Evaluating a Pump Controlled Open Circuit Solution

Transcription

1 Paer Number. Ealuating a Pum Controlled Oen Circuit Solution Kim Heybroek, Jan-Oe Palmberg luid and Mechanical Engineering Systems, Linköing Uniersity, Sweden Johan Lillemets, Martin Lugnberg, Martin Ousbäck Volo Construction Equiment ABSTRACT In this article the authors hae studied a hydraulic system configuration where each actuator/suly system comrises an electrically controlled ariable dislacement um/motor working in an circuit together with four searate electrically controlled ales. Performance, oerability and energy consumtion are ealuated in a wheel loader, first with its original load sensing hydraulic system and then modified with a comletely new um controlled hydraulic system. Measurements ublished in this article demonstrate the adantages and drawbacks of um control in an circuit comared both to a um controlled, closed circuit solution and to a ale controlled, load sensing system. Performance is ealuated by looking at the fuel consumtion, roductiity, and oerability of a wheel loader. Theoretical calculations of energy efficiency including the dominant losses in system comonents are ealuated and alidated. uel consumtion is measured and ut side by side with measurements of the same machine equied with a load sensing hydraulic system. INTRODUCTION There are three energy related key benefits using an circuit um controlled system instead of a traditional ale controlled system in a wheel loader. Obious drawbacks in a um controlled system are the greater number of comonents and the increased comlexity in control. By studying the results from a demonstration machine the adantages and drawbacks of the circuit solution is comared to state-of-theart. In this study the test results and exerience of the hydraulic system imlemented in a medium-sized wheel loader are resented. The emhasis in this study is to comare the erformance in circuit um control to other um controlled solutions as well as to conentional load sensing technologies. The comarison rimarily coers energy efficiency and system comlexity leel. THE OPEN CIRCUIT SOLUTION CONCEPT If the um does not hae a re-defined high and low ressure comartment, it is said to work in a closed circuit. In the other case, when the um only oerates against high ressure on one side, it is working in an circuit. The circuit solution, in this article referred to as OCS, comrises four searate ales combined with a number of sensors on to of an electronically controlled um, used for either ressure or flow control. igure illustrates the simlified schematics of an imlementation of the OCS on two dries. The two functions are mechanically connected in arallel ia the um shaft. There are three ressure sensors for each drie, which is a rerequisite to achiee all the modes of oeration studied in this article. The required controllers for the ums and ales and a suerisory control are also shown in the figure. The OCS was first resented by K. Heybroek, J-O. Palmberg in (). ) No metering losses arise because of unequal drie ressure leels, due to searate ums for each drie. ) The ressure dro oer ales is maintained at a low leel as to ales are not used for flow throttling at any great extent. This includes energy recueration. ) The differential ressure resulting in a meter-in ressure dro, caused by the load sensing architecture is eliminated. Suerisory controller ale contr. um contr. on/off and roortional ales -circuit, cross-center um ale contr. um contr. anti-caitation check ales ressure transducers further dries igure - Simlified circuit diagram of the circuit solution imlemented on two dries.

2 STATE-O-THE-ART Today s mobile machines most often contain ale controlled dries arranged in an loo circuit. or the urose of saing energy the constant ressure systems hae often been relaced by load sensing ums and load sensing ales. Regarding um-controlled solutions, some concets hae reiously been deeloed by, for instance; Habibi, S. and Singh, G. (), Wendel, G. (), Rahmfeld, R. and Iantysynoa, M. (),. or arious reasons, not all of these concets are suitable for mobile alications. LOAD SENSING SYSTEMS In mobile alications the load sensing solutions hae, in the ast, significantly reduced the energy consumtion. The load sensing concet renders a cost effectie and comact solution, as there is usually only a need for one um to ower seeral cylinder dries. The load sensing systems use ariable ressure controlled ums, where the ressure reference is set by the cylinder load haing the greatest ressure magnitude. Howeer, in alications with seeral dries oerating at unequal drie ressure leels, the load sensing systems still result in energy losses, referred to as metering losses. A ressure difference oer the ale is required to create flow which yields a load elocity. This is achieed by the um controller, often hydromechanically setting the suly ressure to the load ressure lus an additional differential ressure, here referred to as LS-. The