Project library

Project evidence for research and industrial R&D.

Case studies grouped by method: high-temperature thermal-fluid validation, industrial decarbonization and electrification, energy/process modelling, experimental coursework, optimisation and reproducible engineering tools.

Case-study structure: technical question -> method -> evidence -> result -> limitations -> role relevance -> links

Energy management

Comparative Energy Audit Methods (consultancy)

Comparative regulatory analysis and tiered audit-method design across multiple EU, Asian and North American jurisdictions. Public-domain methodology summary only — proprietary site detail is not republished.

ISO 50001 / EU EED / Comparative regulation
Industrial energy KPI mapping visual
ISO 50001EU EEDEnergy auditsMulti-jurisdiction comparison
Multi-jurisdictionEU + Asia + North America compared against EU EED Tiered audit methodMatches audit depth to site EMS maturity Method-only summaryProprietary site detail withheld
  • Translated EU EED-aligned audit obligations into a generic tiered methodology.
  • Linked methodology components to EnPI, measurement and follow-up logic.
Energy systems

District Heating Dispatch Optimisation

Stepwise optimisation of a district heating system across base case, future demand, storage integration, cost minimisation and emissions minimisation scenarios.

PuLP / Python / MILP
District heating dispatch optimisation visual
PuLPPythonMILPDistrict heatingStorage
CHP flexible roleBalances demand across the year Woodchip + geothermal baseStable renewable heat supply Scenario comparisonCost and emissions trade-offs
  • Built a stepwise dispatch model covering base, future, storage, cost and emissions scenarios.
  • Used Python and PuLP to compare heat allocation under resource and demand constraints.
Grid modelling

Distribution Grid Study

CIGRE MV network-style power-flow, N-1 contingency and 24-hour simulation study with EV charging, PV and storage impact analysis.

Python / Power flow / N-1
Distribution grid power-flow and N-1 study visual
Power flowN-1EV/PV/storageCIGRE MV
  • Screened line loading, voltage behaviour and contingency sensitivity across the 24-hour demand profile.
  • Compared EV charging, PV injection and battery storage assumptions on network stress indicators.
Software / analytics

EU ETS Exposure Calculator

Excel and Streamlit toolkit for estimating Scope 1 emissions, EUA cost exposure and carbon-intensity trends for industrial sites.

Python / Streamlit / Excel
EU ETS emissions and allowance exposure dashboard visual
EU ETSStreamlitExcelCarbonPython
  • Calculates ETS scenario summaries and EUA price sensitivity from workbook inputs.
  • Includes workbook parsing, CLI runner, PDF export and automated regression tests.
Data / ML

Day-Ahead Heating Demand Forecasting – KTH U-Building

Logistic regression, shallow and deep neural network comparison for day-ahead space heating forecast using Mahalanobis-filtered hourly weather and temporal features.

Python / scikit-learn / Neural networks · KTH
Heating demand forecasting model visual
ForecastingPythonscikit-learnNeural networksMahalanobisBuilding energy
SNN: 93.39% accuracyBest generalisation — 1.02pp test–validation gap DNN: 10× slower7 s vs 0.69 s for only +0.33pp accuracy gain 4,992 rows after cleaningMahalanobis 3.5σ removed 120 outliers
  • Engineered hour-of-day, chronological day and weekday features to capture occupancy and seasonal patterns in KTH U-Building (~3,600 m²).
  • Two-stage pipeline: initial range-based cleaning then Mahalanobis multivariate outlier removal; MinMaxScaler normalisation.
  • SNN (1 hidden layer, 4 neurons) recommended over DNN (6 layers, 8 neurons) on accuracy–compute trade-off.
Energy management

Industrial Energy KPI Toolkit

Python, Excel and Streamlit toolkit for industrial EnPIs, baseline normalisation, anomaly detection and management reporting.

Python / Streamlit / ISO 50001
Industrial energy KPI and normalisation dashboard visual
Energy KPIsISO 50001EnPIReportingStreamlit
  • Generates KPI results, anomaly flags, action logs, plots and one-page PDF reports from meter data inputs.
  • Guided Streamlit flow with diagnostics, baseline normalisation and CI quality checks.
Energy systems

PyNEXUS Green Hydrogen Network Model

Coupled offshore wind, PEM electrolyser and hydrogen-pipeline model with MILP dispatch optimisation and scenario comparison.

Pyomo / HiGHS / MILP
Green hydrogen wind electrolyzer and pipeline model visual
HydrogenPyomoMILPDispatchElectrolyser
  • Models wind-to-hydrogen conversion, pipeline delivery and curtailment KPIs across 8760-hour horizon.
  • Compares cost and emissions objectives with HiGHS/Pyomo MILP optimisation and multi-scenario analysis.
Energy storage

TES Peak Shaving for District Cooling – Ice Storage vs PCM

Python TES simulation and techno-economic comparison of ice storage and salt hydrate PCM for a shopping centre district cooling network, including break-even and lifetime profit analysis.

