The 15th Scandinavian International Conference on Fluid Power, SICFP’17, June 7-9, 2017, Linköping, Sweden

Simulation study of a digital hydraulic independent metering valve system on an

excavator

M. Sc. Miikka Ketonen, Adj. Prof. Matti Linjama

Laboratory of Automation and Hydraulics, Tampere University of Technology, Tampere, Finland

E-mail: [email protected], [email protected]

Abstract
Independent metering valve (IMV) control of working hydraulic systems in mobile machines has been studied for more than
two decades and during the past few years it has also been adopted to commercial excavators. The main advantages of the
IMV systems, compared to load sensing and open centre systems are the possibility of optimizing the pressure losses of the
metering edges and the possibility of re-routing the hydraulic energy between actuators. Energy re-routing can be realized
without storing the energy to accumulators and thus avoiding the losses of additional routing valves and energy conversions.
IMV combined with a hybrid system allows even more improved energy efficiency. Digital hydraulic IMV (D-IMV) allows
additional benefits to the IMV systems with more fault-tolerant operation with robust components, faster and more precise
control and leak-free valves.
The purpose of this study is to apply a D-IMV system to a midsize (20t) excavator and to study the differences in energy
consumption of the working hydraulic system with four actuators. This paper presents the controller designed for the D-IMV
and a simulation model for analysing the system. The energy consumption of the D-IMV is compared to the measured energy
consumption of a state-of-the-art excavator with a load-sensing hydraulic system.
The Controller of the D-IMV system is realized with a sub-optimal mode control, so that instead of calculating and optimizing
the total energy consumption of all possible mode combinations, simpler control logic is defined to remove the complex
structure of the controller and ease the computational burden. Simulation study shows that the hydraulic input energy can be
reduced 28-42% compared to a standard LS-controlled excavator. Detailed analysis of where the reduction emerges is
presented.

Keywords: Digital hydraulics, Independent metering valve, mobile hydraulics, excavator

Asian and American manufactures seem to favour open

1 Introduction centre type valve systems. Main energy consuming features
Demands for the off-road machine builders by markets and of current excavator valve systems are the throttling losses
legislations are driving new technologies to the traditionally occurring when multiple actuators are used simultaneously
conservative field of hydraulic engineering. Many of the and throttling losses occurring when an actuator operates far
construction equipment vendors have brought the hybrid from the designed operation point due to the mechanically
systems to the markets and a few manufacturers are also coupled inlet and outlet metering edges of the spool. Often it
utilizing independent metering valves (IMVs) in off-road is considered a typical LS system lacks the operator feeling
machines. Excavators are the biggest machine type along and response that operators tend to like.
with the wheeled loaders, which makes them perhaps the
most important research subject in the field of mobile Studies recently have also proposed many alternatives for the
machines. open centre and LS systems. These includes, for example
flow demand-based eLS control systems [1], multiple
Most common state-of-the-art excavator valve systems are constant pressure rail systems [2, 3], displacement control
based on open centre flow control (OFC) or a load-sensing systems [4, 5] and different kinds of hydraulic hybrid systems
(LS) principle. There seems to be a trend that LS systems can [6, 7].
be found in excavators below the 25 ton-class and above that
the excavators have an open centre valve system. LS systems, The advantage of an IMV-system compared to other
and especially electronically controlled eLS systems, have technologies is the relatively simple system components. The
proven to be more energy efficient, but have only become system can be improved with only a single pump, keeping the
popular among European excavator manufacturers, while manufacturing costs of the machine low. An IMV allows the

Peer-reviewed Paper, 136 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden
hydraulic system to transfer hydraulic fluid between actuators U U
P P
and actuator chambers through the supply line or through the
return line without additional routing valves. An IMV also
can operate without additional pressure compensator valves.
[8] U
P
U
P

ARM
It is possible to implement the IMV technology to existing
BUCKET
excavators with relatively small changes, and it can further
improve energy consumption of LS systems by using U
P
U
P

differential and regenerative modes when the load force is

suitable. An IMV also allows more flexibility in
programming the boom control as requested by the user. [8,9]
U U
P P

For full performance, IMV systems require relatively high

demands from the valves. Valves need to have high enough BOOM SWING
bandwidth, low pressure losses at high flows, bi-
directionality, good repeatability and robust operation in
changing environmental conditions. [10] P T

