Evaluating Utility-Scale PV-Battery

Hybrids in Operational Models for the

Bulk Power System
Venkat Durvasulu, Caitlin Murphy, and Paul Denholm
National Renewable Energy Laboratory

NREL is a national laboratory of the U.S. Department of Energy Technical Report

Office of Energy Efficiency & Renewable Energy NREL/TP-6A20-78850
Operated by the Alliance for Sustainable Energy, LLC April 2021
This report is available at no cost from the National Renewable Energy
Laboratory (NREL) at www.nrel.gov/publications.

Contract No. DE-AC36-08GO28308

 Evaluating Utility-Scale PV-Battery
Hybrids in Operational Models for the
Bulk Power System
Venkat Durvasulu, Caitlin Murphy, and Paul Denholm
National Renewable Energy Laboratory

Suggested Citation
Durvasulu, Venkat, Caitlin Murphy, and Paul Denholm. 2021. Evaluating Utility-Scale PV-
Battery Hybrid in Operational Models for the Bulk Power System. Golden, CO: National
Renewable Energy Laboratory. NREL/TP-6A20-78850.
https://www.nrel.gov/docs/fy21osti/78850.pdf.

NREL is a national laboratory of the U.S. Department of Energy Technical Report

Office of Energy Efficiency & Renewable Energy NREL/TP-6A20-78850
Operated by the Alliance for Sustainable Energy, LLC April 2021
This report is available at no cost from the National Renewable Energy National Renewable Energy Laboratory
Laboratory (NREL) at www.nrel.gov/publications. 15013 Denver West Parkway
Golden, CO 80401
Contract No. DE-AC36-08GO28308 303-275-3000 • www.nrel.gov
 NOTICE

This work was authored by the National Renewable Energy Laboratory, operated by Alliance for Sustainable
Energy, LLC, for the U.S. Department of Energy (DOE) under Contract No. DE-AC36-08GO28308. Funding
provided by the U.S. Department of Energy Office of Energy Efficiency and Renewable Energy Solar Energy
Technologies Office and the Office of Strategic Analysis. The views expressed herein do not necessarily represent
the views of the DOE or the U.S. Government.

This report is available at no cost from the National Renewable

Energy Laboratory (NREL) at www.nrel.gov/publications.

U.S. Department of Energy (DOE) reports produced after 1991

and a growing number of pre-1991 documents are available
free via www.OSTI.gov.

Cover Photos by Dennis Schroeder: (clockwise, left to right) NREL 51934, NREL 45897, NREL 42160, NREL 45891, NREL 48097,
NREL 46526.

NREL prints on paper that contains recycled content.

Acknowledgments
We express our appreciation to the following colleagues at the National Renewable Energy
Laboratory for their feedback and suggestions on this work: Patrick Brown, Kelly Eurek, Anna
Schleifer, and Jaquelin Cochran. Editing was provided by Mike Meshek and Madeline
Schroeder. We thank LADWP for allowing the use of data sets generated for the LA100 study.

iii
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
List of Acronyms
AC alternating current
BES battery energy storage
CSP-TES concentrating solar power with thermal energy storage
DC direct current
GWh gigawatt-hour
ILR inverter loading ratio
LADWP Los Angeles Department of Water and Power
MWh megawatt-hour
PCM production cost model
PV photovoltaic
SAM System Advisor Model
SoC state of charge
TES thermal energy storage
TWh terawatt-hour

iv
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
Abstract
Systems that combine solar photovoltaic and battery energy storage technologies (PV-BES) are
increasingly being proposed for, and deployed on, the bulk power system. The operation and
value of PV-BES systems have been extensively studied from the perspective of project
developers through analyses that maximize plant-level revenue. However, PV-BES hybrids’
operational characteristics have seldom been studied from the perspective of bulk power system
operators, who seek to optimize the performance of a suite of generation and storage assets that
are connected via the transmission network.

This work presents modeling approaches for representing and evaluating PV-BES hybrids in a
model that optimizes operations across the bulk power system. Its novel contributions include
demonstrating a technique to modify a unit commitment and dispatch model to represent the
operational synergies of PV-BES hybrids. In particular, we describe the challenges of, and an
approach for, representing so-called DC-coupled PV-BES—which use a single bidirectional
inverter—as a dispatchable resource in a commercial, production cost model (PCM), namely
PLEXOS.

We demonstrate this technique in a PCM study of the Los Angeles Department of Water and
Power test system, by replacing existing PV and battery generators on the test system with our
PV-BES hybrids. We then pursue scenario analysis designed to isolate the various drivers of
operational strategies for DC-coupled PV-BES hybrids, including the nature of coupling, PV
penetration on the system, and varying inverter loading ratios (or degrees of over-sizing the
PV field). Results from the analysis include utilization profiles for the PV DC energy across
available pathways, dispatch profiles for the battery component, and the hybrid technologies’
impacts on system-wide production costs. The approach we present can be used in any PCM
study of PV-BES hybrids as a resource in different power system configurations and services.

v
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
Table of Contents
1 Introduction ........................................................................................................................................... 1
2 PV-BES Hybrids .................................................................................................................................... 3
3 PV-Storage Hybrid Production Cost Modeling .................................................................................. 5
4 Capability Demonstration .................................................................................................................. 11
4.1 Simulation Setup ......................................................................................................................... 11
Test System ................................................................................................................................. 11
PV-BES Plants ............................................................................................................................ 11
Scenario Design........................................................................................................................... 12
4.2 Results: PV-BES Modes of Operation ........................................................................................ 14
4.3 Results: Impact of Configuration ................................................................................................ 16
5 Conclusions and Future Work .......................................................................................................... 22
References ................................................................................................................................................. 24
Appendix: Transmission-Line Inverter Topology .................................................................................. 27

vi
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
List of Figures
Figure 1. PV-BES hybrid configurations, including (a) an AC-coupled system with inverter
communication link and (b) a DC-coupled system with grid charging capabilities using a
bidirectional inverter ................................................................................................................ 3
Figure 2. Simulation components that a PV-BES object must represent, along with their efficiencies and
ratings for PCM analysis .......................................................................................................... 6
Figure 3. PLEXOS implementation of DC-coupled PV-BES using the storage generator topology ........... 9
Figure 4. Hourly operation of DC-coupled PV-BES Plant 3 with ILR=2.0 on July 12 and 13 .................. 15
Figure 5. Hourly operation of DC-coupled PV-BES Plant 3 in ILR 1.3 on July 11, 2020 in the 1/2 PV
Scenario (left) and ILR=1.3 scenarios ................................................................................... 16
Figure 6. Battery discharge behavior for the battery component of the DC-coupled PV-BES hybrids with
ILR 1.3 (pink) and 2.0 (green), aggregated over the 1-year analysis period.......................... 17
Figure 7. Destination of PV DC energy collected as a function of ILR for all DC-coupled PV-BES
hybrids in this demonstration ................................................................................................. 19
Figure 8. The distribution of the unutilized PV DC energy due to losses and spillage .............................. 20
Figure A-1. PLEXOS implementation of DC-coupled PV-BES using a transmission line-
inverter topology .................................................................................................................... 27

List of Tables
Table 1. Equations and Values that Define the Efficiencies for Each Available Pathway ......................... 11
Table 2. Capacity Inputs and Generation Outputs for all Generators in the LADWP System ................... 12
Table 3. Details of the DC-Coupled PV-BES Implemented in the Base Case LADWP System ............... 13
Table 5. Utilization Metrics for the DC-Coupled Hybrid-Inverter ............................................................. 18

vii
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
1 Introduction
Solar photovoltaic (PV) systems have experienced rapid and widespread deployment on the U.S.
bulk power system in recent years, and their deployment is expected to continue and accelerate.
Also, interest in, and deployment of, battery energy storage (BES) is growing (Cole et al. 2020;
EIA 2020), in part to mitigate the declining operational value of PV systems with increasing
penetration (Wiser et al. 2017; Mills and Wiser 2012; Hirth 2015; 2013; Mai, Cole, and Reimers
2019)

While BES systems can be deployed independently, recent installations and interconnection
queue data indicate growing industry interest in systems that couple PV and BES through
colocation and perhaps coordinated operations (Wiser et al. 2020). Moreover, market projections
indicate an acceleration in the deployment of coupled PV-BES systems in the United States
(Hledik et al. 2019). These trends reflect both policy drivers (e.g., the federal investment tax
credit can be applied to storage if it charges primarily from colocated PV) and the potential for
technology synergies (Gorman et al. 2020).

It is important to determine the actual value of colocated or coupled systems, particularly relative
to independent deployments and also given the number of options for configurations and
component sizing. Common approaches for quantifying the operational value of generation
assets include price-taker methods and production cost models (PCMs) (Martinek et al. 2018).
The respective merits of PCM and price-taker methods are well established, as are the
requirements for performing simulation and analysis (Martinek et al. 2018).