LS- yields a ressure dro oer the ale and is thereby yet another source of losses. Another inherent roblem with the load sensing solution is the throttling losses associated with meter-out flow control instead of recuerating energy. The schematic of a conentional load sensing regulator is illustrated in igure. In this study the load sensing system is abbreiated LS-S. see igure. Additionally, shut-off ales for load holding in emergency situations, as well as high ressure relief ales are otional. If seeral dries are needed, these can be couled ia the low ressure side, sharing the charge um and the accumulator. The total number of comonents can consequently be ket low. In a similar way to a hydrostatic transmission this circuit handles four-quadrant actuation of the asymmetric cylinder in a simle hydromechanical manner, making the solution robust and reliable. The hydraulic machine works as um or motor deending on the load quadrant, automatically recuerating energy through the um shaft when ossible. In construction machinery the desired lowering seed can be comaratiely high, often twice as high as the lifting seed. This means that the um in a closed circuit must be dimensioned to handle the lowering flow. The closed circuit was imlemented in an O&K wheel loader and ealuated and comared to a standard ale controlled machine (). The measurements from the ealuation demonstrated a reduction in fuel consumtion of % in a truck loading scenario. The energy recueration caabilities are claimed to be the main reason for that reduction. control command EDC Q B, B Q A, A igure - Closed circuit solution schematics (7). LP TARGET APPLICATION THE WHEEL LOADER igure - A load sensing um with a conentional hydromechanical regulator (). PUMP CONTROLLED SYSTEMS The most cometitie um-controlled controlled hydraulic concet comarable to the OCS is the closed circuit solution, here referred to as CCS, resented by Rahmfeld and Iantysynoa. In the circuit, the differential olume of the asymmetric cylinder is balanced on its low ressure side by a charge um and an accumulator, The wheel loader is a true multiurose machine used in all sorts of enironments, igure. Howeer, the main uroses of the machine are often simle; lifting, lowering and transorting loads. To achiee this, both working hydraulics and a roulsion system are essential. These two systems dominate the useful energy consumtion to a large extent the nonuseful energy consumtion in the wheel loader. One reason why the wheel loader was chosen as the target alication of this study is the good ossibility to sae energy in the working hydraulics by energy recueration from lowering motions, energy which today is dissiated as heat.

3 control mode. In the OCS-rototye the ums are mounted in a tandem configuration shown in igure. igure - The target alication is a medium sized wheel loader. THE OPEN CIRCUIT SOLUTION PROTOYPE The OCS is customized to fit onto a medium sized retail wheel loader. The old load sensing system is still resent, but is now only used for steering. The original 8 cc um was relaced with two 8 cc ums, which is more than enough for steering. or the working hydraulics, two 7 cc electrically controlled circuit ums were added to the ower take-out (PTO). A software riority function resembles the system behaior of the reference machine in rioritizing the tilting drie. Also, a torque limitation is imlemented to justify the comarison with a system with less installed hydraulic ower. The working hydraulic LS-ale is relaced with a new ale block, with searated sections for the lift and the tilt functions. The owertrain configuration and auxiliary hydraulics left unmodified. An oeriew of the hydraulic system is illustrated in igure. igure - Electronically controllable ums, % dislacable in both directions, mounted in tandem. Vales A ale ackage for the lift and the tilt drie was manufactured, consisting of eight custom made seat ales, illustrated in igure 7. The ales used for meter-out flow control are roortionally controlled and electrically ressure comensated Valistor ales. igure 7 - Custom made ale block containing eight roortionally controlled oet ales. Tank ressure enhancement When oil is taken from the tank the ressure must be enhanced by about bar to aoid caitiation. In the OCS demonstrator this is achieed by the use of a ressure relief ale, set to a fixed cracking ressure, alternatie a) in igure 8. In alternatie b) the check ale is relaced with an electrically controlled on/off ale. Alternatie c) is a third solution, where the tank is a closed system in which the ressure has been enhanced by other means. Alt. a) Alt. b) Alt. c) P tilt P lift T common P tilt P lift T common P lift T common P tilt igure - Schematics of the hydraulic system of the OCS rototye wheel loader. Pums Two circuit ums ower the lifting and tilting dries resectiely. These ums are electrically controlled and are oerated either in flow or ressure T,enhanced igure 8 Otional solutions to tank ressure enhancement.