IMV systems are still used in only a few commercial

applications; for example, the Caterpillar 336E H excavator, Figure 1. D-IMV valve system in excavator consisting of two
in which the IMV technology, or as they call it, the Adaptive pressure lines, 16 flow metering units and four actuators and
Control System (ACS), is used together with a hydraulic chamber pressure sensors.
hybrid system. [11]
Pressure line P in Figure 1 is an electronically controlled LS-
pump pressure supply line and the pressure line T is a return
1.1 Digital hydraulic independent metering valves
line with a low pressure realized with a spring loaded check
D-IMV systems consist of 4 independently controlled valve or a pressure relief valve. Each of the flow metering
metering edges per actuator. Each of the metering edges units, also known as digital flow control units (DFCUs),
consists of a series of parallel connected on/off-valves. The consists of seven on/off-valves with mixed PCM (Pulse Code
benefits of parallel connected digital valves, compared to Modulation) & PNM (Pulse Number Modulation) coding of
traditional proportional valves, are the fault tolerance, the the valve sizes. In PCM coding, each valve has a different
robustness and the fully repeatable and fast flow control. In flowrate capacity, and the number of different available
some of the previous studies, a fifth flow metering unit flowrate combinations is 2n-1, where n is the number of
between the A and B chambers also has been proposed, to valves. In PNM coding, each valve is equally sized. In
give an extra degree of freedom for the system. [9] practise, the flowrate capacity is adjusted by adding an orifice
in line with part of the valves. Table 1 presents selected valve
In most cases, both IMV and D-IMV systems require pressure sizes for this study. [12]
sensors in actuator ports, in the supply line and in the return
line, and sophisticated control algorithms that estimate the Table 1. Selected flow rate series of each metering edge of the
load force and act upon it. Algorithms of IMV and D-IMV excavator. [l/min @ 0.5MPa]
systems generally consist of the mode selection, and to
calculate a target flow rate for each metering edge. [8] Valve: 1 2 3 4 5 6 7
Boom: PA-edge, PB-edge
2 Digital hydraulic IMV in an excavator & BT-edge. Arm & Bucket: 35 35 35 17,1 8,7 4,3 2,2
all metering edges
2.1 Digital hydraulic valves for an excavator Boom: AT-edge, Swing: all
35 35 17,1 8,7 4,3 2,2 1,2
metering edges
The D-IMV studied in this paper consists of four actuators:
The boom, the arm, the bucket and the swing. Independent
Figure 2 shows the control–flow characteristics of both of the
metering is realized with four flow metering units per
control edge varieties presented in Table 1. The number of
actuator. The system layout is shown in Figure 1.
different possible opening combinations varies from 68 to 96.
Control resolution, i.e., the relation between the maximum
flowrate and the maximum flowrate step between two
consecutive steps varies from 52.3 to 71.5. Resolution was
selected based on simulation responses to handle cylinder
motions without additional oscillation compared to the
measured system and also based on available on/off valves.

Peer-reviewed Paper, 137 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden
 MASTER
CONTROLLER Boom Controller FLOW
pPref CONTROLLER P↔A
Pump mode u_valves
controller FLOW
CONTROLLER A↔T

VALVES
D-IMV
Power MODE
distribution v_ref CONTROLLER FLOW pA
pP CONTROLLER P↔B pB
Parameter pT
Handling FLOW
CONTROLLER, B↔T

VALVES VALVES
ARM Controller

D-IMV D-IMV
Mode Controller

BUCKET Controller
Mode Controller

VALVES
D-IMV
SWING Controller
Mode Controller

Figure 4. D-IMV controller structure.

Figure 2. Possible flow rates of metering edges PA and AT of
the boom at 1.5MPa. The upper-level controller structure presented in Figure 4
includes four actuator controller subsystems and a master
2.2 Operation modes of IMV controller subsystem that includes the pump controller and
the power distribution. Actuator controllers are identical to
The different modes are referred to as: PTe, PTr, TPe, TPr,
each other with a different set of parameters. Parameters are
PPe, PPr, TTe and TTr, in which the first letter describes
tuned in this study so that velocities and velocity oscillations
whether the pump line (P) or tank line (T) is connected to the
are close to the measured responses of the reference excavator
A-chamber, while the second letter describes which line is
with LS-valve system.
connected to the B-chamber. The third letter describes the
direction of the movement, e for extending and r for
retracting. Figure 3 presents the active metering edges with 3.1 Master Controller
each mode. The master controller structure is presented in Figure 5.
Inputs to the controller are the measured supply line pressure
pP and the return line pressure pT, joystick signals and pump
pressure requests from the actuator controllers. Outputs are
the velocity references for the actuators, pump control signal
and the filtered supply line and return line pressures for the
actuator controllers. The Master controller realizes the eLS
pump control by selecting the maximum of actuator supply
pressure requests as a pump pressure reference and turns it
into a control signal for the pump control valve. The Power
and torque distribution block ensures that the engine will not
stall and that the desired operation of actuators is ensured
when the power demand exceeds the available hydraulic
power.

