Diapositiva 1

1
Plate Heat Exchangers in Evaporation
Systems ISHRAE, February 25th Pune, India Dr.
Claes Stenhede
1
2
2
The trend towards indirect refrigeration systems
in large commercial installations such as cold
sto-rages, freezers, supermarkets and air
conditioning plants, often results in only one or
two large eva-porators. These are often built as
thermosiphons, especially if ammonia is used as
refrigerant. The paper describes the correct
installation and ope-ration of thermosiphon
systems using plate heat exchangers and some
pitfalls to avoid.
3
3
1. ?Pfriction 2. ?Pstatic 3. ?Pcomponents 4.
Expansion contraction 5. ?PPHE
4
3
1
4
2
Driving head ??P
Driving head
3
1
Boiling zone zone
5
3
Preheating zone
Principles of a thermosiphon evaporator.
4
4
Entering the separator
Height
Leaving the evaporator
A pinch point is approaching
Boiling starts
Heating fluid
Large driving head
?T
High temperature
Low temperature
The temperature distribution in a thermosiphon
evaporator.
5
? Insensitive to changes of the operating
parameters. ? Even cooling as the refrigerant
always is in a two-phase state. ? Normally
cocurrent flow to give a large starting
temperature difference. ? Countercurrent is
recommended if the starting temperature
difference is large as otherwise back flow could
occur. ? Reasonable constant performance
independent of the circulation. ? The return leg
?P should be low otherwise risk of backflow and
hunting. ? A too low return leg ?P could mean the
vapour having difficulties of lifting the
liquid. ? Never use a control valve in the return
leg, in the drop leg it could be
useful. ? Nonvolatile products entering the
thermosiphon loop remains there and are gradually
concentrated. ? Oil is heavier than and insoluble
in ammonia and is easily drained from the lowest
point. ? An insoluble oil but lighter than the
refrigerant is difficult to drain. ? A soluble or
partly soluble oil is best concentrated by
evaporating the refrigerant and then removed.
General properties of the thermosiphon
evaporator, I.
6
The separator should be placed as low as
pos-sible and with a return leg as short and
straight as possible and with no valves or other
fittings
6
? The preheating zone will be shorter, which
means an increase of the MTD. This is especially
pronounced the lower the temperature is. ? The
driving head does not need to produce a lift of
the liquid in the return leg. ? The separator
must not be placed that low that the circulation
is impeded. ? An evaporator with a horizontal
exit to the sepa-rator, will have excellent part
load properties. ? An elevated separator will
have a very large return leg pressure drop, i.e.
a large difference between the evaporator and
separator temperatures. ? Keep the separator
design as simple as possible. Avoid extra
baffles, distributors, guide vans, etc. ? The
danger of backflow and hunting is diminished.
General properties of the thermosiphon
evaporator, II.
7
The return leg and the flash vapour should
enter at theother end of the separator
The suction pipe and the return oil pipes
should exit at one end of the separator
7
Keep the entrance to the drop leg horizontal in
order to avoid vortices
Keep the return pipe as short and smooth as
possible.
Put an oil drain at the lowest point in an
ammonia system
Overview of the system.
8
3 Bends
2 Bends
1 Bend
0 Bend
2 Vapour exits
2 Vapour exits. Alt.
Unsymmetrical exits. Not recommended
Entrance to separator from above. Flash vapour
into liquid.Not recommended
Design options I.
9
9
?
The pump in a forced flow system should be placed
lower than the evaporator
A downwards loop at the drop leg could prevent
back flow
Injection of the flash vapour by an ejector is
questionable
Injection of the flash vapour by a pipe-in-pipe
system is not recommended
Design options II.
10
10
An ammonia thermosiphon system in a supermar-ket
with an oil recovery system and an oil cooler.
11
11
Oil evaporator heated by hot condensate
Oil cooler with double control valves (one
always, one when needed).
A thermosiphon system for a soluble oil with an
oil evaporator and an oil cooler.
12
12
The cooled liquid is evenly distributed over the
channels.
The ?P increases and the velocity decreases
The velocity decreases Better cooling The
viscosity increases The ?P increases The velocity
increases. Etc., etc., etc.
