The manufacturing of pharmaceutical products is preceded by complex development phases where process design, planning, and scheduling problems must be considered as deeply linked, fact that has not yet been adequately handled by the existing literature. In this perspective, this work discusses the role of the production planning and scheduling decisions in the pharmaceutical industry. It starts by analysing the main aspects that influence planning and scheduling, and defines an extended scope of the related problems, as a way to account for higher levels of integration between process design and operational decisions. We propose a novel conceptual representation, the Delivery Trade-offs Matrix (DTM) to help managing the trade-offs occurring in the drug development process and to expose the factors that affect the performance of these manufacturing systems.
2. June 2015 PSE2015/ESCAPE25 2
Agenda
1. The pharmaceutical industry
2. Critical factors for planning and scheduling
3. Scope of planning and scheduling problems
4. Delivery Trade-offs Matrix
5. Challenges and opportunities
3. June 2015 PSE2015/ESCAPE25 3
1. The pharmaceutical industry
• The shortage in new drug approval applications
• The uncertainty associated to Research and Development
(R&D) and trials I-III phases
• The pressure of generic drugs
Markets tend to become more complex over time, forcing companies to increase
their responsiveness, both in terms of time and cost
4. June 2015 PSE2015/ESCAPE25 4
1.1. Some figures …
• Developing cost grew 481% to $802 million from 1970s to
1990s (DiMasi et al., 2003; Hynes III, 2009)
• The duration of the development cycle remained fairly
stable (roughly 15 years) (Kessel,2011)
• The production of the APIs is considered the rate-limiting
step of the supply chain (Shah, 2004)
The current worldwide
paradigm imposes a reduction
to less than 10 years (Federsel,
2009)
5. June 2015 PSE2015/ESCAPE25 5
1.2. Industry trends
• Adoption of continuous processes (Roberge et al., 2005)
• Process Analytical Technology (PAT) (McKenzie, 2006)
• Advanced optimization tools (Grossmann, 2005)
Production planning and scheduling are systematically considered very difficult
functions to perform
6. June 2015 PSE2015/ESCAPE25 6
1.2. Pharmaceutical Industry System
Main activities
Research and development activities
Development and manufacturing of Active Pharmaceutical
Ingredients (APIs)
Drugs manufacturing
Primary
manufacturing
Storage
Distributers StorageHospitals, clinics,
pharmacies
Raw Material
suppliers
Storage
Storage
Research and
Development
Secondary
manufacturing
(Shah, 2004)
7. June 2015 PSE2015/ESCAPE25 7
1.2. Pharmaceutical Industry System
Main activities
Research and development activities
Development and manufacturing of Active Pharmaceutical
Ingredients (APIs)
Drugs manufacturing
Primary
manufacturing
Storage
Distributers StorageHospitals, clinics,
pharmacies
Raw Material
suppliers
Storage
Storage
Research and
Development
Secondary
manufacturing
(Shah, 2004)
8. June 2015 PSE2015/ESCAPE25 8
Issues
1.Manufacturing in a high
regulated market
2.High variability on the
demand
4.Pressure created by generic
drugs
2. Critical factors for planning and scheduling: Market
3.Large production mixes
in the manufacturing sites
Main Decision Areas
1.Pipeline management
2.Capacity planning
3.Supply chain management
(LaĂnez, Schaefer, & Reklaitis, 2012)
Shah (2004) and Buchholz (2010) pointed out that time-to-market is a critical
driver of the pharmaceutical industry
9. June 2015 PSE2015/ESCAPE25 9
Issues
1.Process topology: sequential
and network processes
2.Regulatory and quality
