30.4 hectares of land, produced 112 tonnes of sweet potatoes; Ward 3, with 18.26 hectares, yielded 69 tonnes; and Ward 7, with 18.58 hectares, resulted in 73 tonnes. The larger land sizes in these wards correlate with higher production levels, yet they also contribute to greater challenges in managing and overcoming agricultural bottlenecks.

Regarding resource availability, Table 11 indicates that more than half of the farmers in Ward 3 do not hire labor for their plantations. Among the three wards facing significant bottlenecks, only farmers in Ward 7 have shown a greater willingness to hire one or more laborers, as illustrated in Figure 10 below. The figure also reveals that, due to limited access to resources, farmers in Wards 2 and 3 struggle to hire more than four laborers, whereas Ward 7 has demonstrated some consistency in hiring labor. This disparity highlights the challenges faced by farmers in accessing sufficient labor resources, particularly in Wards 2 and 3, which exacerbates the bottlenecks in the cultivation process. The hired labor resources have reduced bottleneck stress in Ward 7.

Furthermore, the availability of resources plays a critical role in the cultivation process, significantly influencing the efficiency of sweet potato farming. Among the wards with a high number of hectares targeted in this study, Ward 7 stands out not only for its higher hiring of labor but also for its unique approach to resource utilization. Based on Table 12, farmers in Ward 7, despite having more land to cultivate, use fewer hoes and mattocks—the standard tools in rural farming in Zimbabwe—compared to other wards. This reduction in the use of traditional hand tools is compensated by the adoption of more advanced farming equipment.

Ward 7 is also distinct in that it is the only ward where a limited number of farmers utilize tractors for their agricultural activities. Instead of relying solely on traditional tools, farmers in this ward have increased their use of more efficient equipment, including mouldboard ploughs, scotch carts, and wheelbarrows. These tools are crucial for enhancing the speed and effectiveness of the cultivation process, especially in a region where land sizes are large and manual labor alone may not suffice.

The combination of higher labor hiring and the strategic use of more advanced farming tools in Ward 7 suggests a more resource-intensive approach to agriculture, which may contribute to greater productivity and efficiency. In contrast, the reliance on fewer laborers and traditional tools in other wards, such as Wards 2 and 3, could hinder the farming process, leading to bottlenecks and lower overall productivity. This highlights the importance of resource allocation and the adoption of appropriate agricultural technologies in improving farming outcomes in rural Zimbabwe.

Figure 11, Figure 12 and Figure 13 display the plotted findings from the DES model, highlighting the bottleneck activities within the sweet potato supply chain. A closer examination of the cultivation data, particularly where bottlenecks were identified, reveals a higher efficiency process. Specifically, cultivation in Ward 2 was found to be active and productive 99.63% of the time, with minimal downtime or waiting periods. This minimal waiting time suggests that the number of hectares under cultivation, coupled with limited resources in Ward 2 (as illustrated in Figure 11), has placed significant strain on the farmers. As a result, they are required to operate at near maximum capacity to successfully manage the 30.4 hectares of land. The data indicates that the high level of efficiency observed in the cultivation process is driven by the necessity to maximize output under resource-constrained conditions. This situation underscores the critical need for better resource management and allocation to alleviate the pressure on farmers and ensure sustainable productivity.

In Ward 3, the results depicted in Figure 12 reveal a 13.17% waiting period, indicating that there is room for optimization in the cultivation process. Reducing this waiting time could potentially decrease the backlog of 21 batches that are currently waiting in the queue to be cultivated across the 18.26 hectares of land. This suggests that with improved resource allocation or more efficient scheduling, the cultivation process could be streamlined, leading to more effective use of available land and resources.

In Ward 7, despite the land size being slightly larger than in Ward 3 (18.59 hectares vs. 18.26 hectares), the results indicate a smaller number of sweet potato vine batches queuing before the “Cultivation at Ward 7” activity. This reduced queuing is attributed to the availability of hired labor and mechanized resources, which help streamline the cultivation process. However, the 9.96% waiting period suggests that there is still room for future improvements to further optimize the process and reduce delays (see Figure 13).

