# Slope Optimization (or “Sloop”): Customized Optimization for Road Longitudinal Profile Eco-Design

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

- the optimization problem is fully described;
- the choice of the algorithm is justified;
- the algorithm is described and specially tailored for the problem.

- the evaluation model with respect to the parameters of the longitudinal profiles;
- the optimization with respect to the main construction parameters and traffic assumptions.

## 2. Degrees of Freedom: Parameters of the Longitudinal Profile Model

#### 2.1. Road Longitudinal Profile

#### 2.2. Constraints

#### 2.2.1. Linear Constraints

#### 2.2.2. Non-Linear Constrains

- they are nonlinear;
- they are equality constraints, hence they must be taken into account at each iteration of the optimization process.

## 3. Objective Function

- the assessment of the construction phase;
- the assessment of the operation phase.

#### 3.1. Construction

#### 3.2. Volume Computation

#### 3.3. Assessment of Ghg Emissions from the Volume Computation

#### 3.4. Operations

^{©}library, called VEHLIB [19]. Fuel consumption is derived working backwards from wheel forces to the engine through component sub-models. The chassis model describes the aerodynamic and rolling resistance forces used to solve the fundamental dynamic principle. Wheel forces are then known and relayed to the engine shaft using a power model composed of ratios and efficiencies of the final gear and gearbox. If needed, a clutch model may also be involved (at low speed). A specific fuel consumption map is then used to assess engine fuel consumption, which then gets transposed into GWP (CO

_{2}emissions).

## 4. Optimization

#### 4.1. Choice of Algorithm

**Optimization type**The problem is a finite dimensional nonlinear optimization with various linear and nonlinear inequalities and one nonlinear equality.**Global/Local optimization**The optimized profile cannot be too dissimilar from the initial profile for economic and mechanical reasons: earth movements are costly, embankments too large are not stable. The optimization is rather local.**Initial solution**An initial designed profile is available; it is considered feasible to the extent that it satisfies the constraints.**Smoothness**We analyze this mathematical property for the objective function and for the constraints.- The objective function is relatively smooth as the sum of two smooth functions, i.e.:
- The construction assessment is based on a volume computation. This function has continuous derivatives of all orders (${C}^{\infty}$).
- The operational assessment is the sum of the fuel consumption of various vehicles rolling virtually on the road. This sum numerically regularizes the integrand. However, the integrand is not as smooth as the construction cost. For example, a slight change in slope can lead to a drop in engine efficiency that locally alters the consumption or can lead to mechanical braking, which changes overall fuel consumption on the road. This function is at least ${C}^{1}$.

- The nonlinear constraints stem from geometric considerations; they are ${C}^{\infty}$.

**Computational burden**The construction assessment and nonlinear constraints do not require a large amount of computational operations since they rely on simple geometric analysis. On the other hand, the operational assessment requires the simulation of vehicle dynamics, which is a computation time-consuming task.**Nearly quadratic equality**According to Section 2.2.2, the equality constraint on altitude is nearly quadratic. Given that such a constraint must be verified at each iteration, an efficient algorithm needs to take advantage of this property.

- Generic SQP algorithms do not take the quadratic equality constraints into account. We have developed a dedicated solution to take advantage of this feature.
- SQP imposes that the cost function and constraints can be continuously differentiated twice (${C}^{2}$), whereas the operational assessment is at least ${C}^{1}$. At each iteration, we verify that the quadratic approximation of the Lagrangian is validated by comparing this approximation with the actual value of the objective function.
- SQP requires a computation of the Hessian of the Lagrangian and then of the Hessian of the objective function. The numerical computation of this Hessian is not straightforward because:
- -
- the number of inputs may be high;
- -
- the computation of the operational assessment is time-consuming.