procedures/lots traceability
4.Re-planning and rescheduling are
frequent
2.1. Critical factors for planning and scheduling:
Processes
Main Decision Areas
1.Product Development
2.Process design/scale-up
decisions
3.Processes under development- limited
knowledge of the first batches 3.Advanced Process Control
10. June 2015 PSE2015/ESCAPE25 10
Issues
1.Resources characteristics
2.Batch versus continuous
manufacturing
2.2. Critical factors for planning and scheduling: Plants
Main Decision Areas
1.Plant design: grassroot and retrofit
2.Operating mode: short-term
versus campaign mode
3.Inventory management and
associated policies
3.Reduction of inventory
11. June 2015 PSE2015/ESCAPE25 11
3. Scope of planning and scheduling problems
Process Synthesis
Quantitative
specification of
physicochemical
materials manipulations
Planning and scheduling extended scope
Process
Synthesis
Process
scale-up
Process design Planning Scheduling
Production
execution and
control
“Units
independent”
recipe
Processing
times,
quantities
Recipe
+
Units
R&D
Operations Management
12. June 2015 PSE2015/ESCAPE25 12
3.1. Scope of planning and scheduling problems
Process Synthesis
Quantitative
specification of
physicochemical
materials manipulations
Planning and scheduling extended scope
Process
Synthesis
Process
scale-up
Process design Planning Scheduling
Production
execution and
control
“Units
independent”
recipe
Processing
times,
quantities
Recipe
+
Units
R&D
Operations Management
Process Scale-up and
Process Design
Develop the process
from the laboratory
scale to the industrial
scale
13. June 2015 PSE2015/ESCAPE25 13
3.2. Scope of planning and scheduling problems
Process Synthesis
Quantitative
specification of
physicochemical
materials manipulations
Planning and scheduling extended scope
Process
Synthesis
Process
scale-up
Process design Planning Scheduling
Production
execution and
control
“Units
independent”
recipe
Processing
times,
quantities
Recipe
+
Units
R&D
Operations Management
Process Scale-UP and
Process Design
Develop the process
from the laboratory
scale to the industrial
scale
Planning and
Scheduling
Resources allocation so
as to minimize costs and
increase the plant
output
14. June 2015 PSE2015/ESCAPE25 14
3.3. Scope of planning and scheduling problems
Process Synthesis
Quantitative
specification of
physicochemical
materials manipulations
Planning and scheduling extended scope
Process
Synthesis
Process
scale-up
Process design Planning Scheduling
Production
execution and
control
“Units
independent”
recipe
Processing
times,
quantities
Recipe
+
Units
R&D
Operations Management
Process Scale-UP and
Process Design
Develop the process
from the laboratory
scale to the industrial
scale
Planning and
Scheduling
Resources allocation so
as to minimize costs and
increase the plant
output
Production Control
Production dispatching,
control actions and
quality assessment
15. June 2015 PSE2015/ESCAPE25 15
3.4. Scope of planning and scheduling problems
The scope of the planning and scheduling functions must be extended to account for design
decisions, especially for manufacturing products that are under development
Planning and scheduling extended scope
Process
Synthesis
Process
scale-up
Process design Planning Scheduling
Production
execution and
control
“Units
independent”
recipe
Processing
times, quantities
Recipe
+
Units
R&D
Operations Management
Available alternative units
+ -
16. June 2015 PSE2015/ESCAPE25 16
3.5. Extended scope of planning and scheduling: How?
Flexible graphical representations of
process design and integration with
scheduling
(Moniz, Barbosa-PĂłvoa, Pinho de Sousa, &
Duarte, 2014)