3.4. Modeling Improved Sweet Potato Production

The model built using Simul8 provides a robust framework for testing various interventions aimed at improving the sweet potato supply chain. These interventions include strategies to reduce bottlenecks by adjusting planting schedules, optimizing sorting and grading processes, and enhancing the efficiency of packaging. By offering a detailed analysis of supply chain dynamics, the DES model aids in better decision-making for farmers, suppliers, and policymakers, ultimately fostering a more resilient and efficient supply chain.

The 201 surveys conducted in Goromonzi revealed that the average land size for sweet potato farmers is approximately 0.41 hectares. This presents significant opportunities for optimizing the supply chain and enhancing productivity and efficiency across the board. One key area for improvement is reducing the time required to complete each activity in the supply chain, which is crucial for avoiding bottlenecks. This optimization is directly influenced by the land size, as larger areas require more time and resources to cultivate and manage effectively.

Another critical area for improvement is the availability of resources to complete activities promptly. In the context of cultivation, the increasing climate stresses have resulted in shorter planting windows. As the land size increases, the demand for resources—such as labor, equipment, and inputs—also rises to ensure that planting is done efficiently and yields are maximized. These resources are also essential for protecting the planted vines from pests and diseases during their maturation phase.

Furthermore, the availability of resources plays a vital role in ensuring that mature sweet potatoes are harvested promptly and then processed—either sent to the market for sale or allocated for home consumption. Delays in any part of this process can lead to reduced yields, quality degradation, and financial losses. Figure S2 (Supplementary File)illustrates the improved version of the current model, highlighting how these optimization strategies have been integrated. The improved scenarios have demonstrated an increase in production volume across all five wards, with notable results in Ward 2, which produced 594 tonnes of sweet potatoes on 30.4 hectares; Ward 3, which yielded 387 tonnes on 18.26 hectares; and Ward 7, which resulted in 325 tonnes on 18.58 hectares.

The improved simulation model presented in supplementary file highlights also key areas within the sweet potato supply chain that require targeted interventions to boost efficiency and minimize losses. By reducing the time needed for various activities through the allocation of additional resources, the production volume can be significantly increased. Figure 14 compares the findings from past scenarios, the current scenario based on survey data, and the improved scenario, as depicted in Figure S1 (see Supplementary File). This comparison illustrates the potential impact of implementing these strategic interventions on the overall performance of the supply chain. Figure 14 below show the breaking down of the current and improved sweet potato production chain.

Figure 15 illustrates the evolution of sweet potato production from past practices to current and improved scenarios, with particular attention to the stages of cultivation, harvesting, and packaging. Vine cultivation is assumed to be applicable to all three scenarios. The stages of tuber harvesting and sack packaging, however, have seen noteworthy improvements.

In the tuber harvesting stage, the current scenario’s yield falls within the range of unimproved (0.5 tonnes), tissue-cultured (1.8 tonnes), and irrigated (up to 25 tonnes) sweet potato varieties, as documented in previous studies [43,44]. Figure 15 indicates a significant increase in yield from past practices to the current scenario, despite many farmers still relying on unimproved sweet potato varieties, which are more susceptible to pests and diseases. This increase highlights the partial success of the National Development Strategy 1, which emphasizes the development of the sweet potato value chain in Zimbabwe [15].

The improved scenario in Figure 15 assumes that farmers in the targeted areas have sufficient resources to cultivate white-skinned sweet potatoes under both dry and wet climatic conditions.

The improved production volumes demonstrate significant increases in yield per hectare across all three wards. Ward 3 shows the most substantial percentage increase at 461%, followed by Ward 2 at 431%, and Ward 7 at 345%. These remarkable gains suggest that optimizing practices—such as better irrigation, improved seed varieties, enhanced farming techniques, and more effective use of inputs—can dramatically boost sweet potato productivity.

Currently, yields per hectare are relatively low across all wards, indicating that the land is not being used to its full potential. The optimized scenario uncovers considerable latent capacity for sweet potato production in these regions. Beyond achieving higher yields, this optimization indicates the potential for scaling these practices to improve food security and economic outcomes in similar agricultural settings.