4 Sensors Three ressure sensors are used for each drie, where one is integrated in the um. In addition to ressure, shaft seed and swashlate angle are also measured. ressure sensor failure and the auxiliary control loo handles load holding functionality. In the actual OCS rototye, stes b, and are at this stage not yet imlemented. Control system The software used for control and data acquisition is imlemented in a real-time controller unit oerating at a samle rate of Hz. Mode Selection Power Management Mode Transition CONTROL STRATEGIES Load quadrant definition Power-need ealuation Define method for ressure matching This section describes the desired features of a controller for the circuit solution. The basic controller workflow is shown in igure 9, where the stes are: Mode efficiency ealuation Preliminary mode selection Reise selected control mode 7 8 Suly ressure build-u Deloy ale setting ) Identify the resent load quadrant by looking at load ressure leels and oerator command signals. ) The resent load ressure and the desired actuator elocity are used in an algorithm to select the best mode with regards to energy efficiency. The main comonents in such ealuations for recueratie motions are: a. Minimize losses in ales comaring different control modes. b. Minimize losses in comonents such as ums, cylinders and hydraulic lines, hae comaring different control modes. ) A reliminary mode is selected for further ealuation. ) Inestigate if there is any immediate need for energy to be recuerated and if the workload on the engine can be reduced. ) ind an alternatie control mode that matches the ower need, go back to and iterate oer the mode selection until a match is found. ) Gien the selected mode, define rules for how to match um and load ressure rior to ing ales. 7) Command a um ressure defined by. 8) Execute ale settings defined by the selected control mode. Deending on the chosen mode of oeration the um must be controlled differently. This is handled by the um controller which has the mode and oerator joystick signal as inut. The control signals are sent to the um, which has its own distributed electrical control system, regulating the swashlate angle. The control signal for the um is either relatie dislacement or ressure. A safety control loo handles for examle Pum control Safety functions control Auxilliary control igure 9 - Schematics of a roosed controller for the OCS when used in a wheel loader. THE IMPACT O MODE SELECTION The four ales render a solution ersatile in control, as the cylinder chambers can be connected to um and/or tank as well as be closed at any time. In reious studies the authors has inestigated what influence alternatie control modes hae on energy efficiency (8). What is here referred to as control mode is the way of controlling the load at a requested elocity, i.e. lifting/lowering a hanging load is achieed either in nondifferential or differential mode, either with or without meter-out flow control. The objectie is to recuerate the most energy ossible for a commanded secified motion and still fulfill the lowering seed requirements. Ideally, the recuerated energy equals the energy required to lift the load; in reality, this ratio is much lower due to losses. The losses are mainly related to the efficiency of cylinders, ales and ums (7). Measurement results from the OCS rototye show the differences in otentially recuerable energy in different modes of oeration. In the non-differential and differential control mode the maximum lowering seed is restricted by the maximum um flow. If meter-out control is used the lowering seed is restricted by the flow taken through the meter-out ale. In tests where meter-out is used the mean lowering time is secified as seconds. All tests are erformed with the diesel oerating at idle seed, 7 rm. The efficiency in recueration is here referred to as the recueration ratio, defined in Equation : Erecu recu = () E in where E recu is the recuerated energy outut on the um shaft and E in is the energy inut on the um shaft owering the hydraulics.

5 Non-differential lowering igure illustrates the ale setting and oerating region of the control mode non-differential lowering. Differential lowering igure illustrates the ale setting and oerating region for the differential lowering. Normal working region Differential working region q * n q * d igure - Left: ale configuration for nondifferential control. Right: working region for mode. As the entire load flow is controlled by the um in nondifferential lowering, no throttling losses related to meterout control are resent (the left efficiency lot in igure ). The um/motor efficiency combined with the ale losses in non-differential mode is illustrated to the right in igure. [N] x [N] x.9 igure - Left: efficiency including ale losses. Right: efficiency including ale and um losses. Lowering loads in non-differential mode is associated with unaccetable lowering times, almost seconds. In igure the ower and energy are illustrated for the mode. The bar to the right illustrates the recueration ratio calculated from measured um ressure, relatie um dislacement and calculated um/motor efficiency. The aerage recueration ratio is %. Normalized Power [-] Power Energy Power and energy -. Time [s] igure - Power, energy and recueration ratio. s Normalized Energy [-] recu.9 igure - Left: ale configuration for differential control. Right: working region for mode. Due to the differential couling the limited maximum allowable load force in this mode is decreased. Exceeding this limit requires ressure control in order to stay below maximum system ressure (). The entire load flow is still controlled by the um; the left lot in igure shows that no throttling losses related to meter-out control are resent, although higher lowering seed is achieed. [N] x [N].9 x igure - Left: efficiency including ale losses. Right: efficiency including ale and um losses. Lowering the load at the same engine seed as in the non-differential control mode the oeration now takes seconds instead of. Howeer, the recueration ratio is lower than for the non-differential control mode, aeraging %. Normalized Power [-] Power Energy Power and energy -. Time [s] Normalized Energy [-] recu igure - Power, energy and recueration ratio. s.9