17. June 2015 PSE2015/ESCAPE25 17
3.5. Extended scope of planning and scheduling: How?
Flexible graphical representations of
process design and integration with
scheduling
(Moniz, Barbosa-PĂłvoa, Pinho de Sousa, &
Duarte, 2014)
Frameworks for supporting the overall
decision-making process: product
portfolio, capacity planning, planning and
scheduling, process control and safety
and supply chain management
(Venkatasubramanian et al., 2006)
18. June 2015 PSE2015/ESCAPE25 18
Trials I-III
R&D
Drug
F
Drug
F
Drug ADrug A
Drug CDrug C
Drug BDrug B
Commercialization
Drug DDrug D
Drug
E
Drug
E
Understanding and modelling the
conditions for achieving the global
optimization of these
manufacturing systems is by itself a
hard task
The pharmaceutical industry
should focus on the manufacturing
and delivery issues knowing that
each phase of the development
cycle has different challenges
4. Delivery Trade-offs Matrix
Lot size
Few kilograms
Hundreds of kilograms
19. June 2015 PSE2015/ESCAPE25 19
Uncertainty
Low
High
CostHigh
Low
Short-term
planning
Campaign
planning
Time-to-market Amount delivered
Trials I-III
R&D
II
III IV
Drug
F
Drug
F
Critical leap
Drug ADrug A
Drug CDrug C
Drug BDrug B
Commercialization
I
Drug DDrug D
Drug
E
Drug
E
Exposes the issues that affect the
performance of the manufacturing
systems
Graphical representation of the
portfolio
Helps managing the trade-offs
occurring in the drug development
process
4.1. Delivery Trade-offs Matrix
Lot size
20. June 2015 PSE2015/ESCAPE25 20
Uncertainty
Low
High
CostHigh
Low
Trials I-III
R&D
II
III IV
Drug
F
Drug
F
Critical leap
Drug ADrug A
Drug CDrug C
Drug BDrug B
Commercialization
I
Drug DDrug D
Drug
E
Drug
E
Uncertainty and Costs
1.High uncertainty associated to the
drug structure and to the process design
2.Difficulties in estimating the required
time and resources
3.In the development and
manufacturing of APIs it is common to
allocate production resources 6 to 12
months in advance
4. Critical leap: the first scaled up batch
5. Challenges and opportunities
Opportunity: Simulation-optimization can provide interesting solution methods to combine
the stochastic and deterministic parameters of the planning and scheduling problem
Chen, Mockus et al. (2012), Eberle, Sugiyama et al. (2014), Sahay and Ierapetritou (2014)
Lot size
21. June 2015 PSE2015/ESCAPE25 21
Uncertainty
Low
High
CostHigh
Low
Time-to-market Amount delivered
Trials I-III
R&D
II
III IV
Drug
F
Drug
F
Critical leap
Drug ADrug A
Drug CDrug C
Drug BDrug B
Commercialization
I
Drug DDrug D
Drug
E
Drug
E
Time-to-Market and Amount
Delivered
1. The delivery of products to Trials I-III
phases is of extreme importance
2. Fast and robust scalability of the
production processes
3. The inventory in the supply chain
provides flexibility om what concerns to
delivery dates
5.1. Challenges and opportunities
Opportunity: development of comprehensive methods capable of addressing the long–term
dimension of the design and scale-up decisions
Varma, Pekny et al. (2008)
22. June 2015 PSE2015/ESCAPE25 22
Uncertainty
Low
High
CostHigh
Low
Short-term
planning
Campaign
planning
Time-to-market Amount delivered
Trials I-III
R&D
II
III IV
Drug
F
Drug
F
Critical leap
Drug ADrug A
Drug CDrug C
Drug BDrug B
Commercialization
I
Drug DDrug D
Drug
E
Drug
E
Operating Mode
1.Heterogeneous demand
2.Mixed planning strategies: short
term mode versus campaign mode
5.2. Challenges and opportunities
Opportunity: solution methods capable of addressing mixed planning strategies
23. June 2015 PSE2015/ESCAPE25 23
Final remarks
1. We group the critical factors for planning-scheduling decision-making in 3
categories: market, processes, and plants
2. We propose a representation of R&D and manufacturing trade-offs
3. The integration of process scale-up/design decisions with planning and
scheduling decisions is fundamental to do in the early stages of the drug
development cycle
4. Research addressing integrated decision-making, uncertainty, and knowledge
management is essential to solve extended planning and scheduling
problems
Industrial companies are continuously assessing their operations, as a way to increase the overall effectiveness of the production systems.
Markets where these organizations operate tend to become more complex over time, forcing companies to increase their responsiveness, both in terms of time and cost.
The case of the pharmaceutical industry is a good example on how market is driving the change of drug development cycle and manufacturing activities. Some of the most relevant driving factors are related to:
The current worldwide paradigm imposes a reduction to less than 10 years from pre-clinical development to commercialization (Federsel, 2009).
The critical factors that drive the planning and scheduling functions, in the particular context of the pharmaceutical industry, can be grouped in three categories: market, processes, and plants.
Market factors are related to the specific contextual factors of this industry.
Process factors have to do with the structure of the chemical processes.
Plant factors relate to the operating strategies and resources characteristics of the manufacturing systems
Regulatory agencies such as the US Food and Drug Administration (FDA) or the European Medicines Agency (EMA) impose strict regulations that go from the development to the manufacturing of drugs.