Figure 3. Different operation modes of the D-IMV system.

pPref_bucket
pPref_swing

pPref_boom
pPref_arm

3 Controller user input

Many of the digital hydraulic valve controllers presented in Filter Gain

earlier studies have been designed for powerful computers,
pPfil,
such as the dSPACE Microautobox or PCs. More accurate max pTfil
Qref
and smoother control can be achieved if a more
computationally demanding, model based valve control is Flow
used, but it has been shown in [13] that sufficient control for pPref distribution
mobile application can be achieved also using simpler (Power limit)
controllers that can be realized with commercially available
pPref
controller units.
Pump vrefs
Controller
u_pump

Figure 5. Master controller structure.

Peer-reviewed Paper, 138 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden
3.1.1 Power distribution and limitation To achieve equal operation in simulations and in the reference
measurements, power distribution is limited by only limiting
In case the controller power demand or flow rate demand
the swing torque. In case the boom, arm or bucket actuators
exceeds the machine limits, velocity references are lowered
have bigger pump pressure requests than the swing actuator,
so that the total hydraulic power requirement is within the
the swing torque cannot be limited and only the velocity
power limits of the system and maximum flow rate is not
reference of the swing actuator is reduced.
exceeded. In case the power demand exceeds the power limit,
new velocity references for the actuators are calculated with
3.1.2 Motion Controller
an equation
In the simulation study, cylinder velocity references are
𝑃𝑙𝑖𝑚𝑖𝑡 (𝑛𝑒𝑛𝑔𝑖𝑛𝑒 ) (1) derived from the position and velocity measurements of the
𝑣𝑟𝑒𝑓𝑛𝑒𝑤 = 𝑣𝑟𝑒𝑓 𝑊𝑑𝑖𝑠𝑡 reference machine. Reference velocities for the control
𝑄𝑃𝑟𝑒𝑓 ∙ max(𝑝𝑃𝑅𝑒𝑓𝑠 )
system are calculated with a motion controller realizing the
equation

where vref is the user velocity reference, Plimit(nengine) is the 𝑣𝑟𝑒𝑓 = 𝐾𝐹𝐹 𝑣𝑚𝑒𝑎𝑠 + 𝐾𝑃 (𝑥𝑚𝑒𝑎𝑠 − 𝑥𝑠𝑖𝑚 ) (5)
available power with the current engine speed, 𝑝𝑃𝑅𝑒𝑓𝑠 is the
where KFF is feedforward gain and KP is feedback gain, xmeas
vector of supply pressure references from the actuator
is the reference cylinder position from the reference
controllers, Wdist is a weighting factor for the actuator to
measurements and xsim is the actuator position in simulations.
determine the priority of the actuator velocity reference and
The motion controller is only required for simulation
QPref is the total flow requirement. Wdist is calculated with
purposes.
equation
For the swing actuator, measured pilot pressures of the
̅̅̅𝑑𝑖𝑠𝑡𝑠,𝑘 𝑣𝑟𝑒𝑓⁄
𝑾 (2)
control valve in the reference measurement are used to define
̅
𝒗 𝑚𝑎𝑥𝑠,𝑘
𝑊𝑑𝑖𝑠𝑡 = ∙ the start moment of the joystick actuation and a smooth
̅𝑟𝑒𝑓𝑠
̅̅̅𝑑𝑖𝑠𝑡𝑠 𝒗
∑(𝑾 ⁄𝒗
̅ ) trajectory going from 0 to 1 is applied to mimic the joystick
𝑚𝑎𝑥𝑠
control signal. Control duration is applied so that the steady-
where 𝒗̅𝑟𝑒𝑓𝑠 is the vector of velocity references, 𝒗
̅𝑚𝑎𝑥𝑠 is the state swing positions are equal to the measured reference
̅̅̅
vector of maximum actuator velocities and 𝑾𝑑𝑖𝑠𝑡𝑠 is the positions.
vector including all four weighting gains of four actuators. By
adjusting the relations between the gains, a desired relation of 3.2 Mode controller
the actuator’s velocities can be achieved. 𝑄𝑃𝑟𝑒𝑓 is calculated
The mode controller structure is presented in Figure 6. The
as a sum of actuator flow rate demands, which is calculated
mode controller is built with a sub-optimal principle. It is
using the equation
known that typical pumps used in mobile machines have poor
𝑣𝑟𝑒𝑓_𝑖 𝐴𝐴_𝑖 , 𝑣𝑟𝑒𝑓_𝑖 > 0 efficiency at high pressures and low flowrates, and that high
(3)
𝑄𝑃𝑟𝑒𝑓_𝑖 = { pressures also tend to wear system components and thus to
|𝑣𝑟𝑒𝑓_𝑖 |𝐴𝐵_𝑖 , 𝑣𝑟𝑒𝑓_𝑖 < 0 increase maintenance costs. The differential modes are not
where AA and AB are the cylinder chamber areas or the radian used to calculate the pump pressure reference, but are still
chamber volumes of the swing motor. used when possible to lower the pump flowrate demand.

