Lower heat exchanger maintenance costs

Copying and distributing are prohibited without permission of the publisher

Print Article
Share
- Digg.com
- Technorati.com
- Del.icio.us
- Stumbleupon.com
- Reddit.com
- Fark.com
- Facebook
- Yahoo!
- Twitter
- LinkedIn
Add Comment

03.01.2002 | Andrews, S., Marathon Ashland Petroleum, LLC, Roanoke, Virginia; Matusheski, B., Meridium, Inc., Roanoke, Virginia

Failure analysis was used to predict replacement times, saving millions of dollars per year

Keywords:

One refinery saved nearly $3 million lost opportunity costs in one year by avoiding heat exchanger failures. Similar savings are realized in later years. The methods used modeled the common failure behaviors of the exchangers and then used these models to predict failures. Armed with this information, refinery managers then authorized tube bundle replacements and avoided the expected future failures of these bundles. Here's a description of the methods used to make the decisions that led to the savings.

The reliability group at the Marathon Ashland Petroleum, LLC (MAPLLC) plant conducted the analysis using an enterprise reliability management system. The software facilitated the analysis and provided a computing environment that makes the analysis easy to update when new information becomes available.

In addition to implementing the software, MAPLLC management realized the need to establish a reliability culture at the plant. This shift was driven by the desire for the plant to become a top performer. Market pressures for the products produced at the facility drove management to achieve higher and higher levels of availability and reliability. Today, the plant using this procedure for managing equipment failures is among the top performers in the refining industry.

The motivation to improve reliability was not simply to reduce maintenance costs. Plant management recognizes the need to control reliability for best economic performance. Indeed, least-cost maintenance is not recommended when a plant strives to achieve high reliability. Maintenance costs are relatively small when compared with lost opportunity costs. While these maintenance costs are small, they can have a huge impact on the plant's ability to produce the right products at the right time to achieve high margins.

Availability to produce certain products at certain times is key to capturing high-margin production. The successful plant has control of equipment reliability.

This control stems from an intimate knowledge of the components that will perform reliably during high-demand periods.

Heat exchanger reliability program. Early efforts used to improve heat exchanger reliability were based on procedures developed at Lyondell Chemical Company.¹ While this approach was helpful for aging components, it did not provide sufficient guidance for analyzing relatively new components. Clearly another approach was needed.

The reliability group at the MAPLLC plant began to look at using reliability engineering principles and applying the results to decide which tube bundles to replace.

The heat exchanger reliability effort was further supported by MAPLLC's reliability engineering advisory group (REAG). This group specified that the scope of the reliability analysis should review:

Total number of failures
Years of service
MTBF
Metallurgy
Current bundle age
Fluid velocity
Ops code
Date installed
Date last replaced.

From this, the REAG recommended developing and using:

Tube age factor table
Remaining life factor table
Production criticality factor
Service conditions
Other considerations
- Corrosion mechanism
Analysis value
Bundle replacement costs.

Reliability analysis approach. The following steps are recommended for conducting reliability analysis of heat exchanger bundles.

Step 1. Gather specific bundle failure history. This history is contained in a variety of sources, including:

PMIC - MAPLLC maintenance management system
Inspection records
Turnaround report
Engineering files
Daily ops reports.

Once these data have been collected, the activities need to be organized into chronological order. An example of such a history is shown in Table 1.