Another channel gets a lump of high viscosity
liquid and the process repeats
But suddenly the viscosity increases in one
channel
The liquid is more cooled, the visco-sity
increases
The result will be a severe maldistribution (on
both sides)
Suddenly the entire channel might be blocked
The velocity is further reduced
Maldistribution of a viscous liquid between
parallel channels.
13
13
? The PHE manufacturer is responsible for the
proper design of the unit, but he can only make
as good a design as the data provided by the
contractor. ? Use as simple piping design as
possible above all as few as possible pressure
drop producing items between the evaporator and
the compressor. ? Remember, the return leg ?P
translates to a difference between the
evaporation and suction temperatures. ? Try to
keep return leg ?P lower tan 25 of the total,
otherwise danger of hunting. ? Avoid ejector
designs. ? Put an oil drain at the lowest point
of the loop. This should be immediately before
the evaporator. Use an automatic oil drain. ? If
two exits from an evaporator are necessary, use a
symmetric design. ? Use an exit vapour fraction
of 0.7 to 0.85. Higher means a danger that some
channels are not wetted, lower increases both the
evaporator and the return leg ?P. ? Place inlets
return leg and flash vapour at one end of the
separator and the exits drop leg, suction and
oil at the other. ? To decrease the filling
keep the separator as a separator basically
empty and use a special LR under the separator.
Remember, PHEs have low liquid hold-ups. .
Summary of design considerations, I.
14
14
? If possible use a direct horizontal exit to the
separator from the evaporator. Bends cause
pressure drop. ? The above implies a low static
head e.g. separator, especially important for low
temperature operation. ? Cooling of high
viscosity liquids, e.g. 50 propylene glycol at
- 40 C, is difficult due to the risk of
maldistribution between the channels. Consider
instead potassium formates, ammonium hydroxide,
carbon dioxide and for really low temperatures,
silicon oils. ? For cooling of water to close to
0 C, consider only SWPHE as this is not damaged
by freezing. Try to increase the water velocity
by connecting the water in several passes. This
increases the water K-value and increases the
wall temperature. ? Connect the evaporator in
cocurrent. The high initial temperature
difference facilitates start of the boiling. For
a temperature difference of more than 10 K,
connect in counter current as otherwise the risk
of back flow increases. Note that ammonia is
sensitive to a good starting temperature
difference, other refrigerants less. ? Be
observant of possible water in the ammonia. Water
increases the evaporation temperature and can
form a difficult to remove mud, which settles on
the heating surfaces. Remember that water can be
removed by a pump-down, followed by draining the
water-rich, remaining liquid.
Summary of design considerations, II.
15
15
In the following slides are some actual cases,
which gives a general idea of the problems.
Troubleshooting evaporators.
16
16
Oil drain.
A. Ejector inlet (NH3 in a SWPHE)
B. Pipe-in-pipe inlet (R134a in a CB).
Problem Both flooded flow evaporators above had
injection of flashing refrigerant in the loop in
order to increase the circulation. None of them
gave the intended capacity. A. The low capacity
was at least partly due to oil fouling on the
ammonia side. B. This CB had stability problems
and and occasionally reversed flow.
Discussion Injection of the flashing refrigerant
in the circulation loop can improve the
circulation if a properly designed ejector pump
is used. The ejector in A was a home made design
of questionable functionality. When A was
dismounted, it turned out that there was a lot of
oil in the evaporator, not astonishingly as an
ejector is a good atomizer. A closer inspection
of B showed that there was back flow of
refrigerant through the drop leg back to the
separator. The pipe-in-pipe design had no ejector
function at all. It was easier for the vapour to
pass back through the drop leg than through the
evaporator.
Solution Both were redesigned as normal
thermosiphons. A. The pipe was made slightly
upwards inclining and with an oil drain at the
lowest point. B. A soluble oil does not need an
oil drain. Comments An ejector design can
improve the operation but a correct design is
difficult to make.
Flooded flow evaporators with ejector.