#### 4.2. Description of the Algorithm

#### Sqp Algorithm

- by the descent direction ${p}_{k}$ solving the Karush Kuhn Tucker (KKT) conditions [24] of the quadratic optimization with linear equality constraints. The quadratic program is as follows:$$\begin{array}{c}\hfill \underset{x}{\mathrm{minimize}}\phantom{\rule{0.277778em}{0ex}}\frac{1}{2}{p}_{k}^{t}{\nabla}_{xx}\left(L({x}_{k},{\lambda}_{k})\right){p}_{k}+\nabla \mathrm{J}\left({x}_{k}\right){p}_{k}\\ \hfill \mathrm{subject}\phantom{\rule{4.pt}{0ex}}\mathrm{to}\phantom{\rule{4.pt}{0ex}}\nabla h\left(x\right){p}_{k}=-h\left(x\right)\end{array}$$$$\left(\begin{array}{cc}{\nabla}_{xx}\left(L({x}_{k},{\lambda}_{k})\right)& -\nabla h\left({x}_{k}\right)\\ -\nabla h\left({x}_{k}\right)& 0\end{array}\right)\left(\begin{array}{c}{p}_{k}\\ {\lambda}_{k+1}\end{array}\right)=\left(\begin{array}{c}-\nabla \mathrm{J}\left({x}_{k}\right)\\ h\left(x\right)\end{array}\right)$$
- the linear approximation of the non linear equality constraint on the altitude;
- the linear approximation of the active inequalities, i.e., the inequalities, numbered l, which are saturated and hence become equalities, ${g}_{l}\left({x}_{k}\right)=0$.

This step requires to compute the gradient and hessian of the objective function and constitutes the main computational cost of the algorithm. The gradient is computed numerically by centered finite difference; this requires $(n+1)$ simulations of traffic. We have applied the BFGS approximation to compute the hessian according to the equations given in [23] so as to avoid to carrying out $n\times (n+1)/2$ traffic simulations in order to compute the Hessian by finite difference. - by reducing the length of the increment ${d}_{k}=\mu {p}_{k}$ with $0<\mu \le 1$ being accommodated:
- activation of one non active linear inequality;
- activation of one non active linearized nonlinear inequality;
- an increment can respect linearized nonlinear inequalities while not respecting the nonlinear inequalities themselves. In this case, the increment’s length is reduced until the inequalities are validated, with this reduction factor being due to the linear approximation of non linear constrains;
- along the same line, a reduced increment can output a solution that fails to respect the linear approximation of the non linear equality on altitude. In this case, a reduced factor is computed, as will be described in the next subsection;
- if the reduced increment yields a solution which increases the objective, the increment is again reduced until the objective decreases with this reduction factor being due to the quadratic approximation of the cost;

#### 4.3. Adaptation of Sqp to the Nonlinear Equality on Altitude

## 5. Results

#### 5.1. Case Study

#### Construction Phase Parameters

#### 5.2. Simulation Parameters of the Operations Phase

- number of vehicles/day in each direction over the studied part of the road;
- mass of each type of vehicle (including load and passengers);
- product of the drag coefficient and the front surface area, denoted $S.{C}_{x}$;
- tire rolling resistance coefficients, (${C}_{rr0}$) represents the constant part (${C}_{rr1}$) the part proportional to vehicle speed.

#### 5.3. Optimized Profiles

- 6 abscissas defining the intersection points between two consecutive straight lines; the displacements of abscissas are limited to +/− 100 m in order to obtain a profile not to far from the initial.
- 6 slopes, one per section; Slopes can vary from −5.3% to 5.3%. These bounds are set in order to ensure stability of the embankment.
- 7 radii for the circle linking the sections, since the two radii of the two extremities are also degrees of freedom; Radii vary from +/− 2000 m around the initial radii in order to keep a profile relatively smooth.
- The tolerance on the altitude constrain is +/− 0.1 m. This step with a maximum height of 10 cm is easily flattened during construction.

- ${\mu}_{lin}$: activation of one inactive linear inequality;
- ${\mu}_{nolin}$: activation of one inactive linearized non linear inequality;
- ${\mu}_{nol}$: linear approximation of the non linear equality constraints;
- ${\mu}_{alt}$: compatibility of the quadratic approximation of the altitude constraint (${\mu}_{alt}$ is the solution of Equation (24));
- ${\mu}_{altnol}$: quadratic approximation of the altitude constraint;
- ${\mu}_{obj}$: quadratic approximation of the objective function.

#### 5.4. Sensitivity Analysis

#### 5.4.1. With Optimization Variables

_{2}for the traffic and construction, respectively). They correspond to significant variations for the slopes (variation from 0.18 to 0.54%). The variation of the other parameters are not significant (maximum variation of 2 m for the intersection points and 5 m for the radii). If this result is confirmed on various case studies, the optimization could be simplified by considering only slopes.

