J Fail. Anal. and Preven. (2017) 17:971–978 DOI 10.1007/s11668-017-0336-x
TECHNICAL ARTICLE—PEER-REVIEWED
Root Cause Failure Analysis of Corrosion in Wet Gas Piping D. Ifezue
Submitted: 16 July 2017 / Published online: 22 August 2017 ASM International 2017
Abstract This paper discusses the root causes and operational mitigations of corrosion anomalies reported for the wet gas system of a floating production storage and offloading (FPSO) vessel. This system is classified in general into the flowing and dead leg sections and operated below 62 C, where a protective FeCO3 film will not form. Corrosion which is temperature dependent is general. The root cause of the anomalies in the flowing wet gas system was found to be due to CaN and solid deposition in the LP separator, increasing levels of corrosive condensate, which drips down the pipe walls, resulting in wall losses of less than 50%. Evacuation of deposits in the LP separator and subsequent control of its formation rate would mitigate anomalies attributed to increasing levels of corrosive condensates. The corrosion pattern at elbows, modified by turbulent flow of emergent fluid streams through the complex-geometry elbows, is mitigated by velocity control. The root cause of the inspection anomalies at dead leg locations was due to the lifting of the PSV and/or infrequent flow through a bypass valve and directly to flare. Infrequent transportation of condensed, corrosive water thereby resulted in dead leg corrosion. Proposed mitigation entails replacement of defective PSVs and modifying flow frequencies. This paper summarizes all the identified root causes and proposed mitigations and provides a greater understanding of the internal degradation mechanisms operating in wet gas systems in general. Keywords Wet gas Root cause failure analysis Corrosion FPSO Dead leg
D. Ifezue (&) Global Corrosion Consultancy Ltd, Manchester, UK e-mail:
[email protected]
Introduction This paper illustrates the methodology and results of root cause analysis for a wet gas system. Root cause identification permits specific corrective measures to be targeted for mitigation. UT inspection revealed several anomalies at different levels of severities (\50% wall loss) across the flowing and dead leg subsystems. In order to maintain piping integrity required for continued delivery of the required production levels, the root cause of each identified wall loss was investigated, and specific mitigations were recommended. There are several causative, random variables associated with internal corrosion of a pipe. Only a slight change in one of the causative factors can essentially adjust likelihood and severity. Majority of reported models used for predicting internal corrosion in wet gas systems [1, 2] are based on the principle of mass transport and electrochemistry. These models, however, are still unable to precisely predict, in a repeatable manner, the likelihood of corrosion, rate of corrosion, pattern of corrosion and probable location where corrosion will occur for piping/pipelines seeing the same operating environment/chemistry. Typical anomalies in wet gas systems are within the low-to-medium-severity range (i.e.,\50% wall loss). Although a pinhole leak rather than a rupture will be the likely mode of failure, the safety, health and environment (SHE) consequence is usually in the medium to high severities, with the risk of an explosion depending on proximity to an ignition source (e.g., flare or hot compressor motor). The severity of the overall risk is dependent on the chance of people being in the vicinity during a failure event. Oil/gas topsides/FPSO facilities are increasingly reliant on processed inspection data from a typical integrity database in order to understand corrosion patterns and
123
972
J Fail. Anal. and Preven. (2017) 17:971–978
assess integrity. For a system which requires to be operated continuously with minimum downtime; low safety risks; and low maintenance costs, this can be quite challenging and difficult. Neural networks consisting of inputs, outputs and ‘weighted’ processing elements can be used to solve such complex and highly nonlinear control problems [3]. The network can be trained by adjusting the weights in accordance with results of the processing of the root cause and mitigation nodal values. The root causes of reported anomalies and their respective mitigations were evaluated and identified. These represent the input nodes which can then be processed to determine input values which are then weighted. The variability of the processed weight means that the model output is able to adjust within the input/output range for assessed corrosion rate and integrity of the wet gas system. A significant database can thus be developed which in addition to existing integrity databases (for storing inspection and other anomaly results) could give more accurate prediction of corrosion rates and integrity status of wet gas piping systems.