The price-taker approach is one of the most commonly implemented techniques to estimate the
potential revenue of a single installation, based on historical or forecasted electricity prices in a
region (DiOrio, Denholm, and Hobbs 2020; Martinek et al. 2018; Gorman et al. 2020; Schleifer
et al. 2021). Given their computational efficiency, price-taker methods can be used to study how
the value of PV-BES systems depends on the complex characteristics of the battery component.
For example, recent research has explored how PV-BES value varies when using different
battery dispatch algorithms (Gorman et al. 2020; Mills and Rodriguez 2020), or the when
considering the impact of nonlinear aspects of battery systems (e.g., voltage, current, and cycle
degradation) (DiOrio, Denholm, and Hobbs 2020).

Price-taker methods are useful for exploring how these characteristics influence the value that
can be realized through PV-BES operational synergies, particularly for installations that are too
small to induce changes in system marginal prices. However, they do not capture the impact of
new resources on the rest of the system. In addition, price-taker methods are typically configured
to maximize plant-level revenues, as opposed to system-level benefits.

To understand the value of various resources at the system level, a PCM is typically used. PCMs
optimize the operation of a new resource along with all the existing generators to minimize the
total system variable operating (production) cost. The benefits of a new resource are then
measured in terms of a reduction in production cost relative to an alternative mix of resources
that does not include the added generator (Jorgenson et al. 2018). PCMs are in important tool
used by utilities and other power system planners to develop and analyze resource plans.
Because of the potential role of PV-BES hybrids in the evolving power system, and because of

1
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
the unique characteristics of certain PV-BES configuration, it is important to accurately
characterize these technologies in commercially available PCM tools.

Commercial PCMs can model PV and storage with various levels of detail. However, in our
experience, many commercial PCM software packages do not have a preprogrammed simulation
object that captures all the nuances of the emerging PV-BES technologies, particularly those
associated with DC-coupled systems.

This study focuses on the development of a generalized modeling approach for representing PV-
BES in PCMs, using a specific commercial tool as an example. Section 2 provides a summary of
utility-scale PV-BES, including all its essential components and relevant nuances (compared to
independent PV and battery systems). Section 3 describes a generalized approach to modeling
PV-BES hybrids in a commercial PCM software, and its specific implementation in the PLEXOS
model. Section 4 applies this approach in a model based on the Los Angeles Department of
Water and Power (LADWP) test system. We demonstrate its ability to capture differences
between hybrid and conventional (noncoupled) PV and BES configuration. In Section 5, we
present conclusions and discuss further applications and refinements that would advance the
current understanding of PV-BES impacts and value on the bulk power system. The appendix
describes an alternative approach to implementing the proposed DC-coupled PV-BES in
PLEXOS and other PCMs.

2
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
2 PV-BES Hybrids
A PV-BES system has two basic configurations: AC-coupled and DC-coupled (Figure 1). The
AC-coupled system in Figure 1a consists of a PV system and a storage system connected to a
common point on the AC grid. This configuration has advantages when retrofitting storage to an
existing PV installation because (a) it avoids the need to replace the PV inverter and (b) it allows
the BES to be deployed on the periphery of the previously developed PV system (Fu, Remo, and
Margolis 2018). From an operational standpoint at the system level, an AC-coupled system is
largely functionally equivalent to independent systems. Consequently, there is little need to
develop new models in a PCM environment. 1

(a) (b)

Figure 1. PV-BES hybrid configurations, including (a) an AC-coupled system with inverter
communication link and (b) a DC-coupled system with grid charging capabilities using a
bidirectional inverter

Figure 1b illustrates a DC-coupled PV-BES hybrid, which is the focus of the present study. In
a DC-coupled configuration, the PV and BES are connected on the DC side of a shared inverter
which, in this case, is bidirectional (thus enabling the battery to charge from PV or from the
grid). Because of the shared inverter, this configuration requires additional constraints and
provides additional benefits that are typically not captured in a PCM without modification. The
primary constraint is the shared inverter, which reduces operational flexibility of the PV and
battery compared to independent components. Benefits include increased efficiency when
charging from PV due to fewer conversion losses (DiOrio, Denholm, and Hobbs 2020). In
addition, DC coupling of the PV and battery components enables the ability to recover PV
generation that would otherwise be clipped by the inverter.

A special case of the DC-coupled system is one where a unidirectional inverter is used and the
battery can charge only from the PV system to which it is coupled. This configuration removes
several of the degrees of freedom of operation and is easier to implement. It also shares many
similarities with concentrating solar power with thermal energy storage (CSP-TES), and
therefore can be implemented by adapting existing model formulations for CSP-TES if available
(Jorgenson et al. 2018).

1
The battery inverter could have a communication link with the PV inverter to modify the charge control strategy
based on the PV power availability. Such an operational coupling would require alteration of the PCM formulation.

3
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
In addition to basic configurations, there are several other important aspects of modeling PV-
BES plants. The first is the relative sizing of the components, particularly the impact of the
inverter loading ratio (ILR). The ILR, or DC-to-AC ratio, represents the sizing of the PV panels
(rated in DC power) relative to the inverter output power (rated in AC). The ILR for a typical PV
system is about 1.3, which increases utilization of the inverter and, in turn, reduces the overall
system levelized cost (Good and Johnson 2016; Martins Deschamps and Rüther 2019). Higher
ILRs will increase the amount of time the panels produce more energy than can be used,
increasing clipped energy during the peak-solar hours of the year. A DC-coupled system can
recover this otherwise clipped PV energy and enable even higher ILRs, thus potentially reducing
levelized system costs. Detailed discussion of the components of a PV-BES systems is provided
by (Bullich-Massagué et al. 2020; Wang, Ciobotaru, and Agelidis 2014; Hu et al. 2018;
Miñambres-Marcos et al. 2017; Chen, Huang, and Yu 2013).

A second important issue is the objective of the modeling exercise and consideration of the
perspective of the plant owner/operator. A properly designed PCM will optimize the system
operation to minimize system costs. The PCM does not maximize revenue, nor does it
necessarily consider certain policies that could result in dispatch that deviates from the system-
wide least cost. An example is the impact of the federal investment tax credit: if more than 75%
of the energy discharged from the battery is sourced from a coupled PV system, then a fraction
of the battery’s capital costs can qualify for the investment tax credit. In addition to the credit,
other federal and state incentives might increase the amount of PV charging, including eligibility
for the Modified Accelerated Cost Recovery System and availability of renewable energy credits
(Elgqvist, Anderson, and Settle 2018). Consequently, PCM modeling of hybrid plants must
consider the impact of dispatch decisions and the monetization of services provided by storage.
In particular, a PCM formulation should represent all the components and flows of the PV-BES
hybrid and track the percentage of the energy source charging the BES, in order to represent the
impacts of the various policy incentives. The following section presents a technique to
implement PV-BES in a commercial PCM software.

4
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
3 PV-Storage Hybrid Production Cost Modeling
The objective of a PCM is to minimize the operational (or production) cost of all generators,
such that the generation meets the load requirement at least cost while not violating any system
constraints. The objective function of a PCM can be represented using Equation (1), where 𝑃𝑃𝑔𝑔𝑡𝑡 is
the electricity produced by generator 𝑔𝑔 at time point 𝑡𝑡; 𝐶𝐶𝑔𝑔 is the incremental generation cost
(including fuel and variable operation and maintenance costs); 𝑆𝑆𝑔𝑔 is the cost of starting generator
𝑔𝑔; and 𝑦𝑦𝑔𝑔𝑡𝑡 is a binary variable denoting the startup of the generator 𝑔𝑔. It is further subject to
system constraints, which include (but are not limited to) generator operational constraints
(Equation 2), demand (𝑃𝑃𝑑𝑑𝑡𝑡 ) and generation balance (Equation 3), and transmission line
constraints (Equation 4), where 𝑃𝑃𝑙𝑙𝑡𝑡 is the power flow in line 𝑙𝑙 during interval 𝑡𝑡, and the 𝑃𝑃𝑙𝑙𝑚𝑚𝑚𝑚𝑚𝑚 is
the flow limit of line 𝑙𝑙.

