Professional Documents
Culture Documents
DA in Air Separation Unit
DA in Air Separation Unit
a r t i c l e i n f o a b s t r a c t
Article history: Managing electricity demand has become a key consideration in power grid operations. Industrial de-
Received 19 December 2018 mand response (DR) is an important component of demand-side management, and electricity-intensive
Revised 16 March 2019
chemical processes can both support power grid operations and derive economic benefits from electric-
Accepted 16 March 2019
ity price fluctuations. For air separation units (ASUs), DR participation calls for frequent production rate
Available online 25 March 2019
changes, over time scales that overlap with the dominant dynamics of the plant. Production scheduling
Keywords: calculations must therefore explicitly consider process dynamics. We introduce a data-driven approach
Integrated scheduling and control for learning the DR scheduling-relevant dynamics of an industrial ASU from its operational history, and
Scale-bridging models present a dynamic optimization-based DR scheduling framework. We show that a class of low-order
Demand-side management Hammerstein-Wiener models can accurately represent the dynamics of the industrial ASU and its model
Electricity markets predictive control system. We evaluate the economic benefits of the proposed scheduling framework, and
Air separation units
analyze their sensitivity to electricity price uncertainty.
© 2019 Elsevier Ltd. All rights reserved.
https://doi.org/10.1016/j.compchemeng.2019.03.022
0098-1354/© 2019 Elsevier Ltd. All rights reserved.
C. Tsay, A. Kumar and J. Flores-Cerrillo et al. / Computers and Chemical Engineering 126 (2019) 22–34 23
(Baldea and Harjunkoski, 2014). Conventional methods for com- operating points and identify the optimal production schedule.
puting optimal production schedules therefore implicitly assume Zhao et al. (2019) proposed a state-transition network model for
that a chemical process is at a steady state before and after a scheduling ASUs similar to that of Basán et al. (2018), and ap-
change in production targets. However, electricity prices in current, plied it to the simultaneous scheduling of two multi-product ASUs.
deregulated markets may change at hourly (or faster) time inter- Obermeier et al. (2019) defined a mode-based scheduling approach
vals, and production scheduling decisions must therefore be made to examine relationships between DR scheduling and equipment
over a time scale in which process dynamics and control become fatigue.
highly relevant (Zhang and Grossmann, 2016). Specifically, produc- On the other hand, simplified (e.g., by reducing dimension-
tion targets must change over hourly intervals, and process safety ality) dynamic models can be employed for computationally
and product purity must be accounted for when improving process tractable scheduling calculations as an alternative to assuming
economics by such rapid production modulation. The decision- quasi-stationary modes. Dias et al. (2018) showed that model pre-
making time scale is further shortened in faster electricity markets dictive control (MPC) can be integrated in production scheduling
(e.g., the fifteen minute market), when it is desired to provide an- calculations with low computational effort using a nested decision-
cillary services, or place electricity bids in real-time (Dowling et al., making structure (Zhuge and Ierapetritou, 2015). A simulation-
2017; Otashu and Baldea, 2018; Schäfer et al., 2019b). optimization framework was proposed, where simple state-space
In light of these developments, the integration of scheduling models derived via system identification were employed to predict
and control for chemical processes is an important research area. the open-loop process dynamics. We note that economic MPC can
Recent advances in this field have been the subject of several directly incorporate the economic objectives usually considered in
recent review papers (Baldea and Harjunkoski, 2014; Daoutidis scheduling into the control layer (Baldea and Harjunkoski, 2014);
et al., 2018; Dias and Ierapetritou, 2019), but, for the sake of com- however, optimizing the (large-scale) process model in real-time
pleteness, a brief overview of relevant works is presented here. and considering a time horizon relevant to production scheduling
In an initial effort, Flores-Tlacuahuac and Grossmann (2006) pro- requires some measure of compromise, e.g., a sub-optimal update
posed explicitly including the dynamic process model in the approach (Caspari et al., 2018).
scheduling problem, representing the scheduling and control lay- Recent work from our group has suggested system identifica-
ers as a single, unified, simultaneous dynamic optimization prob- tion as a means to accurately capture closed-loop input-output re-
lem cast as a mixed-integer nonlinear program (MINLP). This lationships with low-order models referred to as “scale-bridging
discretized-time approach was later implemented in closed-loop models” (SBMs), which can then be naturally included in pro-
by Zhuge and Ierapetritou (2012). Beal et al. (2018) provided a fur- cess scheduling calculations (Baldea et al., 2015; Du et al., 2015).
ther extension to allow for time-dependent parameters and con- Pattison et al. (2016) demonstrated these ideas on a simulated,
straints, while still accounting for the full dynamic process model. single-product ASU, including a closed-loop scheduling, moving-
Nyström et al. (2005) simplified the integrated problem by sep- horizon implementation (Pattison et al., 2017). The work was later
arating it into a scheduling master problem and a control sub- extended by Kelley et al. (2018), where the authors expedited the
problem, achieving integrated scheduling and control by iterating computational solution of the scheduling problem by using a fully-
between the two. Similarly, Chu and You (2012) simplified the discretized time domain and an exact MILP reformulation.
