I’m Elias, an energy systems modeller and doctoral researcher working at the intersection of long-term resource planning, open-source modelling and evidence-based decisions.
Selected work on renewable-energy potential, storage, electrification, resource planning and regional power systems.
01
M. E. Islam, P. McWhannel, and T. Niet, “Mapping feasible renewable transition space: Land-use, conservation, and grid-access constraints on wind and solar in British Columbia,” Energy 360, vol. 6, p. 100077, Dec. 2026.
B. Borba, M. E. Islam, P. McWhannel, and T. Niet, “Storage in Long-Term Energy Planning Models with Temporal Aggregation: Best Practices and Algorithms,” SSRN, 2025.
K. Kuling, M. E. Islam, P. McWhannel, E. Kjeang, and T. Niet, “Modelling the Grid Impacts of Electric Vehicle Uptake and Charging Strategies in British Columbia,” SSRN, 2025.
N. Arianpoo, M. E. Islam, A. S. Wright, and T. Niet, “Electrification policy impacts on land system in British Columbia, Canada,” Renewable and Sustainable Energy Transition, vol. 5, p. 100080, Jan. 2024.
Md. E. Islam, Md. M. Z. Khan, D. Chattopadhyay, and J. Väyrynen, “Impact of COVID-19 on dispatch and capacity plan: A case study for Bangladesh,” The Electricity Journal, vol. 34, no. 5, p. 106955, Jun. 2021.
Md. E. Islam, Md. M. Z. Khan, D. Chattopadhyay, and G. Draugelis, “Economic Benefits of Cross Border Power Trading: A Case Study for Bangladesh,” in 2020 IEEE Power & Energy Society General Meeting, Montreal, Canada, pp. 1–5.
Md. E. Islam, Md. M. Z. Khan, C. Nicolas, and D. Chattopadhyay, “Planning for Direct Load Control and Energy Efficiency: A Case Study for Bangladesh,” in 2019 IEEE Power & Energy Society General Meeting, Atlanta, USA, pp. 1–5.
How a messy real-world grid becomes a tidy set of equations a computer can solve — and the one distinction that trips most people up: capacity-expansion versus operational models. Scroll to build the picture.
Part A · building a toy network
Build a toy network, step by step.
Here is the finished model — three buses, some generators, transmission and a city. Below, the same thing assembled one step at a time.
The finished model. Below, how it's built — one step at a time.
01
Draw a network of buses.
Reduce a whole region to a few buses — points where power is injected or withdrawn. Here, three: a generation-rich North, a South, and a City where demand sits.
02
Attach demand.
The City draws power on a daily load profile. Demand is what the whole system exists to serve, so it anchors every other decision.
03
Place the resources.
Wind and hydro at the North bus; solar and gas at the South; a battery near the City. Each is just a short list of numbers: capacity, cost, and — for wind and solar — an availability profile.
04
Wire the buses together.
Transmission corridors link the buses, each with a limited MW rating. Cheap Northern power only helps the City if a line can carry it — so topology shapes what the model can do.
05
Turn the question into a maths problem.
The step most explainers skip: write the whole thing as a Linear Program (LP). The unknowns are how much each unit generates and how much flows on each line; the goal is minimum total cost; the rules are all linear — so an LP solver finds the exact cheapest answer, fast.
Hand the LP to the solver and it returns the cheapest feasible dispatch — power flows, storage charges and discharges — plus the answers: system cost, emissions, and a build-and-run plan. Change a line limit or a fuel price and it all rearranges.
Least-cost dispatch
Results
System cost ↓
Emissions ↓
Build & run plan
Part B · real world ↔ model world
The same grid, told two ways.
A model is a deliberate abstraction. On the left is the world as it is; on the right, the world as the optimiser sees it — every physical asset collapsed into a node, an edge, and a short list of numbers.
Real world
Turbines, a dam, panels, a plant, wires, a city — messy, physical, continuous.
Model world
Three buses, three edges, a handful of numbers. Solvable, testable, reproducible.
Wind / solar farm→a generator with capacity, cost, and a time-varying availability profile
Hydro / gas plant→a dispatchable generator with capacity, fuel cost, ramp and emission factors
A city / region→a demand node with an hourly load profile
A transmission line→an edge with a capacity (MW) and losses
A battery→a storage node with energy (MWh), power (MW) and round-trip efficiency
The skill of modelling is choosing which of these numbers matter for the question — and which real-world detail you can safely throw away.
Part C · the distinction that matters
Capacity-expansion vs. operational models
Both are "energy models," but they answer different questions on different clocks. Confusing them is the most common mistake newcomers make.