#### 5.4.2. On Construction Parameters

_{2}) to the cost of initial profile when the parameters are fixed to their initial values (101,210 Mteq CO

_{2}), we observe that only the change of ${\mathrm{gwp}}_{\mathrm{lime}}$ has a significant impact on the cost function with an increase of 0.4%. This is consistent with the practitioners of Life Cycle Analysis who underline the influence of Lime on GWP [5]. A comparison of the cost of the ${\mathrm{gwp}}_{\mathrm{lime}}$ profile optimized (97,848 Mteq CO

_{2}) with the profile optimized computed with the initial value (97,505 Mteq CO

_{2}) shows a small difference (0.35%) consistent with our expectations: as the cost of construction increased relatively to the operation cost, the optimization process focuses more its effort on the construction cost (3717 Mteq CO

_{2}when the parameter changes and 3917 Mteq CO

_{2}in the initial scenario). The costs of the profile optimized when the other parameters change do not differ significantly with the profile optimized with the initial values (difference < 0.06%).

#### 5.5. Evolution of Vehicle Fleet Scenario

_{2}per kW·h, mean value for the French electrical mix highly composed of nuclear electricity [25]. It is clear that the results are influenced by this value for others countries.

## 6. Discussion

- Since vehicle consumption accounts for roughly 90% of GHG emissions, we anticipated finding an optimized profile with milder slopes than the initial profile because, in this case, consumption would be lower. However, the optimized profile proved to have steeper slopes. This finding can be explained by the fact that the ramps remain sufficiently mildly inclined so as to not significantly modify engine efficiency and not trigger braking. If consumption remains rather stable, the algorithm will minimize construction-related emissions, which are largely due to the lime treatment of embankments as underlined by the sensitivity analysis. For this reason, the optimized profile minimizes the embankment while holding vehicle consumption stable. Yet an optimization that only takes construction into account would output a profile with a steep ramp, as is the case for Iteration 5 in Table 4 in order to avoid the embankment. In this case, the traffic-related emissions increase significantly.
- As expected, according to Table 4, the nearly quadratic constraint on the altitude often limits the descent during several iterations: ${\mu}_{alt}$ lies below 1 for two thirds of the iterations and in half of the iterations, ${\mu}_{alt}$ is the appropriate reduction factor of the increment to validate the altitude constraint (${\mu}_{alt}<1$ and ${\mu}_{altnol}=1$), a finding that underscores the benefit of taking this property into account by the algorithm under the real-world scenario.
- This algorithm explores various feasible profiles with distinct features, thus providing stakeholders with different possibilities. They can refine one of the proposed profiles if deemed relevant. For example, given that the optimized solution includes two successive slopes very close to one another, stakeholders can propose a simpler profile to replace these two successive ramps by just one ramp. In its practical implementation, our algorithm takes this special case into account by automatically equalizing successive slopes when they are nearly equal. This type of ad hoc procedure is not robust (i.e., a discontinuity is artificially created) and difficult to tune (e.g., how to numerically define what constitutes “nearly equal” successive slopes?).
- This last point suggests a limitation of the algorithm, which depends on the number of slopes defined in the initial profile by road designers. An alternative would be to directly optimize the profile in an infinite dimensional space through the use of variational methods. The optimized solution would then be approximated by lines and circles in order to transform this optimized, but not necessarily feasible, solution into a longitudinal road profile that can indeed be built. Research along these lines is currently underway; this approach does not guarantee a feasible solution.
- The degrees of freedom are different in nature and are expressed in different units: angles, abscissa, radii. The implementation of the optimization algorithm does not directly use them: a scaling is first performed.
- The optimization with just a single criterion oversimplifies the problem. One perspective would call for a multi-criteria optimization in order to propose to stakeholders a set of efficient solutions (a Pareto front). It should be noted that stakeholders already have a feasible profile at each iteration in using our algorithm.
- Our vehicle consumption models are more sensitive to slope than the usual models used in environmental assessment. However, they do not take into account the specific deformation of the tire on uphill slopes as described in the article [26]. This phenomenon depends on the tire pressure which is not taken into account in our model. This is a point of improvement in our modeling.
- Our case study does not include civil engineering structures such as bridges or tunnels. Our methodology is based on Project Energy Assessment Method (PEAM) and PEAM also models civil engineering structures as mentioned in [3], so there are no theoretical limits to include civil engineering structures in our methodology. Implementation can be complicated because the construction cost function may not be continuous. We plan to implement them according to users’ requests.