𝑚𝑚𝑚𝑚𝑚𝑚 ∑𝑡𝑡∈𝜏𝜏 ∑𝑔𝑔�𝐶𝐶𝑔𝑔 𝑃𝑃𝑔𝑔𝑡𝑡 + 𝑆𝑆𝑔𝑔 𝑦𝑦𝑔𝑔𝑡𝑡 � (1)

𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑡𝑡𝑡𝑡:

Pgmin ≤ Pgt ≤ Pgmax (2)

∑𝑔𝑔 𝑃𝑃𝑔𝑔𝑡𝑡 = 𝑃𝑃𝑑𝑑𝑡𝑡 (3)

𝑃𝑃𝑙𝑙𝑡𝑡 ≤ 𝑃𝑃𝑙𝑙𝑚𝑚𝑚𝑚𝑚𝑚 (4)

Our implementation of PV-BES hybrids builds on previous work modifying PCMs to represent
CSP-TES systems (Jorgenson et al. 2018; 2013; Mehos et al. 2015). To include PV-BES using
Equation 1, the simulation object must be able to (a) generate a time-series flow of energy from
the PV to BES, PV to the grid, BES to the grid, and grid to the BES, and (b) optimize these flows
to minimize costs. Therefore, the simulation object must constitute separate PV and BES objects
within a generator object in order to calculate these flows. The simulation objects and their
properties that are necessary for DC-coupled PV-BES are presented in Figure 2 (page 6),
including a PV object, a storage object, an inverter (flow object), and a converter (flow object).
𝑚𝑚𝑚𝑚𝑚𝑚
The PV object is characterized by its nameplate DC power (𝑃𝑃𝐷𝐷𝐷𝐷 ) relative to the AC rating of
the inverter (𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖 ), as defined by the ILR in Equation (5).
𝑚𝑚𝑚𝑚𝑚𝑚
𝑃𝑃𝐷𝐷𝐷𝐷
𝐼𝐼𝐼𝐼𝐼𝐼 = (5)
𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖

The bidirectional inverter (blue) is used for (a) PV-to-grid generation, (b) battery-to-grid
discharge, and (c) grid-to-battery charging. The DC-DC converter (purple) is used for (a) PV-to-
battery charging, (b) battery-to-grid discharge, and (c) grid-to-battery charging (via the inverter).
The key characteristics of these devices are power rating and efficiency, resulting in three
effective efficiencies for the different operating modes: PV-to-battery, PV-to-grid, and grid-to-
battery.

5
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 Figure 2. Simulation components that a PV-BES object must represent, along with their
efficiencies and ratings for PCM analysis
Finally, the storage device is characterized by its efficiency, energy capacity, and power
capacity, where the power capacity is ultimately limited by the rating of the DC-DC converter
and inverter.

Commercial PCMs typically include PV and battery objects. However, linking those objects
without modification would likely not result in an adequate PV-BES representation because
of the unique aspects of DC-coupled systems. Therefore, additional modifications (or
parameterizations) may be needed to capture the unique properties of PV-BES. For example:

• PV objects in commercial PCMs are typically represented by a time-series profile of AC

power availability; this profile is generated by an external modeling tool that accounts for
local solar conditions and component sizing, including ILR, inverter efficiency, 2 and
other wiring and interconnection losses based on the system design (panels and inverter
arrangement). To adequately represent avoided clipping and improved efficiencies in PV-
𝑡𝑡
BES, the model needs to use a time series DC profile (𝑃𝑃𝐷𝐷𝐷𝐷 ) from the PV modules, which
helps capture the potential to increase total output by recovering (and shifting) otherwise-
clipped energy.

• Storage objects in commercial PCMs are typically defined by their energy rating (𝑈𝑈𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ),
the power rating of the inverter, and the roundtrip efficiency of the complete system
(including battery, power electronics, and parasitic-related losses). In representing PV-
BES, the battery object would ideally allow for separate efficiencies for each charging
source (grid, local PV), such that the efficiency of the storage (𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ) is represented
separately based on the round-trip battery pack efficiency (including the battery internal
and interconnection losses and certain parasitics not associated with charging). The
battery power must also be limited by the power rating of the DC-DC converter (𝑃𝑃𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ).

2
For example, the inverter model within NREL’s System Advisor Model (Freeman et al. 2018) generates such
profiles, based on libraries of performance data for PV inverters. The underlying performance parameters are
derived from empirical data regarding the relationship between DC input and AC output for grid-connected PV
inverters, as well as the self-consumption of the inverter itself under a range of operating and environmental
conditions (King et al. 2007).

6
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 • Most commercial PCMs do not have separate inverter and DC-DC converter objects, so
they cannot represent the dynamic behavior of these components. In their native forms,
the power rating of the entire PV-BES system would be defined by the hybrid-inverter
(𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖 ); an ideal representation of PV-BES would allow the inverter and DC-DC
converter efficiencies to be considered separately.

There are potential tradeoffs in (a) modeling details associated with the modifications outlined
above and (b) capturing changes in the different operating modes associated with a DC-coupled
PV-BES. Overall, a DC-coupled PV-BES with bidirectional inverter can operate in seven
different modes:

• PV-to-grid
• PV-to-grid and PV-to-battery
• PV-to-battery only
• Grid-to-battery
• Battery-to-grid
• Battery-to-grid and PV-to-grid
• PV-to-battery and grid-to-battery.
𝑃𝑃𝑃𝑃
The efficiency of charging the battery from the PV system (𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ) as in Equation (6) is greater
𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺
than the efficiency of charging from the grid (𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ), which includes the inverter efficiency as
shown in Equation (7).
𝑃𝑃𝑃𝑃
𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 = 𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 (6)
𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺
𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 = 𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 (7)

As a result, a detailed model of a DC-coupled system requires representing both the inverter and
DC-DC converter, which increases complexity relative to the AC-coupled or independent
systems. This increased complexity may not be justified if the magnitude of reduced losses
associated with PV charging are small (DiOrio, Denholm, and Hobbs 2020).

Further complicating the issue is that the introduction of the DC inverter object in a PCM
environment likely results in a loss in modeling fidelity of the PV performance. For example,
detailed inverter models in simulation tools that typically generate AC profiles (e.g., the National
Renewable Energy Laboratory’s System Advisor Model, or SAM) are designed to capture the
complicated nonlinear characteristics of inverters. 3 In this study, the DC module production is
modified by a single inverter efficiency to generate AC profiles, which are separated in order to

3
Empirical evidence indicates that the relationship between measured AC and DC power associated with grid-
connected PV inverters is nearly linear, regardless of operating and environmental conditions (e.g., rapid changes in
cloud cover). However, the inverter efficiency itself (AC power divided by DC power) is nonlinear because of
varying levels of self-consumption at different voltage and power levels (King et al. 2007). The scatter in the
efficiency measurements resulted from a combination of variation in DC input voltage, inherent inverter behavior,
rapidly varying solar irradiance, and measurement error.

7
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
quantify the utilization of energy that would otherwise be clipped (within a detailed inverter
model outside PLEXOS).

To estimate the impacts of a simplified inverter representation, we compare AC output from a

specific PV system using (a) the detailed inverter model included in SAM 4 and (b) outputs based
on the approach proposed in this report. Using the same DC production information for a PV
array with a DC rating of 570 MW and ILR=1.3 in Barren Ridge, California, we find that the
annual AC output is 327 gigawatt-hours (GWh) based on a simulation in SAM, compared to 335
GWh based on the current approach. In other words, the use of a single inverter efficiency value
results in a 2.7% increase in AC output for ILR=1.3, compared to that simulated with a more
detailed inverter model. This difference further depends on the assumed ILR, and it declines to
1.8% for ILR=1.8. Given the magnitude of impact, this limitation should be noted, but it does not
diminish the outcomes or findings of this analysis, which are focused on modeling a system that
can study the flow and interaction between the AC and DC sides of a hybrid setup.

We implement the model presented above in PLEXOS, a commercial PCM. Commercial PCMs
are typically not open source, and changes must typically be used to modify existing capabilities.
Our example uses the PLEXOS commercial PCM and objects and constraints that already exist
in the tool. There can be multiple techniques to implement a PV-BES in a PCM software, and in
this paper, we present an approach using the storage-generator topology, which is similar to the
approach used for implementing CSP-TES (Jorgenson et al. 2018).

In this approach, we use generator objects to represent inverters and hydro-storage objects to
represent the battery and energy inflow (the PV generation). Figure 3 presents the schematic of
the storage-generator topology, using standard PLEXOS objects. 5 The PV component of the PV-
BES is defined by three linked objects:

1. A head storage object represents the DC generation profile; it is set to not store any
𝑡𝑡
energy, and its natural inflow property is linked to the DC rating (𝑃𝑃𝐷𝐷𝐷𝐷 ) of the PV. The
available “flows” for the PV energy, then, include sending it to the grid (through the
connected generator, #2 below) or to the battery (#3 below).

2. A generator object represents the PV-inverter (blue triangles), where the rating property
(𝑃𝑃𝑃𝑃𝑃𝑃 ) is set equal to the inverter rating (𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖 ), and the generator efficiency property is set
equal to the inverter efficiency (𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 ).

4
The simulation uses 2012 weather year data for all PV profiles. The simulation is set up using a detailed PV plant
model in SAM that includes 743 MW of peak PV DC capacity and inverters that are sized to achieve ILR=1.3 (or
570 MW of PV AC capacity).
5
Using PLEXOS terminology, the PV array is represented as a hydroelectric dam with no storage capacity; the
inverter is represented as a hydroelectric generator; the battery is represented as a pumped-hydro system with tail
and head storage reservoirs (where water in the “head” represents energy stored in the battery and water in the “tail”
is equal to the difference between the maximum battery capacity and energy in the “head”); and the DC-DC
converter is represented as a waterway linking the PV hydroelectric dam with the battery “head” reservoir.