Python / TES / PCM · KTH MJ2386
TES peak shaving district cooling simulation visual
TESPCMIce storagePeak shavingDistrict coolingPython
~£2,220/week savedShifting demand below 1,850 kW threshold Ice: 3.5yr payback$404k CAPEX — $1.6M lifetime profit PCM: 14.9yr payback$1.72M CAPEX — 4× higher cost for same capacity
  • Python simulation with 1,850 kW threshold, 12-hour look-ahead and dynamic SoC tracking across 8-day demand profile (198–3,391 kW range).
  • Sized both technologies to 14,750 kWh: ice storage at 351 tons / 320 m³ vs PCM at 246 tons / 273 m³.
  • Concluded ice storage optimal where space is unconstrained; PCM justified only when footprint is a binding constraint.
Energy systems / LEAP

Germany 3E Analysis – Electricity, Cost and CO&sub2; Modelling with LEAP

Scenario-based modelling of Germany’s electricity system to 2045 using LEAP, comparing NDC 2030 targets against nuclear recommissioning on cost and carbon emissions.

LEAP / Scenario analysis / KTH MJ2508
Germany electricity transition LEAP scenario model visual
LEAPScenario modellingNuclearGermanyCO2 emissionsSDG 7
2,488 million GJProjected 2045 demand (+42% from 2022) 14% nuclear shareOwn scenario 2045 vs. 0% at start +2.21% social costNuclear vs. NDC – marginal premium for near-zero CO2
  • Built demand model covering households, industry, transport electrification (EVs, rail) and commercial services to 2045.
  • Compared NDC 2030 pathway against recommissioning 12.1 GW of recently decommissioned nuclear capacity.
  • Contributed Sankey diagram, SDG 7.1 energy poverty analysis and sustainable development section.
Energy management / buildings

Hylkysaari Island – Smart and Sustainable Tourism Energy Design

Building energy simulation, microgrid modelling and economic evaluation for a sustainable tourist island near Helsinki, with a guest-facing energy dashboard business model.

IDA ICE / HOMER PRO / Excel · Valo Finland · KTH MJ2511
Hylkysaari sustainable island energy design visual
IDA ICEHOMER PROSolar PVWaste heat recoveryNPV/BCRBusiness model
3 scenariosBAU, solar PV + heat recovery, solar collectors Solar collectors rejectedDPP equals project lifetime Vihreä ProgramReal-time energy dashboard business model
  • Modelled building energy use in IDA ICE and microgrid in HOMER PRO across three design scenarios.
  • Economic evaluation (NPV, BCR, DPP) showed solar PV + improved insulation + waste heat recovery as optimal combination.
  • Designed guest-facing “Hylkysaari Vihreä Program” business model with real-time energy dashboard and carbon offset scheme.
Techno-economic analysis / solar thermal

Residential Heating Techno-Economics – PV, Heat Pump and Solar Thermal vs District Heating

Four-scenario comparison for replacing district heating at BRF Årstaterrassen (423 apartments, Stockholm) during summer: solar thermal collectors win with a 5-year payback and $128.6/MWh LCoE; PV + electric boiler eliminated at 14% demand coverage.

Solar thermal / Heat pump / SAM · KTH MJ2438
Residential heating techno-economic comparison visual
Solar thermalHeat pumpPVDistrict heatingTechno-economicsSAMLCoE
STC: PBP 5yr — LCoE $128.6/MWh$256k CAPEX, $776/yr OPEX — recommended solution PVEB: 14% demand onlyPV surplus 139.6 kWh vs 994 kWh nocturnal gap — infeasible STC winter contribution: 18.9%188.9 MWh/yr beyond the five-month summer target
  • Demand computed from 2,604.4 MWh/yr annual baseline, scaled to September worst-case of 2,000 kWh/day using district heating monthly demand data.
  • Wagner-Solar C20 AR-M collectors (85.4% optical efficiency, 2.61 m²): 309 units covering 806 m² generate 5,148.5 kWh/day; TES volume 2.95 m³ for nocturnal gap.
  • PV + heat pump (COP = 3) is the feasible electrical alternative: 649 panels, $449k CAPEX, 9-year payback — viable but inferior to STC on all KPIs.
Building energy / simulation

Built Environment Energy Simulation – IDA ICE Supply System Comparison

IDA ICE parametric study (heat exchanger, orientation, location) and 11-configuration supply system comparison on bought energy, primary energy and CO&sub2;, plus India built environment decarbonisation policy analysis.