## 7. Conclusions

## Author Contributions

## Funding

## Acknowledgments

## Conflicts of Interest

## References

- IEA. World Energy Outlook 2020—Analysis. 2020. Available online: https://www.iea.org/reports/world-energy-outlook-2020 (accessed on 1 December 2020).
- FIT. ITF Transport Outlook 2019; OECD Publishing: Paris, France, 2019. [Google Scholar] [CrossRef]
- Bosquet, R.; Jullien, A.; Vandanjon, P.O.; Dauvergne, M.; Sanchez, F. Eco-design model of a railway: A method for comparing the energy consumption of two project variants. Transp. Res. Part D Transp. Environ.
**2014**, 33, 111–124. [Google Scholar] [CrossRef] - Coiret, A.; Deljanin, E.; Vandanjon, P.O. Vehicle energy savings by optimizing road speed-sectioning. Eur. Transp. Res. Rev.
**2020**, 41, 1–15. [Google Scholar] [CrossRef] - Vandanjon, P.O.; Vinot, E.; Cerezo, V.; Coiret, A.; Dauvergne, M.; Bouteldja, M. Longitudinal profile optimization for roads within an eco-design framework. Transp. Res. Part D Transp. Environ.
**2019**, 67, 642–658. [Google Scholar] [CrossRef][Green Version] - Webb, C. The Effect of Gradient on Fuel Consumption and Speed of a Road Vehicle. Proc. Inst. Mech. Eng. Part E
**1952**, 3, 104–117. [Google Scholar] [CrossRef] - Wyatt, D.W.; Li, H.; Tate, J.E. The impact of road grade on carbon dioxide (CO
_{2}) emission of a passenger vehicle in real-world driving. Transp. Res. Part D Transp. Environ.**2014**, 32, 160–170. [Google Scholar] [CrossRef][Green Version] - Ragione, L.D.; Giovanni, M. The Influence of Road Gradient in an Integrated Approach of Real Driving Cycles and Emissions Factors Model. Transp. Res. Procedia TRA-2016
**2016**, 14, 3179–3188. [Google Scholar] [CrossRef][Green Version] - Costagliola, M.A.; Costabile, M.; Prati, M.V. Impact of road grade on real driving emissions from two Euro 5 diesel vehicles. Appl. Energy
**2018**, 231, 586–593. [Google Scholar] [CrossRef] - Monge, G. Mémoire sur la théorie des déblais et des remblais. In Histoire de l’Académie Royale Des Sciences; Imprimerie Royale: Paris, France, 1781; pp. 666–705. [Google Scholar]
- Inyim, P.; Pereyra, J.; Bienvenu, M.; Mostafavi, A. Environmental assessment of pavement infrastructure: A systematic review. J. Environ. Manag.
**2016**, 176, 128–138. [Google Scholar] [CrossRef] [PubMed] - Santos, J.; Ferreira, A.; Flintsch, G.; Cerezo, V. A multi-objective optimisation approach for sustainable pavement management. Struct. Infrastruct. Eng.
**2018**, 14. [Google Scholar] [CrossRef] - Jha, M.K.; Schonfeld, P.M.; Jong, J.C.; Kim, E. Intelligent Road Design; WIT Press: Southampton, UK, 2006; ISBN 978-1-84564-003-3. [Google Scholar]
- Kang, M.W.; Jha, M.K.; Schonfeld, P. Applicability of highway alignment optimization models. Transp. Res. Part C Emerg. Technol.
**2011**, 21, 257–286. [Google Scholar] [CrossRef] - Ferreira, H.; Rodrigues, C.M.; Pinho, C. Impact of Road Geometry on Vehicle Energy Consumption and CO
_{2}Emissions: An Energy-Efficiency Rating Methodology. Energies**2020**, 13, 119. [Google Scholar] [CrossRef][Green Version] - AASHTO. A Policy on Geometric Design of Highway and Streets; Technical Report; American Association of State Highway and Transportation Officials: Washington, DC, USA, 1994. [Google Scholar]
- Capony, A.; Muresan, B.; Dauvergne, M.; Auriol, J.C.; Ferber, V.; Jullien, A. Monitoring and environmental modeling of earthwork impacts: A road construction case study. Resour. Conserv. Recycl.
**2013**, 74, 124–133. [Google Scholar] [CrossRef][Green Version] - Jullien, A.; Dauvergne, M.; Proust, C. Road LCA: The dedicated ECORCE tool and database. Int. J. Life Cycle Assess.
**2015**, 20, 655–670. [Google Scholar] [CrossRef] - Vinot, E.; Scordia, J.; Trigui, R.; Jeanneret, B.; Badin, F. Model simulation, validation and case study of the 2004 THS of Toyota Prius. Int. J. Veh. Syst. Model. Test.
**2008**, 3, 139–167. [Google Scholar] [CrossRef] - Wilson, R. A Simplicial Method for Convex Programming; Harvard University: Cambridge, MA, USA, 1963. [Google Scholar]
- Nocedal, J.; Wright, S. Numerical Optimization, 2nd ed.; Springer Series in Operations Research and Financial Engineering; Springer: New York, NY, USA, 2006. [Google Scholar]
- Gill, P.E.; Wong, E. Sequential quadratic programming methods. In Mixed Integer Nonlinear Programming; Springer: New York, NY, USA, 2012; pp. 147–224. [Google Scholar]
- Hansen, E.L.; Volcker, C. Numerical Algorithms for Sequential Quadratic Optimization. Master’s Thesis, Informatics and Mathematical Modelling, Technical University of Denmark, Lyngby, Denmark, 2007. [Google Scholar]
- Boltyanski, V.; Martini, H.; Soltan, V. Geometric Methods and Optimization Problems; Springer Science & Business Media: New York, NY, USA, 2013; Volume 4. [Google Scholar]
- ADEME. Agency for Ecological Transition. 2020. Available online: https://www.bilans-ges.ademe.fr/ (accessed on 2 December 2020).
- Sina, N.; Nasiri, S.; Karkhaneh, V. Effects of resistive loads and tire inflation pressure on tire power losses and CO
_{2}emissions in real-world conditions. Appl. Energy**2015**, 157, 974–983. [Google Scholar] [CrossRef]