If flowrate demand is exceeded, new velocity references are

calculated with equation

𝑄𝑙𝑖𝑚𝑖𝑡 (𝑛𝑒𝑛𝑔𝑖𝑛𝑒 ) (1)

𝑣𝑟𝑒𝑓𝑛𝑒𝑤 = 𝑣𝑟𝑒𝑓 𝑊𝑑𝑖𝑠𝑡
𝑄𝑃𝑟𝑒𝑓

where Qlimit(nengine) is the available power with the current

engine speed.

In the case of the swing actuator, desired operation is

achieved also by limiting the swing torque. New pressure
reference for the swing actuator is calculated with the
equation

𝑝𝑃𝑟𝑒𝑓_𝑠𝑤_𝑛𝑒𝑤 (4)
Figure 6. Mode controller flow chart.
𝑃𝑙𝑖𝑚𝑖𝑡 (𝑛𝑒𝑛𝑔𝑖𝑛𝑒 )
= 𝑝𝑃𝑟𝑒𝑓_𝑠𝑤 𝑊𝑑𝑖𝑠𝑡
𝑄𝑃𝑟𝑒𝑓 ∙ max(𝑝𝑃𝑅𝑒𝑓𝑠 ) Actuator force or torque is estimated from the measured
chamber pressures and using a filter developed in [6]. This

Peer-reviewed Paper, 139 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden
non-linear filter improves the dynamics of the force 𝑝𝑜𝑢𝑡 = min ( max( 𝑝𝑚𝑖𝑛 , 𝑝𝑇 + 𝑑𝑝𝑜𝑢𝑡𝑟𝑎𝑡𝑖𝑜 (8)
estimation, although a linear first-order filter works well in a ∙ 𝑑𝑝𝑟𝑒𝑓 ) , 𝑝max )
simulation environment.
where pmin and pmax are given lower and higher chamber
Most of the mode selection code is realized with a pressure limits and dpout_ratio is a parameter to lower the
Matlab/Stateflow chart. The mode controller changes state outflow side target pressure difference to improve the energy
from idle to running mode when |vref | is above the tolerance efficiency.
velocity vtol. If the actuator is moving, it goes to the idle mode
when |vref| < vtol/2. Inflow-side chamber pressure is then calculated with the
equation
3.2.1 Target pressure difference
(𝐹 + 𝑝𝑜𝑢𝑡 𝐴𝐵 )/𝐴𝐴 , 𝑣>0 (9)
Target pressure difference over the valves is calculated as a 𝑝𝑖𝑛 = {
(𝐹 − 𝑝𝑜𝑢𝑡 𝐴𝐴 )/𝐴𝐵 , 𝑣<0
function of the velocity reference vref to allow minimum
pressure difference during low and mid-high velocities, and
increased cylinder velocity by increasing the pressure where AA and AB are the piston areas or radian volumes of the
difference over the valves. There are two parameters for motor. From these pPref is calculated with equation
setting the target pressure difference: minimum pressure
difference target (dpnom1) and the maximum pressure 𝑝𝑃𝑟𝑒𝑓 = min(max(𝑝𝑃𝑚𝑖𝑛 , 𝑝𝑚𝑖𝑛 (10)
difference target (dpnom2). dpref is calculated with equation + 𝑑𝑝𝑟𝑒𝑓 ), 𝑝𝑃max, 𝑝𝑖𝑛 + 𝑑𝑝𝑟𝑒𝑓 )

