Approaches and Pitfalls in Reaction Safety Data...

APPROACHES AND PITFALLS IN REACTION SAFETY DATA COLLECTION

Approaches and Pitfalls in Reaction Safety Data Collection

Excellence in process safety performance requires having a wealth of the right information available and accessible, to support and underpin the overarching safety management system. This includes both asset information (P&ID’s, vessel register, instrumentation, piping and physical equipment inventory, safety systems, including safety instrumented systems, etc) and process information (material safety data, reactivity and process safety data, batch manufacturing records, quality / IPC procedures, raw material information, etc). Whilst the requirements for asset information and most process information relating to operations is reasonably prescriptive – and hence can be reliably audited and assured - the specification of required process safety data is far from prescriptive. This often leads to ineffective or inefficient safety management systems and can completely undermine otherwise excellent process safety management systems.

Dr Stephen Rowe, BSc (Hons), PhD, CChem, MRSC, AMIChemE, MIoD

Process safety data includes information on the relevant properties of materials and reactions in the context of the Basis of Safety. For any given unit operation – and indeed, any given piece of equipment – the basis of safety can be different for different processes. This potential “ambiguity” poses the biggest challenge to corporations in ensuring robustness in the basis of safety – and the process safety data requirements that underpin it. In our experience, organisations go about the collation of process safety data in one of several ways, as described below.

Approach 1: Complete Dataset

Some organisations place a blanket requirement to procure a complete set of data on all materials and processes, irrespective of the basis of safety.

Approach 2: Prescribed Dataset

It is not uncommon to find organisations that have a prescribed “list” of data requirements for new or existing materials or processes. The list is often generated from knowledge of typical plant configurations and the type of products manufactured where they are (assumed to be) consistent across a multinational organisation. The specific tests are dictated by the requirements of the Basis of Safety.

Approach 3: Flowchart Approach

In acknowledging that the Basis of Safety may vary between different handling or processing plants, some organisations use a flow chart to guide material testing towards an ultimate basis of safety. Examples of such flow charts can be found in Figure 1 for runaway reaction and thermal instability risks.

1 For more information, see the DEKRA Insight/Chilworth white paper “The Importance of Process Safety Data.”

Figure 1: Data Requirements for the Basis of Safety for Thermal Stability / Reaction Hazards

What are the characteristics of the

normal process?

What hazardous scenarios exist?

What are the consequences of

hazardous scenarios?

Can all hazardous scenarios be avoided?

Reaction Characteristics• Heat of reaction• Kinetics• Gas Generation

Material Stability Characteristics

• Onset temperature• Heat of decomposition• Kinetics• Gas Generation

Hazardous Scenario Characteristics

• Adiabatic simulation

Data Requirement

Decision Basis of Safety

Some

Significant

No

Yes

Minor

None

Process Control

Inherent Safety

Containment

Emergency Venting

Quenching

Reaction Inhibition

Emergency Cooling

Approach 4: Tailored (Case-specific) Approach

Having no prescribed dataset is not uncommon with data requirements for new materials / processes / plant equipment being identified on a case-by-case basis. Where appropriate procedures are followed by competent personnel (ie. those with relevant subject matter expertise), this can be highly effective and targeted. However, across an organisation its robustness becomes a function of local competency and hence it can be prone to inconsistency and resulting variable outcomes.

Approach 5: Prescribed Data + Situational Data Approach

Several key parameters plus selected additional parameters according to processing methods and risks are usually required to define a basis of safety. The key parameters provide initial characterisation of the hazards of the materials / processes involved. This data is utilised in a preliminary risk assessment – the outcome of which may be a requirement for further test data for specification of the basis of safety.

A 6th approach is to assume worst case material / process properties and failure cases and design and implement safety measures accordingly. While simple, this can lead to extreme inefficiencies and / or ineffectiveness.

The inherent advantages and disadvantages of each approach to process safety data collection is provided in Table 1.

Table 1: Advantages / Disadvantages of Various Approaches to Process Safety Data Collection

Approach Advantages Disadvantages

1. Complete Dataset • All bases covered at Day 1• No requirement for retesting for

additional parameters (i.e. supports all bases of safety)

• Transferable as a complete package to customers

• Inefficient and potentially expensive• “Data blindness” – too much data with

important information not clear• Complacency that “having all the data

makes you safe”

2. Prescribed Dataset • Reduced, targeted testing package – can be more cost effective

• Works well for consistent plant and process operations and basis of safety philosophy

• Stops the organisation thinking about data requirements.