Table 1. Heat exchanger summary history for 1E-1A crude exchanger
12/01/68	Complete new exchanger CS tubes
05/01/71	Bun'l retubed CS (eros'n and pit'g under scale); shell-mod attact/ no loss
06/01/74	Bun'l-minor OD pit'g/ no loss; shell-minor pit'g/ GS's ok
04/01/75	Replc'd w/ used 1E-1B bundle w/ 27 new tubes; shell-new deflector plt
10/01/75	New bun'l CS (0.05in. OD loss)/ rerol'd tubes on back end and seal welded 16 tubes
10/01/76	One tube leaked (plug'd 30 tubes)
03/01/77	Tube leaked; replc'd w/ spare CS bun'l (eros'n/corr)
09/01/78	Tube leaked; plug'd 58 tubes
04/01/80	Bun'l leaked; replc'd w/ retubed CS; shell-0.15in. depress'n @ baf'l locat'ns
02/01/81	Tube leaked/ plug'g 52 tubes
05/01/81	2 tubes leaked/ plug'g 11 more tubes for total of 63 plug'g
08/01/81	2 tubes leaked/ plug'g 16 more tubes for total of 79 plug'g
10/01/81	Replc'd w/ retubed CS (under deposit corr); shell-fill welded erod'd > 0.12in.
03/01/82	Tube leaked/ plug'g 126 tubes for "buffer"@ top @ sides of bund'l
06/01/82	Replc'd w/ retubed CS (corr in fouled areas); shell-eros'n @ baf'l areas
01/01/83	Tube leaked/ plug'g 128 tubes for "buffer"@ top @ sides of bund'l
05/01/83	Replc'd w/ retubed CS (5/64in. OD loss); shell-0.13in. pit'g/ 0.06in. recess@baf'l
11/01/83	Tubes leaked/ plug'g 184 tubes for "buffer"@ top @ sides of bund'l
04/01/84	Retubed CS; shell-0.08in. pit'g/ 0.05in. eros'n @ bafl's/ vert'l striations
01/01/85	Tubes leaked/ plug'g 138 tubes for "buffer" @ top @ sides of bund'l
06/01/85	Retubed CS (corr in fouled areas)
09/01/85	New titanium bundle
09/01/86	Pul'd bundle out 2ft only; titanium bund'l-good cond
09/01/87	Shell leaked; UT'ed shell and found loss in 1in. X 6in. area (lap patached area)
10/01/87	New bundle (1301 tubes: 3/4in. X 20ft X 20 BWG SB-338 Gr-12)
04/01/88	Bund'l-no ID or OD corr; shell-mild corr/ no loss/ noz's good
10/01/89	Extrn'l plus UT survey (0.23in. localized loss) no other problems found
01/01/91	UT survey (0.41in. localized loss in 1in. X 10in. area)
02/01/91	One tube leaked (plug'd); shell flange weld leaked (lace welded over hole)*
03/01/91	UT survey-no repairs
08/01/91	Bun'l-tubes good cond/ front TS corr'd 0.09in./ spacer failed; new shell
07/01/92	UT survey-no repairs
06/01/93	UT survey-no repairs
01/01/94	UT survey-no repairs/ 0.50in. max loss on shell
03/01/94	Bun'l-tubes like new/ cage corr; shell-0.625in. grooves @ bafl's (fill welded)/ eroded area @ front fill welded (min wall= 0.50in.)
11/01/94	Bun'l-tubes like new/ cage corr; shell-0.25in. grooves @ bafl's/genr'l corr 0.125in. in upper shell
04/01/09	Plugged (1) leaking tube tubes like new/tie-rod spacers corr/shell corr 0.3125in. @ baffles and inlet *
12/01/97	Tube leaked/plugged one tube*
08/01/00	Hole in shell adjacent to the inlet nozzle; installed new CS shell and shell cover-plugged (52) tubes due to possible mechanical damage was bundle was installed in 1997 and extracted at this time; (1) tube in the middle of the bundle leaked during hydrotest and was plugged*
01/30/01	Plugged (2) leaking tubes; both leaks were in the middle of the bundle*
02/02/01	Plugged (2) more leaking tubes; these were also in the middle of the bundle*

* Used in Weibull analysis.







Fig. 1. Weibull plot of failures for a single bundle.

Step 2. Conduct Weibull analyses on the failure data. This yields an estimate of MTBF, but, more importantly, calculates parameters that can be used to estimate future probability of failure (Fig. 1). The challenge is to determine, based on Weibull parameters, if the exchanger will survive until the next scheduled outage. If not, is it more cost-effective to replace the bundle now or should we wait until the next outage? Can we wait until the following outage?

Step 3. Use the failure probability calculator to estimate current probability of failure. Critical to this analysis is understanding the ops code for the heat exchanger. For heat exchangers that are rated with ops code "A," bundle replacement is recommended whenever reliability falls below 50%. Ops code A requires a total unit shutdown. If the unit feeds other units downstream, then these units may be shut down as well. Ops code B requires production cutback by a certain percentage.

In certain circumstances, flowrates can be increased in other parts of the unit so that production losses are minimized. For heat exchangers that carry ops code C, no impact on production is seen when these heat exchangers fail. Reliability analysis is generally not conducted on these heat exchangers because the most cost-effective method for maintenance of these types of heat exchangers is "run-to-fail."

Heat exchanger reliability analysis. The way in which a Weibull analysis is conducted is affected by both the maintenance activities that ensue after a failure and the failure mechanisms at work. For tube failures that occur because of process corrosion, failures are expected to reveal a "wear-out" pattern indicated by a Weibull beta value greater than 1. If the Weibull analysis results in beta less than 1, the heat exchanger is failing due to lack of quality or some other process that occurred during manufacturing.

If these failures occurred early in the equipment life, then the tube bundle should be considered for replacement at the next available opportunity. The reason for this is that to control heat exchanger reliability, we must be able to accurately predict probability of failure. Infant mortality failures are difficult to predict because they are not time-dependent. If the cause of the infant mortality problems can be ascertained and localized to one area of the tube bundle, then the bundle may be repaired and put back in service.