**Figure 1.**Longitudinal profile (in red) modeled as a sequence of straight lines connected by circular arcs (adapted from [5]).

**Figure 3.**Elementary earthmoving Volume, X is the longitudinal direction, Y is the transverse direction by [5].

**Table 1.**Values of the parameters used to assess the construction phase adapted from [5].

Tasks | GWP Label | Value (kgeq $\begin{array}{c}\hfill {\mathbf{CO}}_{\mathbf{2}}\end{array}/{\mathbf{m}}^{3}$) |
---|---|---|

Extraction | ${\mathrm{gwp}}_{\mathrm{extrac}}$ | 0.753 |

Per km of transport | ${\mathrm{gwp}}_{\mathrm{trans}}$ | 1.235 |

Spread | ${\mathrm{gwp}}_{\mathrm{spread}}$ | 0.913 |

Quicklime | ${\mathrm{gwp}}_{\mathrm{lime}}$ | 2196 |

**Table 2.**Measured traffic for the different vehicle categories (t: metric ton) adapted from [5].

Vehicle Class | Traffic (Per Day/Per Direction) | Distribution | Actual Average Speed km/h (m/h) |
---|---|---|---|

Car | 2419 | 70% | 113 (70) |

Van (<3.5 t) | 130 | 4% | 103 (64) |

Mini-truck | 102 | 3% | 95 (69) |

Truck and articulated vehicle | 802 | 23% | 88 (61) |

Total | 3453 | 100% |

**Table 3.**Vehicle data (t: metric ton) adapted from [5].

Class | Vehicles Per Day | Mass (t) | $\mathit{S}.{\mathit{C}}_{\mathit{x}}$ (m ^{2}) | ${\mathit{C}}_{\mathbf{rr}0}$ (10 ^{−3}) | ${\mathit{C}}_{\mathbf{rr}1}$ (10 ^{−6}m^{−1}s) |
---|---|---|---|---|---|

