The TemperatureScanning PlugFlow Reactor

1
The Temperature-ScanningPlug-Flow Reactor
2
Kinetic measurements the way you alwayswanted
them - FAST and EASY
All of us who study reaction mechanisms using
kinetics, all who test and evaluate catalysts,
who need reliable rate expressions for reactor
design or simulation, know how tedious and
expensive it is to gather the necessary rate
data. Well, here is the solution to these
problems!
3
A New Kinetics Instrument
The Temperature Scanning Reactor A TSR is not
simply a laboratory reactor, it is a kinetics
instrument, capable of establishing the rate
parameters of a rate expression after less than a
working days operation - temperature
coefficients and all - for any candidate rate
model you may wish to propose.
4
Why do Kinetic Studies ?

• Kinetic studies are essential to the
  understanding of reactions. They yield an
  appropriate reaction rate expression. If we know
  the rate expression, we can
• Design better catalyst formulations
• Draw inferences on the mechanism of the reaction
• Quantify the rate of reaction for process
  simulation
• Improve reactor control and design
• Look for optimum reaction conditions

5
Purpose of the TS-PFR

• The TS-PFR can be used to obtain overall
  reaction rates, under non-steady-state
  conditions, at commercially important
  temperatures and pressures. In much less time
  than by conventional means, we are able to
  determine
• Reaction rates.
• Reaction rate coefficients.
• Temperature dependence of rate coefficients.

6
Catalytic Reaction Rates

• Gas-phase catalytic reaction rates are usually
  measured in
• isothermal, or, less commonly,
• adiabatic, plug-flow reactors.
• These two thermal regimes are difficult to
  implement experimentally, especially for highly
  exo/endothermic reactions.

7
Catalytic Reaction Rates

• With the introduction of the TS-PFR, we no
  longer need to operate under idealized and
  hard-to-implement thermal conditions.
• By operating in conformity with certain boundary
  conditions, the TS-PFR can determine an
  arbitrarily large number of reaction rates in one
  (say, 8-hour) TSR experiment, without waiting
  for steady state to be established.

8
Mode of Operation of the TS-PFR

• To do a TSR RUN in the TS-PFR one must
• 1. Load the reactor with catalyst.
• 2. Establish a feed rate at an initial feed
  temperature.
• 3. Ramp the temperature of the feed at a selected
  rate.
• 4. Measure the output composition and
  temperature.
• 5. Terminate the ramp after a pre-selected time.
• One TSR EXPERIMENT requires steps 2 to 5 to be
  repeated using exactly the same ramping procedure
  at a number of feed rates.

9
Mode of Operation of the TS-PFR

• It is important to realize that the data
  obtained by performing steps 1 to 5 only once
  i.e., doing just one TSR run, does not lead to
  interpretable results.
• The TS algorithms must be applied to several
  such rampings, done under appropriate boundary
  conditions. Only data from such a TSR
  experiment will allow the extraction of valid
  reaction rates.
• The rates obtained in this way are the same
  rates one would obtain from conventional
  isothermal experiments.

10
Boundary Conditions

• The boundary conditions which must be obeyed are
  simple to implement
• 1. Exactly the same ramping sequence must be
  used in each run of a TS-PFR experiment
• 2. The reactor must be of uniform effectiveness
  along its length
• 3. Each run must begin with the reactor at the
  same condition throughout the bed (e.g., at
  steady state for the initial conditions)
• 4. The temperature of the surroundings with
  which the reactor exchanges heat must be
  controlled in the same manner from run to run

11
Hardware - General

• The TS-PFR consists of
• Reactor Module Ì Analysis Module

12
Hardware

• The ANALYSIS MODULE may consist of a Quadrupole
  Mass Spectrometer, or any suitable analytical
  instrument.
• The REACTOR MODULE contains hardware, such as
  flowmeters, pressure transducers, the reactor and
  its oven, etc.

13
Software - Reactor Control

• The REACTOR MODULE is controlled by SERs
  CONTROL SOFTWARE, programmed for a Windows
  environment.

14
Software - Reactor Control

• The CONTROL SOFTWARE presents tabs. For
  example, the temperature ramping tab requests
  specifications for each run
• the feed rate
• temperature ramping rate
• initial temperature
• final temperature.

15
Software - Reactor Control

• The CONTROL SOFTWARE displays, in real-time, the
  measured values of
• temperatures.
• effluent composition.
• These values are presented on strip charts, and
  logged to disk.

