This is an excellent and comprehensive project. Based on your detailed M.
Tech report and PowerPoint
presentation, here is a complete, detailed viva script designed for a 20-25 minute defense.
I have structured this script slide-by-slide, anticipating the questions from an external examiner, your
guide, and the HOD. The script includes what you should say and potential Q&A.
Before You Begin: Key Advice
• Be the Expert: You have spent months on this. You know more about this specific model than
anyone in the room.
• Acknowledge Limitations: The examiners will be impressed if you openly state the limitations
(e.g., "Coastal shipping was excluded") and explain why.
• Use Your Pointer: Don't just read the slide. Point to the convergence graph, the congestion
heatmap, or the Pareto front as you explain it.
• Dress & Demeanor: Be professional, confident, and respectful. Make eye contact with all
examiners.
Part 1: Opening & Introduction (Slides 1-6)
Script:
(Slide 1) Good morning/afternoon, respected external examiner, Dr. Sudhakar sir, Dr. Swati madam, and
all faculty members. I am Priyanka Krishna Gharat, and I welcome you to my final viva presentation for
my [Link] in Transportation Engineering and Planning.
(Slide 2) My project is titled "Optimizing the Sustainable Multimodal Freight Transport and Logistics
System in Navi Mumbai using MATLAB Software," and I completed this under the guidance of Dr. S.
Sudhakar.
(Slide 3) The core problem is that Navi Mumbai, being the primary hinterland for JNPT – India's largest
container port – moves 70-80% of its freight by road. This creates immense congestion, high costs, and
pollution. Sustainable modes like rail exist but are underutilized. My research aims to fix this
coordination gap using data-driven optimization.
(Slide 4 - Contents) I will walk you through the Introduction, Literature Review, identified Research
Gaps, my Aim & Objectives, Methodology, the detailed Study Area and Data, Results, and finally my
Conclusions and Recommendations.
(Slide 5 & 6 - Introduction & Problem) As we see here, Navi Mumbai’s freight demand is exploding
due to JNPT expansion and the new NMIA airport. This slide highlights the critical planning gap: rising
freight demand versus a highly inefficient, road-only system. Therefore, my study applies a Genetic
Algorithm to design a sustainable system that reduces cost, time, congestion, and emissions.
Potential Question & Answer (Q&A) here:
Q: Why did you specifically choose Navi Mumbai over other port cities like Chennai or Mundra?
A: That's an excellent question, sir/madam. Navi Mumbai is unique because it is the primary gateway for
India's container traffic via JNPT. Simultaneously, it has the newly operational NMIA, major industrial
clusters like Taloja and Rabale, and an existing but underperforming rail network. This convergence of
road, rail, air, and sea – all within a constrained geography – makes it an ideal 'living lab' for sustainable
multimodal optimization. Other ports may not have this immediate multimodal complexity.
Part 2: Literature, Gaps & Objectives (Slides 10-19)
Script:
(Slide 10-15 - Literature Review) I conducted a systematic review of 10+ key papers. Studies by Okyere
(2022) and Zhao (2018) established that Genetic Algorithms are very effective for this problem. More
recent work by Chupin (2025) and Liu (2025) included sustainability goals. However, you'll notice a
pattern.
(Slide 16 - Inference) Almost all existing studies use simplified, theoretical networks. They lack real-
world congestion data, port delays, and crucially, integrate all three modes – road, rail, and air – into one
framework.
(Slide 17 - Research Gaps) This led me to my specific research gaps. No India-specific study existed
for the complex JNPA-Navi Mumbai region. Real-world constraints like congestion, post-COVID freight
patterns (from 2023 onwards), and Zero Emission Truck (ZET) technology were completely missing
from optimization models.
(Slide 18 - Aim & Objectives) Therefore, my aim is to develop an optimized sustainable multimodal
system using a Genetic Algorithm in MATLAB to reduce cost, time, and emissions. My three core
objectives were:
1. To analyze the existing system using real 2023-2026 data.
2. To develop the GA optimization model in MATLAB.
3. To evaluate three scenarios and provide actionable recommendations.
(Slide 19 - Scope) The scope was clearly defined: geographically Navi Mumbai to major ICDs like Pune,
Delhi, and Bangalore. Temporally, I used 38 months of data from Jan 2023 to Feb 2026. I modeled road,
rail, and air, and evaluated Base Case, Sustainability (75% rail), and ZET 2028 (30% ZET
adoption) scenarios.
