Tight MIP Formulations for Optimal Operation and Investment of Storage Including Reserves

Maaike B. Elgersma,Germán Morales-España,Karen I. Aardal,Niina Helistö,Juha Kiviluoma,Mathijs M. de Weerdt

2024-11-26

Abstract:Fast and accurate large-scale energy system models are needed to investigate the potential of storage to complement the fluctuating energy production of renewable energy systems. However, the standard Mixed-Integer Programming (MIP) models that describe optimal investment and operation of these storage units, including the optional capacity to provide up/down reserves, do not scale well. To improve scalability, the integrality constraints are often relaxed, resulting in Linear Programming (LP) relaxations that allow simultaneous charging and discharging, while this is not feasible in practice. To address this, we derive the convex hull of the solutions for the optimal operation of storage for one time period, as well as for problems including investments and reserves, guaranteeing that no tighter MIP formulation or better LP approximation exists for one time period. When included in multi-period large-scale energy system models, these improved LP relaxations can better prevent simultaneous charging and discharging. We demonstrate this with illustrative case studies of a unit commitment problem and a transmission expansion planning problem.

Optimization and Control

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is the high computational complexity and difficulty in scaling of existing Mixed - Integer Programming (MIP) models when describing the optimal investment and operation of energy storage systems, especially in cases where reserve capacity is included. Specifically:

1. **Simultaneous charging and discharging problem**: When the standard MIP model becomes a Linear Programming (LP) relaxation model after relaxing the integer constraints, it allows simultaneous charging and discharging, which is not feasible in practice.
2. **Computational efficiency problem**: When the existing MIP models are applied in large - scale energy system optimization models, the computational efficiency is low due to the exponential growth of the solution time.

To solve these problems, the paper proposes the following objectives:

- **Derive convex hull**: Derive the convex hull of the optimal operation of the energy storage system within a single time period, as well as the convex hull including investment and reserve capacity problems, to ensure that there are no tighter MIP formulations or better LP approximations.
- **Improve multi - period model**: Apply these improved LP relaxation models to large - scale multi - period energy system models to better prevent the occurrence of simultaneous charging and discharging.
- **Increase solution speed**: By introducing tighter MIP formulations, the solution speed can be significantly increased when solving large - scale MIP models.

### Formula summary

The key formulas involved in the paper are as follows:

#### Basic Operation MIP Model (BO - MIP)
\[
e_t = e_{t - 1}+\eta_C p_C^t\Delta-\frac{1}{\eta_D}p_D^t\Delta\quad\forall t\in T\tag{1a}
\]
\[
E_{\min}\leq e_t\leq E_{\max}\quad\forall t\in T\tag{1b}
\]
\[
p_C^t\leq P_C\delta_t\quad\forall t\in T\tag{1c}
\]
\[
p_D^t\leq P_D(1 - \delta_t)\quad\forall t\in T\tag{1d}
\]
\[
e_t, p_C^t, p_D^t\in\mathbb{R}_{\geq0}\quad\forall t\in T\tag{1e}
\]
\[
\delta_t\in\{0, 1\}\quad\forall t\in T\tag{1f}
\]

#### Basic Operation MIP Model with Reserve Capacity (BOR - MIP)
\[
e_t\geq E_{\min}+\eta_C r_C^{+t}\Delta+\frac{1}{\eta_D}r_D^{+t}\Delta\quad\forall t\in T\tag{2a}
\]
\[
e_t\leq E_{\max}-\eta_C r_C^{-t}\Delta-\frac{1}{\eta_D}r_D^{-t}\Delta\quad\forall t\in T\tag{2b}
\]
\[
p_C^t + r_C^{-t}\leq P_C\delta_t\quad\forall t\in T\tag{2c}
\]
\[
p_D^t + r_D^{+t}\leq P_D(1 - \delta_t)\quad\forall t\in T\tag{2d}
\]
\[
p_C^t - r_C^{+t}\geq0\quad\forall t\in T\tag{2e}
\]
\[
p_D^t - r_D^{-t}\geq0\quad\forall t\in T\tag{2f}
\]
\[
r^-_t = r_C^{-t}+r_D^{-t}\leq R^-\quad\forall t\in T