Primary Reformer - Firebox explosion New BMS and HAZOP Actions overview

In June 2009, a top fired primary reformer exploded in an ammonia plant, YARA Tertre Belgium. Incident investigation concluded that the direct cause of the accident was the introduction, by error, of a large amount of fuel gas through unlighted arch burners. The first part of this paper describes the new Burners Management System (B.M.S) proposed by M.W.KELLOGG which considerably reduces the likelihood of a similar accident.. The second part describes the series of HAZOP actions that were completed prior to start up. This paper highlights major actions realized before the commissioning, permitting to maximized safety of a 41 years old ammonia plant with current available technologies. Aurelien Flamme YARA, Tertre Sammy Van Den Broeck YARA, Sluiskil Wim Versteele YARA, YPO The Tertre ammonia plant (M.W.KELLOGG catalytic high pressure steam reformer natural gas feedstock) has been in service since 1968 with an original design capacity of 910 MTPD. The Primary Reformer is fired by 180 arch burners, 8 tunnel burners and 12 superheater burners. The plant has been revamped several times to achieve an actual maximum daily production of 1250MTPD and a specific consumption in line with European competitors. In June 2009, the top fired reformer blew up during the start-up of the plant

Introduction

T
143

his major accident occurred on Saturday 27 June at 1.00 in Tertre. The root cause was the erroneous introduction of a large amount of fuel gas through the unlit burners of the top fired furnace, which ignited and violently raised the pressure in the box. The blast pressure in the immediate vicinity has been estimated at less than 10kPa. The noise would have been heard some 5 km away.

2010

AMMONIA TECHNICAL MANUAL

The accident resulted in two seriously injured operators and the complete destruction of the reformer. Collateral damages to the neighboring equipment was limited. There were no chemical releases. Sequence of events On Friday June 26, a heavy thunderstorm occurred in the area of Tertre (Bel.). This caused the failure of the external power supply and trip of the ammonia plant at 15:20. When the electrical substation of the ammonia plant switched on again, preparations for restarting the plant were initiated. The ID fan was in operation from 16:30 on Friday afternoon. The start-up sequence of the auxiliary boiler in the ammonia plant was commenced at 19:30 for producing 9 barg steam at 21:00 on Friday evening. The following instructions were given to the night shift: Be ready to produce 38 barg steam in order to start the steam turbine JT-106 for the FD fan at 7:00 on the Saturday morning. Starting this turbine

requires superheated steam, raised in the superheater of the convection section. This superheating occurs in the superheater and tunnel burners. Around 00:30 the pressure of the auxiliary boiler was about 30 barg and the tunnel burners could be started in order to superheat the steam. At 00:55, the block and bleed valves CV103 A/B/C in the fuel gas line were manipulated. This is a common valve system for feeding fuel to the arch burners (180 pcs), the tunnel burners (8 pcs) and the superheated burners (12 pcs). When this block and bleed system is operated, the only valves preventing fuel entering into the furnace are the manual valves on each burner, since the hand control valves for the arch burners (MIC 102-110) are known not to be tight. A check of the position of the manual valves was executed for tunnel and superheater burners. However the check of the isolation valves for the arch burners was not done (check list not followed). From this point, the flow of fuel gas into the furnace (through the arch burners) started. The explosion occurred 35 sec after activation of CV103.

AMMONIA TECHNICAL MANUAL

Figure 1 : Fuel gas distribution overview

144

2010

2010

145

AMMONIA TECHNICAL MANUAL

Figure 2 :Primary Reformer scheme

Extent of the damages

former structure. The north wall bent up to contact with a pressure vessel. Debris was thrown towards the control room at 25 m away from the furnace and beyond.

Figure 5 : North rack view

The North rack showed in figure 5 (supporting ammonia and natural gas lines) was deformed but no leaks occurred on pipelines. Direct causes. The peak flow of fuel gas reached 21000 Nm/h in less than 1 minute. Air and fuel were mixed in a hot environment (300C), above the Low Explosivity Limit (LEL) . The ignition of the mixture could have been caused by: Temperature inside the furnace. However the nitrogen inside the catalyst tubes was only at 300C and the temperature of the circulated air was 250C. This is not sufficient to ignite natural gas (self ignition 537C). Is it however not possible to rule out the possibility of having some residual hot spot in the furnace. A flash of lightening at the stack. However the last lightening was recorded at 16:30 the day before the blast. The electric lighter of the operator. This explanation is the most plausible since one of the injured operators was holding the lighter, ready to light the tunnel burners. This point could not be verified.

