Control Strategies for a Shared Diesel Generator in an Off-Grid Hybrid Power System for Urban Areas in Pakistan

This paper presents control strategies for a hybrid off-grid power system serving an urban residential neighborhood in Pakistan, integrating solar photovoltaic (PV) generation, battery storage, and a shared diesel generator. The primary objective is to ensure power reliability and prevent generator overloading in a multi-household configuration. The system was initially optimized for component sizing and cost using HOMER Pro, followed by the development of a detailed dynamic model in MATLAB Simulink/Simscape. Building on this foundation, two complementary control schemes are designed: one for load distribution among the houses and another for battery bank charging. Simulation results demonstrate that the proposed controls stabilize the load supply—ensuring that the diesel generator consistently maintains single-phase voltage around 220 V while sharing a 2kW load per house without exceeding its capacity—and facilitate efficient battery charging on a common 48 V Direct Current (DC) bus. The diesel generator is time-shared across seven households, delivering approximately 9.65 A Root Mean Square (RMS) per active phase, which prevents overloading and ensures continuous power availability. The batteries are charged in a balanced manner whenever solar energy is insufficient, preserving system autonomy. By sharing the diesel generator, the system reduces battery size requirements, thereby lowering both capital and operational costs. This work offers a scalable, cost-effective control strategy for a PV–diesel–battery hybrid microgrid, providing a viable solution to Pakistan’s urban energy shortages.

Introduction

Energy is fundamental to modern civilization, driving industrial growth, transportation, infrastructure development, and domestic activities, thereby underpinning economic stability and social advancement. This is especially critical in rapidly developing countries such as Pakistan, where increasing industrialization, urbanization, and population growth are intensifying energy demand. Pakistan benefits from considerable solar energy potential, with an average global solar insolation between 5 and 7 kWh/m²/day, positioning solar power as a viable and sustainable renewable resource to address the country’s expanding electricity needs [1].

However, Pakistan’s energy sector is facing significant challenges. Persistent inefficiencies in power transmission and an aging distribution infrastructure contribute to substantial energy shortages. The country’s dependence on imported fossil fuels exacerbates these issues by increasing electricity costs, depleting foreign reserves, and accelerating environmental degradation through elevated greenhouse gas emissions. Moreover, contractual obligations with rental power projects impose capacity payments regardless of actual electricity generation. Combined with subsidy removals and additional taxation, these factors have driven up energy prices, imposing a heavy financial burden on consumers [2].

Despite an installed generation capacity of approximately 45,888 MW as of 2025, Pakistan experiences considerable underutilization, with capacity factors averaging only 33.88%. This underperformance results in frequent and prolonged power outages, particularly during peak demand periods. For example, the April 2025 heatwave led to outages lasting up to 16 hours in some regions, severely affecting industrial operations and economic productivity. Many businesses cite unreliable electricity and high energy costs as major growth constraints [3].

The economic impact of these challenges is further underscored by the accumulation of circular debt, which reached PKR 2.636 trillion in 2024, equivalent to approximately 2.5% of Pakistan’s GDP. This debt reflects structural inefficiencies, including underused generation capacity, onerous capacity payments to Independent Power Producers (IPPs), and ineffective revenue collection. In Fiscal Year 2025, capacity payments alone are projected to amount to PKR 2.1 trillion, corresponding to a tariff surcharge of approximately PKR 17.31 per kilowatt-hour. These financial obligations strain public resources and contribute to elevated electricity tariffs, reducing affordability for consumers and businesses alike [4], [5].

In our previous work, we proposed a hybrid energy system for a small urban community in Karachi that could operate independently of the national grid. This system was designed to be both reliable and cost-effective, leveraging Pakistan’s abundant solar resources while avoiding heavy taxes and regulations associated with utility power. It integrates rooftop solar PV panels at each residence, battery storage banks, and a shared diesel generator as backup. The design and optimization of the system components (PV array sizes, battery capacity, and generator rating) were carried out using HOMER Pro and are presented in detail in [6]. A follow-up study developed a dynamic simulation model of the hybrid system in MATLAB/Simulink to analyze its transient behavior and validate the feasibility of achieving a high renewable energy fraction [7]. Building on these foundations, the focus of this study is to develop control strategies for the hybrid power system that ensure stable operation and prevent overloading of the shared diesel generator.

