Figure 3: Decentralized control based on a network of Orbit controllers of an autonomous integrated downstream process with support systems.
Automated support systems
Automated operation of integrated continuous downstream process for a long time means a lot of support work, like preparing buffers and sample flow path for quality analysis and performing the actual analysis. This will actually generate more work than operate the processing system. For this reason, it is vital to automate these procedures in order to create a sustainable long-term lab-scale operation, which are needed for validation of digital solutions. Automated buffer management system and quality analysis system are briefly presented below.
Automated continuous downstream processes operate 24-hours, 7 days a week in a couple of months. During this time, it consumes a lot of different buffers even in lab-scale. After the automation of the actual processing line, it is important to automate the buffer preparation after demand of the process. One such system was implemented and studied in the lab-scale process and it is called buffer management system (BMS), see Isaksson et al 2023. The BMS is composed of four parts; i) monitoring of the buffer container levels at all clients to detect when they pass below an order level, ii) queue system for orders of needed buffers, iii) buffer preparation unit, BPU, and iv) delivery procedure from BPU to the ordering client.
Based on the digital application configuration of the physical system, Orbit keep track of the active flow path and therefore the size of the flow rate generated by the pumps. Orbit also automatically generate estimators of the levels of all containers in the flow path, like buffer flasks, waste containers and viral inactivation reactors. These estimators integrate the inflow and outflow in open loop. The level in the containers need to be updated based of measurements of the real container regularly, in order not to deviate too much. For the accuracy needed for predicting the buffer flask level this is not an important issue, but for more important processing units it can be. Associated to all container level estimators it is possible to add logic to generate events. One example is to trigger an order of more buffer when the estimated level passed below a given value.
The buffer order generated by the level estimator is sent to the buffer ordering queue system, an orbitX instance in Figure 3. The queue is a first-in-first-out (FIFO), but it is possible to rank the buffers in an importance order. The queue controller adds a new order into the queue and removes an order that is delivered. When the buffer preparation unit is ready it gets the first order in the queue from the queue controller. When the unit has delivered the buffer it communicates to the queue controller that the order has been performed, and the queue controller removes the order from the queue. An 11-days long operation of a lab-scale antibody platform is seen in Figure 4, where 55 order was handled by the buffer order queue controller.
BPU is an ordinary setup with a supervisory Orbit controller that on demand perform a procedure to prepare a specific buffer, following a predefined recipe, based on a set of stock solutions. The BPU used in this study can prepare up to 14 different buffers. The stock solutions are mixed in an intermediate buffer flask. The mixing can be operated in closed loop controlling pH and conductivity, but this feature had not been performed in this study. When the buffer is mixed, it is pumped to the client that ordered the buffer. The receiving client directs the buffer delivery valve to the right buffer flask. This required a communication between the buffer preparation unit and the ordering client over the network, see Figure 3. Each sample point in the downstream process has a collection loop on the preparation system.