Date: Fri, 19 Nov 2004 17:44:17 EST
Reply-To: FrankGRUN@AOL.COM
Sender: Vanagon Mailing List <vanagon@gerry.vanagon.com>
From: Frank Grunthaner <FrankGRUN@AOL.COM>
Subject: On Oil Coolers, Cooling System Design and Engine Life
Content-Type: text/plain; charset="US-ASCII"
There has been a long and voluminous thread running in the Diesel Vanagon
Yahoo Group about the merits of the VW supplied coolant based oil cooler or heat
exchanger, the possible contribution of thermosiphon effects and the utility
of external oil coolers. Implied here is a question of the quality or capacity
of the VW Vanagon coolant design. I have sent many weighty tomes on these
subjects into the Vanagon list archives and they can be found there through a
topical search on the issue as well as my email address. I include two of these
notes at the end of this comment.
For those who have little time, let me summarize my comments: The Vanagon
cooling system is well designed and is overcooled, thermosiphon effects are
negligible to nonexistent, heat exchanger oil coolers seriously and positively
address engine life issues and turbocharged engines need all the help they can
get. So in order of the summary:
1. Vanagon cooling system design. Superb! See the reproduced note on the
Cooling System appended to the end of this comment. Capable of handling a low
thermal efficiency gasoline engine of more than 400 HP, with full thermal
accessories including air conditioning and trailer towing.
2. Thermosiphoning! Aargh! What is the mechanism of this cooling approach
used with moderate success on the Ford Model T and correlated vehicles? Density
differences is the answer. Hot water rises because the density of the liquid is
less than that of colder liquids. This density difference gives a volume
expansion that drives flow. The maximum density of liquid water is achieved at 4
C. At 120 C (roughly corresponding to the boiling point at a 15 psi
overpressure radiator cap, the density has decreased by 5.7%. Comparing 82 C (nominal
cooling system operating temperature) to 4 C, the reduction is a difference of
just 2.9%. This can be viewed as a displacement difference in comparison to a
centrifugal pump. It is equivalent to a VW water pump with the vanes machined
off to 200 micron bumps. The flow through the oil cooler is established by
and completely dominated by the water pump driven hydraulic circuit. (Question:
Why doesn't the G/J coolant flow diagram show arrows for direction of flow
through the oil cooler? Answer: The drawing is too small to be cluttered by
arrows in the area of the oil cooler!). So what establishes flow in the system? The
arrows in any coolant flow diagram show the direction of the pressure
differential! Now the liquid water comments above do not directly apply to the
ethylene glycol mixture used in the cooling system. For a 50/50 mix of water and
ethylene glycol, the density change for the 4 C to 92 C interval is 3.9%.
Remember too, that for temperatures above 2 C the heat capacity of the 50/50 mix is
20% less than for pure water.
3. Flow and pressure differential. As I have often indicated in my missives
on the Vanagon list, I have instrumented my conveyance heavily, and one of my
most useful monitoring devices is my coolant pressure gauge. I can see every
morning that I developed a negative pressure (cooled coolant), can see that the
cap has released to draw in liquid from the overflow reservoir, can tell if
the overflow reservoir is dry, can measure the head pressure of the water pump
(ranges from 4 to 7 psi as a function of idle to 4K), determine the presence of
air bubbles in the system, leaks in the heater circuit and a pleasing
reaffirmation of the amount of work I require the engine to accomplish per unit time.
4. In the various I4 Bentley manuals the purpose of the heat exchanger/oil
cooler is clearly explained as has been repeated in this thread. Coolant heats
up much more rapidly than the oil and is used to accelerate the heating of the
oil to minimize water and sulfuric acid retention. This is particularly
important for city use vehicles which are often used for short trips which generally
never let the oil get over 60 C. Remember, the diesel Vanagon design effort
at VW was initially funded by a development contract for a Post Office delivery
vehicle. This effected things like chosen gear ratios, sound deadening
solutions and top speed levels. When the oil reaches its maximum temperature
(generally well above the coolant steady state temperature), the oil is cooled by the
coolant. Whatever the oil temp is, it would be higher without the heat
exchanger. Ever run a hopped up Corvair engine in a camper bus? Never saw over 300
F! Of course using an air cooled synthetic!
5. My problem with the oil cooler is that there is not quite enough real
estate for truly trying conditions. (Wide out 1.6 D NA, 2nd gear, 5 passengers,
full camping gear, Westfalia going up the Benner Pass from Austria to Italy).
Dieselgeek.com sells a good solution to this problem with an oil cooler/heat
exchanger that has twice the volumetric capacity and swaps in for the stock unit.
the Audi 5000T is a well engineered car with lower thermal margins than the
Vanagon cooling system. Its turbo gas engines and diesel engines used
thermostated external oil coolers, as did the original European GTi. In fact the 5000T
also uses auxiliary oil coolers and secondary radiators. External
thermostatted oil coolers are an excellent solution to the high oil temperature problem.