6 Non-differential lowering + meter-out igure illustrates how increased lowering seed is achieed by meter-out flow control. Differential lowering + meter-out igure 9 illustrates how increased lowering seed is achieed by meter-out flow control. q-contr. n * Normal + meter-out working region q-contr. Differential + meter-out working region q * n q * d max * igure - Left: ale setting for non-differential + meter-out control. Right: working region for mode. In order to achiee the required lowering time of seconds, flow must be throttled oer the meter-out orifice, whereas the loss related to control mode increases dramatically. Only a fifth of the lowering flow goes through the motor, the rest is throttled to tank. The efficiency for this strategy is illustrated in igure 7. [N] x [N] x.9 igure 7 - Left: efficiency including ale losses. Right: efficiency including ale and um losses. The limit where meter-out control becomes necessary naturally deends on um size, engine seed and iston area. In igure 8 the bar to the right shows the recueration ratio for the lowering motion. The aerage ratio oer three cycles is 8%. The achieable lowering seed in this mode is a trade-off in recuerated energy ersus lowering time. Normalized Power [-] Power Energy Power and energy Time [s] igure 8 - Power, energy and recueration ratio. s Normalized Energy [-] recu.9 igure 9 - Left: ale configuration for differential + meter-out control. Right: working region for mode. The differential mode is not sufficient to fulfill the requirement of seconds in lowering time. Howeer, in this mode half of the flow is throttled to tank. Again, the um losses are considered in combination with the losses associated with the control mode. The resulting efficiency is resented in igure. It should be noted that the oerating conditions for the motor hae changed due to the transformation in ressure and flow that come with this mode. [N] x [N].9 x igure - Left: efficiency including ale losses. Right: efficiency including ale and um losses. The calculated recueration ratio is roughly double for the differential + meter mode comared to the nondifferential + meter-out mode. The aerage ratio calculated oer three cycles is % with a lowering time aeraging the targeted seconds. Normalized Power [-] Power Energy Power and energy Time [s] Normalized Energy [-] recu igure - Power, energy and recueration ratio. s.9

7 Conclusions from the mode selection ealuation: It is imortant to note that all calculations are based on measurements from a lifting and lowering oeration with the wheel loader standing still at idle engine seed. At low seed the hydraulic machine has a comaratiely high internal leakage, resulting in oor efficiency both in the lifting and the lowering hase. The fact that the recueration ratio is higher in the nondifferential mode than in the differential mode might aear surrising. This is a result of the additional ressure losses aarent in the differential mode due to its extended flow ath ia seeral shar bends and edges, illustrated in igure. If a dedicated ale, connecting the cylinder chambers, is installed closer to the load, the losses in differential mode will be reduced. Moreoer, the increased ressure leel in differential mode affects the cylinder efficiency, but to what extent has not been further inestigated within the framework of this study. Heay material is loaded and lifted u slowly to a truck, followed by raid lowering in differential mode. igure - Mode selection in different fields of alication. SYSTEM PERORMANCE Normal lowath Differential lowath The goal of this study is to ealuate the OCS with a degree of emhasis on energy efficiency and comare the system to a conentional load sensing system. In order to do so seeral tests were conducted. A first simle test is to erform a single function efficiency test, then roceed to a more comlex loading cycle efficiency test, finally looking at a long distance efficiency test. q igure - low ath in non-differential and differential mode. Imortant in this comarison is how great a recueration ratio can be achieed for a gien seed and force coule If < d * a higher recueration ratio is achieed in differential mode than non-differential mode + meter-out for the same seed q The erformance of the new hydraulic system is ealuated by comaring arious measurement results of two wheel loaders following the same assignment. Besides the modification, the machines must be equialent to ensure a credible comarison between the two. The machine chosen as reference machine is of the same brand and model as the new rototye with identical secifications as to the working hydraulics. igure shows the simlified schematics of the reference machine s hydraulic system. If < n * oeration in non-differential mode yields a higher recueration ratio than differential mode The differential meter-out mode is more energy efficient than the corresonding non-differential mode, which is robably a result of the higher um efficiency in that case. At load forces exceeding d * the differential mode is not an otion unless ressure control is alied. Howeer, the limits in seed and force are determined by the maximum um size and cylinder area roerties that can be otimized gien a secific field of alication. An examle of how different modes can be used to achiee a broader working region with maximized energy recueration is illustrated in igure. In case the wheel loader is used for material handling. High eleation is reached raidly and loaded material lowered carefully in non-differential mode. In case the wheel loader is used in a truck loading cycle. igure - Simlified hydraulic system layout of the reference machine.