8
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 3. A waterway object represents the flow of PV generation to the DC-coupled battery; its
flow limit property represents the DC-DC converter (or charge controller) rating (𝑃𝑃𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ),
and its output scalar property represents the efficiency of the charge controller (𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ).

PV energy is spilled during intervals when additional PV energy cannot be utilized by either of
these two flows (#2 and #3 above).

Figure 3. PLEXOS implementation of DC-coupled PV-BES using the storage generator topology
Inverters are represented using generator objects, and energy sources are represented using hydro-storage objects.
Italicized text represents properties in PLEXOS.
* Equation (8) links the separate inverter objects to emulate a single inverter.

The battery component of the PV-BES is represented using a pumped-hydro generator object,
which can “charge” through two pathways (represented by the arrows in Figure 3). The pump
load property (𝑃𝑃𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 ) represents the load on the system to charge the battery from the grid (red
arrow). Charging from PV through the DC-DC controller is represented by the PLEXOS
waterway object (#3 above). The rating of the pumped-hydro generator object (𝑃𝑃𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ) is set to
equal to the DC-DC converter rating times the inverter efficiency (𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 𝑃𝑃𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ), which defines the
peak discharge capability of the battery at the AC node. This model provides the flexibility to
have a different charge efficiency to charge the battery using each path (i.e., PV-to-battery or
grid-to-battery via the shared inverter). Properties associated with pumped storage plants, such as
minimum generation levels and ramp rate constraints, are set to zero.

The hybrid-inverter (blue in Figure 3) is responsible for PV-to-grid, battery-to-grid, and grid-to-
battery flows. For the model in Figure 3 to represent a single inverter for all the flows, the ratings
of each of the separate objects are constrained using Equation (8) so that any combination of PV
𝑡𝑡 𝑡𝑡 𝑡𝑡
generation (𝑃𝑃𝑃𝑃𝑃𝑃 ), battery discharge (𝑃𝑃𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ), and grid charging (𝑃𝑃𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 ) is less than or equal to
the inverter rating (𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖 ) for all intervals (t) of the simulation T.

9
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 𝑡𝑡 𝑡𝑡 𝑡𝑡
𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑃𝑃𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 + 𝑃𝑃𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 ≤ 𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖 , ∀𝑡𝑡 ∈ 𝑇𝑇 (8)

The battery object efficiency is set to 100%, and the losses corresponding to either energy source
(PV or grid) are incorporated into the charging efficiency of either path (PV or grid). Equation
(6) is replaced with Equation (9) in order to represent the efficiency gains associated with
charging from the PV and discharging to the grid (compared to charging from and discharging
to the grid).
𝑃𝑃𝑃𝑃 2
𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 = 𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 (9)
2
The squared converter efficiency (𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ) comes from the converter's charge and discharge
actions, thus representing a roundtrip efficiency. Equation (7) is replaced with Equation (10) to
represent the round-trip efficiency of the energy from and to the grid; this efficiency is set as the
pump efficiency property for the PLEXOS battery generator object.
𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺 2 2
𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 = 𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 (10)

All the flows through the converter must be less than or equal to the converter capacity (𝑃𝑃𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 )
as described in Equation (11):
𝑡𝑡 𝑡𝑡 𝑡𝑡
𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 𝑃𝑃𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 + 𝑃𝑃𝑤𝑤𝑤𝑤−𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 + 𝑃𝑃𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 /𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 ≤ 𝑃𝑃𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 , (11)
𝑡𝑡
where (𝑃𝑃𝑤𝑤𝑤𝑤−𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 ) represents the flow through the waterway (PV-charging) at an interval 𝑡𝑡. The
entire pump load (grid charging) does not fall on the converter, as some power is lost in the
inverter; to represent these losses, the inverter efficiency (𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 ) is multiplied by the pump load
𝑡𝑡
(𝑃𝑃𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 ). The inverter losses during the discharge cycle are included in the charging efficiency as
described in Equation (9) and Equation (10). The energy discharge through the converter must be
corrected for the inverter losses and, thus, is divided by the inverter efficiency as described in
Equation (11).

The state of charge (SoC) of the battery is derived by dividing the volume in head storage by the
sum of the volume in head storage. The ability to discharge is determined by the volume
available in the tail storage. Because of the two charging sources, the head storage can be filled
without using the pump. A constraint is added such that at every time interval 𝑡𝑡, the volume of
𝑡𝑡 𝑡𝑡
the head storage (𝑈𝑈ℎ𝑒𝑒𝑒𝑒𝑒𝑒 ) and tail storage (𝑈𝑈𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 ) is equal to the battery capacity (𝑈𝑈𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ) for all
intervals of the simulation T, as shown in Equation (12).
𝑡𝑡 𝑡𝑡
𝑈𝑈ℎ𝑒𝑒𝑒𝑒𝑒𝑒 + 𝑈𝑈𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 = 𝑈𝑈𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 , ∀𝑡𝑡 ∈ 𝑇𝑇 (12)

In addition, the tail storage maximum spill property is set equal to 𝑈𝑈𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 so that the energy
balance of Equation (12) can be maintained when the head storage is being charged from PV,
and the equivalent volume is spilled out of the tail storage so that the hydro model can generate
electricity.

An alternative approach that uses the transmission line as an inverter is presented in the
appendix. Though the transmission-line inverter topology approach produces similar results,
we find that it increases the simulation time.

10
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
4 Capability Demonstration
This section presents the validation of the proposed model through simulation of a real power
system in PLEXOS. The simulation setup and the detailed validation results are presented in the
following subsections.

4.1 Simulation Setup

To demonstrate our implementation of a PV-BES hybrid, we begin by defining its performance
characteristics. In particular, we set efficiencies for the inverter (𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 ), converter (𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ), and
battery (𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ) to 97.3%, 99%, and 92%, respectively, following assumptions from the literature
(Schimpe et al. 2018; Gilman et al. 2008; Mongird et al. 2020). See Table 1 for a summary and
details associated with the available energy pathways. The value for parameter 𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 assumes the
battery pack efficiency that includes lithium-ion cell internal efficiency, battery interconnection
efficiency, and the battery pack thermal losses based on (Schimpe et al. 2018).

Table 1. Equations and Values that Define the Efficiencies for Each Available Pathway

Pathway Equation Net Efficiency (%)

PV-grid 𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 97.3
Grid-battery-grid (𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 2 )(𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 2 )𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 85.4
PV-battery-grid (𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 2 )𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 87.7

Test System
Next, we add our PV-BES technology to a test system based on the LADWP generation and
transmission system. In particular, we use a test system that leverages data sets developed for the
Los Angeles 100% Renewable Energy (LA100) Study (Cochran and Denholm 2021) based on
the 2020 capacity mix (first column Table 2, page 12).

The model has 108 nodes and 189 branches with an annual demand of 28.4 terawatt-hours
(TWh), peaking at 6,723 MW. Some of these generation technologies are an aggregation of
subtechnologies; for example, PV includes rooftop PV, distributed PV, and utility-scale PV.
Timeseries data include hourly load (generated as part of the LA100 study), hourly PV
generation profiles from the National Solar Resource Database (using 2012 meteorological
conditions), and hourly wind profiles from AWS Truepower (using 2012 meteorological
conditions). “Demand Response” is flexible demand that can be curtailed or shifted. Performance
of the thermal fleet (e.g., heat rates) are based on data from the LA100 study. The PV-Battery
generators in the base representation of the LADWP system are AC-coupled PV-BES systems
with ILR=1.3.

PV-BES Plants
To implement our new PV-BES hybrid technology, we modified assumptions for five of the PV-
Battery generators on the LADWP test system (Table 2). The combined capacity of these five
PV plants is 1,326 MW (of 1,401 MW for the PV-Battery category), which represents 13% of
the total installed capacity (10,694 MW). The power rating of the inverter and converter, and the
energy rating of the battery for the five existing plants are adopted for this study. Power and

11
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
energy ratings and the nameplate PV DC rating for these five plants are presented in Table 3
(page 13).

All the PV-BES constraints discussed in Section 3 are applied. The peak power from battery is
set to half the peak DC power of the PV (for ILR=1.3), and its energy rating is based on an
assumption of 4 hours of usable peak power; in other words, the battery component will
experience 100% depth of discharge after operating for 4 hours at the peak power rating; or,
more likely, the storage rating can be interpreted as a battery whose nameplate energy capacity
has already been adjusted to account for SoC restrictions. Though this distinction would have
important implications for investment costs, it does not impact the operational characteristics
presented here. This sizing (or any of the other aspects of the systems analyzed) is not intended
to be optimal but is used just to evaluate the performance of the overall modeling approach.