17
17
? A liquid line, passing through a hot area or
exposed to the sun, can be sufficiently heated
for refrigerant to vaporize when the flow stops.
Restarting the flow will cause a vapour plug to
move with high speed along the pipe. The vapour
can easily pass a valve but the subsequent liquid
column cant, i.e. damages and noise. ? If the
pressure is suddenly increased, e.g. a hot gas
defrosting, it can cause a (previous) low
pressure vapour to implode. This is the same
phenomena as cavitation.
? It is not unconceivable to imagine a unit
cooler closing and the resulting liquid column in
the vertical feeder pipe line flow down or
refrigerant to evaporate, with ensuing noise, see
Slide 3. ? However, the noise had a definite
metallic quality. By using a time honoured method
putting an ear to the pipe at various
loca-tions, it was possible to find the place
with the strongest noise. It was at the upwards
bend after the ammonia circulation pumps.
? At the installation, there were probably both
liquid lines at the roof top and vapour lines in
the cold storage. ? Moreover, see the figure
above there were a number flooded flow unit
coolers connected in parallel to the liquid
evaporator. These units coolers were installed at
least ten meter higher than the PHE. It is
difficult, to say the least, to operate parallel
units in general and different types at different
heights can be close to impossible.
? Mechanical. Broken parts, forgotten tools, or
metal pieces can get trapped but not unmovable
and cause the noise. ? An expansion valve is
sometimes, especiallyin DX systems, preceded by
a solenoid valve. If pumpdown is made, the pipe
after the sole-noid valve will be emptied of
liquid refrigerant. When the solenoid valve
opens, a liquid co-lumn with increasing speed
moves towards the TEV, which easily can be
damaged apart from the minor problem of noise.
Install the solenoid just before the TEV. ? In a
vapour line, e.g. for hot gas defrost, pas-sing
through the cold storage, refrigerant can
condense when the line is shut off. At restart, a
high speed liquid plug can damage a valve.
Solution The cause of the noise was not found
until a year later, when the system was opened
for maintenance. A broken valve stem had moved to
the first upwards bend after the pumps. The
liquid velocity was then insufficient to lift it
further, it stayed at the bend, swirled around
and made noise.
Problem Shortly after the installation of a new
PHE evaporator a flooded, forced flow system
loud bangs and noises were heard in the system.
The source of noises in metal piping can be
difficult to find but as the evaporator was the
only new item it was the prime suspect. The
evaporator was installed in a very large
commercial cold storage. Discussion Loud, sharp
noise from an item with no moving parts is
unlikely but then, where does the noise
originate? There are a couple of causes for noise
shown on the next page
Bangs in a pumped flow evaporator.
18
Problem A chiller manufacturer had problemswith
a custom made unit. The nominal temperature
program was 20 Ethylene glycol 3 ? -3
C Propane of -8 (5) ? -8 C. The unit was
assembled from parts pur-chased in some four
different countries. At start-up the end
temperature was about 10 K instead of the nominal
5 K Not surprisingly, all component
manufac-turers claimed that their particular
compo-nent was faultless and worked well in
similar applications. The technical manager was
helpless. Discussion During a technical
seminar, the manager understood that the valve
was not correctly installed he remembered that
the valve was possibly installed before a
bend. Solution The valve position was changed
the same day and nominal duty was obtained.
18
An incorrectly installed thermostatic expansion
valve.
19
Problem This large ammonia thermosiphon
installation in a chemical plant didnt give the
capacity. There was also a problem with
hunting. Discussion An inspection of the plant
showed a very nice installation, everything
including the tightening bolts of the PHE
properly insulated and clad by aluminium sheets.
Unfortunately there were hardly any instruments
or sight glasses, the separator had as only
instrumentation the pressure and the ammonia
level. It was really a black box. A further
inspection showed that there was no automatic oil
drain and the drain point was from the separator
rather than from just before the
evaporator. Discussion with the contractor also
revealed that the separator was full of baffles,
guide vanes, pipe distributors for the flash gas,
protection sheets and a demister.
The final measured was taken six months later at
the annual overhaul. It turned out that the angle
valve was far too small and was ex-changed. This
measure decreased considerably the difference
between the evaporation and suction temperatures.