IDA ICE / HVAC / Building energy · KTH MJ2509
Building energy simulation and supply system comparison visual
IDA ICEGSHPASHPPVBuilding energyHVACIndia policy
Heat exchanger: −26% energy160,733 → 118,890 kWh delivered; AHU −78% GSHP + PV: net export−25,819 kWh bought / −11,436 kg CO&sub2; District H&C: worst config148,461 kWh — 32% above reference
  • Parametric IDA ICE study isolating heat exchanger, window blinds, orientation and Barcelona relocation effects on a ~6% PPD baseline.
  • 11 supply configurations compared: oil boiler reference, GSHP variants, ASHP, district H&C, and GSHP+PV combinations.
  • India policy essay covering ECBC/NBC enforcement, LEED/GRIHA certification, vernacular architecture principles (Nalukettu, Rajasthan) and case studies (Infosys Mysore, Delhi Metro).
Polygeneration / off-grid

Emergency Energy Module – Post-Disaster Polygeneration, Philippines

Modular solar PV, biogas digester and battery system for 3,000 disaster survivors after Cyclone Haiyan, with HOMER PRO economic analysis and a PPP business model.

HOMER PRO / Biogas / Solar PV · KTH MJ2503
Emergency Energy Module polygeneration system visual
HOMER PROSolar PVBiogasBatteryOff-gridDisaster responsePolygeneration
NPC $444k — LCOE $0.063/kWh76% below digester-only baseline IRR 92.3% — 1.08yr paybackPPP-deployable economics 105% renewable coverage47.9 kWh/yr unmet load for 3,000 people
  • Sized solar PV (SunPower Maxeon 7, 24.1% eff, 315 m²), 920 m³ biogas digester and 50 kWh LiFePO4 battery within a 40-foot container constraint.
  • Digester produces 1,400 m³/day biogas covering 886 kWh/day electricity and cooking fuel for 3,000 people.
  • HOMER PRO economic analysis: $444k NPC, 93% OPEX reduction vs. conventional supply, 92.3% IRR.
Test / instrumentation

Rotating Machinery Vibration Mini-Lab

Low-cost vibration measurement workflow using phone accelerometer data, signal filtering, FFT/PSD analysis, spectrograms and automated report generation.

Python / FFT/PSD / SciPy
Rotating machinery vibration frequency analysis visual
VibrationFFT/PSDPythonSpectrogramsSciPy
  • Compares operating points through frequency fingerprints and RMS metrics from filtered accelerometer signals.
  • Packages outputs as HTML reports, plots, spectrograms and summary CSVs.
Energy storage / lab

M-TES PCM Thermal Lab – RT60 HC Discharge Characterisation

Experimental discharge measurement of a latent-heat Mobile TES prototype: 10-sensor temperature logging over 8 hours, three heat-transfer calculation methods, and 99.9% theory–experiment agreement.

PCM / Heat transfer / Lab · KTH MJ2386
M-TES PCM thermal lab visual
PCMLatent heatTESLab experimentPT100Heat transfer
Qtheoretical = 10,049 kJ86.9% from RT60 HC latent heat (8,729 kJ) Method 1: 10,036 kJ0.1% below theory with variable ρ and Cp Tank absorbs +2,250 kJ150 kg steel shell — largest unmodelled term
  • RT60 HC PCM (33 kg, 44 L) discharged from 71.1°C to 40.1°C via RENOLIN THERM HTF at 2 L/min; 10 PT100 sensors logged every 10 seconds for ~8 hours.
  • Three integration methods compared: variable-property (Method 1) and constant-property (Method 2) HTF calculations, with and without stainless-steel tank heat absorption.
  • Propagating thermal front identified in sensor pairs 40 minutes apart; outlet temperature profile confirms integrated heat extraction.
Battery characterisation / lab

Battery Cell Discharge Performance Testing – Ni-MH, Li-Po and Lead-Acid

Seven-test discharge characterisation across three chemistries (Ni-MH 780 mAh, Li-Po 420 mAh, Lead-acid 2500 mAh), three C-rates and three temperatures: voltage, SOC and power profiles with Peukert and thermal effects.

