Improving risk identification on large infrastructure projects

Due to government austerity, governmental agencies are downsizing and the contractors in the construction industry are increasingly responsible for more phases and tasks within construction projects. Not only have they become responsible for the construction phase itself, but also for design, maintenance and in some cases even finance. Furthermore, they are required to provide complete services, which means the combining of different disciplines in a single contract. These increased responsibilities also lead to an increase in risks for which the contractor is responsible. Together with the pressure on turnover in the infrastructure sector specifically, these cause contractors to improve their project management expertise and all relating processes. Recent examples in the industry illustrate the need for a high level of expertise of risk management. These developments led to the goal of VolkerInfra to improve their risk management expertise and knowledge. The research is based around six case studies. These concern six large infrastructure projects of VolkerInfra, of which two are still in construction and four have been recently completed. This approach was chosen because of to the strong project focus of VolkerInfra. To measure success on these projects, usually a set of objectives is developed that the project hopes to accomplish. Part of these objectives are the traditional operational objectives of cost, time and quality. Of these three, cost is the most obvious and consistent measure of project success and therefore this measure is chosen as initial assessment parameter for the effectiveness of risk management. However, an initial review of the cases revealed that there is limited data available on the cases. The data that is available on all cases however, is the amount of identified risks, in both the tender phase and the execution phases of the cases. These revealed that most risks are identified after tender, while ideally they should be identified during the tender phase to prevent surprises when going down the project life-cycle. Due to the limited available other data, a further analysis of cost data was only limited possible. Furthermore, a literature revealed that there was limited attention for the risk identification part of the risk management process. For these reasons it was decided to focus on risk identification. This led to the following objective in the research: "Develop recommendations for the further development of the risk identification process and identify and classify top risks on VolkerInfra projects to enable generic oversight to assist in risk identification on future VolkerInfra projects.”. Interviews were held with the six risk managers of the six cases to determine their preferred risk identification approaches. These interviews revealed that each risk manager has their own preferred approach to risk identification and that this approach developed little over time. It also revealed that there is no company guideline on how to perform risk identification. The approaches of the risk managers were divided in the tender phase and the execution phase. During the tender phase, personal interviews, brainstorm sessions and work groups led by the risk manager were the preferred methods for risk identification. During the execution phase, only the personal interviews were preferred, while few other methods were applied. The literature review also revealed that there are multiple other tools that can be developed to help during the risk identification. These are mainly historic records and checklists, both of which were little used by the risk managers. Historic records are a valuable source of information for risk identification. These historic records are currently unavailable and therefore the goal was to develop the starting point for a database of historic records. The focus lies on occurred risks for the starting point, as these are considered to contain the most valuable information to enable reflective learning from these records. These are however only available on three of the six cases. To be able to maintain an overview of these records, a classification has been developed. This classification consists of a number of criteria, each with a number of underlying categories. Each risk is assigned a category per criterion. The classification will then enable oversight over the projects of the most prevalent classes of risk and thus the identification of the most important points of improvement. Due to the time consuming process of categorizing risks, only the most important occurred risks have been classified. The top fifteen (based on the calculated value, which is an estimate of the financial consequences) occurred risks of the three available cases is classified, giving a total dataset of 45 risks. The most important criteria are the phase identified, nature of the risk and the source of origin criteria. The classification revealed that 27 of the 45 risks was identified after tender. However, further analysis revealed that 15 of these 45 are due to an insufficient risk identification process. The other 12 were attributable to unforeseen scope changes, overarching risks that were identified sooner (so called container risks) and a force majeure and therefore not culpable to a faulty risk identification process. Further it revealed that most risks are of a technical nature, with process and managerial having a lot less risks in the top fifteen occurred. The source of origin criterion revealed that there are a number of risks present due to opportunistic behavior in the form of commercial decisions and a number of risks occurred as a consequence of another risk occurring. Based upon this research it can be concluded that there is a lack of guidelines for risk identification. Furthermore, a lack of data prevents further analysis of the effectiveness of the risk management process. There is also some ambiguity regarding the definition of certain concepts relating to risk management. Based upon this research, ten recommendations are being made to VolkerInfra to develop their risk identification process. First of all it is recommended that more data is recorded of risks, this data should at least consist of: the occurrence of the risk, cost data relating to the risk and a classification in the following criteria: main- and sub-object, nature of the risk and source of origin. When this data is recorded, this will allow for benchmarking of future projects and the setting of goals for improvement. Furthermore, relevant concepts and the exact purpose of risk management have to be clearly defined in order to prevent confusion and differences of approaches between the different risk managers. Fourthly, opportunism due to commercial decision should be prevented and be included as opportunities and not as risks. Due to the revealed link between different occurring top risks, a conditional probability class should be added to the RISMAN categories to define this relation. Guidelines for the risk identification process are formulated in order to develop and improve that process. This guideline makes use of all information gathered from the literature and empirical research parts and proposes the following sequence: Start by collecting all relevant documents and data, this includes tender documents, flowcharts and breakdown structures. If the first recommendation is followed, more data is recorded on the projects that can be used as historic records on future projects, which should be the second step. Then a combination of identification methods should be applied to prevent bias from either ones. This should be followed by a check for opportunism, check for conditional probabilities and then a full risk assessment of the consequences and probabilities. These can then be mutually compared to come to the final risk database for tender. This is a structured approach that is recommended to VolkerInfra as development of their risk identification process.