8 EICIENCY TEST - SINGLE UNCTION It is interesting to comare the theoretical exectations as regards hydraulic energy reduction from energy recueration to the reduction in fuel consumtion in ractice. The test is erformed by simle lifting and lowering a defined load a secified number of times, measuring the fuel tank leel just before and after the oeration. In order to comare the results between the two machines it is necessary to mention a few things about the energy distribution in the wheel loader during this secific test. Only one function is used at a time throughout the test. The significant difference between the OCS-rototye and the reference machine is that energy is recuerated by the OCS machine. Assuming the hydraulic system to be the dominant consumer of useful diesel energy and the losses in the lifting hase are equal for the two machines; the reduction in fuel consumtion deends only on how much energy is recuerated. The ercentage reduction is thereby also directly comarable to the recueration ratio described in the reious section. The energy flow through the reference wheel loader is illustrated to the left in igure. irst, the fuel energy is transformed into heat and mechanical energy in the diesel, which is then transformed into hydraulic ower which is used to actuate the load. When lowering the load again the otential energy is transformed into heat in the hydraulic ales. No work has been achieed. To the right is the corresonding energy flow through the OCS rototye loader. Normalized fuel consumtion [-],,8,,, uel consumtion / max fuel consumtion 9.% decrease Heay w height bucket.% decrease Emty bucket igure - Single function efficiency test results. LS-S OCS The result with weight bucket is a 9.% imroement comared to the reference machine. The result with an emty bucket is a.% imroement. The weight bucket test is erformed in non-differential + meter-out control mode and the emty-bucket test in differential mode + meter-out. Relating to the theories resented in the reious section the test with weight bucket should be comared to the scenario where non-differential lowering with meter-out is used. Howeer, in this case the test is conceied with a greater load and a lower seed, resulting in lower throttling losses, and thereby a higher recueration ratio. The emty bucket test should be comared to the differential + meter out control mode. Also during this test, less flow is throttled to tank comared to the scenario described in the reious section, resulting in a recueration ratio closer to ure differential lowering. LS-S E LS-S,fuel OCS E (- e ) LS-S,fuel recu EICIENCY TEST TRUCK LOADING CYCLE Ed,loss ( E ) LS-S,fuel Diesel E h,other PTO E h hyd. E h,loss Ed,loss ( E ) OSC,fuel Diesel E h,other E h PTO E h hyd. E h,recu = e E h recu E h,loss igure - Illustration of energy saing from recueration in the single function efficiency test. Each cycle of lifting and lowering lasts for seconds and is reeated times for each machine. The test is erformed with two different loads: an emty bucket and a weight bucket. The lowering time is the same for both cases and takes about seconds. The measurement results are resented in igure. To ealuate the working hydraulics efficiency in a tyical field of oeration for the wheel loader, a truck loading cycle is considered. Oeration in the truck loading cycle is often the main urose of these machines; therefore, the cost effectieness of the cycle has economic significance for the entrereneur. The cycle illustrated in igure 7 lasts for about seconds and includes the machine going into a ile of granular material, lifting a full bucket, reersing out from the ile, forwarding to emty the bucket onto a truck, going back out while lowering the emty bucket and then starting the next cycle. The energy distribution in the truck loading cycle differs considerably from the single function test. In this cycle the working hydraulics, transmission, brakes and steering interact continuously with each other. What remains the same for both systems is the actual work being done during the loading cycle. The owertrain, including the roulsion system, is also the same for both systems. In the reference wheel loader about % of the diesel engine s outut ower is used by the roulsion system; the rest is used by the hydraulics.

9 P other Pidle, loss + Pcooling + Pcyl. fric., loss + Pline, loss = () The ower losses integrated oer the loading cycle yield the total energy loss for the load sensing hydraulic system. E loss cycle t.. tot ref = ( P, loss + Pthrottle + PLS Pother )dt () + t= The energy distribution in the reference wheel loader is resented in igure 8. The distribution is based on Equations -, using measurement data from a truck loading cycle. LS-S E LS-S,fuel E in,ls-s igure 7 - The short loading cycle is a tyical field of alication for the wheel loader (9). Diesel hydraulic system Losses in the reference wheel loader: What really differ between the two systems are the losses in the hydraulic system. The redominant losses in the hydraulic system of the reference machine are listed below: Ed,loss ( E ) LS-S,fuel E,loss PTO ro. hyd. E wheel E work E h,loss E work E throttle+metering losses E um losses E LS-D losses E other losses The um losses are calculated from measured ressure, shaft seed, relatie dislacement and maed um efficiency. P um, loss = P, i, D i i, n i= drie, tot ε () The throttling losses are calculated from the energy dissiated oer meter-out orifices during regeneratie motions measuring actuator seed and ressure dro from load to tank side. P throttle = Ai x, i ( L T ), i i= drie () The LS- losses are calculated from the sum of ressure losses in the load sensing ales using measured actuator seed and ressure dro from um to load side. P LS = Ai x i ( P L), i i= drie () All other hydraulic losses are collected into the term P other in this text listed in order of imortance; energy required for cooling, idle losses in ums, cylinder friction losses and line losses. igure 8 - Energy distribution in the wheel loader during the truck loading cycle, using the load sensing system. Losses in the OCS-rototye wheel loader: or the OCS rototye losses in the hydraulic machine working as a motor, occur during the recueratie motions, calculated from measured ressure, shaft seed, and relatie dislacement and maed efficiency. ( ) P motor, loss = m, i m, i D, i n m, tot i= drie ε (7) Howeer the P LS- term is no longer alid and the P throttle term is significantly reduced due to the ossibility in recueration. The P other element has increased due to increased idling losses een though the cooling term decreases with the decrease in total losses. The total hydraulic loss for the OCS rototye is gien by Eq. 8. E loss cyclet.. tot OCS = ( P, loss + Pm, loss + Pthrottle + Pother )dt (8) t= The energy distribution in the OCS-rototye is resented in igure 9.