However, even in these optimized scenarios, where yields reached 1671 tonnes, the production did not meet the 25 tonnes per hectare yield associated with irrigated sweet potato varieties, which would correspond to 2063 tonnes per the literature. This shortfall highlights the ongoing challenges posed by recent climate conditions, which have likely impacted production volumes compared to past standards. Despite the availability of resources, the reduced yield suggests that climate variability continues to be a significant limiting factor in achieving maximum productivity in sweet potato farming.

Overall, Figure 14 and Figure 15 suggest that optimization could significantly boost outputs for market sales and family consumption, despite a concurrent increase in decay and processing capacity. This underscores the advantages of optimization while also highlighting areas needing further attention, such as minimizing decay losses. These improvements indicate that, while cultivation practices have remained stable, substantial progress has been made in harvesting and packaging efficiency, contributing to a more resilient and productive supply chain. This evolution in practices reflects the broader objective of adapting to environmental challenges and market demands, ensuring the sustainability and viability of sweet potato production in regions like Goromonzi.

4. Discussion

The integration of climate data into the simulation model provides valuable insights into the resilience of various sweet potato varieties under different weather conditions. The analysis reveals that yellow-skinned sweet potatoes are the most suitable for harsh climate conditions due to their strong resilience and drought resistance. However, despite these advantages, yellow-skinned sweet potatoes are less popular among farmers compared to white-skinned varieties, primarily due to their reduced taste quality. This finding suggests that achieving a balance between crop resilience and market demand is essential for optimizing production and ensuring that farmers’ efforts align with consumer preferences.

The DES model of the sweet potato supply chain identified several critical bottlenecks and inefficiencies that significantly affect overall performance. A primary bottleneck was found in the cultivation stage, where farmers experienced delays in planting due to limited capacity and resources. This delay caused a backlog that extended throughout the supply chain, leading to inefficiencies in subsequent stages such as processing and storage. The prolonged planting period not only increased the risk of post-harvest losses but also hindered the supply chain’s ability to meet market demand promptly, a challenge similarly observed in other agricultural supply chains [45].

Farmers in Goromonzi are facing an expected shortage in production volume, exacerbated by empty sweet potato ground storage units, commonly known as “pfimbi”, which are traditionally capable of preserving sweet potatoes for up to six months. These storage facilities are currently underutilized due to insufficient harvests. This shortage is largely attributed to recurrent extreme weather conditions, with most farmers (over 97%) opting to plant sweet potatoes only during the summer season due to unreliable rainfall and the lack of irrigation systems in the winter months. This seasonal dependency leaves farmers vulnerable to the impacts of climate variability, which has become increasingly challenging in recent years [46].

Another significant risk identified in the study is the prevalence of pests and diseases. The survey revealed that over 80% of farmers in Goromonzi rely on commercial pesticides, which cost between $5 and $9, to protect their crops. However, the financial burden of these pesticides is a considerable challenge for many farmers, especially those with limited means. Approximately 3% of farmers, unable to afford commercial pesticides, resort to using ashes as a low-cost alternative to combat pests and diseases. This practice, while traditional, is less effective and further contributes to the vulnerability of their crops [14,44].

Additionally, inefficiencies in transportation logistics, such as suboptimal routing and scheduling, resulted in extended transit times and elevated fuel costs—issues that have been documented as key challenges in the agricultural sector [47] (Jones & Brown, 2019). Storage facilities also exhibited significant inefficiencies, particularly in maintaining appropriate temperature and humidity levels, contributing to increased spoilage of sweet potatoes during storage. These findings highlight critical areas within the supply chain that necessitate targeted interventions to enhance efficiency and reduce losses.

Based on the simulation outcomes, several strategic improvements have been identified to enhance the efficiency of the sweet potato supply chain. Addressing the processing bottleneck is a key priority, and it is recommended that investments be made in additional processing equipment and the adoption of best practices in maintenance. This approach aims to increase processing capacity and minimize downtime, which is crucial for meeting market demand and reducing post-harvest losses. These recommendations align with the policy directives outlined in the National Development Strategy 1 (NDS1), which emphasizes the need for improved agricultural infrastructure to enhance value chain efficiency and support economic growth in the agricultural sector [15].