scheduling and control problem by using linear control concepts, In this work, we present a related methodology for creating
computing optimal tuning parameters for a set of PI controllers of- low-order, scheduling-relevant models of a multi-product indus-
fline. Burnak et al. (2018) proposed capturing fast process dynam- trial ASU using historical operating data provided by our indus-
ics with surrogate models in a multiparametric scheduling frame- trial partner. Focusing on demand-response operation, we iden-
work. tify dynamic models for economically-relevant variables (produc-
For large-scale models, such as those that describe ASUs, com- tion rates, power consumption) and the process variables found to
puting schedules through dynamic optimization with detailed first- limit the dynamic agility of the process and its ability to shift op-
principles models is typically computationally intractable in practi- erations to fast-changing electricity price signals. Furthermore, this
cal amounts of time (Dias and Ierapetritou, 2017). To examine the work accounts for the fact that the industrial ASU operates un-
feasibility of DR production-rate modulation in ASUs, Cao et al. der model predictive control (MPC). The industrial ASU also pro-
(2011, 2016a) presented initial results on dynamic modeling and duces multiple products, in contrast to related works investigating
optimization using a first-principles process model by restricting a simulated ASU process that produces only nitrogen (Dias et al.,
the time horizon to individual production rate transitions. The 2018; Pattison et al., 2017; 2016). We incorporate the scheduling-
same authors subsequently examined alternative approaches for relevant models into a DR scheduling problem formulated as a dy-
the collection and usage of liquid storage using collocation-based namic optimization and present an extensive case study to exam-
dynamic models (Cao et al., 2016b). More recently, Schäfer et al. ine the benefits of DR operation. Although re-scheduling (Gupta
(2019a) applied a compartmentalization-based model reduction et al., 2016; Pattison et al., 2017) is not considered in this study,
approach for ASUs and demonstrated improved computational per- the identified low-order models allow the scheduling problem to
formance compared to collocation-based approaches. be solved fast enough for online scheduling and can be easily up-
In an effort to balance computational complexity with captur- dated if the controller and/or the plant are modified.
ing (some of) the relevant process dynamics in production schedul-
ing, many works assume quasi-stationary modes of operation with 2. Background
additional constraints tailored to reflect the transition capabili-
ties of the plant and control considerations (Misra et al., 2017; 2.1. Description of the industrial air separation unit under
Zhang et al., 2015, 2016; Zhou et al., 2017). Zhang et al. (2016) ex- consideration
tended the mode-based scheduling framework presented by Mitra
et al. (2012, 2014) by introducing surrogate sub-process mod- ASUs separate the components of air predominantly for use
els that allow for computationally efficient scheduling of contin- in other manufacturing processes. For example, argon is used in
uous process networks. Misra et al. (2017) used a state-task net- welding, oxygen in steel production, and nitrogen in food and met-
work to model production constraints. With a similar motiva- als processing. High-purity oxygen also has medical applications.
tion, Zhou et al. (2017) defined a set of ASU operating modes Although several air separation technologies are available, the bulk
from historical data, and used convex hulls to reflect feasible of high-purity, high-volume industrial gas production is dominated
24 C. Tsay, A. Kumar and J. Flores-Cerrillo et al. / Computers and Chemical Engineering 126 (2019) 22–34
by cryogenic ASUs, which consume immense amounts of energy tional refrigeration in the MHEX. The waste nitrogen stream is then
and primarily rely on electric compressors to handle and compress vented to the atmosphere. Purified oxygen product is collected at
their air feed streams. The industrial cryogenic ASU considered in the bottom of the UC. A portion of the oxygen product is vaporized
this work produces five products: liquid nitrogen (LN2 ), gaseous ni- through the MHEX, while the remainder is passed through the SH.
trogen (GN2 ), liquid oxygen (LO2 ), gaseous oxygen (GO2 ), and liq- The air fed to the process contains a small amount of argon, which
uid argon (LAr). The process operates under model predictive con- accumulates on an intermediate stage close to the bottom of the
trol (MPC) (Seborg et al., 2010), with the multivariable control sys- UC. At this location, an intermediate vapor stream is drawn and
tem having 12 manipulated variables and 35 controlled variables. fed to the argon column (ARC). The ARC operates at a very high
As previously noted, many previous studies (Cao et al., reflux ratio, and the “crude” argon stream removed at the top of
2011, 2016a; Dias et al., 2018; Pattison et al., 2016; Schäfer the ARC has a very small flow rate. High purity argon is then ob-
et al., 2019a) focusing on ASU scheduling involve a simulated tained in a separate device that is not considered here.