Description of System Figure 1 briefly illustrates the process flow for the wet gas system. Gas released from the low-pressure (LP) separator is stripped of entrained liquids in the LP compressor suction
scrubber prior to the pressure being increased from 0.2 to 3 barg at the LP compressor. The gas is then combined with the gas exiting the intermediate-pressure (IP) separators which is cooled from 127–138 to 62 C at the LP compressor discharge cooler (shell-and-tube-type exchanger). Liquids resulting from cooling are removed at the IP compressor suction scrubber prior to compression from 3 to 15 barg at the IP compressor. Discharge gas from the IP compressor is cooled from 127–138 to 38 C in the IP compressor discharge cooler before flowing through IP compressor discharge scrubber where any liquids are removed. The gas exits IP compressor discharge scrubber to the high-pressure (HP) compression system suction manifold where it combines with gas exiting the HP and test separators. The gas is similarly processed as it flows through the 1st, 2nd and 3rd stages of the HP compression system (distinguished by color
Table 1 Summary of wet gas wraps, replacement packages and wall losses Train 1
Train 2
No of wraps
1
0
Replacement packages
0
0
Severe
[70%
0
1
High
50–70%
0
0
Medium Low
30–50% 15–30%
23 21
7 49
IP Sep Tr 1 & 2
LP Sep
HP Sep Tr 1 & 2
LP Comp Sucon Scrubber
LP Comp
1st Stg HP Sucon Scrubber
1st Stg HP Comp
LP Comp Discharge Cooler
1st Stg HP Comp Dis Cooler
IP Comp Sucon Scrubber
IP Comp
IP Comp Dis Cooler
1st Stg HP Comp Dis Scrubber
Glycol Contactor
Fuel gas Test Sep 2nd Stg HP Comp Dis Cooler
3rd Stg HP Sucon Scrubber
2nd Stg HP Comp
3rd Stg HP Comp
2nd Stg HP Comp Sucon Scrubber 3rd Stg HP Comp Dis Cooler
Gas Li Gas Injecon Gas Export Flare
Fig. 1 Process flowchart of the wet gas system
123
IP Comp Dis Scrubber
J Fail. Anal. and Preven. (2017) 17:971–978
in Fig. 1). After the 1st stage HP compressor discharge scrubber, the glycol contactor removes water from the gas which is then considered sufficiently dry for fuel gas to be taken off, i.e., after the 2nd stage HP suction scrubber. The rest of the gas is further processed through the respective HP 2nd and 3rd stages prior to being supplied as required for gas export, gas lift, gas injection and flare.
Failures and Inspection Findings Table 1 presents the wraps, replacement packages and also the numbers of reported wall losses within each range. As expected for a wet gas system, most of the anomalies are within the medium- and low-severity ranges with only one wrap and one severe anomaly. However, it should be noted that at the current corrosion rates, failure to identify the root causes and the required mitigation will result in rapid deterioration and progression of anomalies into the high and severe ranges.
973
Aqueous CO2 corrosion of carbon steel is an electrochemical process involving dissolved carbon dioxide reacting with iron and producing iron carbonate [1]. The overall reaction is given as: Fe þ CO2 þ H2 O ! FeCO3 þ H2 Prior to dehydration at the glycol contactors, condensation rates of the non-dew pointed wet gas will exceed the threshold of 0.15–0.25g/(s m2) suggested as the value below which sweet top of line (TOL) corrosion does not occur [3]. Condensation rate also increases with turbulence experienced at some elbows. The rate of consumption of carbonic acid by corrosion would exceed the rate at which freshwater condenses thereby limiting corrosion rate by saturating condensed water with corrosion products and further elevating pH. This partly explains why corrosion anomalies reported for the flowing pipe section were of low/medium severities. Wet gas is transported at temperatures exceeding the pipe wall Table 2 Wall loses in the dead leg system
Wet Gas Piping: Flowing System The flowing wet gas piping system consists of all piping from the LP separator to the glycol contactor (Fig. 1).
Corrosion Mechanism The wall loses reported in this wet gas system is attributed to CO2 corrosion caused by corrosive condensate dripping down the pipe wall.