Drawing lessons from neighboring countries like South Africa and Eswatini, where initiatives such as Eswatini’s Tinkhundla system provide platforms for traditional leadership to participate actively in local development, similar bottom-up approaches could be beneficial in Zimbabwe. The Tinkhundla model facilitates local access to resources such as training, equipment, fertilizers, and funding, in partnership with local services and the central government [11,48]. Adopting a similar framework in Zimbabwe could empower local farmers by providing them with the necessary resources and support to enhance their agricultural practices and productivity.

In terms of transportation logistics, encouraging collaboration between the farming community, driver’s associations, and middlemen conducting sales in markets like Doboshawa and Mbare could facilitate the timely shipment of sweet potatoes to the market. Furthermore, the survey revealed that roads leading to the markets are perceived by most farmers as inadequate, discouraging drivers from undertaking the journey to, often distant, farms. Reaching the market promptly would reduce transit times and transportation costs, contributing to a more efficient supply chain. The National Transport Policy (2021) supports such measures, advocating for the modernization of transportation logistics to improve the overall efficiency of agricultural supply chains [49].

Moreover, upgrading storage facilities to include modern climate control systems is essential for reducing wastage and maintaining the quality of sweet potatoes. This recommendation is in line with the National Agricultural Policy Framework (2019–2030), which emphasizes the importance of enhancing post-harvest management to preserve the quality of agricultural products and minimize losses. By improving storage infrastructure, the supply chain can better handle fluctuations in production and market demand, ensuring a more stable and resilient agricultural economy.

Furthermore, the simulation underscores the critical role of promoting community-based approaches that facilitate resource sharing and barter trading to collectively enhance production volumes. In resource-constrained communities, it is common for farmers to exchange services during different phases of the supply chain. For example, one community may assist with planting in exchange for help during the harvesting phase, thereby maximizing collective efforts and optimizing resource use. This reciprocal approach not only boosts production efficiency but also strengthens community bonds, making the supply chain more resilient to external shocks.

Finally, the simulation emphasizes the importance of improving coordination and information sharing among supply chain stakeholders. By aligning supply with demand more effectively, stakeholders can minimize mismatches that lead to either surplus or shortages, thereby creating a more streamlined and responsive supply chain. This enhanced coordination is particularly vital in managing the timing and distribution of sweet potatoes, ensuring that market demands are met without overburdening the supply chain.

These strategic interventions, informed by the insights gained from the DES model, offer a clear pathway for developing a more efficient and resilient sweet potato supply chain. The model’s capacity to guide decision-making and prioritize necessary improvements highlights its value as a tool for policymakers, community leaders, and industry stakeholders. By adopting these recommendations, stakeholders can work towards creating a supply chain that is not only efficient and productive but also resilient to the challenges posed by climate variability and resource constraints.

5. Conclusions

This study employed a Discrete Event Simulation (DES) model to optimize the sweet potato supply chain in Zimbabwe, providing critical insights into the production, distribution, and market dynamics. The DES model identified several key bottlenecks in the supply chain, particularly in cultivation and transportation, that hinder the overall efficiency. Addressing these inefficiencies is crucial to improving the supply chain’s resilience and sustainability, especially under varying climate conditions.

Crucially, the findings highlight the importance of adopting a data-driven approach. Data on factors such as market demand, transportation logistics, and climate conditions must form the foundation for identifying bottlenecks and determining key intervention areas. The integration of real-time data can further optimize decision-making processes, ensuring that improvements to the supply chain are targeted and effective. For example, data-supported adjustments in transportation routes or resource allocation can significantly reduce delays and losses. Additionally, enhancing access to climate resilience strategies, such as improved irrigation techniques and pest management, should be aligned with data insights to bolster both supply chain efficiency and crop yield under adverse environmental conditions.

Grounding future supply chain improvements in rigorous data analysis, stakeholders can ensure more sustainable agricultural practices that not only boost productivity but also enhance food security across regions facing similar climate and supply chain challenges. Future research should focus on integrating more comprehensive data analytics to continuously monitor and address emerging bottlenecks, thereby further optimizing the supply chain.