Turbo's are a particular problem because oil is the coolant for VW's
turbodiesel units. The heating of the oil passed through the turbo shaft bearing
constitutes the most extreme environment for oil in these engines. This heat is then
held in the crankcase until rejection. This leads to a coking problem in the
oil. External oil coolers add a particulate load to the lubrication system. The
Audi 5000T uses an external oil cooler and a dual oil filter arrangement.
Large impact on thermal management and longevity.
6. My solution: I'm currently developing a TDi conversion for my Vanagon,
using many of the systems I have developed for my Audi 2.0L Turbo installation.
These include the windage tray and baffled oil pan (50 degree install) to stop
or minimize oil aeration and cavitation, new oil pick-up geometry, and the
5000T oil filter system adapted to the 1Z/AHU TDi oil filter flange. For this oil
system, I use two oil filters and two stock VW oil cooler/heat exchangers
(one for each oil filter). I also use the external oil cooler arrangement that is
integrated into the Audi oil filter stack. This unit is thermally controlled
with pressure bypass to minimize all oil cooler problems save leaks. The oil
cooler lines from the filter stand to oil cooler are the same as per the
original Audi install. The oil coolers are mounted on the passenger side real well
(described in the archives) and are cooled by a thermostatically controlled
puller fan. That and Mobil 0W-40 will do me nicely. Is it overkill ... probably!
Do i care if someone thinks its unnecessary ... Not a wit!
&. On the Vanagon flow design. There is indeed one (1, uno) difference
between the Vanagon coolant plumbing arrangement and the standard VW watercooling
practice ... Hot water to the bottom of the radiator, cold out the top. Why, not
sure! But it was done with great attention to detail. I suspect it has to do
with head pressure for the pump and a need to keep it primed over any working
angle (parked as well). Other than that, the highest pressure in the system is
in the block/cylinder head and the lowest pressure is at the thermostat entry
port. This is slightly lower than the heater/expansion tank return because of
the thermostat restriction. The expansion tank inlet is directly tied to the
cylinder head output main and the feed to the radiator. On many earlier VW's a
separate line went from radiator overflow to the expansion tank. These points
are hydraulically and thermally equivalent. The bypass line from the head to
the pump inlet port is internally controlled by the thermostat system and does
not offer free flow to the same extent as does the heater/expansion tank
return circuit. The oil cooler is across the main head output to the radiator
output/thermostat input. Other VW implementations take the oil cooler off the
bypass hose. This is hydraulically identical to the Vanagon arrangement. When the
thermostat is closed and the heater valve off, the main cooling circuit is to
the expansion tank and back. When the thermostat opens, coolant will flow
through the oil cooler. Now, VW's implementation of JX turbodiesels in Europe
added an auxiliary pump to the coolant system. The output coolant from the oil
cooler is directed to the aux pump inlet and the output of the aux pump is sent
to the same thermostatted inlet port on the water pump. This pump is run as
needed and after engine shutoff to cool the oil cooler! This is the same system I
use on my TDi conversion, but with a Mercedes recirculating pump.
OK, enough. I now enclose copies of other coolant details from the Vanagon
archives (search time 33 seconds).
______________________
Cooling System Design:
Among those threads that triggered a negative response on my part were
several that addressed the apparently accepted wisdom of the marginal capacity of
the vanagon cooling system. Now some time ago, I put into the archives a
summary of my measurements of the thermal performance of the stock I4 system and
stated that it was actually strongly overcooled. Then again, some individuals
return with the idea that removing the thermostat or switching to a lower
temperature one will give the system the necessary extra capacity it was lacking.
Well, while licking my wounds and sitting here amazed at what beta blockers can
do to your general energy level, I decided not to complete my physicians
recommended text of the day ("35 Delicious Ways to Prepare a High-Fiber Diet of
Sawdust") and returned to some of my cooling system notes. Unfortunately for
you, I decided to share them with list before Tom blocks me for needlessly
verbose posting.
To begin, there are several key issues: 1) the thermal handling capability
of the radiator, 2) the flow rate of the water pump, 3) the flow resistance
of the plumbing going fore and aft and 4) the level of energy generated by the
hot air pump in the rear whether it be a lowly I4, a perverse WBXer or an
exhaulted Subie 6.
1). Radiator capacity. There are well known engineering rules for
determining the necessary effective thermal handling capacity of the closed
loop/water/ethylene glycol cooling system. The rules digest to a volume ratio. The core
volume of the radiator must be at least a multiple of the engine displacement.