Articulated vehicle (40 t) | 198 | 40 | 5.2 | 4.824 | 39.5 |

Articulated vehicle (25 t) | 198 | 25 | 5.2 | 4.824 | 39.5 |

Truck (19 t) | 406 | 19 | 3.27 | 4.824 | 39.5 |

Gasoline-powered Car | 799 | 1.48 | 0.73 | 11.84 | 5.25 |

Small class diesel Car | 540 | 1.12 | 0.68 | 11.84 | 5.25 |

Medium class diesel Car | 540 | 1.23 | 0.97 | 11.84 | 5.25 |

Large class diesel Car | 540 | 1.51 | 0.82 | 11.84 | 5.25 |

Mini-truck | 102 | 3.69 | 3.18 | 11.84 | 5.25 |

Van | 130 | 2.53 | 1.98 | 11.84 | 5.25 |

Iter | $\Delta $ CO_{2} veh In Mt | $\Delta $ CO_{2} cons In Mt | $\Delta $ cost In Mt | ${\mathit{c}}_{\mathit{alt}}$ In m | ${\mathit{\mu}}_{\mathit{ineq}}$ | ${\mathit{\mu}}_{\mathbf{alt}}$ | ${\mathit{\mu}}_{\mathit{altnol}}$ | ${\mathit{\mu}}_{\mathit{obj}}$ |
---|---|---|---|---|---|---|---|---|