10 OCS Ed,loss ( E ) LS-S,fuel E OCS,fuel Diesel PTO ro. hyd. E,loss E h,loss E in,ocs E work E hydraulic system in,ls-s E recu E motor losses E um losses E throttle losses E other losses [N] x recu.9 9% E wheel E work igure 9 - Energy distribution in the wheel loader during the truck loading cycle, using the circuit solution. Switching to the differential control mode in the recueratie hase yields a different oint of oeration in ressure as well as dislacement. This means that the um efficiency at a gien seed and force coule deends on which control mode is alied. In igure the um efficiency during the recueratie hase of the loading cycle is resented, oerating in the two modes resectiely. The efficiency is found through interolation in measurement data deliered by the um manufacturer where ressure leel, relatie dislacement and shaft seed hae been considered in the calculation. As seen in the figure the um efficiency is -% higher in the differential control mode. m,tot Normal state Differential state Time [s] igure - Pum efficiency in the recueratie hase of the loading cycle deending on chosen mode of oeration. The force and seed characteristics of the lifting drie in one tyical loading cycle is illustrated in igure. The seed and force data is deried from ressure- and osition signals, acquired during oeration. The contour lot in the background shows the system efficiency. The efficiency calculation encomasses um losses as well as ale losses gien the best ossible control mode. The bar to the right shows the recueration ratio for the truck loading cycle, in this case 9% of the hydraulic energy inut for the lifting drie throughout the cycle. About the same ercentage alies for the tilting drie, resulting in an aerage of about % of the energy inut being recuerated throughout one cycle. igure - The energy recueration ratio in the truck loading cycle. Lowering an emty bucket in the differential mode is not a roblem, but regarding differential oeration of the tilting drie, a full bucket of grael unfortunately often results in too high a ressure leel. Howeer, choosing the cylinder area ratio differently a higher recueration ratio could also be achieed in this drie. In igure the ower, energy and recueration ratio throughout the loading cycle are illustrated, including both dries. Normalized Power [-] Power Energy Power and energy 8 8 Time [s] Normalized Energy [-] recu igure Power, energy and recueration ratio. The energy consumtion during a truck loading cycle is measured according to Volo standard fuel measurement rocedures. The two machines were drien by an exerienced drier. The tests were erformed during the same day, loading grael from the same sot. or eery six cycles the truck was fully loaded and weighed. Each machine was drien until four truckloads were filled. Each cycle was clocked in order to measure roductiity. The measurement results are resented in igure. The left stale air shows a 9% reduction in consumed fuel olume er loaded mass of grael, reresenting how the efficiency in ure Newtonian work is affected. The middle stale air shows how the system saes % in fuel olume oer time. The right stale air shows a % longer aerage cycle time. The error bars on to of the stales show the maximum and minimum measurement alue throughout the tests. The reduced size of the error bars in measurements from the OCS indicates that the oerability of the machine is good. Tilt Lift.9

11 Normalized indiidually [-],,8,,, 9% less Performance % less % longer fuel olume/loaded mass fuel olume/time cycle time igure - Performance measures from the truck loading cycle. The comarison elucidates some imortant energy related benefits and drawbacks of the two systems, which are summarized in Table. Table - Aarent energy related asects by introduction of the OCS and their imortance. Effected energy asects Imact on fuel by introducing the OCS consumtion in a hydraulic system truck loading cycle Elimination of -margin related to LS-control Elimination of metering losses related to unequal drie ressure leels, introducing searate ums for each drie Reduction in energy required to cool the oil being throttled to tank in lowering motions as well as cooling of engine losses (in case of saing) Introduction of increased losses related to the resence of extra ums Introduction of energy recueration in lowering motions LS-S OCS Medium imact the ums are mostly able to kee this margin excet at maximum flow Low imact because the dries are actuated simultaneously only to a ery small extent Low/medium imact reducing the total energy consumtion imlies less energy required for cooling of diesel losses Low/medium imact increased iscous friction losses according to next section High if the best control mode is chosen according to reious section EICIENCY TEST - TRANSPORTATION DRIVING In both wheel loaders the hydraulic ums are directly owered from the ower take-out unit which is mechanically couled to the diesel engine. Consequently the ums are in motion as long as