plant that produces only nitrogen. Multi-product ASUs have in
some instances been considered, but using reduced-order mod-
els (Cao et al., 2016b) or assuming quasi-stationary operating 2.2. Scheduling under dynamic constraints
points/modes (Misra et al., 2017; Zhang et al., 2016) – in essence,
another dimensionality-reduction device. One should recognize Historically, production scheduling calculations relied on the
that such model reduction measures are computationally efficient, simplifying assumption that the transition of a chemical process
and have even allowed extending production scheduling calcula- from one operating point to another can be defined in terms of
tions to sites that are served by several multi-product ASUs (Zhao a transition period, and that the process predominantly operates
et al., 2019; Zhou et al., 2017); however, to our knowledge, there at steady states corresponding to the aforementioned operating
are no results available in the open literature on the explicit incor- points. Under the assumption that the durations of the transi-
poration of process dynamics in scheduling calculations for multi- tions periods are known, are short compared to periods of steady-
product ASUs. state operation, and do not change in time, scheduling calculations
The flowsheet of the process considered in this work is shown can effectively ignore the dynamics of the process, and schedul-
in Fig. 1. Air enters the feed compressor (FC), which is driven by ing decisions can be separated from control/operational decisions
a large electric motor. Impurities such as water and hydrocarbons (Seborg et al., 2010).
are removed in the prepurifier (PP). The air stream is then cooled Recent works (Baldea and Harjunkoski, 2014; Dias et al., 2018)
in the multistream heat exchanger (MHEX) against warming cryo- have demonstrated that these assumptions may no longer be valid
genic product streams and enters the high-pressure lower column when market conditions change frequently. For DR scheduling of
(LC), where it is separated into nearly pure nitrogen “shelf gas” (at ASUs, this is particularly true given that electricity prices change at
the top of the LC) and an oxygen-rich bottoms stream. A portion frequencies similar to (or faster than) the slowest dynamic modes
of the shelf gas provides the reboiler duty to the low-pressure up- of the process. Fig. 2 illustrates a hypothetical example, where, in
per column (UC) before being returned to the LC as reflux, while schedule 1, a process violates a dynamic constraint for a particu-
the rest of the shelf gas passes through the MHEX and is com- lar state variable owing to operators imposing a large step change
bined with the nitrogen product of the UC at the liquefier (LQ). in operation too quickly. In schedule 2, the production schedule
The oxygen-rich bottoms stream of the LC passes through the ni- accounts for the dynamic agility of the process when making the
trogen superheater (SH) before being used as a condensing utility step change(s). Note the transition times are different in schedules
for the argon column (ARC) and passed to the UC. 1 and 2, and accounting for process dynamics may lead to “slower”
A portion of the combined nitrogen streams is liquefied in the (i.e., with less frequent/abrupt changes), but dynamically feasible
liquefier (LQ), which provides both gaseous and liquid nitrogen schedules. The production scheduling problem considering process
products. A “waste” stream of nitrogen is drawn at an interme- dynamics can generally be stated as a dynamic optimization prob-
diate stage close to the top of the UC in order to provide addi- lem (Pattison et al., 2016):
C. Tsay, A. Kumar and J. Flores-Cerrillo et al. / Computers and Chemical Engineering 126 (2019) 22–34 25
Fig. 2. Left: Hypothetical process schedule that violates a process constraint while tracking a step change imposed in a scheduling-relevant variable. Right: Hypothetical
process schedule with constraints correctly enforced while making the desired transition.
Fig. 3. A comparison between scheduling using detailed closed-loop process models (A) and scheduling with scale-bridging models (Baldea et al., 2015; Pattison et al., 2016)
(B). The dashed box in (A) reflects the system whose dynamics are represented by the SBMs.
data suggest that they do not hinder process agility. Briefly, the ap- scheduled to avoid consuming electricity during price peaks, they
proach for learning SBMs comprises the following steps: were not considered in this study. However, the shutdown of the
liquefier was considered, as described later.
1. Obtain historical process operating data
The data correspond to summer operation, as the summer
2. Identify scheduling-relevant variables to be modeled
months tend to have larger fluctuations in electricity prices. The
3. Determine model form and fit a dynamic model for each
first 88 days (80%) of the data were selected as training data, with
scheduling-relevant variable
the remaining 22 days (20%) used as test data to evaluate the qual-
In the application of the above system identification framework ity of the identified SBMs. While the choice of the functional form
to the industrial ASU considered here, several important factors of data-driven model can be arbitrary, due to the lack of deliberate
had to be considered. On the input side of the models, while the system excitation, we chose to identify SBMs in the Hammerstein-
process operates under MPC with six operator set points (targets), Wiener (HW) form. This choice was supported by successful ap-
we found through trial-and-error that the production rates and plication of HW models to capture the relevant process dynamics
other scheduling-relevant state and output variables could be ac- in a smaller-scale problem focused on DR scheduling of a single-
curately modeled using only two of the six operator set points, product ASU (Pattison et al., 2016). HW models are structured, re-
[SP1, SP2]. This finding suggests that the remaining four operator quiring fewer parameters to be regressed. They comprise a linear
set points either have negligible impact on the variables of inter- dynamic component flanked by static, nonlinear input and output
est for scheduling, or that they have historically been set such that transformations. They can be formulated for multiple inputs and
they are strongly correlated with the first two set points. On the a single output (MISO) with the linear dynamic component repre-
other hand, we found that ambient temperature T, which is a mea- sented as a state-space model:
sured disturbance variable, has a large effect on the process vari-
ables and must be included as an input to the SBMs. Given these hi = Hi (ui ) (5)
findings, we identified multiple-input, single-output (MISO) SBMs.