1 | 93,338 | 7872 | 101,210 | 0 | 0 | 0 | 0 | 0 |

2 | 23.4 | −605 | −581 | 0 | 1.1 × 10${}^{-13}$ | 1 | 1 | 1 |

3 | −44.1 | 38.8 | −5.34 | 0 | 1.6 × 10${}^{-3}$ | 0.53 | 1 | 0.06 |

4 | 2.06 | −85.3 | −83.2 | 0 | 1.7 × 10${}^{-5}$ | 1 | 1 | 1 |

5 | 851 | −2838 | −1987 | 0 | 7.2 × 10${}^{-6}$ | 1 | 1 | 0.25 |

6 | 96.3 | −321 | −225 | −0.01 | 1.7 × 10${}^{-5}$ | 1 | 1 | 1 |

7 | −579 | 528 | −50.9 | 0.01 | 2.5 × 10${}^{-4}$ | 1 | 1 | 0.50 |

8 | −209 | 37.2 | −172 | 0.01 | 1.1 × 10${}^{-4}$ | 1 | 1 | 0.12 |

9 | −119 | −23.1 | −142 | 0.08 | 2.5 × 10${}^{-3}$ | 0.18 | 0.90 | 1 |

10 | 189 | −445 | −256 | 0.10 | 3.2 × 10${}^{-4}$ | 0.40 | 1 | 1 |

11 | −1.23 | −6.74 | −7.97 | 0.10 | 0.01 | 0.01 | 1 | 1 |

12 | 3.02 | −7.67 | −4.65 | 0.10 | 1.2 × 10${}^{-3}$ | 0.04 | 0.73 | 1 |

13 | −0.41 | −0.04 | −0.45 | 0.10 | 0.03 | 2.2 × 10${}^{-3}$ | 0.90 | 1 |

14 | −0.53 | −0.34 | −0.87 | 0.10 | 0.01 | 3.7 × 10${}^{-3}$ | 1 | 1 |

15 | −0.28 | −0.51 | −0.79 | 0.10 | 0.01 | 2.8 × 10${}^{-3}$ | 1 | 1 |

16 | −0.25 | −0.53 | −0.77 | 0.10 | 0.01 | 2.9 × 10${}^{-3}$ | 1 | 1 |

17 | −32.5 | −4.28 | −36.8 | 0.10 | 0.32 | 0.41 | 0.90 | 1 |

18 | −11.4 | −7.04 | −18.4 | 0.10 | 0.17 | 0.12 | 1 | 1 |

19 | 46.6 | −81.6 | −35.1 | 0.10 | 0.08 | 0.38 | 0.21 | 1 |

20 | −8.67 | −13.2 | −21.8 | 0.10 | 0.47 | 0.26 | 0.90 | 1 |

21 | 2.31 | −26.3 | −24.0 | 0.10 | 0.29 | 0.18 | 1 | 1 |

22 | 2.32 | −16.6 | −14.2 | 0.10 | 0.18 | 0.19 | 1 | 1 |

23 | 4.05 | −18.3 | −14.2 | 0.10 | 0.16 | 0.25 | 1 | 1 |

24 | 9.36 | −26.9 | −17.6 | 0.10 | 0.18 | 0.50 | 1 | 1 |

25 | 11.8 | −16.3 | −4.50 | 0.08 | 0.27 | 1 | 1 | 1 |

total | 234 | −3939 | −3704 | 0.08 | 0.27 | 1 | 1 | 1 |

Iter | $\Delta $ CO_{2} veh In Mt | $\Delta $ CO_{2} cons In Mt | $\Delta $ cost In Mt | $\Delta {\mathit{x}}_{1}$ In m | $\Delta {\mathit{x}}_{2}$ In m | $\Delta {\mathit{x}}_{3}$ In m | $\Delta {\mathit{x}}_{4}$ In m | $\Delta {\mathit{x}}_{5}$ In m | $\Delta {\mathit{x}}_{6}$ In m |
---|---|---|---|---|---|---|---|---|---|

1 | 93,338 | 7872 | 101,210 | 0 | 898 | 799 | 629 | 972 | 587 |

2 | 23.4 | −605 | −581 | −0.01 | −0.02 | −0.03 | −0.02 | −0.02 | −0.02 |

5 | 851 | −2838 | −1987 | −0.09 | −0.36 | 0.08 | 1.95 | −0.43 | 0.21 |

7 | −579 | 528 | −50.9 | −8.35 | −13.4 | −6.37 | 5.75 | −5.88 | −6.12 |

Iter | $\Delta $CO_{2}vehIn Mt | $\Delta $CO_{2} consIn Mt | $\Delta $costIn Mt | $\Delta {\mathbf{\alpha}}_{\mathbf{1}}$In % | $\Delta {\mathbf{\alpha}}_{\mathbf{2}}$In % | $\Delta {\mathbf{\alpha}}_{\mathbf{3}}$In % | $\Delta {\mathbf{\alpha}}_{\mathbf{4}}$In % | $\Delta {\mathbf{\alpha}}_{\mathbf{5}}$In % | $\Delta {\mathbf{\alpha}}_{\mathbf{6}}$In % |

1 | 93,338 | 7872 | 101,210 | −1.10 | 0.14 | −1.00 | 1.23 | −0.88 | 0.92 |

2 | 23.4 | −605 | −581 | −0.04 | −0.02 | − 0 | 0.01 | 0.01 | 0.04 |

5 | 851 | −2838 | −1987 | −0.41 | 0.21 | 0.18 | −0.54 | 0.44 | 0.45 |

7 | −579 | 528 | −50.9 | 0.14 | −0.80 | 1.04 | −0.40 | 0.65 | −0.29 |

Iter | $\Delta $CO_{2} vehIn Mt | $\Delta $CO_{2} consIn Mt | $\Delta $costIn Mt | $\Delta {\mathbf{R}}_{\mathbf{1}}$In m | $\Delta {\mathbf{R}}_{\mathbf{2}}$In m | $\Delta {\mathbf{R}}_{\mathbf{3}}$In m | $\Delta {\mathbf{R}}_{\mathbf{4}}$In m | $\Delta {\mathbf{R}}_{\mathbf{5}}$In m | $\Delta {\mathbf{R}}_{\mathbf{6}}$In m |

1 | 93,338 | 7872 | 101,210 | 15,127 | 20,000 | 20,000 | 30,000 | 30,000 | 29,544 |

