(496c) Optimizing the Sustainable Energy Transition for Small Island Developing States Using Multiple Decision Criteria

Trinidad and Tobago is one of the largest per capita emitters of CO₂ globally. A significant amount of these emissions are attributed to the power and petrochemical sectors. The country is rare amongst small island developing states as it is abundant in oil and gas and thus 98% of the country’s power generation is natural gas based. With these abundant resources, the country’s power costs are amongst the lowest in the region, with these low prices linked to subsidies. Trinidad and Tobago has signed agreements like Paris Agreement, which require a reduction in greenhouse gas emissions of 15% by 2030. The country requires a holistic approach to aid its sustainable energy transition. Our study implements mixed integer linear programming (MILP) to model 5 scenarios related to the national power generation system via the General Algebraic Modeling System (GAMS). MILP is commonly used in supply chain optimisation problems across various areas and this framework is tailored to resource availability, local infrastructure, and energy system capability of a SIDS. This is a key novelty of the study as the majority of published work in this field relates to developed nations. The basis of our modelling framework is to examine the efficiency of the current national power generation system using MILP. In accordance with current decision-making, the 2019 power system outlook used a single objective optimization (SOO), whereby levelized cost of electricity (LCOE) was minimized, in line with current national perspective on power generation. To further investigate the effects of process and feedstock constraints on both economic (LCOE) and environmental impacts, based on life-cycle greenhouse gas emissions (cradle-to-power) (GHG_LC) in 2019 and 2030, a multi-objective optimization approach (MOO) was used. The model involves the use of binary integer variables to represent each power plant, thus allowing the model the opportunity to select which ones can be used. Each plant was constrained to a minimum capacity factor. The key variables used included the energy required by each plant for both peak and normal times (assumed to be 12 hours each), along with the natural gas utilized. These variables link directly with costs and emissions. The system was constrained to meet the national electricity demand of 8.87 TWh (2019), with a daily peak load of 1350 MW.

The scenarios examined include a Business-as-Usual case (BAU), which is split into two parts; one including the present power purchase agreement (PPA) implemented, while the second examines the system without the restriction of this agreement, thus allowing the model to choose which power plants can be operated. Scenarios 2 – 5 also allows the model to choose which power plants are used. Scenario 2 examines the upgrade of a large single cycle (SC) power plant to combined cycle power plant, while Scenarios 3 – 5 considers the inclusion of polymer electrolytic membrane produced (PEM) Hydrogen (H₂) and downstream products in the form of green methanol (MeOH) and ammonia (NH₃) from the electrolytic H₂.

Scenario 1B results in an approximate LCOE reduction of $_USD 18 per MWh (29% decrease) compared to Scenario 1A ($_USD 61.72 per MWh). Elimination of the PPA allows the grid demand to be satisfied by only three power plants. The use of more efficient power plants also resulted in a reduction of GHG_LC from 756 kgCO_2eq to 743 kgCO_2eq.

Benefits in both economic and environmental sustainability can be noted after upgrading Plant A (Scenarios 2 – 5) from single cycle to combined cycle. Scenario 2 resulted in a reduction of LCOE of approximately $_USD 25 per MWh (40% decrease) and GHG_LC decreased to 572 kgCO_2eq (24% reduction) compared to Scenario 1A. These improved results were as a result of the improved efficiency due to technological improvements on the power grid.

Scenarios 3-5 demonstrated a similar LCOE reduction as Scenario 2, ranging from $_USD 36 – 37 per MWh. However, these cases required improved commodity prices (H₂, MeOH and NH₃) for the production of downstream products to be considered feasible in this system. Also, GHG_LC increased by 8 – 12% when compared to Scenario 2. Thus, it can be noted that Scenario 2 is the best performing case for both economic and environmental in assessment despite the use of single objective optimization (SOO). Thus, our results illustrate that efficient power generation processes takes precedence over downstream diversification in effectively reducing both energy costs and carbon footprint.