16
Software - Analysis

• The ANALYSIS SOFTWARE controls the Mass
  Spectrometer, deconvolutes the gas composition,
  and sends the results to the CONTROL SOFTWARE.
• Up to sixteen individual components can be
  tracked in the reactor output stream.

17
Software - Interpretation

• The raw data collected from the complete TSR
  experiment are collated and sent to the TS rate
  extraction program. There, rates are extracted at
  selected conditions and used to form X -T- r
  triplets.
• This kinetic data is then downloaded to a
  spreadsheet where the proposed rate expressions,
• r f(X,T),
• can be fitted to the data.

18
Software - Interpretation

• If the rate expression is known, the spreadsheet
  solver is used to evaluate the rate parameters
  using the extracted X -T- r triplets of the
  kinetic data set, the known rate expression, and
  statistical tools.
• If the rate expression is not known, the kinetic
  data set is made available for fitting to
  candidate rate expressions. The success of the
  fit is judged on the basis of the statistics
  supplied by the solver, and by other tests
  appropriate to the system.

19
TSR Simulation

• SER has developed a Temperature-Scanning Reactor
  Simulator, capable of simulating a batch (TS-BR),
  CSTR (TS-CSTR) and plug-flow (TS-PFR) reactor
  operating under temperature-scanning modes. All
  physical aspects are taken into account i.e.,
  reactor materials, all heat transfer processes
  within the reactor system, heat of reaction, etc.
  
• Important note there are no restrictions placed
  on heat transfer in the operation of a TS-PFR.
  We can simulate this point in detail and observe
  the effect on the reaction rates as they are
  extracted by the TS-PFR algorithms.

20
Simulation - Heat Transfer Constants

• In the differential equations describing the
  behaviour of each of the TS reactors, there are
  constants (ki) in the heat balance equations
  which are calculated as functions of the real
  physical properties of the materials envisioned
  for the components of the simulated reactor.
• In a similar vein, all other physical aspects of
  the reactor system, such as mass of materials,
  pressure drop, etc. are included in the
  simulation using realistic values from
  established correlations.

21
Simulation - Rate Equation

• For simulation purposes, a generic
  Langmuir-Hinshelwood gas-solid catalytic reaction
  rate, with adsorption terms for both products and
  reactants, was used to model the kinetics. Each
  rate parameter was assumed to follow the
  Arrhenius temperature behaviour

22
Simulation

• The following data is from such a simulation.
  This allows us to examine a wider range of
  conditions than are approachable in any one
  reaction system.
• In this way we examine, in one unified picture,
  the many phenomena which can arise in all
  systems, but all of which rarely arise at
  approachable conditions in any one system.

23
Simulation - Conversion

• By simulating a TS-PFR, including all heat
  transfer effects, and using the
  Langmuir-Hinshelwood kinetics, conversion as a
  function of clock-time was calculated, and is
  shown below.

24
Conversion

• The raw data shows increasing conversion at each
  constant space velocity as clock time, and
  therefore feed temperature, increases. The lower
  the space velocity, the longer the space time,
  and hence conversion increases more quickly at
  low space velocities (i.e. long space times).
• Notice the data is obtained in a continuous
  fashion. This will allow us to remove error (we
  will call it noise) using sophisticated
  mathematical routines called filters.

25
Simulation - Outlet Temperature

• The corresponding reactor output temperature as
  a function of clock-time is also calculated.
  Notice that the exothermicity of the reaction
  causes the output temperature to differ from the
  input temperature, which followed the upper curve
  in each case.

26
Simulation - Outlet Temperature

• The reactor output temperature is a function of
  clock-time due to
• a) the temperature ramping, and
• b) the exothermicity of the model reaction.
• The output temperature will differ from reaction
  to reaction and confirms the non-ideality of the
  reactor.
• The theory of TSR operation describes how this
  non-ideality can be removed so that correct
  reaction rates can be calculated from TSR data.

27
Simulation - Re-mapping

• By taking X and T points at the same space and
  clock times, we can re-map the data from the
  last two slides onto the reaction phase plane, X
  vs T

28
Re-mapping

• It is in this plane that the presence of
  non-kinetic influences is detected. Catalyst
  aging, diffusion effects and any such distorting
  influences are readily discovered by this
  re-mapping of the raw data. In most cases these
  effects can be quantified by pursuing an
  appropriate experimental program.
• The essence of data treatment and of the
  understanding made available by TSR
  experimentation lies in such re-mappings of the
  data collected.