Scenario Design
Using the LADWP test system, our PV-BES implementation, and our assumed PV-BES sizing,
we define scenarios that isolate drivers of operational strategies for hybrid system, including the
effects of DC coupling, PV penetration, and ILR for the PV-BES hybrids.

The effects of DC coupling are explored with scenarios that compare operational behavior
of AC-coupled and DC-coupled configurations (see Figure 1) for PV-BES with ILR=1.3.
Comparing the operational behavior of scenarios with the same sizing but different architectures
enables verification of the utilization of clipped energy and evaluation of the impacts of varying
efficiencies for each available energy pathway.

Table 2. Capacity Inputs and Generation Outputs for all Generators in the LADWP System

Capacity (MW) Energy (GWh)

Generator Type All Scenarios Base Case 1/2 PV Sensitivity
Biomass 22 1.4 1.8
Coal 1,679 8,578 8,930
Demand Response 44 0.8 1.5
Gas-CC 2,362 4,212 5,545
Gas-CT 881 789 1,040
Gas-Steam 712 82 116
Geothermal 235 1,834 1,838
Hydro 726 1,661 1,633
Nuclear 394 3,271 3,272
Pumped-Hydro 427 17 18
PV 621 1,063 1063
PV-Battery 1,401 5067 2,995
Storage 147 6 3.3
Wind 1,000 2,683 2,685
Total 10,694 29,277 29,142a
a The 1/2 PV Scenario has a lower total generation than the full-PV case for the same demand because of reduced
use of storage that decreases demand on the system. The reduced PV results in increase of coal and natural gas
generation that results in higher prices that discourages use of storage.

12
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
The effects of PV penetration on PV-BES operation and value are explored via the 1/2 PV
Scenario, in which total PV generation (including PV-BES) on the LADWP test system is halved
and all PV-BES hybrids are implemented as DC-coupled systems with ILR=1.3.

The effects of increasing ILR from 1.3 to 2.0 are explored by fixing the AC rating of the inverter
and increasing the PV DC rating, with the resulting nameplate DC ratings shown in Table 3.
Comparing the operational behavior of scenarios with different ILRs enables evaluation of how
excess PV production is utilized in DC-coupled systems.

Table 3. Details of the DC-Coupled PV-BES Implemented in the Base Case LADWP System

Parameter Plant 1 Plant 2 Plant 3 Plant 4 Plant 5

Eldorado Barren Barren
Location Azusa, CA Lockhart, CA
Valley, NV Ridge, CA Ridge, CA
𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖 (MW) 210 250 570 90 205
𝑃𝑃𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 (MW) 137 163 371 59 134
𝑈𝑈𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 (MWh) 546 650 1,482 234 535
𝑚𝑚𝑚𝑚𝑚𝑚
𝑃𝑃𝐷𝐷𝐷𝐷 ILR 1.3 (MW) 273 325 741 117 266
𝑚𝑚𝑚𝑚𝑚𝑚
𝑃𝑃𝐷𝐷𝐷𝐷 ILR 1.4 (MW) 294 350 798 126 287
𝑚𝑚𝑚𝑚𝑚𝑚
𝑃𝑃𝐷𝐷𝐷𝐷 ILR 1.5 (MW) 315 375 855 135 307
𝑚𝑚𝑚𝑚𝑚𝑚
𝑃𝑃𝐷𝐷𝐷𝐷 ILR 1.6 (MW) 336 400 912 144 328
𝑚𝑚𝑚𝑚𝑚𝑚
𝑃𝑃𝐷𝐷𝐷𝐷 ILR 1.7 (MW) 357 425 969 153 348
𝑚𝑚𝑚𝑚𝑚𝑚
𝑃𝑃𝐷𝐷𝐷𝐷 ILR 1.8 (MW) 378 450 1,026 162 369
𝑚𝑚𝑚𝑚𝑚𝑚
𝑃𝑃𝐷𝐷𝐷𝐷 ILR 1.9 (MW) 399 475 1,083 171 390
𝑚𝑚𝑚𝑚𝑚𝑚
𝑃𝑃𝐷𝐷𝐷𝐷 ILR 2.0 (MW) 420 500 1,140 180 410

All simulations are carried out using PLEXOS Version 8.2 using the Xpress-MP solver. The
2020 scenario is chosen as the simulation horizon with one-hour resolution and one-day look-
𝑡𝑡
ahead. All the DC profiles (𝑃𝑃𝐷𝐷𝐷𝐷 ) for the PV-BES are developed using the National Renewable
Energy Laboratory’s System Advisor Model (SAM) (Freeman et al. 2018) based on 2012
weather data.

All renewable energy resources are assumed to supply energy at zero cost. We use variable costs
of 0.1 $/MWh for the battery to avoid degeneracies (related to PV versus battery dispatch) and
unrealistic battery cycling. The relatively low value assumed allows for the optimization to
dispatch the battery as a resource without resulting in curtailment because of cost. These cost
components are not meant to reflect realistic variable costs for battery technologies; rather, they
are used to study the operation of the proposed PV-BES hybrid in a PCM. All other generators
have technology-specific operational constraints as well, based on the native PLEXOS
representation.

Operation of the DC-coupled PV-BES is demonstrated in the following subsections. In Section

4.2, we demonstrate the various modes of operation that are available for a PV-BES hybrid and
discuss the impacts of PV penetration on PV-BES operations and value. In Section 4.3, we

13
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
compare the operations for different PV-BES configurations, including AC-coupled vs. DC-
coupled architectures and the impact of increasing ILRs; the latter presentation includes how
operational strategies evolve with a growing amount of otherwise-clipped energy and
corresponding impacts on production costs.

4.2 Results: PV-BES Modes of Operation

Seven different operating modes are possible for a DC-coupled PV-BES plant. Figure 4 (page
15) illustrates six of these modes during two days of operation (July 12th and 13th) for Plant 3 in
the case with ILR=2.0. 6 For each hour interval, the left-side orange bars represent the available
PV-DC power; the right-side stacked bars represent any one of the seven operating modes for the
PV and/or battery component, including:

1. PV only to grid (dark red bar): This mode is typically observed in hours when the available
PV DC power (orange bars) is lower than the inverter rating (e.g., 5 a.m. on July 12th and
13th).
2. Battery only to grid (teal bar): This mode can be observed in evening hours when PV DC
power is zero and demand is fairly high (e.g., hours 20–22 [8 p.m.–10 p.m.] on July 12th).
3. PV and battery to grid (dark red and teal bars): This mode is observed when the available
solar energy is lower than the inverter rating and load is relatively high (e.g., hour 16 [4 p.m.]
on July 12th, and hour 18 [6 p.m.] on July 13th).
4. PV only to battery (yellow bar): This mode would be expected during times of lower
demand (e.g., spring), and it is not observed on July 12th or July 13th.
5. PV to grid and battery (dark red and yellow bars): This mode is common in peak solar
hours when the available PV DC power is above the inverter rating (e.g., hours 10–14 [10
a.m.–2 p.m.] on July 12th and hours 6-17 [6 a.m.–5 p.m.] on July 13th).
6. Grid to battery (purple bar): This mode is typically observed when no PV is available and
the system foresees a low PV day (e.g., hours 1–4 [1 a.m.–4 a.m.] on July 12th).
7. PV and grid to battery (yellow and purple bars): This mode is observed when the system
foresees lower PV DC power during high load times and charges the battery (e.g., hours 6–8
[6 a.m.–8 a.m.] on July 12th) in preparation for meeting load at a later period.

Figure 4 also indicates several constraints of the system. For example, the system can avoid
clipping when the DC output is greater than the inverter rating (570 MW in Figure 4) by storing
energy (see hours 7–15 on July 15). However, the amount of avoided clipping is restricted by the
size of the battery. PV energy is spilled in hours 11, 14, 15, and 16 on July 15, because the total
available PV DC power on this day is greater than the total power capacity of the battery and
inverter.

Note that the specific daytime hours during which PV energy is spilled is somewhat arbitrary,
due to both the perfect foresight (or “forecastability”) associated with the optimization and
degeneracies in the solution. In particular, the model “knows” that (a) it can achieve a near-100%
SoC via the use of otherwise clipped energy and (b) a non-zero amount of spilled energy is
required (based on the PV-BES energy balance constraints). So, its charging behavior can appear

6
See Table 3 for details about the sizing of this specific setup for Plant 3.

14
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
somewhat erratic as a result of modest changes in the nodal price during the day, whereas similar
overall results would be produced if the spilling occurred during different daytime hours.

Figure 4. Hourly operation of DC-coupled PV-BES Plant 3 with ILR=2.0 on July 12 and 13
The left-side Orange bars represent the available PV-DC power, and the right-side
stacked-bars represent an operating mode of either PV or battery.

Finally, Figure 5 presents the grid charging patterns for a single day (July 11, 2020) from the 1/2
PV Scenario. This scenario is designed to explore different levels of grid charging, which is
sensitive to the level of PV penetration; in other words, increasing PV penetration on the system
more broadly tends to depress energy prices during the daytime hours, thus incentivizing the PV-
BES hybrids to charge from local PV during those hours.