Figure 3 & 4 : Radiant section after the blast

As shown in figure 3 and 4, the east wall of the reformer was destroyed. The other walls were stopped by the adjacent equipment. The south wall was stopped by the convection block. The west wall bulged, but was stopped by the secondary re146

AMMONIA TECHNICAL MANUAL

2010

New Burner Management System (B.M.S)

Tthe intention of YARA Management was to rebuild the reformer and get the Ammonia Plant back online in the shortest possible time. More than this, a major update of the fuel gas management system has been engineered in cooperation with M.W.KELLOGG . A new Burner Management System (B.M.S.) was proposed and accepted by YARA Project Office. This system introduced a new start up philosophy of the Primary Reformer, in line with the current best practices. It includes the introduction of block and bleed systems in each major fuel gas line with a nitrogen leak test associated. Linked with the new Burner Management System, approximately 60 new safety valves have been installed on the new primary reformer along with 58 flame detectors. In addition, it was decided to change the start up philosophy of the FD fan. MS steam export to the turbine allows the startup of the FD fan independently of the reformer light on sequence. This considerably reduces the risk of a massive fuel gas leak through the arch burners during MS steam superheating. Generic Start up procedure of the Primary Reformer The new start up procedure passes through permissives before commencing a leak test (nitrogen) and light on sequence. This procedure for the arch burners can be described as follows: Permissives check Aux. boiler operating Radiant section purged* o Reformer trips active

o Arch burners safe (FG valves closed) o Superheater burners safe (FG main and pilot valves closed) o Tunnel burners safe (FG main and pilot valves closed) o Combustion air flow > purge minimum * conditions satisfied 20min (uninterrupted) prior to
starting

Arch burners FG header depressured Arch burners FG to burners gas valves closed and vent open Arch burners FG to burners rows depressured Arch burner checking gas valves closed Arch Burner checking gas vent open

Start up sequence Check main block valves (no overpressure when block and vent valves closed) If not successful sequence aborted Header and burners rows tightness check o Timed pressurization (nitrogen) o Timed pressure maintain o Depressurization If not successful sequence aborted Vent checking gas (nitrogen) OK to light burners indication starting timer requested to proceed at the light on sequence

Light on sequence Operator starts the light-on sequence. This sets a timed override on the Fuel Gas LL pressure and the flame failure trips for the arch and allows the header fuel gas valve to open and the vent to close. The trip bypass timer lasts for 5 minutes. Flame and fuel gas pressure for at least one row must be established within this time, otherwise the light-off sequence must be re-started.

2010

147

AMMONIA TECHNICAL MANUAL

The Operator can choose any row of the Arch as the first to be lit In the chosen row, the field Operator must now light any two of the burners, fitted with flame scanners (within 5 minutes) , using the portable igniter (4 flame detectors per row) row trip at 2 out of 4 The Operator now proceeds to light the remaining rows in the same manner as the first. The rows can be lit in any order, but in each case two burners fitted with scanners must be lit to keep the row operating (no time constraint anymore) Notes : Only one row may have a gas feed with no flame established at any one time

Nitrogen tightness test in parallel for pilot and main burners Pilots must be light prior to the main burners A fixed igniter is installed at each pilot burner assembly (to be started in the field as well at the control room) Flame detector for each burners

Air/Fuel ratio control A trip is provided to ensure that the reformer is not allowed to operate with insufficient air available for good combustion in the Arch firebox. The amount of air required for combustion of the fuel gas is calculated and a trip is initiated if the actual air flow falls below 3% in excess of the stoechiometric requirement.

