Вы находитесь на странице: 1из 3

PROCESS PUZZLER

Tackle Exchanger Flaws


A variety of issues may contribute to ongoing problems

CHECK A VARIETY OF ITEMS tion of rapid condensation due to THIS MONTHS


Although data are sparse, there are
several items you could look into.
the excess surface area: condensate
builds up in the shell too close to PUZZLER
Temperature gradient. Is the the inlet steam source. Our makeup-air heat exchangers seem
steam superheated or at saturation? Tube supports/baffles/tube thick- to have suffered ruptured tubes. On
Steam at 42 psig has a saturation ness. If the tubes are not prop- the shell side, we use 45-psig steam
temperature of ~289F. Since you erly supported, large temperature reduced from 200-psi boilers. We send
did not mention what the inlet gradients could cause failures along 40% propylene glycol/water through
temperature of the glycol-water the outer surface of the tubes, the tubes; at the steam control valve
mixture is, if you are using a especially at the weakest points, inlet with a 50% load its 42 psig. The
heating media too hot and the which may be the rolled connec- exchangers are horizontal U-tubes
temperature gradient (difference) tions. Have someone experienced with -in. 0.049-in. (16 BWG) cop-
is high, this may be causing addi- in mechanical exchanger design per tubes. The tubes were rolled into
tional stress on the tubes and caus- count the tube supports and check a 316L stainless steel tube sheet and
ing premature failure. I suggest baffle count and spacing to be sure sealed with fluorocarbon gaskets. Both
measuring the temperature differ- they are adequate for the service. If exchangers are 24-in. diameter and 6-ft
ence between the glycol and steam you decide to change the heating long. The exchangers are about 50%
and having a mechanical engineer medium, the baffle count and spac- oversized sometimes while running
determine if there is too much ing may need to change to account about 105% during the winter months.
stress on the tubes, especially at for heat transfer and the potential We run one at a time. Another concern
the rolled connection at the tube resulting vibration. Your tubes is that the tube-side relief valve was
sheet. Dissimilar metals cop- are thin. Have you considered a sized using the old 77% rule is that
per tubes and stainless tube sheet thicker tube wall? Not knowing okay? I think a classic case of water
could also be a concern when your flow, could a thicker tube hammer causes the crushing of a
it comes to high temperature dif- provide the protection you need? number of tubes at the top of the tube
ferences between the fluids. Check Velocity and pressure drop are the bundle that we see. When we operate
the mean metal temperature to keys to determining if you could at only 50%, theres a thermal reservoir
be sure you have not exceeded the afford a thicker tube. in the shell that pulls a vacuum; another
vendors recommendations. Outlet glycol temperature. What engineer believes thats the culprit.
If the steam is superheated, is the outlet glycol/water tempera- What do you think?
the thermal expansion stresses are ture? How is it being controlled?
more significant. You also mention Make sure you have not exceeded
you have significant over surface the boiling point temperature of
availability in the exchanger. the glycol mixture. (Propylene gly-
Condensing steam has a high heat col, 40 wt-%, boils at ~219220F
transfer coefficient, ~ 1,500 Btu/ at 1 atm pressure.) You may want
hr-ft 2-F. I would assume that the to check this because the steam
tube-side coefficient is controlling condensing temperature is above
the overall heat transfer coef- the atmospheric boiling point of
ficient. You may want to consider the glycol mixture. Vaporizing in
using a different heating fluid that the tubes will account for addition-
balances the resistances better. Is al stresses that most likely will not
there a waste heat stream available have been addressed in the original
that could be used instead of the design due to assuming no phase
steam? This could also help mini- change on the glycol side.
mize stresses induced due to high Thermal reservoir. Are you con-
temperature gradients. densing steam at 42 psig or higher?
The hammering may be a func- Im not sure why you are pulling a