2 | 23.4 | −605 | −581 | 0 | 0 | 0 | 0 | 0 | −0.02 |

5 | 851 | −2838 | −1 987 | 0.05 | −0.02 | 0.24 | −0.21 | −0.05 | −5.06 |

7 | −579 | 528 | −50.9 | 1.02 | 1.91 | 2.09 | −1.78 | −3.29 | −13.4 |

Iter | $\Delta {\mathbf{R}}_{\mathbf{7}}$In m | ||||||||

1 | 7789 | ||||||||

2 | 0 | ||||||||

5 | 0 | ||||||||

7 | 0 |

Parameter Increased by 5% | Profile | Traffic (Mteq CO _{2}) | Construction (Mteq CO _{2} | Cost (Mteq CO _{2} |
---|---|---|---|---|

initial values | init | 93,338 | 7872 | 101,210 |

initial values | opt | 93,572 | 3933 | 97,505 |

${\mathrm{gwp}}_{\mathrm{extrac}}$ | init | 93,338 | 7885 | 101,223 |

${\mathrm{gwp}}_{\mathrm{extrac}}$ | opt | 93,552 | 3985 | 97,537 |

${\mathrm{gwp}}_{\mathrm{trans}}$ | init | 93,338 | 7894 | 101,232 |

${\mathrm{gwp}}_{\mathrm{trans}}$ | opt | 93,678 | 3842 | 97,520 |

${\mathrm{gwp}}_{\mathrm{spread}}$ | init | 93,338 | 7888 | 101,226 |

${\mathrm{gwp}}_{\mathrm{spread}}$ | opt | 93,585 | 3972 | 97,557 |

${\mathrm{gwp}}_{\mathrm{lime}}$ | init | 93,338 | 8214 | 101,552 |

${\mathrm{gwp}}_{\mathrm{lime}}$ | opt | 94,130 | 3717 | 97,848 |

Class | Mass (t) | $\mathit{S}\xb7{\mathit{C}}_{\mathit{x}}$ (m ^{2}) | ${\mathit{C}}_{\mathbf{rr}0}$ 10 ^{−3} | ${\mathit{C}}_{\mathbf{rr}1}$ 10 ^{−6}m^{−1}s |
---|---|---|---|---|

Small class electric Car | 1.45 | 0.68 | 11.84 | 5.25 |

Medium class electric Car | 1.68 | 0.97 | 11.84 | 5.25 |

Large class electric Car | 1.98 | 0.82 | 11.84 | 5.25 |

Electric Mini-truck | 4.51 | 3.18 | 11.84 | 5.25 |

Electric Van | 3.35 | 1.98 | 11.84 | 5.25 |

Class | First Year | Vehicles Par Day | |
---|---|---|---|

Scenario 1 In 25 Years | Scenario 2 In 25 Years | ||

Gasoline-powered Car | 799 | 1210 | 1003 |

Small class diesel Car | 540 | 0 | 270 |

Medium class diesel Car | 540 | 0 | 270 |

Large class diesel Car | 540 | 0 | 270 |

Small class electric Car | 0 | 403 | 202 |

Medium class electric Car | 0 | 403 | 202 |

Large class electric Car | 0 | 403 | 202 |

Mini-truck | 102 | 51 | 77 |

Van | 130 | 65 | 98 |

Electric Mini-truck | 0 | 51 | 25 |

Electric Van | 0 | 65 | 32 |

Scenario | Profil | CO_{2} Veh In Mt | CO_{2} Cons In Mt | Cost In Mt |
---|---|---|---|---|

0 | init | 93,338 | 7872 | 101,210 |

0 | opt | 93,572 | 3933 | 97,505 |

1 | init | 85,513 | 7872 | 93,386 |

1 | opt | 85,964 | 3802 | 89,766 |

2 | init | 89,414 | 7872 | 97,286 |

2 | opt | 89,418 | 4110 | 93,528 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Vandanjon, P.-O.; Vinot, E. Slope Optimization (or “Sloop”): Customized Optimization for Road Longitudinal Profile Eco-Design. *Energies* **2020**, *13*, 6575.
https://doi.org/10.3390/en13246575

**AMA Style**

Vandanjon P-O, Vinot E. Slope Optimization (or “Sloop”): Customized Optimization for Road Longitudinal Profile Eco-Design. *Energies*. 2020; 13(24):6575.
https://doi.org/10.3390/en13246575

**Chicago/Turabian Style**

Vandanjon, Pierre-Olivier, and Emmanuel Vinot. 2020. "Slope Optimization (or “Sloop”): Customized Optimization for Road Longitudinal Profile Eco-Design" *Energies* 13, no. 24: 6575.
https://doi.org/10.3390/en13246575