David Ollodart’s Blog¶
Welcome to my website. My writings are on STEM, STEM education, STEM history, and technology for applications in science laboratories.
Motivation¶
The theory of chemical engineering was in bulk developed in the 19th century (thermodynamics, chemical kinetics, and continuum mechanics), with substantial additions in the early 20th (formalization of chemical kinetics into the science of transition state theory and then to chemical reaction engineering, formalization of continuum mechanics into the science of transport phenomena, advent of polymer science).
Improved experimentation is the modern challenge, especially interdisciplinary experiments which require innovative experimental apparatus.
The cumulative count of Nobel prizes awarded to analytical tools and techniques is steeply increasing:
1923: Chemistry for Micro-Analysis of Organic Substances
1931: Chemistry for Chemical High Pressure Methods
1943: Chemistry for Isotope Tracer Studies
1952: Chemistry for Partition Chromatography
1959: Chemistry for Polarographic Methods
1960: Chemistry for Carbon-14 Age Determination
1981: Physics for High-Resolution Electron Spectroscopy and Laser Spectroscopy
1982: Chemistry for Crystallographic Electron Microscopy
1986: Chemistry for Electron Microscopy Optics and Scanning Electron Microscopy
1986: Physics for Electron Optics and Scanning Tunneling Microscopy
1989: Physics for Atomic Clocks and Ion Traps
1992: Physics for Invention and Development of Particle Detectors
1994: Physics for Neutron Spectroscpy and Diffraction
1997: Physics for Laser Cooling and Trapping
2005: Physics for Laser-Based Precision Spectroscopy
2009: Physics for Fiber Optics and Imaging Semiconductor Circuit
2014: Chemistry for Super Resolved Fluorescent Microscopy
2017: Chemistry for Cryo-Electron Microscopy
2018: Physics for Laser Optical Tweezers and Laser High Intensity Ultra-Short Optical Pulses
2020: Chemistry for CRISPR/Cas9 Genetic Scissors
The field of chemical engineering is growing more interdisciplinary. This is evidenced by the large fraction, indeed majority, of the top 100 Chemical Engineers who do interdisciplinary research, as published by AICHE.
The field of chemical engineering has a growing problem of unreproducible results, which I can’t substantiate with data, only having been informed in lectures that this is the case. I also do not know relative to other fields if the problem is greater or if it is growing at a greater rate. In any case, the problem exists, and the solution is not totally field specific.
To address the modern challenge and the reproducibility problem, laboratories should be outfitted with extensive open source hardware, including sensors for ambient conditions and system and instrument conditions. These should be operated in controls by open-source software with data analyzed and archived by open-source software. There is evidently a need for the kind of bending over backwards Richard Feynman advised in his 1974 commencement address to Princeton on Cargo Cult Science. FOSS hardware and software are the means to meet the need.
There is a growing interest in machine learning and then artificial intelligence. Yet many laboratories, even those in high-ranking universities and technical disciplines, are not yet up to speed with the current information management and controls technology. There is more to be gained, more easily, only by making existing technologies easier to implement, as examples:
Information management that is meant only for human beings to analyze, beyond the handwritten paper notebook, such as ditigal notebooks and data archiving.
Controls which use PID routines for experimental variables such as pressure and temperature which run well-established auto-tune procedures.
I suspect the growth of consulting firms in chemical engineering, which was the largest growing sector in the mid 2010s, is mostly due to an inability of the industry to implement well-established best practices. Unlike industry, university labs cannot afford consultants, and so this is an attempt to provide practical advice to implement well-established practices. The posts in this blog should assist a novice student with implementing open source hardware and software. Nothing here is functionally more advanced than what exists and is implemented in some parts, usually high-tech, of industry. The emphasis is on low cost and ease of implementation.
Internal Links¶
See also the sidebar:
External Links¶
Resources I have found useful:
Tera Labs: At-home physics experiments in vacuum, electricity and magnetism. These at-home experiments show the fundamental physics behind apparatus, which too often lab workers are ignorant of.
The Linux Documentation Project: In particular see the Linux Serial HOWTO in the linux documentation project. All links to the “Linux Lab Project” started by Claus Schroeter of Berlin’s Freie Universitaet are broken, which would be most relevant. However guides on networking and serial communications are valuable.
Martindale Center: Website maintained full-time by a former consultant with links to calculators for many applications.
Feldman Consulting Enterprises, LLC: These are industrial solutions which include custom PCBs and robotics which can even be mass manufactured in some instances. Because it is commercial, the level of detail presented is not great. But it shows just how much can be accomplished with FOSS and open-source hardware (Arduino is based on Linux operating systems and its programs are made in C and derivatives). See in particular Feldman’s instrumentation cases.
Dabeaz: David Beazley is a Python educator and has written many python educational texts and articles. His Usenix ;login: contributions are always informative and for FOSS controls in particular those on asynchronous communications are helpful. The approach given here (see FOSS Controls) for effectively asynchronous communication just uses the operating system to manage read/write access to common files. There are performance, scaling, and portability reasons to use asynchronous protocols or multiprocessing within the python application. Threading may increase speed significantly when the program is rate-limited from waiting on I/O calls, and multiprocessing, by spawning truly independent processes on different cores of the CPU, may increase speed in any case, including when the program is rate-limited by its own computations. A highly accessible article is at https://realpython.com/python-concurrency/ and the author (not David Beazley) has several posts on concurrency in python.
Resources that are likely useful:
V. T. Toth: Linux has long been recognized as a superior low cost solution for computer networking, and the recommendations made here to use FOSS (on Linux OS) are far from novel, even for laboratories (see Claus Schroeter’s Linux Lab Project). V. T. Toth authored a book in 1999 on using Caldera OpenLinux to make a computing and networking solution on a ‘shoestring budget’ for a home or small office. In fact, there are many such books about installing Linux which emerged around that time period, and I can’t say whether this or another is best (I am partial to the author for unrelated reasons of his expositions on physics). It is also the case that commercialized Linux, such as Ubuntu, has made set-up of many services much simpler than at the time most of these books were authored. Because laboratory function is a superset of office function, to the extent guides such as Toth’s are not dated they should be useful.