The baseline factor is a multiplier of 2.0 (that is, engine displacement of x,
then radiator core volume of 2x). This factor is then adjusted by known
inefficiencies and additional loads. The inline engine adds 0.1 to the ratio;
outside temperatures above 105 F, add 0.2; for a small engine in a large vehicle,
add 0.2; for air conditioning add 0.3; for a small tight engine compartment, add
0.3; for a standard transmission subtract 0.1; for a full fan shroud,
subtract 0.2; for a horizontal flow radiator, subtract 0.2; for a diesel engine add
0.6; for medium trailer towing, add 0.2. It goes on and on. For my interesting
case (inline, hot, small, AC, tight, manual, shroud and crossflow with tow) I
come up with a factor of 2.8 times displacement. For my application, the 2.0L
Audi 3A motor displaces 121 cubic inches, so I need a radiator core volume of
338 cubic inches. If I had a 2.0L TDi with an automatic, my ratio would be 3.5
and my core requirement would be 423 cubic inches. The Subaru guys don't like
these long winded posts so they will have to do the math.
Ok, so what is the capacity of the Vanagon radiator core. Answer (thank you,
archives), 23.250 (W) x 16.50 (H) x 1.75 (T) for 671.34 cubic inches. I
therefore have effectively twice the cooling capacity in the radiator than I need.
Overcooled says I. Good design says I. Of course, if its broken fix it, but no
need to improve the engineering. Along this line, I should point out that the
factory Ford Mustang with HiPo 302 cid V8 engine ships with 648 cubic inch
radiator core and this calculation suggests that it needs 698 cubic inches. Good
VW design. This discussion actually assumes Al core radiators and Cu would be
somewhat better.
2). Flow rate of the water pump. Well, this has proven hard to lock down. VW
holds that this info is classified and only available in Slovenian! Several
specs are given for industrial engines of 1.0 to 1.4L displacement. These are
given as 15 gallons per minute at 2000 rpm. Several V8 engine design manuals
call for 5 gallons per minute at idle and 22 gpm at 4000 rpm. Well instrumented
tests of GM 4 cylinder engine water pumps give the same flow rates as each
side of the Chevy small block pump: 14.37 GPM at 2000 rpm pump shaft speed and
37.08 GPM at shaft speeds of 5000. All these numbers are without cavitation. I
therefore assume that the VW water pump driven at proper shaft speeds is
generating a flow rate of about 5 GPM at idle and better than 35 at 6000 rpm. This
corresponds to 20 L/Min at idle and 150 L/Min at 6K. That's a complete coolant
capacity exchange every 50 seconds at idle and every 6.5 seconds at 6000 rpm.
More than adequate by any reference.
3). Plumbing flow resistance. Well, for this exercise I did some
measurements on one of my Saturday Morning Junkyard Constitutionals. The entrance and
exit radiator ports on the VW Fox (7 examples) are 25.0 mm in diameter. These
ports on the Audi 5000T (big turbocharged beastie, 2.4 L I think) are 27.5 mm in
diameter and the inlet/outlet tubes on the diesel Vanagon I found ('82) were
43 mm in diameter. Now the resistance to flow of an incompressible fluid
through a pipe is directly proportional to pipe length and inversely proportional to
the internal pipe diameter. Taking into account these relationships and
assuming a to and fro length of 135 inches for the Vanagon pipes, we can compare
the Vanagon plumbing resistance to that of the Audi. Such a calculation says the
Vanagon tubes have a flow resistance of the equivalent of 45 inches of Audi
plumbing. Now the actual Audi run is closer to 24 inches. But the Vanagon tubes
have a lower roughness and frictional loss (metal or hard plastic vs.
rubber). The Audi hoses I saw were corrugated (more losses) and had several complex
bends. In the end the best way to measure reality is to do a pressure vs. flow
plot before bolting up the conversion engine (to paraphrase the elderly
retired over the hill Suburu operating gent). Measurements rule! Nonetheless, I
would argue that the VW engineers sized the Vanagon plumbing for similar flows
(within 20%) of that obtained in the longitudinal and transverse cars in their
stable. Certainly fully compatible with radiator size and pump capacity. Good
design I say. Certainly compatible with anything the Subie or TIICO or
Turbocharged crowd can throw at it.
Of course, this neglects head pressure when the radiator is 45 degrees above
the engine and the driver is holding on with white knuckles, too frightened
to look at the coolant gauge. This also neglects the direct injection of
combustion gases into the cooling jacket.
4). Energy injected. Well, this is actually treated by the general capacity
discussion of part 1. But for reference, consider that the power generated by
the hot air engine is roughly one third of the power processed. One third goes
out the exhaust as heat and about one third goes into the cooling system for
dissipation. So if your conversion pumps out a real level of 30% more horses
than the WBxer, then 30% more heat has to be processed by the cooling system.