The SBM used to predict the behavior of each scheduling-relevant z˙ i = Aizi + Bi hi (6)
variable was modeled with three inputs, [T, SP1, SP2].
While in our previous works (Dias et al., 2018; Pattison et al., yi = Cizi (7)
2016) we relied on relatively small simulated datasets (whose hori-
zons span a few days to a week) for system identification, in this w=W yi (8)
work we use a large set of actual process historical operating data.
Since in the present application the data lack the deliberate ex- where H and W are, respectively, the Hammerstein and Wiener
citation imposed during system identification experiments (as in blocks corresponding to the static, nonlinear input and output
our previous work), we included 110 days of historical operating transformations. Ai , Bi , and Ci are the matrices defining the linear
data from the industrial process to ensure that sufficiently rich in- state-space dynamical system, ui denotes the ith model input (for
formation is available for identifying the desired SBMs. The data the ASU, ui ∈ [T, SP1, SP2]), and w is the model output. The linear
were recorded at one-minute intervals in the process historian state-space system is of order nd,i , with zi ∈ IRnd,i . The structure of
database, during periods of regular, unforced operation. Periods of a generic MISO Hammerstein–Wiener model with three inputs and
start-up, shut-down, and process or measurement faults were ex- one output is shown in Fig. 4.
cluded from this study. These periods were identified easily in the Although the industrial ASU produces five products, we found
historical data, as the sensors are either off or produce readings that the liquid oxygen production rate was relatively constant
that are, e.g., outside the physical bounds for the respective vari- throughout the operating period represented in the data. As ac-
ables. Although shut-downs of the entire ASU could potentially be tual industrial data were used in this study, the argon production
C. Tsay, A. Kumar and J. Flores-Cerrillo et al. / Computers and Chemical Engineering 126 (2019) 22–34 27
Table 1
Details of Hammerstein–Wiener scale-bridging models (refer to Fig. 4 for the model struc-
ture).
rate was not disclosed and modeled for confidentiality reasons. The the connection of adjacent line segments: bpi is the value of
production rates that are therefore considered in DR scheduling the breakpoint at segment i, and pwi is the value of the non-
calculations (and whose dynamic responses to changes in [T, SP1, linear function at the breakpoint bpi . To determine the num-
SP2] are modeled) are thus those of liquid nitrogen (PLN 2 ), gaseous ber of piecewise-linear segments np for each transformation, the
nitrogen (PGN 2 ), and gaseous oxygen (PGO 2 ). The MPC implementa- normalized Akaike information criterion (nAIC) was minimized
tion includes constraints on several process variables, but we found while using a large number of segments for all other piecewise-
that only two controlled process variables, denoted here as CV1 linear transformations and a high-order linear state-space model.
and CV2, closely approach their bounds during operational transi- The order of each linear state-space model was similarly deter-
tions, indicating they are the primary factors limiting the dynamic mined using the nAIC. Table 1 also reports the normalized mean
agility of the ASU. In the interest of protecting industrial intellec- square error (NMSE) for each model on training and test data,
tual property, their true nature cannot be disclosed. Together with defined as:
the power consumption of the plant, the scheduling-relevant out-
xref − x2
puts of the models are w ∈ [PLN2 , PGN2 , PGO2 , CV1, CV2, Power]. NMSE = 1 − (11)
xref − mean xref 2
3.2. System identification results
The SBM predictions for all six variables listed in Table 1 for
a week within the training dataset are plotted in Fig. 5. The
A MISO SBM in Hammerstein–Wiener form (5)–(8) was identi-
low NMSE values are largely due to high-frequency dynam-
fied for each scheduling-relevant variable, and the pertinent details
ics in the historical data: the MPC operates (and the data are
are shown in Table 1. The models were obtained using the Sys-
recorded) on a 1-min interval, while the operator set points are
tem Identification Toolbox in MATLAB. Kelley et al. (2018) found
changed much less frequently. The model predictions are close
that the dynamics of an ASU could be accurately represented us-
to the time-averaged behavior of the process variables in most
ing Hammerstein–Wiener models with piecewise linear input and
cases, and the accuracy of model predictions could most likely
output functions. Consequently, the nonlinearities were all repre-
be improved in future studies by collecting process (or simu-
sented as piecewise linear functions of the form:
lated) data with higher excitation levels such as through sys-
pw j+1 − pw j
Hi (ui ) = (ui − bp j ) + pw j ; ui ∈ [bp j , bp j+1 ) (9) tem identification experiments. Additional data may also help
bp j+1 − bp j improve model accuracy by affording the selection of a differ-
ent model representation/structure. HW models are inherently
pwk+1 − pwk restricted to a linear representation of the dynamics; although
W yi = yi − bpk + pwk ;
bpk+1 − bpk this feature does not pose limitations for this study, it may be-
come limiting in cases where the dominant time constant of
yi ∈ [bpk , bpk+1 ) (10)
the process changes significantly over time or as a result of
where j = 1, . . . , n p + 1 is the set of piecewise linear segments changes in operating regime. We note that all output and in-
in an input nonlinear transformation; k is used to similarly in- put variables listed or plotted in this article have been scaled
dex the output nonlinear transformation. The piecewise linear and filtered to preserve the confidential nature of industrial
functions are each parameterized by np breakpoints that define data.