(6)
𝑑𝑝𝑟𝑒𝑓 = 𝑚𝑎𝑥 (𝑑𝑝𝑛𝑜𝑚1 , min (𝑑𝑝𝑛𝑜𝑚2 , where pPmin and pPmax are limits for the supply pressure.
2
|𝑣𝑟𝑒𝑓 |𝐴𝐴 3.2.3 Calculation of possible modes
( √𝑑𝑝𝑛𝑜𝑚1 ) ))
𝑄𝑑𝑝1𝑛𝑜𝑚_𝑚𝑎𝑥
After pPref is calculated, the current possibility of each mode
is checked. For the positive direction, the possibilities of
where Qdp1nom_max is the flowrate when the valve metering modes are calculated with equations:
edge P↔A is fully open and the pressure difference is equal
to dpnom1. PT_pos = (pPref - pP) < 0.5dpref (11)
PT_notpos = pP < (pPref - dpref)
For the swing actuator sufficient acceleration is done by using TP_pos = F < force(pT - dpref, pP + 0.5dpref)
dynamic target pressure difference calculation that increases TP_notpos = F > force(pT -dpref, pP + dpref)
the dpref value when joysticks move rapidly. This will PP_pos = F < force(pP - 2dpref, pP + dpref)
increase the pump supply pressure more rapidly when swing PP_notpos = F > force(pP - dpref, pP + dpref)
motion is started, and thus will improve the acceleration of TT_pos = F < force(pT - 2dpref, pT + dpref)
TT_notpos = F > force(pT - dpref, pT + dpref)
the swing motion. For the swing motion, dpref is calculated
with the pseudo code:
where the force(p1,p2) calculates the equation
if (vref - vref_previous ) > vref_diff_limit
 dpref=min(dpnom2,dpref_previous+inc_rate)
𝑓𝑜𝑟𝑐𝑒(𝑝1, 𝑝2) = 𝑝1𝐴𝐴 − 𝑝2𝐴𝐵 (12)
else if dpprevious> dpnom2
 dpref = max(dpnom1, dpref_previous –inc_rate/3)
else For the negative directions the possibilities are calculated
 dpref = dpnom2 with equations:

where vref_diff_limit is the parameter describing how fast the TP_pos = ( pPref - pP ) < 0.5dpref (13)
joystick motion must be for starting to increase dpref, and TP_notpos = pP < ( pPref - dpref )
inc_rate is a parameter to adjust the length of the increased PT_pos = F > force( pP - 2dpref, pT - dpref )
dpref duration. PT_notpos = F < force( pP - dpref, pT - dpref )
PP_pos = F > force( pP + 2dpref, pP - dpref )
PP_notpos = F < force( pP + dpref, pP - dpref )
3.2.2 Supply pressure reference TT_pos = F > force( pT + dpref, pT - 2dpref )
The pump pressure reference pPref is calculated for the TT_notpos = F < force( pT + dpref, pT - dpref )
standard inflow-outflow modes (IO-modes), first by
calculating the required chamber pressures from the 3.2.4 Mode selection
force/torque (F) and given limitations. First, the outflow-side
pressure is calculated with the equation: The mode for extending motion is selected in the state
machine presented in Figure 7. For the rectracting motion the
princible is identical. For certain work cycles the controller
cannot utilize all the modes, and thus the tables

Peer-reviewed Paper, 140 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden
enable_ext_modes and enable_ret_modes are declared to 3.3 Flow Controller
allow only certain modes to be used. The logic of the state
The actuator controller includes a flow controller (Figure 8)
machine decides the best possible mode, or if none of the
for each metering edge, to calculate the optimal opening of
modes is possible, the logic selects the hold mode to wait for
the digital hydraulic valves. When the mode selection
the proper supply line pressure.
algorithm sends the flow requirement and target pressures for
the active metering edges, the flow controller calculates the
flowrates for each valve with the equation

𝑄𝑣𝑎𝑙𝑣𝑒𝑖 (𝑝1 , 𝑝2 ) (12)

𝑥
𝐾1𝑖 (𝑝1 − 𝑝2 ) 1𝑖 , 𝑏1𝑖 𝑝1 < 𝑝2 ≤ 𝑝1
𝐾1𝑖 [(1 − 𝑏1𝑖 )𝑝1 ]𝑥1𝑖 , 𝑝2 ≤ 𝑏1𝑖 𝑝1
= 𝑥2𝑖
−𝐾2𝑖 (𝑝1 − 𝑝2 ) , 𝑏2𝑖 𝑝21 < 𝑝1 ≤ 𝑝2
[(1 )𝑝 ] 𝑥2𝑖
{ −𝐾 2𝑖 − 𝑏2𝑖 1 , 𝑝1 ≤ 𝑏2𝑖 𝑝2
where K1i is the flow coefficient, x1i is the exponent and b1i is
the critical pressure ratio of a single valve in the flow
direction 12, and K2i, x2i, and b2i are the parameters for the
opposite direction. For the simulation study, exponent values
of 0.5 and critical pressure ratios of 0.25 were used. Flow
rates of the individual valves are presented in Table 1 in
Chapter 4.