the engine is on. In the reference machine there is only one um owering the working hydraulics, steering and auxiliary functions. In the OCS two ums are used for the working hydraulics and for the steering and auxiliary functions, two small ums had to be installed instead of one big um, due to sace limitations. The result is four ums instead of one. To measure how the increased number of ums affects the energy consumtion, a transortation driing test was arranged. The distance driing test was erformed on an oal test track at the Volo Construction Equiment testing ground. The engine seed was set to rm and the gearbox is locked in the fourth gear for both test objects. The test was carried out during minutes of driing. The measured difference in fuel consumtion is resented in igure. uel consumtion [-],,8,,, uel consumtion in distance.% more fuel LS-S OCS LS-S igure - uel consumtion in a transortation driing test comaring the OCS rototye machine to the reference machine. OPEN VERSUS CLOSED CIRCUIT SOLUTION In measurements from, the CCS demonstrated a fuel consumtion reduction of % in a truck loading scenario similar to the one described in this article. Why the energy reduction resented in this study is lower naturally deends on many factors. One robable reason is the concetual differences in the roulsion system. The CCS rototye uses a hydrostatic transmission whereas the OCS rototye uses a hydrodynamic transmission. In a system where the roulsion requires less energy, the imact of an imroement in the hydraulics is greater. A better way of comaring the erformance of the two solutions is to look at the caabilities of the indiidual systems side by side without a secific alication inoled. In Table the oerall caabilities and thereby similarities, strengths and weaknesses of the systems are listed. OCS

12 Table - Quality measures. Quality measures OCS CCS our-quadrant oeration Yes Yes Energy recueration Yes Yes Recueration in differential mode Yes No Pressure sensors needed Yes No Charge um needed No Yes Accumulator needed No Yes Based on closed loo control No Yes Increased tank ressure needed Yes No Regarding the differences in hardware requirement, the asects of comonent integration must be considered. Both concets require ressure relief functionality, anticaitation and load holding functionality. In the CCS almost all the comonents can be integrated to the um housing, including charge um, ressure relief ales, check ales and load-holding ales. or the OCS it is suggested that a distributed ale configuration be used, mounted as close as ossible to the cylinder dries. Howeer, in the ale configuration, ressure relief functionality anti caitation and load holding functionality are easily incororated (). Regarding four-quadrant actuation, the CCS intrinsically oerates oer all four quadrants; whereas the OCS oerates oer only two quadrants. To make the circuit solution controllable oer all four quadrants, the four ales must be controlled actiely. This of course makes control more comlex, but it also results in greater flexibility when it comes to the selection of modes targeting the highest ossible efficiency. In lifting oerations both concets are more or less equally energy efficient. The dominating loss comonent in this quadrant lies in the um efficiency. In lowering oerations the OCS is caable of oeration at higher efficiency oer a larger region of oeration, due to the ossibility of oerating in the differential control mode. The CCS cannot recuerate energy in the differential mode as a hydraulic couling of the cylinder chambers results in equal ressure on both sides of the hydraulic machine, whereas no torque is generated. or the wheel loader in this study the CCS would require much larger ums to achiee the secified lowering time of seconds. The flow leaing the lift cylinder is about * - m which in terms of dislacement corresonds to a cc um at the gien engine seed. Another almost as big um is required for the tilting motion. Haing such big ums when only one third of that size is sufficient to achiee the secified lifting times, does not render the solution cost effectie. UTURE CAPABILITIES The circuit solution is not restricted to use only with the diesel engine as rimary moer, but would actually benefit from the resence of an electro-hydraulic hybrid ower system. In such a system the diesel engine is relaced by electrical motors to ower each drie indeendently. Moreoer, a cost effectie solution would be to relace the ariable ums with smaller, high seed, fixed dislacement ums, indeendently seedcontrolled by the electrical motors. This would increase the recueration ratio due to greater maximum flow in the differential mode and consequently reduce the throttling losses. Moreoer, a rerequisite to sae energy in the OCS rototye is that the recuerated energy is consumed instantly. In an electro-hydraulic hybrid system with energy storage this would not be an issue. An alternatie to the meter-out flow control would be to use the