Literature Review

This section presents a comprehensive review of existing research on hybrid solar–diesel–battery power systems, highlighting key developments in dynamic modeling, control strategies, and load management techniques that align with the objectives of this study.

Al-Wakeel et al. offer valuable insights into the role of diesel generators as temporary solutions for electricity shortages in areas with unreliable grid supply. Their case study in Iraq demonstrates that, where grid access can be limited to as few as six hours per day, numerous neighborhood-scale diesel generators rated between 100 and 500 kVA have been deployed within isolated mini-grids to provide essential loads such as lighting, fans, and small appliances during prolonged outages. Although these generators meet critical energy demands, this study highlights the significant operational costs and environmental impacts resulting from extended diesel usage. Importantly, the research underscores the advantages of integrating renewable energy sources, particularly rooftop photovoltaic systems, with diesel units to reduce fuel consumption and operational time. This experience strongly supports the adoption of hybrid energy solutions in regions facing persistent grid unreliability [8].

Several studies have adopted a two-stage design and simulation approach for hybrid systems. For example, Oyogbuda and Iqbal designed a DC solar microgrid for a remote Nigerian community, using HOMER Pro for optimal sizing of PV, battery, and generator components, and then validated its performance through dynamic simulations in Simulink. The 48 V DC network supplied nine homes and demonstrated stable power-sharing among houses under varying irradiance and load conditions [9], [10]. This work illustrates the value of combining techno-economic optimization with time-domain simulation to ensure that a hybrid system meets both cost and reliability criteria. Similarly, Salau et al. developed an integrated model of a PV–battery–diesel hybrid energy system for an off-grid healthcare facility and evaluated its dynamic response to fluctuations in the load and solar irradiance. Their simulation results confirmed that coordinated control strategies effectively maintain system stability and optimize energy management under varying operating scenarios [11].

Several researchers in Pakistan have investigated intelligent control enhancements for hybrid microgrids. Khalid et al. proposed a two-step approach for rural electrification. Initially, they demonstrated through HOMER and Simulink simulations that an off-grid PV–diesel–battery system can achieve complete renewable energy penetration while maintaining cost-effectiveness and reliability. This study incorporated hardware-in-the-loop experiments to validate the dynamic responses under real-world conditions, successfully achieving stable operation with varying load and generation profiles [12]. In a subsequent study, Khalid et al. integrated an Internet-of-Things based SCADA system into a hybrid configuration to facilitate real-time monitoring and adaptive control. By employing low-cost microcontrollers, such as Arduino and ESP32, along with the Blynk IoT platform, they developed a remote supervisory control architecture capable of tracking system performance and enabling automated adjustments via GSM and WiFi connectivity [13]. This IoT-SCADA implementation enhanced operational efficiency and user engagement by providing live data visualization and control over distributed components, such as remote battery charging management, thereby improving the overall reliability and maintainability of the hybrid system. The combination of a robust system design with smart monitoring and control features presented in these studies offers a valuable framework for sustainable energy solutions in underserved communities.

Aziz et al. developed a dispatch control strategy for PV/diesel/battery systems in rural Iraq using MATLAB Simulink, optimizing for minimal fuel use and emissions while ensuring load demand satisfaction through dynamic switching techniques and intelligent controllers, such as PID and fuzzy logic systems [14]. Rezk et al. proposed an energy management system (EMS) with fuzzy logic-based decision-making to control the power flow between solar, diesel, and battery subsystems in off-grid setups. Their simulations demonstrated improvements in diesel runtime reduction and voltage stability using MATLAB Simulink and real load profiles [15]. Mohammed et al. comprehensively reviewed hybrid PV–diesel generator systems and identified that the most effective designs integrate model-predictive controllers, hysteresis charging control, and SCADA-based remote monitoring to enhance adaptability in real-time operations [16]. Kharrich et al. presented an equilibrium-optimizer-based control model for hybrid PV/wind/diesel/battery systems. Their results demonstrated improved fuel efficiency and voltage regulation using MATLAB-based simulations in an islanded microgrid context [17]. Amole et al. compared different AC and DC control schemes for stand-alone solar-diesel-battery microgrids. Their study emphasized the use of decentralized agents and programmable controllers for scalable multi-household systems, similar to the neighborhood model in this study [18]. These studies collectively highlight that hybrid microgrids can be made resilient through advanced control logic, precise simulation, and dynamic load-sharing mechanisms. However, a gap still exists in low-cost, modular strategies for real-time generator control with minimal hardware, which is the focus of the present study.