Potential Q&A:
Q: Why did you choose a Genetic Algorithm over other methods like Ant Colony or Particle Swarm
Optimization?
A: *Sir, GAs are particularly well-suited for combinatorial problems like route and mode selection
because they excel at global search and avoid getting stuck in local optima. They also handle multi-
objective optimization (cost, time, emissions) very naturally through a weighted fitness function. While
ACO is good for path-building, GA's population-based approach allows it to explore a much wider
solution space simultaneously, which is critical given our 29 nodes and 60+ OD pairs.*
Part 3: Methodology, Study Area & Data (Slides 20-39)
Script:
(Slide 20-21 - Methodology Flowchart) My methodology had seven phases. It starts with primary data
from JNPA and NLDS, then moves to network modeling, GA design, scenario definition, and finally
MATLAB implementation.
(Slide 22-24 - Study Area Map) Geographically, as this map shows, my study area is the entire MMR,
with Navi Mumbai at the center, connected to JNPA in the south and NMIA near Panvel. (Point to the
map).
(Slide 26 - Internal Connectivity) Internally, I mapped all key road and rail corridors, like the Sion-
Panvel Highway and the Harbour Rail Line.
(Slides 27-31 - Key Nodes) My network model has 29 nodes:
• 6 JNPA Terminals (BMCTPL, APMT, NSICT, etc.)
• 8 CFS Clusters around the port area.
• 10 ICDs across India, including Pune, Nagpur, Ahmedabad, and Delhi.
• 4 Industrial Areas (Taloja, Patalganga, etc.)
• And the NMIA airport, which became operational in Dec 2025.
(Slide 32 - Data Framework) The heart of this project is this multi-source data. I didn't use synthetic
data. I collected:
• 35 months of TEU data (Jan 2023-Feb 2026) from JNPA reports.
• 25 months of operational metrics (dwell time, transit time) from the National Logistics Data
Services (NLDS).
• Cost parameters from KPMG and JNPA.
• Emission factors from the TEMT Tool and GLEC Framework.
(Slide 37 - Cost & Emissions) For example, this table shows that rail is 40% cheaper per tonne-km than
road, and electric rail reduces emissions by 99% compared to diesel trucks. This data directly feeds into
my GA's fitness function.
(Slide 39 - Key Challenges) The data confirmed the core challenges: a severe modal imbalance (70-
80% road), operational inefficiency with high dwell times, and infrastructure gaps. This validated the
need for my optimization model.
Potential Q&A:
Q: You mentioned using NLDS data. What were the biggest data quality issues you faced, and how
did you preprocess them?
A: *Sir, the NLDS data was excellent, but it had gaps. For example, some CFS monthly reports were
missing for Sep-Oct 2025. I handled this using linear interpolation for those months. Also, dwell time data
had outliers due to one-off events like the Gulf crisis in March 2026. I used a z-score method (removing
points with z > 3) to filter those outliers before calculating my average parameters, ensuring my model
represents typical, not crisis, operations.*
Q: Why did you exclude coastal shipping?
A: *That is a key limitation, madam. I excluded it because of low current utilization and limited reliable
data for the Navi Mumbai region specifically. However, as I mention in my Future Scope, integrating
coastal shipping is a direct next step for this research, as its emission factor (0.015 kg/tkm) is even lower
than rail.*
Part 4: MATLAB & Results (Slides 40-50) - MOST IMPORTANT
Script:
(Slide 40 - Results & Discussion) I will now present the core results from my MATLAB implementation,
focusing on GA convergence, scenario comparison, the critical distance-based analysis, and congestion
hotspots.
(Slide 41 - Three Scenarios) I ran the GA for three scenarios: Base Case (current
operations), Sustainability (75% rail shift), and ZET 2028 (30% Zero Emission Trucks). The goal was to
see the optimization potential for each.
(Slide 42 - GA Convergence) This figure is crucial. It shows the GA's fitness value over 50 generations.
As you can see, all three scenarios converged to a stable solution within 50 generations. The Base Case
converged fastest at generation 28 (fewer options), while the Sustainability scenario took longer (42
generations) because the GA had to work harder to find feasible, low-emission routes combining road and
rail. This proves my GA is stable and effective.
