1 reply to this topic

[ianmcq28]

Hello,

 

I am currently involved in HAZOP and LOPA studies involving dry gas seal systems associated with a number of centrifugal compressors we have on site.  All of the seals are John Crane 28AT tandem seals with a process labyrinth and separation labyrinth.  The seal gas supply skids were designed by another vendor. 

 

Brief Description of one Compressor Duty

 

Carbon monoxide leaves a cold box at around 0.53-0.65 barg and is compressed to 31.5 barg by C.803, the CO product compressor. The unit is designed to handle 14,000 kg/hr but can handle up to 19,000 kg/hr.

 

The compressor is a two casing machine with each casing contains nine stages. Intercooling is provided after half the stages in each casing and then again between the casings. The machine is piped such as to give a 'back to back' arrangement, i.e. suction and intercooler return connections at outer ends of casing, discharge to intercooler and casing discharge at the centre.

 

The LP casing raises the pressure from 0.49-0.65 barg to ~7.7 barg. The HP casing raises the pressure from 7.4 barg to 32.2 barg.  The CO is discharged at 30°C after an aftercooler.

 

Brief Description of Equipment

 

The product compressor is a multistage centrifugal compressor with steam turbine driver. The compressor being split into two casings (LP and HP casings).  The turbine and the two compressor casings are mounted 'in-line', being direct coupled by flexible gear type couplings.  The turbine is situated at one end of the assembly with the LP casing in the centre.  The drive to the HP casing is transmitted through the LP casing shaft.

 

The compressor units are of the horizontally split casing design.  Each compressor unit contains nine stages with intercooling after the fifth stage.  Both casings are fitted with dry gas seals to minimise the risk of leakage of carbon monoxide into the compressor house.

 

Machine buffer gas is supplied to the dry gas seals and nitrogen to separation seals to prevent the possible ingress of oil into the process along the shaft.

 

The steam turbine driver is a seven stage impulse turbine which expands steam from 45 barg to 2.8 barg.  It is rated at 2.9 MW at 10,843 r.p.m.  The compressor absorbs 2.5 MW at 10,250 r.p.m.

 

Problem Statement

 

The HAZOP/LOPA action in question concerns hazards associated with a loss of supply of separation seal gas to the separation labyrinth.  The separation labyrinth is intended to maintain a velocity of separation seal gas through the separation labyrinth and prevent ingress of bearing oil into the housing of the dry gas seal which has potential to result in dry gas seal contamination, seal hang-up and failure and hence result in a toxic release of process gas into the compressor house.

 

Our separation seal gas supply is clean and dry nitrogen which is let-down using a direct acting pressure control valve (regulator) (PCV) from 11 barg to a constant pressure of 0.15 barg upstream of the entry into the seal housing/separation labyrinth.  This is sufficient to exceed the minimum velocity requirement across the separation labyrinth and the separation gas is vented via the secondary vent to atmosphere.  The main cause of a loss of separation seal gas has been identified as PCV failing to a closed position.

 

The system is currently designed with a low pressure trip to safely shutdown the compressor in the event of loss of separation gas.

 

We have conducted a LOPA assessment to determine the required probability of failure on demand (PFD) of our safety instrumented system (SIF).  SIF validation calculations (to IEC61511) have determined that the SIFalone can not be considered reliable enough to meet our target mitigated event likelihood.  We have looked into increasing the reliability of the SIF, but this is not practical; largely due to the inherent reliability of the compressor CSEV (combined stop and emergency valve - trips steam to turbine).

 

To close the LOPA gap and meet our target mitigated event likelihood; we require a further risk reduction factor of 10.  This could be achieved by providing additional independent protection layers.

 

Preferred Solution

 

The preferred solution is to provide an independent high priority safety related alarm (SRA) such that; in the event of a loss of separation seal gas supply and the low pressure trip failing to stop the compressors, an operator can be called upon to act to make the situation safe.  It is perhaps reasonable to consider that personal safety monitors and gas detectors in the compressor house may also play a part in preventing serious harm to personnel, but no credit is given as these systems will activate after the event.

 

It was assumed that seal failure caused by a loss of separation seal gas supply will likely be a slow to develop event (will not lead to seal failure and toxic release the instant the separation gas is lost).  Therefore there will be sufficient time for an operator to take appropriate action to make the situation safe within the process safety time.

    

Our proposal is such that if separation seal gas pressure falls to the low pressure trip setting then the system will normally trip and an alarm will be raised.  The alarm response will include:

 

1.            Follow compressor shutdown actions.

2.            Check compressor revs and confirm compressor stopped.  If not;

3.            Check environmental monitors for CO gas escape.

4.            Close CSEV manually.

 

It has been considered that if the compressor continues to run with a loss of separation seal gas; then the alarm response and controlled shutdown of the compressor can be complete within approximately 30 minutes. 

 

Justification?

 

My own survey of industrial standards/references suggest the following recommended control system responses for this scenario:

 

•             API614 suggests an alarm is required in this situation and no automatic shutdown necessary

•             Shell DEPs suggest the same and stipulates that loss of gas supply will not immediately result in a hazardous scenario

•             BP GIS 34-701 suggest an alarm and shutdown is required (which we have)

•             Paper by John Stahley suggests a delayed shutdown is required to give enough time for the operator to intervene and recover the situation / initiate a controlled shutdown http://turbolab.tamu...31/t31pg145.pdf

•             Other internet book references refer to initiation of an alarm and conduct a controlled shutdown within 30 minutes.

 

I would like to call upon the expertise of this forum to determine whether this is a reasonable philosophy please?

 

I am also working with John Crane to determine whether this approach is justified.

 

 

Best Regards,

 

Ian

 

 

Attached Files

•  Basic sketch.pdf   43.68KB   46 downloads

[senthil13]

Hello there,

 

I would like to say this is one of the best explained questions. Thanks

 

To the point, I have seen the alarm set up on seal gas supply failures. You can actually add redundant alarm points just to be on the safer side and make it a priority 1 alarm. followed by a safe shutdown within a period of time ( in your case is within 30 mins)

 

Thanks

Senthil