Lets see, Diesel ... 42 Hp, Turbo Audi about 160 Hp! Glad we have that extra
capacity.
In closing, I want to take a swipe at the thermostat wisdom. It is true that
removing the thermostat leads to block overheating. It is not true that this
is due to higher flow through the radiator - going too fast for cooling. The
reason is that a minimum head pressure must be maintained across the water pump
to eliminate cavitation. Lower temperature thermostats just throw your money
down the drain. Lower engine efficiency. The engine operates more and more
efficiently as the coolant temperature goes up to 105 C (for the ethylene
glycol/water system). The peak efficiency is around 122 C with pure propylene glycol.
Of course, at 87 C, the hot spots in the block are probably well above 170 C!
Small differences in mean control temperature are irrelevant to the thermal
stability of the system ... as I just tried to show.
With apologies for length, I'm glad I finally got that off my chest!
Frank Grunthaner
________________________
I have been following this thread with some amusement. In many of these
discussions it is assumed that the Vanagon cooling system is a weak link. From an
engineering perspective, this is simply not true. Stuart is dead on!
As I prepared for my Diesel to GTi conversion (note that my analysis is
only correct for the diesel cooling system - didn't consider any details of the
waterboxer system), I analyzed the thermal capacity, coolant volume, radiator
surface area, fan-induced air flow, and coolant flow from the engine
compartment to the radiator and back. the numbers took into account the power (heat)
generated by the gasoline engine, the peak thermal inputs of both diesel and gas
engine, etc, etc. The bottom line is that the Vanagon system is seriously
overcooled. The thermal capacity is capable of handling the thermal load
generated by a 400 hp V8 Chevy small block. The peak thermal loads were higher for the
diesel (I was considering both NA and turbo applications) than for the 1.8 L
gasoline engine. This analysis included the additional thermal load introduced
by the addition of Air Conditioning. I also used the capacity recommendations
for high output V8 engines in low relative humidity conditions and desert
ambients (105 F air temperatures).
After the conversion, I have monitored the cooling system under load. The
coolant temperature in the return line from the radiator was more than 20
degrees cooler than the engineering tables suggested as nominal (don't have my
notes here, so I can't be more quantitative). These tests were done on a 100 F
day romping around the Angeles Crest (LA) in first and third with AC on full
bore! As some of you might suspect, I tried to optimize the system. I quickly
found two limitations:1) air in the system and 2) water pump flow capacity.
The first of these is the most important real time variable. Air in the
cooling system seems to segregate to pockets in the cylinder head. Substantial
temperature variations between cylinder #1 and #4. I currently monitor the head
temperature with a Westach gauge with senders under the spark plug for #1 and
#4. With air in the system I can get temperature differences of up to 35 C.
Bleed the system thoroughly! I have designed a bleed screw modification of the
head coolant outlet, but haven't implemented this yet. Trying to follow the
system used by Porsche for the 944 Turbo which also has bleeding problems. With
the coolant pressure gauge, I can readily detect the slow creep of air into
the system (small heater core leaks, air from coolant overflow tank, small hose
leak, etc.). When sealed, the system has to be bleed at several week intervals
to remove trapped air. When fully bleed, no further air introduction happens,
and the temperature differential between cylinders is less that a few degrees
(probably system measurement resolution).
The second issue is water pump flow rate. In the case of the I-4 engines,
the most accessible variable is the pulley diameter, and the radial belt
contact area. From my parts stock, I looked at 4 different pulley diameters (all
used on various implementations of the 1.8 L 8V engine - Golf, Cab, Fox,
Dasher/Quantum). In the end, I equated highest flow with maximum pressure at the
radiator inlet at 6000 rpm. This was achieved with an intermediate size pulley
(not the largest or smallest diameter). It was important to use the pulley from
the three pulley belt design (crank, water pump and AC compressor on one belt,
AC compressor and alternator on the other). I found no pressure difference in
the pumps below 3000 rpm, and concluded that the smaller pulley lead to
cavitation at the pump impeller reducing the net flow.
As to closing off the radiator flow with the spare tire in place, the
swept area feeding the front of the radiator (top and bottom grills) is actually
quite large. The radiator is placed rather forward compared to an American Iron
design minimizing the plenum effect. I found that the area was poorly sealed
at the bottom and the sides. Consequently, I sealed all the area around the
radiator (including the bottom) with thick neoprene sheet (about 1/8 inch
thick). With a similar precaution, I would assume there would be no problem with the
coolant system.
Would never do it myself though, don't like the looks, and it has to
increase the drag (more power, more fuel) to maintain over 50 mph.
Sorry for the length, hope this helps someone,
Frank Grunthaner
| 
