7.

1 Bringing it all together in

Batch Process Engineering
Luuk van der Wielen, UL & TU Delft, Adrie
Straathof (TU Delft) & GTA team

jan 2024 | v3
 Essence of batch process designs
• ‘batch processes’ with multiple parallel and subsequential, independent and dependent
actions (whole or part of chem & pharma industries, but also transport, freight logistics,
computers, university time tables ..). Central question: how does process
performance depend on the scheduling of all these actions ?

• Complexity of scheduling problems (and their solutions) depends on process

structure (topology), equipment assignment and connectivity, inventory storage policies,
material transfer, batch sizes and processing times, demand patterns, changeovers,
resource and time constraints, and costs.

• Process performance: measured in quantitative (economic/ financial, environmental,

…) impacts, as well as qualitative (perception, acceptance, … as indicated with
‘feasible’, ‘high’, ‘good’, ‘low’ etc) impacts. Objective function needs to be defined early.

• In general, the process is limited by some technological, logistical, financial constraints

that also need to be defined early on.
Definition of (complexity of) scheduling problems:

• Set of N products / product batches to be produced.

• Set of M processing units

• Sequence for each product to be processed

• Fixed or variable (batch size) set-up, processing and transfer times

• Constraints related to demand or resource (labor, utilities)

• Nature of intermediates storage between processing steps (for pharma, biotech, food
especially due to stability)

• Processing network – multipurpose (job-shop) or multiproduct (flowshop)

 Simple Economic Analysis of Multi product batch plants
– using ‘economic potential’ (renevue-costs)
(UD’14) Example 5.1: A batch product manufacturer sells products A and B. The
profit from A is $12/kg and from B $7/kg. The available raw materials for the products
are 100kg of C and 80 kg of D. To produce one kilogram of A, the manufacturer needs
0.5kg of C and 0.5kg of D. To produce one kilogram of B, the manufacturer needs
0.4kg of C and 0.6kg of D. The market for product A is 60kg and for B 120kg. How
much raw material should be used to maximize the manufacturer’s profit?
Optimization of ‘simple’ multiproduct plants
requires ‘simple’ tools as linear programming* (LP)

Mathematics for maximising profit by

selecting A (x1) and B (x2) in LP form

max Profit z = 12 x1 + 7 x2
0.5x1 + 0.4x2 ≤ 100 available C
0.5x1 + 0.6x2 ≤ 80 available D
x1 ≤ 60 and x2 ≤ 120
x1, x2 ≥ 0

From U Diwekar Batch Processing (2014) * LP is standard tool in Excel (Solver Add-In)/DynoChem
 Elaborate LP example in
Excel (manually)

Calculate objective function (max

profit Z F23 in example) in Excel as a
function of variable cells F25 (x1) and
F26 (x2).

Create a table x1, x2 and Z manually

for realistic values (x1,x2 >0) and with
market limits (x1≤60 and x2 ≤ 120).

• For which x1, 2 is Z optimal ?

• Are there other maxima ?

• Do all combinations satisfy the set

of constrains ?

Excel can do this automatically !

Elaborate LP example in
Excel/DynoChem Solver

Translate LP model for maximising

profit (F23) by changing variable
cells F25 (x1) and F26 (x2)
 Multi product plants – ‘simple’ non-linear cases

Complexity goes
rapidly up when
systems are non-
linear and involve
discrete decisions
(yes or no; 0 or 1).

Mathematics and
software tools can’t
be missed then !
From U Diwekar Batch Processing (2014)
 Batch Scheduling & Planning – multi
purpose / multi product plants

Optimization of ‘simple’ scheduling requires

common sense – ‘standard’ software tools
(DynoChem,SPD) to calculate & visualise

Job-shop - different products follow

different paths for production.
Flow-shop – different products follow
same path.