Manufacturing in a high regulated market has to deal with additional complexities that do not exist in less regulated markets.
Chemical processes are executed under a close supervision of the regulatory agencies that define procedures to monitor process changes.
Pipeline and development management—this involves the selection of potential drugs to develop further, and the planning of the development activity.
The production process topology strongly determines the scheduling models that can be applied.
In the manufacturing of APIs, for example, processes require numerous production steps with tasks having short and long processing times, usually spanning across several working shifts
Regulatory and quality procedures define the lot size and the changeover requirements that must be rigorously followed in the manufacturing sites, thus introducing additional time to the effective production time.
Stable intermediaries and final products are produced in lots, and therefore lots traceability must be ensured
The first batches after a scale-up are usually more difficult to produce, since this may involve the use of different processing units or even performing changes in the process. For that reason, these processes impose frequent revisions of the production schedule.
The plant structure has also implications on how planning and scheduling are performed.
The characteristics of the plants (such as resources, plant structure, operating mode, and batch/continuous manufacturing) lead as well to specific planning and scheduling problems.
Continuous manufacturing of pharmaceuticals is an emergent process mode that relies on flow reactors and is currently being evaluated for the production of drugs.
A consequence of using flow reactors, instead of batch reactors, is that the production process moves from a batch mode to continuous operating conditions (Buchholz, 2010).
Benefits of continuous manufacturing when compared to batch manufacturing include lower plant and production costs, lower carbon footprint; better quality, higher safety; less costs to scale-up, and higher levels of automation (Roberge et al., 2008; Calabrese & Pissavini, 2011) Nevertheless, existing technological challenges of flow reactors and adaptation of batch processes to continuous processes have made their evaluation and deployment difficult.
Process Synthesis, refers to the quantitative specification of physicochemical materials manipulations that take place, having as output a recipe that is independent of particular processing units
In other words, the recipe describes the chemistry steps required to manufacture the product.
This step complies with the development of the chemical process so as to pass from a laboratory scale to an industrial production dimension, resulting in the determination of the final product quantities (lot sizes) and an initial assessment of the processing times.
Planning and scheduling encompass then the coordination of development and production manufacturing activities so as to pass from the laboratory scale to the industrial scale, resulting in the determination of the final product quantities (lot sizes)
The goal is to find optimal schedules that maximize expected economic value of the investment by considering the resources availability, the probability of success of the clinical trials, and the associated costs.
Production Execution and Control involve the following activities production dispatching, control actions and quality assessment, among others.
The first steps are mainly associated to the R&D functions, while the OM deals essentially with planning and manufacturing.
Nevertheless, decisions should be performed collaboratively in order to ensure that decisions made at each department are properly considered.
Although the figure suggests a sequential and directional decision flow, the different steps are often overlapped and revisited whenever necessary.
We argue that planning and scheduling functions are extended in order to integrate some decisions made in the process scale-up and design steps.
The planning problem, either of long-term or short-term, benefit from considering decisions taken at the scale-up and process design levels, since these decisions have a direct impact on the
determination of the processing units suitable for the process, resulting into different production routes (alternative processes).
On the contrary, after schedule release to the shop-floor, changes on planning and scheduling decisions are very limited, although rescheduling is a common practice.
The same happens with changes in process design decisions that may not be possible or are not desirable to perform.
Despite the significant academic and industrial achievements in this area, there are relevant challenges that make the planning and scheduling decision-making particularly difficult to address.
To address that, in a previous work, we proposed a representation of R&D and manufacturing trade-offs, called Delivery Trade-offs Matrix (DTM)
At the start of a research program, products and processes have not yet been developed, and therefore there is a high uncertainty associated to the drug structure and to the production process. Uncertainty makes planning decisions more complex, since it is more difficult to estimate the required times and resources.
For example, in the development and manufacturing of APIs it is common to allocate production resources 6 to 12 months in advance.
With drug development the uncertainty tends to decrease as product and process characteristics are better understood.
The delivery of the first scaled up batch (usually between 1 to 5 kg), used to support toxicological and formulation studies, along with phase I trials, is on the critical path of the development process
This scale-up is particular difficult to perform since the knowledge obtained at the laboratory scale is seldom sufficient to guarantee a successful process at a plant scale (Federsel, 2009).
