Designing a resilient relief supply network for natural disasters in West Java Indonesia using optimisation-via-simulation: Relief-OpS

In this project, we will develop methods that optimise the delivery of essential relief items following a natural disaster. These include optimising warehouse locations; inventory management strategies particularly for perishable relief items; and robust routing for post-disaster distribution. The optimisation will make use of an agent based simulation model that we will build with input from key decision-makers, designed to mimic the complexities of post-disaster relief efforts.

The Sendai Framework for Disaster Risk Reduction 2015-2030 puts a strong emphasis on Disaster Risk Management, acknowledging the importance of risk management and strengthening resilience against disasters. Based on the Sendai Framework, the Indonesian government produces a national plan for disaster risk management which highlights the needs for Indonesia to improve its preparedness and response to natural disasters. To execute the plan, Indonesia has established the national board for disaster management overseeing several regional boards for disaster management.

DRM activities span four phases: mitigation, preparedness, response, and recovery. Based on the Indonesian national plan, we focus our research on the preparedness phase (activities performed prior to disaster to allow more efficient response activities such as pre-positioning inventory of relief items) and the response phase (activities following a disaster to reduce its impact). The majority of research in DRM solves the preparedness and response phases of DRM separately, leading to suboptimal solutions. Furthermore, the solutions tend to be based on just one type of disaster. Another gap is that most research assumes that relief items are non-perishable, relief distribution is well coordinated, and demand and damages are known immediately after a disaster. Our focus on multi-disaster situations incorporating all of the complexities of multiple organisations providing relief support will result in more realistic models, filling the gaps in the literature in DRM and resulting in more impact. The research also extends previous Optimisation-via-Simulation methodology to multi-objective, multi-level problems.

In this project, we will work with West Java regional board from specifying the requirements for the decision support system to developing realistic scenarios to evaluate our methods. West Java province is chosen because it has the highest multi-disaster risk in Indonesia due to the population size, population density, high contribution to Indonesian GDP and being the centre of rice production in Java island (rice is the main staple food in Indonesia). As we will implement the national standard procedures for disaster preparedness and response, the decision support system can be customised by other regional boards to suit the characteristics of their regions by changing the parameters (for example, map, demography, infrastructure). Therefore, we will also organise a practitioner workshop to present our findings and to provide training for all regional boards to use the decision support system. The decision support system can also be used for training purposes which will strengthen the preparedness of the regional board officials.

Finally, this project will build capacity in this vital area of research. The research assistants in the UK and Indonesia will gain from regular project meetings, interactions with investigators and the scientific advisory board members, and relevant training in Southampton, Padjadjaran or other places in the UK and South East Asia. We will also organise a research workshop for Indonesian researchers who are interested in OvS, ABS and Operational Research in DRM

The proposed project will bring beneficial impact to a wide range of stakeholders that can be grouped into three categories: society, people and knowledge.

1. Impact to society

Our research will enable our model users in Indonesia (Regional Boards for Disaster Management, Regional Boards for Food Security, and Regional Development Planning Agencies) to make better decisions in reducing the risk of natural disasters by pre-positioning warehouses to store emergency food items and in the distribution of the emergency food items after a disaster strike. We will produce a decision support system that can be used to help them make those decisions. This capability will strengthen the Indonesian resilience and response to natural disasters. Hence, the society affected by natural disasters will benefit from better preparedness and response by the authorities. The optimum inventory management will reduce the operational cost of maintaining the perishable relief items which results in better value for money for the tax payers.

2. Impact to People

All investigators will benefit from receiving advice from the scientific advisory board members who are world leaders in their field.

This project provides valuable experience to the research assistants based in the UK (RAUK) and Indonesia (RAI). The RAUK and RAI will gain experience working with UK investigators. The RAUK will also be exposed to solving real-world challenges faced by people in Indonesia and learn from the local experts and model users about the challenges. Both RAs will learn about how to conduct research in a multi-disciplinary and multi-national setting. This includes research ethics and values such as openness, equality, mutual respect, and transparency. We expect that the RAI will pursue a PhD in a related topic upon the completion of this project. The two Indonesian named-researchers will develop their research skills by working with the more experienced investigators.

The project also provides knowledge transfer between investigators as well as between investigators and model users. The ODA investigators will learn about optimisation-via-simulation (OvS), agent-based simulation (ABS) and the research environment in the UK. The UK investigators will learn about local context and new applications of OvS and ABS. The investigators will learn about the challenges in implementing policies and responding to natural disasters from model users. Model users will learn about scientific methods that can help them make better decisions and understand the limitations of the methods.

Finally, the project will provide capacity building for ODA academics and model users on OvS, ABS and their application to disaster risk management through the symposium and workshop organised as part of this project.

3. Impact to Knowledge

Our research will develop an OvS methodology that is suitable for the complex multi-objective, multi-level stochastic optimisation problem of an integrated stochastic disaster risk management, which incorporates a complex ABS of relief distribution. The OvS methodology will also be applicable more widely to other complex stochastic problems. Through publications in academic journals and presentations at conference attended by users (e.g. Winter Simulation Conference and the OR Society conference) we will disseminate these methods to a receptive academic and user community.

The ABS will advance the field of relief distribution modelling by addressing the gap in the literature on relief distribution modelling in which most models assume that the operation is perfectly coordinated, and demand and damages are known immediately post-disaster.

The generic part of our OvS and ABS will be implemented as Python libraries and made available to the public. This will allow other researchers to evaluate, improve or extend our methods and will make them available to countries facing similar threats to Indonesia.