The remaining difference was mainly due to the
internals of the separator. The hunting also
decreased. Comments We here encounter three of
the most frequent problems in a flooded
installation. ? Oil drain. ? The difference
between the evaporation and ? suction
temperatures. ? A separator placed too
high. Conclusion If the large head is not
necessary, place the separator as low as
possible. The capacity of the evaporator in the
system described in slide 3 improved when the
separator was lowered. There are cases when the
evaporator needs a large pressure drop but then,
the entire thermosiphon loop has to be design for
this and not only the evaporator.
There was a strong suspicion that the pressure
drop from the evaporator exit the evaporation
temperature, via the separator the separator
pressure to the compressor suction connection
the suction pressure - could be very large.
Unfortunately, a refrigeration vessel, as
opposite to a chemical vessel, has no man hole.
Inspection of the interior was not possible. The
separator was placed very high, much higher than
the evaporator required. That increases the
subcooling zone, especially at low temperatures,
-25 C, in this installation. Solution Two
measures were taken at site. There was a
possibility to drain the oil from a connection
much closer to the evaporator than the original
oil drain. Draining reveal quite a lot of oil in
the system. This improved the capacity somewhat
but not entirely. An try was made to measure the
evaporation tem-perature. It had to be done
outside the return leg. This is difficult,
especially here, due to the solid insulation. As
a reference, the separator tempera-ture was also
measured in the same way. The re-sult confirmed
the suspicion that the temperature difference was
very large, i.e. the evaporation temperature was
far above the suction temp.
19
Keep the interior of the separator as simple as
possible. You never know if it is built as
conceived.
Lack of capacity in a low temperature evaporator.
20
Ambient temperature
Undersideof pipe
20
Equaliza-tion line
To bulb
Problem The installation was a pretty normal
direct expansion evaporator equipped with a
standard thermostatic expansion valve with
external equalization, but it simply didnt
produce any cooling at all. Discussion A check
with a thermal camera showed some interesting
results. The first pictures were taken from the
front side only as the back side was difficult to
reach. The exit pipe, above marked Ambient
temperature was just that, i.e. no cooling was
discovered. Astonishingly enough, the pipe
marked Cold pipe underside several meters after
the TEV, became blue on the picture somewhat
after the start, indicating a low temperature.
The part just behind the PHE could not be seen.
The equalization line was ambient temperature.
The result was that baffling that an attempt was
made to take pictures from the back side, where
the bulb and the equalization line connection to
the exit pipe could be seen. The start up showed
the sequence below. At first everything was
orange, i.e. ambient temperature. Then - Exit
of EQ line. The connection was slowly turning
blue. - Below EQ connection. A little later, the
bottom part of the pipe below the connection
turned blue. - Further down the pipe. The blue
colour slowly spread downwards.
A check of the valve showed that there was a
leakage in the packing box. Moreover, the
equalization connection at the suction gas pipe
was very small high ?P. Thus - HP condensate
leaks through the packing box into the
equalization chamber. - The diaphragm is pressed
down and the valve closes. - No more refrigerant
leaves the normal exit. - Liquid condensate flows
through the EQ line to the connection, expands
and cools. - Liquid cold refrigerant drops down
to the pipe bottom and cools this. Solution The
TEV was exchanged and everything
worked. Comments A thermal camera is an
extremely useful tool when troubleshooting.
A broken expansion valve.
21
21
Problem The installation was a pumped flow
evaporator. Return line from the evaporator to
the separator was very long and contained a
number of valves, bends diameter changes. The
evaporator didnt give the capacity.
Discussion A check showed that there was an
considerable temperature drop from the evaporator
to the separator
A forced flow evaporator.
22
22
At the delivery the SWP?HE is filled by nitrogen.
The connections are sealed with two plastic
caps. The upper has a valve for the nitrogen
filling.
A forced flow evaporator.
23
23
At the installation, the lower cap had been
removed but not the upper. The valve had blow out
leaving a little hole, large enough for some
ammonia to pass and give signals to various
instruments.
A forced flow evaporator.