Muhammad Abdullah Bin Arif

Mr. Muhammad Abdullah Bin Arif

orcid

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 August 2025 in Annual Methodological Archive Research Review
Verified
Journal Article
Open access
PDF file
SBSaad Khan Baloch
Saad Khan Baloch
MAMuhammad Abdullah Bin Arif
Muhammad Abdullah Bin Arif
BABin Arif
Bin Arif
IKImad Khan
Imad Khan
MWMuhammad Waqar
Muhammad Waqar
HUHassaan Bin Umar
Hassaan Bin Umar
Inclusive of the Vehicle-to-Grid (V2G) technology, bidirectional charging is a paradigm shift in moving toward sustainable and smart energy systems. The paper examines technical, economic, and environmental impacts of employing infrastructures based on smart EV with support to the V2G. An all-embracing simulation approach was used, and the grid interactions, behavior of batteries, and operating conditions were modeled at the low (10%), medium (30%), and high (50%) levels of adoption. The findings indicate that peak load reduction of up to 38 percent, 20 percent improvement in frequency stability and high level of line loss reduction, leading to increased grid reliability and operational performance, can be realized by high levels of V2G participation. The economically proves to be very rewarding to EV owners and aggregators with annual net profits of $120 per EV in low participation going up to $350 per EV in high participation situations and system level savings and payback times that strengthen with scale. Environmental analysis underlines positive efforts in achieving decarbonization since CO 2 can be reduced by 770 kg per E.V each year mainly through the optimal use of renewable energy. Nevertheless, the results also revealed some trade-offs especially in the case of battery degradation which aggravates with increased frequency of cycling, the need therefore is to optimize dispatch strategies and facilitate enabling policies. The study comes to the conclusion that wide use of V2G is possible and beneficial, supported by the intelligent control algorithms and user-integration, standardization and incentives, thus making V2G a key enabler of future-proof and low-carbon power systems. Keywords: Vehicle-to-Grid (V2G), bidirectional charging, electric vehicles, smart grid, battery degradation, peak load reduction, renewable energy integration, grid stability, energy management, sustainable transportation.

30 July 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
MAMuhammad Abdullah Bin Arif
Muhammad Abdullah Bin Arif
Sanchari DebSanchari Deb
Sanchari Deb
A battery swapping station (BSS) exchanges a depleted electric-vehicle pack for a charged one in minutes, and its shelf of batteries doubles as a grid-connected store that can buy energy when it is cheap and sell it back when it is dear. Running the two businesses together is a scheduling problem: when to charge each pack, when to discharge to the grid, and which pack to hand to which customer. This paper sets that problem out as battery resource allocation (BatRes). In its general form the model is a mixed-integer program, because the request-to-battery assignment is binary and a no-simultaneous-charge-and-discharge condition is nonlinear. Under a first-come-first-served rule the assignment is fixed, and because conversion losses are charged on both power directions the nonlinear condition never binds at the optimum; the model that is actually solved is therefore a linear program, solved to global optimality. Two cases are reported under the industrial time-of-use tariff of Pakistan: a small three-request, two-battery, twelve-slot case that can be read decision by decision, and a reproducible synthetic full-day case with thirty requests across eight batteries. In the small case, net profit ranges from PKR 939.2 to PKR 994.7 across the six arrival orders, and adding the swap business raises profit to about four times the grid-energy-services-only level at this tariff and swap price. Sensitivity sweeps over the swap price, tariff premium, inventory size, efficiency, and depth of discharge map how each shifts the result. Every figure is specific to the assumed tariff, service price, deterministic demand, and station parameters, and is presented as such.

23 October 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
MAMuhammad Abdullah Bin Arif
Muhammad Abdullah Bin Arif
SDSanchari Deb
Sanchari Deb
Public charging is where most drivers form their opinion of electric mobility, and drivers differ in what they want from a charge: the cheapest fill, the fastest turnaround, or the gentlest treatment of the battery. Smart-charging research optimises the grid or the bill and treats the driver as an arrival time and an energy demand. This paper asks a policy question instead: on a real public rapid-charging site, how much does matching the charging schedule to each driver’s declared priority actually help, and does an operator need per-driver optimisation to deliver it? We schedule each session with a transparent per-session convex program solved by water-filling, and we test it on 41,213 cleaned charging sessions recorded over five years at the Newcastle Helix site, against five baselines that include the strong ones a sceptic would demand: a single fixed multi-objective policy, the fleet-average preference, and a simple user-selectable mode menu. Matching schedules to declared priorities raises the share of sessions meeting the driver’s own objective from 32% (one-size-fits-all) to 41%, and to 52% on the 39% of sessions that have real dwell slack. The gain comes almost entirely from battery-health-sensitive drivers; the cost and peak objectives are set by the site’s short dwell, not by the scheduler, and no policy moves the average bill (about GBP 7.45) or the peak-window load. A simple user-selectable mode menu reproduces the optimised result exactly, so the benefit is deployable today without per-driver optimisation. Because driver priority, battery size, and state of charge are derived rather than observed, we treat the headline as a conditional estimate and show it depends on the assumed preference mix, ranging from 34% to 62%. The policy reading is that offering drivers a few honest charging modes is worth more than optimisation sophistication on rapid sites, and that genuinely battery-friendly charging is a question of dwell time, and therefore of where chargers are sited, more than of control.

