This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 178821, “Next-Generation Kick Detection During Connections: Influx-Detection-at-Pumps-Stop Software,” by B.A. Tarr, SPE, D.W. Ladendorf, SPE, and D. Sanchez, Shell, and G.M. Milner, SPE, CoVar Applied Technologies, prepared for the 2016 IADC/SPE Drilling Conference and Exhibition, Fort Worth, Texas, USA, 1–3 March. The paper has not been peer reviewed. At least 25% of all influx events on deepwater wells occur while making connections, but few deepwater-rig contractors use kick-detection alarms to alert the driller during a connection. Because of the transient-flow characteristics associated with connections, kick detection during connections is the most challenging to automate effectively. An influx-detection-at-pumps-stop (IDAPS) software was developed to provide early warning of abnormal flowback conditions during connections. Introduction On the basis of an internal review of kick-detection technology, an operator determined that three different well-construction activities must be included in any automated, smart kick-detection system: drilling, making connections, and tripping. A multiyear effort was initiated in 2012 to develop an automated kick-detection system specific to connections. This effort resulted in the development of the IDAPS software that is now in daily use in the operator’s real-time operations centers (RTOCs), where it is part of the routine suite of monitoring programs used during the well-construction phase of offshore wells. Most existing influx- or kick-detection systems designed for RTOC users require frequent interaction with a knowledgeable user, but the IDAPS application was designed to require only minimal interaction to automatically detect connection-related possible influxes, with an associated influx probability from low to confirmed, and to alert RTOC users in real time. IDAPS-Software-Development Process. The software-development program included the following six steps. Definition of functional requirements and performance objectives and establishment of performance metrics. The influx-detection requirements for IDAPS software as defined by the operator as follows are detailed in the complete paper. Provide reliable influx alarms for all historical connection events. Estimate influx volume for all high-confidence influx alarms. Provide clear, brief explanations for data indicating influx alarms. Detect and provide alarm messages for possible sensor problems. Collection of data and labeling “ground truth.” The historical data sets consisted of time-tagged values (at 1- or 5-second sampling rate) of bit depth, hole depth, flow in, flow out, and pit volume for approximately 30 wells drilled over a large range of depth intervals and drilling conditions. Later in the IDAPS development process, a few “blind” data sets were also provided to validate the performance derived from the historical data sets. The provided data included recorded examples for more than 3,000 pumps-off events. However, only 10 of these pumps-off events included a detected influx. Therefore, the most important limitation of the historical data in the software-development program was the limited number of verified connection kicks detected after pumps off.