Evaluating tank farm capacity

Stochastic simulation provides a refinery operator with the information needed to optimise storage in the tank farm

Alastair Painter

Viewed : 9798

Article Summary

Refineries and petrochemical facilities have historically utilised storage to ensure the security of feedstock and product supply, as well as to manage operational issues within the production complex. Storage allows feedstocks to be maintained at levels sufficient to mitigate upstream supply interruptions or to allow production to continue uninterrupted whenever disruptions or upsets occur within the downstream process units. The amount of storage provided for the facility must be adequate for the management of day-to-day routine activities, as well as allowing upsets associated with unplanned events such as shipping and weather delays, process and utility outages, and unit slowdowns to be accommodated.

For many facilities, the response to operational problems experienced within the refinery complex has often been to add more storage. Hence, the number of tanks within the tank farm slowly increases with time, the ways in which they are used and their interconnection with the process facilities becomes more complex, and the management of the tank farm becomes increasingly challenging. However, tanks are expensive to build and maintain, and in today’s difficult economic environment many refiners are asking whether it is possible to reduce the number of storage tanks that they have in operation, thus allowing significant savings on tank inspection and maintenance activities to be realised. The key challenge that the refinery operator considering this option faces is in understanding just how much storage is sufficient.

Risk assessment
Understanding the optimum level of storage to provide for a given service requires a risk assessment approach; that is, both the frequency and consequence of events that affect the utilisation of the storage must be considered.  Ultimately, the problem facing the operator is that of minimising capital and operational expenditure associated with the storage while ensuring that revenue (production) is not compromised as a result of a lack of available storage. The traditional LP model used by refineries for planning purposes is not sufficient to provide a solution to this problem, since it assumes that plant operations are predictable and steady state in nature. However, as noted previously, the utilisation of storage capacity is affected by sporadic events such as process unit outages, export and import delays, as well as other unpredictable events that may occur within the refinery. Hence, the solution adopted must allow the impact of such unpredictable events to be captured and quantified. Stochastic (or discrete event) simulation is one means that has been successfully used to predict the behaviour of complex systems such as a refinery complex.

DNV utilises its total asset review and optimisation (TARO) simulation tool to assist refinery operators, with the quantitative assessment of performance achievable from their facilities. It is a discrete event simulation tool that is capable of quantifying the expected performance of an entire asset (a refinery complex) or of individual equipment items (pumps, compressors, and so on). Figure 1 illustrates the various components that comprise a TARO simulation model. While it utilises the core concepts and algorithms found in traditional reliability, availability and maintainability (RAM) modelling, it differs from such tools in that it has been explicitly designed to allow the performance modelling of refinery and chemical complexes. Hence, while process unit availability is normally an output of the traditional RAM modelling process, for TARO it is actually a starting point; the model comprises information regarding unit availability, unit capacities, stream routings and flow rates, storage tanks, sales strategies, and so on.

The software operates by simulating the occurrence of fundamental events that result in changes to the operational state of the plant over time. Events are described to the simulator in terms of failure and duration distributions that are generated from historical operating information obtained from the refinery or chemicals facility. These virtual events, combined with the simulator’s knowledge of the refinery configuration and flows, permit the performance of the facility to be tracked and reported. By simulating many operational lifecycles of the facility and aggregating the output, the software is able to predict the future performance of both the overall system and the individual components that comprise it. Performance is typically measured in terms of metrics such as annual product volumes, crude charge rates and storage utilisation.

Case study: refinery/petrochemical complex
DNV has been working with a major European refiner over the past few years to deploy performance simulation as a decision support tool at one of the refiner’s principal refining and chemicals production sites. The culmination of this work is an integrated performance simulation model of the entire refinery and chemicals production complex comprising more than 22 process units, utility systems and all intermediate storage tanks (of which there are approximately 50 in total).

Early in 2010, the refiner embarked upon an initiative to drive reductions in operational expenditure through optimisation of the number of storage tanks at the site. DNV assisted with this initiative through application of the simulation model to generate information that could be utilised for decision support purposes.

Naphtha storage
One particular area of concern was that of naphtha storage, located between the crude unit and the downstream process units; a naphtha hydrotreater and an ethylene cracker. The existing storage consisted of four tanks, providing a total of 132 000 m3 of naphtha storage (see Figure 2). Under normal operation, one of the two large 
(50 000 m3) tanks was allocated to storage of naphtha feed to the downstream hydrotreater and cracker respectively.

Tank T-113, which normally stores naphtha feed for the ethylene cracker, had to be taken out of service for inspection and repair in advance of an upcoming turnaround of the cracker. To eliminate the expenditure associated with this work, it was suggested that T-113 could be removed permanently from service. Two alternative configurations were proposed regarding the disposition of the remaining tanks once T-113 was out of service. The system could be operated using the three remaining storage tanks, with the large tank (T-112) dedicated to cracker service and the two smaller tanks (T-107 and T-102) being deployed for hydrotreater feed service. As an alternative, the idea of converting an existing light cycle oil (LCO) storage tank (T-220) to naphtha service was also considered. This would allow an additional 20 000 m3 of naphtha storage to be made available upstream of the ethylene cracker. Table 1 summarises the various operational scenarios considered.

Add your rating:

Current Rating: 3

Your rate:

  • Responsive image OptiDist - benchmark in atmospheric distillation testing
  • Responsive image Follow us on LinkedIn
  • Responsive image NSI Mobile Water Solutions
  • Responsive image Honeywell Sustainable Aviation Fuel
  • Responsive image Lummus Process Technologies
  • Responsive image RVP in process
  • Responsive image Process catalysts
  • Responsive image Shut-off valves
  • Responsive image Asset Management
  • Responsive image Vacuum Systems