Comparison of panels in Figure 5 reveals that the dominant impact of PV penetration is to shift
the timing (and source) of charging for the battery component and, to a lesser extent, its
discharging. In particular, if we assume a significantly lower penetration of PV (left panel), then
battery charging occurs in the non-solar hours. This shift reflects the interactions with energy
prices, as charging the battery overnight (hours 0–6) corresponds to a period when demand on
the system is low, such that additional charging demand does not increase the production cost
significantly (based on Equations (1) and (2)). Some of this stored energy is then used during a
midday drop in solar irradiation (hours 14–16), which results in a lower production cost.
However, in both cases, most of the battery discharging occurs during the evening hours, when
demand ramps up and solar energy reduces.

15
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 Figure 5. Hourly operation of DC-coupled PV-BES Plant 3 in ILR 1.3 on July 11, 2020 in the 1/2 PV
Scenario (left) and ILR=1.3 scenarios

4.3 Results: Impact of Configuration

As previously stated, PV-BES configuration is defined by both the nature of coupling between
and relative sizing of the PV and battery components. Direct comparison between coupling types
is most meaningful when the PV DC rating is fixed at a level that is appropriate for an AC-
coupled system. However, for an ILR of 1.3, there is limited opportunity for the DC-coupled
system to realize its primary benefit (recovering otherwise-clipped generation) over the AC-
coupled system. 7 Table 4 presents a comparison of results for AC-coupled and DC-coupled PV-
BES with an ILR of 1.3, which reveals a difference in production cost of just 0.1% (due to
utilizing the modest amount of clipped energy and a slightly more efficient charging efficiency).
Table 4. Summary of annual parameters for AC coupled, and DC coupled system at ILR 1.3

Parameter AC-Coupled DC-Coupled

Production cost $368,140,000 $367,810,000
PV generation 19,600 GWh 19,000 GWh
Battery discharge 4,950 GWh 4,990 GWh
PV DC charging 0 730 GWh

The remainder of this section explores how higher ILRs will change the operation of the battery
and the resulting benefits to the system. In particular, the ILR associated with a DC-coupled PV-

7
Based on 2012 weather year data and an ILR of 1.3, only 0.4% of the total annual available DC energy produced
by the PV array is clipped (i.e., it exceeds the inverter rating).

16
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
BES influences the amount of energy produced by the DC modules, which influences both
operational strategies and production cost over the course of the simulation period (one year). To
demonstrate this effect, Figure 6 presents the aggregated discharge energy of the battery
component (bars) throughout the year, which is overlaid with the SoC distributions for the
intervals during which the battery component discharges. For the sake of clarity, Figure 6 only
presents the details for the ILR=1.3 and ILR=2.0 DC-coupled PV-BES, because the values of the
other configurations follow a similar pattern and can be understood by interpolating between the
presented results.

Looking across the time intervals presented in Figure 6, it is apparent that most of the battery
discharge for the PV-BES units happens during the evening hours, when the solar irradiance
decreases and energy demand increases. In the absence of storage units, gas generators are often
used during this period. However, in our scenarios, energy from the battery component of the
PV-BES hybrid is utilized instead, which reduces the need for expensive generators and, in turn,
reduces the system-wide production cost.

Figure 6. Battery discharge behavior for the battery component of the DC-coupled PV-BES hybrids
with ILR 1.3 (pink) and 2.0 (green), aggregated over the 1-year analysis period
The bar plot represents the aggregated hourly battery generation; the dotted lines represent the mean of the SoC for
the intervals during which the battery component discharges, and the shaded area is the 95% confidence range of
this price distribution.

Also depicted in Figure 6 are the different SoC paths and operational strategies for the battery
component in PV-BES hybrids with ILR=1.3 (pink) and ILR=2.0 (green). The green dashed line
and narrow distribution around it indicates that the SoC for a high-ILR system follows a similar
path throughout the year, such that the battery component charges from the otherwise-clipped
energy throughout (and almost exclusively) during the daytime hours. Consequently, the battery
component’s SoC increases steadily throughout the day and enters the evening period with an
SoC that is near-100% and consistently higher than PV-BES hybrids with ILR=1.3. The SoC

17
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
path for the PV-BES hybrid units with ILR=1.3 also differs more throughout the year (as
demonstrated by the wider range around the pink dashed line), although the trends are consistent
and similar to those observed for the higher ILR hybrid units.

Looking across all hours of the year and configurations, the utilization of the PV-BES hybrid’s
inverter can be evaluated through a variety of aggregate metrics:

• Capacity factor is defined as the total amount of energy produced by the plant divided by
annual energy at full output rating of the AC inverter. This is essentially the same as the
capacity factor of a conventional power plant.

• The grid charging factor is the ratio of the energy drawn from the grid (AC-to-DC) for
charging the battery to the annual energy at full output rating of the AC inverter.

• The utilization factor is the ratio of the total energy passed through the inverter (AC-to-
DC and DC-to-AC) to the annual energy at full output rating of the AC inverter (sum of
the capacity factor and grid charging factor).

The observed values for these three metrics (Table 5) demonstrate increasing utilization of the
PV-BES inverter with increasing ILR. The higher capacity factor with increasing ILR is a result
of the recovery of otherwise-clipped (or spilled) energy, which is recovered by the DC-coupled
battery and dispatched later, thus increasing the overall energy output. The same drivers
contribute to an increasing utilization factor with increasing ILR, but this increase is partially
offset by corresponding (and expected) decreases in the grid charging factor. In particular, the
grid charging factor declines from 8.1% with ILR=1.3 to 2.1% with ILR=2.0, reflecting the
reduction in grid charging when the PV-BES hybrid has excess DC energy available from the
oversized PV component. The modest amount of grid charging with the highest ILRs reflects
periods of low irradiance.

Table 5. Utilization Metrics for the DC-Coupled Hybrid-Inverter

Capacity Grid Charging Utilization

ILR
Factor (%) Factor (%) Factor (%)
1.3 41.3 8.1 49.4
1.4 43.3 7.5 50.8
1.5 45.2 6.8 51.9
1.6 46.7 5.7 52.4
1.7 48.1 4.7 52.8
1.8 49.1 3.6 52.7
1.9 50.1 2.8 52.8
2.0 50.8 2.1 52.9

Figure 7 represents the destination of all PV DC energy collected over the course of the year for
all the simulated PV-BES hybrids as a function of ILR. In general, PV DC energy can be
categorized as being:

18
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 • Utilized (for direct electricity generation [brown] or charging the battery [yellow])
• Unutilized (due to losses associated with the various conversions [red] or spilled [grey]).
In addition to demonstrating the growing availability of PV DC energy with increasing ILR, the
breakdown of utilized PV DC energy in Figure 7 indicates that most is sent directly to the grid
and 15%–25% is used to charge the local battery.

Figure 7. Destination of PV DC energy collected as a function of ILR for all DC-coupled PV-BES
hybrids in this demonstration
The stacked sum of these bars represents the total DC energy available for each ILR.

The DC energy that cannot be utilized for either electricity generation or battery charging is
spilled (wasted), which becomes more common in the higher ILR (1.8 and above) systems
(Figure 8). For our assumed PV-BES setup, most of the spilled energy occurs because of
insufficient battery energy capacity (for recovering the clipped PV energy). However, the
converter limit is reached when the clipped DC power is greater than the converter capacity,
which is observed during a few intervals in the PV-BES hybrids with ILR=2.0. A significant
share of inverter losses come from PV generation, and from battery discharge as ILR increases.
The increase in converter and battery losses is also observed as the we reach higher ILR.

19
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 Figure 8. The distribution of the unutilized PV DC energy due to losses and spillage
Finally, due to the increasing availability of DC energy with higher ILRs, the production cost of
the system will decline as the utilization of more otherwise-clipped, zero-cost DC energy
displaces the need for higher-cost generation. Figure 9 illustrates this result by showing how the
annual production costs for the entire LADWP system (green bars) decline as the assumed ILR
of the PV-BES units increases.

Figure 9 also shows the incremental value of the additional PV energy produced with higher
ILRs (blue bars), which is evaluated as the ratio of the change in production cost to the change in
PV utilization, using the results for ILR=1.3 as the baseline:
𝑃𝑃𝐶𝐶 −𝑃𝑃𝐶𝐶
𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑃𝑃𝑃𝑃 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑒𝑒𝑖𝑖 = 𝑛𝑛𝑛𝑛𝑛𝑛 𝑃𝑃𝑉𝑉𝑖𝑖 −𝑛𝑛𝑛𝑛𝑛𝑛1.3𝑃𝑃𝑉𝑉 , ∀𝑖𝑖 = 1.4, 1.5 … 2.0 (13)
𝑖𝑖 1.3

where 𝑃𝑃𝐶𝐶𝑖𝑖 corresponds to the annual production cost of the system for ILR=𝑖𝑖, and the 𝑛𝑛𝑛𝑛𝑛𝑛 𝑃𝑃𝑉𝑉𝑖𝑖 is
the aggregated PV energy utilized in a year for the system with ILR=𝑖𝑖 (including PV DC energy
that is utilized for generation and charging the battery. This shows a general downward trend as a
greater fraction of stored energy is discharged during periods of lower value. For ILRs of 1.8 and
above, the additional PV value plateaus as the amount of spilled energy grows, due to limitations
imposed by the assumed battery size.

