Saving Energy through Predictive Control of

0 downloads 0 Views 197KB Size Report
The goal of the project is to reduce the fuel consumption and driving time under the restriction of overall economic considerations in truck operation. ... From literature and application, many different measures to reduce fuel ... are known, see Table 1. ..... Hellström, E. Look-ahead Control of Heavy Trucks utilizing Road ...
Saving Energy through Predictive Control of Longitudinal Dynamics of Heavy Trucks

Energieeinsparung beim Betrieb von Nutzfahrzeugen durch prädiktive Regelungsstrategien in der Längsdynamikregelung Dipl.-Ing. P. Kock, MAN Nutzfahrzeuge AG and Kingston University, London, Dr. H.J. Welfers, MAN Nutzfahrzeuge AG, Dipl.-Ing. B. Passenberg, S. Gnatzig, Prof. Dr.-Ing. O. Stursberg, Technische Universität München, Prof. Dr. A.W. Ordys, Kingston University, London 1. Abstract This paper outlines a joint research project of the Kingston University (London), Technische Universität München, and the MAN Nutzfahrzeuge AG on economically optimal driving strategies. The project deals with the implementation of predictive optimal control algorithms operating the truck at economically favourable operation points by considering the costs of operation (fuel, driver, leasing, maintenance, income) and the dynamics of the vehicle (engine, brakes, fuel consumption, mass etc.). The approach considers GPS positioning and 3D maps for slope, curve and speed limit information of future road segments. It is planned to include radar information (position and speed of preceding vehicle), traffic information (traffic flow and density) and transportation job constraints like fixed arrival times and use of time buffers for optimisation. The goal of the project is to reduce the fuel consumption and driving time under the restriction of overall economic considerations in truck operation. In the following, the current state of the work, some results and outline the future road map is outlined.

2. Kurzfassung Dieser Artikel beschreibt ein gemeinschaftliches Forschungsprojekt der Kingston University (London) und der TU München in Zusammenarbeit mit der MAN Nutzfahrzeuge AG zum Thema ökonomisch optimaler Fahrstrategien. Das Projekt hat die Implementierung von prädiktiven und optimalen Regelungsalgorithmen zum Ziel, die einen LKW im optimalen Arbeitspunkt betreiben und dabei die Betriebskosten (Treibstoff, Fahrer, Leasing, Wartung,

Einnahmen) sowie die Fahrzeugdynamik (Motor, Bremsen, Kraftstoffverbrauch, Masse etc.) berücksichtigen. Der Ansatz berücksichtigt weiterhin die GPS Positionierung und die Verwendung von 3D Karten für Steigung, Kurven und Geschwindigkeitsbeschränkungen für zukünftige Straßensegmente. Es ist geplant den Ansatz um die Einbeziehung von Radarinformationen (Position und Geschwindigkeit des vorausfahrenden Fahrzeugs), Verkehrsinformationen (Geschwindigkeit und Dichte) und logistische Rahmenbedingungen wie späteste Ankunftszeit am Lieferort und Zeitreserven zu erweitern. Das Ziel des Projekts ist die Reduzierung des Kraftstoffverbrauchs und der Fahrzeit (und damit der Betriebskosten) unter den Rahmenbedingungen der ökonomischen Überlegungen beim Betrieb von Kraftfahrzeugen. Im Folgenden werden der Stand des Projekts, einige Ergebnisse und die zukünftig geplanten Arbeitspakete vorgestellt.

3. Introduction

From literature and application, many different measures to reduce fuel consumption or to improve the economic situation for vehicles from different fields of engineering and science are known, see Table 1. Table 1 Overview over different Fuel Saving Measures

Measure

Fuel

Saving Precondition

Potential

in

Trucks Aero Dynamic : Roof Spoiler Optimisation [1]

0.4 L/ 100 km

none

Aero Dynamic : Trailer Rear Optimisation [1]

1.5 L/ 100 km

none

Full Aero Dynamic Packages [2]

10%

none

Electrical Control : Compressor with Cut-Off Clutch [1]

0.5 L/ 100 km

none

Hybrid Drive for Medium Weight Distribution Trucks [1]

6-15%

Many stops during trip

Hybrid Drive for City Busses [1]

20-25%

Many stops during trip

Electrical Control of Auxiliary Loads [3]

