As with most things in life, nothing is simple or easy. That is very true of this machine that was exhibiting a number of issues. The biggest one being the engine would begin to stall, lug down, when a service is deadheaded or the machine was put under load. Left under load, the engine would stall completely but the engine would restart without any issue. Not only this but the hydraulic pumps on this machine were leaking considerably and noisy, which required them to be removed and sent away for refurbishment.
Is it the hydraulics or is it the prime mover? This is the question often raised when dealing with hydraulic systems issues. To make this harder to verify, machines often have complicated feedback systems, software logic and different modes on a machine which can all change how the machine operates. In all cases having a thorough understanding of the machine and its operating options is required before heading down the multiple rabbit holes that are in front of you. You could say that’s a lesson learnt here as the information about how this machine operated was only really established whilst we were trying to diagnose the problem.
The pumps fitted to the machine are the Kawasaki K3V tandem pump where it has two pumps mounted in series to the engine. The input speed of the pumps is directly driven by the engine so pump speed is equal to engine speed. Given that both pumps are controlled in the same way, I have redraw the pump schematic below but showing the front pump only.
I don’t want to go into too much detail on this, as all the information on how the pump works can be found in detail in this forum post - but we will look at how the different modes can affect the flow rate as we go through the diagnosis.
The below image is helpful in order to understand that this system has a certain amount of backpressure. It is generated by the orifice which sits above the direct acting relief valves K where oil has passed all the way through the main control valve from the front and rear pumps. This pressure signal is feedback to the pump which can be seen in the top image where it comes in at point I.
Prior to the pumps being removed a quick check of this back pressure was taken at the pump ports, PT1 and PT2. Engine speed was normal operational speed, around 2000 rpm with no load applied.
Given both pumps should be offering the same amount of flow to each side of the valve block, 10 bar difference is something to note. Again, product knowledge is key here as knowing that there is a path through each of the spools within the main control valve to the back pressure orifice. This is providing that the spools from each valve are in the correct place. Given that the pressure is resistance to flow, PT1 could be losing some flow over a leaking spool, or the pump isn’t working as well as it should. PT2 on the other hand could have a slight blockage which is resulting in a higher pressure.
Either way, with the noise and the leaks, the pumps needed repair before we were to go forward.
With the pumps back they were fitted back to the machine and as you may have guessed, the machine was still stalling when under a high load and getting nowhere near the pressure rating for the system.
Looking around the machine there were more things wrong. The wiring appeared to have overheated and caused the insulation to become brittle and fall away exposing the wire to the elements. Some of the wiring didn’t really go anywhere either. The main control valve, MCV, also had ‘signs’ that someone had previously had a look at it.
In the end the MCV was removed and repaired with the wiring loom tidied to the best of our ability. We did find a lot of wear on the main relief valve which may explain our lack of maximum pressure.
A lot of unknowns with the wiring loom as we are sure that some things were missing but given the machine was operational with no warnings on the dash we were prepared to work with what was in front of us.
MCV back in place and connected, up guess what. Fault is still present although it has changed as we can now achieve a higher pressure thanks to a new main relief valve, and the engine stall is less likely when operating in a mode that activates max flow cut.
Max flow cut actually applies pilot pump pressure to a port H in the below image. This pushes the swash plate back on the pump and so reduces the overall flow rate.
So whilst the engine speed is still dropping when a service is held in a deadhead, we can maintain a certain pressure for a period of time. Given that pressure is just resistance to flow, as the flow rate is already backed off due to max flow cut, we were able to achieve much higher pressure of around 310 bar. Still not quite the peak system pressure but it does mean the engine isn’t stalling. Whilst this is progress in a way, it still means the machine can’t be used to its full output.