Figure 7. Mode selection in the state machine for positive

direction.

3.2.5 Calculation of reference values for the flow

controller Figure 8. Flow Controller structure of Boom PA metering
edge.
After the mode is selected, the controller selects which of the
metering edges are to be used. For these metering edges, To avoid excessive valve switching, the final valve opening
reference flowrates and chamber pressures are calculated for combinations are selected by minimizing the cost function,
each valve. taking into account velocity error and the amount of valve
state switching required. The valve opening combination is
Target flowrates are calculated for PA and AT-metering calculated with the equation
edges with equation
𝑢𝑠𝑡𝑎𝑡𝑒 = min(𝑣̅𝑒𝑟𝑟𝑜𝑟 (13)
𝑄𝑟𝑒𝑓,𝑃𝐴/𝐴𝑇 = 𝑣𝑟𝑒𝑓 ∗ 𝐴𝐴 (13) + 𝑤𝑠𝑤 (𝑄̅𝑁𝑂𝑀
×|(𝑢̅𝑝𝑟𝑒𝑣 ×𝑢̅𝑜𝑛𝑒𝑠
and for metering edges PB and BT with equation − 𝑢̅𝑐𝑡𝑟𝑙_𝑚𝑡𝑟𝑥 )|)
where 𝑣̅𝑒𝑟𝑟𝑜𝑟𝑠 is the array of the velocity errors for each valve
𝑄𝑟𝑒𝑓,𝑃𝐵/𝐵𝑇 = −𝑣𝑟𝑒𝑓 ∗ 𝐴𝐵 (13)
opening combination, 𝑤𝑠𝑤 is the weighting gain for the
switching term, 𝑄̅𝑁𝑂𝑀 is a vector of nominal flow rates of the
Pressure references for the flow controller are set so that valves, 𝑢̅𝑝𝑟𝑒𝑣 is an array of the current state of valves, 𝑢̅𝑜𝑛𝑒𝑠
outflow-side is minimized, and maximum and minimum is an all-ones array with a length equal to the number of
pressure limits are not exceeded. The inflow-side chamber combinations and 𝑢̅ctrl_mtrx is a matrix of valve states with
pressure is set according to the current force level. each combination.

Peer-reviewed Paper, 141 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden
4 Simulation Model & Reference Data The pump is modelled to have typical dynamics for an open
circuit mobile pump, but the efficiency of the pump is
The machine selected for comparison is a 21-ton LS- neglected because it is out of the scope of this study. The
controlled excavator (Volvo EW210C, Figure 9). This return line is modelled as a constant pressure line.
excavator has a single pump for controlling all the actuators
and the travel motors, and additional small auxiliary pumps 4.2 Reference Data
for fan motor, pilot pressure supply and for other necessities.
The excavator was instrumented with a pump flowrate sensor, Reference data include measurements of a simulative 90-
actuator position sensors and actuator-chamber pressure degree truck loading cycle and earth grading cycle that are
sensors. based on a JCMAS H 020 standard (Figure 11). The standard
defines trajectories for comparing fuel consumptions of
different sized machines in real life-like movements without
touching the earth. This standard is selected to avoid the
complex modelling of the earth and to realize as uniform
conditions as possible for repeatable measurements, although
it is does not match exactly the real world test cycles.

Figure 9. Volvo EW210C wheeled excavator used in

reference measurements.

4.1 Simulation model

The simulation model (Figure 10) includes the D-IMV
controller and the mechanical model of the excavator realized
with Matlab Simulink. The mechanical model includes the Figure 11. Truck loading cycle phases 1-5 and the grading
inertias and kinematics modelled with SimMechanics, cycle according to JCMAS H020. [15]
including the cross connection between the actuators. Supply
pressure dynamics, control valves, hoses and actuators are
modelled with Simulink. Model was verified to the 5 Simulation results
measurement data from the excavator.
5.1 Truck loading cycle
Figure 12 presents the actuator positions and velocities in a
single work cycle. There are some differences in velocity
curves, but the final positions of simulations and
measurements match. The simulation tends to oscillate a little
more when hitting the end of the actuators because of the lack
of a cylinder end damping model.