(Slide 43 - Scenario Comparison) At first glance, the aggregate improvements seem small – only a
0.04% reduction in cost and emissions for ZET 2028. But look at the modal share. ZET 2028 achieves a
tri-modal split (40% road / 30% rail / 30% ZET) . The marginal aggregate benefit tells us that single
interventions are not enough. The real gain is in diversification, and a combined ZET + rail strategy
would be much more powerful.
(Slide 44 - Top Performing Routes) This table is where we find actionable insights. The NSDT to Pune
short-haul route (120 km) shows a massive 4.44% cost reduction – that's the low-hanging fruit. For
ZETs, the NSICT to Ahmedabad corridor (530 km) shows a 2.18% cost saving, making it the ideal
pilot route.
(Slide 45 - The ZET "Sweet Spot") This is my single most important finding. I plotted cost and
emission improvement against distance. The 500-800 km range delivers the highest emission reduction at
0.31%. Below 500 km, the benefits are marginal. Beyond 800 km, ZET performance turns
negative due to battery weight and charging constraints. This is the ZET "Sweet Spot."
(Slide 46 - Terminal Performance) The ZET potential also varies by terminal. NSICT (0.58% avg.
improvement) and APMT (0.47%) are leading candidates, while NSDT is not. This suggests infrastructure
readiness differs by terminal.
(Slide 47 - Pareto Front) For decision-makers, I generated Pareto fronts. For the JNPA-Pune corridor,
you can see the trade-off. To go from minimum cost to minimum emissions, you pay an extra ₹0.32
million to save 0.05 tonnes of CO2. The marginal cost of abatement rises from ₹5M/tonne to
₹8.2M/tonne. This allows a policymaker to choose based on their budget and environmental priority.
(Slide 48 - Congestion Hotspots) Finally, I mapped congestion. The export cycle is significantly worse
than import. C1 (JNPA port access) and C7 (Thane-Belapur Road) are critical with congestion levels
of 3.0 in both cycles. C3 (NH-48/Sonari) is also critical. These are where infrastructure investment must
be prioritized immediately.
(Slide 49 - Sensitivity Analysis) Two key sensitivities. First, rail capacity shows diminishing returns
beyond 30 rakes/day. Investing beyond that gives little benefit unless you also improve terminal
efficiency. Second, ZET adoption has a near-linear relationship with emission reduction – each 10%
adoption gives ~0.1% reduction.
(Slide 50 - Key Takeaway) In summary, the ZET sweet spot is 500-800 km, we have 3 critical congestion
clusters, optimal rail capacity is 30 rakes/day, and NMIA integration offers a massive new opportunity for
high-value cargo.
Potential Q&A (Be ready for deep technical questions):
Q: Explain your fitness function in detail. How did you normalize cost, time, and emissions, and
why did you use a weighted sum instead of a true Pareto ranking like NSGA-II?
A: *Sir, my fitness function was F = w1*Cost_norm + w2*Time_norm + w3*Emissions_norm. I
normalized each objective by dividing by a reference value (Cost_norm = Cost/₹10M, Time_norm =
Time/100hrs, Emissions_norm = Emissions/200,000kg) to bring them to a comparable 0-10 scale. I used
a weighted sum because it is computationally efficient and allowed me to easily define scenario priorities
(e.g., w3=0.5 for Sustainability). However, you are right, NSGA-II would provide a richer Pareto front. I
mention this in my future scope as a methodological enhancement. For this study, the weighted sum was
sufficient to identify the optimal distance ranges and top corridors.*
Q: Your 500-800 km ZET sweet spot is very specific. What physical or operational constraint causes
the negative returns beyond 800 km?
A: *Excellent question, madam. Based on literature (Nykvist & Olsson, 2023) and my model
assumptions, the primary driver is the battery weight penalty. For a 40-tonne truck, beyond 800 km, the
battery pack becomes so heavy that it displaces cargo weight, reducing revenue per trip. Also, the lack of
charging infrastructure on long-haul corridors forces detours or long idle charging times, which my model
penalizes via the 'time' objective. Under 800 km, a single depot charge is sufficient. Over 800 km, you
need en-route fast charging, which isn't available yet in India, making the total operational time and cost
higher.*
Q: How did you model the uncertainty of NMIA, given it just became operational in Dec 2025?
