Learnings

Outcomes

Progress 2019-2020

Historical and prospective gas composition data was needed from the Delivery Facility Operators (DFOs) at Bacton and St Fergus to be able to conduct phases 1 of the study. The data is commercially sensitive, therefore a confidentiality agreement (CA) was required, which was finally signed off in February 2020. 

The first phase of the project was to create and validate hydraulic models of the Bacton and St Fergus terminals using SPS software, which is widely used for online and offline modelling of single-phase gas flow within transmission systems. The Bacton model was completed and at the end of 2019/20 the St Fergus model was in progress. 

Closure 2020/21

Phase 1

The model of St Fergus was completed and both terminal models were validated against available recorded data, i.e. actual historical flows and gas quality from the incomers to the terminals was entered and the results observed at the exit points from the terminals.  This simulated gas composition data at the exit points mostly matched the actual measured historical gas composition data for the different validation conditions. Discrepancies were attributed to incomer composition data where a daily average was given rather than the specific hourly values used on other flows.

Phase 2a 

The hydraulic models were then used to run scenarios to blend specific incoming pipelines’ gas qualities.  A mix of projected flow and qualities for gas year 2022-23 and sensitivity cases outside the GS(M)R specification were used as inputs for the inputs from the UKCS terminals and flow configurations through the terminals were defined.  These scenarios included steady state (all incomers at constant rates) and transient cases (starting at a steady-state and then simulating a trip of flow from one incomer). It was recognised that the volume of flows through the interconnectors and whether they are in import or export mode are commercially driven and therefore much more difficult to predict, therefore assumed flows and compositions were derived from historical data. The Bacton scenarios modelled were:

1 Steady state central forecast (GY 2022/23)
2 All low case Wobbe Index, central case flow 2022/3, interconnectors importing
3 As scenario 2, interconnectors exporting
4 Non-compliant low Wobbe Index scenario at Perenco with Shell low Wobbe Index, SEAL low Wobbe Index, using low flow case for gas year 2022/23. IUK and BBL in exporting​ mode. (i.e. the least helpful scenario to blend out non-compliant Perenco gas)​
5a Transient - Scenario 4 plus Shell stops flowing
5b Transient - Scenario 4 plus SEAL stops flowing
6 Scenario 4 plus IUK and BBL importing to GB ​ (i.e. to test if the interconnectors can help blend up the Wobbe Index of gas from the Perenco incomer)​

The results for Bacton showed that the gas quality at all exit points from the terminal for all scenarios was GS(M)R compliant except for the Great Yarmouth power station offtake (GYPS) in scenarios 4-6.  This was because in the flow configurations modelled the GYPS offtake is directly downstream of the Perenco incomer so no commingling with on-spec gas occurred.  Since not all input sources comingle within Bacton terminal,  a homogeneous flow-weighted Wobbe Index at all exit points from the terminal is not achieved, which presents a risk to the operation of a blending arrangement because National Grid could not predict with confidence that all outlet points would be on-spec.  There are several different configurations that flows through Bacton can operate to which could not all be modelled and the uncertainty of the interconnectors’ flow-rate and direction are also significant variables outside of National Grid’s control.  

The St Fergus scenarios modelled were:

1 Steady state central case forecast for flow and quality for all 3 DFOs (gas year 2022/23)
2 Ancala high case Wobbe Index, high flow case, Shell low flow case, high Wobbe Index case, NSMP high case Wobbe Index, low flow case – all 2022/23
i.e. the least suitable flows and quality from NSMP and Shell to blend with Ancala gas
3 Ancala off specification scenario on ICF and Wobbe Index, maximum flow for 2022/23, Shell & NSMP – low case flow 2022/23, high case Wobbe Index 
i.e. the least suitable to blend Ancala gas
4 NSMP off-spec scenario (NSMP provided) on Wobbe Index and flow, Shell and Ancala – low case flow 2022/23, low case Wobbe Index
i.e. the least suitable to blend out of spec NSMP gas
5a Transient - Scenario 3 + NSMP stopping flow
5b Transient - Scenario 3 + Shell stopping flow
6a Transient - Scenario 4 + Ancala stopping flow
6b Transient - Scenario 4 + Shell stopping flow
 

The results for these scenarios all showed that a homogeneous blended gas was present at all terminal outlet points onto the NTS feeders, which were all within the GS(M)R specification.  Unlike Bacton, all the incomer flows were routed through a common mixing area before exiting the terminal which delivers the homogenous blended gas.   Routing through the mixing area is normal operational practice at St Fergus unless there are operational problems or maintenance requirements that impact this. 

Scenario 6a, which simulated the loss of the Ancala supply when off-spec low Wobbe gas was being delivered from NSMP came close to breaching the lower GS(M)R limit but remained just above it.  However, it was noted that the transit time of gas through the terminal is short and that gas analyser sample frequencies may be insufficient to detect an off-spec combination, should that occur within the terminal.  In addition, sample timings across the analysers that measure gas quality as it enters and exits the terminal may not be  harmonised, making an accurate real-time determination of the flow-weighted Wobbe Index and ICF difficult to establish.  