System Description

The study site is an urban neighborhood in Karachi located at latitude 24.919904 and longitude 67.145514, with an average solar irradiance between 5.45 and 5.6 kWh per square meter per day, peaking in May and lowest in December. Flat-roofed houses, none of which currently have solar installations, provide ideal conditions for solar panel deployment. Fig. 1 shows the location of these sites.

Fig. 1. Selected site location.

The system design, component selection, sizing, and optimization were performed using HOMER Pro, and the resulting schematic showing all major components is shown in Fig. 2 [6].

Fig. 2. System model in Homer Pro for each house.

The designed system includes seven residential units in a neighborhood in Karachi, each equipped with an individual photovoltaic (PV) system coupled with battery storage to satisfy their daily energy requirements independently. During periods of low solar irradiation, such as cloudy days or increased energy demand, a centralized diesel generator connected to the DC bus provides backup power, charging the batteries as required, and delivering electricity to each house through single-phase connections and individual breakers. Furthermore, each house is equipped with a dedicated meter to accurately monitor electricity consumption from the shared diesel generator. Fig. 3 shows the block diagram of the system.

Fig. 3. System block diagram.

We specifically address two operational modes of the microgrid: (1) powering residential AC loads via the diesel generator during periods of low solar output, and (2) charging the distributed battery banks using the diesel generator when needed. For each mode, we propose a tailored control scheme that coordinates the generator with the household loads and batteries. These control strategies were implemented and tested using the MATLAB Simulink/Simscape model of the full system.

Diesel Generator

The diesel generator model comprises two tightly integrated subsystems: a mechanical subsystem, which includes the diesel engine and its governor control, and an electrical subsystem, which consists of the synchronous generator and its excitation system. These components were implemented in MATLAB Simulink using Simscape Electrical, employing physical modeling blocks and control structures that closely reflect real-world dynamic behavior. Fig. 4 shows the detailed configuration of the model.

Fig. 4. DG model in Simulink/Simscape.

In the simulated configuration, a four-pole, 50 Hz synchronous machine is rigidly coupled to an inertial representation of a diesel engine through a mechanical shaft. This coupling ensures a consistent alignment between mechanical and electrical angular displacements, maintained by a constant

(1) $$\begin{array}{r}C=\frac{2\pi \ast 50}{4}\end{array}$$

which corresponds to the synchronous speed of 1500 rpm for a four-pole machine at 50 Hz.

The diesel engine converts the chemical energy of the fuel into mechanical torque, which drives the shaft. The rotating inertia of the engine plays a critical role in mitigating the torque pulsations inherent to combustion processes. Additionally, the inertia provides a kinetic energy buffer, which helps maintain speed stability under transient load variations.

The synchronous machine is energized by a direct current excitation of 32 A in the model, producing a rotating magnetic field at a synchronous speed of 1500 rpm, which corresponds to a four-pole machine operating at 50 Hz. This rotating field induces three-phase voltages at the stator terminals. The resulting voltages supply real power to the connected load and, depending on the excitation level and the system loading, enable the exchange of reactive power with the isolated local network.

Active power regulation in the diesel generator system is achieved by controlling the fuel input to the engine, which in turn adjusts the mechanical torque and shaft speed. Reactive power control is governed by the excitation current supplied to the generator's field winding. During electrical disturbances, the generator’s dynamic behavior is significantly influenced by its direct-axis and quadrature-axis reactances, denoted as $X_d$ and $X_q$, along with their respective time constants, ${T^{\prime }}_{d0}$ and ${T^{″}}_{q0}$. These parameters determine the strength and timing of the electromagnetic coupling between the rotor flux and the terminal voltage. As a result, they play a critical role in shaping the generator’s transient response, including voltage overshoot, oscillatory behavior, and overall system damping.