Capacity expansion
What should we build?
Plans investment over decades: which technologies to build, when, and when to retire — minimising total investment plus operating cost across the whole horizon.
hydrowindsolarstorage
Operational · dispatch
How should we run it?
Takes the fleet as fixed and decides, hour by hour, which plants run — respecting start-up limits, ramp rates, reserves and storage cycling.
gasdemand← the sweeping line is the hour being solved
Capacity expansion plans all the way to 2050 in a handful of coarse slices. Pick a single transitional year — here 2035 — and the operational model dissects it into 8,760 hourly steps to check whether the plan can actually be run, hour by hour.
Dimension
Capacity expansion
Operational / dispatch
Question
What and when to build (and retire)?
How to run the existing fleet right now?
Time horizon
Years to decades (e.g. 2025–2050)
Hours to a year, solved chronologically
Time resolution
Coarse — representative time-slices or typical days
Fine — hourly or sub-hourly (8,760 h/yr)
Capacity
A decision variable — the model chooses it
Fixed input — given by the plan
Key detail kept
Investment economics, long-run trade-offs
Unit commitment, ramping, reserves, storage state
Typical tools
OSeMOSYS, TIMES, GenX, PyPSA*
Production-cost / UC models, PLEXOS, PyPSA*
Output
An investment pathway
An operating schedule & reliability check
*Some frameworks (e.g. PyPSA) can be configured for either role — the distinction is the question and resolution, not always the software.
So why bother linking the two?
Here's the catch. To plan decades ahead, the capacity-expansion model has to squeeze a whole year into a few representative slices — otherwise it would be extremely difficult and slow to solve. But that shortcut quietly flatters the plan: it hides the awkward hours, the fast ramps, the reserve shortfalls, the days when storage runs dry. So a build-out that looks beautifully cheap on the planning spreadsheet can turn out to be a nightmare to actually operate.
The way out is to make the two models talk to each other: build the fleet with the expansion model, then stress-test that exact fleet in a high-resolution operational model across all 8,760 hours, and feed what breaks back into the next planning run. Do that well and you get a plan that is both affordable and keeps the lights on.
That coupling — long-term investment on one side, hour-by-hour operational reality on the other — is exactly the problem I work on. I'm exploring the tools that let these two models actually talk to each other, and setting the rules that make their results agree rather than contradict. It's harder (and more interesting) than either model alone.
Milestones & news
Recent highlights.
Selected milestones, features and conversations from my work, research and life in energy systems.
Active model development
Currently developing and improving the following models
Where can wind and solar realistically be built in British Columbia?
This study maps where new wind and solar projects could realistically go in British Columbia.It removes places that are protected, environmentally sensitive, or unsuitable for development.It also checks whether promising locations are close enough to the electricity grid.The results show how land choices can change the amount of renewable energy available.The maps can help planners compare practical pathways toward a cleaner electricity system.
Learning to translate and communicate what my models say
Nancy Baron's Escape from the Ivory Tower is ostensibly about talking to journalists and policymakers, but it quietly rewired how I deliver modelling results. Her core point lands hard for a modeller: a result only its author can read changes nothing — you have to lead with the "so what" and translate rigour into something a decision-maker can actually act on. That is exactly why I build my models in the open, document them obsessively, and treat a clear figure as part of the science rather than an afterthought. The maths earns trust; the story earns action, and a planner's model is only ever as useful as the decision it helps someone make.
Staying honest about what a model can — and can't — tell you
If Baron taught me to communicate results, Erica Thompson's Escape from Model Land taught me to be honest about what those results actually are. "Model-land" is the seductive place where a model's tidy world gets mistaken for the real one — where a least-cost pathway on screen quietly hardens into a promise about the future. Thompson's argument reshaped how I caveat my own work: state the assumptions plainly, test how the conclusions break when the inputs shift, and treat a model as a structured argument rather than an oracle. Often the most useful thing I can hand a planner is the sentence that begins, "here is where this stops being trustworthy."
Modelling a more renewable and adaptable energy future
SFU’s Faculty of Graduate Studies profiled my path from power-utility planning into Sustainable Energy Engineering research. The conversation explores renewable-resource planning, open-source modelling, the value of research collaboration and the people who have supported my graduate journey.
BC Hydro
Economic and Social Commission for Asia and the Pacific · United Nations
Simon Fraser University · ΔE+ Research Lab
Bangladesh Power Development Board
Bangladesh India Friendship Power Company
BPDB · Teletalk · HRC Technologies