The key limitations of the study were:

·         The scenarios assumed fixed flow-rates and gas qualities whereas in reality, natural gas entering the terminals varies in composition at source and resulting from both onshore and offshore processing requirements. As a result, gas quality parameters are in continual flux and can range from normal variances to ‘spikes’ caused by events such as processing failures and occasional gas quality instrument error, which can require a rapid operational response to try to mitigate off-spec gas reaching an outlet point from a terminal.

·         No modelling of process failures upstream of the National Grid terminals has been carried out. These are typically infrequent events but when they do happen could have a material impact on deliverability through the terminal and hence the availability of a blending service. For example, if contaminants are introduced to the terminal from one incomer causing a gas quality analyser to malfunction, it can also impact the measurement accuracy of other analysers that are linked to it by a common ‘slow’ maxum measurement across all the incomers to the terminal.   

·         The practicality of managing gas quality compliance from an operational perspective e.g. systems, processes, measurement requirements, interfaces etc. and any change to current procedures and practices was also outside the scope of this study.

·         The modelling necessarily had to assume particular flow configurations and it is noted that, in relation to Bacton in particular, in operational practice these are subject to change depending on the pressure requirements for feeding gas deeper into the NTS. Due to the location of Bacton and the multiple feeder flow path directions away from the terminal, its configuration is subject to frequent change to  meet wider NTS operational pressure and exit feeder flow requirements as well as to mitigate any potential gas quality excursion, where required. These configurations often require routing and prioritising gas flow from specific incomers either partially or wholly onto specific feeders and different configurations would give different results at the outlet points from the terminal. There is also a requirement to ensure CV variance across the feeders is within tolerable levels for CV shrinkage/billing purposes.

Phase 2b

Phase 2b consisted of a Reliability/Availability/Maintainability (RAM) study of both terminals to determine their expected capability to transmit gas from input to outlet in the 2a operating scenarios.  

For Bacton, the results showed a very high availability percentage when gas being delivered was GS(M)R compliant. This percentage was subject to reductions for the off-spec and transient disturbance scenarios because of the inability of Bacton terminal to supply the GYPS offtake with compliant gas based on the chosen flow configuration and consequent need to curtail flows from the Perenco entry point.  National Grid measures gas quality at the terminals using both ‘fast’ microsam and ‘slow’ maxum instrumentation.  The study assumed that both ‘fast’ and ‘slow’ gas quality measurements would be required to be operational if non-GS(M)R compliant gas was expected to be delivered in a blending scenario.  This resulted in a marginal reduction in deliverability (i.e. the ability of terminal to receive, process and export gas from all DFOs) compared to on-spec scenarios where gas flow can continue to flow if one analyser were to fail.   

Results for St Fergus similarly showed a high percentage availability if all gas coming into the terminal is GS(M)R compliant.  In scenarios where one of the incoming flows is non-GS(M)R compliant, the terminal deliverability is reduced due to tighter restrictions on blending, where there are no options to bypass the mixing area and both GQ analysers (fast and slow) are required on incomers and feeders to monitor gas quality through the terminal. The reduction was smaller than observed at Bacton, however in a scenario where one of the incoming flows is stopped such that a compliant blend cannot be achieved, the deliverability is subject to a much more significant reduction. 

Lessons Learnt

Progress 2019-2020

·         As mentioned above, the Confidentiality Agreement proved logistically challenging to agree and execute by multiple organisations. In future a different approach could be to facilitate a meeting between all the legal teams within the multiple companies to reach consensus as a collective which would reduce delays to the signatures. This would prevent the back and forth email communications which caused delay and could create more immediate decisions on its terms.   

·         Due to the delays with National Grid obtaining data for phase 1, the data for phase 2 was requested earlier than originally planned.  This could have been applied to the data requests for Phase 1 too.  A general learning point for future projects therefore is that identification of data required and requests for that data should be prioritized early in the project.   

Closure 2020/21

·         For future gas quality studies where gas compositions are provided to derive a parameter such as Wobbe Index, it is suggested to check that the gas compositions match the declared Wobbe Index sooner in the study.  This would have enabled engagement with the DFOs to highlight the discrepancies earlier and potentially resolve them before the hydraulic scenarios were modelled.

·         Before embarking on a study such as this in future, it would be beneficial to seek views on the operability of the envisaged revised arrangements.  Scenario modelling can deliver useful outputs to validate hypotheses but it is also important to consider how the proposed revisions to ways of working are achievable in practice. 

·         It would have been beneficial to seek future projections of gas composition from DFOs that were outside the GS(M)R range at an earlier stage of the project.  This was requested after the initial set of compliant data was obtained which resulted in additional work for the DFOs and some delay.

·         It would have been beneficial to give greater consideration to the scenarios to be modelled before the project began.  With hindsight, these were inadequately defined at the start of the project to give meaningful results and had to be re-worked between phases 1 and 2a.  Whilst this did not result in any material delay, it but did require adjustments to be made mid-project and additional engagement with DFOs to sense-check the revised scenarios.