While best practice has leaned toward keeping control and
safety isolated from each other, recent enterprise data integration
and cost control initiatives are providing incentive to
achieve some level of integration. This paper describes three
basic integration models, including an “interfaced” approach,
in which separate control and safety communicate via a
custom built software bridge; an “integrated but separate”
approach, in which the disparate systems sit on the same
network, but share information only across isolated network
channels; and a “common” approach, in which both control
and safety systems share a common operating system. The
authors then compare the three approaches according to
compliance with safety standards and cost efficiencies.
Report
Share
Report
Share
1 of 6
Download to read offline
More Related Content
Integrated Control and Safety - Assessing the Benefits; Weighing the Risks
1. Integrated Control and Safety —
Assessing the Benefits; Weighing the Risks
by Grant Le Sueur, Director, Product Management, Schneider Electric
and Phil Knobel, Director, Product Management, Schneider Electric
Executive summary
While best practice has leaned toward keeping control and
safety isolated from each other, recent enterprise data inte-gration
and cost control initiatives are providing incentive to
achieve some level of integration. This paper describes three
basic integration models, including an “interfaced” approach,
in which separate control and safety communicate via a
custom built software bridge; an “integrated but separate”
approach, in which the disparate systems sit on the same
network, but share information only across isolated network
channels; and a “common” approach, in which both control
and safety systems share a common operating system. The
authors then compare the three approaches according to
compliance with safety standards and cost efficiencies.
2. Integrated Control and Safety — Assessing the Benefits; Weighing the Risks
Safety instrumented systems are industrial safety nets. They must be available 24/7 to provide backup
when something renders a process automation system unable to perform its job of controlling a
hazardous process. To protect the safety instrumented systems (SIS) from faults caused by the same
issue that caused the process automation system to malfunction, good practice has traditionally
dictated strict physical and functional isolation between the two systems. But as increasing business
complexity and global competitiveness drive often conflicting needs for greater enterprise integration,
improved safety and reduced costs, officials at some companies are looking at integration and consol-idation
of safety and control function as a way out of the dilemma.
Safety and risk managers might see the safety system as a goldmine of valuable data, which, if made
more accessible, could help identify leading indicators of future problems. Engineering managers see
redundant effort which can be streamlined. Operations managers see islands of activity which can be
better communicated with each other and with rest of the enterprise. Maintenance managers see vol-umes
of data on machine and system health that can contribute to improved maintenance and lower
maintenance costs and financial managers see redundant capital expenditures and training costs ripe
for consolidation into a single system.
In efforts to address these needs, automation vendors have offered various models for integrated
control and safety systems (ICSSs). This paper compares the benefits and risks at four levels of
integration: complete physical separation, integration via a custom programmed software interface,
integration via isolated subsystems on a client-server control network and integration across a
common control platform.
Although both the process automation and safety instrumented are control systems, they are designed
for fundamentally different purposes. The PAS, which is also often called a distributed control system
(DCS) or basic process control system (BPCS), regulates production based on values of production
variables received from field devices such as pressure and temperature transmitters, via I/O cards
terminating in a control room. A PAS also incorporates an engineering environment and tools used to
configure and maintain it. Users interact via a human machine interface (HMI).
Safety instrumented systems (SISs) also provide control based on signals received from field devices;
but unlike PASs, which are optimized to handle high volumes of complex process logic, SISs are
applied to provide safe and orderly shutdown of operations that might otherwise fall under the control
of the PAS. When applied for this purpose, SISs are also called emergency shutdown systems (ESDs.)
For the highly critical ESD function, SISs are optimized for speed and reliability. The control elements
are usually redundant, high speed, programmable logic controllers (PLCs) that have been heavily
tested and certified for reliability.
Virtually all medium to large companies processing hazardous materials or running otherwise potentially
dangerous operations will implement an SIS to back up their PAS. These systems provide indepen-dent
control of a process operation, typically using dedicated field devices, I/O, networks, engineering
workstations, configuration tools and HMIs. This is by far the dominant approach taken throughout the
world. And more often than not, the PAS and the SIS have been from different vendors.
Efforts to make more strategic use of safety operation information or to save money through consol-idation
of safety and control functions have led to the emergence of a number of integrated control
and safety system models. In its 2013 “Process Safety Systems Global Market Research Study,”
ARC identifies 4 levels of control/safety integration: separate, interfaced, integrated but separate, and
common. We will look at each option in more detail and evaluate it according to its impact on safety,
productivity and cost control.
Introduction
The difference
between a PAS
and an SIS
2
3. Integrated Control and Safety — Assessing the Benefits; Weighing the Risks
Ask most safety engineers for their preferred level of integration and most would opt for no integration
at all. That is what Schneider Electric found in a 2010 survey of more than 200 Schneider Electric (then
Invensys) customers, including 23 of the top 25 petroleum companies and 45 of the top 50 chemical
companies in the world. 78 percent adhered to strict separation of safety and control for safety
protection and 74 percent indicated that independent protection layers (IPL) were critical.
Although the leading standards influencing process safety, IEC 61508 and IEC 61511, have been
somewhat ambiguous regarding integrated control and safety, there is no doubt that implementing
systems separately satisfies requirements for the independent layers of protection necessary to ensure
that a potential hazard could not occur unless both the DCS and SIS fail.