A: *Sir, that's a key limitation. I used secondary data from announced plans – CIDCO reports, FedEx
announcements, and AERA tariff orders for 2025. For my 2023-2026 data period, NMIA's traffic was
projected based on its Phase 1 capacity of 0.5 MMTPA. In my model, I used the published road distances
(14.5 km from JNPA) and standard air emission factors. I acknowledge that real operational data for
NMIA does not exist for my study period, and my results for air-related routes should be seen as a
forward-looking potential analysis, not a validation of current operations.*
Part 5: Conclusions & Recommendations (Slides 51-60)
Script:
(Slide 51) My conclusions are directly linked to my objectives.
(Slide 52) I successfully developed a GA-based model in MATLAB that evaluated 60+ OD pairs using 28
data categories from 2023-2026.
(Slide 53 - Key Scenario Findings) The key findings are summarized here. Sustainability rail share of
75% is achievable, aligning with Gati Shakti. The ZET sweet spot is 500-800 km.
The NSICT→Ahmedabad corridor is the best for ZET deployment (2.18% savings),
and NSDT→Pune is best for short-haul cost reduction (4.44%).
(Slide 55 - ZET Performance) To reiterate, ZET deployment must be distance-based. Don't put them on
short hauls (<300 km) or very long hauls (>1200 km). Target the 500-800 km sweet spot.
(Slide 57 - Congestion) On congestion, the export cycle is in crisis. I've identified three immediate
priority clusters: C1 (JNPA), C3 (Sonari/NH-48), and C7 (Patilpada/Thane-Belapur Road).
(Slide 58 & 59 - Recommendations) Based on these findings, I have a clear, phased action plan for three
stakeholder groups.
• For Infrastructure Planners: Prioritize decongesting C1, C3, C7. Develop the ZET charging
corridor on NSICT→Ahmedabad by 2027. Expand rail capacity to 30 rakes/day.
• For Policy Makers: Phase 1 (2026-2028) – Mandate ZETs on 500-800 km corridors with higher
subsidies. Phase 2 (2028-2030) – Extend to 800-1200 km.
• For Logistics Operators: Pilot ZET immediately on NSICT→Ahmedabad to capture the 2.18%
cost saving. Develop hybrid rail-ZET for long hauls.
(Slide 60 - Final Statement) This research proves that sustainable multimodal optimization is not just a
theoretical concept but is technically feasible and practically actionable for Navi Mumbai. The GA
framework quantifies the trade-offs and provides a roadmap.
Potential Q&A:
Q: What is the single most actionable recommendation from your thesis for the JNPA authority?
A: Madam, if I had to give one, it would be to immediately pilot the NSICT→Ahmedabad corridor for
ZET deployment while simultaneously investing in decongesting the C1 port access node. The ZET pilot
gives a quick, measurable 2.18% cost benefit, proving the technology. Simultaneously, decongesting C1 is
critical because if trucks (ZET or diesel) can't get in and out of the port efficiently, no amount of route
optimization will fix the systemic problem. These two actions – a technology pilot and an infrastructure
fix – must happen together.
Q: How can your model be used in real-time by a logistics operator today?
A: *Sir, in its current form, it's a strategic planning tool. An operator could input an origin, destination,
and cargo volume, and the model would recommend the optimal mode and route based on historical data.
For true real-time use, I would need to integrate live GPS traffic data, port queue lengths, and dynamic
pricing. That is a clear future scope: integrating reinforcement learning with my GA, as proposed by Zhu
& Li in 2023, to create an adaptive system. My model provides the core optimization engine for that
future digital twin.*
Q: You have many strong conclusions. If you had to redo your thesis with unlimited time and
budget, what one thing would you add or change?
A: *That's a great question. I would add a full life-cycle assessment (LCA) of the ZETs. My model
currently uses operational emissions. I would include the 'well-to-wheel' or even 'cradle-to-grave'
emissions – meaning the CO2 cost of manufacturing the batteries, generating the electricity (grid mix),
and recycling the truck at end-of-life. This would give a truly holistic sustainability score. Some studies
show that battery manufacturing is carbon-intensive, and this could slightly shift the optimal distance
range.*
Closing the Viva
Script:
(Slide 61 - References) I would like to thank my guide, Dr. S. Sudhakar, and the department for their
support. I have also published a paper from this work at the ICACE-2026 conference.
(Slide 64 - THANK YOU) I am now ready to answer any further questions you may have. Thank you for
your time and consideration.
(Be silent and wait for their response. Do not fill the silence.)