29
Simulation 3D - mapping
After just ten runs the experimental data
presents enough information to delineate a smooth
(T, t, X) surface allowing for accurate
interpolation.
30
3D - mapping
The curves on the X,T plane are contour lines
from this three dimensional surface. With ten
curves or so there is usually enough data to
delineate the full (T, t, X) surface. This
is the source of the unlimited data available
from a TSR the data are obtained in continuous
fashion allowing sophisticated two-dimensional
filters to construct a smooth surface. The
smooth surface in turn allows any point within
its confines to be accessed.
31
Surveying the Reaction Surface

• We now see that there exists a reaction
  surface which we survey using a chemical
  reactor. In the case of an isothermal PFR, one
  can measure any point on this surface, but in the
  case of a TS-PFR one must follow prescribed
  traverses which restricted movement during the
  survey.

32
Surveying the Reaction Surfaceby Isothermal
Operation

• The conventional isothermal method of surveying
  involves taking a small set of readings, at
  isolated points, along a few isothermal
  traverses. Each of these data points represents
  an independent measurement, with its own error.
• In conventional studies the 10 to 20 points
  collected in this way are used to estimate the
  shape of this surface and then to fit a rate
  equation which reproduces this shape.

33
Surveying the Reaction SurfaceThe TSR Operation

• In contrast, the operation of a TS-PFR is like
  an extensive satellite survey of the reaction
  surface.
• The TS-PFR does this by taking numerous
  prescribed traverses over the surface.

34
Conversion Vs. Residence Time

• On this surface, TSR theory allows us to
  identify the Operating Lines for this reactor.
  Some operating lines are shown below, and are the
  directions on the surface that yield the correct
  rates of reaction.

35
Conversion Vs. Residence Time

• By identifying traverses along the Operating
  Lines for this surface, we can construct the
  correct plots of X vs t. From this data we
  evaluate the reaction rates
• r dX/d?t

36
Extraction of Rates

• To collect the set of (X -T - r) triplets
  required for fitting to a rate expression we must
  therefore
• spline the discrete X-t data using a suitable
  spline function.
• evaluate dX/dt at the desired values of X and t.
• read the outlet temperature at the corresponding
  X and t from the (T, X)t curves.

37
The Triplets

• In this way we form the (X,T,r) triplets
  necessary for the fitting of a rate expression.
  Each
• (X - T - r )
• triplet collected in this way contains all the
  values necessary to fit a rate expression. We can
  collect an arbitrarily large number of such
  triplets from each experiment. With these we can
  proceed to search for the appropriate rate
  expression, to establish its rate parameters, or
  to examine its behaviour visually.

38
The Rate Surface
Now we can construct many new plots, such as
that of r vs X at various space times. Clearly
this is the same data as that discussed before
but seen in a different projection. Much of the
data processing in TS operations consists of such
re-mappings.
39
Outlet Temperature

• Alternatively, we can plot the values of reactor
  outlet temperature observed at various
  conversions and space times, and so on

40
Re-Mappings

• In fact, each data point has associated with it
  the dimensions of
• Conversion
• Inlet Temperature
• Outlet Temperature
• Space Time
• Clock Time
• Reaction Rate
• We can therefore examine TSR data in a large
  variety of 2D and 3D presentations.

41
The Rich Harvest of Rates

• As many of the (X -T - r) triplets as we may
  wish to have are made available by the procedure
  of defining a smooth surface using the dense mesh
  of raw experimental data.
• We now proceed to sieve out appropriate sets
  of data for model-fitting, or any other purpose.
• For example, we could select sets of isothermal
  (constant T) or sets of isokinetic (constant r)
  X-T-r data.

42
Isokinetic Rates

• Here we show isokinetic rates, extracted by
  these procedures from the TSR data. These rates
  are shown overlaid on the corresponding constant
  rate curves generated by the kinetic expression.
  The fit is good.

43
Isothermal Rates
Similarly, unlimited sets of rates, at various
constant temperatures, can be extracted from the
same data. These rates are used for the fitting
of isothermal forms of the rate equation.
44

Accessible Region

• Many more rates can be extracted in this way.
  The grayed-out area on this graph depicts the
  area of the reaction phase plane which we have
  investigated by the TSR experiment and from
  which we can now extract any reaction rate we
  wish.