20
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 Figure 9. Annual production cost of the LADWP for each ILR and the additional value adding PV
brings to the system with respect to ILR=1.3

21
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
5 Conclusions and Future Work
In this paper, we present a method to represent a DC-coupled PV-BES in a commercial PCM,
with PLEXOS as the demonstrated test case. This method can be reproduced in any other PCM
software, and it is a critical step toward being able to evaluate the system benefits that DC-
coupled PV-BES can provide, particularly if it is deployed with higher ILRs in the future. The
proposed model in PLEXOS includes an explicit representation of both sides of the DC-coupled
PV-BES hybrid’s inverter (i.e., separate DC and AC “sides”), which is essential for both
representing the synergies of DC-coupled PV-BES hybrids and validating the model. This model
is developed such that the size of each component (PV, battery, converter, and inverter) of a DC-
coupled PV-BES can be varied and analyzed.

To demonstrate our model, we implemented hypothetical DC-coupled PV-BES hybrids on the

LADWP test system and evaluated the operational behavior and system costs across one year
(2020). This demonstration revealed that each possible flow among all the components of the
DC-coupled PV-BES can be observed, with select days combining multiple modes of operation
that demonstrate the constraints and efficiencies. In addition to observing specific days, an
aggregated analysis of the model demonstrated the capabilities of a PV-BES. The growing
availability of PV DC energy with increasing ILR resulted in (a) an increase in the utilization of
the inverter (e.g., capacity factor), (b) a reduction in grid charging (in favor of charging from the
local PV, which is more-efficient and zero-cost), and (c) a decrease in system-wide production
cost.

Overall, we believe this model will serve the research community with a comprehensive
approach to representing DC-coupled PV-BES in a PCM in order to evaluate the system-level
benefits of such systems. Potential valuable future work includes the following.

• This work demonstrated the operational behavior of DC-coupled PV-BES, whose

contributions to the grid mix inherently reduce system-wide production cost (particularly
for higher ILR systems). Translating this reduced production cost into economic benefit
requires comparing the production cost savings against the corresponding PV-BES
investment costs, which are not considered in this initial demonstration.
• To enable the more detailed treatment of ILR variations, this work explored a single
storage technology with fixed operational characteristics (e.g., round-trip efficiency) and
sizing. Future analysis with this model architecture could explore the effect of increasing
battery size (energy and capacity), including interactions with different ILRs (for the
coupled PV component) and system-wide production costs. Also, a more detailed
representation of when battery-related losses occur (e.g., attributing specific shares of the
round-trip efficiency to the charging and discharging periods) might improve the
precision of this modeling. Finally, different storage and charging technologies could also
be evaluated to understand the ideal configurations and technology combinations for
various power system applications (e.g., regulation), accounting for the potential for
charging speed restrictions.
• This work explored interactions of nodal energy prices and the DC-coupled PV-BES
operations. Additional model development is needed to enable an evaluation of the
capacity and reserve capabilities of this technology, within the context of the price
dynamics that are captured in a PCM.

22
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 • The present PCM modeling demonstrates the value that DC-coupled PV-BES could
provide to a vertically integrated utility. Estimating the revenue associated with such
systems requires combining multiple model techniques, such as estimating revenues from
PCM results or employing a combination of PCM-generated prices with price-taker
modeling of the corresponding DC-coupled PV-BES configuration (Schleifer et al. 2021).
Such an evaluation would help inform investment decisions for potential PV-BES plant
developers and owners in competitive market regions.

23
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
References
Bullich-Massagué, Eduard, Francisco-Javier Cifuentes-García, Ignacio Glenny-Crende, Marc
Cheah-Mañé, Mònica Aragüés-Peñalba, Francisco Díaz-González, and Oriol Gomis-
Bellmunt. 2020. “A Review of Energy Storage Technologies for Large Scale
Photovoltaic Power Plants.” Applied Energy 274 (September): 115213.
https://doi.org/10.1016/j.apenergy.2020.115213.
Chen, Y., A. Q. Huang, and X. Yu. 2013. “A High Step-Up Three-Port DC–DC Converter for
Stand-Alone PV/Battery Power Systems.” IEEE Transactions on Power Electronics 28
(11): 5049–62. https://doi.org/10.1109/TPEL.2013.2242491.
Cochran, Jaquelin, and Paul Denholm. 2021. “The Los Angeles 100% Renewable Energy
Study.” NREL/TP-6A20-79444. https://maps.nrel.gov/la100/.
Cole, Wesley, Sean Corcoran, Nathaniel Gates, Daniel Mai, Trieu, and Paritosh Das. 2020.
“2020 Standard Scenarios Report: A U.S. Electricity Sector Outlook.” NREL/TP-6A20-
77442. Golden, CO: National Renewable Energy Laboratory.
https://www.nrel.gov/docs/fy21osti/77442.pdf.
DiOrio, Nicholas, Paul Denholm, and William B. Hobbs. 2020. “A Model for Evaluating the
Configuration and Dispatch of PV plus Battery Power Plants.” Applied Energy 262
(March): 114465. https://doi.org/10.1016/j.apenergy.2019.114465.
EIA. 2020. “Annual Energy Outlook 2020.” Washington, D.C.: U.S. DOE Energy Information
Administration. https://www.eia.gov/outlooks/aeo/pdf/AEO2020%20Full%20Report.pdf.
Elgqvist, Emma, Kate Anderson, and Edward Settle. 2018. “Federal Tax Incentives for Energy
Storage Systems.” National Renewable Energy Laboratory.
https://www.nrel.gov/docs/fy18osti/70384.pdf.
Freeman, Janine M., Nicholas A. DiOrio, Nathan J. Blair, Ty W. Neises, Michael J. Wagner,
Paul Gilman, and Steven Janzou. 2018. “System Advisor Model (SAM) General
Description (Version 2017.9.5).” NREL/TP-6A20-70414. National Renewable Energy
Lab. (NREL), Golden, CO (United States). https://doi.org/10.2172/1440404.
Fu, Ran, Timothy Remo, and Robert Margolis. 2018. “Evaluating the Cost Benefits of U.S.
Utility-Scale Photovoltaics Plus Energy Storage Systems.” In 2018 IEEE 7th World
Conference on Photovoltaic Energy Conversion (WCPEC) (A Joint Conference of 45th
IEEE PVSC, 28th PVSEC 34th EU PVSEC), 1–4.
https://doi.org/10.1109/PVSC.2018.8547852.
Gilman, Paul, Nathan Blair, M. Mehos, Craig Christensen, Steve Janzou, and Chris Cameron.
2008. “Solar Advisor Model User Guide for Version 2.0,” January.
https://doi.org/10.2172/937349.
Good, Jeremy, and Jeremiah Johnson. 2016. “Impact of Inverter Loading Ratio on Solar
Photovoltaic System Performance.” Applied Energy 177 (September): 475–86.
https://doi.org/10.1016/j.apenergy.2016.05.134.
Gorman, Will, Andrew Mills, Mark Bolinger, Ryan Wiser, Nikita G. Singhal, Erik Ela, and Eric
O’Shaughnessy. 2020. “Motivations and Options for Deploying Hybrid Generator-plus-
Battery Projects within the Bulk Power System.” The Electricity Journal 33 (5): 106739.
https://doi.org/10.1016/j.tej.2020.106739.
Hirth, Lion. 2013. “The Market Value of Variable Renewables.” Energy Economics 38: 218–36.
http://dx.doi.org/10.1016/j.eneco.2013.02.004.