Before going any further, the decision was made to look at the prime mover, the engine. Given this is an older machine, it’s natural to expect some wear on the engine. As this is the one thing that is driving the pumps then surely this would have been the first thing to look at. Well, typically I would say it would be but we had numerous hydraulic issues to start with and without those rectified, the diagnosis wouldn’t be able to continue. Ordinarily, when starting on a hydraulic complaint, I would start with some basic checks on the engine. Relative compression, exhaust pulse, crankcase pressure, cylinder balance using crank maths etc. We did find and adjusted a slightly loose intake valve rocker thanks to the intake manifold pressure test which was resolved. Other than that though, there wasn’t a huge amount to be concerned about.
Moving back to the hydraulic system as flow is what makes the machine go, making sure the pumps are capable of outputting the correct flow rate is important. Given the 3 modes, A - automatic, L - lifting and E - economy all have an effect on the flow rate we set the scope up to capture these changes. In order to make sure the swash plates are held in the maximum position, the feedback lines were removed and capped to prevent the back pressure through the MCV affecting the swash plate angle.
In A mode we are getting 215 lpm which is the maximum rate for this system as stated in the technical information. That’s probably all that was needed but as we were there it was worthwhile capturing the other 2 modes. In L mode, 173 lpm and in E mode 194 lpm. In L mode - lifting, speed is less important and so the flow rate can be lowered. Happy that the flow rates can be easily achieved it was time to get some known good data.
Luckily we found another machine and although the engine was different, it had the same hydraulic system and pump setup. Fitting the flow meter into the rear pump circuit we manually loaded the hydraulic circuit with the loading valve and took note of the PQ curve.
When the pressure hits around 240 bar, the flow rate seems to change and drops at a faster rate. This is different from our problem machine where the PQ curve has a much flatter curve. Utilising viewports and reference waveforms we can look at the two captures at the same time. In the image below the faulty machine is on the left and the known good is on the right.
So what is causing this difference in the PQ curve and how does it affect the load on the engine? It keeps coming back to the statement - pressure is resistance to flow. We can have a 350 bar of pressure with hardly any flow but if the pump is still in a position to provide a maximum flow rate, it’s going to be a lot harder to compress the oil to make it flow at that rate. However, if we reduce the flow rate the effort required to compress the oil in the piston chambers is less due to a lower volume. This way we can maintain a constant pressure, with minimal flow without overloading the prime mover. The K3V pump does this by using the chamber at point E in the image below.
Now comes the part that some people might not agree with. Knowing all the above and theorising around how the system works, we knew it was likely to be an adjustment on the pump. Having spoken to a Kawaski pump supplier, this sort of adjustment is not recommended which is why I’m not going to say how it was performed. Needless to say though, by having Pico connected and making small adjustments and loading the system using the flow meter, we were able to watch how the adjustments were affecting the system.
We finally got to a point where the PQ curve was looking similar to the known good machine and below is the capture before the adjustment and after.
In the before capture on the left, we can see the flat line in the PQ curve, where the overall flow rate doesn’t really drop significantly as the pressure increases. It’s a good sign our pump is healthy but not so great for our engine. If we look on the right, after the adjustment, the curve takes a bigger dip as it approaches the higher pressure reducing the flow rate and therefore the load on the engine. Following this the machine no longer stalled when any service, in any mode, was deadheaded.
What we believe is that the new hydraulic pumps would have been put on a test bench where the pumps would have been set up according to the specifications. What this doesn’t take into consideration is that the prime mover may have lost a few horses over the years and so doesn’t have the same amount of power required to run the pumps, compared to when it was new. Following the adjustment the engine was more than capable of holding a service in a deadhead position which is something it wasn’t able to do at the beginning of this adventure.
Asking the owner of the machine to try it out he told us he wasn’t sure it had ever worked as well as what it did following the work. I don’t think you can ask for anything more than that!
Going forward, the PQ capture obtained once the adjustment was made can now be used going forward to monitor the machine should any further issues arise. Whilst using a flow meter is intrusive and can introduce more problems with dirt ingress, it is invaluable when it comes to diagnosing difficult hydraulic problems. By using Pico we can capture data in real time at high sample rates to ensure we never miss a thing.
