David Fork
David Fork is a Renewable Energy Technologist at Google where he did solar receiver assessment on the Renewable Energy Cheaper than Coal project and intelligent inverter control on the Bottom up Grid (BUG) projects. He also designed the Google Bike. He currently serves within Google’s Climate and Energy Team. Before Google he was a Principal Scientist at the Palo Alto Research Center (PARC) where he was instrumental in several start up ventures, including two in solar energy. He graduated Summa Cum Laude from the University of Rochester in 1987 with degrees in Physics and Electrical Engineering. He completed his Ph.D. from Stanford University in Applied Physics in 1991.
His work on electronic materials and devices includes complex oxide epitaxial thin films, laser crystallized display materials, organic electroluminescent devices, semiconductor LEDs and lasers, electronic imaging systems, micro-electromechanical systems and photovoltaic devices. His work at Google has covered topics including solar energy, power conversion, fusion, low carbon fuels, and remote sensing. Dr. Fork holds over 80 issued US patents and has authored over 100 publications.
Research Areas
Authored Publications
Sort By
General Geospatial Inference with a Population Dynamics Foundation Model
Chaitanya Kamath
Prithul Sarker
Joydeep Paul
Yael Mayer
Sheila de Guia
Jamie McPike
Adam Boulanger
David Schottlander
Yao Xiao
Manjit Chakravarthy Manukonda
Monica Bharel
Von Nguyen
Luke Barrington
Niv Efron
Krish Eswaran
Shravya Shetty
(2024) (to appear)
Preview abstract
Supporting the health and well-being of dynamic populations around the world requires governmental agencies, organizations, and researchers to understand and reason over complex relationships between human behavior and local contexts. This support includes identifying populations at elevated risk and gauging where to target limited aid resources. Traditional approaches to these classes of problems often entail developing manually curated, task-specific features and models to represent human behavior and the natural and built environment, which can be challenging to adapt to new, or even related tasks. To address this, we introduce the Population Dynamics Foundation Model (PDFM), which aims to capture the relationships between diverse data modalities and is applicable to a broad range of geospatial tasks. We first construct a geo-indexed dataset for postal codes and counties across the United States, capturing rich aggregated information on human behavior from maps, busyness, and aggregated search trends, and environmental factors such as weather and air quality. We then model this data and the complex relationships between locations using a graph neural network, producing embeddings that can be adapted to a wide range of downstream tasks using relatively simple models. We evaluate the effectiveness of our approach by benchmarking it on 27 downstream tasks spanning three distinct domains: health indicators, socioeconomic factors, and environmental measurements. The approach achieves state-of-the-art performance on geospatial interpolation across all tasks, surpassing existing satellite and geotagged image based location encoders. In addition, it achieves state-of-the-art performance in extrapolation and super-resolution for 25 of the 27 tasks. We also show that the PDFM can be combined with a state-of-the-art forecasting foundation model, TimesFM, to predict unemployment and poverty, achieving performance that surpasses fully supervised forecasting. The full set of embeddings and sample code are publicly available for researchers. In conclusion, we have demonstrated a general purpose approach to geospatial modeling tasks critical to understanding population dynamics by leveraging a rich set of complementary globally available datasets that can be readily adapted to previously unseen machine learning tasks.
View details
Preview abstract
Engineers: You Can Disrupt Climate Change
Illustrates the scale and nature of challenges, and opportunities for engineers to tackle them.
View details
A human-labeled Landsat contrails dataset
Vincent Rudolf Meijer
Erica Wickstrom Brand
Carl Elkin
ICML workshop on Climate Change 2021 (2021)
Preview abstract
Contrails (condensation trails) are the ice clouds that trail behind aircraft as they fly through cold and moist regions of the atmosphere. Avoiding these regions could potentially be an inexpensive way to reduce over half of aviation's impact on global warming. Development and evaluation of these avoidance strategies greatly benefits from the ability to detect contrails on satellite imagery. Since little to no public data is available to develop such contrail detectors, we construct and release a dataset of several thousand Landsat-8 scenes with pixel-level annotations of contrails. The dataset will continue to grow, but currently contains 3431 scenes (of which 47\% have at least one contrail) representing 800+ person-hours of labeling time.
View details
Revisiting the cold case of cold fusion
Curtis Berlinguette
Jeremy Munday
Matt Trevithick
Thomas Schenkel
Yet-Ming Chiang
Nature (2019)
Preview abstract
The 1989 claim of ‘cold fusion’ was publicly heralded as the future of clean energy generation. However, subsequent failures to reproduce the effect heightened scepticism of this claim in the academic community, and effectively led to the disqualification of the subject from further study. Motivated by the possibility that such judgement might have been premature, we embarked on a multi-institution programme to re-evaluate cold fusion to a high standard of scientific rigour. Here we describe our efforts, which have yet to yield any evidence of such an effect. Nonetheless, a by-product of our investigations has been to provide new insights into highly hydrided metals and low-energy nuclear reactions, and we contend that there remains much interesting science to be done in this underexplored parameter space.
