February 2018

Maintenance and Reliability

Eliminate mechanical seal failure in a crude distillation unit

Multiple mechanical seal failures in a crude distillation unit (CDU) resulted in total losses of $3 MM in a refinery since its startup in January 2015. The maximum seal life achieved did not exceed 6 mos, which was much shorter than the American Petroleum Institute (API) 682 “Advancements in Mechanical Sealing” goal of 3 yr of seal life.

Telmesani, A. A., Yanbu Aramco Sinopec Refining Company (YASREF) Ltd.

Multiple mechanical seal failures in a crude distillation unit (CDU) resulted in total losses of $3 MM in a refinery since its startup in January 2015. The maximum seal life achieved did not exceed 6 mos, which was much shorter than the American Petroleum Institute (API) 682 “Advancements in Mechanical Sealing” goal of 3 yr of seal life.

The root causes of failure for the mechanical seals for 16 hot hydrocarbon service pumps (with operating temperatures in excess of 250°C) are discussed here, and solutions are demonstrated to eliminate failures and improve plant reliability.

CDU hot pumps are equipped with API seal flushing Plan 32 (injection to seal chamber from an external source) and API seal flushing Plan 54 (pressurized external barrier fluid). API Plan 32 utilizes heavy vacuum gasoil (HVGO) that is supplied at 1.7 barg above the seal chamber pressure at 110°C. API Plan 54 utilizes light vacuum gasoil (LVGO) that is supplied at 1.7 barg greater than that of Plan 32 pressure at 110°C. For example, if the seal chamber pressure is 5 barg, Plan 32 should be supplied at 6.7 barg (5 + 1.7), and Plan 54 should be at 8.4 barg (6.7 + 1.7).

The flowrate for each API plan varies, depending on the pump service. The HVGO utilized for Plan 32 is produced from the CDU and supplied to the CDU hot pumps by a common system, as shown in FIG. 1. The LVGO utilized for Plan 54 is produced from the CDU and supplied to the CDU hot pumps by a common closed system (FIG. 2).

Fig. 1. The HVGO utilized for Plan 32 is produced from the CDU and supplied to the CDU hot pumps by a common system.
Fig. 1. The HVGO utilized for Plan 32 is produced from the CDU and supplied to the CDU hot pumps by a common system.
Fig. 2. The LVGO utilized for Plan 54 is produced from the CDU and supplied to the CDU hot pumps by a common closed system.
Fig. 2. The LVGO utilized for Plan 54 is produced from the CDU and supplied to the CDU hot pumps by a common closed system.

Mechanical seal failure data was collected from a computerized maintenance management system (CMSS), and selection criteria were set to concentrate the focus on the “bad actor” pumps. Selection criteria included an operating temperature that exceeded 250°C. The seal failure history is indicated in TABLE 1, which shows a total of 41 seal failures that occurred in 2015 vs. 25 failures in 2016. The large discrepancy in failures between 2015 and 2016 was due to a startup and commissioning period in early 2015, during which time the refinery was suffering from system dirt, operational upsets and various commissioning problems. While seal failures were reduced in 2016, equipment mean time between failure (MTBF) and seal life were unsatisfactory for refinery maintenance key performance indicators (KPIs) and targets.

Root cause analysis

A root cause analysis (RCA) was conducted in April 2016 to identify the root cause of failure for CDU bad actor pumps to eliminate seal failures and enhance equipment reliability. This RCA evaluated all possible seal failure causes including, but not limited to, the following major possible causes:

  • Piping stress
  • Plan 32 barrier fluid (HVGO) incompatibility
  • Plan 54 barrier fluid (LVGO) incompatibility
  • Seal cartridge installation.

Almost 90% of the dismantled failed mechanical seals experienced either coke formation or sludge in the outboard seal side (secondary seal provided with Plan 54 barrier fluid). Most of the findings showed coke particles trapped on the rotating face along the outboard seal side and bellows, as shown in FIGS. 3 and 4.

Figs. 3 and 4. Almost 90% of the dismantled failed mechanical seals experienced either coke formation or sludge in the outboard seal side and bellows.
Figs. 3 and 4. Almost 90% of the dismantled failed mechanical seals experienced either coke formation or sludge in the outboard seal side and bellows.