Moreover, the drug development process requires a series of scale-ups so as to develop an efficient production process.
At the commercialization stage, the need for API or drug products is normally in the order of hundreds of kilograms. The processes are well defined, thus the uncertainty is mainly associated to market parameters such as demand, and to the processing time of the complex production tasks.
The current practice demonstrates that there are large costs and high uncertainty at the R&D and trials I-III phases (see Figure 2 a), with the total estimated cost of bringing a new drug to market being larger than 1 billion dollars (Kessel, 2011).
The total cost of bringing a new drug to market is estimated to exceed 1 billion dollars (Kessel, 2011). In terms of the total cost structure, pharmaceutical R&D costs are around 30% to 35% and clinical trials (typically representing the most significant cost) can be between 35% to 40% of the total (Suresh & Basu, 2008).
Capturing all relevant sources of uncertainty in comprehensive frameworks for the supporting the planning and scheduling decision-making process is still a not solved problem. Simulation-optimization can provide interesting solution methods to combine the stochastic and deterministic parameters of the planning and scheduling problems.
Chen et al. [15] developed a simulation-optimization model for managing the entire trial supply chain, including the planning and scheduling of the Active Pharmaceutical Ingredient (API) manufacturing
Sahay, Ierapetritou [16] have applied an agent-based model hybridized with a linear programing model so as to study different supply chain decision-making policies
Eberle et al. [17] have applied a Monte-Carlo simulation so as to better estimate the production lead time of pharmaceutical processes.
Modeling the uncertainty of real world problems with optimization often results into models with a large number of variables and computationally intractable.
On the other hand, simulation can easily deal with the stochastic nature of the problems and with complicating constraints of the production systems.
Recent methods combine simulation and optimization to propose efficient solution methods.
Simulation can be used to assess solutions provided by deterministic optimization models, by considering the uncertain of the parameters, and to provide additional information to optimization models
It should be noted that from the planning and scheduling perspective, the delivery of products to Trials I-III phases is of extreme importance On the other hand, at the commercialization phase there is more flexibility concerning delivery dates, if there is inventory on the supply chain. According to Shah (2004), the whole pharmaceutical chain stock can represent 30% to 90% of the annual demand in quantity.
Therefore, at this phase, we can say that delivering the right product amounts is relatively more important than respecting delivery dates.
Remember that the production lot sizes at the Trials I-III phases are in the order of few kilograms, while after several scale-up and validation steps, the lot sizes are around hundreds of kilograms. After drug development, the manufacturing costs are lower and tend to decrease with the reduction of the root causes of variability in the production process.
Varma et al. [6] have developed a comprehensive decision-making framework called Sim-Opt for resource management that includes components for stochastic simulation, schedules generation based on a mixed integer linear programming (MILP) formulation, and evaluation of various resource strategies.
Concerning the operating mode, manufacturing sites run in short-term mode to fulfil a small product demand, or run preferably in campaign mode to respond to a regular demand.
Sometimes the short-term mode is also used for manufacturing products that are in commercialization, this naturally resulting in the production of a smaller number of lots.
However, in all cases the process must run with the same lot size as approved by the regulatory agencies.
All the above issues led the pharmaceutical industry to recognize the need for reducing time-to-market, the costs of new drug development, and the manufacturing costs.
The path to efficient R&D and manufacturing activities requires new ways to address uncertainty and reduce costs
Such path must focus on improving the reliability of the drugs delivery by dealing with the uncertainty and the associated costs and account with a heterogeneous demand
This will involve the introduction of new production technologies (Suresh & Basu, 2008), as well as the adoption of innovative process design, planning, and scheduling decision-making tools.
For example, according to Roberge et al. (2005), 50% of the reaction tasks in the chemical-pharmaceutical industry could benefit from the adoption of continuous processes based on the micro-reactor technology.
In what concerns decision-making, the relevance of applying optimization tools and deploying more integrated decision-making processes is being recognized by the industry, despite the challenges that still exist (Grossmann, 2012).
A reduced number of works propose solution methods capable of addressing mixed planning strategies