A Riemann sum is a certain type of approximations for an integral. The easiest and simplest method to use is a rectangle sum; where we sum the area of n rectangles with equal width to approximate the area under a curve; as n approaches infinity, the sum approaches the true value of the net area.
Here's a Riemann sum with 4 rectangles only:

This is just an example. Instead, consider n rectangles.
Let f be a function. Say, we wish to find the area under f from a to b. For the moment, let's only do the case a=0 and b=1. We will generalise later.
Now; imagine x_0, x_1, x_2, ... x_n to be some "marks" on the x-axis such that the width of the i-th rectangle is x_i - x_(i-1) and x_0 = 0.
Let Delta_i = x_i-x_(i-1) be the width of the i-th rectangle. As we want Delta_i to be constant, we can more easily call it color(red)(Delta).
Another property about our x-axis marks is that the lenght/non-width side of the i-th rectangle is going to be f(x_i). Hence, the its area is going to be
"Area"_i = "width" xx "lenght" = Deltaf(x_i)
Since we want the area until b=1, we must have x_n=1.
The areas of the rectangles approximate the areas under the graph of f. If there are n rectangles, then:
int_0^1 f(x) "d"x ~~ sum_(i=1)^n (x_i-x_(i-1))f(x_i) = sum_(i=1)^n Deltaf(x_i)
One way to make Delta_i constant is by defining x_i = i/n. Then,
Delta_i = i/n - (i-1)/n = 1/n, forall i in NN.
By this definition, x_0 = 0 and x_n = 1, as desired.
int_0^1f(x) "d"x ~~ sum_(i=1)^n 1/n f(i/n)
The more rectangles we have, the better the sum approximates the integral. Thus, we can say that
int_0^1 f(x) "d"x = lim_(n->oo) sum_(i=0)^n 1/n f(i/n)
Generalisation: Since we want the area from a to b, the very first rectangle must have lenght f(a) and the last one f(b); to do this, we can change the lower and upper bounds of the sum operator:
color(blue)(int_a^b f(x) "d"x = lim_(n->oo) sum_(i=an)^(bn) 1/n f(i/n))
This is one of the many Riemann Sum formulae for a function f.
In our case, f(x) = x+x^2:
:. int_a^b f(x) "d"x ~~ sum_(i=an)^(bn) 1/n (i/n + i^2/n^2)
:. int_a^b f(x) "d"x ~~ 1/n^2sum_(i=an)^(bn)(i+i^2/n)
if there are n rectangles, as stated already.