From U Diwekar Batch Processing (2014)

Industrial scheduling cases become quickly very complex –
example Resin manufacturing in flowshop

Optimization of complex scheduling

requires complex mathematical model –
nowadays ‘standard’ software tools.
1. Indicate Batch time to complete full ‘recipe’.
2. What determines Cycle time (start of next cycle) ?
 Strategies to solve scheduling problems:

• The Pragmatic Approach …. Critical Path Analysis & Gantt Charts

• Simplifying / approximating / linearising the problem (objective function, constraints,

parameters) to one that can be solved by a (linear) scheduling (optimisation)
algorithm as LP or by scheduling policies. Focus CG5052

• Exact algorithms. Scheduling problem can be formulated as a mixed integer linear

programing (MILP) problem, for instance recurrence relations which are expressions to
define start/finish times of each product in each processing stage.

• Using AI, probabilistic methods or heuristics such as genetic algorithms, which are based
on evolution. Starts with random population (Monte Carlo), where individuals are
evaluated and selected on their ‘fitness’ to produce a new generation that that form a
better situation (solution).

• Flowsheet synthesis methods that include (non-linear) equipment and utilities design –
these are computationally heavy mixed integer non-linear programing (MINLP) problems.
Scheduling ‘policies’ – readily implementable no
guarantee of economic or other optimum

• First Come First Serve (FCFS) Scheduling Algorithm. ...

• Shortest Job First (SJF) Scheduling Algorithm. ...
• Longest Job First (LJF) Scheduling Algorithm. ...
• Priority Scheduling Algorithm in OS. ...
• Round Robin Scheduling Algorithm in OS. ...
• Shortest Remaining Time First (SRTF) Scheduling Algorithm.
 Critical Path Analysis
The critical path method, or critical path analysis, is an algorithm for scheduling a
set of project activities. A critical path is determined by identifying the longest stretch
of dependent activities and measuring the time required to complete them from start
to finish. CPM helps to determine
• the minimum time in which the project can be completed, reducing delays and
increasing efficiencies.
• the sequence and dependencies of activities to be completed on time to complete
the project in time.
• Float calculation: which tasks can be delayed without delaying the project
completion time. Improves resource distribution.

From https://www.workamajig.com/blog/critical-path-method
 Critical Path Analysis (2)
1. Identify activities based on Work Breakdown Structures (hierarchy deliverables)
2. Identify all dependencies
3. Create a network diagram (equivalent to ‘block scheme’)
4. Estimate the (weighted) activity duration (when known: short/average/long)
5. Calculate the Critical Path – longest stretch of dependent activities
6. Determine Floats – difference with 2nd longest stretch

From https://www.workamajig.com/blog/critical-path-method
 Critical Path Analysis (3)

Outcome of 1-5… 6. Determine Floats –

difference with 2nd longest stretch

Activities B, C, D and E are 2nd longest sequence

From
https://www.workamajig.com
/blog/critical-path-method
(90 hours). E is common, so non-common activities
B, C, D in 2nd sequence - have a float of 10.
This allows to allocate resources to Critical Path.
 Gantt Chart builds on Critical Path Analysis
1. Lists activities based on Work Breakdown Structures (hierarchy deliverables)
2. Indicates all dependencies
3. Creates a network diagram (‘block scheme’)
4. Indicates the (weighted) activity duration (short, average, longest)
5. Indicate the Critical Path – longest stretch of dependent activities = batch time
6. Indicate Floats – options for improving resource efficiency, ‘best’ cycle time.

From Mab process tutorial in SPD : critical path dominated by BR-101 (and RBS101)
 From the Mab process tutorial

A 2nd cycle can’t start earlier due to the bottleneck,

but it does start before bottleneck equipment is finished

Gantt chart also gives detail per operation.

 Tutorial on scheduling

• Batch time ?

• Cycle time ?

• … batches / year ?

• Botteneck equipment ?
Throughput adjustment (for scale-up)
Debottlenecking / scaling-up tutorial
V-104
• Example in which 5

staggered (fed)batch

bioreactors (FR-101)

have been added such

that V-105 has become

the bottleneck.
 Manufacturing of an active
pharma ingredient (API) for
skin care applications.
1. Indicate Batch time to complete full
‘recipe’
2. What determines Cycle time (start
of next cycle) ?