The MOO analysis for 2030 allows for insight into the current power system’s outlook and assessment with respect to the national goals. Due to the projected increase in grid demand and peak loads, four power plants were online across all scenarios. The results for 2030 followed a similar trend to 2019. However, there was an approximate increase in LCOE of $_USD (8.5 – 14) per MWh compared to the equivalent scenarios in 2019 SOO. The GHG_LC increased by approximately (8 – 27) kgCO_2eq, when compared to 2019 equivalent, with Scenario 3 being an exception. Scenario 3 demonstrated a reduction in the GHG_LC of approximately 39 kgCO_2eq when comparing 2030 to 2019. This was due to a reduction in electrolyser capacity, as a result of grid constraints. Overall, our results indicate an estimated increase of 29% and 5% for LCOE and GHG_LC, respectively compared to the equivalent scenarios in 2019. These results suggest Scenario 2 as the most sustainable pathway to clean and affordable energy. Despite the increase in GHG_LC impact from 2019 to 2030, the upgrade of Plant A to combined cycle (CC) operations provides a 21% improvement in GHG emissions when compared to BAU operations (Scenario 1A). Thus, the MOO 2030 outlook underlines the importance of increasing efficiency across the grid and utilizing accurate price forecasts to guide decision making.

The system’s sensitivity to natural gas prices, commodity (H₂, MeOH and NH₃) prices and PEM efficiency was examined. All cases are impacted by adjustments in natural gas price as Trinidad and Tobago’s power generation system is exclusively linked to natural gas utilization. The sensitivity to natural gas price is particularly volatile given the global shift away from coal as an energy source. When natural gas prices are set to $_USD 4.00 per MMBtu (57% higher than current prices), LCOE increases over the range of $_USD 11 – $_USD 23 per MWh across Scenarios 1A, 1B and 2. The highest impact was noted in Scenario 1A, which was constrained by the PPA. An increase in natural gas price would increase the country's reliance on its subsidies to maintain its low electricity rates, which is unsustainable.

Our results also highlight potential economic benefits among commodity-based scenarios, specifically when natural gas prices are reduced. For example, for natural gas prices ranging $_USD 1 – 1.5 per MMBtu, there was a 30-43% reduction in LCOE across Scenarios 3, 4 and 5. However, at natural gas prices above $_USD 2.60 per MMBtu, production of H₂, MeOH and NH­₃ were deemed infeasible, as the model chooses to neglect electrolysis- mainly due to high capital charges. Regardless of the natural gas price forecast, increased costs only impact the LCOE, thus leading to potential conflicts within our MOO model. Hence, our results illustrate the importance of using multiple decision criteria when proposing future energy policies that align to both economic and environmental sustainability. It must also be noted that the viability of utilising commodity-based scenarios to reduce LCOE is dependent on an increase in selling price of the products (H₂, MeOH and NH₃).

Improved efficiency of PEM in the system demonstrated minimal benefits to both LCOE and GHG_LC, thus highlighting that the downstream production of electrolytic H₂, MeOH and NH₃ does not necessarily support national GHG reductions targets.

Overall, our results support the need for an upgrade of the current power system, driving down susceptibility to increasing natural gas prices and promoting higher energy and resource efficiencies through technological advancement. This must be coupled with new policies to reduce GHG emissions and incentivize sustainable energy sources. A reconsideration of the present PPA must take place, as well as introduction of carbon taxes and emissions trading. The use of carbon capture and sequestration processes (CCS) should be explored. The introduction of feed-in tariffs to promote alternative energy sources in the power generation system can also be beneficial.

This work highlights the use of multiple decision criteria, through multi-objective optimization, and its relevance in promoting clean and affordable energy among SIDS, using Trinidad and Tobago as a case study. This work demonstrates the ability to link energy supply chains for less developed countries in the Caribbean and globally, thus linking to sustainable development. Our results illustrate that the most optimal scenario would be the improvement of gas turbine technology from SC to CC for the nation’s largest power plant. Our results underpin viable solutions-driving impact and distributing guidance on the sustainable energy transition across the Caribbean Region, through the deployment of clean and affordable energy.