OSESMO

Logo

Inspiration

OSESMO was born out of the California Public Commission's SGIP Working Group, which is evaluating options to ensure that behind-the-meter storage and solar-plus-storage systems are reducing GHG emissions. This is a project I've been working on for about three months.

What It Does

OSESMO is an optimal-dispatch storage simulation model that calculates the maximum achievable customer savings possible when performing time-of-use energy charge arbitrage, demand charge management, and solar self-consumption. It is also capable of co-optimizing for GHG emissions reduction.

How We Built It

The prototype version of OSESMO was built in MATLAB prior to SunCode. At SunCode, we focused on translating the code into Python to make it more accessible. Translating it to Python also could make it possible to create a web app interface with Flask or Django.

Challenges We Ran Into

Our group was fairly inexperienced with Python, and the prototype model is several thousand lines of code long. As a result, we were only able to translate about 30% of the model code. We also were not able to make progress on developing a web-app front-end, beyond a simple mockup created with R's Shiny package also developed prior to SunCode.

Accomplishments That We're Proud Of

We successfully translated all of the function scripts associated with importing load profile, PV production, retail rate, and emissions rate datasets, which represent a substantial portion of the codebase. We also researched optimization-package options for solving large, sparse linear programs, and began setting the problem up using the PuLP package.

What We Learned

This project served as a useful refresher on Python and algebraic optimization modeling.

What's Next for OSESMO

We hope to continue working on translating OSESMO to Python, with today's work session serving as a great first step. We are still very interested in creating a web-app interface and hosting it online. Please let us know if you would be interested in assisting with this effort - we would very much appreciate your help.

Built With

Submitted to

SunCode 2018

Created by

For the future objective of creating a web-frontend, tried to create a package that can be seamlessly integrated into Django Python web framework. The package would have been published at PyPI (https://pypi.python.org), for anybody to be able to download with "pip" command. But while testing the translated codes, found that code typed line by line accessing various source code files worked in Python interactive console up to optimization portion but the code from the top-level script file accessing the same various source code files didn't work with direct execution with "python" command, probably because we didn't fully made the translated Python code follow the Python convention due to the time constraint or because we shared the codes through flash drive that introduced older codes from one person mixed with newer codes from another person, I noticed. Regardless it's close to resolve the issue.

Other than that, mostly learned how optimization is done in the existing MatLab code (I would like to say, thank-you to Ryan) and learned about the industry the team mates are in (Thank you, all).

And I couldn't contribute as much as I wanted to due to time constraint.

It was disappointing that the final code was not pushed to the GitHub repository due to the last minute issue with "git push" (only the team lead had "write" access since the rest of the team participated only one day for this Hackathon) and the codes at GitHub that were available at the end of the Hackathon contained many typos and errors I already fixed in the final code.

Tadatoshi Takahashi
Optimization of Solar Photovoltaic (PV) Systems. Keen on Data Science. Strong in Engineering but not strong in finance
Ryan Mann
Modeler and data analyst focused on dispatchable distributed energy resources and demand flexibility.
Jeffrey Kwan
Prathamesh (Pratha) Pawar
Data Scientist in clean energy industry