The start up of tunnel and superheater burners follow the same sequence as the arch burners. Differences are: One pilot burner linked with each main burner

Figure 6 : Schematic overview of BMS Steps for Arch / Tunnel / Superheater Burners

AMMONIA TECHNICAL MANUAL

148

2010

Major HAZOP Actions overview

Following this major event, a complete HAZOP study of the plant (supervised by M.W.KELLOGG). was carried out parallel tothe Reconstruction Project. 216 actions items came out of this HAZOP. These actions are split into four categories. Class 1 : Need to be implemented prior to restart Class 2 : Need to be implemented in 2010, but can be after restart Class 3 : Can be phased Class 4 : Need further study Class 4 also contains some elements which should be implemented at the first opportunity, but which need more study prior to implementation.

tion given the fact that these tag numbers have been used in several other documents. Guidelines given by ISA 5.1 were applied regarding the update P&IDs update project, applying a more strict method for tag numbering and instrumentation representation. DCS/Safety Management programming A major effort was required to modify the DCS and the SMS system. Main elements were: Implementation of additional alarms and trips Modify trip system such that no by-passing of trips is required during start-up Complete Plant trip on low level of boiler IAutomatisation of major controls (air/gas ratio control, steam to carbon ratio control)

Main Instrumentation improvements requiring hardware The main instrumentation improvements where additional hardware is required were the following:
Figure 7 : Number of actions per category

Improvement of anti-surge control of synthesis gas compressor 103J Improvement of anti-surge control of refrigeration compressor 105J Installation of atrip valve at the outlet of the HP ammonia separator Improvement of fuel gas system to synthesis startup heater to acceptable standards.

Update of PIDs Main Mechanical Hardware requirements A major effort was made to bring the P&IDs up to standard and to make sure that P&IDs reflect the actual situation. P&IDs was problematic due to the tag numbering history. Tag numbers were assigned according to the process equipment the instruments need to monitor and several suffixes are attached to the number making the system complicated. It was however impossible to renumber the instrumentaThe main mechanical modifications where additional hardware is required were the following: Installation of non return valve in semilean (aMDEA) line to absorber, close to the absorber Installation of non return valve in lean line to absorber, close to the absorber

2010

149

AMMONIA TECHNICAL MANUAL

Installation of a vacuum breaker on the strippers

Class 1 actions were subdivided in 9 subcategories: Logic Control Modification Checks/Calculations Update P&IDs/C&E diagrams Production Actions New Primary Reformer Project Central & Local Maintenance Process Instrumentation NDT Trips/Alarms Others Points

Required Inspections The main actions regarding NDT department were the following: Inspection of CO2 section (historical inpection of the absorber and the strippers was unclear) internal and external inspection recommended Review of relief valves capacities-sizing calculations and underlying scenarios

Required Studies For the time being, there is a single non return valve between the process air preheat coil and the secondary reformer. The coil is protected in case of air compressor shutdown by injection of medium pressure steam. This steam should also prevent backflow from the secondary reformer into the process air coil. This system is considered as weak and must be re-enforced in the coming months by a block and bleed system.
Figure 8 : HAZOP Action Management Organisation chart

HAZOP Actions Management To maximize efficiency of resolving Class 1 actions, it was decided to create a HAZOP Action team. In parallel, team members were also involved in the reconstruction project. A HAZOP Actions Manager was dedicated at the organization of a weekly follow up meeting to identify red lines and organize issues to be solved. Completion of these actions was directly linked to the commissioning approval database.

AMMONIA TECHNICAL MANUAL

150

2010

Conclusion
The consequences of the explosion could have been far more serious for the persons injured and catastrophic for the population around the plant if the ammonia pipe on the piperack had ruptured. The direct cause of the destruction of the reformer was a human error, made possible by the lack of rigor in the way the operating procedures / checklists were implemented in Tertre and the absence of others barriers. The root cause was the lack of a safety system on the reformer. It should have been confirmed by safety studies (HAZOP, SIL, ) and dealt with in order to reduce the risk to an acceptable level. Procedures and checklists will always remain, especially during start up or shutdown, the first, and sometimes unique barriers on the way to the accident. The new B.M.S. that has been implemented partially resolves the problem. It fits the plant with the best safety practices with regard to the start up procedure of lighting on a primary reformer. More than improving safety-reliability of the reformer, opportunity was taken to update and resolve weak points of the plant (safety, productivity, reliability point of view) with major technical actions. The HAZOP Actions team must be part of the Rebuilding Project team. This helps by reducing redundant information and gives a more realistic view of the potential bottlenecks. These bottlenecks (or challenges!) must be identified as soon as possible, reducing the risk of material delay and/or technical issues not in line with the Reconstruction planning (= master planning). Putting forces on a well identified weak department/team can definitely improve the probability of success in this HAZOP project concept, reducing stress while improving efficiency of the different actors.

2010

151

AMMONIA TECHNICAL MANUAL

AMMONIA TECHNICAL MANUAL

152

2010