The RCA team also reviewed the design basis and barrier fluid (API Plan 54) selection criteria. It was concluded that LVGO is not a compatible seal barrier fluid, as the most desirable viscosity range for any hydrocarbon seal barrier is between 2 cSt and 10 cSt in accordance with API 682 (shaft sealing systems for centrifugal and rotary pumps). To achieve this viscosity range, the operating temperature for the LVGO was increased during the design phase of the refinery to 110°C, which contradicts the refinery standard that limits the barrier fluid for Plan 54 to a temperature of 70°C.

In addition, barrier fluid selection was reviewed for other process units within the refinery that contain pumps with similar operating parameters and that utilize API seal Plan 32 and Plan 54, such as hydrocracking (HCK) units and delayed coker units (DCUs). Similar pumps within HCK units utilize diesel for both API plans at 50°C–60°C, whereas DCU pumps utilize heavy coker gasoil (HCGO) at 60°C and light coker gasoil (LCGO) at 60°C for Plan 32 and Plan 54, respectively. Seal failures for both units, along with failures modes, were also evaluated and were found to be satisfactory.

The RCA concluded that the direct cause of the mechanical seal failures was due to coke formation in the outboard seal (atmospheric side), which impacted the seal bellows and caused the seal faces to open. It was also concluded that the root cause of failure was due to the improper selection of API seal Plan 54 (LVGO), as this type of hydrocarbon has been proven to form coke at 110°C on the atmospheric side of the seal.

The effect of coke formation in mechanical seals can be understood from a mechanical seal balance ratio. Mechanical seal vendors design the seal face with a balance ratio to minimize heat generation between seal faces. This seal ratio also impacts seal faces opening and closing forces. The seal balance ratio normally ranges from 0.6–0.9.

Seal balance ratio, B, is defined by Eq. 1:

B = (Do2Db2) / (Do2Di2)        (1)

where:

Do = The seal face contact outer diameter
Di = The seal face contact inner diameter
Db = The effective seal balance diameter.

The term (Do2Bb2) represents the closing area, or closing force, while (Do2Di2) represents the opening area, or opening force.

When coke forms and solidifies on the seal faces, the seal outer diameter will increase, resulting in a higher opening force that, in turn, leads to excessive leakage, as shown in FIGS. 5 and 6.

Figs. 5 (top) and 6 (bottom). Coke forming and solidifying on the seal faces leads to excessive leakage.
Figs. 5 (top) and 6 (bottom). Coke forming and solidifying on the seal faces leads to excessive leakage.

Recommendations

The RCA recommended changing the barrier fluid from LVGO to diesel (supplied from the CDU atmospheric column). The API seal Plan 54 common pumps (FIG. 2) were evaluated to accommodate this change, and it was found that diesel was unsuitable for these pumps at high temperatures—the pumps were designed for LVGO, which has a relatively higher viscosity than diesel. To overcome this problem, it was decided to cool the diesel to 50°C to increase fluid viscosity.

Design modifications were made (FIG. 7) by rerouting the coolers piping upstream of the diesel transfer pump to drop the temperature from 140°C to 55°C.

Fig. 7. Design modifications prompted a reroute of the coolers piping upstream of the diesel transfer pump.
Fig. 7. Design modifications prompted a reroute of the coolers piping upstream of the diesel transfer pump.

After the refinery turnaround, the CDU bad actor pumps were monitored closely for 13 mos following the refinery startup in December 2016. Seal failures were determined to be almost entirely eliminated, and seal life was increased. TABLE 2 shows the total number of seal failures since diesel was utilized with API Plan 54. The number of failures was reduced drastically to two, compared to the same period in 2016. For example, pump 110-G-0016 C (which was considered as the most troublesome CDU bad actor pump) had zero failures compared to eight failures since the refinery startup. Although some failures were recorded since diesel was utilized as barrier fluid, it is strongly believed that these failures were due to remaining LVGO traces in the seal internals.

In addition, the RCA identified areas of improvement with regard to quality assurance and quality control procedures, which had been found to be less than adequate. The RCA team developed mechanical seal replacement procedures that focused on improving the following important aspects in accordance with API 686 “Recommended Practices for Machinery Installation” and equipment OEM manuals:

  • The shaft alignment offset must be less than 0.05 mm
  • The motor magnetic center must be verified and adjusted as per motor OEM manual specifications
  • The pump total axial float must be verified and adjusted as per pump OEM manual specifications
  • The distance between shaft coupling will be verified and adjusted as per the coupling drawing. HP

The Author

Related Articles

From the Archive

Comments