Professional Documents
Culture Documents
A Brief Review of Natural Water's Influence On Scale Formation in Heat Exchangers
A Brief Review of Natural Water's Influence On Scale Formation in Heat Exchangers
A
tral and which greatly influences chemistry.
t the two power plants and chemical plant in
which this author formerly worked, the fresh- Atmospheric influences
water-makeup supplies (two lakes and an un- Natural waters absorb carbon dioxide from the atmo-
derground aquifer adjacent to a river) all had a sphere. While it is often argued that much of the CO2
consistent pH at or slightly above 8.0. This pH range is exists as hydrated molecules, the following equations
common for many surface supplies, and typically comes sufficiently represent the chemistry.
from bicarbonate alkalinity (HCO3–) that naturally dis-
solves in these water bodies. But from where does this CO2 + H2O H2CO3 (4)
alkalinity arise? Per the excellent discussion of this sub-
ject in Ref. 1, we briefly explore this issue and also exam- H2CO3 H+ + HCO3– (5)
ine what can happen in heat exchangers without proper
chemistry control. The lowest pH in natural surface waters that can be
achieved by these reactions (excluding acid rain issues)
Geology is one key is around 5.6, but the solution is still acidic, which is very
One of the most common mineral deposits near the earth’s important. Consider again, Equation (3). When the acidity
surface is limestone, whose principal component is cal- generated by CO2 absorption interacts with the alkalinity
cium carbonate (CaCO3). This versatile mineral serves as generated by the fractional CaCO3 dissolution, the hydro-
a raw material for numerous important industrial and infra- gen and hydroxyl ions combine to form water, and per Le
structure applications including concrete, water-treatment Chatelier’s Principle, Reactions (3) and (5) are both driven
chemicals, scrubbing reagent, and simply for gravel roads. to the right. This synergistic effect can produce water with
Many surface waters are in contact with limestone for- a HCO3– concentration of 1 × 10–3 M (equivalent to about
mations, and groundwaters often percolate through lime- 60 parts per million as the species), and “a pH of about
stone and settle in aquifers that are bounded by mineral. 8.3” [1]. The relationship of the carbonate species is clearly
Calcium carbonate has a strong crystal lattice, and thus illustrated in Figure 1.
CaCO3 is only slightly soluble in water. This same acid-base synergy is what makes high-pu-
rity limestone (high CaCO3 content) quite reactive and
CaCO3 Ca2+ + CO32– (1) economical as a scrubbing agent, when ground to very
fine particles, in wet fluegas desulfurization systems.
Ksp(25°C) = [Ca2+][CO32–] = 4.6 × 10–9 Aqueous sulfur dioxide (SO2) is a stronger acid than CO2,
and analytical data have shown nearly complete CaCO3
Per this solubility product, the molar concentrations of reactivity in well-designed scrubbers [2].
calcium and carbonate in neutral water would be 6.8 ×
10–5, which is indeed very slight. But an additional factor
must be considered. CO32– is a fairly strong base and will
hydrolyze water to some extent.
S
numerous manufacturing sectors, process-control platform. The nCon-
ynthesizing principles of including agriculture, water treat- trol system builds latent represen-
engineering and biology, ment, oil-and-gas, cleaning prod- tations of critical information using
Solugen Inc. (Houston; ucts, food additives, personal care empirical observations in the form of
www.solugen.com) has and concrete production. Solugen sensor data, extracting features con-
developed and commercially de- has a location in Slaton, Tex. and tributing to deviation in normal per-
ployed a unique carbon-negative a primary manufacturing plant in formance. It is able to surmise latent
platform for peroxide manufactur- Houston (Figure 1). As with any variables, whose analytical definition
ing. The platform, called Bioforge, new process technology, commer- is nearly intractable, and provide a
employs a renewable plant-based cialization and subsequent stable rich snapshot of the past, present
extract and combines it with de- operation involve many technologi- and future state of the process.
ionized water and compressed air, cal hurdles. This article describes The nControl prediction and con-
which is then processed through the implementation of an advanced trol models operate through dimen-
an enzyme reactor to continually process-control scheme that en- sionality reduction — compress-
extract molecules as the reac- abled Solugen to reduce opera- ing informationally sparse states,
tion is occurring. The intermediate tional inefficiencies and improve or data extracted from disparate
molecules are imbued with more production yield. sensors, into a representation that
compressed air and poured over conserves important attributes of
Solugen’s metal catalysts to create Overcome scalability challenges the observed phenomenon. This
a final product. For more details on Like any new process, scalabil- results in higher informational den-
Solugen’s process, see Enzyme ity is a concern. Within Solugen’s sity for both humans and machines
Engineering Enables Bio-based process, there is critical variation to make operational decisions — a
Hydrogen Peroxide, Chem. Eng., across the catalysts that are de- key benefit in Solugen’s process,
May 2019, p. 7. ployed and the chemistry on which where human interaction is classi-
The molecules Solugen pro- the process relies. This variation cally the first response to varying
requires quasi-human interpreta- process parameters.