excess flow to boost the hydraulic fan motor. In future deeloment of this tye of modeswitching systems one should always consider the differential mode when selecting the cylinder area ratio in order to fully benefit from using the otential load energy in recueratie quadrants on all cylinder dries. CONCLUSION The calculated results from the single function efficiency test showed that the non-differential mode is more energy efficient than the differential mode, but the differential mode soon becomes a better alternatie if meter-out flow control is needed in order to achiee the required lowering seed. The test showed how the circuit solution, within reasonable limits in lowering seed, saes about % fuel comared to the reference machine, equied with conentional load sensing hydraulics. In the truck loading exeriment fuel consumtion was reduced by about % comared to the reference machine. According to the theoretical calculations this result corresonds to a % reduction of losses in the OCS hydraulic system. In a final test the OCS-rototye was drien a longer distance without using the hydraulics. The result showed how the increased number of installed ums increases the fuel consumtion by about %. Comaring the circuit solution to the closed circuit solution, the OCS has a greater otential for energy recueration oer a broader working region while fewer sensors and more actie control are required for the closed circuit. Concerning oerability, where smoothness in control and system resonse is imortant for a wheel loader alication, many of these arameters are easily adjusted in software which makes the hydraulic system usable in many other alications as well. ACKNOWLEDGMENTS We wish to thank the staff at the deartment of Management and Engineering as well as staff at Volo Construction Equiment for heling out with much of the ractical work resented in this article.

13 REERENCES ) K. Heybroek, J-O. Palmberg, and J. Larsson. Oen circuit solution for um controlled actuators. Proceedings of the th PNI-PhD Symosium,. Sarasota,lorida, USA., 7-. ) S. Habibi and G. Singh. Deriation of design requirements of otimization of a high erformance hydrostatic actuation system, International Journal of luid Power, (),,. ) G. Wendel. Hydraulic system configurations for imroed efficiency, Proceedings of the 9th National Conference on luid Power, March, 7 7. ) R. Rahmfeld and M. Iantysynoa, Dislacement controlled linear actuator with differential cylinder - a way to sae rimary energy in mobile machines, th International Conference on luid Power Transmission and Control (ICP ),, 9. ) Krus, P. On load sensing fluid ower systems with secial reference to dynamic roerties and control asects, Ph.D. thesis, 988, Linköing, 7-8. ) R.Rahmfeld, M. Iantysynoa and J. Weber, Dislacement controlled wheel loader a simle and cleer solution, th IK (International luid Power Conference),, Dresden, Germany. 7) C. Williamson and M. Iantysynoa, The effect of um efficiency on dislacement controlled actuators, Proceedings of the Tenth Scandinaian International Conference on luid Power, 7, Tamere, inland, (), -. 8) Heybroek, K. and J-O. Palmberg, Alied control strategies for a um controlled circuit solution, th IK (International luid Power Conference), 8, Dresden. 9) illa, R. An eent drien oerator model for dynamic simulation of construction machinery, The Ninth Scandinaian International Conference on luid Power, June, Linköing, Sweden. ) K. Heybroek, J-O. Palmberg. Mode switching and energy recueration in circuit um control, Proceedings of The th Scandinaian International Conference on luid Power, SICP'7, CONTACT Kim Heybroek, M.Sc., Ph.D.-student at the diision of luid and Mechanical Engineering Systems at the Deartment of Management and Engineering (IEI), Linköing Uniersity, Sweden. kim.heybroek@liu.se, Tel: + () 8 LIST O NOTATIONS Index Exlanation Unit LS- Differential ressure determined by the load sensing controller Pa D Pum dislacement m E recu E in n * d * Total energy outut on the um shaft when um oerates as a motor, measured throughout a secified oeration Total energy inut to the um shaft when uming, measured throughout a secified oeration Load force acting on the cylinder rod Maximum load force in the nondifferential control mode, system ressure at maximum Maximum load force in the differential control mode, system ressure at maximum /m Relatie um/motor dislacement - recu The relation between recuerated energy and energy inut to the hydraulic ums throughout a secified oeration /m Pum/motor efficiency - Actuator elocity m/s n * d * Maximum actuator elocity in the non-differential control mode, um flow at maximum Maximum actuator elocity in differential control mode, um flow at maximum J J N N N - m/s m/s q Pum flow m /s Pum ressure Pa P x Power losses, x=defines the source of the ower loss n Pum shaft seed rad/s /m Pressure difference oer um/motor W Pa (L-T) Load ressure minus tank ressure Pa (P-L) Pum ressure minus load ressure Pa x Piston dislacement m