This model captures both the mechanical behavior of the diesel engine and the electrical dynamics of the generator, making it suitable for studying start-up behavior, load changes, and fault conditions. It provides a useful tool for analyzing and improving control strategies, system stability, and overall performance in hybrid renewable energy systems.

Simulation Design and Control Strategies

For this hybrid renewable energy system, two control strategies were designed in this study.

Control Strategy for Load Supply

One primary operating mode occurs when the diesel generator directly supplies the household alternating current loads, such as during nighttime or extended cloudy periods when the photovoltaic output is low and battery storage alone cannot meet demand. In this mode, the control objective is to distribute the generator’s output among the houses in a manner that prevents any single phase from exceeding the generator’s capacity, while ensuring that each household receives power periodically. This was accomplished through a time-sharing scheme implemented via a switching matrix. Fig. 5 illustrates the complete simulation model for the load-sharing control strategy.

Fig. 5. System model for AC load supply.

The three-phase generator output at 50 Hz, with phases A, B, and C, is connected to an electronic single-pole double-throw (SPDT) switch that dynamically routes each phase to one of the two houses based on a predefined schedule. In this configuration, Phase A alternates between House 1 and House 4, Phase B between House 2 and House 5, and Phase C between House 3 and House 6. Due to its relatively small load, House 7 is combined with House 6 as a single equivalent load on Phase C. At any time, each phase supplies power to either the group of Houses 1, 2, and 3 or the group of Houses 4, 5, and 6. The control system toggles the switches to continuously alternate power delivery between these two groups, effectively multiplexing the generator output among them. The advantage of this scheme is that it limits the concurrent load on the diesel generator. Instead of attempting to supply power to all six or seven houses simultaneously, which would draw a current potentially exceeding the generator’s rating, the generator is subjected to approximately half the total neighborhood load at any one time. This control strategy is inspired by practical load-management techniques in microgrids, such as feeder rotation or load prioritization, which prevent generator overload.

To realistically model the residential loads, each house’s alternating current demand is represented by an equivalent impedance network in Simulink. A parallel resistor–inductor branch model captures both the resistive and inductive characteristics that are typical of household appliances. Specifically, each house is modeled as a 2 kW load at 220 V, comprising a resistive branch drawing approximately 1.5 kW, representing devices such as lighting, electronics, and heaters, and an inductive branch drawing approximately 0.5 kW with a lagging power factor, representing appliances such as refrigerator compressors, fans, and small motors. The resistive branch corresponds to a resistance of 32.3 Ω, drawing approximately 6.81 A at 220 V. The inductive branch is modeled as a 62 Ω resistor in series with a 0.148 H inductor; at 50 Hz, this inductor presents a reactance of approximately 46.5 Ω, resulting in a current draw of about 2.84 A at 220 V, corresponding to roughly 0.5 kW of real power and 0.35 kVAR of reactive power. The combined load per house thus draws approximately 9.65 A rms at a power factor close to 0.95, matching the 2 kW load demand.

All load branches are connected to the respective phase lines through the SPDT switch, which functions as a breaker to control the connection and disconnection of houses from the generator. The system is designed to ensure that no more than 2 kW per phase is drawn simultaneously from the generator, thereby maintaining the total current within the safe operating range of the generator. Section V presents the output waveforms that confirm the generator current remains stable and below its rated capacity under this control scheme.

Control Strategy for Battery Charging

The second mode of operation occurs when the diesel generator recharges the batteries of all houses, typically during extended periods of low solar irradiance, such as night-time or consecutive cloudy or rainy days. During these intervals, the AC loads may be minimal; however, battery replenishment is essential to maintain daytime autonomy if solar generation remains insufficient. The control challenge involves efficiently converting the three-phase AC output of the generator to a regulated DC voltage suitable for simultaneously charging seven battery banks and distributing the charging current effectively. This design employs a centralized AC–DC conversion scheme. The generator’s three-phase output was first rectified using a six-pulse diode bridge rectifier. Subsequently, a buck converter implemented in Simscape with insulated-gate bipolar transistor (IGBT) switches steps down and regulates this DC voltage to a precise 48 V required for battery charging. The buck converter utilizes pulse-width modulation (PWM) control at a switching frequency of 10 kHz. A proportional-integral (PI) controller compares the output voltage against a 48 V reference and adjusts the PWM duty cycle to maintain voltage regulation, ensuring a stable DC supply regardless of input fluctuations or varying battery charging currents.