Figure 10. Hydraulic and mechanical model of the EW210C

in Simulink.

The model of the digital valves includes four valve blocks,

including 4x7 leak free on/off-valves per actuator. Flow rates
of the on/off-valves are adjusted with orifices to
accommodate the maximum velocity of the actuators and a
control resolution that also allows low velocities.

The diesel engine is not modelled because the focus of the

paper concentrates on the hydraulic efficiency comparison.

Peer-reviewed Paper, 142 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden
 small that it can be considered negligible. Most of the
difference comes from the behaviour difference of the swing
drives. The peaks in the measured input power at phase 4 are
due to the arm and bucket actuators hitting the end of the
cylinder. The simulated system does not increase the cylinder
force after hitting the end. In a truck loading cycle, a rapid
force increase is not required when opening the bucket. The
difference in the input energy comes from the lowered flow
rate requirement because of the utilization of different modes.

3 4 5
Figure 14 presents the different modes used during the cycle.
Phase 1 consists mostly of lifting the boom and extending the
arm cylinder. As the boom lifting requires high pressure, the
arm operates at differential modes and thus required input
energy is reduced the around 40 percent. During the closing
of the bucket in phase 2, the bucket cylinder is the only
1 2 actuator and it operates in inflow-outflow mode. During the
phase 3, inflow-outflow modes are used by the boom and the
swing actuator, and in phase 4 the bucket is opened in inflow-
outflow mode. When returning to the starting position, the
boom is lowered with the TTr –mode, in which zero pump
Figure 11. Positions and velocities from the reference flow is used simultaneously with the arm retraction and swing
measurements (dashed line) and from the simulated D-IMV motion, which are done in inflow-outflow mode.
system (solid lines) with different phases of the JCMAS
digging cycle.

The upper graph of the Figure 13 presents the input and

output powers of the simulated D-IMV system and the
measured LS system, as well as the supply line pressure and
flow rates. It can be seen that 19 kJ more work was done in
the simulations than what was measured. This can be
explained with a slight difference in velocity curves and in
force/torque curves, but the relation to the input energy is so

1 2 3 4 5

1 2 3 4 5

1 2 3 4

Figure 13. Hydraulic input powers (PIN) and mechanical output powers (POUT), pump pressures and pump
flow rates of the measured LS reference system (meas.) and the simulated D-IMV system (sim.) in a JCMAS
truck loading cycle (phases 1-5 divided in figure).

Peer-reviewed Paper, 143 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden
 motion. Also, the metering losses are reduced during the in-
to-out motion.

Figure 14. D-IMV Actuator modes in digging cycle

simulations.

Figure 15 presents the cumulative input and output energies

of four truck loading cycles. The total input energy was
reduced 28 percent with D-IMV compared to the measured
LS system.

Figure 16. Simulated (D-IMV system) and measured (LS

system) grading cycle position and velocities and modes used
in simulation.

Figure 18 presents the cumulative input energies in four

consecutive cycles. The total required energy was reduced
41.5% compared to the measured LS system.

Figure 15. Measured and simulated cumulative energy

consumptions in four consecutive truck loading cycles.

Total energy consumption of the working hydraulic circuit

was reduced 28% in the digging cycle.

5.2 Grading cycle

Figure 16 presents the simulated and measured actuator
Figure 18. Measured and simulated cumulative input and
positions, velocities and modes in a single grading work
output energies of four grading cycles.
cycle. The bucket tip remains at a steady height from even
ground surface, and the bucket is drawn from the out position
to the in-position, and back.

Figure 17 presents the measured and simulated input and

output powers, supply pressures and pump flowrates in the
grading cycle. In the grading cycle, the main difference
comes from the arm differential mode PPe used in the out-to-
in motion and the boom TTr-mode used in the in-to-out

Peer-reviewed Paper, 144 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden
 Figure 17. Hydraulic input powers (PIN) and mechanical output powers (POUT), pump pressures and pump flow
rates of the measured LS reference system (meas.) and the simulated D-IMV system (sim.) in a JCMAS grading cycle.