28 C. Tsay, A. Kumar and J. Flores-Cerrillo et al. / Computers and Chemical Engineering 126 (2019) 22–34
4.2. Results for the case of real-time electricity pricing 1994b), using two-hour piecewise constant profiles. As expected,
the optimal schedules found for both P1 and P2 involve over-
The optimization problem in (12)–(18) was solved using the producing (and increasing power consumption with respect to the
prices Celec (t) given by the historical real-time electricity price base case) when electricity prices are low. Conversely, power con-
shown in Fig. 6. As their name implies, these prices are set in sumption is decreased when electricity prices are high. The power
real time and, as a consequence, their values for a 96-h future demand profiles for P1 and P2 are shown in Fig. 7. Although the
time horizon would not be available in practice at the time when identified SBMs are (piecewise) continuous, a visual comparison
scheduling calculations are performed. Rather, predictions would of the results against historical data suggests that the sharp drop-
need to be used to generate the optimal schedule. The bounds de- off in power consumption between hours 40–45 in both computed
fined in the MPC system for CV1 and CV2 were used as their re- schedules corresponds to a liquefier shutdown, where power con-
spective upper and lower bounds in (14)–(15). We will refer to this sumption is drastically decreased at the cost of producing no LN2
problem as P1. The same problem was solved using a 10% backoff (PLN2 ≈ 0).
constraint for both CV1 and CV2 to generate a more conservative The LN2 production profiles for the computed optimal sched-
schedule, noting that the SBMs representing process dynamics may ules are shown in Fig. 8. In the optimal solution for both P1 and P2,
not be completely accurate (Table 1). Such “backoffs” for active only the LN2 production constraint (16) is exactly met, as the other
constraints have been used for SBM-based scheduling (Dias et al., two products are over-produced throughout the four-day window
2018; Pattison et al., 2016), and are in effect implemented in many to allow for excess LN2 production. LN2 production is decreased to
practical situations to avoid constraint violations in the presence of practically zero during the liquefier shutdown episodes, as can be
disturbances and/or model inaccuracy. We direct the reader to the easily seen in Fig. 8. We again note that the variables are scaled
works by Narraway and Perkins (1993) and Aske et al. (2008) for and that a value of PLN2 = 0 actually corresponds to a “negative”
further information. We will refer to the second problem, which net LN2 production rate, i.e., LN2 is transferred from storage to the
includes the aforementioned backoff constraints, as P2. For each UC (Fig. 1) when the liquefier is turned off. The profiles of CV1
variable, the 10% backoff was computed by adding/subtracting 10% at the optimal points are also shown in Fig. 8; CV2 did not reach
of the scaled variable range to/from its lower/upper bound. its bounds in the optimal schedules. It can be seen that using the
Optimization of the operator set points was treated using a tighter backoff constraint results in CV1 reaching its lower bound
control vector parameterization approach (Vassiliadis et al., 1994a; at multiple points in the solution of P2. The backoff constraint
30 C. Tsay, A. Kumar and J. Flores-Cerrillo et al. / Computers and Chemical Engineering 126 (2019) 22–34
Fig. 9. Real-time (top) and day-ahead (bottom) electricity prices under uncertainty.
The shaded area denotes one standard deviation for i ∼ N(0, 50% ). Five Monte
Carlo (dotted) samples are shown.
Pˆj − Pj
j = (19)
Pj
Fig. 11. Power consumption predicted by SBMs for the optimal schedules in the
day-ahead market using the MPC bounds (P3, top) and backoff bounds (P4, bottom).
Fig. 10. Relative costs computed in Monte Carlo simulations for P1 (top) and P2 right, indicating that although the chances of exceeding constant-
(bottom) using σ = {10%, 30%, 50%}. Skewed Gaussian probability density functions
are shown as dashed lines.
operation costs may be low, the probabilities of encountering ex-
treme values are greater on the right-hand side of the distribution.