Nanotronics tion of a given state to successfully Measurements are encoded with
progress the process. Through nonlinear, time-aware and sensor-
continuous use, the particles be- agnostic deep networks, ensuring
come inconsistent in structure and that each transformation embeds
do not perform in agreement with additional, pertinent information
the theoretical ideal. Therefore, op- in its latent representation. These
erators must be able to adjust plant representations are simultaneously
settings to account for discrepan- fed forward for the ultimate pur-
cies in saturation, which can be un- pose of control via artificial intelli-
reliable when left to human capac- gence (AI), but are also interpreted
ity or inference alone. Introducing for immediately useful consumption
a deep-learning control strategy to by human observers, in the form of
the plant was seen as one poten- anomaly detection and alerting, as
tial solution for real-time optimiza- well as quality metrics.
tion of such complex objectives. To The anomaly detection capa-
adjust for variation in particle com- bilities nControl offered to Solugen
position, it is necessary to derive were highly effective in streamlin-
an intelligent control scheme that ing and refining the site’s chemi-
automatically adapts to process cal processes. The platform is
FIGURE 1. At Solugen’s manufacturing plant in fluctuations in-situ. able to adjust for, as well as pre-
Houston, a novel technology uses enzymatic reac- Nanotronics (Cuyahoga Falls, dict, degradation of capital equip-
tors to produce carbon-negative chemical building
blocks
Ohio; www.nanotronics.co) was ment, unanticipated restriction of
42 CHEMICAL ENGINEERING WWW.CHEMENGONLINE.COM FEBRUARY 2024
Nanotronics parametric, meaning that it learns
to recognize new empirical pat-
terns not captured in theoretical
formulations of the production pro-
cess, such as physics-based or
chemistry-based simulations. It is
also inherently multivariate and re-
tains interactions between sensors.
Its multiscale structure enables it
to capture periodicities and other
patterns at different time scales —
predictions can be thought of as a
composite of multiple timelines.
For Solugen, the initial models
from Nanotronics are unsupervised
and do not require labor-intensive
data labeling. The system deftly
handles datasets with significant
variance and can be trained on a
FIGURE 2. The nControl platform provides process variety of operations — startup,
adjustments and predictions related to degrada- shutdown, maintenance and vari-
tion of equipment, such as valves and pipelines,
within Solugen’s plant ous run conditions — and has
successfully predicted processes
the valves and pipes that house across diverse situations, including
feedstocks and atypical chemistry conditions that were not visible in
correlated to exothermic events. training data. ■
Inferred quality metrics allow engi- Edited by Mary Page Bailey
neers and operators to determine
and execute goals in real time by Authors
providing immediate feedback on Damas Limoge is the director of
process deviations with acute ac- applied research at Nanotronics
(2251 Front Street, Suites 108-
curacy (Figure 2). 109, Cuyahoga Falls, OH 44221;
Solugen’s control strategies ben- Phone: 330-926-9809; Email:
efit from a reduction in the dimen- dlimoge@nanotronics.co), focus-
sionality of inferred quality metrics ing on nonlinear system control,
real-time analysis with generative
that highlight the important fea- autoencoders, time-series atten-
tures of both human-legible, use- tion networks, deep reinforcement
ful observations, and the values learning techniques and developing the nControl and
AIPC platforms. He holds a M.S. degree in mechanical
targeted by the control strategy, engineering from the Massachusetts Institute of Tech-
which is able to act faster and more nology (MIT) and has previously worked with MIT’s Lin-
precisely than a human agent. coln Laboratory, Resolute Marine Energy, General Elec-
tric and Goodrich in various mechanical engineering
Such dynamic performance is and simulation design roles.
unique among control systems
for complex chemical processes.
Cecilia McLaren is a marketing
Competing platforms operate as and communications associate at
backward-looking manufacturing Nanotronics (Brooklyn Navy Yard
analytics that do not include real- Building 20, Brooklyn, NY 11205),
where she serves as a writer, edi-
time process control. They may tor and media producer. She is the
provide retrospective reports and producer of Utility + Function,
insights or instantaneous monitor- Nanotronics CEO Matthew Put-
ing and alerting, but they do not man’s podcast, which explores an
interdisciplinary range of subjects,
have the capacity to actively con- including AI’s impact on social, economic and political
trol production processes, and no spheres. McLaren holds a B.S. degree in media, culture
software-only solution is capable and communication from New York University, and has
been involved with the university’s Digital Theory Lab, a
of making these interventions. research cohort convening graduate students and fac-
Typically, factory operations must ulty members from universities across the globe to con-
be networked via a programmable duct multimodal inquiry converging technological arti-
facts and humanities-based methodologies.
logic controller (PLC) to achieve
this effect, whereas the nControl
platform can act on occurring pro- Note: nControl is a registered trademark of
cesses immediately. Nanotronics
Furthermore, nControl is non-