Pit depth: 7.82mm General corrosion (grid): A: 90 x 140mm D:7.82-8.15mm
Pit depth: 7.8mm General corrosion (grid): A- 220 x 65mm D-7.3-8.5mm
General corrosion (grid): A: 40 x 18mm D: 7.94-8.20mm
Pit depth: 7.15-7.89mm General corrosion (grid): A: 235 x 78mm D: 7.43mm
Max wall loss (%) Dead leg lines
Train 1
Train 2
LP scrubber to PSV
15
HP cooler to PSE
16
23.2 39
LP cooler to PSV
48
20.04
LP FG header to PSV
49
39.38
GFU to PSV
No corrosion
45
HP/IP/LP separator to PSV
42.71
20.78
Pit depth: 8.26mm General corrosion (grid): A: 80 x 230mm D: 8.00-8.41mm
Pit depth: 8.19mm General corrosion (grid): A: 10 x 12mm D: 7.61mm
General corrosion (grid): A: 290 x 960mm D: 9.11mm
General corrosion (grid): A: 90 x 140mm D: 8.39-8.47mm
Fig. 2 UT scans indicating location and size of erosion–corrosion at elbows (grid)
123
974
J Fail. Anal. and Preven. (2017) 17:971–978
General corrosion (grid): A: 320 x 80mm D: 4.51mm
Figs. 3 and 4 Schematic showing anomalies in lines from GFU to PSV and from PSV to flare. UT scan showing size of anomaly in elbow from GFU to PSV
Fig. 7 UT scan showing general corrosion with isolated pitting in LP FG header to PSV
Fig. 5 DCS output showing historic intermittent blowdown from LP separator to flare
Fig. 6 UT scan showing general corrosion with isolated pitting in LP separator to flare line. Corrosive water condenses and drips down the pipe wall thereby causing the observed corrosion
temperature. Therefore, prior to reaching the glycol contactor, condensation is expected to occur within the flowing piping.
123
Corrosion rate increases with increase in gas velocity, liquid velocity and CO2 partial pressure [4]. It will also tend to increase exponentially with temperature, the trend being halted after the threshold ‘scaling temperature’ is reached, i.e., the temperature above which protective iron carbonate films form [5]. At temperatures below approx. 60 C, protective films are not formed due to the high solubility of FeCO3. At temperatures between approx. 60 C and up to 80 C, a protective and more adherent layer is formed resulting in a decrease in the general corrosion rate. Above typically 80 C, the solubility of iron carbonate decreases markedly and the resultant high supersaturation leads to the formation of a very dense and protective iron carbonate layer. Since most of the wet gas system operate at temperatures below 62 C where the protective FeCO3 film will not be expected to form, temperature-dependent general corrosion should be expected to be dominant. The mostly low-severity corrosion reported for the flowing sections of Train 1 and Train 2 may indicate moderate pH of the gas fluid, thereby counteracting the lack of a protective FeCO3 film at the operating temperatures (\62 C) reported in the ‘Description of System’ section above.
J Fail. Anal. and Preven. (2017) 17:971–978
975
•
• Fig. 8 UT scan showing general corrosion with isolated pitting in LP FG header to PSV
•
•
the LP separator and flowing through complex-geometry elbows thereby resulting in erosion–corrosion of elbows and pipework immediately downstream d/s (Fig. 2). Turbulence at the complex-geometry elbows d/s of the LP separator increasing condensation which drips down the pipe wall resulting in the observed corrosion (Fig. 2). CaN (calcium naphthenate) deposition in the LP separator resulting in insufficient fluid retention time and inadequate water/gas separation, hence increased wetness of carried-over wet gas. Corrosive water present in wet gas plus heat associated with the compression process resulting in CO2 corrosion and hence wall loss in the compressor to cooler lines. Insufficient design corrosion allowance at current corrosion rate resulting in up to 46.38% WL.
Mitigation: Wet Gas Flowing System • Fig. 9 UT scan showing general corrosion with isolated pitting in 1st stage HP Compressor cooler to PSE
Root Causes: Wet Gas Flowing System No corrosion anomaly was reported in Train 1. The observed 46.38% wall loss (max) reported in Train 2 of the piping (flowing) was caused by: •
Combination of CO2 corrosion due to corrosive condensate and erosion due to wet gas emerging from
•
•
Consider possibility of solids removal/cleanup of LP separator Train 2 in order to improve fluid retention and hence separation efficiencies. Determine entrant velocity that will reduce turbulence and hence erosion at elbows. Consequently, implement velocity control of produced wet gas though the complex-geometry elbows downstream of LP separators. If Train 2 anomalies persist, then consider batch-wise injection of gas-phase corrosion inhibitor/pH neutralizers at locations experiencing condensation and
Fig. 10 Root cause failure analysis—overall mitigation diagram
123
976
J Fail. Anal. and Preven. (2017) 17:971–978
dripping. The objective is to increase pH to 6.5–7.5 thereby facilitating the formation of a protective iron carbonate film on the steel surface and consequently reducing corrosion of the steel [6]. Note that inhibitor injection may not be very effective at elbows due to the turbulent regime leaving velocity control as the most viable mitigation option.
