# How can I formulate an objective function that minimises the number of items required to solve a problem

I am currently trying to solve a problem where I need to minimise transport cost through the choice of vehicle (and how many of each choice) subject to a given demand.

The problem:

There are currently 3 vehicle sizes corresponding to their haulage capacity, an associated daily cost, and a daily demand.

I need an objective function that minimises the cost through the choice of vehicle whilst satisfying the sales demand, however, I do not know how to define an expression that is based on the number of vehicles used * the cost per day, as it depends on the weight.

Data:

# Vehicle capacity
truck_capacity_dict = {
'7.5T': {'capacity': 7500},
'12T': {'capacity': 12000},
'44T': {'capacity': 44000}
}

# Vehicle daily costs
truck_capacity_dict = {
'7.5T': {'rate': 350},
'12T': {'rate': 660},
'44T': {'rate': 2000}
}

# Daily demand
sales_demand_tonnes = {
'2020-01-01': 300,
'2020-01-02': 293,
'2020-01-03': 176
}


Mathematically, this is similar to the below expression where the OF is to minimise the costs based on the choice of vehicle:

$$\min \sum V_{t, v} \cdot C_{t, v} \forall t \subset T, v \subset V$$

However, I do not know how to formulate an expression in python that determines how many vehicles are chosen, as this depends on the weight.

• Are you trying to model a VRP variant or it is about the cargo optimization problem? Dec 16, 2020 at 9:19

## 2 Answers

I suggest introducing $$3n$$ variables. Let $$c_{i}$$ be the daily rate of each vehicle $$i=1,2,3$$.

$$x_{i,j}$$ designates the quantity of vehicles which will be used every day: the subscript $$j$$ indicates the day in interest. Supposing to consider $$n$$ days, we have $$j=1, 2, \ldots, n$$ days. Clearly, $$x_{i,j}$$ is a non-negative integer numbe and $$x_{i,j}=0$$ means that no vehicle of kind $$i-th$$ is used in $$j-th$$ day.

The total cost based on the number of vehicles used is $$\min \sum_{j=1}^n \sum_{j=1}^3 c_i \cdot x_{ij}$$

I am not sure of the bounds of the problem, but I'll try to provide an answer.

Let's consider the following two sets: $$V = \{v: vehicles\} \\ T = \{t: days\}$$ Now, we have the parameters you have provided, which I understand are the following:

$$VCAP_v: \text{capacity of vehicle v} \\ VCOST_v: \text{cost of vehicle v}\\ D_t: \text{demand of day t}$$

The only decision variable is the number of trucks that you have to deploy (each day?) to meet the demands. Therefore.

$$n_{t,v}: \text{integer variable that indicates the number of vehicles v that are deployed on day t}$$

You only have one constraint, which is the demand satisfaction constraint:

$$\sum_{v\in V}{n_{t,v}VCAP_v} \geq D_t \qquad \forall t\in T$$

And the objective function is simply:

$$\min z = \sum_{v\in V}\sum_{t\in T}n_{t,v}VCOST_v$$

If, on the other hand, you cannot change the election of trucks, this means, you have to decide which trucks to buy, the integer variable would lose a set ($$t$$).

If the question was however how to introduce integer variables, it depends on the progamming language. GAMS and Pyomo (I think) can directly establish a variable as integer. Otherwise, I think you have to manually use binary variables to establish the integer variables.