• Non-standard / changed operations may require different data

• No “complete package” of hazards data for suppliers or customers

3. Flowchart Approach • Reduced, targeted testing package – can be more cost effective

• Encourages “thought” on the Basis of Safety – but encourages data consistency to underwrite it

• Non-standard / changed operations or equipment may require different data


• Origin of the flow chart may be lost over time – people don’t continually question its continued validity

• Complacency • Requires re-evaluation for plant changes

4. Tailored (Case-specific) Approach

• Reduced, targeted testing package – can be more cost effective

• Demands rigorous, independent assessment on the basis of safety for all operations


• Requires re-evaluation for plant changes• Strongly dependent on availability of

local competence and subject matter expertise

• Potential for inconsistency in application

Approach Advantages Disadvantages

5. Prescribed Data + Situational Data

• Reduced, targeted testing package – can be more cost effective

• Demands thought on the basis of safety for all operations

• Two-step process consistent with the basis of safety development process (characterisation, risk assessment, basis of safety specification)

• Strongly dependent on availability of local competence and subject matter expertise for situational data specification

• Requires re-evaluation of situational data for plant changes

• Can be more time consuming due to two-step nature of the process

• Can require testing that isn’t necessary in the specific situation

6. Assume Worst Case Characteristics

• Robust and reliable, provided the worst case selection is correct

• Potentially extremely expensive• Requirement for stringent control

systems and maintenance may adversely impact plant efficiency and effectiveness

• Compromised if the assumed worst case is not the real worst case

Of the approaches referenced, the tailored approach, where testing is based on the specific situation and what issues need to be addressed, provides the most relevant data in the most efficient test program. This approach requires design of the testing program by competent professionals who are able to assess the specific facility, operations or reactions and equipment in advance of any testing or risk assessment, and then refine data needs based on initial data acquired. If data requirements are not specified by competent professionals, this approach can result in incomplete or inappropriate data.

The prescribed plus situational approach is the next best choice for striking the balance between robustness, effectiveness and efficiency. The initial prescribed data meets the needs of highlighting the process safety characteristics of the material / process and feeds and informs the risk assessment phase of scale-up. This data must be collected in ALL cases. The subsequent situational data requirements derive from the outcome of the risk assessment (Process Hazards Analysis (PHA)) which focuses on the additional data requirements necessary to specify the basis of safety. The approach is best described schematically in Figure 2.

Common Pitfalls in Reaction Hazards Assessment

Due to the inherent differences in reactions, and the way they are performed, most companies don’t have a robust and rigid (prescribed) assessment process. This lack of rigour is generally more common for reaction scale-up than for powder and flammable gas / vapour handling operations where unit operations are more consistent. This leads to frequent short comings in reaction scale-up, with the most common identified below:

Use of the Wrong Equipment for Data Acquisition

• With a plethora of test methods available, operating companies typically acquire a relatively small number of techniques and attempt to collect all their data from their limited resources. Using DSC (Differential Scanning Calorimetry), for example, as a catch-all solution for thermal instability screening will be reliable for exothermic decomposition detection but will be wholly inadequate for self-heating property assessment, gas generation measurement or safe drying temperature specification. This over-reliance on a limited range of equipment commonly leads to data deficiency, making subsequent decision-making treacherous.

ProcessNew or

Modified

Initial Process

Safety DataDefines the hazard

and informs the risks

Risk AssessmentDefines the risks and informs the Basis of Safety

Specific Process

Safety DataDefines and

specifies the Basis of Safety

Figure 2: “Ideal” Approach to Process Safety Data Collection

No Characterisation Data at PHA• During Process Hazard Analysis, it is critical to be able to contextualise the potential impact of identified deviation

scenarios based on an understanding of the base reaction characteristics. If the heat of reaction and / or the stability of the reaction mixture at elevated temperatures is not known, it is impossible to extrapolate the consequences of a given deviation. This leads to inefficiency (“plugging the gap” post-PHA) or unsafe scale-up (consequences left unknown or incorrectly extrapolated based on insufficient data).

No Characterisation Data at PHA

• During Process Hazard Analysis, it is critical to be able to contextualise the potential impact of identified deviation scenarios based on an understanding of the base reaction characteristics. If the heat of reaction and / or the stability of the reaction mixture at elevated temperatures is not known, it is impossible to extrapolate the consequences of a given deviation. This leads to inefficiency (“plugging the gap” post-PHA) or unsafe scale-up (consequences left unknown or incorrectly extrapolated based on insufficient data).

Lack of Thermal Instability Knowledge

• Many organisations focus on heat of reaction (and associated adiabatic temperature rise) as the primary indicator of reaction hazard without taking into account the rate of reaction but the majority of serious global incidents involving reaction hazards arise from the consequences of secondary reactions or thermal instability. Many of these incidents have been associated with entirely foreseeable decomposition of desired intermediate streams with only a small minority being due to more divergent and less foreseeable ensuing chemistries.

