Apr-2017

Identifying contributors to flaring

Applying an integrated flare and fuel gas monitoring system to identify and eliminate major contributors to refinery flaring.

P SRIDHAR
Indian Oil Corporation

Article Summary

People are often confused when it comes to flare monitoring. In a refinery, where hundreds of control valves, dump valves and pressure safety valves make their way into the list of causes, it can be difficult to monitor and minimise flaring.

It is a very complex task for a person to identify difficult and frequent occurrences of flaring, or to pinpoint contributors to flaring. Operators can come into contact with a huge list of split ranges and dead bands when they look into this matter. In addition, those responsible for flare monitoring may not have an in-depth knowledge of every process involved, thus making the issue more confusing.

Common reasons for flaring
A few common reasons for flaring (apart from acid gas flaring) in petrochemical plants and petroleum refineries are outlined below:
A. Imbalance in the fuel gas system
B. Process requirement where column overheads are operated at very low pressure (about 1 kg/cm2)
C. Dumping from the hydrogen header
D. Process abnormalities leading to flare
E. Process equipment abnormalities.

Point B above is a system requirement to reject lighter gas; it cannot be controlled.
Point E can be solved during a turnaround (if fouled) or by revamping (if under-designed). Points A, C and D can be addressed by effective monitoring of the system.

Proposed methodology
Trivial flaring points might nonetheless be identified or targeted as major contributors to flaring depending on the percent opening of control valves. To avoid any confusion, it is proposed to sort flaring points based on volumetric flow rate using the following formula:

P*D3/100*% op

This can be used to set the order and magnitude of flaring. The formula cannot predict the exact amount of flaring, but it can arrange flaring points in order of priority (see Table 1).

For all contributors to flaring, percent opening can be normalised on the basis of actual percent opening to flare, considering dead band and split range. This method can be adapted wherever a flaring measurement is missing or not reliable. For the above example, trivial Point A can be ignored and Points B and Point C can be focused upon to reduce the level of flaring.

The basis of the above equation to infer flaring is as follows:
At any point, the number of moles being flared through a control valve is α.
For the number of moles present in a control valve port, n:
n = P * V/ (R * T)
n = P * π D3 (6 * R * T)

where P is the upstream pressure,  T is the temperature and V is the volume of gas being flared. D is the diameter of the control valve port.
Because the temperature from the surge drum, from where most of the flaring takes place, is almost constant:

n =K*P*D3  (K is constant)
n/K=P*D3
dividing by 100 on both sides
n/K/100=P*D3/100
n/K/100 represents relative moles

Integrated flare and fuel gas monitoring sheet
Data for an online system can be imported directly from a real time database in Excel format. Flaring data should be retrieved on a five minutes average basis to make the system dynamic.

Flaring control valves can be arranged in order of maximum flaring by using the expression:    
     
Pressure in kg/cm2 g *(c/v diameter in inch) 3*% opening

Fuel gas hydrogen generation and consumption data can be retrieved on a last one average basis to make it more reliable. These data can be sorted on the basis of their deviation from the last one week average. Based on this, deviation in the system can be identified to fix the actual problem.

All of these data can be integrated onto one sheet, an integrated flare and fuel gas monitoring sheet (see Figure 1). Continual updating of these data facilitates easy, one-click understanding of a fuel gas and flare system sheet dashboard.

Action to be taken can be incorporated into the sheet, based on the upper and lower limits of individual consumption points.

Predetermined disturbances should also be considered in order to determine a strategy to mitigate flaring. For instance, delayed coker unit drum change-over and FCC unit catalyst make-up lead to extra gas generation which will go to flare if not absorbed into the system.

Provision should also be made to highlight recovery of poor data as the system depends completely on the quality of data recovered.

Results
The real time trends shown in Figures 2 to 4 demonstrate the effectiveness of a close monitoring system.
 
Conclusion
At present, there is no reliable and effective method for determining flaring from process units. The integrated system for flare and fuel gas monitoring described here will facilitate monitoring of a fuel gas and flare system at a glance.

It will also enhance operational competitiveness among the various operating units for achieving maximum reduction in flaring and fuel gas consumption. Hidden problems will be identified and resolved in an appropriate manner. For instance, a persistent problem may be resolved by the addition of a condenser or by more timely cleaning of a condenser.