All the battery banks are connected in parallel to this regulated DC bus. Although individual battery charge controllers would typically manage charging current in practice, the simulation assumes ideal load sharing, where each battery draws current proportional to its state of charge without mutual interference. Balanced charging is inherently achieved through voltage regulation at the common bus, allowing each battery to draw current until it approaches full charge.

By integrating the rectifier and buck converter stages into a single centralized converter, this approach enables simultaneous, efficient charging of multiple battery banks from the diesel generator, offering improved efficiency compared to sequential battery charging methods.

This operating mode ensures that when photovoltaic generation is insufficient to sustain battery charge—primarily during evening and nighttime hours—the diesel generator reliably recharge all battery banks to maintain off-grid autonomy for the following day. Fig. 6 demonstrates the simulation model for the battery charging strategy.

Fig. 6. System model for battery charging.

Simulation Results and Analysis

The system was simulated in MATLAB/Simulink using Simscape to evaluate the proposed control strategies under typical operating conditions.

Fig. 7a illustrates the current delivered by the diesel generator to the first group of houses—Houses 1, 2, and 3 connected to phases A, B, and C, respectively—while Fig. 7b shows the current for the second group comprising Houses 4, 5, and 6. Each active phase carries approximately 9.65 A RMS, corresponding to the combined resistive (6.81 A) and inductive (2.84 A) components of the 2kW load per house. The current waveforms exhibit a sinusoidal shape at 50 Hz when the phase supplies power and drops to near zero when disconnected, aside from the minimal magnetizing currents. The time-sharing scheme alternates the power supply between the two groups without overlap, effectively halving the concurrent load and preventing generator overload. The current remains stable and free of distortion throughout the simulation, indicating operation within steady-state limits. Additionally, the generator speed remains near the synchronous speed of 1500 rpm, with some minor transient droop during switching intervals promptly corrected by the governor. Overall, the load current behavior confirms successful load management and stable generator operation under the proposed control strategy.

Fig. 7. (a) Load current waveforms for House 1, 2 & 3. (b) Load current waveforms for House 4, 5 & 6.

Figs. 8a and 8b show the voltage waveforms measured at the load terminals for two representative phases. The generator output is regulated to approximately 220 V AC (RMS), corresponding to a peak-to-peak voltage of about 311 V. The simulated waveforms remain nearly sinusoidal, confirming stable voltage levels throughout the operation. Minor voltage dips occur during switching events owing to transient current changes but are rapidly damped by the generator’s excitation system and mechanical inertia. Voltage magnitudes consistently remained within acceptable limits for single-phase appliances in Pakistan, typically specified as 220 V ± 5%–10%. This stability is maintained by the control strategy that limits sudden load changes by supplying only half of the neighborhood at a time. Although the inductive load introduces a slight phase shift between the current and voltage, it does not cause excessive voltage drop or power quality issues, as the generator operates within its reactive power capacity. Collectively, these results validate that the generator supplies a stable AC voltage in a time-shared manner without exceeding operational limits, ensuring the proper functioning of residential loads.

Fig. 8. (a) Load voltage waveforms for House 1, 2 & 3. (b) Load voltage waveforms for House 4, 5 & 6.

Fig. 9 presents the battery charge waveforms for Houses 1, 2, 4, and 5 during diesel generator charging.

Fig. 9. Battery charging profile.

The graph tracks the cumulative charge in coulombs over time, starting from a partially charged state to simulate evening conditions. Once the generator and rectifier-buck converter are activated, the regulated 48 V DC bus supplies charging current to each battery bank. The charge rate is initially high, reflecting a greater current draw by more discharged batteries, and gradually decreases as the batteries approach full capacity, consistent with constant-voltage charging behavior. The smooth and stable waveforms indicate effective voltage regulation and balanced charging among the represented battery banks, ensuring reliable energy storage to support off-grid operations.