0.3-1.6%

none

Longitudinal Predictive Speed Controller Algorithms [4- 1- 3.5%

Slope

or

8]

changes,

speed

vehicle weight

high

The method proposed here consists of a longitudinal predictive speed controller governing the velocity of a vehicle in an intelligent way and can be viewed as an extension to ordinary cruise controllers. The concept optimises online the dynamic speed operation point of a truck and is based on the assumption of changing slopes along the travelled road and speed levels. Among the above methods the longitudinal predictive speed controller has a fuel saving potential of 1% to 3.5% depending on the slope characteristic of the road. This number reflects only the effect of slope without traffic influence known from literature. The technical preconditions of this approach like GPS, 3D maps and radar are available now and online traffic information will be available soon. The costs of these components are low because GPS is a basic equipment nowadays, maps are part of navigation, though additional slope information has to be provided, ACC radar will be a mandatory equipment with emergency brake assistance according to EU plans in 2012 [9]. Online traffic information is already provided to TomTom customers for route planning. Basic obstacles are algorithm complexity, computing power and missing experience with driving behaviour and customer acceptance.

The next table gives an idea of the benefits of 2% fuel reduction for trucks with different fuel consumptions and mileage driven with active cruise control per year: 1

Table 2 Savings for 2% Fuel Reduction at 1.40€/L and active Cruise Control Mileage

50.000km/year

60.000 km/year

70.000 km/year

32 L/100km

448 €/year

536 €/year

626 €/year

34 L/100km

476 €/year

570 €/year

660 €/year

36 L/100 km

504 €/year

604 €/year

704 €/year

With a preferred ROI of less than 3 years, an investment between 1344€ and 2112€ for 2% fuel reduction is reasonable and over a 6 year life time savings between 2688€ and 4224€ with the current diesel price are feasible. In the following, it will be concentrated on the measure of longitudinal predictive cruise control of heavy trucks on long distance highway operation. This measure is reasonable for applications, where the truck is operated on long distances with high speeds and a load from 25 to 40 tons. 1

Savings on 2% reduction  milage per year travelled with cruise control 





! "#$%"&$ '($) *+,-(./01+, 

2

344



! '($) /%1*$ 

2

! 2/10000

New Aspects of this Paper The present approach differs from other approaches known from literature [4-6, 10-13] in the following points: •

Consequent optimisation to economic consideration. The cost function only includes costs for fuel, driver, leasing, maintenance and income. The speed trajectory of the vehicle is a product of pure economic cost factors and hard constraints.



Analytical optimality conditions (Karush-Kuhn-Tucker conditions, Pontryagin Minimum Principle, constraints on control and state variables) and a numerical solution with a multiple shooting algorithm are used for solving the optimisation problem.



Alternative heuristic approach that is based on a simple truck model generates a speed trajectory based on the model, rules and future slope profiles. The heuristic rule and model based controller offers the same performance but is real time capable, easy to implement and less sensitive to errors.

4. Basic Ideas of Predictive Optimal Control Many approaches try to imitate the driving behaviour of an experienced driver with a foresighted driving style. It is assumed that this perfect driver would reduce fuel consumption by avoiding the use of brakes and very moderate use of the accelerator pedal by considering present and future events like actions of preceding vehicles, influence of slope and curves or general traffic conditions. By a combination of today’s technology it is possible to use the class of predictive optimal control algorithms to control a truck in an economically optimal manner by using GPS, 3D maps, ACC radar and online traffic information. Optimal Speed Levels It is known from literature [10, 11] that on flat roads a constant speed is optimal if the engine fuel consumption map is convex. If the engine map has significant local minima, it might be optimal to drive at two changing speed levels if the road is flat [14]. Flat means, that a slope value is below a certain threshold which can be calculated [10, 15]. Two limiting factors of the methods above are that they will not be feasible without a driver assistance system and they are not able to calculate the constant, optimal speed levels. If the road has slopes larger than the slope threshold it is optimal to adjust the speed level of the truck according to the slope shape and magnitude. Many papers describe solutions to this problem [4-8, 10-13, 16, 17].

Optimal Control A crucial point in optimisation is to decide what should be optimised and to determine the constraints for optimisation. In optimal control, a cost functional is introduced to describe the optimisation target: C

9:; :0

Suggest Documents