FG header to the PSV. As explained in the previous section, these anomalies are caused by corrosive condensate dripping down the pipe wall and resulting in CO2 corrosion.
Root Causes: Wet Gas Dead leg System The observed anomalies were caused by: •
Wet Gas Dead leg System The wet gas dead leg system consists of lines transporting wet gas from the scrubbers, coolers, GFUs, HP separator, IP separator, LP separator to the PSVs and PSEs (pressure safety element) and onward to the flare header as required. The reported wall loses for these lines are shown in Table 2 with the max wall loss being 49% in the line from the LP
Operaonal Failure
Corrosion Failure /Anomaly
•
Corrosive condensate (in wet gas lines from scrubbers to PSV, coolers to PSE and GFU to PSV) dripping down the pipe wall and resulting in dead leg (CO2) corrosion (Figs. 4, 5, 6, 7, 8, and 9). GFU to PSV to flare—PSV lifting and/or infrequent flow through middle bypass valve to flare resulting in corrosive condensate accumulating in the PSV to flare lines hence dead leg (CO2) corrosion (Figs. 3 and 4).
Changes in plant
Design Failure
Inadequate corrosion allowance & design detail
Migaon Failure
Chemical injecon failure
Monitoring failure
Fig. 11 Root cause failure analysis—results
123
-D/L corrosion due to PSV passing and infrequent blowdown to flare. -Velocies d/s of LP Sep. -CaN deposion in LP Sep
-D/L CA likely to be consumed by end of design life, at current corrosion rate. -Complex elbow geometries esp d/s of LP Sep
pH neutralizer not injected u/s of coolers to buffer low pH of condensate
-Non-sampling of condensate pH. -Non draining of D/L locaons . -Inspecon triggered when 0.7CA exceeded. -Probe results not acted upon.
J Fail. Anal. and Preven. (2017) 17:971–978
•
•
•
Severity of dead leg corrosion strongly correlating with infrequent blowdown to flare (Figs. 5 and 6); valve condition; pipe elevation; and wall temperatures (condensation). Design corrosion allowance proving to be insufficient when measured against the current corrosion rate (up to 48.58% WL at dead leg line to PSV and assuming elbow WT = pipe WT). Infrequent transportation of removed condensate from the discharge cooler to closed drains lines and from the discharge scrubber to IP separators lines.
Mitigation: Wet Gas Dead leg System •
•
•
•
•
The current inspection frequency should be maintained at once every 2 years on the basis of the estimated time to leak being 9–11 years (i.e., for all locations, at current corrosion rate). Reported UT scan (Fig. 7) indicates that a pinhole leak rather than a rupture will be the likely mode of failure. The SHE consequence severity will therefore be in the lower range. Although this line is located near an ignition source (flare), the chance of people being in the vicinity during a failure event is low. Inspection of all cooler/scrubber lines to PSV should immediately be triggered if probe monitoring results show that 0.7 CA factor of safety has been exceeded. If process conditions change and corrosion rate increases, consider installing injection point for weekly injection of gas-phase corrosion inhibitor into line to the PSV. The appropriate corrrosion inhibitor should be selected in consultation with the chemical vendor. For the GFU line from PSV to flare, check whether PSV is lifting and replace the valve if found to be defective. For GFU line from PSV to flare, check whether flow is intentional and consider reducing frequency of flow through the middle bypass valve to flare; see Fig. 3.
