Jul-2010

Reliability analysis to determine environmental emissions from units

Reliability analysis determines the extent to which plant can meet emission limits and supports estimates of possible financial penalties for non-compliance

Chiron Mukherjee
DNV

Article Summary

The control of environmental emissions is one of the most important current considerations in the design and operation of a refinery. Along with the general aim of limiting harmful emissions to the environment to as low a level as possible, specific legal requirements are now common in many countries. Failure to limit refinery emissions may force refinery or unit closure, or the payment of large financial penalties.

Given these considerations, the quantitative assessment and prediction of emissions are very important values for refinery planning and decision making. Reliability, availability and maintainability (RAM, or simply “reliability”) analysis is often used to assess the performance of a unit or asset in terms of availability and production or throughput efficiency. The methodology used in a typical reliability study can be expanded to include the quantification of environmental emissions, and this process is outlined here.

This assessment enables the probability of a unit or refinery meeting emissions targets to be estimated. Additionally, by quantifying expected financial penalties from excess emissions, a more complete cost-benefit analysis for the selection of alternative design or operational choices can be made. 

The methodology outlined here is based on several consulting projects performed by Det Norske Veritas (DNV) for refineries across the globe. These analyses have used the Total Asset Review & Optimisation (TARO) software package developed by DNV. A specialised RAM tool for the refining and petrochemical industries, it can be used in the analysis of individual standalone process units, but its main value is that it enables the study of the entire refinery or asset in a holistic approach with full interconnectivity between process units, utilities systems and logistics constraints included within a single model.^1,2

The article describes a general methodology for incorporating the assessment of environmental emissions to current reliability analysis studies and is not limited to a specific software modelling or analysis technique.

Description of methodology
Standard reliability analysis
The key input data for a standard reliability analysis of a process unit or facility are events that cause capacity or production losses. The reliability analysis, driven by these events, can be used to determine values such as the availability or production efficiency of a process unit. This can be expanded further to model the performance of an entire facility (such as a refinery or petrochemical complex) by also considering interactions between different units and the effect of each event on the overall system material balance.^1,3

The level of detail of the analysis can be performed at various scales. At an equipment-level detail, unplanned failures of each piece of equipment would be included individually in the analysis as events. In a higher level analysis, each event may correspond to a general unit or subsystem failure. Scheduled maintenance and turnarounds are typically also included within the analysis as events.

Events are characterised by the following input data:
•    Frequency (or probability of event occurring) For unplanned failures, this can be equivalent to the mean time between failures (MTBF), with an associated failure frequency probability distribution such as an Exponential or Weibull function. For scheduled maintenance, events would occur at specific planned times
•    Duration The duration of the event would include the logistics delay between the event occurring and repair starting, which could include mobilisation of people, tools, equipment and parts, and the actual duration of repair for the failure or maintenance activity. This can follow a repair time probability distribution, such as a normal, rectangular or triangular function. Any restart ramp-up delays following repair can also be included within the analysis
•    Impact Each event would have an impact on the capacity or production rate of the unit or system. The event could cause a unit shutdown or a slowdown to a reduced rate. It is possible that different stages of the event would have a different effect on capacity.

An example of an event, with repair duration and slowdown impact on unit capacity with time, is shown in Figure 1.
Based on the input data of failures and scheduled events, reliability analysis of the system can be performed using various alternative methodologies or software methods.

A key result from a reliability study is the expected production from a unit or facility. This can be derived for each product stream and as a function with time. Given expected pricing forecasts for each product type, a revenue forecast can also be projected:

RESULT ,R = ∑ Cij Pij
                                                       ij

In this expression, the total financial result is equivalent to the revenue produced for each product stream, Pj. This is calculated for each time step (i), where product cost forecasts exist for each time step and individual stream (Cij), over a given total design life. This can also include feed costs, equivalent to a negative financial parameter, -C.

The effect of design changes or operational changes on the projected revenue allows for cost-benefit optimisation and decision support. This is based on the predicted financial result, with the expected cost for making a specific infrastructure change, Δ. The effect of cost-cutting measures would be associated with a negative value for Δ:

R = ∑ Cij Pij – Δ
          ij
Some examples of design optimisation include:
• Equipment sparing level
• Debottlenecking analysis
• Cost benefit of unit revamp
• Comparison of alternative designs
• Tank size selection.