Vopak is the world largest independent tank storage provider, specialized in storing liquid chemicals, gasses & oil products and is active in 29 countries. For the storage of liquefied petroleum gas (LPG), a global strategy was constructed in 2010, in which LPG terminal opportunities have been identified and which is now out of date. The problem is delineated, arguing that it is essential for Vopak, as a leading terminal operator, to be able to continuously identify robust hotspots for LPG terminals in order to manage risk whilst growing strongly. It is concluded that consultant reports are not directly suitable for LPG terminal hotspot location and that the LPG market currently is going through a lot of change. Hotspots are defined as a opportunity for a LPG terminal in a certain point in time. LPG is produced as by-product at refineries and at oil and gas production sites and because LPG is a substitute product with multiple end uses. These characteristics make LPG terminal business cases less apparent due to a less structural supply-demand imbalance. It is underpinned that due to high capital expenditures associated with LPG terminals, understanding of LPG market dynamics is needed. The research objective is to develop and execute a strategy to systematically identify potential LPG terminal hotspots with both a base case forecast and with a deep understanding as to how opportunities can develop in different future scenarios in order to make robust investment decisions. In order to reach this objective the research question is defined as follows: “How can LPG market dynamics and terminal hotspots emergence be understood and described?” The overall methodology consists of a modelling strategy. A novel method, recently applied in other theses and under research at the TPM faculty of Delft University of Technology, Scenario Discovery, is the main research approach. Scenario Discovery assists in identifying relevant scenarios for decision makers by applying data mining- and statistical algorithms to large databases of simulation model results. The method has not been applied often to this type of problem in a business environment. The research strategy consists of the creation of a simulation set-up resulting from a literature review of theory, the creation of an LPG model based on what the driving forces in the LPG market are, available data sources and how LPG terminal hotspot conditions can be quantified. Together with the simulation set-up and the LPG model, scenarios can be generated, that can test identified LPG terminal hotspot locations identified in a base case forecast. Together with the base case forecast and the simulation results, conclusions of the research can be drawn. The scientific relevance of the research is to test the applicability of Scenario Discovery to the type of problem at hand, while the practical relevance of the research is a contribution to the 2014 LPG strategy at Vopak by means of supplying a list of LPG terminal hotspot locations. Limitations of the research are the inability to; identify all future states of the LPG market, to assess LPG terminal hub functions in a metric and to assess the effect of action made in the system. The theoretical context is further explored by means of making the problem characteristics explicit and reviewing different scenario schools. It is concluded that the intuitive logic scenario school is most appropriate for the research. Scenario Discovery belongs is appropriate for the research, but needs to be tailored. Methods and steps of Scenario Discovery are discussed, after which the tailored Scenario Discovery approach is presented. This approach takes identified LPG terminal hotspot locations from the base case forecast and tests them, one by one; using data generated by the scenario simulations and identified LPG terminal hotspot indicators. The LPG model is then conceptualized. Production and consumption variables that should be used in the LPG model are identified. Production variables are; associated petroleum gas (APG) flaring reduction, the shale gas effect, LPG from associated & non-associated gas source combined in one variable, and LPG from a refinery source. Consumption variables are; chemical consumption, possible propane dehydrogenation plant projects, domestic consumption, autogas consumption and industrial consumption. The LPG model is constructed in Microsoft Excel and generic rules for model creation as well as regions are established. The balancing mechanism in the model is assumed to be the proportionate distribution of a surplus in a simulation over the global chemical capacity. A number of validation and verification steps are presented that were executed during the research in order to validate and verify the LPG model. Then, the simulation set-up and scenario filtering steps are discussed. 50.000 simulations are executed using a Latin Hypercube sampling strategy. After filtering out ‘logical impossible’ scenarios, 11.800 scenarios remain for analysis. The logical impossible scenarios have to be filtered out due to the chosen modelling strategy and are done so by means of four rules that are the consequence of the modelling approach. The LPG terminal hotspot conditions are constructed and quantified. This is done by means of analysing the business model for an independent LPG terminal and with an expert opinion. The LPG terminal hotspot conditions are classified by 4 rules, representing basic terminal justification, need for new infrastructure, stability of the surplus/deficit and terminal justification on the long term. With use of the base case forecast, the hotspot location shortlist is constructed. The locations are divided in import and export opportunities and are ranked according to the severity of the opportunity with the LPG terminal hotspot conditions in mind. 14 import and 7 export hotspot locations are identified on a shortlist. These hotspot locations are then elaborated on with the base case story line. The locations are hereafter tested on their robustness and causality using the tailored Scenario Discovery approach per identified hotspot location. Then, different parts of the research are critically discussed. The research approach and used methodology are examined. After this, the results of the research are critically reflected upon. In addition, the societal, academic and practical relevance of the research are discussed. The conclusions of the research are drawn by means of answering the main and sub research questions. It is concluded that the hotspot classification rules are not black and white but the process of applying these rules is valuable. The full understanding of hotspot emergency is only partly understood, in the next step of the project management cycle, probabilities should be assigned to the found events. Next to this, it is concluded that the Scenario Discovery process is applicable next to the base case forecast and not separately due to communication strengths of base case results, which could have been more aggregated in this research. In addition, it is concluded that the methodological approach is appropriate and replicable to other products/systems provided that are substitute products, supply driven and require high infrastructure investments. It is then concluded that the chosen driving forces were appropriate, but that a shift in the global chemical consumption gravity point should be examined more closely. In addition, the possible breaking of a hotspot classification rule should be done by hand when examining a potential location. It is recommended for Vopak to examine the base case and Scenario Discovery results of this research first when embarking on a new study in a certain location. If the two rules representing infrastructure need are limiting, it should be checked if the LPG terminal can be combined with something else. If the rule representing new infrastructure need is limiting, an analysis on the competition should be executed. And when the stability of the surplus is limiting, the dialog with consumers that potentially can switch feedstock should be sought, and preferable be included in the business case with a long term contract in order to safeguard stability and prevent substitution. In addition, the LPG model can be updated with new figures. Further research can be done on the creation of a generic Scenario Discovery approach for business trade systems, the balancing mechanism of the LPG model, the scenario filtering process, and on more dynamic LPG market models.