Separate systems also comply most completely with IEC 61511-1 11.2.4 sections that dictate that
the process automation system shall be designed to be separate and independent to the extent that
“the functional integrity of the SIS is not compromised” and IEC 61511-1 clause 9.5, which addresses
the requirements for preventing common cause, common mode, and dependency failures, suggesting
consideration of the following criteria:
• Independency between protection layers
• Diversity between protection layers
• Physical separation between protection layers
• Common cause failures between protection layers and the DCS
But, because separation does require implementing, operating and maintaining two different systems,
it can also be the most costly route. Also, because operating data is so strictly isolated, there may be
lost opportunities for improvements in maintenance, troubleshooting and trend analysis.
Interfaced systems still maintain a high degree of separation, but the DCS and SIS exchange informa-tion
through custom designed interfaces using standard integration protocols such as OPC, Mod-bus,
PROFIBUS, Profinet, TCP and HART. These are used most commonly when control and safety
systems are from a different vendor and the end-user needs the systems to share specified data for a
specified purpose.
Assuming that the systems integrators who build the interface have adequate expertise in working
with safety systems, this could be a very safe approach. However, the information that it yields, will be
limited to the specification. Additional ongoing maintenance and subsequent change could be costly.
And the integrity of the gateway will not likely have been subjected to third party validation.
In the third model, which ARC has labeled “integrated but separate,” the safety and control logic
solvers are deployed on independent buses of the control network. Clients can share process data
across isolated sub networks but do not share control functionality. In Schneider Electric’s Foxboro
Evo™ process automation system, for example, the safety controllers are deployed as peers on a
Foxboro Evo MESH control network (Figure 1).
This model formats all data to flow natively between network channels that are physically isolated with
one-way communications maintained by a communications module (Figure 2). This example is
“integrated” in that companies who want to integrate control and safety data or who want to take
advantage of other productivity and cost efficiencies can do so safely. But it is “separate” in that all
functionality is implemented on separate devices and the system can be configured as an entirely
separate system.
Maintaining
separate control
and safety
architectures
Interfaced ICSS
architectures
3
Integrated but
separate ICSS
architecture
4. Integrated Control and Safety — Assessing the Benefits; Weighing the Risks
Generally, integrated but separate control and safety systems are viewed as compliant with IEC
standards for independent layers of protection, because the network channels are independent and
threats to one system will not affect the other. Safe access to data enhances safety, productivity and
cost savings by providing a fully integrated user experience, including sequence of events recording,
system management, engineering and maintenance.
• Integrated sequence of events repository. Seamless integration of PAS and the SIS enable
shared sequence of event (SOE) logging. In the Foxboro Evo integrated but separate implementa-tion,
for example, sequence of events logs and system diagnostic logs are recorded into the same
data repository managed by the Foxboro Evo enterprise integration control software platform.
Logging all SOE events into the same repository provides end users with a more convenient way
to perform a post trip analysis. They can use common tools to review them and identify the true
root cause of a trip event more effectively.
4
Figure 1
Foxboro Evo integrated
but separate control and
safety system
Figure 2
In the Schneider Electric
Foxboro Evo implementation
of integrated but separate
control and safety, network
channels are physically
isolated with one-way
communications maintained
by a communications module.
Users can choose the level of
integration that meets
their needs, from fully
integrated to complete
and total separation.
5. Integrated Control and Safety — Assessing the Benefits; Weighing the Risks
• Integrated system management. In integrated but separate architectures (Figure 3), all of the
capabilities of field diagnostics and asset management, including partial stroke testing, can be
implemented more effectively, simplifying actuator testing and avoiding false trips. Such extensive
system diagnostics and system management capabilities provide end users a single application
point of view from which they can view the state of the entire system and, if required, acknowl-edge
system alarms. It also minimizes the number of steps it takes to get information from the
safety system to the operator; and the fewer the number of steps, the less likely that mistakes will
occur. This also simplifies operator training.
Management of safety instrumented functions would be easier because diagnostics can be sent
from sensors to control elements. HART device alerts, for example, can be sent to operators and
maintenance personnel as early warning of problems with the device or surrounding process.
Predictive testing can help avoid spurious trips on demand.
• Integrated engineering workflow. Integrated workflow would ensure that changes in any new
tags that might be created in SIS user logic become immediately available to the PAS for use with
linking to graphics or historization functions, or to drive interlocking permissions that the PAS
might use in a broader control scheme.
Project engineers would also enjoy a single point of entry and the use of common tools to
configure both safety and process control systems, reducing time to start up new installations.
Common programming procedures, languages, and installation requirements boost productivity
further. Systems engineers would also enjoy improved alarm handling, time synch, user access
and authorization management; and mapping of data would no longer be necessary.
• Integrated compliance. The repository, system management and workflow function of integrated-but-
separate architectures can also assist with compliance with regulations and standards.
Integrated systems provide better device audit trails, including calibration history, process and
safety configurations, and process and event histories. Both document and change
management will be easier.
Because the integrated but separate approach still requires installing, maintaining and configuring
what are essentially separate systems, there would be minimal cost reduction on the technology end,
although there might be some economies in communications technology. The greatest financial bene-fits,
however, are in attainment of information, configuration, asset management and HMI efficiencies,
without jeopardizing safety.
5
Figure 3
The Foxboro Evo integrated
but separate architecture
provides end users a single
application point of view
from which they can view the
state of the entire system.