977
CI,velocity control and reviewing inspection frequencies. Each node is then processed (e.g., evaluating for condensation) in order to obtain a value which is then multiplied by an assigned weight wi and then summed to give a nodal output Rwixi. Each assigned weight wi is proportional to the contribution of the respective node (e.g., condensation) on the corrosion rate and integrity of the specific piping system being evaluated. The system should then be trained by entering input data in backpropagation mode and within the desired range. This is followed by processing as described above to obtain an output which in turn is used to adjust the weights in accordance with the desired range, thereby changing the corrosion rate value and integrity status of the specific system. When further data inputs are then made, the network operates in forward propagation mode and will not require further training. The ability to intrinsically alter processor weights makes the model to be dynamic and responsive. Repeating this process for each root cause and mitigation node (Fig. 10) makes a significant contribution to development of a database necessary for predicting corrosion behavior and hence the integrity of wet gas system. The next step is to develop a neural network predicting software capable of executing all the features proposed above.
Conclusions and Recommendations The root causes of the reported anomalies are summarized in Fig. 11 using the Energy Institute model [7]. Wet Gas (Flowing) 1.
2. Neural Network Predictive Modeling An outline of the proposed neural network predictive model for the wet gas system is discussed. For a neural network prediction model, output y equals the sum of inputs xi multiplied by the weights wi, i.e., y = Rwixi. For example, in order to design a neural network model for the LP separator to glycol contactor piping (Fig. 10), the inputs xi will consist of each of the causes and mitigation (also referred to as nodes), i.e., condensation, turbulence, CaN deposition, insufficient corrosion allowance, injecting
3.
Corrosion anomalies reported in the wet gas piping (flowing) is as expected mainly within the low and medium severities, i.e., \50%. The morphologies revealed by UT scan show general CO2 corrosion over a wide area with occasional spots indication localized CO2 corrosion. Batch injection of gas-phase corrosion inhibitor/pH neutralizers should be considered at susceptible locations. The root cause of the anomalies is due to CaN and solid deposition in the LP separator. Increasing amounts of corrosive condensate consequently drip down the pipe walls causing corrosion. Solids removal/cleanup is recommended in order to improve fluid retention and hence separation (gas/water) efficiencies. Corrosion at elbows immediately downstream of the LP separator is caused by the turbulent flow of emergent fluid streams through complex-geometry elbows. Velocities of produced wet gas through these elbows should be controlled.
123
978
4.
J Fail. Anal. and Preven. (2017) 17:971–978
Corrosion in the compressor to cooler lines is due to increased condensate and heat from compression.
Wet Gas (Dead leg) 5.
6.
7.
The wet gas dead leg system consists of the lines transporting wet gas from the scrubbers, coolers, GFUs, HP separator, IP separator, LP separator to the PSVs and PSEs and onwards to the flare header as required. Similarly, most of the reported anomalies have less than 50% wall loss. The root cause (for lines from scrubbers to PSV, coolers to PSE and GFU to PSV) is due to condensed, corrosive, uninhibited water dripping down the pipe wall and resulting in dead leg (CO2) corrosion. In mitigation, gas-phase corrosion inhibitor should be considered. The root cause of dead leg (CO2) corrosion in the GFU to PSV to flare lines is likely due to lifting of the PSV and/or infrequent flow through middle bypass valve directly to flare. The severity of dead leg corrosion also strongly correlate with pipe elevation and wall temperatures.
123
8.
The root cause of corrosion in the discharge cooler to closed drains lines and in the discharge scrubber to IP separators lines is due to infrequent transportation of removed, condensed and corrosive water.
References 1. S. Fengmei, A model developed to predict the internal corrosion rates of wet and dry gas pipelines, in CORROSION 2011, 13–17 March, Houston, Texas (2011) 2. R. Nyborg, CO2 corrosion models for oil and gas production systems, in CORROSION 2010, 14–18 March, San Antonio, Texas, NACE-10371 (2010) 3. R. Nyborg, Top of line corrosion and water condensation rates in wet gas pipelines, in CORROSION 2007, 11–15 March, Nashville, Tennessee (2007) 4. Y.H. Sun et al., Corrosion under wet gas conditions, in CORROSION 2001, Paper No. 01034 (2001) 5. Y.H. Sun et al., CO2 corrosion in wet gas pipelines at elevated temperature, in NACE Conference Paper—2002, Paper No 02281 (2002) 6. M. Swidzinski et al., Corrosion inhibition of wet gas pipelines under high gas and liquid velocities, in CORROSION 2000, Paper No 00070 (2000) 7. Guidance for corrosion management in oil and gas production and processing. Energy Institute, Oil & Gas UK (2008)