45
Range of Accessibility

• The grayed-out area presents all the data that
  can be obtained using this PFR.
• The lower bound is defined by a run at a space
  velocity which causes maximum tolerable pressure
  drop through the catalyst bed.
• The upper bound is at a space velocity which is
  at a Reynolds number on the brink of transition
  to laminar flow.
• Between these two limits lies all of the
  performance space accessible to this plug flow
  reactor, for this reaction, regardless of the
  mode of operation.

46
CO Oxidation

• Experimental results, using the real TS-PFR
  described previously, were gathered in a study of
  the catalytic oxidation of carbon monoxide,
  performed on a proprietary automotive catalyst.
• The raw data and results are presented in the
  following slides.

47
CO Oxidation - Results

• Shown here are experimental curves for the
  oxidation of CO as a function of clock time at
  various flow-rates. Note the similarity of this
  data to a truncated section of the simulated
  curves shown earlier.

48
CO Oxidation - Results

• The measured outlet temperature is shown as a
  function of clock time. In this case heat
  transfer was such that the outlet temperature
  tracked the inlet temperature fairly closely for
  most of the ramp i.e. the reactor was isothermal
  up to high conversions.

49
CO Oxidation - Results

• As previously described, by re-mapping the data
  we can produce the X vs T curves in the reaction
  phase plane, as shown below.

50
CO Oxidation - Interpretation
The experimental conversion vs. space time data
must be smoothed, or filtered, in order to
extract rates successfully, since numerical
derivatives of dX/dt will have to be calculated.
The taking of point to point differentials from
noisy data amplifies the noise present in the
original data and will distort the
interpretation. Within the TSR interpretation
software, several data-filtering techniques have
been made available. Many more filters are
available in the literature.
51
Filtering

• The many mathematical filters available are
  designed to deal with specific types of noise,
  each has special merit in particular
  circumstances
• FFT smoothing in the frequency domain
• Savitsky-Golay filtering and
• 2-Dimensional surface smoothing using
• least - squares splines.
• Are available in the TSR software. Once the data
  have been properly filtered, the interpretation
  techniques described earlier can be applied.

52
CO Oxidation un-Filtered Surface

• The un-filtered surface is wrinkled, and slopes
  taken off this surface will produce very
  scattered rates.

53
CO Oxidation The Filtered Surface

• The filtered surface is smooth, though its
  underlying shape has been preserved. Slopes taken
  off this surface will produce consistent rates
  which will in turn generate a smooth rate surface.

54
CO Oxidation The Kinetic Surface

• Once the filtering is done, the (X,T,r) surface
  is smooth. Notice that the surface in this case
  is largely featureless. This is often the reason
  for the difficulty in finding a unique rate
  expression. Many rate equations can approximate
  such a surface.

55
CO Oxidation The Rate Fitting

• Although this particular kinetic surface is
  featureless, it is still generated by a unique
  rate expression.
• In order to identify this expression we need to
  have as much of the surface surveyed as possible,
  and it must be as smooth as possible.
• These are the reasons why the TSR produces
  results which are greatly superior to those
  obtained by traditional isothermal
  experimentation.
• The TSR produces much more data
• The data can be smoothed in a rational way.

56
CO Oxidation The Rate

• In this case it was possible to identify the
  dissociative model of oxygen adsorption as the
  one whose rate expression gives the best fit to
  the data.

57
CO Oxidation The Rate

• This model contains six Arrhenius parameters, as
  well as the two exponents defining the adsorption
  regimes for oxygen and carbon monoxide.
• 
  
  
• Parameter Value
• Ar atm s-1 1.443?1016
• Er J/mol 1.462?105
• 
  
• ACO atm-1 6.832?101
• ?HCO J mol-1 -7.495?103
• 
  
• AO2 atm-1 1.991?10-6
• ?HO2 J mol-1 -8.299?104

58
CO Oxidation The Goodness-of-Fit

• One way of evaluating the goodness-of-fit is to
  look at the parity plot between calculated and
  experimental rates. The fit is clearly excellent.

59
How Robust is this Procedure ?

• If we obey the boundary conditions, then
• heat transfer between system components
• ramping rates
• pressure drop, and
• the absence of a thermal steady state
• will not affect the TS algorithm.
• We will always extract the correct reaction
  rates.

60
Conclusions - What are the Benefits ?

• Compared to a conventional reactor operating at
  isothermal conditions, an automated TS-PFR gives
• a very large number of filtered reaction rates,
• in a very short time.
• As a consequence, data are easier and cheaper
  to acquire. One need no longer be satisfied with
  the limited information available at a standard
  test condition.

61
The Temperature-Scanning Plug-Flow Reactor

• The Kinetics Instrument