Contribution to Increased Efficiency Through Fluid

6 Conclusion Power on Mobile Machines. Fuidotecnica, 345.
D-IMV systems, and IMV systems in general, have great
potential to improve the efficiency of mobile machines that [2] Vukovic, M., Leifeld, R., and Murrenhoff, H., 2016.
often use working hydraulics with several actuators. This STEAM a hydraulic hybrid architecture for excavators.
paper presented study comparing the measurement from a 10th International Fluid Power Conference (10. IFK), 8–
state-of-the art LS excavator and a simulated digital hydraulic 10 March, Dresden, Germany.
IMV excavator. Comparison was done with a controller that
utilizes different modes of the actuator, and thus lowers the [3] Linjama, M., Vihtanen, H-P., Vilenius, M. Secondary
flow rate required from the pump. The controller also keeps controlled multi-chamber hydraulic cylinder, The 11th
the supply pressure lower or at a similar level than that of Scandinavian International Conference on Fluid Power,
current LS-systems. SICFP’09, June 2-4, 2009, Linköping, Sweden

In this study, system modifications other than those involving [4] K. Heybroek, G. Vael, and J.-O. Palmberg, “Towards
the valve system were kept to a minimum so as to compare resistance - free hydraulics in construction machinery,”
the IMV system to a state-of-the-art LS-valve system in a 21- in The 8th International Fluid Power Conference, 8. IFK,
ton excavator. The comparison included only hydraulic input Dresden, Germany, 2012.
energies, and it was shown that even without regenerative
modes, hydraulic energy required for the two standard [5] C. Williamson and M. Ivantysynova, “Power
working cycles can potentially be decreased 28-42 percent. optimization for multi-actuator pump- controlled
systems,” in The 7th International Fluid Power
Research continues in applying D-IMV technology to an Conference, 7. IFK, Aachen, Germany, 2010.
actual test case to validate the results and study the system in
actual work cycles. [6] K. Einola and M. Erkkilä, “Dimensioning and control of
a hydraulic hybrid system of a cut-to-length forest
Acknowledgment harvester,” in The 9th International Fluid Power
Conference, 9. IFK, Aachen, Germany, 2014
This work was supported by the Doctoral School of Industry
Innovations (DSII). [7] Lin, T., Wang, Q., Hu, B., & Gong, W. (2010).
Development of hybrid powered hydraulic construction
References machinery. Automation in Construction, 19(1), 11–19.
https://doi.org/10.1016/j.autcon.2009.09.005
[1] Lettini, A., Havermann, M., Guidetti, M., & Fornaciari,
A. (2010). Electro-Hydraulic Load Sensing: a

Peer-reviewed Paper, 145 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden
[8] Eriksson, B., Palmberg, J.-O., Individual metering fluid
power systems: Challenges and opportunities.
Proceedings of the Institution of Mechanical Engineers,
Part I: Journal of Systems and Control Engineering,
225:196–211, 2011.

[9] Nielsen, B. (2005). Controller Development for a

Separate Meter-In Separate Meter-Out Fluid Power
Valve for Mobile Applications, 237. Available
http://vbn.aau.dk/files/74051350/brian_nielsen.pdf

[10] Heybroek, K., Larsson, J., Palmberg, J.-O., Mode

Switching and energy recuperation in open-circuit pump
control. In The Tenth Scandinavian International
Conference on Fluid Power, Tampere, Finland, May 21-
23 2007. Available http://liu.diva-
portal.org/smash/get/diva2:133010/FULLTEXT01.pdf

[11] Caterpillar website,

http://www.cat.com/en_ID/news/machine-press-
releases/cat-sup-174-sup-
336ehhydraulichybridexcavatordeliversnocompromis.ht
ml , visited 10.1.2017

[12] Linjama, M., Digital fluid power - state of the art. In The
Twelfth Scandinavian International Conference on Fluid
Power, Tampere, Finland, May 18-20 2011, available:
https://pdfs.semanticscholar.org/a1a3/afd27352191866b
5614af9abea93ff5a9cf7.pdf

[13] Huova, M., Ahopelto, M., Ketonen, M., Ahola, V.,

Linjama, M., Huhtala, K., Characteristics of digital
hydraulics with commercial controllers, The Seventh
Workshop on Digital Fluid Power, February 26-27,
2015, Linz, Austria.

[14] Karvonen, M. (2016). Energy Efficient Digital Hydraulic

Power Management of a Multi Actuator System.
(Tampere University of Technology. Publication; Vol.
1384). Tampere University of Technology.

[15] JCMAS H020. Japan Construction Mechanization

Association. Standard for of excavator fuel efficiency
measurement procedure. 2010.

Peer-reviewed Paper, 146 SICFP2017 7-9 June 2017

Accepted for publication on 2017-05-01. Linköping, Sweden