Fig. 13. Relative costs found in Monte Carlo simulations for P3 (top) and P4 (bot-
Fig. 12. Profiles of PLN2 and CV1 predicted by SBMs for the computed optimal tom) using σ = {10%, 30%, 50%}. Skewed Gaussian probability density functions are
schedules in the day-ahead market. shown as dashed lines.
larger than the corresponding values found for the real-time elec-
nitrogen at times of lower electricity price. We note that this is tricity price cases (P1 and P2).
likely strongly related to the large number of liquefier shutdown We again explore the effect of price uncertainty on the sched-
events recorded in the training data set. While this inherently bi- ules using the same Monte Carlo simulation with i ∼ N(0, σ ), but
nary decision is handled using the HW model, a discrete variable with no uncertainty in the first 24 h (where day-ahead prices
could also model the on/off state of the liquefier. The profiles of would be exactly known). The first five samples for σ = 50%
CV1 at the optimal points are also shown in Fig. 12; CV2 is again are shown in Fig. 9 (bottom). Histograms of the relative costs
not found to reach its bounds in the optimal schedules. The backoff found in the Monte Carlo simulations for P3 and P4 using σ =
constraint is met multiple times in the solution to P4. {10%, 30%, 50%} are shown in Fig. 13. The same abscissae as in
We again define CDA 0 as the incurred cost of four-day operation Fig. 10 were used in Fig. 13 to facilitate comparison between re-
in the base case given day-ahead electricity prices. The optimal sults for the real-time and day-ahead markets. A comparison be-
schedule found by solving P3 results in a 3.6% savings compared tween Figs. 10 and 13 reveals that, although the expected relative
0 , while the schedule given by the solution of P4 results in a
to CDA cost of DR operation in the day-ahead market is higher (demand
2.4% savings. Although lower than in the case of the real-time mar- response in the real-time market yields larger savings in the case
ket, these savings still constitute a significant economic advantage of no error in the expected values of electricity prices), the sched-
in the industrial gas sector. The economic benefits are lower when ules based on day-ahead data are less affected by price uncertainty,
the backoff constraint is used, as the tighter bound for CV1 gives a and the distributions of relative costs found for the day-ahead mar-
more conservative measure to the degree to which the ASU power ket have significantly less variance. There is still a small probability
consumption can be modulated (Fig. 12). Using the same storage that the relative cost exceeds 1. The relevant statistical data can be
capacity estimation method as above, the solution for P3 requires found in full in the Supporting information.
a liquid storage capacity of 2.62 times the hourly average produc- The probabilities of relative costs exceeding 1, estimated from
tion, while the solution for P4 requires a capacity of 1.95 times Monte Carlo frequencies for all four schedules, are shown in
the average hourly production. These storage capacities are slightly Fig. 14. The mean relative costs and standard deviations for all four
C. Tsay, A. Kumar and J. Flores-Cerrillo et al. / Computers and Chemical Engineering 126 (2019) 22–34 33
5. Conclusions
Disclaimer
University Graduate Continuing Fellowship at The University of Klus, S., Nüske, F., Koltai, P., Wu, H., Kevrekidis, I., Schütte, C., Noé, F., 2018. Data–
Texas at Austin. driven model reduction and transfer operator approximation. J. Nonlinear Sci.
28 (3), 985–1010.
Kou, P., Liang, D., Gao, L., Lou, J., 2015. Probabilistic electricity price forecasting with
Supplementary material variational heteroscedastic Gaussian process and active learning. Energy Con-
vers. Manage. 89, 298–308.
Misra, S., Kapadi, M., Gudi, R.D., Srihari, R., 2017. Energy-efficient production
Supplementary material associated with this article can be scheduling of a cryogenic air separation plant. Ind. Eng. Chem. Res. 56 (15),
found, in the online version, at doi:10.1016/j.compchemeng.2019. 4399–4414.
Mitra, S., Grossmann, I.E., Pinto, J.M., Arora, N., 2012. Optimal production planning
03.022. under time-sensitive electricity prices for continuous power-intensive processes.
Comput. Chem. Eng. 38, 171–184.
References Mitra, S., Pinto, J.M., Grossmann, I.E., 2014. Optimal multi-scale capacity planning
for power-intensive continuous processes under time-sensitive electricity prices
and demand uncertainty. part i: modeling. Comput. Chem. Eng. 65, 89–101.
Aske, E.M., Strand, S., Skogestad, S., 2008. Coordinator MPC for maximizing plant
Narraway, L.T., Perkins, J.D., 1993. Selection of process control structure based on
throughput. Comput. Chem. Eng. 32 (1–2), 195–204.
linear dynamic economics. Ind. Eng. Chem. Res. 32 (11), 2681–2692.
Baldea, M., Daoutidis, P., 2012. Dynamics and Nonlinear Control of Integrated Pro-
Nyström, R.H., Franke, R., Harjunkoski, I., Kroll, A., 2005. Production campaign plan-
cess Systems. Cambridge University Press, Cambridge, UK.
ning including grade transition sequencing and dynamic optimization. Comput.
