Our client was looking to optimise the performance of its biogas engines. The objective was to reduce natural gas use and biogas flaring, and reduce the instances of engine tripping. The result is the biogas engines can be run on automatic- requiring less operational personnel – and in the main stoppages only result from natural gas supply constraints and maintenance schedules with positive cost benefits.
A full version of this case study including diagrams can be downloaded here.
Experience has shown us most pump stations are equipped with enough measurement instrumentation to make online pump efficiency calculations possible. Therefore, it is a relatively simple process for any business to achieve power savings by monitoring pump efficiency.
Our client wanted us to examine the unstable running of its biogas engines in order to optimise its control system. More often than not the client would run the process manually – causing excessive gas flaring or tripping of the engines due to gas starvation. Our objective was to reduce both natural gas usage and biogas flaring, and so reduce the instances of engine tripping. The result would be a positive cost benefit with less effort required by operational personnel to run equipment.
The problems our client was having stemmed from rapid and uncontrollable fluctuation in the low pressure gas holder. This was a major symptom caused by the blowers ramping up or down, engines starting and stopping, and variability in biogas production.
At startup the control of the blower pressure would ramp up to 80% and a second blower would start at the same output percentage as the first. Almost twice the amount of gas would be transferred from the low pressure to the high pressure gas side causing a sudden and drastic drop in the low pressure gas holder level.
Their engine biogas blending control was based on trying to maintain a set level in the low pressure gas holder. This strategy completely ignored the variation in biogas production and biogas supply pressure.
The sheer number of permutations and combinations of equipment in this complex operation also caused problems. Each permutation and combination has a different open loop transfer function so a control strategy had to be designed to overcome this.
Plus, the equipment was often left in ‘manual’ operation which meant unnecessary flaring of biogas.
Optimising the control system for the biogas engines had two goals:
Process stability – To get the engines running in ‘auto’ and reduce the trips and start ups, and
Process optimisation – Once the process was stable we could minimise biogas flaring and natural gas usage.
Our main constraints were to take into account the protection interlocks built into the blowers, and the way the engines responded to tripping out.
The overall strategy for controlling the blending engines were changed based on the following principle:
The control will balance the biogas inflows to the outflows. Due to errors in the 11 flow meters there will always be an imbalance between the real biogas inflow and outflow when the two totals are reading equal. The physical discrepancy between the two will cause the gas holder level to either increase or decrease.
We added a level controller to offset the difference in measurement between inflow and outflow. Then a level controller output was added to the measured flow rate to produce a set point for the biogas PID loop. The output of the level controller was scaled to produce a trim effect.
Because the open loop transfer function changes depending on which combination of engines are running, five different gains and integral time constants were programmed into the system. The system detects which combination is running and then passes the relevant PID tuning parameters to the controller. This was done for both the biogas and low pressure gas holder level controllers.
Tuning remains a challenge because the system was inherently unstable and the two variables (flow and level) could not be isolated from each other to determine process gain and time constants. So the combinations catered for are:
One blending engine running
One blending and one fixed engine running
Two fixed and one blending engine running
One fixed and two blending engines running and
All four engines running
Getting the blower to respond to the processes better was a major concern. The long pipelines between the blowers and the engines created deadtime so we implemented a cascade controller. Its output was a setpoint to the slave controller which controlled the immediate blower controls looking at the pressure at the blower discharge.
A similar strategy was adopted to overcome the complication where the open loop transfer functionmeant hree different gain and integral time contestants were passed to the slave pressure controller depending on whether one, two or three blowers are in operation.
Interlocks were set to trip out the blowers instantaneously. A de-bounce time of 30 seconds was added to each interlock to ensure the interlock condition was not transitionary. The effect was a 66% reduction in tripping instances.
The end result means the engines are now left in automatic control. Trips of blowers or engines rarely occur, and engines are mainly stopped because of natural gas supply contract constraints or for maintenance reasons.