Lack of Competence in PHA’s in Extrapolating Reaction Safety Data

• Having the appropriate data available at the start of a PHA is a good starting point, but if the composition of the PHA team does not include a specialist in the understanding or extrapolation of this reaction safety data, erroneous decisions can often result. A good example is when considering the scenario of reaction temperature “too low”. To the uninitiated, this means that the reaction system is further away from boiling or decomposition and is thus safer. Whilst true in some cases, in others low temperature can lead to reduced reaction kinetics and accumulation of unreacted materials during semi-batch (feed rate controlled) processes and a higher risk of runaway once the system returns to normal temperature.

Lack of Robust Basis of Safety

• Some organisations collect an excellent portfolio of reaction safety data – often following a prescribed test program – but then do not translate this to safety system design. During the scale-up process it is common to have chemists driving the initial characterisation data acquisition and then a transition of ownership to chemical engineering colleagues for pilot / production scale-up. If the chemical engineers are not adequately competent in understanding the data they receive, scope exists for inadequate safety system design.

• A further scenario that is regularly observed is an over-reliance on procedural controls for safety critical operations. The tragic incident at Corden Pharma in Cork, Ireland in 2008 graphically illustrates the consequence of reliance on procedural control for a safety critical reaction phase.

• Over-reliance on condensers to remove heat without consideration of failure scenarios of the condenser system often occur. A condenser system whilst improving safety of the reaction, is not a fail-safe panacea.

Incorrect Vent Sizing

• There are so many potential errors associated with adequate vent system design for emergency pressure relief for reactive systems that it is hard to condense and capture them without breaking in to a whole new paper! Common deficiencies in vent sizing include:

• Use of data from inadequate tests. Low thermal inertia, adiabatic, calorimeters should be used for the direct simulation of worst case deviation scenarios to provide scalable rate data. Modelling of runaways can be effective but requires significant mechanistic understanding and physical property data – often sacrificed for more empirical and efficient direct simulation techniques.

• Reliance on sizing calculations for single phase (vapour or gas ) only flow when two-phase flow (liquid plus vapour or gas) will occur in practice and almost invariably requires a considerably larger vent diameter

• Incorrect classification of the vented stream as vapour only, gas only or hybrid (combination). Different sizing methods exist for each type – all yielding different calculated sizes for the same base dataset.

• Incorrect bracing of vent lines for the high thrust forces associated with two-phase (liquid / vapour / gas) flow.

Contact:> China : [email protected]> France : [email protected]> India : [email protected]> Italy : [email protected]

> Netherlands : [email protected] > Spain : [email protected]

> Sweden : [email protected] > UK : [email protected] > USA : [email protected]

PS -

UK

- W

P -

143

- 01

CHILWORTH, THE BACKBONE OF PROCESS SAFETY WITHIN THE DEKRA GROUP Established in 1986, Chilworth now forms the backbone of Process Safety within DEKRA Insight. We offer a vast range of process safety services to most of the major companies in the chemical, agro-chemical, pharmaceutical, food-processing and oil industries. Our range of services includes:

• Laboratory testing: dust and gas flammability, thermal stability, chemical runaway reactions, regulatory testing (REACH, etc.)

• Consulting and training in all areas of process safety: management, culture, audits, risk analyses (HAZOP etc.), accident expertise, major hazards, functional safety, emergency relief systems, compliance to major process safety regulations (SEVESO/ATEX/OSHA) and beyond.

Today our worldwide representation means we can provide our expertise locally to our clients wherever they are and also offer the best of the DEKRA services in testing, inspection and certification.

CONCLUSION

There are a variety of strategies that may be used for the collation of process safety data, but getting the right data in the most efficient way requires knowledgeable professionals who understand what data is needed to assess risks and design appropriate mitigation measures. There are a vast range of pitfalls that can be encountered throughout the data collection, extrapolation and use process which require assured organisational competence at each stage of process scale-up to mitigate.

DR. STEPHEN ROWE

Stephen Rowe manages the activities of the UK headquarters of DEKRA Process Safety (Chilworth Technology Ltd). He has a career background in the assessment of chemical reaction hazards and the laboratory assessment of a full range of process safety hazards including dust, gas and vapour flammability and explosives characterization. He is an experienced trainer and regular contributor to national and international process safety conferences and symposia. As a manager, Stephen Rowe focuses on building successful teams and growing the organization in a customer-centric manner. He oversees and is actively engaged in the company’s quality and safety management systems (ISO9001 and OHSAS18001).

Approaches and Pitfalls in Reaction Safety Data...

Documents

Transcript of Approaches and Pitfalls in Reaction Safety Data...