24
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
———. 2015. “The Optimal Share of Variable Renewables:How the Variability of Wind and
Solar Power Affects TheirWelfare-Optimal Deployment.” The Energy Journal 36 (1):
127–62. http://dx.doi.org/10.5547/01956574.36.1.5.
Hledik, Ryan, Roger Leuken, Judy Chang, Hannes Pfeifenberger, Jesse Cohen, and John Imon
Pedtke. 2019. “Solar-Plus-Storage: The Future Market for Hybrid Resource.” December.
https://brattlefiles.blob.core.windows.net/files/17741_solar_plus_storage_economics_-
_final.pdf.
Hu, Jiefeng, Yinliang Xu, Ka Wai Cheng, and Josep M. Guerrero. 2018. “A Model Predictive
Control Strategy of PV-Battery Microgrid under Variable Power Generations and Load
Conditions.” Applied Energy 221 (July): 195–203.
https://doi.org/10.1016/j.apenergy.2018.03.085.
Jorgenson, Jennie, Paul Denholm, Mark Mehos, and Craig Turchi. 2013. “Estimating the
Performance and Economic Value of Multiple Concentrating Solar Power Technologies
in a Production Cost Model.” NREL/TP-6A20-58645. National Renewable Energy Lab.
(NREL), Golden, CO (United States). https://doi.org/10.2172/1260920.
Jorgenson, Jennie, Matthew O’Connell, Paul Denholm, Janna Martinek, and Mark Mehos. 2018.
“A Guide to Implementing Concentrating Solar Power in Production Cost Models.”
NREL/TP-6A20-68527. National Renewable Energy Lab. (NREL), Golden, CO (United
States). https://www.nrel.gov/docs/fy19osti/68527.pdf.
King, David, Sigifredo Gonzalez, Gary Galbraith, and William Boyson. 2007. “Performance
Model for Grid-Connected Photovoltaic Inverters.” SAND2007-5036. Sandia National
Laboratory. https://energy.sandia.gov/wp-content/gallery/uploads/Performance-Model-
for-Grid-Connected-Photovoltaic-Inverters.pdf.
Mai, Trieu, Wesley Cole, and Andrew Reimers. 2019. “Setting Cost Targets for Zero-Emission
Electricity Generation Technologies.” Applied Energy 250 (September): 582–92.
https://doi.org/10.1016/j.apenergy.2019.05.001.
Martinek, Janna, Jennie Jorgenson, Mark Mehos, and Paul Denholm. 2018. “A Comparison of
Price-Taker and Production Cost Models for Determining System Value, Revenue, and
Scheduling of Concentrating Solar Power Plants.” Applied Energy 231 (December): 854–
65. https://doi.org/10.1016/j.apenergy.2018.09.136.
Martins Deschamps, Eduardo, and Ricardo Rüther. 2019. “Optimization of Inverter Loading
Ratio for Grid Connected Photovoltaic Systems.” Solar Energy 179 (February): 106–18.
https://doi.org/10.1016/j.solener.2018.12.051.
Mehos, M., J. Jorgenson, P. Denholm, and C. Turchi. 2015. “An Assessment of the Net Value of
CSP Systems Integrated with Thermal Energy Storage.” Energy Procedia, International
Conference on Concentrating Solar Power and Chemical Energy Systems, SolarPACES
2014, 69 (May): 2060–71. https://doi.org/10.1016/j.egypro.2015.03.219.
Mills, Andrew, and Pía Rodriguez. 2020. “A Simple and Fast Algorithm for Estimating the
Capacity Credit of Solar and Storage.” Energy 210.
https://doi.org/10.1016/j.energy.2020.118587.
Mills, Andrew, and Ryan Wiser. 2012. “Changes in the Economic Value of Variable Generation
at High Penetration Levels: A Pilot Case Study of California.” LBNL-5445E. Ernest
Orlando Lawrence Berkeley National Laboratory. https://emp.lbl.gov/sites/all/files/lbnl-
5445e.pdf.
Miñambres-Marcos, Víctor Manuel, Miguel Ángel Guerrero-Martínez, Fermín Barrero-
González, and María Isabel Milanés-Montero. 2017. “A Grid Connected Photovoltaic

25
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
 Inverter with Battery-Supercapacitor Hybrid Energy Storage.” Sensors 17 (8): 1856.
https://doi.org/10.3390/s17081856.
Mongird, Kendall, Vilayanur Viswanathan, Jan Alam, Charlie Vartanian, Vincent Sprenkle, and
Richard Baxter. 2020. “2020 Grid Energy Storage Technology Cost and Performance
Assessment.” DOE/PA-0204. U.S. Department of Energy.
https://www.pnnl.gov/sites/default/files/media/file/Final%20-
%20ESGC%20Cost%20Performance%20Report%2012-11-2020.pdf.
Schimpe, Michael, Maik Naumann, Nam Truong, Holger C. Hesse, Shriram Santhanagopalan,
Aron Saxon, and Andreas Jossen. 2018. “Energy Efficiency Evaluation of a Stationary
Lithium-Ion Battery Container Storage System via Electro-Thermal Modeling and
Detailed Component Analysis.” Applied Energy 210 (January): 211–29.
https://doi.org/10.1016/j.apenergy.2017.10.129.
Schleifer, Anna H., Caitlin A. Murphy, Wesley J. Cole, and Paul L. Denholm. 2021. “The
Evolving Energy and Capacity Values of Utilierty-Scale PV-plus-Battery Hybrid System
Architectures.” Advances in Applied Energy, March, 100015.
https://doi.org/10.1016/j.adapen.2021.100015.
Wang, G., M. Ciobotaru, and V. G. Agelidis. 2014. “Power Smoothing of Large Solar PV Plant
Using Hybrid Energy Storage.” IEEE Transactions on Sustainable Energy 5 (3): 834–42.
https://doi.org/10.1109/TSTE.2014.2305433.
Wiser, Ryan, Mark Bolinger, Will Gorman, Joe Rand, Seongeun Jeong, Joachim Seel, Cody
Warner, and Ben Paulos. 2020. “Hybrid Power Plants: Status of Installed and Proposed
Projects.” 34197. Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United
States). https://emp.lbl.gov/publications/hybrid-power-plants-status-installed.
Wiser, Ryan, Andrew Mills, Joachim Seel, Todd Levin, and Audun Botterud. 2017. “Impacts of
Variable Renewable Energy on Bulk Power System Assets, Pricing, and Costs.”
Lawrence Berkely National Laboratory and Argonne National Laboratory.
https://emp.lbl.gov/sites/default/files/lbnl_anl_impacts_of_variable_renewable_energy_fi
nal.pdf.

26
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
Appendix: Transmission-Line Inverter Topology
This appendix presents an alternative to the model presented in Section 4 and explored
throughout this report. In this alternative approach, a constrained transmission line represents
the hybrid-inverter, as depicted in Figure A-1. Therefore, we refer to this approach as the
transmission-line inverter topology.

The implementation starts with the addition of a new node to the system, which (a) represents the
point of common connection of the PV object and battery object, and (b) enables an explicit
representation of the DC-side on which the hybrid components are coupled. The inverter is
replaced by a lossless transmission line, for which the line rating property is set to equal to the
inverter rating (𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖 ) and the line efficiency property is set to equal to the inverter efficiency
(𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 ). The PV is represented by a PLEXOS generator object with the rating property (𝑃𝑃𝑃𝑃𝑃𝑃 )
𝑡𝑡 ),
pointing toward an external file containing the time-series PV-DC profile (𝑃𝑃𝐷𝐷𝐷𝐷 which is
defined by the hybrid’s location and size.

Figure A-1. PLEXOS implementation of DC-coupled PV-BES using a transmission line-

inverter topology
In this approach, a transmission line represents the shared inverter in a DC-coupled PV-BES hybrid.

The battery object in PLEXOS includes an inherent representation of the battery inverter. When
using this object in this model, we assume the inverter to be a DC-DC converter. The battery
energy (𝑊𝑊𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ) will remain the same, and the rating property (𝑃𝑃𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ) will be equal to the rating
𝐷𝐷𝐷𝐷
of the converter (𝑃𝑃𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ). The efficiency of the battery (𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ) is the product of the square of
2
converter efficiency (𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 )—which represents the round-trip efficiency (i.e., for charge and
discharge cycle) of the battery-converter combination—and the battery internal efficiency
(𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ), as described in Equation (14). The internal battery losses account for the losses that
occur during the conversion from chemical to electrical energy (and vice-versa).
𝐷𝐷𝐷𝐷 2
𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑡𝑡 = 𝜂𝜂𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝜂𝜂𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 (14)

27
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.
Because the PLEXOS generator and battery objects generate AC power, a transmission line that
links the new node (DC) and the system emulates the function of the shared inverter. The line
parameters are set based on the inverter ratings. The line rating (flow limit) is set to equal to the
inverter rating (𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖 ). The line efficiency property is set to equal to the weighted inverter
efficiency(𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 ). It must be noted that in PLEXOS, this is a dual property (flow and backflow),
so the value must be set for both directions of flow.

Using this model for PCM simulations in PLEXOS requires some additional options to be
enabled in order for it to function as intended:

• Line losses must be enforced so that the transmission line connecting the DC node to the
system emulates an inverter.
• Line limits must be enforced so that the line rating represents the inverter limit.
• The rating file for the PV generator must belong to the DC-rating of the PV setup.
Although this is an issue that is specific to PLEXOS, some of the flows that include the converter
cannot be monitored as the converter is considered as an internal component of the battery.
Because of the added node and transmission line, the transmission-line inverter topology can
increase the complexity of the problem, especially when a large number of PV-BES are added to
the system.

28
This report is available at no cost from the National Renewable Energy Laboratory at www.nrel.gov/publications.