Baldea, M., Du, J., Park, J., Harjunkoski, I., 2015. Integrated production schedul-
Chem. Eng. 29 (10), 2163–2179.
ing and model predictive control of continuous processes. AlChE J. 61 (12),
Obermeier, A., Windmeier, C., Esche, E., Repke, J.-U., 2019. A discrete-time schedul-
4179–4190.
ing model for power-intensive processes taking fatigue of equipment into con-
Baldea, M., Harjunkoski, I., 2014. Integrated production scheduling and process con-
sideration. Chem. Eng. Sci. 195, 904–920.
trol: a systematic review. Comput. Chem. Eng. 71, 377–390.
Otashu, J.I., Baldea, M., 2018. Grid-level battery operation of chemical processes and
Basán, N.P., Grossmann, I.E., Gopalakrishnan, A., Lotero, I., Méndez, C.A., 2018. Novel
demand-side participation in short-term electricity markets. Appl. Energy 220,
milp scheduling model for power-intensive processes under time-sensitive elec-
562–575.
tricity prices. Ind. Eng. Chem. Res. 57 (5), 1581–1592.
Pattison, R.C., Touretzky, C.R., Harjunkoski, I., Baldea, M., 2017. Moving horizon
Beal, L., Petersen, D., Pila, G., Davis, B., Warnick, S., Hedengren, J., 2017. Economic
closed-loop production scheduling using dynamic process models. AlChE J. 63
benefit from progressive integration of scheduling and control for continuous
(2), 639–651.
chemical processes. Processes 5 (4), 84.
Pattison, R.C., Touretzky, C.R., Johansson, T., Harjunkoski, I., Baldea, M., 2016. Optimal
Beal, L.D., Petersen, D., Grimsman, D., Warnick, S., Hedengren, J.D., 2018. Integrated
process operations in fast-changing electricity markets: framework for schedul-
scheduling and control in discrete-time with dynamic parameters and con-
ing with low-order dynamic models and an air separation application. Ind. Eng.
straints. Comput. Chem. Eng. 115, 361–376.
Chem. Res. 55 (16), 4562–4584.
Burnak, B., Katz, J., Diangelakis, N.A., Pistikopoulos, E.N., 2018. Simultaneous process
Process Systems Enterprise, 1997–2018. general PROcess Modeling System
scheduling and control: a multiparametric programming-based approach. Ind.
(gPROMS). http://www.psenterprise.com/gproms.
Eng. Chem. Res. 57 (11), 3963–3976.
Schäfer, P., Caspari, A., Kleinhans, K., Mhamdi, A., Mitsos, A., 2019a. Reduced dy-
Cao, Y., Swartz, C.L., Baldea, M., 2011. Design for dynamic performance: application
namic modeling approach for rectification columns based on compartmental-
to an air separation unit. In: Proceedings of the American Control Conference
ization and artificial neural networks. AlChE J. doi:10.1002/aic.16568.
(ACC), 2011. IEEE, pp. 2683–2688.
Schäfer, P., Westerholt, H.G., Schweidtmann, A.M., Ilieva, S., Mitsos, A., 2019b. Mod-
Cao, Y., Swartz, C.L., Flores-Cerrillo, J., 2016a. Optimal dynamic operation of a high-
el-based bidding strategies on the primary balancing market for energy-intense
purity air separation plant under varying market conditions. Ind. Eng. Chem.
processes. Comput. Chem. Eng. 120, 4–14.
Res. 55 (37), 9956–9970.
Seborg, D.E., Mellichamp, D.A., Edgar, T.F., Doyle III, F.J., 2010. Process Dynamics and
Cao, Y., Swartz, C.L., Flores-Cerrillo, J., Ma, J., 2016b. Dynamic modeling and collo-
Control. John Wiley & Sons, Hoboken, NJ.
cation-based model reduction of cryogenic air separation units. AlChE J. 62 (5),
Tsay, C., Baldea, M., Shi, J., Kumar, A., Flores-Cerrillo, J., 2018a. Data-driven mod-
1602–1615.
els and algorithms for demand response scheduling of air separation units. In:
Caspari, A., Faust, J.M., Schäfer, P., Mhamdi, A., Mitsos, A., 2018. Economic nonlinear
Towler, G.P., Eden, M.R., Ierapetritou, M.G. (Eds.), Computer Aided Chemical En-
model predictive control for flexible operation of air separation units. IFAC-Pa-
gineering, Proceedings of the Thirteenth International Symposium on Process
persOnLine 51 (20), 295–300.
Systems Engineering, San Diego, CA. Elsevier, pp. 1273–1278.
Chu, Y., You, F., 2012. Integration of scheduling and control with online closed-loop
Tsay, C., Pattison, R.C., Piana, M.R., Baldea, M., 2018b. A survey of optimal process
implementation: fast computational strategy and large-scale global optimization
design capabilities and practices in the chemical and petrochemical industries.
algorithm. Comput. Chem. Eng. 47, 248–268.