From DP1 D. Petrides, D. Carmichael

and C. Siletti. BATCH PROCESS
SIMULATION. By Intelligen Inc.
Chapter 15 of Diwekar, U. (2014).
 Manufacturing of an active
pharma ingredient (API) for
skin care applications.
1. Indicate Batch time to complete full
‘recipe’. Start of first, to end of last.
2. What determines Cycle time (start of
next cycle) ? Usually longest duration.

From DP1 D. Petrides, D. Carmichael

and C. Siletti. BATCH PROCESS
SIMULATION. By Intelligen Inc.
Chapter 15 of Diwekar, U. (2014).
R-102
 Monoclonal antibody manufacturing

1. Indicate Batch time to

complete full ‘recipe’.
2. What determines Cycle
time (start of next cycle) ?

DP2 Petrides, D. Bioprocess Design & Economics Book Chapter, DP3 A. Toumi, C. Jurgens, C. Jungo, B.A. Mier, V. Papavasileiou
in Roger Harrison, Paul Todd, Scott Rudge and Demetri Petrides and D. P. Petrides. Design and Optimization of a Large Scale
“Bioseparations Science and Engineering”, Oxford University Biopharmaceutical Facility using Process Simulation and Scheduling
Press in 2015 (ISBN 978-0-19-539181-7). Tools. Pharma. Eng. (2010) Vol 30(2). ISPE.org.
 Batch & Cycle Times: staggering and labor demand

DP6 D. Petrides, A. Koulouris, C. Siletti. Throughput Analysis and Debottlenecking of Product

Formulation and Packaging Lines. SPD use to model, visualize, and optimize product formulation
and packaging lines for food, consumer product, pharmaceutical, beverage, etc industries to
reduce cycle times and increase plant throughput. (part of U Diwekar (2014), Chapter 15.
 Optimizing real Industrial scheduling cases can be done !

1. Indicate Batch time to complete full ‘recipe’. ‘Job shop’

2. What determines Cycle time (start of next cycle) ? ‘Job shop’

Optimization of complex scheduling

requires complex mathematical model
– nowadays ‘standard’ software tools.

Also for non-technical aspects such as

labor, recruiting, upskilling ….
Further model based strategies to
process optimization, scheduling
and synthesis problems.
(beyond CG5052)

Linear Non-linear Integer (Discrete) Mixed

Objective All linear Non-linear

function,
constraints
Variables Scalar, continuous Scalar, continuous Scalar, integers

example scheduling, planning, equipment and control, optimisation, discrete

blending operations design dynamic operation . optimisation
MILP X X scheduling,
planning,
MINLP X X X process synthesis
Optimal X X X X
control
Tutorial: Debottlenecking in the (simplified) Mab case
 Debottlenecking / upscaling Mab process tutorial SPD
• Starting point: Brigthspace’s “2019 solution” of Mab process, producing 213 kg Mab
per year
• Use task “Adjust throughput flow” to produce 4000 kg/year. Note that you need
occasional scheduling.
• Minimise total equipment costs (“FOB costs”) merely by optimising scheduling.
• Award for 1st team with lowest costs, if according to rules. Mention us your equipment
costs; if lowest number, then provide bulletwise improvements made + SPD file.
• Plenary discussion at Tutorial time (qualifying submissions must be received before).

• If you did not finish this exercise before, rather continue with that.
• NB by adding extra equipment, process throughput and costs are increased
• NB there can also be resource bottlenecks (not discussed) due to limited availability
of raw materials, labor, utilities, …
Brightspace

• Central communication tool

• Updated schedule
• Hand-out and Submission of Assignments (Reports)

Reference materials
• Slidedecks and (links to) recorded lectures (videos)
• Diwekar, U. (ed.) (2014-02-25). Batch Processing – Modelling
and Design. CRC Press, (2014) ISBN 978-1-43986119-6.
https://doi.org/10.1201/b16527. £62 eBook or paperback
• E-Binder Intelligen (SPD developer) reading materials or
downloadable at https://www.intelligen.com/download/literature/
• Useful text on Bioseparations (complements Diwekar):
• J.A. Wesselingh, J. Krijgsman, Downstream processing in
biotechnology, 2013, original Delft Student Press Publication
• Few remaining hard copies @ € 18/each. Basis of DSP slides.