This article describes technical causes behind the split multi-section gear pump failure, described in the "Travelift Adventure" post, and also touches a couple of gear pump related myths.
Brief malfunction description:
A five section Parker gear pump,
four sections of which were feeding independent Sauer Danfoss PVG 32
proportional valves, split the bodies and spat out the seals after ten
minutes of operation, causing notorious agitation of standers-by and
appearance of Grey hair on the head of the attending technician
Before going into the main causes of the failure, I
would like to talk a little about the pressure induced forces that
exist in "traditional" gear pumps, as well as to touch a couple of
gear-pump related misconceptions I come across fairly often.
In a common single section gear pump (like this one,
for example) you have two end plates, one with a shaft sticking out of
it and the other one blind, and a body in between, with all the
gears, seals and what not inside. The "sandwich" is held together with
bolts, which either are screwed into the threaded holes of the body, or
pass through the whole sandwich and secure the parts with nuts
on the other end. Each of the plates has an area exposed to the
outlet pressure which produces force directed to separate the plates
from the body, and therefore creates tension stress in the screws. Let
us pass on, now, to the myths, starting with
Gear pump myth number one.
Not a myth, actually, but a
misconception I often see around workshop population, and which
concerns gear pumps that use passing screws. With the threaded body
and short screws the understanding of separating forces is
straightforward and normally causes no difficulties - the
internal pressure acts on the plate, creating a force which equals the
exposed area times outlet pressure. If we had to calculate the internal
stress of the screws, it would be this very force that we'd use for the
calculations. However in case of passing screws (diagram,
internal parts are excluded for the sake of simplicity) it is often
misunderstood that, as the upper plate creates an "area times pressure"
force in one direction, and the lower plate creates an equal force in
the opposite direction, the resulting screw tension force is doubled.
This is wrong - the tension stress of passing screws is the same as
with the threaded body design, and if we were to calculate it, we'd
still use the same one area times pressure, just like with the threaded body design.
Gear pump myth number two.
This misconception boils down to
believing that in multi-section gear pumps, held together with long
threaded rods that pass through the whole assembly, the pressure
induced separating forces of every section add up, causing a much
bigger tension stress in the screws/rods than in the case of a simple
pump. Once again, let us look at the diagram - the section with higher pressure creates higher separating force, which is then transmitted to the screws through the end plate and the body
of the section with lower pressure. The pressure inside this section
relieves the central body from some of the stress due to the separating
force acting on the endplates, but it doesn't add up! If we were to
calculate the tension stress of the screws in a multi-section pump,
we'd have to use the same one area times the highest pressure.
Now that we are clear about gear pump pressure induced separating forces, let us go back to the malfunction and its
1. The first and most obvious cause, of
course, is the four threaded rods used to hold the five bodies together
However, the same rods had been used for ages in single section pumps
without a single problem. So what went wrong?
Two mistakes were made during the rod choice - the
rods were threaded through the whole length, and the rod grade was too
low. The separating forces of a multi-section pump do equal to
the ones of a single pump, however due to the fact that multi-section
pumps use far longer rods, the same amount of tension force results in
a higher rod length increase, which is why the rods suitable for single
section pumps can be not suitable
for multi-section pumps. Rods that are threaded though the whole
length are a poor choice for these units, because the thread reduces
the effective load bearing area of the rod, making it "thinner". For
large multiple gear pumps the use of high grade steel rods, which have
threads only at the ends, is imperative.
2. The second cause was the presence
of high pressure spikes, which caused the rod lengthening and the
consequent "foda-se, caralho!" (Portuguese for "unpleasant") situation.
To trace the origin of the spikes let's look closer at the operation of
the PVG 32 valves used in this system.
On the section view
of a PVG 32 pump side module you can see that the relief function is
accomplished by the compensator spool 6. When the pressure exceeds a
set level, the small poppet of the valve 1 lifts from its seat, venting
the left side of the compensator to tank and causing it to dislocate to
the left, thus connecting the inlet to the tank port. The relief
function is there, but like in case of any spool type relief valve, it
is relatively slow.
In that hydraulic system, each PVG valve had a
ON/OFF module, controlled by a PVEO 24v solenoid. These modules were
used to operate the four hydraulic winch motors. Due to
let-us-get-rid-of-what-we've-had-for-years reasons 100 liter per minute
spools were chosen for the sections, which exceeded the pump supply,
and so when the valve was in ON position, the compensator spool closed
the P to T passage completely. When the solenoid was de-energized, the
distributor spool would return to central position in about one tenth
of a second, which was faster than the pump section compensator could
open the P to T passage, so whenever the ON/OFF solenoid was shut-off
there was a pressure spike in the P line, caused by the combination
of "slow" compensator spool and the fixed displacement gear pump
unstoppable flow. By the way, the choice of a high flow spool can be
gracefully explained by the desire to lower the spool pressure drop and
therefore the consequent heating, as the winching function required no
flow control and was to be operated for long periods of time.
There are several ways to address the problem. The most obvious is installing pre-charged accumulators at the pumps' outlets. It is,
probably, the best technical solution, with the drawback of being expensive and requiring regular maintenance. Another, cheaper solution lies in
installing additional fast acting pressure relief valves in order to
clip the spikes. Yet another - using a longer pressure line hose which will
damp the surges due to the hose accumulator effect. Another one is to
limit the distributor spool travel (or, alternatively, apply a smaller
flow spool) to the point where the compensator starts compensating. In
this type of valve, flow regulation is done by means of metering the
pressurized fluid to tank, from P to T, so when the compensator meters
flow, there is already a small passage open between the inlet gallery
and the tank line, making the high pressure surges unlikely. Yet
another solution would be to find a way to dampen the distributor spool
movement (by installing orifices in the pilot lines, for example), thus
giving the compensator time to open.
In that particular case there were four fast acting
SUN relief valves readily available from the old installation, with all
the fittings, pipes and everything, so I opted for installing them
directly at the pump's outlets. Even when they were adjusted 40 bars
above the PVG relief setting, they'd still spit a stream of oil from
the T line each time the solenoids were disconnected. The pump hoses
were also increased a little.
Lessons to learn:
- Multi-section gear pumps should be assembled with high grade steel rods with thread only at the ends.
- Sauer Danfoss PVG valves, as well as other
proportional valves based on the same design, under certain conditions
can create high pressure spikes in the inlet line. These include: the
combination of flow saturation condition (when the compensator spool
closes the P to T passage completely, which can happen when a spool
flow rate is higher that the pump flow, or when several spools are
actuated at the same time, demanding more flow than the pump can
deliver) and a fast release of the spool/spools to the neutral position
(which is the case of the PVEO on/off solenoid).
- Such pressure surges can be aggravated in fixed displacement pump circuits that use rigid (steel) piping.