View details
Calorimetry under non-ideal conditions using system identification
B. P. MacLeod
B. Lam
C. P. Berlinguette
Journal of Thermal Analysis and Calorimetry, 138 (2019), pp. 3139-3157
Preview abstract
We report a model-based method for quantifying heat flow and storage in thermal systems using data from multiple thermal sensors. This approach avoids stringent requirements on the system geometry and sensor positions and enables calorimetry to be performed under a broader range of circumstances than is accessible with existing calorimeters, such as when nonlinear heat transfer occurs, when spatially separated heat sources are active or when multiple thermal masses partic- ipate. Using experimental data from a model thermal system, this paper provides a tutorial on the construction of nonlinear lumped-element heat transfer models and the use of system identification to estimate the parameters of these models from calibration data. The calibrated models are then used to estimate unknown energy inputs to the thermal system from sensor data. Our best model enabled the measurement of the total input energy with 0.02% accuracy; the instantaneous input power could be measured with a root-mean-square error of 10% of the average input power.
View details
Accurate Coulometric Quantification of Hydrogen Absorption in Palladium Nanoparticles and Thin Films
Rebecca Sherbo
Marta Moreno
Noah Johnson
David Dvorak
Curtis Berlinguette
Chemistry of Materials (2018)
Preview abstract
We report here an electrochemical method for the precise and accurate quantification of hydrogen absorption into palladium materials with the use of an electrochemical flow cell. Chronocoulometry is used to quantify the desorption of hydrogen absorbed in palladium with the electrolyte flowing past the sample to prevent the concurrent hydrogen oxidation reaction (HOR). The HOR can render this coulometric technique inaccurate is not accounted for properly. Performing the electrochemistry measurements in a flow cell improves the accuracy of quantifying the amount of hydrogen absorbed with much higher levels of reproducibility. This absorption quantification technique represents an easily accessible and fast method that can be used for thin films and nanoparticle samples.
View details
High-temperature high-pressure calorimeter for studying gram-scale heterogeneous chemical reactions
B. P. MacLeod
P. A. Schauer
K. Hu
B. Lam
C. P. Berlinguette
Reviews of Scientific Instruments, 88 (2017), pp. 084101
Preview abstract
We present an instrument for measuring pressure changes and heat flows of physical and chemical processes occurring in gram-scale solid samples under high pressures of reactive gases. Operation is demonstrated at 1232 °C under 33 bars of pure hydrogen. Calorimetric heat flow is inferred using a grey-box non-linear lumped-element heat transfer model of the instrument. Using an electrical calibration heater to deliver 900 J/1 W pulses at the sample position, we demonstrate a dynamic calorimetric power resolution of 50 mW when an 80-s moving average is applied to the signal. Integration of the power signal showed that the 900 J pulse energy could be measured with an average accuracy of 6.35% or better over the temperature range 150-1100 °C. This instrument is appropriate for the study of high-temperature metal hydride materials for thermochemical energy storage.
View details
PowerNet for distributed Energy Resources
Anand Ramesh
Sangsun Kim
Jim Schmalzried
Jyoti Sastry
Michael Dikovsky {{+mdikovsky
Konstantin Bozhkov
Eduardo Pinheiro
Scott Collyer
Ankit Somani
Ram Rajagopal
Arun Majumdar
Junjie Qin
Gustavo Cezar
Juan Rivas
Abbas El Gamal
Dian Gruenich
Steven Chu
Sila Kiliccote
Conference: 2016 IEEE Power and Energy Society General Meeting (PESGM), IEEE Power and Energy Society, Boston, MA, USA (2016)
Preview abstract
We propose Powernet as an end-to-end open source technology for economically efficient, scalable and secure coordination of grid resources. It offers integrated hardware and software solutions that are judiciously divided between local embedded sensing, computing and control, which are networked with cloud-based high-level coordination for real-time optimal operations of not only centralized but also millions of distributed resources of various types. Our goal is to enable penetration of 50% or higher of intermittent renewables while minimizing the cost and address security and economical scalability challenges. In this paper we describe the basic concept behind Powernet and illustrate some components of the solution.
View details
Optimal trajectory control for parallel single phase H-bridge inverters
Seungil You
Decision and Control (CDC), 2015 IEEE 54th Annual Conference on, IEEE, pp. 1983 - 1990
Preview abstract
We describe a novel inverter control method that solves an optimization problem during each switching interval to closely follow a virtual impedance control law.
We report droop behavior over a wide range of applied loads and power sharing among multiple inverters.
View details
What It Would Really Take to Reverse Climate Change
IEEE Spectrum December 2014, IEEE, 3 Park Ave, 17th floor, New York, NY 10016-5997, pp. 30-35
Preview abstract
What two Googlers learned from a failed attempt to find the renewable energy source of tomorrow.
View details
