Seismic shifts6 February 2019
Russell King discusses the latest trends in seismic monitoring and protection systems, the technologies adopted and current best practice driven by these enhanced risk management demands.
FOLLOWING THE FUKUSHIMA ACCIDENT IN 2011 the nuclear industry has placed a renewed focus on risk mitigation from extraordinary events caused by nature.
The various regulatory bodies around the world have gone back to basics, not only to consider the design basis for operational sites, but also to revisit the risk analysis, the accident management strategy and the periodic safety review policy.
Considering that many nuclear power stations are located on the coast, especially in the UK, the risk of flooding, caused by seismic events or severe weather, has been a particular focus. Of particular importance is reviewing the methodology used to derive the seismic hazard, and how that hazard has been mitigated through design process and operational systems. Operational monitoring of seismic vibration for structures and equipment is important in providing automatic shutdown protection and recording seismic events for later analysis.
Seismic monitoring approach
No two nuclear plants have the same approach to seismic monitoring and protection. Some sites utilise data from the national network of geophysical instruments, while others have independent monitoring and shutdown on each critical plant item. In each case, the seismic monitoring and protection strategy must support the safety case and provide appropriate risk mitigation.
The structural effects to be expected at a site from an earthquake are caused by vibration, classified in terms of seismic response spectra. This defines the ground acceleration magnitude versus frequency, typically over a range of 0.1Hz to 100Hz. Two types of spectra are specified, the operational basis earthquake (OBE) and the design basis earthquake (DBE), based on a predicted worst case seismic event within a period of time (the OBE may be specified within 100 years). Secondary response spectra are derived from the ground accelerations by modelling to predict the response of each structure and each level within that structure.
A nuclear plant will specify several seismic categories in the design requirements and assess according to the safety class. For example, the most stringent category will demand the equipment or process be tested to the DBE level, plus a margin (+40% is recommended in IEEE-344, Standard for Seismic Qualification of Equipment for Nuclear Power Generating Stations), since the process must still remain operable even if less critical plant processes failed above the OBE level. Any earthquake above the OBE level may result in the plant being shut down until analysis has determined the plant is safe restart.
The challenge is to design and construct in a cost- effective manner to meet seismic requirements and provide sufficient design margin. It may not be possible for all equipment or processes to be fully categorised and this is where independent seismic monitoring systems can be used to detect an OBE event and bring the process to a safe state. These monitoring systems must be robust to seismic events and be available during a beyond DBE event to maintain a valid alarm function. In combination with the seismic requirements, various safety standards are applied to obtain a stated availability. EN IEC 61508 is the most common approach. Meeting this standard provides system reliability and availability while providing an understanding of the systematic failures and ensuring compliance with the 61508 life cycle model.
The starting point with any seismic monitoring design is the sensor. There is a clear difference between the types of sensors that are used for seismic protection and those used for geophysical earthquake monitoring. Geophysical seismic monitors use broadband, magnet and moving coil (electrodynamic) sensor arrangements capable of measuring ‘micro g’ acceleration events with sinusoidal periods of over 100 seconds. In contrast, strong motion sensors for seismic protection applications only need to provide a resolution down to 1mg and a response to 10 seconds. In the past electrodynamic sensors were used, but piezoelectric-based accelerometers are preferred as they match the technical requirement closely. With no moving parts they are also more reliable.
A trend in vibration monitoring is micro electromechanical systems (MEMS) in a wide range of applications. These have an excellent low frequency response and the required dynamic range for strong motion seismic monitoring. MEMS devices have been widely used in civil engineering applications since the 1990s. Their low cost and small size suit applications where many measurement points are required for a short time. The seismic protection market has been slow to adopt it, because reliability and maintainability are its key requirements. MEMS are a fast-evolving technology and where they have limited applications specific sensors may become obsolete and hard to replace.
Significant earthquakes are rare, so how do we verify an installed strong motion sensor is working correctly, especially when the sensor is difficult to access. With broadband seismometers it is common to have a secondary coil arrangement which can be excited to simulate movement of the mass to verify calibration without physical shaking. Sensonics has incorporated a similar mechanism into its piezoelectric based seismic sensors to ensure the measuring element is operating to the correct sensitivity; this self-test feature is a critical requirement.
It is common to use redundant sensor configurations in the overall monitoring system concept (see Figure 1). Three separate physical locations are monitored with triaxial sensors capable of measuring acceleration in the three orthogonal axes. The acceleration of each sensor is processed by a trip amplifier with the overall triaxial unit performing a ‘one out of three’ logic operation to derive the location OBE alarm. The trip alarms from each location are fed back to the central control panel, which performs a subsequent ‘two out of three’ logic operation to determine the final trip result. In this example, the voting logic is also redundant to enhance reliability and maintainability. The final element of the system is connected to the specific plant circuit breakers, or to the emergency shutdown system, to complete the safety loop.
For redundancy, a simple ‘one out of two’ (1oo2) system will usually meet with the reliability requirements. However, this system configuration offers no protection against spurious trips which can result from mechanical interference or sensor failure. Two out of two (2oo2) is an alternative option that can be considered, however on failure of a channel the system defaults to a 1oo1 system, whilst the 2oo3 option on channel failure can revert to either 2oo2 or 1oo2, both of which are preferred over 1oo1, making the 2oo3 system the norm for nuclear applications.
Combine these channels with dual voting arrangements and the inbuilt test function results in a system design that can be fully proof-tested while on line, maximising the availability of the system. Each voting circuit can be isolated and tested in turn through signal injection of each sensor. The sensor will respond to a real seismic event even while under test.
Avoiding ‘smart’ devices within the protection loop eases the analysis burden to meet safety requirements and is the preferred solution for most clients. Separating the protection and event recording functions is a logical step which enables the latest technologies and features to be used for the seismic waveform recording without affecting the protection safety case.
Using proven technologies in combination with measurement redundancy tends to be the industrial norm for modern nuclear applications, with self-testing features and spurious trip performance of particular importance to automatic shutdown systems. Adopting this best practice has become standard for new installations and should also be considered for obsolete seismic monitoring equipment on existing sites.
A stated and demonstrated reliability, minimal spurious trip occurrence, full measurement loop proof testing, maximum design life and maintainability combined with a low demand and high integrity shutdown system is now expected. Typical nuclear industry applications include reactor structural monitoring, fuel handling, waste processing and safe shutdown of crane equipment.
Author information: Russell King, Managing director, Sensonics Ltd