39  CHEMICALPROCESSING.COM JUNE 2015


PROCESS PUZZLER

AUGUSTS
PUZZLER
We would like to increase the cooling water flow to a tower con- point of the pump? Do we really need the balancing orifices?
denser (heat exchanger A in Figure 1) from 2,900 gpm to the ex- Send us your comments, suggestions or solutions for this
changer nameplate rating of 4,500 gpm. A review of the files shows question by July 10, 2015. Well include as many of them as pos-
the heat exchangers and control valves have plenty of capacity. In sible in the August 2015 issue and all on ChemicalProcessing.
fact, we are concerned that so many of our heat exchangers are com. Send visuals a sketch is fine. E-mail us at ProcessPuz-
oversized. Were also looking at ways to reduce the recirculation zler@putman.net or mail to Process Puzzler, Chemical Process-
flow and pressure losses caused by the balancing orifices we ing, 1501 E. Woodfield Rd., Suite 400N, Schaumburg, IL 60173.
may be able to re-use the pump if we can eliminate some of this Fax: (630) 467-1120. Please include your name, title, location
flow and the pressure losses caused by the balancing orifices. and company affiliation in the response. And, of course, if you
Do you think we can increase heat exchanger A to its nominal have a process problem youd like to pose to our readers, send
flow rate? What will this entail? Will this affect the best operating it along and well be pleased to consider it for publication.

REFINERY COOLING WATER SYSTEM


32 in.

Distance from pump outlet 2,900 gpm


A 4,500 gpm @ 5 psig
HX-A: 3,300 ft HX-G: 2,720 ft
HX-B: 4,850 ft HX-H: 5,340 ft
HX-C: 4,990 ft HX-J: 5,840 ft 8 in. 28% 2,600 gpm
HX-D: 6,358 ft HX-K: 6,050 ft
B 4,000 gpm @ 5 psig
HX-E: 6,340 ft HX-L: 6,090 ft 18 in.
HX-F: 5,510 ft HX-M: 6,850 ft 2,100 gpm 24 in.
6 in. 4 in. 46%
C 2,700 gpm @ 5 psig
39%
32% 440 gpm
RO3 3 in.
3 in. 4in. 1,100 gpm @ 5 psig
25,200 gpm @ D
69%
5.7 psid for 20 in.
36 in. 3 in. 440 gpm 3 in.
E 220 gpm
2 in. F
600 gpm @ 4 psig 800 gpm
89%
3,640 gpm @ 559.3
psid for 4.5 in. 400 gpm @ 5 psig
H 2,000 gpm @ 5 psig
RO-min.
24 in. 39%
8 in. 1,100 gpm
29,000 gpm 32 in.
@ 120 psi J 2,200 gpm @ 5 psig
44%
CSO
900 gpm
36 in. 10 in.
600 gpm K 1,800 gpm @ 5 psig
CSO 59% 44% 2,100 gpm
20 in. G
10 in.
6 in.
58% L 3,200 gpm @ 5 psig
600 gpm
Utilities 40%
6 in. 1,400 gpm
61%
RO1 600 gpm 12 in.
8,000 gpm @ 5.8 RO2
psid for 13 in. 6 in.
59% M 4,500 gpm @ 4 psig 6,300 gpm
600 gpm @ 16.8
32 in.
6 in. 45% psid for
600 gpm 14 in. 9 in.
2,800 gpm @ 6 psig

Figure 1. Plant wants to increase flow to top heat exchanger (A) to rated flow.

vacuum if you are condensing at 42 psig. What is Tube-side relief valve. Im not familiar with the
the chest pressure in the exchanger? If you are pull- 77% rule. If the relief valve is not relieving, Im
ing a vacuum and the exchanger is not designed not sure what part it will play in tube failure. You
for vacuum or you do not have a vacuum breaker may want to check to see if it is leaking by. Im
on the exchanger, this could be creating a problem. not sure how much would have to be leaking by
Perhaps you dont need that much energy/heat to cause a problem.
transfer into the glycol stream. Propylene glycol degradation products. This may