Electrochemistry / C-rate / Temperature · KTH MJ2386
Battery discharge characterisation lab visual
Ni-MHLi-PoLead-acidC-rateTemperatureBattery degradationElectrochemistry
2C vs 0.5C: Peukert lossPlateau collapses at 2C; effective capacity reduced by ohmic heating −10°C: severe capacity lossVoltage drops sharply from start; usable charge inaccessible under load Li-Po: 3,500 s vs Lead-acid: 500 s7× longer runtime; flat plateau vs linear lead-acid decay
  • C-rate tests (Tests 1.1–1.3): voltage plateau collapses at 2C; 0.5C maximises runtime and capacity utilisation across all four analysis dimensions.
  • Temperature tests (Tests 2.1–2.2): +40°C improves kinetics and lowers internal resistance; −10°C cuts accessible capacity and power delivery.
  • Chemistry tests (Tests 3.2–3.3): Li-Po holds flat ~4 V plateau until 30% SOC then drops sharply; Lead-acid declines linearly from 1.9 V; Ni-MH is predictable across all conditions.
Waste-to-energy / urban energy

Waste-to-Energy Proposals – Delhi and Kozhikode

Delhi MSW crisis policy roadmap (biomethanation → RDF → gasification) and a Kozhikode 11.11 MW WTE plant design coupling a grate incinerator with a 36,000 m³ biogas digester for 214 buses.

WTE / Biogas / Incineration · KTH
Waste-to-energy plant and MSW processing visual
Waste-to-energyBiogasIncinerationMSWIndiaAnaerobic digestion
Kozhikode: 11.11 MWGrate furnace + cold-condensing steam, 20% eff 32,200 m³/day biogas36,000 m³ digester → 214 biogas buses daily Delhi: 5,533 TPD landfilledPhased roadmap: biomethanation → RDF → gasification
  • Delhi proposal: phased technology roadmap matched to segregation reality — biomethanation first, combustion WTE only after 60–70% source segregation achieved.
  • Kozhikode: 600 TPD split 50/50 between grate incineration (11.11 MW) and anaerobic digestion (32,200 m³/day biogas, 30-day retention, 36,000 m³ digester).
Numerical methods / Python

Numerical Heat Transfer – Finite Difference, Time-Stepping and Advection Schemes

Python assignments covering catastrophic cancellation, numerical differentiation error analysis, 2D Laplace stencil comparison, Euler vs RK4 time-marching and Lax–Friedrichs vs Lax–Wendroff advection.

Python / FD / CFD methods · KTH
Numerical heat transfer and finite difference methods visual
PythonFinite differenceRK4Lax–WendroffNumerical methodsHeat conduction
RK4: 323 stepsvs Euler 742 — 56% fewer to converge Central diff: 36× betterMin error 1.72×10³ vs forward 6.23×10³ 9-pt stencil: 0.038 K errorvs 0.187 K for 5-pt at same Δx=1 cm
  • HA1: floating-point cancellation (100% error in single-precision for large-magnitude case), differentiation truncation–roundoff balance, 2D Laplace 5-pt vs 9-pt stencil grid convergence.
  • HA2: Euler vs RK4 stability limits and convergence; Lax–Friedrichs numerical diffusion vs Lax–Wendroff second-order accuracy for smooth wave advection.
Entrepreneurship / energy tech

GridFlex – ML-Driven Energy Storage Optimisation Concept

KTH entrepreneurship assignment: a predictive ML platform for real-time storage dispatch in renewable grids, addressing 5% curtailment and 30% storage waste — idea screen score 87/100 with full Business Model Canvas.

Business concept / ML / Energy storage · KTH
GridFlex energy storage optimisation concept visual
Machine learningEnergy storageGrid optimisationBusiness modelSaaSRenewable energy
Idea screen: 87 / 100Industry 19 · Market 18 · Idea 17 $120B storage market by 2030CAGR ~20% driven by renewable integration Addresses 5% curtailment + 30% storage wastePredictive ML dispatch vs static/scheduled systems
  • Three-iteration concept development via SCAMPER: rule-based → weather-forecast-driven → real-time ML with historical grid pattern learning.
  • Multi-storage compatibility (Li-ion, pumped hydro, thermal, flywheel) differentiates from Fluence/Stem focus on utility-scale batteries.
  • Business Model Canvas: SaaS subscription primary revenue, premium analytics add-on, one-time integration fees.
B.Tech final year / robotics

B.Tech Final Year Interactive Robot Project

Final-year interactive robot project combining IRJET control-system integration work with published frame and locomotion design: Raspberry Pi/cloud control, surveillance, SOC indication, steering integration and mechanical drivetrain sizing.

Raspberry Pi / CAD / ANSYS · IRJET 2021
Robotic frame and locomotion design visual
Raspberry PiCloud controlCADANSYSAckermann steeringBLDC motor
68 Nm rear-axle torque75 kg system at 30° inclination, 0.5 m/s² Two-stage gear reduction3,016 → 532 rpm via compound system Ackermann lock angles35.26° inner / 71.06° outer at R = 510 mm
  • IRJET DOCX manuscript covers robotic control-system and subsystem integration for the final-year project.
  • Control scope included cloud/voice control, surveillance, SOC indication, environmental sensing and servo steering integration.
  • Mechanical publication covers wheelbase stability, drivetrain torque sizing, compound gear reduction and ANSYS screening.