22 August 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
MAMuhammad Abdullah Bin Arif
Muhammad Abdullah Bin Arif
SDSanchari Deb
Sanchari Deb
Hybrid renewable energy systems (HRES) that put solar, wind, and battery storage under one control scheme are widely presented as the way to keep an inverter-dominated grid stable, yet many studies describe their simulations only in words and report summary indicators without the dynamic quantities that actually decide stability. This paper supplies those quantities through three linked, fully specified studies on named test systems. A reduced-order system-frequency-response model measures the rate of change of frequency (RoCoF), the nadir, the settling time, and the steady-state deviation for four strategies (uncoordinated, adaptive droop, virtual inertia, and a coordinated model-predictive-plus-virtual-inertia scheme) under a contingency, across a penetration sweep, and in a 300-trial Monte-Carlo with one-way ANOVA. A receding-horizon model predictive control (MPC) scheduler is then compared against a greedy rule on curtailment, unserved energy, cost, state of charge, and forecast robustness. A Volt/VAR study on the IEEE 33-bus feeder measures daily voltage fluctuation under real AC power flow. The central finding is metric-dependent and corrects a common belief: adaptive droop, ranked best by summary metrics, does not reduce RoCoF at all (it stays at 2.50 Hz/s, crossing a 1 Hz/s grid-code limit above about 40% penetration), because RoCoF at the instant of a disturbance is set by inertia alone, and only virtual inertia lowers it (to 0.50--0.63 Hz/s). The MPC does not cut curtailment, which is storage-bound, but removes unserved energy (0.77 to 0.00 MWh/day), cuts cost by 37%, and holds the frequency-reserve band on 100% of days against 0% for the greedy rule. Coordinated Volt/VAR cuts the daily voltage swing by 11 to 26%, and the ANOVA confirms every frequency difference at p<10^-6. Results are reported as they came out, including the cases that contradict common assumptions.

22 September 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
MAMuhammad Abdullah Bin Arif
Muhammad Abdullah Bin Arif
SIShahid Iqbal
Shahid Iqbal
SDSanchari Deb
Sanchari Deb
Most smart-charging research serves the grid or the electricity bill and treats the driver as an arrival time and an energy demand. Drivers differ: some want the cheapest charge, some want the car ready soon, some want the battery treated gently. This paper schedules each session with a small convex program whose weights come from the driver’s stated priority, solved by water-filling so it runs on the charger without an external solver. Tested on 41,213 real sessions from the Newcastle Helix site, a non-personalised smart policy meets the driver’s own objective in about 30% of sessions; the user-centric scheme meets it in 55%, holding within one percentage point across 600, 900, and 1200 sessions. Because priorities, battery sizes, and target charge states are absent from the logs, they are assigned by transparent rules; the advantage survives an independent re-drawing and stronger baselines. Personalisation buys driver satisfaction, not a cheaper bill.

27 August 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
MAMuhammad Abdullah Bin Arif
Muhammad Abdullah Bin Arif
SIShahid Iqbal
Shahid Iqbal
Learning controllers for electric vehicle charging are usually taught to respect network limits through a penalty term added to the reward. The penalty is soft, so the constraint can still be violated during training and, less often but more seriously, after deployment. This paper reports a depot charging controller in which the network constraint is instead enforced by a closed-form projection applied to the action before it reaches the plant, and it quantifies what that choice is worth on a measured connection. The problem is posed as a two-pool constrained Markov decision process with separate deadlines for vehicles that stay overnight and vehicles that return at midday, and it is solved with a deep Q-network of 6165 parameters trained without any machine learning framework.On a measured 11 kV connection serving a 94-route institutional bus fleet in Gujrat, Pakistan, the proposed controller reduces daily charging cost by 45.0 percent against uncoordinated charging, from 340,788 to 187,335 PKR per day, and sits 4.4 percent above a perfect-foresight linear programming bound. The uncoordinated schedule breaches the statutory voltage floor six times a day. Every controller carrying the projection layer breaches it zero times across 40 held-out test days, eight independent training seeds, and nineteen hyperparameter settings, which is 38 separately trained networks.The ablation gives the paper its main result, and it is a negative one. With the projection layer active, an agent trained on a physics-informed reward and an agent trained on cost alone converge to the same policy and the same daily cost of 235,544 PKR because the voltage penalty is never incurred and therefore carries no gradient. Reward shaping for constraint satisfaction is redundant once a hard layer is present, and it is only when the layer is removed that shaping earns its place, cutting violations from 1.20 per day to zero at the cost of 34 kWh of undelivered energy. The practical implication is that effort spent tuning penalty weights is better spent on the projection.