JUNE 2015 CHEMICALPROCESSING.COM  40


be a remote possibility, but propylene glycol, in Now, lets consider the 77% rule. Its a dumb
the presence of O2 and metal, e.g., copper, can rule: you should always check for tube rupture in
degrade at high temperatures into lactic acid. If relief calculations. Besides, given the wide variation
the steam pressure is higher than the glycol pres- in flow rate through the exchanger, it seems doubt-
sure, un-deareated condensate might be entering ful that the rule applies, even in its modern form
the glycol mixture, degrading the glycol into lactic of 10/13. According to API-RP-521, section 5.19.2,
acid. I would check to be sure that you are not tube rupture is not a concern if the low pressure is
leaking condensate into the glycol mixture. within 77% of the high pressure: this is based on
Eric M. Roy, principal engineer the temperature-corrected hydrostatic test pressure.
Westlake Chemical, Houston There is some confusion about which tem-
perature to use, so to be conservative, use the
ADD A SMALL EXCHANGER maximum operating temperature for the hot
There are several problems with this application: fluid. Note the rule applies to older vessels;
1) poor selection of heat exchangers; 2) potential in July 2011, ASME changed the over-pressure
corrosion between the copper tubes and stainless requirements for vessels from 150% of maximum
steel tube sheet; and 3) using the 77% rule for allowable working pressure (MAWP) to 130%.
bypassing consideration of tube rupture in relief Regardless of ASME, you should always look at
calculations. tube rupture because it might reveal design errors
Operating a heat exchanger at 50% of rate in the exchanger.
means that it is oversized. You dont need a spare To get a better idea of the root cause of the
exchanger you need a smaller one to meet de- tube failure, check: 1) the type of temperature
mand when it is low. Operating a heat exchanger control valve employed; 2) the condition of the
at a low rate affects the heat transfer coefficient: glycol; 3) the steam supply; and 4) whether a
h50% ~ 0.57hdesign based on Dittus-Boelter. vacuum breaker is used.
Although copper is about four times more con- Typically, globe valves, with their slowly rising
ductive than stainless steel, it is more susceptible stems, are better at steam control than ball valves.
to rupture because it has a quarter of the allowable The latter have been responsible for numerous
strength of type 316 stainless. Assuming 200-psig water hammer problems. You should replace ball
steam, I estimate a temperature of about 333F valves or add snubbers to their air supply so they
based on superheat tables: http://goo.gl/DZ8flA. close slowly. Also, check the tuning of the control
At this temperature, copper has an allowable stress valves, especially at low flow rates where valves
of about 4,300 psi compared to 16,800 psi for tend to hunt for steady state.
type 316 stainless (CASTIs guidebook on ASME It seems possible that plugging in the tubes
B31.3: http://goo.gl/ZAORy8). One table I found may be a contributing factor to the water ham-
showed a maximum allowable pressure for -in. mer. To see if thats the case, take a sample of
tubing of 470 psi. So, based on a pressure of 55 your glycol system. If there isnt a strainer and
psig for the glycol (based on the 77% rule), the filter, get one.
copper should be fine. Although copper will work- If the steam supply pressure dips, the tempera-
harden, strength wont improve much. And then ture control valve goes to 100% and condensate
there is corrosion. Copper and steel, even stain- literally will drop out of the shell with a potential
less steel, although isolated by a nonconductive for causing water hammer. You should review the
gasket will cause corrosion because the propylene isometrics for the system, model it and decide if
glycol/water is conductive: the stainless steel is you need loops around heavy users on the circuit
the anode, the copper is the cathode the anode to maintain a relatively constant pressure in the
dissolves. Copper exposed to steam condensate exchanger shells.
always poses a corrosion concern. It might be Finally, if you dont have vacuum breakers on
possible for the condensate, and the dissolved CO2 the shells get them. Make sure nothing isolates
always present in it, to corrode one side of the them from the shells and there are procedures
tubesheet while the water causes corrosion on the and valves to vent the air that entered the shells.
tubeside. If the steel in the tubesheet is corrod- Trapped air can reduce heat transfer and cause hot
ing, one of the signs will be the presence of H 2 in spots that could damage tubing.
the glycol, assuming that the glycol is a trapped, Dirk Willard, consultant
recirculating liquid. Wooster, Ohio

41  CHEMICALPROCESSING.COM JUNE 2015

Вам также может понравиться