Comput. Chem. Eng. 112, 180–189.
Conejo, A.J., Nogales, F.J., Arroyo, J.M., 2002. Price-taker bidding strategy under price
Vassiliadis, V.S., Sargent, R.W., Pantelides, C.C., 1994a. Solution of a class of mul-
uncertainty. IEEE Trans. Power Syst. 17 (4), 1081–1088.
tistage dynamic optimization problems. 1. Problems without path constraints.
Daoutidis, P., Lee, J.H., Harjunkoski, I., Skogestad, S., Baldea, M., Georgakis, C., 2018.
Ind. Eng. Chem. Res. 33 (9), 2111–2122.
Integrating operations and control: a perspective and roadmap for future re-
Vassiliadis, V.S., Sargent, R.W., Pantelides, C.C., 1994b. Solution of a class of multi-
search. Comput. Chem. Eng. 115, 179–184.
stage dynamic optimization problems. 2. Problems with path constraints. Ind.
Dias, L.S., Ierapetritou, M.G., 2017. From process control to supply chain manage-
Eng. Chem. Res. 33 (9), 2123–2133.
ment: an overview of integrated decision making strategies. Comput. Chem.
Willcox, K., Peraire, J., 2002. Balanced model reduction via the proper orthogonal
Eng. 106, 826–835.
decomposition. AIAA J. 40 (11), 2323–2330.
Dias, L.S., Ierapetritou, M.G., 2019. Optimal operation and control of intensified pro-
Zhang, Q., Grossmann, I.E., 2016. Enterprise-wide optimization for industrial de-
cesses challenges and opportunities. Curr. Opin. Chem. Eng. doi:10.1016/j.coche.
mand side management: fundamentals, advances, and perspectives. Chem. Eng.
2018.12.008.
Res. Des. 116, 114–131.
Dias, L.S., Pattison, R.C., Tsay, C., Baldea, M., Ierapetritou, M.G., 2018. A simula-
Zhang, Q., Grossmann, I.E., Heuberger, C.F., Sundaramoorthy, A., Pinto, J.M., 2015.
tion-based optimization framework for integrating scheduling and model pre-
Air separation with cryogenic energy storage: optimal scheduling considering
dictive control, and its application to air separation units. Comput. Chem. Eng.
electric energy and reserve markets. AlChE J. 61 (5), 1547–1558.
113, 139–151.
Zhang, Q., Sundaramoorthy, A., Grossmann, I.E., Pinto, J.M., 2016. A discrete-time
Dowling, A.W., Kumar, R., Zavala, V.M., 2017. A multi-scale optimization framework
scheduling model for continuous power-intensive process networks with vari-
for electricity market participation. Appl. Energy 190, 147–164.
ous power contracts. Comput. Chem. Eng. 84, 382–393.
Du, J., Park, J., Harjunkoski, I., Baldea, M., 2015. A time scale-bridging approach for
Zhao, S., Ochoa, M.P., Tang, L., Lotero, I., Gopalakrishnan, A., Grossmann, I.E., 2019.
integrating production scheduling and process control. Comput. Chem. Eng. 79,
Novel formulation for optimal schedule with demand side management in mul-
59–69.
ti-product air separation processes. Ind. Eng. Chem. Res. 58 (8), 3104–3117.
Flores-Tlacuahuac, A., Grossmann, I.E., 2006. Simultaneous cyclic scheduling and
Zhou, D., Zhou, K., Zhu, L., Zhao, J., Xu, Z., Shao, Z., Chen, X., 2017. Optimal schedul-
control of a multiproduct CSTR. Ind. Eng. Chem. Res. 45 (20), 6698–6712.
ing of multiple sets of air separation units with frequent load-change operation.
Gupta, D., Maravelias, C.T., Wassick, J.M., 2016. From rescheduling to online schedul-
Sep. Purif. Technol. 172, 178–191.
ing. Chem. Eng. Res. Des. 116, 83–97.
Zhu, Y., 2001. Multivariable System Identification for Process Control. Elsevier Sci-
Hahn, J., Edgar, T.F., 2002. An improved method for nonlinear model reduction using
ence, Oxford, UK.
balancing of empirical Gramians. Comput. Chem. Eng. 26 (10), 1379–1397.
Zhuge, J., Ierapetritou, M.G., 2012. Integration of scheduling and control with closed
Hinkley, D.V., 1969. On the ratio of two correlated normal random variables.
loop implementation. Ind. Eng. Chem. Res. 51 (25), 8550–8565.
Biometrika 56 (3), 635–639.
Zhuge, J., Ierapetritou, M.G., 2015. An integrated framework for scheduling and con-
Kelley, M.T., Pattison, R.C., Baldick, R., Baldea, M., 2018. An MILP framework for op-
trol using fast model predictive control. AlChE J. 61 (10), 3304–3319.
timizing demand response operation of air separation units. Appl. Energy 222,
951–966.