Fig. 2. Failure probability calculator.

The aging process begins when the heat exchanger is first installed. In most cases, chemical corrosion due to contact with the process flow is a major source of tube degradation in refinery heat exchangers. Other causes of tube failure are erosion due to flow and caking of product or byproducts left from cooling water evaporation within the heat exchanger.

For heat exchangers that are used in the process, before correcting problems such as tube leaks, maintenance personnel must first isolate the heat exchanger from the rest of the unit. The exchanger must be allowed to cool before disassembly. Once disassembled, the exchanger must be cleaned so that a detailed inspection and location of the leaking tube can be done.

Conceptually, heat exchangers can be broken down into the constituent parts. Each tube should be looked at as a separate entity. Other component parts that can be identified are the shell and flanged connections. The approach for estimating tube bundle reliability should account for the age of each individual tub that failed. Table 2 shows how this is done. Each tube lifetime is estimated from the installed date of the bundle to the failure date, at which time the failed tube is plugged.

Table 2. Failure data
Entity ID	Event date	Event type
1E-1A	10/01/87	Startup
1E-1A	02/01/91	Failure
1E-1A-1	10/01/87	Startup
1E-1A-1	04/01/97	Failure
1E-1A-2	10/01/87	Startup
1E-1A-2	12/01/97	Failure
1E-1A-3	10/01/87	Startup
1E-1A-3	08/01/00	Failure
1E-1A-4	10/01/87	Startup
1E-1A-4	01/30/01	Failure
1E-1A-5	10/01/87	Startup
1E-1A-5	02/02/01	Failure

Cost benefit analysis. Traditional approaches to understanding the benefit associated with a mitigating action based on data analysis requires that the analyst account for costs associated with the failures that were mitigated and costs for the mitigating maintenance procedures. Difference between the totals of the two values estimated above is known as the benefit differential. This differential is the true estimate of savings realized from the proactive maintenance. This approach was used to estimate benefits for this example. Table 3 shows estimated costs and avoided lost opportunity (savings) for a group of tube bundles scheduled for replacement.

Table 3. Bundle replacements - 2000 shutdown
Heat exchanger ID	Description	Ops code	Bundle replacement costs	Estimated lost opportunity cost*
2E-2B	Condenser	B	$48,660	$162,000
2E-2D	Condenser	B	48,660	162,000
2E-28	Vacuum cooler	A	23,416	590,000
8E-9 ABCD	Alkylate coolers	B	2,790	23,788
8E-14A	O/H condenser	A	30,590	94, 816
8E-14B	O/H condenser	A	30,590	94,816
8E-14	Reboiler	A	9,490	94,816
8E-14	Reboiler	A	20,225	94,816
8E-23	Dryer heater	A	2,666
8E-31B	Feed bottoms	A	11,685	94,816
83E-6A	Main column condenser	A	25,475	323,840
83E-6C	Main column condenser	A	25,475	323,840
83-6D	Main column condenser	A	25,475	323,840
83-19A	Feedwater heater	A	7,726	323,840
83E-43	Pump coolers	B	1,755
84E-8A	Feed subcoolers	A	27,466	40,465
84E-8B	Feed subcoolers	A	27,466	40,465
84E-11	Upper reboiler	A	12,490	40,465
84E-12	Lower reboiler	A	12,490	40,465
84E-37A	Reboiler	B	35,655	35,185
84E-37B	Reboiler	B	35,655	35,185
84E-38A	Product cooler	B	6,295	17,571
84E-38B	Product cooler	B	6,295	17,571
		Totals	$478,490	$2,974,600
		Benefit differential		$2,496,110

* Lost opportunity cost is the lost margin (gross profit) from downtime associated with tube failures.

Savings reported here are significant compared with repair costs. However, improved reliability usually comes at a price. The nature of improved reliability requires that parts be renewed or upgraded once the problems have been identified. If no changes are made, no improvement will be realized.

Literature Cited
¹ Gamino, C. and P., and F. Walter, "Shell and Tube Heat Exchanger Reliability Study," NPRA Maintenance Conference, New Orleans, May 1999. ² Rouch, M. L. and W. Willie, Applied Reliability Engineering, Volume I, Beacon Printing, Waldorf, Maryland, 1999. ³ Matusheski, R. and Holman, M., Predictive Maintenance Guidelines, EPRI TR 103374, August 1994.

Have your say

All comments are subject to editorial review.
All fields are compulsory.
Name: Please enter your name
Email address (not published): Please enter your email Please enter your email
Comments: Please enter a comment