Conclusion

In this study, two new integrated control strategies for managing a hybrid PV–battery–diesel power system were developed and validated through detailed simulations in MATLAB/Simulink. The first strategy implements a time-shared load distribution scheme that sequentially supplies subgroups of residential loads to prevent diesel generator overload. The second strategy manages battery charging through controlled rectification and DC–DC conversion, optimizing the battery recharge cycles during periods of insufficient solar energy. Simulation results demonstrate stable system operation under realistic scenarios, with the diesel generator maintaining nominal voltage and frequency while effectively supplying residential loads and efficiently recharging batteries within the prescribed current limits. Consequently, the proposed hybrid configuration reliably meets residential power demands during grid outages or peak load periods, thereby ensuring system autonomy from the main grid. This approach provides a cost-effective alternative compared to standalone diesel generators, which entail high initial investment and ongoing maintenance expenses, and also outperforms purely battery-based systems, which involve substantial upfront costs and periodic battery replacements approximately every five years. Furthermore, the simplified maintenance requirements of a shared diesel generator present business opportunities for local operators and entrepreneurs, particularly if such hybrid systems are widely deployed in urban regions such as Karachi.

Suggested Future Work

For future studies, several improvements are recommended to build on this work:

Predictive and Adaptive Control: Future work may include the implementation of advanced control strategies such as Model Predictive Control (MPC) or artificial intelligence (AI)-based controllers to enhance system performance. Predictive algorithms can leverage short-term forecasts of the solar irradiance and load demand to proactively schedule generator operations and manage battery charging. For instance, an MPC controller can anticipate evening peak loads and initiate battery charging in advance or preemptively activate the generator before a sudden drop in PV output, thereby minimizing the frequency and voltage deviations [19]. Such intelligent control would reduce generator runtime, improve fuel efficiency, and maintain power quality by addressing disturbances proactively instead of reactively.

IoT and SCADA Integration: While the current design relies on local control for switching and voltage regulation, integration of Internet of Things (IoT) and Supervisory Control and Data Acquisition (SCADA) platforms could significantly enhance real-time monitoring and operational flexibility. By deploying IoT sensors to measure voltage, current, fuel levels, and battery state-of-charge (SOC), and by interfacing with a cloud-connected SCADA dashboard, operators can remotely monitor the system status, dispatch the generator based on real-time conditions, and automate fault detection. As demonstrated in recent studies, low-cost IoT-SCADA implementations using platforms such as Blynk can provide functionalities such as overload alerts and selective load shedding [13]. Integration with smart meters and home automation systems can also enable a demand response, allowing temporary reduction of non-critical loads during peak stress. This would enhance the resilience of the system, reduce downtime, and improve maintainability.

Optimization Under Uncertainty: Owing to the stochastic nature of solar irradiance and variable household loads, robust control under uncertainty is essential. Future research should explore stochastic and scenario-based optimization techniques that incorporate probabilistic models of weather patterns and consumption profiles. Machine learning methods can be used to forecast system conditions and generate optimal schedules for generator usage and battery charging, with the goal of minimizing fuel consumption and emissions. Multi-objective optimization may also be applied to balance the operational cost, system reliability, and component wear. For instance, start-stop thresholds for generator operation could be optimized to prevent overuse while ensuring adequate battery levels. These techniques would allow the microgrid to adapt to seasonal variations and unpredictable demands, thereby enhancing long-term sustainability.

Hardware Demonstration and Field Trials: To validate the proposed control schemes under real-world conditions, a hardware-in-the-loop (HIL) setup or field-scale prototype is recommended. A scaled system comprising a few houses, small PV arrays, battery storage, and a diesel generator can be constructed using programmable logic controllers (PLCs) or embedded microcontrollers. This setup allows the assessment of practical concerns such as switching transients, control loop delays, and fuel efficiency under cycling operation. Data collected from pilot deployments in urban neighborhoods in Pakistan can inform further refinements in algorithm design, such as adding hysteresis to battery SOC thresholds to prevent rapid toggling. Real-world testing is a necessary step toward deployment, ensuring that the proposed strategy is not only effective in simulation but also reliable and robust in operational environments.