Anyone who has ever been involved in planning and executing a major shutdown project will recognise the scene: a dozen teams scrambling over scaffolding to work on tanks, columns and pipelines. Containers are dropped onto any free space as temporary offices, meeting and break rooms. There are makeshift parking spots in front of the site and maybe a large tent to serve as a staff restaurant. All told there might be several hundred or thousand contractor colleagues from various disciplines who have to be accommodated alongside the company’s own personnel. In short, it’s controlled chaos. Every day that a plant is not running costs a lot of money in the form of lost profit margins, so everyone makes every effort to realise the turnaround as quickly as possible. At the same time, if the shutdown period is too short, you run the risk of incurring additional costs, untold production downtime and delivery bottlenecks. Worst case scenario: customers migrate over to the competition and the company is left to pay punishing fines. And if there is one thing we all know, it’s that whenever you have several thousand people working side by side then something is bound to go wrong. A successful turnaround demands reliable and realistic scheduling that reflects the dynamic nature of a shutdown.
What you really need to know is: when exactly does a schedule achieve the perfect balance between the minimum downtime required and adequate, transparent time reserves that allow a response to unforeseen or newly emerging risks? The answer could lie in the tried and tested Monte Carlo simulation. Known for its many casinos, Monte Carlo in Monaco lent its name to this stochastic method that utilizes random numbers. Originally used during the Second World War by US nuclear research scientists, the Monte Carlo simulation is now an established tool in risk and project management for evaluating and quantifying time and cost risks. Taking an existing schedule and entering the project risks into the model, the simulation calculates their impact on the critical pathway and end date. The benefits are obvious. Conventional schedule analysis, detailed as it is, can only provide turnaround managers with insights into a limited number of scenarios, e.g. the best and worst case, meaning the shortest and longest expected downtime. Yet these findings are merely deterministic given that deadlines are usually static. No consideration is given to how milestones and the end date in particular could shift if individual tasks take longer than planned.
This is where a Monte Carlo simulation comes into its own, by working through all potential project scenarios and analysing various types of documented risks. The most critical of these is known as ‘task uncertainty’, meaning the uncertainty that a task could take longer or be completed faster than scheduled. There are also risk events, such as heavy rain that prevents welders from doing their job and delaying everything that follows. For each position, the simulation randomly decides whether a risk arises and what impact it has on subsequent deadlines and the end date. It then creates an individual scenario for each run that describes one potential project outcome. This process can be repeated as often as necessary before statistically analysing the resulting scenarios on a scale from worst to best case. These findings give planners concrete information about the likelihood of meeting individual deadlines and the project end date, for example, that the probability of meeting the project end date might only be 20% and it would take an extra five days to achieve the target probability of 80%. The simulation results also help identify risks that have the greatest impact on the target end date.
As with any mathematical method, the quality of the results is only as good as the quality of the input data. Consequently, a Monte Carlo simulation draws on a detailed timeline for the shutdown that takes into account the distribution of expected completion times for each work step, i.e. an optimistic, realistic and pessimistic duration. This data is not usually on hand and has to be requested from managers and onsite experts. The Monte Carlo method allows you to then link up work packages in a dynamic plan and see, at a glance, by how much the end date will shift if work on a package takes more or less time than scheduled. ‘Time and again we see engineers and technicians having difficulties with relative estimates,’ said Frank-Uwe Hess, Co CEO at T.A. Cook who analyses shutdown projects and supports their optimum preparation, ‘and sometimes they even use pessimistic estimates.’
It is worth all the effort since it gives turnaround managers a solid basis for reviewing their strategy and deciding whether to adjust the time given for the shutdown or to schedule more resources to complete the work faster. Take this example: A Monte Carlo simulation for a shutdown reveals that the delayed completion of a column, even though it is not part of the critical pathway, would result in a 30% probability of meeting the target project end date. The key factors here are problems expected during dismantling, the planned duration of welding activities and the potential scope of unforeseen repairs. Choosing to carry out welding work on the column overnight as well as during the day, as originally planned, decreases the lead time and increases the probability of meeting the project end date to 75%.
Another advantage of this simulation is that it does not only reveal shortcomings but also buffers in scheduling. To fully capture these reserves, which are rarely used productively, turnaround managers can devise optional tasks that are useful but not absolutely essential. For example, workers should only clean certain components of a machine when the project is progressing well and the time buffer is not being used elsewhere. This is an efficient use of any reserves identified by the simulation.
Companies often use the time allocated for a regularly scheduled cleaning shutdown or legally prescribed audit to optimise their plant at the same time. Here too a Monte Carlo simulation can help challenge investment decisions. In cases where it seems as though necessary work is going to prolong the shutdown period, a comparison may show that the extension is perhaps not worth the loss of profit margins. Such was the case for a chemicals corporation with a plant in Europe. This plant had been shut down for mandatory checks, and the company was planning to expand in three areas during that period. 400 work packages had been scheduled: 150 for an audit of the existing plant and 250 for the expansion. After running a Monte Carlo simulation, the company realized that some of the investment related packages would very probably not be completed in time for the authorities’ audit visit. So as to not spin out the shutdown unnecessarily, management decided to cancel three investment projects. According to the Monte Carlo simulation there was a 70% probability that the work on this subproject would have extended the shutdown by up to five days. Each additional day would have added EUR 300 000 to shutdown costs and exceeded the proposed EUR 50 000 investment cost for the expansion project many times over. This is yet more proof that analysing a project by means of a Monte Carlo simulation pays off in many cases. In spite of its name, this tool provides an important basis for decision making and ensures that scheduling and project planning are much less of a gamble.
Written by André Schmidt-Carré, on behalf of TA Cook.
Edited by Claira Lloyd.
Read the article online at: https://www.hydrocarbonengineering.com/special-reports/13112014/ta-cook-anything-but-a-gamble-1531/